Golriver

Fourth Edition
BEHAVIOURAL ECOLOGY An Evolutionary Approach EDITED BY
John R. Krebs
Chid ExeCUlive of tht.' Nalural Environment R~search Council and Royal Society Research ProCessor Department of Zoology
University of Oxford
AND
Nicholas B. Davies Professor of Behavioural Ecology
Dcpanmelll of Zoology University of Cambridge
flJ
Blackwell Publishing
Fourth Edition
BEHAVIOURAL ECOLOGY An Evolutionary Approach EDITED BY
John R. Krebs
Chid ExeCUlive of tht.' Nalural Environment R~search Council and Royal Society Research ProCessor Department of Zoology
University of Oxford
AND
Nicholas B. Davies Professor of Behavioural Ecology
Dcpanmelll of Zoology University of Cambridge
flJ
Blackwell Publishing
@1978, 1984, 1991, 1997 by Blackwell Science Ltd, a Blackwell Publishing company BLACKWELL PUBLiSHING
350 Main Street, Malden, MA 02148-5020, USA 9600 Garsington Road, Oxford OX4 2DQ UK 550 Swanston Street, Carlton, Victoria 3053, Australia
11le right of the Author to be identified as the Authors of this Work has been asserted in accordance with the UK Copyright, Designs, and Patents Act 1988.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical. photocopying, recording or otherwise, except as permitted by the UK Copyright, Designs, and Patents Act 1988, without the prior permission of the publisher. First published 1978 Second edition 1984 Third edition 1991 Fourth edition 1997 11
2008
Library a/Congress Cata/oging-in-Pllblication Data Behavioural ecology: an evolutionary approach I edited by J. R. Krebs, N. B. Davies. - 4th ed.
p. em. Includes bibliographical references (p. ) and indexes.
ISBN 978-0-S6S42-731'{) 1. Animal behaviour. 2. Animal ecology. 3. Behaviour evolution. I. Krebs,}. R. Oohn R.) U. Davies, N. B. (Nicholas B.), 1952-. QL751. B34S 1996 591.5'1 - dc20 96-30486 A catalogue record for this title is available from the British Library.
Set by Setrite Typesetters, Hong Kong Printed and bound in Singapore by COS Printers Pte Ltd The publisher's policy is to use permanent paper from mills that operate a sustainable forestry policy, and which has been manufactured rrom pulp processed using add-free and elementary chlorine-free practices. Furthennore, the publisher ensures that the text paper and cover board used have met acceptable environmental accreditation standards. For further information on Blackwell Publishing, visit our website: www.blackweUpublishing.com
Contents
Contributors, v Preface, vii Acknowledgements, viii
Part 1: Introduction The Evolution of Behavioural Ecology, 3 John R. Krebs & Nicholas B. Davies
Part 2: Mechanisms and Individual Behaviour Introduction, 15 2
Sensory Systems and Behaviour, 19 Riidiger Wehner
3 The Ecology of Information Use, 42 Luc-Alain Giraldeau 4
Recognition Systems, 69 Paul W. Sherman, Hudson K. Reeve & David W. Pfennig
5
Managing Time and Energy, 97 Innes C. Cuthill & Alasdair I. Houston
6
Sperm Competition and Mating Systems, 121 Timothy R. Birkhead & Geoffrey A. Parker
iii
iv
CONTENTS
Part 3: From Individual Behaviour to Social Systems Introduction, 149 7
The Evolution of Animal Signals, 155 Rufus A. Johnstolle
8
Sexua I Selection and Mate Choice, 179 Michael J. Ryall
9 Sociality and Kin Selection in Insects. 203 Andrew F.G. Bourke 10
Predicting Famjly Dynamics in Social Vencbrates. 228 Stephen T. Emlen
II
The Ecology of Relationships, 254 Anne E. Pusey & Craig Packer
12
The Social Gene, 284 David Haig
Part 4: Life Histories, Phylogenies and Populations Imroducrion. 307 13
Adaptalion of Life Histories, 311 Serge Daan & Joost M. Tillbergen
14
The Phylogenetic Foundations of Behavioural Ecology, 334 Paul H. Harvey & Sean Nee
15
Causes and Consequences of Population Structure, 350 Godfrey M. Hewitt & Roger K. BIIIlin
16
Individual Behaviour, Populations and Conservation, 373 John D. Goss-Cuslard & William J. SlItherland References, 396 Index, 447
Contributors
Timothy R. Birkhead
Departmenr of Animal and Plant Saences. University of Sheffield, Sheffield. S10 2TN. UK
Andrew F.G. Bourke
Institute ofZoology, Zoological Soaety of London. Regm!'s Park. London. NWI 4RY. UK
Roger K. Bullin
Departmel1l of Biology. University of Leeds. Leeds, LS2 9JT. UK
Innes C. Cuthill
Behavioural Biology Group. School of Biological Saences. University of Bristol. Woodland Road. Bristol. BS8 1 UG. UK
Serge Daan
Zoological Laboratory. UniversityofGroningen, PO Box 14, 9750AA Haren. The Netherlands
Nicholas B. Davies
Department of Zoology, University of Cambridge, Downin,q Street, Cambridge, CB23£J, UK
Stephen T. Emlen
Section of Neurobiology and Behavior. Division of Biological Sciences, Seeley G. Mudd Hall, Cornell University. Itltaca, NY 14853·2702, USA
Luc-Alain Giraldeau
Departmel1lofBiology, Concordia University, 1455 ouest, Boulevard de Maisonneuve, Montreal, Quebec, H3G JM8, Canada
John D. Goss-C us ta rd
Instilllte of Terrestrial Ecology. Funebrook Research Station. Wareham, Dorset, BH20 5AS, UK
David Haig
Museum ofComparative Zoology. Harvard University, 26, Oxford Street Cambridge. MA 02138. USA
Paul H. Harvey
Department ofZoology, University of Oxford, South Parks Road, Oxford, Ox t 3PS, UK
Godfrey M. Hewitt
School of Biological Sciences, University of East Anglia.
Norwich. NR4 7TJ, UK
v
vi
CONTRIBUTORS
Alasdair I. Houston
Behavioural Biology Group, School of Biological Sciences, University of Bristol, Woodland Road, Bristol, BS8 I UG, UK
Rufus A, Johnstone
Departmentofloology, University ofCambridge, Downing Street, Cambridge, CB2 JEJ, UK
John R, Krebs
Departmelllofloology, University of Oxford, South Parks Road, Oxford, Ox I JPS, UK
Sean Nee
Departmentofloology, University of Oxford, South Parks Road, Oxford, OXI JPS, UK
Craig Packer
Departmelll of Ecology, Evolueion and Behavior, University of Minnesoea, 100, Ecology Building, 1987 Upper Buford Circle, Saint Paul, MN 55108, USA
Geoffrey A, Parker School of Biological Sciences,
University of Liverpool,
Liverpool, L69 JBX, UK
David W, Pfennig
Deparement of Biology, Caker Hall, University of North Carolina, Chapel Hill, NC 2 7599, USA
Anne E. Pusey
Deparemene of Ecology, Evolueion and Behavior, Universieyof Minnesoea, 100, Ecology Building. 1987 Upper Buford Circle, Saint Paul, MN 55108, USA
Hudson K. Reeve
Seerion of Neurobiology and Behavior, Division of Biological Sciences, Seeley G. Mudd Hall. Comell University, IChaca. NY /485J-2702, USA
Michael J_ Ryan
Departmentofloology, University of Texas. Austin. TX 78712.
USA
Pa ul W, Sherman
Seerion of Neurobiology and Behavior, Division of Biological Scif/!ces, Seeley G. Mudd Hall. Cornell University. IChaca, NY 1485J-2702. USA
William J, Sutherland
School of Biological Sciences. Universiey of Eas/ Anglia,
Norwich, NR4 ITJ. UK
Joost M. TInbergen
Zoological Laboraeory, University ofGroningf/!, PO Box 14, 9750 AA, Haren, Tlee Nelleerlands
Rudiger Wehner
Department ofloology, University ofZurich, Winterchurerstrasse 190, CH-8057. Zurich, Switzerland
Preface
As with each of the three previous editions of Behavioural Ecology, we have brought together a completely new sel of chapters for this volume. This is an exciting time for the subject. Stronger links are being forged with studies of mechanisms and how they control and constrain the adaptive behaviour of individuals. New molecular phylogenies are being used to unravel the evolution of behaviour and life histories. Molecolar techniques for measuring parentage have revolotionized empirical sludies o[ breeding systems and have thrown up many surprises, especially the high levels of 'extrapair' matings in species previously supposed to be monogamous. New models are being developed to study behaviour sequences and signalling systems. There are also new ideas linking immunocompetence and mate choice. Atthe same time, the traditional boundaries of the subject are being extended to consider how individual decision making influences population structure and its consequences for lhe conservation of species and habitats. We have tried to incorporate these developments along with the familiar topics from previous editions by organizing the book in three new sections. We have aimed for a fresh approach by choosing new authors or new subjects. We asked everyone to provide a review of the main ideas and empirical data to test them, and to focus on current controversies and unsolved questions. As with the previous editions, Ihe book is intended for graduate and upper level undergraduate courses, where students are already familiar with the basic ideas in behavioural ecology. We hope that these chapters will inspire the next generation of behavioural ecologists to strive for improvements in the theories and in the data to lest Ihem. Since the first edition of this book (1978), we have produced new editions at 6- to 7-year intervals (1984, 1991, 1997). The best measure of Ihe subject's continuing success would be the need for a fifth edilion in another 7 years' time.
John R. Krebs Nicholas B. Davies
vii
Acknowledgements
From Blackwell Science Lid. we thank Robert Campbell and Susan Sternberg for the initial encouragement to edit a fourth edition of this book and Ian Sherman and Karen Moore (or their expert advice and enthusiasm during its
production. Authors of various chapters would like to thank the following for their help. Chapter 2
Simon Laughlin.
Chapter 4
George J. Gamboa. Warren G. Holmes. Laurent Keller. Jill M. Mateo. Karin S. Pfennig.
Chapter 5
John Hutchinson. Sonia Lee. Neil Metcalfe. John McNamara. Mark Witter and the Natural Environment Research Council and Biotechnology and Biological Sciences Research Council.
Chapler 7
Rebecca Kilner and Naomi Langmore.
Chapter 9
George Chan. Tim Coulson. Jeremy Field. William Foster, Laurent Keller. (an Owens, Yves Roisin
and Roland Stark. Chapter 12
Helena Cronin. Alan Grafen. Laurence Hursl. J. McKinnon. Naomi Pierce. S. Porter.
Chapter 16
Paul Marrow.
Once again. we thank Jan Parr for drawing the vignettes which accompany some of the figures in this book.
viii
PART I INTRODUCTION
A starling Stumus vulgaris rtlllrns to its nest to feed its hUlJgry brood. Why art the chicks beggllJg so noisily? What are the proximate and ultimare factors which influence the parent's choice ofprey. irs provisioning rart and its clutch size? IPholograph by Eric and David Hosking.)
Chapter 1 The Evolution of Behavioural Ecology John R. Krebs & Nicholas B. Davies
1.1 Observations and questions All natural history observations begin with a question. AI first our curiosity may be satisfied simply by knowing the species name of Ihe animal we are watching. Then we may want to discover what il is doing and to understand why it is behaving in a particular way. In 1978. we began the first edition of this book by asking the reader to observe a bird. such as a starling (Sturnus vulgaris). searching in the grass for food. The starling walks along. pausing every now and then to probe into the ground. Sometimes it succeeds in finding a prey item. such as a beetle larva. and eventually. when it has collected several prey. it flies back to the nest to feed its hungry brood. For students of behavioural ecology, a whole host of questions come to mind as they observe this behaviour. The first set of questions concern the way the bird feeds. Why has it chosen that particular place to forage? Why is it alone rather than in a flock? What determines its choice of foraging path? Does it colJect every item of food it encounters or is it selective for prey type or size? What influences its decision to stop colJecting food and fly back to feed its chicks? Another set of questions emerges when we folJow the starling back to Ihe nest. Why has it chosen this site? Why this brood size? How do Ihe two parent starlings come to an agreement over how much work each pUIS into offspring care? Why are the chicks begging so noisily? Are they each simply signalling their own degree of hunger or are Ihey competing for food? Why such costly begging behaviour? If we observed Ihe starling over a longer period of lime then we may ask about what determines how much effort il puts into reproduction versus ils own maintenance, about Ihe factors influencing the timing of its seasonal activities. its choice of male. and so on. Behavioural ecology provides a framework for answering these kinds of questions. It combines ideas from evolution, ecology and behaviour and has emerged from five schools of thoughI, developed primarily in the 1960s and early 1970s. We discuss them in turn to provide a brief history of the subject and to show how the ideas have evolved in the lasl20--30 years, and we puint out how this book reflecls recent developments.
3
4
CHAPTER I
1.2 Tinbergen's four questions TInbergen (1963) showed that there are four ways of answering the question 'why?' in biology. Returning to our starling, il we asked why it loraged in a particular way we could answer as follows. I In terms olfunction, namely how patch choice and prey choice contribute to the survival 01 the bird and its offspring. 2 In terms of causation, namely the proximate factors which caused the bird to select a foraging site or prer type. These may include cues to prey abundance, such as type of soil or vegetation, or the activities 01 othet birds. 3 In terms of development. This answer would be concerned with the role of genetic predispositions and learning in an individual's decision making. 4 In terms of evolurionary history, namely how starling behaviour has evolved from ils ancestors. This answer might include an investigation of how the starling family has radiated 10 fill panicular ecological niches and the innuence of competition Irom other animals on the evolution of starling behaviour and morphology (e.g. bill size, body size). Tinbergen's slUdies on gulls aimed to combine the four kinds of answers and he emphasized the need to study animals in their natural surroundings, namely those where their behaviour had evolved. He championed the use of the field as a natural laboratory for observations and controUed experiments and showed how ideas can be tested by collecting quantitative data on behaviour patterns (e.g. Tinbergen, 1953, 1972). Tinbergen's legacy is evidem in current field studies of behaviour, many of which use simple experiments to measure the costs and benefits of traits. A good example is the classic tail manipulation experiments by Andersson and M011er to investigate mate choice in widowbirds and swallows, discussed by Ryan in Chapter 8. However, early studies in behavioural ecology often locused on function and tended to ignore the other three questions. A caricalUre of behaviour studies in the 1930s is one where researchers imagined their animals as little machines, blindly folloWing fixed action patterns in responses to external stimuli. A caricature from the early days of behavioural ecology and sociobiology in the 1970s is the opposite extreme of regarding animals as scheming tacticians, weighing up the costs and benefits 01 every conceivable course of action and always choosing the best one. Current work is leading to an intermediate position. While we expect selection to favour mechanisms that maximize an individual's fitness, we must recognize that mechanisms both constrain and serve behavioural outcomes. A good example to illustrate this point is Lotem's recent studies of how hosts come to recognize a cuckoo egg in their nest. The cuckoo, Cuculus canoms, is a brood parasite which exploits various small birds as hosts to raise its offspring. The female cuckoo lays just one egg per host nest. The cuckoo chick hatches firsl, whereupon iI ejects the host's eggs over the side of the nest. so becoming the sole occupant. Given the cost of parasitism, iI is not surprising
EVOLUTION OF BEHAVIOURAL ECOLOGY
5
that many hosts have evolved defenees sueh as rejection of odd eggs in their nest. Nevertheless. the puzzle is that egg rejeclion rarely reaches 100% in the host population and. furthermore, hosts never reject the cuckoo chick. A consideration of mechanisms may help to solve both puzzles. Experiments show Ihal Ihe defence mechanism used by hosts involves learning the characteristics of their own eggs the !irs! time they breed and then rejecting eggs which diller from this learned set (Lotem et al.. 1995). This makes it unlikely that the host population will evolve 100% rejection of cuckoo eggs because hosts which are parasitized during their !irst breeding allempt will learn Ihe cuckoo egg as part of their own set. Neverthele s, the learning rule works quite well and leads hosts to reject many parasite eggs. At the chiek stage, however, a learning rule does less well than a rule 'accept any chick in my nest'. This is because there is a considerable cost of misimprinting; any host parasitized in ils !irst allempt would learn only the cuckoo chick as its own and would then subsequently rejecl its own young in future, unparasitized, broods (Lotem, 1993). The main message from this study is that it is not very fruitful to discuss the evolution of 'rejection' without specifying the mechanisms, because these will determine the costs and benefits involved. Studies of mechanism and function must go hand in hand. In 1975, Wilson predicted the demise of ethology. with mechanisms becoming the domain of neurobiology, and function and evolution the domain of sociobiology. This prediction was ful!illed until recent years, when Ihere has been a welcome renewed intereSI in linking mechanism and function. We have marked this change by devoting Ihe !irst section of Ihis volume to this fruitful interchange. For example. Giraldeau (see Chapter 3) shows how research on foraging behaviour has stimulated new questions about learning and memory mechanisms. and Sherman et al. (see Chapter 4) point Ollt common features of recognition mechanisms of kin. mates and predators and discuss their functional significance.
1.3 Ecology and behaviour Even before Darwin, biologists often interpreted morphological adaptations in relation to the environment in which the species lived_ Darwin's achievement was to show how these could arise without a Creator. Once the early ethologists, such as Lorenz and Tinbergen, had demonstrated that behaviour pallerns were often as characteristic of a species as its morphological features. allempts were made to correlate dillerences between species in behaviour with dillerences in ecological facrors, such as habitat, food and predation. A pioneering study was that by Cullen (1957), who was a student of Tinbergen. She interpreted the reduced anti-predator behaviour of killiwake gulls, Rissa tridaayla, eompared to Ihe ground-nesting gulls, in relation to their safer nest sites on steep e1iffs. 1vo other early sludies were those by Winn (1958). who linked the reproductive behaviour of 14 species of darter fish (Percidae) 10 their eeology, and by Brown
6
CHAPTER 1
and Wilson (1959), who related the colony size and structure of dacetine ants to their feeding habits, Crook's (1964) study of weaver birds (Ploceinae) has become established as the model for this approach. Crook showed how differences between species in food and predator pressure affeel a whole host of adaptations, including nesting dispersion (colonies versus territories), feeding behaviour (solitary versus nock) and mating systems (monogamy versus polygamy). This comparative method was soon extended to other groups, including primates (Crook & Ganlan, 1966), other bird species (Crook, 1965; Lack, 1968), ungulates (Jamlan, 1974), carnivores (Kruuk, 1975) and coral reef fish (Fricke, 1975). The comparative approach remains innuential today, the main advances being in methodology, particularly the quantification of behavioural and ecological trails, the use of multivariate statistics to tease out confounding variables and methods for identifying independent evolutionary evems (Clutton-Brock & Harvey, 1984). It is now agreed that the ideal way to carry out a comparative analvsis is to reconstruct a phylogenetic tree of the group under study and to use this as the basis for independelll comparisons. Phylogenies not only provide a way of identifying independent evolution but also show the sequence in which traits have evolved within a group. With molecular phylogenies there is also the potential to measure the time-scale of evolutionary change. This new approach to Tinbergen's fourth question is one of the major developments in recent years and is discussed by Harvey and Nee in Chapter 14.
1.4 Economic models of behaviour Many early studies in ethology recognized that behaviour patterns involve costs as well as benefits. For example, Tinbergen tl al. (1963) showed that removal of the egg-shell aher hatching reduced predation of black-headed gull. Larus ridibundus, nests (the egg has a conspicuous while interior). But, leaving newly hatched chicks unattended is costly too, which probably explains why the parent delays egg-shell removal until the chicks have dried out and become less vulnerable to attacks from neighbouring gulls. The pioneer in the use of mathematical models in ecology to quantify these kinds of trade-offs was Robert MacArthur, who first applied the idea of optimal choice in the context of foraging behaviour (e.g. MacArthur & Pianka, 1966; MacArthur, 1972). The argument for using optimality models in behavioural ecology is that natural selecrion is an optimizing agent, favouring design features of organisms which best promote an individual's propagation of copies of its genes into future generations. Behaviour patterns clearly contribute to this ultimate goal. so we expect individuals to be designed as eflicient at foraging, avoiding predators, mate choice, parenting, and so on. Optimality models have three components: (i) an assumption about the choices facing the animal (e.g. prey type); (ii) an assumption about what is being maximized (e.g. rate of energy gain); and (iii) an assumption aboul constraints (e.g. bill size, searching
EVOLUTION OF BEHAVIOURAL ECOLOGY
7
speed). For example, from a knowledge o[ prey available and morphological constraints, we could predict how our starling should select prey so as 10 maximize its rate of food delivery to its brood. If the model fails 10 predict the observed behaviour, we can then use the discrepancies to help identify which of our assumptions was incorrect. Classic early studies include work by Schoener (1971) and Charnov (I 976a,b) on prey choice and patch choice by foragers, and Parker's study of copulation time in the yellow dungfly, Scatophaga stercoraria (Parker, 1970a; Parker & SllIan, (976). The optimality approach has mel with some criticism but in our view it remains the most powerful method [or studying the design of behaviour (for discussion see Maynard Smith, [978; Stephens & Krebs, 1986). Perhaps the main problem facing a behavioural ecologist is that the animals under study are clearly faced with trade-offs not just within a particular activity but also between activities. The slarling has 10 find food, keep an eye out for predalOrs and return 10 the nest 10 keep its brood warm, for example. In Chapler 5, Cuthill and HOUSlOn discuss techniques for considering how different activities combine to influence fimess. In particular, they show how dynamic programming can be used to model sequences of behavioural choices.
1.5 Evolutionarily stable strategies An animal's environment does not consist solely of places to feed, nest, shelter and hide from predalOrs. There is also a liVing environment of competilOrs. Often an individual's best choice will be influenced by what these competitors are doing. Thus, the best place for our starling to feed will depend on where the other starlings go, the best strategy 10 adopt in a fight will depend on what the opponent does, and the best sex ratio for an individual to produce in its offspring will depend on the population sex ratio. Early studies to recognize this important point include Fisher's (1930) explanation for why parents expend equal resources on male and female progeny, Hamilton's (1967) analysis of slable sex ratios under local mate competition, Parker's (1970c, 1974b) field sllldy of how male dungflies distribute themselves across different mating sites and work by Fretwell and Lucas (1970) on habitat choice by birds. All these sludies analysed the problem in terms of which choices would produce an equilibrium distribution in the population, Maynard Smith's (1972, 1982) concept of the evolutionarily stable strategy (ESS) is now widely accepted as the way of analysing decision making where the payoffs are frequency dependent. A strategy is an ESS if, when adopted by most members o[ a population, it cannOl be invaded by the spread of any rare alternative strategy. This idea has been innuential in analysing many problems in behavioural ecology including fighting behaviour and communication (see Chapter 7), mating systems (see Chapter 6) and cooperation and conniet in
social groups (see Chapters 9 and II). [n many cases no single strategy is an ESS, so one of the main messages for field workers has been to expect variability
8
CHAPTER I
in behaviour. Sometimes the variability is between individuals, so there is a polymorphism in the population. Hamilton's (1979) study of dimorphic males in fig wasps provided an early classic example of stable alternative strategies within a species. For recent examples, see Shuster and Wade (199 I), Lank 'I al. (1995) and Sinervo and Lively (1996). More often, individuals vary in their behaviour depending on what their competitors do. A rule 'go to the patch with the greatest number of worm casts' would be fine for a bird if it was the only forager, but in the presence of competition this may not be the best thing to do. The behavioural ecologist's task here is to consider what would be the stable decision rules (see Chapter 3 by Giraldeau for a discussion of this problem). ESS models have been particularly useful in slUdies of signalling systems. In many cases animal displays seem at first sight to be unnecessarily extravagant, for example the stretching, gaping and calling of young birds as they beg for food or the energetic dancinll of males on a lek as they attempt to attract a female for mating. Zahavi's (1975, 1977a) handicap principle proposed that signals are costly to prevent cheating. The key here is that there are often conmcts of interest between signallers and receivers; it pays offspring to beg for more than their fair share of food and it pays even poor-quality males to attract mates. zahavi suggested that if signals were costly then this would enforce honesty so that. for example, only really hungry chicks would gain from begging and only the best-quality males could perform impressive displays. ESS models by Grafen (1990a,b) have confirmed lhat such costs produce a signalling equilibrium. This theory is now stimulating empirical work on the costs and benefits of signalling in relation to individual quality, and Johnstone (see Chapter 7) reviews these studies together with more recelll theoretical develupments.
1.6 Kinship, social evolution and breeding systems up to the mid-1960s, many interpretations of animal social behaviour were in terms of how it was of advantage to the group. Wynne-Edwards (1962) proposed that social behaviour was an adaptaLion for regulating animal populations and many ethologists also used group selection to explain behaviour. For example, Tinbergen (1964) interpreted the mobbing of a hawk by a group of birds as behaviour which, although of danger to the individual. was advantageous to the group. He argued that 'only groups of capable individuals survive - those composed of de!ective individuals do not, and hence cannot reproduce properly. In this way the resull of cooperation of individuals is continually tested and checked, and thus the group determines, ultimately, through its efficiency, the properties of the individual'. Group selection was criticized most cogently by Williams (1966) and Lack (in an Appendix to his [966 book). They showed lhat clutch size and also
EVOLUTION OF BEHAVIOURAL ECOLOGY
9
many social interactions enhanced an individual's fitness and argued that adaptalions evolve for individual benefit, not for the benefit of a group. This lell the problem of altruism, behaviour which increased the fitness of others at a cost to lhe altruist's own personal reproduction. The key insight 10 understanding the evolution of altruism was provided by Hamilton (J 964a,b). He argued that individualscan pass copies ortheir genes on to fUlure generations nol only through lheir own reproduction (direct fitness) but also bi' assisting the reproduction of close relatives (indirect fitness). Hamilton's now famous rule specifies the conditions under which reproductive altruism evolves and there is good evidence, especially from insects (see Chapter 9) and social vertebrates (see Chapters 10 and 11), that kinship provides the key to understanding altruistic behaviour. The huge interest in cooperative breeding during the last 20 years is largely inspired by Hamilton and is a good example of how empirical work is orten driven by advances in lheory. Further impetus to studies of social systems came from pioneering papers by Parker and Trivers. Parker (1970a) recognized the importance of multiple mating for the evolution of reproductive behaviour and coined the term 'sperm competition' [or sexual competition aller maling. when sperm from different males compete for fertilization of a female's ova. Trivers laid the foundations for theories of conflicts in family groups (see Chapters 9 and 10), including male-female conflict and parent-Qffspring conflict (Trivers, 1972, 1974; Trivets & Hare, 1976). He emphasized the importance of the earlier conclusions of Baleman (1948) that differelll factors limit reproduclive rates in males and females. Females tend to invest more in orrspring and their reproduclive rate is usually limited by resourCeS. A male, on the other hand, has the potential to father offspring at a faster rate than a female can produce them. For males, therefore, reproductive success is limited more by access to females. Trivets argued lhal females should be more choosy in mating while males should practice a 'mixed reproductive strategy', both guarding their mates and also attempting to gain eXlrapair matings. Parker (1979, 1984b) also emphasized lhe conflict between lhe sexes, and both he and Maynard Smith (1977) used ESS models to analyse how these may be resolved. [n the last decade, detailed behavioural observations combined with molecular measureS of parentage (e.g. DNA fingerprinting) have confirmed lhe importance of sperm competition and sexual conflict. This has revolutionized our view of mating systems. Just compare, for example, Lack's (1968) conclusion 30 years ago that monogamy predominates in birds because 'each male and each female will leave most descendants if they share in raising a brood' with the currem evidence for widespread mixed paternity and sexual conflict (see Chapter 6). While it is clear thai males compete for mates and females are indeed choosy, there is still vigorous debate about exaclly what benefits females gain from their choice (sec Chapter 8).
10
CHAPTER I
1. 7 Critical views Behavioural ecology has been criticized on a number of grounds during the past 20 years, the main points being as follows.
1.7.1 Determinism Lewontin and colleagues (Lewontin et aI., 1984; Lewontin, 1991) have interpreted the position of behavioural ecologists as implying genetic determinism. Statements such as '...a gene for altruism .. .' could be read as meaning that if an individual carries a certain gene or combination of genes it immutably and irrevocably behaves in a particular way. This would be biologically unsound, since the adult phenotype depends on complex interactions between gene and environment during development. It could also, Lewontin and colleagues argued, be open to an ideological interpretation in which social policies were built around the assumption that humans are purely products of lheir genetic heritage. Although it would be fair to say that behavioural ecologists have generally underplayed the complexities of behavioural development, the phrase' ... a gene for .. .' is never used to imply genetic determinism, but rather as shorthand for '...genetic differences between individuals that are potentially or actually subject to selection'; in other words it implies gene selectionism not genetic determinism (Dawkins, 1982). 1.7.2 Panglossianism A parody of behavioural ecology (and of neo-Darwinism in general) is that every last detail of any organism's behaviour, anatomy, physiology and so on can be explained by natural selection, for example the fact that carrols are orange and parsnips white. Gould and Lewontin (1979), in a classic article, coined the phrase 'The Panglossian paradigm', referring to Dr Pangloss in Voltaire's Candide, who took the view that everything was always for the best. According to its critics, the pure adaptationist approach is flawed for two reasons. First, it ignores the fact that evolution is a historical process influenced by chance, and some of the outcomes would be quite different if the video of life were played again (Gould, 1989). Differences between species or phyla may have no 'logic', they may just be the one chance outcome out of a huge range of possibilities. Second, at anyone moment in time, the degree of perfection of adaptation of behaviour, physiology and so on are constrained by many factors such as developmental flexibility, historical accident and interactions between genes. While these critidsms do not underntine the value of a Darwinian framework as a powerful device for analysing and predicting behaviour, they are reflected in the fact that in this edition of the book we have included grealer emphasis both on the analysis of historical events in evolution, increasingly
EVOLUTION OF BEHAVIOURAL ECOLOGY
11
possible because of new phylogenetic data, and on the constraints that limit adaptation here and now.
1.7.3 Anthropomorphism Kennedy (1992), in a thoughtful and detailed critique of behavioural ecology, points to the dangers of using (often anthropomorphic) linguistic shorthand to describe funclional categories of behaviour. One of his key points is that the use of functional labels such as 'foraging', 'mate searching' and 'parental allocation' tend to become substitutes ror a proper analysis of what is actually going on and may even encourage anthropomorphic interpretations of behaviour. For instance. a behavioural ecologist interested in 'honest signalling' (itself a dangerously anthropomorphic terml) between nestlings and parents in a particular bird spedes might analyse in a model. or by experiments. whether or not •... parents allocate reso'rces in response to the need of individual nestlings .. .' Kennedy would argue that, in fact, parents respond to stimuli, including those emitted by the offspring. and that this is what determines the pattern of feeding. The terms 'allocation' and 'need' are. in effect. terms related to functional considerations of optimal reproductive strategies and should not be taken to constitute causal explanations of behaviour. Kennedy's critique recalls the distinction between two of Tinbergen's four questions, function and causation. and it also serves as a reminder that in carrying OUt experimental manipulations of behaviour one can normally only determine the stimuli to which animals respond. not the functional reasons for a response, which are inferred from the logic of natural selection. The main lessons for behavioural ecologist are these. 1 That functional models should not be taken to imply particular mechanisms or decision rules (in analyses of 'tit-for-tat' as a model of cooperation. the metaphor of the model has been interpreted literally by some authors to mean that animals play the actual game originally modelled by Axelrod and Hamilton - see Chapter II). 2 That care should be taken in experimental analyses of functional models not to connate manipulation of the stimuli to which animals respond. for example prey size or tail length, with the Darwinian interpretation of adaptation. 3 That functional labels should not substitute for a full analysis of the causes of behaviour.
1.8 Looking ahead Sam Goldwyn neatly summarized the dangers of predicting: 'Never prophesy, especially about the (uture.' However. the changes signalled by the new emphases in this edition of Behavioural Ecology coincide with a change in the
12
CHAPTER I
nature of the subject. As we have mentioned earlier, Wilson's (1975) predicted fragmentation of the subject of ethology into functional and causal aspects is not happening; what is now emerging is a new form of integrated study of behaviour. In order for this to nourish, one of the keys will be to embrace the powerful armoury of techniques, from gene splicing to magnetic resonance imaging that have transformed other areas of biology,
PART 2 MECHANISMS AND INDIVIDUAL BEHAVIOUR
Tire dtsert ani Cataglyphis fonis Wtllldt'fS over a ,irmitous l'allt for several }llmdred milUS. yt'( it
call return dirtd/y to ilS ntSI. How dON il achieve t'is amazing '0';90(;01101 [em? (Photograph hy Rudiger wehner.)
Part 2: Introduction
Visitors from another planet would find it easier 10 discover how an artificial object, such as a car, works if they first knew what it was for. In the same way, physiologists are beller able to analyse the mechanisms underlying behaviour once they appreciate the seleaive pressures which have influenced its funaion. For example, the sensory mechanisms of an insea eye are much easier to understand once it is realized that the eye functions not only to detea food, predators and mates but also light from the sky for navigation. Just as function can help an understanding of mechanisms, so mechanism can help us appreciate adaptive design. Behavioural ecologists need to know how underlying mechanisms both serve and constrain the outcomes they measure in terms of fitness costs and benefits. The chapters in this section are all concerned with this fruitful interchange between studies of mechanism and function. [n Chapter 2, Wehner shows how behaviour is affected by constraints from both the physical environment and the animal's own body. For example, how an individual best communicates depends on whether it signals in the air or in water and certain types of behaviour are beyond an animal's capability simply because of scaling faaors. A small animal is unable to generate sufficient force to kick or hit effectively whereas it can bite and crush opponents. In some cases, the sensory system used is nicely adapted to the environment; a good example discussed by Wehner is the different frequencies used by bats for echolocation in relation to foraging height and interference from vegetation. [n other cases, sensory systems impose constraints. Wehner shows that the single lens eyes of spiders and vertebrates have much beller resolving power than the compound eyes of insects and they are also energetically more efficient. This explains why visual communication in insects is restricted to short-range encounters and why other sensory channels (acoustic, olfactory) are used for longer distances. Have insects got stuck with an inferior design feature? Not necessarily. As Wehner argues, compound eyes may have advantages over Single-lens eyes in terms of panoramic vision and navigation.
Wehner's chapter also shows how complex behaviour may be the outcome of simple subroutines. A female paraSitic wasp uses a simple mechanism to
measure the size of a host egg (which determines how many eggs she herself should lay); some migratory birds may use a simple sun-compass mechanism 15
16
PART 2: INTRODUCTION
to fly the great-circle route around the globe (which minimizes the distance between summer and winter quaners): desen ants use path integration combined with landmark memory to find their way home after foraging trips: and male hoverOies use a simple movement rule to guide their flight to intercept a passing female. Giraldeau continues the mechanistic theme in Chapter 3. To behave
adaptively. individuals often need to assess the quality of their habitat or of a potential mate or competitor. How do ecological factors influence the way animals gather. store and process this information? A fundamental problem is how animals measure time intervals. for example to determine foraging rates.
Giraldeau describes experiments which show that birds behave according ro the predictions of scalar expectancy theory. He then goes on ro discuss how individuals estimate patch quality not only on the basis of their own current gain in relation to what they expect, but also by using the 'public information' provided by the behaviour o[ others. Such social learning often involves acquiring information about places and objects but there is surprisingly little good evidence that individuals actually copy the behaviour patterns of others. Giraldeau provides a critical account of whether social learning can give rise to the long-term behavioural traditions which form the ba is of culture. considering some classic examples such as the pecking of milk bottle tops by British titmice and the washing of potatoes by Japanese macaques. Studies of spatial memory provide especially good evidence that ecological factors influence the design of nervous systems. Bird species which cache food and then later on recover their hoards have a larger hippocampus than nonstorers. In polygynous mammals. where males patrol larger home ranges than females, males also have a relatively larger hippocampus. Giraldeau reviews experiments with titmice and corvids which show that the better spatial memory of food-storing species can also be used in non-caching contexts. so it set'lnS La result from an impruvement in general mechanisms for processing sparial information rather than a specialization for rood caching per se. In Chapter 4. Sherman. Reeve and Pfennig consider whether there may be common principles underlying recognition systems. In a wide ranging review they show, for example. that anuran tadpoles prefer to associate with Siblings. that tiger salamanders devour non-relatives but avoid eating close kin, that vervet monkeys give different alarm calls for leopards, eagles and snakes, that female great tits avoid mating with males that sing like their fathers. that female ground sqUirrels are nicer to full-sisters than half-sisters and that female mice. and perhaps humans too. prefer mares with dissimilar major histocompatibility complex genotypes. All these examples involve recognition. whether of mates. kin. prey or predators, and Sherman el al. argue that all involve the same three components. namely: (i) production of a label: (ii) perception of the label; and (iii) an action (acceptance or rejection). Given errors in recognilion systems, selection is expected to favour an optimal balance
between acceptance errors and rejection errors. The chapler discusses whether
PART 2: INTRODUCTION
17
we should expect labels to be genetic or environmental in origin and shows that perception often involves comparison of the label with a learned templale. These lemplates can be learned by self-inspeclion, from nestmales or fmm parents, and experiments show how individuals can be tricked into forming inappropriale templales (e.g. goslings learn Konrad Lorenz as lheir mother or warblers treat a cuckoo chick as their own offspring). Experiments also sugge t lhal acceptance lhresholds become more permissive with increasing costs or decreasing benefits from choosiness. For example, paper wasp workers are more intolerant or unrelated kin on their nest than away from their nt:Sl. female zebra finch become choosier when lhey are exposed to more highquality males and hosts vary their rejection of foreign eggs in relation to their probability of being parasitized by cuckoos. The key to assessing costs and benefits in economic models of behaviour is lhe concepl of lrade-offs. Cuthill and Houston tackle this problem in Chapter 5, focusing on how animals can make best use of lime and energy. Imagine, for example, a starling searching for food during the breeding season. How should il allocate time between feeding and other activities such as singing, male-guarding, avoiding predators or keeping its chicks warm? How fast should it fly OUI to the feeding grounds? Increased flight speed reduces travel time, enabling more time for feeding, bUI the increased energy expended has to be recouped by increased food intake. The chapter shows that animals often ((lpe wilh variations in lhe food available by forming fat stores as an insurance. However, these slores are costly because they reduce locomotory performance, so increasing predation risk and reducing feeding efficiency. Field observations and experiments indicate thaI the optimum fat reserves vary with dominance, predation pressure and patterns of food availability. Cuthill and Houston show how dynamic programming can be used to predict optimal behaviour sequences, such as patterns of feeding and resting during the day. The technique takes account of how the best behaviour varies with lhe animal's Slale (e.g. food reserves), the payoffs from various choices and the time available (e.g. time to dusk). The chapter then considers how short-lerm consequences of behaviour (food gain, mate attraction) can, in principle, be relaled to Iifelime reproductive success. These functional models should help direct fUlllre work on the physiological mechanisms which control food intake. Current textbook accounts ignore the influences of environmental stochasticity, predation and social dominance on the equilibrium level of food reserves.
Darwin coined the lerm 'sexual selection' for selection of traits involved in compelition for males. In a key paper, Parker (1970a) recognized that sexual competition also occurs after mating when the sperm from different males compele for fertilization of a female's ova. He called this process sperm compelition. In Chapter 6, Birkhead and Parker demonstrale that sperm competition is widespread, as revealed by behavioural observations of multiple mating by females and molecular genetic analysis showing multiple paternity.
18
PART 2: INTRODUCTION
They begin the chapter by arguing that sperm competition is likely to have shaped the evolution and maintenance of the two sexes. Models show that the two sexes evolve as the stable outcome in which small gamete producers (males) parasit.ize the resources of large gamete producers (females). In theory, even small levels of sperm competition mean that it pays a male to maximize number of gametes produced (to ensure paternity) instead of contributing resources to the zygote. The study of sperm competition provides an excellent example of how understanding mechanisms (how sperm are stored and compete in the female tract) is vital for an understanding of the function of behaviour (the costs and benefits of muiliple mating for either sex). The chapter reviews Parker's studies of the yellow dungfly and Birkhead's studies of the zebra finch, both ideal model systems for combining field observations with laboratory studies. In both species there is second male advantage in fertilizations of eggs, but the mechanisms differ. In the dungfly the second male advantage fits a model of some displacement of the first male's sperm followed by random mixing inside the female's sperm stores. In the zebra finch, there is passive sperm loss from the female's tract and the second male gains an advantage simply because fewer of his sperm are lost by the time fertilization occurs. The authors discuss the consequences for how males and females should behave to maximize their fitness. Early studies took very much a male-centred view of muiliple mating, focusing on how males maximized paternity through frequent copulation, mate-guarding, the insertion of mating plugs or the production of antiaphrodisiacs in the seminal fluid. Current work suggests that females often control sperm competition and may gain both genetic benefits and better resources by mating with more than one male. The chapter reviews the evidence for this and discusses how mating systems may arise as the outcome of sexual conflict. An unresolved issue is to what extent females can manipulate sperm inside their reproductive tracts so as to favour fertilization by particular males.
Chapter 2 Sensory Systems and Behaviour Rudiger Wehner
2.1 Introduction Behavioural ecologists agree that if they metaphorically regard animals as 'decision-makers' (Krebs & Kacelnik, 1991), they do not imply that the animals decide nn the basis of conscious choices or any appreciation of the computational structure underlying the problem to be solved. Instead, they assume thal simple processes mediate apparently complex behavioural decisions. This assumption flies in the face of what the majority of neuroscientists have thought all along (e.g. Marr, 1982), namely that nervous systems form relatively complex internal representations of the outside world, and then use information derived from these global representations LO accomplish any panicular behavioural task that comes up. This conventional wisdom - the representational paradigm, supponed also by cognitive scientists (GallisteL 1990) - has been challenged recently by the notion that many behavioural tasks may nol require elaborate representations of the external world. By exploiting constraints that are introduced when the animal interacts with its environment, special-purpose task-directed programmes may be able LO solve a given behavioural problem more effectively (Ballard, 1991; Aloimonos, 1993; Churchland etal., 1994). It is here thatlhe ways of thinking of behavioural ecologists and physiologists converge. This convergence, however, has not yet been put into action. Of course, the approaches of behavioural ecologists and physiologists differ in emphasis and focus. While the former -the functional or why-question approach aims at an understanding of the fitness (and hence evolutionary) consequences of a panicular mode of behaviour, the laller -the mechanistic or how-question approach - tries LO understand the physiological machinery mediating that behaviour. Consider, for example, the ease of a foraging honey bee, and in panieular the question when the bee should SLOp collecting nectar and stan LO carry the load back LO the hive. Functional analyses show that under a wide range of ecological conditions crop load can be predicted best by assuming that the bee maximizes energetic eUiciency (energy gain per unit of metabolic cost) rather than net rate of gain (energy gain per unit of time), or any other more complex alternative (Schmid-Hempel et al., 1985). However, economic models of this sort or another do not tell anything about how bees measure variables such as energy gain or foraging costs, and how they integrate these 19
20
CHAPTER 2
measures in order to compute the amount of nectar they should extract from the flowers visited during individual foraging trips. It does not even prove that
the 'currency', which describes the bee's behaviour in economic terms. is actually computed by the animal in the way proposed by the model. Only a physiological analysis can tell what sensory mechanisms a bee employs and what neural computations it performs in order to arrive at what the behavioural ecologist thinks is the currency used in a particular foraging task. Although until recently mechanistic and functional approaches have been entertained by researchers of different camps, they arc in no way mutually exclusive, but complementary. In the example memioned above, knOWing physiologically that worker honey bees are constrained by a limited amoum of flight performance, or flight-cost budget (Neukirch, 1982). may emphasize the economist's finding that energetic efficiency rather than intake rate is the animal's decisive currency. It is upon constraints imposed on behaviour by various sources that this chapter concentrates. One source is the animal's physical environment (see Section 2.2). Certain habitats on the surface of our planet favour particular sensory channels, and as sensory systems differ in their potential for, say, resolving spatial detail, certain behavioural tasks can be accomplished only in one type of habitat or another. When it comes to COnstraints set by the organism itself, body size is an important although widely neglected factor (see Section 2.1). The kind and amount of sensory informatiOn that can be handled and used by a nervous system depends more dramatically on the size rather than the particular design of the system. The latter is responsible for what could be called the fine tuning of behavioural performances (see Section 2.4) - and it is here that behavioural ecologists with their intrinsic interest in microevolutionary processes become most imrigued.
2.2 Constraints imposed on behaviour by the physical environment It goes without saying, but is not always fully appreciated, that the most fundamental functional characteristics of animal design have been shaped by very general properties of the physical world within which an animal lives, moves and behaves. For example. there are much larger fishes in water than there are birds in the air; the body of a fish is more perfectly streamlined than that of bird; for a fish it is more difficult to extract oxygen from its environmem, but less costly to move through this environmcm than it is for a bird. A lillie exercise in physics and physiology will immediately show that all these functional differences between aquatic and terrestrial animals are due ultimately to the way in which water and air differ in such general properties like density, viscosity, oxygen content or gas diffusion rates.
Moreover, the fundamental physical pwperties of air and water are responsible not only for how animals gain and spend their energy, but also for
SENSORY SYSTEMS
21
how they gain the information to move about. 10 detect. localize and recognize objects of interest - in short. to explore their environment (see Dusenbery. 1992. for a thorough treatment of such topics). One simple example might help to make the point. Above the water surface one can see even the islands furthest away on the horizon, but one will not receive any sounds from there. In contrast, under water the visual scene gets blurred and obscured even a few metres ahead of the observer. but one may readily pick up the sounds produced by the engine of a ship too far away to be seen. The reasons for these triking differences are simple and s!raightforward. Owing to the physical properties of lighl, vision is the most accurate source of spatial information that an animal can gain about the world. In both air and water, absorption and scattering of light decrease the brightness and contrast of the image. respectively, but these effects are much stronger in waler than in air (Lythgoe, 1988). Marine fishes. for example. have responded to the strong selection pressures of their dim-light environment by boosting the light sensitivities of their visual systems in various ways (Locket. 1977; Munz & McFarland. 1977; Douglas & Djamgoz. 1990). As depth increases. the spherical lenses of their eyes become more powerful. the photoreceptors increase in size. are arranged in multiple layers or arc combined to functional multi receptor units; screening pigments usually shielding the photoreceptors are lost. and light reflectors underlying the receptor layer are formed - until, at depths of 800-1200 m, eyes disappear altogether (and prevail only in some bioluminescent species). At this 'faunal break' quantum capture rates have become so low that any visual signal gets buried in photon and receptor noise, and finally vanishes. While this suite of adaptations to environmental constraints tells a clearcut story, other seemingly similar specializations are more difficult to interpret in terms o( the optimization towards which any adaptation works. Take. (or example, the spectral absorption characteristics of the rhodopsin photopigments built into vertebrate photoreceptors. The maximal absorption rates of the highsensitive, dim-light receptors, the rods, are tightly clustered around 500 nm (Goldsmith. 1991). This is a reasonably good adaptalion to the spectral light conditions prevailing at depths of about 100 m, but at lower deplhs. as well as at and above the water surface. the photon (lux is greatest al much longer wavelengths (Fig. 2.la). Why have rod photopigments. which are trimmed (or high quantum capture rates. not responded to these strong selection pressures and shifted their maximal spectral sensitivities (their Am.. -values) to longer wavelengths? This question is all the more intriguing as the photopigments of the cone-type receptors containing the same chromophore (retinal-I) and the same opsin-type protein moiety as the rod-type receptors are usually well adapted to the colour o( the water in which their owners live. This holds !rue even (or closely relaled species inhabiting -like the snappers (genus Lutjanus) o( the Australian Great Barrier Ree( - different marine habitats. e.g. the clear blue water of the outer reef. the greener water of the inshore reefs or the more
22
CHAPTER 2 0 10
~
(b)
-
Cones
Rods
0
1D 0
E
ac
100
CT
1000 m
,
.. .
2
~
~ c
3
4
0;
a:
0 400
4SO
SOO
5SO
Wavelength maximum 1~&l{1 of rod rhodopsin (nm
500
450
r--
-
J ,.... --
-
500 550 Wavelength (nm)
600
Fig.2.1 (a) Histogram: absorption maxima (Arnax·values) of 274 phOLOpigmellls (rhodopsins) of venebratc rod pholoreceptors. Curves: relative sensitivity (relative quantum CillCh) of rod rhodopsin as a function of AIll.1lf (abscissa) calculated for various
depths of water (0-1000 mi. (Modified from Goldsmith, 1991.) (b) Absorption maxima (Alllu-values) of the phOlopigments of rods (dark grey area) and cones, i.e. double cones, (black bars) of 12 spedes of teleost fish belonging to rhe genus Lutjanus and inhabiting different marine habitats: I. outer reef; 2, middle reef; 3. inner reef; 4, estuary. The left and right limitations of each bar mark lhe A.rn;n; of the two mt>mbers of the double cones present in the teleost retina. The light grey area represents the range of nil,,·values calculated. for each water type. to confer greater than 90% of the sensitivity of the most sensitive rhodopsin. (Modified from Lythgoe el al.. 1994.)
heavily stained mangrove and estuarine walers (Lythgoe et al' 1994). Again, however, in all these species the absorption spectrum of the rod photopigmem stays put (Fig. 2.1 b). Why does the family of rod pigmems exhibit such evolutionary inertia, while that of the cone pigments does not? I ask this question, at this juncture, not in order to discuss hypotheses about the molecular biology of vision, but to caution against simplistic adaptational explanations. As this particular case shows, molecular constraints might be as significant as ecological ones. More generally, it is difficult to include in any hYPOlhesis all the variables over which adaptation integrates. In the present context, the 'mox -value is cenainly only one of many attributes of the rhodopsin molecule that is sensitive to natural selection. Note, for example, it might already be the high absolute sensitivity of the rods - higher by orders of magnitude than that of the cones - that limits any shift of 'mox to larger wavelenglhs. On the surface of the earth as well as in yellowish freshwater habitats, such shifts would indeed increase the number of quanta absorbed, but they would also increase the rate of darknoise events and hence decrease the signal-to-noise ratio. In addition, rhodopsin is not only a receiver of light. but also a membrane-bound enzyme involved in the phototransduction cascade. As mentioned before, due to the 'veil of scallered light' between the eye and the object, underwater vision is essentially a short-range affair. fn contrast, underwater hearing extends into the far field. Hence, in aquatic as well as in
SENSORY SYSTEMS
23
nocturnal animals acoustic (and especially sonar) systems of orientation are much more eflective than visual guidance schemes. Dolphins and bats oHer prime examples. The propagation of sound pressure waves is almost five times faster in water than in air. Furthermore. the power of sound emission depends on the product 01 the velocity 01 propagation and the density 01 the medium. This productthe impedance of the medium (Michdsen. 1983) - is 3500 times larger for water than lor air. Consequently. low-frequency sounds as produced by baleen whales (about 20 S-I) allenuate very lillie when travelling over large distances. Calculations 01 the transmission losses. which occur at various depths. especially in those layers in which sound is clfectively trapped due to particular temperature conditions. show that fin whales. for instance. might hear each other over distances o[ several hundred kilometres (Payne & Webb. 1971). One only wonders how these whales could usc an acoustic communication system in modern times. when the engines o[ ships produce powerful sounds exactly in the whales' frequency band. Let us now turn to the special case o[ the sonar system. Whatever the medium within which such a system works. there must always be trade-olf between the range and the accuracy of target detection: the higher the frequency 01 the emilled sound. the beller the spatial resolution that can be achieved. but the stronger the attenuation 01 the signal as distance increases. tn evolutionary terms. bats have responded to this trade-ofl situation by choosing their microhabitats and predatory life-styles correspondingly. For example. in comparisons across dil[erent species and genera 01 bats. the frequency of the echolocating sound (and. correspondingly. the best frequency of the auditory system) is inversely related to the height of the preferred foraging area (Neuweiler el 01.• 1984). In other words. the Irequency increases the closer the bats hunt to the ground or to the edges of vegetation (Fig. 2.2). Above the lorest canopy. where potential prey (fiying insects) arc the only objects [rom which sounds can be renected. a premium is paid for far-ranging (Iow-Irequency) signals. while within cluttered environments the detection 01 targets against a noisy background becomes a more severe problem. fn this case. high-accuracy (high-Irequency) sounds arc advantageous. What looks like the exception to the rule is Me9aderma lyra. the lalse vampire (Fig. 2.2. M). However. this bat. which hunts close 10 the ground. detects its prey (beetles. birds. mice. etc.) not by using its sonar system but by listening to the sounds produced by the moving prey itsell. Due to the properties 01 sound transmission in air and water. some details 01 sonar systems should be dillerem in aquatic and terrestrial animals. For example. sounds of constant Irequency exhibit higher velocities and larger wavelengths in water than in air. Hence. the same accuracy 01 orientation (for data on toothed whales see Au. 1988; Wiirsig. t 989) requires that the echolocative sounds are 01 higher frequencies in aquatic than in terrestrial animals. This is indeed what occurs (Nachtigall & Moore. 1988).
24
CHAPTER 2
10
I 'c e •'> ~
.&
•
8
:%
~
T
R
c:;::::::>
6
~ '?!,
4
~
'
.~
:I:
2 M 0
17
24 26
35
H
,, 50 55
Best frequency (kHz)
Fig. 2.2 Relationship between the best frequendes (of 3udiograms recorded in the dorsal midbrain) and the preferred foraging ranges of echolocaling baiS in southern India. The nine species of bats studied belong to the follOWing genera: H. Hipposideros;
M. Megaderma; P. Pipistrellus. R. Rlrinopoma; T. Tadarida. (Modified from Neuweiler tl al., 1984.)
Similar environmental constraints apply to auditory communication, where they have been studied in both vertebrates (Wiley & Richards, 1978) and insects (Romer & Bailey, 1990). They are especially intriguing in the latter. because the frequencies of most insect songs lie in the high sonic or ultrasonic range, i.e. well above those of most vertebrates. The attenuation and degradation of these high-frequency sounds by vegetation poses intricate questions. For example, as insect habitats act as effective low-pass filters (Fig. 2.3a) and as all orthopterans studied so far have the potential for frequency analysis. the frequency-dependent attenuation of sound could by used as a means to estimate the distance between sender and receiver. However, as the amount of frequency filtering depends on the structure and density of vegetation, the frequency content of a Signal does not provide an unambiguous cue. In fact, the spacing of calling bushcrickets in the field varies with the loudness of the calls (with larger animals producing more intense sounds) and the density of vegetation. As the calling males are nut informed aboul either variable. they can maintain only acoustic rather than absolute distances to their neighbours. Males in which the sound output is experimentally reduced. or which live within denser vegetation. move closer together than the controls (Dadour & Bailey, 1990). But, there is yet another and even more severe problem, namely to localize rather than only to detect the sound source. Due 10 multiple renection and scattering of sound in vegetation, the sound field around a listening insect might become ralher diffuse (Fig. 2.3b-
SENSORY SYSTEMS lal
25
Ie)
E
10
0
20
40
30
Distance 1m)
M
Ibl
8
,
L
0
:;
E
o
1.5 m
4
.~
~
0
~
0
:;
lOmm
R
0.
0
0.
Om
E ~
~
•
z
4
Idl
R L
0 90· Left
60·
30·
o· Frontal
30·
60·
go. Right
50ms
Fig. 2.) (a) Frequency-dependent sound awmuation in a bushcrickel habitat. The fom curves refer 10 sounds of 5, 10,20 and 40 kHz. (Modified from Romer & Bailey. 1990.1
(b) Directional selectivity of a pair of auditory interneurons (T-fibresl of the bushcrickt't
Tettigonia vi';dissima. The recordings (see c and d) wert;' laken in dense bushland at a distance uf to m from the sound source either 1.5 m above the ground (upper figure) or on the ground (lower figure). The solid and open circles refer to the right (R) and lefl el) interneuron, respt.·etivt'iy. (c) Recording device: portable 'biological microphone'. The device enables long-term extracellular recordings from single idenlified inlernfurons in Ihe prolhoracic ganglion. The animal is mounted with the ventral side facing upwards A glass-insulated tungsten electrode (El is inserted in the ganglion with a microdrive (M). The socket of the recording contains an amplifier (A). a bandpass filter (F; 0.5-5.0 kHz) and the miniature microdrive (M) by which the preparation can be moved in three dimensions relative to the fixed electrode (E). The portable recording unit is placed within lhe habitat at various dislances. heightS and directions relative to the sound Source. (dl Response of an auditory interneuron (upper tran') as monitored with the portable recording unit in the field. The sound stimulus (lower [race) is the conspecific male stridulatory song. (Modified from RheinJander & R6mt'r. 1986.)
of small animals, face the severe problem to extract information about soundsource directions from weak directional cucs. The notion 'especially in small animals' brings us to our next topic: how body size influences the way animals behave.
26
CHAPTER 2
2.3 Constraints due to one of the most fundamental biological characteristics: body size Animals come in various forms. but functionally even more importantly. they also come in various sizes. From the smallestlO the largest. they span a range of body masses that covers more than 10 orders of magnitude (Schmidt-Nielsen, 1984). Within this range, they are not isometric. even if the organization of their bodies follows the same general pattern (or bauplan). Nearly all morphological and physiological variables change in proportion to each other. as body size varies: relative to body size they are scaled in non-isometric (allometric) ways. Furthermore, the constraints that pertain to body size may become so severe that they can be Overcome only by a novel design. For example. unicellular organisms move by using cilia or flagella, but if animals which are only one or two orders of magnitude larger (e.g. small crustaceans) were covered with cilia. they would get nowhere. As body size increases, a new design is needed -locomotion by movable body appendages. Or. to cite another example: diffusion is an adequate mechanism for supplying oxygen lO all body pans of a small organism (less than about 1 mm in diameter). but it is too slow and hence completely inadequate for oxygen supply to larger animals. A novel mechanism -transport by convection - must be added lO diffusion. As can be inferred already from these two examples. size dependencies in biological phenomena are anything but trivial. In fact. the appearance of the physical environment lO an organism and the organism's evolutionary response depend most strongly not on whether the organism is a bee or a bird. a worm or a whale, but on how big it is (Schmidt·Nielsen, 1984; Vogel. 1988; Pennycuick. J 992).
How does this apply to an animal's behavioural ecology? One of the most fundamental interactions between an animal and its environment is the way in which it moves about within this environment. Here, it is already as simple an aspect of locomotion as tripping that scales dramatically with body size. The bigger an animal is the harder it falls. The momentum when a (large) organism hits the ground is proportional to the fourth power of its length (note that momentum is mass times velocity and that for short falls by large creatures drag is negligible). Hence, small animals can afford to stumble, but large ones cannot (Went. J 968). If this might seem to be loO trivial an example. let us turn to a more intricate mude of behaviour: the throwing of projectiles like SlOnes or rocks. as it is practised by apes bUI not by smaller (although manually dexterous) animals. One might surmise 'hal the smaller animals lack the necessary sensorimotor skills. Be this as it may. Ihe much more fundamental reason is that they jusl cannot impart enough momentum lo a projectile to make it an effective weapon. The momentum of a projeaile thrown by an animal of proportionate mass is again proportional to the fourth power of the animal's length. By the same token. kicking and hitting can be performed only by large animals. while biting. crushing and squeezing will work for small
SENSORY SYSTEMS
27
animals as well (Vogel. 1988). In conclusion, certain types of behaviour are beyond an animal's reach for reasons not (or not only) of neural performance but simply of body size.
2,3, I Visual acuity When it comes to behaviour that depends on the analysis of fine spatial detail. vision provides the most accurate source of information. In accord with this potential offered by the physics of light, simple eye-spots or more advanced types of eye have evolved independently 40-60 times in almost all major groups of animals (Salvini-Plawen & Mayr, 1977), and have led to at least 10 different biological solutions to the physical problem of forming an image (Land & Fernald, 1992). Among those animals which rely most heavily on vision, two types of eye prevail: single-lens eyes and compound eyes. The former occur in a wide variety of taxonomic groups (coelenterate medusae, annelid worms, gastropod and cephalopod molluscs, insect larvae and spiders), whereas the latter are restricted, almost exclusively, to the arthropods. It is intuitively clear - but, can be derived from optical theory as wellthat visual resolution decreases with eye (and body) size, but this size dependency varies dramatically between single-lens and compound eyes. In the former the radius of the eye increases linearly with resolution, while in the latter it is proportional to the square of resolution (Land, 1981; Wehner, 1981). This prediction derived from optical analyses is confirmed by the evolutionary result (Fig. 2.40): compound eyes are rather large and restricted to small animals. Therefore, it is costly to support them and to carry them around (Laughlin, 1995).10 order to acquire a unit of visual information (a pixel), the owner of compound eyes must invest more in terms of energy expenditure than an animal equipped with single-lens eyes (Fig. 2.4b). In other words, for a given size of eye single-lens eyes offer the potential of much better resolution than compound eyes. For example, in the principal (single-lens) eyes of jumping spiders the imerreceptor angle can be as small as 2.4 arc min (Williams & Mcintyre, 1980) and, hence, comes close to the one found in the human fovea (0.6 arc min). As a consequence, the spider can distinguish between conspecifics and similar-sized prey at distances as great as 30 body lengths away (Jackson & Blest, 1982). An insect of about the same body size but equipped with compound eyes must get one or two orders of magnitude closer to the object to resolve the same amount of spatial detail. Why, then, are insects and crustaceans using such an inferior optical instrument? Why have they not replaced their compound type of eye with the single-lens type? Connecting a new set of eyes to an existing neural hardware might not have been a viable option (Laughlin, in press), so that insects might have got stuck with a type of eye that worked well at low resolution, but then could not be changed. However, as many insect larvae show, insects are, in fact, able to build high-performance single-lens eyes and to use them effectively in behaviour (Wehner, 1981).
28
CHAPTER 2
ta) 0.3
~
0.2
Compound eyes Single-lens eyes
-1 A
[0
V
o
:
~.
• •
~
.~
;; a:
.....-...'• •.•. •.•.
•
.~ ~
•
0.1
• ••
0 0 0 0
0
0 0
o
0
o
0 °°00
0
0
0
~~'
o
,0'
'0
0 <9
0
'0'
'0'
Body size (mm)
Ib)
c
.g '0' m
E
.E .£
'0
10'
;; o u
'c
:::l
10
10'
'0'
'0'
,0'
10'
Information uptake
Fig. 2.4 (a) Relative size of compound eyes (1iIled symbols) and single·lens eyes (open symbols) in anhropods (A. squares) and vertebrates (v. circles). Body size is given in linear dimensions ('nominal length', i.e. the cube root of body mal;s). and so is eye size (the
largest diameter of the eye). The relative size of the eye is ddined as eye size divided by body size. (Modified from Wehner. 1981.) (b) The unil cost of visual information (10 ' 11m'
transported retinal mass per pixel) plotted as a function of the lotal amount of information acquired (pixels per steradian of solid angle uf visual space). The cost is defined in terms of lhe energy required to transport the volume of eye that subserves one piXel. siner the metabolic cOSt of phototransduClioll is nrgligible if compared with the transportation custs. (From Laughlin. 1995; and personal communication.)
On the olher hand, behaviour will cenainly have exerted more selection pressures on eye design than merely the need for high visual acuity. For example, compound eyes may be the adval1lageous type of eye whenever panoramic vision becomes important. This is because compound eyes cover
Ihe surface of the head and hence create a convex - rather Ihan concaverelina. The number or single-lens l'yes a spider employs total up 10 eight. yl'l
the spider cannot see all of its surroundings at once (Fig. 2.5a). On Ihe other hand, many insects with only two compound eyes are able to view the entire visual world simultaneously, and at the same time may have al their disposal
SENSORY SYSTEMS
Fig. 2.5 (a) The visual rield of a jumping spider. Metnphidippus ameol/ls. The small
29
ViSUlll
ril'lds of the anlerol'lledian (principal) e;:y~s shuwn in black can be moved over a horizolltal rJnge indicated by lhe heavy broken arrows. Light grey: monocular and binocular fidd'i of vit·w of the anterolateral eyes; dark grey: monocular and binocular fields of view of the
posterolateral eyes. The visual fields of Ihe tiny poslt'rolm~dian eyes have been omitted. (From Wehner & Srinivasan. 1984; based on data from Homann. 1928; Land. 1969.) (b) Visual field of a praying mantis. Tmodera Qustra/asiae. Light and dark grey indicate the monocular and binocular fields of view, respedively. The arrow points forward. The dashed line marks the horizon. (From Wehnl:r & Srinivasan. 1984: based on data from
Rossel. t979.) a huge binocular field to which more than 70% of all ommatidia can cOlllribute (Rossel. 1979; Fig. 2.5b). By monitoring the apparent motion of the environment and the objects within it as the animal moves, compound eyes provide useful information as to the animal's own motion and to the landmark skyline around the moving animal. Systems 01 navigation and course comrol in which such information is used do not necessarily demand high visual acuity. One can obtain sufficiently reliable infonnalion on movement by monitoring the low spatial-frequency content of the environment. Compound eyes, viewed in this light. 'creatively' destroy unwanted information at the very first stage of vision, by using their coarse-grain optics to filter out superfluous spatial detail. In summary, the principal design features of compound eyes - relatively poor resolution and large fields of view - enable insects to perform well in dealing with global aspects of their visual world. Such aspeets are used in course control and navigation; these types of behaviour might have been the ones for which compound eyes have evolved primarily. For performing detailed local analyses, i.e. detecting and identifying objects like conspecific mates, insects must employ complicated anatomical compromises to insert acute zones in their faceted eyes (Land, (989) and must get rather close tnthe object under scrutiny. This may be the reason why in insect communication visual sigoals are usually restricted 10 short-range encounters. Over larger distances, species-specific messages are conveyed through other sensory channels - ollaetory, acoustic, vibrational- which thus playa more signilicant role in insed communication.
30
CHAPTER 2
2.3.2 Sound-source detection The need to localize sounds becomes most apparent in acoustic communication. In this context, only those sounds are localized and recognized that can also be produced by the animal. Here again. physical limitations abound: the smaller the animal, the higher the minimal frequency that can be generated, and the lower the distance over which sounds can be received. This is because maximum efficiency of sound emission requires that the diameter of the sound source is 01 the same order or magnitude as the wavelength of sound. Consequently, insects with body lengths below I cm are restricted generally to ultrasound. but ultrasound is a useful means of communication only in free space or at shon range (Michelsen, 1983). Vel. many small insects communicate over distances of many times thde body lengths. even up 10 metres. For doing so. they must abandon the acoustic channel and switch to substrate-borne vibrations (Markl, t 983). Such signals can travel across the surface of an insect's host plant with rather little attenuation. In addition, the effiCiency of convening muscle power into vibrational power is much higher than that of the conversion into acoustic power, so that it is not only functionally more effective but also energetically less costly for a small insect to communicate through solid substrates rather than through air. If one is small, air offers yet another 'cheap' possibility: communication by near-field air oscillations rather than sound-pressure waves. Such oscillations as caused by wing vibrations are used by Drosophila flies to communicate with females (Bennett-Clark, 1971) and by dancing honey bees to convey their sound message to othcr workers (Michelsen et al.. 1987). At close range these oscillations arc so intense (0.5-1.0 mm peak-to-peak amplitude. 1-3 mm s-'; Michelsen, 1983) that they are able to activate antennal mechanoreceptors of near-by conspecifics. As they decrease with the third power of the distance to the source. they are the signals of choice for 'private conversation'.
2.4 Constraints set by the animal's computational capabilities In trying to understand the computational soflware and physiological hardware of animal behaviour, neuroscientists have often been led astray by their ideal of general, all-purpose designs. The following section of this chapter shall remind us nf what we have known all along, but not always fully appreciated, namely that an animal's solution reflects a unique nervous system with adaptive limitations and particular biases. Formally similar problems may be solved by different animals in different ways depending on the animal's evolutionary history and present-day ecology. Idiosyncrasies in neural circuitry may persist as long as they do their job and as long as the animal has managed to design its way around them. Just recall the example of compound-eye vision: as the need for higher resolution increases, insects squeeze high-acuity zones into
SENSORY SYSTEMS
31
Ihe low-acuity facet arrays of Iheir eyes ralher than exchange their Iype of eye for one that is intrinsically superior in terms of overall acuity.
2.4.1 Coping with spherical geometry: the egg and the globe lchneumunid wasps of the genus Trichogramma lay Iheir eggs into Ihe eggs of other insect species. The number of eggs which are deposited depends on the size of the host egg. In determining the volume of the spherical host. the wasp does not trace out spherical triangles and perform spherical trigonometry. but assumes a particular body poslure. in which the angle between the head and Ihe first segmenl of the antenna (the scapus) is relaled to the radius of the sphere (Fig. 2.6). This angle is probably monitored by the mechanosensory bristles located al the joinl between head and scapus. NOle. however. that Ihis simple method. by which the volume of a sphere is 'compuled' by relying on a simple angular measurement. works only if the wasp adjusls its body position ' as 10 keep IWO olher measures (onslant: Ii) the height of the thorax above Ihe sur/ace; and Iii) Ihe angle hetween Ihorax and head. Both conditions are mel in the animal's behaviour. While parasitic wasps must cope wilh the geometry of minute spheres. migrating birds must trace out navigational Courses across the surface of the
globe. On their spring and autumn migralion even small birds - say. warblers and waders - can Iravel for several thousand kilometres non-stop over
(el 150'
E
'
c
.'
c m
,
~
~
100'
Q.
c
0 (J)
75' 1 M
3
5
Radius of glass bead. r (head units)
Fig. 2.6 Parasiloid wasps, Trichogramma minutum (Ichneumonidae). use the surface curvature of their host eggs to determine the number of progeny allocated 10 the host. (a. b) female wasps examining glass·bead models or diHerem sizes (radius f). (c) Angle.. belwe~n head and scapus observed for different sizes of glass-bead models. 11. scapus-ht'ad angle (angle BCD); M. centre of spherical glass bead. (From Wehner. 1987; based on data from Schmidt & Smith, 1986.)
32
CHAPTER 2
potentially hostile territory like the vast expanses of sea or desert. What are the most convenient routes these migrants should take? With all other things being equal (which they never are on the surface of our planet), the energetically least demanding way of travelling is to follow the great-circle (orthodrome) route. This route (Fig. 2.7) defines the shortest distance between two points, but is cumbersome and difficult to compute, because it intersects successive lines of longitude at different angles. However, there is a short-cut way of travelling along the great-circle route: if the bird followed a sun-compass course, but did not reset its internal clock as it moved eastward or westward, i.e. crossed different time zones, it would automatically fly along that route without having to compute it by spherical trigonometry (Alerstam & Penersson, 1991). Radar slUdies suggest that this mechanism is employed by Siberian waders crossing the Arctic Ocean. In these polar regions, where there are IlO topographical and ecological barriers to cross. the great-circle route is the energetically most efficient one. On the other hand, if brent geese followed the great-Circle route on their way from their spring stop-over sites in Iceland to their breeding grounds in northern Canada, they would have to cross tlte Greenland ice cap where it is steepest and widest. Instead, they take a more circuitous route following more or less a constant-angle (loxodrome) course from Iceland to the east coast of Greenland, turn soulh, stay for 2-7 days within a rather delimited area, and then continue across southern Greenland on a course nearly identical to the one taken at Iceland. It is fascinating 10 hypothesize that the geese use their temporary halt at eaSl Greenland to fesct their infernal clock from local rcelandic to local
Greenlandic time, and then cot11inul'
Oil
the same.' sun-compass course they
have followed previously tAlerstam. 9%). We know from laboratory studies in other species of birds thatllnder ,'xl'tlSllr,' ttl a lll'W 24-h light/dark regimen
..

····
.
.):
...... .. .. ......
Fig. 2.7 Orthodmme (great-
circld and loxodrome (constB11I angle) courses drawn on lht' surface of tilt' globt'. Solid line:,
orlhodrom...: dUlled lox()drol1lt'.
Iillt',
4
SENSORY SYSTEMS
33
it usually takes 3-6 days to recalibrate the sun compass. In conclusion. depending on lhe ecological needs experienced during their evolutionary history. migrating birds might take orthodrome or loxodrome courses and select and maintain either course by rather simple computational means. Moreover. modern large-scale satellite-based radiotelemetry reveals that long-distance migrants do not travel for very long distances on either orthodrome or loxodrome courses. but seem to employ a number of navigational subroutines rather than an all-purpose system of navigation. For example. North American warblers reach their South American wintering sites by follOWing neither orthodrome nor loxodrome courses. but by taking a wide eastward sweep across the Atlantic. Surprisingly as it might appear at first sight. this vast detour is the energetically most economic route. because. it allows the birds to exploit large-scale wind and barometric pressure patterns (Williams & Williams. 1978). In conclusion. during evolutionary time the migration routes of birds have been shaped by a number of quite different selection pressures. e.g. by synoplic weather patterns. large-scale topography. suitability of celestial or magnetic cues. etc. As Alerstam (1996) has succinctly put it. birds travel without any idea in their minds that the earth is a globe. Instead. they have responded to the selection pressures mentioned above by developing a number of sophisticated tools of migration and ways to integrate and adapt these tools in intricate ways. 2.4.2 Reading skylight patterns and landmark panoramas: the insect navigator Insect navigation. although less impressive than bird navigation in spatial scale. is just as intriguing in terms of behavioural sophistication - all the more as in inseet.s some of the underlying mechanisms have recently been
unravelled in unprecedented delail. The best studied and. in fact. most eminent insect navigators are eusocial hymenopterans like ants and bees. These central place foragers (Stephens & Krebs. 1986) continually move to and from their central place. the site of the colony. to retrieve widely scattered food particles from the colony's environs. It is during these foraging endeavours that the spatial coherence of the superorganism - the colony - is relaxed and is reestablished only by the navigational performances of [he individual colony members. An example of this performance is given in Fig. 2.8. While foraging in a circuitous way over distances of more lhan 200 m. COloglyphis ants of the Sahara desert navigate by path integration. Thcy continually measure all angles steered and all distances covered. and integrate these angular and linear components of movement into a continually updated vector always pointing home. This is a computationally demanding lask which Caloglyphis must solve with its small nervous system. and - as recent research has shown (e.g. Wehner el af.. 1996) - it does so by relying on a number of rather simple subroutines.
34
CHAPTER 2
!~
~? J 0
If
!
, I~ 1'
I-
It>
J-
~
.,1
-
iii !Vl r
r
Fig. 2.8 Outward and homeward pathS of an individually foraging desert ant. Caroglyphis fortis (see inset). The start of the foraging excursion (nesling she) and the sile of prt'y capture are indicated by the open and the large Wled circle. rcspeaively. Time marks (small filled circles) arc given every 60 s. Grid width, 5 m; length of outward path (thin
Iint'), 592.1 m; length of return path (heavy line), 140.5 m. (Modified from Wehner & Wehner. 1990.)
In t.he present context, let me focus on the compass used by CQrQglyphis to monit.or the angular components of its movements. This compass is a skylight compass based primarily on a peculiar straylight pattern in the sky, the pattern of polarized light (or E-vector pattern; Fig. 2.9a). At this juncture, it is not important to understand this pattern in any physical detail. Suffice it to say that in any panicular pixel of sky the electric (E) vector of light oscillates in a particular direction. and that the phOloreceptors in a particular region of the ant's (and bee's) eye are sensitive to these osdllations. But. Ihere is more to it. The skylight pattern the insect experiences is not static. but changes with the elevation of the sun above the horizon (compare left and right half of Fig. 2.9a). These dynamics notwithslanding, CQrQglyphis can infer any part.icular point of the compass - say, 30° to the left of the solar meridian - from any particular point illihe sky. This task musl be accomplished when, for example, under cloud cover or due to experimental tricks played by the human investigator, E-vector information can be obtained only from a small gap of dear sky. If a physicisl tried to solve this navigational problem from first principles, he/she would have to run a rather sophisticated series
SENSORY SYSTEMS lal
180' 150'
210'
180'
lb)
210'
150'
= 90'
30'
- -
= 270'
330'
30' O'
35
O'
Fig. 2.9 (a) 1Yo·dimensional representation of (he E-veC10r pattern in the sky shown for twO elevations of the sun (black disc): 25° (left) and 60° (right). The orientation of the Evectors (the directions of polarized light) are represented by the orientation of the blad. bars. The sizes of the black bars mark the degree (percentage) of polarization. The zenjth is depicted by an open circle. 0°, solar meridian; l80o. anti-solar meridian. (b) The ant's internal representation of the sky as derived from behavioural experimems. The open bars
indicate where in the sky the insect assumes any panicular E·vector to occur. This 'template' is used invariably for all elevations of the sun (for details see Wehner, 1994).
of measurements and computations, and use spherical geometry to perform elaborale three-dimensional constructions (Fig. 2.10). The insect navigator, however. comes programmed with a trikingly simple internal representationor 'template' - of the external E-vector pauerns (Fig. 2.9b). This fixed neural template resembles the skylight pallern when the Sun is at the horizon. but differs lrom it for all Dlher elevalions (lor a review 01 the behavioural and neurobiological analysis of Ihe E-veClOr compass see Wehner, 1994). The tantalizing question now is this: how can COlog/yphis navigate correctly by using an internal representation of the sky that is not a correct copy of the external world? In the full blue sky, with the entire E-veClOr pallern available. the best possible match between the external pallern and the internal template is achieved when the inseer is aligned with the solar - or anti-solar - meridian. the zero-point of the compass. (The distinction between these two principal meridians can be made by Dlher means.) The match decreases systematically, as the animal rotates about its vertical body axis, i.e. selects other compass directions. Due lO the discrepancy between the internal template and the external pauern. mismatches occur whenever only parts 01 the skylight pauern are available. For example, an individual E-veeror is matched with its corresponding deteeror in the template only when the animal deviates by a certain angular amount from the solar meridian, so that the zero-point of its compass gets shilted. Consequently, navigational errors arise when the foraging animal experiencing, say. the entire skylight pauern, is suddenly presented with a small patch of sky. In fact, it was from these systematic errors observed in the
36
CHAPTER 2
1'1 -
.zJ;
'.. vY~ -- .xYvz~ ; ZJV
Fig. 2.10 Howa physicist could infer the position of the sun from viewing at least two smdll patches of skylight: First. determine the E-vcClor direction (X) in the two patches of sk} (this is a problem in ilself, which is not discussed here). Then. construct the great circles that run at right angles through the E·veclQrs. The position of the sun (filled circle)
is defined by the intersection of the two great circles. r( only one E-vector is visible. the po~ition of the sun cannol be determined unambiguously. Provided that the elevation of lhe sun is known at the panicular time of day the (two) intersection points of the great cirtle and the parallel of altitude (dotted arc) defined by the elevation of lhe sun yield the correct position of the sun (filled circle) and a second one (open circle) that is separated by the azimuthal distance u. from the correct one. Calaglyphis does not perform such construnions but uses a generalized template of the sky (see Fig. 2.9b). (Modified from Wehner. 1981.) In the lower pan of the figure Caraglyphis inspects a paper of Frantsevich (1982) outlining a model of E-veetor navigation.
insect's behaviour under certain experimental conditions that evidence for th(' internal template could be derived in the first place. Note, however, thai such errors do not occur when the animal is continuously presented with the same patch of sky. It then always uses the same reference direction, be Ihis Ihe actual solar meridian or any other celestial meridian that is characlerized by Ihe currently besl match between the template and the outside world. For comparison. if a human navigator used a magnetic compass in which the needle erroneously but consistently pointed towards east rather than north, this 'defective' instrument would work as a reliable compass as well. In condusion, evolution has managed to build inlo the insed navigator a nervous syslem that indudes only some general knowledge about th.. geomelrical characteristics of the celeslial world. but this panial knowledge
SENSORY SYSTEMS
37
is sufficient if the navigator restricts its field trips to shan periods 01 time. The insect assumes that the celestial hemisphere does not change during any of its panicular foraging excursions. Given its shon foraging times which lie in the range of tens of minutes rather than hours. this is generally a valid assumption.
Similarly. simplified solutions are employed by insect navigators when landmarks are used to back up the noisy path integration system. As shown in Fig. 2.8, in which an ant performed its foraging and return path within the expanses of a nat and featureless Saharan salt pan, the path-integration system worked without the aid of any landmark-based information. However, as this system is prone to cumulative errors, landmark guidance helps to reduce homing time. often substantially. The elfeetive use bees and ants can make of landmarks as visual signposts (Wehner, 1981) has led 10 the assumption that insects are able to assemble map-like internal representations of the landmarks in their nest environs and then use such 'cognitive maps' 10 find their way to a familiar site, even from points at which they have never been before (Gould. 1986). Although this notion has generated a lot of excitement - and controversymore recent research has shown that ants and bees are indeed able to make intensive lISe of landmark information in relocating nesting and feeding sites, but that they do not incorporate such sites into a map-like system of reference (Wehner & Menzel. 1990; Dyer, 1996). The strategies they employ are more straightforward, foolproof and largely sufficient for the task to be accomplished. One !ask, for example, is to pinpoint the nesting site alter the path integration system has led the animal into close proximity of the goal. As suggested by the experiments described in Fig. 2.1 L ants seem to acquire a two-dimensional visual template - or 'snapshot' - of the three-dimensional landmark array around their nest, and later move so as to match this template as closely as possible with the current retinal image. This matching-ta-memory routine can
be studied best by distorting the training array of landmarks and recording the animal's responses to its altered visual world. In these experimental situations particular matching algorithms are able to indicate at which locations a better (partial) match is obtained than at any other location in their immediate neighbourhood, and it is at these locations thai the local peaks in the insect's
search density profile occur (Cartwright & Collett. 1983). This snapshot-matching mechanism used in landmark guidance might have some fundamental neural traits in common with the template mechanism employed in skylight navigation. The obvious difference is that the skylight patterns are predictable, but the landmark configurations are nol. Hence, the E-vector template can be hardwired. as it actually is. but the landmark snapshots must be acquired during the animal's individual foraging life. In conclusion, the insect obtains landmark-based information not by taking a bird's eye - or a bee's eye - view of the terrain Over which it travels. bUI gains this information successively and by egocentric perceptions during the process of palh integration. This context-bound acquisition and retrieval of
38
CHAPTER 2
~Ibl~~.
(al
0.02 [
Fig. 2.11 Search density profiles 01 desert aJ11~. Cauf;j/rplm [orris. uained 10 the (enIH: of an array of three cylindrical landmarks. The test area containing tluee different landmark arrays is shown in the upper figures. (a) landmarks in the training position. In the training area (not shown) the nest is positioned in the centre of an equidistant triangle formed by the three cylinders. (b) Landmarks separated by twice the training distance. The ams behave as though lost. (c) Landmarks twice the training size and separated by twice the training distance. Again. a match between the stored image ('snapshot') and the current retinal image can be achieved when the ants are in the centre of the landmark array. However, due [0 the larger distance of the: landmarks from the goal (as compared 10 (he training situation shown in (a». motion parallax cues are weaker. and hence the search demity profile is broader than in (a). The results in the three experiments arc in full accord with the matching-IO-memory hypothesis. (Modified from Wehner tt al.• 1996,)
landmark information reduces the danger of gelling inappropriately trapped by similar landmark configurations present elsewhere in the animal's environmenl. For example, the snapshot-matching mechanism, by which the ant finally pinpoints its nesting site, is activated only after the path-integration system has been reset to zero (Wehner el al., 1996). In addilion, the insect can take snapshots at various sites and from various vantage points, and can even align them as sequences of visual images like 'beans On a string' along frequently travelled routes. As such routes can be entered - and familiar sites can be approached - from various vantage points, landmark memories are retrieved and used in quite flexible ways. This flexible use of site-recognition and routeguidance mechanisms leads to navigational performances that might give the impression of map-based behaviour.
2.4.3 Computing interception courses: male pursuits and fly-ball catching Another - and beautifully simple - example of how a difficult computational problem is turned into a tractable one is provided by male hoverflies pursuing and finally catching passing females. As Collell and Land (1978) have shown
SENSORY SYSTEMS
19
la' ' '.
5
''
4
3
2 0.2501
Ib)
.............
5 4
3
2
Fig.2.12 (a) Film recording of a hoverny male. Voluce/la ptlluctlls. pursuing his Quarry. Positions of male Hilled symbols) and quarry (in this experimental case a black wooden
block. 1.5 em in diameter; open symbols) are given every 20 ms and numbered every 100 rns. The broken line indicates the line of sight between male and quarry 20 ms before
the fiy accelerates. As shown by the male's trajectory. the male sets out on the proper interception course. (b) Simulation of the male's behaviour on the assumption lhal he
does
nOI
adopt an interception course but tracks his quarry. Le. turns continuously
towards it. This simulation does not describe the ny's real behaviour. (From Collett & Land.
1978.)
by filming hoverflies in the field, a male fly is able to foresee the female's !light path and to compute the proper interception course (Fig. 2.12a). The male initially does not [urn towards his quarry, when the latter is !irst seen, but immediately sets out on an interception course. Theoretically, this task can be accomplished by a simple calculation only if the male 'knows' the absolute size and the absolute velodty of his female target (as well as his own acceleration when speeding up to catch Ihe target) and incorporates these 'biological constants' into his neural computations. If these conslants are given, the male can obey the simple rule that the size of the turn he makes (l~) depends on the initial position (e,) and velocity (a,) of the target image within the male's visual field as follows: l~ = et - 0.1 ± 180·. The data indicate that the initial turns of the males obey this rule and lead to collisions between the male and the target, if - and only if - the target is a conspeci!ic female. As in biological terms there is no need for a hoverfly male to chase anything other than a femalt:, it is rather likely Ihal natural selection has incorporated into the male's nervous system all the information about the female's flight behaviour that the chasing male needs to know. It is not only a male hoverfly that must compute interception courses, but also a human fielder running to catch a cricket high-ball. In principle, the ball's path across the sky could be computed by a set of differential equatiOns based on the observed curvature and acceleration. Obviously, fielders running for a high-ball do not get engaged in such intricate computations. Instead,
a,
40
CHAPTER 2
they seem to follow one or the other simple rule. One hypothesis holds that they select a running path that maintains a linear optical trajectory for the ball relative to the wicket and the background scenery. In short. a fielder is supposed to adjust his speed and direction so that the (apparent) trajectory of the ball looks straight (linear-trajectory hypothesis: McBeath et al., 1995). If the ball is hit directly at the fielder rather than at an angle to either side, another simple rule might be used, namely to select a running path that keeps the apparent speed o( the ball conSlant (zero-acceleration hypothesis: McLeod & Dienes, 1993). Both strategies, which receive support from video recordings of running paths, do not tell the fielder where or when the ball will land. and hence he does not run to Ihe point where the ball will fall. and then wait for it. They simply sel him on a course which will ensure interceptionand this is all that matters.
2.5 Outlook Behavioural ecologists and physiologists share a mutual interest in each other's e(forts. In the case of the (ormer, this interest is obvious. because behavioural ecologists are keen to learn how neural infonnation-processing mechanisms might have constrained the functional design of the behaviour they analyse in economic terms. It is perhaps less obvious that physiologists should be interested in knowing why it is that a particular neural subsystem mediating a particular kind of behaviour has evolved in one way rather than another. Until recently, behavioural scientists have been preoccupied with the belief that physiological mechanisms underlying behaviour have been designed from first principles (e.g. Mittelstaedt. 1985). They have usually aimed at outlining the complete algorithmic solmion to a given behavioural problem, and then asked the physiologist to discover how this solution is implemented in the hardware of the nervous system. This is the classical approach 'neuroethologists' have entertained for decades. However, neurophysiological analysis is technically demanding, and exhaustive recOnstructions of entire neural subsystems are even more so. All too easily does one get lost amidst the hurly-burly of the higher nervous centres. Are such herculean efforts worth it? In this slate of a[fairs. physiologists have learned an important lesson: that the mechanisms they study are adaptations tailored to particular ecological needs rather than general-purpose processing devices. It is these needs that the physiologist should be concerned with. in order to be able to formulate the right questions in the first place. Let me provide an example by going back to the safe ground of my favourite organism. the desert ant Cataglyphis. Only after we had realized that the awe-inspiring navigational performance of a homing ant could be dissected into a number of simpler special-purpose subroutines, each responsible (or a particular aspect of lhe task, were we able to look properly at the underlying sensory and neural mechanisms. A system must be designed to solve the problem in question - but no more; or as
SENSORY SYSTEMS
41
Diamond (1993), while discussing design limits of physiological systems, and especially the question of how physiological capacilies are matched to their expected loads, has neatly put it: 'How much is enough but not too much?' One day, historians of science might well come to the conclusion that the recent developments in behavioural ecology have had an impact on the way physiologists started to think in evolulionary terms and have caused them to promote what could be dubbed - analogous to Huxley's (1940) connotation of 'new systematics' - 'new physiology' (sensu evolutionary physiology). Hence, there is hope that these recent developments in conceptual approaches will help to bridge the gap in our understanding of what is economically desirablein terms of the functional design of a given behavioural trait - and what is, after all, physiologically feasible.
Chapter 3 The Ecology of Information Use Luc-Alain Giraldeau
3.1 Introduction The use of information is of central importance to a number of behavioural systems. Information, for instance, is involved in social decisions (see Chapters 4. 10 and II), communication (see Chapter 7) and selection of mates (see Chapter 8). It can a((ect population structure (see Chapter 15) as well as the distribution of animals over habitats (Lima & Zollner, 1996; see Chapter 16). The goal of this chapter is to introduce behavioural ecologists to a diverse set of questions concerning information use by animals. The chapter explores how an animal's ecology a((ects the way it gathers, stores and processes information. It asks questions about the types 01 information acquired (but, see also Chapter 2) and the temporal scale of their eflects on behaviour. Section 3.2 reviews information usage as it concerns resource estimation problems while Section 3.3 deals with its use in a pre-detection contex!. Subsequent s<'ctions deal with what some argue to be specialized forms of information use. Section 3.4 considers the extensive comparative research on spatial memory, e,pecially in lhe context of avian food caching. Section 3.5 explores social learning that involves the use of information produced by other individuals (public information). Section 3.6 deals with cultural transmission of information, paying spedal attention to the factors that promote or retard the cultural transmission of behavioural innovations. The chapler ends with some hints of the future research directions for the emergence of a behavioural ecology of information use and cognition.
3.2 Quality estimation This section looks at quality estimation models. The dedsion to seule in a habitat, lorage in a food patch or escalate in combat often hinges on an individual's subjective assessment of the quality of the habitat, food patch or opponent, respectively. The models deal with how animals gain subjective assessments of the parameters required to make adaptive decisions. Most examples involve foraging. possibly because the study of learning uses food almost exclusively as a convenient reward, and because the emergence of foraging theory has underlined the importance o[ cognitive constraints in 42
ECOLOGY OF INFORMATION USE
43
economic decision making. Despite an emphasis on foraging decisions. the intention remains that the wpies should be of relevance to decision proce'ses in other behavioural systems. 3.2.1 Estimating time:
scalar expectancy theory (SET) The ability to estimate short time intervals is likely to be of survival value in a number of circumstances. ranging from foraging to predator avoidance and fighting. In foraging theory. an animal's ability [0 measure and remember time
(e)
(al 0.4
> u c e
0.3
...!
0.2
~
< />
Variable interval
S.4
ge 0.3 ~
...!
C'
0.1 0.0
S.4
0.4
0
15 5 10 Remembered time
0.2 0.1
20
350 (b)
C
Idl
~
0
u
0
0.' >
~
e
~
...
CT
!
S.6
0.3 0.2
O. , 20
Fig. 3.1 (a) An example of three memory distributions resulling from the experience {lr three different time intervals (S) to reward as assumed by SET. Keeping the coefficjent~ of variation constant as S increases forces flattening of the distributions. (b) The pooled frequency distributions of relative giving up intervals (GUliS) for six starlings exposed 10 six values of S. NOI(> that the distribulions are similar in mean and variance. especially {or
the three largest values of S. The greater variance observed for the shonest 5 may indicate a lower limit to the variability of remembered dislribution. (From Kacelnik fl al.• 1990.) (c) SET memory distributions for two equiprobablc inlervals (5 = 4 or 8 5). (d) The mixed distribution (thick line~ resulting from the variable intervals in (c) is superimposed lO the distribution resulting from a fixed interval of 5 = 6 s. NOle Ihe slight negative skew of the distribulion for the variable compared (0 fixed interval dislribution. even though both distribulions have the same mean of 6 s. The negative skew makes il more likely thai a shoner value of S will be remembered from the memory of variable intervals.
44
CHAPTER 3
intervals is of particular importance to the estimation of resource quality whether it involves Bayesian (see Section 3.2.2) or linear operator (see Section 3.2.3) models. Of the many theories concerning how animals estimale lime (Gibbon et 01., 1988; Gallistel. 1990) one, SET (Gibbon, 1977; Gibbon et 01., 1988; Gibbon & Church, 1990; Kacelnik et 01., 1990), has been used in behavioural ecology. SET assumes that animals measure and memorize time elapsed (S) to a signilicant event (i.e. food reward). The memory of S, however, is imperfect and for each sample of S experienced a distribution of values centred around S is memorized. When an animal needs to remember S, it draws a value randomly from the distribution so that the distribution's variance determines the range of recall error (Fig. 3.la). lWo aspects of SET are particularly relevant 10 behavioural ecology; I the relationship between the distribution's variance and S; 2 the ellect Ihat variability in experienced S has on recall error. Variance and interval size
SET hypothesizes that the variance of the memorized distribution increases with the magnitude of S but that the coefficients of variation of all stored distributions remain constant. Hence, the memorized distributions for dillerent S values can all be made idenlical by simple scalar Iransformations of each other (Fig. 3.la). For instance, starlings (Sturnus vulgaris) required 10 exploit food patches in operant devices, have a temporal memory thaI conforms 10 properties of SET (Brunner eT 01., 1992). Starlings foraged in an experimental environment where palches proVided prey al a lixed unchanging interval S, but depleted suddenly after some unpredictable number of prey. The only cue that signalled depletion was time since the last prey capture. A bird that had perfect lemporal estimates of S would have left the patch once S had elapsed without a prey. Leaving sooner would have cost some prey, leaving later would have wasted time. The distribution of intervals Ihat starlings waited belore giving up a patch as empty was indeed bell-shaped as SET assumed (Fig. 3.1 b). Moreover, variance in observed waiting times grew with the S experienced and mosl distributions were superimposable following a scalar transformation exactly as predicted by SET (Fig. 3.lb). Variability of experienced S
Up to now we have considered cases where animals experienced constant, unchanging values of S. In most natural situations, however, intervals between significant events such as the appearance of drift prey in a brook or the courtship displays of a mate will vary. For instance, prey may be encountered at variable intervals ranging say, from 2 to 8 s. What effect will this variability have on the memory of intervals for that food patch? SET postulates that Ihe memory
ECOLOGY OF INFORMATION USE
45
distriblllion of time imervals is lhe mixlUre (aggregate) of the distribUlilns corresponding 10 each experienced S. So. for instance. an oplion that is equ.ll1y likely 10 provide rewards after 4 or 8 s will have a temporal memory that is lhe mixture of dislributions oround 4 and 8 s (Fig. 3.1 c). An inevitable consequence of the scalar propen y that forces memory distribulions of longer S to be lIaller. and hence more variable than those for shoner S. is that the mixed distribution will be negatively skewed compared 10 a distribution for a fixed value IIf S wilh the same mean (Fig. 3.1 d). It follows that when animals are confromed with a choice belween a fixed and a variable option with the same mcan imerval. lhey are slightly more likely to recall a shaner interval from Ihe aggregale than from the fixed distribution. SET. therefore. predicts preference for the variable alternative. assuming shoner delays are preferred (Fig. 3.1 d). The consequences of SET can be of broad significance 10 models of palch exploitation (Kacelnik er al.. 1990; Reboreda & Kacelnik, 1991; Kacelnik & Todd, 1992). These have just started 10 be considered in other foraging systems such as risk sensilivity (Reboreda & Kacelnik. 1991). diet choice (Shelliewonh & Plowright, 1992) and sampling (Sheltlewonh et al., 1988). Extending SET to estimalion of reward size (Bateson & Kacelnik. 1995) as well as fighting ond mate choice decisions may prove valuable.
3.2.2 Estimating quality in unchanging environments: Bayesian models Bayesian estimation is a way of forming an estimate of an object's value on the basis of lhe combination of current sample information acquired from the objecl and a prior expectation of object value distribulion in the environmcnt. The exact way in which currenl and prior information are combined characterizes the Bayesian updating process (McNamara & Houston. 1980; Stephens & Krebs, 1986). Transposed to a patch foraging context, Baye,ian updating involves combining the prior expectation of the dislribulion of prey in patches with current patch-sample information 10 generate an updated estimale of the number of prey currenlly remaining in the patch (McNamara & HouslOn. 1980; Green, 1980. 1984; Iwasa et al.. 1981). A simple graphical illustration of Bayesian updating under three differenl expected patch quality distributions should suffice 10 make lhe poinl qualitatively. Consider three possible types of environmems. In one, prey are underdispersed such that the number of prey per patch (quality) is highly variable with patches containing either many or few prey. In another, prey are overdispersed such thaI each palch contains relatively similar numbers of prey. In a lhird, prey are randomly and independently dispersed. For all three types of environment, time elapsed while in the patch and number of prey encounlered are suffident patch-sample statislics foreffident Bayesian estimation of current patch quality, and the updated estimates must decline with increa,ing time in the patch (Iwasa et al., 1981; Fig. 3.2). Depending on Ihe prior expected
46
CHAPTER 3 101
Ibl
Prey underdispersed
Prey overdispersed
Ie) Prey dispersed independently
= . . .. _--l.
.........................................=: .....:::: .....
~
lime in patch (I)
Fig. 3.2 Graphical represcOlation of Bayesian estimation of the number of prey items Jert in a patch in environments that differ in the way prey are distributed. The thick lines shuw how the estimated number changes over time and the number of encounters (X) with prey. The dashed horlzonlalline gives the critical number at which an organism ought to abandon the patch (arrow). Note that in all three cases the expected number of prey remaining always declines with time spent in a patch. The difference lies in rhe effea of prey en QUlllers on the eSlimated number remaining. When prey are undcrdispersed (a), each encounter with an item raises the expected number of items remaining in the patch. be ause patches tend to contain many or very few items. However. when prey are ovtrdispersed (b). each cncoumer makes the expected number of prey remaining dedine. because patches vary little in prey number. Hence. the only useful information provided here is thatlhe patch nuw contains one Jess prey item. When prey are dispersed randomly and independently (c). prey enCOunters say nothing about the number of prey remaining becausc prey arc dispersed independently. In Ihal situation the best foragers can do is 10 decrease their estimate of the number of prey remaining as a function strictly of time in the patch. (Modified from Iwasa et al.. 1981.)
ECOLOGY OF INFORMATION USE
47
distribution of prey, however, encountering a prey has radically different consequences on the updated estimate of the number uf prey remaining in the patch. For the underdispersed environment, each prey encounter increases the likelihood that the patch contains still more prey because patches comain eithcr very many or very few prey. Su, the updatcd estimate of thc number of prey remaining in the patch increases following each encounter (Fig. 3.2a). In the overdispersed environment, whcre all patches are of similar qualily. encountering a prey now means Ihat the patch likely contains one less prey. So. thc updated estimate of the number of prey remaining in the patch now declines wilh encounters (Fig. 3.2b). In the environment where prey arc distributed randumly and independently, epcountering prey says nothing about the number of prcy remaining in the patch and so has no effect on the updated estimatc. Here. the best policy is to decreasc thc estimation with patch time inc pective of encounters (Fig. 3.2c). Because updated estimatcs incrcasc with prey encounters in underdispersed environments, thc use of Bayesian estimation in those cases predicts that the mosl successful individual: I has the highest updatcd estimate of the number of prey remaining in the patch; 2 exploits the patch more extensively; 3 requires longer unrewarded search before its updated estimate declines to the threshold for patch abandonment (Fig. 3.2a). Predictions for an overdispersed environment arc the exact opposite (Fig. 3.2b). The patch departure of budgerigar (Melopsi/lacus undulatus) dyads foraging in the laboratory for millcl seeds hidden in fine gravel follow predictions of Bayesian updating (Valone & Giraldeau, 1993). The birds foraged in environments where seeds were either under- or overdispersed. In underdispersed environments the birds' Success predicted their order of departure from the patch. A bird left first mostly when it was the least successful member of the pair and last when it was the most successful. In addition, the most successful member of the pair tolerated longer runs of unrewarded search before leaving than the less successful one. When the same birds foraged in overdispcrsed environments, their individual success no longer predicted the order of patch departure nor the length of unrewarded search (Valone & Giraldcau. 1993). Behaviour consistent with Bayesian eSlimation has also been reported in desert granivores (both birds and mammals: Valone & Brown, 1989). inca doves (Columbina inca: Valone, 1991), black-chinned hummingbirds (Archilochus alexandri: Valone. I992a), cranes (Grus grus; Alonso et al.. 1995) and bluegill sunfish (Lepomis macrochirus: Wildhaber el al.. (994). One non-foraging example concerns the male amphipod (Gammarus lawrencianus) that is increasingly reluctant to give up a female the longer he has amplexed with her, as if upgrading her reproductive value during amplexus (Hunte el al., 1985). Future work should explore how animals construct prior distributions, how and
48
CHAPTER 3
whether they update these priors, huw they collect patch sample information and the frequency with which they update their estimates. Bayesian estimation may be a powerful tool for estimating patch quality, especially when prior distributions do not change quickly over time. Changing priors makes updating patch sample informalion messier because individuals must alter their prior expectation as they update their current estimates. Unear operator models may offer simpler devices for tracking changing environmental conditions.
3.2.3 Tracking quality: linear operator models Here we consider situalions where resource distributions change such that no stable prior expectation can be acquired. To track change in a useful way, individuals must distinguish between local slOchastic fluctuations in parameter values from more significant directional movement. They must also take into account the declining reliability of information over time. Tracking models usually reduce the influence of local irrelevant fluctuations by integrating information (i.e. averaging) over some time·span. Declining reliability is generally dealt with either by discarding or devaluating outdated information. Discarding is used by memory window models where means of environmental parameters are estimated over a range (window) of experiences, dropping the oldesl events as new ones are added to the estimate (Cowie, (977). Despite their appeal, memory window models may be difficult to lest (Cowie & Krebs, 1979; Mackeney & Hughes, 1995). The alternalive approach, which involves devaluating oUldated information, is more commonly invoked and forms the basis of linear operator learning rules (Kacelnik 'I al., t987). Studies of linear operator models (also called exponentially moving weighted averages: Devenport & Devenport, 1993, 1994) usually involve two components: (i) an algorithm that updates information (the linear operator); and (ii) one that prescribes lhe behavioural decision (the decision rule) once estimates of the alternatives have been updated (Houston & Sumida, 1987). A number of different linear operator rules have been proposed as updating algorithms (reviewed by Kacelnik et al.. 1987). One, the relative payoff sum (RPS) (Harley, 1981), has historical importance if only because it had initially been described as a serious candidate for an eVOlulionarily stable learning rule, a rule that if adopted could not be outcompeted by any other updating algorithm (Harley, 1981; Maynard Smith, 1982). Whether the RPS is an evolutionary stable strategy (ESS) rule, however, remains controversial (Houston & Sumida, 1987; Tracy & Seaman, 1995). Nonetheless, the rule has been tested experimentally with some qualitative success (Milinski, t 984). so it is worth going over it in more detail, if only as an illustration of how it works. The updating part of lhe RPS rule, like all linear operators, partitions the experience or each behavioural alternative into past and present. The past and
ECOLOGY OF INFORMATION USE
49
lhe present are given a relative weight and then summed to yield an updaled estimate for a given behavioural alternative. The updating process is repeated for each pOlential alternative. The decision component of the RPS rule uses the updated eSlimates 10 delermine which behavioural alternative should be expressed in lhe future. For the RPS Ihe decision rule is to use an option in proportion to its relative value, matching: V(I) P'(t + I) = --'-.':'-'--
I.
Vj(l)
J-I
where P,(I + I) is the probability of responding to option i on trial I + I. The numerator gives the value of alternative i after the last trial I, while lhe denominator is the sum of values for aLI k alternative after lrial I. Whelher malching is an appropriate decision rule is not our concern here (Houston & Sumida, 1987). The main focus is nn how RPS updates estimates of V,(I). The RPS rule updates the value of alternative i after trial I as a linear opera lor of the form: Vi(l)
= exvi(I-I) + (I
-ex)ri + Qi(l)
where ex (0 < ex < I) the memory parameler. selS the weight of past (ex) and present (I - ex) experience. Qi(t) is a measure of the rewards obtained from alternative i during trial I. ri acts as a prior or residual value for alternalive i (V,(O) = r,). The peculiarity of RPS is that without rewards, the updated value of i converges to ri rather than to zero as it would in all other linear operator models. The residual ensures some non-zero chance that any of the k alternatives will be used, even when one alternative is overwhelmingly more valuable, acting somewhat like an insurance against overlooking a previously depleted resource that has unexpectedly recovered. To date, empirical support for the RPS rule is weak at best (Milinski, 1984; Kacelnik & Krebs, 1985). Some sludies provide results lhat are partly consistent with linear operator learning rules. For instance, bOlh lracking of travel times by foraging slarlings (Culhill el al., 1990. 1994), and monitoring of interprey encounter limes in pigeons (Shettleworth & Plowrighc 1992) seem to follow linear operator rules. In addition, linear operator rules are also consistent with the observation that distance travelled to next nower by bumble bees (Bombus bimacularus) depends on the sequence of the last three nowers visited (Dukas & Real, 1993). In all of these recent cases, foragers devalued past experience strongly, sometimes responding only to the very last event (Kacelnik & Todd, 1992; Cuthill el al., 1994). Future work should explore whelher the weight of past versus corrent experience is a fixed cognitive cunstraint adapted to a species' ecology (Mackeney & Hughes, 1995) or possibly a decision variable that can be adjusted to each specific tracking problem encountered by individuals (Valone, 1992a; Devenport & Devenport, 1993; 1994; Cuthill et 01., 1994).
50
CHAPTER 3
3.2.4 Estimating quality in social environments:
using public information Conceivably, in social environments, the activity of other individuals can provide 'public information' (Valone, 1989) concerning food location or quality. Because the rate of Bayesian updating (see Section 3.2.2) is limited by an individual's sampling rate, a group of G individuals using public information can sample at up to G times the raw of solitary individuals. reaching reliable estimates of patch quality G times faster or G times as reliably for any given sampling period (Clark & Mangel. 1986). Public information. therefore, can provide an advantage to group foraging behaviour (Clark & Mangel. 1984). Starlings are social foragers and sample a food patch by probing their bill into the substrate (Tinbergen. 1981). Public patch-sample information, in this context. is provided by other individuals' probes of the patch. Templeton (1993) used starlings 10 explore whether public information is used in a patch-sampling context. In one laboratory experimcnt (Templeton & Giraldeau. 1996). starlings foraged on artificial patches that offered 30 probe sites. The birds expccted patches to be either partly or completely empty so they had to probe a number of empty sites before deciding that a patch was empty. To avoid effects of food depletion the birds were always tested while probing totally empty patches. The number of probes made before giving up a patch was measured for each subject as it foraged alone then in the company of a partner. lWo types of partners were provided. A low public information partner had been trained to probe just a few holes while a high public information parmer had been trained to probe all 30 sites. Starlings altered their probing in a way that is consistent with the use of public information. Individuals probed most when foraging alone. As expected from public information use they sampled fewer holes with the high public information than the low public information partner (Fig. 3.3). Since no food was present during testing. the difference cannot be due to faster food depletion with the high public information partner.
Fig. 3.3 Templeton and 30
~
~
.8
e
Q.
'0
20
.8E ~
c c c c
to
Giraldeau's (1996) experimental evidence of public information
use by starlings faced whh a palch.qualityestimation
problem. The mean (± SEM) number of empty sites probed by five starlings before giving up a patch is highest when tested alone. and lower when foraging with a partner. Moreover. when
:l:
foraging with a partner. a subject 0
Alone
Low Treatment
High
samples more when paired with ' low-information panner than
with a high·in(ormalion partner.
ECOLOGY OF INFORMATION USE
51
Templeton's (1993) starling work forced individuals into subject and partner roles. However. in normal drcumstances sampling within groups will likely be a frequency-dependent game with some individuals exploiting the public information provided by others without contributing any in return. Indeed. when pairs of starlings are confronted with the problem of tracking the quality of a variable alternative food patch. individuals spedalize into samplers and non-samplers (Krebs & Inman. 1992). Consequently. foraging groups may generate much less public information than was antidpated initially (Clark & Mangel. 1984. 1986).
3.3 Detecting cryptic prey: search image or search speed? Prey items can pose detection problems for a number of different reasons. two of which have been the focus of considerable study: crypticity and mimetism. Research on model-mimic systems focuses mostly on information processing and decision-making aspects (Getty. 1985). Most of its development involves the use of signal detection theory. a variam of Bayesian statistical decision theory (Getty. 1985; Getty & Krebs. 1985; Getty et al.• 1987). Behavioural studies involving the detection of cryptic prey. on the other hand. focus mostly on the perceptual (Tinbergen, 1960; Dawkins, 1971) and attentional (Dukas & Ellner, 1993) processes as well as optimal search tactics (Gendron & Staddon, 1983; Gendron. 1986; Reid & Shettleworth, 1992: Getry. 1993; Getty & Pulliam, 1993; Dukas & Clark, 1995). This section concentrates on detection of cryptic prey. Tinbergen (1960) suggested that foragers confronted with cryptic prey develop a 'search image' that enhances their detection. The formation of a search image. which is thought to involve only the detection process, requires successive encounters with the prey. and strongly imerferes with the concurrent acquisition or use of a different search image (Pietrewicz & Kamil, 1979; Lawrence & Allen. 1983; Gendron. 1986; Guilford & Dawkins. 1987; Reid & Shettleworth. 1992). Support for the search image hypothesis has been reported in a number of observational and experimental studies (references in Guilford & Dawkins, 1987; Endler. 1991; Reid & Shettleworth, 1992). Most evidence consists of apostatic prey selection, I.e. that a cryptic pre~' type is practically excluded from a forager's diet when it is rare. but is over-represented above some threshold abundance. Dawkins (1971) cautions that alternative hypotheses can account for apostatic prey selection. For instance, learning where or how to search for prey. improving capture Or handling efficiency. and changing prey preference can all lead to apostatic selection without a search image. Apostatic selection can also result when cryptic prey cause foragers to reduce their search rates in order to increase the likelihood of detecting cryptic prey (Gendron & Staddon. 1983; Gendron, 1986; but see also Getty, 1993; Getty &
52
CHAPTER 3
Pulliam. 1993). So. clear. irrefutable evidence for a search'image remains elusive (Guilford & Dawkins. 1987). The reduced search rate hypothesis has emerged as a serious challenger to the search image hypothesis (Gendron & Staddon. 1983; Guilford & Dawkins. 1987). It slates that predators include more cryptic prey in their diet. not because they detect them more efficiently. but because they search more carefully. The search rale and search image hypotheses make a number of common predictions. but the crucial prediclion is lhat reduction in search rate allows a predator to detect any other equally cryptic prey type whereas a search image supports only one prey type. So. a predator offered two equally cryptic prey types concurrently will choose only one if it forms a search image but will be equally likely to find eilher if it reduces its search rate. This prediction was tested with a series of elegant experiments involving pigeons (Columba livia) (Reid & Shellieworth. 1992). The experiments consisted of presenting pigeons with wheat grains dyed either yellow. brown or green on a multicoloured aquarium gravel background. Preliminary experiments showed lhat brown and green seeds were equally likely to be selected by pigeons when presented in equal numbers on the same multicoloured background. They were. therefore. considered equally cryptic on that background. Yellow seeds. however. were conspicuous. Birds pecked in an operant device to gain access to a small piece of multicoloured substrate. First. the substrate contained only one seed. Then. after a number of single-seed presentations'il contained two. If pigeons form a search image during Ihe starting series of encounters with a prey type. then when offered a choice between IwO equally cryptic types they should only peck at the seed type used in the initial series. If. inslead. they reduce their search rate then Ihey should be equally likely to peck at either cryptic seed type when offered a choice. Pigeons pecked more frequently than expected by chance at the prey type used in lhe inilial run. as predicted by the search image hypothesis. Familiarity with a prey type acquired during the inilial runs cannO! explain the preference because no preference was found when the same seed types were presented against a uniform beige background thaI made them conspicuous. As predicted by the search image hypothesis. pigeons developed a search image that was specific to a Single cryplic prey type al a time and only developed such an image if the prey was cryptiC. The cognitive level at which search images arc formed remains uncertain. Do search images involve perceplUal or 3nenlional mechanisms? A plausible cognilive account of search images involves predators learning to allend to a specific feature. or combination of features of a prey type thaI enhance detection (Reid & Shettleworth. 1992). The definition supposes that search images can allow the detection of several prey types if they happen to share Ihe fealures the predator learns to attend to. The extent of interference bel ween the concurrent operation of different search images and the factors that govern the encounter rates required for search image formation remain open questions.
ECOLOGY OF INFORMATION USE
53
3.4 Spatial memory as an adaptive cognitive specialization Here we explore the possibility that the use of information in some cases qualifies as a specialized adaptation thai involves qualitatively or quantitatively different learning and memory processes. Two distinct approaches can be used to address the issue: comparative and experimental. The comparative approach asks whether similar selection pressures gave rise. through convergent evolution. to common behavioural and neuroanatomical adaptations (Krebs. 1990; see Chapter 14). The experimental approach. on the other hand. compares the performance of different species confronted with a common laboratory problem. The objective is to establish whether different cognitive adaptations involve augmentation or reduaion of common cognitive sys-
tems or. instead. are supported by separately evolved dedicated systems (Shetlleworth. 1990).
3.4.1 Comparative approach There is little doubl that an animal's ecology has placed seledive pressures on its neuroanatomy. especially when sensory and motor processes are involved (Krebs. (990). Now there is increasing evidence that selective pressures influence brain strudures normally associated with higher levels of information processing. For inslance. Pagel and Harvey (1989) found that insectivorous and frugivorous bats and primates tended to be more encephalized than folivorous ones. The difference is thought to be related to the larger ranges required by insedivores and frugivores coupled to the spatial and temporal unpredictability of those food types compared to leaves. Similarly. birds with larger song repertoires possess larger higher vocal centres and brain nuclei associated with song production. Crepuscular and nocturnal birds have relatively larger olfactory bulbs than diurnal birds (Healy & Guilford. 1990). Finally. birds with larger relative forebrains are more frequently reported to use novel foraging techniques (Lefebvre et al.• 1996a). Of all areas comparing ecology. behaviour and neuroanatomy. the use of spatial memory for food caching represents the most extensive research programme involving the greatest diversity of taxa (e.g. Krebs il al.. 1989; Sherry et al.. 1989. 1992; Shettleworth. 1990. 1995; ClaylOn. 1995a; Jacobs. 1995). The following seclions.therefore. focus on spatial memory in the context of retrieving cached food. Note Ihat parallel research programmes also occur in bird song ( ordeen & Nordeen. 1990; Marler. 1991; Nottebohm. 1991; Catchpole & Slater. (995).
54
CHAPTER 3
Avian food caching: spatial memory alld the hippocampus
Fuod storing has evolved in a wide range of taxonomic groups (Vander Wall. 1(90), but careful comparative work in birds has focused on two groups of passerines: the Paridae (titmice and chickadees) and Corvidae (jays and nutcrackers), In parids, single pieces of food (seeds, insects) are stored for a few hours to a few weeks in hundreds of scaltered cache sites that are never re-used (Krebs, 1990). Corvids store many thousands of seeds for longer times, up to 7-11 months in Clark's nutcracker (Nucifraga columbiana: Vander Wall, 1990). Monocular occlusion experiments have provided elegant demonstrations oj the use of spatial memory in food-storing birds. The procedure has animals storing and then recovering food while only one eye is available (Sherry er al.. 1(81). It provides evidence for memory because, in birds, visual pathways of the two eyes cross completely at the optic chiasm such that, for the most pan, each hemisphere receives input from Ihe contralateral eye (Clayton & Krebs, 1993, 1994a,b). So, marsh tils (Parus pall/stris) allowed to store seeds while one eye is covered with a small, opaque, plastic cup fail to recover food above chance levels within the first 3 h when a different eye is available for retrieval. indicating that information stored in the other hemisphere (memory) was required to recover the seeds (Sherry et al., /981; Clayton & Krebs, 1993). Similar results were also obtained with corvids (Clayton & Krebs, 1994b). Given that avian food caching requires remembering the spatial locations 01 hundreds to thousands of caches for periods ranging from days to months, the question remains whether some specialized neuroanatomical structure is required to accomplish the task. [n mammals, considerable experimental evidence points to the hippocampus as a brain structure associated with higher cognitive functions such as memory and processing of spatial information (Squire, 1992). Lesion sludies with black-capped chickadees (P. atricapillus) provide indisputable evidence that the avian hippocampus, which includes the hippocampus and area parahippocampus, are involved in spatial memory (Sherry & Vaccarino, 1989). Birds that were either unlesioned, lesioned in the hippocampal region or Iesioned in the hyperstriatum ventralis, a brain area unrelated to spatial memory, stored seeds normally in aviaries containing artificial trees. However, only unlesioned birds and those with non-hippocampal lesions showed memory for cache sites. Hippocampal-Iesioned birds could not recover their seed caches above chance level, although their ability to learn other tasks such as colour discriminations appeared totally unimpaired (Sherry & Vaccarino, 1989). Studies of avian hippocampus development also demonstrate involvement of the avian hippocampus in the retrieval of cached food (Clayton & Krebs, 1994c; Clayton, 1995a,b; Healy et aI., 1995). For instance, the volume of the hippocampal area relative to the rest o[ the relenceph;llon in the food-storing marsh til depends upon actually retrieving
ECOLOGY OF INFORMATION USE
55
slored food. The act of retrieving stimulates growth uf hippocampal volumes while deprivation of opportunities to cache and retrieve induces loss of hippocampal volume (Clayton & Krebs. 1994c). There is even some suggestilln that the onset of seasonal hippocampal neurogenesis described in some foodstoring birds may be triggered by the seasonal need to cache and recover stored food (Krebs el al.• 1995). Comparative studies involving both New and Old World avian taxa show that food caching is associated with enlarged hippocampus (Fig. 3.4). Hence. selection pressures that give rise to food caching also lead to reallocation of nervous tissue in favour of the hippocampus. Note that increased hippocampal volume does not necessarily imply an increased total brain size (Harvey & Krebs. 1990). Some unknown factor. therefore. appears to constrain a species' total brain volume. As a consequence. selection for increased hippocampus volume might have imposed some potential. yet undocumenlt:'d. cost to other neural functions (Krebs. 1990). Nonetheless. these comparative avian studies show a pOsitive relationship between relative hippocampus volume and food-sloring behaviour in two phyletically different groups (Fig. 3.4).
Turdidae Troglodytidae
(8)
0 0
Sylviidae
0
Sturnidae
0
Ploceidae Fringillidae Sinidae
0 0
Paridae
0
Corvidae
0
•
•
·0.15 ·0.10 -0.05
•
0 0.050.100.150.200.25 Relative hippocampus size
(b)
Passeridae Fringillidae Emberizinae Cardinalinae
Parulina8 Sturnidae Mimidae 0
Turdina8
Sylviinae Troglodytidae Corvidae Paridae
0 0 0
0
0 0
0 0 0
Sittidae
-0.2
·0.1
0
0.1
• •• 0.2 0.3
0.4
Relative hippocampus size
Fig. 3.4 Residual varia lion in hippocampal
volum~ after body size and telencephalon volume effects have been removed through multiple regression. Dark squares are food storers, open squares non-food-swrers. (Data from (a) Krebs tl al., 1989; (b) Sherry tt ai, 1989. Figure taken from Krebs. 1990.)
56
CHAPTER 3
Sex differences in spatial memory and hippocampal size The spatial memory of voles has been studied extensively because some species show dramatic sexual dimorphism in home-range sizes depending on their mating system (Gaulin & FitzGerald, 1986, 1989). For instance. adult males of the polygynous meadow vole (Microtus pennsylvanicus) have home ranges in the breeding season that are four to five times larger than those of adult females or immatures (Gaulin & FitzGerald, 1989). By contrast, the monogamous prairie (M. ochrogaster) and pine (M. pinetorum) voles show no sexual dimorphism in home-range size (Gaulin 5' FitzGerald, 1986. 1989). Polygynous males are sexually selected to occupy larger ranges because the greater a male's range, the more females it interacts with (Jacobs, 1995). In line with expectations, the hippocampus 01 males of the polygynous species are significantly larger than the female's, while no such sexual dimorphism is apparent in the monogamous species (Jacobs et al.. 1990). Moreover, the meadow vole's sexual dimorphism in hippocampal size disappears outside the breeding season, a time when the male's home range is of comparable size to the female's. That means that hippocampal cells are grown each year prior to the breeding season and then reabsorbed after breeding; a seasonal neurogenesis also reported for vocal centres of canaries. Increased hippocampal size, therefore, can also be a sexually selected secondary sexual character (Jacobs, 1995). The size of the hippocampus can be an adaptation to specific ecological conditions. Size, however. is only one representation of a neuroanatomical structure's information processing and storage capabilities. Finer level cytoarchitectural and neurochemical changes could also influence competence. sometimes without affecting a structure's size (Krebs, 1990). The future will no doubt reveal the extent to which ecology affects finer levels of the brain's structure.
3.4.2 Experimental approach The spatial memory used by food-caching animals could be a cognitive specialization in two different ways. First, it can be qualitatively different from other forms of spatial learning and rely on a separate and entirely dedicated cognitive process. Alternatively, it may simply have improved the all-purpose mechanisms that are commonly used for the processing of any spatial information. Determining which type of specializalion is involved requires making different kinds of comparisons. To establish that the memory is a distinct and dedicated cognitive process one must compare food-storing animals on food-storing and non-food-storing spatial tasks. Establishing whether il is a general enhanced spatial competence requires comparing sloring and non-storing species on some common non-storing spatial task.
ECOLOGY OF INFORMATION USE
57
Avian food-slorer spalial memory 1101 a dedicaled food-caching specializalion If spatial memory used by food-storing animals is a distinct and dedicated adaptive spt'cialization, then food storers would only exhibit their enhanced
spatial memory in the food-caching context. Any other spatial task would tap into a more mundane cognitive system probably shared with non-caching
animals. Experiments designed in part to answer this question have been conducted with corvids mostly by Kamil, Balda and colleagues, while equivalent work on parids has been carried out by Krebs, Sherry, Shettlewonh and colleagues (reviewed by Krebs, 1990; Shettleworth, 1990, 1993, 1995). While several North American corvid species cache food, each relies on it to a diflerent extent. At one extreme, Clark's nutcracker and pinyon jay (Gymnorhinlls cyanocephaills) inhabit cold places and rely heavily on stores of pine seeds during the winter and spring. In comparison, both scrub (Aphelocoma coerulescens) and Mexican (A. ultramarina) jays inhabit warmer climes and rely much less on food stores (Balda & Kamil. 1989; Balda er al., 1996). If these animals' spatial memory is specialized only for use in cache recovery, then when tested on other types of spatial tasks their ability should be uncorrelated to their rankings on a food-caching task. In a test that involved caching food and then retrieving it within an indoor arena, nutcrackers and pinyon jays outperformed scrub jays although pinyon jays outperformed nutcrackers becallse they tended to hide seeds in clumps facilitating recovery (Balda & Kamil, (989). In a non-caching task, the birds had to return to sites whcre they encountered food earlier within a radial-ann maze analogue (Kamil el al.. 1994). The task differed from caching because the birds did not store the seeds they encountered. The birds were trained to expect a buried seed in each of 12 sand-filled cups arranged in a circle on the floor of an aviary. They were then allowed into the aviary while only four randomly chosen cups were uncapped and hence available for exploitation. Following a retention interval. the birds re-entered the aviary. This time eight cups were uncapped, the same four they exploited in the initial phase plus another random set of four. The efficiency with which the subjects directed their search to the new set of cups measured their spatial memory. The ranking of spatial memory in both Sloring and non-storing spatial tasks were similar: nutcrackers and pinyon jays were more efficient than scrub and Mexican jays. So, the spatial memory used by food-caching corvids was unlikely to he a distinct, dedicated cognitive specialization. Instead, it was more likely supponed by an enhanced general spatial cognitive system. Parallel experimental results have been obtained for two species of foodstoring parids: black-capped chickadees and coal tits (P. arer: Shettleworth er al., t 990). Both species were as efficient at returning to locations where they encountered seeds they had not cached as they were at returning to cache locations. Hence, spatial memory used both by food-storing parids and
58
CHAPTER 3
corvids seems to be an enhanced wrsion of a general spatial cognitive ability that can and is used in a non-caching context.
Evidence that avian food stOre/'S have superior spatial memory Because both corvids and parids use their food-storing spatial memory for non-storing spatial tasks, it should be simple to compare storing and nonstoring species on a common spatial problem. If storers have a better spatial memory than non-storers, they should exhibit anyone or more of the following features: learn spatial tasks [aster, memorize more spatiai information and support longer retention intervals without loss of memory. In corvids, champion storers like Clark's nutcracker and pinyon jay are compared to less assiduous storers like scrub and Mexican jays on a radial arm problem (Kamil et al.. 1994). As expected, nutcrackers and pinyon jays are able to remember more spatial information than Mexican and scrub jays but inexplicably they also forget it faster. After 300-min retention intervals, species no longer differ. This is especially surprising when one considers that retention intervals for Clark's nutcrackers in the wild is on thc order of months. These surprisingly small differences in spatial memory among corvid sp,'cies are comparable to those found between storing and non-storing parids. The spatial task commonly used in parid comparative experiments is called 'window-shopping'. The birds are allowed into an aviary where a number of locations contain a peanut placed bchind a window. Then, after a retention interval the windows are removed and the animal allowed to return. When subjected to these kinds of window-shopping tests, differences in performances between the storing coal tit and non-storing great tit (P. major) were small, although they consistently favoured the storing species (Krebs et al.. 1990). Except perhaps for the Clark's nutcracker, the small differences observed between the spatial memory of storing and non-storing species is remarkable given that the natural history of food-storing birds suggests an ability to memorize a staggering number of distinct spatial locations (Shettlewonh, 1995). A more spectacular set of quantitative differences in spatial memory may have been in order especially since food storing is associated with relatively larger hippocampus. Some factors may have prevented food-storing species from shining as well as they could when compared to non-storing species (Shettleworth, 1990, 1995). Jt is also possible that the major difference in the spatial cognition of storing and non-storing species is more subtle and qualitative, affecting mostly the kinds of information animals pay attention to and the way it is stored rather than the quantity (Brodbeck, 1994; Clayton & Krebs, 1994b,d; Brodbeck & Shenleworth, 1995). For instance, monocular occlusion experiments suggest that long-term spatial memory is lateralized (i.e. located in one hemisphere) in food-storing birds but not in non-storing ones (Clayton & Krebs, 1993, I 994a,b).In marsh tits, for instance, memory of
ECOLOGY OF INFORMATION USE
59
100d locations transfers from the left to the righl hemispheres 3-7 h after storage, a time during which elliciency of cache retrieval declines. The long-term storage of spatial memory of food-storing birds, therefore, is located in the right hemisphere (Claylon & Krebs, 1994a). Moreover, lhe left eye system of food-slOring birds appears lO specialize on storing spatial information while the right eye system specializes on non-spalial cues, both in parids and corvids (Clayton & Krebs, 1994d). Research on the spatial memory of food-storing birds, therefore, suggests lhal selective pressures may have altered the mechanisms of memory processing and storage more than its capacity.
3.5 The ecology of social learning The expression 'social learning' has been used 10 mean a wide range of behaviour (Gale!, 1988. provides an exhaustive list). Here, it refers to the use of public information (see Section 3.2.4) about places, objects and behaviour: area, object and behaviour copying, respectively.
3.5,1 Area copying Area copying, also called local enhancement (Thorpe, 1956; Gale!, 1988), occurs when an individual directs its behaviour towards the place where others are currently aClive. In addition to birds (Krebs et al., 1972; Barnard & Sibly, 198 I), area copying has been reported in social spiders (Ward, 1986), fish (Pitcher el al., 1982) and mammals (Gale£, 1990). Although common, area copying has generated little interest in its ecological delerminants. For instance, it is conceivable that the area to which individuals are auracted depends on the size and distribution of resource clumps (Krebs, 1973; Barnard & Sibly, 1981). Moreover, how long should an individual's behaviour be influenced by the sight of another and how is this related to the animal's ecology? Krebs and colleagues' parid studies report that area copying effects are detectable for only moments, while McQuoid and Gale! (1992) report that Burmese jungle!owl (Gallus gallus spadiceus) show significant area copying effects 48 h after the initial observation of other individuals foraging in a given place. ft would be use!ulto explore how factors such as food patch ephemerality affect the longeVity of area copying.
Adaplive hYPolheses Area copying can allow animals to avoid dangerous places. For instance. in red-Winged blackbirds (Agelaius phoeniceus), flocks that contained experienced individuals induced naive birds to forage first in places with safe food but it failed to teach them to avoid places with poisoned food (Avery, 1994). Area copying can increase foraging gains, whether in terms of foraging rate or reduced risk of starvation (Caraco, 1981; Hake & Ekman, 1988). For instance,
60
CHAPTER 3
area copying allows grealtits foraging in flocks of four to find food more quickly than in flocks of two or when foraging alone (Krebs cr al., 1972). The same was observed for flocks of green finches (Cardue/is elrloris) (Hake & Ekman. 1988). Area copying is often portrayed as an information-sharing system where
individuals maintain contacr with their group mateS as they search for their own food (Ranta et al., 1993; Ruxton er al.. 1995). However, keeping up bOl h activities Simultaneously can be difficulL under a wide range of circumstances (Vickery er al.. 1991). Area copying. therefore. can also be modelled as individuals ailernating between searching for (producing) their own food. and exploiting (scrounging) the food discovered by others: a producer-scrounger game (Barnard & Sibly, 1981; Caraco & Giraldeau, 1991; Vickery er al.. 1991). Whatever the scenario. the payoff of area copying likely declines with the frequency of its usc. whether benefits are expressed in terms of individual feeding rate (Giraldeau er
3.5.2 Object copying Object copying, also called stimulus enhancement (Gale!. 1988) or 'releaserinduced recognition' (Subowski. 1990), is similar to area copying in that it directs behaviour. However. the behaviour is directed to an object that matches the type attended by others, rather than a place. It differs from area copying in that the effect is not constrained in space but instead is generalized 10 all objects of that type wherever they occur. Unlike area copying, object copying has been implicated in a number of non-foraging systems. For instance, it is described in predator and enemy recognition (see Chapter 4; Curio, 1988; Mineka & Cook, 1988; Chivers & Smith, 1994) and could be implicated in mate copying (see Chapter 8; Gibson & Hoglund. 1992). The focus here is on foraging examples, specifically object copying as it relates to acquisition of food preferences and the ability to forage from new types of food. Object copying has been described in the context of food choice in many animals. For instance. red-winged blackbirds acquire food aversions and preferences socially (reviewed by Mason, 1988), pigeons avoid seed types chosen by flock mates (Inman fI al.. 1988), while woodpigeons (Columba pal,lIl1bus) (Murton, 1971) and greenfinches (Klopfer, 1961) prefer food chosen by flock companions. The mllst detailed sludies of object copying in rhe conlext of food choice have, no doubt, been conducted by Galef and
ECOLOGY OF INFORMATION USE
61
colleagues with Norway rats (Rnttus norvegicus) (reviewed by Gale!, 1990). Rats eat copious amounts of an unfamiliar food type if they have had the opportunity of smelling that food on another colony member's breath. Object copying in rats is so strong that it can even reverse a conditioned food aversion, an effect also reponed in spotted hyenas (Crocula crocula) (Yoerg, 1991). Object copying has been implicated in a number 01 studies 01 behaviour copying (imitation) in non-human primates. Unfortunately, in those cases it
is often considered a consolation prize lor failure to demonstrate behaviour copying and so attracts little further study (Whiten & Ham, 1992). Adaplive hypotheses
Group-living itself may promote sodallearning (Klopfer, 1961). The underlying argument implies that only social animals have the opponunity of incorporating public information, and so only social animals should have evolved the ability to use it. The hypothesis may be tenuous since, on occasion, non-social animals could also benefit from using public information. Nonetheless, it has been tested by comparing the social and non-social learning perfonnances of taxonomically related social and non-social animals: the pigeon and the Barbados zenaida dove (Zenaida aurila). respectively (Lefebvre el 01.. 1996b). Both species are dietary opportunists but pigeons are gregarious, both nesting and leeding in 1I0cks, while Barbados zenaida doves defend year-round territories. More pigeons learned and did $0 Inure quickly in the presence of conspedfic tutors
than territorial zenaida doves. However, the same dilferences were noted for non-social tasks, so it is unclear whether sociality has selected specifically for social learning or just learning in general. Reliable conclusions will require comparisons involving more than one phyletic group. 3.5.3 Behaviour copying
Behaviour copying. also called imitation (Gale!. 1988; Heyes, 1993), occurs when a topographically novel behaviour pattern is acquired by seeing another individual use it. The behaviour must be topographically novel in order to distinguish behaviour copying from situations where behaviour patterns already in an animal's repertoire are simply redirected to new objects, places or used in a different context as would result from area or object copying (Sherry & Gale!, 1984). It is generally undisputed that behaviour copying occurs in the vocal learning of birds (Catchpole & Slater, 1995). In many songbirds. for example. the development of normal male song requires exposure to the song of other males and sometimes even social interaction with them. Social vocal learning, however, has almost always been regarded as a special case of behaviour copying because, among other reasons, it is dedicated to acquisition of specialized vocal motor patterns that can be acquired only within specific developmental windows (Whiten & Ham, 1992).
62
CHAPTER 3
Visual imitation, on the other hand, is often seen as Ihe cogntllve 'Lamborghini' of the sodaI learning world. The assumptions are that it is fast but, unlike area or objecl copying. requires such elaborate and cosIly cognitive machinery thaI only a few select species can afford i1. Admilledly, over 100 years of research into visual behaviour copying, moslly in non-human primales, has yielded surprisingly few instances Ihal could not be accounted for by other mechanisms such as area and objecl copying. If behaviour copying occurs, il is rare (Gale!. 1976; Whiten & Ham, 1992; Heyes, 1993; Byrne & Tomasello, 1995). Commonly (bul nOI universally) accepted examples of behaviour copying involve anecdOlal evidence in chimpanzees (Pan troglodytes) (Whilen & Ham, 1992), orangutans (Pongo pygmaeus) (Russon & Galdikas, 1993; 1995), experimental evidence in budgerigars (Dawson & Foss, 1965; Galef et 01., 1986) and possibly rats (Heyes & Dawson, 1990). OUlside of vocal learning in birds, therefore, it is safe to say thaI behaviour copying is rare.
Adaptive hypotheses Given that accepted inslances of behaviour copying are so rare, it is nOI surprising thaI adaptive hypotheses have received Iiltle formal tesling. Boyd and Richerson (1985) propose Ihat behaviour copying is a faster, more economical means of acquiring behaviour and as such is expected to occur when: 1 non-social learning is hazardous (e.g. learning 10 recognize poisons or predators); 2 large environmental changes are predictable; 3 social learning is more accurate Ihan non-social learning. An ahernative view portrays behaviour copying as a specialized rewardindependent learning mechanism thaI is likely 10 evolve when the rewards required for behaviour acquisilion by non-social learning mechanisms are unavailable, either because Olher knowledgeable individuals have already depleled the reward sites, or because it is more expedient to feed from olhers' discoveries than 10 learn 10 discover through non-social mechanisms (Giraldeau, 1984; Beauchamp & Kacelnik, 1991; Giraldeau etal., 1994b). In conclusion, il is apparent Ihal a number of animals are capable of area and object copying but many fewer seem capable of behaviour copying. We know some of the ecological faClors Ihat promote the use of area copying, bUI Ihe ones that promole either object or behaviour copying remain uncertain. The comparalive work in columbids appears to suggeslthal selection does nol operate independently on social and non-social learning. If Ihis is so, Ihen social learning may not be a distinct adaptive specialization (see Section 3.4) but simply a Side-effect of selection for learning in general. Naturally, many more phyletically independent comparisons are necessary 10 support any conclusion.
ECOLOGY OF INFORMATION USE
63
3.6 The ecology of cultural transmission This section first looks at populalion studies that are meant to describe and predict how the use of public information can lead to the spread of behavioural innovations within populations. Then, it considers studies of Ihe transmission
process itself. emphasizing factors that govern the rate and efficiency of information transkr. Finally. it looks at an emerging approach that asks whether social learning mechanisms alone (see Seclion 3.5) can sustain long-lived behavioural traditions that form the basis of culture. 3.6.1 The population approach Modelling the spread of behavioural phenotypes within populations can be analogous to modelling changes in gene frequencies. especially when the trail is acquired from parents (Cavalli-Sforza & Feldman. 1981). Unlike genes. however. cultural trailS can be acquired horizontally within generations. or obliquely through collaleral kin. In any case. they are assumed 10 spread wilhin populations following a logistic progression: slowly at first when demonstrators are rare. exponentially as demonstrators become common and then slowly again as naive observers are rarer (Cavalli-Sforza & Feldman. 1981). Re-analysis of a number of now classic sludies of innovalions spreading within populalions provides some support for the accelerated spread (Lefebvre. 1995a.b). For instance. the number of sites where birds have been reported to open milk bottles to drink the supernatant cream spread exponentially Ihrough the UK in the first haLf of this cemury (Fisher & Hinde. 1949; Lefebvre. 1995a; Fig. 3.5a). Potato and wheat washing both spread exponentially wilhin a provisioned populalion of Japanese macaques (Macaca fusca/a) (Kawai. 1965; Lefebvre. 1995b). Potato washing involved dipping potatoes in the sea. presumably to remove sand. while wheat washing was an efficient means of separating wheat from sand by throwing a handful of mixed sand and wheat in the sea and scooping up the floating grain. Unwrapping and eating of caramels. eating beached fish and washing apples in Japanese macaques. as well as mango and lemon eating in chimpanzees. all spread exponentially (Fig. 3.5b--g). Obviously. the pattern of spread aLone is only weak support for cultural diffusion. Non-sooallearning in a populalion with normally distributed learning speeds can also produce an accelerated function (Lefebvre. 1995b). Stronger support requires establishing the social learning mechanisms underlying diffusion models. Unfortunately. these mechanisms in both avian and primate examples have remained unspecified. righlfully leading some researchers to question whether social learning was involved at all (Galer. 1976; Sherry & Galef. 1984).
64
CHAPTER 3 400
10)
40
~E 200 '00
,c
30
.,.~•
20
~
'0
U
(bl
oo~~~:=~-:::---:::-----: 5 10 '5 20 25 30 20
(el
4
2
6
~
, c
60
Idl
.~
, ,
40
.!!!
E u
20
°0~-'·'--7----':----:----:8 246
20
Time (years)
1.0 0.8
E 0.4
.,;;
,cE
c ';:: 0.6
&.
'----
Time (years)
,E
•
,E
u
0.00'--':2--'4'---':-6--:8,---..,'.,.0-':':2,----'4
~
'00
'5
.~ 10
50
80
20
lei
o
0.2
60 Time (months)
' ~
Q.
•
•
E
•
10
8
Time (years)
Time (years)
10
20
30
• 40
50
60
Time (sl
Igi
40-
c
.,.~•
,E
u
100
200 300 lime (days)
400
Fig. 3.5 Cumulative spread of foraging innovations over lime. (a) The number of sites reponing avian bottle opening in Ihe whole of the UK betwet:n 1921 and 1947.
Cumulative number of Japanese macaques (Mocoea/useDra) al Kashima Islet: (b) washing potalOes; (e) washing wheat; (d) eating beached fish. (e) The cumulative proportion of
Japanese macaques at Takasakiyama unwrapping and ealing caramels. The cumulative number of chimpanzees (Pall troglodYlts) at Mahale (f) eating mangoes and (g) lemons. (Avidll data taken rrom Lerebvre. 1995a. and primate dalJ rrom Lefebvre. 1995b.)
ECOLOGY OF INFORMATION USE
65
3.6.2 Factors affecting the rate of spread
The population studies show that a skill may spread, sometimes slowly, other thnes explosively. However. the determinants of the rate remain unspecified. At least two factors have been implicated in the regulation of cultural diffusion: (i) demonstrator and observer densities; and (Ii) the pOiemial of feeding from the tutor's rood discoveries (scrounging).
Demonstrator and observer densities Logistic spread assumes that the rate of transmission of a skill is proportional to the product of the number of demonstrators and observers. Increases in either, therefore, should enhance diffusion rate. Although for pigeons the rate of acquiring a food·finding skill increases with the number of individuals simuLtaneously demonstrating the skill. it declines as the number of un· informative bystanders placed around a demonstrator increases (Lefebvre & Giraldeau. 1994). The observers may have difficulty identifying the individuaL that prOVides useful information so, by increasing the number of bystanders, the chance or observing the correct individual during a demonstration declines. It follows that logistic spread may not be an accurate depiction of a culturally diffused trait. Instead, compared to a logistic spread, the trait may propagate more slowly at first because of the depressing effect of bystanders while later. when demonstrators become common, the spread may be faster because the loss of bystanders is compounded 10 the increased number of demonstrators.
Scrounging In a duplicate cage procedure. a caged demonstrator performs a skill while a caged subject observes. In pigeons, the procedure has consistently shown that skills are more quickly acquired when demonstrators are provided, although the exact social learning mechanism involved is rarely identified (Palameta & Lefebvre, 1985; Giraldeau & Lefebvre, 1986, 1987; Giraldeau & Templeton, 1991; Lefebvre & Giraldeau, 1994). Surprisingly, similar skills spread very little when demonstrators were available within foraging flocks (Lefebvre, 1986; Giraldeau & Ldebvre, 1987). One [actor that has been implicated in the slow spread within flocks is the availability of scrounging opportunities where observers get to eat some of the food discovered by the demonstrator. To test this hypothesis Giraldeau and Lefebvre (1987) simulated the effect of scroung· ing in a duplicate cage procedure. They found that demonstrators failed to enhance an observer's learning when observers passively received some or most of their demonstrator's food. suggesting that scrounging may have prevented the spread of the skill within the foraging flocks (Giraldeau & Lefebvre. 1987; Giraldeau & Templeton, 1991). Whether scrounging also reduces the spread of innovations in many other species remains to be shown. For instance, it
66
CHAPTER 3
does not alter the spread of skills in domestic chickens (G. gallus domesticus) (Nicol & Pope. J 994). However. observations consistent with the negative effects of scrounging have been reported in capuchin monkeys (Ctbus apella) (Fragaszy & Visalberghi. 1989. 1990). jackdaws (Corvus montdula) (Partridge & Green. 1987) and zebra finches (Taeniopygia gll/rara) (Beauchamp & Kacelnik. 1991). 3.6.3 Factors affecting longevity of traditions
Boyd and Richerson (1985. 1988) argue that even if social learning occurs. only behaviour copying (see Section 3.5.3) can give rise to culture. The reason is that acquisition of a trait in both other forms of social learning (area and object copying) rely strongly on an individual's rewarded performance of the behaviour. a step that allows for individual variation in its final form. Heyes (1993) goes even further. arguing that imitation itself may be insufficient to sustain cultural traditions because nothing stops individuals from subsequently modifying the socially acquired cultural phenotype through individual experience (i.e. non-social learning). hence putting an end to tradition. Chain procedures where a previously naive observer is used as a tutor for the next naive observer have shown that in enemy recognition. cultural traditions can be maintained lor several generations (Curio. 1988; Mineka & Cook. J 988; Chivers & Smith. 1994). The chain procedure may artificially enhance the fidelity of transmission because of the paucity of alternatives to copy. A more realistic means of testing fidelity was proposed by Galef and Allen (J 995). They investigated experimentally whether object copying by rats in a lood selection context (see Section 3.4.2) could give rise to long-lived traditions. They trained a number 01 individuals to avoid one of two equally unpalatable food flavours - cayenne pepper or Japanese horseradish - and used these animals to form homogeneous founder groups that all avoided one or the other flavour. The fidelity to the lounders' original food preference was challenged by replacing lounders. one at a time. at regular intervals by naive subjects. The founders' arbitrary preference could be maintained over generations. giving rise to lung-lived traditions that survived beyond the founders' tenure within the groups. but fidelity of the cultural tradition was sensitive to the extent to which rats had access to the alternative flavour and hence the opportunity for non-social learning (Fig. 3.6). In conclusion. therefore. research on cultural transmission remains patchy. no doubt in part because of the real problems it poses to experimentalists. Nonetheless. because cultural transmission frees behavioural evolution from slower Darwinian processes. il presents a major challenge to Darwinian-based behavioural ecology. Clearly. better documented examples of culturally transmilled traits based on solid evidence 01 the lransmis ion mechanism would be very valuable. Moreover. determinants of the rate of spread and the fidelity of transmission clearly need to be explored more extensively and systematically (Galel, 1995).
ECOLOGY OF INFORMATION USE
67
(a)
c
100
S
~
~
;;
'
80
~
0. 0. ~
0. ~
60
c c
~
>
~
•'C>
40
S c
•~
•
20
0.
•• 0
::;;
c
~
• ;;
' ~
0
'00
3
2
4 5 TIme (days)
7
6
8
(bl
~
80
pepper
0. 0. ~
c c > ~
• •' E
60
• :• :;; ~
or Japanese
a naive individual such that
40
from day 5 on, none of the original founders is present.
C> ~
(el
horseradish (_) as a function of time. Each day one of the original rounders is replaced hy
~
0.
Fig. 3.6 Mean per cent (± SE I of cayenne pepper diet eaten by Norway rats (R. norveg;cus) in groups of fOUT individuals whose founders ate only either cayenne
20
(a) represents an experiment
0-
C
~
0
2
4 3 TIme (days)
5
6
with ad libitum access 10 food while groups in (b) had access to food only} h a day. (From Gale! & Allen. 1995.)
3.7 Future directions The informal ion processing ability of animals is clearly emerging as an important component of an increasing number of behavioural ecological questions. However, it will probably become an important area of behavioural ecology in ils own right only once research interests swilch from merely cataloguing animals as having this or that ability to investigating the ecological circumstances under which any given cognitive process is used. Only once information use is studied as a decision rather than a constraint will it be possible to trace the ecological inOuences on an organism's cognitive architecture (Real, 19911. A number of studies have already engaged in this direction. For instance.
instead of asking whet.her memory parameters should differ between species (see Section 3.2.3) one could ask whether the memory parameters of any
68
CHAPTER 3
organism should change according to the ecological problem at hand. Valone (1992b). for instance. applies a memory window modd to the issue of economic patch exploitation and finds that rate maximization occurs when the size of the memory window used to estimate patch quality increases with the duration of travel between resource palches (see also Devenport & Devenpon. 1994; Mackeney & Hughes. 1995). Similarly. public information should be envisaged as a process that is most profitable when individual sampling is costly and/or lime available for resource exploitation is short (Clark & Mangel. 1986). Not surprisingly. in starlings it is used only when accurate personal patch-sample information is difficult to obtain and when acquiring public information does not impose any reduaion in personal patch sampling (Templeton & Giraldeau. 1995. 1996). Research has established successfully that both natural and sexual selection can promote the relative size of the hippocampus when ecological condilions call for enhanced spatial memory. but comparative experimental work has not revealed large quantitative differences in spatial memory. Differences may lie more in the quality of the information processed than its quantity. Alternatively. size of a neuroanatomical struaure like the hippocampus may be a poor approximalion of its processing ability. Future work. no doubt. will go both in the direaion of analysing the Iype of informarion used in spatial memory (Clayton & Krebs. 1993; 1994a.b.d; Brodbeck. 1994; Brodbeck & Shettlewonh. 1995) and increasing the detail with which neuroanatomical struaures are studied. The piaure that is emerging from the ecological study of information use opens the way to research that may reveal that animals. rather than being capable of using informalion or nol. simply use it when it pays and in ways Ihat are adaptive.
Chapter 4 Recognition Systems Paul W. Sherman, Hudson K. Reeve & David W. Pfennig
4.1 Introduction What do these diverse situations have in common? 1 A liver-transplant patient's body rejects the new organ. 2 Bdore matin&- male and female fireOies engage in a predsely timed flash dance. 3 A male bird disregards a singing neighbour but attacks a stranger singing from the same spo!. 4 A hungry mouse and an iridescent green beetle ignore each other. 5 A carnivorous tadpole engulfs a smaller sibling, but immediately spits it out. The answer is that each involves recognition - of self, spedes and males, neigh-
bours, prey and predators, and kin, respectively. Are there common principles governing the evolution of these varied recognition systems? This chapter explores this question and provides a unified evolutionary framework for understanding discriminative behaviour. After briefly discussing the forms and functions of recognition systems, we show how they can be analysed in terms of three component pans. We then: (i) identify the central problem that recognition systems have been designed by nalUral selection to solve; (ii) discuss principles that relate the evolution of each component to lhe central problem; and (iii) illustrate these principles with evidence from a variety of organisms. Kin and mate recognition are highlighted, because they are major foci of mechanistic and functional research (Pfennig & Sherman, 1992, 1995). Finally, we address failures of and ntisunderstandings about kin recognition and suggest some profitable avenues for future research on recognition systems.
4.2 Forms and functions of recognition We define different forms of recognition by the nature of the objects being discriminated, nol by the functional significance or proximate mechanisms of recognition. Thus, kin recognition is differential treatment of conspecifics (including self) differing in genetic relatedness (Sherman & Holmes, 1985; Waldman el al.. 1988; Gamboa el al.. 1991c). Surprisingly, the fitness consequences of kin recognition have rarely been documented. The traditionally hypothesized 69
70
CHAPTER 4
benefits are dispensing nepotism (Hamilton. 1964a.b) and optimiZing the balance between inbreeding and outbreeding (Bateson. 1978; Shields. 1982). However. kin discrimination al 0 functions in other contexlS. For instance. many anuran tadpoles associate preferentially with siblings (Blaustein & Waldman. 1992). In some cases this may renect nepotism (Waldman. 1991) or learning kin phenotypes for oplimal outbreeding (Waldman el al.. 1992). but in others it may indicale only a tendency 10 associate with any conspecifics. including siblings. that smell like the (safe. food-rich) natal sile (Pfennig. 1990). Another function of kin recognition may be disease avoidance. Cannibalistic Arizona tiger salamander larvae (Ambysloma ligrinum Ilebulosum) feed voraciously on non-relatives but avoid eating close kin (Pfennig el al.. 1994). This either represents nepotism or avoidance of a deadly bacterium that is efficiently transmitted through cannibalism (Pfennig el al.. 1991. (993). Infections may be especially transmissible among close relatives because they generally have similar immune systems (Pfennjg el al.• 1994). Male recognilion encompasses many types of recognition which differ in the nature of the potential mates being discriminated. including heterospedfics versus conspecifics (species recognition) and. among the lauer. individuals differing in sex (sex recognition), relatedness (kin recognhion), genetic quality or attractiveness (mate-quality recognition) or parental resources (mate-resource recognition). There is a fundamental difference belween the function(s) of mate recognition and nepotislic kin recognition. In the lauer, discriminative acts increase transmission of the recognition-promoling alleles by favouring relatives likely carrying copies of those alleles. In Dawkins's (1982) terminology, kin-selected kin recognition is aimed at detecling replicalors. In mate recognition. by contrast, discrimination increases transmission of the recognition-promoting alleles by favouring genetically compalible or superior mating partners likely to transmit beneficial alleles or resources to offspring that also receive the recognitionpromoting alleles. Thus, mate recognition is aimed at improving IIthic/es. In this chapter we use 'recognition' and 'discrimination' interchangeably, although they are not synonymous when recognition refers to an internal neural process that underlies, but can occur without. detectable behavioural discrimination (Lacy & Sherman, 1983; Byers & Bekofr. (986). However, if discrimination never occurred, recognition (as a strictly internal process) would be an empty concept. The real point behind the distinction is lhat lack of discrimination in one context does not imply its absence in another.
4.3 Components of recognition systems Any recognitjon process involves an aclor and a recipient (Fig. 4.1); these usually are different individuals (except in self-recognition). All recognition systems can be partitioned conceptually into three component parts (Sherman & Holmes, 1985; Waldman, 1987; Reeve, 1989; Gamboa elal.. 199Ic).
RECOGNITION SYSTEMS
71
1 Production: the nature and developmenl of labels (cues) in recipienls that actors use to recognize them. 2 Perception: the sensory detection of labels by actors and subsequenl phcnotype matching; i.e. comparison of labels to a template (internal represenlation) of the phenotypic attributes of desirable (fitness-enhancing) or undesirable (fimess-decremenling) recipients; the ontogeny of templates is also part of this component. 3 Action: the nature and determinanlS of actions performed. depending on the similarity between actors' templates and recipients' labels. Discriminative actions need nO( be slriClly behavioural. as illustrated by two examples: (a) Arizona tiger salamander larvae transform into physically different cannibal morphs more frequently when they are reared among non-kin than among kin (Pfennig & Collins. 1993); (b) in montane larkspurs (Delphinium nelsonii). genetic similarity declines with distance between plants. and pollen-pistil interactions result
Production
~ (TemPlate
Mal.'s call
~ Degree of match
Fig. 4.1 Recognition sYSlern~
comprise three components: produaion. perception and action (indicated by separate boxes). Any recognition pr()(-ess involves an actor and a reOplcnt;
Action
Decision rule ~ Behaviour
here. the aaor is a female (wg (middle) who is anempling 10 recognize a potcmial male (lOp) based on a label (cue) contained in the pauern of notes in the
Actor
Recipient
male's advertisement call. In lhis case there is a slight mismatch belween the male's cue and The female's template: the degret' of mismatch affects whether or nOI discriminative behaviour occurs.
72
CHAPTER 4
in plants 'preferring' as mates distant relatives over close relatives or nonrelatives, as indicated by the number of pollen tubes that the pistil allows to grow to its ovary (see Chapter 15; Price & Waser, 1979; Waser & Price. 1993). Characteristics of components (1-3) should vary according to ecological and social circumstanccs that arrect the costs and beneHts of discrimination. Knowledge or the components can therefore illuminate thc selective rorces shaping discrimination, and vice versa. Understanding the evolution or the components requires appreciating the central problem confronting all recognition systems. 4.3,1 The central problem for recognition systems
Selection should favour individual organisms (and cells) whose recognition mechanisms classiry recipients without error. tn nature, however, desirable and undesirable recipients orten exhibit overlapping cues (Getz, 1981, 1991; Lacy & Sherman, 1983). In addition, undesirable recipients may benefit by either mimicking the cues of desirablt' recipients or scrambling cues to prevent discrimination (Reeve, 1997al. Consequently, the central problem is how to optimize the balance between acceptance errors (accepting undesirable recipients) and rejection errors (rejecting desirable recipients). Imagine graphing, ror each kind or recipient, the frequency distribution or the diHerence between the recipients' cucs and the actor's template (Fig. 4.2). The shapes and overlap between the frequency distributions ror desirable and undesirable recipients depends on which cues and templates are used and how the match between them is assessed (Crozier & Dix, 1979; Lacy & Sherman, 1983; Getz & Chapman, J 987). Probabilities or each type or recognition error also depend on the position of the acceptance threshold (Reeve, (989). Alternative designs for each recognition component aHect the balance or errors diHerently, and the design that optimally balances these errors is predictable from knowledge of the organism's environment. 4.3.2 The production component
General prindples Actors should use recognition cues that maximize the separation of templatecue dissimilarity distributions ror desirable and undesirable recipients (Fig. 4.2). Suppose the choice is between 'D-present' cues, possessed by nearly all desirable recipients but also by some undesirable recipients, and 'U-absent' cues. possessed by some desirable recipients but rarely by undesirable recipients. Use of the former leads to few rejection errors but some acceptance errors, and vice versa for the latter. Accordingly, actors should use D-present cues when: (i) interaClantS are most often desirable recipients; (ii) Hmess benefits
RECOGNITION SYSTEMS
73
of accepting desirable recipients are high; and (iii) costs of (mistakenly) accepting undesirable recipients are low. For example. in the neotropical ant Ectatomma rllidllm (Breed et al., 1992) and the desert woodlouse Hemilepistlls reallmeri (Linsenmair, 1987), nestmates share a chemical label that is readily transferred by direct contact, enabling young colony members to acquire it rapidly. Sometimes, foreigners also acquire
Accept
Rejecl
Desirable
o Acceptance error •
recipients
Rejection error
,
Acceptance threshold
Template-cue dissimilarity
Evolution of recognition system
I
Change in cues used (production component> or in template/matching rule (perception component)

Change in acceptance threshold (actIon component)
Fig.4.2 Evolution of recognition systems. (Upper) When there is overlap in the labels of desirable and undesirable recipiems. selection will favour an optimal balance between accepting undesirable recipients (acceptance errors) and rejecting desirable recipients (rejection errors). The figure iIIuStrales lhe frequency distribution. for desirable and undt.>sirable redpients. of the difference between their cues and the actor's template. For both desirable and undesirable reopients. the probabilities of each type of recognition error depend on the shapes of their frequency distribulions and the pOSition of .he acceptance lhreshold. (Lower) Selection can alter the magnitudes of and balance between acceptance and rejection erIOrs by changing the recognition cues lIsed (production component). the recognition template or matching algorithm (perception component) or the decision rule - e.g. acceptance threshold. as shown here (action component); see also Box 4.1. (Modified from Reeve. 1989.)
74
CHAPTER 4
the cue. enabling them 10 enter the nest to steal food. Persistent use of Dpresent transferable labels implies that benefits of always accepting colony mates outweigh costs of occasionally being robbed. Actors should use U-absent cues under the converse of circumstances (i-iii). For example, to avoid accepting any pathogen-infected cells (i.e. costly acceptance errors). vertebrate immune systems use molecular markers of individuality that are possessed rarely by undesirable recipients. If antigens pmduced by infected (or possibly cancerous) cells are not represented among T- (immune) cell surface antigens, rejection occurs. Surface antigens are indiVidual-specific, because loci in the major histocompatibility complex (MHC), which produces the amigens, arc highly polymorphic (Brown & Eklund. 1994). A potential cost of using U-absent cues in self-recognition is self-rejection. One clue as to why lupus and rheumatoid anhritis arc so common (i.e. > I in 200 of the US population) is Ihat sufferers experience reduced risk of some cancers (Gridley el al.. 1993; Duquesnoy & Filipo. 1994). especially those of viral origin (Kinlen. 1992). Whereas cancer represents a potentially deadly acceplance error, autoimmune diseases represent costly rejection errors. Selection should favour an optimal balance between these reciprocally related errors. but the near equivalence of slightly different balances of errors may help explain interindividual variation in susceptibilities. Elevated immune responsiveness may protect againsl cancers. and perhaps pathogens and environmental toxins (venoms. plant secondary compounds), but at the expense of increased likelihood of autoimmune diseases.
The production component ofkin recognition Kin-recognition cues may be any aspect of the phenotype that signifies kinship reliably. Chemical cues arc widely used (Beecher, 1988; Halpin, 1991; Waldman, 1991). They are potentially information-rich (owing to their three-dimensional molecular structure), while often requiring lillie energy to obtain or produce (Alcock, 1993); e.g. when Ihey are present in food or nesting materials, or generated as metabolic byproducts. Moreover, a matching system exists in most organisms to delect and decipher chemical subslances (the immune system). Multiple sensory modalities may be employed in recognition. however, depending on the context. For example, as tadpoles. American toads (Bufo americanus) use chemical labels to school preferentially with siblings (Waldman, 1986). but as adults females may choose 'optimally related' mates based on their vocalizations (Waldman et al.. 1992). Labels used in kin recognition can be of genetic or environmental origin. and may be produced endogenously by actors or acquired from their environment (Gamboa el al., 1986a). As an example of genetic endogenous labels, larval Botryllus schlosseri tunicates settle near and sometimes fuse with individuals that carry the same allele at One histocompatibility locus (Grosberg & Quinn.
RECOGNITION SYSTEMS
75
1986). Larvae settle closer 10 non-relatives that bear the same allele than to relatives that carry a different allele. Use of this D-present cue suggests that interactions with non-relatives are infrequent and/or costs of acceptance errors (fusing with a non-relative) are low. Because these histocompatibility loci are highly polymorphic (58-306 alleles/locus; Rinkevich et al., 1995), chemically similar individuals usually are close kin. Genetic, endogenous cues mediate self-recognition in flowering plants (Charlesworth, 1985), where proteins produced by multiallelic recognition loci in the pistil (e.g. the'S' locus; Lewis et al., 1988) are matched with pollen surface proteins. In many outcrossing species, only pollen bearing an allele that does not match the plant's own alleles will be accepted, thereby minimizing costly inbreeding (Waser, 1993). Some plants (e.g. Collomiagrandijlora) produce both outcrossing (chasmogamous) and selfing (c1eistogamous) Dowers (Lord & Eckard, 1984); the former reject while the latter accept their own pollen. In outcrossing Dowers, the self is the (strongly) undeSirable recipient and recognition cues are U-absent, whereas in c1eislOgamous flowers the self is the desirable recipient and recognition cues are (presumably) D-presenl. Genetic endogenous labels have been implicated in parent-offspring recognition in various birds and mammals (Beecher, 1991; Halpin, 1991), including humans (poner, 1991; Christenfeld & Hill, 1995), and in nestmate recognition in social inseds (e.g. Michener & Smith, 1987). Preferences for unfamiliar paternal half-siblings over unfamiliar non-siblings in tadpoles (Rana cascadae: Blaustein & O'Hara, 1982; R. sylvatica: Cornell et al., 1989), Belding's ground squirrels (Spermophilus beldingi: Holmes, I986b) and house mice (Mus musculus: Kareem & Barnard, 1986) also imply genetically encoded recognition labels because these kin share 50% of the male parent's genes by descent, but none of the environmental factors that full-siblings or maternal half-siblings share (e.g. in tadpoles cylOplasm, egg jelly or oviposition site cues). Other organisms, such as paper wasps (Polistes: Gamboa et al., 1986a; Gamboa, 1996) and honey bees (Breed el al., 1995), discriminate colony mates using acquired labels that may be genetic and/or environmental in origin. Paper wasps absorb hydrocarbons from their nest at eclosion (Espelie & Hermann, 1990; Singer & Espelie, 1992), and these serve as recognition labels (Gamboa el al., 1996). The relative importance of odours derived [rom plant fibres in the nest and genetically encoded odours applied to the paper by the wasps themselves is uncertain. However. environmental cues theoretically could rival genetic polymorphisms in diversity if different. colonies use different mixtures of plants in nest construction (Gamboa et al., 1986b). AcqUired, environmental cues may playa role in vertebrate kin recognition. Tadpoles of wood frogs (R. sylvatica) and common frogs (R. temporaria) prefer environmentally acquired odours 10 which they were experimentally exposed as embryos, sometimes even after metamorphosis (Waldman, 1991; Hepper & Waldman, 1992; Gamboa et al., 199Ia). These tadpoles can also use genetic endogenous labels in kin recognition (Waldman, 1981; Gamboa el al., 1991a;
76
CHAPTER 4 Wood frogs
American toads
(a)
1.0 '0
.~
p< 0.0001
P< 0.0005
~
0.8
C
o ~
.§ '0 c
o
0.4
& e
0.2
.j;
'-
10
0.6
5
oLl_'-..:..._:L-_J...:._ _-LPaternal half-sibs
Non-kin
oL l_ _-,'='---L--'-_='Siblings
Paternal half-sibs
Ibl
o
0' 0 0
Wood frog communal egg mass
0,
•
American toad egg string
Fig. 4.3 An organism's environment can influence whether labels pradu cd by genetic loci or environmental cues arf' more reliabre indicalOrs of kinship. (a) In laboratory choice tests, wood frog tadpoles (left) school preferenlially with paternal half-siblings over non-kin. This implicates genetically encoded cues, because paternal half-siblings share only their father's genes by descent. (Data from Cornell tt al.• 1989.) American lOad tadpoles (right). which inhabit the same ponds. discriminate lull·siblings from paternal half-siblings. which are treated like non-siblings, implying thai recognition cues may come from the environment (e.g. egg jelly). (Data from Waldman, 1981.) (b) Female oviposition behaviour may explain why wood frog tadpoles, but not American toad tadpoles, rely heavily on genetically encoded cues to identify their kin. wood frogs oviposit in communal clumps (left). presumably to insulate their eggs against the cold. Different clutches experience the same microhabitat. and only labels produced by polymorphic loci would enable different sibships 10 aggregate assonalively. American toads breed in warmer water and individual females often deposit strings of eggs separalely (right). Each clutch experiences a slightly different chemical milieu and labels acquired from lhe (micro-) en'ironment through diffusion into the egg jelly permit accurate discrimination of siblings.
RECOGNITION SYSTEMS
77
Fig. 4.3). In the fonner experiments, environmental cues apparently 0 erwhelmed genetic labels. The relative importance of genetic versus environmental labels in kin recognition theoretically should optimize the balance of recognition errors (see Fig. 4.2). Relying solely on cues acquired from the environment might increase acceptance errors (mistakenly assisting non-relatives that share the same environment), whereas relying solely on genetic endogenous labels might increase rejection errors, because genetic recombination and multiple mating by females ensures that non-clonal family members will not be genetically identical for segregaling alleles (Gamboa el 0/.. 1986b; Waldman, t987; Pfennig & Sherman, 1995). The frequencies and costs of recognition errors, and thus the optimat cue system, should depend on the spedes' genetic system and ecology (Fig. 4.3). Endogenous genetic labels are most useful for organisms that occur in homogeneous (chemical) environments, as docs a group of tunicales on a rock. For organisms that occur in mOre diverse environments, such as paper wasps or tadpoles, additional reliance on labels acquired from the environment may reduce rejection errors without substantially increasing acceptance errors.
Assessing the relative importance of different kinds of cues in kin recognition can be dillicult because, as Breed (1983) and Carlin (1989) noted, whenever experimental subjects are reared in uniform environments, where the only detectable differences are in gene products, investigators may erroneously conclude that they use only genetic labels to recognize kin. The issue that must be addressed is whether such cues would be supplemented, or even supplanted, in nature by variable environmental cues.
The produdion component ofmale recognilion Selection should always favour effident recognition of males through uSt' of detectable, discriminable and memorable cues (Guilford & Dawkins, 1991) thai reveal the properties individuals (especially females) are seeking. In <ex and spedes recognition, females should minimize acceptance errors by using V-absent cues and minimize rejection errors by favouring males whose cues deviate least from the population mean. As an example of the former, female pied flycatchers (Ficedu/a hypo/euca) use sexually dichromatic plumage markers to recognize males (S<etre & Slagsvold, 1992); these cues are V-absent because they are restricted to one sex. Regarding the laller, species recognition in fruit flies involves female choice of males that produce stereotypic courtship sounds (Kyriacou & Hall, 1986; Hoy el a/., 1988), odours (Spiess, 1987) and behaviours (Hoikkala & Welbergen, 1995). These cues consist of heritable elements whose number and complexity correspond to the frequency of exposure to heterospecifics, and all of them must match the mate-recognition template for copulation to occur (i.e. they also are V-absent).
78
CHAPTER 4
Cues used in discriminating among potential mates, by contrast, are more variable intrasexually and intraspecifically. Phonotaxis experiments on tungara frogs (Physalaemus puslulosus: Rand el al., 1992) suggest that mate-quality recognition is based on the variable presence of one to six chucks and their pitch (females prefer the lowest frequencies, which are produced by larger males), whereas species recognition is based on a relatively invariant character, the fundamental frequency of the first 0-100 ms of the 'whine' pan of Ihe male's call. In many mate-quality recognition systems, females prefer signals that deviate most from the population mean (Ryan & Keddy-Hector, 1992; Andersson, 1994), in contrast to sex- and species-recognition systems. Fisherian models of sexual selection (see Chapter 8) suggest that these preferences are maintained because sons of males that exhibit exaggeraled traits will inherit these traits and thus be attractive to females, whereas in 'good genes' models preferences are maintained because offspring of males with exaggerated traits will be highly viable, since only males with high-viability genes can afford the costs of expressing exaggerated traits (see Chapter 8; Pomiankowski, 1988; Grafen, 1990b). In either case, females who prefer exaggerated traits minimize acceptance errors. Open-ended female preferences for exaggerated male traits should occur when Ihe benefits of reducing acceptance errors consistently exceed the increased cumulative costs of rejection errors. The directness of the connection between mate-recognition cues and the vehicle-enhancing qualities they signify depends on what females art' attempting to obtain from males. In mate-resource recognilion, stimuli from the resource ilself are the cues. Thus, female blue-headed wrasses (Thalassoma bi!ascialum) choose good spawning sites (Warner. 1987), and female hanging flies (Hylobit/aeus apiealis: Thornhill, (981) and katydids (Conoeephalus nigropleurum: Gwynne, 1988) choose the size and quality of males' nutritional o[ferings, not males' phenotypes per se. In lepidoptera, chemical materecognition cues often are metabolic derivatives of males' nuptial gifts (Conner el 01., 1990). Some female birds use cues that correlate with paternal effort, such as male body size (Petrie, 1983), courtship feeding rale (Wiggins & Morris, 1986) or colour (Hill, 1991). When females seck high-viability genes or sexually attraClive trailS for offspring, they must rely on a correlation between those benefits and one or more aspects of males' phenotypes (e.g. Petrie, 1994). For example, in greatreed warblers (Aerocephalus arundinaeeus) post-nedging chick su rvival is correlated with paternal song repertoire size. Females engage in eXlrapair copulations only with neighbouring males that have larger song repertoires Ihan their social mate (Hasselquist el 01., 1996). The connection between mate-recognition cues and benefits of discrimination is most indirect when lemales seek pathogen-resistant genes for their offspring, because rapid coevolution makes it unpredictable as to which genes confer resistance to prevailing pathogens. Females would
RE COG NIT ION S YST EM S
79
frequently commit both acceptance and rejection errors by using as a recognition cue only one trait (Le. the prnduCls of a few resistance genes). Selection has favoured two opposite solutions to this probkm: (i) mating with many males. thereby increasing the likelihood that some patrilines will be resistant (e.g. social insects: Sherman et 01., 1988; Schmid-Hempel, 1994); or (ii) choosing a mate based on multiple condition-dependent traits that would be altered by any disease (i.e. cues that are connected to many potentially relevant genes). Thus. female barn swallows (Hirundo fIlstica) prefer males with long symmetrical tail leathers. cues that correlate with age and parasite resistance, which is heritabk (M..ller, 1994). Indeed. females in many species of birds (reviewed by Andersson. 1994. pp. 74-6) and fish (Houde & Torio, 1992; Kodric-Brown. 1993) prefer vignrously courting males with elaborate ornamentation. characteristics Ihat are inversely related to parasitism and.
pre umably. genetic susceptibility to it. 4.3.3 The perception component Gmeral principles
Selection may shape the perception component by modifying both the recognition template and the matching algorithm. Templaces
Templates are internal representations of the characteristics of desirable or undesirable recipients (see Fig. 4.1). Recognition occurs when phenotypes of recipients match these templates closely enough. Generalized templates are lavoured when appropriate responses to all undesirable (or desirable) recipients are the same. despite variation in their exact cues. For example, Belding's ground squirrels usually give multiple-note trill vocalizations to terrestrial predators and single-note whistles to aerial predators (Sherman. 1977. 1985). However, the squirrels trill at walking hawks and whistle at running coyotes. Apparently the rapid approach of any large heterospecific, regardless of its exact phenotype, represents imminent danger; Evans eC 01. (1993) present similar data for chickens (Gallusgallus). By contrast. vervet monkeys (Cercopithecus aeclriops) give struClurally different calls to leopards, lions and hyenas. hawks. snakes. baboons and unfamiliar humans (Cheney & Seyfarth, 1990), suggesting that templates may be more specific when appropriate responses are not the same for all recipients (e.g. predators). Template formation involves various degrees of learning. On the one hand. learning is disfavoured when recipients are nOt reliably present for template formation before discrimination is necessary or when template learning (particularly of undesirable recipients) might increase mortality. For example, naive (hand·reared) mot mots (Eulllomota superciliosa) are instantly repulsed
80
CHAPTER 4
br models resembling venomous coral snakes (Smith, 1975), and newborn garter snakes (Thamnophis sirtalis) respond strongly to odours of their primary prey; the latter preferences are heritable and differ among populations (Arnold, 1981 ).
Genetically encoded templates are inferred when artificial seleaion on male in changes in female mate-seJeaion criteria as a correlated response (reviewed by Bakker & Pomiankowski, 1995). For example, Wilkinson and Reillo (1994) selected male stalk-eyed flies (Cyrtodiopsis dalmanni) for large (l) or small (S) eye span relative to body size. After 13 generations, females from l-lines preferred l-males (as did unselected controls). but S-Iine females preferred S-males, indicating a genetic correlation between females' templates and males' traits. On the other hand, templates must be learned when the characteristics 01 desirable or undesirable recipients vary uver space or lime. Thus, juvenile Belding's ground squirrels (Mateo. (996), vervets (Cheney & Seyfarth. 1990) and other vertebrates (Curio. 1988; Mineka & Cook, 1988) learn the characteristics of predator contexts by observing anti-predator responses of adults (often parents). When templates are learned, the objects or individuals that provide information about the characteristics of desirable or undesirable recipients are called referents. Selection favours use of referents that maximize the separation of template-cue dissimilarity distributions for desirable and undesirable recipients (see Fig. 4.2). Sometimes, actors serve as their own referents (e.g. in bat echolocation or the immune system). More often, learning occurs by; (i) associating characteristics of individuals with the consequences of interacting with them directly; (ii) observing their interactions with others; or (iii) learning the characteristics of individuals that are likely, by virtue of their location in time or space, to be desirable or undesirable recipiellls. The timing of template learning depends on when the most informative referents are available and when discrimination is first adaptive. In some species, templates are learned ('imprinted') during a short sensitive period early in life, presumably because useful referems (parents) are predictably present and their characteristics reflect those of desirable recipients throughout an actor's Iile (Bateson, 1979). In many species, a second sensitive period occurs at reproduction, when parents imprint on their newborn offspring (Colgan, tr~its results
1983).
When there is'a lengthy association between individuals whose charaaeristics may change through time, recognition templates may be 'updated'. Template updating is what enables uS to recognize our own teenage offspring regardless of how they looked as babies. In the many territorial species that discriminate neighbours from strangers (Temeles, (994), template updating must occur whenever a new resident immigrates (Catchpole & Slater, 1995; Lambrechts & Dhondt, 1995). Sometimes, immigrants also update their song templates, because they match their new neighbour's songs (e.g. in village indigobirds.
RECOGNITIO
SYSTEMS
81
Viduo cholybeoto; Payne. 1985). Finally. lemales that copy the mating choices of other females (e.g. Gibson elol.. 199 I; Ougatkin. 1992b) update their matequality recognition template whenever a new male is frequently chosen by conspecifics. Matching algorithms
In general. selection should favour matching algorithms. which are internal weighting schemes. that optimally balance acceptance and rejection erro'. If rejection errors are costly. cues that characterize nearly all desirable recipients (i.e. O-present cues) should be disproportionately weighted. even if some undesirable recipients also will match closely the template. Thus. myrmecophilous beetles (Vander Meer & Wojcik, 1982) and parasitic ant queens (Franks el 01.. 1990) gain access to host ant nests because they possess a chemical cue that is heavily weighted by the host species in nest mate recognition. Conversely, if acceptance errors are costly. actors should disproportionately weight cues that are possessed rarely by undesirable recipiel1s (U-absent cues), even if some desirable recipients will not match closely the template. Thus, female birds avoid diseased males by crutinizing secondary sexual characteristics associated with health, such as the colour and size of wattles. contbs and snoods (Zuk elol.. 1992; Buchholz, 1995). Cues associated with male health are weighted so heavily that males with visible parasites (Borgia & Collis. 1989) or those that merely smell parasitized (Kavaliers & Colwell, 1995) may not match females' templates for acceptable partners. The perception componel1/ of kin recognilion
There are no clear examples of genetically encoded kin-recognition templates (Alexander, 1990; Pfennig & Shennan, 1995). This is probably because, as a result of meiotic shuffling of genetic cues and spaliotemporal variation in environmental cues. the characteristics of desirable recipients (kin) will differ for different actors, rendering genetically encoded templates unreliable. In addition, intragenomic conflict should thwart expression of selfish 'recognition alleles' (template loci linked to both recognition cue and decision rule loci: ,ee Alexander & Borgia, 1978; Alexander. 1979; Ridley & Grafen, 198 I; Chapter 12). Organisms that live in sufficiently diverse environments (see Section 4.3.2) can learn templates partly or wholly from environmental features that reliably correlate with kinship. For example. paper wa ps learn recognition odours from their nest and not directly from nest mates or themselves. Non-relati'es can be fooled into accepting each other by experimentally exposing them to different fragments of the same foreign nest (Pfennig el 01.. 1983). If wasps are isolated from their comb and nestmates when they eclose. or are exposed only to nestmates but not to nests, they later treat all conspecifics as nestmates, regardless of relatedness (Shellman & Gamboa, 1982).
82
CHAPTER 4
Kin-recognition templates also are learned diredly [rom parents or neSlmates (Buckle & Greenberg, 1981; Blaustein et al., 1987). Thus, female great tit chicks (Parus major) learn their father's songs and, as adults, females avoid pairing with males whose songs match this template (McGregor & Krebs, 1982). By contrast, female Belding's ground squirrels learn kin-recognition cues from nestmares. fn this species, mothers and daughters and sisters behave nepotistically: they warn each other of predators and protect their own and each others' pups against infanticide by establishing and jointly defending terrirories (Sherman, 1977, 198Ia.b). Cross-fostering studies in the field (Holmes & Sherman, 1982) and laborarory (Holmes, 1984, 1986a, 1994) indicate thar nestmate females learn each other's odours just before weaning (when lillers normally begin 10 mix), and later treat each other as siblings, regardless of their actual relatednesses. Socially learning these kin-recognition templates does not increase acceptance errors in nature because unrelated
pups rarely cohabit the same nest burrow (Sherman, 1980). Kin-recognition templates may also be learned via self-inspection. For example, the immune sysrem learns the identity of the body's cells (von Boehmer & Kisielow, 1991). Alexander (1990, 1991) argued that genes promoting selfimpection are genetic 'outlaws' when their interests conflict with the rest of the genome. However, genes enabling formation of the molecular selfrecognition template in immune recognition are not outlaws because the entire genome benefits from identifying and disabling foreign organisms (pathogens) or substances (environmental toxins). Selfish outlaw genes that caused recognition solely of products of linked genes would endanger the organism by causing serious autoimmune diseases.
Outlaw genes promoting self-inspection are more likely to spread when selfish targeting of linked produdion-<:omponent genes does not endanger the organism, e.g. in decisions about nepotistic allocation of resources. Even in this case, however. Alexander's argument implies only that intragenomic
conOid will suppress such outlaws; it does not imply that perception-<:omponent genes causing self·inspection of the products of un/inked genes cannor evolve. This was Hamilton's (1987a, p. 423) point: 'so long as the genes involved in all pans of this discrimination are well spread among the chromosomes ... we need not expect that other elements in the genome will evolve 10 suppress the effect.' Self-inspection is favoured when template learning is desirable, but opportunities 10 directly assess relatives' phenotypes are limited. For example, selfinspection may occur in Botry//us sch/osseri tunicates, the larvae of which are planktonic (Grosberg & Quinn, 1986), and in female Gry//us bimacu/atus crickets and parasitoid wasps, which lay their eggs singly so larvae develop alone (Loher & Dambach, 1989; Ueno, 1994). In G. bimacu/atus, females learn their own cuticular pheromones and later reject similar-smelling suirors. Unfamiliar unrelated pairs mate more quickly than do unfamiliar maternal half-Siblings or full-siblings (Simmons, 1989). Since rhis recognition functions 10 improve
RECOGNITION SYSTEMS
83
vehicles (oIfspring) by reducing inbreeding, genes enabling formation of the self-recognition template are again nOl outlaws. Finally, self-inspection may enable kin recognition when available relatives would yield error-prone templates as, for example, when nestmates are fulland half-siblings due to multiple mating by females (Sherman, 1991). There are two possible examples of what Dawkins (1982) dubbed the 'armpit effect'. I Honey bee (Apis mel/ifera) queens mate with seven to 17 drones, leading to mixed paternity (Page, 1986). Isolated workers can use their own phenotype as a kin-recognition template (Getz & Smith, 1986; Getz, (991), and may be able to discriminate full- from matcrnal half-sisters (Getz & Smith, 1983). Whether nepotism occurs within the hive is controversial (see Visscher, 1986; Page et al.. 1989 versus Oldroyd et al., (994). A possible reason for this controversy is that discrimination may only occur when competition is severe (e.g. resources are limited) but discrimination does not jeopardize colony reproductive efficiency (Sherman, 1991; Ratnieks & Reeve. 1991). 2 Female Belding's grouod squirrels mate with one to eight males and most litters are multiply sired (Hanken & Sherman, 1981). In the field. females are slightly less likely to attack full-sisters than maternal half-sisters, and more likely to share territories with full-sistcrs (Holmes & Sherman, 1982). Discrimination of nestmates from non-nestmates and full- from half-sisters among neSllnates suggests that a female's own smell. as well as the odours of nestmates. both contribute to her kin-recognition template (Sherman & Holmes, 1985). In apparent support of self-referent phenotype matching, in laboratory studies paternal half-sisters also recognized each other (Holmes. 1986b; but see Alexander, 1991).
The perception component of mate recognition Genetically encoded templates should be more common in species recognition than in kin recognition. because within populations there often will be little spatiotemporal variability in the characteristics of desirable recipients. Indeed. genetically encoded templates have been inferred in acoustically mediated mate recognition in crickets (Hoy et al., 1977; Doheny & Hoy. 1985) and frogs (Walkowiak. (988), based on intermediate responses of hybrids to calls of parental species. Of course, genetic encoding of species-recognition templates does not necessarily imply a lack of interpopulational variability in templates. In grey tree frogs (Hyla chrysoscelis) risks of species-acceptance errors have apparently influenced the evolution of female templates and matching algorithms across populations (Gerhardt. (994). In recognizing mates. females weight the pulse repetition rate, which is unique to conspecific males. more heavily than call duration in areas of sympatry with H. versicolor. In areas where only H. chrysoscelis occurs, females weight call duration (an indication of mate quality) more heavily in mate choice. Thus, reproductive character displacement is expressed in the perception and, potentially, the action components of recognition.
84
CHAPTER 4
By contrast, templates for mate-quality recognitIon should more often be learned, particularly in inbreeding avoidance, because characteristics of desirable recipients vary spatiotemporally. In house mice, the MHC genotype is expressed in urinary odours; individuals can detect the influence of even a single mUlation (Brown & Eklund, 1994). Buth sexes prefer tu mate di,assortatively (Beauchamp ,t al., 1988; Boyce el al., 1991), but preferences of males for females with MHC genotypes different from their own can be reversed by cross-fostering male PUP5 to parents with different MHC genotypes (Fig. 4.4). This implies that males learn their template from parents. Sex-recognition templates also are often learned from parents. Male zebra finch chicks learn their mother's plumage and bill colours and later recognize females based on those characteristics (Yos, t994, 1995a,b). Female zebra finches, in contrast, apparently identify males by their courtship behaviour (Ten Cate, 1985; Yos, 1995a). Thus, a male's sex-recognition template directly in!onns him about a conspecific's sex, and indirectly informs females aboul sex via the males' behaviour.
Genotype of subject
Rearing
environment (dam, sire)
(a) Templates learned from parents IMHC preferences of male mice) Percentage of males mating with bib (vs. klkl females
o
bib. bib bibd'
k/k, k/k
C I
., .'
B
'
~
'
75
I I
100
(bl Templates not learned from parents (t·allele preferences of female mice)
Percentage of females that preferred +/+ over +It males
Fig. 4.4 Mate·recognition Icmplalcs may he..' learned fmm social associates or hy selfinspeclion. (a) Male house mice prdt..·r (emall's whose MHC genOlVpcs differ from those of the male and female who re 86
CHAPTER 4
the idea that accurate perception of species- or sex-recognition cues results alilOmatically in 'hidden preferences' for the same cues in mate-quality choice contexts (e.g. Enquist & Arak. 1993. 1994; Johnstone, 1994). Based on simulation studies. these allthors argued that female preferences (templates) for extravagant and symmetrical male charaCleristics are indeed unscleCled byproducts of species or ex recognition. However, as Dawkins and Guilford (1995) pointed OUI, these neural nelworks contain so few elements and conneclions that they bear only superficial resemblances to aaual neural circuitry, and lheir behaviour is unlike real recognilion systems in many imponant ways. Hidden biases may thus be anifacts of the models' simplicity. and their direction may depend on the 'training' regimen. Moreover, these models only suggesl a possible origin for mate-quality recognition templates. Such templates will be maintained only if the resuhing preferences consistently yield the predicted benefits - e.g. more amaaive sons or heahhier. more viable offspring (see Reeve & Sherman. 1993; Watson & Thornhill, 1994).
4.3.4 The action component Gmeral principles
Given a panicular degree of match between a recipient's cues and an aClOr's template, whal action should the aClOr take? Individuals might perform a continuous range of actions (or illlensilies of a single aclion), with responses changing gradually as the cue-template match changes. Alternatively, actions may be all-or-none, as when there is a Ihreshold above which all recipients are accepted and below which they are rejected (see Fig. 4.2). Many decision rules involve binary actions (e.g. attacking or ignoring potential prey. mating or not mating). so we reslrict our attention to Ihis conceptually simpler ca<;e.
The optimal acceptance threshold will depeod on the relative rates of interaction with and the fimess consequences of accepting and rejecting desirable and undesirable recipients (Reeve, 1989). For example. the threshold should become more restriCl;ve as the fitness COSIS of accepting undesirable recipients increase. However. the oplimal threshold can become either more or less restrictive as the benefits of accepting desirable recipients increase, depending on whether there are limits on the number of recipients an aClor can accept (Box 4.1). If acceplances are essenlially unlimited (e.g. a nestentrance guard in a social inseCl colony can admit all neSlmale foragers), the threshold should become more permissive as the desirable recipient's value increases. because the aClor should accept as many desirable recipients as possible. Conversely, if an aClor can afford 10 accepi only one or a few recipients (e.g. a female io a lck-breeding species searching for her male for the season), the lhreshold should become more restrictive as the desirable recipient's value increases because a restrictive threshold enhances the probability that the
RECOGNITION SYSTEMS
87
Box 4.1 Shifting acceptance thresholds in guard and search contexts Natural selection may favour diHercni
aCCepldJ1Ce
thresholds in dHfcrem
contexts. Suppose g(x) and b(x) are the probability densily functions (p.d.!.' describing the frequency dislribulions of template-recognition cue dissimilarity x for desirable (0) and undesirable (U) recipienls. respeclively (see Fig. 4.2). The probability of accepting a recipient given a threshold I is:
,
J
g(x)dx
=G(I)
o
for desirable recipients. and:
,
J
b(xldx = B(I)
o
for undesirable recipients. Let Pbe the fraction of Os which each cause a fittles' change F if accepted. and I - P be that for Us. which yield a filness chang' f < 0 if accepted. The optimal acceptance threshold 1* usually cannot be explicitly obtained. because the precise shapes o[ the p.d.!.'. lypically are unknown. However. we can predict how the threshold will shift as. say. F changes. We seek the sign of aW(I)
dl*/dF. which is the same as thai o[ - - at 1=1*. where wIt) relates the aFiJt
actor's fitness to
t.
Case 1: guard context Suppose an actor must decide whether to accept n:cipients that approach it. and multiple acceptances are possible. Assuming additive ritness changes. tht' actor's threshold should maximize the mean fitness increment per interaction: W(I)
=PFG(I) + (I -
P)fB(l)
iJW(I)
Since - - > 0 at 1= 1*. it must be that dl*/dF> 0 and the threshold become, aFal
more permissive (moves right) as F increases. Case 2: search context Now suppose an actor searches for a single recipient (at a cost c per search). The threshold maximizes: w(t) =
PFG(I) + (I - P)fB(l) - ( PG(I) + (I - P)B(t) OmtlffUlJMp.1lJ.
88
CHAPTER 4
Box 4.1 Conl'd.
Here, ~~? < 0 at 1=1*. so dr*/dF < 0 and the threshold becomes more restrictive (moves left) as F increases. Thus, increasing value of Os causes greater choosiness when only one recipient can be accepted and less choosiness when multiple recipients can be accepted (Reeve. 1989).
chosen recipient is desirable. This theoretical result supports and generalizes to all recognition systems Bateman's (1948) principle (Trivers, 1972; Hubbell & Johnson, 1987) that in mate choice, males should emphasize quantity whereas females should emphasize quality. When the optimal acceptance threshold varies in different recognition contexts, organisms should assess environmental cues that distinguish such contexts (Hamilton, 1987b). These have been termed 'Iocational' (Holmes & Sherman, (982) or 'indirect' cues in the case of kin recognition, as opposed to 'direct' or phenotypic cues (Waldman er al., 1988). Non-phenotypic cues may signify limes or places where desirable or undesirable recipients are likely to be found. Phenotypic and non-phenotypic cues are not mutually exclusive; spatial and temporal cues often supplement phenotypic labels. As an example of a temporal cue, when female acorn woodpeckers (Melanerpes formicivorus) oviposit communally, each ejects eggs until she begins to lay (Mumme el al., 1983; Koenig el al., 1995); females thereby avoid destroying their own eggs. Locational cues are used by parents in some birds (e.g. bank swallows, Riparia riparia: Hoogland & Sherman, 1976) and mammals (Belding's ground squirrels: Sherman & Holmes, 19851 to recognize offspring. Young remain in their natal nest for several weeks after hatching (birth), so cues of the nest or burrow define a context in which the frequency of interaction with desirable recipients (offspring) is so high that universal acceptance is favoured (I.e. a maximally permissive acceptance threshold). This may seem paradoxical because parents should almost always have the opportunity to learn phenotypic cues that are probabilistically associated with their own offspring. However, selection might never favour use of these cues if they lead even infrequently to rejection of offspring, given the fitness cost of such rejection. Parent swallows and mother ground squirrels stop accepting foreign young when broods .begin to mix in nature. At that point, temporal cues indicate a new COntext in which the frequency of interaction with undesirable recipients
(i.e. non-offspring) is high enough that a more restrictive acceptance threshold is optimal. Thus, parents switch to using phenotypic cues, chicks' vocalizations in the swallows (Beecher er al., 1981 a,b) and pup odours in the ground sqUirrels (Holmes, 1984). The ontogeny of sibling recognition in the laller also involves a change from universal acceptance to discrimination against non-nestmates
(Holmes & Sherman, 1982; Holmes, 1986a, 1994).
RECO G N ITION S YSTE MS
89
The action component of kin recognition Kin-recognition acceptance thresholds shilt among COlllexts in the ways predicted if animals facultatively optimize the balance of acceptance and rejection errors. For example. paper wasp workers are more imoleralll of unrelated non-nestmates that arrive on their nest. where brood theft and usurpation occur (Kasuya et al.. J 980; Klahn, 1988), than they are when they meet the same non-relatives elsewhere (Gamboa el al.• 199Ib). probably because benefits of defensive aggression are low away from the colony (Fig. 4.5). On the nest, queens have more restrictive thresholds than workers. as evidenced by greater aggression than workers toward both nestmates and non-nestmates (Fishwild & Gamboa, 1992). Such a caste-mediated threshold shift makes adaptive sense because queens are more likely to illleract with non-nestmates that intrude on the nest than workers (Fishwild & Gamboa, 1992), and queens may have more to lose reproductively than workers from accepting them. Dramatic illustrations of cOlllext-dependent acceptance thresholds come from species that facultatively produce distinct morphs which diUer in their
ta)
10
,
7.5
.. ~
0
c
1.0
lii £
0.8
>.
~' 0.6 0..5: -:0 0._ c· .2 ~ 0.4
> ~
(b)
5
~
,,~
'
8.~
,2 2.5
~
*
Q.
0
Nestmates
Non-kin
Fig. 4.5 Examples of context-dependent kin discrimination. (a) In paper wasps, resident females are equally tolerant of nestmates on (grey bars) and off (white bars) the nest comb. By contrast, resident females are Significantly less tolerant of unrelated females on the comb, where brood theft and usurpation occur (Klahn. 1988). than when they meet the same non-relatives off the comb. (Data from Gamboa etal.. 199Ib.) (b) When satiated. carnivorous Plains spadefoot lOad tadpoles eat significantly fewer kin than non-kin; when they are very hungry. however, they engage in indiscriminate cannibalism. The experiment was done sequentially over 96 h. with food withheld for 24 h, 48 h, and then 24 h again. At the end of each fasting period. the carnivore's cOl1sumplion of kin and nOI1kin was assessed. An asterisk indicates that the observed value was significantly less than random expectation (indicated by the horizontal line). (Data from Pfennig et al., 1993.)
90
CHAPTER 4
ability 10 harm relatives. Plains spadefoot lOads (Scaphiopus bombifrons) breed in temporary desen pools. All tadpoles begin life as omnivores, feeding primarily on detritus. Occasionally, however, individuals eat freshwater shrimp. triggering a series of changes in morphology and dietary preference. Changed tadpoles become exclusively carnivorous, often feasting on conspecifics (pfennig, 1992). Omnivorous and carnivorous tadpoles have different context-specific decision rules (Pfennig et al.. 1993). Omnivores are allraeted to and school preferentially with siblings (i.e. kin are desirable recipients), while carnivores generally avoid relatives (i.e. kin are undesirable recipients of cannibalism). When carnivores arc very hungry, however, they associate wilh and eat kin and non-kin at random - i.c. their acceptance threshold becomes more permissive because risk of starvation increases their personal cost of rejecting related prey (Fig. 4.5). Hokit et al. (1996) present a dramatic example of context-specific changes in aggression toward kin in larval salamanders.
The action component ofmate recognition As with kin recognition, a recipient's cues may trigger dHferent actions by an actor depending on the mate-recognition context and its associated decision rule. There are at least three general Iypes of mate-quality decision rules. Individuals (usually females) may choose: (i) the best mate available among those they can sample; (iI) the first male they encounter whose characteristics exceed some threshold; and (iii) the mate chosen by other same-sex individuals (copying). Analogous decision rules apply 10 all types of recognition systems (Reeve, 1989). Janetos (1980) suggested that a 'best-of-It' decision rule (choosing the best mate from a sample of size nl yields higher fitness than a 'one-step' strategy (searching terminates only when the quality of an encountered mate exceeds the expected quality of subsequent mates). Parker (1983) then showed that the optimal threshold quality for acceptance depends acutely on search costs. Real (1990) demonstrated that, when search costs are included, a sequential sampling strategy (searching ceases only when the quality of an encountered mate exceeds some minimum threshold) generally yields higher fitness than does best-of-n; Real's model is similar to Reeve's (1989) repeated search model, which incorporates recognition errors and makes similar predictions. Empirical studies of mate-choice decision rules have barely begun (see Wiegmann et al., 1996). Nonetheless, it is already clear thaI females often do sample multiple males. Thus, female great snipe (Gallinago media: Fiske & Kalas. 1995), cock-of-the-rock (Rupicola rupicola: Trail & Adams, 1989) and Australian brush turkeys (Alectura lathami: Birks, 1996) visit up 10 12 males before returning 10 copulate with One. This indicates either a best-of-n decision rule or a sequential search in which the acceptance threshold becomes more permissive as the lime spent searclting increases (Fiske & Kalas, t 995). Female fruit flies
RECOGNITION SYSTEMS
91
require four to six bouts of courtship before becoming receptive, which enables them to sample the relative frequency of male phenotypes in the population. and to choose the rarest for mating (Spiess. 1987). Under the models of Parker (1983), Reeve (1989) and Real (1990), optimal mate-acceptance thresholds become more permissive with increasing search costs or decreasing benefits for chousiness. In support, female guppies (Paecilia
Cal House mice Choices of lactating females
Choices of oestrous females
1.8 c
0.6
-
1.4 LJ_ _..L-L_ _.LNot Chosen chosen as as
0.4 '-'-::-:--L---'--,,-_-'--Different Same from female's
d'
as female's
nestmates
nestmates
MHC similarity among
MHC type
<;'>-9
(bl Humans
Preferences of cycling ~
6
Preferences of women taking contraceptives
women
-
';
.'
-~
~
0
•
w
-
-;;;
E '0 4
•
E '0 4
w w
~-
~ ~
~
c
c
C
C
•• •
•
~
~
n:
-
6r
0
0 .' 0
~
n:
~
2
Different from female's
d'
Same
a. female's
MHC type
2
DIfferent from female's
d'
Same
a.
female's
MHC type
Fig. 4.6 Examples of facuhalive shifts between recognition systems in house mice and humans. (a) Sexually receptive female mice avoid males with similar MHC genotypes (thereby promoting outbreeding~. whereas pregnalll and lactating females are attraaed 10 females with similar MHC genotypes (With whom they form nepotistic communal nesting associations). (Data from Egid & Brown, J989: Manning tl al.. J992; Model 3: MHC prererences based on social learning.) (b~ Women find odours of males with different MHC genotypes more pleasant than odours of males with the same genotype unless they are laking oral contraceptives; Ihen they prefer males with similar genmypes. (Data from Wedekind' al.. 1995.)
92
CHAPTER 4
mictllata) collected from a river in which predators are common reduce both
their overall level of sexual activity and their preference for brightly coloured males in the presence of predators (Godin & Briggs, 1996), and female peacock wrasses (Symphodus linca) are choosiest about whelher they lay Iheir eggs in nests lended by a male, and aboul which nesls to lay in. when the fitness payoff for laying eggs in male-tended nests is highest (due 10 predalion) and sedech COStS are lowest (Warner el al.. 1995). Female zebra finches become choosier after Ihey are exposed 10 males with high display rates (Collins, 1995) implying thai. in accordance with Iheory (sec Box 4.1), their mate-acceplance thresholds become more restrictive as the assessed frequency of high-quality mates increases. The third decision rule. male-choice copying, occurs in many vertebrate (Gibson elol.. 1991; Dugatkin, 1992; Hoglund & Alatalo, 1995). Copiers may save time and reduce predation risks by not sampling every male; informalion about which males other females have accepted or rejected also enables individuals 10 minimize erroneous choices. Mate-choice copying also may affect the perception component of mate recognition by causing template updating in Iighl of evidence of allraCliveness to olher females (see SeCtion 4.3.3). Clearly, male-acceptance thresholds can be context dependent. In addition, which recognilion system is in operalion depends on the context. For example, female house mice prefer as mates males with dissimilar MHC genotypes (Egid & Brown, 1989; Potts et aI., 1991; Fig. 4.6), apparently to avoid inbreeding (Potts el al., 1994). However, when females are pregnanl or lactating, they foml communal nesting and nursing associalions with females that have similar MHC genotypes, which are usually close kin (Manning el al., 1992, 1995). Inlerestingly, human females also prefer odours of males with dissimilar MHC genotypes, except when Ihey are laking birth control pills; in that case, Ihey prefer odours of males with similar genotypes (Fig. 4.6). Wedekind el al. (1995) suggested thai Ihe basis of the former choice is oplimal outbreeding. whereas Ihe latter indicates nepotistic kin recognition by (pseudo-) pregnant females.
4.4 Current topics in recognition research 4.4.1 Failures of kin recognition Many organisms occasionally make acceptance or rejection errors. There arc at leasl three evolutionary reasons why Ihis may be so. First, circumstan cS favouring recognilion may be rare, or may have been rare until recently. Female Belding's ground squirrels behave nepOlistically (lnly to daughters and sislers. More distant relalives (granddaughlers, second cousins) are treated like non· kin (Sherman. (980). These distant kin are infrequently alive simultaneously (Sherman, 198Ib). Either eleclion has not favoured abilities 10 learn templates to recognize distant rdatives (perception cumponent). or the rate of interaction with them is so low that the optimal acceptance threshold is reslriClive.
RECOGNITION SYSTEMS
93
excluding distant kin (action component). As another example, bird species thaI have long been exposed 10 brood-parasitic cowbirds or cuckoos consistently reject parasite eggs, whereas species whose nesting habitats have been recently invaded (e.g. due to forest fragmentation) tend to accept all eggs in their nest, including those of parasites (Mayfield, 1965; Davies & Brooke, 1989). Second, errors may persist because the error-related costs of kin discrimination outweigh the benefits. Recall that recognition errors occur because of overlap in labels. If either rejection or acceptance errors become too costly, selectioll may favour universal acceptance Or rejection, respectively, of kin
and non-kin (Reeve, 1989). For example, male red-winged blackbirds (Agelaius pllOeniceus: Wesmeat et al., 1995) and house martins (Delichon urbica: Whillingham & Lifjeld, 1995) reed all the chicks in their nest, although 25-30% are unrelated due to extrapair copulations by Iheir mother. Since the chicks are of the same age and species, they closely resemble each other, but they look and sound nothing like adults. Apparemly it is more efficiem reproductively for males to feed all the chicks than to risk rejection errors, thereby allowing their progeny to starve (Wesllleat & Sherman, 1993; Kempenaers & Sheldon, 1996). Imerestingly, in several birds (e.g. dunnocks, Prunella modularis: Davies et al., 1992; alpine accenlOrs, P collaris: Hartley et al.. 1995) males do adjust how much they feed the entire brood based on a non-phenotypic (time-specific) cue: the amount of exclusive sexual access they had 10 the female (which correlates with degree of paternity). Brood parasitism by cuckoos also illustrates this cause of errors. Normally, first-lime parents learn what their own eggs I,)ok like and rejeC! eggs from subsequem clutches rhat do not march this template (Lotem et ai' 1995). The cost of the learning rule is that if a host is parasitized during its first breeding attempt, it will learn the parasite egg and be doomed 10 accept parasite eggs forever. When parasitism is sufficiently rare, the benefits of correct impriming exceed the costs of misimprinting, so acceptance errors persist (Lotem, 1993). Hosts minimize these acceptance errors by varying their rejedion rate in relation to their probability of parasitism (Davies et 01., 1996). Third, when recipients benefit from the absence of discrimination they will be favoured 10 hide their true kinship by 'muting' or 'scrambling' recognition labels (Reeve, 1997a), For example, extrapair copulations set in motion a coevolutionary race between males attempting to discriminate their progeny and unrelated juveniles attempting to dupe the male by not revealing their genotype via their phenotype (Beecher, 1991). If young are initially uncertain about whether the resident male is in fact their father, they all should hide, change or mix their recognition cues (e.g. via rapid growth and feather developmem) because the fatal COSI of being rejected, even if improbable, may exceed the small benefit, even if likely, of receiving extra food from a male that has identilied them as his progeny. Juveniles have an upper hand in this coevolutionary struggle because, on average, they have more to lose by being recognized and rejected than males lose through misdirected nepotism.
94
CHAPTER 4
4.4.2 Misunderstandings about kin recognition Kin recognition, like kin selection (Dawkins, t 979), is often misunderstood. We therefore highlight and attempt to clarify several semantic and conceptual bsues.
Misunderstanding 1: kin recognition favoured by kin selection must be mediated by genetic cues The flaw in this proposition (Grafen, 1990a, 1992) becomes clear when we view kin recognition as a strategy by which recognition-promoting alleles target copies of themselves in recipients (Reeve, 1997b). The recognition-promoting allele in the aaor does not direaly 'see' copies of itself but nevertheless promotes its own spread through a chain of positive statistical associations linking the recognition cues in the recipient with possession of a copy of the allele in the recipient. Whether labels are prodoced by genetic loci or acquired from the environment, the recognition-promoting allele spreads via an indirect association between cues and copies of the recognition-promoting allele in the recipient. The statistical association between cues and alleles is nOt necessarily less (and may often be greater) for environmentally acquired than for genetically specified cues (Gamboa etal., 1986b, 1991c; see also Section 4.3.2).
Misunderstanding 2: non-phenotypic recognition is not ·true' kin recognition Halpin (l99, p. 220) suggested 'animals that rely only on 'spatially-based recognition' are actually incapable of recognizing kin from non-kin, and it is precisely because of this that they are forced to rely on spatial cues to determine who will be treated as if they were kin: This argument also fails to appreciate that recognition-promoting alleles spread because of indirect statistical associations between the recognition cue (i.e. the location) and presence of the recognition-promoting allele in conspecifics. When frequencies of interactions with relatives at a panicular location (e.g. a nest burrow) are suffiCiently high, selection may favour universal acceptance at that location (see Seaion 4.1.4). Rarity of acceptance errors and costs of rejection errors, not mechanistic inability to use phenotypic labels, drives the evolution of kin recognition via non· phenotypic cues. In species exhibiting location·specific behaviour, parems occasionally do rear non-kin due to mix ups (in bank swallows: Hoogland & Sherman, 1976; Belding's ground squirrels: Sherman, 1980; paper wasps; Gamboa et al., 1986b), but these recognition errors are rare. Moreover, as we have seen, recognition systems based on phenotypic cues of genetic or environmental origin are not immune to errors either.
RECOGNITION SYSTEMS
95
Misunderstanding 3: kin recognition is an epiphenomenon ofspecies or group recognition According 10 Grafen (1990a, 1992) most examples of kin recognition are epiphenomenaL meaning thaI they are nOI maintained by kin selection bUI rather by selection for species or group member recognition. There are two problems with the alternative functions Grafen identified. First, in general, kin arc inappropriate templales for species recognition because if individuals learned to recognize all conspecifics based on their resemblance to kin only, they would acquire a very restriclive species-recognition template, and many rejeclion errors would result. Second, group member recognition is adually kin-selecled kin recognition in the majority of social insects, birds and mammals because group members arc close relatives (Sherman et al., 1995). Grafen's general poim is instructive, however, because some forms of kin discrimination fundion in contexts that do not involve kin selection or mate complementarity Isee Section 4.2).
4.5 Conclusions This chapter presents a synthetiC framework for understanding recognition systems by conceptually partitioning them into produdion, perception and action components. Charaderistics of these components vary predictably with the recognition context. Knowledge of each component illuminates the selective forces that shape recognition and behavioural discrimination, and vice versa, because selection favours components thaI enable individual cells and organisms 10 achieve an optimal balance between rejection errors and acceptance errors. These principles apply to all recognition systems, including recognition of mates, kin, habitats (e.g. Jaenike & Holt, 1991; Chivers & Smith, 1995), hosts (Honda, 1995). reciprocators (Reeve, 1997b), dominance (Butcher & Rohwer, 1989; Drickamer, 1992), predators and prey (Stephens & Krebs, 1986) and individuals (Caldwell, 1992; Temeles, 1994). Understanding recognition systems is important to developing and testing hypotheses about such diverse topics in evolutionary biology as mate choice, nepotism, intragenomic conflict, immune responsiveness and learning. Exciting areas for future research include the following. 1 Quantifying the fitness consequences of recognition in nalure, especially kin recognition (sec Section 4.2). One promising approach is to create behavioural 'mutants' that differ in their discriminatory abilities and then to compare, within natural populations, the inclusive fitnesses of individuals that discriminate, e.g. kin (individuals that were allowed to imprint on kin) versus individuals that do not discriminate kin (because they were forced to imprint On non-kin). Moreover, the widespread occurrence of facultative changes in the
96
CHAPTER 4
components of recognition, such as context-dependent acceptance thresholds (s~e Seclion 4.3.4, Box 4.1), has opened a new avenue of research into the functions of recognition systems. For example, the hypothesis that discrimination between dc::sirable and undesirable recipients functions to obtain some benefit' X' or avoid some cost' Y can be tested by assessing whether the
recognition cues, template weighting scheme and/or acceptance threshold (inferred from the probability of acceptance errors) vary in predictable ways as the magnitude of X or Y changes across recognition contexts. 2 Determining the occurrence of self-referent phenotype matching in Ih~ context of mate recognition (e.g. brood parasitic birds, t-alleles in mice) and nepotism (e.g. full- versus half-sibling discrimination among nestmates in birds and mammals or within insect colonies, and paternal discrimination of offspring versus unrelated young; see Sections 4.3.3 and 4.4.1). Is the armpit effect rare, or just subtle and hard to detect? The answer will enhance our understanding of the evolution of self-inspection and intragenomic connic!. 3 Determining how mechanisms underlying perception and action components develop (sec Sections 4.2, 4.3.3 and 4.3.4). Recent advances in understanding the neurophysiology of signal perception (e.g. Barlow, 1995) and processing (Schildbergerel al.. 1989; Suga, 1995) notwithstanding. for behavioural ecologists recognition mechanisms remain 'black boxes'. Studies of the neurophysiological processes underlying the development of recognition abilities will provide deeper insights not only into the mechanisms of recognition, but also into the more fundamental processes of learning and memory.
Chapter 5 Managing Time and Energy Innes C. Cuthill & Alasdair 1. Houston
5.1 Introduction All behaviour takes time. All behaviour consumes energy. But. the behavioural ecologist's interest in lime and energy does not simply lie in the lact that behaviour utilizes these key resources. Also important is the lact that they cannot be allocated 10 all behaviours simultaneously. The concept 01 tradeoffs is thus a central pillar in the evolutionary approach to behaviour. While one can see the imprint 01 trade-oIls in observed pallerns 01 behaviour and in comparative swdies. the most direct 1001 for examining them is the optimality model. Economic models force one to be explicit about the nature of the trade-offs involved and to explore their putative impact on the design 01 the behaviour. In this chapter we concentrate on lunctional (ultimate) aspects 01 the allocation of time and energy but. as we shall see. proximate and ultimate approaches to this question can. and should. be interlinked. We start with some classic economic models of decision making that make simplilying assumptions about the value 01 energy. then progress to models that can accommodate changes in value as a result o[ current energetic state. When should an animal feed? When energy or some other dietary component is required? When food is most abundant? Even the simplest lunctional hypothesis would have to take into account the fact that feeding is incompatible with many Dlher important behaviours. all of which require energy. Even when two behaviours can be performed concurrently. there will usually be reduction in efficiency compared to when each behaviour is performed in isolation (Futuyma & Moreno. 1988: Leigh. 1990). So. an organism must
acquire and store reserves not only for occasions when food is not available. but to fuel periods when it cannot leed due to the performance of other behaviours. A lull analysis 01 daily leeding pallerns is thus likely 10 have to take into account not only temporal variation in the costs and benefits of feeding. but temporal variation in the costs and benefits of other behaviours. But. how can we evaluale lhe fimess consequences 01 different feeding strategies?
97
98
CHAPTER 5
5.2 The functional approach The lunctional approach involves comparing actions in terms 01 their con· tribution to future reproductive success. In principle, this gives us a common currency for comparing aC1ions whose consequences might differ in a variety of ways (e.g. in amount of lood obtained and probability 01 being killed by a predator). In practice it is often difficult to assess future reproduC1ivc success. and so attention has often been locused on simple currencies which have the lollowing properties: I they are easy to measure; 2 maximizing them will maximize fitness. In the context of an animal's foraging decisions. the net rate of energetic gain Yis an obvious currency. Maximizing y maximizes the amount of energy obtained from a given time spent loraging. Maximizing yalso minimizes the lime required to obtain a given amount 01 energy (Schoener, 1971). We illustrate this using data on the modes 01 hunting of the kestrel (Falco tinnul1culus). Kestrels may hunt for prey by !lying or by sitting on a perch. Masman el al. (1988) refer to these methods as !light hunting and perch hunting. respectively. and provide information on their energetic consequences. In both summer and winter. Dight hunting has both a higher rate of energy intake and a higher rate of energy expenditure. In winter. the gross rate of gain b, from !light hunting is 13 1.8 kJ h-'. whereas the gross rate 01 gain bp lrom perch hunting is 13.2 kJ h-I. The corresponding rates 01 expenditure are er = 52.2 kJ h-I and ep = 6.5 kJ h-', respectively. Thus, !light hunting has a net rate of gain y, =b,- e, =79.6 kJ h-' and perch hunting has a net rate of gain yp =bp ep = 6.7 kJ h-'. The kestrel will maximize the energy gain in I day by using !light hunting for all the available time, and will minimize the time to balance its energy budget by Dighl hunting lor a time I,. where Iry, is the energy spent on other activities. Flight hunting is indeed the predominant foraging behaviour. but the fact that kestrels use perch hunting at all is not consistent with the simple predictions of raLe maximizing or time minimizing given above. (For
funher discussion and parameter values for summer foraging. see Masman et at.. 1988). The kestrel story illustrates how a rate-maximizing model can be applied to animal decision-making. It also indicates that such models may not be able to explain all aspects of foraging behaviour. The maximization 01 y forms the basis 01 many optimal foraging models. including the standard models of prey choice and patch use (see Stephens & Krebs, 1986). Although net rate is an obvious and appealing currency. in some cases it has been shown that another currency gives a better description of the data. Some of these examples involve parent birds bringing lood to their young. A bird's !light speed determines not only Ihe time Ihat the bird takes to travel a given distance, but also the rate of energy expenditure. By increasing its flight speed. a bird reduces the journey time (which increases the rale of delivery of food to the young), but will also usually increase its rate of energy
MANAGING TIME AND ENERGY
99
expenditure. Norberg (1981) introduced a way of calculating the costs and benefits of flying at a given speed. If a bird flies a distance D at speed v, then the time to complete the journey is simply Dlv. If this is the only effect that has to be considered, then the bird should fly as fast as possible (in order to reduce this time). Norberg's insight was that a bird will typically have to repay the energy that it spends on flight, and this will take time. If the bird spends energy at a rate Ply) and can replace energy at a rate g, then the lime to replace the energy spent is P(v)D
gv In this model, the best speed is the one that minimizes the total time
PlY)) -D( 1+-v
9
of the two components. This will maximize the overall rate to the young, subject to the constraint that the parent balances its energy budget. Norberg introduced this idea in the context of a parent bird bringing food to its young (in which case D is the round-trip distance from the nest to the feeding site and back again), but the principle is general and has been used in the context of migration by Alerstam and Lindstrom (1990). Mclaughlin and Montgomerie (1990) studying lapland longspurs (Calcarius Iapponicus) and Welham and Ydenberg (1993) studying black terns (Chiidonias nigra) report that the flight speed of parent birds does not maximize overall delivery rate but is close to the speed that maximizes efficiency. Houston (1993) and Hedenstrom and Alerstam (1995) show that this is to be expected if the parents' energy expenditure is up against a constraint. Hedenstrom and Alerstam (1995) give a general review of the currencies that have been used for assessing flight speed and also establish some new general results. Efficiency is the ratio of energy gained to energy spent and hence has no dimensions. It is obvious that if an animal has a fixed amoum of energy to spend, then maximiZing e[ficiency will maximize the amount of energy
obtained. This argumem has been suggested as an explanation of why honey bees (Apis mellifera) appear 10 maximize efficiency (Schmid-Hempel et al., 1985), but what mallcrs in this case is not the energy brought in by each bee over its foraging life, but the gains to the colony from energy brought in and the cost to the colony of a bee dying. Even if each bee has a fixed amoum of energy that it can spend, maximizing the energy brought in by maximizing efficiency will not necessarily maximize the reproductive success of the colony (Houston etal., 1988). One reason why foragers might not maximize long-term rate depends on the idea of constraints on the amount of energy that an animal can acquire or spend in a given time. The ability of an animal to assimilate energy is likely to
100
CHAPTER 5
be limiled by ils physiology IKirkwood, 1983; Weiner, 1992). In addilion 10 Ihis limil, it has also been argued Ihat there is an upper Iimil to the rale at which an animal can spend.energy (Drent & Daan, 1980). These constraints ml'an that an animal may not be able to forage for all the time available 10 it (Ydenberg et al., 1994; Houston. 1995). As a resull. Ihere can sometimes be an advantage in choosing an oplion Ihal does not have the highest value of y. If Ihere is a foraging oplion which also has a lower rale of energelic expenditure than the oplion with Ihe highest y. then the low-cost option can be adopted for a longer lime before Ihe constraint on energetic expenditure is reached. However, if no foraging oplions allow the animal to escape the constraint on energetic expenditure then it should use the option that maximizes a modified form of effidency for as long as possible and spend the rest of ils lime resting (see Houston, 1995). II is easy enough to lrunk of biologically plausible silualions when Ihe simple currencies considered above are highly correlated wilh fimess. However, il is even easier to think of cases when gathering energy at the maximum possible rale. or at the lowesl unit cost. are nOI sensible slrategies. Good feeding siles may incur a rugh predalion risk. Mighl nOI Ihe value of energy change with lime and circumstance? Is il nOllikely. for example, Ihat a dominant individual. sure of displacing conspccifics from good feeding siles, mighl show different feeding pallerns from a subordinate? And Ihat being as fal as possible is not necessarily a good thing? How can we incorporale such biological realism into our models? To calculale Ihe value of energy, we need to know whal it is going to be used for and when. Then we need to know the COSIS of acquiring and maintaining energy reserves. Finally, we need a way of evalualing Ihesc benefits and costs such Ihat we can predicl the effeds of changing lime and circumstance on foraging pallerns.
5.3 The benefits of energy reserves Ullimately. Ihe value of feeding is to provide Ihe raw material for selfmaintenance, growlh and reproduction. In this chapler we shall confine ourselves to Ihe use of energy, bUI the same principles can be applied to other dietary components. These may be required on different lime scales (e.g. Ihe season-specific requirements of small birds for calcium to produce egg-shell; Graveland & Vangijzcn, 1994; Krementz & Ankney, 1995) and the oplimal diet may ilself involve Irade-offs belween different dielary requirements (Belovsky, 1978; Murphy, 1994; Raubenheimer & Simpson, 1995), but the modelling approach is the same. No animal can fuel expenditure simply from current income. There will always be limes when it is nOI feeding (because it cannOl, or Ihere are beller things 10 do), so il has 10 store reserves for Ihese occasions. The lime-scale and magnitude of requirements can vary markedly, from sufficient fuel to survive
MANAGING TIME AND ENERGY
101
an emire winter or to complete a migratory flight, to overnight survival. Such periods of enforced fasting are to an extent predictable, although variability in weather conditions, and changes in windspeed or direction, make the energetic requirements hard to specify exactly in advance. But, short-term changes in weather. prey behaviour and simple lack of omniscience as to prey distribution, also render moment-to-moment foraging success stochastic. Although even small animals (e.g. shrews) are unlikely to die if they fail to feed for a few minutes, it only needs several successive runs of bad luck for a disastrous cumulative effect on survival prospects. What is the evidence that animals alter foraging behaviour to anticipate either predictable shortfalls in food availability or to buffer themselves against environmemal predictability? For animals at high latitudes. winter represents a deterioration in the foraging environment for several reasons. Nights arc longer with. for diurnal animals, the double penalty of reduced foraging time during the day and an increased period of enforced fasting at night. Temperatures are lower, increasing basal metabolic expenditure. Food is both lower in mean availability and, perhaps, the variance in food intake may be higher. Foraging time may be shortened and rendered more variable by periods of bad weather, such as storms or snowfall. All such effects should favour storage of greater energy reserves (Lima, 1986; McNamara & Houston, 1990; Houston & McNamara, 1993; McNamara el al.. 1994), but this need not imply that animals respond to these factors as proximate cues. For example. as hibernators must acquire large reserves well in advance of winter, a hyperphagic response based on decline in current food availability or temperature would be maladaptive. Thus, for example, captive alpine marmots (Marmofa marmola) show a 100% increase in food intake even under constant feeding conditions (Kortner & Heldmaier, 1995). However, animals that remain active throughout winter could plausibly rely on direct response to local conditions, use of short-term predictive cues (such as changes in weather) or long-term predictive cues (such as photoperiod). Which strategy, or combination of strategies, is best will depend on the predictive power of the cues. the cost of tracking them, the benefit of energy storage come the change in conditions and the cost of storing energy should conditions fail to change (see also Witter & Cuthill, 1993; Rogers et al., 1994). Direct response to local deterioration in environmental conditions has heen demonstrated for a variety of variables. Laboratory studies on several species of passerine bird have shown thai they store more fat at low temperatures (Komogiannis, 1967; Kendeigh el al.. 1969; Ekman & Hake. 1990), and Bednekoff el al. (1994) found higher evening masses in great tits (Parus major) exposed to variable overnight temperatures than to constam conditions. However, the latter result seemed to be more a response to the coldest night experienced than a response 10 variability per se; without a factorial design one cannot separate these effects. Bednekoff el al. (1994) argue that their birds use daily
102
CHAPTER 5
~ .:,;.
(al
1.07
.~
1.05 irl G E C>
c
'E
..
0
E
G
>
':ii
Long days
0;
a:
0.97l-------J-----~----____:!
6
16
~
~
Day
(b)
7
6 ~
5
G G
G
E .S G
' .c C G
u
4
3
2
o l--.l-_-'-_-'-_...J...._-'-_-J_----'_ _'--Control
Afternoon
Variable
Deprivation treatment
Fig. 5.1 (a) Great tilS were switched from a 9-h feeding day ('long days') 10 a 6·h feeding day .'shon days') while holding photoperiod constant. On short days. birds gained mass relative to their mass during Ihe first phase of Ihe experiment (scaled 10 a value of I); on
switching back to long days they rapidly lost mass. (From Bednekoff & Krebs. t995.) (b) Starlings were exposed to fixed length interruptions in their fceding day. These interruptions could occur in the morning. in Ihe afternoon. in either the morning or afternoon wilh equal probability rvariable') or not al all (control). Birds in all experimental groups increased mass relative to controls. wilh afternoon and variabll' interruplions producing Ihe largest mass gains. (From Witter rt al.. 1995.)
temperatures to predict overnight enc:rgy requirements, but that if the nigh I differs from expectation, they regulate overnight metabolic expenditure to match their reserveS.
MANAGING TIME AND ENERGY
103
Various forms of interruption from feeding have also been shown to produce a direct response in avian fat storage. While several field studies have indicated that photoperiod is a good predictor of body mass in birds (reviewed by Blem, 1990), Bednekoff and Krebs (1995) demonstrated a direct response to manipulation of the length of the time available to feed each day. Great tits whose feeding day was shortened by 3 h rapidly gained mass (Fig. 5.la). probably through reducing energetic expenditure rather than eating more. In a second experiment, birds gained more mass on days of unpredictable length than days of predictable length equivalent in mean duration. The data of Bednekoff and Krebs (1995) suggest a differential response 10 short and long feeding days, with the former having a stronger effect; as with Bednekoff ef al.'s (1994) temperature experiment. there is no evidence of a response to variability per se. tnterruptions from feeding during the day have also been shown to promote increased fat storage in greenfinches (Cardl/elis chloris: Ekman & Hake. 1990) and starlings (SlIIrnus vulgaris: Witter el al.. 1995). Witter ef al. (1995) sought to discriminate effects of unpredictability from the actual time of interruption: fixed length interruptions could occur in the morning, in the afternoon, either in the morning or afternoon with equal
probability or not at all (control). Birds in all experimental groups increased mass relative to controls, with afternoon and variable interruptions producing the largest mass gains (Fig. 5.1 b). Afternoon interruptions might be expected to have a larger effect than morning interruptions, as there is less time
(until dusk) in which to replenish lost reserves (McNamara ef al., 1994). Winer (1995) results support this predicted differential response to the time at which interruptions occur but. again. no additional response to variability was found. Species may dHfer in their responsiveness to manipulations of feeding conditions depending upon the predictability of the food supply to which they are adapted (Rogers & Smith, 1993; Rogers ef al., 1994). In meadow voles (Micro/us pennsylvanicus) there appear to be dHferent phenotypes with respect to response to winter conditions; some become non-reproductive and reduce body mass to save energy, others maintain body mass and breed opportunistically (Bronson & Kerbeshian, 1995). There is also evidence that the response to feeding interruptions is seasonally modulated. Witter e/ al. (1995) found that photosensitive starlings increased body mass in response to food deprivations, but that phOlorefractory birds maintained a constant body mass without increasing it. The difference seemed linked to photorefractoriness (lack of gonadal response to long days; Dawson ef al., 1985) rather than the period of moult. with which it overlaps. but the adaptive signi!icance o! such modulation is unclear (see discussion in Witter ef al., 1995). Several authors have !ound differences in mass and fat correlated with factors expected to affect the predictability of food intake but, in light of the experimental evidence above, we should be cautious in interpreting such results as a response to variability itselL Thus, ground-feeding birds, in habitats prone
e/ al.'s
104
CHAPTER 5
to sudden snowfall, store more fat than tree-feeding species (Stuebe & Kenerson, 1982; Rogers, 1987; Rogers & Smith, t993), but it may not be the differences in the predictability of interruption that arc critical. Alternatively, birds in such cooditions may have mechanisms that provide adaptive responses to feeding unpredictability without cueing on predictability itself. More problematic is the interpretation of effects of social dominance on fat storage. Higher fat storage by subdominant than dominant great tits (Gosler, 1996) and willow tits (Porus man/anus: Ekman & Lilliendahl. 1993) has been interpreted by these authors as a response to a less predidable food supply, due to displacement of subdominants from feeding sites. This may well be the case, and manipulations of dominance by removal of birds also produce the predicted changes in body mass (Ekman & Lilliendahl, 1993: Willer & Swaddle, 1995). However, dominance may alter not only the mean and variance in food gain, but energetic expenditure (Hogstad, 1987; Bryant & Newton, 1994) and predation risk (Ekman, 1987), all of which are predicted to affect optimal fat storage (Willer & Cuthill, (993). The degree of competition for food has also been shown to ailed the optimal fat reserves of dominants and subdominant to differing degrees (Willer & Swaddle, (995), so in fad any relationship between social dominance and fat storage is possible, depending upon ecological circumstance (see discussion in Willer & Swaddle, 1995). Such arguments based on the dominance-dependent costs and benefits of fat storage go some way towards helping us understand the widely varying pallerns seen in field data (see Witter & Cuthill, 1993; Witter & Swaddle, 1995), but we are a long way from understanding the proximate mechanisms by which such differences come about.
5.4 The costs of energy reserves It may seem obvious to a behavioural ecologist, schooled in cost-benefit
reasoning, that fat torage must have costs. But, many analyses of intraspecific variation in body fat, or mass, seem to neglectlhis point. Most obvious is the use of fat or mass, usually scaled to control for skeletal size variation, as a measure of condition. Animals with naturally high levels of fat. or high relative body mass, are considered to be in 'good condition' or of 'high quality'. Such data arc then correlated with other fitness components, such as mate acquisition, breeding success or survival. They may also be used as a bioassay, of sorts, for habitat quality. Yet, for fat or relative mass to be positively correlated with individual or habitat quality would seem to suggest that fat storage is limited only by food availability or the ability to acquire it. Is this the case? The most extensive data sets come from avian morphometric studies and these provide substantial. if circumstantial. evidence that typical levels of energy storage are not food limited. First, temperate zone species are fatter in winter than summer, and often fattest in the coldest periods of the winter (references in Blem, 1990; Witter & Cuthill. 1993). Winter days are shoner
MANAGING TIME AND ENERGY
105
and temperatures lower, leading to higher metabolic costs and lower food availability, so winter is precisely the lime of year when wc might expect birds to have lowest energy reserves. The fact that they store less fat in summer suggests that fat storage is not (usually) limited by food supply (King & Murphy, 1985). Second, body mass changes prior to migration or breeding suggest strategic regulation independent of current food supply (Mrosovsky & Sherry, 1980; Sherry et al.• 1980; Gwinner. 1990; Wingfield et al.. 1990). That energy reserves are usually maintained below the physiological maximum, or those dictated by food availability, suggests that the acquisition and storage 01 energy has co ts. 5.4.1 Acquisition costs
There are two types of cost associated with feeding: (i) acquisition of the food; and (ii) maintenance 01 the energy reserve once secured. Acquisition costs intrinsic to the foraging process itself. such as search and handling time, have always been part of classical optimal foraging models (Stephens & Krebs. 1986). That different behaviours involve different metabolic costs has also been considered by simple optimality models based on maximi7.ation of net energy intake rate (see Section 5.2; also Houston, 1986). However. foraging animals are also likely to experience greater predation risk (Lima, 1986; McNamara & Houston. 1987). Through being active, a forager spends a greater amount of time exposed to predators than an inactive animal (a suggested functional explanation for the initial evolution of sleep: Meddis. J 993). Also. as foraging usually requires close attention to the task (detection of cryptic prey. stealth in approaching prey. pursuit of lIeeing prey), there is liable to be a higher predation risk per unit time. due to decreased vigilance (e.g. Milinski, 1984). Although predation risk is the most widely considered cost. there may be other costs associated with heightened activity. such as risk of injury (Cuthill & Guilford. 1990) or parasitism (e.g Schmid-Hempel & Schmid-Hempel. 1993). 5.4.2 Storage costs
Food, once acquired, can either be transported to another location for subsequent processing or storage. or consumed immediately. These different immediate fates have different attendant costs, which will in turn inlIuence the optimal acquisition strategy. Assuming that the ultimate consumer is the forager itself (foraging for dependent young or as part of a cooperative group comprises a vast literature in itself). should an animal store energy as fat or in an external larder or cache? It may be that the amount of energy required is greater than the maximum fat storage that is physiologically possible. In th.is case. the only options are caching or reduction of the energy requirements through torpor. Some species do not build large caches for use in subsequent months. but scatter-hoard small amounts for use on a much shorter time-
CHAPTER 5
106 9
15
15
10
10
~
c ~ > m
'c ~
:§
m 0
~
'0
§ ~ ~
5
5
:;•
~
-' E ~
z
0
low
High
0
Variability in food supply
Fig. S.2 Hudy (1992) measured s~ed hoarding and body mass changes in captive marsh tits by means of infrared detectors and compuler·controlled balances. When an automatic feeder delivered seeds at highly unpredictable times, the birds hoarded more seeds per day. bUI did 1101 increase body reserves. Greater hoarding under variable ft:t:ding conditions is predicted when birds minimize the chance of an energetiC shortfall during the day or
overnight. The (act that this extra energetiC buffer is stored as a hoard rather than as fat. suggests that the lauer is more costly than the former in this particular situation. (Rt'ploned from data in Hudy. 1992.)
scale. Marsh tits and black-capped chickadees. for example. relocate and eal hidden food a few hours or days aller storing it (Sherry. 1985; Stevens & Krebs. 1986); why not SlOre these small amoums of energy in the body as fat? There are clearly special costs of storing energy in an external cache. These include the fact that hidden food can be pilfered. travel costs in taking the food. and subsequently returning. to the cache site. mistakes in relocating the cache (McNamara (I al.• 1990; Lucas & Walter, 1991) and costs of memorizing the location and subsequemly accessing that information (see Chapter 3). If one is considering the evolution of hoarding as a strategy. rather than the decisions of a species which already has the capacity to slOre food. there are the harder to quantify costs of maimaining extra neural tissue for spatial memory tasks (see Chapter 3 and Sherry (I al.. 1992). However. storage of energy in the body, as fat for example. also has distinctive costs (see below). An examination of the relative costs of hoarding versus fat storage can shed light not only on why some species have evolved food caching as an energy storage strategy. but also the fine details of individual hoarding and retrieval decisions (McNamara el al., 1990; Lucas & Walter. 1991; Hurty, 1992; Fig. 5.2). The nature of the costs of fat storage are dealt with in more detail below. as they have implications for a wider range of animals and behaviours. 5.4.3 The costs of being fat
The most obvious costs of fat storage in human life are the various pathological conditions associated with obeSity: heart disease, gall and kidney SlOnes. diabetes and various fonns of cancer and arthritis (see, e.g. Bjorntorp & Brndoff. 1992). But. while human males can be considered obese if fal exceeds 20%
MANAG[NG TIME AND ENERGY
[07
of their body mass, pregnant female polar bears (Ursus marilimlls) enter their winter dens carrying I kg of fat for every I kg of lean mass (Atkinson & Ramsay, (995) and some birds also double their mass in pre-migratory fattening (Lindstrom & Alerstam, 1992; Lindstrom & Piersma, 1993). So, a high level of fat storage is part of the life history of many animals. 'Obesity' can also aris<' as a byproduct of maintaining the correct nutrient balance when faced with a diet of suboptimal composition. Locusts fed on low-protein diets compensate for this deficiency by increasing intake rates, at the expense of excess carbohydrate consumption and elevated body mass (Raubenheimer & Simpson, 1995). There are almost no data on pathological effects of fat storage in wild animals (perhaps understandably: if migrating birds have heart allacks, how do we find out?), but even moderate levels of fat storage can have tangible ecological costs. These arise from the elevated body mass that fat storage entails, either directly from the adipose tissue, or indirectly through any changes in postural or locomotor muscles necessary for transporting the extra fat (e.g. Marsh, 1984; Piersma, 1990; see also Piersma, 1988). These 'mass-dependent' costs include elevated metabolic expenditure, particularly during locomotion, and reduced locomotor performance resultmg in increased predation risk and reduced foraging efficiency (reViewed in detail by Witter & Cuthill. 1993). The key issue is not whether elevated body mass has such costs, but whether they are of sufficient magnitude to be biologically imeresting. For example, does the approximately 5% increase in body mass of a green finch in response to a temperature drop (Ekman & Hake, 1990) a[fect its predation risk? Conversely, would changes in predation risk be suffidem to affect patterns of fat storage? The effects of changes in body mass on energetic expenditure arc derived largely from laboratory measurements of natural variation, and [rom theoretical calculation. That subcutaneous fat can reduce heat loss is well established for many mammals (see, e.g. Young, 1976; Pond, 1978) and various non-hibernating spedes may save total energy expenditure in winter by reducing body mass (Held maier & Steinlechner, 1981; Heldmaier el al' J 989). However, adipose tissue is relatively metabolically inert, so it is active melabolism, through mass-dependent locomotor costs, that is liable to impose the most substantial costs. The metabolic costs of walking or running increase linearly with load size in mammals (Taylor et al., 1982) and the crossspecies relationship between body mass and energetic costs of terrestrial locomotion are similarly linear for birds and mammals (Taylor et al., 1980; Peters, 1983). However, for Oight, theoretical calculations suggest that the power requirements should accelerate with changes in body mass across species (to an exponent of about 1.5; Pennycuik, 1989; Rayner, 1990). But, there arc lew experimental data on the effects of loading on metabolic expenditure in [light, and even less on the relationship between within-individual changes in body mass and flight costs (see Bryant & Tatner, 1991: Witter & Cuthill, 1993). This is an area ripe for empirical investigation. It seems likely that fat storage, and the increased body mass that results,
108
CHAPTER 5
may have decremental effects on locomotor performance, panicularly in nying animals. Theoretical models of night suggest reduced performance on various manoeuvres (Norberg, 1990; Alerstam & Lindstrom. 1990; Hedenstrom. 1992) and. not unreasonably, modellers of behaviour have assumed that predation risk is likely to be increased by such predicted deleterious effeCls on mobility (Uma. 1986; McNamara & Houston, 1990; McNamara il al., 1994). But. collecting empirical evidence for mass-dependent predation risk is not straightforward. First. natural correlations between body mass (or fat) and survival probability are confounded by variation in escape ability or risk-taking behaviour (e.g. 'good' individuals store more fat, take less risks and arc better at escaping predators). Second. predation rates need not be a good predictor of. or even positively correlated with, predation risk (McNamara & Houston, 1987; McNamara. 1990; Walts. 1990: Abrams, 1993a; McNamara & Houston. 1994). Animals trade off predation and starvation risk so. for example, individuals with low energy reserves may need to feed in more risky locations in order to avoid starvation, but in doing so expose themselves to higher predation risk. Fat individuals can afford to feed in safer sites, sacriIicing intake rate for safety so, dependent upon the assodation of patch quality with predation risk. it is possible to generate a variety of relationships. positive or negative, between energy reserves and predation rates (McNamara. 1990; Abrams, 1993a). Thus, evcn field measurement of survival rates in response to experimental manipulations of body mass may give spurious estimates of predation risk. A more indirect approach is required: evidence that increased body mass alters those aspects of locomotor performance likely to affect escape ability. evidence that increased body mass is assodated with increased anti-predator behaviOurs such as vigilance, and evidence that animals adjust body mass in response to differences in predation risk. There are few experiments on the effects of fat storage on locomotion. but several showing that other causes of mass increase, such as carrying eggs or artifidal loads. have significant effects. One must interpret these with caution. as it may not be the mass per Se thai is responsible for the observed changes. but such data give one confidence that fat may have quaHtatively similar dfl'cts. Mass increases associated with being graVid have been shown to impair mobility and reduce escape speeds in several species of lizard and snake (Shine, 1980; Andren, 1985; Madsen, 1987; Siegel il al., 1987; Cooper i/ al.. 1990). For example. gravid female Sceloporus occidinlalis lizards show lower sprint speeds (Snell el al.. 1988; Sinervo e/ al.. 1991) and yolkeClomies show that females w'ith reduced broods survive better than control females (Sinervo fI al.. 1991; Landwer, (994). In birds. carrying eggs reduces feeding efficiency in great tils (Parus major; Krebs, 1970), rate of ascent in sand martins (Riparia riparia; Jones. (986) and take-off angle in starlings (S/urnus vulgaris; Lee e/ al.. 1996). Jones (1986) demonstrated the same decreased Oight performance when females were experimentally loaded by injecting an eqUivalent mass of water into their body cavity (the mass gain is temporary as the water is naturally expelled).
MANAGING TIME AND ENERGY
109
This suggests that it is the added mass of the eggs or developing follicles which is responsible for the decrease in flying ability. But, does fat storage have similar effects, given that small birds show seasonal changes in flight muscle composition dUring egg-laying (Jones, 1991; Houston ft al., 1995a,b)? Witter et al. (1994) manipulated both natural body mass and artificial loads to investigate the effeet of fat reserves on escape performance in starlings. Ascent angle at take-off and manoeuvrability through an obstacle course were both reduced by increased body mass, whether natural or artificial (Fig. 5.3a.b). Although lat and artificial loads have differem distributions around a bird's cemre of gravity, the facl that the ellects on flight are of similar type and magnHude suggests that there is a definite cost of increased mass per sc. Strung
la) 50 r
,
T o
•
0.5
B
Mass added (g)
Ibl 50 r
T T
Fig. 5.3 (a) Adding artifidal masses to starlings result~d in a shallower angle or ascent when escaping from a simulated predator; (b) reduction of nalllral
f-
o
Control
body mass. by means of shonterm food deprivation, also showed that heavier birds had
Reduced
Mass treatment
lower take-orr angles. (From Willer rt at.. t 994.) (e) The nalUral daily variation in bod}
lei 0.'
...
• •
0.3
!1 ~
E
F
• • •
0.2
O. I 13
14
• •
.
15 Mass (9)
.. .
16
mass of zebra finches is correlated wilh the time they lake to ascend a given vertical
17
height. Lal~r in lhe day. birds are heavier and ascend morc slowly. The data plotted are from a single individual. but the trend was similar (and highly significanu across eight replicate birds. (From Metcalfe & Ure, 1995.)
liD
CHAPTER 5
drcumstantial support comes from the fact that diurnal vanatlon in body mass is correlated with diurnal variation in lake-of( ability in zebra linches (Taeniopygiaguttata: Melcalfe & Ure, 1995; Fig. 5.3c). II escape ability is reduced by increased body mass then we might expect animals to respond by increasing compensatory anti-predator behaviours. Schwarzkopf and Shine (1992) showed that although female southern water skinks (Eulamprus tympanum) suffer a decrease in mobility while gravid, these females also change their anti-predatory behaviour, relying on crypsis to avoid detection rather than active escape. Relating changes in, say, vigilance to fat storage is more problematic, as manipulation of fat reserves necessarily alters not just body mass but energetic stale. Thus, a fat animal may be heavier (reduced escape ability and hence higher predation risk), but it is also likely to ta)
1.0 0.8
Hawks
Hawks
decreasing
absent
2J: 2
0.6
:§ w w
M
E
,
0;
'0
0.4
0.0
.;;; a:
-0.4
'
-0.6
Hawks
established
i
oW~~
0.2
-0.2
j Hawks : : increasing:
f2
g
~
'0
-0.8 -1.0 48
52
56
50
64
68
72
16
80
84
88
92
96
'(ear
lb) 3.25 2.75 0 ~ w
2.25
'iii
~
1.15 1.25
I
i
~
r!f 1.0
r! 2.0 Cover score
I
3.0
FiB· S.4 (3) In years when sparrowhawks were absent from Wytham Woods in
Oxfordshire, the resident great tits maintained higher body mass than when the predators were present. (From Gnsler er al., 1995.) (b) Slarlings were assigned randomly 10
experimental aviaries diHering in the amOUIlI o( protective cover Ihey contained. As would be predicted from mass-dependent predation risk, birds in aviaries with more awt>r. and
thus a reduced predation risk, stored more fat lhan birds in less proteded aviaries. (From Willer et al.. 1994.)
MANAGING TIME AND ENERGY
III
favour vigilance over feeding because gaining energy is relatively less important. Both Ihe predation risk and starvation risk hypotheses predict the same change in vigilance behaviour, so the experiment in which fat levels are manipulated is worthless (but, see Willer et al' 1994). While increased anti-predator behaviours may compensate somewhat for increased vulnerability to predators (at the expense of decreased time allocated to other behaviours), animals might also regulate fat reserves in response to predation risk. Correlational data comes from changes in great til body mass over a 40-year period in Wytham Woods, Oxfordshire (Gosler el al., 1995). During years when the major predator, the sparrowhawk (Aaipiter nisus) was absent due to organopesticide poisoning, tits maintained higher relative body masses Ihan they do today (Fig. 5.4a). Such 'natural experiments' are open to confounding variables - populalion densilies may have changed, and competilion for food is known to affect rat reserves (Witter & Swaddle, 19951 bul. this is the best available field evidence to dale. Willer et al. (1994) sought to manipulale predalion risk ·in Ihe laboratory by randomly allocating groups of starlings to aviaries with different amounts of prolective cover. As predicted, birds in aviaries with more cover stored more fat (Fig. 5.4b).
5.5 Relating short-term behaviour to lifetime reproductive success Having considered the costs and benefits of energy acquisition and storage, how do we use this information to construcl more realistic models of foraging than those based on simple rate maximization? It is a central problem in behavioural ecology to evaluate Ihe payoff from various pallerns of behaviour over a short period of an animal's life. Yel, a functional explanalion of behaviour is based on Ihe contribulion Ihal behaviour makes to fitness, which can Iypically only be assessed by looking at an animal's Iifelime reproduclive succes,. It turns out thaI, at least in principle, the problem of relaling short-term consequences to lifetime reproductive success is easy to solve. Each feature oj an animal that is an important determinant of its reproductive success is taken to be a 'stale variable'. Such variables might be internal (e.g. body temperature, parasite load) or external (e.g. territory size). An animal's behaviour wiU typically change its state (e.g. foraging increases energy reserves, basking increases body temperature). I! we look at an animal at some time T, its expected reproductive success alter Ihis lime will depend On ils state at this time. We now wish to evaluate behaviour over a period of time that ends at time T. The relationship between state and reproductive success at this final time is known as the terminal reward. This provides the link between behaviour over the short time period and lifetime reproductive success. Even if the animal is not reproducing during the period, its choice of behavioor will innuence its state at T, and hence its future reproductive success. This can be illustrated with various simple examples based on foraging; having understood
J 12
CHAPTER 5
the principles we can then move on to more complex, and realistic, problems. We assume that the only important component of state is the animal's level of energy reserves, x. We start by assuming that foraging is deterministic and that the level of reserves has no costs in terms of metabolic rate or predation risk. We also assume that there is no predation risk while the animal forages. The animal has a range of foraging options, and under none of them will it reach the upper limit to its possible energy reserves by the fioal time T. Por each foraging option i, there is a net rate of energetic gain Yi • If the animal's level of reserves is X o at time 0, and it adopts option i lhroughoutthe time period, thea its state at T will be:
Jt follows that as long as the future reproductive success increases as the level of reserves at Tincreases (Le. the terminal reward is an increasing function 01 state) then the option with the highest net rate of gain should be chosen. VVe now consider various modifications.
Tlte terminal reward is a step function and foraging is dangerous Instead of assuming that future reproductive success always increases with reserves at T, we assume that future reproductive success is 0 if reserves are 12
8 ~
~ ~ ~
•
•
~
~
Rest
•
10
6
. •
a:
4 Forage
2 0
x, 0
2
4
6
8
10
12
14
16
18
20
lime (arbitrary units.
Fig. 5.5 The optimal policy for an animal over a s~ason (arbitrarily 20 time units). where the decision whether 10 feed or rest depends on both cncrgclic state and time. Here we as~ume that futuce reproductive success is 0 if reserves are less Ihan some critical level x<; and is maximal if reserves are equal to or greater than x(' Foraging can increase energy reserves, but exposes the animal to predation risk: resting is safe. but resling metabolism steadily burns up energy. The optimal polity for reaching the end of the season with reserves of al least Xc is to forage when reserves arc below the boundary and to rest when ahove. NotC' that the shape of the optimal policy of resting and foraging is such that an animal losing energy early in the season (left-hand squares) gains from resuming foraging, while later in the season (right-hand squares) an animal with lhe same or lower initial r~serves. losing exactly the same amount, does bettt'r by continuing to resi. Such an effect has been demonstrated experimentally in overwintering juvenile salmon (Bull etal.. 1996).
MANAGING TIME AND ENERGY
113
less than some critical level x, and is I if reserves are equal to or grealer Ihan x, (Fig. 5.5). All foraging oplions are dangerous, in Ihat there is a probability lhat the forager will be killed by a predator. We assume that this probability per unil lime spent foraging is the same for all the foraging options. If the animal decides not to forage. it can rest without incurring any predation risk. The expected future reproductive success associated with any behaviour between t: 0 and I : Tis given by the probability of surviving until Tmultiplied by the terminal reward. There is no advantage in having reserves greater than . at T, and survival probability decreases with time spent foraging. It follows thai the best behaviour is to minimize the time spent foraging, subjecl to lhe condition Ihal the level of reserves at T is x,., This is achieved by using Ihe foraging option with the highest possible mean net rale of gain for just long enough to ensure that reserves reach the critical level x, al T, and spending the remainder of the interval resting. Note that this characterization does not specify Ihe exact sequence of foraging and resting that should be observed. All sequences that resull in the optimal overall allocalion 01 time are equally good. Animals wilh very high levels of reserves may be able to survive by resting throughout the foraging period. In such circumstances, redUcing an animal's reserves by a given amount early in Ihe foraging period may result in an increase in foraging compared to reducing the reserves of an animal (wilh the same or even lower initial level of reserves) later in the foraging period (Fig. 5.5). This effect has been found by Bull et al. (1996) in a study of juvenile salmon (Salmo
salar). In general. foraging oplions will differ in both their nel rale of gain and their predalion risk. Gilliam and Fraser (1987) analyse Ihe optimal choice in such a caSe. They show that when the forager can choose between a habitat with no predation risk (a refuge) and foraging oplions that differ in lerms of nel rate of gain and their predation risk, Ihen the oplimal behaviour is 10 use Ihe refuge and the habilal with Ihe lowesI ralio 01 predalion rale 10 intake rale. This crilerion is able to predici the behaviour of juvenile fish (Semolilus alromaculatus, Cyprinidae) foraging in an experimental slream.
The lerminal reward is a step function, there is no predation risk but foraging is stochastic As in Ihe previous case, there is a crilical level of reserves x, Ihal mUSI be auained. Foraging is nOI dangerous, and there are only IWO foraging oplions. One option is determinislic wilh a mean net rate of gain y. The olher is stochaslic, with Ihe same mean nel rate of gain. Assuming Ihal the animal will use one of Ihe oplions for Ihe whole lime period, it is clear that il is best 10 use the deterministic option if the expected level of reserves at T is at leasl x, i.e. il:
xo+yT?x,
114
CHAPTER 5
and to use the stochastic option if Ihe expected level of reserves al T is less than x' This is the expected energy budget rule for risk-sensitive foraging (Stephens, 1981). When we move to considering an environment in which there is both predation risk and stochasticity (perhaps both in food intake and metabolic requirements), we rapidly move to models which are analytically intractable. Fortunately, there is a computational technique which can cope with this complexity, and capture the biological realism we require: the technique is stochastic dynamic programming. 5.5,1 Static versus dynamic models
So far we have assumed that the animal will use the same foraging option throughout the period under consideration. There are clearly cases when this sort of behaviour will be best (e.g. energy maximization), but there are other cases in which it is likely that the optimal behaviour will depend on either the animal's state, or the time remaining until final time T. or both. Such cases require an analysis that is dynamic. as opposed to the static analysis that underlies rate maximization. In finding the best behaviour, we now need to consider all possible ways in which behaviour could be specified by state or time over the time period. This is a daunting task, but there are two reasons why we need not be completely discouraged. 1 Under some circumstances, a potentially dynamic problem collapses to a static problem. 2 There is a simple and general computational technique (dynamic programming) for solving dynamic optimization problems. The first point can be illustrated in the context of foraging. Assume that an animal can forage with intensity u. This intensity determines both the mean net rate of energetic gain and the rate of predation. Both these rates increase with u, and the predation rate is accelerating. As a result, the animal can only increase its intake rate by increasing its probability of being killed, and each increase in foraging intensity is more costly in terms of predation than the previous one. In trying to find the foraging intensity that the animal should use if it is attempting to change its state from Xu at lime t =0 to x T at time 1 = T it turns out that under some circumstances we need only consider behaviour in which foraging intensity is constant. In particular, Houston et al. (1993) show that if: (i) foraging is deterministic; (ii) foraging is not subject to interruptions; and (iii) the animal's level of energy reserves does not influence its rate of gain or rate of predation, then u should be kept at the constant value that just gets reserves to xTat T. Houston el al. (1993) call this the risk-spreading theorem because it states that the predation risk is spread out equally over the time available. Given a terminal reward, we can then use this result to find the best value of xat T. But, in nature foraging is likely to be stochastic, interrupljons
MANAGING TIME AND ENERGY
115
Box 5.1 Stochastic dynamic programming (SOP) SDP is a numerical technique lor linding the optimal behavioural decision as a function of state and time. The animal is characterized by one Or morc state
variables (e.g. energy reserves. immunocompetence, knowledge) and a sel of possible behavioural options. The resull of its behaviour is to change its statc(s).
usually in a stochastic way. For example, loraging does not always result in food. The goal is to find the optimal sequence of decisions that maximizes
luture reprodudive success; the problem is the vast combination 01 possible actions as a lund ion 01 state and lime. SDP tackles this by working backwards Irom some linal time (T) at which the relationship between state and luture reproduoive success is known. Given this relationship then. for each state at
time T- I, the optimal choice lor this linal step can be lound readily. In doing so, one also obtains the expected future reproductive success associated with
each state at time T- 1. One proceeds to step T - 2 in an analogous way;
ont~
calculates the optimal choice that maximizes expected reproductive success at time T - I. Repeating this process for all dedsion times yields a matrix or decisions
Ihat conslitutes the optimal policy. The optimal policy specifies the best decision lor all states and times. Following this policy forward from the lirst time step. lor given initial state(s), generates the expected behaviour. For technical details see Mangel and Clark (J 988). may be common, and predation risk is likely to be mass-dependent (see Sections 5.3 and 5.4). We tackle such a situation in the next section.
5.5,2 Dynamic programming in action Dynamic programming is a general technique lor finding a solution to a dynamic optimization problem (Box 5.1). It was lirst applied systematically to behavioural problems by Mangel and Clark (1986) and McNamara and Houston (1986; see also Mangel & Clark 1988; Houston 'I al., 1988). The approach is especially useful when: (i) the relationship between behaviour and its consequences for the animal's state is stochastic rather than deterministic; and/or (ii) Ihe animal's state inlluences the possible actions or their consequences. Both conditions are likely to hold for a wide variety 01 behaviours. An example with both the above leatures is the analysis of singing versus loraging in songbirds (McNamara 'I al., 1987). In Ihis model, a male bird is characterized by his level 01 energy reserves, x. He dies of starvation il x falls to O. During the day, Ihe male can choose between singing and loraging; at night, he rests. Singing gives the male a chance to attract a mate, but uses up energy. By loraging the male can increase its energy reserves, but the amount 01 energy obtained is variable (with a specified distribution). The male's rate 01 energy
116
CHAPTER 5
expenditure increase with body mass and hence with energy reserveS. In some runs of the model. Ihe amount of energy needed to survive the night was fixed; in olher runs it was variable. This difference turn oulto be imponant in determining the optimal pallern of behaviour. McNamara et al. (1987) use dynamic programming to find the behaviour thai maximizes a male's expected value of the terminal reward at the end of a period of several days. If Ihe male is dead at this time, Ihen the reward is zero; if he is alive bUlunpaired he gains a positive reward; and if he is alive and paired his reward is greater still. It turns OUI that the exact values given to the two cases in which the male is alive are not crucial for determining the optimal behavioural rule. The optimal rule specWes, for each level of reserves and time of day, whether the male should sing or forage. Because of the danger of starvation males tend to forage when reserves are low, but above a critical level of reserves the male sings. As the end of the day approaches, Ihe critical level above which the male sings rises. This enables the male to build up reserves for the period of darkness when it cannot forage. When the energy used overnight is not variable males build up just enough energy to survive the night. tn contrast, when the energy used overnight is variable, males need to go to roost with reserves of more than the average overnight expendilure if they are to have a reasonable chance of survival. On most nights nOI all of this energy will be required. As a resull, the male will have energy 'left over' al the end of Ihe night and can best utilize this energy by singing. Thus, variability in overnight energy expenditure can produce a 'dawn chorus', i.e. a burst of song at dawn. This is a striking example of how a daily roUline can emerge even when the consequences of activities do not depend on time of day (for further investigations of this problem and how it relates to honest signalling, see Hutchinson el al.. 1993). In many circumstances, the succcs of a male that is calling to attract females will depend on Ihe behaviour of Olher males. This requires a game-theoretic analysis (e.g. Lucas & Howard, 1995). The importance of energy reserves for avian song has been demonstraled by Reid (1987) in the Ipswich sparrow Passerculus sandwichensis princeps and by Cuthill and Macdonald (1990) in the blackbird Turdus merula. Reid found that males sang less after cold nights, when overnight energy expenditure would have been high. Both studies found that the level of song in the morning could be increased by providing males with food. Allhough the model we have described is relatively simple, it illustrates some important general points. One is that the timing of an activily (song in this case) may only be understandable in the broad context of an animal's overall time budget. Another is lhal a selective force may be small but yet exert considerable innuence. The rouline of singing and foraging is driven by the threal of starvation, but under the optimal policy few bird starve. The general point is that the absolute magnitude of a cost does nOl tell us about the importance of the cost in determining optimal behaviour (McNamara & Houston, 1987; Abrams, 1993b; Houston &
MANAGING TIME AND ENERGY
117
McNamara, 1993). Thus, just because starvation is rare, it does not mean thaI we can
undt~rsland
behaviour without taking it into account.
5.5.3 Investigating behavioural routines
We have seen how dynamic programming can tackle the problem of optimal behavioural routines in the specific case of male birds singing to attract mates. The approach is. however, of general applicability and can be used to investigate routines of other behaviours and on different time-scales. For example, McNamara et at. (1994) consider whether a smaU bird in winter should forage or rest at panicular times of day. The resulting optimal daily routines are driven by the conflict between keeping reserves high to avoid starvalion, and keeping reserves low to avoid mass-dependent costs (see Section 5.4). If the bird is free to forage at any time of day, when food availabilil y is low, the optimal routine is to forage fairly intensively throughout the day (Fig. 5.6). When food availability is higher, there tends to be a burst of foraging at the start of the day and also in the aflemoon. Such patterns are often observed in nature (see McNamara et a/., 1994, for references). However, it is also realistic to consider the situation where a bird can be prevented from foraging by, for example, periods of inclement weather (see Section 5.3). fn SUdl an environment, it may be optimal to concentrate foraging a dawn and dusk, as long as an evening interruption cannot carryover to the next morning. These predictions remain untested because, as yet, we have insufficient information about environmental variation as experienced by birds in nature. 1.0
' 's. c
0.8
~
.E 0.6 'E ~
:0
'0 c .2
0.4
..e
0.2
t:
0 0.
o0~--'l':0--2:'0:---::'30=---'4':-0--:50~-60::7---=70=----::80':----'90~--:-!100 Dawn
Time of day
Dusk
Fig. 5.6 The predicted proportion of birds foraging .11 dilferent times of day. from a dynamic programming model where birds can either forage to gain energy or rest to avoid predation. In the case illustrated. foraging can occur continuously without interruption (if thl.:' bird chooses). When food availability is low, birds forage at a high level ror most or the day (Opt.'1l circles). When food avaiiabililY is higher. the overall level of foraging is lower and there is a burst of foraging at the stan or the day and in the afternoon (rilled circlesl. (Based on McNamara etal.. 1994.)
118
CHAPTER 5
Other routines that have been investigated include daily patterns of food hoarding (McNamara er al., 1990), daily patterns of venical migration in fish (Clark & Levy, 1988; Burrows, 1994), seasonal body mass changes (Bednekoff & Houston, 1994) and annual patterns of moulting (Holmgren & Hedenstrom, 1995). The ultimate application of dynamic programming would be to consider the optimal sequence of behaviour over an entire life cycle. This is where optimality modelling meets life-history theory (see Chapter 13) and is a nat ural extension of the approach outlined in this chapter. Questions that are traditionally the domain of life-history theory, such as optimal clutch size, diapause and phenotypic plasticity, have all been tackled in this way (see McNamara & Houston (1992), McNamara (1994) and Houston & McNamara (1992), respectively). The important messages lrom this chapter are, first, that current behaviour can have repercussions on survival prospects (or more generally fitness) at times quite removed from present actions. Second, that the best time to perform a particular behaviour can olLen depend on the costs and benefits of other behaviours at this and other times. Because all behaviour takes time and uses energy, an animal's energy reserve is a natural state variable to consider when modelling behavioural routines. Dynamic programming, because it allows one to model state-dependent decision-making and to link current actions to future payoffs, is a powerful tool for analysing the usc of time and energy.
5.6 Final considerations 5.6.1 Is your dynamic programme really necessary? The state-dependent approach that we have described has the advantage of producing models that can include a range of features that are biologically imponant. Furthermore, analysis of behaviour over a relatively short period of an animal's life is placed in the context of the whole life history (although this can be done with other methods; e.g. Abrams, 1982). The disadvantages of the approach are that quite a large amount of information may be needed in order to construct such models, and that the optimal policy will typically have to be found by a numerical technique such as dynamic programming. These advantages and disadvantages have been discussed several times (e.g. Houston & McNamara, 1988; Gladstein er al.. 1991; Houston ef al.. 1992; Abrams, 1995). Instead of going over all the arguments again, we will confine ourselves to a few points which are undeniable. 1 Dynamic programming is a particular tcchnique for finding optimal solutions. It is not always necessary to use this technique just because a simple rate-maximizing approach is inadequate. For example, the trade-oil between gaining energy and avoiding predators can often be analysed with static models (e.g. Abrams, 1982). 2 Stodtasticlty can be analysed without using dynamic programming. When
MANAGING TIME AND ENERGY
119
there is a single decision or a fixed sequence of decisions, it may be relatively easy to get analytic results about optimal behaviour. Tlte daily energy budget rule (Stephens, 1981) is an obvious example. When an animal makes repeated state-dependent choices in a stochastic environment, analylic results will be difficult to obtain, even in fairly schematic models (e.g. Houston et al' 1992). 5,6.2 Which state variables are important? In principle, one of the benefits of dynamic programming is that one can explicitly link funclion and mechanism by incorporating realistic state variables which can be measured empirically. For an animal to make adaptive decisions based on energy and time, one needs to know how each state is represented. This is a major challenge for behavioural biologists, as experimemal studies indicate that animals may respond to differemtypes of energetic stress in differem ways on different time-scales (e.g. lucas et al., 1993). However, this challenge is worthwhile as a proper understanding of behaviour must integrate functional and mechanistic accounts. We know that deterioration in the foraging environmem under many, but not all, situations leads to an increased storage of energy reserves (see Section 5.3). Functional models based on dynamic programming tell us why this is adaptive, and finer details such as the predicted pallem of weight gain on time-scales ranging from the day to the emire wimer (e.g Bednekoff & Houston, 1994). However, we know much less about rhe actual mechanisms by which 'environmental deterioration', or key facrors such as predation risk, are assessed and integrated. As we have seen earlier (see Section 5.3), while theory has predicted heightened fat levels in response ro unpredictability of the food supply, several experiments suggest it is the duration and timing of interruptions, as much as their unpredictability, which is important. This suggests a common mechanism. That mechanism is likely to involve the adrenal stress hormones, but very lillie work has been done on endocrine responses of wild animals under field conditions. Plasma corticosterone has been found to be higher in wild dark-eyed juncos (Junco ilyernalis) after recent snowfall, as was total adrenal mass (Rogers et al., 1993). As fallevels were correlated with these changes, and implanted corticosterone promotes feeding behaviour in previously food-deprived birds (Astheimer er al., 1992), the implication is that corticosterone mediates the influence of nutritional stress on energy srorage. Exogenous corticosterone also reduces overnight energy expenditure (Astheimer et al' 1992). Elevation of plasma glucocorticoid hormones, such as cortisone and cortisol. is the main endocrine response rolJowing capture stress in verlebrales, which suggests a mechanism by which predation risk could also affect fat storage, but heightened feeding under predation risk is precisely the opposite of what functional models would predict (Lima, 1986; McNamara & Houston, 1990; McNamara et al., 1994). [merestingly, the corticosteroid response to capture stress has been found to be negatively correlated with fat reserves in several species of birds (Smith el
120
CHAPTER 5
al.. 1994; Wingfield ef al.. 1994a.b). but we need
to know much more about the proximate factors affecting feeding behaviour and energy storage. their integration and the extent to which they are seasonally modulated (e.g. Wingfield el al., 1992). Functional models. such as the dynamic ones we have discussed in this chapter. are useful in indicating the key environmental and state variables to which we should direct our allention in such a search for mechanisms. But. more than this, they help us to understand the tradeo((s and conslrail1ls under which these mechanisms have evolved and are
maintained.
Chapter 6 Sperm Competition and Mating Systems Timothy R. Birkhead & Geoffrey A. Parker
6.1 Introduction 6.1.1 Definition Sperm competition is the competitIon between the ejaculates of different males for the fertilization of a given set of ova (Parker, 1970a). Thus. sperm competition forms a part of sexual selection (see Chapter 8), and includes the adaptations which arises as a result of it; e.g. any behaviour. morphology or physiology associated wilh multiple mating by females, paternity guards and ejaculate characteristics, all viewed from both a male and female perspective (Birkhead, 1996). Over Ihe last 25 years it has become clear that sperm competition is a remarkably powerful selective force. shaping life-hislory characteristics such as the body size, morphology, physiology and behaviour and even the evolution of the two sexes. Our aim in this chapter is first to consider how sperm competition may have driven the evolution of males and females (see Sections 6.1.2 and 6.1.3). Although sperm competition occurs in both external and internal fertilizers. in this chapter we will concentrate on species with internal fertilization. We discuss why females should bother to mate with more than one male, and how the resulting competition between ejaculates shapes male and female strategies (see Section 6.2 and elsewhere). [n Section 6.3 we consider the taxonomic incidence of sperm competition and show that it is virtually ubiquitous. Our main focus, however, is on the underlying mechanisms that determine which copulations result in fertilization (see Section 6.4). Finally, we consider the role that sperm competition has played in sexual conflict (see Section 6.5) and the evolution of mating systems (see Section 6.6). 6.1.2 Why two sexes? It would be hard nol to notice that most organisms exist in two forms:
(i) males. which produce tiny gametes (microgametes); and (Ii) females, which produce much larger gametes (macrogametes). Indeed. virtually all complex multicellular organisms exist as either separate male and female individuals (dioecious species). or less frequently as male and female ill the 121
122
CHAPTER 6
same individual (simultaneous hermaphrodites). In contrast, many unicellular forms produce just one size of gamete size (microgametes). II we define a sex in terms of the gamete size an individual produces, then there are either one or two sexes, broadly relating to the complexity of the organism. Why? There have been various theories, The earliest was that the rate of fusion in an external medium is maximized when the lusing gametes are wry diflerent in ,ize (Kalmus, 1932, Scudo, (967). At lea t in its simplest form this requires group selection. More recently, Hurst (1990) has suggested that a strong division into two gametic sizes reduces the possibility 01 transfer of cellular parasites during fertilization. Here we discuss the theory of Parker, Baker and Smith (1972; referred to from now on as PBS) since it specifically deals with the role of gamete competition in the evolution of males and females. PBS showed by computer simulation that in some hypothetical externally fertilizing ancestor, coexistence of males and females would be an evolutionarily stable strategy (ESS; Maynard Smith, 1982) when there is a high advantage in provisioning the zygote, and that producing microgametes is the ESS when the advantage of provisioning is weaker. Specifically, in a population in which all individuals produce gametes of equal size (isogamy), two gamete sizes (anisogamy) are favoured il the relationship between zygote fitness and zygote size is accelerating (at least over part of its range). Large gametes result in zygotes which survive well, but can be produced only in small numbers because they are costly, Small gametes contribute little to zygote survival. but obtain vastly more fusions, Starting from an isogamous population, disruptive selection quickly produces two sexes (males and lemales) because small-gamete producers and largegamete producers are simultaneously favoured; producers of intermediatesized gametes quickly become extinct. In a sense, males succeed by parasitizing females: sperm become smaller and smaller in order that more and more of them can be produced, In the original PBS model. fusion between gametes was random. Because of relative numbers, most fusions are between sperm, but since sperm-spenn zygotes have negligible viability, mechanisms to prevent such fusions are likely to evolve quickly, leaving a vast predominance of sperm in the gametic pool. Almost all ova are snatched immediately by the ubiquitous sperm, and sperm-ovum zygotes have good prospects. However, ova do better if they are able to fuse with other ova, since ovum-{)vum fusions have much higher fitness, especially if sperm are tiny. There is a good reason why ovum producers cannot 'retaliate'. Parker (1978b) showed that if sperm have become small (and hence numerous) enough before an ovum producer retaliates by mutating in such a way that it permits only ovum-{)vum fusions, then all the wild-type, randomly lusing ova will be grabbed by sperm before they have any significant prospect of collision with mutant ova. Thus, the mutant ova remain unfertilized, or are forced eventually into self-fusions. ll'retaliation' occurs before sperm have become small, then sperm may be lost. But, renewed 'sperm drives' from within the retaliator
SPERM COMPETITION
123
female population would eventually proceed to a point where sperm become too small and numerous to make retaliation [avourable. The ESS is stable coexistence of males and females. At this point, ova are expected to lose their motility and to channel the savings into resources for zygote survival. effectively 'settling' for sperm-ovum fusions (Parker, 1984). Some evidence for PBS comes from algae and protozoans (Knowlton, t 974; Bell, 1978, 1982). One would expect that the importance of zygote size would increase with complexity and size; Knowlton (1974) showed that in the volvocine algae there is a general transition from isogamy in the simplest unicellular forms to anisogamy in the more complex of the multicellular forms. Bell (1978) found a similar (but less distinct) correlation in other groups of chlorophyte algae (see also Bell. t 982). 6.1.3 Sperm competition, sexual selection and sexual conflict
A primordial form of gamete competition mal' then account for why there are two gamete-producing morphs (males and females), not three, four or five. Furthermore. once selective fusion has evolved. PBS generates males and females in a I : l sex ratio. The reason for equal production of the two sexes is due essentially to an argument originally put forward by Fisher (1930). Suppose that all eggs are fertilized (males are never very rare). Each female gains a fixed number. n. of progeny whatever the sex ratio. Bul, the expected gains per male are the total progeny divided by the number of males: i.e. (n x females)/males. If males are less common than females. each male gains more offspring than each female by fertiliZing the progeny of more than one female. If males are more common than females, the expected number of progeny per male is less than that per female because more than one male must share each female's progeny. The ESS - for an autosomal gene determining the sex of the progeny - is at: I sex ratio, although different genetic mechanism and deviations (rom random mating can generate adaptive sex biases (Hamilton, 1967). If a single male fertilizes a given set of ova (as may be common with internal fertilization). there is no longer any advantage in prodUcing huge numbers of microgametes to compete for fertilizations, since there are no other microgamete-producing competitors. PBS depends upon sperm competitionit cannot generate anisogamy when gametes from just two adults fuse (although it could then arise by Kalmus' model without invoking group selection). Instead, there should be a return to isogamy withjusl one sex. with the two spawning individuals sharing the burden of the cytoplasmic resources in eggs. each producing gametes of 50% of the optimal size (Parker, 1982). If this means that males cannot persist with internal fertilization. PBS is clearly wrong - because they clearly do! In fact, sperm competition is commonplace in internally fertilizing species. and the original definition of the term and development of the concept (Parker, 1970a) was actually for an internally fertilizing group - the insects. It appears that only very low levels
124
CHAPTER 6
of sperm competition are necessary to maintain males. especially if eggs and sperm are very different in size (Parker, 1982). To see why this is so. consider the cow's ovum. It is roughly 20000 times the size of a bull's sperm. If a mutant bull were to halve ils sperm number and so double sperm size. then (assuming that normal sperm contribule nothing to the zygote but deoxyribonucleic acid (DNA» this would increase the cytoplasmic reserves of the egg by only one unit in 20000. an entirely trivial benefit. It nevenheless generates a huge cost whenever two bulls mate with the same cow (relative fertilization probability drops from one-half to one-third). Parker (1982) has shown that it will not pay to increase cytoplasmic reserves in the sperm if Ihe probability of sperm competition (between two bulls) is greater than (roughly) four times the ratio of minimal sperm size divided by the optimal ovum size. In most species, this ratio is a very tiny value. so that anisogamy will be highly stable (Fig. 6.1). So. sperm competition may not only have both produced the two sexes. it may also currently maimain them. This is hardly trivial- bUI. as we shall see. it also does much more than this. We therefore end up with males and females in a 1 : I sex ratio. Darwin (1871) poimed out that this, coupled with the relative cheapness of ejaculates compared to egg batches, typically means that males compele vigorously for females. he termed adaptations arising through this process sexual selection (see also Bateman. 1948; Trivers. 1972; Chapter 8). Males may compete directly (male-male competition) or indirectly, by appealing more effectively 10 females (female choice). Sperm competition can be seen as a form of sexual selection occurring even after mating (Trivers. 1972). Note that the intereSlS of the two sexes need not be coincident. and many situations involve sexual conflict. e.g. concerning whether maling should take place or how much parental care should be given by each panner (Trivers. 1972; Parker, 1979).
go .~
E
0.80
~
:c g
'0
~
:a=
0.40
Anisogamy stable
..e .D
Sperm size: ovum size
Fig. 6.1 Stability of anisogamy in relation to sperm size: ovum size ratio and the probability or double maling. Above the line. anisogamy is slabIe. SO that in systems with a pronounced gamete dimorphism only tiny amOunts of sperm competition will stabilize
anbogamy. Below the line. anisugamy becomes unstable, but Ihis requires less pronounced gamete dimorphism and higher probabililies of double mating. (Calculaled from the exaa equal ion for stability in Parker. 1982.)
SPERM COMPETITION
125
6.2 Why should females copulate with more than one male? The benefits to males of copulating with several females are obvious: they father more offspring. It is much less clear why a female should copulate with more than one male. espedally since in many species a single insemination can fertilize all her ova. In some cases there may be no advantage: males may simply force copulations on females (e.g. some lizards: Olsson. 1995; ducks: McKinney el 01.• 1983). or the costs of resistance may be greater than acquiesence (e.g. some insects). In other cases. however. females appear to aaively seek multiple males. indicating that they obtain some benefit from copula ling with more than one male. The benefils can be either direct or indirea (genetic) and are summarized in Table 6.1. Perhaps the most Widespread explanation is that copulations with different males occur to ensure an adequate supply of sperm (Parker. 1970a). yet there is surprisingly little evidence to supportlhis (see Woodhead. 1985). The potential indirect benefits obtained through multiple matings by females comprise: genetic diversity. altracliveness genes (Fisher'S runaway sexual selection) and good genes (superior genetic quality
Table 6.1 Some of the main benefits to fc::males of mating with multiple males.
Reference
Benefit
DirtCI bmefits Fertility insurance
Insects: several Birds: no convincing eviden e
Thornhill & Alcock (1983) Wenon & Parkin (1991) Birkhead & Fletcher (1995a)
Acquisition of nutrients
lnsecls: many spedes Birds: one example
Thornhill & Alcock (1983) Mills (1994)
Paternal care
Birds: cooperative polyandrous species Mammals: marmosets
Davies (1992); Faaborg (', al.
Avoidance of harrassment
lnseelS: dungfly
Parker (1970a)
Change of long-term panner
Birds: oystercatcher
Ens'.t. (1993)
(1995); Davies Goldizen (1988)
'.1. (1996)
IndirtCf benefilS Genetic diversity
o known examples
Genelic quality
Attractiveness Viability
Birds: several Fish: guppy Reptiles: adder sand lizard Birds: blue tit
Saino' ./. (1996) Reynolds & Gross (1991) Madsen tt al. (1992)
Olsson rt al. (1994) Kempenaers rt a/. (1991)
126
CHAPTER 6
[or survivorship) (see Chapter 8). Genetic diversity is not generally regarded as a convincing explanation [or muhiple mating since copulation with extra males produces very little additional genetic diversity over mating with a single male (Williams, 1975). The acquisition o[ attractiveness genes and good genes (see Chapter 8) is controversial since it is generally believed that traits closely associated with fitness have low heritabilities. [n other words, sexual characters lack additive genetic variance. This creates the 'paradox of the lek' (see Chapter 8): why do [emales chose to copulate with panicular males if their choice does not innuence fecundity? A resolution to this paradox has been provided by Pomiankowski and M011er (1995), who showed that contrary 10 previous ideas, both phenotypic and additive genetic variation are actually relatively high in sexually selected traits and this is because they have been under long-term directional selection. Moreover, the results [rom several sludies of socially monogamous birds appear to suppon the 'good genes' hypothesis since females o[ several species preferentially select males that are more attractive than their partner [or extrapair copulations (e.g. Kempenaers it al., 1992; Saino et al., (996). On the other hand, Sheldon (1994) has pointed out that if morphological or behavioural traits covary with ejaculate features this is also consistent with females acquiring direci fenility benefits. However, in the only test of this hypothesis to date there was no evidence that these fealures covary (Birkhead & Fletcher, ) 995a). In conclusion, where females obtain direct benefits, such as food, sperm or parental care as a consequence of copulating with di(ferent males, the advantage to the female are clear. However, in other cases the precise nature of the indirect benefits females obtain remain 10 be established, and this continues to be an imponant [unctional question.
6.3 Detection and incidence of sperm competition Since sperm competition occurs when a female is inseminated by
twO
or mort:
males, sperm competition can be inferred from either Ihe detection of mixed paternity or from direct observation of copulations. These two methods provide estimates of the incidence of sperm competition, because: (i) not all cases of sperm competition can be detected by paternity analysis (in some cases one male will 'win' and father all offspring); and (ii) not all copulations result in insemination. The most informative studies have been those which have used both methods in combination. This is desirable in field studies, but is often hard 10 achieve, whereas for laboratory studies it is much easier and such studies have been essential for revealing the mechanisms of sperm competition (see Section 6.4.2). It is ironic that Darwin should be among the first 10 document sperm competition since this is a topic he aSSiduously avoided pursuing and hence is absent from his writings - presumably because of social mores. Darwin (1871)
SPERM COMPETITION
127
described how a female domestic goose Anser allSer produced a brood comprising offspring fathered by two different males, her partner (anolher domestic goo e) and a Chinese goose A. CYlloides. Naturalists subsequently observed females occasionally copulating whh more than one male (e.g. praying mantids: Pabre. 1897: ducks: Huxley. 1912). but such behaviour was wrillen off as aberrant. Il was not until Parker (1970a) and Trivers (1972) focused on the evolutionary implications of multiple mating that biologists started 10 think about the adaptive significance of sperm competition. Even before this. however, there had been experimental studies of multiple mating in insects. Many species can be reared easily in the laboratory and paternity was assigned using eitlter genetic markers. such as a difference in colour (e.g. Hunter-Jones, 1960) or tlte irradiated male technique (Parker 1970a). The sperm of males subject to irradiation are capable of fenilizing ova. but have so many lethal mutations lltatthe zygote never develops. Hence. by mating females with a non-irradiated and an irradiatcd male, and recording the number of developing and nondeveloping eggs. respectively. paternity can be assigned. Most insect studies involved reciprocal matings to quantify the effect of mating order on paternity. and most revealed a strong pallern of second (or last) male sperm precedence (see below). Only after a substantial body of work on insects and other invenebrates was complete did biologists start 10 look for evidence of sperm competition in other taxa. Serious observational studies of sperm competition in birds staned in the late 1970s. and somewhat later genetic evidence for multiple paternity within a brood was obtained, albeit with difficulty. using allozymes (e.g. Westneat, 1987). The discovery in 1985 of multilocus DNA fingerprinting which could assign parentage with a high degree of accuracy (Jeffreys el aI' 1985). was an important contributory faclOr in tlte growth of avian paternity studies (Burke & Bruford. 1987; Quinn el al.. 1987; Welton el al.• 1987). The field of sperm competition in birds developed rapidly (Birkhead & M0l1er. 1992) for several reasons: 1 birds are relatively easy 10 observe in nature; 2 small blood samples yield DNA (from nucleated red blood cells) for paternity analyses; 3 it is relatively easy to find nests. capture the putative parents and obtain blood samples from entire families; 4 the fact that the majority of birds are socially monogamous with male parental care raises the question of the relationship between paternity and paternal care. The extent of extrapair paternity in socially monogamous birds is remarkable. ranging from zero in a number o( marine birds (e.g. Hunter el al.• 1992). to over 50% in some passerines (Dixon el al.• 1994; Mulder el aI' 1994). There now exists information on the level of extrapair paternity for over 100 bird species. providing a unique opportunity to explore the factors responsible (or this variation (e.g. M011er & Birkhead. 1994). Evidence for sperm competition
128
CHAPTER 6
in mammals was more difficult 10 oblain panly because most mammals are difficult to observe and rartly because mnlecular ralernity lechniques took longer to develor (Hanken & Sherman. 1981; M'ller & Birkhead, 1989; Pemberton 'I at.. 1992). Palernity assignment using increasingly refined molecular techniques (such as DNA amplification - see Wesmeat & Websler. 1994; SchierwatCf 'I at., 1994), has revolutionized Ihe delection of srerm comretilion, and techniques have been adapted for a range of taxa and there is now evidence for sperm competition in virtually every animal axon (Table 6.2). Although molecular techniques have revolulionized the study of srerm competilion, palernity analyses tell us only about the outcome of sperm compelition -the end result of a succession of behavioural and physiological processes. Behavioural observations are essential for telling us about the way multiple paternity arises. For example, not all cases of extrapair pat emily in birds are Ihe outcome of extrapair corulalion: in some cases mixed paternity may arise from rapid male switching (Pinxten el at.. 1993). Even in those ca'ies where t'xlrapair paternity results from extrapair copulations, it is
important to know whether these are initiated by females. or forced on females Table 6.2 Taxonomic Occurrl'I1Ct= or sperm fOlTIpt'lilio!1 in the animal kingdom. Taxon
Examples
Rdcrenccs
COidarians
Corals·
Levitan & Petersen () 995)
Platyhdmintht's
Flalworms
M(lllu~
Galilrorod Aritlltla SPI'.
Peters' al.( 1996) Barker ( 1994) Chen & Bam (1993)
Annelids Arlhropods
Paluln worm·
Levitan & Peh.'fSt.'n (1995)
( hcliceratt's
Spiders. milt'S
Austad () 984), Radvan Hmckmann fltJI. () 994)
C fUSl(lCeanS
Hnrsl'shoc crah l.imulm-' Ghost crah Inadllls spp.
Insects
Numerous species
Thornhill &- Alcod< (1983)
flillipedes
Dungfly Scatophaga AJloporus 'ndnatus
Barnell etal. (1995)
Fish
Blur·ht.'dded wrasst.'
Shapi',al. (1994)
Amphihians
TIUllossowa' Fmg Clliromamirl<
ASlheiminthes
ematodc..'s
& Siva·Jt>lhy (1996)
Replilt's Birds Mammals
Dit'it.'l (1990) Parker 1970a (St'c It.'XI)
Adda VipertJ bfrus
Halliday & Verrell (1984). Je:'nnions & Passmore:' (1993) Madsen tl al. ( 1992)
Sand li7..u d LArrrftl QHiJis
Olsson ff al. (1994)
Milny
Jl~sscfincs
Birkhead & Moller (1992)
Many specit's
Weslneal & Wt.'hstt'r (1994) Ginshcf)t & Huck () 989)
Humans
Baker& Bellis (19951
SlockJey & PUfvis (1993)
•
Inllicale~ eXIt.'rnal ft:nilizers.
SPERM COMPETITION
129
by males. A combinalion of behavioural observation, paternity analysis and a knowledge of mechanisms provides the best opportunity for understanding the selective forces and constraints resulting in the evolution of particular trailS.
6.4 Patterns of paternity 6.4.1 General Except for spawnings in which several males ejaculale simultaneously (e.g. in some marine invertebrates and some fish), Ihne is a time delay between the ejaculations which compete for a given set of ova. Students of spenn competition have been much concerned with the pallern of paternity in internally fenilizing species - i.e. with the proportion of off pring fertilized by males in relalion to their order of mating. When twO males mate, the proponion subsequemly sired by the second male is usually referred to as the P2 value (Boorman & Parker, 1976) - although P2 is oflen taken simply as the proportion of the progeny gained by the last male to mate. Can we predict any general adaptive rule for P2? This may be dilficult. Parker (1970a) argued that there would both be selection on males both: (i) to oust the sperm of previous males, so as to favour self's spenn; and simultaneously Iii) to prevent self's sperm being ousted by future males. These two adaptations are obviously in connicl, and female interests apart, the P2 attained can be seen as a balance between these two opposing interests. Much will depend on the ease with which each adaptive trend can be achieved. Indeed. P2 can range from very low (i.e. first-male precedencespiders: Austad, 1984; the Adder Vipera berus: Hoggren, 1995), through 'mixing', where competing males gains are roughly ewn (e.g. rats: Dewsbury, 1984; field crickets: Parker el al., 1990), to very high (cenain butternies: Gwynne, 1984; migratory locusts: Parker & Smith, 1975; hirds: Birkhead & M011er, 1992, see below). In mammals a range of precedence and mixing patterns occur because fertilization is determined by the timing of insemination relative to ovulation (Ginsberg & Huck, 1989). In insects, both sorts of adaptation ((i) and (ii) above) occur. For example, the last male to mate may displace (e.g. dungHies: Parker & Simmons, (991) or replace (dragonflies: Waage, 1979) the previous sperm so that P2 is high. But, in some species the first male to mate leaves a plug or a part of the spermatophore within the female tract which may make sperm transfer by subsequenl males dilficult, resulting in lower P2 values than when no obslacle is present (see Parker & Smith, 1975). With strong constraints 30ing against the transfer of sperm after a previous mating, P2 may be low. Bu!. if there is little constraint acting against displacement of previous sperm, we may expect a high P2, which is often the case, especially in insects and birds. P2 may be high for purely mechanistic reasons. If ejaculales comain constant sperm numbers, then a high P2 can simply be the result of passive loss of the
130
CHAPTER 6
first ejaculate before the second occurs (Lessells & Birkhead, 1990). It would al>o occur if there is little sperm mixing. and if sperm from the last male to mate is necessarily deposited closer to the site of fertilization (Smith, 1979). There is now growing evidence that ejaculates vary adaptively depending on the risk or intensity of sperm competition (e.g. Gage, 1991): a high P2 could then be the result of an adaptive escalation in sperm numbers (Cook & Gage, 1995). Female interesls may also be important in delennining paternily. If a difference in 'genetic quality' between IWO males can be detected. selection may favour females exerting a preference for the sperm of the best male and aflect the pattern of paternity (see Chapter 5, Section 5.2). For example, if females adopt a strategy of remating only if the male is of higher quality than the one she has previously mated with. then preference for the best male would be manifesl as last male sperm precedence. 6.4.2 Models and mechanisms of last male precedence
The processes underlying ferrilizalion are typically complex and take place at a microscopic level. so it is difficult 10 observe what is happening direclly. An allemalive approach is 10 construct mathematical models which assume a parricular mechanism of sperm competition, and predict outcomes in terms of paternity which, as outlined above, can readily be measured. We can thus allemptlO deduce how ejaculates compete by comparing the observed paternity wit h that predicted under different mechanism assumptions. This can then form the basis of more delailed empirical investigation.
InSi!cr models Parker etal. (1990) proposed a series of linear mathematical models for analysing sperm competition data in such a way as to be able 10 predictlhe mechanism of sperm competition. If Ihe relative number of spenn transferred by each of two males is known, then P2 can be predicted under the assumptions of: 1 Ihe 'fair raffle', in which sperm from the two males mix randomly to generate the 'fertilizalion set' (the pool of sperm from which the ferrilizing speml are drawn); 2 the 'loaded raffle', in which a proporrion of one or other male's sperm are discounted (see also Sakaluk & Eggen, 1996) before the fertilization set is generated, as may be the case where there is a competitive race, or passive sperm loss before the second mating; 3 sperm displacement with instant mixing (see also Parker & Simmons, 1991), in which sperm from Ihe fim male are displaced by incoming sperm with which Ihey swiftly mix randomly to form the ferrilization sel; 4 'penn displacement with mixing after displacement, in which there is random mixing only after sperm from the first male are displaced by incoming sperm.
SPERM COMPETITION
131
Parker el al. (1990) demonstrated the use of such techniques by analysing data from two insect species. In the yellow dungny, Scalophaga slercoraria, the male delivers free sperm during copula. The female's sperm stores, the spermathecae, are chitinous structures of fixed volume, and the P2 data fit well with the model of sperm displacement with random instant mixing (see below). In the field cricket, Gryllus bimaculatus, sperm transfer does not occur during copulalion itsell, which serves to place an external spermatophore, from which the sperm later drain. The single spermatheca is an elastic membranous sac, and sperm from successive copulations simply add to its volume (Simmons, 1986). P2 concurs well with the fair raflle or random mixing (Parker & Simmuns, 1991 j. The predictions of the sperm displacement model have been tested most extensively on the yellow dungfly, a ubiquitous inhabitant of canle pastures where male flies assemble around fresh droppings to search for and mate with the females as they arrive to lay their egg batch in the dropping. Virtually all incoming females contain enough sperm to fertilize their clutch, but nevertheless mate belore ovipositing. By remating they ensure that they are guarded until the end of oviposition by the male that has just mated - a feature which greatly reduces harassment from searching males. The guarding behaviour is likely to have arisen as a paternity assurance mechanism: since the last male to mate fertilizes over 80% of the current egg batch, so the potential mating losses due to some 16-min guarding is well repaid by the protection of paternity (Parker, 1970b), However, guarding is not always effective - if an attacking male is sulliciently large a take-over and copulation may ensue (Sigurjonsdullir & Parker, 1981), Sperm competition in dungflies seems to fit a model of constant random displacement with instant spenn mixing, To explain, imagine a tank of sperm, representing the fenilization sel, which has an input pipe and an outlet pipe, During copulation, sperm flow at a constant rate into the tank through the inlet and out (by displacement) through the outlet. First imagine that the sperm entering the tank do not mix with the sperm already present, which are pushed towards and out from the outlet. The new sperm displace only the old sperm, so that the proportion of sperm from the last male (and hence P2) rises linearly al a rate equal to the input rate (volume entering per unit time) divided by the total volume of the tank, But, now suppose that there is swilt random mixing of the incoming speml with the previous sperm in the tank. At first the sperm displaced from the outlet will be only the old sperm. As the last male's sperm build up in the tank, some of the displaced sperm will be his own (',elfdisplacement'), By the time most of the sperm in the tank is new, most of the olllflow will represent sell-displacement. In fact, the proportion of sperm from the last male in the tank (and hence P2) will increase wilh exponentially diminishing returns (Parker et ai' t990; Parker & Simmons, 1991), P2 in dungllies lits such a relationship quite closely (Parker & Stuart, 1976; Parker & Simmons, 1991, 1994); other mechanisms may generate similar results, but this is the
132
CHAPTER 6
simplest explanation of Ihe dungfly data. What is not yet cenain is how this analogue fils in with the analOmy of Ihe female Iract. It seems rather unlikely thai the tank is the spermathecae, because sperm are deposited in the bursa and sperm reach Ihe spennathecae via long thin ducts in which simultaneous innow and outflow currents are difficult 10 envisage, especially for a viscous fluid. A knowledge of the probable mechanism of sperm competition in dungflies has helped 10 make and lest predictions aboul adaplalion. II has long been known Ihat Ihe copula duralion of male dungflies is close to that predicted from an optimality model. The approach used is the marginal value theorem (Charnov, 1976; Parker & Stuart, 1976). The gains to a male (P2) are ploued against the lime a male spends rna ling and displacing rival sperm (Fig. 6.2a). The average time it takes a male 10 find and guard a female (the search time) is known from field work (Parker, 1970c). The oplimal copula duration is given by the tangent 10 the P2 curve drawn from a distance on the x-axis equal 10 Ihe search time; Ihe slope of this tangent line gives Ihe maximum rale of gain in fitness anainable from the system as a whole. The predicted copula duralion is quite close 10 Ihe observed average duralion (Fig. 6.2a; see Parker & Sluart, 1976; Parker, 1978a), and either feeding-time costs of replenishing Ihe sperm (Parker, 1992), or the effeas of take-overs on reducing search time (Charnov & Parker. 1995) could account for Ihe discrepancy between the observed and predicted values. Rather Ihan estimate the average solution. the most recent work (Parker & Simmons, 1994) has investigated how males of differing sizes should behave. Copula duration decreases with Ihe size of the male (Fig. 6.2b). Why? Firsl. il is nOI surprising 10 find thai the bigger the male. Ihe fasler Ihe rale of flow of spenn during copulation and Ihe higher the displacement rate (Fig. 6.2c). Bigger males show P2 curves which rise more steeply, and we would expealhallhe langentline would give shorter optimal copula durations for large males. BUI, thtTe is a second reason why large males should show shorter copula durations: Ihey experience shoner search limes because they gain more lake-overs 01 ovipositing females (Fig. 6.2d). These two e!feas (higher displacement rale. Fig. 6.2
(opposit~)
Optimal copula duration in S£acophaga. (a) Marginal value approach. The J - exp(-
curve is calculated from P2 =
estimate of the displacement rate (Parker & Simmons, 1991). The predicted optimal copula duration (p) is approximately 42 min. while the observed duration (0) is 35.5 min (see also Parker & Stuart, 1976; Parker, 1978a). (b) Observed copula duration of males in relation 10
their size (measured as the cube of the hind tibia length). Dotted lines are 95% confidence intervals. The duration decreases with male size. (From Parker & Simmons. 1994.) (c) Sperm displacement rate of males in relation to size. estimated from P2 data. (From Parker & SImmons. 1994.) The di placement rate increases with male size. (d) The relative chance of gaining a fe'malc by a take-over (upen circles) increases with male size and su the search and guard time (filled circles) between successive m3tings is reduced as males get higger (calculated from various field data; see Parker & Simmons, 1994). (e~ The effect of displacement rate (line I) and take-over Wne 2) on the predictt'd optimal copula duration. compared with the observed relationship (dOlled line ± 95% confidence intervals).
SPERM COMPETITION Ibl 200 180
lall.0
:g c
0
140
.2 120 ~ 100
B
,
~ 0.5
~
0
C 160
'3 ~
80
'-
60
0
u
0 ...............c ...~.........
40
20 0
0.0.'15'0--.-=-100=---.=c:----'O:--t+::----:-: 100 50 50 p Search and guard time (min)
133
5
10
15
20
40
45
50
o
Copula duration (mini
lei 0.6
1.2
(d)
200
o
0.5
1.0
C
~
~ 0.4
150
o
C
}
'E
0.3
o
~
'E
~
~
u
0.6
~
0.2
~
100
~
a::
0.4
10
20
30
40
50
60
0.0
70
0
40
20
In 100
(0)100 80
,
C
:g
..... ~~::~::::~::: ..•.._.. ....... .......
..,
80
c
.2
......
....... ........
-.. -......
2
'3 ~
'-
<3
0 80
60
Male size
Male size
~
~
Jl
0.2
.0.1 0
~ 40 '3
'c' ~
50
o
60 -
~
~
V>
.~
,
~
0;
it o. I
:g .§
0.8
u~ c
~
~
55
40
'0
u
-'-'
20
10
15
20
25
30
35
Male size
40
45
50
55
20 0
5
10
15
20
25
30
35
40
45
Male size
Fig. 6.2 (continutdj (From Parker & Simmons. 1994.) In thiS figure the effects of displacement rate and take-uver are trealed separatt'1y. whereas in (f) they are combined. (£) The effect of combining Ihe two size·dependent eHects (displacement rate and take-overs, to predict rhe optimal copula duration (solid line ± 95% confidence intervals). compared with the observed relationship. (From Parker &- Simmons, J 994.)
50
55
134
CHAPTER 6
shorter search time) can be evaluated independently (Fig. 6.2e): optimal copula duration is much more sensitive to the variation in displacement rate. When both effects are included, the fit is a good one across most of the male size range, although the smallest 10% of.males should copulate longer (Fig. 6.2f). Interestingly, it turns out that (despile larger males achieving higher gain rates in fitness) the two effects interact so that the P2 gain per mating is approximately equal for all males. Suppose there is a phenotypic characteristic that correlates with foraging efficiency (for dungflies this is male size). If uptake rate is directly proportional 10 the foraging efficiency characteristic, and search time is inversely proportional to it, then we get the rule that the cumulative gain from the patch should, at the optimum, be equal for all phenotypes (Chamov & Parker, 1995). Thus, in dungnies P2 is found 10 be constant and independent of male size as the rule predicts. Bird models
Although there are many fewer studies of birds than insects, high P2 values appear 10 be the rule in birds (Birkhead & M0l1er, 1992). One of the most influential studies, bOlh in terms of the magnitude of the last male effect and the mechanism by which it occurred, has been that of CompIOn el al. (1978). They used genetic markers to assign paternity in domestic fowl and artificially inseminated equal numbers of sperm from two different genotypes 4-h apart. Regardless of the order of inseminations the second male fertilized most eggs (P2 =0.77) and they suggested that this occurred because sperm from successive inseminations remained stratified in the blind-ending sperm slOrage tubules (Fig. 6.3), so that a last in-first out system operated. In an earlier study Martin et al. (1974) found that when domestic fowl were inseminated once with a ntixture of semen from two different genotypes, paternity was directly proportional 10 the relative number of sperm from each genotype. On the basis of these two studies the outcome of sperm competition was thought 10 be determined by the interval between successive inseminations: when the interval was less than 4 h, sperm mixed before going into the sperm slOrage tubules and paternity was proportional to the relative number of sperm from each male, but when the interval was greater than 4 h, stratification of ejaculates within the sperm storage tubules resulted in last male sperm precedence (Cheng el al., 1983; McKinney el al., 1984). Although this scenario seemed reasonable, a number of inconsistencies subsequently emerged, e.g. if stratification accounted for lasl male sperm precedence, as sperm from the second male were utilized, those of the first male should be uncovered and result in an increase his paternity. However, in Compton et al.'s (1978) experiment the relative success of the two inseminations remained const.ant over time (see Birkhead
& M0l1er, 1992).
Lessells and Birkhead (1990) constructed a series of mathematical models to find a plausible mechanism which would account for the P2 value of 0.77
SPERM COMPETITION
135
Fig. 6.3 One of 1500 sperm storage tubules from a female zebra finch. Several spermalOwa can be St'cn in the lumen or the tubule. The scale bar = 50 ~m.
reponed by Compton el at. (1978) in Ihe domeslic fowl. They considered three main models (Fig. 6.4). 1 SLralificalion in which separate ejaculates remain layered within the sperm storage Lubules. This model failed because: (a) the observed rale of spenn loss from the female IraCI (see Wishan, 1987) was 100 high 10 maintain Ihe long-Ienn levels of spenn precedence observed; (b) Ihe model predicts a decrease in lasl male precedence as sperm from the firsl inseminalion are 'uncovered', bUI Ihe dala showed Ihat this did nOL occur and Ihe ratio of offspring remained conslant over lime. 2 Passive spenn loss in which sperm are losl from Ihe female Iract al a constant rale and the second (or lasl) male's sperm have precedence simply because fewer of Ihe second male's sperm are lost by the lime fertilizalion occurs. This is similar to Parker it at.'s (1990) 'loaded rarne' (above). This model also failed: to achieve P2 = 0.77, Ihe rate of sperm loss from Ihe female Iracl would have 10 be much higher than recorded. 3 Sperm displacement in which it is assumed Ihal space for stored sperm in the female tract is limited and that incoming sperm displace those already present. This model is similar 10 Parker it at.'s (1990) 'displacement with mixing after displacement' (above), and was able to account for Complon et at.'s (J 978) results.
'st
(a)
2nd
$)
..
i.)
__ ._._. __
1.0
~)
.4.~
.~
,
E 0.5 ~ m
,
,
a.
--------Time
1st
(b)
§.)
'.0
2nd
1-~ ~~-:.?
' d~
... E
1b
'
(/) C
'.
Time
#--------------------------------
~
'E ~
0.5
m
a.
...
... 'st
lei
Time
2nd
J-_:_-:~ - - -1.0
~
.~
~
0.5
m
a.
Time
Fig. 6.4 Sperm competition in birds: three models of mechanisms that would result in last male sperm precedence. In each case there are two inseminations (black triangles) made at different times. At the top of each ligure the position of sperm in a typical sperm storage tubule (SST) is shown after the first and second insemination. Lines indicate the proportion of offspring fathered by the first (solid line) and second (dashed line) inseminations. (a) Stratification. The sperm of a second insemination overlie those of the first and being closest to (he SST exit the second insemination fertilizes most eggs. Once this sperm is used up, however. the firsl male starts to fertilize eggs again. (b) Passive sperm loss. By the time the second insemination occurs some sperm from the first have been losl. The longer the interval between inseminations the greater the last male effect. Once sperm are in the SSTs they mix, so the ratio of offspring remains constant over lime. The inset shows the Jog number of sperm in the female tract following two inseminations (arrows); at any point in time after the second insemination there are more sperm from the second male. (c) Displacement from a limited sperm store. The second insemination displaces the sperm stored in the SSTs and Ihe level of P2 depends on the degree of displacement. (Based on Lessells & Birkhead. 1990.)
SPERM COMPETITION
137
However, when Birkhead el al. (1995a) repeated Compton el al:s (1978) experiment, they found no marked last male precedence with a 4-h interval between inseminations. Instead, their results were more consistent with the passive sperm loss model. The discrepancy occurred because Compton et al.'s first inseminations lOok place close to the time of egg-laying when the uptake of sperm by the sperm slOrage tubules is greatly reduced (Brillard el al., 1987), thereby giving the second insemination a greater fertilization success - hence the high P2 value (Birkhead el al., 1995a). This also: (i) accounts for why Lessells and Birkhead (1990) were unable to explain Compton el al. 's (1978) results in terms of the passive sperm loss model; and (ii) shows that the 'success' of the displacement model was fortuitous. Indeed, empirical observations of the numbers of sperm in storage tubules now show that the displacement model's assumption that storage space is limited is biologically implausible in most situations. The predictions of the passive sperm loss model have been tested in the zebra finch. This is a small (12-15 g), socially monogamous, colonial passerine living in the more arid parts of Australia. In the wild, extrapair courtship and copulation attempts are frequent, although extrapair paternity is relatively low: 2.4% (two out of 82 offspring, from 25 families: Birkhead el al.. 1988, 1990). Males attempt to protect their paternity by following their partner during the period before and during egg-laying, and to a lesser extent by copulation, which occurs about 12 times for each clutch. Using genetic plumage markers to assign paternity Birkhead el al. (1988) conducted two sperm competition experiments in the laboratory. I Mate switching experiment: the female copulated with two different males in succession and both males obtained a similar number of copulations; the second male fathered most offspring, P2 = 0.75 (95% confidence limits: 0.65-{).83). 2 Single extrapair copulation experiment: the aim was to test simultaneously the efficacy of an extrapair copulation, and the magnitude of the last male effect seen in the first experiment. Females copulated nine times on average with one male over several days and then finally once with a male of the other genotype. Overall, P2 =0.54 (95% confidence limits: 41.6-66.1). Thus, the extrapair copulation fertilized significantly more eggs than that expected from the 9 : I ratio of copulations, demonstrating both a strong last male effect and that single extrapair copulations can be disproportionately successful (Birkhead el al., 1988). To ascertain whether the passive sperm loss model (Fig. 6.4b) could account for these two sets of results required information on the relative numbers of sperm from each male in the female tract at the time of fertilization. This is determined by three factors: (i) the numbers of sperm inseminated by each male; (ii) the timing of inseminations; and (iii) the rate of loss of sperm from the fcmale tract. I Ejaculate size was determined by persuading males to copulate with a freezedried female fitted with a false cloaca. This revealed that for males which had
138
CHAPTER 6
not copulated in the previous 7 days the initial ejaculate was relatively large (7.8 x 10· sperm). but subsequent ejaculates were mu h smaller (1.7 x 10· sperm) (Birkhead 'I al.. 1995b). 2 The timing of inseminations was determined directly by observation and continuous video recording. 3 The instantaneous per capita rate of loss of sperm from the female tract (0.026 sperm h- I ± 0.007 SE) was dctermined by counting the number of sperm trapped on the outer perivitelline layer of the yolk of successively laid cggs alter copulations had ceased (Birkhead 'I al.. 1993). The levels of paternity predicted by the passive sperm loss model using lhese variables were very similar to those observed (Fig. 6.5). The main reason the single extrapair copulation was so successful was a combination of it being last (since much of the previously inseminated sperm had already been lost from the female tract) and because it contained a relatively large number of sperm. This in turn was a consequence of the way male zebra finches produce and store sperm in their reproductive tract and the fact that the time since last ejaculations is important in determining the number of sperm ejaculated (Birkhead 'I al.. 1995b). In the wild. female zebra finches appear to be particularly choosy about with whom they perform extrapair copulations and thus extrapair offspring are rare. Mate choice experiments in captivity have shown that female zebra finches prefer males with certain trailS (e.g. high song rates: Collins 'I al.• 1994). Since they do not obtain direct fertility benefits from extrapair copulations with such males (Birkhead & Fletcher, 1995a), this indicates. by default, that they obtain indirect benefits, although there is no direct evidence for this as yet (Birkhead, 1996).
Mate switch
100
Observed
Model
Single EPC Observed
Model
-
Fig. 6.5 Sperm competition in Ihe zebra finch. Observed P2 values in two experiments and predicted P2 values derived from the passiv~ sperm loss model. In the mate-switch experiment each male copulated several times with the female. In the single extrapair copulation experimenl the first male obtained nine copulations. followed by a single
copulation from an 'extrapair male'. In both experiments the observed values are similar to those predicted by the passive sperm loss model. {Redrawn (rom data in Colegrave et a/.. 1995.)
SPERM COM PETITIO N
119
6.5 Effects of mechanisms on behaviour 6.5.1 Sperm expenditure theory Sperm expenditure is the amount 01 reproduaive effort spent by a male on a given ejaculate. A series of 'sperm competition games' has been devised to predict patterns of sperm expenditure in relation to the intensity of sperm competition and the information available to a male at the time 01 mating (e.g. Parker, 1990a,b). The models assume that ejaculates are expensive (Dewsbury, (982) in the sense that ejaculate expenditure trades off against mate-searching expenditure (Parker, 1982). Increased ejaculate effort increa,es the gains /rom a given mating if there is sperm competition, but reduces the number of matings that can be achieved. The models also Iypically assume that sperm competilion follows some form of raffle, so that a given male's gains depend on the strategy played by his competitors. The predictions depend on the conditions. Suppose that sperm competition is rather rare so that most females mate just once for a given set of eggs, and the rest (proportion p) mate twice. If a male has no information about p when he mates with a female, the ESS sperm expenditure (for low PI, as a proportion of the total reproductive effort per mating, is roughly p/4 (Parker, 1982, 1990a); sperm expenditure should increase with the probability of sperm competition. The observation that, viewed across spedes, sperm expenditure should increase with the probability or imensity of sperm competition is a universal finding of such models. This prediction is supported by various data, although direct evidence (sperm numbers per ejaculate increase as sperm competition increalies across species) is much rarer than the indirect result that across species relative
testis size increases with sperm cumpetition risk. (The first suggestion for such a correlation was made by Short (1977) for the great apes and has subsequently been shown to occur in several other taxa - see Birkhead, 1995.) But, now suppose that a male has information that he will face sperm competition from the ejaculate of another male and, further, suppose that mating first or second carries a relative disadvantage if equal ejaculates are delivered (the 'loaded rarne'). Thus, one 'role' (first or second to mate) is favoured, and each male 'knows' which role he occupies. What is the ESS sperm npendilure pattern? If the roles are occupied randomly so that a given male is equally likely to be first or second to mate, then sperm expenditures should be equal (Parker, 1990a). But, if given males are more likely to occupy given roles, then the male in the disfavoured role should spend more on the ejacula Ie. Cases resembling the loaded raffle may occur in nature, and Stockley and Purvis (1993) have found support for its predictions in a comparat ive study across mammal species. In the t3-lined ground squirrel. SpermophiJus tridecemJineatus, males (typically two) 'queue' for mating with an oestrous female (Schwagmeyer& Parker, 1987). There is a firsl-male advantage (Schwagmeyer & Parker, 1990): sperm competition may resemble a raffle loaded in favour o[
140
C HAP T E R 6
the first male. Although sperm numbers were not measured directly (which would have required killing the females), all behavioural measures that might correlate with sperm numbers (e.g. numbers of mounts, copulatory attempts, time of longest copulation, etc.) failed to show any differences between first and second males. The circumstances appear consistent with the loaded raffle with random roles (Schwagmeyer & Parker, 1994). Similar sperm competition games were analysed for monogamy with extrapair copulations, and for 'sneak-guarder' situations (Parker, 1990b). In the extrapair copulations model, each male copulates with his mate, and has a low probability p of achieving an extrapair copulation with somebody else's mate. The ESS is to transfer much more sperm at an extrapair copulation (which faces certainty of sperm competition) than to one's mate (which faces sperm competition only with probability pl. Indeed, bird studies show that high paternity sometimes accrues to extrapair copulation matings (Birkhead et al., 1988) and that sperm numbers in extrapair copulations are likely to be higher (Birkhead et ai., 1995b). In the 'sneak-guarder' version, each male is either a 'sneak' (which achieves a mating with low probability p), or a 'guarder' which always mates. The same finding applies: sneaks should expend more on sperm. Evidence is found in fish, where the typically smaller sneaks produce relatively and sometimes even absolutely greater ejaculates than the guarding males (e.g. Shapiro et al., 1994; Gage et al., 1996). Most recently, sperm competition games have attempted to make predictions about the size of sperm under internal (Parker, 1993) or external fertilization (Ball & Parker, 1996), although why a few species should produce sperm with giant tails (such as the 58-mm longtail in the 2-3-mm male Drosophila bifurca) (Pitnick et al., 1995) remains a mystery. Work is currently in progress to explain sperm polymorphisms such as those found in lepidopterans: the apyrene sperm are typically smaller and much more numerous than the eupyrene sperm which are transferred in the same ejaculate. Only the eupyrene sperm contains the DNA to fuse with the female pronucleu . Most birds, many mammals and a few insects show a pattern of multiple ejaculation into a given female, rather than transferring sperm as a single ejaculate. An early sperm competition game (Parker. 1984) investigated two strategies: S (single ejaculation, in which a single large dose of sperm are transferred at the start of oestrus), and M (multiple ejaculation, in which I/nth the dose of sperm are transferred at n equal intervals throughout oestrus). Conception occurs randomly throughout oe trus, and sperm competition follows the passive sperm loss model: a con tant proportion of each ejaculate is lost at each time unit. S always outcompetes M unless the rate of sperm loss is very high; M wins only if more than (roughly) 90% of sperm ejaculated at the start of oestrus would be lost by the end of oestrus.
SPERM COMPETITION
141
6.5.2 Female behaviour The traditional view of sperm competition has been one of sexually enthusiastic males and passive or reluctant females, mainly because selection was thought to operate more intensively on males than females (Parker, 1970a, 1984). Recent empirical observations, however. especially on birds, reveal that females sometimes actively seek additional copulation partners; this has resulted in a recent reappraisal of female roles in sperm competition (e.g. Hunter 'I al.. 1993). In contrast to the situation where females acquire direct benefits from mating with multiple males, where indirect benefits are concerned, kmales seek additional partners specifically in order to have their ova fertiJized by them. There are two ways that females can control the paternity of their offspring: either overtly through their behaviOur, or covertly through physiological means, a process referred to as cryptic female choice (Thornhill. 1983). Intuitively, behavioural methods appear to offer the most direct way for females to control paternity. However. it may not be this simple: a female's priority must be to ensure that her eggs are fertilized. Her best strategy might therefore be to copulate with one male and then if she encounters a more preferred male, to subsequently copulate with him. fn this situation females would benefit from being able to favour the sperm of the second (or last) male. In birds the last male to inseminate a female has the best chance of fertilizing her eggs and the longer the interval between the previous pair copulation and the extrapair copulation, the greater are the chances (above). This may explain why females of many bird species terminate copulations with their partner soon after egglaying has started but before they cease to be fertile. By doing this females get the best of all possible worlds: they have sufficient sperm from their partner to fertilize the clutch. but retain the option of engaging in an extrapair copulation should they encounter an appropriate male (Birkhead & M01ler, J 993). The females of those species subject to forced copulations (abow) would benefit from some physiological control over which sperm fertilize her ova. Although the idea of cryptic female choice is inherently appealing, and there arc some intriguing possibilities. there is as yet very little convincing evidence
for it (see Eberhard. 1996; Simmons 'I al., 1996).
6.5.3 Males seeking extra pair copulations Under a range of sperm competition models (abow) the more spern, a male inseminates the greater are his chances of fertilizing ova. In some species males have the physiological ability to adjust the number of sperm ejaculated and ejaculate relatively large numbers of sperm in sperm competition situations (e.g. Mediterranean fruit fly Ceralilis capilala: Gage, 199t: and the reef fish Thalassoma bi[asc;atum: Shapiro 'I al., 1994; Warner 'I al.. 1995). However, there is no evidence that male zebra finches can adjust the number of sperm they transfer (Birkhead & Fletcher, I995b), instead they maximize
142
CHAPTER 6
the number of sperm they transfer through the timing of extra pair activities and seeking extrapair copulations once their own pair-copulation period is over and their sperm supplies have recovered and are maximal (Birkhead et aI., 1995b),
6.5.4 Paternity guards Males exhibit a range of morphological and behavioural traits which minimize the likelihood of their female being inseminated by another male (Parker. 1970a), The most obvious of these traits is mate-guarding, in which the male remains either in physical contact or in close proximity to the female, Guarding may be pre- or post-copulatory depending on whether females store sperm and the interval between copulation and fertilization, Contact guarding occurs in the amphipod Gammarus pulex, in which the male carries the female for several days prior to copulation. Females do not store sperm so the first male to copulate fertilizes all eggs. hence pre-copulatory guarding (Birkhead & Pringle. 1986). Contact guarding also occurs in the dungfiy (Parker, 1970b). Here. although the insemination-fertilization interval is short. males can displace previously stored sperm (see Section 6.4.2), so guarding is postcopulatory. Non-contact guarding also occurs in some insects, such as cenain dragonnies (Jacobs, 1955). In birds, guarding is non-contact. and both preand post-copulatory: the male follows his partner from several days before she first ovulates until all her eggs are fertilized. This is necessary because females store sperm and because each egg of the clutch is fertilized separately. usually the day before it is laid (Birkhead & M0l1er. 1992). Male mammals also guard oestrous females, a behaviour usually referred to as consorlship (Packer & Pusey. 1983; Sherman. 1989). Other paternity guards include the deposition of mating plugs (e.g. in some nematodes, arachnids. insects. snakes and mammals), antiaphrodisiacs within the seminal fluid (Reimann et al.. 1967) and frequent or prolonged copulation (Birkhead & M0l1er, 1992).
6.6 Sexual conflict and mating systems 6.6.1 Sexual conflicts and their outcomes
Sexual conflict occurs if the evolutionary interests of males and females do not coincide. Conflicts of this sort can apply in various contexts, the most obvious being parental investment (it may pay one sex to reduce parental investment) or mating decisions. We discuss the latter here. and only briefly. If a male and female meet, it could pay both to male, or neither to mate. There is then no sexual conflict. But. if it pays one sex to mate and the other not to mate. we can say that there will be sexual conflict over mating decision (Parker. 1979, 1983). In mating conflicts, because of the disparity in parenlal inveslment. males usually occupy the role in which mating is favourable and females the
SPERM COMPETITION
143
role in which it is unlavourable, although occasionally roles may be reversed. Con[ia can occur over mate quality, incest, matings between ecotypes or sihling species, etc. Parker (1979) identified thresholds at which the three possible outcomes would occur -the range 01 conditions lor conllict is likely to be greatest when the male makes no parental contribution to ollspring. What about the outcomes of mating conflicts? Clutton-Brock and Parker (1995) review game-theory models that have been applied: the general determinants of outcomes in all the games depend on the balance (see Parker, 1979) between what may be termed 'power' (which concerns the relative contest costs, the ability to inflict damage, or the relative cost of enlorcing viaory for the sexes) and the 'value 01 winning' (which concerns the relative fitness difference between mating and not mating lor the sexes). Any asymmetry in the value 01 winning is usually in lavour 01 the male, because the difference between mating and not rna ling represents n offspring, rather than just the variation in the quality 01 the 11 olfspring. But, asymmetry in power may be in favour 01 the female - it may be easier to prevent a mating than to enforce it, unless males are much bigger. Thus, general outcomes of mating conflicts are not easy to predict. One possible way that females could avoid costs is simply to allow mating, but prevent sperm lrom fertilizing eggs. 6,6,2 Sexual conflict: empirical observations
The traditional view of mating systems is that 'what you see is what you get'; in other words, if a particular bird species is socially monogamous, it was also sexually monogamous and often had biparental care. But, this seems () be true in only a small proportion of cases. In mosr seabirds, for example, there is no conflict over copulation or parental care because the interests of the sexes are similar and partners depend upon each other to rear a single olfspring. Females of these species rarely seek or engage in extrapair copulations and extrapair paternity is low (Birkhead & M0l1er, 1996). In other socially monogamous bird species the interests of each sex differ and paternity analyses show that both sexes routinely copulate with and are lertilized by other males (Birkhead & M0lJer, 1992). A conflict thus exists over parental care; being less certain of their parent.age than females, a male may be less willing to invest. especially if their female has copulated with another male. In some species infidelity is costly for females because in these circumstances males reduce their care, but in others females succeed in securing their partner's full assistance in rearing ollspring lathered by other males (Westneat & Sargent, 1996). All mating systems may be the outcome 01 intra- and intersexual conflicts (Davies, 1992; Reynolds, 1996) and the resolution of conflicts olten represent a compromise, with neither of the participants achieving their preferred optimom. In socially monogamous birds, lor example, this occurs because male-guarding constrains females and prevents them from timing extrapair copulations optimally and hence reducing their effectiveness (Birkhead, 1996). This in
144
CHAPTER 6
turn must also account for the considerable intraspecific variability in the outcome of exuapair copulations (e.g. Birkhead & Moller, 1992). The conllias and dynamic interplay between male and female decisions which result in a compromise can mosl clearly be seen in Ihe polygynandrous mating system of the dunnock and alpine accentor Prunella collaris. In both species females oblain direct benefils in the form of male care from securing a threshold number of copulations with particular males. and to this end females solicit copulations from all males in Ihe group. Dominant males. on the other hand, auempt to monopolize females to maximize their own reproduclive success. The outcome of Ihis conflict is relleaed in the high rates of copulation solicitation by females, the refusal of many solicitalions by males. but also Ihe hi~h rates of copulation, which for some females may number over 1000 per clutch. High copulation rates in turn are associated with relalively enormous testes and seminal glomera (speml slores) in the males of both species (Davies et 'I., 1996). In some cases one sex may gain the upper hand.ln the externally fertiliZing blue-headed wrasse Ihe most sexually successful males produce fewer sperm per mating. and hence fertilize a smaller proportion of each female's ova (93%) than less aC1ive males (96%). In this way males increase their reproduC1ive OUlput, but by choosing these panicular males females suffer the cost of having fewer of their eggs fenilized (Warner et 01.. 1995). In Drosophila there is last male sperm precedence (P2 > 90%) and males benefit from repealed mating with dinerent females because it maximizes the number of offspring they [ather (Bateman. 1948; Gromko el 01.. 1984). For females, however, mating is costly; the more females copulate the shoner their lives. This occurs because, prior lO introducing sperm during inseminalion, Ihe male transfers a cocktail of other substances in Iheir seminal fluid designed lO maximize his reproduaive success. These substances speed up the rate of oviposition. decrease the female's receptivity lO other males and, perhaps most significantly, disable any previously slOred sperm in her tract. It is thought that the proteins which disable slOred sperm damage the female and significantly reduce her lifespan (Fowler & Partridge, 1989; Harshman & Prout, 1994; Chapman et 01., 1995; Clark et 01., 1995).
6.7 Conclusion A number of fundamental queslions in the field of sperm competition remain 10 be answered. For example, we still need lo determine the funC1ion of multiple mating by females. We also need to know a great deal more aboul Ihe underlying mechanisms of sperm competition. and in this respeC1 the continued interaction between empirical sludies and theoretical models will increase our understanding of sperm competition adaptations. Empirical investigations of sperm quality and the use of labels to identify the sperm of panicular males promise 10 be panicularly revealing. The continued integration of mechanism
SPERM COMPETITION
145
and function is imporlanr since it will allow us to make better predictions aboUl the optimal behaviours of males and females. The field of sperm competition is a broad one spanning behaviour. physiology anatomy. molecular biology and is intimately associated with fitness. Sperm competition is such a ubiquitous and pervasive force in evolution thal it will almost certainly continue to reveal exciting insights into sexual behaviour and gamete biology.
PART 3 FROM INDIVIDUAL BEHAVIOUR TO SOCIAL SYSTEMS
Male lions Panthera leo cooptratt to take owr and defend prides, yet thert is ofttn a high degru of skt'W in their individual uproduaivt surass. How. then. can cooperativt behaviour bt stable? IPhotograph by Craig Packer. (
Part 3: Introduction
Most animals spend time interacting with others of their own species. They attract mates, fight rivals. group together for safety. look after offspring. and so on. How can sodal groups (male-female pairs. families or colonies) be stable despite selection on individuals to maximize their own fitness. often at the expense of others in the group? The chapters in this section are all concerned with the mixlUre of cooperation and conflict seen in animal societies. In Chapter 7. Johnstone explores the evolution of signals used in communication. How should we expect individuals to advertise their quality to mates or rivals? How should prey signal that they have seen a predator? How should offspring advertise their hunger to their parents? To understand signal evolution we need to consider selective pressures on both signallers and receivers and to recognize that there will often be a conflict of interest between them. There will be selection on signallers to make their signals detectable and stimulating to receivers. On the other hand, it will pay receivers to extract reliable information from the signals. The problem is that because individuals usually have conflicting genetic interests it will potentially pay signallers to misrepresent their true quality. Johnstone argues that the outcome of this conflict is the evolution of costly signals which enforce honesty. an idea originally proposed by Zahavi (1975). Honesty can be maintained by physical constraints (e.g. the carotenoid pigment used in displays by some birds and fish can be obtained only from food) or because inferior competitors find the signal more costly to produce or mainta;n (e.g. only good quality male swallows can bear the handicap of a long tail). The chapter reviews the theory and evidence for this conclusion and suggests two interesting problems for fUlUre work. Why do animals often use multiple signals and how does competition among signallers influence signal form and receiver response? The most elaborate signals are concerned with mate choice. as exemplified by the tail of the peacock or the extraordinary dances of some birds of paradise. Darwin (1871) suggested these evolved through female choice. although he had lillie evidence for choice and did not explain why females might have such strange preferences. The last decade has seen an explosion of interest in these questions and Ryan reviews the findings in Chapler 8. Sometimes female choice is easy to understand because the female gains beller resources which improve her fecundity. for example a beller place to nest or a beller male 149
150
PART 3: INTRODUCTION
provider for herself and her oUspring. In other cases however, the male's only contribution to the oUspring is his sperm. It is these cases that have orten Jed to the most elaborate male displays. Ryan provides a critical discussion of how female choice and male displays evolve in these circumstances. Sometimes females gain belter fertilization of their eggs through choice, or they avoid males who may pass on diseases during the act of mating. In other examples, females may have a sensory bias selected for in other contexts (e.g. feeding) and males exploit this through their displays. Some phylogenetic st udies show that the female preference evolved prior to the sexually selected male trait (e.g. the choice of call in the tungara frog or the female preference for swords in swordfish). In other examples, there is evidence that the female preference and the male trait have coevolved. as predict.ed by Fisher's 'runaway' hypothesis (coloration in sticklebacks and guppies). Finally. there is growing evidence that a female'S choice for mating with particular showy males may improve the genetic quality of her ofrspring (Marion Petrie's study of peacocks). Perhaps the main conclusion is that there will be different benelits of choice in difrerent cases and that several evolutionary mechanisms may interact. For example, Ryan points out that female preference evolved under sensory bias may later give rise to elaboration under a runaway process. This echoes the message from the previous chapter that we need to distinguish explanations for Ihe origin of signals from those for their maintenance. The next three chapters consider the evolution of social behaviour. Work in this field gained enormous impetus from Hamilton's (1964a,b) insight that individuals can pass on copies of their genes not only by producing ofrspring (direct fitness) but also by helping close relatives to breed (indirect fitness). This theory of 'kin selection' specifies, in Hamilton's rule, the conditions under which reproductive altruism will evolve. The social insects have proVided particularly good opportunities to test this theory, espedally the eusodal societies in which there is cooperative brood care with some individuals (workers) largely sacrificing their own chances of reproduction in order to help the queen produce offspring. In Chapter 9, Bourke reviews the theory and evidence that kin selection can explain reproductive altruism, focussing on both the genetic predispositions for helping behaviour (relatedness to beneliciaries) and the ecological faclors which innuence its COSts and benefits. Bourke discusses the famous and brilliant suggestion made by Hamilton, that haplodiploidy may explain both the prevalence of eusociality in the Hymenoptera (bees, wasps and ants) and the fact that only females in this group of insects show worker behaviour. In haplodiploidy, males develop from unfertilized eggs, and so are haploid, while females develop from fertilized eggs and are diploid. If the queen mates just once, then this genetic system creates 75% relatedness among sisters, and so seems to predispose females towards rearing reproductive sisters rather than daughters (to which they are related by 50%). The problem with this argument is that the stable sex ratio (or more strictly, investment ratio) from a worker's pOint of view is 3: I
PART 3: INTRODUCTION
lSI
reproductive females: males (Trivers & Hare, 1976). At this ratio, which is often achieved in nature, sib-rearing yields the same fitness returns as offspringrearing. Therefore haplodiploidy may not, afterall, provide a special genetic predisposition for the evolution of eusoaality. Bourke discusses the ecological factors which may have facilitated eusoaality in the Hymenoptera, including the benefits of group-living and a large nest, which favours joining a group rather than breeding solitarily, and the possession of a sting which enables workers to contribute effectively to nest defence, Although the cooperalion in bee and ant societies is impressive, Bourke shows that connicts of interest are also rife, Workers control sex allocation in the colony, against the queen's best interests, and recent studies show that they do this by selectively destroying male eggs or larvae. In addition, workers often lay unfenilized (male) eggs. Theory predicts when workers should police such cheating by their fellow workers, and instead favour the queen's male eggs. Recent experiments provide impressive suppon for the theory. The chapter also reviews 'reproductive skew' theory which attempts to explain whether dominant individuals in the colony should monopolize all the reproduction or share it with others. In Chapter 10. ErnJen discusses the social tensions in venebrate family groups. Like the previous chapter, he uses kinship theory and reproductive skew theory to predict the conditions under which individuals should help others or allempt to breed themselves. Among the most important factors determining whether an individual should stay at home and help or disperse and breed are its relatedness to the young produced at home, its chances of successful dispersal and the quality of the territory available at home compared with that for independent breeding. Experiments with several cooperatively breeding birds (e.g. Seychelles warblers. white-fronted bee-eaters) show that young vary their decisions adaptively and are more likely to leave home when their genetic profit from breeding exceeds that from helping. Emlen discusses how the conflicts predicted by kinship theory actually occur in practice in several birds and mammals. For example: when a breeding female dies, her sons may compete with their father for matings with the new female: replacement males kill young sired by the previous male; offspring are less likely to help at home when a parent is replaced; when their father dies, his sons may evict their mother and altract a new breeding female; fathers may disrupt their son's breeding allempts to cause their son to COme home to help instead. It is exciting to see the same theories applied successfully to the ants and bees of Chapter 9 and the birds and lions of Chapter 10. Early game-theory models of contests considered how single encounters were resolved by asymmetries in fighting ability or the value of a resource. In social groups, however, individuals often meet repeatedly to contest for food, space or mates. In Chapter II, Pusey and Packer consider how these repeated interactions may give rise to stable dominance hierarchies and cooperation. They review evidence from primates showing that alliances among buth
152
PART 3: INTRODUCTION
relatives and non-relatives may have various effects on the stability of social groups. Then they provide a critical discussion o( the 'prisoner's dilemma' as a model (or the evolution o( cooperation through redprodty. Although this model has become a (avourite o( theoreticians. Pusey and Packer point out that it (ails to capture the key (eatures o( most real-life examples o( cooperation. Real-lile contests are rarely between pairs o( individuals who make simultaneous decisions between just two options (cooperate versus de(ecl), with symmetrical payoHs (or each contestant. Instead, they suggest that much o( the cooperation in nature renects mutualism in which both individuals gain the highest payoff from cooperation, so there is no temptation to cheat. In other cases, some powerful individuals may be able to impose cooperation on others by the threat o( punishment. The chapter reviews three empirical studies o( cooperation. In the authors' own long-term studies of A(rican lions, individuals cooperate in hunting, care o( young and territory defence but there is no him that reciprocity is involved in this behaviour. Rather, it involves kin-selected altruism and mutualism. Predator inspection in fish has been claimed as an example of 'tit for tat' cooperation between pairs o( individuals. Pusey and Packer discuss the evidence (or this and the alternative hypothesis that individuals approach and reHeat together simply to seek the safety o( shoaling. Finally, some clever experiments with blue jays, using operant (eeding devices, show that pairs readily cooperate in mutualistic games but not in a 'prisoner's dilemma'. Pusey and Packer conclude that the evidence (or redprocity in nature is not sHong. In the final chapter o( this section, Haig shows that just as connicts are rife between individuals within pairs, (amilies and larger social groups, so they occur among the genes o( an individual's genome. The common metaphor o( an organism is o( a well-designed machine with the genes cooperating to maximize that individual's reproductive success. Instead, Haig suggests that we might think of the genes as members o( social groups with the organism's properties renecting the same mixture o( cooperation and internal conllicts whidl underlie the societies discussed in previous chapters. Selfish genetic elements provide a clear example o( how a gene's interests may not coincide with that of the individual. A gene which submits to the meiotic lottery in reproduction will get through, as a copy, to 50% o( the o((spring. However, some genes (segregation distorters) kill off gametes which do not contain copies of themselves and so now get through to all the offspring. Provided they do not reduce the parent's (ertility by 50%, the gene will spread. Another example of gene selfishness involves 'genomic imprinting' in which the expression o( a gene varies depending on whether it is inherited from the mother or father. Paternally imprinted genes include 'growth factors' in mice which cause embryos to take more resources from the mother. This makes good sense because. giv~n multiple
paternity, paternally derived genes are Jess likely to be present, as copies, in siblings than maternally derived genes and so they should behave more selfishly. Haig discusses how internal connicls within the cell nucleus are limited by
PART 3: INTRODUCTION
153
the procedures o[ fair segregation and recombination. Just as the rules which govern an individual's social behaviour in societies may have evolved as the evolutionarily stable ways to sellie conflicts. so mechanisms which regulate the cell cycle may reflcclthe stable outcomes of conflicts among genes within the genome.
Chapter 7 The Evolution of Animal Signals Rufus A. Johnstone
7.1 Introduction This chapler is concerned with the evolution of animal signals; that is, with traits that are specialized for the purpose of communication. The diversity of signals is enormous, ranging from the bright colours typical of many birds and bUlternies, to the calls of frogs and crickeLs, from Lhe pheromones released by moths, ants and many other insects. LO the aggressive posturing of lizards and fishes. They may serve to amact a mate or to deLer a rival. to warn conspecifics of an approaching predator or alert o(fspring to the return of a food-bearing paren!. Faced with this bewildering array of traits, so varied in their appearance and their function, any altemptto deduce general principles of signal evolution may seem doomed Lo failure. However, animal displays in all their variety form a coherent and distinctive class of characters, because the selective pressures that innuence their design are different from those lhal act on other traits. Communication occurs when the actions of (or cues given by) one animal influence the behaviour of another (Wiley, 1983; Endler, 1993). Consequently, the properties of that other individual. the receiver, exert strong selective pressures on signal design. Furthermore, signal design exerts reciprocal selective pressures on receiver behaviour. On the one hand, natural selection favours signallers who elicit favourable responses - those who are belter able to intimidate opponents, for example, or to atLract maLing partners. On the other hand, it favours receivers who can accurately deduce the nature and intentions of signallers from their displays - those who can best determine whether an opponent's threat is real and not just a bluff, for instance, or whether a potential maLe will make a good paren!. To understand the process of signal evolution (commonly referred to as rilllalization), we have to bear in mind both of these sets of selective pressures. The structure of the chapLer reflects Lhis dual emphasis: the first secLion deals wiLh the selective pressures aCling on signallers; the second with the selective pressures acting on receivers. Theories of signal evoluLion Ihat emphasize the former are referred to as 'efficacy based', while those that emphasize the latter are termed 'strategic'; the third and final section altempts to reconcile the apparently conlrasting conclusions of the efficacy based and strategic approaches. 155
156
CHAPTER 7
7.2 The signaller's perspective From the signaller's perspective, a display is a means of manipulation. Il serves 10 inlluence the receiver's behaviour in a way lhat benefits the signaller (whether or not the receiver also benefits is a dif(erent matter, 10 which I will turn in Section 7.3). Selection favours individuals whose displays are more eHective at eliciting beneficial responses. At the same time, however, signalling is likely to be costly. The calls of frogs and crickets, for example. which are so effective in allracting mates, may also draw the undesired attention of predators and parasites (Cade. 1979; Tunle & Ryan. 1981; Sakaluk & Belwood. 1984). Funhermore. they are likely to involve considerable energetic expenditure (Ryan. 1988; Prestwich, 1994). Selection thus favours individuals whose displays incur less risk and are energetically cheaper to produce. The signals We see in nalure should be those lhal strike the optimum balance between lhese two conOicting pressures for greater ef(ectiveness and lower fitness cost, i.e. those that are most efficient (Wiley, 1983, 1994; Endler. 1992. 1993).
7.2.1 Getting the message across The first requirement of an dfective signal is that it should be detectable by receivers. Communication, however, often takes place over long distances, in a noisy environment. Signals may therefore be severely anenuated and de· graded by the time they reach the receiver. and may have been mixed with irrelevant stimuli. This makes it difficult for recipients to distinguish them from spurious stimulation. or [rom other kinds of display. A number of design features. common to many different kinds of signal. serve to increase effectiveness by making the task of detection easier. Fleishman's (1988a,b, 1992) studies of lhe visual 'head-bobbing' displays of anoline lizards provide a nice illustration of some of these common properties. Territorial male anoles employ lwo basic types of head-bobbing display. The 'challenge' display is usually given after a direct approach toward an intruder. and is often accompanied by postural modifiers lhat appear to be associated Wilh different levels of aggressiveness. The 'assertion' display, which is not accompanied by sueh modifications, is given spontaneously from elevated perches within a male's territory, and does nOl appear 10 be directed at specific individuals. To be noticed by viewers that are relatively far away, and unlikely to have lheir gaze directed at the signaller, the assertion display must be visible and anention-catching at long range. Detailed examination of this display in AnoJis aura/us. a grass anole from Panama, reveals that the signal is designed to function effectively in this way. First, as shown in Fig. 7.I.the display begins with a series of head movements that are of high acceleration, velocily and amplitude compared with the rest o[ lhe display. These movements are conspicuous, both because of their amplitude and speed. and because they involve frequency components distinct from those present in the common background motion
EVOLUTION OF ANIMAL SIGNALS
157
Head
position
Dewlap extension
-'---Head
~~7'~~'< />
I15mm
Fig. 7.1 A schematic diagram of how Ihe assertion display of A. auratus is created, together wilh two typical examples of the display. (Modified from Fleishman. 1992.) The lizard
moves ils head up and down and, independcnlly. expands and contracts its dewlap. Each display record shows the poSition of a fixed point on the head along the venicalline L. and the vertical distance from the bonom point of the head 10 the lowest portion of the dewlap. as a function of time (in the upper diagram. the lizard's head is shown at limes t.
and
'z'
with the resulting record to the right of the picture). Each display is divided into a
high·amplitude. introductory ponion (I), and a laller ponion (II).
typical of windblown vegetation. The potenliallor background motion otherwise to mask the display is illustrated by the cryptiC movemenltypical of the vine snake Oxybelis amel/S, the anoles' major predator in Panama. While the lizards' head-bobbing contrasts with spurious background movement, the snake superimposes on its forward travel a genlle rocking motion that is similar in frequency to the swaying of a branch blowing in the wind (a type of camouflage that is rendered all Ihe more effective because the snake moves preferenlially when the surrounding vegetalion has been put into motion - Fleishman, 1985).
Second. the anoles' assertion display is repetitive. with the head being raised and lowered several times during the initial, high-amplitude portion (Fig. 7.1). Repetition is just one form of redundancy. a more general property (hat can be
158
CHAPTER 7
defined as the existence of predictable relationships among different parts of a display. Redundancy helps with deteClion because it allows the receiver 10 reconstruct the correCl signal from an imperfectly received one (another Cl)m.mon instance. besides repetition, is the use of multiple signal components in parallel. a striking example of which is provided by the sexual advertisement of male Jackson's widowbirds, Eupleetesjacksoni, which comprises an elongated tail, performance of a jump display and conslruction of a courtship bowerAndersson, 1989, 1991, 1993). Third, the anoles' signal is stereotyped, particularly in its latter part, which features the species-specific pattern of movement known as the 'signature bob'. Stereolypy reduces the difficulty of the task faced by receivers by minimizing the number of categories into which they must classify incoming signals (Cullen, 1966). Instead of producing displays that vary in complex ways, Signallers increase the probability of deteClion by employing only one or a few standardized signals. Finally, Ihe assertion display illustrates the use of alerting components. That is, it begins (as described above) with components that are relatively variable and encode little information, but which are highly deteClable compared to the rest of the display. This initial section increases the chance that a recipient will notice and recognize the subsequent (information-bearing) parts of a signal. by specifying the interval of time during which it can expect to receive them. Conspicuousness. stereotypy. redundancy and the presence of alerting components are properties common to many different kinds nf signal. because Ihey serve very generally to increase the reliability of detection (Wiley. 1983). Further consideration of the selective pressures acting on signallers can also help to account for the differences in signal design among species (or among the different displays of a single species). This is because the physical environment in which communication occurs and the audience of potential recipients differ from onc signalling system to another. and these factors have an important influence on the relative efficiency of different kinds of display. What makes for an efficient signal in one eon text will not necessarily do so in another. Sections 7.2.2 to 7.2.4 illustrate how a consideration of the communicalOry context can help to explain detailed aspects of display design. beginning with the effects of the physical environment in which signalling occurs.
7.2.2 The influence of the physical environment
The influence of the physical environment on the design of animal displays is one of the most well-studied aspects of signal evolution, particularly with regard to long-range acoustic and visual displays (for reviews of the subjeCl, see Chapter 2 and Wiley & Richards. 1978; Gerhardt. 1983; Alberts, 1992; Romer. 1993; Forrest. 1994). 'JWo sample studies, one of visual and one of acoustic signaUing. will serve to illustrate how interspecific variation in the magnitude and form of signals can be explained by variation in environmental conditions.
EVOLUTION OF ANIMAL SIGNALS
159
Consider, first, the study by Marcheni (1993) of the plumage panerns of eight species of warblers of the genus Phylloscoplls, which breed in the forests of Kashmir, India. All of the species arc small and greenish in colour, but they possess varying numbers of pale colour patches on their wings, crown, rump and tail. These conspicuous plumage traits were found to playa role in intraspedfic communication, particularly in the establishment and maintenance of territories. Experimental manipulation of male coloration in P. inomatlls (using paint to enlarge or reduce wing-bar size or add a new colour patch) directly aliened territory size, with males that were rendered more conspicuous obtaining largerterritories than control indiViduals, who in turn obtained larger territories than males rendered less conspicuous. Comparison of all eight species then revealed that they could be unambiguously ranked from duller to brighter according to the number of colour patches they possessed (the order being: no colour patches, one wing-bar, two wing-bars, two wing-bars and a crown stripe, two wing-bars and a crown stripe and a rump patch and, finally, all of these plus white outer tail feathers). Species with more patches tended to occupy dark, dense habitats, while those wilh fewer patches tended to breed in open areas. In other words, there was a negative correlation between habitat brightness and species brightness (which remained significant when controlling for the effects of phylogeny). !L thus appears that properties of the physical environment occupied by these species have influenced the evolution of their plumage signals. Because the ability to perceive visual displays depends on lhe amount of light in the environment, species thal occupy darker habitats have evolved brighter coloration for intraspecific communication, while those that breed in open areas, where they are visible even without conspicuous plumage panerns, have evolved duller coloration. The above example shows that the physical environment can influence the evolution of signal intensity (i.e. conspicuousness). A broader comparative study by Wiley (1991) of the male territorial songs of eastern North American oscine birds shows that it may also affect the particular form that a signal takes. The study considered 120 species of osdnes, for each of which a number of song properties were recorded. Three of these properties concerned the temporal structure of the song, and were indicative of rapid modulation: the minimal period of repeated elements (such as syllables in a trill), the presence or absence of one or more buzzes (notes with a wide bUl band-limited spectrum) and the presence or absence of 'side-bands' on a song spectrogram. The habitats occupied by territorial males of the various species were then classified into six categories, three of which (coniferous forest, mixed forest and parkland) could be grouped as forested habitats, and three (shrubland, grassland and marsh) as open habitats. Statistical analysis (incorporating various measures to control for lhe e[fects of phylogeny) revealed a pronounced association between the physical environment and the lemporal properties of song, with shorl repetition periods, buzzes and side-bands less common among birds of forested habitats. All three of these features render a song easily degradable by
160
CHAPTER 7
reverberation, which is more common in scattering environments such as forests than in open habitats. Wiley's results thus suggest that the temporal propenies of the song of many oscines have evolved to reduce the effects of reverberation in forested environments. 7.2.3 The influence of the audience
While the physical environment can have strong effects on Ihe design of animal displays, it is the influence of receiver properties on their evolution that distinguishes signals from other trailS. To be effective in eliciting favourable responses. a signaller must employ displays that are well suited to detection by the sensory and neural mechanisms of intended receivers (Ryan, 1990; Guilford & Dawkins, 1991; Endler, 1992, 1993). Moreover. in order to minimize costs resulting from the attraction of predators and parasites, a signaller should employ displays that are ill-suited to detection by these unwanted eavesdroppers (ibidem). Every member of the audience of potential receivers. not just lhose that the signal is designed to reach. can influence the evolution of signalling behaviour (McGregor, 1993). Studies by Endler (1978. 1980, 1991. 1992) of the coloration of male guppies (Poecilia reliculara) provide some of the most detailed evidence for the matching of display design to receiver sensory capacities. Wild populations of guppies have complex and polymorphic colour patterns. comprising patches of eight major types: red--<>range, yellow. bronze-green, cream-white. blue. silver, brown, black and body colour. These colours. restricted to adult males. playa role in mate choice. Female preferences vary among populations, bUl typically favour males with more carotenoid (red, orange and yellow) patches. more structural colour (blue, green and silver) patches, and patterns that contrast in colour or patch size with the visual background (see also KodricBrown, 1985; Houde. 1987; Long & Houde. 1989; Houde & Endler. 1990). At the same time. conspicuous and highly colourful fish suffer greater risk of predation. Endler determined. for the various light environments typical of the species' habitat, the relative conspicuousness of a number of colour patterns, both to other guppies and to potential predators. The colour patterns considered were those typical of lhree different guppy populations, while the predators included the most dangerous diurnal visual predator, Crenicichla alra (Cichlidae). a less dangerous diurnal fish, Rivulus harlii (Cyprinodontidae) and the moderately dangerous diurnal freshwater prawn, Macrobrachium crenulaltlm. Endler's results suggested that selection had favoured those colour patterns thaI were most conspicuous to guppies while being least conspicuous to predators. For instance. in one population where prawns were the major predalor, the most abundant hue was orange. a colour to which the prawns are not very sensitive. Where fish were the major predalor, by contrast. all colours were more evenly distributed. Moreover. the timing of guppy counship appeared to be adapted for efficient communication. Guppies court
EVOLUTION OF ANIMAL SIGNALS
161
most frequently early and late in the day. when the colour of ambient light is different to that typical of other times. The net effect of courting under this condition was found to be better than under any midday light environment. Courting under the latter conditions would lead to decreased conspicuousness to guppies or increased conspicuousness to predators (or both). The above studies demonstrate that the mating displays of male guppies and the sensory capacities of females are well matched. but they do not reveal whether this is the result of evolutionary change of the former to match the latter. or vice versa (or whether it is the result of coevolutionary change of both traits), Recently. however, evidence has begun to accumulate that some male mating displays have evolved to take advantage of pre-existing sensory properties of females, which evolved [or reasons unrelated to communication.
This possibility is referred to as 'sensory exploitation' or 'sensory trap' (Ryan, 1990; Ryan & Rand. 1990, 1993; Christy. 1995; see also Chapter 8). The courtship display of the water mite Neumannia papillator. studied by Proctor (199 I l, provides a good example. Mites of this species are ambush predators of copepods. They assume a characteristic 'net-stance' posture while hunting. in which they rest on their hind four legs on aquatic vegetation, with their first four legs held out in the water column. From this position. they orientate to and clutch at the vibrations caused by swimming prey. A male searching for mates, once he has located a female, will walk slowly around her while vibrating his legs, a display referred to as 'courtship trembling'. She, in turn. will olten respond as if to prey. by orientating to the source of vibration and clutching the male in her forelegs. Male leg-trembling frequencies are well within the range of vibrations produced by copepods; moreover, experimental feeding and starvation of females revealed that hungrier individuals were more likely to orientate to and clutch at courting males. It thus appears that male mites are capitaliZing on female sensory adaptations for the detection of prey. Additional support for this suggestion is provided by reconstruction of the historical pattern of preference and display evolution. Proctor (1992) has shown. by cOnstructing a c1adogram of N. papillator and other mites in the same family (Unionicolidae), that the female sensitivity to leg trembling (associated with the use of the net stance during hunting) may have originated prior to the evolution of the male display. in which case it cannot initially have served any role in male choice.
Unfortunately. while studies like Proctor's show that some male displays take advantage of pre-existing female biases. there is not yet enough evidence to assess the frequency of sensory exploitation (see Johnstone. 1995a). There are good theoretical reasons to expect that sensory biases will be common. The sense organs and nervous system of an organism serve many functions besides the detection of and reaction to any particular signal. and it is unreasonable to assume that each function can always be optimized independently. Consequemly. selection for sensory capacities useful in non-mating contexts (such as vibration sensitivity in water mites) is potentially a widespread source
162
CHAPTER 7
of pleiotropic mating biases. In addition. Arak and Enquist (1993) have pointed out that recognition of any given signal (such as a male sexual display) can be achieved by a large number of equally efficient mechanisms that differ only in their response to stimuli outside the normally occurring range. Because selection is blind to responses that are provoked by such stimuli. the precise mechanism used in Signal recognition is thus subject to change by random drift. This may lead to a change in the display form that is optimal in stimulating a response. giving rise to hidden preferences that are open to subsequent exploitation (sec also Krakauer & Johnstone. 1995). However. the fact that hidden preferences arc likely to be widespread does not necessarily imply that most existing male display traits have evolved to take advantage of such preferences. If sensory exploitation is widespread, then it may offer a new explanation for some of the common features of signal design. Bmh lhe coloration of guppies and the leg-trembling behaviour of mites are matched to unique sensory and neural properties peculiar to these species. Common display features. however, may serve to lake advantage of widespread sensory properties that are shared by many different receivers. Ryan and Keddy-Hector (1992). for example. have shown (in a review of published studies) that in spedes where females prefer male trailS that deviate from the population mean, they tend to favour traits of grealer quantity. even if these lie outside the range of displays normally employed by conspecific males. In the case of visual traits. for example. females tend to prefer larger. more actively displaying, more colourful males. while in the case of acoustic display, they generally favour calls that are longer. of greater intensity and delivered at a higher rate. The exaggerated nalure of many display traits might thus be attributable, in part. to a widespread bias of receivers toward greater neural stimulation. A related argument is that of Searcy (1992). who has suggested that the song repertoires of many male birds (see also Searcy & Andersson. 1986; Catchpole, J 987) may have evolved because variation in singing behaviour helps to reduce habituation on the part of the receiver, and thus to elicit a stronger response. These possibilities are discussed in more detail in Chapter 8. 7.2.4 The influence of other signallers
If a signal is to be readily detectable, it must (as previously mentioned) contrast with spurious stimulation reaching the receiver. One of the most common sources of such stimulation is the displays of other signallers attempting to communicate with other receivers in the vidnity. Species that are active at the same season and in the same habitats should employ distinctly different signals. in order 10 reduce the impact of this kind of interference. A recent study of reproductive character displacement (Loftus-Hills & Littlejohn. 1992) in toads provides some illustrative evidence of this possibility. The two species of narrow-mouthed toad. Gaslrophryne Ci1rolinensis and G. olivacea, are widely distributed in the southern US. The former occurs in the southeast,
EVOLUTION OF ANIMAL SIGNALS
163
the laller in the southwest, but there is a zone of sympatry in eastern Texas and eastern Oklahoma. Males of both species rely on an advertisement call to attract females. Recordings of calls were obtained from a number of diflerent localities, which could be grouped into four areas: sympatric and allopatric for G. carolinensis, sympatric and adjacent allopatric/shallow sympatric for G. olivacea. Comparison of calls among these groups (taking into account the influence of body size and of water temperalllre on calling behaviour) revealed that while the values for the dominant frequencies of the calls did not overlap between the species, those of sympatric G. carolinensis were displaced away from those of both groups of G. olivacea, which were very similar to each other. It is not only the displays of other species that can influence signalling behaviour. In many signalling systems, groups of conspecifics interact in their attempts to allract or deter receivers (Greenfield, I 994a,b). Examples indude the interaction among males anempting 10 obtain mates in lekking birds and mammals or chorusing frogs and crickets, the interaction among young in a brood allempting to obtain food from their parents, or the interaction among a group of prey individuals attempting 10 divert the allention of predators away from themselves. Sometimes, there are clear conflicts of interest among the signallers. In other cases, the interaction appears cooperative (at least in the short term), as in the dual-male courtship of the long-tailed manakin (Chiroxiphia linearis), where a bew male assists with the display of an alpha male, although claiming virtually none of the resulting matings (McDonald & Potts, 1994). Whether the signallers cooperate or compete, however, the interaction can strongly influence their display behaviour. The effects of signaller interaction (in conjunction with the properties of receivers) on display are nicely illustrated by Greenfield and Roizen's (1993) sllldy of chorusing in the neotropical katydid, Neoconocephalus spiza. To all ract mates, males of thiS species produce loud advertisement 'chirps' by forewingforeWing stridulation. An isolated male wilJ usually maintain a regular chirp rhythm for several minutes, but those calling within 10 m of each other roughly synchronize their chirps, with the leading role in a chorusing pair usually alternating from one chirp to the next. This synchronization appears to be the result of a phenomenon known as inhibitory resetting. A male is inhibited from calling by sound initiated more than a very short time before his anticipated chirp, and this inhibition continues umil the sound ends, Immediately after release from inhibition, the insect's next chirp is slightly advanced. When two males who sustain similar chirp rates call together, runs of synchrony will inevitably result (although if one male chirps faster, it will end up doing most of the calling, while the slower individual remains silent, due to repeated inhibition). Why should males exercise inhibitory reselling and thus call in synchrony? Females of the species show a preference for the leading call in a closely synchronized sequence. This presumably imposes strong selection pressure on males to adopt a mechanism that improves the chance of calling slightly before
164
CHAPTER 7
a neighbour, and inhibilory resetting does just thaI. Greenfield and Roizen (1993) compared, in simulation, the performance of males employing the resetting strategy and males employing a hypothetical 'independent' strategy (in which an individual's calling behaviour is uninfluenced by his neighbours), They showed that because of the female preference for leading calls, reseuing males would be favoured over independent callers, whichever strategy the two were competing with. It thus appears that synchronous chorusing in this species is the product of competitive interaction among males attempting to jam each other's signals.
7.3 The receiver's perspective As described above, both the physical environment and the psychology of receivers exert a strong influence on the design of animal displays. While the selective pressures associated with the physical environment, however, are relatively static. those associated wilh receiver psychology may change in response to lhe evolUlion of signal design. This is because receivers are themselves under selection to respond appropriately to the behaviour of (and cues given by) other individuals. From the signaller's perspeclive, a signal may be a means of manipulation, but from the receiver's perspective, it is a potential source of information. There are many situations in which an individual stands to gain from a knowledge of the physiological or motivational state of others. When a parent bird returns to the nest with food. lor example, it is laced with a brood of gaping, jostling nestlings all fighting for attention. Knowledge of which chick is hungriest would allow the parent to allocate the food where it is most needed. Similarly, when coursing predators such as wild dogs approach a group of prey, they must Single out one individual to pursue. Knowledge of which prey animal is weakest would allow them to minimize pursuit costs by avoiding less vulnerable targets. Selection thus favours receivers who can better adjust their behaviour in response 10 the information provided by signals (which extends 10 ignoring displays thai are uninformative). The rest of Section 7.3 deals wilh the implicalions of this selective pressure for signal evolution. 7.3.1 Cooperation and conflict in signal evolution In a cooperative signalling system. selection acting on receivers works in concert
with selection acting on signallers, in both cases favouring efficient communication. This will occur whenever the signaller benefits by eliciting a response that is also to the advantage 01 the receiver. One of the best-known instances of cooperative signalling is the use of a dance 'language' by honey bees to convey to fellow hive members the exact location of food. Because a worker returning from a food source to the hive benefits by directing fellow workers to the resource, JUSt as they benefit from the information it has to provide, selection on the receivers works in concen with selection 011 the signaller. This
EVOLUTION OF ANIMAL SIGNALS
165
is an unusually precise form of communicalion, in that different aspects of the dance convey information about the distance 1:0 the food source, and about the direction that must be taken to reach it from the hive (von Frisch, 1967). A simpler and more widespread example of cooperative communication is the advertisement of species identity during courtship. Both males and females share an interest in pairing with members of their own species (although, as discussed below and in Chapter 8, females may also benefit by exercising choice among conspecifics), and as a result males typically have evolved distinctive, species-specific displays, while females have evolved efficient mechanisms for their location and identification. Although some signalling systems are cooperative, however, many kinds of communication involve a conlliet of interest between signaller and receiver. Selection acting on each individual then opposes selection acting on the other,
because the signaller stands to gain by provoking a response that is not to the advantage of the receiver. The exchange of threat displays during aggressive interactions provides a dear example of this kind of conllict of interest. Each participant would benefit from an accurate assessment of its opponent's fighting ability and motivational state, but each also stands to gain by misleading that opponent about its own ability, so as to more ef!cctively deter resistance (Maynard Smith, 1974). Sexual signalling is anOther case in point. Although males and females share an interest in locating and pairing with conspecilics, males are typically under stronger selection to acquire many mates, while females are under stronger selection to acquire superior mates (Trivers, 1972; Clutton-Brock & Vincent, 1991; Clulton-Brock & Parker, 1992; Johnstone er aI., 1996). Females would thus benefit from an accurate assessment of male mate quality (allowing them to reject inferior partners), while males would benefit by misleading potential mates as to their own quality (so as to gain more matings). As a final example, although parents would benefit by allocating food in relation to the hunger or need of their young, individual offspring stand to gain by misleading the parent as to their own level of need, because they are selected to acquire more food than the parent is selected to give (Trivers, ) 974; Parker & MacNair, 1978; Godfray, 1995a). In situations like these, communication is best viewed, not as an harmonious exchange of information, but as the focus of an arms race between signallers as manipulators, and receivers as 'mind-readers' (Dawkins & Krebs, 1978; Krebs & Dawkins, ) 984). Early proponents of the 'arms race' approach suggested that, on an evolutionary time-scale, informative or honest signalling was unlikely to endure for very long. Maynard Smith (1974; Maynard Smith & Price, 1973; Maynard Smith & Parker, 1976), for example, argued that threat displays conveying information about aggressiveness or level of cscalation were unlikely to be stable. Suppose a population existed in which individuals did convey information about their intentions. II an individual found that its opponent was announcing a higher level of escalation than its own, it would pay to retreat at once. Consequently, a 'deceitful' mutant that invariably announced a very
166
CHAPTER 7
high level of aggressiveness, regardless of its true intentions, would be favoured by seleClion because its opponents would always back down, Before long, everyone would be lying, and it would then pay 10 ignore the signal altogether. The same argument can, with slight modification, be applied 10 any other situation in which there is a conflict of interest bctween signaller and receiver. Whenever a correlation exists between signalling behaviour and the underlying state of the signaller (Le. whenever the signal is informative), the population appears vulnerable to invasion by a 'lying' mutant that adopts the signals typical of individuals in a different state, and t.hereby elicits a more favourable response. There are, however, properties of signal design that can help 10 maintain honesty, even in the face of a confliCl of interest, by making it impossible or unprofitable for signallers to employ a display that is not representative of their state. It is not necessary, therefore, to abandon the idea that animal signals convey information of value to receivers. Instead, one can argue that the displays we see in nalUre should simply be designed in such a way that they are not vulnerable 10 corruption by deceit. This offers a new 'strategic' perspeClive on signal evolution, which differs from the 'efficacy based' approach detailed in SeClion 7.2 (see Guilford & Dawkins, 1991). If seleClion favours reliability as well as efficiency, then some aspects of signal design may have evolved to ensure honesty rather than 10 facilitate detection. SeClions 7.3.2 to 7.3.6 outline the properties that can make a display reliable, and consider the evidence that animals make use of informative signals that possess these properties. 7.3.2 Physical constraints on deceit
The simplest mechanism for the maintenance of honesty is physic.,1 constraint. If there is a direct material link between a signal and some underlying aspect. of the signaller's state or condition, then lying may be physically impossible. Carotenoid-derived coloration is a good example of an 'assessment signal' of this kind. Most animals cannot synthesize carOlenoids for themselves, but must ingest them. Consequently, the intensity of an animal's carotenoid colours provides information about its foraging success and nutritional
stat us. An individual that has had poor success in finding food simply cannot produce the carotenoids necessary 10 maintain intense coloration. Hill and Montgomerie's (1994) study of plumage colour in the house finch (Carpodacus mexicanus) illustrates the effectiveness of this mechanism in maintaining honesty (see also Hill, 1992). Male house finches display ornamental carotenoid coloration that varies from pale yellow to bright red among individuals in a single population. This coloration is deposited in the feathers at the time of the annual moult. During moult (which lasts aboUl 105 days), leathers grow in regular daily cycles, with darker material deposited at night and lighter material in the day, giving rise to 'growlh bars'. The width of these bars can be
EVOLUTION OF AN[MAL SIGNALS
[67
used to infer the nutritional condition o[ an individual over the moult period, because leathers grow more slowly (yielding narrower bands) during episodes o[ food stress. Hill and Montgomerie examined four separate populations, and in each of them found a significant positive correlation between male plumage brightness and mean growth bar width. In other words, males that grew redder and more intensely pigmented plumage also grew feathers laster. Moreover, there was a positive correlation in all cases between the hue 01 growing feathers and the extent o[ moult of individuals examined, indicating that males that grew redder plumage tended to begin moult earlier. Both observations suggest that carotenoid-based plumage colour provides information about nutritional condition during moult. Hill (1990, 1991) has also shown that females make use of this information during mate choice, preferring to pair with redder males. In doing so they gain substantial benefits in the form or increased paternal care, because males feed their mates and offspring during incubation at a rate that was found to be positively correlated with their level of coloration. To sum up, it appears thaI females use a reliable assessment signal to acquire mates 01 superior quality. 7,3.3 Strategic constraints on deceit While some displays appear to be honest for reasons of physical necessity, there are many signalling systems in which there is no fixed link bel ween the signals used and any underlying aspects of the signaller's condition. An animal's choice of one threat display rather than another, for example, is not physically constrained by its fighting ability or motivational state. Equally, the length of a bird's tail is nOI fixed by any aspect of its condition. The maintenance of honesty under these circumstances requires a strategic justification, which can explain why 'inferior' signallers do not adopt misleadingly impressive signals even though they are physically capable of doing so. The 'handicap principle' or zahavi (1975, 1977a,b) provides a potential explanation of this kind. zahavi suggested that a signal could provide honest information about the 'quality' of the signaller, even in the race of a connict or interest, provided that it was costly to produce. His argument can be illustrated by reference to sexual display, the context in which it was first discussed. Suppose that a display used by males to advertise their value to females is cosIly to produce, particularly lor inferior males. Honesty may then be stable, because only superior individuals stand to gain a net benefit rrom display. A 'deceitful' mutant that opted to use the COSIly signal even though it was of low quality would suffer a loss of fitness, because the high cost involved would oUlweight the benefits to be gained from auracting more mates (see Fig. 7.2a). The handicap principle was greeted with considerable scepticism when first proposed, but the development of rormal models of honest signalling have since shown that the idea is plausible theoretically (Enquist, 1985; Grafen, 1990; Godfray, 1991; Maynard Smith. 1991; Johnstone & Grafen, 1992; Vega-
168
CHAPTER 7
(0)
/
Costs for / low- quality / signaller /
Benefits
/
...
~
Costs for
high-quality signaller
Signal intensity
Ibl
Benefits for high-need
Signalle;.r ............
',
-- --
~
~
, I
,,
,
, /1 /
,
,~
,, ,
--- -----Benefits for low-need signaller
......,.I~ /
,'/
Signal intensity
S·high
Fig. 7.2 An illustration of the way in which diHerential costs or benefits of display can maintain honesty. In (a) all signallers gain the same benefits, but low-Quality individuals
pay higher costs for a given level of signal intensity than do high-quality individuals. Under these circumstances, the optimal signal intensity (which maximizes net benefits) for
the former. s* /ow' is less than the optimal signal intensity for rhe laner.
Si~h'
In (b) all
signallers pay the same costs, but low-need individuals stand to gain less benefit from a
given kvel of signal intensity than do high-need individuals. Once again. the optimal signal intensity for the former,
s*/(fW
is less than the oplimal signal intensity for the latter,
SI:~h.
Redondo & Hasson, 1993). Moreover, the same argument can be applied to many different kinds of signal, not just to sexual display. Vulnerable prey individuals, for instance, would benefit by misrepresenting their status to predators, but a display that entails a costly increase in the risk of caplUre could serve reliably to advertise escape capacity, because only superior individuals could afford to emplOI' it (Vega-Redondo & Hasson, 1993). Similarly, weaker individuals would benefit by bluffing during conflict, but a signal that increases the danger of escalated fighting could proVide honest information, because only strong competitors could bear the risks involved in its use (Enquist. 1985). As a final example, offspring would benefit by simulating a dishonestly
EVOLUTION OF ANIMAL SIGNALS
169
high level of hunger in order 10 solicit additional food from parents, but a costly begging signal could provide honest information about their need for food, because only for truly hungry individuals would the benefits of additional resources outweigh the costs involved in production of the display (Godfray, 1991). This last case dif[ers from the previous examples in that signallers are assumed to differ, not in their ability to bear the costs of display, but in the benefits they stand to gain by eliciting a favourable response (sec Fig. 7.2b). The handicap principle, in other words. allows for honest advenisemem of 'need' as well as of 'quality' (see Maynard Smith. 1991; Johnstone & Grafen, 1992). While there is now widespread agreemem that the handicap principle is plausible theoretically, few empirical studies have demonstrated clearly that signal costs do playa role in the maimenanee of honesty. Perhaps the best evidence of this kind is provided by M0ller's (1988, 1989, 1994; M0ller & de Lope, 1995) studies of the tail ornamems of male barn swallows (Hirundo rusrica). Adult males of this monogamous species arc similar to females inmost respeds, the exception being that their outermost tail feathers are much longer. Field experimems in which male tail length was manipulated by cutting out sec!ions of feather and reglueing the tail together (or cutting and adding additional sections), together with simple observation, revealed that males with longer tails lend to be preferred by females during mate choice. As a result. they tend to gain mOre rapid access to a mate during the breeding season, and to acquire a higher qualily mate, increased maternal investment and a higher frequency of extrapair copulations. Given these benefits, why do some males only grow shon tail feathers? Males with experimentally elongated tails also tended to capture smaller, less profitable prey, and perhaps as a result of this exhibited lower survival rates (while males with shortened tails enjoyed improved survival prospeds). Moreover, nalllrally shon-tailed males were less able to survive with elongated tails than were naturally long-tailed males. M0ller's results thus suggest that the natural tail length of male swallows refleds their ability 10 bear the costs of an elaborate omamem. Only superior (naturally long-tailed) males can afford to grow long tails. because the cost of doing so would be prohibitively great for inferior (naturally shan-tailed) individuals. It must be admitted, however, that adding a set length to a shan tail results in a proportionally greater increase in size than adding the same length 10 a large lail, which might lead one to expect a greater impact on flight performance in the former case even if there were no relationship between natural tail length and male 'quality'. The above studies of swallow tail ornaments are, unfortunately, unique in their altempt to investigate whether the costs of display are greater for some signallers than others. There is considerable evidence, from a wide range of different signalling systems. that animal displays are expressed in a conditiondependent manner even when they are not physically constrained to be so (see Johnstone, 1995a, 1996a for reviews). However, while many of these
170
CHAPTER 7
displays appear to be costly, experimental evidence that this is the case is less common, and evidence for condition-dependent COSt almost completely lacking. A study by Cresswell (1994) of singing by skylarks (Alauda arvensis) in response to allack by merlins (Falco columbarius) serves to illustrate the kind of evidence that is available generally, Observation of natural allacks revealed that merlins were probably not using the occurrence of song to choose which skylarks to anack, but that they did chase non- or poorly singing skylarks for longer periods compared to skylarks that sang well (thiS finding being based on a comparison of attacks that were not lerminated by capture). In addition, song appeared to provide honest information about the birds' escape ability, since a merlin was more likely to catch a non-singing than a poorly singing than a full-singing skylark. The predator's decision as to whether or not it should continue pursuit thus seems to be based on a reliable signal of prey vulnerability, Reliability may be ensured by the energetic cost of singing while being chased, which is likely to be prohibitively great for birds that are in poor condition, and hence more vulnerable. However, there is no definite evidence that the display does involve high levels of energetic expenditure, or that it would have a more detrimental effect on the performance of non-singing than of fully singing birds, One cannot firmly conclude, therefore, that honesty is maintained by signal cost in this case, 7.3,4 The possibility of rare deception
While most studies inspired by the handicap principle have focussed on reliable signallin&- and the costs that may be necessary for its maintenance, it is worth noting that models of cornnlUnication based on zahavi's theory do not always predict universal honesty (Dawkins & Guilford, 1991; Johnstone & Grafen, 1993; Adams & Mesterton-Gibbons, 1995). For a signal to be evolutionarily stable it is necessary, as discussed above, that it should provide sufficient information for selection to favour response by receivers. 11 it were always misleading, the argument runs, then it would soon fall into disuse, as receivers would evolve to ignore il. Provided, however, that the signal is honest 'on average', this does not rule out the pOSSibility of occasional deceit. For example, if certain signallers find display cheaper (for a given level of quality or need) than do other individuals, then they will be able to afford to produce the display typical, in most cases. of a higher level of quality or need than their own. In doing so, they will elicit more favourable responses from receivers than would otherwise be the case. At an evolutionary equilibrium. receivers will devalue their responses to take into account what fraction of individuals signalling at any particular level belong to this class. However, as long as such individuals are rare enough, and the disadvantage (from the receivers' point of view) of responding to Ihem inappropriately is slight enough, they need not disrupt the signalling system altogether (Johnstone & Grafen, 1993), Evidence for occasional deception can thus provide support for the handicap
EVOLUTION OF ANIMAL SIGNALS
171
principle, if it can be shown Ihal deceitful signallers do differ from othe' in the net benefit they stand 10 gain from display. A good example is provided by Redondo's studies (Redondo & Castro, 1992; Redondo, 1993) of begging by magpie (Pica pica) and great spoiled cuckoo (Clamalor glandarius) chicks. The intensity with which magpie nestlings beg can be estimated by combining measures of the duration, latency and posture of the begging display, and the emission and duration of begging calls (all of which are strongly correlated with each other). Redondo and Castro first used anificial feeding 10 equalize Ihe begging intensity of chicks in naturaL broods of four or five individuals, and then recorded both their mass gain due to parental feeding over the next hour. and Iheir begging intensity at the end of this period. The chicks that begged most intensely were found 10 be those that had received smaller quantities of food from parents (relative 10 their body mass), indicating that the begging display of nestling pagpies provides reliable information about hunger. Moreover, a parallel experiment suggested that parents make usc of this information when distributing food. After anificial feeding had been used [0 enhance within-brood differences in begging, those chicks that djsplaved most intensely were found to receive more food (relative 10 their body mass) during the following hour. Honesty may be maintained in this case by the energetic costs of begging, as the signals employed by hungrier nestlings involved a higher degree of muscular activity. The above reliable system of parent-offspri ng communication, however, is exploited by the chicks of great spoiled cuckoos, obligate brood parasiles that are raised alongside hosl chicks, and severely depress the breeding success of magpies. Formal models of signalling suggest that individual young who are less closely related to others in the brood, or 10 parents, should be selected to use more costly begging behaviour, because they have less 10 lose by depriving nestmates of additional food (Godfray, 1995b). In accordance with this prediction, laboratory experiments conducted with magpie and greal spotted cuckoo chicks of a similar developmental stage, kepI in isolation under controlled conditions of food supply, revealed that cuckoo chicks exhibit exaggerated begging behaviour. For a similar degree of need, the parasite nestlings begged for longer, and gave more calls per unit time than host chicks. Funhermore, field experiments revealed that magpie parents, when given a choice between a cuckoo and a magpie chick, tended 10 preferentially feed the cuckoo. It thus appears that cuckoo chicks 'cheal' magpie parents by means of dishonest begging signals. This kind of deceit is, however, prevented from completely disrupting the signalling system by COnStritints on the frequency of brood parasitism (e.g. high rales of cuckoo egg mortality, in pan due to nest defence and rejection of foreign eggs by hosts). 7,3.5 Strategic explanations for signal diversity
Studies of the kind described in Sections 7.3.2 to 7.3.4 have shown that animal
172
CHAPTER 7
signals often do convey reliable information about the state or condition of the signaller (at least 'on average'), and that in some case they are designed in such a way as to prevent this reliability being compromised by deceit. There have been few attempts, however, to account for differences in signal design among species (or among the different displays of a single species) in terms of selection for reliability. Zahavi (I 977b, 1987) has argued that costs of signal production can only help to maintain honesty if the ability to bear those costs depends on the aspects of signaller condition that are being advertised. Consequently, he suggests there is a necessary link between the form of a display and the information that it conveys. Honest advertisement of nutritional condition and energy reserves, for example, should require a signal that is energetically costly, while advertisement of escape ability should require a Signal that entails a costly increase in the risk of capture by predators. However, this approach has not been used yet to generate testable predictions, for two main reasons. I The requirement that a signal hould entail a certain type of cost does not impose much of a restriction on its design, and so does not allow one to draw detailed conclusions about its likely form. Advertisement of nutritional condition, for example, may require an energetically costly signal. but this could take the form of vocal display, strenuous posturing, or many other types of behaViour. 2 The costs involved in the production of many signals may be imposed by the behaviour of receivers, rather than being physically concomitant upon display. In these cases, the design of the signal is 'conventional', because the appropriate receiver response (imposing whatever kinds of cost are necessary for the maintenance of honesty) can be elicited by a display of any kind (Guilford & Dawkins, 1995). A more successful attempt to explain signal diversity in strategic terms is that of Briskie et al. (1994 l, who investigated the begging behaviour of I I species of passerine birds, which differed in their frequency of extrapair paternity. A higher level of mixed parentage means that young in a brood are, on average, less closely related to each other. Consequently, greater costs should be reqUired to maintain honest signalling of offspring need in these species (sec Godfray, 1991, 1995b; Johnstone & Grafen, 1992). Just as a cuckoo chick begs more loudly than a host chick because it has nothing to lose by depriving its nest mates of food, so chicks in a species with low relatedness among brood members should beg more loudly than chicks in a species with high relatedness. In accordance with this prediction, comparative analysis (using various methods to control for the effects of phylogeny, brood size and body mass) revealed that the loudness of neStling begging calls among the sample species was positively correlated with the percentage of extrapair young, as illustrated in Fig. 7.3. It thus appears that interspecific differences in signal intensity can, in this case, be explained in strategic terms, as a consequence of differences in the degree of relatedness among competing signallers.
EVOLUTION OF ANIMAL SIGNALS
173
,-__...!O!f---- 1 Tachyc;neta bieolor ' - - - - 2 Hirundo rustica
b
1-------- 3 Sialia sialis 1-------- 4 Prunella modularis , - - - - - - 5 Passer;nB cyanes c
r---- 6
f----
Melosp;za me/odis
7 Zonotrichia Jeucophrys
8 Calcariu$ lapponicus
9 C. pictus
• r - - - - 10 Agela;us phoeniceus L
11 Molothrus atef
o
'a.
c
< />
d
.~
3~4
b
~
~
Fig. 7.3 A phylogeny of the bird species considered by Briskie et
6~7
80
•
100
-40
-30
.9
al. (1994), IOgether with a paired
.11
comparison of begging loudness bel ween species wilh high (open
-20
-10
Begging loudness IdB)
circles) and low (filled circles) levels of relatedness within a
brood. (From Briskie tt al.. 1994.)
7.4 Conclusions and prospects Sections 7.2 and 7.3 outlined two different perspectives on signal evolullon. The efficacy based theories described in Section 7.2 suggest that as a result of the selection pressures acting on signallers. animal displays will be designed so as to eHiciently eliot a response. The strategic theories described in Section 7.3, by comrast, suggest that as a result of selection pressures acting on receivers, displays will be designed so as to prevent their corruption by deceit. Many common display propenies can thus be explained in two different ways. The striking and costly nalure of many animal signals is a case in point. On the one hand. extravagance may serve to make a display more easily detectable or stimulating to receivers; on the other hand, it may make a display more reliable by ensuring that only superior (or more needy) signallers can aHord the cost of its produclion. How can these two different views of signal evolution be reconciled? From the evidence presented. it is clear that neither approach can be entirely abandoned in favour of the other. As Marcheni's (1993) studv of plumage brighmess in warblers (described in Section 7.2.2) showed, some of
174
CHAPTER 7
the variation in signal lorm seen in the natural world can be explained only in terms 01 ellicacy. Phylloscopus SpeciL'S liVing in darker habitats have nOl evolved brighter coloration as a result 01 selection lor reliability. Equally, the study by Briskie er at. (1994), described above, showed that some dillerences in signal lorm can be explained only in strategic terms. Chicks 01 species with higher levels 01 extrapair paternity do not beg more loudly because 01 selection lor detectability. To date, however, the ellicacy based perspective has proved lar more successlul in accounting lor the diversity 01 natural signal lorm. This suggests that although selection lor reliability may lavour the evolution 01 more costly, elaborate displays, it has lillie innuence on the detailed design 01 a signal. Rather, it is the environment in which communication lakes place, and the audience to which a display is direCled, that exert the strongest influence on its appearance (Guilford & Dawkins, 1991). Animal signals, in other words, must be both ellicient and reliable, bUl it is the lormer condition that places the greatest constraint on their design. The issue is, however, complicated by the lactthat dillerent selective pressures may come into play at dillerent times during the evolution of a trail. Studies 01 sensory exploitation, lor example, have revealed that signals can arise as a means to exploit a pre-existing bias in receivers (see Section 7.2.3). In such cases, strategic considerations are irrelevant 10 the origin 01 the display. Once signallers have begun to take advantage 01 the bias, however, receiver selection wiU start to playa role. Strategic considerations may thus be of great importance with regard to the maintenance of the display, as a preference that leads to inappropriate responses could prove short-lived (see Arak & Enquist, 1995; Krakauer & Johnstone, 1995). Consider, for example, Ryan's (1983, 1985,1990; Ryan & Rand, 1990, 1993) studies 01 sexual communication in the tungara Irog. Physalam,us pllsrllloslis (discussed in more detail in Chapter 8). Females 01 this species prefer male calls containing low-frequency chucks, a bias that appears to have evolved prior to the chuck itself. Although the preference cannot originally have served an adaptive role in mate choice, Ryan has shown that in current populations it results in females mating with larger males that lertilize more of their eggs. The maimenance of the bias may thus depend on the lact that the preferred trait provides reliable inlormation about size.
7.4.1 Unanswered questions As the above discussion indicates, there are many aspects of signal evolution that are still poorly understood. II seems appropriate, therefore, 10 end lhis chapter by drawing allention to some 01 the queslions that remain unanswered. Two issues that clearly illustrate the wealth of possibilities for future research, bOlh empirical and theoretical, are the use and evolution 01 multiple signals, and the consequences of competition bel ween signallers lor information transler during communication.
EVOLUTION OF ANIMAL SIGNALS
175
Multiple signals Why are animal displays so complex? Many displays, in contexts ranging from courtship and mating (Meller & Pomiankowski, ) 993; Johnstone, 1995a) to parent-offspring communication (Kilner, 1995), involve a number of distinct signal components, and may often combine different sensory modalities. One striking example was mentioned earlier: male Jackson's widowbirds employ a suite of sexual signals comprising an elongated tail, performance 01 a jump display and construction of a courtship bower (Andersson, 1989, 1991, 1993). To date, the evolution of multiple signals like these has received little attention. Models of signalling, [or example, have typically assumed thaI only a single form o[ display is available to signallers; it is only recently that they have begun 10 address the possibility of mullicomponent displays (Schluler & Price, 1993; Johnstone, 1995b, 1996b; Iwasa & Pomiankowski, 1995). Equally, empirical data regarding the way in which receivers assess multiple signals are scarce (although see Zuk it al., 1990, 1992; Andersson, 1991; Kilner, 1995; Scheffer it al., 1996 [or examples). Explanations for the evolution o[ complex displays [all into two main categories: (i) those which suggest that multiple display components serve to provide Ihe receiver with additional in [ormation about the signaller; and (ii) those which suggest that Ihey serve to [acilitale delection, or to take advantage o[ arbitrary preferences. Theories of Ihe former kind include the 'back-Up signal' hypothesis, which proposes that multiple signals allow more accurate assessment o[ one aspect o[ the signaller's condition (Schluter & Price, 1993; lwasa & Pomiankowski, 1995; Johnslone, 1996b), and the 'multiple message' hypothesis, which proposes that dillerent signals convey information about dillerent aspects o[ condition (Johnstone, 1995b, 1996b). Theories o[ the latter kind include the suggestion that multiple displays have evolved to exploit pre-existing sensory biases (Ryan & Rand, 1993), or (in the context o[ sexual signalling) Ihat they are the product o[ Fisherian runaway evolution (Heisler, 1985; Tomlinson & O'Donald, 1989; Pomiankowski & Iwasa, 1993; see also Chapter 8). Un[onunately, while theoretical analyses suggest that all these explanations are plausible. there is little empirical evidence against which to test them. Meller and Pomiankowski (1993), in a comparative study (based on measurement o[ museum spedmens) o[ avian taxa with and without apparent multiple male [eather ornaments, round that ornament size was negatively correlated wit h asymmetry only in those spedes with a single display trail. They interpreted this as evidence that single ornaments are expressed in a condition-dependent manner, whereas multiple ornaments are l1ot, and suggested that the latter were most likely to be the product o[ Fisherian runaway evolution. However, although their results provide some suppon [or 'uninformative' theories o[ mulliple signal evolution, evidence based on correlations between measures o[ quality (such as ornament symmetry) and display can be problematic (see
176
CHAPTER 7
Johnstone, I995a). Furthermore, experimental studies of individual species seem to paint a more complex picture. In some cases, experimental evidence suggests that multiple signal components provide linle information, but serve instead to render a display more detectable or stimulating to receivers. Consider. for instance. a recent study by Scheffer el al. (1996) of sexual signalling in the brush·legged wolf spider Schizocosa ocreala (Araneae: Lycosidae). The male courtship display of this species involves both visual and vibratory dements: synchronized tapping, waving and arching of the forelegs, which are decorated with a conspicuous tuft o[ bristles, and production of substratum-coupled vibrations by stridulatory organs in the pedipalps. Mating experiments in which a female was offered a choice between two males, one whose foreleg tufts had been shaved off and one who was intact, suggested that the visual display component does not playa role in mate assessment. Females were most likely to mate with the individual that first captured their attention. regardless of his condition. However, the tufts may serve to render the courtship display more delectable in situations where vibratory communication is constrained. When females were paired with single
males in an artificial arena that eliminated vibratory communication, they responded less often to individuals whose tufts had been removed. Since the complex leaf-liner habitat of S. ocreara is ill-suited to the transmission of vibratory signals, the visual display elements may often be necessary for effective communication at a distance. In this regard, it is interesting that males of the congeneric species S. rovneri. which occupies compressed leaf-litter habitats better suited to Vibratory communication, lack foreleg tufts and have a much simpler visual display than S. ocrtala. In other cases, however, multiple display components do seem to provide information about the condition of the signaller. A good example is provided by the courtship display of male guppies (Poecilia reliculala). As discussed in Section 7.2.3, males of this species exhibit complex polymorphic colour patterns that play an important role in mate choice, with females typically favouring males that have more carotenoid and structural colour patches, and those that contrast more strongly with the visual background (Kodric-Brown. 1985; Houde, 1987; Long & Houde, 1989; Houde & Endler, 1990). Colour is not. however, the sole determinant of male mating success. Several studies have round that male display rate also influences mate choice (Farr. 1980; KodricBrown, 1993; Nicoletto, 19931, with females favouring males that exhibit higher rates of sigmoid display (a rapid, highly energetic 'S-shaped' movement of the male's body). Moreover, both ornamentation and display activity are related to male condition. as measured by sustained SWimming performance, by the relationship of body mass to length and by parasite load (Kennedy el al., 1987; Kodric-Brown, 1989; McMinn. 1990; Houde & Torio, 1992; Frischknecht, 1993; Nicoletta, 1993). Experimental infection of males with the (naturally occurring) gut nematode Camallanus cotti, for instance. resulted in lowered display rates and reduced attractiveness to females in choice tests (Kennedy et al., 1987;
EVOLUTION OF ANIMAL SIGNALS
177
McMinn, 1990), while lemporary infeclion wilh Ihe (naturally occurring) monogenean ectoparasile Gyrodactylus tumbulli led 10 the orange carolenoid SpOIS of male hosls becoming paler and less saluraled, with similar consequences for Iheir anracliveness (Houde & Torio, 1992). fn guppies, Iherdore, mulliple sexual display components do appear lo provide females wilh informalion about Ihe displaying male (see JohnslOne, 1995a, for addilional example, of Ihis kind). Given the constrasling resulls of differem studies, more data is needed regarding Ihe occurrence of complex displays. and Ihe way in which receivers assess their various componel1ls. Only wilh addilional informalion of Ihis kind will it prove possible 10 lesllhe predielions of the various models and hypotheses discussed above.
Multiple signallers A second topic Ihal is poorly understood al presem is the inlluence of competilion among signallers on informalion Iransfer during communicalion (see Godfray, 1995a,b). The slralegie view discussed in Section 7.3 suggests Ihal receivers can acquire reliable information about signallers by focusing on costly displays that only superior or more needy individuals can afford to use. However, where several signallers vie for anemion, competition has Ihe potenlial to imerfere with receiver assessmem of signaller condition. To date, this issue has received mosl a([enlion in the context of chick begging (see, e.g. Godfray, 1995b; Kilner. 1995; Kacelnik et 01., 1995; Price et 01.. (996). As discussed in Sections 7. }.3 and 7.3.4 Ihe solidlalion behaviour of neslling birds appears (alleasl in some cases) to vary in relation 10 Iheir level of hunger, and hence 10 provide parems with informalion Ihal allows them 10 dislribute food adaptively among the brood (see, e.g. Redondo & CaSlfo, 1992). Compelition between brood members may. however, confound Ihe relalionship between signal and Slate, a possibilily thaI is nicely illustrated by a recent study of begging in canary (Serinus cananus) chicks (Kilner, (995). Observation of visits by parent canaries 10 artificial nests revealed that nestling position relative 10 Ihe parem strongly inlluenced food distribulion, with Ihe chiek whose mouth was c10sesl 10 the parent tending 10 be fed lhe mos!. Proximily, in lurn, was inlluenced by size and by hunger. Because of a hatching spread of approximalely 36 h, chicks in Ihe experimemal broods (comprising Ihree individuals) could be ranked in a clear size hierarchy, and larger nestlings tended bOlh 10 maintain a position closer to the parem and to obtain more food. Experimemal manipulation of Ihe level of food deprivalion, however, could temporarily over-ride Ihe effects of Ihe size hierarchy, and led 10 the mosl deprived chick being fed most ollen. Chick posilion thus appears 10 provide parem canaries with some infonnation aboUl the condilion of their young, allhough the efleas of hunger are panially masked by the competilive advamage that larger chicks enjoy in joslling for position. Possibly as a resull
178
CHAPTER 7
of this masking. it appears that parents also respond 10 other cues in addition to chick position when allocating food. When the positions of the canary chicks were controlled experimentally (by use of Plexiglas panitions within the nest). food-deprived nestlings were still favoured. suggesting that parents responded to additional faclors such as chick posturing and possibly calling. Given the potential impact of sibling competition on parental allocation of resources, one might expect that individual nestlings should adjust their behaviour in response to the level of competition with which they are faced. A recent analysis of competitive solicitation by Godfray (l995b), for instance, which considers the division of a fixed quantity of resources between two chicks, suggests that each should adjust its level of solicitation in relation to both ils own condition and that of its nestmale. The model predicts. for example. that an individual should beg more intensely when paired with a hungrier competitor (who should tend to beg more than a well-fed chick). A study by Price et al. (1996) of the begging tactics of yellow-headed blackbirds (Xanthocephalusxanthocephalus) provides empirical suppon for this idea. Individual chicks (of similar size) were paired, in artificial nests, with nestmates of four types: (i) a big satiated nest mate; (ii) a big hungry nestmate; (iii) a small satiated neslmate; and (iv) a small hungry nest mate (hunger levels having been experimentally manipulated). Both the size and the hunger of the nestrnate were found to inOuence various aspects of the vocal begging behaviour of the experimental subjects. Chicks begged significantly more loudly. more vigorously and for longer when their nestmate was big rather than small. and for longer when their nestmate was hungry rather than satiated (and Ihe nestmates themselves begged more vigorously when hungry). Given the evidence that competition can affect both the begging behaviour of chicks and their success in obtaining food from parents, models of parentoffspring communication (and of olher forms of signalling) need to follow Godfray's (1995b) lead in allowing for interaction between signallers who may differ in competitive abilily. as well as in their need for additional re ources. Only by taking into account the complexities revealed by empirical studies of communication will it prove possible to obtain testable predictions that apply to real biological signalling systems.
Chapter 8 Sexual Selection and Mate Choice Michael J. Ryan
8.1 Introduction Sexual selection poses a simple but central question that Darwin addressed initially in On the Origin of Species by Means of Natural Selection (1859) and more extensively in The Descent of Man and Selection in Relation to Sex (1871): why do the sexes differ and why is the male usually the more elaborate sex? Darwin was concerned about the evolution of sexual dilferences other than those of reproductive organs, which he and others referred to as secondary sexual characters (e.g. see Fig. 8.1). Darwin was convinced that natural selection alone could not bring about such differences. but instead posed an alternative selection force. sexual selection, which •... depends on the advantage which certain individuals have over others of the same sex and species solely in respect of reproduction' (1883. p. 209. 3rd edn). Sexual selection. Darwin suggested. can favour the evolution of traits useful in combat between members of the same sex. or of traits that increase the allractiveness of individuals to members of the opposite sex. It is the laller subject I treat here - sexual seJeclion and mate choice. Darwin's simple enquiry has given rise to one of the most active disciplines in behavioural ecology and evolutionary biology (Gross. 1994). This field has anraCled the allention of behavioural ecologists as well as those working in population genetics theory. parasitology. neurobiology. molecular genetics. phylogenetics and artificial intelligence. Indeed. one of the more recent developments in this (jeld is its increasingly integrative texture ,see Chapter I). A central concern in sexual selection is understanding how female mating preferences evolve. In many instances it is clear, as a female's mate choice influences her immediate reproductive success, Le. her fecundity. Other times. however. the logic of female mate choice is more opaque. This is especially true in lek or lek-Iike mating systems. In these systems males aggregate when advertising for females. females are unimpeded in their choice of mates. and males are thought to make no contribution to the females' effort besides their sperm. Since males make no contribution to parenting. it had appeared that the choice of a mate would not influence a female's fecundity. Thus. the paradox of the Jek - how can female mating preferences evolve if there is no effect of preference on female (jlness (Kirkpatrick & Ryan. 1991; Ritchie. 1996)? 179
180
CHAPTER 8
Until recently. much of the debate in sexual selection has centred on two hypotheses, runaway sexual selection and good gene, as the only alternatives for the evolution of female maling preferences in lek-like mating syslems. Closer attention to the natural hislory of sexual selection. and a better under·
standing of the neural mechanisms underlying preferences, however, have bOlh shown thaI seleclion can have direct. although previously unappreciated eflects on preference evolution. Furthermore. phylugenetic reconstruction, lahoralory experiments lhat probe preferences wiIh novel stimuli and the use of neural networks 10 model preferences suggest that pre-existing or hidden preferences might play an important role in generating selection on male display traits. Finally. rigorous [ield sludies, seleClion experiments and investigation of molecular recognition systems have all combined to provide strong empirical support for the role of good genes in the evolution of female preferences. The field of sexual seleaion has grown so large we can review neither its breadth nor depth. Fortunately, Andersson (1994) has recently published an excellent. detailed review of most aspeas of the field in his Sexual Selection. Any serious student of sexual selection must read this book. Also. the historical development of sexual seleaion and the debate surrounding female choice are recently chronicled in The Ant and the Peacock (Cronin, 1991).
8.2 Sexual selection by female mate choice Although lhe role of female mate chuice in Ihe evolution of male traits was controversial (Wallace. 1889; Noble & Bradley. 1933; Huxley, 1938; Trivers. 1972; Cronin. 1991), a large number of studies in the past two decades have demonstrated that female mate choice is influenced by variation in male traits (Ryan & Keddy-Hector, 1992; Andersson, 1994). Traits as diverse as the elaborate trains of peacocks (Petrie. 1994), the hyperdeveloped tail fins forming the sword of swordtails (Basolo, 1990a). the intricately constructed bowers of bower birds (Borgia, 1985), the huge eye spans in stalk-eyed mes (Wilkinson & Reillo. 1994; Fig. 8.1), the complex acoustic repertoires of songbirds (Searcy, 1992) and lungara frogs (Ryan. 1985). and pheromones of mOIhs (Eisner & Meinwald, 1995) are all traits that females auend to in deciding Iheir future males. Once the phenomenon of female mate choice became firmly established. however. there immediately raged debates as to the forces that have led to the origin and maintenance of such preferences. especially when these preferences are [or traits that increaSe male mortality. For many years. much of the debate contrasted Fisher's (1930) theory of runaway sexual selection with various manifestations of 'good genes' hypotheses. There has been no final resolution, and additional hypotheses have further complicated malters. As a preview to a more detailed discussion of sexual selection and mate choice, [ will summarize some of the major components of the phenomenon in the context of the different approaches that have been used to study it (Fig. 8.2).
SEXUAL SELECTION
181
Fig. 8.1 A lek of stalk-eyed !lies. A male (lOp) and a group of three female slalk.eyed flies roosting on a root hair. There i..
clear sexual dimorphism in eyt' span. (From Wilkinson & Reillu, 1994.)
Funclional approaches are concerned with the behavioural interactions that bring about mate choice and how this generales selection on male trails and female preferences. Female preferences can guide mate choice towards males with certain trails and thus generate direci selection on Ihese Iraits (Fig. 8.2(i). If there is a genetically heritable component 10 trait variation, Ih<'re will be an evolutionary increase in the preferred trail (Fig. 8.2(ii». Female preferences can also be under direct selection if the preference increases Ihe female's reproduclive success either by increasing her survivorship or fecundity (Fig. 8.2(iii». This could result directly from effects of mate choice, or from pleiotropic effects, such as sensory adaptations for predator avoidance or prey detection, which can then affect mating preferences. Alternatively, preferences can evolve by indirect selection because they are genetically correlaled with the male trait under direct selection (Fig. 8.2(iv». Mechanistic studies allemptto identify the physiological processes underlying mating preferences (Fig. 8.2(v). Many of these studies have concentrated on the neural mechanisms, including peripheral. central and cognitive processes, that gUide female mating preferences. Although physiological processes mighl regulate preferences they probably serve other functions as well. Thus, certain biases in sensory syslems can be favoured by selection outside of the context
182
CHAPTER 8
o O. ( ; ) f9 Or'>
I
I
I
Mechan;sm
I
(v)
OV)1;
iL.......
~m
X
O'
{=~
{=
Y?
I;;I~
-Key:
~I
I=} J=}
selection Iv;) ::;=~ ~I constraints
Iv;)
''>
Behavioural and historical interactions Indirect selection
~ Direct selection P
Female preference
T
Male trait
Ph
Physiological
RS
Reproductive success
mechanisms
Derived state
Fig. 8.2 Some components of sexual selection and approaches to their study. Roman
numerals refer to explanations in text.
of mate choice because they increase female survivorship and thus reproductive success (e.g. Fig. 8.2{iii», but these biases will also have incidental influences on mating preferences (Fig. 8.2(i»). Historical approaches acknowledge that extant trails and preferences have experienced a long history of selection and constraints Ihat inOuence their current expression (Fig. 8.2{vi»). By allempting to reconstruct Ihe historical pallern by which Irails and preferences have evolved, these studies strive to delermine if preferences and trailS coevolve due to genetic correlations or if traits evolve 10 exploit pre·existing sensory biases. The laller case is illustrated in Fig. 8.2{vii). in which the physiological mechanisms responsible for generating the preference existed prior to Ihe evolution of the preferred trail. In such a case the sensory biases did not coevolve with the trait.
8.3 The null model for the evolution of female mating preferences Kirkpatrick and Ryan (1991) reviewed the theoretical models for Ihe evolution
SEXUAL SELECTION
183
•
Mean female mating preference
Ibl
•
2
-----+ ••
ll~ Mean preference
lei
=---------'1'--.
.....•......... -...............
Resistance
.... p
•
Fig. 8.3 Three hypotheses for the evolution of female malin~ preferences. Under direct seledion (a). mating preference genes also affed female survival or fecundity. This determines the eqUilibrium for the preference and, through it. the male trail. In a runaway process (b), the equilibrium curve becomes unstable. Seleclion on the male trait (1) also causes rhe evolution of the female preference through a genetic correlation (2) with lhe resuh that the tfait and the preference are exaggerated indefinilely as they evolve away from the equilibrium curve. In the parasite hypothesis (e) a genelic correlation is established betwt:en resistance to parasites and fem.3le preferences for a male display (rait. Directional evolution of resistance (dashed arrow) results in evolution of the preference. and this causes the preference and trait to coevolve towards greater exaggeration along the equilibrium curve (solid arrow).
184
CHAPTER 8
of female mating preferences, and that general outline is followed here (Fig. 8.3). The assumptions of these models.indude that fitness is measured as the number of o(fspring, and traits and pref~rences are heritable and thus evolve in response to selection. In the simplest or null model, female preferences bias mate choice towards males with certain (raits. Preferences are often for traits that reduce male survivorship, such as the complex calls of tungara frogs which increase predation by the frog-eating bat (Ryan, 1985). But, preferences can counter this natural selection cost by increasing the mating success of males bearing these traits. In this null model the male trait evolves to some compromise between the different optima favoured by natural and sexual selection; that is, a compromise between maximizing a male's own survivorship and his attractiveness to females. Assuming a constant natural selection cost to the male trait, there will be a line of equilibrium that predicts the male trait which evolves when a given preference is exhibited in a population. This line represents the balance between natural and sexual selection and is a crucial component of all the models of preference evolution (Fig. 8.3). The null model assumes, as did Darwin (1871), that the female preference is a constant which predicts trait evolution; neither the null model nor Darwin addresses the evolution of these preferences. A large number of sexual selection studies now address that question. A female's preference might influence her fecundity, and thus her fitness (Fig. 8.3). If so, one would expect rhe preference to evolve under direct selection. Alternatively, there could be no influence of male choice on fecundity, but genetic variation in the male trait and lhe female preference might be correlated statistically, i.e. in linkage disequilibrium. Such a situation would allow both characters to evolve as a correlated response 10 the evolution of the character with which it is correlated. Since the trait is under direct selection by the preference, genetic correlations between trait and preference would cause preferences to evolve under indirect selection. Considering how female preferences can evolve under direct and indirect selection is a useful way to proceed.
8.4 Direct selection on female mating preferences 8.4.1 Fecundity effects as a direct consequence of mate choice Given the debate surrounding the evolution of female choice, it is easy to miss the consensus for the controversy. In many cases a female's mate choice has an immediate effect on her reproductive success; when males offer parental care, defend young or feed their males, prudent mate choice can be rewarded by increased fecundity. In some insects, for example, females feed on the male's spermatophore, and female choice of males with superior spermatophore quantity and quality have increased fecundity (Thornhill, 1976, 1980; Gwynne. 1984; Wedell, 1994). In resource-based mating systems, direct selection on
SEXUAL SELECTION
185
preferences is expected to be the rule rather than the exception. When under direct selection. we expect the female preference to evolve to an optimum that maximizes her fecundity (Fig. 8.3a). The paradox of the lek is that a female's choice of mates does not influence her fecundity. but female mating preferences have evolved under these circumstances. Recent studies have shown, however, that this crucial assumption about leks might not be true. Male choice on the Iek can have an immediate effect on female fecundity. thus preferences might be under direct selection. although in subtle ways. These cases are especially interesting because they suggest that often the lek might not be a paradox. only that researchers have not examined il closely enough. One set of examples suggesting direct selection on mating preferences in leks involves variation in the male's ability to fertilize the female's complete
compliment of eggs. Robertson (1990) and Bourne (1993) both offer compelling evidence for two frog species (Uperolia laevigata. and Ololygon rubra, respectively) that females select mates of a size that maximizes fertility rates. This is probably due to the size difference between the male and female affecting the mechanical efficiency of external fertilization. li'ansmission of parasites or venereal diseases could also generate direct selection on mating preferences if male phenotype reveals parasite load or disease condition. A number of slUdies addressing the good genes-parasite hypothesis (see Section 8.4) have shown that females can sometimes assess male parasite load (e.g. M0l1er, 1990a; Milinski & Bakker. J 990), bUI fewer studies have shown that females aClUally increase their reproductive success by avoiding parasitized males. One such example is the bush cricket (Requena verticalis). Simmons et al. (1994) showed that the male's ability to donate nutrients through the spermatophore was influenced by the presence of gut parasites and. in turn, females mating with more parasitized males had reduced fecundity. Even resource-based mating systems can generate selection on preferences in subtle ways. Eisner and colleagues (reviewed in Eisner & Meinwald. 1995) have shown that in a moth. Utethasia. both adults and eggs are protected against predation by pyrrolizidine alkaloids (PAs). These chemicals are not synthesized by the moth but instead are sequestered from the plants it feeds upon as a larva. When males mate with females they transfer some of their PAs to the female via the spermatophore. and these alkaloids are eventually transmitted to the eggs. Males use pheromones to attract females. the pheromone is derived from these alkaloids. and the presence of this pheromone is necessary for mate attraction. Thus. males advertise their alkaloid load to females and, it is speculated, female choice might be influenced by alkaloid levels. Although there is yet no evidence that females actively choose mates, there is some evidence that multiply inseminated females preferentially utilize the sperm of large males. The cost of searching lor a mate on the lek is an additional opportunity for direct selection on mating preferences (Parker. 1983). In this case. female
186
CHAPTER 8
survivorship ralher than fecundity is affected (see Fig. 8.2(iii). Some studies have shown that females assess mates in a manner consistent with attention to search costs. Female pied flycatchers are more likely to mate with preferred (as shown in other studies) males when males are more clumped in space (Alalalo el al., 1988), and female crickets show increased choosiness in an arena when there are more refugia from predators (Hedrick & Dill, 1993).
8.4,2 Pleiotropic effects on mating preferences A mate preference can evolve under direct selection, but selection might not be involved in male choice. Pleiotropy can have important effects on female mate preferences, and sensory systems used in mate choice seem to manifest such effects sometimes. Sensory systems have undoubtedly evolved under both selection and constraints that can be independent of mate choice (sec Chapter 2). In some cases, sensory systems might be used to locate food, assess habitat potentia!. detect predators and migrate. Selection to enhance performance of any of these functions could have effects on female preference for stimuli associated with mate choice that are best viewed as incidental consequences rather than evolved functions (Williams, 1966). Although such concepts have predecessors in ethology (Ryan, 1990, 1994), the interaction of selection forces and constraints lhat influence the performance of lhe sensory system in different situations has recently been termed sensory traps (WestEberhard, (979), sensory bias (Endler, 1992), and also has been discussed in the context of 'receiver psychology' (Guilford & Dawkins, 1993). All of these hypotheses recognize the variety of factors lhat influence sensory systems and emphasize that sensory perceptions implicated in mate choice, or in other functions, might be influenced by sensory biases that evolved in other contexts (see also Kirkpatrick, 1982). The specific hypothesis that males evolve traits to exploit such pre-existing biases Or preferences is known as sensory exploitation (Ryan, 1990). An example of mating preferences being influenced by selection in contexts besides mate choice is perhaps best illustrated by water mites. They locatc small prey items by detecling water-surface Vibrations, but this results in the females being mated by males that mimic these vibrations (Proctor, 1991). This mating preference for males prodUcing vibrations is an incidental consequence of selection for prey recognition (see also Chapter 7). It has been suggested that selection on anolis lizards (Fleishman, 1992) to be sensitive to patterns of prey movement, and selection on fiddler crabs to detect predators (Christy & Salmon, 1991; Christy, 1995) biases the traits of males Ihat successfUlly attract females. General properties of nervous systems might further affect mating preferences. For example, sensory systems can habituate to repeated stimulation, and some have suggested that songbirds have evolved complex song repenoires to release both male (Hartshorne, (956) and female (Searcy, 1992) receivers from such habituation.
SEXUAL SELECTION
187
Mate choice for the correct species can also have incidental effects on mate choice within the species. Reproductive character displacement has been proposed 10 explain the evolution of enhanced specific differences in male courtship in areas of sympatry relative to areas of allopatry. Preferences as well as traits. however. could exhibit character displacemenl. Recently. Gerhardt (1994) argued that female mating preferences have been displaced in sympatry in grey treefrogs. Such an effect should reduce the error rate of females mating with heterospecifics but. as Gerhardt showed, this can have an incidental effect of generating sexual selection on conspecific calls. 8.4.3 Historical studies of sensory biases and sensory exploitation The sensory exploitation hypothesis predicts that the evolution of sexually selected traits is influenced by these pre-existing sensory biases. This generates a prediction about the historical pattern of trait-preference evolutionpreferences evolved prior to the sexually selected traits - that is distincl from the prediction of good genes and runaway sexual selection in which the preference and trait evolve in concert (see Figs 8.2(vi, vii) & 8.3c). To the extent that this hisLOrical pattern can be reconstructed, one can discriminate between these two sets of hypotheses: sensory exploitation versus indirect selection (good genes and runaway; Ryan, 1990, 1996; Shaw, 1995). although these two processes could interact (see Section 8.7). Several studies offer strong suPPOrt for the sensory exploitation hypothesis. The tungara frog, Physalaemus pustulosus. produces a call consisting of a whine and a chuck. The whine is always present and is necessary and sufficient to elicit mate recognition from females. The chuck is not always produced, but when added LO the whine it increases the attractiveness of the call to females as well as to frog-eating bats (Ryan, 1985). This species is a member of the P. pustulosus species group which contains six species, three constitute one monophyletic group on the western side of the Andes and three another monophyletic group in Amazonia and Middle America (Fig. 8.4). In the AmazonianMiddle American clade. P. puslulosus and an undescribed spedes that is sister to P. petersi add suffixes. while P. petersi does nol. It is not known if these suffixes are homologous; it is equally parsimonious to assume either that these suffixes were derived independently in the two species. or that the suffix was derived once at the base of this clade and subsequently lost in P. petersi. But, any reasonable interpretation suggests the ability to add the suffix was derived after the divergence of the two clades within the species group. Male P. coloradorum do not produce chucks, but when chucks are digitally added to their calls females prefer the conspecific caU with chucks to the normal conspecifie whine. The most parsimonious explanation for the occurrence of a chuck preference in both clades within the species group is that the preference is shared from a common ancestor. which would have had to exist prior to the
188
CHAPTER 8
pusru/osus
peters;
species 8
species b
coloradorum
~ I
pustulatus r:l
I
u
f--l
200 ms
Fig. 8.4 Phylogenetic relationshi!)s of frogs of the P. puslulosus species complex showing the advertisement calls of each spedcs. Call attributes unique to each taxa are indicated in brackels. (From Ryan & Rand. 1993. 1995.) This phylogeny. based on morphology. allozymes and deoxyribonucleic acid (DNA) sequence of the 125 milochondrial gene
(Ryan & Rand. 1995), differs in some details from a previous preliminary phylogeny (Ryan & Rand. 1993), although the interpretations are the same.
divergence of the two clades (Ryan & Rand, 1993). If so, the preference for chucks existed prior to the chucks, and thus males that evolved chucks were favoured by a pre-existing preference. There also are call components that occur in the western Andes clade that are absent in the other clade: a strongly amplitude-modulated prefix in P. pustu/a/us and the ability to rapidly produce calls in groups of two or three in P. c%radorum (Fig. 8.4). Although male P. pustu/osus lack these call components, females prefer conspecificcalls 10 which these componems are added, ollering additional evidence for the role of sensory exploitation. Furthermore, the sensory luning thaI results in female tlmgara frogs preferring lower frequency calls of larger males did not evolve to guide !cmale lungara frogs to larger maks but, instead, appears to be an ancestral trait (Ryan et a/., 1990a), and even 'species-recognition' preferences are strongly inlluenced by what are estimated to be calls of ancestral pecies (Ryan & Rand, 1995). The series of studies by Ba 010 on pre-existing preferences for swords in fishes also offers strong support for the sensory exploilalion hypothesis. Swordtails (genus Xiphophorus) are characterized by an extension of several lower rays of the caudal fin which gives the appearance of a sword. Female X. heJleri prefer males with longer tails (8asolo, 1990a). Platyfish are members of the same genus but tend 10 lack elaboration of the caudal rays. When artificial swords are appended to X. maculatus and to X. varia/liS (Basolo, 1990b, 1995a) females preferred the artificially adorned conspecific males. Since swordtails were thought to be monophyletic. Basolo reasoned Ihat the shared preference for swords between both swordtails and platys resulted from a pre-existing
SEXUAL SELECTION
189
bias for swords. Meyer et al. (1994) challenged this illlerprelation when their molecular phylogeny of the genus suggested that swordtails were not mOIlOphyletic and thus the sword evolved more than once. But, this criticism now seems irrelevant for two reasons. Basolo (1995b) showed preference for swords in the closely related but sword less genus Priapetla, and a recent molecular phylogeny by Borowsky et al. (1995) supports the classical interpretation of the monophyly of swordtails and rejects the hypothesis o[[ered by Meyer el 01. (1994). Although there are other cases in which a phylogenetic interpretation suggests pre-existing preferences for sexually selected traits (Ryan & Wagner. 1987; Proctor. 1992; Christy, 1995) the phenomenon is certainly not universal. For example. Hill (1991) had shown that female preferences for plumage coloration in house finches is under direct selection as brighter males are better fathers. (Hill also speculated that females might gain a genetic benefit for their offspring but this was 1I0t documellled.) There is population variation in the size of a patch of ventral coloration; a population phylogeny suggests that the larger patch size is ancestral, but this reduction in patch size is independelll of any change in preference for larger patch size and brighter males (Hill, (994). This is similar to a situation in swordtails (Xiphophorus). Female X. nigrensis and X. multilineatus prefer the large-size genotype of males (Zimmerer & Kallman, t989; Ryan el al.. 1990b); large size has been lost in X. pygmaeus but females retain the preference for large size (Ryan & Wagner, 1987). In both the finches and swordtails the loss or reduction of a sexually selected trait probably results from increased natural selection against the trait or through genetic drift. In both of these cases the stage would be set for sensory exploitation; if males were to evolve the lost trait again it would be favoured now by the pre-existing preference; of course, natural selection against the trait could still be suffiCiently strong to keep it from evolving. Enquist and Arak (1993) argued that general properties of learning and nalUral selection could result in 'hidden preferences' which could give rise to sensory explOitation. They used neural net models in which the network was trained to recognize a 'species-specific' shape. After the model was trained, subsequent simulations revealed a variety of shapes with which the m'del had no previous experience that were equally or more attractive than the initial shape (but, see Dawkins & Guilford, 1995). Staddon (1975) and Weary el al. (1993) also argued thaI a general property of some kinds of learning might result in greater preferences for unknown stimuli due to a phenomenon called 'peak shift learning'. Direct selection on female mating preferences in resource-based mating
systems has never been controversial and had tended to be overshadowed by the paradox of the lek. A growing number of studies, however, now illustrate that direct selection can be more subtle than previously thought, involving fertility effects, avoidance of sexually transmiut,d parasites and diseases, crossgenerational transmission of chemical resources, search costs and pleiotropic
190
CHAPTER 8
effects of sensory biases. Unlike good genes and runaway sexual selection. there has never been any debate as to whether direct selection could operate. There is a consensus among theoreticians that it can. and there now appears to be a growing body of empirical evidence that it probably does in many situations not previously appreciated.
8.5 Indirect selection on mating preferences: Fisher's runaway sexual selection A lack of direct selection on a character does not exclude that character from evolving. Besides stochastic processes such as drift. characters can also evolve under indirect selection if they are correlated with traits under direct selection. Fisher's (1930) brief discussion of runaway sexual selection is notorious for both its insight and impenetrability. O'Donald (1962). Lande (1981) and Kirkpatrick (1982) presented formal models of Fisher's theory that clearly elucidated some of its basic principles. Again, the question is how do female preferences evolve if the preferences do not resuh in variation in female fecundity? All of the models of runaway assume that there is heritable variation in both trails and preferences. and assonalive mating generates a statistical association between trait and preference. Initially. the assonative mating could come about by chance or. as Fisher suggested. by a fecundity advantage to females exercising the preference. In a simple example (Arnold, 1983). assume that males have short or long tails and females either prefer long tails or ignore tail length when choosing their mates (see Fig. 8.3b). Individuals have genes for both the trait and the preference but only express the character appropriate for their sex. After an episode of mating. the relative mating success of longtailed males will be greater than that of short-tailed males due to the female preference for long tails. Furthermore. alleles that determine tail length and preference for long tails will be more likely to be found together in the same offspring than other allelic combinations of the male trait and female preference; assortative mating results in a genetic correlation of trait and preference. Thus. as the frequency of tail length evolves due to female preference for longer tails. so will the preference itself evolve as a correlated response to evolution of the male trait. The stronger the preference. the greater the evolution of the trait and. through linkage disequilibrium. the greater the evolution of the preference (see Fig. 8.3). This is why the term 'self-reinforcing choice' is often used to describe the runaway process (Maynard Smith, 1978). The term runaway should be reserved for this specific model of trait-preference evolution. and should not be used to merely define the elaboration of male traits. Fisher's theory of runaway sexual selection has been a djffjcuh one to evaluate empirically. Ideally. empirical support would come from data showing lhat male traits evolve under the opposing forces of natural and sexual selection and this results in the correlated evolution of the female preference (see Figs 8.2ilv) & 8.3b).
SEXUAL SELECTION
191
There is no empirical support for Fisher's hypothesis of runaway sexual selection. This might be because the runaway process is very rapid and thus unlikely to be observed. Also, in many of the systems in which sexual selection by female choice has been well documented. it is not feasible to conduct the experiments necessary to validate runaway. In many cases. runaway sexual selection is merely considered the null hypothesis and is accepted if a good genes eHect can not be demonstrated (see above; M011er & Pomiankowski, 1993b); this should not be considered adequate proof of runaway because runaway and good genes are not the only forces that can inlluence female preferences.
8.5.1 Genetic correlation of trait and preference Mos! empirical studies that have attempted to assess Fisher's theory have tested the assumption of a genetic correlation between trait and preference. either by examining patterns of geographical variation in trait and preference or selecting on the trait and measuring a correlated evolutionary response in the preference. Guppies are one of the most polymorphic vertebrates and much of this variation can be partitioned among high- and low-predator populations (Endler. 1983). Stoner and Breden (1988) (see also Breden & Stoner. 1987), Long and Houde (1989), Houde and Endler (1990) and Endler and Houde (1995) all showed that there is a correlation between the average amount of orange in males in the population and the strength of female preference for orange. Houde (1993) has pointed out that other processes besides runaway sexual selection can result in spatial covariation of trait and preference; for example. habitat diHerences that favour transmission of different light spectra. Nevertheless, these studies of geographical variation are at least consistent with a prediction of runaway. Selection experiments have also been used to demonstrate a genetic correlation between trait and preference. and some have shown that evolution of a trait can result in an evolutionary response in the preference (see Fig. 8.2(iv)). The bright red nuptial coloration of male sticklebacks has been part of ethological lore since Tinbergen's (1953) account of red postal vans acting as sign stimuli releasing male aggression. Milinski and Bakker (1990) documented female preference for redder males in sticklebacks, employing some clever experiments in which the male's apparent nuptial coloration was manipulated by changing ambient light conditions. There is considerable geographical variation in the intensity of the male's nuptial coloration (McLennan & McPhail. 1990; Milinski & Bakker. 1990), and this variation in male ornament might be correlated to ornament preference. Bakker (1993) crossed wild-caught male sticklebacks whose nuptial coloration was either intense red or dull. The sons' intensity of red was correlated with that of their fathers'. Daughters of red males preferred red over dull males, while daughters of dull males tended to
192
CHAPTER 8
show no preference for nuptial coloration. Thus, there was a positive genetic correlation between trait and prderence. A genetic correlation between trait and prelerence was also shown in studies of stalk-eyed flies (see Fig. 8.). These small flies have huge eye spans that can exceed their body length. They lorm nocturnal roosts which consist of a single male and several lemales perched on root hairs. Males with relatively larger eye spans are accompanied by more lemales, suggesting the action 01 lemale choice on eye span. Wilkinson and Reillo (994) conducted bidirectional selection experiments on eye span. Alter 13 generations, lemales from long eye-span lines and Irom unselected lines both preferred long eye-span males in mating experiments, but lemales Irom the short eye-span lines prelerred males with short eye spans.
8.6 Indirect selection on female mating preferences: good genes hypotheses An alternative to the arbitrary lemale prelerences postulated by Fisher is a more utilitarian lunction 01 female choice. Zahavi'S'( 1975) handicap principle popularized the notion lhat females preler to mate with males who have demonstrated their superior genetic qua lily lor survivorship. Females can evaluate a male's survival ability, zahavi suggested, by assessing the magnitude 01 the handicap with which he is able to survive. There is no disagreement that many sexually dimorphic traits can increase male mortality, and zahavi suggested that these handicaps to survival evolve as honest signals, allowing lemales to assess male genetic quality; thus, the term lemale choice for 'good genes'. There are several variations 01 zahavi's original handicap proposal (Maynard Smith, 1991; Harvey & Bradbury, 1991).ln the original model, the handicap is fixed and it acts as a liIter to survival. If a male has a large handicap and has survived, he is likely to have high intrinsic survival ability (as opposed to merely good luck). Males without handicaps cannot be judged, and thus are assumed to be ignored by females. A more plausible variant of the original handicap model is the condition-dependent handicap. Here, the investment in the handicap varies with the male's condition such that he is optimizing the trade-all between mate amaction and survivorship to maximize his reproductive success. When zahavi firs! suggested the handicap model it was enthusiastically embraced by many field biologists but harshly criticized by most population geneticists. Early population genetics models were unable to demonstrate the spread 01 alleles for preferences under the original lormulation 01 Zahavi's model. At the risk 01 oversimplilying a quite substantial. technically complicatt'd and emotionally charged area 01 research, it appears that good genes hypotheses lor the evolution 01 lemale preferences gained wide support when it was shown that preferences could evolve il they were genetically correlated to the good gene (Pomiankowski, 1988; Gralen, 1990).
SEXUAL SELECTION
193
The good genes hypothesis, like Fisher's theory of runaway sexual selection, also relies on a genetic correlation, but in this case between the preference and the male's 'good genes', which is taken to mean some heritable component
for viability (see Figs 8.2(iv) & 8.3c). Many slUdies that have alleged to demonstrate female choice for good genes do so based merely on the logic that if females prefer traits that are costly to male survival, these traits are both handicaps and reliable indicators to females of a male's genetic constitution
for survival. Costly displays by themselves should noL be taken as evidence for female choice for good genes. Most male ornaments are used in communication.
A purpose of any signal is to increase the conspicuousness of the signaller against background noise (see Chapter 7). Conspicuousness often incurs an immediate viability cost resulting, for example, from unimended advertisement Lo predators and parasites, Furthermore, all signals involve physiological costs, either in the growth of a structure used as a signal or in displaying the signalit costs to grow a long tail and to make a loud call. Since mate choice involves communication it involves costs. The existence of a costly ornament is hardly
serious evidence that female preferences have evolved to assess a male's genes for survival. Testing the hypothesis 01 good genes requires documeming the effect of a female's mating preference on offspring viability. Ideally, such a study would demonstrate a genetic correlation between a female preference and male genes that inlluence viability, evolution of the viability genes and correlated evolution in the female preference. This is a tall order that is clearly not possible in most systems. There have been a number of studies recemly, however, that have documemed a relationship between female mating preferences and offspring viability that appears to result from a paternal genetic effecl. The focus of most good genes studies can be classified imo four categories: Ii) the general relationship between ornament elaboration and offspring viability; (ii) the hypothesis of female choice for parasite resistance genes; liii) female choice of trait asymmetry as an indicator o[ genetic quality; and Liv) choice lor genetic complementarity. Thc [irst and foorth categories offer some of the most compelling evidence for good genes. The great tits of Wytham Woods near Oxford have been cooperative subjects 01 a large number of slUdies in behavioural ecology, Norris (1993) took advantage of this well-known system to gather some strong empirical evidence for the operation of good genes. Female tits prefer to mate with males having larger black breast stripes. Cross-fostering experiments revealed that stripe size was heritable and there was a significam relationship between stripe size of the putative father and the proportion 01 male offspring surviving from a brood; no such relationship existed between stripe size of the foster father and survivorship, Surprisingly, the survivorship benefits were not shared by female offspring. M011er (1994) also showed that offspring viability was correlated to the degree of male ornamemation in barn swallows, His study was conducted with a
frec~ranging
population and the lack of any influence of parental care
194
CHAPTER 8
on offspring survival could only be argued on statistical grounds and thus is somewhat less compelling than if this variable were experimentally controlled. A potential problem with the studies of tits and barn swallows is that highquality females might have had preferential access to the more attractive males, and thus viability differences among offspring might have been due to maternal rather than paternal effects, although Norris and M011er were unable to detect assortative maling. Petrie (1994) controlled for maternal effects in her study of mate choice in peacocks. Previous studies had shown that the number of eye-spots in the train predicted a male's mating success, and experimental reduction of spot number was shown to influence male mating success between years (Petrie el af.. 1991). In her study, Petrie randomly assigned females their mates, offspring were raised under common conditions and then were released in a park, alleged to reflect almost natural conditions. Male attraction was assessed as the mean area of the father's eye-spot (studies have not shown that this character per se was under selection). There was a significant association between this measure of male attractiveness and size of offspring at 84 days, and also with the survivorship of a sample of these offspring after 24 months. These studies of tits and peacocks offer some of the strongest evidence, to date, that females obtain heritable viability benefits for their ofIspring through mate choice. Other studies of good genes have nol fared so well. For example, many species exhibil a large-male mating advantage (Ryan & Keddy-Hector, 1992; Andersson, 1994). In ectothermic vertebrates, growth in males is often indeterminate, thus larger males might be older and beller able to accrue resources. Howard el af. (1994) tested the hypothesis thai larger male toads (Bllfa americanw;), which are preferred by females, sire offspring with superior larval characleristics - age and size at metamorphosis. Although both traits showed significant heritability among sires, there was no effect of sire size on either age or size of their offspring at metamorphosis. Toads might have good genes bUI females are not selecling them. Other teslS of the good genes hypothesis have concerned more specific effects. Hamilton and Zuk (1982) suggested that bright plumage in birds indicates genetic resistance to parasites and that by preferring brighter males females increase offspring viability due to this inherited resistance. Furthermore, coevolutionary cycles of hosts and parasites should maintain genetic variation for fitness. Initial tests of this hypothesis used interspecific comparisons, testing the prediction that plumage dimorphism (and song complexity) should be greater in species with higher parasite loads. The results were equivocal; some data sets initially supporting the hypothesis did not properly control for phylogenetic effects and failed to support the hypothesis once proper phylogenetic comrols were used (e.g. Read & Harvey, 1989; reviewed by M0l1er, I 990b; Kirkpatrick & Ryan, 1991). Many studies of single species have demonstrated or suggested that male parasite loads can influence female mating preferences (e.g. Clayton, 1990;
SEXUAL SELECTION
195
Houde & Torio, 1992; Zuk er al., 1993). Since parasite load can have a variety of effects on male morphology, physiology and behaviour, it is hardly surprising that parasites influence a male's ability to attract mates; in the most extreme,
females do not mate with dead males. Furthermore, female preference for less parasitized males could be under direct selection, as discussed above. Studies have not shown that female preference for less parasitized males increases offspring survivorship, bUl M011er (1990a) has conducted a series of elegant studies with barn swallows strongly suggt'ting such an effect. Previous studies had shown that females prefer males with longer tails (M0l1er, 1988). In nature, males with longer tails have fewer parasites. In cross-fostering experiments the number of mites of a putative male parent is correlated with the number of miles on his offspring, whether the offspring are raised by him or by a foster father. There was no correlation between mites on the male parent and other 'unrelated' offspring in his nest. Furthermore, by artifidally manipulating mite loads, M011er (1990a) showed a detrimental effect of mites on growth rate. This study supports some of the major assumptions of the Hamihon-Zuk hypothesis: parasiles are detrimental. parasite resistance is heritable and parasite load is correlated to traits preferred by females. Plumage brightness might indicate the presence of one type of good gene, parasite resistance, but some researchers have suggested that a single phenotypic measure might indicate overall genetic quality. Fluctuating asymmetry (FA) is the deviation from symmetry in otherwise bilaterally symmetrical traits. Soule (1982; Soule & Cuzin-Roudy, 1982) theorized that FA is a measure of an individual's ability to develop in the face of environmental and genetic stresses. Thus, il has been suggested that FAs might be reliable indicators used by females to assess a male's overall genetic Quality (M0l1er, 1993; M011er & Pomiankowski, 1993a; Watson & Thornhill, 1994). Female mating preferences are sometimes correlated with male FA. In barn swallows, tail symmetry and tail length interact in innuendng male attractiveness, and parasites can increase tail FA (M0l1er, 1992). Scorpion mes are attracted to a pheromone, some Quality or Quantity of which is correlated wilh FAs of various morphological traits (Thornhill, 1992). Such a correlation between attractive traits and morphological FAs did not hold in a study of mating calls and FAs in cricket frogs (Ryan er al., 1995). If FAs are indicative of male quality then one would expect this information to be especially reliable in sexually selected traits. M011er and Hoglund (1991) compared patterns of sexually selected characters with other morphological characters. Sexually selected characters tended to have higher levels of FA. Also, sexually selected characters showed a negative relationship between character size and FA, e.g. longer tails were more symmetrical. TWs differed from the reIationsWp in other characters in which average size was more symmetrical and asymmetry increased with both positive and negative departures from average. The differences between sexually selected traits and other traits is taken to suggest that FAs are reliable indicators of genetic quality.
196
CHAPTER 8
The slUdies dted above show that there can be preference for symmetry, but that result per se might not be unexpected. It is possible that there is a sensory bias to prefer more symmetrical structures (Jennions & Oakes, 1994; Ryan el al., 1995). Symmelry is one of the 'laws' of Gestalt perception in humans, and symmetrical palterns are more likely to be identified accurately than asymmetrical ones (Pomerantz & Kubovy, 1986). The bias toward symmetry might be even more general. In a pair of SlUdies in which neural networks were trained for arbitrary paltern recognition, there emerged an overall preference for symmetry (Johnstone, 1994; Enquist & Arak, 1994; but, see Dawkins & Guilford, 1995, for a cautionary note). Furthermore, no studies showed either that the preference for more symmelrical mates is under direct selection due to a positive innuence on female fecundity, or that it is under indirect selection via good gcnes due to more symmetrical males fathering more viable offspring. M011er el al. (1995) have shown that fluctuating asymmetry in human breasts is correlated to fecundity - womcn with morc symmetrical breasts are more likely to have children. But, there is only anecdotal evidence that breast symmetry influences male choice of mates. Most studies of good genes are based on females identifying absolutely superior genotypes in males. But, a series of recent studies has examined mate preference based on genetic complementarity. The major histocompatibility complex (MHC) exhibits high levels of heterozygosity - up to 60 alleles at some loci - and this heterozygosity appears to be maintained by selection (Brown & Eklund, 1994). These genes influence immune recognition by coding for ccll-surface glycoproteins. The molecules are characterizcd by a small cleft or basket that contains a peptide or amigen of about 10 amino acids in length. If thc cell presents a 'self-peptide' in this basket it is recognized as self by T cells, but if the cell is infected by a virus a small peptide from this virus is placed in the basket and the cell is thcn auacked by T cells. Brown (1983) speculated that this cell recognition system might also function at the behavioural level in influencing kin-directed behaviours, including inbreeding avoidance in mating (see Chapter 4). In 1976 Yamakazi el al. showed that mate choice in mice was mediated by variation in MHC; there was disassortative mating by haplotype. A series of 'lUdies has further confirmed that mice mate disassortatively by MHC haplotype and this preference is mediated by olfactory cues in the urine (reviewed in Brown & Eklund, 1994). Nor are these results restricted to the congenic laboratory strains used in the initial studies. For example, Pous el al. (1991) showed a 27% deficiency of homozygous MHC offspring in mice under seminatural conditions; al least some of this deficiency could be explained by females seuling on territories of males with different MHC haplotypes, but not by selective fertilization or abortion. In a recent study, POltS el al. (1994) suggest the major advantage of disassonative mating by MHC genes was due to the fitness decline associated with inbreeding rather than resistance to infectiou diseases.
SEXUAL SELECTION
197
The r-complex is another well-known. highly polymorphic genetic system in mice that has been implicated in mate choice. The allelic variants are relerred 10 as either wild lype (+) or '1-' haplotype (I); there are at least I; 1haplotypes. Animals that are homozygous for the same I-haplotype die before birth. Animals that have two different r-haplotypes have more or less decreased survival and males are sterile. LeninglOn el al. (1994) showed that females mate disassortatively by haplotype and their degree 01 discrimination is related 10 the fitness advantage of disassortalive mating (see Chapter 4). These detailed studies 01 mate choice lor genetic complementarity are possible in rodents because of the wealth of information on the molecular genetics 01 the MHC and I-complexes. Other systems. however. as diverse as cnidarians (Grossberg. 1988) and wads (Waldman el al.. 1992) suggest similar palterns.
8.7 Interacting forces Direct selection. runaway selection and selection lor good genes are often portrayed as mutually exclusive hypotheses for the evolution ollemales preferences. This need not be the case. In fact. Fisher (1930) suggested that the runaway process could be initiated by females preferring males with higher survival abilities. There are numerous scenarios by which selection forces could interact to influence the evolution 01 female preferences. In most species. females have strong preferences for conspecifics over heterospecifics. This is because direct selection will lavour conspecific preferences given the reduced lecundity typically resulting from hybrid matings. These conspecific mating preferences could also result in sexual selection on males. Females might prefer those males that least resemble heterospecifics. despite any genetically based differences in the viability 01 conspecific males that would be correlated with their resemblance to helerospecifics. This could lead 10 assorlative mating. linkage between preference and trail genes and. ultimalely. runaway sexual seleclion. Thus. direct selection for species recognition could generate a bout of runaway (Ryan & Rand. 1993). Searcy (1992) provides another. more specific example of how diflerent forces can interact to influence preference evolution. Because of inherent processes of the auditory system. such as habituation. Searcy posited that male songbirds have exploiled a pre-existing preference lor complicated acoustic stimuli. Thus. preference for song repenoires did not coevolve with repenoires. as would need 10 be the case if these preferences evolved under indirect selection. as by runaway or good genes processes. Among songbirds studied. however. there is a positive correlation between strength of preference and repertoire size. Searcy's conclusion is Ihal song repertoires were initially favoured by a pre-existing preference for complex auditory stimulation. but larger repertoires and the preference for them coevolved through a runaway or a good genes processes.
198
CHAPTER 8
In the above examples, diflerent forces act sequentially on preference evolution. Some population geneticists have recently allempted to ascertain the relative strengths o[ these dillerem forces when they act simultaneously. Kirkpatrick and Barton (in press) provide an equation that allows an estimate 01 the impact of good genes preferences on the evolution o[ the overall mating preferences. Evolution is quamified as the change in mean preference as measured in phenotypic standard deviations (AI). The parameters of interest are the square root of the heritability o[ the trait (!l.r), the heritability of the preference (h' pI, the phenotypic correlation between the female's preference and the trail o[ the male she actually mates in nature (PPT)' the correlation between the male's trait and his genetic quality for survival (rTW ) and a measure o[ the genetic variance for IOtal fitness (G): AI: 'p,.,.rTWh!'pJGw
There are no single mating systems in which there are su!liaent data to solve this equation. But, some of these parameters have been measured, and these can be used to estimate the upper limits o[ indirect effects (e.g. Bakker, 1990; Bakker & Pomiankowski. 1995; Pomiankowski & M0ller, 1995). KirkpatriCk and Barton (1996) used estimates of ..JGw (0.25), hT (0.7) and h'p (0.4) from the literature. Furthermore, they assumed that the trait was a perfect indicator of male fitness (rTW : 1.0) and that females always mated with males that perfectly matched their preference (PPT: 1.0). Given these eSlimates, the change in preference is only 3.5% of the preference's standard deviation per generation. This effect is small relative to the effect Ihat other selection forces are known to have on phenotypic change (Burt, 1995). Realistically, il will be even smaller in nature since it is unlikely that the male trait will indicate perfectly his fitness and that female preferences and phenotypes of mated males are not compromised by less than perfeci female discriminabilily or alternative male mating strategies. This preliminary analysis suggests thai preferences for good genes work, but when competing with direct selection they do not work very well. This analysis also provides a tractable method for measuring the strength o[ indirect and direct selection on preferences in real mating systems. Although debates about preference evolution have previously contrasted mutually exclusive hypotheses, there is lillie doubt that multiple forces interact in their influence on the evolution of preferences. Two challenges are 10 ascertain how these forces might act sequentially through a species' hislOry and the relative importance of each force when they act simultaneously.
8.8 Multiple traits and their preferences A limitation of most treatments o[ sexually selected traits is the locus on single traits. In many cases, however, males have suiles o[ sexually dimorphic traits thaI appear 10 be under sexual selection, such as the myriad of dillerent colours and shapes o[ spots than adorn male guppies (Endler & Houde, 1995). Some
SEXUAL SELECTION
199
sludies have demonstrated empirically that females do attend to more than one trait (Zuk el al., 1990; Kodric-Brown, 1993; Collins el al., 1994; Brooks & Caithness, 1995; Endler & Houde, 1995). M011er and Pomiankowski (1993b) indicated that multiple sexual ornaments are more likely in lek breeding and other polygynous species of birds while single sexual ornaments seem to be the rule in monogamous species. To what extent might runaway sexual selection and good genes preferences account for these differences? M011er and Pomiankowski measured the length of sexually dimorphic feathers in both lekking and monogamous birds to estimate the degree of sexual ornamentation. They assumed that traits wilh larger asymmetries were honest signals of good genes, and preferences for such traits evolved for choosing good genes. Furthermore, they assumed that if a trait had a low level of FA then both the trait and the preference for the trait evolved as a result of runaway sexual selection; in Ihat sense, runaway is the null hYPolhesis. Their results showed that single sexual ornaments had higher levels of FAs while multiple sexual ornaments had lower levels, suggesting that in monogamous species preferences evolved for indicators of good genes while in lekking and other polygynous species traits and preferences evolved by runaway sexual selection.
One potential problem with this interpretation of multiple trait evolution is the assumption Ihat preferences can only evolve for good genes or by runaway sexual selection. It is known, however, that a variety of other forces such as direct selection on preferences and pleiotropic effecls can inOuence preference evollllion. Another problem is that FAs are greater in monogamous species, in which males contribute substantially to offspring survival. Thus, FAs might be more indicative of male paternal quality rather than genetic quality. If so, the preference is under direct selection. The patterns of feather FAs and numbers of sexual ornaments found by M011cr and Pomiankowski is in accord with recent theoretical predictions by Pomiankowski and Iwasa (1993; Iwasa & Porniankowski, 1994). They modelled Ihe evolution of preferences for multiple sexual ornaments under a good genes and a Fisherian process. Under good genes processes, multiple preferences can evolve only if the joint cost of choice based on multiple traits is small. If there is an increased cost to assessing multiple traits then only a single preference is stable. Interestingly, the establishment of a single prelerence can preclude the later evolution of preferences for traits that are even better indicators of male genelic quality - prelerence for good genes precludes preferences for better genes. The cost of choice does not have such an effect on preferences that evolve by runaway sexual selection. These theoretical results thus suggest that preferences for good genes is unlikely to explain the evolution of multiple sexual ornaments, and thus do not explain preferences in lek-like mating systems. BOlh the empirical and Iheorelical treatments suggest a dire need for more data. Although the exislence of multiple sexual ornaments suggests multiple
200
CHAPTER 8
prderences, this crucial assumption needs to be determined empirically. Mullivariate statistical analysis is one approach, but a more satisfactory one is experimental manipulation of independent traits. Thls has always been tractable in studies of acoustic signals (e.g. Gerhardt, 1992), but recent advances in video animations show this is also feasible in visual communication systems as well (Clark & Uetz, 1993). Such studies would not only allow precise demonstration of mulliple preferences. but would also show how strengths of preferences might vary as a function of traHs that have been hypothesized to evolve under diUerent types of preferences. The assumption that trait asymmetry reveals why the preference for such traits evolved is also an assumption that needs to be evaluated critically. Understanding the evolution of preferences requires study of not only traits but traits and preferences (see Chapter 7 for further discussion o[ the evolution of multiple signals).
8.9 Mate copying: a new horizon? O[ the newest developments in sexual selection studies, many o[ which could not be covered in this chapler. mate copying might be one o[ Ihe more exciting. Most considerations o[ male choice assume thaI a female's mating preference is intrinsic, either because it has a strong genetic component (Bakker & Pomiankowski. 1995) or because il is learned in some early, sensitive period o[ development (Marler, 1991), and thaI its expression is independem o[ the chnice o[ others. Recent studies have now shown Ihal mating preferences can be innuenced by social cues; specifically, it has been suggested that females copy the mate choice o[ olhers (Gibson & Hoglund, 1992). Mate copying has been supported by a variety of observations of birds (Hi)glund el al.. 1990; Pruelt-Jones, 1992), but it has been perhaps most clearly demonstraled in guppies. In a series o[ laboratory studies Dugatkin (Dugatkin & Godin, 1992, 1993; Dugatkin, 1996) has allowed a female guppy to choose between a pair of males. This focal female is then constrained in a SitlJalion in which she observes the previously unpreferred male consorting with a model female and the male she preferred wilhout a female. When the choice test is repeated, females show a significam change in their preference, increasing the time spent with the previously unpreferred male. Dugatkin showed that younger females arc more likely to copy the choice o[ older females, and that to some extent [emale copying can reverse genetically determined preferences [or the amount o[ orange coloration. Mate copying also can occur between species. Schlupp el al. (1994) studied mating preferences in a sexual-asexual species complex o[ mollies, the saillin molly, Poeci/ia lalipinna and the Amazon molly, P.formosa. The Amazon molly is an all-female gynogenetic species; it reproduces clonally but must rely on sperm [rom heterospecilic males, in this case sallfin mollies, to initiate embryogenesis. Why should male saillin mollies male with Amazon mollies? The assumption is that such a mating incurs only costs and no benefits for the male. Schlupp
SEXUAL SELECTION
201
el 01. showed, however, that female saillin mollies exhibit mate copying even
if the female is an Amazon molly. So, by mating with Amazon females, sailfin mollies increase their attracliveness to their own females. Studies of mate copying arc in their infancy and a number of questions demand attention. Although mate coping occurs in the laboratory how common is il in nature? Why du females copy; is there an advantage to copying per se (e.g. reduced search costs) or is Ihis a specilic example of a more general social facilitation of behaviour (e.g. conspecific cueing; Stamps, 1991)? To what extent can copying reverse strong intrinsic preferences among conspecifics and beIween conspecifics and heterospecilics? Is there a copying 'preference function' Ihat can evolve in response to selection? How does copying influence variance in male maling success and correlated evolution of trait and preference (e.g, see Pruett-Jones & Wade, 1990; Kirkpatrick & Dugatkin, 1994)?
8.10 Summary Sexual selection studies have moved from the single debate over Fisher's runaway sexual seledion versus seleClion for good genes. Direct selection had
always been considered an importanl influence on preferences in resourcebased mating systems. bUI this fact was not relevant to attempts to understand Ihe extreme dimorphism found in lek-Iike species. Now it is clear that direct seleclion acts in lekking species as well, either by subtle but important influences of males on female fecundity or by direct selection on sensory systems thaI have pleiotropic effects on mate choice. The former effect has been revealed by more detailed appreciation of the natural history of mating systems and the laller by increasing our understanding of mechanisms guiding mate choice. Both of these avenues of research offer promise for future studies. Recent studies have probably only scratched the surface in revealing the subtle influences that mate choice has on female fecundity. Future sludies should continue to investigate in great detail the varielY of means by which female fecundity can be subtly affected by mate choice; parasite and disease transmission are likely 10 have important consequences. Also, as fulure studies utilize more sophisticated analyses
or
the sensory and cognitive processes
underlying mate choice. we will have a better nOtion as 10 the degree to which selection in other contexts influences mating preferences. or to which mating preferences and other functions can be compartmentalized and thus each possibly oplimized. There is now a consensus that in theory both runaway sexual selection and good genes can work. wilh both relying on genetic correlations between male characters and females' preferences. There is some strong support for the genetic correlations belween trait and preference. and a major advance is the demonslration that male traits can influence heritable variation for viability in offspring, either because of the male's superior or complementary genotype. An interesting new avenue of research is quantifying the relative consequences
202
CHAPTER 8
of direct and indirect selection when the two interact in preference evolution. For much of the history of sexual selection, theory and empirical sludies have been conducted with lillie interaction. There would be great benefits derived from an empirical system that would allow, for example, the precise measurements required to estimate the interaClion of direct and indirect selection on the evolution of traits and preferences. Although there have been a few empirical studies demonslrating selection fur good genes, soch analyses would also be fadlitated by more tractable model syslems. Development of such empirical systems should be a high priority. Finally, historical approaches have introduced a new dimension in attempting to reconslruct the history of trails and preferences. These studies suggest that in some cases female preferences are broader than the phenotypes exhibited by their males, and thus represent a selection force that will act on novel male traits, eilher positively or negatively, as they arise. The modern phylogenetic approach (see Chapter 14) has had an important influence on much of biology and this will continue to prove true in sexual selection as well. Future slUdies in sexual selection, and in behavioural ecology in general. should continue to add a historical analysis using the most sophisticated analyses available. This will undoubtedly require extensive collaboration with phylogenelicists. The major advance in sexual selection is one of approach. There is a growing recognition that mate choice is a complex phenomenon Ihat not only can influence male and female fitness, but also is guided by sensory mechanisms whose history has been characterized by a variety of selection forces and constraints. This integration is leading to a more general appredation of the biology of sexual selection.
Chapter 9 Sociality and Kin Selection in Insects Andrew EG. Bourke
9.1 Introduction This chapter deals with what the social insects can teach us about social evolution and kin selection, since it is in these areas that social insect research has made some of its most fundamental and exciting contributions to behavioural ecology as a whole. To begin with, it examines the classical problem of the origin of eusodal societies. In the traditional definition, such societies are those wllich. alongside cooperative care of the brood and an overlap of adult generations. exhibit a reproductive division of labour (Wilson, 1971). This means that some members are specialized for reproduction (queens or kings), whereas others devote themselves to foraging, nest construction, defence and brood-rearing (workers). Since workers sacrifice their own offspring production to help boost that of others, they are said to exhibit reproductive altruism (e.g. Trivers, 1985). The second topic covered by this chapter is the evolution of a stable reproduaive skew (sharing of reproduction in multiple-breeder groups). The final part discusses sex ratio evolution and conflicls of interest within eusocial colonies. Further coverage of the perennially fascinating subject of insect sociality may be found in the reviews by Wilson (197 t). Hamilton (1972), Lin and Michener (1972), Alexander (1974), West-Eberhard (1975). Crozier (1979), Starr (1979), Andersson (1984), Brockmann (1984), Alexander el 01. (1991), Seger (1991), Krebs and Davies (1993) and Crozier and Pamilo (1996). A specific reason for concentrating on these three topics is that kin-selection theory may be applied and tested in all of them. In addition, each is relevant to the general study of behavioural ecology. For example. the evolution of social behaviour and the sex ratio represent active areas of study in all sort, of organisms (e.g. see Chapters 10 and 13). Moreover, theories of the evolution of a stable reproductive skew have been applied to birds and mammals with cooperative breeding (see Chapters 10 and II; Keller & Reeve. 1994). Kin conflict is likewise a universal concept, applying across all scales of organization from the intragenomic evelupwards (e.g. see Chapler 12; Ratnieks & Reeve, 1992; Godfray. 1995a; Maynard Smith & Szalhmary. 1995). Another reason for focusing on kin conflicl is 10 stress that kin selection underpins competition between relatives as well as cooperation among them (Trivers. 1974; Trivers & 203
204
CHAPTER 9
Hare, 1976; Seger, (991). According 10 the theory, kin groups, like a family gathered for Christmas, are forever uneasily poised between twin impulses of cooperation and conniC!.
9.1.1 Which are the eusodal insects? Most eusocial insects belong 10 two orders, the Hymenoptera (ants, bees and wasps) and the Isoptera (termites). All ants and termites are eusociaL but many bees and wasps are solitary or exhibit other grades of social organization (Wilson, (971). A number of aphids (Hemiptera) are arguably also eusocial (Aoki, 1977; Benton & Foster, 1992), as are some beetles (Kent & Simpson, (992) and thrips (Crespi, 1992). How 10 define eusociality has become a source of controversy (Gadagkar, 1994; Crespi & Yanega, 1995; Sherman et al., 1995). However, the essential point from the Viewpoint of kin-selection theory is that all eusocial societies, wherher broadly or narrowly defined, exhibit some degree of reproductive altruism. Some of the other key fealUres of the eusocial insects are outlined in Table 9. I.
9,1.2 Kin selection and Hamilton's rule The idea at the heart of the modern understanding of the evolution of altruism and sociality is Hamilton's (I 964a,b) theory of kin selection (West·Eberhard, 1975; Michod, 1982; Trivers, 1985). The theory (also known as inclusive fitness theory) concerns the conditions under which genes for social actions spread through pOJlulations. Social actions are ones that increase or decrease the offspring production of conspecifics. Hamilton realized that their evolution would be affected by relatedness. Imagine an altruist helping to rear b (for benefit) extra offspring of another individual (the beneficiary), while incurring a loss of, (for 'OSI) of its own offspring. According to Hamilton's rule, the gene for altruism undergoes selection if the condition rib - r,e > 0 is satisfied, where r, and r, are the altruist's relatedness to, respectively, the beneficiary's offspring and its own offspring. West-Eberhard (1975) shows how this form of Hamilton's rule easily converts to another common version, 'help if rb -, > 0', where r is the relatedness of the altruist to the beneficiary.
Relatedness, under its formal definition, is a regression coefficient (Grafen, 1985; Table 9.2). Informally, it measures the probability, over and above the average probability (which is set by the gene's average population frequency), that a gene in one individual is shared by another. This allows us to see intuitively why Hamilton's rule works. Altruism can evolve because an altruistic individual can more than make up for losing c offspring with a chance r, of bearing the gene for altruism, if it adds to the population b individuals with a chance r, of sharing the gene, where rib exceeds r,' (or r,b - r,' > 0). In short, the gene for altruism spreads because it promotes aid to copies of itself (Dawkins, 1979; Grafen. t99t).
KI
SELECTION IN INSECTS
,
c
z
z
~ c c
<
, 1 o
i o
t
,E
~
5 z
.~
i'
205
IV
Table 9.1 (Continlled)
a a-
Colony site
Degree of Number of
Group
eu§OCial specks
Range
Typical gt'nera
Tropics
Mrlipona,
quC':en-worker
Number of
caSlt'
Modt'of colony founding
dimorphism
qut'ens in mature colony
Wax and resin comb In tree
Fission
High
On.
Soil mounds: tunnds In soil and wood; plillll
Haplometrosis.
High
One-many
High
Ont'-many
High
On.
Nest
type
Totally 5U:'rlle workers in
to nl~art'st ordN 01
n ::t:
»
magnitude
Life cycle
some species?
100-100 000
Perennial
No
tT1 ::< />
Perennial
V'
'
Perennial
V'
AnnulIl.
Yo,
.'
-J
StIngless bees (MelijlQninael AnlS
m
Trijon4
8804-20000
Worldwide
{Formlcidae)
AtloR'. Campemma.
feiton, FontlicQ,
uuiru. uplothorax.
~11'1ll';
10-10000000
plromctrosis.
blldding. !lssion
('IInnn
Linrpilf1tmQ, MjlTJIf;(4,
(HeaphyI/o,
Phridolr.
So/mopsi! Isop/fra
Termites
2200
Marr'ttrmrs. Nasutitrrmrs
Soil mounds; tunnels in soil and wood; carton
SouthcaSI Asia. NOrthern
Co/'phitla.
Plant galls
hemifPhcrc
PsruthJrr~ma
TropiC'S
Amittrffln,
Kafoumlts.
Hapto01etrosis.
100-[ 000000
pleomeuosis
Htmipttra Social arhid~ IHonnaphidldae. remphigidael
43
Prmphi~us.
Haplomeuosis
10-100000
perennial?
Colonies may be founded by a single queen (haplomelrosis), by multiple queens (pleometrosis), by a single queen plus workers (fission), or by multiple queens plus workers (budding). Paper and cartOn are nest materials manufactured from chewed plan! maller; a comb is an array of cells. (Prom: W.A. Poster, personal communication (social aphids); Wilson, 1971; Brockmann, 1984; Winston, 1987; Bourke, 1988; Myles & Nutting. 1988; Engels. 1990; Holldobler & Wilson. 1990; Seger, 1991; Ross & Matthews. 1991; Keller & Vargo, 1993; Stern & Foster, 1996. One volume references on the basic biology of some major social insect groups include Holldobler and Wilson (1990) on ants, Ross and Matthews (1991) on wasps. Roubik (1989) on tropical bees and Winston (1987) on honey bees.)
KIN SELECTION IN INSECTS
207
Table 9.2 Relatedness levels in a sodal insect colony. Haplodiploids
Regression
Diploids
Actor, redpient
relatedness
[ife·for·life relatedness
Regression and life·for-Iife relatedness
Queen, daughter Female. son Father. daughter Queen's mall'. Queen's son Sister. sister Sisler, brother Brother. sister BrOlher, brolher
0.5 1.0 0.5 0.0 0.75 0.5 0.25 0.5 0.5 0.75
0.5 0.5 1.0 0.0 0.75 0.25 0.5 0.5 0.25 0.375
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.25 0.25
Queen. grandson
Female. nephew
These relatedness values are for colonies with single. once-mated. outbred queens (Fig. 9.1). Imagine a population of pairs or groups of social interactants (potential donors of a social action. or actors, and potential reopients of it). Regression rtlattdness al a gene locus is formally obtained when the gene frequency in the potential reopients is regressed on the gene frequency in the potential actors across all the pairs or groups. The slope of the resulting regression line (the regression coefficient) equals regression relatedness. By definition, it gives the probability that a recipient and actor share the focal gene independently of their average probability of doing so (which is set by the gene'saverage frequency in the population). Life-for-Ufe relatedness equals regression relatedness multiplied by relative sex·specijic reproductive value. which is the scx·specific reproductive value of the recipient divided by the sex·specific reproductive value of the actor. Reproductive value measures the contribution of a class of individuals to the future gene pool in Ihe absence of selection, mutation and drift (Grafen. 1986). In haplodiploids, females have twice the sex-specific reproductive value of males because females are diploid and males are haploid. Put simply. females arc illlrinsicaJly twice as reproductively valuable as males because they carry twice the number of genes. So. for example. in haplodiploids sister-brother life-for-lire relatedness (0.25) equals sister-hrother regression relatedness (0.5) x relative sex-spccilic reproductive value (0.5). In diploids, the female: male ratio of sex-specific reproductive values is I, because borh sexes have identical ploidies. Therefore, in diploids, Jife-for·life and regression relatednesses are the same. Several classic studies of kin selection in social insecrs used life· for-life relatedness (e.g. Hamilton, 1972; Trivers & Hare, 1976). However, the regression definition of relatedness needs considering for two reasons. The first is that it is the relatedness deduced from formal population genetiC proofs of Hamilton's rule (Grafen. 1985). The second is that some social traits (for example. extensive worker male production) alter relative sex-specific re· productive value but not relatedness (e.g. Boomsma & Grafen. 1991). meaning that these two quantities are beSt kept conceptually distinct. (After Grafen. 1986, 1991; Bourke & Franks. 1995.)
As well as being the gUiding prindple for [he evolutionary study of so ial behaviour, Hamilton's insight also solved the historical 'problem of altruism' (Cronin, 1991; Alexander et al., 1991). Under an individual-based view of natural selection, the evolution 01 a slerile, altruistic caste is a paradox. How can reproductive restraint ever evolve if individuals are selected to maximize their offspring output? This difficulty, which puzzled Darwin. disappears under
208
CHAPTER 9
the gene selectionist view embodied in kin-selection theory. For, according to the theory, altruism appears at the individual level because of self-interest at the gene level. This is why kin selection still represents the flagship of the 'selfish gene' view o( adaptive evolution (Dawkins, 1976).
9.2 The origin and evolution of eusociality 9.2.1 The necessity of kin selection Kin-selection theory states that the altruism o( helpers in all eusodal animals must have evolved under the conditions set out in Hamilton's rule. A form of altruism may, in some instances, evolve by reciprocity (an individual performs costly aid today in the expectation of receiving such aid tomorrow) (Trivers, 1971; see Chapter II). But, reciprocal altruism is not 'true' altruism. For it to evolve the net benefit in terms of personal offspring number must eventually fall on the 'altruist' itself, which is why aid is delivered to begin with. True altruism, defined as occurring when the altruist suffers a net loss of offspring over its lifetime, requires kin selection. As far as can be judged, eusociality usually entails true altruism (and must do so if workers are totally non-reproductive). Hamilton's theory therefore specifies that relatedness (between the altruist and the beneficiary) must have been greater than zero in groups evolVing eusociality. However, relatedness could have been quite low, proVided the benefit of helping (b) was high and the cost (c) was low. In addition, as Hamilton (1972, p. 198) put it, 'insisting on the necessity of relatedness in no way precludes other (actors as necessary or contributory.' In fact, Hamilton's rule shows that both genetic factors (affecting relatedness) and ecological ones (affecting benefit and cost) must be important in promoting eusocial evolution (e.g. West-Eberhard, 1975). The kin-selectionist interpretation of eusocial evolution has traditionally been contrasted with two additional ideas. The mutualism hypothesis emphasized that social groups may have originated as mutualistic aggregations (Lin & Michener, 1972; Ito, 1993; see Chapter II). The parental manipulation hypothesis proposed that workers were initially forced to act altruistically by a dominant parent (Alexander, 1974). However, neither of these ideas represents a genuine alternative to kin selection. The mutualism hypothesis helps explain the early stages of group-living. But, since both parties in a mUlUalistic interaction must, by definition, evenlUally gain in average offspring number, it cannot explain altruistic behaviour (Crozier, 1979). Parental manipulation, on the other hand, involves interactions among kin and so falls within kinselection theory. In addition, there is no universal reason why workers cannot resist manipulation if it pays them to do so (e.g. Craig, 1979; Michod, 1982; Kdler & Nonacs, 1993). Parental manipulation is therefore just one type of kin conflicl.
KIN S E LEe Tl 0 N I N 1N SEC T S
209
9.2.2 Hymenopteran eusociality and the haplodiploidy hypothesis
The frequency ofellsocial evolution in the Hymenoptera Eusociality has evolved independel1lly many. times among the insecls. For example. Wilson (1971) eSlimated'thatthere have been II origins of eusociality in the Hymenoptera, and one in the termites. However. subsequent phylogenetic analysis has indicated that in termites reproductive altruism evo'ed at least twice. once in the forerunners of the workt,[ caste and once in the defensive caste of soldiers (which. unlike ant soldiers, are not simply modified workers) (Noirot & Pasteds. 1987. 1988). Similarly. a phylogeny based on mitochondrial DNA sequences suggests that the aphid family Hormaphididae evolved a soldier caste at least five times (Stern, 1994; Stern & Foster. 19961. There[ore. eusociality has had multiple origins in groups oUlside as well as inside the Hymenoptera. This finding means that unusual relatedness levels in the Hymenoptera cannot be essenlial for the evolution of eusociality. as Hamilton himself cautioned (Hamilton. 1972). However. high relatedness levels should cenainly facilitate eusocial evolulion. For example, social aphids probably live in clones (or mixtures of clones) of genetically identical individuals. because foundress females reproduce asexually (Stern & Fo. te... 1996). So. high relatedness presllmably underpins the ease with which non-reproductive soldiers have evolved in Ihis group (e.g. Andersson. 1984; Hamilton. 1987b). In faa. sterile soldier aphids in a clone would be almost exactly analogolls to the defen ive polyps of colony-liVing. clonal marine invenebrates (Stern & Foster. 1996).
Relatedness levels in the Hymenoptera In the Hymenoptera. males are haploid and develop from unfertilized eggs, but females are diploid and develop from fertilized eggs. As is well known, this hap/odip/oidy leads to unusual relatedness levels among members of a family (Table 9.2). Specifically. sisters share any gene borne by their father. but have a chance of 0.5 (as in diploids) of sharing a maternal gene (Fig. 9.1). So. relatedness among siSlers (calculated by averaging across the paternal and maternal halves or the genome) is (I + 0.5)/2. or 0.75. By COlllrast, sibling-sibling relatedness in diploids is 0.5 (see Table 9.2). As well as devising kin-seleaion theory, Hamilton (l964a,b) suggested that the three-quaners relatedness among sisters in thc Hymenoptera could promote eusocial evolution in the group. Other things being equal. a female would be more strongly selected to raise a sister (of relatedness 0.75) than a daughter (of relatedness 0.5). This would explain both the relatively high prevalence of ellsocial origins in the Hymenoptera, and the restriction of worker behaviour to females (in the entirely diploid termiles. workers are of both sexes).
210
CHAPTER 9 Queen
Queen's sons
Male
At'
Workers' sons
Fig. 9.1 Sisters are highly related in haillodiploids because they share all.heir father's genes. The figure shows the pedigree of a Hymenopteran colony. in which a single queen
mates with a single male and produces sons and daughters (workers and new queens). wilh the workers in turn producing sons. A. Band C are alleles on representative chromosomes (vertical bars). (From Bourke & Franks. 1995,)
The haplodiploidy hypothesis generated a massive amount of interest because it seemed a triumphant testament to the explanatory power of kin selection. To some, however, the idea seemed too mathematical and simplistic (and too clever by three-quarters, perhaps). The whole issue has been reviewed by many authors, including Wilson (1971), Hamilton (1972), Alexander (1974), West-Eberhard (1975), Andersson (1984), Seger (1991) and Bourke and Pranks (1995). Here, just a few crucial points need making. The first is that the haplodiploidy hypothesis is a subset of kin-selection theory (e.g. West-Eberhard, 1975). Hamilton's rule can easily hold without a potential helper being more closely related to siblings than offspring. Purthermore, the association of haplodiploidy with eusociality could be coincidental. In short, the validity of kin selection does not rely on the haplodiploidy hypothesis being correct.
Split sex ratios and the origin ofeusociality Another important point about the haplodiploidy hypothesis, which for a long time eluded recognition, is that, according to genetic models, it is not correct to say without qualification that haplodiploid relatedness levels promote the evolution of female workers in the Hymenoptera. Trivers and Hare (1976) showed that the evolutionarily stable population sex ratio for workers (assuming a simple colony kin structure) is 3 : I females: males. At this sex ratiO, the mating success of females (their per capita number of mates) equals one-third that of males (if females have one mating on average, the average male must have three matings). This exactly cancels out the benefit that workers receive from rearing greater numbers of their highly related sisters.
KIN SELECTION IN INSECTS
211
In fact. assuming a simple colony kin structure, equal elliciency at raising brood and a uniform population sex racio. sibling-rearing yields the same fitness payoff as offspring-rearing in the Hymenoptera (Craig. 1979; Grafen. 1986; Bourke & Franks. 1995). In response to this finding. theoreticians sought conditions under which a high relatedness among sisters might promote worker behaviour. Such conditions arise if there are split sex rarios. These occur when separate classes within a population contribute broods with systematically different sex ratios to the same generation (Grafen. (986). Specifically, sibling-rearing females have higher fitness than offspring-rearing ones if. at the outset of eusocial evolution. they raise broods whose sex ratio is more female-biased than the population sex ralio (Seger. 1983; Grafen. 1986; Pamilo. 1991a; Krebs & Davies. 1993). To see why. consider a population of solitary bees in which mothers rear daughters that mate then disperse. The population sex ratio is assumed to be the mothers' stable value of I : I. Imagine that a few mutant colonies arise in which daughters Slay at home and help rear a female-biased brood of Siblings. The mating success of females and males will be unchanged because the mutant colonies are 100 rare to affect the population sex ratio. Consequently. female mating success no longer offsets the fitness gain to workers of rearing a bias of their highly related sisters, so worker behaviour will spread.
Other factors in the origin of Hymenopteran eusociality Although it is theoretically sound. whether the haplodiploidy hypothesis correctly accounts for the nature and frequency of Hymenopteran eusociality is uncertain. Confinning that facultatively social Hymenoptera (those whose populations contain both solitarily nesting individuals and colonies with workers) exhibit split sex ratios would strengthen the view that females can exploit their high relatedness with sisters 10 evolve worker behaviour. The problem is to disentangle such effects from others that could be involved. For one thing. the Hymenoptera share features aside from haplodiploidy that may have facilitated eusocial evolution. These include the habit of solitary wasps and bees of building nests (providing a resource worth inheriting), and the possession by females of a sting (a useful weapon for a defensive caste) (Alexander. 1974; Andersson. 1984). Several authors have laid great stress on such nongenetic predisposing factors (e.g. Evans. 1977; Alexander et al., 1991; Crespi. 1994). In addition. haplodiploidy may have genetic consequences helpful 10 social evolution independently of its effects on relatedness. For example. a model by Reeve (1993) showed that dominant genes for sibling care are less likely to be lost from small populatiOns by genetic drift (random loss of alleles) under haplodiploidy than under diploidy (the protected invasion hypothesis). A number of authors have proposed that life-history effects also favuur worker behaviour in both diploids and haplodiploids (Queller, 1989, 1994; Gadagkar. 1990. 1991; Nonacs, 1991). The essential idea is that early mortality
212
CHAPTER 9
damages the fitness gain of a solitary breeder more than that of a worker. If a worker dies early in its active life. it may still have accrued some fitness. One reason is that a worker can begin rearing brood the instant it becomes active. because larvae requiring aid (produced by the queen) are likely to be present on the nest already. This is Queller's (1989) 'reproductive head stan' advantage. Another is that, even after the focal worker's death, olher workers can raise partly-reared brood to adulthood. In Gadagkar's (1990) phrase, a worker therefore has 'assured fitness returns'. By contrast, a solitary breeder that dies early may not yet have started rearing brood (for example, if it dies while nest-building); or, if it has started but not finished, will leave no companions to complete the job. This wealth of possible innuences on eusocial evolution in the Hymenoptera makes testing the haplodiploidy hypothesis rather intractable. So, perhaps the time has come to abandon the attempt. Instead, it seems more fruidulto refocus on the deeper issue -the application of Hamilton's rule, and its more sophisticated variants incorporating sex ratio and life-history effects, to natural populations.
All example of the application of Hamilton's rule in a facultatively social insect Hamilton's rule has previously been applied to the evolutionary decisions of Polistes wasp queens over whether to join other queens in nest-founding (e.g. Metcalf & Whitt, 1977a; Noonan, 1981; Grafen, 1984). But, polistine wasps are eusocial; after the founding stage, queens produce worker broods. To apply Hamilton's rule in the context of the origin of eusociality requires slUdies of the decisions of females in populations of facultatively social bees and wasps. Surprisingly, there have been few slUdies of this type. However, a beautifully revealing example comes from the work of Roland Stark on the large carpenter bee Xylocopa sulcatipes (Xylocopinae) in Israel. Femaies nest singly or in pairs in hollow plant stems. From long-term observations of marked individuals, Stark (1992a) deduced that pairs consist of mother and daughter, two sisters or two individuals from different broods. Behaviour inside the nest was observed by the ingenious method of fixing small lead shapes to the bees and watching them through the nest wall with an X-ray machine. This way, Stark et al. (1990) and Stark (1992a) found that one bee of a pair is reproductively dominant and eats the eggs of the other bee. The dominant bee is also the chief forager, whereas the subordinate acts as a nest guard and so is effectively a nonreproductive helper. Observations also revealed that X. sulcatipes neslS risk usurpation, or takeover by a foreign conspecific bee, who if successful destroys all existing brood (Stark, 1990). Stark's (1992a) data allow Hamilton's rule to be applied to a bee's decision over whether to nest alone or be a guard (Table 9.3). The analysis followed here differs slightly from Stark's own. Only mOlher and daughter pairs are
KIN SELECTION IN INSECTS
213
Table 9.3 Nesting success and relatedness in the carpenter bee X. sulratifNS. (After Slark. 1992a.)
1986
female
Mother plus daughter
Single female
37 18
18 17
26
Single
Number of nests observed (Nil Number not usurped (Nl~ Number of orfspring per surviving nest (N}) Average number of offspring per
foundress (N, = IN, x N,J/N,) Relatedness to offspring reared ('1)
Bendit (b) (= pair'S N4 - singleton's N,) Relatedness 10 offspring losl (rot) CoS! (e) (= singlelOn's N.) Helper's net payoH (= rib - 'IC)
1987
33
Mother plus
daughler 14 12
3.0
5.8
5.8
6.5
1.5
5.5
4.6
5.6
0.5
0.5
4.0 0.5 1.5 1.25
1.0 0.5 4.6 -1.80
In two-female nests. number of oHspring per foundress refers 10 the oHspring of the mOlher.
since she founds the nest and daughters produce no adult offspring. The '1 term is calculated as lht: average of a female's relatedness with full-sislers (0.75) and brothers (0.25), which is 0.5 (mothers mate only once). Stark (19923) weighted the guard-sibling relatedness i)y Ihe population sex ratio, which was femalc·hiased (and equal 10 the colony sex ratio in all Iypes of nest) (Stark. 1992b). However. a female-biased population sex ratio means thai Ihe malt.-s' higher mating success cancels out the advantage of rearing more sisters (see Section 9.2.2. 'Split sex ratios and the origin of cusociality'). Since the net effect on lhe fitness payoff is as if Ihe sex ratio were unbiased. un weighted relatedness values are used here. This does not affect Stark's conclusiuns, since the female bias was slight.
consid~r~d in detail. because th~y are the most common typ~ (in such pairs. the guard is always the daught~r). A guard's benefitt~rm is calculated easily as the number of extra offspring her presence confers on a pair. Since guards produce no adult offspring themselves, h~r cost term equals the number of offspring sh~ could hav~ had as a single nester. The benefit and cost terms ar~ then weighted by th~ guard's relatedness to. respectively, siblings and potemial offspring (Table 9.3). Th~ analysis shows Ihat sel~ction favoured helping (guarding) by daught~rs in the first year of Ihe study but not in Ihe s~cond (Table 9.3). The same was also tru~ for guarding by sist~rs (Stark. 1992a). It was unclear if joining a he~ from a different n~sl. was ever favoured. because the relatedness of these pairs was unknown. However, even helping an unrelated bee could sometimes have been profitable, if the guard slOod a chance of evemually inheriting the nesl (Stark, 1992a; Hogendoorn & Velthuis. 1995). These results help explain why bOlh helpers and single nesters coexisled in the population (Stark, 1992a). They also show exactly why helping paid
214
CHAPTER 9
off. Clearly. the guard's presence meant that pairs of bees suffered far fewer usurpations than singletons (Table 9.3). In addition. having a guard boosted the brood size of non-usurped nests (Table 9.3). almust certainly because the dominant bee could spend longer foraging (Stark. 1989; Hogendoom & Velthuis. 1993. (995). The conclusion is that sociality can both protect the nest against intraspecific attack and allow an efficient division of labour. These represent two very general promoters of social evolution (Lin & Michener. 1972; Oster & Wilson. 1978). Moreover, helping was more profitable in the first year because single nesters were usurped more often that year and. if they survived. were relatively unproductive (Table 9.3). Therefore. the selective pressures maintaining sociality could be those that make solitary nesting costly as much as those affecting groups directly (e.g. Emlen. 1991; see Chapter 10). This exemplary study could have been extended in several ways. One would involve confirming the inferred levels of relatedness among pairs of bees using molecular techniques, for example allozyme analysis (e.g. Metcalf & Whitt, 1977b). DNA fingerprinting (e.g. Mueller etal' 1994) or the analysis of microsatellite DNA variation (Queller el al' 1993). Another line of investigation would be to test the assumption that guard bees would have as many offspring as single nesters if they nested alone. Perhaps guards are intrinsically less fertile, making helping more likely to evolve (West-Eberhard. 1975). The issue could be settled by measuring costs and benefits experimentally. for example by forcing helpers to nest alone (Queller & Strassmann, 1989). Finally. Stark's (1992a) findings pose the question of precisely why single nesters were so unproductive in the first year, suggesting a need for the measurement of ecological variables. 9,2.3 The origin of eusociality in the termites
The termites have long been recognized as providing a testing ground for theories of social evolution wholly independent from that represented by the Hymenoptera. Entomologists used to believe that termites evolved from ulckroaches, and pointed to the wood roach CryptocerclIs. a group-living cockroach found in logs, as a likely approximation to the termites' ancestor (Seelinger & Seelinger. 1983; Nalepa, 1984). However, subsequent phylogenetic evidence suggests a more distant link between cockroaches and termites (Thorne & Carpenter. 1992; DeSalle et al.• 1992; KambhampatL 1995). On the other hand. the ancestor of termites may still have resembled CryplOcerclIs behaViourally (Noirot, 1989), since occupying and eating logs is the hallmark of many termites to this day. Previously. in the flush of excitement caused by the haplodiploidy hypothesis. biologists sought genetic effects in the diploid termites that might mimic the influence of haplodiploidy on Hymenopteran relatedness levels. 'Ivo types of haplodiploid analogy were proposed. The first was prompted by the occurrence of chromosome rings incorporating the sex chromosomes in some species (Luykx & Syren. 1979; Lacy, 1980), the second by alternations of
KIN SELECTION I
INSECTS
215
inbreeding and outbreeding in the typical termite life cycle (Hamilton, 1972; Banz, 1979). Both these traits could in some circumstances make potential workers more closely related with siblings than with offspring. However, further investigation failed to confirm an important role in the origin of termite eusociality for either phenomenon. For example, their assumptions are not universally met (chromosome rings seem 10 be of recent origin, and the required type of cyclic inbreeding is absent from many termite species), and neither are their predictions always fulfilled (workers in species with chromosome rings do not favour same-sex siblings as expected) (Crozier & Luykx, 1985; Luykx etal.. 1986; Hahn & Stuart, 1987; Myles & NUlling, 1988). In any case, the search for a haplodiploidy analogy in termites was unnecessary. Termites live in extended families that are typically founded by a single, monogamous pair. Furthermore, termite workers most probably evolved as sibling altruists in monogamous ancestors (Nalepa & Jones, 1991). Typical relatedness levels between early termite workers and the brood they reared would therefore have been at least 0.5, the level for diploid siblings (Hamilton. 1972; Nalepa & Jones. 1991). In fact, in one of the few allozyme analyses of termite kin structure, Reilly (1987) found a relatedness among nestmates of 0.57. So, the condition in Hamilton's rule would have been satisfied in a prototermite simply by a benefit-to-cost ratio greater than unity. This seems an easy condition to fulfil when considered alongside other aspects of termite biology. For one Ihing. the work in the so-called 'lower' lermites is performed by immatures (unlike the Hymenoptera. lermites have a continuous metamorphosis), and these typically retain the ability 10 moult to a reproductive form (Myles & NUlling, 1988). Sterile, morphologically distinct workers are a later development (Noirot & Pasteels, 1987; Higashi et al.. 1991, 1992; Roisin, 1994). Therefore. the reproductive sacrifice made by early termite workers may not have been very greal. On top of this, logs represem a patchily distributed habitat, suggesting thai allempting to disperse and breed alone (and risk failing to find a suitable log) was costly in early termites (Andersson. 1984; Nalepa & Jones, 1991; Alexanderecal.. 1991; Nalepa. 1994). Both helper and defensive behaviour (as found in termite soldiers) are similarly favoured in other organisms that live and feed in resource-rich patches. Prime examples are gall-dwelling aphids and thrips, and the famous naked mole-rat (Alexander et al.. 1991; Crespi. 1992, 1994; Stern & Foster, 1996). In summary, the unusual ecology of termites, combined with their ordinary diploid relatedness levels, seems entirely capable of tipping the balance in favour of eusociality.
9.3 The evolution of a stable reproductive skew 9.3.1 Ecological constraints and skew theory Social insects vary greatly in the number of queens within their colonies (see
216
CHAPTER 9
Table 9.1). They are typically classified as having one queen per colony (monogyny) or several (polygyny) (e.g. Wilson, 1971). In polygynous species, variation also exists in the level of reproductive skew (the degree of sharing of total reproduction among individuals). Por example, in some multiple-queen ants, only a single queen lays eggs at anyone time (functional monogyny: Heinze & Buschinger, 1988). In others, all queens lay eggs. Respectively, these represent cases of high and low reproductive skew. The queslion is, what accounts for such differences? Skew theory is a recently revived sel of ideas, based on kin-seleclion theory, thaI allempts 10 address Ihis problem (Emlen, 1982a, 1982b; Vehrencamp, 1983b; Reeve & Ratnieks, 1993).11 is important because it integrates in a single explanatory framework ecological. genelic and social factors. It also potentially applies to many kinds of social organism (e.g. Keller & Reeve, 1994: see Chapters 10 and II). In the previous section, the analysis (using Hamilton's rule) of Stark's (1992a) data on the bee x. sulcatipes lead to the conclusion Ihat helping behaviour by the non-reproductive subordinate within pairs was favoured (in the first year of Stark's study). Another way of viewing this finding would be to conclude that, in the conditions of that year, the stable skew among pairs was high (indeed maximal, since the dominant bee produced all of a pair's adult offspring). Skew theory allows investigalOrs also to analyse cases less extreme, in which several individuals in a group share reproduction (but not necessarily evenly). So, another merit of the theory is helping to unify in a single conceptual framework the evolution of eusociality and communal breeding (Vehrencamp, 1983b; Keller & Reeve, 1994; Sherman et al., 1995). Imagine a newly adult queen. Should she set up a colony of her own, so leading 10 monogyny, or join an existing colony, leading to polygyny? Clearly, joining behaviour (including remaining 10 breed in the natal nest) would be favoured if ecological circumstances made attempting to nest alone very difficult. In other words, joining should occur when there are high ecological constraints on solitary breeding (e.g. Emlen, 1982b; see Chapter 10). This is why, for example, the readoption of daughters and polygyny should evolve in ants when unoccupied nest sites are scarce (e.g. Herbers, 1993; Bourke & Heinze, 1994; Keller, 1995). A corollary of this reasoning is that societies with multiple breeders are unstable if joiners could receive greater fitness payoffs (assessed via Hamilton's rule) from dispersing to breed alone. This forms the basis of skew theory. Consider a queen whose colony is joined by a newcomer. A critical assumption is that the resident queen benefits from the joiner's presence (for example, because the joiner boosts the resident's produclivity). Another key assumption is that the residenl can conlrol the joiner's offspring Uulput (for example, hy physical domination). Given these conditions, one would expect the resident 10 skew reproduction in her own favour only up to the Ihreshold beyond which the subordinale would be selected to abandon her and neSI alone. This
KIN SELECTION IN INSECTS
217
is because. if the dominant queen left too small a share of the colony's reproduaion for the subordinate. the subordinate's departure would deprive the dominant of the benefit of her presence. There!ore. the stable level of skew is determined by those faaors that areea the relative allraaion of remaining in the colony to subordinates (Yehrencamp. 1983b: Reeve & Ratnieks. 1993; Keller & Reeve 1994). Skew theory suggests several factors of this type (fig. 9.2). To stan with, a rise in either the joiner's contribution (via aid) to colony productivity, or ils relatedness to the dominant, or the level of ecological constraint, permits a rise in the stable level of skew (Vehrencamp, 1983b). The reason is that these changes either make slaying in an association more profitable to a joiner (rising productivity and relatedness), or make leaving it less profitable (rising ecological constraints). Therefore, the dominant can take a greater share of the colony's reproduction without risking the loss of the subordinate. Relatedness also affeas skew through the generational structure of the group. Subsoaal societies (mother-daughter associations) are expeaed to show greater skew
+
Fig. 9.2 Several factors affect
lh~ stable level of feproductive skew. which itself influences the expected lewI of within·sroup aggression. Each + sign means that a rise in that Jaoof
causes a rise in the one at which lhe arrow is pointing; the - sign means that a rise in one factor causes a decrease in the other. (From the models of Yehreocamp. 1983b: Reeve and Ramicks. 1993.) The level of ecological constraints can influence relatedness (lOp afro') if some dispersing individuals enter unrelated groups. Then. as ecological constraints rise and dispersal becomes less frequent, average relatedness within groups rist's (Bourke & H{'inzt'. 1994). Insel: dominance bailie between PolislfS wasp foundresses. (From Ross & Matthews. t991.)
218
CHAPTER 9
than semisocial ones (sister-sister associations). This is because daughters are (on average) equally related 10 their own and their mother's progeny, and so are essentially indifferent between rearing siblings or prodUcing offspring. But. females in sister-sister groups are more closely related to offspring than to nephews and nieces, and so are less Willing to let anolher individual reproduce in~tead of themselves (Reeve & Keller, 1995). A final fador affecting skew is the subordinate's fighting ability; as this rises, skew should decrease (Reeve & Ratnieks, 1993). This happens because a relatively strong subordinate is beller placed to challenge the dominant for the right to reproduce and, (0 avoid this, the dominant must offer the subordinate a greater share of reprodudion. Reeve and Ratnieks () 993) also argued that the degree of skew itself affects the expected level of aggression. When skew is high, the prize from usurping the dominant is also relatively high, leading to aggressive probing of the dominant by the subordinate. When skew is low, becoming the dominant yields lillIe extra reproduction, so aggression should be muted. Similarly, inside human societies one might exped the level of ,ocial discontent to depend not on awrage earnings but on the variation in incomes -the gap between rich and poor.
9.3,2 Skew evolution in the polistine wasps The polistine wasps form one of the main groups with which social insect biologists have atempted to test skew theory. However, all tests to date have been indirect. This is because to measure skew accurately one needs to allocate offspring to their parents, which requires molecular genetic methods of high resolution. Happily, the development of techniques for analysing microsalellite DNA variation promises to provide such a method (Evans, 1993, 1995; Hamaguchi et al., 1993; Queller et al., ) 993; Peters eI al., 1995). Reeve and Nonacs (1992) conduded a simple experimental test of one of the assumptions of skew Iheory. Say a dominant attempled 10 skew reprodUd ion beyond the 'agreed' level by eating 'some of the subordinate's eggs. The theory assumes thaI the subordinate can deteci such cheating and retaliate. Reeve and Nonacs (1992) removed eggs from field nests of Polis/es [uscatus wasps. Although the experimenters did not know the maternity of the removed eggs, they argued that subordinates would perceive any loss of eggs as potentially due to egg-eating by the dominant. As expected, alpha queens showed no consistent response 10 egg removal. but beta ones became more aggressive on average. Furthermore, this occurred only when eggs destined to be reproductive were taken. When worker-destined eggs (which are of far lower fitness value) were removed, betas showed no significant rise in aggression. Reeve and Keller (1995) performed a comparative test of the idea thaI the generational structure of colonies affects the stable skew. They compiled skew estimates from the lileralUre for a set of social insect species (moslly polistine
K[N SELECTION [N [NSECTS
2[9
wasps) in which some colonies were semisocial and others were motherdaughter groupings. In line wilh the theory, skew in the second type 01 colony exceeded that in the firs! in [3 01 17 cases. Finally, in the North American species P. domi/lul's. Nonacs and Reeve (1995) lound that multiple-Ioundress associations with a small size dillerence among their members were less productive lhan ones with a large size di((erence. This matched the idea that. as subordinates become closer in size to dominants (and therefore more similar in fighting ability). they should challenge the dominants more vigorously lor supremacy. This would lead to higher aggression which, in llIrn, could account lor the observed 10' 01 produclivity. 9.3.3 Skew evolution in the leptothoracine ants
Skew lheory predicts an associalion bel ween low ecological constraints. low rclatednes , low skew and low aggression inside colonies on lhe one hand. and high ecological constrainls, high relatedness. high skew and high aggression on the other (Reeve & Ratnieks, 1993; Fig. 9.2). Bourke and Heinze (1994) found evidence in the leptolhoracine ants for these associations. Facultatively
polygynous Leplolhorax live in eXlended unilorm habitats such as pinewoods. Costs of dispersal are therefore relatively low, because dispersal in any direction still keeps a young queen within the habitat. In addition, all queens lay eggs (so skew is assumed to be low), and queens live together peaceably. By contrast, lunctionally monogynous leptothoracines live in patchy habitats such as rocky oUlcrops. This raises dispersal COSIS. because iI means that departing queens risk failing to find another suitable patch. In facl, dispersal is so costly in these ants lhat some have evolved a morph 01 queens that is permanently wingless and therelore incapable of dispersing lar. Moreover. skew is high by definition (only one queen 01 the several present lays eggs). and aggression is also high (the egg-laying alpha queen physically dominates the others) (Bourke & Heinze, (994). However, genetic studies have revealed thaI, in some populations. cyclical changes in queen numbers over time may lead to a decoupling between lhe sodal structure 01 leplOthoracine colonies and their genetic structure and level of skew (Heinze. 1995; Heinze et al., 1995). Such phenomena complicate the testing 01 skew theory. On the other hand. as expected. between-queen relatedness in the single functionally monogynous species lor which a measure exists is higher lhan the average value lor lhe lacultalively polygynous species (Heinze. 1995)_ In conclusion, skew lheory helps explain a large amount of diversity in the sodal behaviour and colony structure of wasps and ants. Nonetheless. owing to its scope and power, additional experimental tests 01 the assumptions and predictions of the theory are undoubtedly required.
220
CHAPTER 9
9.4 Sex ratio evolution and kin conflict in social insects 9.4.1 Sex ratio theory for the social Hymenoptera In a very innuential paper. Trivers and Hare (1976) applied Fisher's (1930) sex ratio theory to the social Hymenoptera. This work opened up one 01 the most intriguing areas in social insect research: the study of sex allocation and connicts 01 interest within colonies. Many imponant insights into kin seleaion and sex ratio theory have llowed lrom this work (Charnov. 1982; Trivers, 1985). Fisher's (1930) theory states that the sex ratio is evolutionarily stable il whoever comrols the sex ratio derives as much fitness. per unit of e!lon, lrom producing a [emale as lrom producing a male (e.g. Trivers & Hare, 1976; Benlord. 1978; Trivers. (985). The reason is as lollows. rr the equal returns condition holds, there is no selection lor systematic overproduction o[ either sex ([emales and males yield the same payo!l). Therefore. the sex ratio remains unchanged. and is by definition stable. II the condition does not hold. one sex yields greater fitness and so will be produced in comparative excess. This will be the sex that is initially rare relative to its stable frequency. The reason is that members 01 the rarer sex have a greater mating success. For example, if there are two lemales lor every male in a population. then the average male has twice the number 01 mates 01 the average lemale. Therelore. any deviation lrom the stable sex ratio is self-correcting; when the equal returns condition docs not hold. the comparatively rare sex is overproduced. so returning the sex ratio to its stable level. The fitness payoff from producing a member 01 either sex is defined as the product. regression relatedness (r) x relative sex-specific reproductive value (V) x mating success (MS) (e.g. Grafen. 1986; see Table 9.2). Therefore. algebraically, the sex ratio is stable when: ('FV,.MS.)/c= 'MVMMSM
where the subscripts F and M denote female and male. respectively, and c equals the cost ratio, defined as the average energetic cost 01 producing a female divided by that 01 producing a male (Boomsma. 1989; Boomsma et al., 1995). Costs need considering becau e the sexes may diller in their energy demands, for example if there is sexual size dimorphism. Let the stable numerical sex ratio be X : I lemales : males. Then, il a female has one unit 01 mating success. a male must have X units. Therefore, MSF/MS M = IIX. so Xc = 'FV.I'M VM . This last quantity (the ratiD o[ life-lor-life relatednesses; see Table 9.2) is termed the relatedness asymmetry (BDDmsma & Grafen. 1990). So. Fisher's theory can be rephrased as the statement that the stable lemale : male sex investment ratio (Xc. the ratio of energy spent on lemales and energy spent Dn males) equals the relatedness asymmetry. Note that this refers to the sex investment ratio lor the pDpulation, and not necessarily lor individual broods. This is because.
KIN SELECTION IN INSECTS
221
assuming random maling, it is the relative abundance of the sexes in the population as a whole thaI determines their mating success. This reasoning can be applied to yield many familiar results in sex ralio theory. Take the standard case of diploids with parental control of sex allocation, random mating and no sexual size dimorphism (c = I). Then, X equals (lifefor-life relatedness to daughters)/(life-for-life relatedness to sons), which is simply 0.5/0.5 (see Table 9.2). This is fisher's .:I 930) finding that the stable ex ratio in this case is 1 : 1. Now, consider social haplodiploids. for queens, the relatedness asymmetry is still I : 1 (queens are equally related to daughters and sons), so the stable sex ralio must also be I : I. But, for workers, the relatedness asymmetry equals (life-for-Iife relatedness to sisters)/(life-for-life relatedness to brothers), which is 0.75/0.25 (see Table 9.2), or 3: I. This is Trivers and Hare's (1976) classic result; assuming a simple kin structure (monogyny, singly mated queens, sterile workers and random mating), workers in eusocial Hymenopteran colonies favour a population sex investment ralio (among the sexual forms) biased 3 : I in favour of females. Since queens favour a I : I ratio, there is a queen-worker conflict over sex allocation. fisher's (1930) lheory, with Trivers and Hare's (1976) modifications. is very general. Using the approach above, it has been extended to cases either where the kin structure is not simple, so the relatedness or reproductive value termS change (for example, if there is multiple mating, worker reproduction or polygyny), or where some of the underlying assumptions arc altered Ifor examplc, mating is not random) (e.g. Pamilo, 1990, 1991 b; Bourkc & Franks, 1995; Crozier & Pamilo, 1996). 9,4.2 Tests of sex ratio theory in social Hymenoptera
Population sex ratio data Trivers and Hare (1976) argued thaI the workers should usually win the conJlict with queens over sex allocation. because they both outnumber the quecn and rear the larvae, and so have the opportunity to bias the composition of the reproductive brood in their interests. In other words. there should be worker control of sex allocation. Consequently, in monogynous ants one would expect a 3 : I population sex investment ratio. In polygynous ants, the expected ratio should be close to I : I even under workcr control (assuming queens within colonies are related). This is because adding extra queens to a colony reduces the workers' relatedness asymmetry, as proportionately fcwer and fewer females will bc full sistcrs (Pamilo, 1990). Howevcr. in monogynous slave-making ams, one should also cxpcct I : I investment. Slave-makers are social parasites whose workers steal pupae from nests of othcr species, which after maturing raise all the slavc-makers' brood. So, Trivers and Hare (1976) argued. slavemaker workers lack the praclical power necded to control sex allocation. This leaves the slave-maker queen frcc to influence brood composition unimpeded
222
CHAPTER 9
by her workers (for example, by laying equal numbers of male and queendestined eggs). Therefore, in slave-makers one expects unbiased, queencontrolled sex allocalion. These predictions are borne out by the data (Trivers & Hare, 1976; Nonacs, 1986; Boomsma, 1989; Pamilo, 1990; Bourke & Franks, 1995). The average sex investment ratio across 40 species of monogynous ants is significantly female-biased (0.63, expressed as the proporlion of investment in females), whereas polygynous ants (25 species) and slave-makers (three species) have sex investment ratios of 0.44 and 0.48, respectively, which are indistinguishable from 1 ; 1 or 0.5 (Pamilo, 1990; Bourke & Franks, 1995). These findings are especially striking in that the sex ratios are true investment ratios; numerical sex ralios in ants tend to be male-biased (Nonacs, 1986; Bourke & Franks, 1995). Overall investment can nonetheless be female-biased in monogynous ants because queens receive large fal reserves to function as an energy supply during colony foundation (in ants, fat is a feminine tissue). One might ask why in monogynous ants the average ralio is not closer to the expected 0.75 (3 ; I). The likely explanation is that few of the species in the dataset exactly meet all Trivers and Hare's (1976) assumptions. Instead, there is probably a low level of multiple mating, worker reproduction or polygyny within their populations, all of which cause the workers' stable sex ratio to decline (Pamilo, 1990, 1991b). Conceivably, there could also be a degree of queen control in some species (Bourke & Franks, 1995). A few researchers have lested Trivers and Hare's (1976) theory by measuring relatedness asymmelry directly wilhin populations (using allozyme methods), and then checking the predicled sex investment ralio against Ihe observed one. This approach has been used in four monogynous ants, with results Ihal largely confirm Trivers and Hare's (1976) idea of worker control given the errors likely 10 be involved in measuring both relatedness values and sex ratios (Table 9.4). However, in al leasl one population of Lasius niger, sex investment was unexpectedly male-biased (Table 9.4). A possible reason is that this population was poor in resources (Van der Have et oJ.. 1988). Nonacs (1986) suggested thaI resource levels have a proximatc effect on sex investment ratios. Underfed colonies mighl redirect investment from queens into workers (for example, to enhance the colony's chance of surviving 10 beller times), so reducing female bias. Field experiments have confirmed such an effect in one case (Formica podzoJica; Deslippe & Savolainen, 1995) but not in another (Leptothorax Jongispinosus; Backus & Herbers, 1992). Alexander and Sherman (1977) argued that, contrary 10 Trivers and Hare (1976), female bias in the sex ratios of monogynous ants arises because of local mdte competition. This occurs whcn related males compete for mates (so violating Ihe usual assumplion of random mating). Female bias is expected because Ihere are then diminishing returns on producing males but not females (Ihere is no point prodUcing many males bearing similar genes who will compele with one another, whereas females compete with non-relalives for new nest
KIN SELECTION IN INSECTS
223
Table 9.4 Tests of sex ratio theory within monogynous, non-parasitic ant species. Population sex investment ratio (fraction of
invcstmen( in females Number of
Species
Expected
Observed
Colobopsis nipponicus Formica truncorum Leptothorax tuberum
0.75 0.63 0.70
0.75 0.65 0.75
22 47
0.72 0.79 0.59
0.68 0.65 0.36
125 26 50
colonies
Lasius niger
Population I Population 2
Population 3
The source references are J1asegawa (1994) for C. nipponicus. Sundstrom (J 994) for F. truncorum. Pearson et al. (1995) for L. wberum and Van dec Have n aJ, (1988) {or L. niger.
The observed sex ratios for F. tnmcorum and L. niger were recalculated by Bourke and Franks (1995). The expected sex ratios under worker control were derived
by the original authors
from the measured workers' relatedness asymmetries (expressed as fractions). The relafedness asymmetries feU below 0.75 in some cases because of partial multiple mating or worker reproduction. The expected sex ratio in F. trUtlcorum is also calculated on the basis of extensive sex ratio splitting in this population (see text). Under queen control. the €'xpeeted sex ratios were 0.5 in every case. because the queen's stable sex ratio is unaffected by multiple mating and worker reproduction in her presence (Pamilo. 199Jb; Bourke & Franks. 1995).
sites) (Hamilton. 1967). However. ams typically mate in large mating swarms drawn from many colonies (Holldobler & Wilson. 1990). and so lack a mating system involving an appreciable degree o[ local mate competition. Therefore. female bias cannot usually arise from this cause. On the other hand. in a few cases where conditions are appropriate. local mate competition leads. as predicted. to a sex ratio more female-biased than that expected under random
mating. One example is the socially parasitic am Epimyrma kraussei. whose sexuals mate exclusively in the nest (Winter & Buschinger. 1983; Bourke. (989). Another is the non-parasitic harvester am Messor aciculatus in which. although mating occurs outside the nest. genetic evidence suggested that each mating swarm is composed of sexuals from just a lew colonies (Hasegawa & Yamaguchi. 1995). The honey bee (Apis mellifera) exemplifies another type of violation of a Fisherian assumption. Colonies reproduce by splitting imo two (colony fission). with the young queens of a hive competing among themselves to head one 01 the daughter colonies. These queens therefore experience local resource competition (competition among relatives lor resources). Males. by contrast. compete [or mates with all other males in the population. as in the standard case. So, there are now diminishing returns on investment in new
queens. but not on producing males. This makes the predicted numerical ,ex
224
CHAPTER 9
ratio among honey bee sexuals extremely male-biased, and this is what is found (Bulmer, 1983; Pamilo, 199Ib). Split sex ratios in eusocial Hymenoptera
Fisher's (1930) theory specifies only a stable sex ratio at the level of the pupulation. In a very large population at sex ralio equilibrium. lhe sex ratio
of individual colonies can vary randomly (because, by definition. females and males yield equal returns). However. in nature, between-colony sex ratio variation is frequently greater lhan random. with many colonies specializing on a single sex (e.g. Nonacs, 1986). Boomsma and Grafen (1990. 1991) proposed an explanation for such split sex ratios based on Trivers-Hare theory. Suppose that the workers' relatedness asymmetry varies among colonies in a population (due to, say, queens heading some colonies being singly mated and queens heading other colonies being multiply mated). Then. according to Boomsma and Grafen (1990, 1991). workers in colonies in which their relatedness asymmetry is high compared to the female: male population sex ratio should concenlrate on female production. and workers in colonies in which their relatedness asymmetry is comparatively low should produce mainly males. The following example shows why. Say a colony exists in which the workers' relatedness asymmetry is 3 : 1. while the population sex ratio is 1 : l. The level of the population sex ratio means that the mating success of females equals that of males. But. the level of the workers' relatedness asymmetry means that workers are three times mOre closely related to females than to males. Therefore, workers must derive three times more fitness from producing a female than from producing a male (recall that fitness equals the product of life-for-life relatedness and mating success), and so should rear females alone. Boomsma and Grafen (1990, 1991) also showed that. when the workers' relatedness asymmetry varies between classes of colonies, the stable population sex ratio either equals one of the relatedness asymmetries. or lies between them. Therefore, at least one colony class (the one whose relatedness asymmetry is unequal to the population sex ratio) should always cOncenlrate on producing a single sex. In populations of the halictine bee Augochlorella striata. some colonies are mother-daughter associations (eusocial colonies). but others have lost the foundress and consist of sister-sister groups (parasocial colonies). So, the eusocial colonies should produce female-biased broods (Ihe worker' relatedness asymmetry is relatively high because workers are rearing fullsiblings). whercas the parasocial colonies should produce male-biased broods (relatedness asymmetry is relatively low because workers are rearing nieces and nephews). Mueller (1991) tested Ihis with a simple but clever experiment. He removed the foundress from one set of colonies (so creating parasocial colonies). and a random worker from another sel (so leaving them eusocial but cOnlrolling for the reduction in colony size). The result was as expected.
KIN SELECTION IN INSECTS
225
with the fraction of investmem in females being 0.67 among Ihe eusocial colonies and significantly lower (0.31) among the parasocial ones. Mueller et al. (1994) later confirmed the genetic structure of each type of colony using DNA fingerprinling. Another test of split sex raliotheory was carried out by Sundstrom (1994) in a monogynous population of the wood am Formica truncorum. She established, using allozyme analysis, llial some colonies had a multiply maled queen Iso reducing the workers' relatedness asymmetry), and others had a singly mated queen (leaving relatedness asymmetry unaffected). As predicted, the first type of colony produced mostly males and Ihe second type produced mainly females. Among polygynous ants, where workers' relatedness asymmetry varies because of variations in queen number, split sex ratios also occur (e.g. Herbers, 1990; Chan & Bourke, 1994; Evans, 1995). An assumption of Boomsma and Grafen's (1990, 1991) theory is that workers in a colony can assess Iheir relatedness asymmetry. Note that this would not require Ihe ability to disCJ'iminate among different kin within the colony. Instead, workers have to judge whether, for example, their queen is singly or multiply mated, or whelher the colony is monogynous or polygynous. This could plausibly occur if workers assessed the genetic variability of the brood. To lest this idea, Evans (1995) experimentally added unrelaled larvae 10 colonies of the polygynous spedes Myrmica tahoensis. As expected from the idea of worker assessmem, the result was that the colonies reared more male-biased broods. Although other explanations for split sex ratios have been proposed (e.g. Frank, 1987), none predicts an assodalion between a colony's sex ratio and the workers' relatedness asymmetry. Such an assodation is not predicted if queens control sex allocation either, because queens are symmetrically related to daughters and sons. Therefore, the overall conclusion from sllldies of splil sex ratios is that workers largely control sex allocation as Trivers and Hare (1976) predicted. In ants, sllldies of population sex ratios also suppon this conclusion (e.g. Table 9.4). The confirmation of worker control is imponant because, by establishing that workers bias investment according to their relatedness to the sexes, it suppans both Fisher's sex ratio theory and Hamilton's kin-selection theory. A final question concerns how workers manipulate sex allocation. Using histological methods to count the chromosomes inside individual ant eggs (e.g. Aron et al., 1994, 1995), Sundstrom el al. (1996) have recently shown that all queens in a monogynous population of Formica e;uecla contributed a similar fraction of haploid eggs 10 their colony's egg pool. However, workers in colonies headed by singly mated queens raised a female bias of adult sexuals, whereas workers under multiply maled queens raised a male bias, so achieving a split sex ratio among adults identical to that found in F. truncorum. These findings suggest that. in colonies where their fitness interests dictate (ones with a singly mated queen), workers selectively destroy male eggs or larvae. They therefore provide excellem evidence of one way in which workers can facultatively manipulate brood composition to achieve their sex ralio preferences.
226
CHAPTER 9
9.4.3 Kin conflict over worker reproduction
Contrary to what is often thoughL. workers in many social Hymenoptera can produce male offspring from unfertilized eggs (Bourke. 1988; Choe. 1988). Another important type of kin conflict occurs over worker reproduction (Trivers & Hare. 1976; Ramieks & Reeve, 1992). In a colony with a single, once-mated queen, the workers are more closely related 10 sons (life-for-life relatedness = 0.5) than to nephews (relatedness 0.375) or brothers (relatedness 0.25). The queen, however, is more closely related 10 her Sons (relatedness 0.5) than the workers' sons (relatedness 0.25) (see Table 9.2). Therefore, each party favours the production of iLs own male offspring, leading 10 queen-worker conflict (Trivers & Hare, 1976). This arises even if no single worker can monopolize male production, since in this case any given worker will still be more closely related to the average worker-produced male (a nephew) than 10 a brother. Connict over male production could account for the otherwise puzzlingly high level of friction inside bumble bee (Bombus) colonies, which typically have singly mated queens (e.g. Page, 1986). In Lhese, the queen aLlacks laying workers and eats Lheir eggs, while the workers Lry to eat some of the queen's eggs (e.g. Van Honk e/ al.. ) 981; Owen & Plowright. 1982). This happens most often towards the end of the colony's life, when the queen is more likely to be laying male eggs. In fact, the escalating violence may even lead to the workers killing their own queen (Bourke, ) 994). Workers of monogynous ants wilh singly mated queens are usually nonreproductive in the queen's presence (Bourke, 1988; Choe, 1988). This is despite Lhe prediction of Ihe simple relatedness arguments that these workers should favour rearing worker-produced males, and despite the queen's lack of aggression to workers. A previous explanation for this finding was that the queens suppressed worker reproduction with chemical secretions (Wilson, 1971; Bourke, 1988). However, Seeley (1985) and Keller and Nonacs (1993) proposed that, instead, workers do not benefit from aLlempting to reproduce in the queen's presence. Worker reproduction could reduce colony productivity, or workers might sometimes be incapable of singling out the queen's male eggs for replacement (Cole, 1986; Nonacs, 1993). ff so, the workers' interests would be served beSL by rearing the queen's offspring alone, but oniy so long as the queen remained fecund and healthy. So, the queen's pheromones could be an honest signal of her vigour, rather than an instrument of coercion. This signalling hypothesis for the function of queen pheromones is an open issue aL present (Bourke & Franks, 1995), and deserves testing. Multiple mating by queens alters the reproductive behaviour expected among the workers (Starr, 1984; Woyciechowski & Lomnicki. 1987; Ramieks, t 988). Specifically, when the number of mates per queen (k) exceeds two, the workers' relatedness 10 the average worker-produced male (nephews, relatedness equals 0.25[0.5 + Ilk]), falls below worker-brother relatedness (0.25). Therefore. under a multiply mating queen. workers are expected to favour
KIN SELECTION IN INSECTS
227
rearing the queen's males (assuming that no single worker can dominate worker reproduction). To achieve this, they should actively prevent each olher's reproduclion, a phenomenon that Ratnieks (1988) termed worker policing. The honey bee provides striking confirmation of the worker policing hypothesis. Honey bee queens mate up to 17 limes (Winston, J 987). Genetic studies also indicate that only about one in 1000 adult males is worker-produced (Visscher, 1989). This is because although a few workers lay eggs in colonies with a queen, other workers selectively eat these eggs (Ratnieks & Visscher, 1989; Ratnieks, 1993). Furthermore, workers with developed ovaries are physically attacked by their nestmates (Visscher & Dukas, 1995). In short, worker policing both occurs in honey bees and leads to the almost exclusive rearing of queen-laid males. On the other hand, additional genetic evidence suggests that, on rare occasions, workers from one patriline (paternal lineage) can somehow evade policing and undergo a dramatic burst of worker reproduction (Oldroyd el al., 1994). Most remarkably, honey bee queens apparently chemically label their male eggs, so allowing 'police' workers to discriminate between these eggs and those of the laying workers (Ratnieks, 1995). This arrangement is evolutionarily stable because it benefits both queens and the average worker. Honey bees therefore show how mlllual inhibition through 'policing' can lead to harmony inside Ihe colony despite the potemialfor kin conllict (Ratnieks & Reeve, 1992).
9.5 Conclusion The study of social evolution in insects is a dynamic area of research. The kin-selection approach has clearly been uniquely fruitful in improving our understanding of the evolution of both cooperation and conflict in insect societies. It also provides the best basis for integrating the study of social evolution in venebrates and invenebrales (e.g. see Chapter 10). Yet. in every area covered by this chapter, unanswered questions remain. For example, how well do the genetics, demography and sex ralios of facultatively social bees and wasps meet the assumptions and predictions of models for the origin of eusocialily invoking split sex ratio and life-history ellects? Are the quantitative predictions of skew theory met? What mechanisms apart from selective egg-eating do workers use in kin conllicts over sex allocation and male parentage? How widespread is worker policing? Other unsolved issues include two not dealt with in this chapter: (i) the adaptive signHicance of within-colony kin discrimination (Carlin, 1989; Gralen, I 990a; see Chapter 4) and (ii) the evolution of multiple mating (Crozier & Page, 1985; Bourke & Franks, 1995). Social insects Iherefore promise to remain a rich source of ideas and discovery in behaviour, ecology and evolution.
Chapter 10 Predicting Family Dynamics in Social Vertebrates Stephen T, Emlen
10.1 The changing scope of cooperative breeding research The focus of research erforts on cooperatively breeding birds and mammals has changed dramatically during the past two decades. Cooperative breeding refers 10 breeding systems in which adults provide significant care to young that are not their own genetic offspring. Such systems are now known to occur in roughly 3% of bird and mammal spedes. That number will undoubtedly increase as additional field studies are conduaed. At the lime of publication of the first edition of this volume in 1978, interest centred on verifying the exiSlence of 'helping' behaviours, on describing the diversity of types of helping associalions found in nature and on explaining the paradox that these seemingly altruistic behaviours presented. Did 'helpers' really contribute significant assistance to the breeders that they altended? If they did, how could helping others (rather than breeding oneself) be recondled with the tenets of natural selection? By the lime the second edition was published in 1984, a sullident number of long-term field studies had been conduaed to provide a generally allirmative answer to the first question, and to develop a framework for addressing the second. The most common form of group found among cooperative vertebrates was that of grown offspring helping their parents 10 rear younger Siblings. The second question thus became partitioned into two. Why did offspring remain with their parents rather than disperse and attempt 10 breed independently on their own? And why did such grown offspring help, rather than ignore (or even hinder), the breeding ellorts of other adults in their group? Ecological conslraints theory provided the most plausible answer to the first. Ollspring stayed home when opportunities for successful dispersal and independent breeding were 'conslrained' - when such opportunities were temporarily non-existent or of poor quality relative 10 the situalion at home. Answers to the second were more diverse. It turned out that helpers themselves benefited in several additive ways by their helping actions. They increased their own probability of becoming breeders in the future (by expanding or inheriting the parental territory itself). They were beller parents when they 228
FAM[LY DYNAM[CS [N SOCIAL VERTEBRATES
229
did breed (having benefited from the experience of helping). Finally, they increased their inclusive fitness by helping to rear close genetic relatives. The early 1980s also bore witness to a growing realization that connict goes hand in hand with cooperation, and that even in the most highly cooperative of societies, genetic connicts of interest are inevitable. [n particular, competition will exist between group members over who breeds and who does not. A general theory of such connici and its resolution, now termed reproductive skew theory, was in its early stages of development. These models (Emlen, [982a, [984; Vehrencamp, 1983a,b; Emlen & Vehrencamp, 1983) attempted to predict: (i) when dominant members in a group would monopolize reproduction rather than share it; and (ii) when shared, with whom and how equitably it would be shared. By 1991, the year of the third edition of this volume, numerous long-term studies had provided rigorous data on the actual costs and benefits of helping, bolh to breeding recipients and to helpers themselves. The original paradox of cooperative breeding largely disappeared with the widespread confirmation that: (i) helpers frequently do improve their chances of becoming breeders by staying at home and helping temporarily; and (ii) they frequently do obtain large indirect genetic benefits by helping to rear collateral kin. For a summary of such data, the reader is referred to Emlen (1991). With the original set of questions largely answered, I wish to shift my attention in this current edition to the topic of social dynamics within cooperative societies, and particularly to family dynamics. There are several reasons [or an emphasis on the family. First, the vast majority of birds and mammals that exhibit cooperative breeding do, in fact, live in multigenerational family groups (Box 10.1). Second, families provide an excellent arena [or developing Box 10,1 Family definitions In this chapter. the definition of families is restricted to cases where offspring continue to interact, into adulthood, with their parents. I categorize families
as simple or extended depending on whether reproduction is totally monopolized (skew = I) or is shared (skew < I). fn simple families only a single male and female group member breed, while in extended families, two or more group members of one or both sexes reproduce.
The presence of a breeding male is not essential to the definition of a family. Rather, the presence or absence of reproductive males
(OnTIS
the basis of a second
partitioning into biparental (also called nuclearor conju,qaf), versus matrilineal, families. It is usefulLO further differentiate between inlaCI families, those where the
original breeders are still the reproductives, and replacemene (or step-) famllie, where, because of death, divorce or departure, a breeder has been replaced. Despite these differences, most vertebrate families form in the same way when offspring delay dispersal and continue
10
reside wiLh their parent(s).
230
CHAPTER 10
and testing evolutionary social theory because families are comprised of genetic relatives tbat vary both in their degrees of relatedness and in their sodaI dominance ranks relative to one another. Relatedness and dominance are expected to be major predictors of the social dynamics of any group. By focusing on families one seeks similarities. rather than differences. among cooperalive species. In most cooperatively breeding birds. adults form long-term pairbonds and males contribute considerably to the care of dependent young. Similar groupings occur in some canids. bUI pairbonds among mammalian cooperative breeders more typically are lacking or are of short duration. Fathers can even be absent from the group during the time of offspring dependency. Despite these differences. social groupings in both avian and mammalian cooperative breeders typically form via the retention of grown young wilh their parent(s). The result is a tight kin group. as well as a group with a builtin generational dominance asymmetry. These common characteristics feature prominently in the predictions developed later in this chapler. Finally. focusing on the family struclure of many animal societies highlights parallels with the early social organization of our own species. It raises the question of whelher findings emerging from animal studies can usefully be extrapolaled to better understand both Ihe past and the current human condition. Many of the ideas expressed here have been developed elsewhere (Emlen, 1994, 1995a.b). A list of specific predictions concerning the social dynamics of kin groups is reproduced in Appendix 10.1 at the end of this chapter.
10.2 Ecological constraints and the formation of family groups In most organisms. young disperse from their area of birth well before or at least by the age of sexual maturity. In seeking independent breeding opportunities, they cease interacting with their parents. If offspring and parent come into contact later in life. they show no signs of recognition or preferential interaction Wilh one anOlher. What sets species that form multigenerational families apart is the tendency for offspring 10 remain in association with their parent (s) beyond the age of sexual maturity and, commonly. lhroughout their lifetimes. The key to understanding Ihe evolution of families is understanding delayed dispersal. Why should a growing offspring poslpone its dispersal? Imagine you are a Darwinian accountant. Your lask is 10 maintain a ledger sheel on which you record the probable COSIS and benefits associated with the alternalives of: (i) remaining al home and continuing to associale Wilh parent (s); or (ii) leaving home and attempling 10 reproduce independently. Families are expecled 10 form only when lhe expected lifetime fitness associated with the option of Slaying home exceeds that associated Wilh early dispersal. What factors might tip the balance in favour of staying?
FAMILY DYNAMICS IN SOCIAL VERTEBRATES
231
Most mature birds or mammals that remain in their natal group do not reproduce; their breeding is suppressed by their more dominant parents. There is thus an almost automatic cost to staying home - the forfeiture of fitness associated with missed reproductive opportunities (Emlen, 1994). All else being equal, offspring would be expected to leave home. But, all else is not always equal. Suppose there are very few 'vacancies' for dispersing individuals, or that the vacancies are of poor quality. Competition for the few good vacancies will be intense, lowering the probability of successful establishment by a dispersing youngster. Add to this a heightened mortality risk associated with dispersal itself, and the benefits on the 'leave home' side of the ledger sheet begin to pale. As ecological constraints on the dispersal option intensify, it takes fewer benefits in the 'stay home' ledger column to tip the balance in favour of remaining with one's parents. Some researchers have emphasized the importance of the constraints on leaVing; others the benefits of staying home. I view the two as complementary. (For further discussion of this issue, see Woolfenden & Fitzpatrick, 1984; Brown, 1987; Stacey & Ligon, 1987, 199 I; Zack, 1990; Emlen, 1991, 1994; Koenig el 01.. 1992; Mumme, 1992a; lack & Stutchbury, 1993). Together these arguments form thc basis of an economic model of family formation (Emlen, 1994, 1995a). Statcd simply: I delayed breeding occurs when the production of mature offspring exceeds the availability of acceptable opportunities for their independent reproduction; 2 under such circumstances, some offspring must postpone personal reproduction until acceptable breeding opportunities arisc and they are able to successfully compete to obtain them; 3 families wiU fonn when such waiting is best done at home, when remaining on the natal territory and/or associating with one's family members somehow augments the offspring's inclusive fitness. Family groupings are thus expected to be inherently unstable. They should form and expand when independent breeding opportunities are constrained, but decrease in size and break up as outside reproductive opportunities improve. In essence, family social organizations can be viewed as 'solutions' to the often temporary problem of a shortagc of acceptable breeding opportunities. If offspring are assessing the relative fitness profitabilities of staying home versus dispersing, then offspring residing on territories of high quality should require higher quality outside reproductive opportunities to induce them to leave home. The result will be greater stability of families that control highquality resources.
The following studies illustrate the predictability of family formation and dissolution. Acorn woodpeckers, Melonerpes!ormicivorus, live in groups of two to 12 individuals and typically occupy pennanent, year-round territories in the American Southwest. High-quality territories arc in short supply, and most offspring become reproductives by waiting for an established breeder on an existing territory to die, and then successfully competing to fill the breeding
232
CHAPTER 10
vacancy (Koenig & Mumme, 1987). With assistance from numerous researchers, I (Emlen, 1984) compiled field data from several locations across many years on: (i) the proponion of offspring that remained on their natal territories; and (il) the local availability of breeding vacancies. As predicted, the data show a clear, increasing tendency for yearlings to delay dispersal and remain home with decreasing local availability of breeding vacancies (Fig. IO.la). All family members in acorn woodpeckers help build and defend large granaries in which thousands or acorns are stored. These acorns provide an essential resource during times of food shortage (Koenig & Mumme. 1987). By constructing granaries, the birds increase the real estate value of their territory. The quality of territories can vary greatly within a local area. Both nesting success and adult survival are positively correlated with granary size. Stacey and Ligon (1987) found that offspring from high-quality territories were unlikely to disperse to fill vacancies in lower quality areas. Specifically, a higher percentage of yearlings remained (eventually ascending to breeder status) on territories with large granaries. Thus. family stability was greatest on the best quality territories (Fig. 10.1 b).
lal >
~
'00
•' ••
0; '0
;0
£
80
•
w
::< />
'-
60
>0 o~
40
C~
~~ ~
E
..
. .0 o
~
C>
•
20
~
~ ~
n.
0
80 40 60 20 Proponion of territories becoming vacant
o
Fig. 10.1 Ecological constraints
and family formation in acorn woodpeckers. (a) The proportion
of yearlings that remain al home,
(b)
plotted as a function of the 52
severity of territorial shortages. (From Ernlen, 1984.) (b) The
likelihood thaI yearlings stay al home. ploued as a function of the quality of their natal
40
rerritories. Numbers above
'
31
I
histograms are sample sizes of yearlings in each territory quality category. (Data from Slac~y &
·1 Low
Medium Natal territory quality
High
Ligon. 1987. from a popuJalion experiencing moderale terrilorial shorlagt'.)
FAMILY DYNAMICS IN SOCIAL VERTEBRATES
233
Superb fairy-wrens, Malurus cyaneus, live in souIheastern Australia in families consisting predominantly of parents and grown sons. Shortages of both territories and mates (females) have been suggested as possible constraints to independent breeding (Rowley, 1965; Emlen, 1984; but see Rowley & Russell, 1990). Pruett -Jones and Lewis (1990) tested these ideas experimentally. By removing breeding males from nearby territories, they created breeding vacancies where none previously had existed. The result was the dissolution of virtually all family groups under observation. Thirty-one of 33 mature sons left home to fill the newly created breeding vacancies. Analogous results have recently been reported for red-cockaded woodpeckers, Picoides borealis, by Walters et 01. (1992). The Seychelles warbler, Acrocephalus sechellensis, is a formerly endangered species whose range is restricH::d to a few small islands north of Madagascar. In
1960, when the population consisted of just 26 individuals, habitat reslOration programmes were implemented. Over the following 30 years, the population grew impressively. No family groups were reported among Seychelles warblers until 1973, roughly the time at which all suitable breeding habitat became occupied (Fig. 10.2a). In essence, families formed when acceptable breeding opportunities first became constrained (Komdeur, I992). In subsequent years, the population of mature birds has consistently exceeded the number of occupied territories. Family groups have been the norm.
To enhance the numbers of this endangered species, birds were introduced onto two nearby, previously unoccupied, islands. To obtain birds (or these
[a)
200
400
160
300
:!
'.
'2 200 iii 100 ~
120 80
FamIly
formation
0
1960
1970
1980
40 0 1990
Year
a_ 100
41 26
75
>~
''' -' ;«; ~~
e.<:
~-
Q.
25
.~
.!
Seychelles warblers. (a) The numbt'r of individuals and occupied rerrilorics on Cousin
Island between 1959 and 1990. Family formation was first observed when all available territories became filled. (Modified from Komdeur,
J 992.)
stay at home, plotted as a function of the quality of their nalallerrilories. Numbers above
c~
:':0
-:0'~ >- 50
~ ~
·0
(b) The likelihood that yearlings
Ib)
m8.
g
Fig. 10.2 Ecological constraints and raOli1y rormation in
hislOgrams are sample sizes of yearlings in each territory quality
56
cal~gory.
(Data frum Komdeur,
1992, from a populalion
OL-J-,---'--,:..,-..,--'--'-:--:-LMedium
High
Natal territory Quality
experiencing severe lerrilOrial shortage (Cousin Island. 1986-
90).)
234
CHAPTER 10
transfers, Komdeur removed breeding adults from occupied territories on the original island. He thus experimentally created breeding opportunities in a manner analogous to Pruett-Jones and Lewis (1990). As with the fairy-wrens, the manipulations caused widespread family dissolution. All experimental vacancies were rapidly filled by malure offspring thaI previously had been residing with their parents. Komdeur had also independently evaluated the quality of all territories bOlh in lerm of adult survival and nesting success. He discovered lhat dispersals to his newly created vacancies followed the same pattern as in acorn woodpeckers - offspring only filled vacancies thaI were of equal or higher quality than their natal situations. Offspring from families residing on highquality areas thus had fewer 'acceptable' oUlside opportunities; consequently they remained home for longer periods of time (Fig. 10.2b). If this result proves general. families controlling the highest quality resources will be those with the greatest lemporal stability. One result of prolonged residency at home is an enhanced probability of ascending to the parental breeding position itself, in effect inheriting the family resources. Such inheritance of the family terrilory has been reporled in a large number of species and represents a major route to achieving breeding status among family dwelling species of both birds and mammals (see Macdonald & Moehlman, 1982; Emlen, 1984; Brown, 1987; Stacey & Koenig, 1990; Solomon & French 1996). When offspring from successive generalions stay home to inherit the family holdings, the result is the continuous occupancy of the same area by a genetic lineage - the (ormation o( a family dynasty. As long-term monitoring sludies of familial spedes continue, 1 expect the discovery o( dynasties to become commonplace. It will then be possible to leSI the prediction lhat dynastical inheritance will occur preferentially in those family groups that control the highest quality resources.
10.3 Kinship and the tendency to cooperate Families are fundamentally different from other forms o( soda I groupings because they form by the retemion of grown young Wilh their parents. As a result, families are comprised primarily of close genetic relatives. Inclusive fitness theory explains how an individual may enhance its fitness in two ways, directly through the production of its own offspring, and indirectly through its positive ellects on the reproduction o( relatives (Hamilton, 1964). Kin selection has long been hypothesized to be a selective factor (avouring the evolution of cooperalive breeding (e.g. Hamilton, 1964; Brown, 1978; Emlen, 1978, 1984; Vehrencamp, 1979). It predicts that assistance in caretaking should be more prevalent in family groups than in groups of less related individuals. More specifically, within families, such assistance should be expressed to the greatest degree between those individuals that are the closest genetic relatives.
FAMILY DYNAMICS IN SOCIAL VERTEBRATES
235
With help from N.J. Demong. I searched the Iiteralure for all case, of reported family social structure in birds and mammals. We found 112 species of birds and 63 species of mammals (excluding primates) reported to live in multigenerational family groups. Of lhese. fully 96% of the birds and 90% of lhe mammals exhibit cooperative breeding (Emlen. 1995a). Because of the difficulty in generaling a complete list of family-living species these figures are undoubtedly overestimates. but their true values will remain high. Assistance in rearing young appears to be the norm within vertebrate family groups. We next examined all species for which cooperative breeding has been reported and asked what percentage are family-dwelling. Fully 88% of the birds and 95% of me mammals that breed cooperatively live in multigenerational family groups (Emlen. 1995a). Cooperative breeding as a reproductive system in social vertebrates is thus largely restricted to familial societies. To test the prediction that individuals preferentially assist their closest relalives. I examined data from species that live in extended families. Multiple females (or pairs) can breed simultaneously in such families. thereby providing potential alloparenls with choices in whom to aid. Data available to date provide strong support for the prediction. White-fronted bee-eaters. Mtrops bullockoides (Emlen & Wrege. 1988). Galapagos mockingbirds. Nesomimus parvulus (Curry. 1988). bell miners. Manorina mtlanophrys (Clarke. 1984. 1989). noisy miners. Manorina mtlanocephala (PoIdmaa. 1995). pinyon jays. Gymnorhinus cya110cephaius (Marzluff & Balda. 1990) among birds. and lions. Pa11lhtra leo (Pusey & Packer. 1994). brown hyenas. Hyaena brunnea (Owen & Owen. 1984). and dwarf mongooses. Helogale parvula (Creel et al.. 1991) among mammals. all show the predicted preferential allocation of aid. Only one species. the Mexican jay. Aphelocoma ultramarina. is reported to show no apparent kin favouritism; breeders provision fledglings from other nests in their family as much as their own (Brown & Brown. 1990). My own work on while-fronted bee-eaters (Emlen & Wrege. 1988) provides a clear example of a species whose behaviour supports these predictions. These birds live in extended family groups in which up to four pairs may breed simultaneously. Helpers are non-breeding individuals that join one of the active nests and aid in incubation as well as neslling and fledgling care. Only 50% of Ihe non-breeders become helpers; the rest sit out the season as non-participants. Kinship proved to be a strong predictor of bom: (i) whether a given individual becomes a helper; and (ii) to whom it provides aid. Non-breeders are most likely to become helpers when the breeding pairs in their family are close genetic relatives. When faced with a choice of potential recipient nests. they preferentially help the breeding pair to whom they are most closely related (94% of 115 cases; Fig. 10.3). There is an additional genetic reason for increased amicable behaviour within intact families: sexual competition is predicted to be largely absent. One of the major recent discoveries in the area of maling system research has been the high frequency of extrapair fertilizations in species previously thought
236
~
~
u
'0
'u
(; ~
'a
C ~
CHAPTER 10
100
-
~
~
~
u
'0
'u (; ~
'a
C ~
'-
-
~ ~
23
25 0
100 75 50 25 0
';
Fig. 10.3 The importance of kinship lO helping decisions in white-fronled bee-eaters. Graphs show choices of nests actually
50
~ ~
24 r-
75
~
'-
30
assisted by helpers Ihal had
,---, 0.50 0.25
0.50 0.12 21
15
r-
-
n
0.25 0.12
0.50 0.00
multiple recipient nests availahk
-
within their family groups. Data are presellled as dyadic
4
comparisons of nest choices ploned according to the helper's relatedness to the recipient nestlings. Numbers above histograms are sample sizes of helpers in each choice
I,...., 0.25 0.00
0.12 0.00
Coefficients of relatedness
comparison. (Data from Ernlen & Wregc. 1988.)
10 be monogamous (e.g. Birkhead & M0l1er, 1992). This risk of cuckoldry has selected for intense mate-guarding and other forms of aggressive defensive behaviour by males during their females' fertile periods. Within family groups. however, most potential extrapair sexual partners are close genetic relatives. It is well documented that incestuous matings between close kin have deleterious genetic consequences in most normally out bred species (Ralls ef al., 1986, 1988; Thornhill, 1993; Jimenez ef al., 1994; Keller er al., 1994). For this reason, natural selection is expected to have fostered the development of inbreeding avoidance mechanisms. Thus, sons are rarely expected 10 compete with their fathers, or daughters with their mothers. lor sexual access to the parent of the opposite sex. Neither will siblings compete for sexual access with one another. Instead, mating partners will be selected from outside the family group. Family-dwelling species provide an excellent testing ground for incest avoidance predictions because mature offspring remain in close social contact with their parents and siblings throughout subsequent reproductive episodes. They thus have unparalleled opportunities to interact sexually with other family members. A review of the literature indicates that, despite such opportunities. inceslUous matings within families (parenl-offspring or sibling-sibling) are statistically rare. Fully 18 of 19 avian, and 17 of 20 mammalian (non-primate). familial species lor which relevant data are available show strong tendencies to pair exogamously (Em len, 1995a). Mate-guarding, courtship disruption and other
FAMILY DYNAMICS IN SOCIAL VERTEBRATES
237
forms of sexually-related aggression are consequemly expeded to be reduced within multigenerational families, promoling more harmonious imeraclions within the group. Incest avoidance, however, is not universal among vertebrales. As ecological constraims on the option of independent breeding become increasingly severe, a poinl may be reached where it is better 10 breed incestuously than to risk not breeding al all. These conditions have been modelled by Bengtsson (1978) and Waser el al. (1986).
10.4 Conflicts with changing family composition The death, divorce or departure of a breeding parent, and its replacemem from outside the group, will alter the basic genetic and dominance strudure of the family unit. As a result. many aspects of the resulting social dynamics of replacemem families (the equivalem of slepfamilies in the case of socially monogamous species) arc predicted to be differem from those of biologically imact families. A replacement mate will typically be unrelated to extam family members. As such, it will be exempt from incest rest rid ions. The arrival of such an individual creales pOiemial reproductive oppurtunities for subordinates that were closely related to the deceased breeder. For example, a son could obtain a share of personal reproduction by mating with its stepmother, or a daughter could do likewise by reproducing with its stepfather. While such shared mating might be advamageous to the replacemem mate as well as 10 lhe subordinate. it generally will be disadvantageous to lhe surviving parent. Males will incur lost paternity; females will incur reduced male contribution 10 parental care. The surviving breeder is thus expected 10 strongly oppose extrapair mating auempts both by subordinates and by its new mate. The result is a predicled increase in sexually-related aggression. Siripe-backed wrens, Campylorhyl1chus nuchalis, provide an example of this change in sexual dynamics following a paremal repairing. These birds live in nuclear families of two to seven individuals in the savannas of Venezuela. Piper and Sialer (1993) contrasted lhe behaviour of sons in intact versus replacement families. Sons displayed no sexual interest in their mothers, but lhey frequently engaged in courtship activities with their stepmothers. Paternity dala showed that sons never sired offspring wilh their biological mothers. but they did with lheir stepmothers. This increase in sexual-behaviour was accompanied by an intensification of sexual competition between fathers and sons. When sons attempted to consort sexually with their stepmothers. fathers upped their mate-guarding behaviours and became increasingly aggressive toward their sons (Fig. 10.4). Analogous changes occur when a breeder takes a new male in socially monogamous. extended family situations. Among white-fronted bee-eaters, male offspring pair with unrelated females but bring their males home 10 breed
CHAPTER 10
238
100
•
0
80
• '0 60 •0> 40 E •~ 20 • 0
.J!..
~
..
l
8
• • 0
~ 23
o tal Courtship (by son)
(
--!2..
lb) Male-
guarding
(by father)
0
6
~
'0 4
..,~
E ~
'
Ic) Direct aggression
(father-sonl
Z
2 0
Cd) Sirings (by son)
Fig. 10.4 Differences in sexually related behaviours within intact (open bars) and
replacemenl (shaded bars) families of stripe· backed wrens. Histograms plot the likelihood that: (3) sons engage in courtShip adivilies with their mothers or stepmothers; (b) fathers guard their original or new mate; and (c) falhers engage in direct aggression against their sons. Numbers above histograms are sample sizes of cases. (d) The actual number of detected cases of sons siring offspring with their mothers or stepmothers. (Data from Piper & Slater, 1993.)
within their natal family groups. Sons show no sexual interest in Iheir mothers, but brothers do allempl sexual acLivily wilh their sislers-in-Iaw and falhers occasionally copulate with their daughters-in-law. Males maled 10 such females closely monilOr allofeedings 10 Iheir males and physically disrupt mounting allempls by any extrapair males. These behaviours are in stark contrast 10 the dynamic between males when the breeding female is a close genelic relalivc of the eXlrapair males. In Ihis context, the female is nol mate-guarded (i.e. falhers do nol monitor feedings by sons 10 Iheir mothers). Replacement males, for their pan, are nOI expected 10 invest substantially in dependent young remaining from previous breedings because they are genetically unrelated to such oUspring. Consequently, they will gain lillie filness benefil by investing in the continued care of such individuals. In fact, continued care for extant young may be costly 10 a new, incoming male if such care significantly delays ils own reproduction or decreases Ihe survival probabilily of its own, future offspring. When Ihese COSIS are sufficiently large, the replacement mate may benefit by permanently evicling, or infanticidally killing, dependenl young remaining from a previous breeding. A dependenl offspring whose parent lakes a new male is Ihus prediclcd 10 be at increased risk for neglect, abandonmenl and/or even dealh caused by Ihe step-parenl. This risk will be greatesl when Ihe step-parenl is of Ihe physically dominanl (Iypically male) sex. This poinl has been repealedly confirmed in studies of family-dwelling rodems, carnivores and primales (reviewed in Hausfater & Hrdy, 1984), as well as in many species of birds (e.g. Slacey, 1979; Slacey & Edwards, 1983; Rowher, 1986; Emlen et at., 1989; Koenig, 1990). Offspring in replacemenl families also suffer a reduction in Ihe indirect filness benefit available 10 them from helping lO rear fUlure young, should the
FAMILY DYNAMICS IN SOCIAL VERTEBRATES
239
new pair (parent plus step-parent) breed. The offspring will share only 25% of their genes with future half-siblings, in contrast to the 50% shared with fullsiblings (produced by both biological parents) or with offspring of their own, should they leave and attempt to breed independently. All else being equal. this reduction in indirect benefits is predicted to lead to decreased family cooperation: o((spring will exhibit a reduced tendency to provide assistance in the rearing of half-siblings. The available data pertinent to this prediction are mixed. White-fronted bee-eaters, Florida scrub jays and Seychelles warblers exhibit the predicted adjustment in helping behaviour (ST. Emlen, N.J. Demong and P.H. Wrege unpublished data; Mumme, I992b; Komdeur, 1994, respectively). These studies contrasted the likelihood that non-breeding family members would feed young at nests where both parents were breeding with those where one, or both parents had been replaced. Each study found that the proportion of non-breeders which helped decreased when step-parents became breeders (Fig. 10.5). In the Seychelles warblers, those individuals that did help also decreased both their level of provisioning and the number of days they provisioned in families with replacement mates. Red-cockaded woodpeckers and stripe-backed wrens, in contrast, did not alter their feeding rates when provisioning full- versus half-siblings (Rabenold, 1985; Walters, 1990). In the wrens, the prediction of reduced helping is confounded by the recent finding that offspring engage in copulations with their stepmothers (see Fig. I DAd; Rabenold et al., 1990; Piper & Slater, 1993; W.H. Piper, personal communication). Such shared breeding may tum out to be common in replacement families. When this occurs the amount of provisioning ta)
~ 5
10
~
Full Sibs
Half Sibs
Non-kin
{bl
Fig. 10.5 The likelihood that non-breeding family members
100
75
will serve as helpers. plotted as a function of family Iype (full-
53
50
siblings = intact families: halfsiblings and non-kin = replacement families with one, and no biological
--l.!.-
parent remaining. respectively).
(a) florida scrub jay. (Data frum Mumme, 1992b.) (b) Seychelles
25 0
0 Full sibs
Half sibs
Non-kin
warbler. (Da13 from Komdeur.
1994.)
240
CHAPTER 10
expected may depend on the subordinate male's probability or parentage (since the dependent young would now consist of a mix of half-siblings and offspring of the 'helper'). Additional data on changing patterns of helping behaviour following mate replacement, as well as information on the likelihood of helper reproduction with a step-parent, are greatly needed. Finally, replacement families are predicled 10 be less stable than intact, biological families. This will be true whether instabilit y is defined in terms of increased dispersal tendencies of mature offspring (family dissolution) Or in terms of increased separation rates of re-paired breeders (parental divorce). When reproduction occurs in replacement families, the result is the coexistence of offspring from different sels of biological parents (Fig. 10.6a). A simila r mixing occurs when the replacement mate brings offspring of its own into the new pairing (creating what sociologists call 'blended' families; Fig. I O,6b). Such mixing is prediCled to intensify conflicts between different family members. Trivers (1974) was the first to stress how asymmetrical levels of kinship inevitably produce evolutionary conflicts of interest between parents and their offspring. Because offspring, on average, share only 50% of their genes with their siblings, they will have been selected to devalue the welfare of siblings, relative to themselves. Parents, because they are equally related to all of their offspring, will have been seleCled to counter this devaluation by encouraging offspring to behave more altruistically toward siblings than is in the genetic interests of the offspring.
la) Stepfamily
lb) Blended family
Fig. 10.6 Generalized genealogical diagrams of two forms of replacement family. (a) A slt'pfamily formed from a formerly intact oucit'ar family when. following the d~alh of till'
original breeding female. the male parent takes a m:w male and reproduce'S again. (b) A 'blended' family in which twO formerly nuclear families. each with oHspring. join when lhe widowed male parent from one pairs wilh lhe widowed female of Ihe other. Diagrams shnw breeders (lOp line) and their offspring (c;ymbols connected by line'S In the breeders).
Individuals with an X through (hem an.' dece-ascd. DilIerem rows represent uHspring from sllt.'Ct.'ssive breedings. Arrows denote immigration into the family unil. Symbol shape' d~l1otes sex (SQuare'S. males; drdes. females). A represemative offspring. Ego. is shown as a filkd square. Letters within symbols denole relalionships of different family members to Ego. as folluws: FS. full-sibling. , = 0.5; HS, half-Sibling, , = 0.25; P. hiological parent. r = 0.5; SP. stl·p-parent. r = 0; S5. stepsibling. r = O. Familit.'s cuntaining members with many hi~h relatedness asymmetries are predicted tn have high levels of conf!icl and social streS'.
FAMILY DYNAMICS IN SOCIAL VERTEBRATES
241
Consider Ihe magnilUdes of thesl' conflicts when young from IWO or more different pairings arc pooled. Each surviving parent continues to be equally relaled 10 all of its offspring, bUI each offspring now shares only 25% 01 its genes wilh its half-siblings, and none at all with its slepsiblings. Offspring arc predicled to behave progressively less benevolently IOward half-siblings and slepsiblings, respectively, resulling in greater levels of conflict both between the sels of offspring and between the parents and young (see Briskie il ,,/.'s 1994 sludy, discussed in Chapler 7). The replacement parent is also expected to 'disagree' with its matt: over the allocation of its investment in its own,
versus ils stepoffspring. If these darwiniao prediclions arc valid, manifestaliuns of Ihis conflict should be detectable in higher rates of aggression belween predictable dyads of family members, leading to higher divorce and dissolution rates in replacement. ill comparison with intact, families. To my knowledge.
such data have not yet been repurled systemalically fur any familial species or bird or non-human mammal.
10.5 Conflict over who reproduces The preceding seaion deah with situalions where Ihe dominant pair monopolizes breeding (or the dominant female does. in the case of matrilineal families). BUI, subordinates would ohen increase their fitness if they could become reproductives themselves. A dynamic tension thus is expected in many family systems. a tension resulting from conflict over who will and who will not reproduce. There are two ways a subordinate may become a breeder within its natal
group. FirSl, it can simply replace the dominant breeder either by successfully challenging and displacing Ihe dominant, or by waiting to inherit the position following the lann's death or disappearance. Alternatively. it can share reproduction alongside the dominant. This laller produces a more egalitarian. extended family struclure. Consider the first option of simple breeder replacemenl. When a parent in a socially monogamous nuclear family dies, divorces Or disappears, a breeding vacancy is created. Recall that families form when independent breeding options arc constrained, and that a major tOute for becoming a breeder is 10 inherit the breeding position. The Joss of a parent creates Ihe opporlunity for a samesex offspring 10 ascend to breeding status on its natallerrilOry. There remains. however. a major obstacle - the surviving parent. The surviving parent usually allempts to retain Ihe breeding position itself. The seemingly obvious SolUlion of Ihe uffspring inceslUously pairing with its parent virtually never occurs (Emlen, J 995a), presumably because of the high costs of close inbreeding. Instead, whichever individual succeeds in filling the breeding position takes a new replacement mate from outside the family group. The loss of a parelll is thus predicted to result in overl competition between the surviving parent and its opposite-sex nlature offspring over the
242
CHAPTER 10
former's retention, versus the laner's assumption, of breeder status. The conflict may be severe, and can extend to interactions between the challenging offspring and any potential replacement mates that are courting the surviving parent. The outcome of such conflict will depend, in part, on Ihe relative dominance and fighting abilities of the participants. In most birds and mammals, dominance is influenced by gender and age. Males are Iypically dominant over females, and older individuals are dominant over younger ones. Daughters in socially monogamous families thus will seldom be able to challenge successfully and displace their widowed fathers for breeder slat us. Sons, however, frequently will be able to challenge and displace their widowed mothers. Vivid accounts exist of within-family power struggles for breeding status after the loss of a parent (e.g. Hannon el al., 1985; Zack & Rabenold, 1989; Zahavi, 1990). In avian nuclear family species, sons replace fathers as breeders much more often than daughters replace mothers (see Brown, 1987; Stacey & Koenig, 1990). In red-cockaded woodpeckers, whenever a father died, the mother dispersed and Ihe son assumed breeder slatus (23 instances; Walters, 1990). Active, forceful eviClion of mothers by sons following the death of fathers has been observed in Seychelles warblers p. Komdeur, personal communication), Arabian babblers, Turdoides squamiceps (Zahavi, 1990) and white-breasted robins, Eopsalrriageorgiana (M.N. Brown & R.J. Brown, personal communication). Analogous conflicls are expected in most matrilineal families. The death of a breeding female creates an opportunity for a subordinate female to assume the breeding position, and an increase in aggressive challenges is predicted at this time. Such power struggles are widespread in matrilineal family-dwelling mammals following the death of a dominant breeder (e.g. MacDonald & Moehlman, 1982; Solomon & French, 1996). The second way in which a subordinate can become a breeder at home is to share reproduction with Ihe dominant. This can occur if a subordinate either obtains a mate of its own and continues to live and reproduce within the original group, or if it shares sexual access to the mate of the dominant, leading to the produclion of broods or litters of mixed parentage. Each increases the direct fitness of the subordinate. Each, however, typically reduces the fitness of the dominant. When, then, will a dominant share reproduction wilh a subordinale? What determines how eXlensive such sharing will be, and among whom the sharing will occur? These questions are the domain of reproductive skew Iheory. (For a more formal treatment of this theory, Ihe reader is referred 10 Emlen, 1982a, 1984, 1995a, 1996; Vehrencamp, 1983a,b; Emlen & Vehrencamp, 1983; Reeve & Ratnieks, 1993; Keller & Reeve, 1994; Reeve & Keller, 1995; see also Chapter 11.) Skew refers to the distribution of direct reproduction among same-sex individuals within a group, and can vary from zero to one. Socielies with high skew are those where a few dominant individuals monopolize breeding. whereas societies typified by low skew have more egalitarian breeding. My
FAMILY DYNAMICS IN SOCIAL VERTEBRATES
243
categorization of families into simple and extended is based on the distinction of whether reproduction is totally monopolized (skew = I) or is shared (skew < I). This categorization, however, belies the fact that skew will often be a flexible attribute, with family structure changing predictably as ecological. demographic and social conditions change. To understand skew theory, we must rcturn to the basic idea that families fonn when grown off pring delay their dispersal and remain with their parents. Such dispersal decisions may be reversed at any timc, however, and it is likely that offspring reassess their oplions frcquemly. The profitability of dispersing (relative 10 staying home) is expected to change with changing conditions. The stability of the lamily unit, and the distribution 01 reproduclion among its various members, are expected to change as well. The presence of helpers often increases the fitness of the breeders they assist (their parents, in the case of nuclear families). Helpers may enhance the ability of the familial group to hold or enlarge ilS territorial holdings. By their alloparental assistance they may reduce the work load and thereby increase the survivorship of the breeders. Finally, lheir direct helping contributions may result in the production of increased numbers of offspring. Given the above circumslances, parents, not surprisingly, may be expected to cngage in behaviours thallead to the prolonged retention of lheir offspring in the group. One way a parent could influence the dispersal decision of its offspring would be to increase the payoff associated with the offspring's Slaying home. One way 01 doing this would be to relax its monopolization on breeding and allow the offspring to reproduce concurrently within the group. In the initial paper outlining lhe ideas of reproductive skew, I Slated that' ...breeders might be selected who yield a ponion of lheir fimess to auxiliaries... to lhe point where the fitness gained by the breeder via the assistance of the helper...equaled the fitness forfeited to insure the retention of that helper in the group' (Emlen, 1982a, p. 46). More recemly, Reeve and Ralllieks (1993) speak of the dominant offering a 'staying incentive'to keep the subordinate. The key point is that if a parent (or other dominant) can still realize a higher inclusive limess while sharing reproduction than it would if it monopolized reproduction bUI the ofIspring (or other subordinate) left the group altogelher, the result can be a 'win-win' situarion for both participants (Emlen, 1995a). Skew models identify four parameters which, together, specify both the conditions under whi h reproductive sharing should occur and the amount of sharing expected. These are: 1 the magnitude of any benefit realized by the dominant if the subordinate should stay; 2 the expecred success of the subordinate if it should leave; 3 the genetic relatedness between potential cobreeders (lhe dominanl and subordinate); 4 the relative asymmetry in dominance between them.
244
CHAPTER 10
Each parameter influences Ihe relalive payoffs of the staying versus the leaving option for both the dominanl and subordinate. Collectively. they delermine the leverage that the dominanl has in 'withholding', and the subordinate has in 'demanding', (anthropomorphically speaking) a share of reproduction. Predicting the outcome of conflicts over reproductive sharing requires the integration of all four parameters. However. for simplicity, I discuss the influence of each onc separately. 10.5.1 The benefit of group-living This provides the adaptive explanation for why dominant individuals share breeding at all. In the absence of such benefit a dominant is expected to be indifferent 10 the dispersal decisions of subordinates. Family dissolution will occur before reproductive sharing is expected. However, if the dominant realizes a sufficiently large benefit from the pre ence of the subordinate, it can forfeit some of its own direct fitness through shared reproduction and still do beller Ihan if it retained its monopoly on reproduction but lost the subordinate from tht, group. Provided that dispersal is a viable option for a subordinate, the greater tht· magnitude of the group-living benefit realized by the dominant, Ihe greater the potential leverage wielded by the subordinate. 10.5.2 The probable success of attempted independent reproduction This specifies Ihe profitability of the dispersal oplion available to a subordinate family member. The relative magnitude of this payoff. in cumparison to that of Slaying at home, is Ihe second major determinant of a subordinate's leverage. Consider the case where the chance of successful independent reproduction is near zero. (This might be the case when there are no nearby available breeding vacancies.) In such instances, parents have eXlreme leverage because an offspring has little option but to remain at home. Assuming parents are physically dominant over their young (see Section 10.5.4), strong monopolization of breeding is expected. The situation changes when opportunities for independent reproduction improve. As ecological constraints become relaxed, a threshold is reached where the fitness associated with dispersal exceeds that of remaining at home as a non·reproduclive. The offspring's leverage is enhanced: unless il can gain some direct reproduction within its natal group, it should leave. Two additional conditions typically must be met, however, before reproductive sharing is expected. 1 The parent(s) must bendit sufficiently from thc continued retention of the offspring. 2 A non-incestuous mating opportunity must exist for the subordinate.
FAM I LY DYN A M ICS IN SOC IA L VE RTE B RATES
245
Interestingly, if ecological constraints become too benign, it will often be in the interests of all parties for the subordinate to disperse, and families will again dissolve. There are two reasons for this: 1 the benefits of independent reprodudion for the subordinate may now exceed those of staying, even with shared reproduction; 2 the group productiVity benefit to the dominant may decrease to the point where there is no incentive for further retention of the subordinate. Reprodudive sharing thus is expected primarily at intermediate levels of severity of ecological constraints, when conditions afford viable independent reproductive opportunities for subordinates, but are not so benign that the benefits of continued group-living for dominants disappear altogether. 10.5.3 Kinship
Genetic relatedness between group members is the third critical parameter in skew models. One might expect that when a dominant breeder shares reproduction. it will do so preferentially with its closest genetic kin. In fad. exactly the opposite is predicted. This seemingly counter-intuitive result becomes understandable if we contrast the indired fitness benefits of remaining home as a helper for two subordinatc individuals that differ in their degrees of genetic relatedness to the dominant breeder(s). Let uS assume that in the absence of personal reproduction, the direct benefits of dispersing slightly exceed those of staying home. The close genetic relative gains a larger indirect benefit through its helping activities such that it requires a smaller 'staying incentive' (if any) to keep it in the family group. Conversely. the more distant or unrelated individual gains lillie or no indirect benefit. It therefore requires a larger amount of personal reproduction before it will pay it to remain. This prediction has been confirmed in numerous spedes. ranging from lions (Packer el al., 1991) and dwarf mongooses (Creel & Waser. 1991; Keane el al., 1994). to Pukeko. Porphyrio porphyrio. (Jamieson el al., 1994) and bee-eaters (Emlen & Wrege. 1992a.b). fn fact. the most egalitarian of all cooperatively breeding species are non-familial; cooperative groups of Galapagos hawks, Buleo galapagoensis, and groove-billed anis, Crolophaga sulciroslris. are comprised largely of unrelated individuals (Faaborg el al.. 1995; Koford el al.. 1990. rcspectively). Kinship considerations also predict that reproduction will be shared more equitably in sibling-sibling associations than in parent-{)ffspring groupings (Reeve & Keller. 1995). (Same-sex siblings become potential cobreeders in social vertebrates when they disperse as a coalition to fill a breeding vacancy, or when they competc to inherit a breeding slot following the death of a parent.) This is because the genetic relatedness (r) between each sibling and the offspring of the Olher will be symmetrical (r = 0.25 for each). All c1Sl' being equal, under conditions where reproductive sharing is favoured by one sibling, it will also be favoured by the other.
246
CHAPTER 10
In contrast, reproductive sharing generally will nOI be favoured between mothers and daughters (or fathers and sons). Assuming mate fidelity on the part 01 the parents, an offspring will be more closely related to its parents' future offspring (its full-siblings, r = 0.5) than the parents will be 10 the offspring's ol'fspring (their grandoffspring. r = 0.25). This reduces the offspring's leverage by creating a situation where parents have more to gain from withholding shared reproduction than the offspring do from demanding it (Reeve & Keller, 1995). Two addilional faclOrs reinforce the prediction of reduced reproductive sharing in parent-of[spring associations (Emlen, 1996). 1 The age asymmetry between parents and their young assures a dominance asymmetry as welL decreasing the likelihood thaI an offspring can directly challenge its parent for a share of reproduction. 2 In intact biparental families (where both biological parents are still a mated pair), the option of gaining sexual access to the dominant opposite-sex breeder is unavailable because of selection to avoid incestuous matings. Because these effects are additive, lhe contribution of each to Ihe maintenance of high skew in parent-offspring groupings will be difficult 10 determine (Emlen, 1996).
10.5,4 Social dominance and fighting ability Dominance hierarchies exist in most vertebrate families. Dominant individuals enjoy certain privileges (e.g. breeding status), but Ihey are always at risk of losing their 101' position. When a challenger is successful it typically ascends 10 breeder status. The costs of such dlallenges can be high for both participants, however, because fights may lead to injury, eviction and even death. All else being equal. the benefits of such challenges for a subordinate, as well as the risks for a dominant, will be greatest when the disparity in their fighting abilities is leas!. When the risk is suffiCiently greal. it will become advantageous for the dominant to share reproduction as a 'peace incentive' with its potential challenger (Reeve & Ratnicks, 1993). Such sharing increases the profitability 10 the subordinate of staying and continuing 10 cooperate within the group.
10.6 The myth of the stable family? It is the dynamic interaction of all four variables of reproductive skew theory that forms the basis of an integrated Iheory of family social dynamics (Emlen, 1995a). Such Iheory predicts that both family structure and social interactions among family members will change as conditions vary. Specifically, predictable changes should occur: 1 as the benefits of large group size wax and wane; 2 as ecological opportunities for independent breeding increase and decrease;
FAMILY DYNAMICS IN SOCIAL VERTEBRATES
247
3 as breeder deaths and replacements. as well as immigrations. alter family composition; 4 as the sodal dominance of individuals changes with age and experience. The idea of the stable family may thus be largely a myth. Even in dynastic siluations. where inheritance of the high-quality parental territory is favoured as the primary route to becoming a breeder. severe connict is expected among potential inherilees. Here 100. reproduction may become increasingly or decreasingly shared as outside opponunities. kinship and dominance factors change. Thus. my inilial categorization of family types (see Box 10.1) into simple versus extended. or intact versus replacement. should nOI be taken to imply fixed entities. but rather to describe various 'family states' that are expected to change in predictable ways with varying circumstances. The explanatory potenlial of reproductive skew theory in understanding these changes is illustrated below in three case studies. 10.6.1 Babblers (Zahavi. (990) Arabian babblers live in family groups on year-round territories in the Israeli desen. Ecological constraints are severe. and independent breeding opportunities are exceedingly scarce. Males often live their entire lives within their parental territory, and successful groups may pass their territories thruugh the male line over many generations (dynasties). Groups range from two to 22 individuals, with larger groups benefiting by greater stability and growth of their territories as well as by reduced nesling losses caused by intruding bands 01 floaters. Roughly 50% of groups are simple. nuclear families comprised of parents plus their various aged offspring. The remaining groups are extended families. most commonly consisling oltwo or more males sharing reproductive access to a single breeding female. Incest is strictly avoided. Thus, transitions in breeding structure typically occur following the dealh of a parenl. At such times severe fights may break out. during which the loser is either killed or permanently evicted from the group. zahavi witnessed 'five cases in which an offspring killed its parent of the same sex. follOWing the demise of its other parenl' (1990. p. 123). More typically. the loss of a breeding female is followed by acceptance of one or more new females into the group. Subsequently all adult male babblers have the oplion 01 breeding (because incest restrictions disappear) and several may copulate with the same lemale(s). The amount 01 mate-sharing depends upon rank. Old birds (typically fathers) do not share reproduction with younger males (typically sons), bUllWO males close in age and dominance (often siblings) do frequently engage in mate-sharing. These flexible changes in family structure. from nuclear to extended and from monogamy to mate-sharing. conform 10 the predictions of skew theory. Breeding individuals reap advantage from gruup-Iiving. while the options
248
CHAPTER 10
or subordinates for independent reproduction are severely constrained. As expected, dominant pairs typically mOnopolize breeding within the family group. Challenges for breeding staltts occur: 1 lollowing the death of a parent; 2 when the incest restriction is lifted; 3 when the indirect benefits 01 continued residence as a non-breeder would otherwise decrease. Furthermore, the challengers (the individuals that often become cobreeders) are more often of similar age (equal dominance) than of different ages. Specifically, lather and son rarely mate-share. whereas same age brothers frequently do. 10,6,2 Bee-eaters (Emlen. 1982b, 1990; Emlen & Wrege, 1992a,b) White-fronted bee-eaters live in socially monogamous, extended lamily groups or from two to 17 individuals. There is a strong group advantage in that helpers have a large effect on increasing the number of young produced. In our study area in central Kenya, the severity of ecological constraints varies unpredictably across years. In benign years, a large proportion of pairs breed. When conditions are harsh, breeding is more skewed and a larger proportion of family members act as helpers for the breeding few. Interestingly, not all such helping is voluntary. Dominant breeders harass subordinates, disrupting the lallers' breeding efforts. Natal members of such disrupted pairs typically then become helpers for their harassers. Skew theory predicts that dominant breeders have their greatest leverage over those individuals for whom the indirect benefits 01 staying are greatest (their closest genetic kin), and over those with whom the disparity in fighting ability is greatest (their youngest subordinates). Both are true. The most COmmon category of harassers are fathers disrupting the breedings of their youngest sons. Harassment 01 distant kin, and of similar aged individuals, is practically non-existent. As a result of changing social and ecological conditions, family dynamics are highly lIuid with the degree of reproductive skew changing in a predictable manner given the ecological constraints operating in each breeding season.
10,6,3 Mongooses (Creel & Waser, 1991; Creel et at.. 1991; Keane et al.. (994) Dwarf mongooses live in family groups of variable composition comprising three to 18 individuals. Dominant individuals benefit from living in large groups because non-breeding adults assist in various caretaking activities. The dominant female typically suppresses reproduction in subordinates. Occasionally, however, another female becomes 'pseudopregnam', undergoing hormonal changes resulting in the production of milk which is thell IIsed to
FA MILY D YN A M I C SIN SOC I AL V E RT E BRA T E S
249
nurse the young of the dominant female. These 'superhdper' females are virtually always [ull siblings of the young they nurse; as such they are the family members that stand to gain the largest amount of indirect fimess benefit from engaging in such costly help. At the other end of the spectrum are females that co-reproduce along with the dominant breeder. Only 13% of subordinate females produce litlers, but when they do they den their young with those of the dominant and all young arc reared communally. Skew theory predicts that dominant females should share reproduction preferentially with those females that either pose the greatest threat to their own status, or are at greatest risk of dispersing. Again, both are true. The few subordinate females that breed are among the oldest (I.e. the most physically dominant) and the least closely related to the dominant breeders (I.e. the least likl'iy to gain indirectly from staying in the group). DNA fingerprinting data has shown lhat reproductive sharing also occurs among male mongooses. Subordinate males sire 24% of offspring, mostly by copulating with the domina.nt female who then produces a litter of mixed
paternity. Again, as predicted by skew theory, 'those subordinates that reproduced were of high social rank and tended lO be distantly related to the same-sex dominant' (Keane el al., 1994, p. 65).
10.7 Toward a unified evolutionary social theory Wilson, writing over two decades ago (1975, p. 4) stated '...when the same parameters and quantitative theory are used to analyze both termite colonies and troops of rhesus macaques, we will have a unified science of sociobiology.' In my opinion, we arc very close to reaching that goal. The building blocks of such a unified science, as described in this chapter, consist of ecological constraints theory, kin-selection theory. social dominance theory and reproductive skew theory. Together they specify the conditions leading lO family formation and describe the faclOrs that influence family dynamics, structure and stability. Cooperatively breeding species have played, and will continue to play, a pivotal role in the development of evolutionary social theory. In the past twO decades, studies of such species have largely answered our initial questions concerning the evolution of altruism. In the decades ahead, they will proVide answers to a wider array of questions concerning social dynamics. I believe that the intellectual excitement of studying cooperative species no longer lies in their hallmark behaviour of helping, per se, but rather in the opportunities they provide for understanding the complex workings of kin-structured societies. As the search [or general principles continues, there will be a need for greater cross-taxonomic comparison. Too often researchers have partitioned
themselves as sllldying eililer social insects, social birds and lower mammals,
250
CHAPTER 10
or social J>rimates. This anilicial dislinction must be broken down so that social insect perspectives are incorporated into social venebrale studies and vice versa. Chapter 9, on social insccts, proVides grounds for optimism that similar theory applies to all social taxa (see also Reeve & Ratnieks, 1993: Bourke & Heinze, 1994; Keller, 1995). There is also a need for workers to re-examine existing data for evidence either supponing or refuting the lypes of predictions outlined here. When searching the literature for such evidence, I was disappointed by the surprising paucity of such analyses (Emlen, 1995a). Knowing that many long-term studies of familial species are underway or already completed, 1 believe that considerable unpublished dala already exiSI that could be used to test and improve existing theory. I urge my colleagues with comprehensive databases to reexamine them with this goal in mind.
Finally, I suggest to researchers planning future venebrate studies that they focus on species exhibiting shared reproduction (species living in extended family groupings). Since most of the currently available data come from nuclear family situations, information describing other family syslems will be panicularly valuable. Studies of complex, kin-structured societies are imponant fur an additional reason. They provide the best available models for understanding any heritable social predispositions that humans may possess. Not all persons will agree with this position. It is based on three assumptions (developed more fully in Emlen, 1995b). 1 The expression of many social behaviours is governed, at least in part. by heritable assessment algorithms and decisiun rules lhal have been shaped by natural selection. 2 Some of the variance in the expression of human social behaviours is influenced by decision rules that were selected during our long evolutionary history of living in family kin groups. If one is willing to accept assumption number two, where docs one look for animal models to provide hints of the types of decision rules which we humans might be predisposed to employ? 3 My third assumption is that organisms living in similarly structured socielies are those most likely to have evolved similar sets of algorithms and rules. Most anthropologists and evolutionary biologists believe that during our recent evolutionary history, humans have lived in multigenerational family-based societies (Lee & DeVore. 1968; Lovejoy. 1981; Foley & Lee. 1989; Smith & Winterhalder, 1992). The closest venebrate parallels. in terms of societal structure, arc not most primates, bUI rather arc family-dwelling, cooperatively breeding, birds and mammals. To the degree that early humans were socially monogamous with males proViding significant parental care, complex avian societies (where shared care is the norm) and lhe socielies of social canids will provide our best models. As anthropologists learn more abOUI the 'environment of evolutionary adaplation' of early humans (e.g. Alexander. 1979; Irons. 1990; Wright. 1994), we
FAMILY DYNAMICS IN SOCIAL VERTEBRATES
251
will be beller able 10 select the optimal animal analogucs. But, unless our conccpt or our ancestral social cnvironment changes drastically, the sllldy of complex. multigencrationat kin-structured societies will provide the looking glass by which we will corne to see and understand beller the human social condition.
Appendix 10.1 Evolutionary predictions of living within family kin groups (Emlen, 1995a)
Prediction I Family groupings will be inherently unstablc. They will form and expand when therc is a shortage of acceptable reproductive opportunities for mature offspring, and they will diminish in size or dissolve (break upi as acceptable opportunities become available.
Prediction 2 Families that cOlllrol high-quality resources will be more stablc than those with lower quality resources. Some resource-rich areas will support dynasties in which one genetic lineage continuously occupies the same area over many successive generations.
Prediction 3 Assistance in rearing offspring (cooperative breeding) will be more prevalent in family groups than in otherwise comparable groups comprised of nonrelatives.
Prediction 4 Assistance in rearing orrspring (cooperative breeding) will be expressed to the greatest extelll between those family members that are the closest genetic relatives.
Prediction 5 Sexually-related aggression will be less prevalelll in family groups than in otherwise comparable groups comprised of non-relatives. This is because opposite-sex close genetic relatives will avoid incestuously mating with one another. Mating parmers will be selected [rom outside the natal family group (i.e. pairings will be exogamous).
252
CHAPTER 10
Prediction 6 Breeding males will invest less in olfspring as their certainty of paternity decreases.
Prediction 7 The loss of a breeder will result in family conflict over the filling of the resulting reproductive vacancy. In the specific case of simple conjugal families. the surviving parent and it mature opposite-sex offspring will now compete for breeder status. The connid will be especially severe when olfspring are of the dominant sex. and when resources controlled by the family are of high quality.
Prediction 8 Sexually-related aggression will increase after the re-pairing of a parent. In the specific case of simple conjugal families. the surviving parent and its mature same-sex olfspring will now compete for sexual access to the replacement mate (step-parent). This conflict will be especially severe when the asymmetry in dominaoce between the surviving breeder aod its same-sex olfspring is small.
Pred iction 9 Replacement breeders (step-parents) will invest less in existing offspring than will biological parents. They may infanticidally kill current young when such adion speeds the occurrence. or otherwise increases the success. of their own reproduction. This will be more likely when the replacement mate is of the dominant sex.
Prediction 10 Non-reprodudive family members will reduce their investment in future offspring following the replacement of a closely related breeder by a more distantly or unrelated individual.
Prediction II Replacement (step-) families will be inherently less stable than biologically intact families. This will be especially true when offspring from the originally intact family are of the same sex as the step-parent.
Prediction 12 Reprodudion within a family will become increasingly shared as the severity
FAMILY DYNAMICS IN SOCIAL VERTEBRATES
253
of ecological constraints decreases, that is, as the expected profitability of the subordinate's option of dispersal and independent reproduction increases.
Prediction 13 Reproduction within a family will become increasingly shared as the asymmetry in social dominance between potential cobreeders decreases.
Prediction 14 Reproduction within a family will be shared more equitabLy when the potential cobreeders consist of siblings than when they consist of parent(s) and grown offspring.
Prediction 15 Reproduction will be shared most with t.hose family members to whom the dominant. breeders are least. closely related. In species in which dominant.s actively suppress reproduction by subordinates, such suppression will be greatest in those subordinat.es to whom the dominant is most closely related.
Chapter 11 The Ecology of Relationships Anne E. Pusey & Craig Packer
11.1 Introduction Whenever two individuals meet. Iheir behavioural interaction may have consequences that will influence all of their subsequent interadions. Each animal may obtain information about its companion's competitive ability or its propensity to cooperate, or one animal may behave in a manner that raises or lowers its companion's chances of survival. As a result, behavioural ecologists have long realized that social behaviour must often be interpreted in the context of long-term sodaI relationships (Hinde, 1983), and that these relationships arc likely to be complex. By living in groups, companions may benefit from reduced predation risk, improved defence of resources or communal rearing, but they also suffer from increased competition over critical resources. Relationships will therefore be strudured both by cooperation and by competition, and, because selfinterests will never entirely coincide, sodal relationships are expected to reflect a certain degree of coercion or compromise. Indeed, individuals can benefit by manipulating not only their own relationships, but also the relationships between their companions (de Waal. 1982; Cheney & Seyfarth, 1990). What theory helps specify the general principles governing the form of social relationships? Most models of social behaviour have been restricted to assessing lhe net effed of isolated behavioural interadions. However, treatments of repeated interadions are becoming increasingly common. In this chapter, we first consider how competition structures relationships and then examine the long-term consequences of cooperation.
11.2 Competitive relationships Animals generally compete for resources on an individual basis. However, the summation of competitive relationships within any given group can lead to complex patterns, and these have often been the subject of research in their own right. We start with the simplest pairwise interadions, discuss how pairwise relationShips may lead to hierarchies, then discuss how these hierarchies vary in form and intensity across species. 254
ECOLOGY OF RELATIONSHIPS
255
11.2.1 Competition in pairwise contests Single enccunlers If opponents meet for only a single encounter, game-theoretical models predict
that contestants will be more likely to engage in escalated fighting when the COStS of injury are low relative to the value of the resource (Maynard Smith & Price, 1973). However, when fighting has significant costs, disputes are more likely to be settled according to certain asymmetries between the contestants (Parker. 1974a; Maynard Smith & Parker, 1976). These asymmetries may be based on differences in resource holding power (RHP) such as size, strength or fighting ability, or they may be uncorrelated with RHP (Hammerstein, 1981). Detectable differences in RHP are expected to lead to some form of assessment, where individuals 'size each other up' before committing themselves to an escalated light, and the inferior competitor retreats before losing a costly battle. But, when assessment cannot provide a reliable prediction of the outcome and the costS of fighting are especially high, animals may settle their disputes according to an arbitrary cue. For example, individuals following the 'Bourgeois strategy' escalate if they are the first to reach the resource and retreat if they are a latecomer. Finally, if the value of the resource varies between individuals, the hungrier animal will be expected to tolerate greater COStS and thus fight longer or harder (Parker, 1984; Houston & MacNamara, 1988). Empirical studies have shown that all these factors influence fighting behaviour (assessment: Davies & Halliday, 1978; Clutton-Brock & Albon, 1979; Austad, 1983; bourgeois: Davies, 1978; Packer & Pusey, 1982; Waage, 1988; differences in resource value: Krebs, 1982; Rodriguez-Girones el al.. 1996), and Maynard Smith and Riechert (1984) have demonstrated how these different asymmetries interact (e.g. large rival versus small owner, etc.). Repeated encounters: dominance relationships and winner-loser effecls
In stable social groups, the same two individuals are likely 10 compete repeatedly. If the difference in RHP between each opponent is consistent and detectable, the superior competitor should consistently win each contest and the inferior should defer 10 its opponent. The pair may Lhen be said to have a dominance relaLionship. Dominance relationships pervade animal societies (reviewed in Wilson, 1975; Smuts et al., 1987; Langen & Rabenold, 1994; Fournier & Festa-Bianchet, 1995), and they have been measured by the outcome of fights in dyadic encounters, the direction of approach-retreat interactions and the direction of threats and/or submissive gestures. These measures are generally correlated.
However, certain features of dominance relationships suggest that Lhey involve more complicated processes than the simple summation of repeated
256
CHAPTER I I
contests over resources. First. although dominance relationships are frequently correlated with some measure of competitive ability (Wilson. 1975; Fournier & Festa-Bianchet. 1995) (Fig. 11.1a). they often appear to be more clear-cut than expected from the individuals' relative RHP. and some are even based on traits that are uncorrelated with RHP (e.g. seniority or age). Second. the pallern of interactions between pairs often changes through time. Initially. fighting may occur even in the absence of a resource. and the subordinate subsequently defers to the dominant without a figh!. These observations suggest that individuals may compete for dominance per se and learn from the first few encounters with their opponent (see Huntingford & Turner. 1987; Clullon-Brock & Parker, 1995). Both processes are well illustrated by the development of dominance relationships in blue-footed boobies (Drummond & Osorno, 1992). Two chicks hatch asynchronously in each nes!. The older chick frequently pecks and jostles the younger for the first few weeks even in the absence of food and then maintains dominance by less frequent but daily aggression. The younger chick responds to aggression with submissive gestures and rarely challenges the older. Thus, the older chick establishes dominance and subsequently gains greater access to food. The younger chick usually behaves submissively even if it grows larger than the older (females grow more quickly and are eventually larger than males). The importance of early experience is illustrated by a series of experiments that paired chicks with different backgrounds. When chicks that had been raised alone (and thus lacked any experience of dominance or subordinacy) were introduced to other singletons. the pairs established dominance relationships solely on the basis of size. BlII. when birds that had been raised in pairs were given new companions, dominant chicks maintained their dominance
even when they were smaller than the new subordinate. Although large subordinates challenged the dominant chicks more often, they did not fight as tenaciously as the dominants and were more likely to adopt submissive postures. Such winner-loser effects are common in species as diverse as crickets and mice (Humingford & Turner. 1987) and may explain why dominance is sometimes determined solely by age or seniority rather than by differences in RHP. In some species, dominance changes with age according to a bell-shaped function which reflects the greater RHP of prime-aged animals (Fig. J l.Ia). However, in other cases. dominance appears to increase continuously with age (e.g. dwarf mongooses: Creel ft al.. 1992; cooperatively breeding birds: sec Chapter 10), although this may still reflect rising RHP if very old individuals are not able 10 survive in the population. Among male hyenas (Smale el al.. 1996). dominance is determined by the length of time an individual has tl'sided in the group (Fig. 11.1 b). Older or senior individuals presumably enjoy an initial competitive advantagt' so that the younger or newer individuals learn to defer and then continue
(a) Male olive baboons High
Low
Young
Old
Age lb) Male sponed hyenas
•
High
•
• •
•
•
•
•
•
• • • Old
Short
1
Duration of residence
Long
------1f---<11
2
3 - - - -....
• ----------40
•
5 -----,-----41
~ ::;:=;~'----.----j
B
g---
Old
10 - - , - - - - -.... 11 • 5 6 7 B 9 10 11 12 Young Old
Age
Fig. 11.1 The relationship between age and dominance in three diHerent mammalian species. (3) In immigrant male baboons. dominance shows a bell-shaped relationship because prime~aged males have the highest RHP. (Modified from Packer, 1979.) (b) In immigrant male hyenas. dominance is nOI correlated with age. but instead increases with the length of residence in Ihe dan. (left-hand graph provided by K, Holckamp & L. Smale. personal communication; right-hand graph is modified from Smale et al.. 1997,) (c) In female rhesus macaques. dominance is unrelated to age except as revealed by birth order. In the righHland graph. each female's age is indicated by the length of a horizomalline. and her maternity is indicated by each vertical line. Six females were daughters of a deceased female that is marked with a circle in the top-right corner, Note that each daught.er outranks all of her older sisters with the exception of Ihe second-ranking female. Also note that each daughter ranks beneath her mother but above every adult that ranks lower Ihan her mother. (Both graphs modified from Hinde. 1983.)
258
CHAPTER II
to do so. However, there are currently no theoretical models 10 explain why some species show such persistent deference and thereby form breeding 'queues'. Specifically, we need to know why RHP provides the basis of pairwise dominance relationships in some species whereas other species form long· term queues tbal are uncorrelaled wit h RHP (see Ens et at.. 1995). Indeed, persislent respect of an arbitrary asymmetry is unexpected unless latecomers gain some incentive from waiting their turn (see Grafen, 1987; Schwagmeyer & Parker, 1987), although there may be some insights into this pattern from the study of alliances (see Section 11.2.2).
11,2.2 Dominance hierarchies
Hierarchies as the outcome ofpairwise interactions Dyadic dominance relationships can often be arranged in a linear hierarchy in which individual A is dominanlto the rest of the group, B dominates aU but A, C dominates all but A and B, and so on (Schjelderup·Ebbe, 1935; Walters & Seyfarth, 1987). This strict linearity is difficult to explain on the basis of pairwise contests. Unless individual differences in RHP are extreme, the outcome of a dyadic fight will sometimes differ from the direction of RHP differences, especially between animals with near-average abilities. Consequently, the correlation between dominance and RHP will be imperfect, and, since the probability that a hierarcby will be perfectly linear depends on the product of all these dyadic probabilities, linearity should decline rapidly with increasing group size (Landau, 1951; Chase, 1974). Using a game-theoretical model, Mesterlon-Gibbons and Dugatkin (1995) show that the probability of linearity is increased where individuals assess their relative RHP in each dyad and fight only if this exceeds an evolutionarily stable threshold. This threshold increases with the magnitude of variation in RHP, the reliabilily with which RHP predicts the outcome of a light and Ihe ralio of the COSt of fighting to the value of winning. Such assessment reduces the number of fights and Iheir associated uncertainly of outcome. Nevertheless, their model still predicls that linearity should decrease with group size, becoming highly unlikely in groups larger than about nine individuals. When the observed degree of linearity is greater Ihan predicled by these models, dyadic dominance relationships are likely 10 be ordered by additional factors beSides differences in RHP. These may include psychological reinforcement of 'losing status' through winner-loser effects (see Section I 1.2.1) as well as the ordering of relationships through alliances.
Alliances Some animals (e.g. female Old World monkeys and spotted hyenas) form stable linear hierarchies in groups much larger than to (reviewed in Chapais, 1'992;
ECOLOGY OF RELATIONSHIPS
259
Frank el at' I995a).ln most cases. these hierarchies are not based on individual RHP. Instead, daughters rank directly below their mother in reverse order of age. and the whole matriline ranks above the next most dominant matriline (Fig. I l.lc). Rank inheritance is achieved by mothers supporting their daughters against females from lower-ranking families and by supporting their youngest daughter against her older sisters (Kawai, 1958; Cheney, 1977; Datta, 1983). Females also intervene in disputes between non-matriline members, allying against individuals that rank lower than themselves (Chapais. 1983; Hunte & Horrocks. 1986; Netto &van Hoo!!, 1986). The consequences o((emale alliances on rank inheritance and maintenance is demonstrated by a recent series of experiments with Japanese macaques (Chapais. I 988a,b; Chapais el al., 1991)(Box Il.I).
Box 11.1 Experiments on rank detennination in Japanese macaques
Chapais and his colleagues performed experiments with a captive group of Japanese macaques consisting of three unrelated matrilines. A (1973)
C 11971)
B 11971)
1981
Cl
1982 1983 1984 1985 1986 811
1987
87
1988
Fig. 8.11.1 Composition of the group at the end of the study period (summer of 1989). Circles, females; squares. males. The alpha male is not represented. (From
Chapais (( al.• 1991.)
In the intact group, members of the A matriline ranked above the Band C matrilines. and 8 ranked above C. Chapais created a variety of experimental groups to examine how alliances in£luenced the inheritance and maintenance of dominance rank.
Inheritance of maternal Tank Lone females from high-ranking matrilines were placed with several females from lower-ranking matrilines. and an ally of the high-born female was later added. For example, female Al was placed with the C matriline. She dropped Co',inwrd on p 160
260
CHAPTER I I
Box 11.1 Conl'd.
in rank until she ranked below all the members of the C matriline. When her mother (A) was added to the group. A mail11ained her rank above the C matriline, and defended her daughter against members of the C matriline. so that Al eventually regained her rank above the C matriline. In all other cases. mOlhers or adult daughters were also able to help their daughlers or younger sislers to regain their rank (Chapa is. 1988b). Maintenance of rank In a series of 58 experiments, a female from a high-ranking matriline was deprived of her allies and placed with: I a single bUI older/larger low-born female; 2
two or three low-born sisters; 3 a complete subordinate matriline. OUI of 148 dyads of high- and low-ranking females. dominance reversals occurred in 56%. The likelihood thaI the female lost her rank depended on two faclors: (i) her absolute age; and (ii) the number of lower-ranking females. Females were most likely to maintain their rank when they were older and were placed with few rivals. For example. when a 3-ycar-old member of the A matriline (A3) was placed with a B female that was 1 year older, A3 maintained her rank, and she was also able 10 maintain her rank above a pair of B sislers. But, when she was placed with two C sisters, One of whom was 2 years older, she did not maintain her rank. Whcn a female aged 4 years or older from the A or B matriline was placed with an entire lower-ranking matriline. the female only managed to maintain her rank in five of J 2 cases. Forexample, the matriarch A, maintained her rank over the entire B matriline. but B was not able 10 maintain her rank above the C malriline (Chapa is. t 988a). Chapais concludes the following.
Competition for rank is ubiquiLOus in matrilineal hierarchies, with low-
ranking females constituting a potential and constant threat for any high-
ranking female. 2 Kin can form revolutionary alliances. 3 Rank maintenance is conditional on the high-ranking fcmale having enough alliance power 10 counter revolutionary alliances. 4 Compctilion for rank is somewhat constrained. In cases where Ihe highranking female was outnumbered, rank reversals only occurred when the power asymmetry (relative size/age or relative alliance power) was pronounced. Chapais refers to Ihis as a 'minimal risk strategy' of compelition for dominance. Omlrmud 0' p.161
ECOLOGY OF RELATIONSH[PS
26[
Box 11.1 Conl'd.
Non-kin alliances
Fig. B. 11.2 Distribution of non-kin interventions among the three malrilines. The
thickness of the arrows is proponionallo the frequencies of interventions (numbers). Support given to a matriline (direction of arrow) is necessarily against the third matriline. (From Chapais et at., 1991.)
Figure B.l1.2 shows lhe pattern of support between malrilines in the intact group. Members of A and B matrilines consistently supported each other against
C. and rarely supported members of the C matriline against the other matriline. In a series of experiments, a single female from the A or B matrilines was placed with the two other matrilines. [n each case. the higher-ranking matriline supported the lone female against the C matriline rather than vice versa. For
example, B was placed with the A and C matrilines. When B was alone with the C malriline she lost her rank (see p. 260). However. when the A matriline was also present. they supported her against the C matriline so that she maintained her rank above the Cs. When A1 was placed with the Band C matrilines, the B matriline outranked A I. but supported her against the C matriline. rather than supporting the C matriline against her. Thus. she maintained her rank above the C matriline (Chapaisetal.. (991).
Chapais (1992) suggesls that matrilineal hierarchies evolved from an initial state in which females were ranked by age or individual RHP. Females protected their daughters when they were most vulnerable and helped them side-step the lowest rungs 01 an age-graded hierarchy. Because mothers would be expected to support their daughters against more distant relatives. females eventually banded together with their daughters and granddaughters to overpower any high-ranking lemale that attempted to act on her own. A linear hierarchy. however. would be unstable illemales only supported members 01 their own matriline because a large low-ranking lamily could
262
CHAPTER I I
ovenhrowa small high-ranking Iamily. However, upheavals in rank (e.g. Erhardt & Bernstein, 1986; Samuels el al., 1987) are rare. and stability apparently results Irom the paltem 01 non-kin suppon. Consider a descending hierarchy 01 three matrilines: A, B, C. 1I1emaies supponed members 01 any non-matriline, matriline B could be threatened by 'bridging' alliances between A and C. Also, matriline A could be threatened by 'revolutionary alliances' between Band C. However, non-kin generally only aid each other against matrilines that rank lower than either olthem (Le. A and B against C) (Box I).)). By assisting higher-ranking matrilines against lower-ranking lamilies, midranking lemales may prevent the lormation of bridging alliances, and, by supporting the most dominant of lower-ranking opponents, the highest-ranking females create a state 01 dependency that forestalls the lormation 01 revolutionary alliances and maintains the status quo (Chapais, 1992). In other cases, alliances can destabilize hierarchies and prevent them from being linear. Male chimpanzees live in a lission-fusion sodety in which males spend considerable periods apart. Male rank depends heavily on alliances, and il male A relies on the support 01 male C to dominate male B, he may be defeated by B when C is absent (Bygon, 1979). In addition, subordinate males are sometimes lickle in their alliances, with C sometimes supporting A against B and sometimes supporting B against A, thereby indudng a state of dependence in each and gaining greater access to resources (de Waal. 1982; Nishida, 1983). Again, however, there are no satislactory explanations lor why alliances act to stabilize the status quo in some species whereas they are destabilizing in others.
11.2.3 Benefits and costs of dominance Stable dominance relationships may benefit both dominants and subordinates by minimizing the incidence 01 serious fighting. A classic study in chickens showed that hens in groups with stable hierarchies lought less and laid more eggs than those with unstable hierarchies (Guhl el al., 1945). Also, damaging fights in a colony of chimpanzees were five times more Irequent when male ranks were unstable (de Waal. 1982). Nevertheless, dominance relationships are generally expected to result in inequitable access to resources, and thus high-ranking individuals should enjoy greater reproductive success than subordinates. Although there is considerable evidence that dominant individuals do gain greater access to scarce or monopolizable resources, such as mates, lood or sale reluges (reviewed in Huntingford & Turner, 1987; Langen & Rabenold, 1994; Fournier & Festa-Bianchet, 1995), high rank is not always associated with higher reproductive success. This sometimes results from the existence 01 equally successlul alternative strategies, such as satellites or sneaks (Gross, 1996), which opt out 01 the competition altogether, Or because there are also costs associated with maintaining high rank or high RHP. We consider two examples where high rank confers costs as well as benefits.
ECOLOGY OF RELATIONSHIPS
263
Spoiled hyenas live in clans of up to 80 individuals in which all adult females rank above the adult males, and females form a stable matrilineal hierarchy (Frank it al., 1995a). Hyenas feed on medium-sized antelope. and clan members compete intensely for carcasses with the result that high.ranking animals gain greater access to food (Frank, 19861. High-ranking females reach sexual maturity earlier, have slightly shoner interbinh intervals and recruit more offspring of each sex (Frank it al.• 1995a). However. female aggressiveness is associated with a syndrome that includes large body size, clevated levels of androgens and dramatic masculiniz.1tion of the external genitalia. Female masculinization is suggested to have evolved because of the advantages of aggressive competition (Frank. 1986). bUI the syndrome carries considerable costs. Females give birth through a penis. and primiparous mothers suffer a high incidence of birth complications that may reduce their lifetime reproductive success by as much as 25% (Frank it al.• 1995b). Female savanna baboons also form matrilineal dominance hierarchies (Cheney, 1977; Hausfaler il al.. 1982; Johnson, 1987). although a small number of females are able to rise in rank above their matrilines (Samuels it al., 1987; Packer it al., 1995). High-ranking females enjoy higher infant survival, shoner interbirth intervals and younger ages at sexual maturity (Packer il al.• 1995). However. rank is not significantly correlated with lifetime reproductive success in some populations because high-ranking females suffer more miscarriages and occasionally suffer from reduced fertility (Wasser. 1995; Packer it al.. 1995; but, see Altmann it al.. 1995). The source of these costs is unknown. but highranking female baboons show an imbalance in their ratio of oestrogen and progesterone (Wasser. 1995), high-ranking juvenile females have higher levels of androgens (Altmann et al.. 1995) and. in social carnivores. high-ranking animals show elevated levels of stress hormones (Creel it al.. 1996). Packer it al. (1995) suggest that traits conferring high rank in females are subject to strong stabilizing selection because of their potentially negative effects on reproductive physiology. Although individuals that invest too much in aggressive competition suffer these costs, those that opt out of direct competition would suffer reduced feeding success. Thus. dominance behaviour will be expected to persist even if high rank does not confer significantly higher reproductive success.
ll.2.4 Alternatives to dominance; ownership When contestants are evenly matched and the costs of fighting are high compared to the value of the resource. animals may settle contests on the basis of asymmetries that are not corrclated with RHp, including prior ownership of the resource (the bourgeois strategy: Hammerstein, 1981; Maynard Smith, (982), provided that the contest does not approximate a war of attrition (Hammerstein & Parker. 1982). fn social groups. this convention would translate into a state-dependent dominance that lasted only as long as one animal
264
CHAPTER II
retained ownership or the resource, and no persistent dominance relationships would be expected. A[rican lions providean apparent example. Female lions compete with their pridemates for access to meat but respect each others' ownership o[ specilic reeding sites at each carcass (Packer & Pusey, 1985). Their prey have tough hides, and meat can be extracted most readily from certain pans of thc carcass. The first [emale to feed from each site reluses 10 yield to any other female in her pride, lunging and snarling if another ventures 100 close. Females 'respect' cach pridematc's ownership, and roles are revcrsed as active feeders bccome temporarily sated and move away from the kill. Breeding appears to be egalitarian, and dominance hierarchies have never been reported in female lions (Packer el al., 1988) - in marked contrast to every other species of social carnivore (e.g. hyenas: Frank el al., 1995a; mongooses: Creel el al.. J 992; wild dogs: Malcolm & Manen, 1982). Male lions form coalitions that compete intensively against other coalitions for access to female prides, but coalition partners also compete for access to individual oestrous females (Packer & Pusey, 1982). When one male forms a consortship with an oestrous femalc, his coalition partners respect his 'ownership' of the female, avoiding the pair and retreating from the threats of the consorting male. Again, dominance is temporary and state dependent, with an owner on one occasion becoming a rival on another. Individual mating
success is unequal in coalitions where companions vary in age or size but is equivalent in coalitions where companions arc evenly matched. However, paternity tests revealed a considerable skew in reproductive success in larger coalitions (Packer el al.. 1991), even in groups that showed no obvious dominance hierarchy. We do not know whether ownership rules are ignored by those few males who do enjoy high reproductive success, or if reproductive skt·w results [rom some other form of male-male competition or from the preferences of the females. We have argued that lions respect ownership both because contestants are usually evenly matched in age and size and because of the high costs of fighting. Besides having sharp teeth and strong jaws, lions can easily blind an opponent with their claws. However, hyenas are also formidable fighters yet show strict dominance (see Section 11.2.3). An additional cost of fighting to lions is that companions are vital alliance partners [or terrilOrial delence and damaging or killing a companion would be highly detrimental (see Section 11.3.3).
11.2.5 Dominance and models of reproductive skew Models o[ reproductive skew (see Chapter J 0) may help explain the occurrence, strength and intensity of dominance hierarchies. Assuming that the dominant can control the subordinate's reproduction, these models predict that [our parameters will affect the extent to which reproduction will be skewed:
ECOLOGY OF RELATIONSHIPS
265
I Ihe probability a subordinale can breed successfully on ils own elsewhere; 2 the extent to which the subordinate can increase the dominant's produclivily; 3 the genetic relaledness of the pair; 4 their relalive fighling abililies. Under cenain circumstances. dominants are expected to cede reproductive opponunilies 10 subordinales 10 iliduce Ihem to stay (slaying incentives). or to induce them to cooperate wilhoul fighling (peace incentives). In general, Ihese incentives will increase wilh increasing prospects for breeding alone (since leaving becomes more advantageous for the subordinate). increasing produclivilY by the dominant (since the dominant can a[[ord to surrender more to Ihe subordinate). and decreasing kinship (since the subordinale gains fewer inclusive fitness e[[ecls from assisling Ihe dominanl). Peace incemives will increase as fighling ability becomes more similar. 1I0w do Ihese incentives influence dominance behaviour? If dominanlS cede more reproduction to subordinates. do Ihey merely penni! subordinates more acce s to food or males, while continuing to enforce slrid dominance, or do dominance relalionships Ihemselves become less clear-cuI? Keller and Reeve (1994) suggesllhal as the incentives increase, Ihe frequency and intensily of dominance interaClions should decrease, not only because subordinales gain a lower payo[[ from 'testing' Ihe fighling abililY of Ihe dominanl, bUI also because dominants have less to lose from such challenges. Dominance should therefore be mOSI pronounced when skew is high, bUlthis does nOI appear to be Ihe case in several mammalian species. Male lions generally exhibil high skew bUI inconspicous dominance, while female baboons exhibil relalively low skew but pronounced dominance. However. much more research is necessary 10 quantify Ihe severilY of dominance interadions across species and to measure Iheir associated skew.
11.3 Cooperation Although kinship plays an essenlial role in Ihe evolulion of cooperative behaviour (as reviewed in Chaplers 9 and 10), Ihe formalion of cooperalive relalionships belween non-relalives has also generaled considerable lheorelical interest and extensive empirical research. Cooperative behaviour is often
assumed to involve a short-Ierm cosl (if only Ihrough an 'opponunilY COSI' from failing 10 behave selfishly), bUllhe long-Ierm consequences of repeated cooperalion may be fundamental to underslaoding social relationships. In fact. mosltheorelical work explicilly recognizes thaI cooperalion involves repeated interaclions between the same pair of individuals.
11.3.1 Reciprocity The basic paradigm for moSl evolutionary models has been Ihe 'Prisoner's
266
CHAPTER II
dilemma' (Box 11.2) where individuals have only two alternatives: to cooperate or to defect. By definition, mutual cooperation gives a higher payoff than mutual defection. However, in the prisoner's dilemma, a defector gains an even higher payoff when paired with a cooperator. Thus, although everyone would do best if everyone cooperated, a selfish mutant could always invade a cooperative population.
Box 11.2 Reciprocity: the basic problem Considerable theory has been developed to specify the conditions when individuals might form long-term cooperative relationships despite short-term advantages from selfish behaviour. Most models specifically deal with illleractions between two unrelated individuals where each individual only has two options, either to cooperate or to defect.
Against
Payorr to
Cooperate
Defect
Cooperalc Defect
R T
s p
Models of reciprocity are based on a specific set of payoffs called the 'prisoner's dilemma'. If both players cooperate, they both receive the reward (R) from mUlual cooperation, but if both defect they receive the punishmelll (P) from mutual defection. If the opponent cooperates, there is a temptation (T) to cheat which causes the lone cooperator to receive the sucker's payoff (5). Because T> R> P> 5, defection confers the highest short-term benefits. However, repeated interactions theoretically permit the evolution of cooperation through some form of reciprocity. Such strategies cooperate in
the first encounter but only continue to cooperate if their opponent also cooperated and to defect if their opponent defected. The simplest strategy involVing reciprocity is called tit-for-tat (TFT). TFT initially cooperates then behaves according to its opponelll's behaviour during the prior move. In an iterated game between two individuals playing TFT, both contestants start out cooperatively and both colllinue to cooperate in aU subsequent interactions: 1'5 Player Player 2'5
payoff 1, TFT 2, TFT payoff
RRRRRRRRRRRRRR... CCCCCCCCCCCCCC . CCCCCCCCCCCCCC .
RRRRRRRRRRRRRR... C'nti'utd ~n p. 267
ECOLOGY OF RELATIONSHIPS
267
Box 11.2 Con I'd. In an iterated game of N encounters. each individuallherefore receives a
cumlilalive payoff of RN. [[TFf meets a pure defector (ALL-D), TFfsuffers in the initial encounter, bUlthen gains the same payoff as ALL-D in all subsequent illleractions: l's payoff
SPPPPPPPPPPPPP...
Player 1: TFT
CDDDDDDDDDDDDD•••
Player 2: ALL-D 2's payoff
DDDDDDDDDDDDDD .
TPPPPPPPPPPPPP .
An iterated prisoner's dilemma gives a cumulative payoff matrix of:
Against PayoH to
TFT
ALL-D
TFT ALL-D
RN T+ P(N-I)
S+P(N-t) PN
In a population o[ TFf. ALL-D cannol invade i[ RN > T + PIN - t), which is most likely when N is very large. Although TFf can never invade ALL-D. as N ..... -. PN ~ S + PIN - 1), and TFf may persist long enough to gain a foothold in the population, then spread if it can primarily inleract with other TFf Slrategists (Le. if the population is spatially structured).
Although selfish behaviour is expected to dominate whenever individuals meet for only a single enCOUlller. repeated interactions allow the evolution of cooperation through some form of reciprocity (Trivers, 1971). By basing its own behaviour on the prior behaviour of its opponent, an individual can benefit from mutual cooperation while preventing cheats from prospering. Following Axelrod and Hamilton (1981). numerous authors have investigated the conditions where reciprocity can evolve. Their original analysis suggested that animals might follow a simple slrategy called 'til-for-tat' (TFf) which shows an initial bias towards cooperation then copies each of its opponent's moves. Box 11.2 presents a simplified version of the iterated prisoner's dilemma and outlines the major features of the game. We seek a strategy that is not only able to withstand invasion by a selfish strategy ('all defect' or ALL-D) but that can also invade a purely selfish population. This latter point is importalll since selfishness is generally assumed to be the ancestral condition.
268
CHAPTER I I
Although TFT has often been considered an evolutionarily stable strategy (ESS: sensu Maynard Smith, 1982), there is no ESS in the iterated prisoner's dilemma. This partly results from the mathematical pathologies of the repeated game: in a population of TFT, an unconditionally cooperative strategist (ALLC) would gain lhe same overall payoff and thus be able to persist by drifl. Once enough ALL-C have appeared, the population could be invaded by ALLD. In addition, certain slrategy combinations can replace TFT. A population playing TFT can be invaded by a pair of mutant strategies called lil-for-twotats (TF2T), which waits for its opponent to defect twice in a row before retaliating, and suspicious TFT (STFT), which has an initial bias to defeci before playing TFT (Boyd & Lorberbaum, 1987). The most important weakness of TFT, however, is that any mistakes can lead to a permanent breakdown in cooperation (Box 11.3). Thus, if both
Box 11.3 The problem of mistakes
As originally fonnulated, the iterated prisoner's dilemma a sumed that animals never made any mistakes in judging their opponent's behaviour, bUI TFT is so responsive that a single mistake leads to a long series of mutual retaliation: Mistake
J, 1'5 payoff
RRRRRRRRTSTSTS .
Player 1: TFT Player 2: TFT
CCCCCCCCCDCDCD .
2'5 payoff
RRRRRRRRSTSTST .
CCCCCCCCDCDCDC .
By the convention of these models (T+ 5)/2 < R, so these over-reactive responses confer lower fitness than a strategy that can reCOver from such mistakes and resume mutual cooperation. The simplest of these is TF2T which only stops cooperaling after two successive defections by its opponent: l' 5 payoff
SSPPPPPPPPPPPP .
Player 1: TF2T
CCDDDDDDDDDDDD .
player 2: ALL-D
DDDDDDDDDDDDDD .
2'5 payoff
TTPPPPPPPPPPPP .
TF2T is not provoked into defecting by a single mistake: Mistake
J, l' 5 payoff Player 1: TF2T
CCCCCCCCDCCCCC .
Player 2: TF2T 2'5 payoff
RRRRRRRRSRRRRR .
RRRRRRRRTRRRRR...
CCCCCCCCCCCCCC . Conli'lII~d on
p. 169
E COL 0 G Y 0 F R E L ATI 0 N S HIP S
269
Box 11.3 Conl'd. The most robust of these mistake-correcting strategies is called 'Pavlov' be-
cause it changes its behaviour alter receiving a poor payoff (P or Sj, but repeats its behaviour after receiving Tor R. Pavlov also has an initial bias to cooperate: Mistake
! l's payoff
RRRRRRRRTPRRRR...
Player 1: pavlov
CCCCCCCCDDCCCC ..
Player 2: Pavlov 2's payoff
CCCCCCCCCDCCCC . RRRRRRRRSPRRRR...
NOle that the individual receiving T defected again in the following mOve
whereas Ihe individual receiving S changed its behaviour to delect. But, once both received p. both changed their behaviour 10 cooperatc, and mutual
cooperation was thereby restored. Pavlov can also exploit an unconditional cooperator (ALL·C). Whereas
TFr orTF2T would quickly return to mulual cooperalion after a single round 01 exploiting ALL-C, Pavlov continues 10 defect and therefore 'never gives a sucker an even break': Mistake
! l' 5 payoff
RRRRRRRRTTTTTT .
Player 1: Pavlov
CCCCCCCCDDDDDD .
Player 2: ALL-C 2' s payoff
CCCCCCCCCCCCCc... RRRRRRRRSSSSSS ...
Although Pavlov performs beller than TFf in an impedect world. it fares
poorly in a population of ALL-O (only receiving SPSPSPSP... ). Against Pavlov
ALL-D
Pavlov
RN
N(S
ALL-D
N(T+ P)/2
PN
PayoH to
+ P)12
Thus. Nowak and Sigmund's (1993) analysis suggests that Pavlov is most likely 10 appear in a populalion aller TFr has replaced ALL-O and can only persis I if R> (T + Pj/2. opponents play TFT, a single mistake sets up an alternating patlern of retilliation wilh each animal changing ilS behaviour in response to its partner until a second mistake either restores mutual cooperation or leads to mutual
270
CHAPTER II
defection. Box 11.3 outlines the various strategies that might permit reciprocity in the face of mistakes. These include TF2Twhich 'forgives' a single mistake and 'Pavlov' which repeats its prior move after receiving a good payoff (R or 1), but changes its behaviour after a poor payoff (5 or Pl. If cooperation breaks down, Pavlov results in mutual defection for just one move before cooperation is restored. Nowak and Sigmund's (1993) analysis suggests that while TFT may often be the first form of cooperation to appear in a selfish population, it will generally be replaced by Pavlov. It is important to note, however, that all these analyses are based on quite restrictive assumptions. Pairs of animals are assumed to make simultaneous decisions without any information about the opponent's current move (for a rigorous examination of alternating decisions see Nowak & Sigmund, 1994). Contests are always between pairs of individuals; reciprocity based on communal resources is unlikely to evolve in groups larger than two (Boyd & Richerson, 1988). Payoffs are assumed to be perfectly symmetrical and to remain unchanged through evolutionary time. Asymmetrical payoffs may mean that one partner faces a prisoner's dilemma while the other does not (Packer & RUllan, 1988), or that a dominant individual might have the option of manipulating its subordinate partner (Clullon-Brock & Parker, 1995; see also Section 11.3.2). Even in games that allow mistakes (e.g. Nowak & Sigmund, 1993), the mistake rate is low, the game is assumed to be infinitely iterated and players ean only remember the opponent's behaviour in the prior move (thus excluding tolerant strategies such as TF2T). The assumption of an infinite iteration allows the outcome of the initial encounter to be ignored. Thus, TFT behaves the same as ALL-D in a population of ALL-D and so can persist until drift enables a critical mass of TFT cooperators to interact and reap the rewards of mutual cooperation; TFT then permits the coexistence of ALL-C or Pavlov, but ALL-C is prone to invasion by ALL-D, whereas Pavlov is less so. Infinite iteration also assumes that animals will not 'discount' future payoffs, and thus that they will forego immediate payoffs for higher future rewards, whereas empirical studies show thai animals have strong preferences for immediate gain (Stephens el al., 1995). The theoretical literature on the iterated prisoner's dilemma may therefore be summarized as supporting a general tendency IOward some form of cooperation in very long-term relationships. The conditions where cooperation is actually expected, however, are extremely limited - provided that animals are only allowed two options in each interaction: either to cooperate or to behave selfishly.
11.3.2 Other routes to cooperation Given the substantial difficulties facing the evolution of reciprocity, it is important to consider alternative explanations for cooperalion, especially when evaluating empirical data. We outline three of these in Box 11.4.
ECOLOGY OF RELATIONSHIPS
271
80x 11.4 Other routes to cooperation A: Short-term mutualism CooperaLive interactions do not necessarily involve a prisoner's dilemma. Two heads may be beuer than one (with both individuals receiving the largest
award from mutual cooperation). and there may be no advantage [rom exploiting a companion's behaviour. Against
Payoff 10
Cooperate
Defect
Cooperate
R=6
5=3
Defect
T=2
p
In this case. cooperation is expected even when opponents meet for only a
single encounter. By definition. R> P. but if P> S. defection is also an ESS. and cooperation is most likely to evolve in a structured population (where
families of cooperators typically receive R, while families of defectors mostly receive Pl.
8: Long-term models of mutualism Payoffs may conform to the prisoner's dilemma in the short term but become rnutualistic in the long term. This transition. however, does not result from
following complex behavioural strategies (as is the case for reciprocity). but from the rebounding consequences of selfishness. Consider a daily game where the payoffs depend not only on the behaviour of each contestant. but on the number of animals playing the game each day. and each animal receives a higher payoff as a member of a pair than as a solitary. Let the pairwise payoffs follow the prisoner's dilemma, but assume that defection increases the partner's risk of mortality. The survivor
subsequently becomes a solitary who receives a daily payoff of X, where X < S. In the extreme case, a single defection kills the panner, and the panner cannot be replaced. Thus, the payoffs are accumulated as follows: l's payoff
s
Player 1: ALL-C
C
Player 2: ALL-D 2's payoff
TXXXXXXXXXXXlOC
DDDDDDDDDDDDDD..•
When both defect, one or both is killed. Payoffs are either: l's payoff
P
player 1: ALL-D
D
Player 2: ALL-D
DDDDDDDDDDDDDD .
2's payoff
pxxxxxxxxxxxxx .
272
CHAPTER 11
Box 11.4 Cont'd.
Or:
1'5 Player Player 2'5
payoff 1: ALL-D 2: ALL-D payoff
P D
D P
But when both cooperate:
1'5 Player Player 2' 5
payoff 1: ALL-C 2: ALL-C payoff
RRRRRRRRRRRRRR...
CCCCCCCCCCCCCC . CCCCCCCCCCCCCC . RRRRRRRRRRRRRR. ..
Thus, the cumulative payoffs are: Against Payoff to
ALL-C
ALL-D
ALL-C ALL-D
RN
S
T+X(N-l)
PorP+X(N-l)/2
If N > 1 and R > (T + X)/2, the most likely outcome would be unconditional cooperation. However, the population would have to be structured for cooperation to invade a population of pure defectors. Note that in this case, pure cooperation dearly fares better than any conditional trategy (such as TFT) if animals risk mistaking their companion' behaviour. Group-living often confers advantages strong enough to counter the shortterm advantages of selfish behaviour. A good example is Lima's (1989) model of cooperative vigilance. Each individual must trade-off Vigilance for foraging. A more Vigilant individual protects itseJ( and its companion from predation, but spends less time foraging. The short-term seJ(ish optimum should therefore involve relatively little Vigilance. However, because the predator is assumed to capture only one prey at a time. an animal's survival greatly reduces its companion's risk of predation (due to the dilution effect). Thus, in a repeatedencounter game, animals will have to improve their companion's chances of survival so as to minimize their own long-term risk of predation.
c: Producers versus scroungers Obtaining resources an often entail an inherent cost, but no resources will be
ECOLOGY OF RELATIONSHIPS
273
-~
Box 11.4 Cont'd.
available unlt:ss someom: works to extract them. In this case, there may he a lemptation to defect (T> R), but cooperators are al an advantage when lhey are rare.
Against Payoff to
Cooperall'
Defect
Coopt'rate
VI2 - C/2 VI2
o
Defect
VIZ - C
1vo individuals compete for a resource of value. V; extracting thl' resnuro: incurs some COSI, C. As long as someone works to exlract the resource. it is divided between bOlh contestants. If bOlh work together. each pays 50% of the costs. As long as VJ2 - C> 0 cooperators (produCl'rs) can invade ddcclors (scroungers). but if defectors gain equal access 10 the resource withollt paying any costs. scroungers can always invade producers. In this caSe cooperation will be frequency dependent. and in large gruups subsets of cooperators Illay coexist wilh a proportion of defectors.
D, Coerced cooperation Most models of cooperation assume that animals have only two options, eilher to cooperate or to defecl. Ln addilion, these models assume that both players possess exactly the same skills and RHP. However, most animals live in groups with definite dominance hierarchies and their relationships will generally be asymmetrical (see Section 11.2). Thus. a difference in RHP or dominance rank may permit one animal to coerce its companion into behaving coopt'ralivdy (elutton-Brock & Parker. 1995). Note that the following analysis now describes phenotype-limited stralegies. Consider a two-person game with the same basic structure as the prisoner's dilemma. However, now allow a dominant animal to punish any subordinate: that fails to behave: cooperatively. Le:t d be the costs to lhe dominant of punishing lhe subordinate, and i the costs to lhe subordinate of being punished. If S > P - i. then a defecting subordinale will do better to change its behaviour 10 cooperate than 10 risk further punishment. If the subordinale subsequently stans 10 cooperate and T-d> P, the dominant benefits by punishing Ihe subordinate every time it defecls:
274
CHAPTER II
Box 11.4 Con I'd.
Punishment
1'8 payoff
p-
i SSSSSSSSSSSSS .
Player 1: subordinate
D
CCCCCCCCCCCCC ..
Player 2: dominant
D
DDDDDDDDDDDDD .
2' 8 payoff
P- d TTTTTTTTTTTTT .
However, if S < P-i, the subordinate would be better off incurring repealed punishment rather lhan submitting to the demands of a tyrannical despol. [n which case, the dominant might be expected to behave as a benevolent despot, habitually cooperating and forcing lhe subordinate to cooperate as well: Punishment
l' 8
payoff
T-
i RRRRRRRRRRRRR .
Player 1: subordinate
D
CCCCCCCCCCCCC ..
Player 2: dominant
C
CCCCCCCCCCCCC .
2'8
payoff
S - d RRRRRRRRRRRRR .
Here the subordinate will cooperate if R> T -i, and the dominant will be expected lO punish each defection if R-d > S.
The most obvious explanation for cooperation is that when an individual meets a cooperative partner, it gains a higher payoff from cooperating than from defecting. In the absence of any temptation to cheat. cooperation is dearly an ESS even in a single-encounter game and no elaborate safeguards against cheating will be expected. The set of payoffs described in Box Il.4A are ohen described as 'byproduct' mutualism (West-Eberhard, 1989) because cooperation is the best strategy regardless of the opponent's behaviour, and thus cooperation is merely a byproduct of following the optimal strategy. Mutualistic advantages can also arise in long-term interactions, even if the payoffs from any single encounter are consistent with the prisoner's dilemma. In Box I I.4B, we present a simplified example in which individuals gain a short-term advantage from cheating, but this selfish behaviour results in the loss of their companion with the consequence that they must subsequently live alone. With strong enough advantages of grouping, unconditional cooperation will be favoured in an iterated game (Lima, 1989). Again, in such situations, no elaborate cooperative strategies are necessary: the well-being of one's companions has an important effect on personal fitness. However, the outcome of mare realistic games will depend on the ease with which partners can be replaced.
ECOLOGY OF RELATIONSHIPS
275
Cooperation may not always be two-sided or ubiquitous. 'Producers' may extract more resources per capila than 'scroungers', yet a population of producers will often be prone to invasion by scrounging and vice versa (see Barnard, 1984). In pairwise interactions, this may result in one individual cooperating while the other defecls (Box IIAC), bUI in larger groups a subset of individuals would be expeded to cooperate while the remainder defect (see Sed ion 11.3.3). Thus, a majority of individuals may cooperate unconditionally. Finally. several recent models have examined the possibility that cooperalion may be imposed by coerdon (e.g. Boyd & Richerson. 1992; Clutton-Brock & Parker, 1995). This approach is especially relevant here, since dominance relationships are explicitly incorporated into the structure of the payoffsand dominance relationships are nearly universal in animal societies. The important addition of these models, however, is that individuals can physically punish defectors whereas earlier models only permit TFT or Pavlov to withhold further cooperation. As outlined in Box 11.40, dominant animals may often be expected to coerce subordinates to behave cooperalively. For simplidty, we have assumed that a single punishment induces the subordinate into cooperating in all subsequent encounters. In reality, of course, the evolution of 'enforced cooperation' will depend on the speed with which the subordinate 'learns' to cooperate in response to the dominant's behaviour (Clutton-Brock & I'arker, 1995) as well as on the rate of recidivism. If dominanls can impose sufficiently heavy costs they may be able to exploit the subordinate and maintain a highly one-sided relationship. However, in other circumstances lhe dominant would be forced to settle for a mutually cooperative relationship (also see Section 11.2.5 and Chapter 10). In summary, none of these routes to cooperation requires strategies as elaborate as TFT or Pavlov, and in many cases such 'reactive' strategies would be highly disadvantageous compared to a simple strategy of 'all cooperate'. Coercion may be reactive (i.e. the subordinale is punished if it does not cooperate), but only leads to mutual cooperation when the dominant cannot maintain a purely exploitative relationship with the subordinate.
11.3.3 Empirical studies Several recent reviews (e.g. Packer & Ruttan, 1988; Emlen, 1991; Clements & Stephens. 1995) have concluded lhat reciprocity is rare to non-existent in nature, while others suggest that reciprocity is widespread (e.g. Dugatkin et al.. 1992). Reciprodty remains one of the most beguiling concepts in behavioural ecology, but the most formidable obstacle to proving its existence stems from the difficulty of measuring short-term payoffs. Does a particular example of cooperation really follow a prisoner's dilemma or is it mutualistic?
276
CHAPTER II
Here we outline three different studies of cooperative behaviour that highlight the difficulties in disentangling the alternatives.
Cooperation in African lions Field sludies conlirm that African lions are highly cooperative. Lions hunt in groups (Scheel & Packer, 1991; Stander, 1992), females raise lheir young in a communal creche (Pusey & Packer, 1994) and both sexes defend joint territories against like-sexed individuals (McComb et al., 1994; Heinsohn & Packer, 1995; Grinnell et al., 1995). Lions of each sex gain higher reproductive success by living in groups (Packer et al., 1988): males in larger coalilions gain higher per capita reproductive success lhan those in smaller coalitions, and females in moderate-sized prides have higher fitness Ihan those in very small prides. These advanlages, however, are the summed outcome of a variety of behaviours, and we discuss the evolutionary basis of each in turn. Group hunting. By hunting together, individuals can increase their joint hunting success so that R> P. However, an individual joining a hunt is likely to incur
costs, whether from the energetic effort required to capture the prey or from the risk of injury while subduing the prey. When the prey is large enough to feed an entire group, the decision to join a group hunt may follow a prisoner's dilemma (Packer & Ruttan, 1988). If the success rate of a lone hunter is already high, a second hunter may not be able to improve the group's success rate sufficiently to overcome its own costs of huming, in which case T> R. However, if individual success rate is very low, two heads may be so much better than one that the improved chances of prey capture can overcome the personal costs of joining the hunt. in which case R> TLions, like most other group hunters (Packer & Ruttan, 1988), show the
clearest evidence of cooperation when R is likely to be greater than T, and thus when cooperation is mutualistic. Scheel and Packer (1991) found that individual behaviour during group hunts could be classified into several discrete strategies, one of which, 'refraining', involves almost no participation in the hunt.
'Refrainers' remain stationary or move only a few paces towards a distant prey animal and do not join in the final charge. Other lions clearly cooperate. actively stalking the prey and participating in the linal charge. Consistent with predictions of mutualism, the probability thal Serengeti lions cooperated during a group hunt depended on two factors relating 10 hunting success. I Lions were most likely to cooperate when the prey was difficult to capture, and showed higher levels of 'refraining' when their companions pursued prey that was more easily captured. 2 Since male lions are less skillful hunters than females, they are less able to improve the females' success rate, and they 'refrained' more than females. Perhaps the most impressive data on cooperalive hunting in lions comes from Stander's (1992) studies in Namibia. These animals demonstrated a clear
ECOLOGY OF RELATIONSHIPS
277
division of labour, with certain individuals habitually stalking to the left, others habitually stalking to the right and the remainder moving directly towards the prey. However, they hunted fleet-footed prey in open habitat, and individual hunting success was close to zero. Thus, their cooperative hunting was highly likely to have been mutualistic. Across a variety of species, cooperative hunting appears to be restriaed to circumstances of simple mutualism (see Caro, 1994; Creel & Creel. 1995), even though cooperation might theoretically evolve by reciprocity. However, it might often be impossible to detect whether a hunt has failed due to the elusiveness of the prey or to the defection of another hunter (Packer & Ruttan, 1988). Strategies such as TFT and Pavlov are sensitive to the mistake rate (see Box 11.3), and thus reciprocity may not evolve in situations where failure is the most likely outcome of a cooperative interaction.
Although cooperative hunting appears to be largely mutualistic, this con· clusion must be considered provisional until experimental tests of reciprocity can be deVised. All published studies are strictly correlative; no field study has ever measured T, R, Sand P.
Communal cub·rearing. Female lions pool their cubs in a 'creche' and raise tbem communally, even nursing each other's cubs. In the short term, communal nursing may involve a prisoner's dilemma (see Caraco & Brown, 1986). A defector directs all her investment to her own cubs while a cooperator allows all cubs 10 nurse equally. Thus, the defeaor pays some small costs of vigilance to ensure that only her cubs nurse from her, but her cubs gain exclusive access to her milk. A pair of cooperative females avoid the costs of excluding other cubs (R> PI, but when a defector is paired with a cooperator, the defeaor's cubs gain all their mother's milk plus a portion of the cooperator's (T), while the cooperator's cubs only gain a portion of their mother's milk (S). With trivial costs of enforcement, T> R> P> S. However, females do not form creches in order to engage in communal nursing. Creches are proteaive coalitions against infanticidal males, and groups of mothers are more e[[ective than solitaries (Packer et al., 1990). Once together, though, the mothers have to balance a number of diHerenl activities, and cubs often try to nurse when their mothers are resting or sleeping. Thus, mothers are constrained to be gregarious and communalnllTsing is largely a consequence of their cubs' parasitic behaviour (Pusey & Packer, 1994). It is therefore misleading to view communal nursing in isolation. Instead, losing milk to nonoffspring appears to be a cost of forming the creche, with individual females adjusting their milk distribution according to their personal costs. Field observations revealed that female lions nur ed indiscriminately only when paired with their closest female kin. When their companions were distant relatives, mothers directed most milk to their own oHspring. Lion Iiller size varies from one to four, and milk production appears to be independelll of liuer size. Females with small litters were more generous than mothers
278
CHAPTER II
of larger litter.;, presumably because of the lower costs of milk loss (Pusey & Packer, 1994). Although the precise payoffs are impossible to measure, there is no evidence that communal nursing in lions involves any form of reciprocity; after kinship and litter size were controlled for, females did not nurse their companions' cubs according to the extent to which their own cubs were nursed by their companions. Group territoriality. Lions maintain joint territories that persist over generations, and successful reproduction requires a high-quality territory. The lion's roar is a territorial display, and both sexes respond cooperatively to the pre-recorded roar of a like-sexed stranger, approaching the speaker and even auacking a SlUffed lion hidden nearby. Playback experiments reveal that lions are most likely to respond when they outnumber their opponents and that they often monitor each other's behaviour while approaching the speaker (McComb el al.. t 994; Grinnell etal.. 1995). Territorial delence is likely to involve a shortterm prisoner's dilemma, with joint defence being more effective than non· defence (R > P) and a defector suffering fewer risks of injury than a cooperator during an intergroup encounter (T> S). A single delender can repel a lone intruder, so T> R. However, several studies suggest that lions cooperate unconditionally, rather than basing their responses on their companions' behaviour. Grinnell et al. r1995) found that males would approach the peakef when their companions were abseot, thus cooperating even when their companions' response could not be monitored. Even more strikingly, Heinsohn and Packer (1995) found that certdin females habitually lagged behind their companions during the approach to the speaker. 'Leaders' could recognize whether their companions were also leaders or if the' were 'laggards'. However, when paired with a laggard, a leader would continue toward the speaker, arriVing considerably earlier than the laggard. Thus, the leaders cooperated rather than 'defected' in response to their partner's defection, nor did they physically punish a laggard for failing to cooperate. As in communal nursing, a focus on shon-term payoffs is probably inadequate: male lions enjoy only a brief tenure within a pride and they need their companions for future interactions (Grinnell et al.. 1995). Males must therefore behave cooperatively at every opponunity to enjoy long-term mutualistic advantages (see Box 11048). For females, the territory is a longterm resource that must be defended habitually and any failure to defend the territory today will result in fewer resources tomorrow (Heinsohn & Packer, 1995). However, as long as enough females maintain effective defence of the territory, a proportion of laggards can take advantage of their companions' behaviour, resulting in a mixture of producers and scroungers (see Box lIAC). In summary, the precise evolutionary basis of the lions' cooperative behaviour is difficult to determine because of their complex social system and the long-term consequences of their short-term decisions. Furthermore,
ECOLOGY OF RELATIONSHIPS
279
behaviour in one context may well rebound on another, with foraging success influencing territorial defence, and so on. Nevenhele s, it is striking that one of the most cooperative of all mammalian species shows no obvious adherence to the rules of reciprocity.
Predator approach behaviour Many species of [ish approach and orientate towards a novel predator, and this behaviour has been dubbed 'predator inspection' whereby individuals obtain information about the predator's hunger and aggressiveness (Pitcher el 01., 1986; Milinski. 1987). Repeated tests have shown that experimental groupings of sticklebacks, guppies and mosquito fish will move into close proximity of the predator, and they often do so in pairs (Milinski, 1992; Turner & Robinson, 1992). Milinski (1987) was the first to suggest that predator inspection involved a prisoner's dilemma, defining any move towards the predator as 'cooperation' and any movement away from the predator as 'defection'. Cooperalive pairs gain R because of the advantages of inspecting the predator from the comparative safety of a group. Defecting pairs only gain P because they learn nothing about the predator's state. Single defectors gain T by remaining at a safe distance and watching th' predator's response to the cooperative companion. Finally, the lone cooperator gain S from incurring the risk of being selected by the predator. Thus, T> R> P> S. Milinski (1987, 1990) and others (Dugatkin, 1988, 1991; Masters & Waite, 1990; Huntingford el 01.. 1994) have shown that sticklebacks typically copy each other's behaviour while approaching a predator, and most have interpreted this paHern as evidence of TFT. However. several other authors have viewed the same behaviour quite differently, interpreting it inslead as 'pursuit deterrence' (e.g. Lazarus & Metcalfe, 1990), and one study has shown thaI non-approaching fish suffer higher risk of predation than the approaching partner (Godin & Davis, 1995a,b). This result is consistent with a broad literature suggesting that cryptic predators are less likely to attack once they have been detected (e.g. FitzGibbon & Fanshawe, 1988). If this is also generally !rue in predatory fish, the short-term payoffs from approaching the predator would be mutualistic, with R> S> P> T. Copying the partner's behaviour in this case can simply be explained by shoaling: individuals benefit from the dilution effect by staying as close together as possible, whether moving toward or away from the predator. Despite nearly 10-years research in a highly tractable system, where s!raightforward manipulations can be performed with mirrors, models and trained fish. it is still not clear which of these alternative viewpoints is more n'arly correct. While it seems intuitive that R> P (the fish do typically approach the predator in pairs, so the behaviour must confer some benefit), it has proven remarkably difficult to measure Sand Tlo everyone's satisfaction.
280
CHAPTER 11
Godin and Davis's (1995a,b) data suggest that S> T, but their methods have been criticized by Milinski and Boltshauser (1995), and data collected by Dugatkin (1992) suggest that. in mixed groups, individuals that had previously been classified as inspectors suffered greater predation risk than non-inspectors. In the absence of any consensus concerning the payorrs, it is still possible to test hypotheses that distinguish between TFT and shoaling. Assuming that the fish are in a prisoner's dilemma and that they are playing some form of reciprocity, they should copy each other's behaviour according to certain rules (Lazarus & Metcalfe, 1990; Reboreda & Kacelnik, 1990; Huntingford et aI., 1994; Stephens et al., 1996). If both animals cooperate on one move they should continue to cooperate on the following move; if both defect, both should either defect again (if playing TFT) or cooperate (if playing Pavlov). If one cooperates and the other defects, then the behaviour of one or both will be expected to change depending on whether the animals are playing TFT or Pavlov. The predicted transition matrices for these two strategies are given in Table II. t, as well as empirical data collected by Stephens et al. (1996) who took as separate 'moves' the animals' net movement every 1.09 s. These data show
that while mutual cooperation is followed by further cooperation 57% of the time, and that mutual defection is typically followed by further defection (consistent with TFT), the most important predictions of TFT and Pavlov are not met. Whenever a non-cooperaror is paired with a cooperator, the fish were not reactive but instead tended to repeat their prior move (the most common response to CD was CO, etc.). Consistent with shoaling, however, fish were most likely to copy each mher's movements when they were In close proximity and to move towards each other when farther apart (even when this meant moving in opposite directions with respectlO the predator). Thus. a major goal of their movements seems to be to maintain close proximity with their companion, rather than to maintain
a close check on each other's cooperative tendencies. Critics of the shoaling hypothesis counter that if there was an inherent. mutualistic advantage of grouping, then every fish in the tank should approach the predator together whereas they typically approach in pairs (Milinski er al., 1990; Dugatkin, 1996). Pairwise approaches are also considered evidence of reciprocity since TFT is most likely to evolve in small groups (Boyd & Richerson, 1988). Proponents of shoaling counter that the advantages of predator deterrence are likely to be frequency dependent, so that the optimal group size for approaching the predator may only be one or two (Godin & Davis, 1995b; Stephens et al., 1996). The chief attraction of this system compared to most field studies has been that shon-term payoffs are likely to be the sole determinant of the fishes' behaviour. However, it is disturbing to realize how inconclusive the predator
inspection/deterrence controversy has been. Perhaps it is futile to study the
ECOLOGY OF RELATIONSHIPS
281
Table Il.l Transition matrices for predalor·approach behaviour predicted by TFf. Pavlov and empirical data collected by Stephens I!t al. (1996). The unique predictions of TFf an.'
printed in bold. and of Pavlov in ;/(J/irs. By TFT. if both defea (DO) in the first move then bOlh should continue 10 defecl (DO) in the next move; if one deft-ds when its partner cuoperates in the first move (DC), Ihe two sl10lild reverse their behaviour (CD) in lhe
second. and vice versa (CO
--i'
DC). while a cooperative Ilair (CC) should continue to
couperate (CC). With Pavlov. mutual defection (DO) should provoke mutual cooperation (DO -t ee). as should mutual coopcratiqn. whereas defection hya single partner should lead 10 mutual ddeaio!l (CD ---+ DO and bc -+ DO).
To From
DD
DC
CD
CC
1 0 0 0
0 0
0
J
0
0 0
0 0 0
0
0 0 0 0
0 0 0 0
0.156 0.}62 0.167 0.135
0.147 0.117 0.359 0.204
Tit-[or.tar
DD DC CD CC
J
I
Palliov
DD DC CD CC
I I
0
J
0 0 I
Observed
DD DC CD CC
0.609 0.266 0.240
0.090
0088
0.256 0.234 0.571
evolutionary mechanism of cooperation when short-term payoffs cannot be experimentally controlled and manipulated. Cooperative key-pecking in blue jays
In order 10 clarify Ihe role of short-term payoffs in eliciling cooperative behaviour, Clements and Stephens (1995) conducted a series of operant feeding experiments where pairs of blue jays were each allowed to press one of two coloured keys. One key had the effect of being cooperative because if both cooperated, both received R, whereas if both pressed the opposile key, both received P, with R > P. The payoffs for these experiments (measured in numbers of food pellets) either conformed to a prisoner's dilemma or to simple mutualism (Fig. 11.2) and were held constant until the birds showed the same re ponse more than 90% of the time for 4 consecutive days. The birds Iypically completed 200 trials per day, and all three pairs showed Lhe same result (Fig. 11.2): afIer a brief period of random pecking, they reacted to the prisoner's dilemma by settling into a pattern of mutual defection. Once Lhe payoffs were changed to mutualism, the birds showed mutual cooperalion.
(b) Clear partition
(a) Opaque partition
':1(' r~~~ ;C~'I l' [ 'i I, . 1:[:[ ( X l~ 1: if K: ~,J_ ':[[ b'i l~ FIrst prisoner's dilemma
Second prisoner's dilemma
Mutualism
First prisoner's dilemma
,i
iii
'ii, i,
,.
Second prisoner's dilemma
Mutualism
t,·
i Ii Ii
Ii
Ii,!'
~
0-
o
'S
!0. E
o
u
'0 ~
0>
S
~
100
50
,
1
[
o ~~
)[ ; /
V
J
j-
;
0

.
~.
C
~
of
o
.L ,
I'
I
,
1
'
II
14
21
1
14
Day in condition
21
~~·_'.':''~':L'':.'
•
1
'4
21
'''II'
1
11
'II
22
I,
Ii
i
1
'i'!t'
'1
i
22
'itt
,
11
iii'
r
22
Day in condition
Fig. 11.2 Cooperative behaviour in thr~c pairs of blue jays during operant feeding tests (Clements & Stephens, 1995). Black circles plot the percentage of trials each day In which the birds showed mUlual cooperation (Cel. open circles show mutual defection (DO), dotted lines show CD plus DC. When the payoffs followed a prisoner's dilemma. the number of food pellets received by each bird was T= S. R:= 3. p= 1. S = 0; when mUluallstic, R = 4. T= S = I, p= O. The birds were separated by a glass partilion In one treatment and by an opaque partition in the other. but the nature of the partition had no effect on the results.
' '
00
n
:I:
>
...,
.'
m
::c
ECOLOGY OF RELATIONSHIPS
283
Then. most intriguingly. when once again faced with the prisoner's dilemma, the birds quickly switched back to mutual defeclion, even though they entered the final phase playing mlllual cooperation. Although these results derive from a highly artificial experimental design, the stlldy clearly shows that animals may not be as predisposed toward reciprocity as has often been claimed. Because the birds were separated by a partition, there was no scope for punishment (see Box I lAC), and it would be interesting to test whether birds might enforce cooperation if they could physically punish non-cooperation by their partner. Future studies may seek to find a more naturalistic system bm the first concern should be to have
absolute control over the four possible payoffs from a pairwise game. In summary. these studies emphasize the importance of carefully controlled experiments in determining whether cooperation results [ronl mutualism or reciprocity. In practice, it may be impossible to provide definitive tests in
natllralistic settings: animals do not behave in a vacuum and behaviour often has long-term as well as short-term consequences. There are many putative examples of reciprocity, but almost all suffer lrom the same general problem: does the behavioural exchange truly involve a prisoner's dilemma? For example, because of their greater cognitive abilities, non-human primates are widely believed to exchange altruistic acts (de Waal & Luttrell, 1988) and yet claims of reciprocity in these species have also been contested (see Bercovitch. 1988; Noe, 1990; Hemelrijk, 1996). One commonly cited context has been the exchange of altruistic behaviour for social grooming, yet there is no convincing evidence that grooming is itself costly (Dunbar, 1988; Hemelrijk, 1994). Witham a clearly designed test where R, S, T and P are all known with certainty, any claims of reciprocity SeCI11 increasingly tenuous.
Chapter 12 The Social Gene David Haig
12.1 Introduction The complex behaviours and structures that have evolved by the process of natural selection can be viewed as adaptations for the good of the relevant genes ('replicators') rather than for the good of individual organisms ('vehicles'), A common criticism of this view is that organisms are integrated wholes, in which no gene can replicate without the assistance of many others. The implicit metaphor is of the organism as a machine. with the genes as instructions ror
the assembly of its component parts. But, an alternative metaphor is possible: genes as members of social groups. Societies, like machines, can display intricate mutual dependence and elaborate divisions of labour; but. onlike machines, societies are not designed. Cooperation and coordination cannot be assumed; when present, they must be explained. Social theories vary in the causal relationships they posit between individual and society: some emphasize the power of individual actions to shape society whereas others emphasize the social constraints on individual freedom. This chapter views properties of organisms as social phenomena that arise from the actions of individual genes, and explores the internal conflicts that can disrupt genetic societies and the social contracts that have evolved to mitigate these conflicts. lWo disclaimers are necessary, First, the chapter emphasizes the current state of genetical systems, and asks why conserved features of organisms are evolutionarily stable relative to conceivable alternatives. Phylogenetic questions, although important, are not my principal concern. Equilibria can often be reached by multiple paths and, in this sense, are independent of history. Thus, too narrow a phylogenetic approach runs the risk of becoming Whig history, telling the story of the winners, and not the innumerable losers, those ephemeral less [unctional genes that strut and fret their hour upon the stage and then are heard no more, Second, gene-centred theories are often reviled because of their perceived implications for human societies. But, even though genes may cajole, deceive, cheat. swindle or steal. all in pursuit of their own replication, this does not mean that people must be similarly selfinterested. Organisms are collective entities (like firms, communes. unions.
charities, teams) and the behaviours and decisions of collective bodies need nOt mirror those of their individual members. As I write this paragraph, my 284
SOCIAL GENE
285
replicators - my genes and my memes - are in constant debate, even dissension, yet somehow I muddle through. I am glad I am not a unit of selection.
12.2 Genes as strategists Genes are catalysts. They facilitate chemical reactions but are not themselves consumed. A gene influences its own probability of replication by the reactions it catalyses, usually indirectly via transcripts and translated products. These effects can be likened to the gene's strategy in an evolutionary game. When, where and in what quantity the gene is expressed is part of that gene's strategy to the extent that changes in the gene's sequence (mutations) could produce a different pattern of expression. The evolutionary theory of games has illuminated many issues in behavioural ecology (Maynard Smith, 1982), but has usually been phrased in terms of payoffs to individuals rather than their genes. This sleight of hand is possible because outcomes that enhance an individual's reproductive success also enhance the transmission of most (if not all) of the individual's genes. Individual and genetic payoffs are no longer in such close harmony when an individual's actions affect the reproductive success of relatives or when conflicts occur within an individual's genome. Interaoions among relatives
can be reconciled with an individualistic perspective by recourse to the concept of inclusive fitness (Hamilton, 1964; see Chapter 9), but intragenomic conflicts pose a Olore intractable problem because an individual's fitness. inclusive or otherwise, is ill-defined when different genes have different filllesses. Such conceptual difficulties do not arise if genes, rather than individuals, are treated as the strategists. (See Hurst et al. (1996) for a recent review.) Why use strategic thinking, which anthropomorphizes genes, instead of the well-developed infrastructure of population genetics? My reasons are pragmatic. Molecular biology reveals genes that are much more sophisticated than the stolid dominant or recessive caricatures of classical genetics. A gene may be expressed in some tissues. and some environments, but not in others; may have multiple alternative transcripts; may respond 10 signals from other genes; may have a history (be expressed when maternally derived but silent when paternally derived); and so on. Such complexities are diffieultto model by traditional genetic methods. Game theory, however, allows evolutionarily stable strategies to be selected from among a large range of alternative patterns of gene expression. The realism of a strategic analysis depends on the realism of the set of alternatives from which candidate strategies are chosen. Some conceivable strategies may be unavailable in the real world, but too restricted a set of alternatives can also mislead.
286
CHAPTER 12
12.3 Variant genes There are at least tWO defensible answer,; to the question 'How many words in this chapter?' The first is the number tallied by the word processor of my computer. In this answer, a word is a string of characters terminated by a space or punctuation mark. Each time 'vehicle' appears it is an extra word added to the tally. The second is the size of my vocabulary. In this answer, 'replica tor' counts as a single word no maller how many times it appears. Gene has a similar ambiguity. It can refer to the group of atoms that is organized into a particular deoxyribonucleic acid (DNA) sequence - each time the double helix replicates, the gene is replaced by two new genes - or it can refer to the abstract sequence that remains the same gene no maller how many times the sequence is replicated. The materia/gene (first sense) can be considered to be a vehicle of the informationa/gene (second sense). Debates about the 'units of selection' are interminable, partly because different meanings of 'gene' are connated. When hierarchical selectionists (e.g. Wilson & Sober, 1994) describe the gene as the lowest level in a nested hierarchy 01 units (species, populations, individuals, cells, genes), their sense is closer to the material gene, whereas, when gene selectionists (e.g. Dawkins, 1982) refer to the gene as the unit of selection, their sense is closer to the informational gene. The laller may be materially represented at multiple levels of the vehicular hierarchy, but is not itself a level of lhe hierarchy. The informational gene, however, is not precisely the meaning of gene selectionists. I will call their gene the strategic gene because their sense corresponds to the gene that is a strategist in an evolutionary game. Every genetic novelty (new informational gene) originates as a modification of an existing gene and is initially restricted to a few vehicles at lower levels of the malerial hierarchy, solely because it is rare. Therefore, the gene's material copies will interact with each olher only when they are present in different cells of the same body or in the bodies of closely related indiViduals. If such a gene is ever to become established, it must be able to increase in frequency under these circumstances. As the gene's frequency increases, its fate may be influenced by selection at higher levels of the hierarchy, but it will still retain the features that ensured its success when rare. Thus, the gene can be said to commit itself to a strategy when rare that it must maintain at all frequencies. The phenotypic effecls of successful genes will consequently appear to be adaptations for the good of groups of material genes that interact because of recent common descent. A strategiegene corresponds to such a set of material genes and can be considered the unit of adaptive innovation. In philosophical jargon, lhe strategic gene is a concept intermediate between the type (informational gene) and its tokens (material genes). The meanings of words (like genes) evolve and it would be futile to legislate a single meaning of 'gene', just as it would be futile to legislate a single meaning of 'word'. Semantic nexibility can even be useful when precise distinctions
SOCIAL GENE
287
are unimportant, because it allows subtle shifts of sense without becoming embroiled in long terminological explanations. Occasional inconsistency is somelimes the price of brevity.
12.4 The reach of the strategic gene A strategic gene is defined by the nature of the interactions among the copies of an informational gene that influence transmission of its sequence when copies are rare (Fig. 12.1). If all' copies of the informational gene acted in isolation, the only phenotypic efleas that would promote its transmission would be effects that directly promoled the replication of its individual copies. A strategic gene would then be coextensive with a material gene. Material genes need not act in isolation. For example. material genes that are expressed in
the soma of multicellular organisms do nOI leave direa descendants but promote the lransmission of their replicas in lhe organism's germ line. The strategic gene now corresponds to an organism-sized cluster of material genes. Similarly, a gene in the soma of one individual may promote the transmission of its copies in the germ lines of relatives. In this case, the strategic gene becomes
Fig. 12.1 An informational gene is an abstract sequence embodied by material genes (its physical copies). Material gen~ can be arranged in nested hierarchies: multiple copies
within each multicellular pea; multiple peas within a pod; multiple pods on a plant; multiple plantS within the local population; and so on. In the diagram. filled circles
represent the copies of a gene that 3re identical by (('cent common descent (rBRCD); 0PC' circles represent other alleles (the multiple copies of a gene within each pea are nOI represented). The definition of a strategic gene depends on how the phenotypic effeds of material genes influence the transmission of their TBReD copies. The effects of a materidl gene could affect the transmission of ils JBRCD copies solely via flowers of the seedling Ihat develops from its own pea; or could influence the transmission of its tBRCD copies via other peas in the same pod; or could influence the transmission of its IBRCD copies via I>t.'as in other pods; and 50 on. These possibilities are arranged in order of increasing numbers of material genes being luml>cd together in the concept of the strategic gene. Gene se1cetionisls and hierarchical selectionists use different terminology 10 explain tht' same reality. Protagonists of both schools want the other side to admit that their terminology is wrong.
288
CHAPTER 12
a cluster o( material genes distributed among some, but not aiL o( the members o( a (amily. Such a gene's strategy could be 'treat all offspring equally', not because all carry its copies, but because the gene has no way o( directing benefits prelerentially to the offspring who do. H a gene copy confers a benefit 8 on anolher vehicle at COSt C to its own vehicle, its costly action is strategically beneficial i( pB> C, where p is the probability that a copy o( the gene is present in the vehicle that benefits (see Chapter 9). Actions Wilh substantial costs therelore require signilicant values o( p. Two kinds o( (actors ensure high values o( p: relatedness (kinship) and recognition (green beards). The action o( kin seleclion is distinguished (rom a green beard e((ect because, under kin selection, a gene's stralegy is blind to the outcome o( each toss of the meiotic coin. Thus, the lrealment o( the individual members o( a class o( relatives does not depend on which genes they actually inherit, and p corresponds to a conventional coellicient of relatedness. By contrast, green beard eI(ects will discriminate between brothers with and without the relevant gene. Genes recognize kinship by historical continuity: a mammalian mother learns to identify her own o((spring in the act of giving birth; a male preferentially directs resources to the offspring of mothers with whom he has copulated; the other chicks in a nest are siblings; and so on. A green beard e((ect is in operation j( genes are recogoized directly by their phenotypic effects. Green beard e((ects gained their name (rom a thouglu-experiment o( Dawkins (1976), who considered the possibility o( a gene that caused its possessors to develop a green beard and to be nice to other green-bearded individuals. Since theo, a 'green beard e((ect' has come to reler to (orms o( genetic sel(-recognition in which a gene io one individual directs benefits to other individuals that possess the gene. The recognition 01 sel( and the recognition o( non-self are two sides of the same coin (see Chapter 4). Thus, the rejection 01 individuals that do not possess a label can also be conSidered a green beard e((ect, if the absence of the label is correlated with the absence of the genes responsible for rejection. Green beard e((ects have often been dismissed as implausible because a single gene has been considered unlikely to specify a label. the ability to recognize the label and the response to the label. However, these functions could also be performed by two or more closely linked genes. If X and Yare in linkage disequilibrium (Box 12.1) and X caUSeS its vehicle to treat vehicles with Y differently lrom vehicles wilhout Y, then (on average) X causes its vehicle to treat vehicles with X di((erently from vehicles without X. These concepts are applicable to many kinds of genetic interaction. When
homologous centromeres segregate at anaphase I o[ meiosis, their orderly behaviour is made possible by the prior recognition of some degree of sequence identity between homologous chromosomes (a green beard eHect), whereas when sister centromeres segregate at anaphase II, recognition is 1101 necessary because Ihe centromeres have been physically associated since their joint origin (rom an ancestral sequence (kinship). Similarly. the physical cohesion
SOC I A L G ENE
289
Box 12.1 Linkage disequilibrium and recombination Consider two informational genes Xand Y, If p(X) and plY) are the probabilities that Xand Yare present in a vehicle and p(XY) is the probability that they are jointly present. (hen the distributions of X and Yare statistically non-
*
independent if p(XY) p(X)p(YI, In other words, knowing whether Yis present provides information about X, If X and Yare alleles at different loci within a species, this l1oo-indt'pendence i known as linkage disequilibrium. but the ddinition of linkage disequilibriulll can be generalized to refer to all caseS of
non-independence whclher these occur within or between species, and whether or nor the concept of alleles at a locus (positions in a team) is well
defined, Species boundaries are a major cause of linkage disequilibrium (using the hroad definition above). For example, lhe genes of the introduced grey sqUirrel
havt' displaced the genes of the indigenolls
ft'd
squirrel from many English
forests, Some genes from red squirrels might do just fine in grey squirrel bodies. but they never get the chance. Thus. alleles al twu loci may be in linkage equilibrium within a species but in linkage disequilibrium between species. Chromosomal inversions can thus hav(;.' similar genetic consequences 10
speciation events if they restrict recombination between inverted and uninvcrted chromosomes. Linkage disequilibrium rcsults from one of three causes. A ncw mutation initially occurs in a single vehicle. on a single genetic
background. Recombination will spread the gene to new backgrounds. but the approach to equilibrium will be slow for closely linked genes, and cannot occur where there arc absolute barriers 10 recombination.
2
Epistatic seleclion generates linkage disequilibrium because il causes a gene
to leave more descendants when it is present in some combinations than in
others, The strength of disequilibrium will be determined by the balanre between the selective elimination of less favoured combinations and their
by rt:combination. 3 Sampling effects can cause linkage disequilibrium in small populations. regeneration
of the body is made possible because sister cells have remained in intimate contact since their origin from a common zygote (kinship), but this rich source of nutrients is defended against interlopers by an immune system that distinguishes self from non-self (a green beard effect),
12.5 The prokaryotic firm: managing a cytoplasmic commons Genetic replicalion makes use of energy and substrales that are supplied by the metabolic economy in much greater quantities than would be possible withoUi
290
CHAPTER 12
a genetic division o[ labour. These materials are common goods, available to every gene in the cytoplasm. Thus, genetic communities are potentially vulnerable to [ree-riders, genes that take more than they contribute, and the gains of trade from biochemical specialization would nut have been possible without the evolution of institutions and of procedures that limit the opportunities for social exploitation. In particular, striet controls arc expected on access to the machinery of replication. DNA-based replica tors are believed to have evolved from ribonucleic acid (RNA)-based replica tors, possibly because DNA is copied with greater fidelity than RNA (Lazcano et al.. 1988). The change also had implications for cellular security. Communities in which RNA polymerases were responsible for both replication and transcription would have been less easily policed than communities in which replication ('self-aggrandizement') was performed by DNA polymerases and transcription ('communal labour') by RNA polymerases. As a bonus, a community that modified its own genes to DNA, and periodically cleansed its cytoplasm with ribonucleases, would in the process eliminate most RNA-based parasites. An effective way to manage the cytoplasmic commons is to link genes to a single origin of replication, and to exclude non-members from the cytoplasm. The chromosome becomes a team whose members' interests coincide. The solution is egalitarian, at least within the group. Each gene that joins the chromosome is given equity and replicates once per cycle, no matter what its contribution during that cycle. Efficiency might conceivably be improved if genes that contributed more to productivity in the local environment were rewarded with increased copy number, but this argument ignores the costs of negotiating fair shares and of policing complex rules. The suppression of internal conflict by replication from a single origin has a price because genes can be copied more quickly from multiple origins than [rom one (Maynard Smith & Szathmary, 1993).
12.6 Dangerous liaisons Cosmides and Tooby (1981) called a set of genes that replicated together, and whose fitness was maximized in the same way, a coreplicon. They argued that intragenomic conflicts are likely if an organism contains more than one coreplicon, because the members of a corepUcon will sometimes be selected to maximize their own propagation in ways that interfere with the propagation of other coreplicons. Selection for short-term and long-term replication may be opposed. A coreplicon that replicated faster than other coreplicons within its cell lineage would increase in frequency. However, if differential replication were costly for cell survival, the cell lineage would eventually be eliminated in competition with other lineages. Thus, the long-term interests of corepJicons that share a cell lineage will coincide if they never have opportunities to form new combinations with other genes in other lineages. Recombination
SOCIAL GENE
291
decouples genes' fates and is therefore essemial for the indefinite persistence of imragenomic conflict (Hickey. 1982). Many bacteria comain muhiple circular genomes. By convemion. one of these circies is designated the bacterial chromosome. and the extra circles are called plasmids. Plasmid replication consumes energy and substrates. Whether a plasmid pays for its keep - from the perspective of chromosomal genes - depends: (i) on the metabolic skills that its genes bring 10 the cell; (ii) on whether these skills are required in the current environment; and (iii) on the degree uf coadaptalion between plasmid and chromosome. Many of the genes for amibiotic resistance that are the scourge of modern hospitals are carried on plasmids. Such a plasmid may be esselllial in the presence of amibiotics but a burden in their absence (Eberhard. 1980). Most plasmids promote conjugation between their host and other bacteria, although some smaller plasm ids rely on larger plasmids for these functions. In the process, a copy of the plasmid is retained by the donor cell and an uninfected bacterium acquires a plasmid. Thus. conjugation allows plasmids to colonize new cytoplasms. Chromosomal genes. by contrast. are usually not transferred. Therefore. the chromosome bears the costs of replicating the donated plasmid, and the costs of increased exposure to viruses during conjugation. bill seemingly gains lillIe in return. If the plasmid encodes beneficial functions these are transferred to a potelllial competitor. Perhaps, if a plasmid is a burden to chromosomal genes. the chromosome benefits from sharing its cold with rivals. Plasmids cannot be categorized simply as parasites or mutualists. For example, a plasmid that initially reduced its host's competitiveness enhanced fitness after plasmid and chromosome were propagated together for 500 generations (Lenski et al.• (994). A plasmid has two modes of transmission, vertical and horizontal. and selection can favour its propagation by either path. Selection for greater horizontal transmission, at the expense of vertical transmission, will generally increase the costs of the plasmid to chromosomal genes. whereas selection for increased vertical transmission will generally benefit the chromosome. In the limit, when there is no horizontal transmission of the plasmid. the long-term fates of chromosome and plasmid are inexorably linked and they effectively become a single coreplicon. Similar arguments apply to the viruses, transposons and other coreplicons that populate bacterial cytoplasms.
12.7 Plasmid protection rackets Plasmids, once acquired. are difficult to discard. Plasmid genes encode multiple functions that ensure their stable transmission within an infected lineage (Nordstrom & Austin. 1989). Many plasmids encode a persistent 'poison' and its short-lived 'antidote'. Thus. if a cell segregates without the plasmid. it is cut off from its supply of antidote and succumbs to the poison. The gene for the
292
CHAPTER 12
poison can be said to recognize the presence or absence of the gene for the antidote. Because poison and antidOle are inherited as a unit. the plasmid can be said to recognize itself (a green beard effect). Such protedion rackets take many forms (Lehnherr el al., 1993; Salmon el al.. 1994; Thisted et al.. 1994). Some plasmids, for example. encode a methylase and its matching restriction enzyme. The methylase modifies baCierial DNA so that it is protcded from the restriction enzyme, which cleaves unmodified DNA. Bacteria that lose the plasmid die, because methylalion must be restored each time the chromosome replicates. The restriction enzyme simultaneously defends its cytoplasm against viruses and rival plasm ids that lack the appropriate methylase, just as a gangster defends his turf (Kusano et al.. 1995). Mitochondria of trypanosomes contain a large DNA maxicircle andt'nany small minicircles. The maxidrcle encodes essential genes in garbled form. whereas the minicircles encode guide RNAs that edit the otherwise unreadable transcripts to yield translatable messenger RNAs (Benne, 1994). Could RNA editing have evolved as a minicircle-maintenance system? If so. one would predict that minicircles can also edit DNA and encrypt maxicircle genes in ways that only they can decipher. Minicirdes have been observed to spread from one mitochondrial lineage to another (Gibson & Garside, 1990). strengthening their analogy with baCierial plasmids.
12.8 Team substitutions A non-recombining bacterial chromosome is a learn thai does not change its members (except by mutation). Its social cuntract is 'all for one, and one for all' not 'every gene for itself.' Chromosomal recombination occurs on rare occasions as a coincidental side-effect of the horizontal transfer of plasmids (conjugation) or viruses (transdudion). Some bacteria. however. have evolved mechanisms by which DNA is taken from the environment and used to replace homologous sequences of the chromosome (natural transformation). Transformation. unlike conjugation and transdudion. is controlled by chromosomal genes (Stewart & Carlson. 1986). Uptake of DNA is induced under conditions of nutritional stress and may have evolved primarily as a means of gaining nutrients (Redfield. 1993). Nevertheless. the expression of DNA-binding proteins that prevent the degradation of the donor sequence and the induction of the enzymatic machinery of recombination (Stewart & Carlson. 1986; Lorenz & WackerknageJ, 1994) suggest that recombination is not a mere side-effect but has been positively selected. Why should a team replace one of its members? The repair hypothesis views transformation as a means of replacing injured team members (damaged DNA). However, repair is unlikely to be the principal fundion of transformation because uptake of DNA is not induced by damage to the chromosome (Redfield, 1993). The recombinant-progeny hypothesis views transformation as a means of trying out new players. Replacement of one gene by another occurs in a
SOCIAL GENE
293
single cell of a clone and does not commit other team members to the new combination, because Ihe old combination survives in other cells of the clone. A team's chances of remaining successful in a changing environment will presumably be improved by some degree of experimentation with new combinations. The problem is that, for each member of the team, Ihe advantages arise only from changes at positions other than its own. Genes would be selected to increase their chances of replacing an established gene on other chromosomes, but to decrease their chances of being so replaced. The important social question becomes whether some positions are privileged and not subjea to replacement; in panicular, whether the genes responsible for transfomlation are themselves transformed.
12.9 Multicellular corporations The development of resistant spores by Bacillus sub/ilis iIIuslrates the differentiation of soma and germ line in simple form. A bacterium undergoes an unequal cell division 10 produce a mother cell (soma) and a prespore (germ line). The mother cell engulfs the prespore, assists in formation of the spore coat and is then discarded. The process is coordinated by an exchange of signals between mother cell and prespore (Errington, 1996). The genes of the mother cell sacrifice themselves for their replicas in the spore. Some bacterial somas are more complex. Myxococcus xan/hum is a motile, predatory prokaryote that forms a multicellular fruiting body. Individual myxobacteria forage and divide in the soil but, when nutrients become scarce, they aggregate to form a structure in which sacrificial slalk cells (soma) raise myxospores (germ line) above the substrate (Shimkets, 1990). Organisms develop somas to gain the benefits of a cellular division of labour. A soma, however, is a rich resource that can be exploited by genes of other germ lines. Therefore, the advantages of somatic specialization can be realized only if the genes of the soma have some degree of confidence that their copies are represented among the beneficiaries of their labour. The simplest means by which genes in somatic cells ensure that their effons are well directed is physical cohesion between snma and germ line. The genes of the Bacillus mother cell can be assured Ihattheir copies are present in the prespore because cell division and sporulation take place within an enclosed sporangium that excludes outsiders. However, as somas become larger and more complex, interactions between somatic ceUs and germ cells become less direct. This creates additional opponunities for parasites 10 misappropriate somatic effon, and necessitates more elaborate security systems to protect the soma from exploitation. The genes of my liver are almost cenain to have copies in my testes (because of my body's physical cohesion), but my lump of lard and mass of meat would not lasL long without a sophisticated system of immune su rveillance.
294
CHAPTER 12
Uninterrupted physical cohesion cannot protect the genes of somatic cells from exploitation if sister cells become detached to forage (as in MyxocolXUs) or to form complex organs (as in multicellular animals). Some form of cellular recognition is required. When two cells meet, their responses can be influenced by what they learn about each other. Molecules on their surface can provide clues about which genes are present in a cell and whether the cell is friend, foe or indifferent. lWo categories of molecular interactions can be distinguished. Homotypic interactions occur between identical molecules on the two cells and are a particularly direct means for a gene to recognize itself in other cells. Heterotypic interactions occur between molecules encoded by different genes and can also provide a gene with information about its presence or absence in another cell iI there is linkage disequilibrium between the interacting genes (Haig, 1996). Thus, green beard effects may play an importanl role in the somatic security systems of multicellular organisms (particularly making use of the complele linkage disequilibrium between genes of different species). The origin of molecules that were able to discriminate between themselves and closely similar molecules greatly expanded the strategies available to genes and made possible the evolution of large multicellular bodies. The ancestor of the immunoglobulin superfamily probably interacted with itself in homotypic adhesion or signalling, but the family now includes many heterophilic adhesion molecules as well as the T-cell receptors, major histocompatibility complex (MHC) antigens and immunoglobulins of the vertebrate immune system (Williams & Barclay, 1988). The cadherins, to take another example, are a family of cell-surface proteins that bind to copies of themselves on other cells. Nose et al. (1988) introduced the genes for P-cadherin and E-cadherin into a cell line which lacked cadherin activity, creating two sublines that were identical except for this single gene difference. When the cells were mixed, they spontaneously segregated into discrete populations, like oil mixed with water. Cadherins playa pivotal role in organogenesis, but similar mechanisms could clearly be used to distinguish self from non-self.
12.10 A chimeric menagerie Slime moulds are eukaryotes with a life cycle remarkably similar to Myxococcus (Kaiser, 1986). They feed as unicellular amoebae, but aggregate when starved to form a fruiting body with a simple division of labour between spores and somatic stalk. Slime moulds are thus particularly vulnerable 10 somatic exploitation because there is no guarantee that the amoebae who respond to an aggregation signal are members of the same clone, or that a predator will not use the signal to lure amoebae to their doom. The dangers are real, although somewhat mitigated by the mechanisms of cell-surface recognition discussed in the previous section. Dictyoscelium caveatum is a predator that responds 10 the aggregation signals of other species and devours their amoebae before forming its own fruiting body (Waddell, 1982). Zygotes of D. discoideum produce
SOCIAL GENE
295
the aggregation signal and devour haploid amoebae of their own species as they respond to the signal (O'Day, 1979). Some strains of D. discoideum form chimeric fruiting bodies with amoebae of other strains without contributing to the stalk (Buss, 1982). Chimerism between members of a single species has also been described in animals. Vascular fusion [requently occurs between neighbouring genotypes of the colonial urochordate Botryllus schlosseri. The progenitors o[ germ cells circulate in the blood and will colonize, and in SOme cases totally replace. the gonads of the neighbouring soma (Pancer et al' 1995). As another example, 'hermaphrodite' females of the haplodiploid scale insectlcerya purchasi are host to spermatogenic cells derived from sperm that entered the cytoplasm of an egg, but which failed to fertilize the egg nucleus because they were pre-empted by another sperm (Royer, 1975). Thus, a sperm that fails to fertilize the eggs of the mother can try again with those of the daughter or granddaughter, or persist as a permanent haploid inhabitant of female somas. Occasionally winged males are produced from unfertilized eggs (Hughes-Schrader, 1948), but must compete for fertilization with the 'reduced males' resident in female gonads. Marmosets and tamarins regularly produce dizygotic twins in a uterus ancestrally designed for singletons (i.e. in a simplex uterus which lacks long uterine horns to keep squabbling offspring apart). The placental circulations of the twins fuse, with the result that each adult marmoset carries blood cells derived from its twin (Benirschke et al' 1962). If germ cells were also transferred, and equally mixed between twin brothers, the genes of their respective somas would be indj[ferent about which brother copulated, although competition within the brothers' testes and ejaculates could be intense. Chimerism between dizygotic twins is the rule for marmosets but is the exception for human twins (van Dijk et al., 1996). Chimerism is common, however, between human mothers and their offspring. Fetal cells circulate in a mother's blood from the early weeks of pregnancy and descendants of these cells may persist in a mother's body [or decades after the child's birth (Bianchi et al., 1996). Are these cells simply lost, or do they manipulate the maternal soma [or the offspring's benefit? Exploitation of host somas by pathogens and parasites remains a major problem for multicellular organisms, This seoion's colleoion of intraspeCific chimeras emphasizes that the risk of somatic exploitation is not restricted 10 members of different species. Of course, somatic exploitation within species usually involves the everyday strategies of coercion and deceit.
12.11 The nuclear citadel The spefd of replication limits the amount of DNA that can be copied efficiently frem a single origin of replication. The chromosome of Escherichia coli takes aboul40 min to replicate (Zyskind & Smith, 1992). If the I OOO-foid larger
296
CHAPTER 12
genome of Homo sapiens were similarly organized as a circular chromosome with a single bidirectional origin of replication, it would take almost 1 momh to replicate (comparative genome sizes: Monon, 1991; Fonstein & Haselkorn, 1995). Humans and other eukaryotes avoid this problem by using multiple origins of replication. The auendant risk that some parts of the genome will replicate faster than others is exacerbated because the alternation of gametic fusion and meiotic segregation creates ample opportunities for rogue elements to colonize new genomes (Hickey, 1982). For these reasons, eukaryotes are expected to have evolved sophisticated systems for controlling unauthorized replication. Two characteristic features of eukaryotic cells probably contribute to replicative security. The first is the separation of the machinery of protein synthesis (in the cytoplasm) from the genetic material (in the nucleus). Passage of large molecules to and from the nucleus is controlled at the nuclear pore complex. Before a protein can dock with this complex, it must possess nuclear localization signals that are recognized by docking molecules in the cytoplasm (Davis, 1995; Hicks & Raikhel, 1995). The second is the eukaryotic cell cycle. Replication is confined to a specific S phase. Before DNA can replicate, it must acquire a 'replication licensing factor' that authorizes it to replicate once, but once only, per cycle (Rowley et oJ., 1994; Su et oJ., 1995). The origin recognition complex (ORC) that marks a site for future initiation of replication causes transcription to be silenced in its vicinity (Rivier & Pill us, 1994). Thus, genes near an ORC are prevented from producing locally acting RNAs that could tamper with the genes' own replication. The bread mould Neurospora crassa has evolved a highly effective defence against genetic elemems that replicate more than once in a cell cycle. If a sequence is repeated within a haploid nucleus, both copies are inactivated by methylation and subject to a process of repeat-induced point (RIP) mutation until their sequences have diverged sufficiently to be no longer recognized as similar (Selker, 1990). Thus, if a DNA sequence replicates faster than other members of its collective, both the additional copies and the master sequence are corrupted by a process of programmed mutation. Vertebrates compartmentalize their DNA into active regions and methylated regions that are maintained in a compact transcriptionally inactive state (BeslOr, 1990; Bird, 1993). The inactive ponion of the genome often contains large amounts of simple repetitive sequences that do not encode proteins and which are subject to high rates of sequence turnover because of replication slippage and unequal crossing over (Dover, 1993). This arrangemem may function. in part, as a system of defences against intragenomic parasites. First. a higher proponion of insenions will occur in non-critical sequences. Second. foreign DNA (once inserted) is transcriptionally inactivated by a methylation process which may speCifically recognize structural features of parasitic DNA (BeslOr & Tycko, 1996). Third. insened DNA is subjeclto sequence degradation by lhe processes of genomic turnover.
SOCIAL GENE
297
11 would be misleading to argue that the sole function of the organizational changes of the eukaryotic nucleus has been internal security. Even when agel1ls have idemical interests there is still a problem of coordination. The genome of E. coli contains about 4000 genes whereas the genomes of humans. mice and pufferfish contain about 80000 different genes. Bird (1995) has argued that the nuclear envelope and hislOne proteins of eukaryotic cells. and the extensive methylation of vertebrate chromosomes. are adaptations for the reduction of the transcriptional noise assodated with larger genomes. New security measures would have evolved hand in hand with new mechanisms of control.
12.12 The sexual revolution Bacterial recombination involves the formation and dissolution of partnerships between coreplicons or the substilution of one gene for another in a process that has clear winners and losers. By contrast. meiotic recombination involves a symmetrical relationship in which two lemporary teams come together. swap
members and form new temporary teams. The members of successful teams gel 10 play more often in the next generation than members of unsuccessful leams. Thus. a successful player is one who performs well as a member of many differemteams. and the system favours teams of champions rather Ihan champion teams. Team members pursue the same goals. nOI because their long-term destinies arc indissolubly linked. but because the rules of meiosis ensure that all receive the same opportunities if only their team can make il through to the next 10llery. If there were completely independenl assortment of genes at meiosis. players could not form long-term partnerships because any two players present in a haploid team before gametic fusion would have an even chance of parting at meiosis. This 50% probability of recombination per generation applies to almost all randomly chosen pairs of genes in organisms with multiple chromosomes. Genes that are linked on the same chromosome can expect 10 remain associated for longer periods. If some combinations of linked genes work more effectively IOgelher than others. these combinations will tend 10 occur in successful teams and leave more descendants than less favoured combinations. By this process. selection generates non-random associations of players (i.e. linkage disequilibrium). but these assodations are constantly being disrupted by recombination. One of the major preoccupatinns of evolutionary genetics has been the question why so many genetic collectives regularly break up successful teams to take a chance on untried combinations. ZhivolOvsky et al. (1994) summarized numeroos models lhat reached the conclusion: in a random mating population. if a pair of loci is under constal1l viability selection (the same in both sex's). with recombination between them controlled by a modifying gene. and if this system
298
CHAPTER 12
anains an equilibrium at which the major genes are in linkage disequilibrium, then new alleles at the modifying locus can invade only if they reduce the rate of recombination between the major loci. A similar principle applies for an arbitrary number of lod (ZhivolOvsky et al.. J 994). The intuitive explanation of this 'reduction principle' is that new teams generated by recombination will, on average, be less successful than existing teams that have survived a generation of selection. Therefore, individual players arc more likely to be successful in the next generation if there is less recombination of thdr current team.
Recombination is widespread in nalure, and one or more of the assumptions of models that predict selection for reduced recombination must be violated. Genes that increase recombination can be favoured if a population has not reached selective equilibrium because recombination increases the efficiency with which currently favoured players are broughtlOgether in the same team. Technically, the advantage a team gains from haVing both gene A and gene B must be less than the sum of their individual contributions 10 team success (Barton, J 995). A similar process favours increased recombination if the cost of injury to A and B (i.e. mutation) is greater than the sum of the costs if A and B were damaged individually (Charlesworth, J 990). In both these examples, increased recombination improves the effidency of selection because it reduces the risk that inferior players will 'hitch-hike' on the success of their team mates or, what amounts 10 the same thing, that superior players will be dragged down by lesser players. Theories that ascribe the adaptive advantage or recombination to increased resistance lO parasites arc explanations of this sort because recombination rates evolve in a constantly deteriorating environment in which the most favoured allelic combination is always in nux (Hamilton etal., 1990).
12.13 The open society and its enemies The reduction principle also breaks down in the presence of multilocus green beard effects. Green beard effects allow genes 10 direct benefits 10 teams in which they have a high probability of being present. As we have seen, a gene (or coalition of genes) can profit from conferring a benefit B on another team at cost C to its own team, if pB> C where p is the probability that the gene (or coalition) is present in the team that benefits. If this probability were the same for all genes in the donor team, all members would gain equally from the transaction. But, if the probabilities differ - as they do when benefits are directed 10 green-bearded relatives at the expense of other relatives (Ridley & Grafen, 1981) -some team members will lose while others gain. Linkage disequilibrium can enable small coalitions of genes to conspire against the common good, but high levels of recombination will disrupt the persistent non-random associations on which multilocus green beard effects depend.
SOCIAL GENE
299
Other team members would suspect the motivation of a group of players who were simultaneously members of a rival team, and can benefit from disrupting cliques before they form. The best-studied conspiracies are systems of meiotic drive. A haplotype in a heterozygous diploid causes the failure of gametes that do not carry its copies, usually by means of a two-locus poison-antidote mechanism. If the haplotype does nol go to fixation. it must be associated with countervailing fitness costs that will be experienced in full by team members that are unlinked to the haplotype. Therefore. selection at unlinked loci favours increased recombination to disrupt the conspiracy and separale lhe poison from its antidote (Haig & Grafen, 1991). Leigh (1971; see also Eshd, 1985) has compared the genome to 'a parliament of genes: each acts in its own self-interest. bOl if its actS hurt the others. they will combine together to suppress it.' Segregation distortion and related phenomena arc departures from fairness. Leigh (1971) suggested: The transmission rules of meiosis evolve as increasingly inviolable rules of fair play. a constitution designed to protect the parliament against the harmful acts of one or a few.... Just as too small a parliament may be perverted by the cabals of a few, a species with only one. tightly linked chromosome is an easy prey to distorters.
12.14 The eukaryotic alliance Most internal conflicts within the nucleus are defused by the procedures of fair segregation and allelic recombination. However, eukaryotes also contain genes, in mitochondria and plastids. that are nOt part of the meiotic compact. The eukaryotic cell originated as an alliance between nuclear genes and the genes of symbiotic bacteria. Many of the laller eventually joined the nuclear firm, but some retained a limited independence as the mitochondrial and plastid genes of today. We do not fully understand why some genes have accepted (or been granted) nuclear equity whereas others have maintained a separate contractual arrangement, nor why these shifts of allegiance have been predominantly one-way, from organelle to nucleus. Nuclear and organellar genes are mutually dependent, yet their different rules of transmission ean be a source of conflict in their partnership. If different organellar lineages occupied lhe same cytoplasm after gametic fusion, the lineages would be expected to compete for occupation of the cytoplasm. with concomitant costs to nuclear genes. Cosmides and Tooby (1981) suggested that nuclear genes have been selected to minimize conflicts among organellar genes by causing the destruction of the organelles of one gamete, either before Or after syngamy. For this reason, they proposed, the nuclear genes of one kind 01 gamete (sperm) discard their organellar partners before krtilizing a different kind 01 gamete (eggs) which retain their organelles (see Hastings. 1992; Hurst & Hamilton. 1992; Law & Hutson, 1992 lor related
300
C H A PTE R 12
arguments). Nudear-enforced suppression of cytoplasmic conflict may thus have been the key factor in the evolution of eggs and sperm. with all other differences between the sexes arising from this initial dichotomy. In support of this conjecture, Hurst and Hamilton (1992) have noted that morphologically distinct sexes are absent in taxa that exchange nuclear genes without cytoplasmic fusion. Uniparental inheritance of mitochondria and plastids resolves one conflict but creates another. Nuclear genes are transmitted by sperm and eggs whereas organellar genes are transmitted by eggs alone. Organellar genes would therefore benefit from preveming reproduction by male function. if this increased the resources available for female function. Cytoplasmic male sterility has evolved many times in flowering plants. In aU well-studied cases. male sterility is caused by mitochondrial genes but their effects are often countered by nuclear genes which restore male fertility. Chloroplasts also have predominantly maternal inheritance. but chloroplast genes are not known to cause male sterility (Saumitou-Laprade et af.. 1994). The plastid genome may lack mechanisms to abon male function. or. if such mechanisms exist. they may be circumvented easily by nuclear genes. Despite its internal conflicts. the eukaryotic alliance has been an outstanding success. The philosopher Daniel Dennett (1995. PI'. 340-1) considers humans to be a radically new kind of entity. comparable in importance to the eukaryotic cell. In his view. we are a symbiosis between genetic replicators and cuhural replicators (memes). Just as eukaryote cells cannot survive without both nucleus and organelles. we cannot survive without both genes and memes; neither genes nor memes are dispensable; and neither genes nor memes can claim priority as representing our true selves. Genes and memes have very different rules of transmission and a meme cannot simply be incorporated into a chromosome where it follows the rules of meiosis. Conflicts are therefore expected. Some people will die for an idea. Others will abandon their faith for a brief sexual encounter.
12.15 Sex chromosomes An average gene from a species with two sexes spends equal time in male and female bodies because every individual has a mother and a father. However. at a selective equilibrium some genes (or combinations of genes) may be more successful than average in one sex and less successful than average in the other. Such sexually antagonistic genes will benefit from being associated with other genes that bias sex determination towards the sex in which they have a relative advantage (Rice. 1987), whether this is a conventional fitness advantage (viability) or a segregational advantage (meiotic drive). The process is self-reinforcing because genes that influence sex determination spend more time in one sex than the other and can thus persist in linkage disequilibrium with genes whose disadvantage in the less frequem sex is greater than their
SOCIAL GENE
301
advanlage in Ihe more frequent sex. Genomes lhus have a tendency to split into factions that spend equal time in the two sexes (autosomes) and factions thai specialize in one sex or the other (sex chromosomes). Meiotic drive in spermatogenesis or oogenesis (but not both) can favour the evolution of sex chromosomes. because a distorter has an advantage in one sex that is abselll in lhe other. Segregation distorters will also be favoured if they arise on existing sex chromosomes. Associations between agents of meiotic drive and the genetic determiners of sex resuh in biased sex ratios. but these biases will be opposed by the parliamenl of genes (or at least by ils autosomal majorily). Half of the genes in the next generation will come from males and haii from females. This means that members of a minority sex will leave more descendants. on average. than members of the majority sex. and autosomal genes will benefit from being present in the minority sex (assuming that lhe sexes are equally costly). Autosomal genes arc expected to enforce fair segregation of sex chromosomes in the heterogametic sex because neither sex will then be in a majority. Hamilton (1967) recognized that autosomal genes will somelimes favour biased sex ratios. He considered a model in which small numbers of unrelated females founded local populalions. their offspring mated among themselves. and the newly mated females dispersed to found new local populations. If males were heterogametic and segregation was strictly Mendelian. the sex ratio in the global populalion would be very close 10 unily. with some variation among local populations because of random fertilization by x- and V-bearing sperm. The expected fitness of a female offspring would be the same as the average female fitness for the global populalion, regardless of the local SeX ratio. bUI the expected fillless of a male offspring would increase with the local proportion of females. Therefore. an autosomal gene that caused itself to be present in female-biased local populations would have higher than average fitness. and a balanced sex ratio would no longer be the unbeatable strategy. In this example. the parliamenl of genes contains a number of parties with different policies concerning the sex ralio. The X party. the autosomal party and the mitochondrial party would enter into coalition against the Y party to force a female-biased sex ratio among offspring. but the coalition partners would lack unanimity about precisely which ratio (Hamilton, 1979). Sexual politics can profoundly destabilize the 'parliamentary rules' of meiosis (Haig. 1993).
12.16 Genomic imprinting and the altercation of generations Relatives arc genetic collectives lhat share some. bUI not all, of their members. A gene can benefil from employing a contingent strategy Ihaltreats collectives differently depending on information about the probability r that a collective includes one of its copies. This section will assume lhat a gene's only infornlaliun
302
CHAPTER 12
about r comes from the family tree (pedigree of collectives) and the Mendelian probabilities associated with the pedigree, in some cases supplemented with knowledge of parental origin and the uncertainty of paternity. Green beard eHects will not be considered. The simplest relationship between a diploid mother and her sexually produced diploid oHspring is one in which the mother produces a series of eggs that are provisioned, fenilized and scattered, without subsequent maternal care. Each gene in the mother has an equal probability of being present in each oHspring - determined by a flip of the meiotic coin - and the quantity of yolk received cannot be influenced by genes expressed in oHspring because provisioning is completed belore meiosis. Genes of the mother can do no better than produce the size and number of eggs that maximize the mother's expected lifetime reproductive success. In a simple model in which eggs are produced sequentially on a production line from a fixed quantity of resources, maternal fitness is maximized when each egg is supplied with an amount of resources such that the marginal benefit (liB) that an oHspring would gain from a little bit extra committed to its egg would equal the marginal cost of these resources (0C) to another offspring that develops from an egg at the end of the line. Marginal costs and marginal benefits are given equal weight because a gene in the mother has the same chance of being present in either offspring. The relationship becomes more complex if offspring receive post-zygotic maternal care, because the amount of care received can be influenced by genes expressed in offspring. A gene expressed in the current offspring gains the full marginal benefit of extra resources received from the mother but has only a probability r of being present in the offspring that experiences the marginal cost. Therefore, genes expressed in offspring will favour receiving extra resources as long as OB> rOC, whereas genes expressed in the mother will favour terminating investment once oC> OB. Thus, parent-offspring conflict exists whenever OC> liB > rOC (Trivers, 1974; Haig, 1992). This conflict arises from the difficulty of making binding agreemellls. Even though genes in a parent would agree among themselves to terminate investment in each offspring when OB = OC, the agreement is generally unenforceable once genes
find themselves in offspring. All genes would do better if offspring demanded less, but unilateral restraint will be exploited. The probability r is 50% for maternal genes expressed in oHspring. whereas rwill generally be less than 50% for paternal genes. This is because the offspring that gains a marginal benefit from extra maternal resources may have a different father from the offspring who suffers the marginal cost. Therefore, paternal genes in offspring are predicted to make greater demands on mothers than are maternal genes in the same offspring. Such conditional strategies are made possible by genomic imprinting which causes genes to have different patterns of expression depending on whether they spelllthe previous generation in a male or female germ line (Moore & Haig, 1991). For example, during murine development, insl/lin-like growth factor 2 (/g/2) is expressed from
SOCIAL GENE
303
the paternal allele and the maternal allele is silent, whereas the insulin-like growth fador 2 receptor (/g[2r) has the opposite pattern o( expression_ Mice with an inactivated paternal copy o( /g[2 are 60% normal size at birth, whereas mice with an inactivated maternal copy o[ Ig[2r are 20% larger than normal at birth. Birth weight is normal in mice that inherit the inactivated genes [rom the opposite parent (DeChiara etal.. 1991; Lau eta/., 1994).lgf2r has been proposed to [unction by degrading the products o[ /912 (Haig & Graham, 1991). Thus, Ihese genes employ a conditional stralegy: 'make greater demands when paternally-derived than when maternally-derived: The relatedness asymmetry betwcen maternal and paternal genes is maximal [or hall-siblings, but most kinds o( relatives will have different degrees o( maternal and paternal relatedness. These patterns can be quite complex. Consider a hypothctical social system in which males disperse and [emales remain in their natal group. Jl all ollspring within a group are fathered by a single male who maintains a monopoly on sexual access to the females until he is displaced by a new unrelaled male, female group members of different ages will be closer relatives on the maternal side Ihan on the paternal side becausc o( (emale philopatry, whereas ollspring of the same age will be either full siblings or paternal hal[-siblings. It is not known whether genes have evolved conditional strategies that take account of such pallerns of relatedness. All the well·studied cxamples of genomic imprinting so (ar appear to be simplc conditional stralegies o( the form 'do One thing when maternally derived and something else when paternally derived: More complex conditional strategies are logically possible - for example, 'do one thing when derived (rom an egg; something else when derived (rom a sperm o( a resident male; and something else when derived (rom a sperm o( a cuckolding male' - but, whether such logical poSSibilities are actually realized depends on COSIS, benefits and Ihe existence of appropriate mechanisms. In the social system of the previous paragraph, a gene's probability of being present in other [emale group members increases as it passes Ihrough successive (emale germ lines. One could imagine a gene subject 10 a cumulative imprinting el[eClthat was reset to zero every time the gene passed through a male germ line.
12.17 Reprise Our intuitive concept of the genetic boundaries o( an organism approximates Ihe membership of a coreplicon (i.e. of a sel o[ genes thai are transmitted by the same rules). A coreplicon evolves as a unit with common goals because its members benefit (rom the same outcomes, whereas the genes o( dille rent cnreplicons can have conOicling interests. Thus, prophage genes inserted into a bacterial chromosome are distinguished from 'true' bacterial genes because o( Iheir alternative mode o( transmission. They can be mobilized to outreplicate their companions; package themselves into resisLant phage particles; and lyse
304
CHAPTER 12
their host. The relationship between coreplicons need not, however, be strictly adversaria!. One coreplicon can obtain a good from another by trade, as well as by theft, but with room for haggling over the price. A coreplicon, however, functions as a commonwealth, without an internal market. Its members thus avoid the transaction costs of finding buyers; of learning the prices and quality 01 the goods on offer; and are protected from hucksters and frauds in the marketplace (d, Coase, 1993). Bacterial cells usually contain a small number of coreplicons (sometimes only one). Recombination between bacterial chromosomes is rare and, when it occurs, is a substitutional process in which one gene (a winner) is substituted for another (a loser). By contrast, eukaryotic recombination is much more frequent, and is a segregational process in which genes are swapped between chromosomes, without winners and losers. The coreplicon is nO longer a group of non-recombining genes who cooperate because their long-term fates are intimately bound, rather it becomes a group of temporary associates who obey the same rules and who gain an equitable division of resources when their
ephemeral partnership is dissolved. High-frequency recombination creates a market (of a son) for currently favoured team players. Recombination creates relatives - genetic collectiv.es that share some, but not all, of their genes. Interactions with relatives are a potential source of internal dissension within the collective, because some members of the collective can gain at the expense of others, for example, by sabotaging other members' gametes or by favouring some offspring over others. High levels of recombination can be a partial solution to the connicts created by lower levels of recombination, because randomizing devices disrupt the 'cabals of the few'. The reasons for the eukaryotic sexual cycle of gametic fusion, recombination and meiotic segregation remain somewhat unclear, but the process probably enhances individual genes' chances of long-term survival in a changing selective environmenL.
PART 4 LIFE HISTORIES, PHYLOGENIES AND POPULATIONS
Habitat quality and ,vmpn;tioll with otlter bird)' d('lerminf how an oysrercaldler Hacmawpus oSlralegus chooses where 10 Iud and where to breed. Can we liSt! an understanding ofindividual decisio,,·makinglo prrdicI how habitat/oss will i,,/lllmce populatioll number? (Photograph by Jan van de Kam.)
Part 4: Introduction
How should individuals allocate resources to their own growth and survival versus reproduction? When they reproduce. how should they divide their effort between current and future reproduction or between number and quality of offspring? These trade-offs form the basis of life-history theory. discussed by Daan and Tinbergen in Chapter 13. Much of the current research is still stimulated by David Lack's classic studies in the 1950s and I 960s on optimization of clutch size. Today's studies use more sophisticated measures of energy expended by parents in raising broods of different sizes and test how these influence lifetime reproductive success. Daan and Tinbergen emphasize that life-history trade-offs cannot be studied by using natural variation in dutch size and parental survival because this often reflects adaptive decisions made by individuals of different quality. Instead. experimental manipulations are needed where. for example. clutches of different sizes are allocated randomly across individuals to measure the erfects of reproductive effort on adult survival. Some of these experiments provide good evidence for optimization of reproductive effort in relation to individual quality. Comparative studies across species show that many aspects of life history. from longeVity to heart beat rate. scale with body mass. Traditionally. these scaling relationships have been assumed to reflect basic physiological constraints within which organisms are forced to operate. Daan and Tinbergen point out that this cannot explain why species may vary by a factor of 10 from the allometric prediction and favour the view that the relationships may arise as a consequence of resource allocation between growth and reproduction to optimize life-history trade-offs. As well as providing a clear account of current theoretical issues. the chapter summarizes some n'at recent experimental work on life histories in lizards and birds. revealing that individuals exhibit 'phenotypic plasticity', namely a single genotype can produce
308
PART 4: INTRODUCTION
to unravel how evolutionary change in behaviour has been influenced by changes in ecology. The main advantage of this approach (the method 01 independent contrasts) is that it can be osed to identify independent evolutionary events and so 10 leSt statistically whether various characters change in relation to each other (e.g. warning colours and gregariousness in insects
or sexual dimorphism and extrapair mating in birds). However. as Harvey and Nee point out. the problem is that the conclusions may vary depending on the model assumed for evolutionary change, on the exact phylogeny used and on the statistical mel hods. Faced with this minefield 01 complexity, some workers have used the simpler technique of 'pairwise' comparisons between closely related taxa (e.g. species in the same genus) to provide independel1l evolutionary events. Finall}, Harvey and Nee discu s some recent exciting studies 01 lizards in the Caribbean and warblers in the Himalayas where phylogenies can be used to infer how ecological niches are filled as speciation occurs. The importance 01 evolutionary history for interpretation of current adaptation is continued as the theme in Chapter I S, where Hewitl and Botlin consider how the genetic strudure of populalions is influenced by both historical factors and currem selective pressures. Behavioural ecologists often assume that the organisms they study are well adapted to local conditions and they then ask how various behaviour patlerns might maximize an individual's fitness. However, Hewitt and Butlin poim out that the distribution of many organisms in north temperate regions has been influenced dramatically by recent ice ages. Colonization patterns lollowing the retreat 01 the ice sheets have been complex so that populations now close to each other in space may have distant geographical origins and diverged genomes (e.g. the grasshoppers in Fig. I S.2). Many populations are still evolving after their latest colonization and so traits may be changing under selection and, furthermore, may have different genetic bases in other parts of the species' distribution. Hewitt and Butlin provide some examples where restrided gene now may permit local adaptations. However, in other cases dispersal from neighbouring populations may prevent this. Daan and Tinbergen discuss a possible example from a study of great tits on the island of Vlieland in the Netherlands, where clutch izes are larger than is optimal probably because of immigration from mainland populations where larger clutches are selected lor. Environmental change, much of it caused by humans, is having profound effects on animal numbers and species diversity. Conservation efforts will be enhanced greally if we can predict what will happen to populations when a patch of forest is destroyed or a barrage is built across an estuary, for example. In the final chapter, Goss-Custard and Sutherland show how studies 01 individual decision making can help us 10 understand how populations will respond to changed environments. Population ecologists usually take various parameters as given, such as birth rate, death rate and dispersal. and then explore the consequences for population numbers and distribution.
PART 4: INTRODUCTION
309
Bl'havioural ecologists, on the other hand, are intt'rested in how these various parameters have evolved and how individuals vary their lire-history decisions adaptivdy in rdation to sdective pressures. Tlw advantage uf this additional knowledge is that it may then be possible to predict what will happen to populations when condit inns change and individuals vary their behaviour. Goss-Custard and Sutherland build on the idea of the 'ideal free distribution' to predict how individuals will distribute in relation to resource abundance and competition with others. They consider huw the stable distribution across habitat patches will be influenced by interference and unequal competitive ability. Then they discuss the problems of mea
Chapter 13 Adaptation of Life Histories Serge Daan & Joost M. Tinbergen
13.1 Introduction Life history is the distribution of major events over the lifetime of individuals. Life-history studies concern the timing and the intensity of reproduction. as well as the processes generating this temporal distribution. They analyse life span. age and size at maturity, the trade-o[fs between somatic growth, maintenance and repair versus reproduction, the decisions on number and size of the offspring. the investment in current offspring and in future reproductive attempts. Life-history research aims to reveal why this temporal organization varies among species - between fruitnies which start breeding after 12 days of life and giant tortoises which wait until they are 30 years old. Also. it aims to understand variation among individuals in a species: why does one pair of great tits lay twice per year. and another pair in the same population only once? The 'why' here is the general evolutionary question: how has natural selection led to variation in adaptive strategies, both between and within species? Life-history theory proVides the functional framework to evaluate results and generate hypotheses. It aims to explain why evolution produced particular life histories out of innumerable possibilities, and why a diversity of strategies persists in a population even if they yield different reproductive outputs. Life-history theory is concerned with the consequences of temporal patterns for the fitness of the individual. It is formulated largely independent of the genetic and physiological processes causing these patterns. Its predictions are thereby fairly general and applicable to a wide array of organisms. Life-history theory is most advanced in the analysis of reproductive decisions. where the consequences of di[ferent strategies are 1110st readily related to fitness. For other aspects of Iile-history organization, such as ageing and sex ratio control. separate bodies of theory have been developed. Eventually these will all have to be integrated in life-history theory. Two elements are crucial in the process 01 natural selection and in the variation 01 phenotypes: environmental and genetic variability. In the present context, environmental variation is the more general. Even in the absence of genetic variation. Iile-history strategies may vary between members of a species as phenotypic adaptation to di[ferent environments. In the absence 3 11
312
CHAPTER 13
of environmental variation, different strategies may also stably persist due to genetic variation, but only under the specific conditions defined by evolutionarily stable strategies. We therefore choose to emphasize environmental adaptation rather than the genetic basis of the slrategies. In this chapter, we summarize the main elements of life-history theory, and illustrate these with recent experimental studies designed to test the hypotheses. We funher outline questions poorly addressed by current theoretical or empirical work. We start by clarifying our usage of several concepts which form the core of life- history theory.
13,2 Concepts in life-history theory 13.2.1 Traits A trait is any quantitative property of a liVing organism. Life history concentrates on traits such as age of firSI reprodudion, clutch size and sex ratio of the offspring.
13.2.2 Fitness Life-hislOry theory is based on the idea that the temporal dislribution of life events affecls the contribution of an individual animal 10 the gene pool in Ihe nexl generalion. This contribution represents Ihe (evolutionary) filness of that individual. At any age I, where a life-hislOry 'decision' is taken, the expeded rate of gene propagation by an individual over the rest of ils life is equal 0 Fisher's (1930) reprodudive value:
where Ix is the expected fecundity (zygotes produced) at age x (> I), and 'x is the probabilily of survival until age x. In the case of sexual reproduction V, has 10 be multiplied by O.S 10 lake Ihe contribution of male and female 10 each zygole into account. Clearly, the product I),. when inlegraled over the lifetime of an individual (from t = 0) represents the total produclion of copies of its genes. To oblain the rate of change in relative frequency which lhese individual copies represent, the product at each age has to be weighted by the general change in frequency for all individuals in lhe population. This weighting faclOr is A-X, and I., the innate rate of increase of the populalion, is found by solVing the Lotka-Euler equalion:
ADAPTATION OF LIFE HISTORIES
313
r.
In practice. and I. are often determined in population studies by measuring the number of offspring produced per age class x. and the survival from one age class to the next. The innate rate of increase is ofren implicitly assumed to be I. in which case reproductive value would equal the lifetime reproductive suaess (LRS) of an individual. Since populations on which empirical demographic parameters are based may well he stable in numbers, but yet have surplus emigration (A. > I) or immigration (A. < I), LRS gives only an approximation of fitness.
13.2,3 Trade-offs Since fitness is a complex measure, based on multiple components, a change in slrategy may have negative consequences for one component. and positive dfects on another. Such consequences determine a trade-off. e.g. between current and future reproduction. or belween Ihe number of offspring and their reproductive value. We use the concept of trade-offs in the evolutionary or functional sense. However, trade-offs have Iheir roots in the physiology of the organism. Due to physiological limitations, variation in the value of one trait which affects fitness may have direct consequences for the value of another trait. This would appear to lead to a Irade-off between two or more traits. but in fact is better considered as a constraint (see below) on the system.
13.2.4 Optimization By identifying trade-offs. we may hope to define models predicting optimal values for particular life-history traits; optimal at least under panicular conditions. These values are those which maximize fitness. A fitness maximum usually emerges as the consequence of a trade-off. We emphasize thaI optimization refers to the individual in ils particular situation. and that the trait value which leads to maximal fitness in a population need not be the one which is optimal for every individual.
13.2.5 Decision rules and reaction norms A reaction norm (Stearns & KoeHa. 1986) describes the variation in Irait values as a function of environment and/or condition. Decision rules refer to the mechanism of response to these conditions. In essence the two concepts are Ihe same. Reaction norms or decision rules are optimal if they maximize fitness for each environmental condition.
13.2.6 Constraints The use of the conSlraint concept is confounded in life-history theory, and Stearns (1992. pp. 16-18) teaches us that we should not expect biologists to
314
CHAPTER 13
Environmental value
Fig. 13.1 Graphical r('peesemation of some basic concepts in life-history theory. In the presence of environmental variation (here simplified as unidimensional on the abscissa) diferent trait values lead to different fitness contours. The dashed line is the optimal reaction norm connecting the optima for diHerem environments. The option set is the
collection of trait values possible in the species. A boundary of the option set forms a constraint; when the optimal trait value is outside the option set it constrains the optimum. The constrained optimal reaction norm is given by the solid line.
agree on a certain definition. We use the word constraints in the sense of the boundaries of the option set or parameter space, i.e. the extreme values a trait can assume. These boundaries are set, e.g. by physical restrictions. When discussing individual phenotypic plastiCity, we should be aware that individual decisions may in addition be constrained by the genetic make-up of the spedes and thus by historical limitations. The optimal solution is located either inside or outside of the option set. If it is outside, maximization (or minimization) of the trait may represent the optimal realizable solution. Figure l3.1 shows the connection between several of the concepts.
13.3 Growth and maturation The first major life-history problem an animal faces is when to start reproduction? Most animals are much larger as adults than the zygote they have grown from. Some time must elapse for the organism to grow, even with the simplest form of reproduction, which is cell division. There is always some mortality, however. The longer maturation is delayed, the lower the probability of reaching the reproductive stage. The dilemma is a clear example of a trade-off; fecundity often increases with size but there is a lower probability of surviving to a larger size. The expected duration of reproductive life is thus negatively as ociated with age of maturity. The theory on optimal age at maturity is basically due tu Gadgil and Bossert (1970). Their fundamental idea of a !rade-off between fecundity and survival was later expanded to provide quantitative predictions. Such predictions were. for instance, derived and tested for age and size at maturity in different fish
ADAPTATION OF LIFE HISTORIES
315
species by Rolf (1984). Stearns and colleagues went a step funher by developing predictions for intraspecific varialions. given different growth and mortality rates (Stearns & Crandall, 1981. 1984; Stearns & Koella. 1986). The resulting optimal reaction norms were based on maximization of the innate rate of increase A as a measure of fitness. Predicted reaction norms could be tested, for instance with experimental data on age and size of pupal eelosion in Drosophila (Gebhardt & Slearns, 1988). Kozlowski and colleagues (Kozlowski & Wiegert, 1986; Kozlowski, 1992) have provided a formulation of the theory based on the allocation of energy to growth and reproduction. Their models yield optimal Solulions by maximizing LRS. The direct link with energy allocation gives these models additional power in lerms of predictions of other life-hiStory pallerns. We therefore choose 10 illustrate the main line of reasoning using Kozlowski's approach. Imagine an organism at birth. II starts acquiring energy from the environmenl at a rate A (e.g. in wallS =joules per second) and spending energy for maintenance and acquisition at' rate R. The excess energy (P = A - R) can be used for growth and/or reproduction. II is assumed thai P initially increases with increasing body size (w). The larger an animal grows the more spare energy it will have 10 turn into reproduction; Ihe potential fecundity II w) increases as a function of age (x), as depicted in the backplane of Fig. 13.2. If there were no monality, il would always pay to keep growing. However, since there mUSI be some mortality, life expectancy at birth is finite. In Fig. 13.2, mortality is assumed to be independent of age. Expected duration of reproductive life decreases with increasing age of first reproduction (ex). This is in fact true no maller whelher mortality is dependent or independent of size. The prominent result of Kozlowski's and other allocation models is that it is oplimalto invest each spare joule of energy in growth as long as this increases expected fulure reproduction by more than I J, and otherwise each spare Fig. 13.2 Optimization of age
and size at maturity in Kozlowski's model. The curve in the groundplane relates the
probability
(/~)
to survive until
age x. The curve in the backplane relates the potential fecundity fr (0 the age of malUrity a (switch from growth to reproduction). Expected lifetime reproduction for three different ages at
maturity, aI' U z and Q y is represented by (he three solids. The switch from growth to
reproduction at age «2 yields a higher lifetime reproduclion than at «lor aI' Thus. ther~ is an intermediate optimal u,
316
CHAPTER 13
joule should be allocaled 10 current reproduction. After Ihe optimal age of maturity is reached there should be no funher growth, since the condition remains fulfilled. The necessary existence of a single switchpoint from growth 10 reproduction for a plausible assumption of diminishing returns was derived mathematically by Ziolko and Kozlowski (1983). A constant rate of reproduction afler the switch point means thattolal expected lifetime reproduction is represented by f(a))Ix' or by a volume in thef, t. I space. Three such volumes are shown in Fig. 13.2 for dilferem ages of maturity (a' a a,). Clearly. lhere is an intermediate age. al' where the volume is maximal.' Maturation al this age maximizes fimess. Since' there is no funher growlh beyond age al' the model simultaneously yields a prediction for Ihe optimal size of malllre animals. The general prediction that adult size remains constant after reaching maturity is violated by indeterminate growers, such as many fish, molluscs and other poikilotherms. These continue 10 grow after their first reproduction. However, the model so far did not take into account variation in environ-
mental conditions. In variable environments. for instance due to seasonal changes in energy expenditure (R) and energy intake (A), the fecundity curve of Fig. 13.2 is no longer monotonic. Instead. during the wimer after first reproduction. fecundity is low and. moreover, expected fUlllre reproduction gradually rises as the season of high mortality passes by. Under such conditions. LRS maximization predicts that animals revert 10 allocating energy 10 growth ralher than reproduction alter initial reproduction. Instead of continuous reproduction beyond age a, there is an alternation of reproductive and growing seasons. Kozlowski and Uchmanski (J 987) have shown thaI under such conditions the optimal solution is a gradual increase with age in the annual allocation of resOurces to reproduction and a decrease in growth rate. Indeed, they were able 10 predict growth and fecundity curves, e.g. for wild populations of ArClic charr (Fig. 13.3). These growlh curves can be closely fitted by the well-known BertaJanffy curve, l, = l_ (I - exp(-kx)), where l, is length al age x, and k is the growth coefficient. Kozlowski (1996) has recemly shown that such growth curves may be the general outcome of maximizing LRS lhrough energy allocation.
• •
60
E .!!
~ 0
-'1 >-
'C
0 lD
50
•
40
••
••
30
Fig. 13.3 Body mass as a function of age in female Arctic
20 -
charr Salve/inlls a/pinus in
10
Labrador. (Symbols: data rrom Dempson & Green. 1985; line:
0
0
2
4
6
8
10
Age (years)
12
14
16
18
prediClion from the resource allocation model of Kozlowski & Uchmanski. 1987.)
ADAPTATION OF LIFE HISTORIES
317
The resource allocation approach to body size still has significam tasks waiting ahead. It hould evenlually be able to predict which drcumstances should lead 10 determinate growth. which to indeterminate growth and which to pronounced (seasonal) fluctuations in body mass such as observed in many small rodents (Heldmaier, 1989). It should further lead to precise predictions for some species where the necessary field data on size dependence of energy allocation and mortality can be obtained. Such information is now coming within reach with the developmem of isotope techniques for the assessment of energy turnover in nature (Nagy, 1980).
13.4 Scaling of time and energy Many aspects 01 life history are known to scale in a systematic fashion with body mass between species (Fig. 13.4). Traditionally, explanations for such panerns have been sought in a physiological dependence 01 time and/or energy constraints on body size. Such explanations face particular problems, such as the difference of scaling factors when individuals within species or spedes within higher taxa are compared (Heusner, 1987), or the large variance around the regression lines. Above, we have seen that body size itself may be optimized dynamically, along with measures of time (age of maturity) and energy (fecundity). In this section we discuss to which insights the optimi7.ation approach leads with respect to problems of scaling (Kozlowski & Weiner, 1996). We address the question of how random variations in rates of mortality (m) and in rates of reproduaion (n, as may occur between species, will affect optimal size.
Century 10'
Year
10'
Month
Day Hour
Minute
10 3
Second 10-5
10'
10'
10' Body mass (gJ
10'
10'
Fig. 13.4 Allometry of the 'rate of living'. Interspedfic regressions of hean beat lime. breath time. gut beal time. sleep cycle. age at maturity and maximum life span on body mass in mammals. For age at maturity. the data points on which the regression is ba'ied are indicated. Numbers indicate the slope of the regressions (exponent of body mass). Note: (i) that the exponenls are similar. varying between 0,26 and 0,31: and (ii) Ihat the variation around the regressions is in the order of one log unit. (Modified from Daa n & Aschoff, t 982.)
318
CHAPTER 13
In lhe model of Kozlowski and Weiner (1996) both respiration and assimilation rates are assumed to be power functions of body mass for individuals within a species: A = awl', and R = '..Jl. For simplicity we assume that b = ~, so that production P (= A - R = dw/dt) also scales as a power function of mass: P = cW'; where e = a -'. We further assume that mortality (m = dPldw) is mass independent (Kozlowski and Weiner explore consequences of different assodations belween mass and mortality). We keep parameter b, the exponent 01 the intraindividual power function, constant at an arbitrary value of 0.5. Different spedes may have slightly dilferent parameters e and m, which can be interpreted as different rates of food intake or energy expenditure (el and of mortality (m). To simulate such variations, we take for 50 species random values of e and m from normal distributions around mean m = 0.0002, and mean e = 0.015. These means were again arbitrarily chosen. Their particular value does affect the mean size and mean life-history traits. We shall see later what a change in the mean values brings about. For each combination of e and m, the switch criterion (wa = (belm) I/O -') is calculated. This yields the optimal adult size w a' as well as the age of maturity lX = wan-b'le( I - b), and production P = cwa'. Figure 13.5 shows the results of three such simulations in double logarithmic plots of P (energy) and lX (time) against mass (w). In the (jrst simulation (filled squares), the coe!licient of variation (cvm) of m was set at 0.2, while cv, = O. All the solutions are distributed on a straight line with slope 0.5, both for log P and for log lX. This can easily be understood: since e does not vary between species, they all follow the same growth curve, determined by b = 0.5, hence the switch points mUSI be on this curve. [n the second simulation (filled circles), cv, was set at 0.2, while cv•• = O. Now the slopes of the distributions are changed: age at maturity is the same for all 'spedes' (slope 0), since it is solely determined by m. Production rate (log P) increases proportionally with mass (slope 1.0), i.e. rwice as steep as the intraspedfic curve (slope b = 0.5). The highest values of e (largest difference between energy intake and energy expenditure) lead to the largest optimal adult sizes. In the third simulalion, e and m were both allowed to vary independently, with cV,= cv m = 0.2. In this case, the data (circles) show more scatter on all three axes of mass, production and maturation age. The regressions of log P and log lX on log ware now intermediate between the previous regressions (where either e or m was constant): log P = -2.76 + 0.80 log w (,-'=0.90) and loglX=3.07+0.2010gw (,-'=0.36). The associations of production P and age al maturity lX with mass reflect allometries of energy turnover (power) and time in general. The length of different intervals in the lives of birds (Lindstedt & Calder, 1981) and mammals (see Fig. 13.4), such as heart beats, gestation and life span, all scale with exponents similar to age of maturity. Predictions of age at maturity thus have implications also for patterns on other time scales. Keeping this in mind, we should make three general points about the results of Fig. 13.5. I Environmental variation in e leads to a steeper allometric relationship
ADAPTATlON OF LIFE HISTORIES (01
0.4
319
Energv
0.2 0.0
c .2 ;;
·0.2
C-
-0.4
, 'e C>
....0
-0.6 -0.8 ·1.0 '--'----'------'----'----'---'
2.4
Ibl
4.4
2.6
2_8
3.0 3.2 Log mass
3.4
3.6
3.8
Time
Fig. 13.5 Production (3) and age
4.2 .~ ~
;;
E ;; ~
0>
of maturity (b) ploued againS! body mass for 50 'species' in a
4.0
resource allocation model (simplified from Kozlowski & Weiner. 1996). with (and m
3.8
drawn from normal distributions. (Mean m = 0.0002; mean c= 0.015 Ihroughout.) Squart's
3.6
~
0>
....0
3.4
(c constant. m varying). coefficient of variation of c
3.2
=
3.0 '--'--'----'----'----'----'---' 2.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8 Log mass
=
lev,) 0; cV m 0.2. DolS Ie varying. m COnSI3nl). cv( = 0.2; (v', = O. Circles (c varying. m varying), cv( = 0.2; cv., = 0.2.
belween produclion (energy intake - expendilure) and mass Ihan there is along Ihe individual growlh curve. This provides a general explanation [or the fact Ihal the scaling of melabolic rates between species is typically steeper Ihan within species (e.g. Heusner, 1987). 2 In all Ihree simulalions, the slopes of Ihe regressions of log P and log a on log w add up 10 I, which means that Pa is proportional to WI, or Pa/w is a constant. This result. noted first by Kozlowski and Wiegert (1987), is contingent on Ihe basic assumption in the model that Wo = 0, or Ihe calculalion of oplimal size stans al size O. Pa/w has been considered a 'life-hislory invariant' (Charnov, 1993), and has the dimensions power x lime/mass, or energy/mass. Its size independence among animals, predicled by Kozlowski's theory, possibly provides a general explanation for the facI, fir I nOled by Pearl (1928) in mammals, Ihal Ihe tolal mass-specific melabolism integra led over Ihe lifetime (or over a heart beat) is weight independent; per unit of tissue, the same number of joules are consumed in a lifetime from mouse to elephant.
320
CHAPTER 13
Speculations on the nature of this phenomenon have concentrated on specific mechanisms relating metabolism to the process of ageing and senescence (see references in Rose. 1991). The size independence of the relationship is. however. a consequence of evolutionary theory without specification of any particular biological mechanism. 3 Due to independent variance in both production and mortality. the slopes are intermediate between those generated by each source of variance separately. Since the maximum value of b must be I. this means that the observed interspecific slope of energy metabolism on mass should be less than I. or. in other words. larger animals should have lower mass-specific metabolic rates than smaller animals. We now expand our exploration of the Kozlowski-Weiner approach by introducing two other groups of 50 imaginary ·species·. These are characterized by the same cv, = cvm = 0.2 as in the third simulation of Fig. 13.5. but one group (black circles in Fig. 13.6) has a higher mean ( (= 0.03). the other (triangles in Fig. 13.6) a higher mortality (mean m = 0.0004). Increasing mean (causes the whole distribution to shift 10 larger optimal body mass. The average production for the group is increased (Fig. 13.6a. black circles). In contras!. the average age at maturity has remained the same (Fig. 13.6b). Within the group. the regressions have retained the same slopes: log P = -2.61 + 0.79 logw ('=0.91) and !OglX=2.91 +0.21 logw ('=0.44). There is a range of overlapping body masses between the two groups with mean (= 0.015 and c = 0.03. In this mass range the 'species' from the lalter group have higher production rates. and lower lX. This result may be interpreted in various ways. The (= 0.03 group may represent. for example. a phylogenetically or ecologically coherent group of birds adapted to exploit a high-protein energy source. such as shorebirds which have typically high-energy turnover for their mass (Kersten & Piersma. 1986). In yet another group. mortality m is drawn from a distribution around m = 0.0004 instead of 0.0002. In this case. we find a reduction in the mean age at maturity. as well as in the adult production rates (Fig. 13.6 triangles). Again. the same regression slopes are found: log P = -2.52 + 0.78 log w (' = 0.92) and Jog lX = 2.82 + 0.22 log w (' = 0.51). Again. there is a range of overlap in body mass with the m = 0.0002 group. In this range the species from the highmortality group have increased production rates and decreased ages of maturity. In this case. we may anribute the increased mortality in the group. for instance. to the increased degree of seasonality in the temperate zone compared to tropical areas (Curio. 1989). Thus. life-history theory readily gives an explanation for increased r>roductivity and metabolism of similarly sized animals in temperate compared to tropical zones of the globe. Such global clines are well known for mammalian Iiller size (Lord. 1960) and avian clutch size (Klomp, 1970) and metabolic rates (Weathers, 1980; Ellis. 1984). Increased r>roductivity for a given body mass away from the equator is further rellected in an increase in the size of the metabolic machinery such as found in heart, liver and kidneys (Rensch
ADAPTATION OF LIFE HISTORIES 10)
1.2
321
Energy
0.8
•
0.4 c
• •
.2 1l 0.0
, e
'0
~ -0.4
0
-'
-0.8 0
·1.2 ·1.6 1.6
2.0
2.4
(b) 5.2
2.8 3.2 log mass
3.6
4.0
4.4
Time
4.8
.~
3
.. ~
E ~
' ~
4.4
Fig. 13.6 Production (al and age 4.0
of maturity (b) ploued against body mass for 50 'species' in a resource allocation model lev, = 0.2; (v' m =0.2 throughout). Open symbols: same simulations as circles in Fig. 13.4 (mean m = 0.0002; mean ( = 0.015). Dots: mean
0
3.6
0>
0
-'
3.2 2.8 2.4 1.6
2.0
2.4
2.8
3.2
Log mass
3.6
4.0
4.4
m = 0.0002; mean c = 0.030. Thiangles: mean m = 0.0004: mean c = 0.015.
& Rensch, 1956; Daan et al., 1990b). This illustrates how deeply into morphology and physiology the life history of animals eventually is reflected. The general importance of the approach is that it removes the problem of scaling lrom the realm 01 physiological constraints to that of optimization. For over one century theoreticians have concentrated on the allometric exponents of the relationships with body mass. Much of this work was based on the inwition that these exponents provide a set of physiological 'laws of nature', of constraints within which organisms are forced to operate. Rubner (1893) postulated that energy metabolism necessarily scales with body mass to the power 2 !> because of the su,face : volume ratio. When more data became available, this exponent wmed out to be closer to 'I. (Kleiber, 1947), and this led to other more complex theories of dimensionality (e.g. Stahl, 1962; McMahon, 1973; Heusner, 1987). These are all based on the assumption of
322
CHAPTER 13
constraints restricting the design of animal physiology. They all treat body mass as the independent variable on ;'hich all the derived variables are dependent. The resource allocation theory suggests that body mass itseJ[ may be dynamically optimized along with other biological variables. It also makes clear that interspecific allometric relationships can vary a great deal between groups of animals, depending on the variances and mean values of production and mortality rales, even with allocation models simplified to such a degree as the ones used here. Physiological 'laws' predict neither that species may deviate by a factor of 10 from allometric prediction nor the fact that the allometric exponent itself varies with the taxonomic level of analysis (Elgar & Harvey, 1987; Bennett & Harvey, 1988).
13.5 Parental effort and investment Williams (1966) advanced the seminal idea that some of the effort animal parents dedicate to the production and growth of their offspring is at the expense of later reproduction. This cost o/reproduction renects a trade-off between current and future reproduction and is a key concept in life-history theory. The reduction in future reproduction which is the result of the current effort has been termed parental investment by Trivers (1972). Neither Williams nor Trivers had strong evidence for a fitness cost of reproduction, let alone for particular mechanisms generating such a cost. Their hypothesis led to a new field of theoretical and experimental enquiry. One powerful strategy has been to manipulate parental effort by modifying the family size. Reduction of the effort is usually possible by taking dependent offspring away from the family. Increasing the effort has at least proved feasible in some altricial birds thal accept extra young in the nest as their own. Three studies are now available in which lhe parental effort has been assessed in terms ofthe mass-specific daily energy expenditure (DEE; Fig. 13.7). In all three, DEE/mass increosed with the experimental brood size raised.
7 6 5
---------+ Parus major.n .38) ...------ ..
----40
;:----
0. ;. • ~ 3 ~
w
~
PassercuJus sandwichensis (n-(OJ
Fig. 13.7 Effects of brood sizt' manipulations on mass-specific ~ Falcotinnunculus(n.25)
2
1 Reduced Control Enlarged
DEE in three species of birds: Ihe
savanna sparrow. Passermlus sand',;c/rms;s (Williams. 1987), kesl reI, Falco li',runt1ilus (O«renberg' 01.. 1995) and greal til. Parus major (J. M.
Tinbt·rgen. in Ilrt'paraliun).
ADAPTATION OF LIFE HISTORIES
323
However. the increase may be either due to mass loss (great tit: J. M. Tinbergen. in preparation) or to increased DEE (kestrel: Deerenberg et 01.• 1995). Hence. We should not take it for granted that parents in all species necessarily respond in the same way to brood size manipulations. or that such experiments always test Williams' theory. The measured consequences of brood size manipulations in birds for different fitness components have been summarized by Dijkstra et 01. (1990) (see also Stearns, 1992). Of 14 studies analysing the fate of parents after brood manipulations five have reported significant reductions in local survival, and eight out of 14 reported reductions in later fecundity (e.g. Gustafsson & Sutherland, 1988). Reduced local survival may be due to increased mortality or increased emigration from the study area. To distinguish between these possibilities, we have recently studied the time of death among 39 kestrels that had raised manipulated broods and were subsequently reported dead by the general public. mostly away from the study area. The results demonstrate that monality in winter is sharply increased in birds that had extra nestlings to raise in the previous summer (Daan et 01., 1996). The rates of survival and future reproduction of parent birds raising reduced, control or
enlarged broods are integrated in the reproductive value (Vp) in Fig. 13.8. The curve, approximated by an exponel1lial function as proposed by Kacelnik (1989), suggests that there is a non-linear reduction in Vp with increased parental effon. These data lend quantitative suppon to Williams' original proposition, but they do not identify the causal chain of events between parental effon and mortality. The time of death. half a year after the experiments, is inconsistent with any theory invoking enhanced predation due to increased effons. It rather points in the direction of physiological impairment. Recent research on the cost of reproduction focuses on enhanced parasite infection and susceptibility to disease. Indeed, in great tits, Porus major. raising enlarged broods increased
c:
.. ?
~
Reduced
~
10
> ~ > ~
8
'C
6
~
4
.. ~
.,
~
.'a: 'C
Vp_l0.46 _ eCl.23F.3.671
12
Control
Enlarged
2
~
0
0
2
3
4
5
6
Parental effort: flight hours/day IF)
Fig. 13.8 Kestrel residual reproduoive value plotted as a function of their parental ('Hon (hours of flight per doy). (From Doon tI al., t 990a.)
324
CHAPTER 13
rates of infection - with malaria (Richner fI al.. 1995) and haematozoans (Norrise/ al.. 1994) - have been found. Such effects may be caused by increased rates of comaCl with parasites or by reduced effeCliveness of the immune system. Studies by Deerenberg (1996) measure immunocompetence by challenging Ihe immune system with a standard antigen - sheep red blood cells - and assessing amibody titres se'Cral days later. This approach has so far been restricted to laboratory ilUations. Zebra finches. Taeniopygia gulla/a. show increased breeding imervals in response 10 brood size enlargements (Deerenberg. 1996) and reduced reproduClive success if the efficiency of food acqUisition is reduced (Lemon. 1993). Similar manipulations of brood size and daily work schedules reduce the formation of amibodies against the sheep red blood cells (Deerenberg. (996). This is a promising direClion of research. which may lead to the specification of (physiological) allocation of energy to reproductive effort versus repair and maimenance; and of the corresponding trade-off between currem offspring and expected future reproduction. We are dealing here more generally with plasticity of energy allocation. The allocation models assumed thaI assimilation A and respiration Rare inflexibly delermined by body size. But. there is evidence that both may be fleXible. A. the total energy collecled. may be increased during reproduClion. as we know from the brood manipulation slUdies. R. or the DEE. is panly channelled into foraging aClivities. and this pan cannot be reduced without incurring a penalty in a reduClion of A. The remainder is associated with maintenance and repair processes. including the immune system, and this pan may well be more fleXible. Resting metabolism can. for instance. be reduced in response to low planes of nutrition (Daan el 01., 1989b). and 10 increased work rates in daytime (Deerenberg. 1996), and increased by low-temperature acclimatization (Konarzewski & Diamond, 1994). Such flexibility may be the key 10 the question why reproduClion is nO! cominuous after maturation as predicted by the allocation models.
13.6 Seasonal timing David Lack was the first 10 discuss seasonal tinling in a life-history context. Lack (1950) proposed that breeding seasons in birds have evolved in such a way that, on average. the peak demand of the nestlings coincides with the peak in food supply. In this manner the average bird would raise the largest number of offspring. Variations around the optimal time would be retained in the population as a balanced polymorphism by year-to-year variations in Ihe liming of food supply. Population slUdies inspired by Lack's ideas later showed that the earliest breeders are typically Ihe most produclive, both in terms of offspring produced and recruits surviving (Klomp, 1970). Thus, the average bird in the population appeared to behave suboptimally. The patlern of seasonal clmch size variation can be reconciled by the concept of individual optimization (Hoegstedt. 1980; Drem & Daan, 1980). Declining prospects of offspring with
ADAPTATION OF LIFE HISTORIES
325
progressive date of birth lead 10 optimal breeding times preceding the lime when the maximum number of young can be raised. The trade-off between raising fewer superior offspring early in the year and more inferior offspring late in the season would determine the optimal time. Optimal limes would vary between individuals, from early in rich territories 10 late in poor territories. Drent and Daan (1980) and Rowe et al. (1994) saw female body condition as a proximate constraint on number of eggs produced, and its vernal increase as the cause of the seasonal increase in the maximum raisable number of offspring. However, the trade·off causes a seasonal decline in optimal clutch size-date combinations regardless of the phase (laying eggs or raising nestlings) at which constraints operate (Daan et al., 1989a). This can be illustrated using the relationship between parental effon and residual reproduClive value (Vp) in Fig. 13.8. Let us consider first a pair of birds raising a single nestling. Parental effon for one offspring is minimal at the time when food supply is maximal. If we represent the seasonal environment by a sine wave in foraging yield (y), then parental effort will be minimal (and hence Vp maximal) when it coincides with the time of maximum food supply (Fig. 13.9). This reflects Lack's original argument. However, as we have seen, the earliest born young often have the best prospects to become recruits. The most general reason is probably that by their earlier birth they gain an advantage over their year class in competition for food, territories, mates, etc. The reproductive value of the young (Vo) therefore declines with date of binh. Vo has to be added 10 the residual reproductive value of the parent (Vp )' and the sum Vis maximized (circles in Fig. 13.9) at a date preceding the 'Lack date'. In Fig. 13.JO(a,b) we have further introduced differentiation between habitats by varying the amplitude of the seasonal yield (y) curve. Maximizing Vp in all cases leads, of course, to reproduction at the seasonal peak (Fig. 13. lOa). Maximizing Vleads to earlier optimal dates in the richer habitats (Fig. 13.1 Db). We have extended the same argument to broods of more than one offspring in a formal optimization model (5. Daan et at., in preparation). In Fig. 13.1 Olc,d) the optimal c1utch-date combinalions are calculated, again for three levels of lZ
v
10
Fig. 13.9 Seasonal timing:
Vp
(sinusoidal) seasonal variation in
food supply (yieldlleads to
z
Vo ,,
,
,
,
,,
Yield
100
,; Effort
,
,
50
...
150
200
DAte
250
(the 'Lack date'). A seasonal
, .. . 300
minimal parental eHon and maximal residual reproductive value (VJI) at the yield maximum
350
dedine in offspring reproductive value Vo leads to an optimal date where the slopes of Vp and VI' cancel each other (circles), Le. before the yield maximum.
326
CHAPTER 13
(al
~
.~
-'
,
~
U
(bl
Vo constant
25 20
20
15
15
10
10
5
5
00
50
100
150
200
Vo declining
2 25
250
300
350
Single
egg
0
100
(e)
(d)
25
2 25
20
20
2
150
'. 250
200
300
350
0
2
'. V~''
'.
15
Multiple
eggs
10 5 '.
50
100
150
200
250
300
200
'.
250
300
Date
Fig. B.10 Optimal date and clutch size combinations obtained from the model in Fig. 13.9. when there is variation betwt:en territories in environmental quality (three amplitudes of seasonal yield curves; solid lines): (a.b) For single egg clutches; (c.d) variable clutch size. (a.c) For constant V~ (b.d) for seasonally declining Yo' The solid dots give the solution for Ihe richest. the open symbols for the poorest territory. with shaded symbols intermediate. In (a) the three solutions are identical.
food supply (territory quality). The optimal clutch size now increases wi!h increase in y. The optimal date is again the 'Lack date' if seasonal change in Vo is not taken into account (Fig. I 3.lOc). II is advanced relative to the peak food supply when Vo declines with progressive date of birth (Fig. 13.IOd). The important point here is that the advance is greater in the richer territories. Toge!her, these effeds cause a seasonal pattern of declining clutch sizes in Ihe population, without invoking any particular constraint such as female condition. Thus, the variation in brood size and timing of breeding within populations can, in principle, be explained on the basis of individual optimization. The optimal combinations have been worked OUI for the kestrel (Daan etal., 1990a; Lessells, 1991), and were found to be in reasonable harmony with observations. In this species, the seasonal decline in Vo was based on correlative ra!her !han experimental data, and, hence, we cannot be sure that for an individual Vo would indeed change along this same line with date as it does between individuals. This particular weakness of the analysis was resolved in a recent study of the European COOl, Fll/ica atra (Brinkhof. 1995). Brinkhof studied the life-history consequences of a change in the date of
0 350
ADAPTATION OF LIFE HISTORIES
327
laying by exchanging whole clutches of eggs which dillered by 10 days in laying date. By this manipulation he either advanced or delayed lor the parents the date at which their brood halched. The experiment is suitable to resolve two different issues (Fig. 13.11). 1 A seasonal trend in any component 01 litness may be caused by a general environmental variable assodated with date and aaing on aU birds in the same manner (date hypotheses). Alternatively, such a trend may be caused by dillerences in quality 01 the birds or their habitat which simultaneously also aflea the date of breeding (quality hYPolheses). When parent birds are conlronled with earlier or later broods lhan they chose themselves, the same trend in the fitness component as observed in the population should be found under the date hypothesis, but deviations Irom this trend under the quality hypothesis. This issue is important lor the derivation of optimal solutions for individuals. 2 If individuals optimally tune lhe date of breeding to their Own circumSlances, advance and delay manipulations should both lead to a depression in litness (Fig. 13.11). Note that this question of individual optimization cannot be answered for single components of litness but only for total reproductive value. Brinkhof followed post-hatching growth and survival of the young COOlS as well as survival of Ihe female parents of his experimental broods, and calculated the offspring and parental components of reproductive value. The results are as follows: for Ihe survival rate of hatchlings until 7 weeks of age (So) and the probability that parents produce a second clutch (P2)' the varialion in the populalion is apparently fully due to date ilself. Indeed, Brinkhof could identify the seasonal change in insea food as the cause of the variation in So' The seasonal variation in the survival of young coots until age 1 year Delay
Advance Date
;'
5
o:
'l<
'v
Quality
''/ e_
Individual optimal
date
Quality
/
Date or individual optimal
•
date Date
Fig. 13.11 Expected resulls of manipulations of date on fitness components under different hypotheses. Date hypothesis: advanced and delayed broods are expected 10 follow the seasonallrend observed in the population. Quality hypothesis: advanced and delayed broods are not expected to differ from controls. Individual optimization hypotheSiS (only for total reproductive value): advanced and delayed broods are expected to have lower ritness Lhan conLrols. (From Brinkhor. 1995.)
328
CHAPTER 13
(L I ) was attributable to quality (of the step-parents!). not to date variation.
Finally, the survival of the female parent (Lpl )' which did not vary with date in the population, remained unaffected in parents raising a delayed brood. However, it was sharply reduced compared to controls in parents raising an advanced brood. The cause of this increased mortality remained unidentified. It is possibly associated with increased rates of parental effort due to the fact that advanced dates caused more young to survive the early stage of post-hatching development. Be this as it may, on th~ basis of these analyses Brinkhof was able to reconstruct the final picture for total reproductive value V (Fig. 13.12). This shows Ihat. except for lhe very earliest birds in the population, V was negatively affected by both advanced breeding dates (due to reduced parental survival) and delayed dates (due to reductions in hatchling survival and the probability of second broods). The study - unique in its completeness - thus strongly supports the notion thaL variation between (al
~
..
35 30
,,--'- '
~
;;; > ~ >
25
~
20
~
15
{2
10
'fl
'ea. ei
-----=.=
.~
5 70
90
110
130
150
130
150
laying date (b)
~ ~
;;; > ~
35 30 25
.~
u ~
'ea.
20
~
15
{2
10
ei
5 70
90
110 laying date
Fig. 13.12 Estimates of individual reproductive values of cont Fu/ica arra parents plus brood as a function of laying date in the natural population (open circles) and following experimental IO-day shifts (arrows) of the hatching date of the brood (filled circles). Both
delay shifls (a) and advance shifts (b) reduce fitness for mosl laying dates, favouring the individual date optimization hypothesis. (From Brinkhof. 1995.)
ADAPTATION OF LIFE HISTORIES
329
individuals in the timing of breeding reflects variation between individuals in the optimal solutionc. At each date, the birds staning a dutch would have done worse in terms of reproductive value if they had chosen an earlier or a later date. The most imponant features of seasonal timing from a life-history perspective are thus twofold. 1 Optimal timing may often be based on the core life-history trade-off between current and future reproduction; in this case between reduced parental fitness (due to increased effort) and increased offspring fitness at earlier dates. This trade-off explains why most animals concentrate reproduction in spring, at the beginning rather than the height of the good season. 2 The general seasonal decline in clutch and litler size is best understood on the basis of differentiation between individuals in optimal timing.
13.7 Offspring size and numbers Another fundamental trade-off in life-history theory is that between the number of independent offspring produced per attempt and their size. Generally, the prospects for each offspring will be positively associated with size, but size is negatively associated with the number of offspring over which each unit of parental investment is distributed. The life-history problem is to find the number of offspring per reproductive event which maximizes the total reproductive value for the parent. Size at independence is determined both by the energy stores allocated to the eggs, and - in species with parental care -the energy invested in raising and protecting the brood. The simplest situation is offered by species without parental care. We illustrate the experimental analysis of the size versus number trade-off with a study on eggs of the side-blotched lizard Ula stansburiana by Sinervo el al. (1992).
0.6 f.
0.5
f.. '.
0.4
Fig. B.B Number of eggs per
C
>
,
.~
t.
0.3
en
0.2
(J:J. survival of hatchlings
(I..> and the fitted function through their produetJic- as an
estimator of fitness. in a population of eggs of manipulated size of the lizard
O. ,
0.0
clutch
U. stansbur;ana. Bars show the 0.2
0.4
0.5 Egg mass (g)
0.6
0.7
natural variation in egg mas'. (From Sinervo ttal., 1992.)
330
CHAPTER 13
Sinervo et al. (1992) collected near-term gravid females of this small iguanid lizard in the inner coastal range of California. They measured egg sizes and established that there is generally a negative association between the number of eggs laid by a lizard and their average size. This is represented in Fig. J 3.13 by the linear regression of clutch size U,) on egg mass in a sample from one study site (1989, Del Puerto, late clutches). The authors further experimentally assessed the consequences of egg size for post-hatching survival. The experiments made use of two techniques: 1 egg size was reduced after laying by extracting some of the egg yolk with a syringe; 2 in some of the growing follicles in early vitellogenic females the authors stimulated extra yolk deposition by removing yolk surgically from other follicles. After laying, eggs were incubated under standard conditions and the hatchlings released in the study area. One month aher release the study area was thoroughly searched for survivors. The fraction I, of surviving offspring, ploned in Fig. 13.13, was usually positively associated with egg mass, although in some samples a decline was observed for the largest hatchlings. The product f/. is an important component of the fitness of the offspring, and indeed may have an over-riding effect on total fitness. This component showed a maximum at intermediate egg mass in seven out of eight repeats (two locations, 2 years, two halves of the breeding season) of the experiment. A close match was found in most cases between observed and predicted mean egg mass. This ingenious study did not relale Ihe consequences of the experimental modification 10 the original unmanipulated egg mass. Testing whether interindividual variations in a trait reflect variations in the optimal value for this trait would require the comparison of manipulated with control groups. Only by experimental manipulation can we be sure that the Iitness consequences for individual alternatives are measured. Such experiments are often diflicult, and assessing the full consequences for fitness is tedious, but without them it is not possible to evaluate whether Iife-hislOry strategies represent optimal solutions. A study where Ihe difference between descriptive and experimental analysis has been worked out is that of the decision on the ntlmber rather than the size of eggs in the great tit, PartlS major (Tinbergen & Daan, 1990; Verhulst, 1995). The analysis is based on a population of great tits breeding in nestboxes in a woodland 'Hoge Veluwe', which has been under study for nearly 50 years. Tinbergen and Daan (J 990) analysed the reproductive value of parents raising clutches of different size, as well as of their offspring. The results, summarized in Fig. 13.14(a), show that in the population both ~,and V.. are positively associated with clutch size. This is due 10 the fact that parent birds laying larger clutches have a higher probability of producing a second brood (hence V. is larger), and raise more fledglings. Over 5 years, brood size was experimentally manipulated in 341 nests by either taking 50% of the nestlings out
ADAPTATION OF LIFE HISTORIES 25
(a) Hoge Veluwe
331
Ie) Vlieland Natural variation
Natural variation
20 -
2
15
• 25
5
6
7
8
9 10 11 12 13 14 15
Ibl
20 15 10 5
3
4
5
6
7
8
9
10 '
12 13 14 15
Idl Artificial variation
v~
Artificial variation
2 .
VP~ <S:~::::8
vc~
oL-'--.l.-.l--'----..L---'-----'-----'----'---'---'----.J 3
oL-.l.-.J....L-Ll....L..1...J....J..LL..LLL-'='='-'---'-_
•
5
6
7
8
9 10 11 12 13 14 15
Clutch size
oL-.l-..L---'---'---'----'--J'--'-.l-..L---'--~ 3
•
5
6
7
8
9 10 11 12 13 1. 15
Clutch size
Fig. lJ.14 Natural and anificial variation in clutch size in two populations of great tits P.
major. (a) Mainland study sile the Hoge Veluwe. Lines show reproductive values of parent (V p) and dutch (V.). and their summation V as fined functions of natural clutch size. Bars: distribution of natural dutch sizes. (b) Anifidal reduction and enlargement of dUlches reduce fitness. (Modified from ltnbergen & Daan. 1990.) (c) Island population Vlieland. Open symbols: lifelime production of recruits by parems of different dutch sizes +/- one standard error of the mean. line: fined quadratic funaion. (d) Lifetime produClion of recruils for parents wilh conlrol (average 9 eggs) and reduced dUlche (average 4 eggs). Arlifidal reduClion of clutch size enhances fitness in this population. (From Verhulst. 1995.)
or adding 50%. Such experimental variation in brood size leads to a wholly different picture (Fig. 13.14b). Now, Vp declines with increasing brood size and V, shows a curvilinear relationship with brood size. On the basis of the descriptive analysis one might surmise that the dutch size associated with maximum fitness (V) is 15 nestlings. The experimental analysis shows that optimal brood size is in fact the control brood size (which for practical reasons was around the most common brood size), since both enlargements and reductions led to a reduction in fitness (V). Hence, clutches manipulated from nine to 13 represenllower fimess than natural clutches of 13. The implication is that natural variation in brood size must have fitted the variation in optima. This is another instance of individual optimization as we encountered earlier in the kestrel and COOl. More recently, Verhulst (1995) has applied the same approach to another population of great tits, living on the DUICh island of Vlieland. He used lifetime
332
CHAPTER 13
production of recruits rather than reproductive value as a fitness measure, and applied only brood reductions, not brood enlargements, as the experimental intervention, Interestingly, the Vlieland results deviate from those in the mainland population: in the descriptive analysis, a curvilinear relationship was found, with maximum fitness associated not with the maximal clutch size but with clutches of about eight eggs (Fig. 13.14c). In the experimental analysis, artificial reduction of brood size led to a significant enhancement of LRS (Fig. 13.14d). This means that the observed clutches on Vlieland were on average larger than is optimal. Parents in the long run would contribute more genes to the population if they would lay fewer eggs. This discrepancy shows clcarly that the method docs not necessarily lead 10 a fit between average and optimum, bUI that it is possible also to detect maladaptive features in life history. Verhulst (1995) has argued in this case that probably the tendency to lay too many eggs is maintained in the population by immigrants from the mainland. Such gene now would prevent the population from becoming fully adapled to the local environmental circumstances.
13.8 Population heterogeneity in life histories Variation of life-history paramelers within animal populations has stimulated the analysis of their adaptive significance. When there is such variation, we cannot directly assess whether this variation is either distributed along the same reaction norm for all genotypes in Ihe population (i.e. be solely based on phenotypic plasticity) or represents purely genetic diversity, or any mixture of environmental and genetic diversity. (n the early days of life-history studies it was more or less implicitly assumed that most of the variation is genelic. In Ihis vein, Lack's (1950) original propositions were that the average clutch size and the average date of laying would be the trailS maximizing fitness in stable populations, with stabilizing selection weeding out the extreme genotypes. Population studies in birds, mostly inspired by Lack's ideas, have not confirmed these predictions. In general. the earliest breeders, and Ihose wilh the largest c1ulches, contribute moslto the next generation. If these trends would reflect directional selection, the genotypes leading to late and small clutches would soon disappear from the evolutionary scene. Yet, population geneticists claimed, on the basis of mother-daughter correlations in reproductive traits, that a considerable pan of the variation in traits such as c1ulch size and laying date was of genetic origin (e.g. Van Noordwijk et al.. 1981). Current views on how environment interacts with genotypes leave more scope for environmental erfects on variation in life-history traits via phenotypic plasticity or reaction norms. Sucll variation may nOt be related exclusively to the environment of the adult animal but also to Ihe conditions during early ontogeny. Grafen (1988) has coined the term 'silver spoon effect' 10 indicate the lifetime benefits an animal may gain from growing up in rich nutritional conditions. Examples of
ADAPTATION OF LIFE HISTORIES
333
such dlects are now slowly accumulating in the literature (e.g. Huck et oJ.. 1987; TInbergen & Boerlijst. 1990). If proving general. they may help explain mOlher-<:laughter correia lions in reproduclive traits as well as lhe high degree of variation in lOlallifctime reproductive success. which is now eOlerging as a
general phenomenon in long-term population sludies (Newton. 1989). Be this as it may. a sizeable extent of environmemal variation. whether acting al the lime a reproductive decision is taken or early in ontogeny. implies that we
cannot find quantitative solutions of any optimization problem withoUI recourse to experimental manipulations of the traits. By their very nature such trailS are concerned with rare life events. and it will not usually be possible 10 use individuals as their own control in experiments. This and the fact that fitness COSIS and benefits have to be assessed in the nalUral environment makes it necessary to perform large numbers of experiments and renders analytical progress in this field slow. This reasoning should nol distract from lhe [act that some of the variance may be of genolypic origin. From the experimenls on the mainland great lits we have seen that the mean c1l1lch size closely corresponds with the mean oplimal solution. Yel. in the case of the Vlieland great tits. clutch sizes were larger than optimal. and such deviations may be attributed to genetic constraims imposed by gene flow from a SOl/ree populalion (Verhulst. 1995). Life-history research faces important challenges in unravelling - both theoretically and empirically -the degree to which genelic and environmental trait variation is found. With a strong unidirectional gene now from sources to sinks. we intuitively expect more directional selection in the sinks than in the sources. Following this reasoning. one may predict more genetic diversity in trailS slrongly affecling fitness to be maintained in sinks by the combinalion of directional selection and gene now. Furthermore. the degree of predictability of the temporal environment is bound to affeclthe contributions of genetic and environmental variance. In a perfectly predictable. but spatially heterogeneous environment. we expect that animals can individually optimize their decisions beuer. and the observed variation will largely reflecl environmental components. In a temporally unpredictable environment. variation may be more prone to containing a strong
component of balanced genetic polymorphism. The time-scale of these variations is necessarily relative to Ihe time-scale of individual lives. Evolution must have found many differem Solulions for adjusting the traits to whatever environmental predictor is available. The ubiquitous photoperiodic response is a physiological mechanism suppressing reproductive activity when the harsh winter season approaches for many short-lived animals; it triggers reproduction in those large mammals where winter gestation is required to synchronize parturition with the neXI springlime. The interrelationships between genetic and environmenlal variation. between physiological comrol mechanisms and their fitness consequences. and belween temporal charaaeristics of lhe environment and of a species' rate of living will be fascinating future foci of life-history research.
Chapter 14 The Phylogenetic Foundations of Behavioural Ecology Paul H. Harvey & Sean Nee
14.1 Introduction The vibrant synergism between studies of behavioural ecology and evolulionary biology is well iIlustrated by Ihe results of using phylogenetic Irees to help lackle problems in the two areas. Evolutionary biologists want to understand the reasons for the diversily of species. Phylogenetic trees describe that diversity, but it is becoming increasingly obvious that behavioural ecology can help explain why that diversity arose. In particular, spedes wilh particular behavioural ecologies are more likely to speciate so Ihal each of several bushy areas of phylogenetic Irees may contain species with similar life-styles. On the other hand, behavioural ecologisls want to undersland the intimate links berween behaviour and ecology. Mapping character states on to phylogenetic Irees proVides crilical insighls into how behaviour and ecology have shown correlated evolutionary change, or when new associations between the two have come aboul. This chapter is divided into four seclions. The first examines how species dilferences in behaviour and ecology can help explain differences in the structure of evolutionary trees. The second seClion asks how charaaer change can be mapped on to phylogenetic trees and used to test ideas aboUl how particular behavioural ecologies evolved and whether some Iypes of character evolve more rapidly than others. The third seclion moves from the consideration of single characters 10 correlations among characlers, and shows how, if we are to understand why characler covarialion across laxa occurs, phylogenetic tree structure must be an integral part of the analysis. Phylogenetically based comparative analysis has proved a powerful 1001 for generaling and tesring ideas about the links between behaviour and ecology. The fourth and final section integrates the first three by showing how, during Ihe course of evolution, speciation may be accompanied by prediclable changes in behavioural ecology. The consequence is that similar sequences of phenotypes may evolve independently al diIferent times and places. This chapter is constructed around iIlusrralive examples taken from Ihe recent lireralure. We should emphasize from the start an assumplion that is common 10 those examples, bur which may nOI be correa in some cases and wiIl almOSI certainly nor be rrue in others. When a slalisticallest is performed 334
PHYLOGENETIC FOUNDATIONS
335
in the various examples described, it is often assumed that the phylogenetic tree used for the analysis is perfectly correct (always in topology, and occasionally in branch length also). In fact, phylogenetic trees are working hypotheses generated from limited data. As the dataset improves, so does the likelihood that the working hypothesis will be correct (for consistent methods of tree construction see Huelsenbeck el al., J 996). We are lucky to be working at a time when there has been a revolution in the reconstruction of phylogenies which show the relationships among contemporary species (Hilliselal., 1996). This revolution results from technical advances (notably the polymerase chain reaction) wltich makes the sequencing of large sections of genetic material (deoxyribonucleic acid (DNA) and ribonucleic acid (RNA)) routine. Sequence data have added enormously 10 our ability to reconstruct more accurate phylogenies.
14.2 Tree structure 14,2, I Key innovations and cichlid fishes Occasionally, a phylogenetic lineage or clade branches at a higher rate than others. For example, cichlid fishes are extraordinarily speciose, as are phytophagous insects. If we are interested in why particular clades are very speciose we might attempt to identify some 'key innovation' common to members of the clade but which is not shared by its close relatives. 'Innovation' must be interpreted broadly as simply any biological feature of a clade which either promotes speciation or, for example, simply allows speciation to occur if the clade finds itself in a situation where competition is absent, leaVing many niches empty. Liem (1973, 1980) may have identified such an innovation. He pointed out that a small sltift in position of a single muscle aLLachment ultimately allowed the pharyngeal bones of cichlids to manipulate their prey items while still holding them. As a consequence, the pre-maxillary and mandibular jaws were freed to evolve along new routes that did not involve manipulating prey. This, he argues, allowed cichlids to evolve a whole new diversity of feeding mechanisms and there seems little doubt that the adaptive radiations of cichlids in African lakes, in the face of competition from other fish families, does result from their evolved diversity of feeding mechanisms. However, Lauder (1981) pointed out that while such explanations may well be correct, any other characteristic common to cichlid fishes but which differs from a sister group could, in principle, be the key innovation which resulted in high speciation rates. Conclusions are strengthened if several sister-group comparisons can be demonstrated where the same key innovation is associated with high species diversity (Heard & Hauser, 1995). Such comparisons control for the many differences between sister lineages because any particular dHference apart from the one under test is less likely to be partitioned in the same manner among different sister taxa. We now use two examples, one
336
CHAPTER 14
from insects and the other from birds, to show how such sister comparisons are performed and how they have been used in hypothesis testing.
14.2.2 Phytophagy and insect diversification Insect species that feed on vascular plants are termed phytophagous. Half the world's insect species are phytophagous, but they are restricted 10 only nine of the 30 orders of insects. Southwood (1973) imagined that there were considerable barriers to an insect group evolving phylOphagy. Behavioural and morphological adaptations would be required: (i) to reduce the risk of desiccation; (ii) 10 remain allached to the host; and (iii) to deal with lownutrient food. Once those problems were overcome, Strong et at. (1984) suggested that the diversification of phytophages would be promoted by the great diversity of plants species and diverse plant parts as well as the absence of competitors. As a test of the idea that phytophagy promotes insect diversification, Miller et al. (1988) were able to perform 13 sister taxon comparisons between the diversity of a phytophagous clade and its non-phytophagous sister clade. Since sister clades originated at the same time, they have had idemical times for diversification. In 11 of the 13 comparisons, there were more species in the phytophagous clade than in the sister group, and in each of those cases the difference was greater than twofold.
14,2.3 Mate choice and speciation in passerine birds It has been suggested frequently that sexual selection by female choice might
increase the rate of reproductive divergence between populations and thereby increase the speciation rate of a clade (e.g. Darwin, 1871; Lande, 1981; Dominey, 1984; Carson, 1986; Schluter & Price, 1993). Barraclough et al. (1995) tested this idea using passerine birds as a case study. It is generally accepted that mate choice is responsible for the evolution of sexual dichromatism in passerine birds (Andersson, 1994), and so Barraclough et al. used dichromatism as an indirect measure of the importance of mate choice for a species. With few exceptions, passerine species can be scored unambiguously as dichromatic or no!. On significantly mote than 50% of occasions, the clade with a higher proportion of sexually dichromatic species was more speciose than its sister taxon, thereby supporting the idea that mate choice can increase speciation rate.
14.3 Character change How and why do charaaers evolve? We can begin to answer these questions if we have: (i) a phylogenetic tree showing the relationships among a series of
contemporary species; (ii) character states for those species; and (iii) a model for character evolution.
PHYLOGENETIC FOUNDATIONS
337
14.3.1 Mapping character change on to the phylogenetic tree There are various ways of mapping character change on to the phylogenetic tree. and the one we choose will depend on the model of character evolUlion Ihat we have in mind. Maddison and Maddison (1992) give a clear exposition of the pranicalities associated with choosing and employing a model of charaner change. Although their discussion is given in pan with panicular reference to Iheir computer program MACCLADE. the major issues are well described. The basic 1001 is the 'slep matrix' which specifies how easy il is 10 go from each character state to any other. They discuss. for example. how a slep matrix can be derived from a matrix of probabilities of change between states. However. so lirtle is generally known about probabilities of character transitions lhat some sort of parsimony or minimum-evolution criterion is used for ancestral character-stale reconstrunion. If required. probabilities of changes can then. themselves. be estimated from the tree which provides ancestral character stales at the different nodes.
14.3.2 Tracing character evolution: the evolution of parental care It is arlen of interest to behavioural ecologists to trace behavioural change on
the phylogenetic Iree. For example. some models for the evolution of parental care assume that evolutionary transitions involving a near-simultaneous change
in parel1lal behaviour from care 10 no care (or vice versa) by both parents is extremely unlikely. and Girtleman's (1981) analysis of the phylogeny of parental care in fishes provides support forthat assumption (Fig. 14.1). Of 21 transitions. all were between biparental and maternal carc. biparental and paternal care. no care and maternal care. or no care and paternal care; none was between biparcnlal care and no care. or maternal care and paternal care. The most common Iransilions were from no carc to male care 10 biparental care to female
/
• ,
~
.---.---+ ,
~
t
/
Fig. 14.1 The directions of change..' in paremal (are in fishes. Four states are possible. The
solid arrowed lines are hypothesized transitional routes and the dashed lines. on the other
hand. du
nOI
accord with the hypothesis thai simultaneous change by bolh parents is
unlikely. All transilions ubserved
by Ginleman (1981) were in accord with the hYPOlhe~is.
338
CHAPTER 14
care. Subsequently and similarly, Carpenter (1989) tested West-Eberhard's (1979) model for the evolution of patterns of social behaviour in vespid wasps; in that example some transitions did not accord with expectation, although others did. A more recent and thorough analysis of evolutionary transitions returns to parental care, this time in shorebirds, which for birds show an astonishing diversity of incubation and brood-rearing patterns, ranging from pure maternal through pure paternal to biparental. There had been two schools of thought concerning the evolutionary transitions likely to be responsible for the diversity. The first is that biparental incubation was ancestral, and uniparental care followed depending upon the costs of care and the benefits of desertion to each parent (lennL 1974; Pitelka et at.. 1974; Emlen & Oring, 1977). The second suggestion involved the evolution of biparental care from male care. as males reduce their contribution (van Rhijn, 1985. 1990). To help distinguish between the two scenarios. Szekely and Reynolds (1995) used a phylogeny based on Sibley and Ahlquist's (1990) DNA-DNA hybridization data, adjusted to account for recent critidsms and improvements. (Fortunately, their results
seem robust to alternative likely phylogenies.) Parental care was claSSified for almost 50% of the 203 known species in two 'ays: (i) which parent. if either, is involved in most care; and (ii) what is the duration of parental care for each parent. Parsimony was used to determine ancestral character slates. although some branches and nodes had equivocal states (e.g. biparental and maternal care equally likely). The major results to emerge (Fig. 14.2) were that most
(a)
Full biparental
~
5-11 0
More male
6-8
Ib)
«
No male care
lei
~
Male only or biparental
I
•
I
More female
Partial
male care
0-2
Female only or biparental
~
9-10
2-2 0
I
~4
0-2
0-2
No female care
~
Panial female care
Fig. 14.2 The minimum and maximum number of evolutionary changes in pattl'rns of parental care in a sample or shorebirds according to Szekely and Reynotds (1995). la) The number of transitions belwt'en care types. (b.c) Transilions in duration of care within each s~x.
PHYLOGENETIC FOUNDATIONS
339
evolutionary transitions in the shorebirds had been towards a reduction in
paternal care, which was sometimes but not always compensated for by an increase in maternal care. Szekely and Reynolds (1995) conclude that their analysis provides more support for the idea that males have been reducing their parental support in comparison with females over evolutionary time. They argue that in shorebirds the COsls of caf(~ per clutch are so high that males never guard more than one clutch at once. At the same time, there is always the possibility for the male of extrapair fertilization or multiple pair bonds. Accordingly, they reason, there has been continued selection for reduction in male care, which leaves them the problem of why there are actually more contemporary species in which males provide more care than females. Presumably, one might answer, either the rate of speciation has been higher among lineages in which males provide more care than females or the rate of extinction has been higher in lineages in which females provide more care than males.
14,3.3 Rates of character evolution: behaviour evolves rapidly Given that we can trace character evolution on a phylogenetic tree, it should be a fairly straightforward exercise to examine rates of character change. The idea that some types of character are evolutionarily more labile than others is frequently assumed but rarely tested. In particular, Gittleman et al.'s (1996) review of the literature indicates a general belief 'that behavioural traits are more labile than morphological or physiological traits'. There are a variety of ways evolutionary rates can be measured (Gingerich, 1993), the more sophisticated of which incorporate the phenotypic standing variation of the character and generation time. In their attempt to test the idea that some traits have more rapid rates of evolution than others, Gittleman et al. (1996) could not use such sophisticated measures owing to paucity of data, and made do with Haldane's (1949) crudest measure, the 'darwin'. A rate of I darwin corresponds to a change of I logarithmic unit per million years. They examined variation in two behavioural characters (group size, home-range size), two life-history characters (gestation length, birth weight) and two morphological characters (adult brain weight, adult body weight) in each of eight mammalian taxa for which there were reasonable phylogenies. [n summary, the behavioural characters seem to have been more evolutionarily labile than the morphological characters, which were more labile than the life-history characters. Although Gittleman et al. (1996) wisely urge caution over their results, which must be regarded as preliminary, the analysis does point the way forward to the use of phylogenetiC information for determining which characters show more evolutionary inertia than others. As that information accumulates, general patterns may well emerge which will suggest why some characters evolve more rapidly than others.
340
C H A PT E R 14
The fact that some charactcrs are usually more evolutionarily labile than others, does not necessarily mean that rates of character evolution remain the same in different lineages. Garland (1992) has developed statistical tests designed to compare rates of character evolution in different clades. As an illustration of the method, Garland uses data on metacarpal: humerus ratio (an index of cursoriality), nwtatarsal : femur ratio, body mass, maximal sprint running speed and home-range size from 16 spedes of Carnivora and 27 species of ungulates whose phylogenetic relationships are reasonably well known. As expected (Janis, 1996), the limb proportions evolved more rapidly in thc ungulates than the Carnivora, but the other three variables seem to have evolved at very similar rates in the two groups. A comparison of overall rates of change of the various characters is also 'consistent with the lung-standing idea that behaviour generally evolves more rapidly' (Garland, 1992, p. 516) than other characters.
14.4 Correlated character change Behavioural ecologists are frequently interested in the reasons why some species behave differently or occupy different habitats from other species. One perfectly reasonable way to proceed is to search for correlates of the difference(s) we are interested in. However, two variables can be correlated because the first has influenced the second, because the second has influenced the first or because both have been influenced by an unknown third variable. Biologists are fortunate because many third, or confounding, variables can be eliminated effectively from their comparative analyses by careful use of phylogenetic information.
14.4.1 Why incorporate phylogeny? We know that closely related species are similar among themselves, and they differ in many ways from other less closely related species. The similarities among closely related species are often said to arise from 'phylogenetic constraint', but that can be a misleading term because of the different processes which may be involved. For example, new species are likely to invade niches very similar to those occupied by their immediate ancestors because those niches provide environments to which the species is well adapted; here, the 'constraint' is adaptive resulting from similar selective pressures. Alternatively,
there may not be genetic variance available for new characters to evolve, in which case the 'constraint' is genetic. These and other reasons for similarity to b,· retained by descent through common ancestry are discussed elsewhere (Harvey & Pagel. 1991). The evolutionary perspective provided by phylogenetic information allows comparative biologists to identify cases where change in one character or habit is accompal)ied by change in another. It is then possible to use independent evolutionary events to statistically test the
PH YLO G E N ETI C FOUN DA nON S
341
null hypothesis that two or more characters change independently of each other.
14.4.2 Types of character change There is a useful but somewhat artificial distinction to be made bel ween continuous and calegorical characters. Examples of continuously varying characters are body mass or amount of time spent foraging. while typical categorical characlers might be presence versus absence of eyes or type of habitat occupied (woodland. hedgerow. pool. stream). Continuous variables are. in facl. discrele variables where the widlh of lhe discrete units is vanishingly small. Nevertheless. different campara live methods have been developed lO deal with the differenl types of character. Whatever method is used. an important point to grasp is that it will incorporate a model of character change. For continuously varying characters. the mOSl usual model is that of Brownian mOlion or a random walk in which. for example. it is supposed that the average body size of a species changes continuously and randomly over time. An alternative. punctualiona!. model is that characters change only at speciation events (I.e. at nodes on the tree). In any evenl. our null model would be that any two characters we were interested in changed independently of each other through the tree. We shall consider the most widely used method of testing for correlated change in continuously varying character below. Before considering comparative methods per se. it is important to return to the point lhat characler change in either direction may not be equally likely (see Sections 14.3.1 and 14.3.2). For example. Dollo's law Slates lhat complex characters are more frequently lost in evolution than they are gained. The question is. how much more frequenlly? If we were dealing with the presence and absence of eyes and we had considerable variation among contemporary species and a well-resolved phylogeny, we might impose the restriction that eyes are homologous organs which were present in the common ancestor of the group in question and evolved just once: all evolutionary transitions would be towards the loss of eyes. Similarly, although less extreme. there appear to be many lineages in which size increases more frequently than il decreases through time (Cope's law), and we might wish to impose some differential likelihood of size increase versus size decrease on our characterstate reconstruction.
14.4.3 The method of independent contrasts For continuou Iy varying characters, Felsenslein (1985) developed the comparative method of independent contrasts. The key realizalion here is that differences in character Slates between each pair of sister taxa in a phylogeny evolved independently of differences between all other pairs of siSler laxa. Each node in a bifurcating phylogeny subtends two sister taxa. and so there
342
CHAPTER 14
are as many comparisons possible as there are nodes in the phylogeny. Each comparison is termed a contrast. For a bifurcating phylogeny, the number of
independent contrasts adds up to the number of species minus one. After estimating ancestral character stateS for each node, contrasts are calculated fnr each character being considered. The direction of comparison is arbitrary, although il is conventional to order the taxa so that all contrasts for the independent variable are positive (or zero). The contrasts for the dependent variable(s) then use the same order of taxa and can take both positive and negative values. The contrasts for the dependent variable(s) can then be compared with those for the independent variables, and convelllional statistical ll'chniques (correlation or regression) are used to test for correlated evolutionary change. The null hypothesis is that change is uncorrelated, and significant depanures from the null hypothesis can indicate either positive or negatively correlated character change (Box 14.1).
Box 14.1 The independent comparisons method for two characters in a single phylogeny Under a Brownian motion model of evolution, dL d2 and d3 provide independent comparisons. Path length differences are ignored in this illustration. Values of
7
9
r
Values of x
20
14
8
17
20
24
40
30
35
22
+-d2--+
y
+--d3--+
d1
10
~m .~m .~
m >
d2
6
4
d3
9
10 13
od3
~
o
odz
a.
E
°dt
0
>'u
o
x
2
15
x variable comparisons
PHYLOGENETIC FOUNDATIONS
343
Independent contrasts have been used to test fairly sophisticated ideas. For example. in 1927 the ecologist Charles Elton argued that birds feeding on small prey items are likely to provide less food for their offspring than species that select large but rare prey. A similar suggestion was subsequently made by Sutherland and Moss (1985). Saether (1994) found that. indeed. a smaller amount of food was brought back to Ihe nest in taxa of altricial birds feeding on small prey items. Saether reasoned thai if those species feeding on larger prey items could bring back more food. then they could either provision a larger dutch or fledge their young earlier. In either event. their reproductive success would be increased. However. species that feed on larger prey tend themselves to be larger bodied. oILen produce smaller clutches and need to provision larger offspring. As a consequence. Saether needed to control for the correlations wilh adult body size when examining the relationship between prey size and clulch size or incubation period. This was done by regressing the contrasts of prey size. clutch size and incubation period on the contrasts for adulL body weight and calculating the deviations from the regression line. Those deviations removed the correlations with body size and were then plotted against each other. In fact. it turned out that when altricial birds feed on relatively large prey they produce relatively large clutches (Fig. 14.3). but they do not have a more rapid nestling growth rate or earlier fledging time. Saether's (1994) analysis. like mOSl others in the recent literature. assumed that the different variables being considered evolved according to a Brownian motion model of evolution. Statistical conclusions can. in fact. change when different models of character change are entertained. For example. Harvey and Purvis (1991). report a study of thl' evoluLionary relationship between a measure of running activity and relative foreleg length (foreleg length corrected for body size) in a group of Allalis lizards. Three models of evolutionary
0.4
0.2 ~
';;
'5 u .~• 0
0.0
~
0;
a:
-0.2
-0.4
-2
•
.
• • •• , • ••• • • •• •
•
-1
0 Relative prey size
Fig. 14.3 Birds which. for their
2
3
budy size. feed their (hicks larger prey items lend to also have relatively larger clutches. (Adapted a1ler Saelher, 1994.)
344
CHAPTER 14
change were considered (Brownian motion. punctuation and minimum evolution), only the first of Vltich produced a significant relationship between contrasts for the two variables being studied (p < 0.025 compared with p > 0.10 for the other two models). While independent contrast methods are proving an extremely powerful tool for investigating correia led character change, apan from the need to consider alternative models of character evolution, it is important to bear in mind two other factors which sometimes lead to faulty or overstated conclusions. First. when comparing contrasts. linear regression models are usuaIly used and. because the expectation of no change in one character is associated with the expectation of no change in the other character, regressions are forced through the origin. Such procedures may hide important features of the data. For example, larger bodied spedes typicaIly achieve lower population densities in favoured habitats. and contrast analyses are expected to reveal a negative relationship belween population size and body mass. However. when closely relaled bird species are being compared, the relationship can be reversed (Nee el a1.. 1991): larger bodied species have larger populations (for reasons which will be considered below: see Section 14.4.5). If a linear regression is forced through the origin of a contrast plot, thiS interesting phenomenon might weIl be overlooked (Fig. 14.4). The second note of caution when using independent contrast methods concerns error in the phylogeny being used. Independent contrast methods assume that the phylogeny is correct but. in fact, many phylogenies can mislead. For example. branch lengths may be incorrect, the extreme being the representation of a dichotomous phylogeny as a Slar phylogeny where all
z:-
2
'0;
c
~
'C
c
.51
0
;; -1
'5 Q.
It
.S ;;
1? E 0
u
·2 ·3
-.
-5
• 0
2
3
Contrast in body mass
•
5
Fig. 14.4 An independent comparison (or romrasl) analysis of (he relationship between
population densily and body mass among eight hypotht!lical bird spedes (which gives seven independent contrasts). When contrasts of population density are plotted against contrasts of body mass, the relationship is non· linear and a model I regression line forced
through the origin fails to distinguish the inlt~resling faci that. when contrasts are small. population densily 3C1ually increases with body mass. Small contrasts are belwcen closely relaled species. such as congeners.
PHYLOGENETIC FOUNDATIONS
345
internal branch lengths are zero, while node to tip lengths are the same. Alternatively, and ac!dilionally, the wrong topology may be used so that ralher than representing a coarsened variant of the true phylogeny, what we are actually analysing is the wrong phylogeny. Whatever the sources of error, as they are increased the chances of failing 10 reject an incorrect null hypothesis are also increased. When designing comparative methods, it is always necessary to test for acceptable statistical properties using simulation studies. Essentially, characters are evolved along the branches of phylogenetic trees according to set levels of character covariation. The comparative method is then applied 10 the resulting species data, and both type I and type 2 statistical errors are estimated. There have been several such simulation studies which reveal thaI. on the whole, independent contrast methods have acceptable statistical properties (e.g. Martins & Garland, 1991), and are a vast improvemelll over cross-species studies which treat species values as independent items of information. Although evolutionary biologists have now largely accepted this fact, there is still considerable resistance to it from some ecologists (Harvey, 1996).
14.4.4 Methods for categorical variables Starting with Ridley (1983), there have been several methods proposed (or analysing character covariation between pairs o( categorical variables. When the different methods have been applied to comparative datasets in order 10 test particular comparative predictions, almost invariably there has been a subsequent reply in the literature seeming 10 suggest that the original conclusion was incorrect because the method used had faults. Then, an 'improved' method is used 10 produce differelll conclusions. Cases relating to behavioural ecology include the relationships between warning coloration and gregariousness in insects (Harvey & Paxton, 1981; Silien-lUliberg, 1988, 1993; Maddison, 1990), and between plumage dimorphism and lekking in birds (Hoglund, 1989; Oakes, 1992; Hoglund & Sillen-Tullberg, 1996). This is dearly an unsatisfactory state of a((airs. Furthermore, there is a view that many popular methods (or the comparative analysis o( discrete traits, even though they were expliCitly constructed to 'take account o( phylogeny', do not solve the problem o( phylogenetic non-independence at all (Read & Nee, 1995; Ridley & Gra(en, 1996). Although there is a general consensus that contrast-based methods (or continuous characters are basically sound, there is no hint of a consensus about any aspect of comparative methods for discrete characters. Why is lhe status of the comparative method (or discrete characters so differelll to that (or continuous characters? One simple reason is as (ollows. With continuous characters, as we have seen. analyses begin by construcling contrasts and, almost always, such contrasts do, indeed, exist; (or continuous characters, even closely related species differ 10 some extent. This is no longer
346
CHAPTER 14
the case for discrete characters. Typically, large areas of the tree show no variation in one or both characters of interest. If, and how. such regions of the tree are to be incorporated into the analysis is a very contentious issue. More subtle problems that arise in the study of discrete characters are discussed in Ridley and Grafen (1996).
14.4.5 Sister taxon and other pairwise comparisons It is not at all surprising that working biologists often back away from performing a full-blown comparative analysis. The literature is peppered with reanalyses using different statistical methods and models of evolution. so that few conclusions seem robust. and evermore rococo comparative methods continue 10 prolilerate. Yet. it is now widely acknowledged that comparative biologists seeking correlated character change must take phylogenetic information into account: statistical inferences based on cross-species comparisons art' virtually worthless. Fortunately, it is possible to perform reasonably valid comparative analyses that do not require elaborate models and Statistics. So-called pairwise methods simply compare taxa whose separate evolutionary histories traced from a common ancestor are not shared with other taxa being compared. (Note that the taxa being compared could be two populations of a species, or two species. but that if intraspecific comparisons are made. those same species cannot be used in an interspecific comparison.) The advantages of pairwise methods are frequently bought at a considerable cost. The advantages are that ancestral character states need not be inferred and. consequently. precise models of character evolution necd not be specified. The cost is that a lot of information is ignored; frequently. the vast majority of variation in character states is found among the most distantly related taxa. and that is the variation that tends to be ignored with the consequence that very few degrees of freedom are available for statistical analysis. In short. the methods lack statistical power and may find it difficult to reject the null hypothesis of no character covariation. However. despite having high type 2 statistical error rates, pairwise comparisons are expected to have low type I ermr rates; they are unlikely to reject a correct null hypothesis. M011er and Birkhead (1992) have championed the use of pairwise comparative methods, and illustrate their use by testing the idea that high copulation rates among birds are a 'paternity insurance' mechanism. First. they asked
whether rales of extrapair copulation increased with local population density, which they did in all eight possible comparisons, which is a highly significant resllit. (Note: 15 potential comparisons are possible with 16 taxa when ancestral states are considered.) Second. they asked whether birds living in colonies had higher rates of in/ropair copulations, which they did in 12 of 13 comparisons. Unfortunately, in this the results arc not as striking as they first appear because sometimes the same bird species was used in an interspecific and intraspecific comparison. For example. solitarily and colonially breeding Hinmdo rustico were
PHYLOGENETIC FOUNDATIONS
347
compared, and then colonially breeding H. rllstica was compared with solitarily breeding H. dallrica. A fine example of the use of pairwise comparisons in behavioural ecology is provided by Robinson and Terborgh's (1995) observational and experimental study of Amazonian bird species. The territories of more than 330 bird species were mapped along a successional gradiem by the side of an Amazonian river. Species pairs from more than 20 genera were found to hold non-overlapping but comiguous territories, while others held partially or totally overlapping territories. Congeneric species pairs were then chosen for reciprocal heternspecific song playback experimems. One species of a pair which held non-overlapping territories typically approached the speaker aggressively, and when Ihis happened it was always the heavier species. Robinson and Terborgh argue that the larger congeneric species pairs thereby occupy the more productive end of habitat gradients, and thai it is the marked successional gradients typical of Amazonia which thereby helps to explain the increased congeneric species richness of Amazonian bird communities. Where they occur, the larger species
will be at higher densities as a consequence of contest competition for resources; in this case, territories in the higher productivity habitats. Robinson and Terborgh's sludy provides a test of one proposed explanation for the unexpected positive correlation between populalion density and body size mentioned above (see Section 14.4.3). Finally, we will mention an imriguing pairwise behavioural analysis that has provided an insight into population genetic history. Until the Panama seaway closed about 3 million years ago, the Caribbean and eastern Pacific were connected. The final closure of the seaway split many species of snapping shrimps of the genus A/pheus in two (among many other species, of course). For each of seven pairs of such species, Knowlton et 0/. (1993) studied the level of intolerance [snapping and other aggressive contacts) shown between individuals, one from each species in the pair, when placed in a glass dish. They also studied the degree of mitochondrial DNA divergence and allozyme divergence between the two species in each of the seven pairs. Treating degree of intolerance as a divergence measure, they found that all three measures are highly correlated, providing confidence in their final conclusion that, for three of the pairs, gene flow was disrupted millions of years before the final closure of the seaway.
14.5 Character change and tree structure combined Phylogenies can be used to ask a series of questions which integrate evolution, behaviour, ecology and community structure. For example, when speciation occurs, how and to what extent is the old ecological niche expanded to accommodate the two daughter species, and how and to what extent is it partitioned between them? 11, by chance, a species invades a new location,
348
CHAPTER 14
to what extent is its evolutionary IUlUre predictable? These and other similar questions are now being tackled by a combination of techniques which involve: (i) phylogenetic analysis; (ii) morphological, behavioural and eCological comparisons between Contemporary species; and (iiil long-term lield experiments. Losos and colleagues' work on Anolis lizards in I he Greater and Lesser Antilles islands in the Caribbean provides an excellent case-slUdy. There are about 150 species 01 Anolis lizards in the Caribbean, but only six 'ecomorphs': phenotypically similar species adapted to particular ecological niches. The ecomorphs are mainly described by the habitats which they exploit: twig, crown-giant, trunk-<:rown, trunk, trunk-ground, grass-bush. Morphological and ecological comparisons show that Ihe ecomorphs are a valid classificatory tool because they tend to form cluslers in multidimensional space when data are summarized by principal components analysis (Losos, 1996). However, the same ecomorphs have ollen evolved independently by parallel or convergent evolution, so that dilferent ecomorphs from the same island may be more closely related to each other than Ihe same ecomorph from different islands. Phylogenetic trees for the two independent radiations on Puerto Rico and Jamaica are reasonably well established. When Lo os mapped ecomorphs on to phylogenetic tree structures, the two islands revealed very similar pictures (Losos, 1992). The first speciation evem resulted in a generalist arboreal lizard and a twig species. The generalist then produced daughter species, a crowngiant and a trunk-ground ecomorph. The trunk--<:rown ecomorph was the nexi to evolve (on Puerto Rico this evolved from the trunk-ground species, and on Jamaica it evolved from the crown spedes). The evolution 01 the trunkcrown species on both islands was followed by speciation inlO a larger and a smaller lrunk--<:rown species. Finally, on Puerto Rico but not on Jamaica, the trunk-ground lorm spawned a daughter species that lives on grass and bushes independently 01 trees. A statistical randomization test demonstrated that the chances 01 two such similar sequences occurring by chance alone is very small indeed. Instead, the palterns point to a common process at work which Losos imerprets as niche partitioning (the treel with the occasional invasion of new niches (such as the ground, grass and bushes). The supposed role of niche partitioning in the adaptive radiations implies that interspedlic competition is likely to have been an important mechanism forcing species 10 occupy narrower niches, and the dilferences between the ecomorphs suggest that particular morphological and behavioural changes should occur over the generations when species invade new habitats. A number 01 experimental field studies have been used to lest these ideas (reviewed in Losos, 1996). 1 When two species were placed eilher alone or in combination on unoccupied islands, each species achieved higher population densities when alone than when in combination as predicted if interspecific competition is imponanl.
PH YLO G E N ETl C FO U N DATION S
349
2 Almost 20 years ago T.W. Schoener illlroduced one species from a known habitat onto a number of unoccupied islands. The change in habitat (from trees to scrub
v~gelati()n)
has been accompanied over about 40 lizard gener-
ations by expected changes in limb lenglh which are significalllly correlated with Ihe availability of perches with different diameters among the islands. 3 Some 'natural experiments' have occurred where unoccupied small islands have been invaded by specialized Irunk-ground and rwnk-crown species. In some cases the species have evolved to become less specialized. but in mher cases the specialization seems to have been retained even in the absence of competitors. Another notable study thai takes a phylogenetic approach to evolutionary ecology is thaI of Richman and Price (1992) on Old World leaf warblers. In Kashmir. there are eight sympatric warbler species of the genus Phylloscopus which constitute the insectivorous leaf-gleaning guild. Using Felsenstein's (1985) comparative method (see Box 14.1), Richman and Price first show thaI there are significant correlations between morphology, on the one hand, and behavioural differences in feeding behaviour and habitat choice, on the olher. So, for example, species with relatively longer tarsi breed in deciduous, as
opposed to coniferous, woodland and species which have a greater tendency
to f1ycatch have wider beaks. They also show that these correlations are robust in the face of uncertailllies in the phylogeny. An analysis of the contrasts suggests that the ecological diversificalion of the warblers can be characterized as an early differentialion of feeding method and prey size. with changes in habitat selection being more recent; closely related species tend to differ primarily along the niche axis of habilat choice and lillie on other axes. These studies, in addition to the others reviewed in this chapter, illustrate the celllral role that phylogenetic analysis can play in studies of behavioural ecology. Gone are the days when phylogenies were used merely to describe the genealogical relationships of taxa. Now we know that a phylogeny is a necessary component of any scientific allempt to integrate rhe resulrs of any biological study which is based on differences among populations or species.
Chapter 15 Causes and Consequences of Population Structure Godfrey M. Hewitt & Roger K. Butlin Genes exist within individuals and individuals make up populations which are geographically distributed in the habitats to whi h they become evolutionarily adapted through time. The behaviour of an organism, like other aspects of the phenotype, will be a product of evolution in lhis geographically and temporally struclllred context. It can, in turn, aflea the distribulion, interaClion and evolution of genotypes. In this chapter we wish to explore Ihe relationship between population geneticstruaure and behaviour. Both fields are currently producing and incorporating new ideas and approaches, but until recently have for the most part developed separately. We will consider how ancient and modern history has forged the present patchwork of diverse genomes within species, the effects of gene flow and population size, and how we may measure these. We will then consider some ways in which population structure and behaviour might interact. Does population history compromise currem adaptation? How does dispersal behaviour evolve when there is local adaptation, and how is local adaptation influenced by dispersal? Does divergence in mating behaviour restrict gene [low? Pue small. isolated populations sill'S for rapid evolution of behaviour and does this contribute to speciation?
15.1 What is population structure? We may define population structure as the spatial variation in density and genetic composition of individuals in a species. It is clearly a considerable task to describe this adequately within even one species. There are a number of components to populalion struaure that need to be considered and measured: primarily these include genotype variation, spatial hierarchy, life histories and temporal changes. First, the genotypes of individuals need to be identified. Under some circumstances one local population may contain only one or two genotypes while another of similar size may contain hundreds. Quantifying genotypic variants in a population is a mountainous task compared with simply counting the number of individuals but, as we shall see, even a small input of genetic information can greatly increase our understanding. Second, there is a hierarchy to the population struclllre of a species. Individuals and their genotypes exist in families (or clones), and one to several 350
POPULATION STRUCTURE
351
or these may be represented in the local deme. within which there is interbreeding and interaction. These demes are found in patches of suitable habitat:
the number or demes in any patch depending on the size and structure or the patch. Patches are usually clustered in panicular geographical areas, and there may be little or no movement between them. Figure 15.1 provides an example or this hierarchy for alpine Podisma pedes/ris, somelimes called the brown mountain grasshopper. There may be genetic differentiation at the level of denles or patches or clusters of patches.
Geographical areas orten contain distinct races that are phenotypically and genetically different (Mayr, 1970). Funhermore, where their ranges meet they mate and hybridize, producing hybrid zones that are usually relatively narrow (Hewitt, 1988). These narrow strips typically represent considerable genetic transitions between the racial genomes, which themselves may cover a considerable area. It has become apparent in the last decade that such genetic subdivision or species ranges into races separated by hybrid zones is common; indeed, the use o[ genetic markers such as allozymes, chromosomes and deoxyribonucleic acid (DNA) sequences has revealed much more cryptic subdivision than was apparent rrom external phenotypic characters (Bullin & Hewitt, I 985a,b; Cooper el al' 1995). Third, Ihe life history of each species determines a panicular age structure, class or caste specialization and sex ratio or its populations. At one end of the scale are the annual plants and univoltine insects that form a major part of the biota of temperate and boreal regions. In keeping with the seasons, they have one generation a year, so that when sampling adults in the summer they are all roughly the same age. Synchronization is also found in some tropical organisms like army worm caterpillars and locusts where, even though continuous breeding is possible, good weather conditions instigate development and breeding. At the other end are redwoods, oaks, whales, elephants ,1I1d man, that are long-lived with several overlapping generations and individuals of all ages. Differences in the age, caste, morph and sex structure of populations and species can have imponant effects on the genetic variation present and significant consequences ror the evolution or behaviour. In sampling populations for genetic and behavioural studies, one must be fully aware of these factors. Fourth, temporal variation must be considered explicitly in describing population structure. Some temporal variation is implicit in the life-history properties intrinsic to the organism. Temporal changes in distribution, density and genotype or individuals and populations are also caused by extrinsic factors of climate and habitat variation. These occur over time-scales from days to ages. For shon-Iived organisms that can complete a generation in a few weeks, seasonal conditions produce large changes in numbers and genotypes, and, in fruit flies like Drosophila, changes occur as Ihe rruit rots over a few days. While changes over a rew years can be studied first hand, longer time series require historical evidence, rossils and inrerence. Volcanic islands, such
:.1ao
·:11oom 1750m '1800m
....... -:7:::'0':'[email protected]• •-i:..... '.:.•'
Key:
Grasshoppers/10 m 2
00.5
o
Vegetation patch Scree
Grasshopper
' T,..
Fig. 15.1 Geographical distribulion of Podisma pedNlr;s. (a) This Wingless bort'o-alpinc grasshopper is found over much of northern Europe and Asia. and in the high mountains of southern Europe and Asia. However. its dislribulion is very inhomogeneous. (bJ Adapted to colder c1imarcs. in the Alps il lives on blocks of mountains over aboul 1500 m up to 2800 m. often associated with alpine shrubs like wonleberry. where it can be quite abundant. It finds hot. dry slopes and herbless scree inhospitable. This density map is otround Chabanon. Seyne-Ies-Alpes. France. (c) This aS5nciaiion wilh hahiral is apparent down to Ihe local populalion level. wilh each patch of suilable vegelation containing one or more demes. The dispersal distance of Podisma is 15-20 m generation-I. so while occasional migrants move between neighbouring patches 01 suitable vegetalion, crossing th... valley shown is most unlikely_ This location is between Tt'te Grosse and us Tumples.
POPULATION STRUCTURE
353
as Hawaii. provide a link through from recem eruptions to several million years ago. Lava floW', wipe out most vegetation and leave small isolated refugia, or 'kipuka'. They also dissect the range of species. New islands are produced over a 10'_106 year-scale and may be coloni7.ed progressively from neighbouring ones. Such processes must modify population structure enurmously, and there are some outstanding studies of the effects of historical and ancient events on I,he genetic structure and behavioural adaptations of Drosophila (Carson et at..
1990).
In summary, the population structure of a species can be seen as spatial variation in density and genetic composition of its individuals. It has a hierarchy from families to geographical races and subspecies. The life history of the organism imposes an age structure, sex ratio and polymorphism within this. and
environmemal fluctuations and catastrophes modify it through time. In terms uf ecology and behaviour, such features have significance for lucal adaptation 1.0 particular conditions, the form and consequence o[ metapopulation dynamics, the spread of adaptations and the shifting balance theory, the persistence of varieties and species with their adaptations, and the evolution of racial and specific charaClcrs.
15.2 What shapes population structure? tn the previous section we explained that population structure changed drastically over longer periods of time, laking hundreds and thousands of generations, caused by large climatic and habitat fluctuations. In the present, over a sciemific lifetime, we may study and experiment with the effects of habitat patchiness, local adaptation, population size. dispersaL extinction and colooization, and their interactions on demographic and genotypic changes. Having considered both past and present processes, it will be pussible to attempt to clarify which effects are dominant in particular cases and their relative importance under different scenarios.
15.2.1 Past processes The past 2'/2 million years -the Pleistocene period - is characterized by a series of major climate oscillations recognized as ice ages in temperate regions, but whose effects were global. These have a cycle of 100000 years driven by the orbital eccentricity of the Earth round the Sun. Lesser cycles of climatic change of 41 000 and 23000 years arc overlaid on this. There were lesser fluctuations nested within these, some very rapid, and fortunately these can produce brief warm interglacials, such as the period in which we live. The physical causes and detailed effects of these processes are currently the subject of much research and debate (Hewitt, (996). The effects on species distribution and population structure were enormous. Pollen and fossil analysis shows that organisms now in northern Scandinavia were in southern Europe,
354
CHAPTER 15
and those in northern Canada were south of the Great Lakes and the ice sheets. As the climate warmed around 15 000-8000 8r these species moved nonhward and up mountains, while others invaded from the south, where desens expanded. At the level of demes, this clearly involves much colonization and extinction, dispersal and adaptation. The distinct geographical races and genotypic forms that currently inhabit these regions. and are often separated by hybrid zones (Hewitt. 1988), must have moved in during this post-glacial period, the Holocene. Consider the example of the meadow grasshopper Chor/hippus parallelus where DNA sequence analysis identities tive subspecitic genomes in Europe (Cooper el al., 1995). Most genetic variation is found in the distinct taxa in Spain, Italy, Greece and Turkey. while the DNA sequences of the tifth taxon show that it spreads from the Balkans to France. UK, Scandinavia and western Russia (Fig. 15.2). Suitable vegetation for C. parallelus existed in the far south of Europe during the ice age; these southern countries have kept their refugial
:/
..
.....'.
....
• •
Fig. 15.2 Europe. shOWing the putdtive ice age relugia (shaded) from fossil evidence, and possible post·glacial expansion roules of the grasshopper C. paraUe/us as deduced from DNA sequences from the sample sites marked (black squares). Narrow hybrid zones occur in the Pyrenees and Alps. and possibly also in Greece. Turkey and Caucasus. where div('cged genotypes meet. Thus. C. J'Qralldlis is a patchwork of distinct genomes with considerable diversity in the south.
• • • •
POPULATION STRUCTURE
355
genomes - now up mountains - while a small sample from the Balkans expanded north, west and east to fill the rest of Europe. The dynamics of such colonization may well have reduced and restructured the genetic variation further (Hewitt, 1996). A number of other species provide similar evidence, although their detailed geographical distributions are not identical. For example, allozyme data on the Norway Spruce indicate that different genotypes colonized different parts of Europe during iLs post-glacial expansion from
refugia in the Dinaric Alps, Carpathians and Russia (Lagercrantz & Ryman, 1990). Several species show western and eastern forms in Europe, with hybrid zones running roughly north-south down from Scandinavia to the Alps, indicating southwest and southeast glacial refugia from which they have expanded - e.g. crow, mouse, newt, toad, snake, oak - and many others in Europe and North America have distributions which suggest similar histories (Hewitt, 1996). Consequently, populations that are presently close to each other in space may have quile distant geographical origins and diverged genomes. If these hybridize, the offspring may be more or less unfit and form a hybrid zone, and there may be no obvious morphological sign that this is occurring. Furthermore. while separale during Ihe ice age the two populations may have
developed differenl adaptations which may be reflected in their behaviour. Clear differences in morphology and behaviour between such parapatric populations have been found in many cases where hybrid zones have been studied in detail; for example, Spanish and French C. parallelus in the Pyrenees differ in their courtship behaviour, songs and pheromones (Hewitt, 1993a; Neems & Butlin, 1994). The yellow and red fire-bellied toads Bombina bombina/ variegata have distinct pond preference, mating calls and egg-laying strategies (Szymura, 1993). and similar examples occur in crickets, butterflies, mice, birds and flowering plants (see Harrison, 1993). Besides being excellent places to study the evolution of such characters, such cases demonstrate that the geographical history of the populations being studied can be of primary importance in understanding behavioural differences and adaptations. Methods for establishing such phylogeography are being developed using DNA sequences and offer exciting prospecls (e.g. Avise, 1994). Such range changes consequent on climatic and environmental fluctuations have some interesting population genetic consequences (Hewitt, 1993b). The colonization of new territory involves founder events by small numbers of individuals, which will reduce allelic diversity and reorganize gene interactions. Where the expansion is very rapid, this process will be repeated and accumulative, and results from an increasing number of species agree with this expectation. Classic allozyme data on lodgepole pine in northwest America coupled with radiocarbon dated fossils show how its post-glacial expamion northwards caused loss of genetic diversity and change in dispersal adaptation (Cwynar & MacDonald, 1987). The same pattern is seen for DNA sequence in C. parallelus in northern Europe (Hewitt, J 996).
356
CHAPTER 15
In recent limes, we have seen a number of range expansions, some caused by man, and indeed of man - out of Africa (200000 years), across Europe (5000 years) and around the world (300 years), The expansion of the collared dove out of Turkey/Balkans to progressively fill most of Europe this century is well documented (Hengeveld, 1989), The similar spread of the oak gall wasp (Stone & Sunnocks, 1993) also showed a reduction in allelic diversity as predicted with rapid coloni7111ion, The spread of the rabbit by Romans, Normans and British all over the world, and the house mouse with agriculture across Europe and then in ships to colonies, involved founder events, so that the populations in Australia and North America carry only a small proportion of the Old World genomes, Insect pest species and diseases of plants, animals and humans show great plagues and epidemics which have changed the ecology and genetics of many if not all species, and will continue 10 do so, One corollary of such Holocene expansions is that species have very different genotype composition and variation across their range. In Europe, southern populations in areas of glacial refugia may well contain greater diversity, which will affect their adaptIve response to natural selection, and is of relevance 10 conservation strategies. Another corollary is that equilibria, both numerical and genetical. may be hard to lind. Many populations are probably still evolving after their latest colonization or catastrophe, and the adaptations we study may be ongoing and genetically various. A life-history component measured in one population, such as number of eggs laid by a grasshopper, may still be changing under selection, and have a different genetic basis in another part of the species' distribution. 15.2.2 Present interactions
The habitats in which species live are more or less patchily distributed. Even before urbanization the population of generalist man was concentrated in more suitable environments. This localization is all the more apparent in flightless insects or small plants living in discrete areas on high mountains, such as the alpine biota of Europe. Such patchiness may be equally marked for isopods living in Scandinavian lakes, and at an even finer scale for insects specialized to a particular plant host that is scallered occasionally through a tropical forest. Populations on small isolated patches of suitable habitat will be subject to genetic drift. Over the generations Iheir allelic frequencies will vary stochaslically and alleles will be lost, leading in principle to increased homozygosity. Thb neutral divergence will occur in the ab ence of selection for different alleles, but it may also accelerate any differentiation due to natural selection. While small isolated populations will continue to diverge, in general the closer they are the more likely they will receive dispersing individuals from each other. These migrants will tend to delay or prevent divergence by exchanging genes among the patches and demes. The interplay of population size, gene
POPULATION STRUCTURE
357
now and selection, and its complexities, has received much theoretical attention with the produdion of many population genetic models (Wright, 1978; Slatkin, 1985). For a good basic text see Hanl and Clark (1989). The 'island models' arc most widely used, where a series of finite populations with N individuals each exchange migrants equally at a rate of m each generation. Drift will increase homozygosity (identity by descent) in each population and divergence among them. This divergence is measured by F', which can be estimated from the difference between the heterozygosity observed in the subpopulations and that expected from random mating across the whole population. Its value ranges between 0 and I. At equilibrium between drift and migration F' = 1/( 1 + 4Nm), and so Nm - the number of migrantscan be estimated. It is surprising how little gene flow can prevent significant genetic divergence among the island populations - e.g. even when there is only one migrant (Nm I) each generation, F' 0.20 at equilibrium. Another commonly used 'approach is the 'stepping stone model' where migration is allowed between neighbouring populations only. Under these assumptions too, the subpopulations become well differentiated when Nm < I, and when Nm > 4 they behave essentially as one population (Tables 15.1 & 15.2). Many organisms are not distributed simply as if on small islands, but more
=
=
Table IS.I Estimates of the number of migrants per generation between subpopuiations (Nm) from genetic estimates of inbreeding in 5ubpopularions (Fst ) calculated from observed and expected heterozygosities. (From Slatkin. 1985.~
Species
Organism
Nm
F.
Mytillls tdulis
Mussel Fly
42.0
0.006 0.025 0.102 0.106 0.200 0.212 0.225 0.532
Drosophila willisto'i Ptromyscus califomims Pln/rodon ouachitar Drosophila pwudoobscura Podisma pedislris Thomom)'$ bOllOt PlttJrodon cintrtJls
9.9
Mouse Salamander
Fly Grasshopper Gopher Salamander
2.2 2.t 1.0 0.9t 0.86 0.22
Table 15.2 Estimates uf neighbourhood size (Nb) from measures of dispersal (s) and population density (d), where Nb =- 4n.s1d, and s is the stJndard deviation in dislance between parent and offspring in metres. Species
Organism
s (m)
d (individuals m-I)
Podisma ptdatris Bombina bombino
Grasshopper Toad
Mus muscuills
Mouse
Dendroica (oronota lkmbidum viUQlI/m
Warbler
20 100 430 900 2
0.1 0.01 0.05 0.005 26
Intertidal snail
Nb (indiViduals)
503 t257 116176
50894 t44105
358
CHAPTER 15
continuously for at least some of their ranges, and models of genetic differentiation due to limited dispersal. or 'isolation by distance' have been employed. Genetic studies using allozymes and. lately, DNA markers generally show that subpopulation differentiation (F~) is higher in subdivided habitats as compared with continuous ones. It is also higher in poorly dispersing spedes than in similar good dispersers. However, a number of cases where values of F~ are not apparently compatible with the observed gene now remind us of the simplicity of these models and the complexity of factors affecting population structure.
In recent years, models of local extinction and recoloni7.ation have been examined - often called metapopulation dynamics - and have generated much debate concerning their effects on genetic variation and evolution (McCauley, 1991; Harrison & Hastings, (996). Most theory has revolved around the classic Levins metapopulation model. where each of the set of demes has equal likelihood of extinction and recolonization, and the question is whether such population turnover generates morc or less variation among the demes. If the recolonizers are few in number and from essentially one deme, then the dillerentiation among demes can increase, otherwise if the recolonizers are many and varied in origin then extinction/recolonization promotes gene !low and homogeneity among demes. The ecological reality of these melapopulation models is questionable, and other types of metapopulation may be more common in nature. Further insight may come from computer simulation, integrating genetic and ecological features. Most population genetic models concerning gene !low and drift assume no selective difference among the alleles, and for many milOchondrial and nuclear non-coding regions this Seems reasonable. However, there is increasing concern that occasional mutations causing increased litness may allow an allele and genes linked to it to spread through the population and even the range of the species. There is good evidence for such selective sweeps in Drosophila melanogasler, where a number of loci show reduced polymorphism in their region of the chromosome (Kreitman & Akashi. (995). In reverse. when deleterious mutations arise they will cause selective loss of their regions of the chromosome, and so reduce polymorphism. Consequently. as a population adapts 10 local condilions. alleles will be selected and polymorphism around lhem redoced; how much of the genome will be affecled will depend on how many genes are selecled and the tightness of the linkage. This should not affect direct measures of population structure but due caution is needed in deductions from indirect models based on neutral theory. Phylogeographical conclusions should also be robust 10 such effects. The great changes in distribution and population structure caused by major climatic oscillations have been emphasized. and such extinction and recolonization occurs right down to the local patch level over a few generations. It is therefore difficult to see thatlhe equilibrium conditions assumed for most oflhese models can pertain in almost any spedes. Furthermore. the
POPULATION STRUCTURE
359
approach to equilibrium is very slow in many models, particularly where population size is large; the clines in blood group alleles in man across Europe remain from the Neolithic agricultural colonization (Ammerman & CavalliSforza, 1984). Indeed. it has been known for some time that local areas of distinct allele frequency that arose by drift can persist for hundreds of generations. as Endler (1977) showed through simulation. The spatial pattern of neutral allele frequencies sel up during the colonization of an area can often include large patches of distinct genomes. which then remain for a long time under stable population dynamics (Ibrahim et 01., 1996). Consequently, a population genome is not only affected by current changes, but also carries residual effects of history.
15.2.3 Current versus historical effects Until recently the study of population structure has concentrated on current distribution and determinams. i.e. deme sizes, gene [Jow and seleetion. but historical evems like range expansion and colonization can produce major lasting effects on the dislribution of genotypes. For example. populations scattered over a wide range may be similar genetically. indicating high gene [Jow among them; on the other hand. they may have been put in place by a recent expansion and colonization from a source population, which could produce the same features but not involve current gene flow. How can we determine the importance of past and present events? As Nichols and Beaumont (1996) put it while considering data from man. Drosophila and Podisma - ·[s it ancient or modern history that we can read in our genes?' Differences in allele frequency across a geographical distribution have been analysed by principal components and spatial autocorrelation to measure the genetic similarities and relate these to known historical events, notably in man's colonization of Europe (Ammerman & Cavalli-Sforza. 1984; Sokal etal., 1989). Typically. blood groups and allozymes provide the data for such analyses. The availability of DNA sequences now provides not only allelic informalion in the form of haplotypes, but also the genealogy of these haplotypes. which can be expressed as a phylogenetic tree. When the tree is combined with the geographical distribution of the haplotypes it gives a phylogeography (Avise. J 989). For example. the related sequences on major branches of the tree can occur in separate parts of the species range, and the subset from a younger nested branch may be confined to a small area within a major one. Such relationships between geographical distributions and phylogenetic position allow one to distinguish among hypotheses about the historic events that could have produced such a distribution; in this case. it suggests an older geographical separation and divergence. with a younger muration undergoing local range expansion (Avise. 1994). These phylogeographies contain further information which can be mined by the development of suitable analyses combined with good data sets. and
360
CHAPTER 15
recently Templeton et 0/. (1995) have produced a major methodological advance in this endeavour. They developed a nested cladistic analysis of geographical dbtances among haplotypes, and applied this to mitochondrial DNA (mtDNA) data from the tiger salamander Ambysloma lig,illlilll from the midwest US. This
method introduces statistical rigou r to procedures for discriminating restricted gene now, past population fragmentation. colonization and range expansion.
In the tiger salamander very restricted gene flow is evidenced by the tips nf the mtDNA cladogram being geographically localized within the interior clades. and this fits well with its biology - it is a strict pond breeder, often paedomorphic and returns to its home pond. Range expansion is also apparent where a vounger clade of haplotypes is more widespread than it ancestral clade, particularly where it is found in areas thaI were affected by glaciation. This is the pattern noted in C. para/lelus in Europe (Cooper et a/.. 1995), and the Anrbystonra dala reveal two clear examples of such Holocene range expansions. Good background knowledge of the species biology and distribution, adequate sampling of the range, and informalion on the Pleistocene hislory allow the be,t use of such phylogeographical data and deductions.
15.3 Testing predictions about population structure Another important research endeavour is to teSI the measures and predictions in natural populalions of particular species, where Ihe various life history and ecological parameters can be specified and quantified to some extent. We will demonstrate this through a series of cases which cover many, but not all, or the major approaches.
15.3.1 Turkey oaks: neighbourhood size Several studies have attempled to discnn and measure neighbourhood size and structure. particularly in plants, but with somewhat variable success, and
it seems that adequate sampling in the field is crilical. Studies in a natural forest of Turkey oak (Querats :aevis) in South Carolina. assessing ils population genetic structure for the effecls of isolation by distance, provide an instructive example (Berg & Hamrick, 1995). Nine polymorphic allozyme loci were typed in 3400 trees in a 160-m' plot in an old stand. This impressive dataset comprising allele and genotype frequencies distributed spatially was analysed by a ballery of techniques, including priilcipally number of alleles in common (NAC). hierarchical estimation of differentiation (C'. a variant of F,,) and inbreeding (F,J, spalial autocorrelation ,'f allele frequencies and of genotypes. Control comparisons were provided by computer simulalions of a 10000 tree sland (100 x 100 grid) run for 10 000 years (10 years minimum generation time) with different sets of values for pollen and seed dispersal (which define neighbourhood size). This was ilerated 99 limes for each set of dispersal values.
POPULATION STRUCTURE
361
and the simulated data sets analysed by the same methods. Individuals within 10m have similar genotypes as shown by a positive correlation coefficient, but nol further away indicating a limit on gene flow above this distance. In the case of these oaks, this is probably due to heavy acorns falling near the parent tree. However, no larger patch structure of genotypes was revealed, indicating that gene flow over larger scales was sufficient to overcome any significant isolation by distance. The simulations gave the best fit with the observed statistics when the seed and pollen dispersal levels were high, encompassing a neighbourhood size of 440 individoals. Most acorns may fall near the tree but there is also clearly effective long-distance dispersal due to pollen and probahly some seed. Such leptokurtic dispersal is commonly described in field work with many organisms. The application of detailed ecological knowledge and complementary genetic analysis in this species has provided a much clearer understanding of its population structure.
15.3,2 Intertidal snails: gene flow The neighbourhood size in oak was deduced by comparing genetic evidence with simulations, and the dispersal was inferred indirectly. Dispersal could also be estimated directly by mark recapture experiments. although such ecological measures can be very difficult Or impossible in some species, as {or example those with highly pelagic larvae. Such direct and indirect estimates (Slatkin. 1985) were compared in the intertidal snail Bembicium viI/alum in the Houtman Abrolhos islands off western Australia (Johnson & Black. 1995}. These islands provided a comparison between a linear continuous II-km shore and a linear series of islands over similar distance and north-sout.h disposition. Along the continuous shore. based on mark-multiple-recapture experiments, the variance in dispersal distance (s') of snails was calculated over 15 months. Dispersal was leptokunic and provided estimates of neighbourhood size of 7000 adulls in one location and 1600 in another. Neighbourhood size was also calculated from genetic data in the form of 13 polymorphic allozyme loci in samples colleCled from Ihree nested transects of different length; several thousand individuals comprise this large dataset. This analysis showed Ihat subdivision began to rise between 150 and 300-m distance between samples, and therefore estimates neighbourhood size at 1665033300 in the dense location and 3900-7800 in the sparser one. More indirect estimates of neighbourhood size (Slatkin, 1993) from the alJozyme data based on a stepping stone model gave values in the range 22-38. These are very low and cause concern about the model's suitability in this case. In comparison the isolation by distance measured by G' in the series of separate islands was twiCe as high as the continuous population for those I-km apan, and did not increase with distance. Furthermore, alJozyme frequencies varied considerably. Clearly, small water channels between islands are effectively redUcing gene flow in this snail. This is understandable as the species does not have pelagic larvae.
362
C HAP T E R 1 5
These experiments and genetic analyses define the amount of gene flow, neighbourhood size and isolation by distance in continuous and discontinuous habitats and they provide a firm basis for understanding population structure and ils consequences.
15.3.3 House mice: DNA markers, breeding units and colonization In reCent years, with the invention of polymerase chain reaction and many new techniques for analysing DNA effiCiently. a much more powerful range of genetic markers is becoming available for population and ecological studies (see Avise, 1994). The millions of base pairs of DNA sequence in an individual genome comprise an enormous reservoir of information, some parts of which
are very variable and ideal for population studies (e.g. non-coding regions and microsalellites), while others are highly conserved over time. The raw data from DNA-based approaches often require new staListical and analytical methods (Slatkin, 1993, 1995). A recent study of populaLion subdivision and gene flow in house mice in Denmark utilizes both mtDNA control region sequences and microsatelJite loci to good eUect (Dallas el al' 1995). Previous studies have caused some debate as to whether gene flow is very restricted due to mice living in closed breeding units, or whether it occurs more widely as suggested by the spread of introduced alleles. There is certainly major geographical structuring, wilh hybrid zones, and isola lion by distance over several kilometres from allozyme and mtDNA studies. Dallas el al. (1995) sampled 11 farm populations in two clusters with the farms generally just a few kilometres apart. They used five microsatellite loci with 135 alleles altogether. and 888 base pairs of mtDNA sequence. Both Hardy-Weinberg and F statistics showed that within these populations mating was essentially at random. but 9 (an estimator of F,,) was significant indicating that the farms are to some extent genetically separated. A plot of 9 against distance beLween farms suggested some isolation by distance, with some gene flow among neighbouring sites. Assuming an equilibrium between migration and drift the level of 9 observed would be produced by Nm = 1-5 migrants each generation for each farm. Thus, gene flow appears limited and probably occurs by active migration between neighbouring farms. A further interesting inference is possible with such data. The effective population size (N,) can be calculated from populations collected over short periods and compared with the long-term N, I.e. the N,. necessary on coalescence theory to explain the nucleotide divergence (dx) in the control region mtDNA within a population that has been separate since its founding. possibly thousands of generations ago. These do not agree in this mouse example; Lhe current population size is much smaller than that needed to explain the survival of the current genetic diversity unaugmented by migration. [L strongly supports
POPULATION STRUCTURE
363
Ihe proposal that recenl gene flow or colonization by genetically varied mice has occurred. Such examples demonstrate how DNA data with suitable analysis can provide real insights into population Slructure, from family behaviour through to historical range expansions.
15.4 Interactions between behaviour and population structure The models of population structure that we have described so far, and which arc typically used to make inferences from genetic data, assume that dispersal is random with respeclto both habital and genotype, that all individuals have equal dispersal tendencies and that immigrants have the same mating success as residents. Clearly, Ihere is much scope for the real behaviour of animals and plants to cause departures from Ihese assumptions, and for selection to operate on behaviour in ways that mighl depend on population structure. We will consider some of these issues but, at present, the number of studies thaI explicitly deal with both behaviour and population structure is limited. 15,4.1 Non-random gene flow
fn Ihe consideration of population Slructure, migrams are typically assumed to be a random subset of the available individuals in each generation and the errect of migration is to reduce differemiation among subpopulatinns. However, it is easy to see thai selective migration can have the opposite effect of accenluating divergence, for some loci at least. Habitat preference is the simplest example: alleles that increase movement out of one habilal Iype and/or into another will quickly become differentiated among habitats. The resulting tendency for reduced movement among habitat types, and for mating encounters among individuals originating from the same habitat type, will favour local adaptation (Diehl & Bush. 1989). The apple maggot ny, Rhago/elis pomol1eUa, proVides an excellent example of genetic divergence favoured by habilat, in this case host-plant, fidelity. The apple maggot fly infests both apple and hawthorn fruits in the eastern US. Populalions on the IWO hosts exhibil consistent allele frequency differences at eleclrophoretic loci (Feder & Bush, 1989) and genetically based differences in emergence time that represem adaptations to the different phenologies of the two hosts (Feder el a/.. 1993). The contribution of host fidelily to Ihe maintenance of this differentiation has been studied at a site in Michigan where the two hosts grow togelher, well within the dispersal range of the adult flies (Feder el af.. 1994). Active host preference, a tendency to move to the nearest tree following adult emergence and matching (}f emergence time to host fruiting time all contribute to reslriction of interhost movement. However, Ihese experimemal results still predict about 6% gene exchange belween host races per generation. enough to remove the observed allozyme
364
CHAPTER 15
differemiation rapidly in the absence of other effects. most probably hostassociated selection (Feder el al.. 1994). The restriction of gene exchange facilitates further adaptation and thus greater isolation. However. it is not yet clear whether this process. which has come a long way in the short time since apples were introduced to the US in the mid-19th century. will continue 10 complelion and thus produce speciation. An interesting pOSSibility is thai the initial divergence was promoted by non-random gene exchange without genetic divergence (Butlin. 1990). II the population comains heritable variation in emergence time. and development is slowed on the new. presumably suboptimal host. then matings on the new host tend to involve late developing individuals from the original host and vice-versa. Thus. alleles for late development flow preferentially 10 the new host. and for early development 10 the original host. reinforcing the initial divergence without the involvement of selection. The structure of a population in a patchy habitat will be influenced by behavioural charaCleristics that influence the probability of an individual leaving a patch or being allraCled to another patch. There are few empirical data on these elleCls but a new study by Kuusaari elal. (1996) on the Glanville fritillary bUllerfiy. Melilaea cillxia. provides valuable pointers. BUllerflies introduced into an empty network of patches on a Finnish island were less likely to leave large patches. palches with many conspecincs. with many nectar·bearing flowers or with a high proportion of the boundary forested (Le. very distinci from the habitat within the patch). They were more likely to enter new patches that were large and had many nectar sources. Females were more likely 10 emigrate and moved further than males. perhaps because females were searching for oviposition sites while males were searching for mates. 15.4.3 Variation in dispersal Dispersal behaviour is a major determinant of population structure but it need not be a static characteristic of a species: il can evolve. Inbreeding depression will favour increased dispersal away from natal groups. while local adaptation will increase the costs of dispersal. Patchy environments. where there is a risk of extinction in anyone patch. will favour coloniZing ability. Adaptations for dispersal. such as wings and flight muscles. may be costly in themselves. movement may expose an individual to greater predation risk and there may be a signincant chance of failing to find suitable habitat after dispersal. Compromises among these various factors may be found in many poS'ible ways. for example in mammals it is common for males to disperse and females 10 remain in their nalal group. although there are several alternative explanations [or such a pallern (Greenwood. 1980; Johnson & Gaines. 1990). More powerful tests of hypotheses about the costs and benenls of dispersal can be made where there is marked dispersal polymorphism within species. Denno (1994) has used the widespread wing dimorphism of delphacid
POPULATION STRUCTURE
165
planthoppers to test everal hypotheses. Delphadds are small sap-sucking inseCls in which many species have fully winged macropterous forms capable of long-distance migration that appear in otherwise brachypterous populations when population densities are high or hostplant quality is low. Short-winged flightless brachypterous females reproduce earlier and are more fecund that macropters. Brachypterous males are often more successful in competition for mates than macropters and may live longer. Thus. macroptery must have benefits in some drcumstanccs to outweigh these costs. Denno (1994) prediCled that the frequency of macroptery would decrease with increasing habitat persistence. This prediction is very strongly supported by across spedes comparisons (Fig. 15.3) and the association remains significant when only phylogenetically independent contrasts are used (see Chapter 14). In species occupying persistent habitals. both males and females show marked density dependence in the frequency of macropters. whereas in temporary habitats males are macropterous at both high and low densities. This effect may be due to the need for mobility in mate location. One would dearly predict that the genetic strUClllTe of planthopper populations would also vary wilh the frequency of macroplery. and that local adaptation would be stronger in sedentary species. but these predictions have yet to be tested. In some insect species, especially holometabolous inseClS. wing dimorphism has a simple mode of inheritance. one locus with two alleles and brachyptery dominant. but in other well-studied examples. such as crickets, plant hoppers and gerrid pondskaters. a polygenic threshold inheritance pattern has been documented (RoH. 1994). There appears 10 be an underlying distribution of 'liability', wilh both genetic and environmental contributions. and a threshold liability above which development follows the macropterous pathway. Where it has been measured, the additive genetic contribution to variation is substantial. with heritabilities from 0.30 to 0.98. Thus. there is ample opportunity for wing dimorphism to evolve in response to local selection pressures.
~
..•
~
80
E .:!!
60
~
40
.:
~
li
eu
•
20
•
~
::;
~
100
0
1
•
•
•
10
• 100
1000
10000
•
Habitat persistence
Fig. 15.3 The relationship between female rnacroplery in natural populations and the persistence of the habitat among species of delphacid planthoppers. Habilat persistence
IS
estimated by multiplying habitat age in years by the number uf generations per year fur each species. Data from 4 J ropulalions tlf}5 speot.'S are included. The relationship
remains significant when
corrl~et('d
ror phylogenetic indl'pendence. (From Denno. 1994 )
366
CHAPTER 15
In isolated patches of suitable habitat, dispersal is effectively selective death, and dispersing individuals are unlikely to be replaced by immigrants. One therefore expects to find lhat species living on islands. or in isolated habitats like mountain tops, have lost night ability. even in the absence of flight polymorphism in other populations. This effect has been Widely documented, ahhough it can be difficult to separate direct selection against dispersal from selection on other life-history traits in the absence of predators. for example (McArthur & Wilson, 1967). BUllerOies in isolated patches have been observed to evolve toward smaller size and there is some evidence that larger individuals are more likely to leave patches (Kuusaari ft al., 1996). At the other end of the scale, patchy and ephemeral habitats will seled for high dispersal ability. Given a trade-off between dispersal and fecundity. one should find a correlation between habitat stability and dispersal ability within species analogous to the correlation across species in planthoppers. Such a correlation has been observed in the African armywomt moth. Spodopffra exempta, by Wilson and Gatehouse (1993). This mOlh is a serious pest on pasture grasses and crops that depend on sporadic rainfall. Migratory potential. as measured by pre-reprodudive period (PRP), shows heritable variation within populations and significant differences among populations which correlate with the frequency of rainfall in the predicted direction: shon PRP where rainfall is frequent. long where it is infrequent. A further prediction is that longer established populations will have lower frequencies of dispersive morphs or lower dispersal tendencies (Olivieri et al., 1995). This has been documented in thistles, Cardllus spp.• which produce two types of seed: with or without the pappus thaI enables them to be dispersed by wind. Older populations produce a lower proporlion of seeds with the pappus. As in wing-dimorphic insects. selection within a population would result in zero dispersive seeds but seledion at the level of the metapopulation maintains dispersal since new sites can only be colonized by migrants. 15.4.4 Mating success of migrants
Dispersal is not equivalent to gene now: it has no genetic consequences unless immigrants leave offspring and to do so they have to achieve matings. However, selection may favour either avoidance 01. or preference for. immigrants as mates. This can best be illustrated with plant populations. Electrophoretic analysis of plant populations frequently demonslrates sl(ong spatial genetic Strudure even where distributions are fairly continuous. The spatial scale may be small. only a few metres in some species (Waser, 1993). Thus. there is potential for outbreeding depression either due to genetic incompatibility or because offspring are poorly adapted (0 either of their parents' local environments. On the other hand. normally out breeding plants typically sufler strong inbreeding depression if selfed or crossed to close relatives (Barrell & Harder, 1996).
POPULATION STRUCTURE
367
In the slow-growing. perennial larkspur, Delphinium nelsonii, long-lerm experiments by Waser and colleagues (Waser, 1993; Waser & Price, 1994) have explored the fitness consequences of crosses 10 male parents at varying distances from the maternal plan!, when the ollspring are planted close 10 the maternal parent. Crossing distances were chosen on the basis of the observed movement palterns of hummingbird and bumble bee pollinaLOrs. Seed dispersal is less Ihan I m. The resulls are summarized in Table 15.3. There is evidence for both inbreeding and outbreeding depression, and selection is strong in bolh cases. The typical pollen and seed dispersal distances are such that a randomly chosen mate at 1-01 distance is roughly equivalent LO a lirst cousin, a result consislent wilh electrophoretic data on population struclure, and yet litness reduction is still about 0.5 relative 10 the to-m crosses. Outbreeding depression is even grealer. It is very unlikely 10 be due 10 genetic incompatibility at this scale but there is evidence, from reciprocal lransplanl experiments (Waser & Price, 1985), for adaplalion to local environmemal conditions on a comparable scale. The genetic struclure of Ihe population will clearly be influenced by Ihe reduced IiIness and Ihus reduced genetic contribution of progeny from both shon- and long-distance migrams. Funher local adaptalion will tend to be favoured. One might also expectlhe evolUlion of slrategies on Ihe part of the female parent to avoid outcrossing. In D. nelsonii Ihere is, indeed, evidence for such an effect: pollen lube growlh is slower for pollen from flowers al 100m Ihan from those at 10m from Ihe malernal parem, and this has the expected ellect on fertilizalion success (Waser & Price, 1993). This is analogous 10 the comroversial process of 'speciation by reinforcemem', selection for assonalive mating in response to the production of unlit 'hybrid' progeny (Bullin, 1995a). In this case, it forms a step in a feedback loop: local adaptalion produces oUlbreeding depression which favours mechanisms LO reduce wide OUlcrosses and thus facilitales funher local adaptation. How commonly Ihis loop operales is nOl clear. In principle, il can cenainly apply to animal populations as well and there are some dala to suppon Ihe idea of Table 15.3 EHt'et of crossing distance on fitness of offspring in D. ne(soIJii. (From Waser & Price, 1994.)
Crossing
Progeny
Flower
distance (m)
Progeny size·
lifespan (years)
productiont
Overall fitness:t
I 3 10 30
78.0t 122.9 t 131.1 t 64.4 t
3.33 ± 0.36 3.77 ± 0.32 4.27 t 0.32 3.11 to.32
4 6
0.30 t 0.086 0.42 ± 0.086 0.65 ± 0.086 0.08± 0.086
• leaf area in
mOl!
19.9 23.2 t9.9 24.2
8 2
5 years aher planting.
t Number of sibships (lowering within t: A of the lnlka-Euler equation.
7 years. oul of 13 in each case.
368
CHAPTER 15
optimal outbreeding (e.g. in crickets: Simmons. 1991; see also Chapter 4). A similar feedback loop can be produced by the interaction between population structure and cooperative breeding (Breden & Wade, 1991; see also Chapter 10). Detecting the process is. however, critically dependent on studies at the correct scale and with thorough documentation of fitness in the appropriate habitat (Waser, 1993). An indirect way of assessing the likely genetic contribution of immigrants is to examine geographical variation in mating signals. Both mating signals and, to a lesser extent, associated preferences have been shown to vary among populations within species (Butlin, 1995b; Bakker & Pomiankowski, 1995). However, the geographical scale of this variation is much greater than the dispersal distance so that its impact on gene now must be negligible. Even in hybrid zones between races that differ in mating signals and preferences, where the steepest transitions would be expected, clines for these characters are wide. For example, in the fire-bellied toads, Bombina bombina and B. variegata, clines in song characters are in the region of 10-km wide, wider than the clines for allozyme markers and much wider than the dispersal distance of about 1 km (Sanderson et al., 1992). Similarly, in the grasshopper, C. paralMus, clines in song characters and female preference are wide relative to the dispersal distance of about 30 m (BUllin & Ritchie, 1991). Assortative mating in a steep cline for direction of coil in the land snail, Partula sU/llralis, on the Pacific island of Moorea has little, if any, impact on gene exchange as measured by allozymes (Johnson et al.. 1993).
15.5 Evolution of behaviour in small isolated populations A long-standing debate in evolutionary biology concerns the role of founder events, where small numbers of individuals initiate isolated populations, in divergence and speciation. The alternative points of view were crystallized in a seminal pair of papers in 1984 (Barton & Charlesworth, 1984; Carson & Templeton, 1984). The debate concerns the likelihood that the extreme sampling effects in the foundation of such a new population will cause sufficient genetic change for it to come under the domain of allraction of a different adaptive peak, in the framework of Wright'S 'adaptive landscape'. This debate continues; see Whitlock (1995) for example. 'Peak shifts' in general are beyond the scope of this chapter, but specific effects of founder events and evolution in small populations have been proposed for mating behaviour and the consequent origin of pre-zygotic reproductive isolation. 15.5,1 Additive genetic variance in courtship Characters that are closely correlated with fitness are unlikely to maintain substantial additive genetic variation within populations and, furthermore,
POPULATION STRUCTURE
369
founder events are likely to involve loss of additive variation. As a result. the potential for evolutionary response in such characters in recently founded
populations is expected to be low. However, non-additive variation, due to dominance and epistasis (interactions between alleles at different loci), can be maintained under selection and may be 'convened' to additive genetic variance during founder events as a result of random changes in allele frequencies (Goodnight, 1988). The simplest example would be a recessive allele at low frequency in a source population: an increase in frequency in a founder event would cause the allele 10 cOlllribute much more to the additive genetic variance.
Counship characlt:rs arc clearly fitness-related traits. There is evidence for low additive genetiC variance. but signiricant dominance variance in natural
populations for such traits as male counship vigour or female mating propensity (Meffen, 1995; see also Butlin, 1995b; Bakker & Pomiankowski, 1995, with respect to mating signals and preferences). Epistatic effects are harder to detect. If significant additive variance is released following founder events, this could contribute to the evolution of mate discrimination and ultimately specialion. Some assonative mating has been observed in experimentally bOlllenecked laboratory populations, mainly in Drosophila, although the eUects are not strong (Rice & Hosten, 1994). Direct allempts to measure the expected release of additive genetic variation have been successful for other characters, but a study of these effects for counship repenoire gave equivocal results (Meffen, 1995). Meffen (1995) compared estimates of additive genetic variance, from parenHJffspring covariances, in six bOlllenecked and two control lines of house fly, Musca domesr;ca, for II components of counship behaviour, both male and female. The six bOlllenecked lines showed concened phenotypic shifts over 10 generations in seven of the II characters and significalllly increased addilive genetic variance in some cases. ThiS could be explained by directional dominance but there was no close link between increased variance and phenotypic shifts. Thus, some other mechanism, potentially epistasis, must be involved. The measuremelll of these couTlship characters is, however, seriously complicated by the influence of the behaviour of the maling panner. For example, male courtship may appear more vigorous and more complex when a male encounters an unreceptive female than when the same male encounters a receptive one. Since effects of bOlllenecks are not likely to be coordinated across counship elements, the most common effect is likely to be a reduced success rate of courtship. Typically, this will be analogous 10 inbreeding depression: it will increase the risk of extinclion of the newly founded population. However, it may be Ihat. rarely, a resolution of this incompatibility can lead to a realignment of the courtship elements in a novel way that creates incompatibility with the source population and thos contributes to speciation.
370
CHAPTER 15
15.5.2 The 'Kaneshiro' hypothesis The evolutionary scenario originally envisaged by Kaneshiro (1976) had much in common with the above argument. Kaneshiro observed thai pre-mating isolation was asymmelrical between pairs of species in the planitibia subgroup of Hawaiian picture-winged Drosophila. Females of D. pla'itibia (P) and D. difJerens (D) discriminale againsl males of D. silvestris (S) and D. heleronellra (H) more slrongly Ihan females of S or H do againsl males of P and D in IwO male: one female laboralOry tesls. Sand H occur on the youngest Hawaiian island, Ihe 'Big Island' of Hawaii ilsdf. whereas P and D occur on the older islands of the Maui complex to the northwest. Chromosomal phylogeny suggests that Sand H are derived from Por D by a process that certainly involved colonization of the Big Island and may have been driven by founder events. Kaneshiro, therefore, proposed that the founder event may have lead to simplification of the courtship repertoire of the derived species through loss of behavioural elements in the founder event and, furthermore, that this pattern could be used to infer the direClion of evolution where asymmetrical isolation was observed. If derived males lack courtship elemems typical of the ancestral species, ancestral females will discriminate strongly against them while ancestral males will more than satisfy the requirements of the derived species. The pallern of asymmetry is supported by observations on natural and laboratory variation within species (Kaneshiro, 1989), although scenarios that do not involve founder events may predicl different asymmetries. Purther study of courtship in P and S (Hoikkala & Kaneshiro, 1993) has begun to elucidate the behavioural basis of the asymmetrical isolation. The presence of the same set of courtship elements in each species initially appears to contradiCl Kaneshiro's original hypothesis, but Hoikkala and Kaneshiro argue that critical transitions between male and female behavioural elements have been lost. The connection between founder events and loss of behavioural elements has never been very clear. ft is a less critical part of a more recent formulation of the same underlying idea (Kaneshiro, 1989) which concentrates allention on the changes in selection pressures associated wilh colonization events, rather than genetic changes. Source populations are supposed to exist al relalively high population densities and in more complex communilies than are newly founded populations. Kaneshiro argues tbal high densities produce high rates of encoumers between potenlial males, while complex communilies may contain relaled species so Ihat the risk of interspecific courtship is also high. In these conditions, seleclion will favour discriminating females with slringent courtship requirements since the risks associated with rejecting a male are small and an encounter with another male can be expected quickly, while the risks of indiscriminate maling are high; a randomly selecled male may well be of the wrong species. If females are very discriminating in male choice, males will be selected for high courtship vigour and elaboration of mating signals. The opposite conditions apply in small newly founded populations. Population
POPULATION STRUCTURE
371
densities are low, at leasl initially, and the new habitat may not have been colonized by related species. Relatively indiscriminale mating by lemales is favoured by selection. Male courtship vigour will decline and signals may degenerate. These difrerences alone are sufficient to generate asymmetrical isolation. The vigorous males of the ancestral population easily satisfy the weak conditions imposed by the derived females, whereas the derived males cannot provide sufficient stimulation for the ancestral females. Kaneshiro further argues that relaxed seleaion will allow greater genetic variation in the derived population. In time, the new habital is likely to fill, and the pattern of selection to re!Urn to the conditions of the source populalion, but it is possible that a novel syslem of signals and responses will crystallize from the variable and loosely coordinated courtship of the derived species, eventually leading to symmetrical isolation. Some of the necessary elements of this process have been demonstrated. For example, where female discrimination has been separated from female preference it has been found to be more variable, and with a larger heritable component, than preference (Bullin, 1993). Drosophila melal10gasterstrains from Africa, the ancestral range of the species and its close relatives, have higher courtship vigour in males and lower mating propensity in females than French strains (Cohet & David, 1980). However. it remains imponant that the mechanism is modelled explicitly and further investigated empirically. 15.5.3 Parallel speciation These ideas view mating preferences and the traits on which they operate as if they evolve independent of the environment and of other adaptations of the organisms concerned. By contrast, Paterson (1985) has argued that maling signals and responses are closely adapted to the environment in which they operate and that colonization of novel environments is the primary driving force fo[(heir evolutionary divergence. Rice and Hostert (1994), from a review of laboratory experiments on speciation, conclude that reproduaive isolation is most likely to evolve as a pleiotropic effea of adaptation to divergent habitats. Small isolated populations will invariably be in habitats that are distinct from those of the source populations so it is dangerous to conclude that an association between speciation and founder events is necessarily due to bottlenecking effeas rather than seleaion. Schluter and Nagel (1995) have recently drawn attention to the fact that a large ancestral population may often give rise to many new populations in similar environments, especially in periods of population expansion like the phase of global warming at the stan of the Holocene. Such repeated colonizations can provide powerful evidence for the role of seleaion in both adaptation and speciation if the same patterns appear in independent replicates. Consideration of one of the possible examples offered by Schluter and Nagel will make the mechanism clear.
372
CHAPTER 15 Al the end of the PleislOcene, new freshwater habitats became available
in many areas of the northern hemisphere and were colonized from a largt:
ancestral marine I>opulation of the three-spine stickleback, GasterosteLls arlltealLls. Adaptation to the freshwater habital in most cases involved a substantial reduction in adult body size, although in some lakes twO forms evolved: a small-bodied limnetic form dnd a large-bodied benlhic form. Since mating is strongly size-assortative, these adaptive changes in body size incidelllally produced pre-mating rel>roduclive isolation bOlh bel ween Ihe small-bodied fn:shwater populations and their large-bodied marine ancestors. and belwet'n Ihe pairs of benthic and limnetic populations within lakes. The slOchastic effecIS 01 rounder events are mOSI unlikely 10 have produced these parallel pallerns, whereas they are readily explicable in terms of adaptation. Indeed, the pairs of forms within some lakes call inlO question Ihe need for geographic separation, let alom:: botllcllecking, in lht' origin of repruductive isolation.
15.6 Future prospects The great majority of examples presented in Ihis chapler have been either examples of genetic analysis of population structure or studies of behaviuur. Very few organisms have been considered from both points of view and yel Ihere are clearly good reasons why they should be. The history of many species may compromise the optimality approach if populations arc still evolving local adaptations; the current struclUre of populations will influence Ihe precision of local adaptalion while gene [low determines Ihe scale of variation in behaviour. Behavioural variability and evolution in response to the costs
and benefits of dispersal similarly compromise the simplifying assumptions of population genetics. Key qoestions that remain to be answered, and where the combination of modern genetic techniques wilh progress in theoretical
modelling and computer simulalion may permit progress, are to determine Ihe scale of local adaptation in behaviour in relation to the genetic structure of populations (determined from pUlatively neutral loci), and Ihe in[luence of history and contemporary dispersal and gene flow on local adaptation. Of particular interest are processes that might involve feedback loops, such as the interaClion bel ween optimal outbreeding and local adaptation, and those that involve conflicts between selection pressu res al different levels, such as dispersal polymorphisms. There are greal opportunities for collaboration between populalion geneticists and behavioural ecologists, but the magnilUde of the undenaking should nm be underestimated.
Chapter 16 Individual Behaviour, Populations and Conservatiori John D. Goss-Custard & William J. Sutherland
16.1 Introduction Throughout the world. habitats are being lost and degraded and natural populations are in danger of being overexploited by the ever increasing demands placed upon them by our expanding and resource-hungry population. There is widespread concern about the efleet of this on species diversity and on the abundance of particular organisms. One of the many factors that will influence how energetically such concerns are translated by decisionmakers inlO conservation-friendly action is the credibility of the predictions made by scientists as to the consequences of allowing present trends to continue; few will take seriously doom-laden forecasts that are scientifically suspect. In many of the cases where natural populations need to be managed rather than their habitats merely secured. quite sophisticated quantitative predictions may be required. For example. when some people harvest a species that other people like to watch. a conflict of interests arises that may require policy-makers to make trade-offs and scientists to provide quantitative predictions for a number of policy scenarios. The need for a strong scientific underpinning of decision making in such circumstances remains as strong as ever. The scientific challenge arises because ecologists are increasingly being asked to predia outside their direct experience. As scientists, we are trained 10 treat with suspicion extrapolations beyond the empirical range. Yet. this is precisely what we must do if we arc to forecast to the novel circumstances brought about by changes in such fundamentally important global systems as the weather. In the case of vertebrate populations. the subject of this chapter, this means that we need to forecast for the new environments the foml and parameter values of the important demographical funaions. such as density dependent recruitment and mortality rates (Goss-Custard. 1980. 1993). It is difficult enough 10 measure such funaions in present-day circumstances, yet we are required to understand their basis sufficiently well to predict their properties in novel ones. But. unless we can du so, conservation and resource-management strategies may fail because key components of our population models do not apply in the new environments for which predictions arc needed. 173
374
CHAPTER 16
Our contention is that behavioural ecology has an important role to play in making predictions at the population level, even though few behavioural ecologists so far have attempted to do so. We argue that it is possible to derive demographical functions in new environments through the study of competing individuals (Goss-Custard, 1985, 1993). We focus on predicting demographical functions because quantitative population prediction requires us to use population models that include demographical functions appropriate to the new environments. The many other ways in which behaviour may have important consequences for population and conservation issues are therefore outside the scope of the chapter. As examples: the social system has considerable consequences for effective population size (Primack, 1993) and thus the amount of genetic variation that will be retained; the dispersal pattern will influence both the extinction and recolonization rate of populations; foraging behaviour can be used to examine the behaviour of hunters (Wimerhalder, 1981), fishers (Abrahams & Healey. 1990) and whalers (Whitehead & Hope, 1991). It is not our aim to discuss how discoveries at the behavioural level have implications for understanding processes at the population level. Rather, it is to illustrate how behavioural slUdies of individuals enable us to derive demographical functions that can be inserted into population models in order to make quantitative predictions of population responses to new environments. Our fundamental point is that this approach provides a reliable basis for prediction because the choices made by the individual animals are based on decision principles, such as optimization, that are unlikely to change in the new environments, even if the exact choices made by animals, and thus their chances of surviving and of reprodUcing, do so.
16.2 Importance of density dependence When populations are at high levels, individuals are more likely to starve, fail to breed or raise fewer offspring if they do breed. This simple concept of density dependence is of fundamental importance in population biology and has been known since Malthus's (1798) suggestion that increasing human populations will lead to increasing 'misery and vice'. It is, however, surprising how many behavioural ecologists. conservationists and even some ecologists have yet to grasp the widespread implications of this simple idea. If conservationists wish to predict the consequences of exploitation, changing the mortality rate caused by predators, disease, pollution or habitat loss, then it is usually necessary to understand the role of density dependence. We concentrate in this chapter on predicting the consequences of habitat loss and change. For sedentary species, the consequences of large-scale habitat loss are relatively easy to predict. Destroying a large block of habitat containing 50% of a spedes of snails is likely to halve the population because the demographical functions in the area that remains have not been affected by the adjacent habitat loss. Note, however. that this is only likely to be true if the
BEHAVIOUR AND CONSERVATION
375
scale of habitat loss is much greater than the individual animal's home range. If the habitat loss occurs on a small scale. the results can be hard to predict without in-depth studies of species requirements. For example. with smallscale forest clearing. some species may be indifferent to. or even benefit from. the creation of open patches. But. others may show greater declines than the proportion of the habitat that is lost because of increased predation along forest edges (Wilcove. 1985). changes in microclimate or as a result of the dispersal paltern being affected (Henderson it at.. 1985). For migratory species. predicting the consequences of habitat loss is much more complex and we devote much of the rest of the chapter to tlUs problem. An understanding of density dependence is now crucial to understanding the consequences of habitat loss (Goss-Custard. 1977). If there is no density dependence. habitat loss does not maller providing that displaced individuals can locate the remaining habitat and move to it. The population can then simply crowd together in the areas that remain and fare just as well. Although the local density is increased. total population size is unaffected. At the other extreme. there is perfectly compensating density dependence. such that the addition of one individual leads to another individual either dying. emigrating or failing to breed. In this case. a loss of habitat leaves population density the same as it was before in the areas that remain. but the total population size is reduced. The extreme pred,ctions are thus that habitat
(b)
(81
E'
E
Ie)
E
(dl
Total population size
Fig. 16.1 (a) The equilibrium population size E is that at which the net birth rate (thin line) equals the winter death rate (thick line). In this casc, both net binh rale and winter death rate are assumed to change with density. (bl A loss of winter habitat shifts the winter mortality curve to the left and results in a lower population. (c) The c:onsequenct's of a loss of breeding habitat. (d) The consequencf'S of a loss uf both.
376
CHAPTER 16
loss has no impact on the population (as some developers assert) or that the decline in population size is in direct proportion to the amount of habitat that is lost (as some conservationists assert). One general way of considering the issue of habitat loss in migratory species is to consider the relationship between total population size and per capita population change on both the breeding grounds and the wintering grounds (Fig. 16.1). Habitat Joss can then be expressed in terms of a shift in these density dependent relationships and 80x 16.1 shows how the consequences
Box 16.1 The relationship between habitat loss and population size for migrating animals. (After Sutherland, 1996b.) Consider a small loss o( winter habitat so that the winter mortality curve is
shifted very slightly by an amount population size (see Fig. 8.16.1). From simple algebra: xw Lw
;:: _d' _
Lw '
This results in a reduced equilibrium
(8.16.1)
d'+ b'
where Xw is the decline in population size and Lw is ,he loss of habitat. so that xwlLw is the relationship between population decline and habitat loss. As shown in Fig. 8.16.1, b' is the slope of the relationship between net birth rate and density and d' is the slope of the relationship between winter mor-
tality rate and density.
E' E Total population size
Fig. B.16.1 How the consequt'nces of a small loss of winter habitat can be predicted. As in Fig. 16.1. habitat loss shifts the willler mortality curve (thick line). After hahil31
loss the population declines from E to E'. The circle to the right shuws the critical pan of the figure on a larger scale. Note Ihal il is assumed the slopes are linear over such a small scale. xw' decline in population; Lw ' the change in Ihe lOlal population size required for the populalion to remain the same in Ihe remaining patches (hence. L,./E equals proportional habilJI Joss). 4From Sutherland. 1996b.)
BEHAVIOUR AND CONS ERVATION
377
Box 16.1 Con I'd. A value xwlL w of I means that the population declines in direct proportion
10
the extent of habiull loss: a 2% loss of habitat then results in a 2% population decline. A low value of x,/ Lw means Ihat Ihe population decline will be much less than the habitat loss. Almost exactly the same approach can be used to predict the population decline x, as a consequences of the loss of breeding habitat. L~: b' . .:1.= __ L~
d'+ h'
(B.16.2)
Thus. if the density dependence can be measured in both the winter and
breeding grounds. the consequences of habitat loss of average quality can be considered. The value of d'derived from studies of oystercatchers on the Exe
estuary (see Section 16.4) is 0.00011 and the value of b' for studies in Schiermonikoog (see Section 16.3) for the same-sized populalion is 0.00005. Incorporating these values into Equation B.16.1 shows Ihat a I % change in winter habitat will result in a population change of 0.69%. while incorporating them in Equalion B.16.2 shows that a 1% change in breeding habitat will result in a population change of 0.31 %. Resident species an be considered as a special case of this in which the wintering and breeding habitats are the same. It thus follows by combining Equations B.16.1 and B.16.2 that xw.J Lw., = l. The populalion decline will be in direct proportion to the extent of habitat loss.
of the loss of habitat of average quality can then be delermined. The impact of habitat loss is more severe for the season in which the density dependence is stronger. That is. if density dependence is weak in one season bUI strong in Ihe other. Ihe loss of the habitat where the density dependence is stronger has grealer impacl. At first. il may seem surprising thai the impaci of habitat loss is only affected by the ratio of lhe strengths of lhe density dependence in the two seasons (Fig. 16.2). The consequences are as great in a species with strong density dependance in bOlh the breeding and wintering grounds as in one with weak dependence in both. Most importantly. Box 16.1 shows that the consequences of habitat loss on population size can only be predicted if the strengths of the density dependence in both the breeding and wintering areas are known. Much of the rest ofthis chapter describes how to estimate the strength of density dependence and to predict any changes in function form that might arise as a result of habitat loss and change.
378
CHAPTER 16
la)
Fig. 16.2 TWo examples to illustrate that it is the relative
slopes of the winter (thick lines)
(bl
and summer (thin lines) density
dependence that is critical. and not the absolute values. In (a), Ihe density dependence in both the wintering and breeding areas is sHong. while in (b) it is weak
in both. The consequences of losing 25% of the wiOlering
habitat leads to the same proportional decline in each casco E'
E
Total population size
After habitat loss the population declines from E to l!.
16.3 Determining the strength of density dependence Occasionally, it is practicable to determine density dependance by manipulating densities experimentally (Alatalo & Lundberg, 1984). Bul, the usual approach is to estimate the Iife-hiSlory parameters or mortality rate or birth rate over a naturally occurring range of population densities. Although this method has been used successfully (Sinclair, 1989), there can often be problems with this direct approach. In most vertebrates, population size is unlikely to vary enough for the density dependent [uncI ions to be defined, especially if there arc large annual variations in survival or birth rates due to weather and other factors. A more fundamental problem is that a populalion is likely to be temporarily high in the first place for reasons which affect life-history parameters; for example. a reduction in parasites which affect mortality or breeding output. The observed density dependence may then differ from the response to density that would apply under more typical conditions. Furthermore. it may be necessary following habitat loss to extrapolate the density dependent relationships to densities that are conSiderably higher or lower than lhose which occurred during the sludy ilself. Finally, empirical estimates of density dependem relationships will only apply to the conditions under which the slUdy took place. If habitat of above or below average quality is IOSI. for example. new density dependent relationships will apply, thus making
BEHAVIOUR AND CONSERVATION
379
present-day functions inappropriate (Goss-Custard 'I al.• 1995c). Removing the best or worst patches changes both their average quality and spatial arrangemem. An approach is needed to show how individuals would respond to thi and how it would arrect their chances of survival at dillerem population sizes. 16.3.1 A behavioural approach to density dependence An alternative approach to determining density dependence which allows this to be done is to understand the decisions individuals make. specially in their choice of feeding or breeding location. and then to create game-theory models based on these decisions to determine the probability of starving or breeding successfully at different population sizes (Goss-Custard. 1985. 1993; Sutherland & Parker. 1985). Game theory is necessary because the decisions individuals make in response to the decisions made by other individual~have importam limess consequences. For example. more individuals may decide to de-rer breeding as population size increases and competition intensifies.
thus redUcing per capita reproductive performance across the population as a whole. By running game-theory models that incorporate such processes at difFerem population sizes. it is possible to predict how mortality or breeding output changes with total population denSity. These behavioural models have the advantages that a change in the environment. such as in the average quality of the habitat. can also be incorporated; thus. the effect of both habitat loss and habitat change can be explored. Two main classes of models arc available. Ideal free-distribution models describe the distribution of individuals between patches within a site (Fig. 16.3). Assume that patches difler in a quality that afFects their 'suitability'; for example. they
A
B
Worst
Number of competitors in patch
Fig. 16.3 The ideal free distribution. This shows three patches which differ in 'suitability';
thiS could. for example, be intake rate. As the density of competitors increases the suilability dedines. The best patch is occupied until a density 'A' is reached where the suitability is the same in the best patch with many competitors as in the intermediate patch with none. At this point the intermediate patch will sian being used. As the denSity in these patches increases further, 'B', the suilability. declines until the point is reached when it is worth individuals occupying the worst patch. (Modified from Fretwell, 1972.)
380
CHAPTER 16
may differ in food density such that average intake rate differs between patches. Assume also that there is some form of negative feedback so that the suitability declines as the number of competilOrs increases. At low population sizes, all individuals occur in the best patch. As the population increases, the suitability in the best patch, where there are many compelitors, becomes equal to thai in the nexl palch. where there arc none. From then on as total numbers increase still further, individuals occupy both patches and the mean rewards at any given point arc equal in both. The ideal despotic distribution (Fig. 16.4) is similar but subtly different. In this case, the order in which individuals occupy a patch is imponant because of. [or example, territorial defence or dominance hierarchies. In this ca'e, the mean rewards may differ but the suitabililies for the best territory of the next position in the hierarchy should be equal in all patches. Negative feedback is an essential component of these two models and can have a number of origins, such as terrilOrial exclusion, resource depletion and interference. Both models assume that individuals possess perfect knowledge of the quality of each patch. In reality, animals never possess perfed knowledge, but [or some species this approximation is probably much benerthan [or others, depending both on Iheir memory and their ability 10 sample the environment. For example, wading birds may return to the same estuary year after year and, by capturing numerous prey, build up a reasonable picture of the habitat. By contrast, some invertebrates may live only [or a few hours or days and catch or parasitize a few victims. This is important, as the ideal free and despotic distributions assume that individuals are able 10 respond 10 bOlh patch quality and competitor density. If they arc unable to sample or evaluate the alternatives available, the models are not appropriate. 11 is probably much easier for individuals 10 evaluate feeding areas in the non-breeding season than breeding areas because the quality of breeding habitat
Order of occupancy in patch
fig. 16.4 The ideal despotic distribution. This is similar to (he ideal free distribution except that individuals defend areas within a patch and exclude others. Hence. as further individuals settle in a patch, their SUitability will decline ~ven though the suitability of the first to settle will be unaHeeted. The dashed line shows when individuals would gain as much (rom settling in the beSt area within the worst patch as (rom a poor area in the best patch. (Modified from Fretwell. 1972.)
BEHAVIOUR AND CONSERVATION
381
may only emerge a long time after the decision is taken. Breeding locations are often chosen early in spring, yet the most important facLOr determining breeding success may be the abundance of food or predators or diseases later in the summer when the young are being raised. Although yellow-headed blackbirds, Xanthocephalus xanthocephalus, feed their young largely on odonates, the location of males and the choice of female terriLOries are unrelated LO prey abundance, perhaps because the terriLOries are chosen before the odonates emerge (Orians & Wittenberger, 1991). Similarly, the risk of eggs and young being taken by predaLOrs is probably difficult for individuals LO determine when they are choosing where LO breed because the predaLOrs may be nocturnal and scarce and predation risk may depend upon factors, such as the structure of the environment and the abundance of alternative prey, some weeks later. Kentish plovers, Charadrius alexandrinus, in Hungary switch back and forth between alkaline grasslands and drained fishponds even though the breeding success is twice as high in rhe grasslands due to the lower predation risk (Szekely, 1992). The suggested explanation for this is that the food availability is much greater in the fishponds and the birds are unable LO assess the higher nest predation risk. Wheareas feeding arcas are sampled continuously, nest predation risk, being the probabUity of a single catastrophic cvent occurring, is hard to estimate, espccially a fcw weeks hcnce. This differs from the non-breeding scason whcre dccisions tcnd to have immediate consequences. Such problems in evaluating breeding season habitat quality not only present difficulties for the animals themselves but also make breeding season slUdies more difficult for the behavioural ecologist interested in deriving density dependent functions. Most progress has therefore been made for the nonbreeding season where the payoffs seem more immediate and easierlo evaluate. We therefore illustrate the approach LO measuring density dependence from empirical game-theory modelling by refercncc to work done in that season.
16,3,2 Measuring density dependence in the non-breeding season
Introduction When the breeding season ends and young are eirher completely independent or sufficiently mobile to keep up with the group. many species abandon their breeding territories. This section is concerned with such animals; for rhose that remain terriLOriaL we assume that the ideas developed later in this chapter apply. Although the parents in many free-ranging species continue to care for young, we assume for simplicity that the main goal of individuals is JUSt to survive tht:: non-breeding season even though, in reality, they must
do this in sufficiently good condition to return to the breeding art'as. For the same reason, we discuss food acquisition with lillie reference to the
382
CHAPTER 16
trade-offs animals must often make between foraging and redudng the risk from enemies. If the population of either herbivores or carnivores is large enough, depletion will reduce food density by the end of the non-breeding season to the lhreshold level at which foragers will be unable 10 feed fast enough 10 survive; in Fig. 16.5, this is where the initial population is 3'. If all individuals are equally e[fidem at foraging, all will starve simultaneously when the threshold is reached. With smaller initial populations, the threshold will nOl be reached and all individuals will survive. With larger initial populations, all will again starve but, as depletion is more rapid, they will die sooner. Such scramble competition through depletion gives the all-or-nothing density dependelll mortality function, with a slope of infinity, shown in the inset to Fig. 16.5(a). In reality, all individuals will seldom be identical. [f some animals are able to secure food at the expense of others, so that individuals vary in competitive abitity, contest competition will arise. In foraging, compelitive ability comprises
fa)
fl[
(b)
n 2030405n
elL n 203n4n5n
Initial population
Initial population
n n 2n . - Threshold intake rate --.
Start
End
Start
End
Stage of non-breeding season
Fig. 16.5 How introdudng variation between individual animals reduces the slope of the density dependence. In (a), all indlviduals are equally eHident at foraging and contesling food during the non-breeding season. The result is scramble competition in which all
animals starve once the population at the start of the non-breeding season is large enough (3n) 10 deplete the food to the level at which the animals cannot feed fast enough to meet their energy requiremenlS. Once the initial population size reaches this point. all individuals die and the slope of the density dependence is infinity (insel). In (b). there is variation between individuals. as shown by the hemispherical frequency distributions of individual intake rates. As the food is depleted and the average intake falls. some poorly performing individuals starve (shown by shading). thus relieVing the pressure on the food supply. With such contest competition. some good performers survive the non-breeding season even though greater than 3n individuals occupied the habitat at the ~tart. The slope of the density dependence is thus reduced (inset).
BEHAVIOUR AND CONSERVATION
383
two characteristics. Foraging efficiency is the intake rate an individual achieves in the absence of competitors (Fig. 16.6a). Inefficient foragers will reach the threshold first and die. The subsequent rate of depletion and/or interference will then slow down, enabling the more efficient individuals to survive; case 4n in Fig. 16.5(b). Mortality is still density dependent, but the slope is no longer vertical (inset to Fig. 16. Sb). Individuals also vary in their susceptibility to interference, the rate at which intake rate changes as competitor density increases (Fig. 16.6b). Dominants may not be affected by interference because other animals avoid them; only the subdominants thus bear the time-cost of increased interaction rates as forager density increases. Where there is food stealing (kleptoparasitism), dominants may even increase their intake rate as forager density increases because there are more opportunities to rob subdominants. This again increases the variation in performance and reduces the slope of the density dependent function. By allowing subdominants to escape
Fig. 16.6 How individual foragers may vary. Each line
refers to the intake rate of an individual at different densities of
competitors. In (a), all
individuals are equally susceptible to interference once the density of competitors has reached the density at which
(bl
intake rate is reduced. Individuals differ in foraging erficiency because. before the
threshold is reached. their intake rales differ. In (b), all individuals have the same foraging effidency but their response to increasing competitor density above the threshold differs so their individual susceptibilities to interference differ. The positive slope refers to an individual which is able to steal food increasingly often as density. and Low
Competitor density
High
thus the opponunity to steal, increases.
384
CHAPTER 16
areas of high competitor density by moving to poorer feeding areas with few competitors, spatial variation in food density increases the variation in individual performance and further reduces the slope of the density dependent function. An individual's foraging elficiency determines how successful it is in responding to a reduction in the density of available prey. By definition, its susceptibility to interference will determine its response to interference. Behavioural ecology provides a framework for exploring the costs and benefits of such individual variation in behaviour through the principle of individual optimization. The optimal solution for one individual in a given environment differs from that of another because its own particular combination of efficiency and susceptibility to interference subjects it to different constraints. The following section illustrates individual variations in foraging efficiency and susceptibility to interference and their filness consequences.
Individual variation in foraging efficiency This arises from differences between classes of animals within the population and from differences between individuals within a class. Class differences have been widely studied and have been related in birds, for example, to age (e.g. Sutherland el al., 1986) and sex (Durell el al.. 1993) and are often connected with morphological factors (e.g. Gosler, 1987) that affect efficiency (Marchetti & Price, 1989). Individual variation in foraging efficiency has been slUdied in birds (Partridge, 1976), but not frequently. An example from other vertebrates is provided by bluegill sunfish, Lepomis macrochirus, a generalist freshwater predator. In an experiment, fish learned to increase their search speed in open water but to reduce it in vegetated areas so as to match the requirements of the particular food supply present. In open water, the prey (Daphnia) are quite conspicuous and unable to avoid being attacked by the fish, which learned to pursue Daphnia quickly. But, in vegetation the Daphnia are cryptic and become motionless or hide when fish are nearby and can also escape when detected by fish which therefore learned to hunt more stealthly. Individual fish were not equally nexible in adapting to the change in behaviour required and thus differed in the searching tactics they adopted which, in turn, affected the efficiency with which individuals foraged in the two habitats (Ehlinger, 1989). In an experimental study of zebra finches, Taeniopygia guttala, individuals varied considerably in how successfully they selected between patches that provided quite different rates o[ energy gain. This between-individual variance was partly accounted by a heritable component (Lemon, 1993). In vew of their scarcity, more experimental studies of the causes of such individual variation would be worthwhile. lvo field studies on mammals have stressed the fitness consequences of foraging efficiency. Free-living Columbian ground squirrels, Spermophillls columbial1us, varied in their ability to select the diet that maximized daily energy consumption. Body size and the feeding time available constrained what was
BEHAVIOUR AND CONSERVATION
385
Ihe optimal diel for each individual. 63% of whom consumed Ihe diel Ihat maximized Iheir daily intake. The remaining 37% approached an energy maximizing diet bUl made some incorreCI foraging decisions, and so failed 10 reach it. These individual differences, which were consislelll across seasons and were nOl relaled 10 social faCiors, seemed large enough 10 have fimess consequences, ahhough these could nOI be studied in detail (Ritchie, 1988). However, individual differences in foraging efficiency in Soay sheep, Ovis aries, during an overwinter populalion crash caused by overgrazing, did have dear fimess consequences (Illius et al.. 1995). Individual varialions in survival were related to variations in the breadlh of Ihe incisor arcade; a big moulh is advalllageous when food densily is low. In accordance with Ihe increasing number of studies poinling to Ihe imponance of parasiles 10 hosl fitness (Nelson, 1984; M01ler et al.. 1993; Loehle, 1995), survival during Ihe crash was also rdaled 10 Ihe gUl parasile burden. 1Ilius et al. (1995) argue Ihal narrow incisor arcades nonetheless persisl in the population because, when food is more plellliful allow population densily, Ihey can increase thei.r fitness by feeding more precisely, and thus more seleclivelY, Ihan individuals wilh a broad incisor arcade. These two sludies illuslrale thaI funher work on the foraging differences bel ween free-living individuals, and thus on Iheir variable response 10 increased rales of deplelion and inlerference as populalion size increases, would be wonhwhile.
Individual variation in susceptibility to interference In kleptoparasitic syslems, the success of individuals stealing food will, by definition, be relaled to dominance, since dominance is measured as the percelllage of encoulllers won. Pusey and Packer (see Chapler 11) discuss the many correlates and Ihe COSIS and benefits of dominance. Suffice to say Ihat fitness benefits in Ihe non-breeding season have oflen been identified. For example, the more dominalll families in wintering geese feed in Ihe more profitable feeding areas and parts of the nock and, apparently as a consequence, leave for Ihe breeding grounds in beller condition, Ihereby increasing their subsequent chances of survival and of breeding successfully (Ebbinge, 1989; Black & Owen, 1989; Prop & Deerenberg, 1991; Black et 01., 1991). However, Ihe possibility that dominant individuals are also particularly efficielll at foraging has nOl been investigaled in such sludies. Allhough many slUdies have relaled intake rate 10 dominance, few have described the interference funclion and eSlimaled susceplibilily 10 interference. In seed-eating snow buntings, Plectrophenax nivalis, the relalionship appears dome-shaped. At low densilies, birds feed ralher slowly, perhaps because of the need to be vigilalll, whereas al higher dt:nsities, imake rale declines. Imerference apparently arises because birds supplam olhers from feeding siles. Interference onLy occurs when food is scarce, presumably because it is not cosl-effeclive, or worth the risk of being damaged, for dominams 10 supplam olhers when food is abundant (Dolman. 1995).
386
CHAPTER 16
For understanding density dependence, the important question is how susceptibility to inlerference varies between individuals. So far, Ihis has only been measured in oyslercalchers, Haematopus ostralegus, ealing mussels, Mytillls edulis. Dominants attack subdominants increa ingly as bird denSity rises and often sleallheir mussels. Intake rate in subdominams decreases with compelitor densily buI, as in Dolman's (1995) study, only above a Ihreshold densilY. The Ihreshold is unrelated to bird dominance but the slope decreases with domi· nance and even becomes positive in top dominants (Stillman et al.. 1996). Interference does not arise solely because more mussels are stolen as density increases. Ralher, il is mainly due to Ihe lower rate at which subdominants capture mussels in the first place (Ens & Goss-Custard, 1984), perhaps because they spend more time avoiding olher birds (Vines, 1980). There is a clear need for more empirical studies of how interference really operates (Ens & Cayford, 1996). Habitat quality
Habitats will, of course, vary in qua lily with some patches being beller than olhers. The intake rate Ihat individuals experience will depend upon a complex interaction of the quality of the chosen patch, ils foraging eHiciency, the amount of competition within the patch and its susceptibility to interference. The ideal free distribution (see Fig. 16.3) describes patch choice and incorporales differences in competitive ability. There is a range of models based on the ideal free distribution which make differenl assumptions about the shape of Ihe interference function (Ruxton et al., 1992; Holmgren, 1995; Moody & Houston, 1995; Goss-Custard et al., 1995a,c; Clarke & Goss-Cuslard, 1996) and the manner in which individual differences are incorporaled. One way 10 model such differences is to incorporate the decisions of individual animals (Holmgren, 1995; Moody & Houston, 1995; Goss-Cuslard et al.. 1995a). Another is to consider classes of animals (for example, individuals differing in dominance or competitive ability) and determine the strategy by which individuals of each competilive class will obtain the highesl intake rate (Sulherland & Parker, 1985; Parker & Sutherland, 1986; Sutherland, 1996a). All individuals of a given class will oblain the same intake rale - otherwise they would movebut individuals of different classes will differ in intake rale. 8UI, however it is modelled, interference and depletion will be grealer at high population levels and most or all individuals will obtain a lower intake rale as a result of a combinalion of both increased interference within each palch and a greater use of poorer palches (Goss-Cuslard, 1993; Sutherland & Dolman, 1994). Deriving density dependent functions from ideal free models
Models that predicl forager mortalily rate as
BEHAVIOUR AND CONSERVATION
387
1994; Goss-Cuslard et aI., I995a,b) or use up aillheir energy reserves (GossCustard el al., 1995c). Such models generate density dependent mortality functions by calculating mortality al a range of initial population sizes. In an empirical individuals-based model of oysLercatchers eating mussels on the Exe estuary in England, each bird has its unique combination of foraging efficiency and dominance (Clarke & Goss-Custard, 1996). Some birds starve late in the non-breeding season because of depleted food stocks and high, temperaturerelated energy demands. Individuals with quite high rank but low efficiency die, whereas those with above-average efficiency seldom do so (Fig. 16.7). This suggests that efficiency is often more important than dominance in determining a bird's survival chances, as empirical studies have already indicated (Goss-Custard & Durell, 1988). This underlines the need for further studies, like that of IIlius el al. (1995). on the selective processes determining average efficiency and on the causes of variation between individuals. Surprisingly little is known about the selective constraints on further increases in foraging efficiency and about the causes of individual variation. As population size, when food is limiting, is so sensitive LO the foraging efficiency of individuals
1.0
*
J
0.8
.. ''cc
E 0
..' .0 0
'
*-l9'
~
0.6
0.4
0.2
0.0
* *
*
~~ *
~*
*** * L-,:-.::*''::! -3
-2
-,
0
,
2
3
Foraging efficiency (SO from the mean)
Fig. 16.7 The characteristics of lhe individuals that slarve during the winter, as predicted by the individuals·based model of oystercalchers foraging on the mussel beds of the Exe
estuary. Each symbol shows the dominance. and thus susceptibility to interference. and foraging efficiency of a bird that starved to death at some point during the winter when its energy reserves fell to zero. An individual's eHiciency is measured in empirically determined standard devialions of the interference-free intake rates of a sample of individuals feeding on the same food supply. Global dominance is a hypothetical. estuarywide quality that. for example, may reflect an individual's fighting ability. The model assumes that the value for each bird stays the same on all mussel beds but determines it's local dominance, and thus itS susceptibility to interference, on anyone mussel bed. An individual's local dominance increases as, on the mussel bed where it is feeding. the proportion of birds on the same bed Ihat have lower global dominance scores to its own increases. (From Goss-Custard et oJ., 1995a.)
388
CHAPTER 16
(Goss-Custard it al., 1996a), more studies on the d~terminants of foraging efficiency, to parallel the effort invested in studies of dominance, are required if behavioural ecologists are to contribut~ all they can to predicting population responses to habitat loss and change. The mortality rate from starvation increas~s with initial population size in the oystercatcher-mussel model (Fig. 16.8). The density dependence is quite gradual, as would b~ expected where individuals vary so widely in competitive ability, and patch quality varies threefold. For ecological prediction, this function would only apply if habitat of average quality was lost, because it changes considerably if habitat of above- or below-average quality is lost (Fig. 16.8). As more of the better-quality habitat is removed, d~nsity dependence begins at a lower density. and mortality rate rises more steeply. Clearly, it is important to know whether habitat of average or atypical quality is to be lost if population predictions are not to be nawed seriously.
100
30% best
80
. , . .
60
20
.
'
30%wOrS! •• ,
Present
·--~
jJ
.-....
f
dav
Daily consumption
, ,
. '~
f
Dailv consumptIon
2000
4000
6000
8000
10000
12000
Initial population size
fig. 16.8 Density dependent starvation functions as predicted by the game-theoretic
model of oystercalchers feeding on the mussel beds of the Exe estuary. The solid line shows the percentage of birds thaI, with different initial population sizes. are predided to
die at some point during the winter with the presenl·day mussel beds of the estuary. The frequency histograms show the distribution of the daily consumption rales of all
individuals in the population, averaged over a full spring-ncaps' 14·day cycle in October, before many birds have died. As the initial population size increases, an increasing proponion of birds have low consumption rales in autumn and so fail to accumulate the energy reserves needed to survive the poor feeding conditions that arrive later in the wimer; (he numbers that die are indicated by the hatching. The dashed Jines show the density dependent functions Ihal arise if 60% Or 30% of the beSt feeding areas. or 30% of the worsl, are removed. (From Gos~-Custard tl aJ.. 1995c. 1996a.)
BEHAVIOUR AND CONSERVATION
389
16.3.3 Measuring density dependence in the breeding season
In many species, individuals either defend space and exclude others or exhibit dominance hierarchies within groups. The ideal despotic distribution (see Fig. 16.4) is often an appropriate framework for considering such breeding systems. As densities increase, individuals may: (i) defend smaller territories; (ii) occupy poorer breeding areas; or (iii) refrain from breeding altogether. All of these may result in reduced breeding output per capila at high population sizes. There are field examples of each of these three mechanisms that lead to density dependent production. The red grouse, Lagopu.s lagopu.s, studied by Watson and Miller (1971) provide an example of territory size decreasing with population size. All the suitable habitat was occupied by grouse at both low and high population densities; thus, at high densities individuals possessed smaller territories than at low densites. Evidence for the second mechanism, the buffer effect (Brown. 1969), comes from a number of species. For example, when nuthatches, Sitla europea, are scarce the mean territory quality is greater than when the population is high (Nilsson, (987). Similarly, female'red squirrels, Sci/aus vulgaris, only occupy poorer quality territories at high population densities, and these territories produce lewer yOllng (Wauters & Dhondt, 1989). A number of studies have shownthatjuveniks and poorer quality individuals tend to occupy the poorer territories, further reducing per capila reproductive rate at high population sizes, For example, blackbirds, Turdus merula, breeding in small patches of woodland were more likely to be juveniles and sllffered greater nest predation and starvation than the birds breeding in large palches (M011er, 199 I). Refraining from breeding at high population sizes has also been recorded (Smith & Arcese, 1989). For example, many species, especially mammals, breed in social groups in which only the highest ranking females breed (Macdonald, 1979), with the result that, as the populalion increases, a lower proportion of individuals breed. In many cases, there may be more than one of these mechanisms opera ling to produce densily dependence. In oystercatchers breeding in territories on the island of Schiermonnikoog in the Netherlands (Ens el 01., 1992, (996), both buffering and rdraining from breeding occur. There are two main strategies of territorial defence. Some birds, known as residents. defend lerritories along the edge of the saltmarsh and walk with their chicks straight onto the mudflats to feed. Others, known as leapfrogs, ddend two territories: one on the saltmarsh but inland of the resident territories and another on the mudflats but seawards of the resident territories. Leapfrogs thus have to fly back to the nest with each food item to feed the chicks and, as a result, have a considerably lower breeding success (between one-half and one-sixth, depending upon the year) of the residents, many adult oystercatcher birds refrain from breeding even though they are physiologically capable of doing so (Ens el 01., 1992). The explanation seems to be that some birds defer breeding in order to obtain a high-quality territory. Ens el 01. (1996) show that most of these delayed breeders are birds
390
CHAPTER 16
that eventually become residents. The mean expected lifetime reproductive success is very similar for the two strategies. The leapfrogs. on average. breed for more years but with a low average success. whereas the residents usually delay breeding but then. on average. have a higher annual success. These strategies are then frequency dependent. At the evolutionary stable strategy. the mean lifetime reproductive success will be the same for residents and leapfrogs. If a higher proportion of individuals were to wail 10 become residents. for example. the leapfrog strategy would become more advantageous and more birds would then become leapfrogs. The data from this study can be used to derive a densily dependent breeding function for oystercatchers (Sutherland. 1996a). Assume that there is a fixed number of resident and leapfrog territories. By assuming that individuals adopl the strategy that results in the highest lifetime reproductive success. the number of individuals waiting for an opportunity to gain a territory of either type and to breed. along with the numbers breeding in resident or leapfrog territories. can be calculated [or a range of population sizes. At low population sizes. all individuals are residents and thus have a high per capita breeding success but. as numbers increase. some individuals become leapfrogs and others defer breeding. Both of these strategies result in a reduction in per capita breeding success as numbers increase and thus lead to a density dependent reproductive rate.
16.4 Population size and the effects of habitat loss As Ihe effect on equilibrium population size of habilat loss in one season depends on the strength of the density dependence in the other. the winter density dependent function does nor. of itself, tell us how the population will respond. From the equations in Box 16. I. the theoretical expectation is that the population should decrease in the proportion d'/(b' + tf). The values for oystercatchers of tf (for the Exe estuary) and b' ([or Schiermonnikoog) in the vidnity or the present-day population size were estimated from the work reviewed above. and are 0.000 I t and 0.00005. respectively (Sutherland. 1996b). The effect of the loss of winter habitat of average quality on population size is therefore 0.69 and is thus subproporrional to the proportion of the winter area lost. This is confirmed by a simulation slUdy of the world population of the Palearetic subspedes of oystercatcher. H. ostra/egus os/ralegus (Coss-Custard et al.• 1995d.e). Under a wide variety of assumptions of winter denSity dependence, rhe reduction in equilibrium population size is subproportional 10 the amount of winter habitat lost across the whole range of habitat removal. at least until all has gone and the population goes extinct (Fig. t6.9). The impact of winter habitat loss is influenced by the strength of the summer density dependence (Fig. 16.9b). but. in these particular simulations, by rather a small amount (Coss-Custard et al.. 1995e).
BEHAVIOUR AND CONSERVATION la} 100
A __
,:
8 -----
,..~..~.~/
, ....,.,'
D··········
~
.~
;)~.. .. /
c
.,;; .2
0. 0 0.
.~ c
60
, ~.... ;
40
0
'13
:>
.'
•
'
:
c--
~ 80 -
;
;
;
;
;
;
'
.'
'
.
;
;
,.~./
~
E __
;
F ----G·_-
80 -
~
;
H··········
.~
c
g
~
.,,:
;
60
0. 0 0.
.~
c
40
;
0
B :>
•
.'
'
.-
; ;
;
20 ;
0
;
;
:>
;
20
Ibl 100
391
0
;
;
;
;
;
;
;
;
;
;
;
;
;
;
;
......
,,' :.~.:., ..
.
,,' :
.
.' .
~.',~
20
40
60
80
100
Loss of winter habitat ('0)
Fig. 16.9 The proportional reduction in the equilibriunl population size of oyslerc3lchers resulting from an increasing loss of winter habitat. as predicted by an empirical demographical model of the European population (a,b) and of the Exe estuary (c). The heavy, dashed diagonal line shows the line of proportionality along which a given percentage reduction in habitat area results in the same percentage reduction in equilibrium population size. In (a). summer density dependence. bT, is set at 0.5 but the strength of Ihe winter density dependence. bW, is increased from 0.001 (A) through 0.01 (B) and 0.05 (C) to 1.0 (0), fn (b). the winter density dependence is set at 0.1 and the strength of the summer density dependence is increased from 0.3 (E) through 05 (F) and 0.7 (G) to 0.9 (H). In (c) the change in population size resulting from reducing. in successive steps of 10%. the area of the mussel bed feeding grounds. starting with either the worst or the best quality beds. At each step, the appropriate density dependent wimer mortality function for that food supply, as derived from the game-theoretic model simulations illustrated in Fig. 16.8. was insened into a demographical model of the population in which the strength of the summer density dependence. bT, was assumed to be 0.5 (From Goss·Custard
~I
al.. 1995d,e. I 996b, where Illodel construction. paramett'r
values and the unils used in (a-c) are detailed.)
392
CHAPTER 16
The conclusion that the reduction in equilibrium population size is subproponional to lhe amount of winter habitat lost depends critically. however, on the assumption that habitat of average quality is removed. The effect of removing habitat of above- or below-average quality was simulated by inserting the density dependent functions derived from the individuals-based model for oystercatchers (see Fig. 16.8) into a demographical population model (GossCustard el al.. 1996b). With the besl habitat being removed firsl. the reduction in equilibrium population size now becomes supraproportional (Fig. 16.9c). This simulation shows the inadequacy of using present-day densily dependent functions 10 make predictions when habitat other than of average quality is being lost. and underlines the importance of being able to devise a methodology to predict the form and parameter values of density dependent functions in new environments (Goss-Custard el al., 1995c).
16.5 Flexibility and constraint
in response to habitat loss
Global environmental change is probably proceeding at an ever increasing rate, resulting in both habitat loss and habitat deterioration. There is also the possibility of latitudinal shifts in ecosystems due to global wanning. One key issue is the extent to which populations can respond to these changes and. in the present context in particular. whether populations can readily alter migration roules where necessary. The world environment is in a continual slate of gradual nux, for example, due to lhe movement of continental plates and the succession of ice ages and il is inevitable thaI species have undergone dramatic changes in distribution and migration routes. The populations of most birds in tempera Ie regions. for example. must have arrived since the last ice age 10 000 years ago, and the current migration routes must have evolved in
I hat
time so evolution in migration routes must be possible over such
relatively short periods. There are examples both of very rapid changes in migratory behaviour and of the apparent failure of species to adapt. The clearest example of how a species may respond to environmental change is the study by Berthold et at. (1992) of the the blackcap. Sylvia alricapila. Berthold el al. showed that much of the variation in migratory behaviour between or within populations is determined genetically and thus subject to selection. By an eXlraordinary coincidence, over the period during which the capacity of blackcaps to change migratory behaviour was being studied, a change did occur. Blackcaps are common breeders in Brilain but. until recent decades, were scarce in winter. The wintering blackcaps were initially assumed to be British breeding birds which had become resident. but ringing studies showed thaI, in fact, they bred in Germany and Austria. By capturing blackcaps wintering in Britain, Bert hold el at. (1992) showed thaI their migratory direction differed from lhuse
BEHAVIOUR AND CONSERVATION
393
from southwest Germany. They then showed that this orientation behaviour was genetically determined because crossing the two produced offspring that orientated in an intermediate direction (Helbig el al., 1994). Olher species have also changed their migratory behaviour. Over recent years, an increasing proportion o( great crested grebes, Podiceps crislalus, in the Netherlands have become sedentary (Adriaensen el aI., 1993). Serins, Serinus serillus, have expanded their range this century (rom the Mediterranean to central and northern Europe and in doing so have changed /rom being sedentary to migratory (Berthold, J 993). Such studies may suggest that changes in migratory behaviour can occur both readily and rapidly. There is, however, evidence that other species may be very slow to aller migration routes. All red-backed shrikes, Laniuscollurio, migrating (rom Europe 10 A(rica cross the eastern side o( the Mediterranean, including those in Spain that start their migration to Africa by /lying northeast! Perhaps there is an adaptive explanalion for Ihis migration route but another possibility is that they have simply (ailed to evolve a new and more direct route. Similarly, all wheatears, Omamhe omanlhe, migrate to A(rica in the winter even though the species has spread in both directions (rom its European breeding ground to eastern Canada in the west and across Asia to western Canada in the east. Were they to migrate directly to Asia or central America, respectively, these newly-established populations could halve their migration distances. Although A(rica may be the only continent with suitable wintering habitat (or wheatear, perhaps a more likely explanation is that they have not evolved a change in route. We therefore do not have a clear picture yel o( how species will respond to changes in the distribution o( their habitats at a global scale and it does not necessarily (ollow that animals will be able to adjust rapidly to a change in the world distribution o( their resources.
16.6 Discussion The science o( predicting densily dependent functions for the present day and the novel environments broughl about by habitat loss and change is still in its infancy and the range of empirical and theoretical sludies with which to illustrate the approach is limited. However. the increasing demand from Ihe community at large that ecologists should more willingly address environmental concerns may stimulate more interesl in the population consequences of behavioural decision rules, the identification of which has dominated the first 25 years of modern behavioural ecology. This subject began largely because Robert MacArthur, an evolutionary ecologist, wanted to make sense of such ecological concepts as the niche. This reqUired a quantitalive understanding o( the rules of animal design and their fitness consequences. Since then, much of the ecological lalent of a generation has been devoted to pursuing this aim. The increasing interest in using the resulting understanding for the purpose o(
394
CHAPTER 16
predicting animal numbers in new environments should bring closer together the two disciplines of population (and community) and behavioural ecology. as MacArthur always intended. Attempts to do this will undoubtedly run into the major dilemma in ecology. Ecological models that are surnciently detailed to predict the population consequences of environmental change in a particular case are usually complex. contain estimates of some of their many parameters that are no better than guesses and give predictions that are lhought to be situation-specific. Models that are general yet simple and claim to capture lhe essence of the problem are difficult to test and generate predictions that many do not believe because. they suspect, some vital part of species natural history has been ignored for convenience and analyticaltractabilily. There seems to be no solution to this dilemma other than through the pragmatic and heuristic interchange of results between theoreticians and empiricists. As this chapter has shown, the main conclusions of an empirical model and those of a highly simplified general model can be encouragingly similar. Although the empirical model summarized here is rather complex compared with many theoretical models, it is nonetheless still very simple in an absolute sense. The need for lime-consuming simulations arises [rom the decision to follow individuals (Goss-Cusrard et at.. 1995a). Individual-based models are becoming popular in population ecology. partly because modern computing power is available and panly because, for many scientists. they do provide the most convincing way in which to represent lhe real world. Even so, the simulations are cumbersome, slowing down progress. The potential for adding further realism and complicating models still further is considerable. Por example, introducing state-dependent decision making in a stochastic environment (Mangel & Clark, 1988) will complicate the models enormously, making their operation and properties diflicult to understand. Nonetheless, it will be important in models that predict mortality rates to include the trade-orrs animals make between, for example their rate of food intake and risk from predation, and how the trade-off is affected by their body condition (see Chapter 5; Houston, 1993; Goss-Cuslard et al.. 1995c, 1996a). (Clearly, such trade-offs also have important implications for how individual variations in dficiency and susceptibility to interference are measured in the field and included in empirical models.) The solution to this continuing dilemma between simplicity and realism may lie in using behavioural ecological models to derive the decision rules to be used by animals in other models that are designed to predict the population consequences of environmental change. By providing a firm basis for predicting how individual animals would be expected to behave in such new environments, behavioural ecology will be able in lhis way to contribute significantly to the dirncult lask of ecological prediction. However. we finish with a note of warning. Malthus was an early ecological predictor who underestimated the capacity of humans to increase lhe world food supply. This illustrates one major problem of prediction. It assumes that
BEHAVIOUR AND CONSERVATION
395
we understand how individuals will respond (() a changed environment. In fact. a spedes may be able to draw on adaptalions that it has nOI yet displayed. In the 1960s. the populalion of the dark-bellied brent goose. Branta bernicla bernicla. was very low and was restricted 10 feeding on inIenidal plams. especially Zostera spp. (Ogilvie & Mathews. 1969). II was argued during the 1970s that the proposed airport on the Foulness intertidal flats in England would have had serious consequences for the population as it would have resulled in the loss of a substantial proportion of the feeding area in Britain. Subsequent work has shown Ihal the population was low at that time because of hunting and it has since increased markedly as a result of conservation measures (Ebbinge. 1991). As the populalion increased. the geese fed on a wider range of species and have. indeed. started to reed inland. Their abilily to do this had not been foreseen. Such an example of unexpected shihs in behaviour may be the exception but it is clear that the approach described in this chapter not only requires appropriate modelling but also a thorough understanding of the adaptive reperloire. or the nawral history, of the species.
References
Abrahams. M.V. & Healey. M.e. (1990) Variation in
the competitive abilities of fishermen and its influence on the spatial distribution or the British Columbian salmon troll fleet Can J F;sh Aquat Sci. 6.1116-21. Abrams. P.A. (J 982) Functional responses of optimal
foragers. Am Natur. 120. 382-90. Abrams. P.A. (1993a) Why predation raLe should not
be proponional to predator density. Ecology. 74, 726-
n.
Abrams. P.A. (1993b) Optimallrails wh~n there are several costs: the interaoion of mnnalily and energy costs in determining foraging behavior. 8ehav E(ol. 4,246-53. Abrams. P.A. (1995) Overestimation versus under·
estimation of predation risk: a reply to Bonskila et al. Am Natur, 145. 1020-4. Adams, E.S. & Mesterton·Gibbons. M. (1995) The COSt of threat displays and the stability of deceptive communication. J Thtor Bioi. 175.405-21. Adriacnsen. F.. Ulenaers. P. & Dhont. A.A. (1993) Ringing recoveries and th~ increase in numbers of European great crested grebes (Podiceps cristatus). Ardra. 81, 59-70, Alalalo, R.V. & Lundberg. A. (1984) Density dependence in breeding success of pied flycatcher. J Ani', Em/, 53. 969-77. Alatalo. R.Y.. Carlson, A. & Lundberg. A. (1988) The search cost in mate choice of the pied t1ycatcher. Anim Bthav, 36, 289-91. Alberts, A.C. (1992) Constraints on the design of chemical communication systems in terrestrial vertebrates. Am Nnwr, 139. S62-589. Alcock, J. (1993) Animal &Jzavior. 5th edn. Sinauer. Sunderland, Massachusetts. Alerslam. T. (1996) The geographical scale factor in orientation of migraling birds. .J Exp 8iol. 199,919.
Alerslam. T. & Lindstrom. A. (I 990) Optimal bird migration: the relative imporlance of time, energy and safelY. In: Bird Migration: Physiology and
396
Ecophysiology (ed. E. Gwinner). pp. 331-5 I. Springer· Verlag, Berlin. A1er.Ham, T. & Pcltersson, S.-G. (1991) Orienlalion along great circles by migrating birds using a sun compass. J Tilt'or BioI. 152. 191-202. Alexander. R.D. (1979) Danvinism and Human Affairs. University of Washington Press, Seattle, Washinglon. Alexander. R.D. (1990) Epigenetic rules and Darwinian algorithms: Ihe adaptive study of learning and development. Ethol Sociobiol, 11, 241-303. Alexander, R.D. (1991) Social learning and kin recognition: an addendum and reply 10 Sherman. Ethol Sociobiol, 12, 387-99, Alexander, R.D. & Borgia. G. (1978) CroliP selection. altruism. and the levels of organization of life. Am' R,v Ecol Syst. 9, 449-74. Alexander. R.D. & Sherman, P.W. (1977) Local mate competition and parenral investment in social insecls. Stitna. 196. 494-500. Alexander. R.D., Noonan. K.M. & Crespi. B.J. (1991) The evolUlion of eusodality. In: Tht Biology of tltt Naktd Molt-Rat (eds P.W. Sherman. J.U.M. Jarvis & R.D. Alexander). pp. 3-44. Princeton Universily Press, Princeton. New Jersey. Aloimonos, Y. (1993) Adive Perception. L. Erlbaum As.. Hillsdale. Alonso, J.e., Alonso. J.A .. Bautista. L.M. & MuiiozPulido, R. (1995) Patch use in cranes: a field teSI of optimal foraging predictions. Anim &hav, 49. 136779, Altmann. J., Sapolsky, R. & licit •. P. (1995) Baboon fenililY and social status. Naturi Lond. 377, 688-9. Ammerman, A.J. & Cavalli-srorza, L.L. (1984) The Ntolithi' Transition and the Genl'lics of Populations ;11 Europe. Princeton University Press. Princeton. New Jersey. Andersson, M. (1984) The evolulion of eusodality. Ann Rtv Eco! Syst. 15, 165-89. Andersson, M. (1994) Sexual Srlertion. Princt'tlln University Press. Princelon. NJ. Andersson, S. (1989) Sexual selection and cues
REFERE for female choice in leks of Jackson's Widowbird
£liplfdfsjacksoni. Behov Ecol Sodobiol, 25, 403-10. Andersson, S. (1991) Bowers on the savanna: display couns and mate choice in a lekking widowbird.
&hav Ecal, 2, 21 ()-18. Andersson, S. (1993) Sexual dimorphism and modes of sexual selection in lekking Jackson's Widowbird Eupltatsjacksoni. BioI) Unn Soc Lond. 49. 1-17. Andren. C. (1985) Risk of predation in male and female adders, Viptra berlls (Linne). Amphib ReptilM, 6,203-6.
Aoki. S. (1977) Colophi'a clemaris (Homoplem. Pemphigidae), an aphid species wilh ·soldiers'. Konlyii. 45, 276-82. Arak, A. & Enquist. M. (1993) Hidden preferences and rhe evolution of signals. Phil Trans Roy So< Lend B, 340, 207-13. Arak, A. & Enquist, M. (1995) Conflict. receiver bias and the evolulion of signal form. Phil Trans Roy Soc
umd B. 349, 337-44. Arnold, S.J. (1981) The microevolution of feeding behavior. In: Foraging &havior: Ecological. Ethological and Psychological Approaches (cds A.C. Kamil & T.D. Sargent), pp. 409-53. Garland STPM, ew York. Arnold. SJ. (1983) Sexual selection: th~ interface of Iheory and empiricism. In: Mate Choice (cd. P. Baleson), pp. 67-107. Cambridge University Press, Cambridge. Amn. S.. Passera. L.&Keller.l. (1994) Queen-worker connict over sex ralio: a comparison of primary and secondary sex ratios in the Argentine ant, lridomyrmrx Irumilis. J Evol Bioi. 7, 403-18. Aron, S.. Vargo. E.L. & Passera. L. (1995) Primary and secondary Sex ratios in monogyne colonies of the rire anl. Allim Behav, 49. 749-57. ASlheimer. L.B., Butlemer, W.A. & Wingfield, J.e. (1992) Interactions of corticosterone with feeding. activity and metabolism in passerine birds. Omis
Scand, 23, 355-<;5. Alkinson, S.N. &- Ramsay, M.A. (1995) The effects or prolonged fasting on the body composition and reproduaive success of female polar bears (Vrsus maritinws). FUllet Ecol. 9. 559-67. Au. W.W.L. (1988) Sonar target detection and recognilion by odontocclCS. In: Animal Sonar (cds P.E. Nachligall &- P.W.B. Moore), pp. 451-<;5. Plenum Press. New York. Austad, S.N. ( 1983) A game theorelical interpretation of male combat in [he bowl and doily spider. Frontintlla pyramitela. Anim Bellav, 31. 59-73. Austad, S.N. (1984) Evolulion of sperm priority patterns in spiders. In: Sperm Competition and Tilt' Evolution of Animal Maring Sys1l'ms (ed. R.L. Smith). pp. 223-49. Academic Pre-ss. London.
CES
397
Avery. M. (1994) Finding good food and avoiding bad food: does il help to associate wilh experienced flock males? Auim Behav, 48. 1371-8. Avise. J.e. (1989) Gene trees and organismal hislOries: a phylogenetir approach to population biology. Evolution. 43. 1192-1208. Avise, J.C (1994) Molecular Markers. Natural HiSlOT)' and £,,'Olution. Chapman & Hall. New York. Axelrud, R. & Hamilton. W.O. (1981) The evolution of ronperation Science, 211. 1390-6. Backus. V.L. & Herbers. J.M. (1992) Sexual allocation ratios in rorest ants: food limitation does not explain observed pallcrns. Behav Ecol Sodobiol, 30, 425-9. Baker. R.R. & Bellis. M. (1995) Human S,¥rn, Com/Nlition. Chapman & Hall. London. Bakker. T.e.M. (1990) Genelic variation in female mating preference. NetherlJ Zoo/, 40. 617-42. Bakker, T.C.M. l1993) Posilive generic correlation between female preference and preferred male ornament in slicklebacks. Nature. 363, 255-7. Bakker. T.CM. &- Pomiankowski, A. (1995) The genetic basis of female mate preferences. J EvolBiol. 8, 129-71.
Balda. R.P. & Kamit A.C. (1989) A comparalive study uf cache recovery by three (orvid species. Anim
<nv, 38, 486-95. Balda, R.P., Kami!. A.C &- Bednekoff. P.A. (1996) Predic..,ing cognitive capacities from natural histories: examples from four species of corvids. Curr Ornitllol (in press). Ball. M.A. & Parkl'r. G.A. (1996) Sperm competilion games: external fenilization and 'adaptive' infenililY. J Theor Bioi. 180. 141-50. Ballard, D.H. (1991) Animate vision. An Inrrll. 48.5786.
Barker, D.M. (1994) Copulatory plugs and paternity assurance in the nematode Camorltabditis elegam. Anim Behov, 48, 147-56. Barlow. H. {I 9951 The neuron doctrine in perception. In: Tilt' Cognitiw Neurosciences (ed. M.S. Gazzalllga), pp. 415-35. MIT Press. cambridge, Massachusells. Barnard. CJ, (J 984) Producers and ScrouIJgers. Chapman & Hall. New Yurk. Barnard. CJ. &- Sibly, R.M. (1981) Producers and scmungers: a general model and its application IU caplive flocks of house sparrows. Alii', Bellav. 29. 543-50.
Barnell. M.. Telford, S.R. &-l1bbles, BJ. (1995) Female mediation of sperm competilion in the millir~ede Alloporus unictltlls (Diplopuda: Spirostreptidae). Btllav Emf Sorichiol, 36, 4 13-19. Barraclough. T.G.. Harvey. P.H. & N<..'l'. S. (1995) Sexual selection as a cause of laxonomic divt'rsily in passerine birds. Proc Roy Soc u:md B. 259. 21 1-15.
398
REFERENCES
Barrett. S.C.H. & Harder, L.D. (1996) Ecology and t'volution of plant mating. Trmd! Erol £10/. 11. 739. Barton, N.H. (1995) A general model for the evolution (If recombination. Gmtl Res. 65, 123-44. Barton, N.H. & Charlesworth, B. (1984) Geneli r('volutions. founder effects and spedation. Alln Rev Eeol Syst. 15. 133-64.
Bartz, S.H. (1979) Evolution of eusodality in termites. Proc Nat Acad Sci USA, 76. 5764-8. 8a<;010. A.L. (l990a) Female preference for male sword length in the green swordtail. Xiphophoru5 helleri (Pisces: Poeciliidae). Anim Bthav, 40. 332-8. Ba~oJo, A.L. (1990b) Female preference predates the evolution of the sword in swordtail fish. Samet. 2'0, 808-10.
8a5010. A.L. (1995a) A further examination of a
pre-existing bias favouring a sword in the genus Xiphophorus. Ani', Bthav. 50. 365-75. Ba!oolo, A.L. (1995b) Phylogenetic evidence for the role of a pre-existing bias in sexual selection. Proc Roy Soc Lond 8.259. 307-11. Bateman. A.J. (1948) Intra-sexual selection in Drosophilia. Heredity, 2. 349-68. Bateson, M. & Kacelnik, A. (1995) Accuracy of memory for amount in the foraging slarling. StUTntiS vulgaris. Anim BtJrav. 50. 431-43. Bateson, P.P.G. (1978) Sexual imprinting and optimal outbreeding. Nalure. 273, 659-60. Bateson, P.P.G. (1979) How do sensitive periods arise and what are they for? Anim Bthav, 27, 470-86. Beauchamp, G. & Giraldeau, L.-A. (1996) Group fnraging revisited: infonnation-sharing or producerscrounger game? Am Natur. 148,738-43. Beauchamp. G. & Kacelnik, A. (1991) Effeas of Ihe knowledge of partners on learmng rates in zebra finches Tatniopygiasultata. Anim Behav. 41, 247-53. Beauchamp. G.K .. Yamazaki, K.• Bard, J. & Boyse, E,A. (1988) Preweaning experience in the control of mating preferences by genes in the major hlslOcompatibility complex of the mouse. Belrav Gmel, 18, 537-47. Bednekorr. P.A .. Biebach. H. & Kreb<;, l.R. (1994) Greal tit fat reserves under unpredictable temperatures. J Avian Bioi, 25, 156-60. Bednekoff. P.A. & Houston, A.1. (1994) Optimizing fat reserves over the entire winter: a dynamic modd. Oikos, 71. 408-15. Bednekoff. P.A. & Krebs, l.R. (1995) Greal Ii' fa' reserves: effects on changing and unpredictable feeding day length. Funct Ecol, 9, 457--62. Beecher, M.D. (1988) Kin recognition in birds. Behav Gmt/. 18.465-82. Beecher, M.D. (1991) Successes and lailures of l)clrent-
offspring recognition in animals. In: Kin Recognition (ed. P.G. Hepper). pp. 94-124. Cambridge University Press, New York. Beecher. M.D.. Beecher, I.M. & Hahn. S. (198Ia) Parent~ffspring recognition in bank swallows (Riparia riparia): II. Development and acoustic basis. Anim BtJrav, 29, 95-101. Beecher. M.D.. Beecher, I.M. & Lumpkin. S. (198Ib) Parent-orrspring recognition in bank swallows Riparia riparia: 1. Natural history. Anim Behav, 29. 86-94. Bell, G. (1978) The handicap principle in sexual seleclion. Evolulion, 32, 872-85. Bell. G. (1982) The MaJterpieeeo!Narure -Ihe Evolution and Genelies of Sexuality. University of California Press. Berkeley. Belovsky. G.E. (J 978) Diet optimization in a generalist herbivore: the moose. Theor Popul Bioi. 14, 105-34. Benford, EA. (1978) Fisher's theory of the sex ralio applied 10 the social Hymenoptera. J Theor Bioi. 72. 701-27. Bengtsson. B.D. (1978) Avoiding inbreeding: at what cost? J Theor Bioi. 73, 439-44. Benirschke, K.. Anderson, J.M. & Brownhill. L.E. (1962) Marrow chimerism in marmosets. Scienet, 138. 513-15. Benne. R. (1994) RNA editing in trypanosomes. £/lr J Biochem, 221, 9-23. Bennet!, P.M. & Harvey, P.H. (1988) Active and restive melabolism in birds: allometry, phylogeny and ecology. J ZooIILond). 213. 327-63. Bennett-Clark. H.C. (1971) Acoustics of insect song. Nahlre, 234, 255-9. Bemon. T.G. & Foster, W.A. (1992) Altruistic housekeeping in a social aphid. Proc Roy Soc Loud B, 247. 199-202. Bercovitch. EB. (1988) Coalilions. cooperation and reproductive tactics among adult male baboons. Anim lkhav, 36, 1198-209. Berg, E.E. & Hamrick. J.L. (1995) Fine-scale genelic structure of a turkey oak forest. Evolution, 49, lID20, Berthold, P. (1993) Bird Migration. Oxford University Press, Oxford. BenhoJd, P.. Helbig. A.l.. Mohr. G. & Quemer. U. (1992) Rapid microevolulion of migratory behaviour in a wild bird species. Nature. 360, 668-70. Bestor. T.H. (1990) DNA methylation: evolution of a baclerial immune function into a regulator of gene expres ion and genome structure in higher eukaryoles. PlJil Trans Roy Soc Lond B. 326. 17987. Bestor, T.H. & lYcko. B. (1996) Crea'ion of genomic methylation patterns. Nature Gmtl, 12. 363-7.
REF ERE NeE S Bianchi. D.W.. Zickwolf, G.K.. Wei!, G.J .. Sylvester. S. & DeMaria. M.A. (1996) Male fetal progenitor cells persisl in malernal blood for as long as 27 years postpanum. PrO( Nat Acad Sa' USA. 93. 705-8. Bird, A.P. (1993) Functions for DNA methylation in venebrates. Cold Spring Harbor Symp Quam 8iol. 58. 281-5. Bird. A.P. (1995) Gene number. noise reduction and biological complexity. Trends Gmt', 11. 94-100. Birkhead. T.R. (1995) Spenn competition: evolutionary causes and consequences. In: 'Sewtrth Inttmational Symposium on Sptm,atology: Plmary Papers'. Repro· duetion. Fertility and Devtlopment. 7. 755-75. Birkhead. T.R. (1996) Sperm competilion: evolution and mechanisms. Curr Topics Dtvtlop BioI. 33. 10358. Birkhead. T.R. & Fletcher. F. (1995a) Male pheno,ype and ejaculate quality in the zebra finch Taeniopygia guttata. PrO( Roy SO( Lend B. 262. 329-34. Birkhead. T.R. & Fletcher. F. (1995b) Depiction determines sperm numbers in male zebra finches. Anim Behav. 48. 451-6. Birkhead. T.R. & M011er. A.P. (1992) Sperm Competition in Birds: Evolutionary Causes and Consequences. Academic Press. London. Birkhead. T.R. & Moller. A.P. (1993) Why do male birds SlOp copulating while their partners are still fertile? Anim Behav. 45. 105-18. Birkhead. T.R. & M0l1er. A.P. (1996) Monogamy and sp~rm competition in birds. In: Parllltfships in Birds: tile Ecology of Monogamy (ed. J.M. Black). Oxford University Press. Oxford. Birkhead. T.R. & Pringle. S. (1986) Multiple mating and palernity in Gammarus pulex. Anim Behav. )4. 611-13. Birkhead. T.R.. Pellall. EJ. & Fletcher. F. (t993) Selection and utilization of spermatozoa in the reproductive tract of the female zebra finch Taeniopygia guttata. J Reprod Fert. 99. 593-600. Birkhead. T.R.. Pella II. J.E. & Hun,er. F.M. (1988) EXlra·paircopulation and sperm competilion in the zebra finch. Nature Lond. 334. 60-2. Birkhead. T.R .. Burke. T., lann. R.. Hunter. EM. & Krupa. A.P. (1990) Extra-pair palernity and intraspecific brood parasitism in wild zebra finches Taeniopygiagurtata. revealed by DNA fingerprinting. Behav E(ol Sociobiol. 27, 315-24. Birkhead. T.R.. Wishan. GJ. & Biggins. J.D. (1995a) Sperm precedence in Ihe domestic fowl. Proc Roy Sot Lend B. 261. 285-9. Birkhead. T.R.. Fletcher. F.. PellaH. E.J. & Staples. A. (1995b) Ejaculate Quality and the success of extra· pair copulations in the zebra finch. Nature (Lend). 377.422-3.
399
Birks. S.M. (1996) Reproductive behavior and paternity in the Australian brush turkey. Alectura lathami. Ph D dissertation. Cornell University. Ithaca. New York. Bjorn.orp. P. & Brodofl. B.N. (1992) Obt'ity. J.B. Lippincoll Company. Philadelphia. Black. J.M. & Owen. M. (1989) Agonistic behaviour in barnacle goose flocks: assessment. investment and reproductive success. Anim Bthav. 37. 199-209. Black. J.M .• Carbone. c.. Wells. R.L. & Owen. M. (1991) Foraging dynamics in goose nocks: the cost of living on the edge. Anim &hav. 44. 41-50. Blaustein. A.R. & O·Hara. R.K. (1982) Kin recognition in Rana cascadae tadpoles: maternal and paternal effecls. Anim Ikhall. 30.1151-7. Blauslein. A.R. & Waldman. B. (1992) Kin recognition in anura' amphibians. Anim Behav. 44. 207-21. Blaustein. A.R .. Bekoff. M. & Daniels. T.J. (1987) Kin recognition ill vertebrates (excluding primales): empirical evidence. In: Kin Recognition in Animals (eds OJ.c. Flelcher& C.O. Michener). pp. 287-331. John Wiley & Sons. Chichester. Blem. C.R. (t990) Avian energy storage. CurrOrnitllol. 7. 59-113. Boomsma. J.J. (1989) Sex-investment ratios in ants: has female bias been systematically overestimated? Am Nawr. 133. 517-32. Boomsma, J.J. & Grafen. A. (1990) Intraspecific variation in ant sex ratios and the Trivers-Hare hypothesis. EVlllulion. 44. 1026-34. Boomsma, J.J. & Grafen. A. (1991) Colony-level sex ratio selection in the eusocial Hymenoptera. J Evol Bioi. 3. 383-407. Boomsma. JJ .. Keller. 1. & Nielson. M.G. (1995) A comparative analysis of sex ratio investment parameters in ants. Fund Beal. 9. 743-53. Boorman. E. & Parker. G.A. (1976) Sperm (ejaculdte) competition in Drosophila melanogaster. and the reproductive value of females to males in relation to female age Jnd mating status. Ecol Entomol. I. 145-55. Borgia. G. (1985) Bower quality, number of decorations and mating success of male satin bowerbirds (PtilonorJryndJus violactus): an experimental analysis. A'im Bthav. )3. 266-71. Borgia. G. & Collis. K. (1989) Femal(' choi~ for parasile·free male satin bowerbirds and the evolution of bright male plumage. &hav Ecol Sociobiol. 25. 445-54. Borowsky. R.1.. McClelland. M.. Cheng. R. & Welsh. J. (1995) Arbitrarily primed DNA fingerprinting for phylogenetic reconslruction in vertebrates: the Xiphophor', model. Mol Bioi Evol. 12. 1022-32. Bourke. A.EG. (1988) Worker reprodu
400
REFERENCES
Bourke, A.F.G. (1989) Comparative analysis of sex·
investment ralios in slave-making ants. Evolutioll. 43.913-18. Bourke. A.EG. (1994) Worker matricide in social bees and wasps. J Theer BioI, 167, 283-92. Bourke. A.F.G. & Franks, N.R. (1995) Social Evolurion in Ants. Princeton University Press. Princeton. New Jersey. Bourke. A.EG. & Heinze. J. (1994) The ecology of communal breeding: (he case of multiple-queen It:plOthoracine ants. Plril TrailS RiJY Soc Lond 8. 345. 359-72. Bourne. G.R. (1993) Proximate costs and benefits of mate acquisition at leks of the frog Ole/ygon rubra. Anim Brhav. 45.1051-9. Boyce. E.A.. Beauchamp, G.K., Yamazaki, K. & Bard. J. (1991) Genetic components of kin
recognilion in mammals. In: Kin Recognition (ed. P,G. Hepper). pp. 148-61. Cambridge University Press.
Cambridge. Boyd. R. & Lorberbaum. J.P. (1987) No strategy is evolulionarily stable in the repeated prisoner's dilelnma game. NalUre Lond, 327, 58-9. Boyd, R. & Richerson. P. (1985) Culture and the Evolutionary Process. University or Chicago Press. Chicago. Boyd. R. & Richerson, P. (1988a) An evo!ulionary model or social learning: lhe effects of spatial and temporal variation. In: Social Learning: Psychological and Biological Perspectives teds T.R. Zenlall & B.G. Galef Jr) pp. 29-48. LawrenceErlbaum Assodalions, Hillsdale. New Jersey. Boyd, R. & Richerson. P.J. (1988b) The evolulion of reciprocity in sizable groups. J Theor BioI. 132, 33756. Boyd. R. & Richerson. P.J. (1992) Punishment allows the evolution or cooperation (or anything else) in sizable groups. Ethol Sodobiol. 13, 171-95. Breden. E & Stoner, G. (1987) Male predation risk determines female preference in the Trinidad guppy. NalUrf. 329. 831-3. Breden, F. & Wade. M.J. (1991) 'Runaway' sodal evolution: reinforcing selection for inbreeding and altruism. J Theor BioI, 153, 323-37. Breed, M.D. (1983) Nestrnale recognition in honey bees. AnimBehav. 31, 86-9L. Breed. M.D., Garry. M.F.. Pearce, A.N .. Hibbard, B.E., Bjoslad, L.B. & Page, R.E. Jr (1995) The role of wax comb in honey bee nest male recognition. Anim B'hav. 50. 489-96. Breed. M.O.. Snyder. L.E.. Lynn. T.!.. & Morhart. J.A. (J992) AcqUired chemical camouflage in a tropical ant. Anim Behav, 44, 519-23. Brillard. J.P., Galul. O. & Nys, Y. (1987) Possible causes
of subfertility in hens following insemination nt~ar lhe time of ovulation. Brit Poult Sci. 28,307-18. Brinkhof, M.W.G. (1995) Timing of reproduction. An experimenlal study in coots. PhD thesis, University or Groningen. Groningen. Briskie. J.V.. Naugler, C.T. & Leech. S.M. (1994) Begging intensity of nestling birds varies with sibling relatedness. Proc Roy Soc Lond B. 258. 73-8. Brockmann. H.J. (J 984) The evolution of social behaviour in insects. In: Behavioural Ecology: all Bvolutionary Approach, 2nd edn (eds J.R. Krebs & N.B. Davies), pp. 340-61. Blackwell Scienti(ic Publications, Oxrord. Brockmann, H.J.. Colson, T. & POliS, W. (1994) Sperm competition in horseshoe crabs (Limulus polyphemus). 8ehav Scot Sociobiol. 35, 153--60. Brodbeck. O.R. (1994) Memory for spalial and local cues: a comparison of a storing and nonstoring species. Anim Learn Behav. 22. 119-33. Brodbeck, D.R. & Shettleworth, 5.J. (1995) Malching location and color of a compound stimulus: comparison of a food sloring and non·::.toring bird spedes. J Exper Psychol Anim Behav Proces. 21( 1). 6477.
Bronson, EH. & Kerbeshian. M.e. (1995) Reacrions of reproductively photoresponsive versus unresponsive meadow voles to simulated winter conditions. Can J Zool. 73. 1479-88. Brooks. R. & Caithness, N. (1995) Female choice in a leral guppy population: are there multiple cues? Anim Behav. 50, 301-7. Brown. J.L. (1969a) The buffereffecr and produaivily in tit populations. Am Natur, 103. 347-54. Brown. J.L. (1969b) Territorial behavior and populalion regulation in birds: a review and re-evaluation. Wilson Bull, 81, 293-329. Brown. J.L. (1978) Avian mmmunal breedi.ng systems. Ann Rev Ecol Sysl. 9, 123-55. Brown. J.t. (1983) Some paradoxical goals of cells and organisms: the role of MHC. In: Ethical Questions if! Brain and Behavior (ed. D.W. Pfaff), pp. 111-24. Springer. New York. Brown, J.1. (1987) Helping and Communal Breeding ill Birds. Princeton University Press, Princeton. NJ. Brown. J.L. & Brown, E.R. (t990) Mexican jays: uncooperalive breeding. In: Cooperatiw Breeding in Birds: LOJJg-term Studies of Ecology and Behavior (cds P.B. Stacey & W.O. Koenig). pp. 267-88. Cambridge University Press, Cambridge. Brown. J.L. & Ecklund, A. (1994) Kin recognition and the major histocompatibility complex: an imegrative review. Am Natur, 143.435-61. Brown. w.t. & Wilson. E.O. (1959) The evolution of
REFERENCES lhe dacetine anls. Quart Rtv BioI. 34. 278-94. Brunner. 0 .. Kacelnik. A. & Gibbon. J. (1992) Optimal foraging and timing processes in the starling. Sturnus vulgaris: effect of inter-capture interval. A'im Behav. 44. 597-613. Bryant. D.M. & Newton. A.V. (1994) Metabolic costs of dominance in dippers. (indus dndus. Anim BelJav. 48.447-55. Bryant. D.M. & Tatner, P. ( 1991 ) Inlraspecie-s variation in en('rgy expenditure: correlates and constrainiS. Ibis. 133. 236-45. Buchholz. R. (1995) Female choice. parasi'e load and male ornamentation in wild lurkeys. Anim &hav, 50.929-43. Buckle. G.R. & Greenberg. L. (1981) Nestmatt: recognition in sweat bees (Lasio!Jlossum zephyrum): does an individual recognize its own odour Or only odours of nestmatcs? Anim Behav. 29. 802-9. Bull. C.O .. Me'calfe. N.B. & Mangel. M. (1996) Seasonal matching of foraging and anticipatt>d energy requiremenls in anort>xic juvenile salmon. Proc Roy Soc Lond B. 263. 13-18. Bulmer. M.G. (1983) Sex ratio theory in social insects with swarming. J Theor Bioi. 100. 329-39. Burke, T. & Bruford. M.W. (1987) DNA fingerprinting in birds. Nature Lond, 327. 149-52. Burrows, M.T. (1994) An optimal foraging and migration modd for juvenile plaice. Evol £'01. 8, '25-49. Burt. A. (1995) The evolution of fitness. E'olutioll. 49.1-8. Buss. LW. (1982) Somatic cell parasitism and the evolution of somatic tissue compatibility. PrO( Nat Acad Sd USA. 79. 5337-41. Butcher. G.S. & Rohwer. S. (1989) The evolution of conspicuous and distinctive coloration for communication in birds. CurrOrnit!rol. 6.51-108. Bullin. R.K. (1990) Divergence in emergence time of host races due 10 differential gene flow. Heredity. 65.47-50. BUllin, R.K. (1993) The variability of mating signals and preferences in Ihe brown plant hopper. Nilaparvata lugens (Homoptera: Delphacidae). J InS«l
Bl'hav. 6,
12~0.
ButJin. R.K. (1995a) Reinforcement - an idea t'volving. Trends Erol Evol. 10.432-4. Butlin, R.K. (1995b) Genetic variation in mating signa 1<; a nd responses. 1n: Speciation and l!rr Recognition ConCfpt: Theory and Applirations (eds D.M. Lambert & H.G. Spencer). pp. 327-66. Johns Hopkins Univt'rsily Press. Baltimore and london. Butlin. R.K. & Hewiu. C.M. (198501) A hybrid zone between Chonhippus paral/elus paral/elu5 and Chorthippus paral/elus eryt!rropus (Orthoptera:
401
Acrididae) I. Morphological and electrophoretic characters. Bioi J Unn Soc. 26. 269-85. Bu,lin. R.K. & Hewitt. G.M. (I 985b) A hybrid zone between Chorthippus parallelus parallr/us and Chorthippus parol/ellis erythropus (Onhoptcra: Acrididae) II. Behavioural characters. Bioi J Linn Soc, 26.287-99. Butlin. R.K. & Ritchie. M.G. (1991) Variation in female mate preference across a grasshopper hybrid lone. J EvoI Bioi. 4. 227-40. Byers. J.A. & Bekorr, M. (1986) What does 'kin recognition' mean? Ethology. 72, 342-5. Bygoll. J.D. (1979) Agonistic behavior. dominance. and social structure in wild chimpanzees 01 the Gombe National Park. In: The Gr~at Apes (eds D.A. Hamburg & E.R. McCown). pp. 405-27. BenjaminJ Cummings. Menlo Park. California. Byrne. R. W. & Tomasello. M. (1995) Do rats ape? Anini BtIrav.50.1417-20.
Cade. W.H. (1979) The evolution of alternative male reproductive strategies in field crickels. In: S<xual Selection and Reprodud;ve Competition in Insects (eds M.S. Blum & ~.A. Blum). pp. 343-79. Academic Press. London. Caldwell, R.l. (1992) Recognition, signalling and reduc<.'d aggression between former mates 111 a slOmaTopod. Anim Behav. 44. 11-19. Caraco. T. ( 1981) Risk-sensitivity and foraging groups. Ecology. 62. 527-34. Caraco. T. & Brown, J.l. (1986) A game bet'een communal breeders: when is food-sharing stable? J TIrtor Bioi. 118. 379-93. Caraco, T. & Giraldeau, L.-A. (1991) Social foraging: producing and scrounging in a slochastic environment. J Th~or Bioi. 153. 559-83. Carlin. N.F. (19891 DiscriminaTion between and within colonies of social insects: two null hypotht'ses. Netll' J Zool. 39. 86-100. Caro. T.M. (1994) Cheetahs oflhe Seren!Jeri Plains.' Group Livin9 in an Aj(Jcial Species. University of Chicago Press, Chicago. Carpemer. J.M. (1989) Testing scenarios: wasp social behaviour. Cladistics. 5. 131-44. Carson. H.l. (1986) Sexual selection and speciation. In: Evolutionary' PrO(~ss and Theory (eds S. Karlin & E. Nevo). pp. 391-409. Academic Press. Orlando. Carson. H.L. & Templeton. A.R. (1984) Genetic revolutions in relation to speciation phenolnt:na: the founding of new populations. Ann Rl'v Ecol Syst. 15.97-131. Carson. H.L.. Lockwood. J.P. & Craddock. E.M. (1990) Extinction and recolonisation of local populauons on a groWing shield volcano. PrO( Nat Acad sa USA. 87.7055-7.
402
REFERENCES
Cartwright, B.A. &- Collen. T.S. (1983) landmark learning in bees. Experiments and models. J Comp Physiol. 151, 521-43. Catchpole. C.K. (1987) Bird song, sexual selection and female choice. Trends Eeol Evol. 2. 94-7. Catchpole. C.K. &- Sialer. P.J.B. (1995) Bird Sollg. Biological Themesand Van'alions. Cambridge University Press. New Vork.
Cavalli-Sforza. L.L. &- Feldman. M.W. (1981) Culrural Transmission and Evo/miol1: a Quantitative Approach.
Princeton University Press, Prin(eton. Chan. G.L. & Bourke. A.F.G. (1994) Split sex ratios in a multiple·queen ant population. Prof Roy Soc Land 8.258. 261-6. Chapais. B. (1983) Dominance. relatedness and structure of female rhesus relationships. In: Primal( Sodal Relationships (ed. R.A. Hinde), pp. 208-17. Sinauer. Sunderland. Massachusetts. Chapais. B. (1988a) Rank mainlenance in female Japanese macaques: experimental evidence for social dependency.lkhaviour. 104.41-59. Chapais. B. (1988b) Experimental matrilineal inheritance of rank in female Japanese macaques. Anim Behav. 36.1025-37. Chapa is. B. (1992) The role of alliances in social inheritance of rank among female primates. In: Coalitions and Alliances in Humal1.~ and other Animals leds A.H. Harcourt & EB.M. de Waal). pp. 29-59. Oxford University Press. Oxford. Chapais. B.• Girard. M. & Primi. G. (1991) Non·kin alliances and the stability of matrilineal dominance relations in Japanese macaques. Anim Behav. 41. 481-91. Chapman. T.. Liddle. L.E. Kalb. J.M .. Wolfner. M.E & Partridge, L. (1995) Cost of mating in Drosophila mtlan09aster females is mediated by male accessory gland produas. Nature Lond. 373. 241-4. Charlesworth. D. (1985) Distribution of dioecy and sel£·inrompatibility in angiosperms. In: Evolurion: Essays in Honor of John Maynard Smith (eds P.J. Greenwood. P.H. Harvey & M. Slatkin), pp. 23768. Cambridge University Press, Cambridge. Charnov, E.L. (1976a) Optimal foraging: attack strategy of a mantid. Am Nawr. 110, 141-5 [. Charnov. E.L. (1976b) Optimal foraging: the marginal value theorem. Theor Popul Bioi. 9, 129-36. Charnov. E.L. (1982) The Theory of sex AI/oearioll. Princeton University Press. Princeton, ew Jersey. Charnov. E.L. (1993) LIft History Invariants. Some
Explorations of Symmttry in Evolutionary £cology. Oxford Series in Ecology and Evolution. Oxford University Press. Oxford. Charnov, E.L. & Parker, G.A. (1995) Dimensionless invariants from foraging theory's marginal value
theorem. PrO( Nat Acad Sci USA. 92, 1446-50. CharlesVonh. B. (1990) Mutation- election balance and the evolutionary advantage of sex and re· combination. Genet Res, 55. 199-221. Chase. 1.0. (1974) Modds of hierarchy formation in animal societies. BtJravSO, 19, 374-82. Chen. X.F. & Baur. B. (1993) The effecl of multiple mating o~ female reproductive success in Ihe simultaneously hermaphroditic land snail ArialJla arbustorum. Can J Zool. 71. 2431-6. Cheney, D.L (1977) The acquisition of rank and the development of reciprocal alliances among freeranging immature baboons. Behav £col Sodobiol. 2. 303-18.
Cheney, D.l. & Seyfanh. R.M. (1990) How Monkrys See lhe World. University of Chicago Press, Chicago. Cheng. K.M .. Burns. J.T. & McKinney. F. (1983) Forced copulation in captive mallards. Ill. Sperm competition. Auk, 100. 302-10. Chivers. D.P. & Smith. J.P. (1994) Fathead minnows. Pimephales promtlas, acquire predator recognition when alarm substance is associated with the sight of unfamiliar fish. A,lim Bella v, 48. 597-605. Chivers. D.P. & Smith. R.J.F. (1995) Fathead minnows (Pimephales promelas) learn to recognize chemical stimuli from high·risk habitats by the presence of alarm substance. Behav £col. 6. 155-8. Choe. J.e. (1988) Worker reproduction and social evolution in ants (Hymenoptera: Formicidae). In: Advances in Myrmecology (ed. J.e. Trager). pp. 16387. E.J. Brill. Leiden. Christenfeld. N.J.S. & Hill, E.A. (1995) Whose baby are you? Naturt. 378. 669. Christy. J.H. (1995) Mimicry. mate choice. and the sensory trap hypothesis. Am NaWr. 146. 17)81.
Christy. J.H. & Salmon. M. (1991) Comparalive studies of reproductive behavior in mantis shrimp and fiddler crabs. Am ZOOI. 31. 329-37. Churchland. P.S.. Ramachandran. V.S. & Sejnowski. T.J. (1994) A critique of pure vision, In: Large-Scale Neuronal Theories of/he 8rain (eds C. Koch &- J.H.L. Davis). pp. 23-60. MIT Press, Cambridge. Clark. A.G., Aguade. M.. Prout. T.. Harshman, L.G. & langley. C.H. () 995) Variation in sperm displacement and its association wilh accessory gland protein in Drosophila mtlalJogaster. Gmetics, 139. 189-201.
Clark. C.W. & Levy. D.A. (1988) Die! vertical migralions by juvenile SOCkeye salmon and the antipredation window. Am Nawr. 131, 271-90. Clark, C.w. & Mangel. M. (1984) Foraging and flocking strategies: information in an uncertain environmenl. Am Nawr. 123. 626-4 t.
REFERENCES Clark. c.w. & Mangd. M. (1986) The evolutionary advanlages of group foraging. Th~or Popul Bioi. 30. 45-75. Clark. D.L. & Uetz. G.W. (1993) Signal efficacy and the evolution of male dimorphism in the jumping spider. Mantia inclemens. Proc Nat Acad Sci USA. 90. II 954-7. Clarke. M.E (1984) Co-o~rative breeding by the Australian Bell miner.' Manon·na mtlanophrys latham: a test of kin seledion theory. Behav ccol Sociobiol. 14, 137-46. Clarke. M.E (1989) The pallern 01 helping in ,he Bell miner Manorina mtlanophrys. Ethology. 80. 292306. Clarke. R.T. & Goss-CuSlard. J.D. (1996) The Ext' t'stuary oyslercateher-mussel Illodel. In: The Oysttrcatchtr:!rom Individuals to Populations (ed. J.D. Goss-Cuslard). pp. 390-93. Oxford University Press. Oxfurd. Clayton. D.L. (1990) Mate choice in experimentally parasitized rock doves: lousy males lose. Am ZOOI. 30.251-62. Claylon. N.S. (1995a) The neuroethologicai development of food-storing memory: a case of use it or lose it! &hav Brain Res. 70. 95-102. Claylon. N.5. (1995b) Development of memory and the hippocampus: comparison of food·storing and nonstoring birds on a one-trial associative memory task. J Neurosci. 15.2796-807. ClaYlOn. N.S. & Krebs. J.R. (1993) Lateraliza,ion in the Paridae: comparison of a storing and a non· swring species on a one-trial associative memory task. J Comp Physiol A. 171.807-15. Claylon. N.S. & Krebs. J.R. (1994a) LalcraHzation and unilat~rallransferof spatial memory in marsh tits: art' two eyes better than one? JComp Physiol A. 174, 769-73. Clay,on. N.S. & Krebs. J.R. (1994b) Memory for spatial and object-specific cues in food·swring and nonstoring birds. J Comp Physiol A, 174. 371-9. ClaylOn. N.S. & Krebs. J.R. (l994c) Hippocampal growlh o1nd anrilion in birds affected by experience. Pro< Nar Aald sa USA. 91. 7410-14. Clay'on_ N.S. & Krebs. J.R. (I 994d) One-trial associative memory: comparisons of food-storing and non-sloring species of birds. An;m !.Lam Behav. 22.366-72. Clements. K.C. & Stephens. D.W. (1995) Testing models of non·kin cooperation: mutualism and the prisoner's dilemma. Anim Behav. SO. 527-35. Clutlon-Brock. T.H. & Albon. S.D. (1979) The roaring of red deer and the evolution of honest advertisement. Behaviour. 69. 145-70. Clulton-Brock. T.H. & Harvey. P.H. (1984)
403
Comparative approaches to investigating adaptalion. In: BehaviiJural Ecology:an ElIOlutionary Approach (eds J.R. Kr<=' c.ond=' b.=' 259.=' 223-8.=' colgan.=' p.=' (1983)=' comparativr=' recognition.=' john=' wiley=' sons.=' :-.lew=' collet!.=' t.s.=' undo=' m.e=' (1978)=' how=' hoverflies=' compule=' inlerceplion=' courses.=' j=' camp=' physio'.=' 125.=' 191-204.=' collins.=' s.a.=' (1994)=' male=' displays:=' cause=' or=' erleet=' female=' preference?=' belrav.=' 48.371-5.=' effect=' recent=' experience=' on=' choice=' finches.=' an;m=' &hav.=' 49.=' 479-86.=' s.a..=' huhbard.=' c.=' houtman.=' a.m.=' mate=' finch:=' effeet=' beak=' colour=' song.=' bthavecol=' sodobiol.=' 35.=' 21-5.=' compton.=' m.m=' ..=' van=' krey.=' h.p.=' siegel.=' p.b.=' filling=' ('mptying=' ihe=' uterovaginal=' sperm·=' hosl=' glands=' lhe=' domeslic=' hen.=' poult=' sci.=' 57.=' 16961700.=' conner.=' w.e=' roach.=' b..=' benedia.=' e..=' meinwdld,=' j.=' eisner.=' t.=' (1990)=' courtship=' pheromone=' produaion=' body=' size=' as=' correlates=' larval=' diet=' ardiid=' moth.=' utttheisq=' omatnx.=' ch,m=' eal=' 16.='>
404
REFERENCES
Cook, P.A. & Gage, MJ.G. (1995) Effects of risks of sperm competition on the numbers of eupyrene and apyrene sperm ejaculated by the male moth Plodia htterpllndella (Lepidoptera: Pyralidae). Behav Beol
Sociobiol. 36. 261-8. Cooper, S.J.B., Ibrahim. K.M. & Hewitt, G.M. (1995) Postglacial expansion and subdivision of Ihe grasshopper Chorthippus parallel'ls in Europe. Mol Ero/. 4, 49-60. Cooper. W.E.• Viti, L.J .• Hedges. R. & Huey, R.B. (1990) locomotor impairment and defence in gravid lizards (Eumeces laticeps): behavioural shih in activity may
offset costs of reproduction in an active forager. Seha', £co/ Sociobiol. 27. 153-7. Coopersmith, C.B. & Lenington, S. (1992) Female preferences based on male quahty in house mice: interaction between male dominance rank and (complex genotype. Ethology. 90. 1-16. Cornell. T.J.. Beeven. K.A. & Gamboa, G.J. (1989) Kin recognition by tadpoles and [eoglets of the wood Crog Raila sylvatica. O~colo9ia, 78. 312-16. Cosmides. L.M. & Tooby. J. (1981) Cytoplasmic inheritance and inuagenomic conflict. J Thear Bioi.
89.83-129. Cowie. R.J. (1977) Optimal foraging in great tils (Parus major). Nature, 268, 137-9. Cowie. R.J. & Krebs. J.R. (1979) Optimal foraging in
patchy environments. British ecological Society Symposium on Population DY'dmics (cds R.M. Anderson, R.D. Turner & L.R. Taylor), pp. 183-205. Blackwell Scientific Publications. Oxford. Craig. R. (1979) ParcllIalmanipulalion. kin selection. and the evolution of altruism. Evolwion. 33.319-34. Creel. S.R. & Waser. P.M. (1991) Failures of reproductive suppression in Dwarf mongooses (Helogale parvula): accident or adaptation? Behav Eml Sociobiol, 2, 7-15. Creel, S. & Creel, N.M. (1995) Communal hunting and pack size in African wild dogs Lycaon pic/us. Anini Bellov, SO, 1325-39. Creel. S.• Creel. N. & Monfor!, S.L. (1996) Social stress and dominance. Nalure Lond. 379, 212. Creel,S.. Creel. N.. Wildt. D.E. & Monfor!. S.L. (1992) Behavioural and endocrine mechanisms of reproductive suppression in Serengcti dwarf mongooses. Anim Behav, 43. 231-45. Creel. S.R .. Mon[on, S.L.. Wildt. D.E. & Wasec. P.M. (1991) Spontaneous laclation is all adaplivt' result or pseudo-pregnancy. Nature. 351. 660-2. Creo;pi. B.l. (1992) Eusodality in Auslralian galllhrips. l'a1llre, 359, 724-6. Cn:.. pi, BJ. (1994) Three condirion~ ror the evolution of eusocialilY: are they surficient? insect Soc. 41. 395400.
Crespi. BJ. & Vanega, D. (1995) The definition of eusociality. Behav Ecol. 6. 109-15. Cresswell, W. (1994) Song as a pursuit-deterrent signal. and its occurrence relative to other anti predator behaviours of skylark (Alauda tlrvmsis) on attack by merlins (Falco eoillmbarius) Behav Eecl Sociobiol. 34. 217-3. Cronin. H. (1991) The Am and the Peacock. Cambridge University Press. Cambridge. Crook, J.n. (1964) The evolution of sodal organisation and visual communication in the weaver birds (Ploeeinae). 8ehaviou' 10 (Suppl.). 1-178. Crook. J.H. (1965) The adaptive significance of avian soo(ll organiSiltion. SympZoo! Soc Lond, 14, 181-218. Crook. J.H. &Gartlan. J.S. (1966) Evolulion or primate societies. Nature Lond. 210, 1200-3. Crozier, R.H. (1979) Genetics of sociality. In: Social Insects. Vol. 1 (cd. H.R. Hermann), pp. 223-86. Academic Press. New York. Crozier, R.H. & Dix. M.W. (1979) Analysis of two genelic models for the innate components of colony odor in social Hymenoptera. Behav Ecol Sociobiol. 4. 217-24. Crozier, R.H. & Luykx. P. (1985) The evolulion of termite eusociality is unlikely to have been based on a male haploid analogy. Am Natur, 126, 867-9. Crozier. R.H. & Page, R.E. (1985) On being the right size: male contributions and multiple mating in social Hymenoptera. Behov Eco/ Sodobiol, 18. 10515. Crozier. R.H. & Pamilo. P. (1996) Evo/lIlion of Social [meet ColonliS: Sex Alloearion and Kin Sfleetlon. Oxford University Press, Oxford. Cullen. E. (1957) Adaptations in the kittiwake to cliff nesting. Ibis. 99. 275-302. Cullen. J.M. (1966) Reduction of ambiguily through ritualizalion. Phil Trans Roy Soc Lond B, 251, 363-4. Curio, E. (1988) Cullural lransmission of enemy recognition by birds. In: Social Leamin9 (eds T.R. Zentall & B.G. Galer. Jr). pp. 75-97. Lawrence Erlbaurn Associates, Hillsdale. New Jersey. Curio. E. (1989) Some aspects or avian 1110nality patterns. Miueil Zeal Museum Berl. 65{Suppl.). 474
4
70.
Curry. R.L. (1988) Influence or kinship on helping behavior in Galapagos mockingbirds. Bel1av Ecol Sodobiol. 22, 141-52. Cuthill. I.e. & Guilford. T. (1990) Perceived risk and obstacle avoidance in flying birds. Anini Bellav. 40. 188-90. Cuthill. I.e. & Macdonald. W.A. (1990) Expt:rilllt:lllal manipulation of the dawn and dusk chorus in th~ blackbird Yurdus merula. Behav Ecol Sociobiol, 26. 20916.
REF E RE N CE S Cuthill. I.e.. Haccou. P. & Kace1nik. A. (1994) Starlings (Stumus vulgaris) exploiting patches: response to long-term changes in travel time. Bthav Ecol. 5. 8190. Cuthill. I.C.• Kacelnik. A.. Krebs. J.R .. Haccou. P. & ' .....asa. Y. (1990) Starlings exploiting patches: the eHeel of recent experience on foraging dedsions. Anim &hav. 40. 62540. Cwynar. L.c. & MacDonald. G.M. (1987) Geographical variation of lodgepole pine in relation to population history. Am Natuf. 129.463-9. Oaan. S. & Aschoff. J. (1982) Circadian contributions to survival. In: Vertebrate Circadian Systems (eds J. Aschoff. S. Daan & G. Groos). pp. 305-21. SpringerVerlag. Berlin-Heidelberg. Daan. S.. Deerenberg. C. & Dijkstra. C. (1996) Increased daily work precipitates natural death in thl: kestrel. J Ani', Ecol. 65.539-44. Daan. 5 .. Dijkslra. C. & llnbergen. J.M. (I 990a) Family planning in the kestrel (Falco linnunculus): the ultimate control of covariation of laying dale and dutch size. Ikhaviour. 114. 83-116. Daan. S.• Masman. D. & Groenewold. A.A. (1990b) Avian basal melabolic rates: their association with l.>ody composition and energy expendilurt: in nature. AmJ Plrysiol. 259. R333-40. Daan. S.. Dijkstra. C .• Drenl, R. & Meijer. T. (I 989a~ Food supply and the annual liming of avian reproduction. In: Acta XIX Congress us Intemationalis Ornilil%gid. Vol. I (ed. H. Ouellet), pp. 392-407. University of Ottawa Press. Ottawa. Daan. S.. Masman. D., Strijkstra. A.M. & Verhulst. S. (1989b) Intraspecific allomelry of basal metabolic rate: relations Wilh body size. temperalure. comlx>silion and circadian phase in the kt.'StreL Falco ';'Imlneulus. J Bioi Rhythms. 4. 267-83. Dadour. I.R. & Bailey. W.J. (1990) The acoustic behaviour or males Mygalopsis marki (Copiphorinae; TC'nigoniidae). In: The Tettigoniidae: &haviour. Systmratics. Evolution (eds W.J. Bailey & D.e.F. Rentz). pp. 98-11 I. Crawford House Press. Balhurst. Dalla. J.E. Dod. B., BoursOl, P., Prager, E.M. & Bonhomme. F. (1995) Population subdivision and gene (low in Danish house mice. Mol £rol. 4.31120. Darwin. C. (1859) On the DriBin ofSpeaes by MeallS of Natural Sdection. Murray. London. Darwin. e. (1871) The Descent of Man. and Sllenion ill Rtlation to sex. John Murray. london. Darwin. C. (1883) rhe Dtsanr of Man and Selection iI' Rl'lalion 10 $lx. lrd edn. Murray. London. Oalla. 5.B. (1983) Relative power and the acquisilion 01 rank. In: Primate Social Relationships (ed.
405
R.A. Hinde). pp. 93-103. 5inauer. Sunderland. Massachusetts. Davis. l.l. (1995) The nuclear pore complex. Ann Rev BiocIJem. 64. 865-96. Davies, N.B. (1978) Territorial defence in the speckled wood butterlly (Pararge at9'ia): the resident always wins. Anim &hav, 26. 138-47. Davies. N.B. (1992) Dunnoc:k &haviour and Social E.. .,olution. Oxford University Press. Oxford. Davies. N.B. & Brooke. M. de L. (1989) An eXI>erimental study of co-evolution between the cuckoo Cumlus canorus and itS hOSlS. I. Host egg discrimination. J Anim Erol, 58. 207-24. Davies. N.B. & Halliday, T.R. (1978) Deep croaks and fighling assessment in wads Bufo bufo. Nature. 274. 683-5. Davies. N.B' Brooke. M. de L. & Kaceinik. A. (1996) Recognition errors and probability of parasitism determine whether reed warblers should accept or reject mimetic cuckoo eggs. Proc: Roy Soc Lend B. 263. 925-31. Davies, N.B., Hartley, I.R .. HatchwelL B.J & Langmore. N.E. (1996) Female control or copu· latiOns to maximize male help: a comparison of polygynandrous alpine accelllors. Pnmella COl/lITis. and dunnocks P. modularis. Anirn Behav, 51, 27-47. Davies. N.B .• Hatchwell. B.J.• Robson. T. & Burke. T. (1992) Paternity and parental cHon in dunnocks Prunella modularis: how good are mal~ chick-feeding rules? Allim Beitav. 43. 729-45. Dawkins. M. (1971) Perceptual changes in chicks: another look 011 the 'search image' concepr. Anim Bellav. 19. 566-74. Dawkins. M.S. & Guilford. T.e. (1991) The corruption of honest signalling. Anirn Btlrav. 41. 865-73. Dawkins. M.S. & Guilford. T. (1995) An exaggerated preference for simple neural network models of signal evolution? Proc Roy Soc Lend B. 261, 357-60. Dawkins. R. (1976) The Selfish Gene. Oxford University Press. Oxford. Dawkins. R. (1979) 1velve misunderstandings of kin selection. Z Tierpsychol, 51. 184-200. Dawkins. R. (1982) Tht Exlmdtd Phmofypt. W.H. Freeman. San Francisco, california. Dawkins, R. & Krebs. J.R. (1978) Animal signals: information or manipulation? In: Behavioural Ecology: an E,,'Olutionary Approach. lSI edn (eds J.R. Krebs & N.B. Davies). pp. 282-309. Blackwell Scientific Publications. Oxford. Dawson. A.. Goldsmith. A.R. & Nicholls. T.J. (1985) Thyroidectomy results in termination of photorefractoriness in starlings (Stumus vulgllfis) kept in long day lengths. J RtprDd Fen. 74. 52733.
406
REFERENCES
Dawson. B. V. & Foss. B.M. (1965) Observational learning in budgerigars. Anim &!hav. 13.470-4. De-Chiara. T.M., Robertson, E.J. & efstratiadis. A. (1991) Parental imprinting of the mouse insulinlike growth faetor II gene. Ctll. 64. 849-59. Deereoberg. C. (1996) Partlllal tIItf9J and pruess costs
m birds. PhD thesis. University of Groningen, Groningen. Dt'eren~rg, c.. Pen, I., Dijkstra. c.. Arkies. B.·J.• Visser, G.H. & Daan. S. (1995) Parental energy expenditure in relation to manipulated brood size In Ihe European kesuel Falco titmUllculus. Zoology;
Analysis Compltx Systems 99. 39-48. Dempson. J.B. & Green. J.M. (1985) Life history of anadromous arctic chare. SalvelflJlIs a/pinus. in the Fraser River. northern Labrador. Can Zoot. 63. 31524. Dc.-nneu, D.C. (1995) DatWin's Dallgtrous Idta. Simon & Schuster. New York. Dr-nno. R.F. (1994) The evolUlion of dispersal polymorphisms in insects: the influence of habitats. host plants and mates. Rts Popul £col. 36. 127-35. DeSalle, R.. Gatcsy, J .. Wheeler. W. & Grimaldi. O. l1992) DNA sequences from a fossil termite in Oligo-Miocene amber and their phylogenetic implications. Sdmu, 2'7. 1933-6. Deslippe. R.J. & Savolainen. R. (1995) Sex investment in a social insect: the proximate role of food. Ecology. 76.375-82. Devenport. J.A. & Devenport. l.D. (1993) TImedependent decisions in dogs (Canis fami/iaris). J Compar Psychol. 107. 169-73. Devenport. L.D. & Devenport. J.A. (1994) 11medependent averaging of foraging information in least chipmunks and golden-mantled ground squirrels. Anim &Jtav, 47. 787-802. Dewsbury. D.A. (1982) Ejaculate cost and male choice. Am Natur. 119.601-10. Dewsbury. D.A. (1984) Spenn competition in murid rodents. In: Sptrm Compttition and Tht Evolution 0/ Animal MatingSysttms (ed. R.L. Smith). pp. 547-71. Academic Press. Orlando. Diamond. J.M. (1993) Evolutionary physiology. In: The Logic of Ufe (eds C.A.R. Boyd & D. Noble). pp. 89-111. Oxford University Press. Oxford. New York. Diehl. S.R. & Bush. G.L. (1989) The role of habitat preference in adaptation and spedarion. In: Spenation and its COnsequenas (eds D. Oue & J.A. Endler). pp. 345-65. Sinauer. Sunderland. MA. Diesel. R. (1990) Sperm competition and reproductive success in the decapod Inluhus phala'gium (Majidae): a male ghost spider crab that seals off nvals' sperm. J ZOOI Lond. 220. 213-23.
Dijkstra. C.. Bult. A.. Bijlsma. S.. Daan. S.. Meijer. T. & Zijlstra. M. (1990) Brood size manipulations in the kestrel Falco tinnunotlus: erreets on offspring and adult survival. J Anim Ecol. 59. 269-86. Dixon. A.. Ross. D.. O·Malley. S.L.e. & Burke. T. (1994) Paternal investment inversely related 10 degree of extra-pair paternity in the reed bunting (EmbtrizQ schomiclus). Nature (Loud). 371. 698700. Doherty. J. & Hoy. R. (1985) The auditory behavior of crickets: some views of genetic coupling. wng recognition. and predator detection. QIIQrt Rtv Bioi. 60.457-72. Dolman. P.M. (1995) The intensity of interference varies with resource density: evidence from a field slUdy with snow buntings. Pltetrophtnax 'ivalis. Dec%g;a. 102. 511-14. Dominey. W.J. (1984) Effects of sexual selection and life histories on speciation: spedes flocks in African ckhlids and Hawaiian Drosophila. In: Evolmiouo/Fi,fh Spedes F/ocks (cds A.A. Echclle & I. Kornfield). pp. 231-49. Orono Press, Maine. Douglas. R.H. & Djamgoz. M.B.A. (1990) Tilt Vi'l/al Sysltm ofFish. Chapman & 1·lali. London. New York. Dover. G.A. (1993) Evolution of genelie redundancy for advanced players. Curr Opin Genet Devtlop. 3. 902-10. Drelll. R.H. & Daan. S. (1980) The prudent pan:llI: energeric adjustments in avian breeding. Ardea. 68. 225-52. Drickamer. L.e. (1992) Oestrus female house mice discriminate dominanl from subordinale males and sons of dominant from sons or suoordinate males by odour cues. Anim Behav. 43. 868-70. Drummond. H. & Osorno. J.L. (1992) Training siblings to be submissive losers: dominance between booby nestlings. Anini BtJrav. 44. 881-93. Dugatkin. l.A. (1988) Do guppies play til·for-tat during predator inspection visits? eehav £rol Sodobiol, 23. 395-9. Dugatkin. l.A. (1991) Dynamics o( the tit for tal strategy during predator inspection in guppil's. &hav E(ol Sodobiol. 29, 127-32. Dugatkin. LA. (1992a) Tendency to inspect predators predicts mortality risk in the guppy (POfrilia retiwlata). &hav E(ol. 3. 124-7. Dugatkin. LA. (1992b) Sexual selection and imitation: females ropy the male choice of others. Am Nowr. 139. 1384-9. Dugatkin.l.A. (1996a) lit for tat. by-produn mutualism and predator inspection: a reply to Connor. Ani' 8
REFERENCES male choice in Poecilia reticulata. Proc Nat Acad Sd USA. 93. 2770-3. Dugatkin. L.A. & Godin. J.·G.J. (1992) Reversal 01 female mate choice by copying in Ihe guppy (Poediio reliculata). Proc Roy Soc Land B, 249, 179-84. Dugatkin. LA. & Godin. J.-G.J. (1993) Female mate copying in the guppy (Poecilia reticulata): agedependent elfects. Behav ECQI Sociobiol. 4. 289-92. Dugatkin, L.A .. Mestenon-Gibbons. M. & HouslOn, A.1. (1992) Beyond the prisoner's dilemma: towards models to discriminate among mechanisms of cooperation in nalure. Trmds Ecol Evo!. 7, 2025. Duquesnoy. E. & Filipo. R.M. (1994) Polyarthrite rhumatoide et neoplasies. Rev Rhum Ed Fr, 61, 1947.
Dukas. R. & Clark. C.W. (1995) Searching for cryptic prey: a dynamic model. Ecology. 76, 1320-6. Dukas. R. & Ellner, S. (I993llnforrnation processing and prey detection. Bcology, 74, 1337-46. Dukas, R. & Real. L.A. (1993) learning constraints and floral choice in bumble bees. Anim BeJzav, 46, 637-44. Dunbar, R.I.M. (1988) Primate Sodal Systems. Croom Helm, london. Durell, S.E.A. le V. dit. Goss-Custard, J.D. & Caldow, RW.G. (1993) Sex-related differences in diet and feeding method in the oystercatcher Haemaropus osrralegus. J Anim Ecol, 62, 205-15. Dusenbery. O.B. (1992) Sensory Ecology. Freeman. New York. Dyer. F.c. (1996) Spatial memory and navigation by honeybees 011 the scale of the foraging range. J Exp 8iol. 199. 147-54. Ebbingt.'. B. (1989) A multifactorial explanation for variation in breeding performance of brenr geese Branca bernic/a. Ibis, 131. 196-204. Ebbinge, B. (1991) The impact of hunting on monality rates and spatial distribution of geese wintering in Ihe Western Palearctic. Ardea. 79, 197-210. Eberhard, W.G. (1980) Evolutionary consequences of intracellular organelle competition. Quart Rev Bioi, 55.231-49. Eberhard, W.G. (1996) Female Concrol: Sexual Selection by Cryptic Female Choice. Princeton University Press, Princeton. Egid, K. & Brown, J.L. (1989) The major histocompatibility complex and female mating preferences in mice. Anim Behav, 38, 548-50. Ehlinger, T.J. (1989) Learning and individual variation in bluegill roraging: habitat-specific techniques. Anim Behav, 38, 643-58. Eisner, T. & Meinwald, J. (1995) The chemistry of sexual selection. Proc Nat Acad Sci USA, 92, 50-5.
407
Ekman, J.B. (1987) Exposure and time use in willow tit flocks: the cost of subordination. Anim Behav. 35. 445-52. Ekman, J.B. & Hake, M.K. (1990) Monitoring starvation risk: adjustments of body r{'serves in greenfinches Cardtllis chlolls during periods of unprt'dictable foraging success. Behav Ecol, 1. 62-7. Ekman, J.B. & LilJiendahL K. (1993) Using priority to food access: fattening strategies in dominancestructured willow tit (Parus mOn/anus) flocks. Bt'hav Eeo/.4. 232-8. Elgar. M.A. & Harvey, P.H. (1987) Basal metabolic rates in mammals: Allometry, phylogeny and ecology. Puna Eeal, I, 25-38. Ellis, H.1. (1984) Energetics of free· ranging seabirds. In: Seabird Energetics, (eds G.c. Whitlow & H. Rahn), pp. 203-34. Plenum Press, New York. Elton, C. (1927) Attimal Ecology. Macmillan, New York. Emlen. S.T. (J 978) The evolulion of cooperalive breeding in birds. In: Btltavtoural Ecology: an evol'riollaryapproach, (edsJ.R. Krebs & N.B. Davies), pp. 245-81. Blackwell Scientific Publkations, Oxford. Emlen, S.T. (1982a) The evolution of helping. H. The role of behavioral conflia. Am Natur, 119, 40-53. Emlen. S.T. (1982b) The evolution of helping. t. An ecological constraints model. Am Nawr, 119, 2939. Emlen, S.T. (1984) Cooperative breeding in birds and mammals. In: Behavioral Ecology: an Evolutionary Approach. 2nd edn (eds J. Krebs & N. Davies), pp. 305-35. Blackwell Saemilic Publications. Oxford. Emlt:n, S.T. (1990~ The white-fronted bee-ealer: helping in a colonially nesting species. In: Coopertltiw Breeding in Birds: Long-Term Studies of Ecology and Behavior (cds P. Stacey & W. Koenig), pp. 305-35. Cambridge University Press, Cambridge. Emlen, S.T. (1991) Evolution of cooperative breeding in birds and mammals. In: Behavioural Ecology: an Evolutionary Approach (cds J. Krebs & N.B. Davies), 3rd edn. pp. 301-37. Blackwell Sciemific Publications, Oxford. Ernlen, S.T. (1994) Benefits, constrain IS, and Ihe evolution of the family. Trends Beol Evol, 9, 282-5. Ernlen, S.T. (1995a~ An evolulionary theory of Ihe family. Proe Nat Acod Sci. 92. 8092-9. Emlen. S.T. (1995b) Can avian biology be useful (0 Ihe social sciences? J Avian BioI. 26, 273-6. Emlen. S.T. (1996) Reproductive sharing in different lypeS of kin associations. Am Nawr, 148. 756-63. Emlen, S.T. & Oring, L W. (1977) Ecology, sexual seleClinn. and the evolution of mating systems. Science, 197, 215-23. Emlen. S.T. & Vehrencamp, $.L. (1983) Cooperative breeding snategies among birds. In: PerspectiVe'S in
408
REFER EN C ES
Omithology (ed, A.H. Brush & G.A. Clark. Jr). pp. 93-133. Cambridge University Press. Cambridge. Emlen, S.T. & Wrege. P.H. (1988) The role of kinship
in helping decisions among white-fronted beeeaters. Behav Ecol Sociobiol. 23. 305-15. Emlen. S.T. & Wrege. P.H. (1992a) Parent-offspring connia and the recruitment of helpers among beeealcrs. Nature. 356.331-3.
Emlen. S.T. & Weege. P.B. (1992b) Parent-offspring conflicr. Nature. 3'9. 24. Emlen. S.T.. Demong. N.J. & Ernlen. O.J. (1989)
Experimental induction of inlanticide in female w3uled jacanas. Auk. 106, 1-7. Endler. l.A. (1977) Geo9raphic Variation, Speciation. and Clines. Princeton University Press. Princeton. NJ. Endler, l.A. (1978) A predator's view o( animal colour patterns. Evol BioI. 11. 319--64. EndJer, J.A. (1980) Natural selection on colour patterns in Poecilia rtrieulala. Evolurion, 34, 76-91. Endler. J.A. (1983) Natural and sexual sekction on color pallerns in poeciliid fishes. El1virol1 Bioi Fisl1~s, 9. 173-90. Endler. J.A. (1991 a) Interactions between predators and prey. In: Behavioural Ecolclgy An Evoltttionary Approach (eds J.R. Krebs & N.B. Davies). 3rd t:dn. pp. 169-96. Blackwell Scielllific Publications.
Oxford. Endler. J.A. (1991 b) Variation in the appearance of guppy colour patterns to guppie5- and their predators under different visual conditions. VIS R~s. 31. 587608.
Endler. J.A. (1992) Signals. signal conditions and (he direction of evolution. Am Nartlr. 139. S 125-553. Endler. J.A. (1993) Some general comments on rhe evolution and design of animal signalling systems. Phil TrailS Roy Soc Lelld B. 340. 215-25. Endler. J.A. & Houde. A.E. (1995) Geographic variation in female preferences for male traits in Potcilia reticulala. Evolution. 49. 456-68. Engels. W. (cd.) (1990) Sociallns~cts.- all Evolutionary Approach to Castts and Reprodueticl11. Springer- Verlag, Berlin. Enquist. M. (L 985) Communication during aggressive interactions with particular reference (0 variation in choice of behaviour. Anini Behav. 33, 1152-61. Enquist. M. & Arak. A. (1993) Sele<1ion or exaggerated male trailS by female aesthetic senses. NalUre. 361, 446-8.
Enquisl. M. & Arak. A. (1994) Symmetry. beauly and evolution. Nature. 372. 169-72. Ens. B.J. & Cayford. J.T. (1996) Feeding with other oyslercatchers. In: The Oystercatcher:from I1rdividuals to Popularions (cd. J.D. Goss-Custard). pp. 77-lO4. Oxford University Press. Oxford.
Ens. BJ. & Gnss-Cuslard. J.D. (1984) Imerlerenee amongoystercatchers. Haematopusostrale9us. feeding on mussels. Mytilusedulis, on the Ext' estuary. J An;m
£col. 53. 217-31. Ens. 8.J.. SafrieL U.N. & Harris. M.P. (1993) Divorce in the long-lived and monogamous oyslerc,uc:her, Haemotopus ostral~gl/s: incompatibility or ('hoosing a better option? Ani', Behav. 45. 1199-217. Ens. B.J., Weissing, EJ. & Drent. R.H. (1995) The despotic distribution and defem:d malurity: two sides of the same coin. Am Nawr. 146. 536-64. Ens. B.J., Kersten. M .• Brenninkmdjt:r. A. & Hulscher, J.B. (t 992) Tcrritory Quality. parclllal effort and reproduaive success of 0yslercatchers (Haematoplls oSlralegus). J Anim Erol. 61, 703-15. Erhardt. c.L. & Bernstein, l.S. (1986) Matrilineal ovenhrows in rhesus monkey groups. 1m J Prima to!. 7.157--81.
Errington. J. (1996) Determination of cell fale in
Bacillus subrilis. Treuds Gentt. 12. 31-4. Eshe!. L (1985) Evolutionary genetic stahility of Mendelian st.>gregation and Ihe role of frct' recombination in the chromosomal system. Am
Nallir. 125.412-20. Espclie. KE. & Hermann. H.R. (1990) Surface lipids of the social wasp Po!istts mm'laris (L.) and its Ilest and nest pedicel. J Chtm Eeol. 16. 1841-52. Evans. C.S.. Macedonia. J.M. & Marler. P. (1993) Effeas of apparent size and speed on the response of chickens, Gallus galll/s. to computer-generated simulations of aerial predators. Anim Behav. 46. II1 Evans. H.E. (1977) Extrinsic versus intrinsic factors in the evolution of insect sodality. Bioscience. 27. 613-17.
Evans. J.D. (1993) Parentage analyst·s in ant colonies using simple sequence repeatloct. Molec Ecol. 2. 3937. Evans. J.D. (1995) Relatedness threshold for the production of female sexuals in colonies 01 a polygynous ant. Myrmica tahoel1sis. as revealed by microsatellite DNA analysis. Proc Nat Aead Sci USA. 92.6514-17.
Faaborg, J .. Parker, P.G., Delay. L et al. (1995) Confirmation of cooperative polyandry in the Galapilgos hawk (Bureo galapagoemis). Behav Ecol
Soc/obiol. 36. 83-90. Fabre. J.H. (1897) Souvenirs Eltlomologiques (A.T. de Manos. Trans). C. Delgrave. Paris. Farr. J.A. (1980) Social behaviour patterns as determinations of reproduaive success in the guppy Poeci/ia retio/lara Peters (Pisces. Poeciliidae): An experimental sludy of the effects of intermale competition. female choice. and sexual selection.
REFERENCES Bthawour. 74. 38-91. Feder. J.L. & Bush. G.L. (1989) Gene frequency clines for host races of RhagoJttis pomonella (Diptera: Tephrilidae) in the midwestern United States. Heredity. 63. 245-66. Feder. J.L.. Hunt T.A. & Bush. G.L. (1993) Th~ effects of c1imalt.'. host phenology and host fidelity on the genetics of apple anu hawthorn infesting races o( Rhagolrris pomollella. Entomol Exp Appl. 69. 117-
n.
Feder. J.L.. Opp. S.B.. Wlazlo. B.. Reynolds. K.. Go. W. & Spisak. S. (1994) Host fidelity is an effective premating barrier between sympatric racC's of the apple maggot ny. ProcNat Acad Sci USA. 91. 7990-4. Felscnstcin. J. (1985) Phylogenies and the comparative method. Am Natur. 125. 1-15. Fisher, J. & Hinde. R.A. (1949) The opening of milk bOllle by birds. Br Birds. 42. 347-57. Fisher. R.A. (1930) .Thr Getlflical Throry of Natural Selecrioll. Clarendon Press, Oxford. Fishwild. T.G. & Gamboa. G.J. (1992) Colony defense against conspecifics: caste-spectfic differences in kin recognition by paper wasps. Polistes fuscarus. Anim BeJrav. 43. 95-102. Fiske. P. & Kalas. J.A. (1995) Mate sampling and copulation behaviour of greal snire females. Anim Behav.4·9. 209-19. FitzGibbon. CO. & Fanshawe. J.H. (1988) Stolling in Thomsun's gaz(lIes: an honest signal of condition. Bthav Ecol Sociobiol. 23. 69-74. Fleishman. L.J. (1985) Cryptic movemem in the vim' snake Oxybelis aeneus. Copeia. 1985. 242-5. Fleishman. L.J. (1988a) Sensory influences On physical design of a visual display. Anim 8elrav. 36. 142~.
Fleishman. L.J. (1988b) Sensory and environmental influences on display funnin AnolisaurQ/us. a grass anole from Panama. Behav Bcol Sodobio/. 22.309-16. Fleishman. L.J. (1992) The influence of the s~nsory system and the (Ilvironment on motion patlcrns in Ihe visual displays of anoline lizards and other vertebrates. Am Natur. 139.536-561. foley. R.A. & Lee. P.e. p 989) Finite social space. evolutionary pathways. and reconslruCling hominid behavior. Sdence. 243. 901--6. Fonstein. M. & Haselkorn. R. (1995) Physical mapping of bacterial genomes. J Baaeriol. 177. 3361-9. Forrest. T.G. (1994) From sender to receiver: propagation and environ menial effects of acoustic signals. Am Zool. 34. 644-54. Fournier. F. & Fesla-Bianchet. M. (1995) Social dominance in female mountain goats. Anim &hav. 49. 1449-59. Fowler. K. & Partridge. L. (1989) A cost of maling in
409
female fruilflifs. Nalllrt umd. }18. 760-1. Fragaszy. D.M. & Visalberghi. E. (1989) Social influences on the acquisition of tool-using behaviors in tufted capuchin monkeys (Cebus apella). J Comp Psychol. 103(2). J 59-70. Fragaszy. D.M. & VisalberghL E. (1990) Social processes afkcting the appearance of innovative behaviors in capuchin monkeys. Folia Primatol. 54. 155-65. Frank. L.G. (1986) Social organization of the slx'lIted hyena (Crocufll 'ocuta). It. Dominance and reproduClion. Anim Behav. 34. 1510-27. Frank. L.G.. Holekamp. K.E. & Smale. L. (19950) Dominance. demography. and reproduClive success of spotted hyenas. In: Serengeti 11: Dynamics. Managfnlent, and Conservation of an Ecosysttm leds A.R.E. Sinclair & P. Arcese). pp. 364-84. University of Chicago Pre~s. Chicago. Frank. L.G.. Weldele. M.L. & Glickman. S.E. (19~5b) Masculinization costs in hyenas. Nature Lond. 377. 584-5. Frank. S.A. (1987) Variable sex ratio among colonies of ants. Bthav Ecol Sodobiol. 20. 195-201. Franks. N.R .• Blum. M.. Smith. R.-K. & Allies. A.B. (1990) Behavior and chemical disguise of the cuckoo ant Ltptothorax kutten' in relation to Its host Leptothorax acervorum. J C/lem £00/. 16. 1431-44. Framsevich. 1.1. (1982) A compound eye model for the orientation wwards IlOlarized light. Sms Sy5r (Ltl/i'grad).82. 170-83 lin Russian). Frelwell. S.D. (1972) Populations in a SeasJllal Environment. PrincetOn University Press. Princrton. Fretwell, S.D. & Lucas. H.L. (1970) On territorial behaviour and other faClors innuendng habilat distribution in birds. Aaa Biolheoretica. 19. 16-36. Fricke. H.W. (1975) Sozialstruktur und okologische spezialisierung bei verwandten fischen (Pomacentridae). Z rierpsycitol, 39. 492-520. Frischknecht. M (1993) The breeding colouration or male three-spined sticklebacks (Gosrt'rosteus aculearus) as an indicator of energy investment in vigour. Eval Ecol. 7. 439-50. FUluyma. OJ. & Moreno. G. (1988) The evolution of ecological speciaUsalion. Ann Rev Ecot Syst. 19. 20733. Gadagkar. R. (1990) Evolution of eusocialilY: the advantage of assured fitness relUrns. Phil Trans Roy Soc Lend B. 329. 17-25. Gadagkar, R. (1991) Demographic predisposition to the evolution or eusodality: a hierarchy of models. Proc Nat Acad 5..'; USA. 88. 10 993-7. Gadagkar. R. (1994) Why the definition of eusociality is not helpful to understand its evolution and what should we do about it. Oikos. 70. 485-8.
410
REFERENCES
GadgiJ. M. & Bossen. W. (1970) Life history consequences of natural selection. Am Natur. 104. 1-24. Gage. M.J.G. (1991) Risk of sperm competition directly affecls ejaculate size in the Mediterranean Jruil fly. Aninr Bthav. 42. 1036-7. Gage. M.J.G.. Stockley. P. & Parker. G.A. (1996)
Effects of alternative male mating strategies un charaaerislics of sperm production in the Atlantic salmon (Salnro safar): theoretical and empirical inv~ligati()ns. Phil Trans Roy Soc Lend 8, 350. 39l-9. Galer. B.G. Jr (1976) Social transmission of acquired hehavior: a discussion of tradition and socialleaming in venebrates. In: AdvanctS in lh~ Study of Bthavior. Vol 6. (eds J.S. Rosenblall. R.A. Hinde. E. Shaw & C. Beer). pp. 77-99. Academic Press. New York. Galer. B.G. Jr (1988) Imitation in animals: history. definition, and interpretation of dala from the psychological laboratory. In: Social Leamillg: Psychological mId Biologilal P~rspt(tivtS (cds T.R. Zentall & B.G. Galef Jr). pp 3-28. Lawrence· Erlbaum Associates. New Jersey, Galer. B.G, Jr (1990) A historical perspeClive on recent sludies of social learning about foods by Norway ratS, Can J Psychol, 44.311-29. Gald. B,G, Jr (1995) Why behaviour patterns that animals learn socially are locally adapLive, A'im Behav.49. 1325-34. Galer. B.G, Jr & Allen. C. (1995) A new model systcm for studying behavioural traditions in animals. Auim Behav. 50. 705-17. Galer. B.G. Jr. Manzig. L.A. & Field. R.M. (1986) Imitation learning in budgerigars: Dawson and Foss (1965) revisited. BehavProclss. 13. 191-202. Gallistel. C.R. (1990) The Org01,izaIii'n ofLearning. MIT Press. Cambridge Massachussets. Gamboa. G.J, (1996) Kin recognition in social wasps. In: Natural History and Evolution 0..( Paper Wasps (cds S. Thrillazzi & M.J. West-Eberhard). pp. 161-77. Oxford University Press. Oxford, Gamboa. G.J .. Reeve. H.K .. Ferguson. 1.0. & Wacker. T.t. (1986.1) Nestmate recognition in social wasps: the origin and acquisition of recognition odours. Aninr Bthav. 34. 685-95. Gamboa. GJ .. Reeve. H.K. & Pfennig. D.W. (I986b) The evolution and ontogeny of neStmate recog· I11tion in social wasps. Ann Rev Entomol. 31. 43154. Gamboa. G.J .• Berven. K,A,. Schemidt. R.A .. Fishwild. T.G. & Jankens. K.M. (199Ia) Kin recognition by larval wood frogs (RaIJQsylvQtica); e[(ectsof diel and prior exposure to conspecifiC'S. Oecologia. 86. 31924. Gamboa. G.j,. Fo ter. R.L.. Scope. J.A. & Binerman. A.M. (I991b) Erreds of S1age of colony cycle.
conlexl. and intercolony distance on cunspecific tokrance by paper wasps (Polistes fuscallls). Bthav £col Sociobiol. 29. 87-94. Gambua. G.J .• Reeve. H,K. & Holmes. W,G. (1991c) Conccptual issues and methodology in kin· r.:cognilinn research: a critical discussion. EtJroIOf/Y. 88. 109-27. Gamboa. G.J .. Grudzien. T.A .. Espelie. K.E, & Bura. E.A. (1996) Kin recognition pheromones in soda I wasps: combining chemical and behaviural evidence, Anim Bthav, 51. 625-9. Garland. TJ. (1992) Rate testS for phenotypic t'volulion using phylogenetically independent contrasts. Am Natur. 140. 509-19. Gaulin. SJ.C. & FilzGerald. W. (1986) Sex ditrerenCt.'S in spatial ability: an evolutionary hypolhesis and ICSt. Am Narur. 127.74-88. Gaulin. SJ.c. & FitzGerald. W. (1989) Sexual selcction forspalial·learning ability. Anim Ikhav. 37. 322-31. Gebhardt. M,D. & Stearns. S,C. (1988) Reaaion norms for developmental lime and weight at celosion in Drosophila meractorUnl. J Eva! Bioi. I. 335-54, Gendron. R.P. (1986) Searching for cryptic prey: t'vidence furoptimaJ search rates and the formation of search image in quail. Anim Bt!hav. 34. 898-912, Gendron. R.P, & Staddon. J.E,R. (1983) Searching for cryptic prey: lhe effects o( search ralc. Am Nnwr. 1Z1.172-86. Gerhardt. H.C. (1983) Communication and 'he environment. In: CommuniCtJtiol1 (eds T,R. Halliday & P.J.B. Sla,er). pp. 82-113. Blackwell Scientific Publications. Oxford. Gerhardt. H.C. (1992) Conduding playback experiments and interpreting their results. In: Playback and Studies of Animal Commu'ica/;orl (ed. P.K, M<.:Gregor). pp, 59-77, Plenum Press. New York. Gerhardt. H.C. (1994) Reproductive charaaer displacement of female mate choice ill the grey treerrog, flyla chrysoscelis. An;m Bthav. 47. 959-69, Getty. T. ( 1985) Discriminability and the sigmoid func· tiona I response: how optimal foragers can slabilizc model·mimic systems, Am Natllr. 125.239-56. Geny. T. (1993) Search taClics and frequency· dependent prey deteaion. Am Natur. 141,804-11. Getty. T. & Krebs. J.R. (1985) Lagging partial preferences for cryptic prey: a signal detection analysis or great tit foraging. Am Natur. 125. 39-60. GeIlY. T. & Pulliam. H.R. (1993) Search and prey deteClion by foraging sparrows. Ecology. 74. 734-42. GeIlY. T.. Kamil. A.C. & Real. P.G. (1987) Signal detection theory and foraging for cryptic or mimetic prey. In: Foraging Ikhavior (eds A,C. Kamil. J.R. Krcbs & H.R. Pulliam). pp. 525-49. Plenum Press. New York,
REFERENCES Ge17.. W.M. (1981) Genelically bascd kin recogniTion systt'ms. J Theor Bioi. 92. 209-26. Getz, W.M. (1991) The honey bee as a model kin recognilion system. In: K;n Reco!lnilion (ed. P.G. Hepper). pp. 358--112. Cambridge UniversilY Press, Cambridge. Ge,z. W.M. & Chapman. R.E (1987) An odor discrimination model with application 10 kin recognition in social insects. 1m J Neurosci. 32. 96378. Get7. W.M. & Smith. K.B. 11983) Genetic kin recognilion: honeybees discriminate between full and half sisters. Nature. 302. 147-8. Getz. W.M. & Smith. K.B. (1986) Huneybee kin recognition: learning seJr and nest mate phenotypes. Anim BtlJall. 34. 1617-26. Gibbon. J. (1977) Scalar expectancy theory and Weber's law in animal timing. Psy(hol Rell, 84. 279325. Gihbon. J. & Church. R.M. (1990) Represenlalion of lime. Cognition. 37. 23-54. Gibbon. J .. Church. R.M .. Fairhurst. 5. & Kacelnik. A. (t 988) Scalar expectancy theory and choice between ddayed rewards. Psycho/ Rev. 95.102-14. Gibson. R.M. & Hoglund. J. (1992) Copying and sexual selection. Trends Erol Evol. 7. 229-32. Gibson. R.M .. Bradbury. J.W. & Vehrencamp. S.L. (1991) Mate choice in lekking sage grouse revisited: the roles of vocal display. female sill' fiddity. and copying. &hav £eol. 2. 165-80. Gibson. W. & Garside. L. (1990) Kinctoplast DNA minidrdes are inherited from both parents in genetic hybrids of Trypanosoma bniCei. Molec Biochem Parasirol. 42. 45-54. Gilliam. J.F. & Fraser. D.F. (1987) Habitat selection under predation hazard: test of a model with foraging minnows. Ecology. 68. 185~2. Gingerich. P.O. (1993) Quantification and comparison of evolutionary rates. Am J Sci, 293-A. 453-78. Ginsberg, J.R. & Huck. V.W. (1989) Spcnn rom poet ilion in mammals. 7tmds Ecol Eval. 4.. 74-9. Giraldeau. L.-A. (1984) Group foraging: the skill pool effect and frequency-dependent learning. Am Natur. 124.72-9. Giraldeau. L.-A. & Livoreil. B. (1997) Game theory and social foraging. In: Advances i' Game Thforyalld tilt Study of Animal Bthaviour (eds L.A. Dugatkin & H.K. Reeve). in press. Oxford University Press. Oxford. Giraldeau. L.-A. & Lefebvre. L. (1986) Exchangeable producer scrounger roles in a captive flock of reral pigeons. An;m &hav. 34. 797-803. Giraldeau. L.-A. & Lefebvre. L. (1987) Scrounging prevents cultural transmission of food·finding
411
behaviour in pigeons. An;m Behav. 35. 387-94. Giraldeau. L.-A. & Templetun. J.J. (1991) Food scrounging and dirfusion of foraging skill~ in pigeons. Columbalivia: the importance of tutor and ubserver rewards. ErholO!JY, 89. 63-72_ Giraldeau. L.-A.. 5005. C. & Beauchamp. G. (1994a) A test of the producer-scrounger foraging game in captive flocks of spice finches. LonchurD pUI1Cfulata. Bthav Eeol So(iobiol. 34. 25l-6. Giraldeau. L..·A .• Caraco. T. & Valone. T. (1994b) Social foraging: individual learning and cultural transmission of innovations. Behall £col. 5. 35-43. Gittleman. J.l. (1981) The phylogeny of parental care in fishes. Anim Behav. 29. 93~1. Gittleman. J.L.. Anderson, C.G.. Kot. M. & Luh. H.· K. (1996) Comparative tests of evolutionary lalility and rates using molecular phylogenies. In: NfW Uses for New Phyl09t11ia (cds P.H. Harvey. AJ. Leigh Brown. J. Maynard Smith & S. Neel. pp. 289-l07. Oxford University Press. Oxford. Gladstein. D.S.. Carlin. N.F. & Austad. S.N. (1991) The need for sensitivity analyses of dynamic optimization models. Oikos. 60.121-6. Godfray. H.C.J. (1991) Signalling of need between parents and offspring. Nalure Lond. 352, 328-30. Godfray. H.C.J. (1995a) Evolutionary theory of parent-offspring conflict. Nalllre Lond. 376, 133-
8. Godrray. H.C.J. (1995b) Signalling of need between parents and young: parem-offspring conflict and sibling rivalry. Am Natur. 146. 1-24. Godin. J.·G.J. & Briggs. S.E. (1996) Female mate choice under predation risk in the guppy. Anim &hav. 51.117-30. Godin. J.-GJ. & Davis. S.A. (1995al Who dares. benefits: predalor approach behavior in the guppy (PoedJia rerieu/ala) deters predator pursuit. Proc Roy Soc Lond B. 259. 193-200. Godin. J.-G.J. & Davis, S.A. (I995b) Boldness and predator deterrence: a reply to Milinski and Boltshauser. Proc Roy Soc Lond B, 262. 107-12. Goldizen. A.W. (1988) Tamarin and marmoset mating syslt'ms; unusual flexibility. Trends Eco/ £vol. 3. 3640. Goldsmith. T.H. (1991) The evolution of visual pigments and colour vision. In: Vision and Visual Dysfunction. Vol. 6 (ed. P. Gouras). pp. 62-89. Macmillan Press. Houndmills. London. Goodnight. C.J. (1988) Epistasis and 'he effen or founder events on [he additive genetic variance. Evolution. 42. 441-54. Gosler. A.G. (1987) Patlem and process in the bill morphology of the great tit Parus major. Ibis. 129. 451-76.
412
REFERENCES
Goslee. A.G. (1996) Environmental and social determinants of winter fal storage in the great tit Parus major. J Anim Eeo/. 65. 1-17. Gosler, A.G., Greenwood, J.J.O. & PecTins. C.M. (1995) Predation risk and the cost of being fat. Nature, 377.
621-3. Goss-Cuslord. J.D. (1977) The ecology of Ihe Wash. 111. Density-related behaviour and the possible effects of a Joss of feeding grounds on wading birds (Charadrii). J Appl Eeol. 14. 721-39. GOSs-Cu5Iard. J.D. (1980) Competition for food and interfer~nce among waders. Ardt'Q. 68. 31-52. Goss·Custard, J.D. (1985) Foraging behaviour of wading birds and the carrying capacity of estuaries. In: Behavioural Ecology: Ecological Consequences of Adaprive Behaviour (cds R.M. Sibly & R.H. Smith),
pp. 169-88. Blackwell Scientific Publications, Oxford. Goss-CusLard, 1.D. (1993) The effect of migraLion and scale on the study of bird populations: 1991 Wilherby Lecture. Bird Study, 40, 81-96. Goss-CuSlard. J.D. & Durell. S.E.A. Le v. dil. (1988) The effect of dominance and feeding method on the intake rates of oystercatchers, HCwl1atopus ostralegus, feeding on mussels, Mytilus edulis. J Arlim Eeol. 57. 827-44. Goss-CuSlard. J.D .. Caldow. R.W.G .. Clarke. R.T.. Durrell. 5.E.A. Le V. dit & 5u.herland. W.J. (1995a) Deriving population parameters from individual variations in foraging behaviour. I. Empirical game theory distribution model of oystercatchers Haemaropus ostraleglls feeding on mussels Mytilus tdulis. J Anim Ecol, 64, 265-76. Goss-CuSlard. J.D .. Caldow. R.W.G.• Clarke. R.T. & West, A.D. (1995b) Deriving population parameters from individual variations in foraging behaviour. II Model tests and population parameters. J Anim Eeol. 64. 277-89. Goss-CuSlard. J.D.. Caldow. R.W.G .. Clarke. R.T.. Durell, S.E.A. Le V. die Urfi, A.l. & West, A.D. (1995c) Consequences of habitat loss and change lc) populations of wintering migratory birds: predicting the local and global effects from studies of individuals. Ibis. 137. 556-66. Goss-Custard, J.D., Clarke, R.T., Briggs, K.B. et aJ. (1995d) Population consequences of winter habitat loss in a migratory shorebird: I. Estimating model paramelers. J Appl Eeol. 32. 317-33. Goss-Custard, J.D., Clarke, R.T., Durell, S.E.A. Le V. dit, Caldow, R.W.G. & Ens, B.J. (1995e) Population consequences of winter habitat loss in a migratory shorebird: II. Model predictions. lAppl Bcol, 32, 33448.
Goss-Custard, 1.0., West, A.Do, Caldow. R. W.G ..
Clarke. R.T. & Durell. S.E.A. Ie V. dil (1996a) The carrying capacity of coastal habitats for oystercatchers.ln: The Oystercatcher:from Individuals to Populations (ed. J.D. Goss-Cuslard). pp. 327-51. Oxford University Press. Oxford. Goss-Custard. J.D., Durell, S.E.A. Ie V. Oil. Clarke. R.T. et al. (1996b) Population dynamics of the Oystercatcher. In: The Oystercatcller:from Individuals 10 Populations (ed. J.D. GOSs~Cuslar'), pp. 352-83. Oxford University Press. Oxford. Gould, J.L (1986) The locale map of honey bees: do insects have cognitive maps? Science. 232. 8613.
Gould, S.l. (1989) Wonderful Life. Hutchinson. London. Gould. S.J. & Lewol1lin. R.C. (1979) The spandrels of San Marco and the Panglossian paradigm: a critique' of the adaplationist programme. Proc Roy Soc Lond B. 205.581-98. Grafen, A. (1984) Natural selection, kin selection and group selection. In: Behavioural Ecology: an Evolutionary Approach, 2nd edn, (cds 1.R. Krebs & N,B. Davies). pp. 62-84. Blackwell Scientific Publications, Oxford. Grafen, A. (1985) A geometric view of relatedness. In: Oxford Survt'ys in Evolutionary Biology, Vol. 2 (eds R. Dawkins & M. Ridley), pp. 28-89. Oxford University Press, Oxford. Grafen, A. (1986) Split sex ratios and the evolutiunary origins of eusodality. J Theor BioI, 122, 95-121. Grafen, A. (1987) The logic of divisively asymmeuic contests: respect for ownership and the desperado effect. Anim Behav. 35, 462-7. Grafen, A. (1988) On the use of data on lifelime reproductive success. In: Reproductive Success: Studies of Individual Variation in Contrasting Brttding Systems (ed. T.H. Clutlon-Brock), pp. 454-71. The University of Chicago Press. Chicago. Grafen. A. (1990a) Do animals really recognize kin? Anim Behav, 39, 42-54. Grafen, A. (1990b) Sexual selection unhandicapped by the Fisher process. J Theor BioI. 144. 473-516. Grafen, A. (1990c) Biological signals as handicaps. J Theor BioI, 144, 517-46. Grafen, A. (1991) Modelling in behavioural ecology. In: Behavioural Ecology: an Evolutionary Approach, 3rd edn. (eds J.R. Krebs & N.B. Davies). pp. 5-31. Blackwell Scientific Publications, Oxford. Grafen. A. (1992) Of mice and the MHC. Nalllre, 360. 530. Graham, D.S. & MiddlelOn, A.L.A. (1989) Conspecifk recognition by juvenile brown-headed (·owbirds. B;rd Sehav. 8. 14-22. Graveland, J. & Vangijzen T. (1994) Arthropods and seeds are nut sufficient as calciull SOurces for shell
REFERENCES formation and skel('tal growth in passerines. Ardea. 82.299-314. Green, R.F. (l980) Bayesian birds: a simple example of Oaten's stochastic model of optimal foraging. Theor Popul Bioi. 18. 244-56. Green. R.E (1984) Stopping rules for optimal foragers, Am Natur. 123. 80-90. Greenfield, M.D. (1994a) Cooperation and conflict in the evolution of signal interaaions. Ann Rev Bcol SySl. 25. 97-126. Greenfield. M.D. (1994b) Synchronous and alternating choruses in inseCls and anurans: common mechanisms and diverse functions. Am Zoo/. 34, 605-15. Greenfield. M.D. & Roizen. I. (1993) Katydid synchronous chorusing is an evolutionarily stable outcome of female choice. Nature Lend, 364. 61820. Greenwood, P.J. (1980) Mating systems. phiJopatry and dispersal in birds and mammals. Anim &hav. 28. I 14(H;2. Gridley. G.. Mclaughlin. J.K .. Ekbom. A. tlal. (1993) Incidence of cancer among patients with rheumatoid iJrthritis. J Nat Canctr Inn. 85, 307-11. Grinnell. J .. Packer. C. & Pusey. A.E. (1995) Cooperation in male lions: kinship, reciprocity or mutualism? Anim Bella v, 49. 95-105. Gromko. M.Ho, Gilbert. D.G. & Richmond. R.C. (1984) Sperm transfer and use in the multiple mating system of Drosophila. In: Sperm Competition and the Evolution ofAnimal Mating Systtms (ed. R.L. Smith). pp. 371-426. Academic Press. Orlando. Grosberg. R.K. (198g) The evolution of alleorecognition specificity in clonal invertebrates. Quart RnI Bioi. 63. 377-412. Grosberg. R.K. & Quinn. J.E (1986) The genetic control and consequences of kin recognition by the larvae of a colonial marine invertebrate. Nature. 322, 457-9. Gross. M.R. (1994) The evolution of behavioural ecology. Trends Ecol Evol. 9. 358--60. Gross. M.R. (1996) Ahernative reproduClive strategies and tactics: diversity within sexes. Trends Ecol Eva/, II. 92-8. Gohl. A.M .. Collias. N.E. & Allee. W.C. (1945) Mating behavior and the social hierarchy in small flocks of white leghorns. Plrysiol Zoo/. 18. 365-90. Guilford. T. & Dawkins, M.S. (J 987) Search images not proven: a reappraisal of recent evidence. Anim Hehav. 35. 1838-45. Guilford. T. & Dawkins. M.S. (1991) Receiver psychology and the evolution of animal signals. Anim Bthav. 42. 1-14. Guilford. T. & Dawkins. M.S. (1993) Receiver
413
psychology and the design of animal signals. Trmds Neurosa. 16.430-6. Guilford. T. & Dawkins. M.S. (1995) What are conventional signals? Anim &hav. 49, 1689-95. GUSlafsson. L. & Su.herland. W. J. (1988) The costs of reproduClion in the collared Oycatcher Fictdula albicollis. Natu'. 335. 813-15. Gwinner. E. (1990) Circannual rhythms in bird migration: control of temporal patterns and interaction with photoperiod. In: Bird Migratiol1. Physiology and Ecology (ed. E. Gwinner). pp. 25768. Springer-Verlag. Berlin. Gwynne. D.T. (1984) Courtship feeding incrcdses female reproductive success in bush·crickelS. Nature. 307.361-3. Gwynne. D.T. (1984) Male mating effort. confidence ofpalemity, and insect sperm competition. In: Spmn
Comperition and rht Evolurion ofAnimal Maring Systems (ed. R.L. Smith), pp. 117~9. Academic Press. Orlando. Gwynne, D.T. (1988) Counship reeding and the fitness of remale katydids (Onhoptcra: Tcttigoniidae). Evolurion, 42,545-55. Hahn, P.O. & Stuan. A.M. (1987) Sibling inter· anions in IwO species of termites: a lcst of the haplodiploid analogy (Isoptera: Kalotermitidae; Rhinotermilidae). Sociobiology, Il, 83-92. Haig. D. (1992) Genomic imprinting and the theory of parent-offspring conflict. Semin Dewlop Bioi, l. 153-00. Haig. D. (1993) The evolution of unusual chromosomal syslems in sciarid flies: intragenomic confliCl and the sex ratio. J Evel Bioi, 6. 249-61. Haig. D. (1996) Gestational drive and the green-bearded placenta. PrOf I'al Acad Sci USA 93. 6547-51. Haig,. D. & Grafen. A. (199 I) Genetic scrambling as a defence against meiotic drive. J Theor Bioi. 15l. 53158. Haig. D. & Graham. C. (1991) Genomic imprinting and the strange case of the insulin·like growth factor-IJ receptor. Cd/. 64. 104S-{). Hake. M. & Ekman. J. (1988) Finding and sharing dt:plctable patches: when group foraging decreases intake rales. Ornis &and. 19. 275-9. Haldane. J.B.S. (1949) Suggestions as 10 quantitative measurement of rates of evolution. Evolurion. l. 516.
Halliday. T.R. & Verrell. P.A. (1984) Sperm compe.ition in amphibians. In: Spurn Competition and The Evolution ofAnimal Mating SysltmS (cd. R.L. Smith), pp. 487-508. Academic Press. Orlando. Halpin. Z.T. (1991) Kin recognition cues of venebrates. In: Kin Recognition (ed. P.G. Hepper). pp. 220-58. Cambridge University Press. cambridge.
414
REFERENCES
Hamaguchi, K., Ito. Y. & Takenaka, O. (1993) GT dmucleotide repeat polymorph isms in a polygynous ant, uptothorax spinosior and their use for measurement of relatedness. NaturwWf1JSfha!tm. 80. 1798.
Hamilton. W.O. (1964a.b) The genetical evolution of social behaviour. I. 11. J Thear BioI. 7. 1-52. Hamilton. W.O. (1967) Extraordinary sex ralios. Sritnct. U6.477-88. Hamilton. W.O. (1972) Altruism and related phenomena. mainly in social insects. Ann Rev Erol Syst. 3. 193-232. Hamilton, W.O. (1979) Wingless and fighting males in fig wasps and other inseas. In: Sexual ~/tct;olJ and Reproductive Competition in blseC1S (eds M.S. Blum & N.A. Blum). pp. t67-220. Academic Press. New York.
Hamilton. W.O. (1987a) Discriminating nepotism: expectable. common. overlooked. In: Kin Rtcognit;on ill Animals (eds D.J.C. Fletcher & C.O. Michener), pp. 417-37. John Wiley & Sons. New York. Hamilton. W.O. ( I987b) Kinship. recognition. disease. and intelligence: constraints of social evolution. In: Animal Sodttia: TJreori~ and FaCTS (eds Y. Ito. J.l. Brown & J. Kikkawa), pp. 81-102. Japan Scientific Societies Press. Tokyo. Hamilton, W.O. & Zuk. M. (1982) Heritable Hue fitness and bright birds: a role for parasites? Sdmce, 218, 384-6. Hamilton. W.O .. Axelrod. R. & Tanese. R. (1990) Sexual reproduction as an adaptation to resist parasites (a review). ProcNat Acad Sa' USA. 87. 356673. Hammerstein, P. (1981) The role of asymmetries in animal contests. A,rim lkJrav. 29. 193-205. Hammerstein. P. & Parker. G.A. ( 1982) The asymmetric war of aUrition. J Theor Bioi. 96. 647-82. Hanken. J. & Shennan. P. W. (1981) Multiple paternity in Belding's ground squirrel litters. Science Wash. 212. 351-3. Hannon. 5.1.. Mumme. R.L.. Koenig, W.O. & Pitelka, F.A. (1985) Replacement of breC:'ders and within· group conflict in the cooperatively breeding Acorn woodpecker. Behav Eml Sociobiol, 17. 303-12. Harley, C.B. (1981) Learning the evolutionarily stable sirategy. J Thtor Bioi. 89. 611-33. Harrison. R.G. (ed.) (1993) Hybrid lolles and II' Evolutionary Pr«us. Oxford University Press. New York. Harrison, S. & Hastings. A. (1996) Genetic and evolutionary consequences of metapopulation structure. Trends Erol Evol. II. 180-3. Ha~hman. L.G. & Prout. T. (1994) Sperm dis· placement without sperm transfer in Drosophila
mtlanogasltr. Evolution, 48. 758-66. Hanl. D.l. & Clark. A.G. (1989) PrinaplesafPapulatian Gtntlics. Sinauer Associates, Sund~rland. MA. Hartley, I.R., Davies, N.B.. Hatchwelt S.J.• Desrochers. A.. Nebel. D. & Burke. T. (1995) The polygynandrous mating system of the alpine accentor. Prunella col/aris. II. Multiple paternhy and parental effon. Anim IkJrav. 49. 789-803. Hartshorne. C. (1956) The monotony threshold in singing birds. Auk, 95. 758-60. Harvey. P.H. (1996) Phylogenies for ecologists. J Anim £rol. 65, 255-63. Harvey, P. & Bradbury, l.W. (1991) Sexual selection. In: Behawoural Ecology. An Evolutionary Approach. 3rd edn. (eds J.R. Krebs & N.B. Davies). pp. 203-33. Blackwell Sdentific Publications. Oxford. Harvey. P.H. & Krebs. J.R. (1990) Comparing brains. Soma. 249. 140-6. Harvey, P.H. & Pagel. M.D. (1991) Tire Comparative Method in Evolutionary Biology. Oxford University Press. Oxford. Harvey. P.H. & Paxlon. R.J. (1981) The evolution of aposematic coloration. Oikos. 37. 391-3. Harvey. P.I-!. & Purvis. A. (1991) Comparative methods for explaining adaptations. Nawrt, 351. 619-24. Hasegawa, E. (1994) Sex allocation in the ani Colobopsis nipponiws (Wheeler). I. Population sex ralio. Evo/wion, 48.1121-9. Hasegawa, E. & Yamaguchi. T. (1995) Population struaure.local mate competition, and sex-allocation pattern in Ihe ant Messor acicillatus. Evolurion. 49. 260-5. Hasselquist. D.. Bensch. 5. & Von Schantz,. T. (1996) Correlation belween male song repenoire. extra· pair paternity and offspring survival in thl:' great reed warbler. Nature. 381, 229-32. Hastings. I.M. (1992) Population genetic aspects of deleterious CYloplasmic genomes and their effect on the evolution of sexual reproduction. Genet Res. 59. 215--25. Hausfater, G. & Hrdy. 5.B. (1984) IlIfantiridt:Comparatiw and Evolutionary Ptrsptctives. Aldine Press, NY. Hauslater, G.. Altmann. l. & Altmann, S. (1982) Longterm consislency of dominance relalions among female baboons (Papio cynocephalus). Sdenct. 217. 752-5. Healy, D.O., Clayton. N.S. & Krebs. l.R. (t995) Devcfopmenl of hippocampal speaalizalion in two specics of tit (Porus spp.). &lrav Brain Res. 61, 23-8. Healy, S. & Guilford. T. (1990) Olfactory bulb size and nOCIurnality in birds. Evolution. 44. 339-46. Heard. S.B. & Hauser, D.L. (1995) Key innovations and their ecological mechanisms. Histor Bioi. 10. 151-73.
REFERENCES Hedenstr6m. A. «(992) Flight performance in relation to fuel load in birds. J Thtor BioI. 158.535-7. Ikdenstr()m. A. & Alerslam. T. (1995) Optimal flight speed of birds. Phil Tran.< Roy Soc umd B, 348,47187. Hedrick. A.V. & Dill. L.M. (1993) Mate choice by female crickets is innuenced by predation risk. Anim Ikhav, 46, 193-6. Heinsohn. R. & Packer. C. (1995) Complex coo)>t>rative strategies in group· territorial African lions. Selma, 269, 1260-2. Heinze, J. (1995) Reproductive skew and genetic rdatedness in uplothorax ants. PrO( Roy Soc Lond B, 261,375-9. Heinze. J. & Buschinger, A. (1988) Polygyny and hmclional monogyny in Leptothorax ants (Hymenoplera: Formicidae). Psy
Hddmaier. G. (1989) Seasonal acdlmatization of energy requirements in mammals: functional significance of body weight control, hypothermia. torpor and hibernation. In: Entrgy Transformation in etlis and Organisms (eds W. Wieser & E. Gnaiger). pp. 130-9. Thieme, Stuttgart. Heldmaier, G. & Steinlechner. S. (1981) Seasonal control of energy requiremenls for thermoregulation ill the Djungarian hamster (Phodopus sungoms) living in nalUral photoperiod. J Comp Physiol. 142. 42937. Heldmaier. G., Klaus. S., Wiesinger, H.. Friedrichs. U. & Wenzel. M. (1989) Cold acclimation and thermogenesis. In: Living in thtCold II (eds A. Malan & B. Canguilhem), pp, 347-58. John Libbey, Eurotext Ltd london-Paris. I-lemclrijk, C.K. (1994) Support for being groomed in long. tailed macaques. Macaca fasdcularis. Allim Be'av, 48, 479-81. Hemelrijk, C.K. (1996) Reciprocation in apes: from complex cognition to self-structuring. In: Grim Ape Sodetits (eds w.e. McGrew, 1.F. Marchant & T. Nishida), pp. 185-95. University Press, Cambridge. Henderson, M.T.G .• Merriam. G. & Wegner, J. (1985) Patchy environments and species survival: chipmunks in an agricultural mosaic. Bioi ConStrv, 31,95-105.
415
Hengeveld. R. (1989) Dynamics of Biological Jnva~ions. Chapman & Hall. London. Hepper, P.G. & Waldman, B. (1992) Embryonic olfactory learning in frogs. Quarl J Exp Psychol B. 44, 179-97. Herbers, J.M. (1990) Reproductive investment and allocatiol~ ratios for the a III LtpIOrhorax longispinosus: orting out the variation. Am Natur. 136, 178-208. Herbers, J.M. (1993) Ecological delerminants of queen number in ants. In: Quem Numbtr and Sociality in InYclS (ed. L. Keller), pp. 262-93. Oxford University Press. Oxford. Heusner, A.A. (1987) Whal does the power fundion reveal about structure and function in animals of different size? Am Rev Physiol. 49, 121-33. Hewitt, G.M. (1988) Hybrid zones - natural laboratories for evolutionary studies. Trends Bcol Eva/, 3, 158--67. Hewitt. G.M. (1993a) After the ice - paral/elus meets eryrhroplls in the Pyrenees. In: Hybrid ZontS and th( Evolutionary Pr.1aSS (ed. R.G. Harrison), pp. 14D-64. Oxford UniversilY Press. New York. Hewitt. G.M. (1993b) Postglacial diStribution and species substructure; lessons from pollen. insects and hybrid zones. In: Evolutionary Patrtrns and Processes (eds D.R. Lees & D. Edwards). pp. 97-123. Linnaean Society of London. Academic Press, London. Hewitt. G.M. (1996) Some genetic consequences of ice ages, and their role in divergence and spedalion. Bioi J Linn Soc. 58, 247-76. Heyes. C.M. (1993) Imitation, culture and cognition. Anim lkhav. 46, 999-1010. Heyes. CM. & Dawson, G.R. (1990) A demonstration of observational learning using a bidirectional ronnol. Quart J Exp Psychol. 428, 59-71. Hickey, D.A. (1982) Selfish DNA: a sexually transmiued nuclear paraSite. Genetics, 101. 519-31. Hicks. G.R. & Raikhel. N.V. (1995) Protein import into the nucleus: an integrated view. Ann RevCtl/ Dewlop 8iol, II. 155-88. Higashi. M., Abc, T. & Burns, T.P. (1992) CarbonnitrOgen balance and termite ecology. PrO( Roy Soc Lond B, 249, 303-8. Higashi. M .• Yamamura. N.. Abe. T. & Bums, T.P. (1991) Why don't all termite species have a sterile worker casle? Proc Roy Soc Lond 8. 246. 25-9. Hill, G.E. (1990) Female house finches prefer colourful males: sexual selection for a condition-dependent trail. Anim Bthav. 40, 563-72. Hill. G.E. (1991) Plumage colouration is a sexually selected indicatOr of male quality. Nalure Lond, iSO, 337-9. Hill, G.E. (1992) The proximate basis of variation in
416
REFERENCES
carotenoid pigmentation in male house finches. Auk. 109. 1-12. Hill. G.E. (1994) Geographic variation in male ornamentation and female mate preference in the house finch: a comparative test of models of sexual sdeC1ion. Bthav ficol. 5. 64-73. Hill. G.E. & Montgomerie. R. (1994) Plumage colour signals nutritional condition in the house finch. Proc R>y So< Lond B. 258. 47-52. Hillis. D.M.. Moritz. C. & Mable. B.K. (1996) Molecular SystematiC$. 2nd edn. Sinauer. Sunderland. Hinde. R.A. (1983) A conceptual framework. In: Primou Social Relationships (ed. R.A. Hinde). pp. 17. Sinauer. Sunderland. Massachusetts. Hogsledl. G. (I980) Evolution of clutch size in birds: adaptive variation in relation 10 territory quality. Sl'imre. 210. 114l1-50. Hogendoom. K. & Velthuis. H.H.W. (1993) The sodality of Xylocopa pubeSClllS: does a helper really help? Behav Ecol Sociobiol. 32. 247-57. Hogendoorn. K. & Vellhuis. H.H.W. (1995) The role 01 young guards in Xylocopa pubtsftlls. Imler Soc. 42. 427-48. Hoggren. M. (1995) Mati119 strategies and sperm competition ill the adder (V;pera herus). PhD thesis. Uppsala University. Hoglund. J. (1989) Size and plumage dimorphism
in lek·breeding birds: a comparative analysis. Am Nalllr, 134.72-87. Hoglund. J. & AJalalo. R.V. (19951 Leks. Princeton University Press. PrincelOn. New Jersey. HOglund. J. & Silien-n,lIberg. B. (1996) Does lekking
promote the evolution of male biased size dimorphism in birds? On the use of comparative
approaches. Am Nalur (in press). HOglund. J .• Alalalo. R.V. & Lundberg. A. (1990) Cupying the mate choice of others, observation on female Black grouse. lkhavr·our. 114. 221-31. Hogslad. O. (1987) II is expensive to be dominant. Auk. 104. 333-<;. Hoikkala, A. & Kaneshiro, K.Y. (1993) Change in the signal-response sequcnce responsible for asymmctric isolation between Drosophila planitibia and Drosophila silvtstris. Proc Nat Acad USA. 90. 5813-17. Hoikkala. A. & Wei bergen. P. (1995) Signals and responses of females and males in successful and unsuccessful courtships of thret· Hawaiian lek· mating Drosophila spedes. Anim Behav. SO, 177-90. Hokil. D.G .• Walls. S.C. & Blaustein. A.R. (1996) Context-dependem kin discrimination in larvae of the marbled salamander, Ambystoma opa(um. Anim &hav, 52, 17-31. Hiilldobler. B. & Wilson. E.O. (1990) Tire An'. Springer-Verlag. Berlin.
Sa
Holmes. W.G. (1984) Ontogeny of dam-young recognition in captive Belding's ground squirrels. J Comp Psychol. 98. 246-56. Holmes, w.e. (19863) Kin recognition by phenotype matching in female Belding's ground squirrels. Anim Behav. 34, 38-47. Holmes. W.G. (1986b) Identification of paternal half siblings by captive Belding's ground squirrels. Anim Behav, 34, 321-7. Holmes, W.G. (1994) The development of Iittermale preferences in juvenile Belding's ground squirrels. Anim Behav. 48. 1071-84. Holmes. W.G. & Sherman. P.W. (1982) The ontogeny of kin recognition in two species of ground squirrels. Am Zool. 22. 491-517. Hulmgren. N. (1995) The ideal Iree distribution 01 unequal competitors: predictions from a behaviourbased functional response. J Aninr Em/. 64... 197-212. Holmgren, N. & Hedenstrom, A. (1995) The scheduling of molt in migratory birds. Evel £col. 9. 354-68. Homann, H. (J 928) Beitrage zur Physiologie der Springspinnen. Z VeTgl Physiol, 7. 201-68. Honda, K. (1995) Chemical basis of differential oviposition by lepidopterous insects. Arch Insrc[ Biochem Physiol. 30, 1-23. Hoogland. J.L. & Sherman. P.W. (1976) Advantages and disadvantages of bank swallow (Riparia riparia) coloniality. Ecol Monogr. 46. 33-58. Houde, A.E. (1987) Mate choice based upon naturally occurring colour pattern variation in a guppy population. Evolmian, 41. l-lO. Houde, A.E. (1993) Evolution by sexual seleC1ion: what can population comparisons tell us? Ani Natur, 141. 796-803. Houde. A.E. & Endler, J.A. ( 1990) Correlated evolution of female mating preferences and male color patterns in the Guppy Potd/ia retieulata. Sdtna, 248. 140~. Houde, A.E. & Torio. AJ. (1992) Effect of parasitic infection on male color pattern and female choice in guppies. 8rhav £(01, 3, 346-51. HOuSton, A.I. (1986) The optimal night velocity for a bird explOiting patches of food. J Thtor Bioi. 119, 345-<;2. Houston, A.l. (1993) The eHiciency of mass-loss in breeding birds. Proe Roy Soc Lond B. 254. 221-5. Houston. A.1. (1993) The importance of state. In: Diet Selection: an Interdisaplinary Approach 10 Foraging Behaviour (ed. R.N. Hughes). pp. 10-31. Blackwell Sdentific Publications, Oxford. Houston, A.1. (1995) Energetic constraints and foraging efficiency. &hav Beol. 6. 393--6. Houston. A.I. & McNamara, J.M. (1988) Fighting for food: a dynamic version of the Hawk-Dove game. Evol Em/, 2,51-64.
REFERENCES Houston. A.1. & Mc amara. J.M. (1988) There's no such thing as a free lunch. Behav Brain Sri. II. 1549.
HouslOn. A.1. & McNamara. J.M. (1992) Phenotypic plasticity as a state·dependent liCe·hislory dedsion. £ivai &01. 6. 24l-53. Houston. A.1. & McNamara, J.M. (1993) A theoretical investigation of the (at reserves and mortality levels of small birds in winter. Ornis SCQnd. 24. 20519. Houston, A.I. & Sumida. B,H. (1987) Learning rules, matching and frequency·dependence. J Tlltor 8iol. 126.289-308. Houston. A.I., McNamara. J.M. & Hutchinson. J.M.C. (1993) General results concerning the trade·off between gaining energy and avoiding predation. Phil Trons Roy Soc Lolld B. 341. 375-97. Houston, A.I.. McNamara. J.M. & Thomson, W.A. (1992) On the need for a sensitive analysis of optimization models, or 'the simulation is not as the former'. Oikos. 63,513-17, Houston, A.I.. Schmid· Hempel, P. & Kacelnik. A. (1988) Foraging strategy, worker monality. and the growth of the colony in social insects. Am Natur, 131. 107-14. Houston. A.I., Clark, C.W.. McNamara, J.M. & MangeL M. (1988) Dynamic models in behavioural and evolutionary ecology. Nalure, 332. 29-34. Houston. D.C.. Donnan, D. & Jones. P.J. (1995a) Use of labelled methionine to investigate the contribution of muscle proteins lO egg produaion in zebra finches. J Comp Physiol B. 165. 161-4. Houston. D.C.. Donnan. D.• Jones. P.. Hamilton, I. & Osborn. D. (1995b) Changes in the muscle condition of remal~ Zebra Findles Potphila gut/ala during egg laying and the role of protein storage in bird skeletal muscle. Ibis. 137. 322-8. Howard. R.D .. Whileman. H.H.&Schueller. T.l. (1994) Sexual selection in American toads: a test or a good· genes hypothesis. Evolution, 48, 1286-1300. Hoy. R.R .. Hahn. J. & Paul. R.C. (1977) Hybrid cricket auditory behavior: evidence for genetic coupling in animal communication. Sdmce. 195.82-4. Hoy, R.R .. Hoikkala, A. & Kaneshiro. K. Y. (1988) Hawaiian counship songs: evolutionary innovation in communication signals in Drosophila. Sdence. 240, 217-19. Hubbell. S.P. & Johnson. L.K. (1987) Environmental variance in Iiretime mating success. mate choire and sexual selection. Am Nalur. no. 91-1 12. Huck. U.. Labov. J. & Usk. R. (1987~ Food-restricting first generation juvenile female hamsters (MtsOCrictlUS auratus) aUtOs sex ratio and growth of third generarion oHspring. BioI Reprod. 37. 612-17.
417
Huelsenbeck. J.P.. Bull. J.J. & Cunningham. C.W. (1996) Combining data in phylogenetiC analysis. Trmlfs £icol £ivai. II. 152-8. Hughes-Schrader. S. (1948) Cytology of coccids (Coccoidea - Homoptera). Adv Gmtl. 2. 127-203. Hunte. W. & Horrocks. J. (1986) Kin and non-kin interventions 111 the aggressive disputes of vervet monkeys. Behav Ecol Sociobiol. 20. 257-63. Hunte, W.• Myers, R.A. & Doyle. R.W. (1985) Bayesian mating dedsions in an amphipod. Gammarus lawrendanus Bousfield. Anim Behav. 33. 366-72. Hunter. F.M.. Burke. T.A. & Walls. S.E. (1992) Frequent copulation as a method of paternity assurance in the Nonhero Fulmar. Anim Beha-', 44. 149-56. Hunter, EM.. Pelrie. M.. Otronen. M., Birkhead. T.R. & Meller. A.P. (1993) Why do females copulate repeatedly with one male? Trtnds Ecol Evol. 8. 21-6. Hunter·Jones. P. (1960) Fertilization of eggs of the desen locust by spermatozoa from successive copulations. Nalure (Lond), 185. 336. Huntingford, F.A. & Turner. A,K. (1987) Animal Conflict. Chapman & Hall. London. Huntingtord. F.A .. Lazarus, J .• Barrie, B.D. & Webb, S. (1994) A dynamic analysis or cooperative predator inspection in sticklebacks. Ani', Behav. 47. 41323. Hurly. T.A. () 992) Energetic reserves of marsh tits (Parus paIIlSlris): food and fat storage in response to variable food supply. &hov £icol. 3. 181-8. Hurst. LO. (1990) Parasite diversity and the evolution of diploidy. multicellularity and anisogamy. J Thtor Bioi. 144.429-3. Hurs•. L.D. & Hamillon. W.O. (1992) Cytoplasmic fusion and the nature of sexes. PrO( Roy Six Ltmd B. 247. 189-94. Hum. L.D.. Allan. A. & Beng..son. B.O. (1996) Genetic conflicls Quart RtlI BioI, 71. 317-64. Hutchinson. J.. McNamara. J.M. & Cuthill.I.C. (1993) Song. sexual selection. starvation and strategic handicaps. Amm Behav, 45. 1153-77. Huxley. 1.5. (1912) A 'disharmony' in the reproductive habits of the wild duck (Anas bose/ras L.). BioI Zmtralbl. 32. 621-3. Huxley. J.S. (1938) Darwin's theory of sexual selection and the data subserved by it. in the light of recent research. Am Nalllr. 72. 416-33. Huxley. J.S. (1940) The New Systtmalics. Clarendon Press. Oxford. Ibrahim. K.M .• Nichols. R.A. & Hewitt. G.M. {1996} Spatial patterns of genetic variation generated by different forms of dispersal during range expansion. Htrtdiry. 77. 282-91. lIIius. A.W.. Albon. S.D.. Pemberton. J.M .. Gordon.
418
REFERENCES
I.J. & elutton-Brock, T.H. (1995) Seleaion for
foraging efficiency during a population crash in Soay sheep. J Anim Ecol. 64. 481-92. Inman. A .• Lefebvre, L. & Giraldeau. l.-A. (1988)
Individual diet differences in feral pigeons: evidence for resource panilioning. An;m &hav, 35. 19023. Irons. W (1990) Let's make our perspective broader rather (han narrower. mhol Sodobiol. 11. 361-74. Y. (1993) Behaviour and Sodal Evolution of Wasps: the Communal Agsregation Hypothesis. Oxford University Press, Oxford. Iwasa, Y. & Pomiankowski. A. (1994) The evolution of mate preferences for multiple sexual ornaments. Evolution, 48. 853-67.
Ito.
Iwasa. Y., Higashi. W. & Yamamura. N. (1981) Prey
distribution as a factor determining the choice of optimal foraging strategy. Am Nalur, 117, 71023. Jackson, R.R. & Blest, A.D. (1982) The distances at which a primitive jumping spider, Portia fimbriata, makes visual discriminalions. J Exp Bioi, 97. 441-5. Jacobs, L. (1995) The ecology of spatial cognition. In: Bthovioral Brain RtStarch ill Nalllralish'c and Smri· naturalistic sellings (cds E. Alleva, A. Fasolo, H.P. Lipp.l. Nadel & l. Ricceri), pp. 301-22. NATO ASI Series, Kluwer Academic Press. Dordrt'cht. Jacobs, L.E, Gaulin, S.J.C., Sherry, D.E & Hoffman, G. (1990) Evolution of spatial cognition: sex-specific patterns of spatial behavior predict hippocampal size. Pro< Nat Acad Sa USA, 87, 6349-52. Jacobs. M.E. (1955) Studies on territorialism and st'xual selection in dragonOies. ECoJlogy. 36. 566-86. Jaenike, J. & Holl. R.D. (1991) Genetic variation for habitat preference: evidence and explanations. Am Namr. 137(Suppl.), S67-S90. Jamieson. LG., Quinn. J.S., Rose, P.A. & White. B.N. (1994) Shared paternity among non~relalives is a resull of an egalitarian mating system in a communally breeding bird, the pukeko. Proc R Soc L'd B, 257, 271-7. JanclOS. A.C. (1980) Strategies of female mate choice: a theoretical analysis. Bthav Bcol Sociobiol. 7, 10712. Janis, C.M. (1996) Do legs support the arms race hypothesis in mammalian predator/prey relationships. In: Vertebrate Behavior as Dfrivtd from lilt Fossil Rtcord (eds J.R. Horner & L Ellis), (in press). Columbia Universily Press. New York. Jarman, PJ. (1974) The social organisation of antelope in relation (0 their ecology. Behaviour, 48, 2 I 5-67. Jeffreys, A.J., Wilson, V. & Thein, S.L (1985) Hypervariable 'minisatellite' regions in human DNA. Natu' Lond, 314. 67-73.
Jenni, D.A. (1974) Evolution of polyandry in birds. Am ZOOI, 14, 129-44. Jennions. M.D. & Oakes. E.J. (1994) Symmetry and sexual selection. Trends E(ol Eval. 9, 440. Jennions, M.D. & Passmore. N.!. (1993) Sperm competllion In frogs: testis size and a 'sterile male' experiment on Chiromantis xeramptlina (Rhacophoridae). Bioi J Linll Soc. 50. 21 1-20. Jimenez, J.A .. Hughes. K.A .. Alaks. G., Graham, L & Lacy. R.C. (1994) An experimental study or inbreeding depression in a natural habitat. Science, 266.271-3. Johnson, J. (1987) Dominance rank in juvenile olive baboons. Papio anubis: the influence of gender. size. maternal rank and orphaning. Anim Bthav, 35. 1694-1708. Johnson, M.L & Gaines, M.S. (1990) Evolution of dispersal: theoretical models and empirical tests using birds and mammals. Ann Rev EaJI Syst, 21. 44980. Johnson, M.S. & Black, R. (1995) Neighborhood size and lhe importance of barriers to gene now in an intertidal snail. Htrtdity. 75, 142-54. Johnson, M.S., Murray, J. & Clarke, B. (1993) The ecological genetics and adaptive radiation of Partula on Moorea. OxfSurv Eval Bioi. 9. 167-238. Johnstone, R.A. (1994) Female preference for symmetrical males as a by-produd of selection for male recognition. Nature, 372. 172-5. Johnstone. R.A. (19953) Sexual selection. honest advenisemem and the handicap principle: reviewing the evidence. Bioi Rev. 70, 1-65. Johnstune. R.A. (1995b) Honest advertisemenl of multiple qualities using multiple signals. J Th~or 8;01. 177,87-94. Johnstone. R.A. (1996a) Game theory and communication. In: Gome Thtory and tht Study ofAnimal Behaviour (eds L.A. Dugalkin & H.K. Reeve). Oxford University Press, Oxford (in press). Johnstone, R.A. (1996b) Multiple displays in animal communicalion: 'backup signals' and 'multiple messages'. Phil Trans Roy Soc Lond B. 351, 329-38. Johnstont:. R.A. & Grafen, A. (1992) The mnlinuous Sir Philip Sidney game: a simple model of biological signalling. j Thtor BioI, 156, 215-34. Johnslone. R.A. & Grafen. A. (1993) Dishonesty and the handicap principle. Anim &hav. 46. 759-64. Johnstone, R.A., Reynolds, J.D. & Deutsch, J.e. (1996) Mutual mate choice and sex differences in choosiness. Evolution. 50. 1382-91. Jones, G. (1986) Sexual chases in sand marlins (Riparia ripon'a): cues for males to incrt.'ase lheir reproductive success. &hav Ecol Sodobiol. 19, 17985.
REFERENCES Jones. M.M. (1991) Muscle protein loss in laying House sparrows Passer domestialS. Ibis. 133. 193-8. Kacelnik. A. (1989) Short-term adjustment of paren'al ef(on in starlings. In: Acta XI)( CongrtsSus Intn-nationalis Omithologiri. Vol. n. (ed. H. Ouellet). pp. 1843-56. University or Onawa Press. Onawa. Kacelnik. A. & Krebs. J.R. (1985) Learning 10 exploit palchily distributed prey. In: lkhQviotlfQI Ecology. British Ecological Society Symposium (eds R.M. Sibly & R.H. Smith~. pp. '89-205. Blackwell Scientiric Publications. Oxrord. Kacelnik. A. & Todd. LA. (1992) Psychological mechanisms and the marginal value theorem: errect or variability in traveltime on patch exploitation. Anim Bthav. 43. 313-22. Kacelnik. A .• Brunner. D. & Gibbon. J. (1990) Timing mechanisms in optimal roraging: some applications or scalar expectancy theory. In: Behavioural Mechanisms of Food Selection. NATO ASI series Vol. G20 (ed. R.N. Hughes). pp. 61-82. Springer-Verlag. Berlin. Kacelnik. A.. Krebs. J.R. & Ens. B. (1987) Foraging in a changing environmenl: an experiment with starlings (Sturnus vulgaris). In; Harvard
Symposium Oil the Quantitative Analysis of Behavior Vol. 6 Foraging (cds M.L. Commons. A. Kaceh'lik & S.J. Shettleworth), pp. 63-87. Lawrence·Erlbaum Associates. Hillsdale, NJ. Kacelnik, A.• COlton. P.A., Stirling, L. & Wright, J. (1995) Food allocation among nestling starlings: sibling competition and the scope orparemal choice. Proc Roy Soc Land B. 259. 259-63. Kaiser, D. (1986) Control or multicellular development: Di'yosteJium and Myxocoaus. Ann R~vGm~t, 20,53966. Kalmus. H. (1932) ueber die Erhahungswen der PhanoIypischen (Morphologischen) Anisogamic und die EnlStehung der Erstem Geslechtsunterschiede. Bioi Ztntralblatt, 552, 716--26. Kambhampati, S. (1995) A phylogeny or cockroaches and related insects based on DNA sequence or mitochondrial ribosomal RNA genes. Proc Nat A(ad Sri USA. 92. 2017-20. Kami. A.C .. Balda. R.P. & Olson. D. (1994) Performance or rour seed-caching corvid species in the radial-arm maze analog. J Comp Psychol. 108. 38593. Kaneshiro, K.Y. (1976) Ethological isolation and phylogeny in the planitibia subgroup or Hawaiian Drosophila. Evolwion. 30, 740-5. Kaneshiro, K. Y. (1989) Dynamics or sexual selection and rounder cHern in species rorrnation.~Gent'tics. Speaationand th~ Founder Prindplt (eds L.V. Giddings, K.Y. Kaneshiro & W.W. Anderson), pp. 279-96.
419
Oxrord Universlly Press, New York. Kareem. A.M. & Barnard. C.J. (1986) Kin recogni'ion in mice: age, sex, and parental dreas. Anim Btltav. 34.1814-24. Kasuya. E.. Hibino. Y. & Ito. Y. (1980) On 'intercolonial' cannibalism in Japanese paper wasps. Polistts chinensis antennalis Perez and P.jadwigot Dalla Torre (Hymenoplera; Vespidae). Rfi Popul £col (Kyolo). 22. 255-62. Kavaliers, M. & CUlwell. D.O. (1995) Odours or parasitized males induce aversive responses in remale mice. A'im Bthav, 50.1161-9. Kawai, M. (1958) In the system of social ranks 10 a nalUraltroop of Japanese monkeys (I); basic rank and dependent rank. Primates. 1, 111~8. Kawai. M. (1965) Newly acquired pre·cultural behavior of the natural troop or Japanese monkeys on Koshima Islet. Primates. 6. 1-30. Keane. B.. Waser, P.M., Creel. S.M .• Creel. N.M., Elliotl. L.F. & Minchella. OJ. (1994) Subordinale reproduction in dwarf mongooses. Anim Behav. 47. 65-75. Keller. L. (1995) Sodal lire: the paradox of multiplequeen colonies. Trends Eco! Evol, 10, 355-60. Keller. L. & Nonacs. P. (1993) The role of queen pheromones in social insects: Queen control or queen signal? Anim Behav, 4', 787-94. Keller, L. & Reeve. H.K. (1994) Partitioning or reproduction in animal societies. Trends £col Evol, 9.98-102. Keller, L. & Vargo, E.L. (1993) Reproductive structure and reproductive roles in colonies or eusorial insects. In: Queen Numb~r and Sociality in Inseds (ed. L. Keller), Pl'. 16--44. Oxford University Press, Oxrord. Keller. L.E, Arcese. P., Smith, J.N.M., Hochacka, W.M. & Stearns, S.c. (1994) Selection against inbred song sparrows during a natural population bottleneck. NOlUrr. 372. 356-7. Kempenaers, B. & Sheldon, B.C. (1996) Why do male birds not discriminate between their own and extra-pairorrspring? Anim &hav. 51, 1165-73. Kempenaers, B.. Verheyen, C.R .. Broeck, M.V.D .. Burke. T.. Broeckhoven. C. V. & Dhondt. A.A. (1992) EXlra·pair patc:rnity results from remale prererence ror high-quality males in the blue til. Naturt, 357, 494-6. Kendeigh, S.c.. KOnlogiannis, J.E .. Maza, A. & Roth, R. (1969)~onmental regulation or rood intake 5y birds. Comp Biochtm Physiol. 31. 941-57. Kennedy. C.E.J.. Endler. J.A .. Poyton. S.L. & McMinn. H. (1987) Parasite load predicts mate choice in guppies. Behal' Ecol Sodobiol, 17. 199-206. Kennedy. J.S. 11992) The New Anthropomorphism.
420
REFERENCES
Cambridge University Press. Cambridge. Kent. D.S. & Simpson. l.A. (1992) Eusociality in
the beetle Austroplatypus incompenus (Coleoptera: Curculionidae). Naturwissenschaften. 79. 86-7. Kersten, M. & Piersma. T. (1987) High levels or energy
expenditure in shorebirds: metabolic adaptations to an energetically expensive way of life. Ardea. 7S. 175-88. Kilner. R. (1995) When do canary parents respond 10 nestling signals of need? Proc R'y Soc umd B, 260. 343-8. King, J.R. & Murphy. M.E. (1985) Periods 01 nutritional stress in the annual cycles of endOlherms: fact or finion? Am ZOOI, 25. 955-64. Kinlen, LJ. (1992) Malignancy and autoimmune diseases. J Autoimmun. 5(A), 363-71. Kirkpatrick, M. (1982) Sexual selection and the evolution of female choice. Evolution. 36, 1-12. Kirkpatrick. M. & Dugatkin, L.A. (1994) Sexual
selection and the evolutionary effects of copying mate choice. /khav Ec:ol SOdobio/. 34. 443-9. Kirkpatrick, M. & Barton. N.H. (1996) The strength of indirect selecrion on female mating preferences. Proc Nat Arod sa (USA), in press. Kirkpatrick, M. & Ryan, M.J. (1991) The paradox of the lek and the evolution of mating preferences. Natu,.. 350. 33-8. Kirkwood, J.K. (1983) A limit to metabolisable energy intake in mammals and birds. Comp Biochtm Physiol. 75A,I-3. Klahn. J.E. (1988) Intraspecific comb usurpation in the social wasp Polistes fuscatus. Behav Bcol Sodobiol. 23,1-8. Kleiber. M. (1947) Body size and metabolic rate. Physiol Rev. 27, 5 tl-41. Klomp. H. (1970) The determination of c1Ulch·size in birds. A review. Ardea. 58. 2-123. Klopfer, P. (1961) Observationalleaming in birds: the establishment of behavioral modes. Ikhaviour. 17. 71-9. Knowlton. N. (1974) A note on the evolution of gamete dimorphism. J Theor Bioi, 46. 283-5. Knowlton, N.. Weigt, L.A.. Sol6rzano, L.A.. Mills. O.K. & Bermingham. E. (1993) Divergence in proteins. mitochondrial DNA. and reproductive compatibility across the Isthmus of Panama. Science, 260, 162932. Kodric~Brown, A. (1989) Oierary caroleooids clOd male mating success in the guppy: an environmental component to female choice. Bthav Bcol Sodobiol, 25, 393-40 I. Kodrlc-Brown, A. (1985) Female preference and sexual selection for male colouration in the guppy Poedlia rerieu/ata. Bthav £col Sociobiol. 25, 393-40 I.
Kodric-Brown, A. (1993) Female choice of multiple male criteria in guppies: interacting effects of dominance. coloration and courtship. Bellav £col Sociobjol. 32. 415-20. Koenig, W.O. (1990) Opportunity of parentage and nest destruction in polygynandrous acorn woodpeckers. &hav Eral, I, 55-61. Koenig. W.O. & Mumme, R.L. (1987) Population ecology of the cooperativt bruding acorn woodpecker. Mono· graph in Population Biology, No. 24. Princeton University Press, Princeron, New Jersey. Koenig. W.D.. Mumme. R.L., Stanback. M.T. & Pitelka. EA. (1995) Patterns and consequences of egg destruaion ~mongjoint~ncstingacorn woodpeckers. Anim &hav. 50. 607-21. Koenig. W.O., Pitelka, F.A.. Carolen. WJ.. Mumme, R.L. & Stanback. M.T. (1992) The evolution of delayed dispersal in cooperative breeders. Quart Rev 8iol. 67, lit-SO. Koford, R.R., Bowen, B.S. & Vehrencamp, S.L. (1990) Groove-billed Anis: joint-nesting in a tropical cuckoo. In: Cooperativt Breeding in Birds: Long-ttrm Studies in Ecology and Bthavior (cds P.B. Sracey & W.O. Koenig). pp. 335-55. Cambridge University Press, Cambridge. Komdeur, J. (1992) Imponance of habitat saluration and terrilory quality for evolution of cooperative breeding in the Seychelles warbler. Nature. 358, 493-5. Komdeur, J. (1994) The eneet of kinship on helping in the cooperative breeding Seychelles warbler (Acrocepha/us seychellemis). Proc Roy Soc umd B. 256. 47-52. Konarzewski. M. & Diamond, J. (1994) Peak sustained metabolic rale and its individual varialion in cold· stressed mice. Physiol Zool, 67.1186-212. Konlogiannis. J.E. (1967) Day and night changes in body weight in the white·crowned sparrow (ZOnotrichia. albicollis). Auk, 84, 390-5. Koops, M. & Giraldeau, L.-A. (1996) Producerscrounger foraging games in starlings: a test of mean-maximizing and risk-minimizing foraging models. Anim BeJrav. '1, 773-83. Konner. G. & Heldmaier. G. (1995) Body weight cycles and energy balance in the alpine marmot (Mannota mannota). Physiol ZOOI, 68, 149-63. Kozlowski, J. (1992) Optimal allocation of resources w growth and reproduction: implications for age and size at maturity. Trends £co/ Eval. 7. 15-19. Kozlowski. J. (1996) Optimal allocation of resources explains interspedfic life-history patterns in animals with indelerminale growth. PrO( Roy Soc Lond B. 263. 559-66. Kozlowski, J. & Uchmanski, J. (1987) Optimal
REFERE N CE S individual growth and reproduction in perennial species with indeterminate growth. Evol Eeo!, 1. 214-30. Kozlowski. J. & Weiner, J. (1996) Interspecific allometries are byproduets of body size optimization. Am Natur (in press). Kozlowski. J. & Wiegert. R. (1986) Optimal allocation of energy 10 growth and reprodudion. Theor Popul Bioi. 29. t6-37. Kozlowski. J. & Wiegerl, R.G. (1987) Optimal age and size at maturity in annuals and perennials with determinate growth. Evol Eeol, 1. 231-44. Krakaucr. D.C. & Johnstone, R.A. (1995) The evolution of exploitation and honesty in animal communication: a model using artificial neural networks. Phil Trans Roy Soc Lond B. 348. 355-61. Krebs, J.R. (1970) The efficiency of courtship feeding in blue tits. Ibis. 112. 108-10. Krebs, J.R. (1973) Social learning and the significance of mixed-species nocks of Chickadees. Can J Zool, 51. 1275-88. Krebs, J.R. (1982) Territorial defence in the great tit Pams major. do residents always win? Behav Ecol Sociobiol, 11. 185-94. Krebs. J.R. (1990) Food storing birds: adaptive specialization in brain and behaviour. Phil TraIlS Roy Soc l.alrd B. 329. 153-60. Krebs, J.R. & Davies, N.B. (1993) An Jlltrodu,tion to Behavioural Ecology, 3rd edn. Blackwell Scientific rublicalions, Oxford. Krt:'bs, J.R. & Dawkins. R. (1984) Animal signals: mind-reading and manipulation. In: Behavioural Ecology: an Evolutionary Approach, 2nd edn (cds J.R Krebs & N.B. Davies). pp. 380-402. Blackwell Scit:'ntific Publications. Oxford. Krt:'bs. J.R. & Inman. A.J. (1992) Learning and foraging: individuals, groups and populations. Am Nalur, 140, 563-584. Krebs, J.R. & Kacelnik, A. (1991) Oedsion-making. In: Behavioural Ecology. An Evolutionary Approach (cds l.R. Krebs & N.B. Davies). 3rd edn. PI'. 105-36. Blackwell Scientific Publications, Oxford. Krebs. l.R .. Healy. S.D. & Shenleworlh. S.l. (t 990) Spalial memory of Paridae: comparisun of slUring and non-storing species. the coal tit. Parus ater. and Iht' grt:'al til. P. major. Anim Behav, 39. 1117-37. Krebs. J.R .. MacRoberts. M.H. & Cullen. J.M. (1972) Flocking and feeding in the great tit Parus majoran experimemal study. Ibis. 114.507-30. Krebs, J.R .. Clayton. N.S .. Hampton. R.R. & Sht:'ulewonh, S.J. (1995) Efft'ctS uf photoperiod on food·storing and the hippocampus in birds.
Neuroreport.6.1701-4. Krebs. J.R..
herry. D.F.. Healy. S.D.. Perry. V.H. &
421
Vaccarino, A.L. {1989} Hippocampal specialization of food storing birds. Proc Nat Aead Sa' USA, 86, 138892. Kreitman, M. & Akashi, H. (1995) Molecular evidence for natural selection. Ann Rev Eeol Syst. 26, 403-22. Krememz. D.G. & Ankney. C.D. (1995) Changes in lotal·body calcium and diet of breeding house sparrows. J Avian BioI, 26. 162-7. Kruuk, H. (1975) Functional aspects of sodal hunting by carnivores. In: Function and Evolution in Behaviour:
Essays in honour of Professor Niko Tinbergen FR.S. (eds G. Baerends, C. Beer & A. Manning), pp. 11941. Oxford University Press. Oxford. Kusano, K.. Naito. T.. Handa, N. & Kobayashi. l. ( 1995) Restriction-modification systems as genomic parasites in competition for spedfic sequences. Proc Nat Mad Sci USA. 92. 11 095-9. KuusaarL M., Nieminen, M. & Hanski. I. (19961 An experimental study of migration in the Glanville fritillary butterfly, Melilaea cinxia. J Ani', feol, 65. 791-801. Kyriacoll, c.P. & Hall. J.c. ( 1986) Interspecific genetiC control of courtship song production and reception in Drosophila. Science. 232, 494-7. Lack, D. (1950) The breeding seasons of European birds. Ibis. 92. 288-316. Lack, D. (1966) Population Studies of Birds. Clarendon Press. Oxford. Lack. D. (1968) Ecological Adaptations for Breeding in Birds. Methuen, London. Lacy. R.C. (1980) The evolution of eusociality in termites: a haplodiploid analogy? Am Nalllr, 116, 449-51. Lacy. R.C. & Sherman. P.w. (1983) Kin recognition by phenotype matching. Am NQlur, 121. 489-512. Lagercrantz. U. & Ryman. N. (1990) Genetic slruclure of Norway Spruce (Pieea abies~: concordance of morphological and allozymic variation. Evolution, 44.38-53. LambrechlS. M.M. & Dhondt. A.A. (1995) Individual voice discrimination in birds. CurrOrnithol, 12, 11539. Land. M.F. (1969) Structure of the retinae of the principal eyes of jumping spiders (Salticidae: Dendryphantinae) in relation to visual optics. J Exp Bioi. 51. 443-70. Land. M.F. (1981) Optics and vision in invenebrales. In: Handbook of Sensory Physiology. Vol. VII/6B (ed. H. Autrum). Pi>. 471-592. Springer-Verlag. Berlin. New York. land. M.F. (1989) Variations in the STructure and design of compound eyes. In: Facets of Vision (eds D.G. Stavcnt:a & R.e. Hardie). pp. 90-111. Springer· Verlag. Berlin.
422
REFERENCES
Land, M.E & Fernald, R.D. (19921 The CvohHion of
t'yes. Ann Rev Nturosd. 15. 1-29. Landau. H.G. (1951) On dominance relations and the ~trueture of animal societies. J. Effect of inhercnl
characteristics. Bull Math Biophysics. 13, 1-19. Lande. R. (1981) Models of speciation by sexual !ieleccion of polygenic traits. Pro,; Nat Acad Sci USA.
78,3721-5. Landwer. A.J. (1994) Manipulation of egg production
reveals cost of reproduction in the tree lizard (Vrosaurus omatus). OtcoJogia, 100,243-9. Langen, TA & Rabenold, K.E. (1994) Dominance and diet selection in juncos. Behav EI.'{)/. 5, 334-8. Lank. D.B.. Smith. C.M.. Hanolu~. 0 .. Burke. T. & Cooke. F. (1995) Genetic p,»)ymorphism for
ahernative mating behaviour in lekking male ruff Philomaclrus pugnax. Nature Lond. 378, 5962. Lau, M.M.H., Stewart. C.E.H., Liu, Z.• Bhan, H., Rotwein. P. & Stewart, c.L. (J994) Loss of the imprinted TGF2/cation-independent mannose 6phosphate receptor results in fetal overgrowth and perinatal lethality. Genes Develop, 8, 2953-63. Lauder, G.V. (1981) Form and function: structural analysis in evolutionary morphology. Paleobiology, 7,430--42. Laughlin. S.B. (1995) Towards the cost of seeing. In: Nervous Systems and Behaviour (eds M. Burrows, T. Mallhews, P.L Newland & H.J. Schuppe), p, 290. Thieme, Stuttgart, New York. Laughlin. S.B. (in press) Observing design with compound eyes. In: Optimal Animal Design: Fact or Fancy? (eds E.R. Weibel & L. Bolis). University Press. Cambridge. Law. R. & Hutson. V. (1992) Intracellular symbioms and the evolution of uniparental cytoplasmic inheritance. Proc Roy Soc Lond B. 248, 69-77. Lawrence, E.S. & Allen, J.A. (1983) On the term 'search image'. Oikos.40. 313-14. Lazarus. J. & Metcalfe, N. (1990) TIt for tat cooperation in sticklebacks: a critique of Milinski. Anim Behav, 39.987-8, Lazcano, A' Guerrero. R. Margulis, L & Or6, J. (1988) The evolutionary transition from RNA to DNA in early cells. J Molec Evol, 27, 283-90. Lee. R.B. & DeVore. I. (eds) (1968) Man thr Hunter. Aldine Publishing Company, Chicago. Lee, S.J., Witter, M.S' Cuthill, LC. & Goldsmith, A,R. (1996) Reduction in escape performance as a cost of reproduction in gravid starlings (Sturnus vulgaris). Proc Roy Soc Lond B, 263. 619-23. Lefebvre. L. (1986) Cultural diffusion of a novel food-finding behaviour in urban pigeons: an experimental field test. Ethology. 71, 295-304.
lefebvre, L. (1995a) The opening of milk bottlcs by birds: evidence for accelerating learning rates. but against the wave-of·advance model of cultural transmission. Bchav Process, 34, 43-54. Lefebvre, L. (1995b) Culturally-transmillcd feeding behaviour in primates: evidence for accelerating learning rates. Primates, 36(2). 227-39. Lefebvre, L. & Giraldeau, L.-A. (1994~ Cuhural transmission in pigeons is affected by the number of wlors and bystanders present during demonstrations. Anim Behav. 47,331-7. Lefebvre, L.. Whittle, P.. Lascaris, E. & Finkelstein, A. (1996a) Feeding innovations and forebrain size in birds. Anim Behav (in press). Lefebvre, L.. Palameta, B. & Hatch, K.K. (1996b) Is group·living associated with social learning? A comparative test of a gregarious and a territorial columbid. Behaviour, 133,241-61. Lehnherr, H .. Maguin. E.. Jafri. $. & Yarmolinsky, M.B. (1993) Plasmid addiction genes ofbaeteriophagc PI: doc, which (aust's cell death on curing of prophage. and phd, which pr~vents host death when prophaglt is retained. J Molec Bioi. 233, 414-28. Leigh, E.G. (1971) Adaplation and Diversity. Freeman, Cooper & Company, San Francisco. Leigh. E.G. (1990) Community diversity and environ· meOlal stability: a re-examination. Trends Bcol Evel, 5. 340--4. Lemon. W.e. (1993a) The energetics of lifetime reproductive success in the zebra finch Taen;opygia gUlIata. Physiol Zool, 66, 946-63. Lemon, W.e. (1993b) Heritability of selectively advantageous foraging behaviour in a small passerine. Evol E',I, 7, 421-8. lenington. S. (1991) The t-complex: a Story of genes, behavior, and populations. Adv Stud Behav. 20, 5186. Lenington, S., Egid, K. & Williams, J. (1988) Analysis of a genetic recognition system in wild house mice. 8ehav Genet, 18, 549-64, Lenington, S.• Coopersmith. C.B. & Williams. J. (1992) Genetic basis of mating preferences in wild house mice. Am lool, 32, 40-7. Lenington, S., Coopersmith. C.B. & Erhardt, M. (1994) Female preference and variability among t·haplotypes in wild house mice. Am Natur. 143. 766-84. Lenski, RE., Simpson, S.c. & Nguyen, T.T. (1994) Genetic analysis of a plasmid-encoded, host genotype-specific enhancement of bacterial fitness. J Bacterial. 176, 3140-7. Lessells, C.M. (1991) The evolulion of life histories. In: Behavioural Ecology. An Evolut;onary Approach (eds J. Krebs & N. Davies~, Pl'. 32-68. Blackwell Scientific
REFERENCES Publications, Oxford. Lessells. C.M. & Birkhead. T.R. (1990) Mechanisms of sperm competition in birds: mathematical models. BtJtav Ecol Sociobiol, 27, 325-37. Levitan. D.R. & P~lersen, C. (1995) Sperm limitation in the sea. Trtnds Ecol Evol, 10. 22&-31. Lewis. 0 .. Verma. S.c. & Zuberi. M.l. (1988) GamelOpbylic-sporophytic incompatibili,y in Ihe Cruciferae Raphanus Stllivus. Htredity, 61. 355-66. Lewontin. R.C. (1991) Biology as Ideology. House of Anansi Press. Concord. Omario. Lewonlin. R.C .. Rose. S. & Kanin. L.J. (1984) NOlin our Genes. Pantheon. New York. Liem. K.F. (1980) Adaptive significance of intra- and inlerspecific differences in Ihe f~eding repenoires of cichlid fishes. Am Zool. 20. 295-314. Liem. K.F. (1973) Evolutionary strategies and morphological innovations: cichlid pharyngeal jaws. Syslem Zool. 22. 425-4 I . Lima. S.l. (1986) Predation risk and unpredictable feeding conditions: determinants of body mass in wintering birds. Ecology. 67. 377-85. Lima. S.L. (1989) Iterated prisoner's dilemma: an approach to evolutionarily stable cooperation. Am Natur. 134.828-44. Lima. S.L. & Zollner. P.A. (1996) Towards a behavioral ecology of ecological landscapes. TrendJ Ecol Bvol. II. HI-5. Lin. N. & Michener. C.D. (1972) Evolution of sociality in insects. Quart Rev Bioi, 47. 131-59. Linsenmair. K.E. (1987) Kin recognition in subsodal arthropods, in particular in the desert isopod Hrmiltpistus reaumuri. In: Kin Rtcognition in Animals (eds DJ.C. Fletcher & C.D. Michener). pp. 121-208. John Wiley & Sons, New York. LindSledl. S.L. & Calder, W.A. (1981) Body size, physiological time. and longevity of homeothermic animals. Quan Rev Bioi, 56. 1-16. LindSlrom, A. & Alerslam. T. (1992) Optimal fat load in migrating birds: a test of the time minimisation hypothesis. Am Natur, 140.477-91. lindstrom. A. & Piersma. T. (1993) Mass changes in migrating birds: the evidence for fat and prOlein storage re-examined. Ibis. 135. 70-8. Locket. N.A. (J977) Adaptations 10 the deep-sea environmenl. In: Handbook ofSensory Physiology. Vol. VI1/5 (ed. F. Cresdtelli). pp. 67-192. SpringerVerlag, Berlin, New York. Loehle. C. (1995) Social barriers 10 pathogen transmission in wild animal populations. Ecology. 76. 32635. Loftus-Hills. J.J. & Liltlejohn. MJ. (1992) Reinforcement and reproductive character displacement in Gastrophrynt carolinens;s and G. olivQaQ (Anura:
423
Microhylidae): a reexamination. Evolution. 46. 896906. Loher. W. & Dambach. M. (1989) Reproductive b~havior. In: Cricket Behavior and Neurobiology (eds F. Huber. T.E. Moore & W. Loher), pp. 43-82. Cornell University Press. Ithaca, New York. Long. K.D. & Houde, A.E. (1989) Orange SPOlS as visual cues ror female mate choice in the guppy (POfcilia reticulara). Ethology. 82. 316-24. Lord. E.M. & Eckard. K.J. (1984) Incompatibility between the dimorphic flowers of Coll'mia grandlf1ora. a c1eislogamous species. Seima, 223. 695-6. Lord. R.D. (1960) Litter size and latitude in North American mammals. Am Midland Natur, 63. 48899. Lorenz. M.G. & Wackerknagel. W. (1994) Bacterial gene transfcr by natural genetic transformation in Ihe environment. Microbiol Rev, 58. 563-602. losos, J. (1992) The evolution of convergent structure in Caribbean Altolis communities. System Bioi. 401. 403-20. Losos. J.B. (1996) Community evolution in Greater Antillean Anolis lizards: phylogenetic patterns and experimentallesls. In: New Uses for New Phylogenies (eds P.H. Harvey. A.J. Leigh Brown. J. Maynard Smith & S. Nee). pp. 308-2]. Oxford University Press. Oxford. Lotem. A. (1993) learning to recognize nestlings is maladaptive (or Cuculus canorus hosts. Nature, 362, 743-5. lOt em. A.• Nakamura. H. & Zahavi. A. (1995) Constraints on egg discrimination in cuckoo-host evolution. Anim &hav. 49, 1185-1209. lovejoy. C.O. (1981) The origin of man. Scimct. 211. 341-50. Lucas. l.R. & Howard, R.D. (1995) On alternative reproductive tactics in anurans: dynamic games with density and frequency dependence. Am Natur, 146. 365-97. Lucas, J.R. & Walter. L.R. (1991) When should chickadees hoard rood? Theory and experimental results. Anim Behav. 41. 579-601. Lucas, J.R .. Peterson. L.J. & Boudinier. R.L (1993) The effects of time constraints and changes in body mass and satiation on the simultaneous expression of caching and diet-choice dedsions. Anim Bthav. 45.639-58. Luykx. P. & Syren. R.M. (1979) The cytogenetics of Indsitermes schwam and other Florida termites. Sodobiology.4.191-209. Luykx. P.. Michel. J. & Luykx, J. (1986) The spa'ial distribution of the sexes in colonies of the termite IncisiltrmtsschwaniBanks (Isoptera: Kalotermitidae).
424
REFERENCES
Insea Soc. 33. 406-21. lythgoe. J.N. (1988) Light and vision in the aquatic
environment. In: Stnsory Biolog)' of Aquati, Animals (cds J. Alcma, R.R. Fay, A.N. Popper & N. Tavoga). pp. 57-82. Springer- Verlag. New York. lythgoe. J.N .• Muntz. W.R.A .• Pan ridge. J.e.. Shand. J. & Williams. D.McB. (1994) The ecology of the visual pigments of snappers (Lutjanidae) on the Great Barrier Reef. J Comp PhysioJ A. 174.461-7. MacAnhur. R.H. & Wilson. E.O. (1967) Tht Thtory of Island Biogtography. Princeton University Press.
Princeton. NJ.
MacAnhur. R.H. (1972) Gtographjca/ Ecology. Harper & Row. New York.
MacArthur. R.H. & Pianka. E.R. (1966) On the optimal
use of a patchy environment. Am Natur. 100.6039.
McBeath. M.K .• Shaffer. D.M. & Kaiser. M.K. (1995)
How baseball outfielders determine
whcr~ 10
run
to catch fly balls. Sdtf7ct. 268. 569-73. McCauley. D.E. (1991) Genetic consequences of local population extinction and recolonization. Trends Ecol Eval. 6. 5-8. McComb. K.E .. Packer. C. & Pusey. A.E. (1994) Roaring and numerical assessment in contests l'letw.een groups of female lions. Panthera leo. Anim Ilthav. 47.379-87. McDonald. D.B. & POtts. W.K. (1994) Couperative display and relatedness among males in a lek· mating bird. Science. 266, 1030-2. Macdonald. D.W. (1979) Helpers in fox society. Nature, 282, 69-71. Macdonald. D.W. & Moehlman, P.O. (1982) Cooptration. altruism, and restraint in (he reproduction of carnivores. Perspec Etho!. 5.433-69. McGregor, P.K. (1993) Signalling in territorial systems: a context for individual identification. ranging and eavesdropping. Phil Trans Roy Soc l..Dnd S, l40, 23744. McGregor. P.K. & Krebs. J.R. (1982) Song types in a population of great tits (Partls major): their distribution. abundance, and acquisition by individuals. Bthaviour. 79. 126-52. Mackeney. P.A. & Hughes. R.N. (1995) Foraging behaviour and memory window in sticklebacks. Bthaviour, 132, 1241-53. McKinney. E, Cheng, K.M. & Bruggers. D.J. (1984) Sperm competition in apparelllly monogamous birds. In: SfNrm Compttition and the Evolution of Animal Matins Systems (ed R.L. Smith), pp. 52~5. Academic Press, Orlando. McKinney. F.. Derrickson, S.R. & Mineau, P. (1983) Forced copulation in waterfowl. Behaviour, 86, 25094.
Mclaughlin. R.L & Montgomerie. R.D. (1990) Flight speeds of parent birds feeding nestlings: maximization of foraging effictency or food delivery rate? Om J ZIJo/. 68. 2269-74. Mclennan, D.A. & McPhail. J.D. (1990) Experimental investigations of the evolutionary significance of sexualJy dimorphic nuptial colouration in Casterosttus amltQIUS (l.): the relationship between mate colour and female behaviour. Can J Zool, 68. 482-92. Mcleod, P. & Di~nes, Z. (1993) Running to catch the ball. Nalllrt. 362. 23. McMahon, T. (l973) Size and shape in biology. Scimrt, 179.1201-4. McMinn. H. (1990) Effects of the nematode parasite Cama/lanllscoltiin sexual and non·scxual behaviours in the guppy (Pouilia rtrieulata). Am ZOOI. 30, 2459. McNamara, J.M. (1990) The starvation-predation trade-off and some behavioural and ecological consequences. In: Bthaviollral Mechanisms of Food Stltctioll. NATO ASI SeritS A. LIft Somees (ed. R.N. Hughes). pp. 39-59. Springer-Verlag. New York. McNamara, J.M. (1994) TIming of entry into diapause: oplimal allocation to 'growth' and 'reproduction' in a stochastic environment. J Theor Bioi, 168,2019. McNamara, J.M. & Houston, A.1. (1980) The application of statistical dedsion theory to animal behaviour. J Theor Bioi. 85, 673-90. McNamara. J.M. & Houston. A,l. (1986) Thc common currency for behavioral decisions. Am Nafttr, 127, 358-78. McNamara, J.M. & Houston. A.1. (1987) Starvation and predation as factors limiting population sizt'o Ecology. 68. 151 5-19. McNamara, J.M. & HOliSton. A.t. (1990) The value of fat reserves and the trade·off between starvation and predation. Acta Biot/reor, lS. 37-61. McNamara, J.M. & Houston. A.I. (1992) Statedependent life-hislory theory and ils implications for optimal clutch size. Eval Ecol. 6. 17Q-85. McNamara, J.M. & Houston, A.I. (1994) The effect of a change in foraging options on intake rate and predation ralc. Am Natllr. 144,978-1000. McNamara, J.M .. Houston. A.I. & Krebs, J.R. (l990) Why hoard? The economics of food storing in tit.... Parus spp. Bthav E(OI, L 12-23. McNamara, J.M., Houston, A.l. & Lima, S.L. (1994) Foraging routines of small birds in winter: a theoretical investigation. J Avian Bioi. 25, 287302. McNamara, J.M., Mace. R.H. & Houston, A.1. (1987) Optimal daily routines of singing and foraging. Bthav Ecol Sociobiol, 20, 399-405.
REFERENCES McQuoid, L.M. & Galef, B.G. Jr (1992) Social innuencesof feeding site selenion by Burmese fowl Gallus gallus. J Omlpar Psychol. 106, 137-41. Maddison, w.P. (1990) A method for t<Sting the correlated evolution of two binary characters: are gains or losses concentrated on certain branches of a phylogenetic Iree? Evailltion, 44, 539-57. Maddison. W.P. & Maddison. O.R. (1992) MocClod<:
Analysis of Phylogeny and Character EvolutiolJ. Version J. Sinauer. Sunderland. Madsen. T. ( J 987) Cost of reproduction and female Iife-
history ladies in a population of grass snakes. Natrix narrix. in southern Sweden. Oikas. 49. 129-32.
Madsen. T.. Shine. R.. Loman, J. & Hakansson. T. (1992) Why do female adde' copulate so frequently? Nature Lend. 355,440-1.
Malcolm. J.R. & Manen. K. (1982) Natural selection and the communal rearing of pups in African wild dogs (Lyeooll pietlls). /kirov /!col Sodobiol, 10. 1-13. Malll)us. T.R. (1798) An Essay on tire Principle of Population. Reprinted by Macmillan, New York. Mangel. M. & Clark. C. W. (1986) Towards a uniried foraging theory. £cology, 67. 1127-38. Mangel. M. & Clark, C.w. (1988) Dynamic Modellillg in Behavioural Ecology. Princeton Universily Press. Princeton. Manning, c..I., Wakeland, E.K. & POllS, W.K. (1992) Communal nesting patterns in mice implicate MHC genes in kin recognition. Nature. 360, 581-3. Manning, CJ .• Dewsbury, D.A.. Wakeland. E.K. & POtlS, W.K. (1995) Communal nesting and communal nursing in house mice. Mus musOilus
domesticus. Anim /khav, 50, 741-51. Marchcni. K. (1993) Dark habitats and bright birds illustrate the role of the environment in species divergence. Nature Lond. 362. 149-52. Marchetti. K. & Price, T. (1989) Differences in the foraging of juvenile and adult birds: the importance of developmental constraints. BioI Rev, 64,51-70. Mark!. H. (1983) Vibrational communication. In: NeuroetJrologya'ld Behavioral Physiology (eds F. Huber & H. Markl). pp. 332-53. Springer-Verlag, Berlin, Heidelberg. Marler, P. (1991) Song-learning behavior: Ihe interface wilh neuroelhology. Trends Neurosd, 14. 199-206. Marr, D. (1982) Vision. W.H. Freeman. San FrancisCt). Marsh. R.L. (1984) Adaptalions of tht.' grey calbird Dumetella carolinmsis to long-distance migralion: nighl muscle hypertrophy associated with elevated body mass. PlrysiolZooI, 57,105-117. Martin, P.A., Reimers. T.J .. Lodge. J.R. & Dziuk. P.J. (1974) The eHect of ratios and numbers or spermatozoa mixed from two males on proportions of uffspring. J Reprod Fert. 39. 251-8.
425
Martins. E.P. & Garland, T.H. (1991) Phylogenies and the evolution of continuous characters. Evolution, 45, 534-57.
Manlun, J.M. & Balda, R.P. (1990) Pinyon jays: making the best or a bad situation by helping. In: Cooptrar;ve Breeding in Birds: Long-tenn Srudies of Ecology alld /kirov/or (eds P.B. Stacey & W.O. Koenig). pp. 197-237. Cambridge University Press. Cambridge. Masman, D.. Daan. S. & Dijkstra, C. (1988) Time allocation in the kestrel (Falco finmmculus). and the principle of energy minimization. J A'im Eeol. 57. 411-32.
Mason, R.J. (1988) Direct and observationc:llearning by redwinged blackbirds (Agelaius photniceus): the imporlan~ of complex visual stimuli. In: 50dal fLaming: Psychological an.d Biological Perspecriws (eds T.R. Zelllall & B.G. Galer Jr), pp. 99-115. LawrenceErlbaum Associates. New Jersey. Masters. W. & Waite. T. (1990) Tit-for-Ial during predator inspection, or shoaling? Anim &hav. 39. 603-4.
Mateo. J. (1996) The development of alann-call response behaviour in rree-livingjuveniJe Belding's ground squirrels. Anim Behav. 52, 489-505. Mayfield. H. (1965) The brown-headed cowbird, with old and new hosts. Living Bird. 4, 13-28. Maynard Smith. J. (1972) On Evolution. Edinburgh University Press. Edinburgh. Maynard Smith, J. (1974) The theory of games and Ihe evolution of animal conflicts. J Thear Bioi. 47, 209-21. Maynard Smith. J. (1977) Parental investment - a prospective analysis. Anim lkJrall, 25. 1-9. Maynard Smith, J. (1978) Tire Eva/llrioll of Sex. Cambridge University Press, Cambridge. Maynard Smith, J. (1978) Optimization theory in evolution. Ann Rev £rol 5yst, 9. 3 I-56. Maynard Smith. J. (1979) Game theory and the evolution of behaviour. PrOf Roy Soc Loud B. 205, 475-88.
Maynard Smith. J. (1982) Evolution and lJr~ TJr~ory of Games. Cambridge University Press, Cambridge. Maynard Smith. J. (1991) Honest signalling: the Philip Sidney game, ,4nim Behav, 42, 1034-5. Maynard Smith,.I. (1991) Theories of sexual seleCtion.
Trends Ecol Evol, 6. 146-51, Maynard Smith. J. & Parker, G.A. (1976) The logic of asymmetric conleSIS. Anim !khav. 24, 159-75. Maynard Smith, J. & Price. G.R. (973) The logic of animal conflict. Naturt Lolld. 246. 15-18. Maynard ·Smith. J. & Riechert. S.E. (1984) A conflicting-tendency model of spider agonistic behavior: hybrid-pure population line comparisons. An;nI Bthav, 32. 564-78.
426
REFERENCE S
Maynard Smith, J. & Szathmary. E. (J 993) The origin of chromosomes. I. Sel~clion for linkage. J Theor Bioi. 164.437-46. Maynard Smith. J. & Slalhmary. E. (1995) Tire Major Transitions in Ew/ution. W.lf. Freeman, Oxford. MayT. E. (1970) Populations. Specia, and Evolution. Belknap Press of Harvard University Press. Cambridge. MA. Mt'ddis, R. (1983) The evolution of sleep. In: Sleep Mechanisms and Functions (ed. A. Mayes). pp. 57106. Van Noslrand Reinhold, London.
Mdfen. L.M. (1995) Bottleneck effects on genetic variance for counship repenoire. Gmenrs, 139, 36574. Meslc-Twn-Gibbons. M. & Dugalkin. L.A. (1995)
Toward a theory of dominance hierarchies: eHcets of assessment. group size. and variation in fighling ability. Bthav Ecol. 6. 416-23. Melcalf, R.A. & Whitt. G.S. ( I 977a) Relative inclusive fitness in the sodal wasp Polistes Jnelricus. Behav Ecol Sociobiol. 2. 353-60. Metcalf. R.A. & Whit!. G.S. (1977b) Intra-nest relatedness in the social wasp Polisles metricllS. A genetic analysis. &hav Eeol Sodobiol. 2. 339-51. Metcalfe. N.B. & Ure, S.E. (1995) Diurnal variation in flight performance and hena' predation risk in small birds. Proc Roy Soc Lond B. 261. 395-400. Meyer. A.• Morrissey. J.M. & Schartl. M. (1994) Recurrent origin of a sexually sdeeted rrait in Xiphophorus fishes inferred from a molecular phylogeny. NaTUre. 368. 539-42. Michelsen. A. (1983) Hearing and sound communica· lion in small animals: eVOlutionary adaptations 10 the laws of physics. In: Tht Evolutionary Biology of Hearin9 (eds D.B. Websier. R.R. Fay & A.N. Popper). pp. 61-77. Springer· Verlag. New York, Berlin. Michelsen. A.• Towne. W.F.. Kirchner, W.H .. Kryger. P (1987) The acoustic near field of a dandng honeybee. J Camp Plrysiol A. 161.633-43. Michener. C.O. & Smith, B.H. (}O987) Kin recognition in primitively eusorial insects. In: Kin Recognition in Animals (eds D.J.C. Fletcher & C.D. Michener), pp. 209-42. John Wiley & Sons. New York. Michod, R.E. (1982) The theory of kin selection. Ann Rl'VccoISyst. 13.23-55. Milinski. M. (1984) Competitive resource sharing: all experimental tesr of a learning rule for ESSs. Afl;m B_lrov. 32. 233-42. Milinski. M. (1984) A predator's cost of overcoming the confusion effect of swarming prey. Ani', &hav. 32.233-42. Milinski. M. (1987) Tit for tat and the evolulion of cooperation in sticklebacks. Nature Lond, 325. 4337.
Milinski. M. (1992) Predator inspection: cooperation or 'safely in numbers'? Anim Behav, 43. 679-80. Milinski. M. & Bakker. T.C.M. (1990) Female sticklebacks use male coloration in mate choice and hence avoid parasitized males. Naturt. 344~ 330-3. Milinski. M. & Bakker. T.c.M. (1993) COStS influence sequential mate choice in sticklebacks. Proc Roy Soc Land B. 250. 229-33. Milinski. M. & Boltshauser. P. (1995) Boldness and predator deterrence: a critique of Godin and Davis. Proc Roy SiK umd B. 262. 103-5. Milinski. M.. Pfluger. D.. Kulling. D. & Kettler. R. (1990) Do sticklebacks cooperate repeatedly in pairs? Bellav Ecol Sociobiol. 27, ) 7-23. Mills. J.A. (1994) Extra·pair copulations in the redbilled gull: females wilh high Quality. attentive mail'S resist. Behaviour. 128.41-64. Mineka. S. & Cook. M. (1988) Soda I learning and the acquisition of snake fear in monkeys. In: Social Learning: Psychological and Biological Persptctivrs (eds T.R. Zentall & B.G. Gale!. Jr). pp. 51-75. Lawrence Erlbaum Assodate5. New Jersey. Mittelstaedt. H. (1985) Analytical cybernetics of spider navigation. In: Nflirobiologyo! Arachnids (l'd. EG. Barth). pp. 298-316. Springer.Verlag. Berlin. Miner, C.. Farrell, B. & Wiegmann. B. (1988) The phylogenelic study of adaptive zones: has phyw· phagy promoted diversification? Am Nat/If. 132. 107-28. Moller. A.P. (1988) Female choice selc:.-cts for male sexual tail ornaments in Ihe monogamous swallow. Nature Lolld. 3]2. 640-2. Moller. A.P. (1989) Viability costs uf male tail ornamelllS in a swallow. Naturt Lond. 339. 132-5. Meller. A.P. (1990a) Effects of a haemalOphagolis milt' on the bam swallow (Himndo nlsIica): a test of till' Hamilton and Zuk hypothesis. Evolutioll, 44.771-84. Muller. A.P. (1990b) Parasites and sexual selection: currelll statuS or the Hamilton and Zuk hypothesis. J Evol Bioi. 3. 319-28. Meller. A.P. (1991) Clutch size. nest predation and distribution of avian unequal comperitors in a palchy environment. Ecology. 72. 228-34. Moller. A.P. (1992) Parasites differentially increase Ihe d~gree of nUCluating asymmetry in st'condary sexual characters. J Eval Bioi. 5. 691-9. M0l1er. A.P. (1993) Developmental siability. s~xual selt'etiun. and Ihe evolution of secondary sexual characters. E(ologin. 3. 199-208. MI~lIer. A.P. (1994) Male ornament sizt' as a rdiahh: cue to enhanced offspring viabilily in Ihe ham swallow. PrOf Nat Acad Sri USA. 91. 6926-32. Meller. A.P. (1994) S{xual Selection and ,hr Bam Swallilw. Oxford Univl.'rsity Press. Oxford.
REF E REN C E S Moller. A.P. & Birkhead. T.R. (1989) Copula.ion behaviour of mammals: evidence that sperm competition is widespread. B;ol J Linn Soc. 38. 11931.
Muller. A.P. & Birkhead. T.R. (1992) A pairwi,e comparative method as illustrated by copulation frequency in birds. Am Narur, 139,644-56. Moller. A.P. & Birkhead. T.R. (1994) The evolution or plumage brightness in birds is related to extra·pair paternity. Evolution. 48. 1089-100. Moller. A.P. & de Lope. F. (1995) Differential COSl' of a secondary sexual character: an experimental lest of the handicap principle. Evolution. 48. 167683. Ml1ller. A.P. & HOglund. J. (1991) Pattcrnso( fluctuating asymmetry in avian fcalher ornaments: implications fur modt:ls of sexual selection. Proc Roy Soc Lond 8, 245. 1-5. Moller. A.P. & Pomiankowski. A. (1993a) Fluauating
asymmetry and sexual selection. Gtntt;ca. 89. 26779. Moller. A.P. & Pomiankow,ki. A. (1993b) Why have birds gOI multiple sexual ornaments? Behav Erel
SQciobiol.32. 167-76. Moller. A.P.. Durva. R. & Allander. K. (1993) Para,ite, and the evolution or hosl social behavior. AdvStudy Beltav. 22.65-102. Meller, A.P.• Soler, M. & Thornhill. R. (1995) Breast asymmetry. sexual selection, and human reproductive success. Ethol Sodobiol. 16,207-19. Moody. A.L & Houston, A.I. (1995) Interference and the ideal free distribution. Anim Behav, 49, 106572. Moore. T. & Haig. D. (1991) Genomic imprinting in mammalian development: a parental tug-of-war. Trends Genet. 7. 45-9. Monon, N.E. (1991) Parameters oCthe human genome. Proc Nat Acad Sci USA. 88, 7474--6. Mrosovsky. N. & Sherry. D.E (1980) Animal anorexias. Science. 207. 837-42. Mueller. U.G. (1991) Haplodiploidy and the evolulion of facultative sex ratios in a primitively eusodal bee. Sdena. 254. 442-4. Mueller. U.G .. Eickwort. G.e. & Aquadro. e.F. (1994) DNA fingerprinting analysis of parent-offspring confliCl in a bee. Proc Nat Acad Sci USA. 91. 51437. Mulder, R.A .. Dunn. P,O.. Cockburn, A.. LazenbyCohen. K.A. & Howell. M.J. (1994) Helpers liberate female fairy-wrens from constraints on extra-pair mate choice. Proc Roy Soc Lond B. 255. 223-9. Mumme. R.L. (1992a) Delayed dispersal and cooperative breeding in the Seychelles warbler.
Tm,ds £<01 Eval.
7. no-I.
427
Mumme. R.t. (1992b) Do helpers increase reproductive success? An ('(penmen tal analysis in the Florida scrub jay. Bthav Ecol SOdobiol. 31, 319-8. Mumme. R.l.. Koenig, W.O. & Pilelka. EA. (1983) Reproductive compel ilion in the communal acorn woodpecker: sisters destroy each mher's eggs. Naturr. 306. 583-4. Munz. F.W. & McFarland. W.N. (1977) Evolutionary adaptations of fishes 10 the photiC environment. In: Handbook of Smsory Physiology. Vol. VlI15 (ed. F. Crcscitelli). pp. 193-274. Springer-Verlag. Berlin. New York. Murphy. M.E. (1994) Dietary complementation by wild birds - considerations for field studies. J Biosd. 19.355-68. Munon. R.K. (1971) The significance of a specific search image in the feeding behaviour of the woodpigeon. Behaviour. 40, 1042. Myles. T.G. & NUlting. W.L. (1988) Termite eusudal evolution: a re-examination of Bartz~s hypothesis and assumptions. Quart Rnt BioI. 63. 1-23. Nachtigall. P.E. & Moore. P.W.B. (1988) Animal Senar. Plenum Press. New York. Nagy. K. (1980) CO 2 production in animals: analysis of potential errors in the doubly labeled water method. Am J Pltysiol. 238. R466-R73. Nalepa. C.A. (1984) Colony composition. protozoan transfer and some life history characteristics of Ihe wood roach CryptocerOls punclulatus Scudder (Oictyoptera: Cryptocerddae). &hav Ecol Sodobiol. 14.273-9. 'alepa. CA. (1994) Nourishmcm and the origin of termite eusociality. In: Nourishment and Evolul;(ln in Insect Societ;es (cds J.H. Hunt & C.A. Nalepa). pp. 57-104. Westview Press. Boulder. Colorado. Nalepa. e.A. & Jone,. S.e. (1991) Evolution of monogamy in termites. BioI Rev. 66. 83-97. Nee. S.. Read. A.F.. Greenwood. J.J.D. & Harvey. P.H. (1991) The relalionship between abundance and body sizt' ill British birds. Nature. 3S1. 312-13. Neems. R.M. & Butlin, R.K. (1994) Variation in cuticular hydrocarbons a ross a hybrid zone in the grasshopper Chorthippus paralltlus. PrO{ Roy Soc Lend B.257. 135-40. Nelson. W.A. (1984) Effects of nUlrition of animal on their ectoparasites. J Med Enlamol. 21. 621-35. Netto. W.J. & van Hoofr. J.A.R.A.M. (1986) Connid interference and Ihe development of dominance relationships ill immature Macaca fascicularis. In: Pn'mate Omogeny. Cognition and Sodal Behaviour leds J.G. EI,e & P.e. Lee). pp. 291-300. Cambridge Universily Press. Cambridge. Neukirch. A. (1982) Dependence of the lire span of the honeybee (Apis mellifera) upon flight perlonmance
428
REFER EN CE S
dnd energy consumption. J Camp Physio/. 146. 3540. Nt'uweiler. G., Singh, S. & SnpalhL K. (1984)
Audiograms of a Soulh Indian bat community, J Comp Physiol A, 154. 133-42. NewlOn, I. (ed.) (1989) LIfetime Reproduction in Birds.
Academic Press. London. Nichols. R.A. & Beaumont, M.A. (1996) Is it ancient Of
modern history that we can read in our genes?
In: Aspects afthe Gmesis and Mainrmanu of Biological Diversity (eds M.E. Hochberg. J. Clobert & R. Harbault). pp. 69-87. Oxford University Press.
Oxford. NkoL C.l. & Pope. S.J. (1994) Social learning in small flocks or laying hens. Anim Behav, 47. 1289-96. Niroletlo. P.E (1993) Female sexual response TO condition-dependent ornaments in the guppy. Paecilia re(feu/ala. Anim Bthav. 46. 441-50.
Nilsson, s.G. (1987) Limitation ilnd regulation of population density in the nuthatch Sirca eliropaea lAves) breeding in natural cavities. J Anim Bcol. 56, 921-37. Nishida. T. (1983) Alpha status and agonistic alliance in wild chimpanzees (Pan lroglodytes schei'furthii). Primates, 24, 318-36. Noble. G.K. & Bradley. H.T. (1933) The mating behavior or lizards; ilS bearing on Ihe theory of s~xual selection. Ann N YAcad Sci, 35.35-100. Nne. R. (1990) A veto game played by baboons: a challenge IO.the use of lhe prisollt>r's dilemma as a paradigm for reciprocity and coopNali()l1. Anim
Bellav, 39, 78-90. Noirot. C. (1989) Social slrUClUre in termife socielies. Ethol £col £vol. I. 1-17. Noirot. C. & Pasteels, J.M. (1987) Ontogenetic development and evulution of the worker caslC in Icrmites. Experientia. 43, 851-60. Noirot, C. & Paslee)s, J.M. (1988) The worker caste is polyphyletic in tnmites. Sociobiology, 14, 15-20. NOllacs, P. (1986) Ant reproductive strategies and sex allocation theory. Quart Rev 8iol, 61. 1-21. Nonacs. P. (1991) Alloparenlal care and eusocial evolution: the Iimils of Quelkr's head-start advantage. Dikos. 61, 122-5. Nonacs, P. (1993) Male parentage and sexual deception in Ihe social Hymenoptera. In: Evolution and Diversity of Sex Ratio in Insects and Mites (eds D.L. Wrensch & M.A. Ebbert), pp. 384-40 I. Chapman & Hall. New York. Nonacs, P. & Reeve, H.K. (1995) The ecology of cooperation in wasps: causes and consequences of ahernative reproductive decisiuns. Ecology, 76, 95367.
Noonan. K.M. ( 1981 ) Individual strategies of indusiv~-
fitness· maximizing in polisres fuscatus foundresses. In: Natural Selection and Social Behavior (eds R.D. Alexander & D.W. Tinkle), pp. 18-44. Chiron Press, New York. Norberg. R.A. (1981) Optimal flight speeds in parent birds when feeding young. J Anim £eol, 50, 473-7. Norberg. U.M. (J 9901 Vertebrate Flight. Springer-Verlag. Berlin. Nordeen, E.). & Nordeen, K.W. (1990) Nt>urogenesis and sensitive periods in avian song learning. Trends Neurosci, 13, 31-6. Nordstr(jm. K. & Austin, s.J. (1989) Mechanisms that contribule to the stable segregation of plasm ids. Ann
Rev Genet. 23, 37-69. Norris. K. (1993) Heritable variation in a plumage indicalOr of viability in male great tils Parus major. Nature, 362, 537-9. orris. K' Anwar. M. & Read, A.F.J. (1994) Reproductive effort influences the prevalence of haemalozoan parasites in grealtits. J Allim Ecol. 63, 601-10.
Nose. A.. Nagafuchi. A. &TakeichL M. (1988) Expressed recombinant cadherins mediate cell sorting in model systellls. CiJl, 54. 993-1001. Nottcbohm, F. (1991) Reassessing Ihe Illechanisms and origins of vocal learning in birds. Trends Neurosci, 14. 206-10.
Nowak. M. & Sigmund, K. (1993) A strategy of winstay, lose-shift Ihat olilperforrns til-for-tat in the prisoner's dilemma game, Natllre LOIld, 3M, 56-8. Nowak, M. & Sigmund, K. (1994) The alternating prisoner's dilemma. J Theor Bioi. 168, 219-26. Oakes, E.J. (1992) Lekking and Ihe evolUlitln of sexual dimorphism in birds: comparalive approaches. Am Natuf. 140.665-94.
O'Day, D.H. (1979) Aggregalion during sexual dev· e!opment in DictyosreliunI discoideum. Can J Microbial. 25. 1416--26.
O'Donald, P. (1962) The theory of sexual selection. Heredity. 17. 541-52. Ogilvie, M.A. & Matthews, G. V.T. (1969) Brent geese. mudflats and Man. Wildfowl, 20, 119-25. Oldroyd, B.P.. Rinderer, T.E .. Schwenke, J.R. & Buco, S.M. (1994) Subramily recognition and task specialisation in honey bees (Apis melllfera l.) (Hymenoplera: Apidae). 8elwv Bcol Sociobiol,. 34, 169-73.
Oldroyd, B.P., Smolenski, A.J., Coenuel. J.-M. & Crozier, R.H. (1994) Anarchy in the bet'hive. NaIUrl', 371. 749.
Olivieri. J.. Michalakis, Y. & Gouyon, P.·H. (1995) Metapopulation genetics and the evolution of dispersal. Am Namr, 146, 202-28. Olsson, M. (1995) Forced copula lion and costly femalt'
R E FE R EN C E S rc,i ...tance behavior ill the Lake Eyre Dragon OmopltMlIs macliloslIS. HerpetoloHica. 51. 19-24. 01...... 011. M .• Gullbcrg. A. & Tegelstrorn. H. (1995) SIll'rrn competilion in the sand lizard. LAcerta agilis. Allim Beltav. 48. 193-200. Oh'un. M.. Gulllxrg. A.. Tegelstrom. H.. Madsen. T. fi Shine. R. (1994) Can female adders multiply? Nature>. 369. 528. Orian>. G.H. & Witten~rger. J.F. (1991' Spatial and h.·mpmal scale in habilal seleaion. Am Nawr. 137. 529-49. OSlcr. G.F. & Wilsun. E.O. (1978) caste and Ecology in tltt Social lnstets. Princeton University Press. PrincelOn. New Jt:'rsey. Owen. D.O. & Owen. M.J. (1984) Helping behaviour in bmwn hyaenas. Nature. 308. 843-6. OWCIl. R.E. & Plowrighl. R.C (1982) Worker-qul'en conflict and male parentage in bumble bees. Bellav Ecol Sodobiol. 11. 9 1-9. OWl'ns. I.P.F., Dixon, A.. Burke. T. & Thompson. D.B.A. (1995) Sirategic paternity assurance in lhc ~t.·x-rote reversed Eurasian donerel (Charadrius marinelJus): behaviural and genetic evidence. Bdtav Ecol. 6. 14-2 I. Packrr, C. (1979) Inter-troop transfer and inbreeding avoidance in Papio anubis. Anim Behav. 27. 1-36. Packer. C. & Pusey. A.E. (1982) Cooperatioll and ('nmpetition within coalitions of male lions: kin sell'dion or game (henry? Nawre Lend. 296. 7402. Pa<..'kcr, C. & Pusey. A.E, (1983) Adaplations of (emale lions tu infantio{!e by incuming males. Am Nawr. 121.716-28.
Packer. C & Pusey, A.E. (1985) Asymmerric contests in social mammals: n.>sllCCI. manipulation and age· spedfic aspects. In: ElIOllllion: Essays in Honour a!101m Maynard Smilh (cds P.J. Greenwood & M. Slatkin). pp. 173--86. Cambridge University Press. Cambridge. Packer. C & Ruttan. L.M. (1988) The evolulion of cooperative hUllling. Am Natllr. 132. 159-98, Packer. C. Scht.'el, D. & Pusey. A.E. (1990) Why lions furm groups: food is nOI enough. Am Nawr, 136. 119. Packer. C. Collins. O.A.. Sindimwo. A. & Goodall. J. (1995) Reprodudive constraints on aggressive cumllt'tition in female baboons. Namre umd. 373. 60-3.
Packer. C. Gilbert. D.• Pusey. A.E. & O·Brien. S.J. (1991) A nwit.'cular genetic analysis of kinship and cooperation in African lions. Nawre umd. 351. 562$.
Porker. C. Herbst. L.. Pusey. A.E. tl al. (1988) Reproduaive success of liollS. :n: Reprodlletive Success (cd. T.H. Clullon-Brock). pp. 363-83. University uf
429
Chicago Press. Chicago. Page. R.E. (1986) Sperm utilization in sodal insl'cts. An' Rev £ntomC'/' 31. 297-320. Page. R.E .. Robinson. G.E. & Fondrk. M.K. (1989) Genelic specialists. kin recognition and nepotism in honey bee colonies. Nam'. 338. 576-9. Pagel. M.D. & Harvey. P.H. (1989) Taxonomic diffcrences in the scaling of brain on body weight among mammals. Sdma. 244. 1589-93. Palameta. B. & Lefebvre. L (1985) The social tr,1OSmis ion of a food-finding technique in pigeons: what is learned? Anim Btlrav. 33. 892-6. Pamilo. P. (1990) Sex allocalion and queen-worker conflin in polygynous ants. Behav Ecol Sodobiol. 27. 31-6.
Pamilo. P. (199Ia., Evolution ufthe slerile caste. JTI,ear Bioi. 149. 7$-95. Pamilo. P. (1991 b) Evolulion of colony characteristics in ~(Kial inseds. I. Sex allocation. Am Nawr. 137. 83-107.
PallCer. Z.. Gershon. H. & Rinkevich. B. (1995) Coexislcnce and possible parasitism of somatic and germ cell lines in chimeras of the colonial urodlordalc BiJtryllus sclrlossen·. Bioi Bull. 189. 10612. Parker. G.A. (1970a) Sperm competition and its evolutionary consequences in the insedS. 8iol Rev. 45.52$-67. Parker. G.A. (1970b) The reproductive behaviour and tht.· nature of sexual seledion in ScalophaBa s/('rcoraria L. L.Oiptera: Scatuphagidae). VII. The origin and evolution of the passive phase. Evolution, 24.774-88. Parker. G.A. (1970c) The reprodudive beha, iour and the nature of sexual selection in S<-alophagQ strrroraria L. (Diptera: Scatophagidae).II. The fertilizaliun rate and spatial and temporal relationships of each sex around (he sile of mating and ovi· position. J Anim £(01. 39, 205-28. Parker. G.A. (1974a) Assessment strategy and the evolution of fighting behaviur. J TJrtor BioI. 47, 22343.
Parker. G.A. (1974b) The reproductive behaiour and the nature of sexual selection in Scatophaga sler(oraria. LX. Spatial distribution of fertilisation rates dnd evolution of male search strategy within the rt>produaive area. Evollition. 28. 93-108. Parker. G.A. (1978a) Searching for mates In: Behavioural Ecology: an £vollltionary Approach (eds J.R. Krebs & N.B. Daries). Blackwell Scientific Publica lions. Oxford. Parker, G.A. (1978b) Seleaion on non-random (lsion of gametes during the evolution of anisogamy. J rlltor Bioi. 73. 1-28.
430
REFERENCES
Parker, G.A. (1979) Sexual selection and sexual conflict. In: Stxual StJulion and RtproduL
Parker. G.A. (1982) Why are there so many tiny ~pt'rm7 Sperm competition and the maintenance of two sexes. J Tlteor Bio/. ~6. 281-94. Parker. G.A. (1983) Matequalily and mating decisiuns. In: Matt Choi,t (ed. P. Bateson). pp. 141-66. cambridge University Press. Cambridge. Parker. G.A. (1984a) Evolutionarily ~table sIr3tt'gies. In: &haviournl Eco!ogy:an EvolmicnaryApproad, (cds J.R. Krebs & N.B. Davies). 2ncl cdn. pp. 30-61. Blackwell Scienlific Publications. Oxford. Parker. G.A. (1984b) Sperm competition and Ihe
t'volution of animal mating strategies. In: Sperm Competition and the Evolution ofAnimal Matil19 Systems (ed. R.L. Smith), pp. 1-60. Academic Press. Orlando. Parker, G.A. (1990a) Sperm competition games: rarnes and roles. Proc Roy Soc Lend B. 242. 120-6. Parker. G.A. (1990b) Sperm competition games: sneaks and eXlra·pair copulations. PrO( Roy Soc Lond B. 242. 127-33. Parker. G.A. (1992) Marginal value theorem with exploitation time costs: diet. spt~rm reserves, and optimal copula duration in dung flies. Am Natur, B9. 1237-56. Parker. G.A. (1993) Sperm competition games: sperm size and sperm number under adult comro!' Proc Roy So, Lend B. 253. 245-54. Parker, G.A. & MacNair. M.R. (1978) Models of parentoffspringconnia.1. Monogamy. Allim Behav. 26. 97110. Parker. G.A. & Simmons. loW. (1991) A mudel of constant random sperm displacement during mating: evidence from Scatophago. PrO( RoyS« Lund B.246.107-15. Parker, G.A. & Simmons. L.W. (1994) Evolution of phenotypic optima and copula duration in dungflies. Nature Lond. 370. 53-6. Par'.er. G.A. & Smith. J.L. (1975) Sl'term competition and the evolution of the precopulatory passive phase behaviour in Locusta migfQwr;o migratorioidts. J Eneomol (A). 49. 155-71. Parker. G.A. & Stuan. R.A. (1976) Animal behaviour as a strategy optimizer: evolution (If resource assess· ment strategies and optimal emigration Ihresholds. Am Nawr, 110, 1055-76. Parker. G.A. & Slilheriand. W.J. (1986) Ideal free distribution when individuals differ in competitive ability: phenotype-limited ideal free models. Anim Behav. 34. 1222--42. Parker. G.A .. Baker. R.R. & Smilh. V.G.F. (1972) The origin and evolution of gamete dimorphism and
the male-ft'male phenomenon. J Theor 8iol. 36. 529-53. Parker. G.A .. Simmnns. L. W. & Kirk. H. (1990) Analysing sperm mmpt'litio!l data: ..imple modl'ls fc)r IlrediCiing mechanisms. Behav £Col Sotiobiol. 27. 55---{;5. Partridge. l. (1976) Individual diHerences in feeding efficiencies and feeding preFerences of ('aplive gn.'at tits. Anim BelJav, 24, 230-40. Partridge. L. & Green. P. (1987) An advantage for s~dalisl feeding in jackdaws Corvus monedula. Anim 8ehov, 35. 982-90. Paterson. H.E.H. (1985) The recognition concept of SI>ecies. In: S/Hcies and Speciation (ed. E.S. Vrba). pp. 21-9. Transvaal Museum, Pretoria. Payne:. R.B. (1985) Behavioral continuity and change in local son1o: populations of village indigobirds Viduo r!talybtata. Z n'erpsydrol, 70. 1-44. Payne. R.5. & Webb. D. (1971) Orientatiun by means of long range acoustic signalling in Baleen whales. Ann N Y Acod Sci, 188. 110-41. Pearl. R. (1928) Tht Ratr of Livil1!1. Alfred A. Knopf. ew York. Pearson. B.• Raybould. A.F. & Clarke. R.T. (1995) Breeding behaviour, relatedness and sex-investmelll ratios in Leptotllorax tIIberum Fabridus. B',omol Expt'r Appl. 75. 165-74. Pemberton, J.M., Albon. S.D.. Guinness. F.E .. ClullonBrock. T.H. & Dover, G.A. (1992) Behavioural estimates of male maling success leMed by DNA fingerprinting in a polygynous mammal. &hav £col. 3.66-75. Pennycuick. C.J. (J 989) Bird Flight pffformanct: a Practical Calculation Manual. Oxford Scientific Pub· Iicalions. Oxford. Pennycuick. CJ. (1992) Nnvton Rilla Biology. Oxford University Press, Oxford. Peters, A.. Streng, A. & Michiels. N. (1996) Maling behaviour in a hermaphroditic flatworm with reciprocal insemination: do they assess Iheir mates during copulation? Ethology. 102. 236-51. Pel~rs. J.M .. Queller. D.C. Strassmann. J.E. & Solis. CR. (1995) Maternity assignmelll and queen replacement in a social wasp. Proc Roy Soc Lolld B. 260.7-12. Pett.'rs. R.H. (1983) The Ecologirallmplicatiolls of Body Size. Cambridge Universily Press. Cambridge. Petrie, M. (1983) Female moorhens compelt' rorsmall fat males. Science, 220. 413-15, Petrie. M. (1994) Improved growth and survival of oHspring of peacocks with more elaborale lrains. Naturt. 371. 598-9. I)elrie. M .• Halliday. T. & Sanders. C. (1991) Peahens prefer peacocks with elaborate trains. Anim Bthav.
REFERENCES 41. 323-31. Prennig. D.W. (1990) 'Kin recognition' among spadefoot toad tadpoles: a side·effect of habitat seleaion? Evolulion. 44, 785-98. Pfennig, D.W. (1992) Polyphenism in spadefoot load tadpoles as a locally-adjusted evolutionarily siable strategy. Evolution. 46, 1408--20. Pfennig. D.W. & Collins. J.P. (1993) Kinship affects morphogenesis in cannibalistic salamanders. Nature, 362.836-8. Pfennig. D.W. & Sherman. P.w. (1992) fdentifying relatives. Scienu. 255, 217-18. Pfennig. D.W. & Sherman. P.W. (1995) Kin recognition. Sci Am. 272. 98-103. Pfennig. D.W.. Loeb. M.L.G. & Collins. J.P. (1991) Pathogens as a factor limfting the spread of cannibalism in tiger salamanders. Oecologia. 88, 16l-6. Pfennig. D.W.• Reeve. U.K. & Sherman. P.W. (1993) Kin recognition and cannibalism in spadefool lOad ladpoles. Anim BeJrav, 46. 87-94. Pfennig, D.W.. Sherman, P.W. & Collins, J.P. (1994) Kin recognition and cannibalism in polyphcnic salamanders. Beltav Ecol. 5,225-32. Pfennig, D. W., Gamboa, G.J., Reeve, H.K., Reeve. J.S. & Ferguson, I.D. (1983) The mechanism of nestmate discrin;inalion in social wasps (PoUstts. Hymenoptera: Vespidae). 8thav Erol Sodobiol. 13. 299-305. Pic:rsma, T. (1988) Breast muscle atrophy and constraints on foraging during the flightless period of wing moulting Great Cresled Grebes. Ardea, 76. 96-106. Piersma. T. (1990) Pre-migratory 'fattening' usually involves more than the deposilion of fat alone. Rbtg Migr. II. 113-15. Pielrewicz. A.T. & Kamil, A.C. (1979) Search image formation in the blue jay (Cyanocitta cristata). Science. 204. 1332-3. Pinxten, R, Hanolte. 0., Eens. M., Verheyen. R.F.. Dhondt, A.A. & Burke, T. (1993) Extra-pair paternity and intraspecific brood parasilism in Ihe European starling, Sturnus vulgaris: evidence from DNA fingerprinting. Anim Behav, 45, 795-809. Piper, W.H. & Slater, G. (1993) I'olyandry and incest avoidance in lhe cooperative stripe-backed wren uf Venezuela. Behawour, 124, 227-47. Pilcher. T.J .. Grt.'en. D. & Magurran, A.E. (1986) Dicing Wilh death: predalor inspection behavior. J Fish BioI, 28. 1439-48. Pitcher, T.J., Magurran, A.E. & Winneld, LJ. (1982) Fish in larger shoals find rood raster. Bthav Ecel Sociobiol. 10. 149-51. Pitt'lka. EA .. Holl11t:'s. R.T. & Maclean. S.F.J. (1974) Ecology and evolution of social organisation in Arctic sandpipers. Am Zool, 14, 185-204.
431
Pilnick, S., Spicer, G.S. & Markow. T.A. (1995) How long is a giant sperm? Nature, 375, 109. P01dmaa. T. (1995) Parenll1ge, kinship and tht significana
of Ittlping behaviour in the cooperatively breeding noisy minor, Manorina melanocephala. PhD thesis, Queens Universily, Kingston. Ontario. Canada. Pomerantz. J.R. & Kubovy. M. (1986) Theoretical approaches to perceptual organization, simplicity and likelihood principles.I.n: Handbook a/Perception and Human Performance, Vol. n. Cognitive Processes and Performance (eds K.R. Boff, L. Kaufman & J.P. Thomas). pp. (36) 1-(36)46. John Wiley and Sons. New York. Pomiankowski, A.N. (1988) The evolution of {emale mate preferences (or male genelic quality. Oxford Surv Evol Bioi, S. 136-84. Pomiankowski, A. & Iwasa, Y. (1993) Evolution of multiple sexual ornaments by Fisher's process or sexual selection. Proe Roy Soc Lend B, 2.53, 173-81. Pomiankowski, A. & Meller, A.P. (1995) A resoluLion or the lek paradox. Proc Roy Soc Lond B, 260, 21-9. Pond. C.M. (1978) Morphological aspects and the ecological and mechanical consequences ot fat deposition in wild vertebrates. Ann Rev Eco/ Sy:it. 9, 519-70. Poner. R.H. (199]) Mutual mother-infant recognition in humans. tn: Kin Recognition (ed. P.G. Hepper), pp. 413-32. Cambridge University Press. Cambridge. POllS, WoK .. Manning. c.J. & Wakeland. EX (1991) Mating patterns in seminatural populations of mice influenced by MHC genotype. Nature, 352. 61921. POItS, W.K., Manning. c.J. & Wakeland, E.K. (1994) The role of infectious disease. inbreeding. and mOl ling preferences in maintaining MHC genelic diversity: an experimental test. Phil Trans Roy Soc Wild 8. 346. 369-78. Prestwich. K. (1994) The energetiCS of acoustic signaling in anurans and inseas. Am Nawr. 34. 6253. Pritt', M.V. & Waser, N.M. (1979) Pollen dispersal and optimal outcrossing in Delphin;um nelson;. Nature. 277.294-7. Price. K.. Harvey. H. & Ydenberg. R. (1996) Begging tactics of nestling yellow-headed blackbirds, Xamhocephalus xanthocephalus. in relation to need. Allim Behav, 51, 421-35. Primack, R.B. (1993) Essentials o/Constrvation Biology. Sinauer, Sunderland, MA. Proctor, H.C. (1991) Courtship in the water mitt: Neumauia papillator. males capitalize on female adaptations for predation. Anim Bthav. 42, 58998. PrOClOr. H.C. (1992) Sensory exploitation and the
432
REFERENCES
evolution of malt' mating behaviour: it cladistic lesl u~ing
water miles (Acari: Parasitengonit). Anim
£kilnv. 44. 745-52. Prop. J. & DcerenheTK- C. (1991) Ft'cding constraints In spring slaging hrent geese and the impaci of dit't on Ihe accumulation of lxKiy reserves. Ofcologia. 87. 19-28.
PruclI-jones. S. (1992) Indepcndenl versus nonIndependent mall' choice:: do ft:'malt's copy l'itch tHher? Am Natur. 140. 1000-9. Prut;'II·Jones. S.G. & Lewis. M.J. (1990) Hahital Iimil3tion and sex ralio pnU1l0leS delayed dispersal in Superb !airy·wrens. Nalllr~. 348. 541-2.
PTlIt'tl-Jones, S.G. & Wade. M.J. (199m Felllale lopying increases Ihe variance in mall' lll
CNlservalioll BioIOSY: the Scienu ofScara'ly and OiwNily fed. M. Soule). pro 35-56. Sinauer AS'iorialt's. Sunderland. Massachusells. Rall!'-. K.. Ballou. J.D. & Tcmplelon, A. (1988) Estimatcs of lei hal equivalents and tht· COSt or inbfl'cding in I11Jlllmals. CelllStrll BioI. 2. 185-93. Rand. A.S .. Ryan. M.J. & WilC7ynski. W. (1992) Signal
rt'dundancy and receiver permis'iivt'lless in a(()ll~lir matt' reco~nilion by the tungara [ro~. Physalaemtls plistulosus. Am Zool. 32. 81-90. Ranta. E.• Rila. H. & lindslr6m. K. (1993) (nmpeliliol1 versus cooperation: success o[ individuals foraginR al()nt~ and in groups. Am Nawr. 142.42-58. R~lInieks. F.L.W. (1988) Rl'producliv(' harmony via mutual r'>t)licing by wt)rkel'S in euS()(;al Hymel1fliHera. Am Nailir. 132. 217-36. Ratnic:ks. F.l.W. (1993) Egg-Iayin~ t'AA-rt·l11oval. alld ovary developmenl hy wtlTkers in Qllct'nright h()llt·y hl'e colonies. &hav Etol Sociobiol, 32. 191-8. Raillieks. F.L.W. (1995) Evidcm-(.· for a <jueenproduced t'gg-marking r)!1erol1lnnl' and its USt' in worker Jlolicing in the honey bet·. J Ari(L/11 Res. 34, 31-7. Ralnieks. F.L. W. & RC:l.'ve. H.K. (1991) The t'volution uf Quc:en-rt:aring nepolism in social Hynu'noptcra: dft'rts of discrimination costs in swarming specks. .J £vol 8iol. 4. 93-115. Ralnieks, F.L. W. & Reeve. H.K. (1992) Conflici ill single-qut't:'n Hymenuptcran socielies: the.: SlrUClurc: ur CunniC1 and pnKesSt.'S that reduct> confliCl in advanced eusocial spt'cies. J Theoret BioI, 158.33-65. Rall1ic:ks. F.L. W. & Visscher. P.K. (1989) Worker policing in Ihe honeybee. Naiurt. 342. 796-7. Raubenhdmcr, D. & Simpson. S.J. (1995) A lllultilt:wl analysis or feeding hehaviour: Ihl.' ~l'(}metry of nutritional decisions. Phil Trans Roy Soc l.and B. 342. 381-402. Rayner, J.M.V. (1990) The mechanics of flight and hird migration perfurmance. In: Bird Migrmioll. Physiology and Ecology fed. E. Gwilll'wr), pp. 28399. Springer-Vt.'rlag. Berlin. Read. A.F. & Harvey. P.H. (1989) Re.Js'l·ssment or comparative evidence for Hamilton and luk Iheory on the evolution or secondary sl'xual character:.. Nawrt. 339. 618-20. Read. A.F. & Nee. S. (1995) Infc:renCl.' frolll hinary comparativc data. J Theor Bioi. 173, 99-108. Real, L.A. (1990) Search theory and lllale choict;'. I. Models of single-sex discrimination. Am Namr. 136. 376-404. Real. L.A. (1991) Animal choice bc:havior and Iht' l'vollllion of ('ognitiv(.' archill'Clurt'. $tit'/1t('. 253, 98(}'6. Rehoreda. J. & Kacelnik. A, (1990) On n}operatioll, tit-for-Iat and mirrors. Anim &hav. 40. 1188-9. Roomeda. J.e. & Kacdnik. A. (1991) Ri,k sonsi'ivily in STarlings: variability in looel amUlIlH and ftKld dday. £kilov Erol. 2. 301-8. Redfidd. R.J. (1993) Genes of hrcakfast: tht' havt'you r-ca h'-and -t:ilt -il-loo of ha(.kriall Tallsfofmmit 111. J Hered. 84. 400-4.
REFERENCES Redondo. T. (1993) Exploitation of host mechanisms lor parental care by avian brood parasites. Eto!ogia. 3, 235-97. Rt'dondt). T. & Caslro. F. (1992) Sigllallingof Ilutriliollill need by magpie nestlings. Ethology. 92, 193-204. Ret've. H.K. (1989) The evolution of cnnspcciric acceptance: thresholds. Am Natur, 133.407-35. Reeve, H.K. (1993) Haplodiploidy, eosociality and allst.'llce of male parental and alloparelllal care in Hymenoplera: a unifying genetic hYPOlhesis distincl from kin seleclion theory. Phil Trans Roy Soc Lond 8. 342, 335-52. Rel.'Vl'. H.K. (1997a) Game theory. reluoductive skcws and nelwtisJll. In: G'ame Theory and &havior (l'ds L.A. Ougatkin & H.K. Reeve). Oxford University Press. Oxford. Reeve. H.K. (1997h) Acting for till' good of othl'rs: kinship and redpmcity. with some new twiSts. In: Evalmiol! and Humall Behavior (cds C Crawford & D. Kn:hsJ. in press. Lawrenct' Erlbaum Associatcs. Hillsdale. Nl'W Jt'rst..'y. Recvt'. H.K. & Kdlt'r. L. (1995) Partitioning of rellroductiol1 in mOlher-daughlcr versus sibling associations: a lest of optimal skew lheory. Am Natllr, 145, t 19-n. R(:cvt'. H.K. & Nonacs. P. (1992) Social COlllracts ill wasp socielies. Nature. 359. 823-5. R4;'4;'vc.', H.K. & Ratnicks. EL.W. (1993) Quecn-quel:n (:onfliCis in polygynous sodelit:s: mutual tolcranct' and rcproductive skew. In: Quem Number and Sociality in Insects (t:d. L. Keller). pp. 45-85. Oxrord University Press. Oxford. Rcc.'vc. H.K. & Sherman, P.W. (1993) Adaptalion and the goals (If evolutionary research. Quart Rev BioI. 68, 1-32. Rdd. M,L. 11987) Cosllincss and reliahility in Ihe singing viglHlr of fpswich sparrows. Anim lkhav. 35. 1735-43. Reid. P.l. & Sht:ltlcworth. S.J. (1992) DeteCliol1 or nyplic prt'Y: search imagl' or search ralc.'? J Exper P.~ychol Animal BeJrav Process. 18. 273-86. Reilly, L.M. (1987) Measurements of inbreeding and average relaledness in a termite populalinn. Am Nawr. 130,339-49. Rdmann. J.G .. Moen. D.O. & Thorson, B.J. (1967) Female Illonogamy and its control in lht: housefly. Mm-ca domestira L. J Insect Physiol. 13.407-18. Rensch. I. & Rensch, B. (1956) Relalivl-' organmaszt:
lid tropischen warmbllilern. Zool Am'. 156. 106-24. Rcynolds. J.D. (1996) Animal breeding systems. Trends Heal F.m!, 1 I. 68-72. Reynolds. J.D. & Gross. M.R. (1992) Fcmlllt: mal<:' preft:'renct' cnhan(cs offsprin).! growth and rc11Wdu<'-lion in a fish. Poecilia relicu!ata. Proc Roy SOl'
433
LOlld B, 250, 57-62, Rhdnlander. J. & Romer. H. (1986) Insect hearing in thl' fidd, J Com,O Physiol A. 1S8. 647-51. Rice. W.R. (1987) The accumulation of sexually antagonistic genl's as a selectivc agent promoting the cvohuioll of redu('cd recombination between primitive sex chromosomes. Evolution. 41. 911-14. Rice. W.R. & Hostert. E.E. (1994) Perspl'ctive: laboralory expi.:'rimelllS on speciation: what have we learned in rony years? Evolution. 47, 1637-53. Richman. A.D. & Price. T. (1992) Evolution of emoto:ical diff~rences in the Old World leaf warblers. Na/Ure. 355, 817-21. Riehm·r. H.. Christe. P. & Oppliger. A. (1995) Paternal in'eSlll1l~nt afft.'cts prevalence of malaria. Proc Nat Acad Sci USA, 92, 1192-4. Ridley. M. (19831 The I::.xplauotion of Organic Diversity:
the Comparative Method and Adaptatiom for Maring. Oxford University Press. Oxford.
Ridley. M. & Grakn. A. (1981) Are green beard gc.'nes outlaws? Anim Behav. 29, 954-5. Ridley. M. & Grakn. A. (1996) How to study discrete comparative methods. In: Phylogenies and the ComparmiVf Method in Animal Behavior (ed. E.P. Marlins). pr. 76-1 03. Oxford University Press. New York.
Rinkcvich. Bo, Porat. R. & Goren. M. (1995) AllnreC()~nili(ln dements on a urochordate hislOcompatibility locus indicate unprecedented extensive polymorphism. Proc Roy Soc Land B. 260. 319-24. Ritchil'..1. (1996) What is 'the paradox of the lek'? Trmds Ecol Evol. 11, ) 75. RilChic, M.E. (1988) lndividllal varialion in tbe alliJity o[ Columbian J::Hlund sqUirrels to select an optllllal die!. Evol Etol. 2, 232-52. Rivier. D.H. & Pillus. L. (1994) SilenCing speaks up. ed/, 76, 963-6. Robertson, J.G.M. () 990) Female choice increases fertilization success in the Auslralian frog Uperofia laevigata. Anim Behav. 39, 639-45. Rollinson. S.K. fr Terborgh. J. (1995) Interspecific aggrcssion and habitat selection by Amazonian birds. .J Anim Ecol. 64. I-II. Rodriguez-Girol1l~s.M.A .• Drummond. H. & Kacelnik. A. (1996) Elfcn of food deprivalion on dominance status in blue-fOOled booby (Sula nebouxii) broods. Behav E(OI. 7, 82-8. Roff. D.A. (19841 The evolulion oIBfe hislOry paraIlleters in lclco~ts. Call} Fis' Aquat Sci, 41. 989-1000. Roff. O.A. (1994) Why is there so much genetic variation for 'ing dimorphism? Res Popu! Ecol. 36, 145-50. Rogers. CM. (1987) Predation risk and fa~ling capacity: do ...... intt=r birds maintain oillimal hody
434
REFERENCES
mass? Ecology, 68, 1051-61. Rogers. C.M. & Smith. J.N.M. (1993) Life-history theory in the non-breeding pt'riod: trade-offs in avian fal reserves. Ecology. 74. 419-26. Rogers. C.M.. Nolan. V. & Kelterson. E.D. (199~) Winter fanelling in the dark-eyed junco: plasticity and possible interaction with migration uade-oHs. Oeco/ogia. 97. 526-32. Rogers. C.M .• Ramenofsky. M., Kenerson. E.D .. Nolan. V. & Wingfield. J.e. (1993) Plasma-corticosterone. adrenal mass. wimer weather. and season in nonbreeding populations of dark-eyed juncos (Junco hyemalis). Auk. 110.279-85. Rohwer. S. (1986) Selection for adoption versus infantidde by replacement mates in birds. Curr Omithol. 3. 253-336. Roisin, Y. (1994) Intragroup conflicts and the evolution of sterile castes in termites. Am Natur. 143.751-65. Romer. H. (1993) Environmental and biological constraints for the evolution of long-range signalling and hearing in acoustic insects. Phil Trans Roy Soc Lond B. 340. 179-85. Romer. H. & Bailey. W.J. ( 1990) Insect hearing in the field. Comp Biochem Pllysiol A. 97. 443-7. Rose. M.R. (1991) Evolutionary Bio/ogyofAging. Oxford University Press. Oxford. Ro>s. K.G. & Mallhews. R.W. (eds) (1991) The Social Biologyo!Wasps. Comstock, Ithaca. Rosse!. S. (1979) Regional differences in photoreceptor performance in the eye of the praying mantis. J Comp Physio/. Ill. 95-112. Roubik. D.W. (1989) Ecology and Natural History of Tropical8ees. Cambridge University Press, Cambridge. Rowe. L. Ludwig. D. & Schluter. D. (1994) TIme, condilion and the seasonal decline of avian clutch size. Am NaluT, 143. 698-722. Rowley. A.. Dowell. S.J. & Dimey. J.F.X. (1994) Recent dcvelopmenlS in the initiation of chromosomal DNA replication: a complex picture emerges. Biochim Biophys Ada. 1217. 239-56. Rowley, I. (1965) The life history oflhe Superb blue wren. Malurus cyantus. Emu. 64. 251-97. Rowley. I. & Russell. E. (1990) Splendid fairy-wrens: demonstrating the importance of longevity. In: C(loperQ/ive Breeding in Birds: LOllg-unn SlIldies Of Etologyand Ekhavior. (cds P.B. Stacey & W.O. Koenig),
pp. 1-30. Cambridge University Press, Cambridge. Royer. M. (1975) Hermaphroditism in insects. Studies on lcerya purchasi. In: fnlersexuality in Ihe Animal Kingdom (ed. R. Reinboth). pp. 135-45. SpringerVe-rIag. Berlin. Rubner, M. (1893) Oberdcn einnuss der korpergrosse auf stoff· und kraftwechsel. leiLSfh BioI. 19, 53562.
Russon. A.E. & Galdikas. B.M.F. (1993) Imitatiun in free· ranging rehabilitant orangutans (Pongo pygmaClls) J Camp Psycho/. 107. 147-61. Russon. A.E. & Galdikas. B.M.F. (1995) Constraints on great apes' imitation: model and action selectivily in rehabilitant orangutan (Pongo pygnlacus) imitation. J Comp Psychol. 109. 5-17. RuxlOn. G.D.. Gurney. W.S.c. & de Roos. A.M. (1992) Interference and generation cycles. Theor Popul BioI. 42.235-53. Ruxton. G.D .. Hall. S.J. & Gurney. W.S.C. (1995) Anraction toward feeding conspedfics when food patches arc exhaustible. Am Natur, 145. 653-60. Ryan. MJ. (1983) Sexual seJection and communicatiun in a neoITopical (rog. Physalaemus puslulosus. Evoluaon. 37,261-72. Ryan. M.J. (1985) the Tungara Frog. a SlIIdy in Sexual ~l«tion and Communication. Chicago University Press. Chicago. Ryan. M.J. (1988) Energy. calling and selection. Am Natur. 28. 885-98. Ryan, M.J. (1990) Sensory systems, sexual selection. and sensory exploitalion. Oxford SUTV Eval Bioi, 7. 157-95. Ryan. M.J. (1994) Mechanistic sludies in sexual selection. In: Ekllavioral Mrchanisms ill Evolwionary Ecology (ed. L. Real). pp. 190-215. University or Chicago Press, Chicago. Ryan. MJ. (1996) PhylogenetiC'S and behavior: some cautions and expectations. In: Pllylogmies and Ihe Comparative Method in Animal &havior (ed. E. Manins). pp. 1-21. Oxford University Press. Oxford. Ryan. M.J. & Keddy-Hector. A. (1992) DireClional patterns of female mate choice and the role of sensory biases. Am Nalur. 139. 54-535. Ryan. M.J. & Rand. A.S. (1990) The sensory basis of sexual selection for complex calls in the tungara frog, Physalaemus puslUlosus (sexual selection for sensory exploitation). Evallllion. 44.305-14. Ryan. M.J. & Rand. A.S. (I 993a) Sexual selection and signal evolulion: the ghosl of biases past. Phil Trans Roy Soc ~Ild B. 340. 187-95. Ryan, M.J. & Rand, A.S. (1993b) Species recognition and sexual selection as a unitary problem in animal communication. Evalulion, 47, 647-57. Ryan. M.J. & Rand. AS. (1995) Female responses to ancestral advertisement calls in the lungara frog. Science. 269. 390-2. Ryan, M.J. & Wagner, W. (1987) Asymmetries in mating preferences between spedes: rcmale swordtails prefer heterospedfic mates. Samee. 236. 5957. Ryan. M.l.. Hews. O.K. & Wagner. W.E. Jr (1990b)
REFERENCES S(.'xual sdCClion on allelc.-'s Ihat determini.:' hody ~izc in the swordtail Xiphopl1orus nigrensis. Bfhnv Etol Sociobiol. 26. 231-7. RYi'ln. M.J., Fox, J.H., Wilo.ynskL W. & RSaino. N.. Primmt'r. C. Ellegren. H. & Ml,lIt'r. A.P. (1996) Srxual ornamentation affects paternity in Ihe socially monogamous barn swallow Hirmldo rustica. Evolwio,., (in press). Sakaluk. S.K. & Belwood. J.1. (1984) Gecktl phonotaxis tucrickct calling song - a case of satellite:' predation. Anim BflJav. 32. 659-62. Sakaluk. S.K. & Eggert. A.·K. (1996) Fcmak contrul of sperm transfer and intraspecific variation in :>perm prt'cedence: antt'Ct.'dents 10 the evolution of a courtship food girl. El/olwion, 694-703. Salmon. M.A .. Van Mddden, L., Bernso,
Schierwater, B.. Streit. B.,
Wa~nt'r.
435
G.P. & DeSallt', R.
(Eds) (1994) Molecular Ecology and Evolution: Approaches al1d ApplicQtiollS. Birkhauser Verlag. Basc.::l. S(:hildher~t:r, K.. Huber, F. & Wohlers. D.W. (1989)
Ct'nlr.11 auditory pathway: neuronal correlates or phOlltJliKtic behavior. In: Cricket Behavior and Neurobiology (cds F. Huber. T.E. Moore & W. Loher).
pp. 423-58. Cornell University Prt.'ss, Ilhaca, .few York. Schjeldt'rup-Ehbe. T. (1935) Social behavior of birds. In: Handbooko[Social Psychology (t'd. C. Murt·hisun). pp. 947-72. Clark University Prt'ss. Worcester. Massachusetts. St'hlupp. I.. Marler. CA. & Ryan. M.J. (1994) Benefit to male sail fin mollit'sof mating: with heterospeciric females. Sdence, 263. 373-4. Schluter. D. & Nagel, L.M. (1995) Parallel speciation by natural selection. Am Nawr. 146.292-301. Schluler, D. & Price. T. (1993) HoneSty. perceplion and population divergence in sexually sde(·tt'd traits. Proc Roy Soc Umd 8, 253. 117-22. Schmid· Hempel. P. (1994) Infection and colony variahilily in social insects. Phil Trans Roy Soc Lond B. 346. 313-21. Schmid-Ht'll1pd, P. & Schmid-Hempt'i.. R. (1~93) Transmission of a pathogen in Bombus terrestris, with a note on division or labour in social insects. BdtQV Ecol Soriobiol. 33. 319-27. Schmid-l-il'mpeL P., Kact'lnik. A. & Houston, A.1. (1985) Honeybees maximize efficiency by not filling lheir nop. Bellav Ecol Sociobiol. 17. 61-6. Schmidt. J.M. & Smilh. J.J.B. (1986) Currdations hetween llOdy imgks and suhslrate curvatUre in the parasif()id wasp Trichogramma milllltmn: a poslible mechanism or host radius measurements. J Exp Bioi. 125.271-85. Schl1lidt·Niclscn, K. (1984) Scalil1.9. Cambridge University Press. Cambridge.
Schot'ncr. T.W. (1971) Theory or fceding strale~ies. Ann R~v Ecol Sysl. 2. 369-404. Schwagmeyer. P.L. & Parkt::r, G.A. (1987) Queuing ror mates in thirteen-lined ground squirrels. Anim
8eltov.35.1015-25. Schwagmeyer. P.L & Parker. G.A. (1990) Male male ch()icc as predicted hy sperm competition in thine-enlined ~round squirrels. NaWrt (Lond), 348. 62-4. Schwagmeyer, P.l. & Parker, G.A. (1994) Matc-quilting rules for lIlale thineen-Iined ground squirrels. Behav Ecol. 5. 142-50. Schwarzkopf. L. & Shine. R. (1992) Costs orrcprodlll.... lion in 1i7..:'lrds: escape tactics and susceptibility to predation. Behw Ecol Sociobiol. 31. 17-25. Scudo. EM. (1967) The adaptive value of scxuill dimof!)hism. I. Anisogamy. Evolwion. 21. 285-91.
436
REFERENCES
Searcy. W.A. (1992) Song repenoire and matt' choice in bird. Am ZO<='>
Sherman. P.W. (1989) Mate guarding as paternity insurance in Idaho ground squirrels. Nature Lond. 338.418-20. Sherman. P.W. (1991) Mllhi~)le maling and kin reco};nition by sclf-inspeaion. EtJrol Soeiobiol. 12, 377-86. Sherman, P.W. & Holmes, W.G. (19851 Kin recognition: issues and evidence. In: Exptrimenltll Behavioral Ecology and Sociobiology (cds B. Holldobkr & M. Lindauer), PI>. 437-60. Sinallcr. Sunderland. Mas)achust:l1s. Sherman. P.W.. Seeley. T.O. & Reeve. H.K. (1988) Parasites. pathogens, and pulyandry in social Hymenoptera. Am Natur. 131. 602-10. Sherman. P.W.. Lacey. E.A.. Reeve. H.K. & Keller. L. (1995) The cusociality continuum. Brl1av £col. 6, 102-8. Sherry. D.E (1985) Food storage hy birds and mammals. Adv Stud Behav. 15, 153-88. Sherry. D.E & Galer. B.G. Jr (1984) Cultural trans mission withoul imitation: milk bottle opening by birds. Ani', &hav. 32. 937-8. Sherry, D.F. & Vaccarino. A.l. (1989) Hippocampus and memory for caches in the black-capped chickadee. Behav Neurosd. 103. 308-18. Sherry, D.E. Jacobs, L.E & Gaulin, S.J.C. (1992) Spatial memory and adaptive specialization of the hil)Jlocampus. Trends Nturosd. 15. 298-303. Sherry, O.F.. Krebs. J.R. & Cowie, R.J. (1981) Memory for the location of stored food in marsh tits. Ani', 8thav.29. 1260-6. Sherry. D.F.. Mrosovsky. N. & Hogan. J.A. (1980) Weight loss and anorexia during incubalion in birds. J Comp Physiol Psychol. 94. 89-98. Sherry. D.E. Vaccarino. A.L.. Buckenham. K. & Herz. R. (1989) The hippocampal complex of food storing birds. Brain Bthav Evol, 34, 308-17. Shcttlewonh, S.J. ( 1990) Spatial memory in food storing birds. Phil Trails Roy Soc LOlld 8, 329, 143-51. Sht'ulewonh. S.J. (1993) Where is the comparison in comparative cognition? Alternative research programs. Psyehol Sri. 4. 179-84. Sheltlewonh. sJ. (1995) Comparative studies of memory in food storing birds: from the field to Ihe Skinner box. In: &havioral Brain Rtstarch in Naturalistic and Stmi naturalistie Sittings (eds E. Alleva. A. Fasolo. II.P. Lipp. L. Nadel & L. Ricceri), pp. 159-92. NATO AS} Series, Kluwer Academic Press, Dordrecht. ShelleIVorlh. S.J. & Plowrighl. C.M.S. (1992) How pigeons estimate rates of prey encounter. J Exper Psychol. 18. 219-35. Shettleworth. S.J.. Krebs. J.R .. Healy. 5.0, & Thomas. C.M. (1990) Spatial memory of food-storing tils 4
4
REFERENCES (Parus atu and P. tHricapiJIus): comparison of slOrilll? and nun-storing tasks. J OJmpar Psychol. 104.71-81. Shetllc-worth. SJ.. Krebs. J.R. Stephens. D. W. & Gibtxlll. J. (1988) Tracking a fluctuating environment: a ~tudy of sampling. Anim Btlzav. 36. 87-lO5. Shields. W.M. (1982) Philopatry. Inbr~fdi'Jg. and tltt Evollllion ofSex. Stale University of New York Prt>Ss. Albany. New York. ShimkclS. L.J. (1990) Social and developmental biol(lgy of Ihe myxobacteria. Microbiol Rev. 54. 473-501. Shint:'o R. (1980) 'Costs' of reproduction in rcrtill'S. Ot(o/ogie. 46. 92-100. Short. R.V. (1977) Sexual selection and Ihe desct'11l of man. In: ProcudinfjS oftht CtJnbtrra Symposium on Rrprodunion and Evalwion. Australian Aradtmy of Scit1Jer. pp. 3-19. Canlx'rra. Shuster. S.M. & Wade. MJ. (1991) Equal rna'ing success among male reproductive strategies in a marine isupod. Nawrr Lond. 350. 608-10. Sibley. e.G. & Ahlquist. J.E. (1990) Phylogwy and Classifiratio' of Birds. Yale University Press. New Hiwc:-n. Siegel. R.A .. Huggins. M.M. & Ford. N.B. (1987) Reducliun in locumotor ability as a cost of repro· duction in gravid snakes. Orcologia. 7}, 481-5. Sigurjonsdottir, H. & Parker. G.A. (1981) Dung fly struggles: evidence for assessment strategy. 8elJav Erol Sodobiol. 8. 219-30. Sillen·Tullbcrg. B. (1988) Evolution of gregarious· ness in aposcmatic butterfly larvae: a phylogenetic analysis. Evolution. 42.293-305. Sillt'n-1Jllberg. B. (1993) The effect of biased inclusion of taxa on the correlation between discrete characters in phylogenelic trees. Evolution. 47.1182-91. Simmons. L. W. (1986) Female choice in the field cricket. Gryllus bimaculatus (De Geer). Ani', &hav, 34. 1463-70.
Simmons. L. W. (1989) Kin recognition and its influence on mating preferences of the field cricket. Gryllus bimaculaws (de Geer). Anim &hav, 38, 68-77. Simmons. L. W. (1991) Female choice and the relalt'dIll'SS of mates in the field cricket. Gryllus binraculatus. Anini Bthav. 493-50 I. Simmons. L.W.. Stockley. P.. Jackson. R.L. & Parker. G.A. (1996) Spenn cOJ1l~lition Or sperm selection: nu evidence for female influence over paternity in yellow dung nies Scatophaga sttrcoraria. Bthav £eol
4'
Sariob;o/. 38. 199-206. Simmons. L.W., llorens, T., Schinzig.. M.. Hoskcn. D. & Craig, M. (1994) Courtship role reversal in bush crickets: anolher role for parasites? Alii', Brhav. 47, 117-22.
Sinclair. A.R.E. (1989) Population regulation in animals. In: Ecological Conupts (ed. J.M. Cherrett).
437
pr. 197-241. Blackwell Scientific Publica lions, Oxford. Sinervo. B. & Lively. C.M. (1996) The rock-paperscis'iors game and the evolution of alternative male strategies. Nature Lond, 380, 240-3. Sinervo. B.. Hedgl'S. R. & Adolph. S.c. (1991) Decreas<.-d sprint speed a~ a cost of reproduaion in the lizard SetloPtmlsOlddmralis: variation among populations. J Exp Bio/. ISS. 323-36. Sinervo. B., Doughty. P.. Huey. R.B. & zamudio, K. (1992) Allomelric engineering: a causal analysis of natural selection on offspring size. Scitnet. 258. 1927-30.
Singer. T.L. & Espelie. K.E. (1992) Social wasp' use nest paper hydrocarbons for nestmate recognition. Ani', Bf'hav. 44. 63-8. Slatkin, M. (1985) Gene now in natural populations. AIIII
Rtv E(o/ Sysl. 16. 393-430.
Sial kin, M. (1993) Isolation by distance in equilibrium and non·equilibrium populations. Evolution, 47. 264-79.
Slatkin, M. (1995) A measure of population subdiVIsion based on microsatellite allele frequendes. Gtntries. 139. 457-62.
Smale. L. Nunes. S. & HoJekamp. K.E. (1997) Sexually dimorphic dispersal in mammals: patterns, primal callses and consequences. In: Advances in the Study ofBelravior. Vol. 26 (eds P.J.B. Slater. J.S. Rosenhlatt. C.T. Snowdon & M. Milinski), pp. 181-250. Acadc:-mic Press. london (in press). SOli,h. E.A. & Wimerhalder. B. (eds) ( 1992) Evo/utirnery £Cologyand Human Behavior. Aldine De Grllyter. New York. Srnilh. G.T.. Wmgfield. J.c. & Veit. R.R. (1994) Adrenocortical response to stress in the common diving petrel. Prlrcanoides urinatrix. Physiol Zool. 67, 526-37.
Smith. J. . M. & Arcese. P. (1989) How fit are floaters? Consequences of alternative territorial beha, iour in a nonrnigralory sparrow. Am Nawr. 133.830-45. Smilh. R.L. (1979) Repeated copulation and sperm precedence: paternity assurance for a male brooding water bug. Scil·lJct. 205. 1029-31. Smith. R.L. (cd.) (1984) Sptr71l OJmpth'tion and tht Evolution ofAnimal Mating Systtms. Academic Press. Orlando. Smith, S.M. (1975) Innate recognition of oral snake patterns by a l>ossibie avian predator. Sdmet, 187. 759-{,0.
Smuts. B.B., Cheney, D.l., Seyfarth, R.M., Wrangham. R.W. & Slruhsaker. T.T. (1981) Pn'matt SoCIl!ti~. University of Chicago Press. Chicago. Snell. H.. Jennings. R.D .. Snell. H.M. & Harcoun. S. (1988) Intra·population variance in preddwr·
438
REFERENCES
dvoidance pedormance of Galapagos lava lizards: the interaction of sexual and natural selection. Evol Ecol. 2. 353-69. SokaL R.R .• Harding. R.M. & Oden. N.L. (1989) Spa'ial jJatlerns of human gene frequencies in Europe. Am J Phys Anthrop. 80. 267-94. Solomon. N. & French, J.A. (1996) Coopcratiw Brttding JnMammals. Cambridge University Press. Cambridge. Soule. M. (1982) Allometric variation. I. The theory
and some consequences. Am Nalur, 120. 751M. Soule, M. & Cuzin-Roudy. J. 11982) Allometric variation: 2. Development insldbilily of extreme phenotypes. Am Natur. 120. 765-80. Southwood. T.R.E. (1973) The insect/plant
relationship - an evolutionary perspective. Symp Roy Entomol Soc Lond. 6. 3-30. Spiess. E.B. (1987) Discrimination among prospective males in Drosophila. In: Kin Rec(lgnirion in Animals (eds D.J.C. Fletcher & C.D. Michener). pp. 75-119. John Wiley & Sons. New York. Squire. loR. (1992) Memory and the hippocampus: a synthesis from findings with rats. monkeys. and humans. Psychol Rtv. 99. 195-231. Stacey. P.B. (1979) Kinship. promiscuily and communal breeding in the acorn woodpecker. Behav EroISOOobiol.6. 53-66. Stacey. P.B. & Edwards. T.e.. Jr ( 1983) Possible cases of infanticide by immigrant females in a groupbreeding bird. Auk. 100.731-3. Stacey. PB. & Koenig. W.O. (1990) Cooperative Breeding
ill Birds: Long-Itr', Studits of Ecology and Behavior. Cambridge University Press. Cambridge. Stacey. P. & Ligon. J.D. (1987) Territory quality and dIspersal options in the Acorn woodpecker. and a challenge to Ihe habitat-saturation model of cooperative breeding. Am Natur. 130,654-76. Stacey. P. & Ugon. J.D. (1991) The benefils of philopatry hypothesis for the evolution of coopcralive breeding: variation in territory quality and group size effects. Am Natur. U7. 831~6. Staddon. l.E.R. (1975) A nOle on the evolutionary significance 01 supernormal Slimuli. Am Natur. 109. 541-5.
Stahl. W. (J 962) Similarily and dimensional methods in biology. Science. 137. 205-12. Stamps. l.A. () 99)) The ('[feet of conspecifics on habilat selection in territorial species. &hav ECDI Sodobiol. 28. 29-36. Stander. P.E. (1992) Cooperative hunting in lions: the role of the individual. Ekhav Ecol Sociobiol. 29. 44554.
SLar'-, R.E. (1989) Beobachtungen zur Nestgriindung und Brutbiologie der Holzbiene Xylocopa sulcalipts
Maa (Apoidea: Anthophoridae). Mitteilungen der Deutschen Gtsellschaft fUr allgtmeine und an9fwandtt Enlomol09;e. 7. 252-6. Stark. R.E. (1990) Intraspecific nest usurpation in the large carpenter bee Xylocopa sulcatipes Maa (Apoidea: Anthophoridae). 1n: Sodal/1IMClS and tht Environment (eds G.K. Veeresh. B. Mallik & C.A. Viraktamath). pp. 165-6. Oxford & IBH Publishing Co. Pvl. Ltd. New Delhi. Stark. R.E. (1992a) Cooperative nesting in th~ multivolt inc large carpenter bee Xylocopa sulcatipes Maa (Apoidea: Amhophoridae): do helpers gain or lose to solitary females? Ethology. 91. 301-10. Stark. R.E. (1992b) Sex ralioand maternal investment in the multivoltine large carpenter bee Xylocopa sulcalipts (Apoidea: Anthophoridae). Ecol Entomol. 17.160-6.
Stark, R.E.. Hefetz. A., Gerling. D. & Velthuis. II.H. W. (1990) Reproductive competition involving oophagy in the socially nestjng bee Xylocopa sulcatipes. Naturwisst1Jschaften. 77. 38--40. Starr. C.K. (1979) Origin and evolution of insect sodality: a review of modem theory. In: Social/~cts. Vol. I (ed. H.R. Hermann). Pl'. 35-79. Academic Press. New York. Starr. C.K. (1984) Sperm competition. kinship. and socialily in the aculeate Hymenoptera. In: Sperm Competition and the Evolmion ofAnimal Mating Systems (ed. R.L. Smith). pp. 427-64. Academic Press. Orlando. Stearns. S.c. (1992) Th, Evolution of Lif' Histori'. Oxford UniversilY Press:. Oxford. Stearns. S.c. & Koella. J. (1986) The evolution 01 phenotypic plasticity in Iife·history trailS: predictions of reaction norms for age and size in maturily. Evolution. 40. 893-913. Stearns. S.C. & Crandall. R.E. (1981) Quan'itative predictiOns of delayed maturilY. Evolution. 35. 45563.
Steams. S.C. & Crandall. R.E. (1984) PlastidlY for age and size at sexual maturity: a life·history adaplation to unavoidable stress. tn: Fish Rrprodurt;on (eds G. PollS & R. Woolton). pp. 13-33. Academic Press. New York. Stephens. D.W. (1981) The logic of risk·sensitivc foraging preferences. Anim &hav. 29. 628-9. Stephen. D.W. & Krebs, J.R. (1986) Foragi'g Thtory. Princeton University Press. Princeton. Stephens. D.W.. Anderson. J.P. & Toyer. K.B. (1996) On the spurious occurrence of til·for-tat in pairs of predalOr-approaching fish. A'im &hav (in press). Slephens. D.W.. Nishimura. K. & Toyer, K.B. (1995) Error and discounting in the iteraled prisoner's dilemma. J Thtor Bioi. 176.457-69.
REFERENCES Stern. D.L. (1994) A phylogenetic analysis of soldier evolution in the aphid family Hormaphididae. PrO( Roy Soc umd B. 256. 203-9. Stem. D.L. & Foster. W.A. (1996) The evolution of soldiers in aphids. 8io/ Rev. 71. 27-79. Stevens. T.A. & Krebs. l.R. (1986) Retrieval of Slored seeds by marsh tits. Parus pa/ustris. in the field. Ibis. 128. 513-25. Stewart, G.J. & Carlson. C.A. ([ 986) The biology of natural transformation. Ann Rev Microbiol. 40.211-
}5. Stillman. R.A .. Goss-Custard. J.D .. Clarke. R,T. & Durell. S.E.A. Ie V. dit (1996) Shape of the interference function in a foraging vertebrate. J Anim Ecol. 65, in press. Stockley. P. & Purvis. A. (1993) Spenn competilion in mammals: a comparative study of male roles and relative invcslrncm in sperm production. Fund Ecol, 7. 56()-70. Stone. G.N. & Sunnocks, P. (1993) Genetic consequences or an invasion through a patchy environment - the cynipid gallwasp Andricus quuOISCQUS {Hymenoptera: Cynipidae}. Mol £col. 2. 251-68. Sloner. G. & Breden. F. (1988) Phenotypic difrerentiation in remale preference related to geographic variation in male predation risk in the Trinidad guppy (Poedlia reticillata). Behav BcolSodobiol. 22. 285-91. Strong. D.R.. Lawton. J.H. & Southwood. T.R.E.(1984)
Insects on PlanlS: Community Patterns and Mechanisms. Harvard Univesity Press. Cambridge. Massadluselts. Stuebe. M.M. & Kettersoll. E.D. (1982) A study or fasting in tree sparrows (Spizella arboreal and darkeyed juncos (Junco hyemalis): ecological implications. Auk. 99. 299-308. Su, T.T., Follette. P.J. & O'Farrell, P.H. (1995) Qualirying for the license 10 replicate. ~ll. 81. 825-8. Subowski. M.D. (1990) Releaser-induced recognition learning. Psychol Rev. 97(2). 271-84. Suga, N. (1995) Processing of auditory inrormalion carried by species-specific complex sounds. In: rhe Cognitive Neurosciencts (cd. MS. Gazzaniga). pp. 295) 13. MIT Press. Cambridge. Massachusetts. Sundstrom. L. (1994) Sex ratio bias, relatednessasymmetry and queen mating rrequency in ants. Nature, 367.266-8. SundSlrum, L.. Chapuisat. M. & Keller. L. (1996) Conditional manipulalion of sex ralio by ant workers: a lest or kin selection theory. Scimet. 274. 993-5. Sutherland. W.J. (1992) Game theory models of functional and aggregalive rt'Sponses. O(colog;a, 90, 15()-2.
439
Sutherland. W.J. (1996a) From Individual Bthaviour to Population Ecology. Oxford University Press. Oxrord. Sutherland, WJ. (1996b) Predicting the consequences of habilat loss for migrating populations. PrOf Roy Soc Wild B. 26}. I H5-7. Sutherland. W.J. & Dolman. P.M. (1994) Combining bt:haviour and population dynamics with applications (or predicling the consequences of habitat loss. Proc Roy Soc B. 255. 133-8. Sutherland, W.J. & Moss. D. (1985) The inactivity of animals: influence or stochasiticity and prey size. Behav;our. 92. I-B. Sutherland, W.J. & Parker, G.A. (1985) Distribution or unequal competitors. In: Bthavioural EcoloSY: Ecological Cortwl/Ul'nCl'So!Adaptivt &haviour (eds R.M. Sibly& R.H. Smith). pp. 255-74. Blackwell Scientific Publications,Oxrord. Sutherland. W.J .. Jones. D.W.F. & Hadfield. R.W. (1986) Age dirrerences in the feeding abilities of moorhens GalHnllla chloroplls. Ibis. 128.414-18. Szekely, T. (1991) Reproduction of Kentish Plover, Charadrius alexandrinus, in grasslands and fishponds: habilal mal-assessment hypothesis. Aquila, 99. 5968. Szekely. T. & Reynolds. J.D. (1995) Evolotionary transitions of parental care in shorebirds. Pror Roy Soc LOlld B. 262. 57-64. SZylllura, J.M. (1993) Analysis or hybrid zones with
Bombina.ln: Hybrid ZOnes and the Evolut;onary Process (ed. R.G. Harrison), pp. 261-89. Odord University Press. New York. Taylor. C.R .. Heglund. N.C. & Maloiy. G.M.O. (1982) Energetics and mechanjcs or terreslrial locomotion. 1. Metabolic energy consumplion as a funClion of speed and body size in birds and mammals. J Exp Bioi. 97. 1-21. Taylor, C.R., Maloiy, G.M.O., Weibel. E.R. ttal. (1980) Design of the mammalian respiratory system. 3. Scaling maximum aerobic capacity to body mass: wild and domestic animals. Resp PJrysiol. 44.. 25H. Temeles. E.J. (1994) The role of neighbours in territorial systems: when arc they 'dear enemies?' Anim Behav, 47, 339-50. Templeton, A.R .. Routman, E. & Phillips. C.A. (1995) Separating population structure from population history: a cladistic analysis of the geographical distribution of mitochondrial DNA haplOlYpt~s in the tiger salamander. Ambystoma tigrinum. Genet;cs, 140. 767-82. Templeton. J.J. 11993) The use 0/ personal and public ;n/omlalion in!ltras;ngj1Ofks o/Europtan starlings. PhD thesis, Concordia University. TemplelOn, J.J. & Giraldeau. L.-A. (1995) Patch
440
REFERENCES
assessment in foraging flocks of European starlings: evidence for public information use. Btlrav Eco!. 6. 65-72. Templeton. J.J. & Giraldeau. L.-A. (1996) Vicarious sampling: the use of personal and public information by starlings foraging in a simple patchy environment. 8thav Erol Sodobiol. 38. 105-13. len Calc, C. (1985) On sex differences in sexual imprinting. Anim BeIJav, 33.1310-17.
Thisted. T.• S0rensen. N.S .• Wagner. E.G.H. & Gerdes. K. (1994) Mechanism of posl-segregational killing: Sok antisense RNA interacrs with Hok mRNA -ia its 5' -end single-stranded leader and competes with the 3'-end of Hok mRNA for binding to the 'wk translation initiation region. E.fBO J. 13. 1960-8. Thorne. B.L. & Carpenter. J.M. (1992) Phylogeoy of the Diclyoptera. System Bntomo/, 17.253-67. Thornhill. N. W. (cd.) (1993) The Natural History of
1nbretding and Outbreeding: Theorl'l;col and Empirical Perspectiws. University of Chicago Press. Chicago. Thornhill. R. (1976) Sexual selection and nuptial feeding behavior in Billocus apicaJis (Insecta: Mecoptera). Am No',r. 110. 529-48. Thornhill. R. (1980) Sexual selection in thc blacktipped hangingny. Sri Am. 242. 162-72. Thornhill. R. (1981) Panorpa (Mecoptera: Panorpidae) sl'orpionWes: systems for understanding resourcedefense polygyny and alternative male reproductive efforts. Ann Rev EcoJ Syst. 12. 355-86. Thornhill. R. (1983) Cryptic female choice and its implications in the scorpionny Harpovillacus n;grictps. Am Natur. 122. 765-88. Thornhill. R. (1992) Fenlale preference for the pheromone of males with low fluauatingasymmt:'try in the Japanese scorpionrJy (Panorpa japonica: Mecoplera). Behov Eco1. 3. 277-83. Thornhill. R. & Alcock. J. (1983) The Evolution of1nsut Mating Systems. Harvard University Press.. Cambridge. Massachusetts and London. Thorpe. W.H. (1956) Learning and Instinct in Animals. Methuen & Cu.. London. Tinbergen. J.M. (1981) Foraging decisions in starlings (Sturnus vulgaris L.). Ardea. 69. 1-67. linbergen. J.M. & BoerlijSl. M. (1990) Nestling weight and survival in individual great tits (Parus major). J Allim EcoJ. 59. 1113-27. Tinbergcn. J.M. & Daall. S. (1990) Family planning in the Greallit (Parus major): optimal clutch size as integration ofparenl and offspring fitness. Bdmviollr. 114. 161-90. Tinbergen. L. (1960) The natural control of insects in pine woods. I. Factors influencing the intensity of predation by songbirds. Arch Netrhmd Zool. 11.265343.
Tinbergen.. N. (1953) Social BeJurwouritt Animals. Butler and Tanner. London. Tinbergen, N. (1953) The Hern'g Gull's World. New Naturalist Series. Collins. London. Tinbergen. N. (1963) On aims and methods of ethology. Z Titrpsychol. 20, 410-33. Tinbergen. N. (1964) Social &Izaviour in Animals, 2nd cdn. Chapman & Hall. London. TIn bergen. N. (1972) Tht Animal in its World. Vol. I. .Fitld Studies. George Allen & Unwin. London. Tinhergen. N.. Brockhuysen. G.J.. Feekes. E. Houghlon. J.C.W.. Kruuk. H. & Szulc. E. (1963) Egg shell removal by (he black headed gull. LAms ridibu'dlls: a behaviour col11)xment of camouflage. Behaviour. 19.74-117. Tomlinson. l. & O·Oonald. P. (1989) The c04evolulioll of muhiple female mating preferences and prelem:d male characters: [he ·gene·ror-gene· hypmhcsis or sexual selection. J Tlleor Bioi. 139. 219-38. Tracy, N.D. & Seaman. J.W. Jr (1995) Properties of evolutionarily stable learning rules. J TI'lor 8iol. 177. 193-8. Trail. P.W. & Adams. E.S. (1989) Active mate choiceat rock-of-lhe·rock leks: ladies of sampling and comparison. Bellav Ecol Sodobiol. 25. 283-92. Trivers. R.l. (1971) The evolution of reciprocal ahruism. Quart Rev Bioi. 46.35-57. Trivers. R.t. (1972) Parental investmcl11 and sexual selection. In: Sexual Selection aud tlte [Hscem of Mall 1871-/97/ (ed. B. Campbell). pp. 136-79. Aldine. Chicago. Trivers. R.L. (1974) Parem-offspring conflict. Am Zool. 14.249-64. Trivers. R. (1985) Social Evolution. Benjaminl Cummings. Menlo Park, California. Trivets. R.l. & Hare. H. (1976) Haplodiploidy and the evolution of the social insects. Samet. 191, 24963. 'furner. G.F. & Robinson. R.L. (1992) Milinski's tit~for·lat hypothesis: do fish preft:'rentially inspeCl in pairs? Anim Bellav. 43. 677-9. lJltle. M.D. & Ryan. M.J. (1981) Bat predation and theevolution of frog vocalisatilms in the eotropies. Scimce. 214. 677-8. Ueno, T. (1994) Self-recognition by the paraSitic wasp ltop1ectis narallyae (Hymenoptera: Ichneumonidae). Oikos. 70. 333-9. Valone, T.J. (1989) Group foraging. public inform31illll and patch eSlimation. Oikos. 56. 357-63. Valone. T.J. (1991) Bayesian and prescienl aSSCSSI11t'nt: foraging with pre·harvest informalion. Allim lkl,m·. 41.569-77. Valone. TJ. (1992a) Information for patch a<;~sl1lel1l: a field investigation Wilh black-chinned humming-
REFERENCES birds. /khav Erol. 3. 211-22. Valone. TJ. (l992b) Patch estimation via memory windows and the eHeet of traveltime. J Thtor Bioi. 157.243-51. Valone. TJ. & Brown. J.S. (1989) Measuring patch assessment abilities of desen granivores. Ecology. 70. 1800-10. Valone. T.J. & Giraldeau. L.·A. (1993) Patch estimalion in group foragers: what information is used? Ani', 8ehav. 45. 721-8. Van dec Have, T.M .• Boomsma. J.J. & Menken. S.B.l. (1988) Sex-investment ratios and relatedness in the monogynous ant Lasius niger (L.). Eva/urian. 42. 160-72. Vander Meer. R.P. & Wojdk. D.K. (1982) Chemical mimicry in the myrmechophilus beelle Mym,ecaphodius txcQvarico/lis. Sdmet. 218. 806-8. Vander Wall. S.B. (1990) Food Hoarding In Animals. Chicago University Press. Chicago. van Dijk. B.A.• Boomsma. 0.1. & de Man. A.J.M. (1996) Blood group chimerism in multiple human births is not rare. Am J Med Genfl. 6' 264-8. Van Honk. CG.J.. Roseler. P.-F.. Velthuis. II.II.W. & Hoogeveen. J.e. (1981) Factors influencing (he egg laying of workers in a captive Bombus terrestris colony. Bella! Beol SodobioJ. 9. 9-14. Van Noordwijk. A.J., VanBalen. J.H. & Scharloo. W. (J 981) Genetic variation in the timing of repro~ duction in the Great til. Oecologia. 49. 158-66. van Rhijn. J.G. (1985) A scenario fur the evolution of soci(ll organisation in ruHs Philomachus pllgnax and other Charadriiform species. Ardea, 73. 2537. van Rhijn, J.G. (1990) Unidireclionality in the phylo~ geny of social organisation. with special reference to birds. Behaviour. 115, 153-74. Vega· Redondo. F. & Hasson. O. (1993) A gametheoretic model of predator-prey signalling. JTlleor Bioi. 162. 309-19. Vehrencamp, S.l. (1979) The roles of individual, kin. and group selection in the evolution of sociality. In: Handbook of Behavioral Neurobiology Vol. 3 (cds P. Marler&J.G. Vandenbergh). pp. 351-79. Plenum Press. New York. Vehrencamp. S.L. (J 983a) A model for the evolution 01 despotic versus egalitarian societies. Anim BeJrav. 31.667-82. Vehrencamp. S.l. (1983b) Optimal degree of skew in cooperative societies. Am ZOOI. 23, 327-35. Verhulst, S. (1995) R~prodllcti!e decisions ill Grtat tits. A' optimality approach. PhD thesis. University of Groningen. Groningen. Vickery, W.l.. Giraldeau. L.·A.. Templeton. J.J .. Kramer. D.L. & Chapman. C.A. (1991) Producers.
441
scroungers. and group foraging. Am Natur, 137.84763. Vines. G. (1980) Spatial consequences of aggre'Ssive behaviour in flocks of oystercatchers Haematopus ostraltgus. Anim &hav. 28. 1175-83. Visscher. P.K. (1986) Kinshipdiscrimination in Queen rearing by honey bees (Apis me/lifero). /khdv Ecol Sociobiol. 18. 453-60. Visscher. P.K. (1989) A quantitative study of worker reprodudion in honey bee colonies. Behav Eco! Sodobiol. 2S. 247-54. Visscher. P.K. & Dukas. R. (1995) Honeybees recoKnize developl1lcni of nestmates' ovaries. Anim BehQ·. 49. 542-4. Vogel. S. (1988) Lift's Devices. Princeton University Press. Princeten. von Boehmer. H. & Kisielow. P. (1991) How the immune system learns about self. Sci Am. 265(4). 7481. von Frisch, K. (1967) The Dana Langua!Jeand Orientation o/Bees. Harvard University Press, Cambridge, MA. Vos. D.R. (1994) Sex recognition in zebra finch males results from e.:lrly experiences. Behaviour, 128. 114. VO'S. D.R. (1995a) The role of sexual imprinting for sex recognition in zebra finches: a difference be~ tween males and females. Anim Behav, SO. 645-53. Vos. D.R. (1995b) Sexual imprinting in zebra finch females: do females develop a preference for males that look like their father? Ethology. 99. 252-62. Waage. J.K. (1979) Dual function of the damselny penis: Spero1 removal and transfer. Science Wash, 203, 916-18. Waage. J.K. (1988) Confusion over residence and the escalation of damselfly lerritorial disputes. Anim /khav. 36. 586-95. de Waal. F. (1982) Chimpanzee Politics: Power and ~x Among Apes. Harper and Row, New York. de Waal. F.B.M. & Luttrell. L.M. (1988) Mechanisms of social redprocity in three primate species: sym metrical relatiOnship characteristics or rognil ion? Behdv Erol Soc/obiol. 9. 101-18. Waddell, D.R. (1982) A predatory slime mould. Nature, 298. 464--6. Waldman. B. (1981) Sibling recognition in toad tadpoles: the role of experience. Z Tferpsychol. 56. 341-58. Waldman. B. (1986) Chemical ecology of kin recog· nition in anuran amphibians. In: Chemical Signllls in Vmebrales. VolA (eds D. Duval. D. MolJer·Schwarze & R.M. Silverstein). pp. 225-42. Plenum. New York. Waldman. B. (1987) Mechanisms of kin recognition. J Tlteo, Bioi. 128. 159--85. Waldman. B. (1991) Kin recognition in amphibIans. In: Kin Rtcognition (ed. P. Hepper). pp. 162-219.
442
REFERENCES
Cambridge University Press. Cambridge. Waldman. B.. Frumhorr. P.e. & Sherman. P.W. (1988) Problems of kin recognition. Trmds Era! Eval, J. 813.
Waser, N.M. & Price. M.V. (1994) Crossing-distance effecls in Delphinium nilson;i: outbreeding and inbreeding depression in progeny filness. Evolution.
Waldman. B.. Rice. J.E. & Honeycutt. R.L. (1992) Kin
Waser. P.M .• Austad. S.N. & Kena. B. (1986) When should animals tolerate inbreeding? Am Natur. 128.
recognition and incest avoidance in lOilds. Am Zool. 32. 18-30.
Walkowiak. W. (1988) Neuroethology of anuran call recognition. In: The Evolution of the Amphibian Audilory System (eds B. Fritzsch. M.J. Ryan. w. Wilczynski. T.E. HerheringlOn & W. Walkowiak), pp. 485-509. John Wiley & Sons. ew York. Wallace. A.R. (1889) Darwinism. 2nd ron. MacMillan.
london. Wahe'. J.R. (1990) Red-cockaded woodpeckers: a 'primitive' cooperative breeder. In: Cooperative Bruding in Birds: l.ong-ltml Studies of Ecology and Bthavior (eds PB. Stacey & W.O. Koenig). pp. 67102. Cambridge University Press. Cambridge. Walters. J.R. & Seyranh. R.M. (1987) Connict and cooperation. In: Primate Societies (eds B.B. Smuts. D.L. Cheney. R.M. Seyfanh, R. W. Wrangham &T.T. Struhsaker). pp. 306-17. University of Chicago Press. Chicago. Waite'. J.R .. Copeyon. e.K. & Caner. I.J.H. (1992) T~st of the ecological basis of cooperative breeding in Red-cockaded WOOdpecker. Auk, 109.90-7. Ward, P.r. (1986) Prey availability increases less quickly than nest size in the social spider StrgodypJws mimosawm. Behaviour, 97. 213-25.
Warner. R.R. (1987) Fcmale choice of sitcs versus mates in a coral reef fish. Thala.~oma bJfasciatum. Anim Behav. lS. 1470-8. Warner, R.R.. Shapiro. D. Y.• Marcanato. A. & Petersen. C.W. (1995) Sexual conflict: males with highest mating success convey the lowest fertilization bcnefits to females. PrO( Roy Soc Lond B. 262. 1359. Warner. R.R.. Wernerus. E. Lejeune. P. & van den Bcrghe. E. (1995) Dynamics of lemale choice for parental care in a fish species where care is facultative. Bthav £(01,6. 73-81. Waser. N.M. (1993) Population slrueture. optimal ombreeding. and assonative maling in angiosperms. In: The Natural History of Inbruding and OutbrudinfJ (cd. N.W. Thornhill). pp. 17}--99. Unive'ily of Chicago Press. Chicago. Waser. N.M. & Price. M. V. (1985) Reciprocal Iransplllni experiments with Delphinium nt/sonii (Ranunculaceae): evidence for local adaplalion. Am J Bol. 72. J 726-32. Wascr. N.M. & Price. M.V. (1993) Crossing distanet: eflects on prezygotic performance in planls: an ar>:ument for female choice. Oikos. 68. 303-8.
48.842-52.
529-37.
Wasser. S.K. (1995) COSIS of conception in baboons.
Nature Lond, 376. 219-20. Watson. A. & Miller. C.R. (1971) Territory size and aggression in a fluctuating red grouse populalion. J Anim £eol. 40. 367-83. Watson. PJ. & Thornhill. R. (1994) Fluctuating asymmetry and sexual selection. Trmds £COl £1.'01.9.215. Watts. B.D. (1990) Cover use and predator-relaled mortality in song and savannah sparrows. Auk. 107. 775-8.
WaUlers. L. & Dhondt. A.A. (1989) Body weight. longevilY and reproductive success in red squirrels (Sciurus vulgaris). J Anim Eeel. S8. 637-51. Weary. D.M.. Guilford. T.e. & Weisman. R.G. (1993) A product Or discriminative learning may lead 10 female preferences for elaborale males. Evolution. 47.333-6.
Weathers. W. W. (1980) Seasonal and geographic variation in avian standard metabolic rate. Pro, 1111 Ornithol Congress Berl. 283-6. Wedekind. c.. Seebeck, T., Bcnens. F. & Paepke. A.J. (1995) MHC-depcndent mate preferences in humans. Proc Roy Soc Lond B. 261. 245-9. Wedell. N. (1994) Variation in nuplial gift quality in bush crickets (Onhoptera: Tenigoniidea). Bellav £rol. 5. 418-25. Wehner. R. (1981) Spatial vision in arlhropods. In: Handbook of St/l.Sory Physiology. Vol. VII/6C (ed. H. Autrum). PI'. 287-616. Springer-Verlag. Berlin. New York. Wehncr. R. (1987) 'Matched filters' - neural models of the external world. J Comp Physiol A. 161. 51 131. Wehner. R. (1994) The polarization-vision project: championing organismic biology. In: Neural Basis of Bthavioural Adaptation (eds K. Schild berger & N. Elsner). pp. 103-43. G. Fischer. S'utlgan. New York. Wehner. R. & Menzel. R. (1990) Do insects have cognitive maps? Ann RnJNeurosd. 13.403-14. Wehner. R. & Wehner. S. (1990) Insect navigation: use of maps or Ariadne's thread? Ethol Ecol Evol. 2. 27-48.
Wehner. R. & Srinivasan. M.V. (1984) The world as the insect sees it. In: In~a Communication (ed. T. Lewis), pp. 29-47. Academic Press. London. Wehner. R.. Michel. B. & Antonsen. P. (1996) Vhual
REFERENCES
navigation in insects: coupling of ~guccntric and gt:ncentric information. J Exp Bioi. 199. 129-40. Weiner. J. (1992) Physiological limits 10 suslainable t.'nergy budgets in birds and mammals: ecological implications. Trmds Ecol Evol. 7. 384-8. Welham. C.V.J. & Vdenberg. R.C. (1993) Elfidencymaximizing flight speeds in parent black terns. Erology. 74. 1893-190 I. Wenl. F.W. (1968) The size of man. Am Sa'. 56. 40013.
West, M.J. & King. A.P. (1988) Female visual displays aHect the development of male song in the cowbird. Natllre. 334, 244-6. West·Eberhard. M.J. (1975) The evolution uf social hehavior by kin selection. Quan Rev Bioi, 50. 1-33. West· Eberhard. M.J. (1979) Sexual seleccion. social competition and t'volution. Proc Am Philosoph Soc. 123.222-34.
West·Eberhard. MJ. (1989) Phenotypic plaslicity and the origins of diversity. Ann Rev Ecol Syst. 20. 24978.
Westneat. D.F. (1987) EXlra·pair fertilizations in a predominantly monogamous bird: genetic evidence. Anim &hav. 35. 877-86. West neat. D.E & Sargent. R.C. (1996) Sex and parenting: the erreets of sexual ronflja and parentage on parental strategies. Trmds Hcol Evol. II, 87-91. Westneat. D.F. & Sherman. P.W. (1993) Parentage and the evolution of parental behavior. Bthav Eco!, 4. 66-77.
WeStneal. D.E & WebSter. M.S. (1994) Molecular analysis of kinship in birds: interesting questions and useful techniques. In: Molfcular Ecology and Evolu,;on: ApproachN and Applications (eds B. Schierwater, B. Streit. G.P. Wagner & R. DeSalle). pp. 91-126. Birhauser Verlag. Basel. WeSllleal. D.E. Clark. A.B. & Rambo. K.C. (1995) Within·brood panerns of paternity and paternal behaviOr in red-winged blackbirds. Belrav Etol Sociobiol, 37, 349-56. W~lIon, J.H. & Parkin, D.T. (1991) An association between feniJity and cuckoldry in the house sparrow Passer domes';cus. Proc Roy Soc Lond 8.245. 227-33. Welton. J.H .. Carter. R.E .• Parkin. D.T. & Wallers. D. (1987) Demographicstudy of a wild house sparrow population by DNA fingerprinting. Nature Lond. 327. 147-9.
H. & Hope. P.L. (199)) Sperm whalers off the Galapagos Islands and in the Western North Pacific. 1830-1850: ideal free whalers. Ethol
Whilt~head.
Sociobiol. 12. 147-61. Whiten. A. & Ham. R. (1992) On the nature and evolution of imitation in lhe animal kingdom: reappraisal of a century of research. AdvStudy &hav.
443
21.239-83.
Whi.lock. M.C. (1995) Variance-induced peak shifts. EvollUion. 49. 252-9. Whillingham. L.A. & Lifjeld. J.T. (1995) High paternal investment in unrelated young: extra·pair paternity and male paremal care in house marlins. Beha' Ecol Sociobiol. 37. 103-8. Wiegmann. D.O.• Real. L.A.. Capone. T.A. & Ellner. S. (1996) Some distinguishing features of models of search behavior and mate choice. Am Natur, 147. 188-204.
Wiggins. D.A. & Morris, R.D. (1986) Criteria for female choict'of mates: courtship feeding and paternal care in the common tern. Am Nalllr. 128. 126-9. Wil(()ve. D.S. (1985) Nest predation in forest tracts and the decline of migratory songbirds. Ecology. 66. 1211-14.
Wildhaber. M.L.. Green. R.E & Crowder. L.B. (1994) Bluegill continuously update patch giving up limes based on foragll1g experience. Anim /Nltav, 47. SOI13.
Wiley. R.H. (1983) The evolution of communication: information and manipulation. In: Communication (eds T.R. Halliday & P.J.B. Sialer). pp. 82-113. Blackwell Scienlific Publications. Oxford. Wiley. R.t!. (1991) Associations of song propenies with habitats for territorial osonl? birds of eastern North America. Am Natllr. 138. 973-93. Wiley. R.t!. (1994) Errors. exaggeration and deception in animal rommunication.ln: Behavioural Mechanisms in Evolutionary Ecology (ed. L.A. Real). pp. 157-89. University of Chicago Press. London. Wiley, R.H. & Richards. D.G. (1978) Physical constraints on acoustic communication in the atmosphere: implications for the evolution of animal vocalizalion.
lithav £col Sodabiol. 3. 69-94. Wilkinsun. G.S. & Reillo. P.R. (1994) Female choice response to artifidal selection on an exaggerated male trait in a stalk-eyed fly. PriX Roy Soc Lond 8. 255. 1-6.
Williams. A.F. & Barclay. A.N. (1988) The immuno· globulin superfamily - domains for cell surface recognition. Ann Rev Immunol, 6. 381-405. Williams. D. & Mcintyre. P. (1980) The prindpal eyes of a jumping spider have a telephoto component. Nature. 288, 578-80. Williams. G.c. ( 1966a) Adaptation and Natural Sele(non. a Critique of Some Curre', Evolutionary Thougltt.
Princetoll University Press, Princeton. Williams. C.C. (1966b) Natural selection. the COS(S of reproduction. and a refinement of Lack's principle. Am Natur. 100.687-90. Williams. G.c. (1975) Sex and Evolution. Princeton University Press. Princeton.
444
REFERENCES
Williams, J.B. (1987) Field metabolism and food consumption of savannah sparrows during the breeding season. Auk. 104.277-89. Williams. T.e. & Williams. J.M. (1978) Orkntation of transatlantic migrants. In: Animal Migrarion. Nav;9atiofl, and NOtHing (cds K. Schmidt-Koenig & W.T. Keelon). pp. 239-51. Springer-Verlag. Berlin. Wilson. D.S. & Sober. E. (J 994) Reintroducing group sdeaion 10 the human behavioral scicmct's. ]klrav Brain Sci. 17. 585-654. Wilson. E.O. (1971) TIr~ Justa 5«iil;('S. Harvard
niversity Press. Cambridge. Massachusells. Wilson. E.O. (1975) Sodobiology. Harvard
niversity
Press. Cambridge. Mass. Wilson. K. & Galehnuse. A.G. (19931 Seasonal and geographical variatiun in lhe migratory pOlential of outbreak populations of the African armyworm moth. Spodoplfra extmpto. J Anini Erol. 62. 169-81. Wingfield. J.e.. Schwabb. H. & Mattocks. P.W, Jr (1990) Endocrine mechanisms of migratiull. In: Bird Migration. Plrysiology and Ecology .ed. E. Gwinner). pp. 232-56. Springer VerJag. Berlin. Wingfield, J.e.. Suydam. R. & Hunt K. (1994b) The adrenocortical responses to Stress in snow buntings (Plmrophmax nivalis) and lapland longspurs (Calm rillS lapPOl1iclls) at Barrow. Alaska. Comp Biorhem Pllysiol C. 108,299-306. Wingfield, J.C.. Vleck. C.M. & Moore. M.C. (1992) Seasonal changes of the adrenocortical response to S1ress in birds of the Sonoran desert. J Bxp Zool. 264. 419-28. Wingfield. J.C .. Deviche. P.. Sharbaugh. S. et 01. (1994a) Seasonal changes of the adrenocmtical responses to stress in red polls. Acamllis flammia. in Alaska. J Exp laal, 270, 372-80. Winn, H.E. (1958) The comparative ecology and r{'productive behaviour of fourteen species of darter (Percidae). Ecol Monogr. 28. 155-91. Winston. M.L. (1987) Tile Biology of Ihi J-ImtiY Ike. Harvard University Press. Cambridge. Massachusetts. Winter. U. & Buschinger. A. (1983) The reproductive biology of a slavemaker ant. Epimynna rawJUxi. and a degenerate slavemaker. E. kraussti (Hymenoptera: Formiddae). £momol GenfTal. 9. 1-15. Winterltalder. B. (1981) Foraging strategies in the boreal forest: an analysis of Cree hunting and gathering. In: Humer-Galhertr Foragi'9 Sirategirs (eds B. Winterhalder & E.A. Smith). pp. 66-98. University Chicago Press. Chicago. Wishart. G.J. (1987) Regulation of the length of the fertile period in the domestic fowl by numbers of oviductal spermatozoa as reflected by those trapped in laid eggs. J Reprod FiTt. 80. 493-8. Winer. M.S. & Cuthill. I.E. (1993) The ecological costs 4
of avian fat storage. Phil Trans Roy Soc Lond B. 340. 73-90. Witter, M.S. & Swaddle. J.P. (1995) Dominance. competilion and energy reserves in the European ~tarling. Sturnus viligaris. Brlwv E(OI. 6. 343-8. Willer. M.S .. Cuthill. I.e. & !lonser. R.H.e. (1994) Experimental investigations of mass-dependent predali<Jn risk in the European starling. Stlimus vulgaris. Anim Bthav. 48. 201-22. Willer, M.S.. Swaddle, J.P. & Cuthill. I.e. (1995) Pcritldic food availability and strategic regulatiun of body mass in the European starling. SllImus vulgaris. FiOla Ecol. 9. 568-74. Wnodhcad. A.P. (1985) Sperm mixing in the cockroach Diplopti'ra punctala. Evolulion. 39. 159-64. Woolfenden, G. & Fitzpatrick. J. (1984) Th' Florida Scrub J,~y: Drmograplzy of a Coopuatiw Britdi'9 Bird. Princcton University Prl'ss. Princeton, NJ. Woyciechowski. M. & Lomnicki. A. (1987) Multiple mating of queens and the sterility of workers among cusacial Hymenoptera. J Tlrlor Bioi. 128. 317-27. Wright. R. (1994) Tlrt Moral Animal: Ilrt Ntw SdmCf 0/ Evolmiollary Psychology. Pantheon Books. New York. Wright. S. (1978) Evollilion and the Genetics of Populations. Vol. 4. Variability within dnd among Populations. University of Chicago Press. Chicago. Wijrsig. B. (1989) Cetaceans. Science. 244. I 55
REFERENCES lahavi. A. (I975) Mate selection - a seledion for a handicap. J Thtor 8iol. 53. 205-14. Zahavi. A. (1977a) The cost of honesty (further remarks on the handicap principle). J Tlffor BioI. 67. 60~-5. Zahavi, A. (1977b) Reliability in communication systems and the evolution of altruism. In: Evolutionary Ecology (eds B. Stonehouse & C. Perrins). pp. 2539. Macmillan. London. lahavi, A. (1987) The theory of signal selection and some of itS implications. In: Inttrnational Symposium o[Biological Evolution (ed. V.P. Delfino). pp. 305-27. Adriatica Edilrice, Sari. lahavi. A. (1990) Arabian babblers: the quest for social status in a roopt>rative breeder. In: Cooptrativt Breeding in Birds: LA.'9-tenn Studies o[ Ecology and Behavior (cds P.S. Stacey & W.O. Koenig). pp. 10330. Cambridge University Press, Cambridgt.~. Zhivotovsky. L.A.. Feldman. M. W. & Christiansen. EB_ (1994) Evolution of recombination among multiple sekcted loci: a generalized reduction principle. Pr()( Nor Acod Sci USA. 91. 1079-83.
445
Zimmerer, E.J. & Kallman. K.D. (1989) Genetic basis for alternative reproductive tactics in the PYRmy swordtail. X;plt(lpllorus nigrens;s. £IOlul;o'. 43, 12981307. Ziolko. M. & Kozlowski. J. (1983) Evolution of body size: an optimization model. Mathemal Biosci. 64. 127-43. Zuk. M.. Ligon. J.D. & Thornhill. R. (1992) Effeas of experimental manipulation of male secondary sexual characters on female mate preference in red jungle fowl. Anim Bihav, 44. 999-1006. Zuk. M.. Simmons. l.W. & Cupp. l. (1993) Calling characteristics of parasitized and unparasitlzed populations of the field cricket Tf/fogrylllls oceamcus. Beltav E(OI Sociobiol. ]}, 339-43. Zuk. M.. Thornhill. R.. Ligon. J.D. er oJ. (1990) The role or male ornaments and couflship behaviour in female mate choice of red jungle fowl. Am Natur. 136.459-73. Zyskind. J.W. & Smith. D. W. (1992) DNA replication. the bacterial cell cycle. and cell growth. Cell, 69. 58.
Index
Note: page numbers in italics refer to figures. those in bold to tables. acceptance errors 72. 92-3 acceptance thresholds context.dependent (shifling) 87-8,89-90,95-6 optimal 86, 88. 90. 91-2 Acdpittr nisus (sparrowhawk) /10. III acorn woodpeckers su Mtlanerpes
formidvorus
acoustic communication 30 acoustic signalling 159-60. 162-4
Acrocepha/us arundinaceus
(warbler, great reed) 78
Acroaplralus sichel/ens;s (warbler.
Seychelles) 233-4, 239 aClion. recognition component 71-2,86-92 adaptive behaviour 16 additive genetic variance. in counship 368-9 adrt:nal stress hormones 119 Age/aius phoetJiuus (blackbird. red-winged) 59,93, /73 a~ression, sexually-related 237. 251,252 air oscillation. near·field 30
A/auda arvtns;s (skylark) 170
A/taura latham; (brush turkey) 90 alerting components, assertion display 158
tiger) 360
Antbystoma tigrimlln ntbulosllm 70. 71 animals. as dccision·makers 19 anisogamy 122, 123, 124
Anolis auralus 156-8 AI/olis spp. (lizard) 343, 348 Allser anser (goose, domestic) 127 Anser cynoides (goose. Chinese) 127 anthropomorphism II AplreJoronta cotrulesans (jay. scrub) 57, 58, 239 Aphelocoma ul'ramarina (jay, Mexican) 57, 58, 235 Apis mtllifira (bee, honey) 30, 83, 99, 164-5, 227 apostatic prey selection OJ 1-2 Archilocus alexandri (hummingbird, black-chinned) 47 area copying 59-60 'annpit eHeCl' 83 assertion display 156-8 assortative mating 190.368 auditory communications 24,25 Augochlorella striall/ (bee, halietinc) 224 Bacillus slIbti/is 293 'back-up signal' hypo'hesis 175 Bayesian models 45-8 behavioural models 379-80, 386, 389 ideal free-distribution models 379-80, 386 behavioural roulines, investigations 117-18 behaviour anti-predatory 109-10, III complex 15-16 constraints imposed by physical environment 204.26 and ecology 5-6 economic models of 6-7, 17 eJrecls of last malt precedence mechanisms 139-42 evolution or in small isolated populalions 368-72
female. and sperm competition 141 interactions with population structure 363-8 males seeking exuapair copulalions 141-2 selfish 266, 267 shon-term consequences of 17 and lirelime reproductive success 111-18 Vigilance behaviour 110 set also social behaviour behaviour copying 61-2,66 behavioural dedsion rules, population consequences of 393 behavioural ecological models 394 Belding's ground squirrd ste
SpertnopJrilw beldingi &mbidum vitlalUm (snail) 357.
361 bird navigation 31-3 blue jays 281-3 body mass changing prior to migration 104-5 as function of age 316.316 increased, and flight 107-8 and photoperiod 102-3 power functions of 318 prOOuclion and age of maturity 318-19.3/9. 320,31/ body size constraints due (0 26-30 and sound-source detection 30 Bombin. bombina (load) 357, 368 Bombino bombinolvaritgola (load, fire-bellied) 355 Bombus (bee, bumble) 226 Bolryllus schlosseri (urochordatt'1 82, 29S Brama btmida bernicla (goose, dark-bellied brent) 395 breeders dominant, harassment by 248
447
448
INDEX tracing of 337-8
early breeders 324. 332 replacement breeders 241-2. 252 breeding
Chamdrius aJexandn'nus (plover.
Kemish) 381
communal 216
delayed 231. 389-90 multiple 216 optimal timing 325 shared 239-40 ue also cooperative breeding breeding season, measurement density dependence in 389-90
breeding season habitats.
or
nOI
easily evaluated 380-1 breeding success 381 breeding units. closed. and gene flow 362 brood size 330-1 brood size manipulations 322-4 buffer eUecl 389
Bullo galapagoensis (hawk. Galapagos) 245
Bufo anrtriCQflUS (toad, American)
74.194
Callarius lapponicus (longspur. Lapland) 99. /73
Cilkarius picrus J 73 Carnal/amis colt; (nematode) 176
Campylorhynchus nuchalis (wren. stripe-backed) 237
Cardutlis citloris (greenfinch) 60, 103
CarJuus spp. (thistle) 366
Carpodacus mexiconus (house finch) 166 Calliglyphis ants. navigation
by 33--6.40 CatQg/yphis fortis (desen ani) 34. 38 categorical variables. methods for 345--;; causation 4, 11 Cebus apdfa (monkey.
capuchin) 66 cell-surface recognition 294 cemral place foragers, navigation of 33--;; Ctrat;tis capitQIQ (fruit ny, mediterranean) 141 CUGJpithtcus atthiops (monkey. vervel) 79
chain procedures. maintaining cultural traditions 66 challenge display 156 character change 336-40 combined with tree Structure 347-9 correlated 340-7 Iypes of 341 charaet~r
evolution
rates of 339-40
cheating 149 chick begging 167-9. l71. 172. 177 chick position, and food allocation 177-8 chimerism 295 Chirox;pltia linear;s (manakin. long-tailed) 163 Chlidonias nigra (tern. black) 99 choice. self·enforcing 190
Chortllippus paralltills (grasshopper, meadow) 354-5. 368 chorusing. synchronous 163-4 chromosomes bac.erial 291.292.303.304 sex 300-1 C/amator glandarius (cuckoo. great spotted) 171 clutch laying dales. Ii£c·history consequences of changes 327-8 clutch size. variation in 330. JJ/ dUleh-date combinations. optimal 326 coercion 273-4. 275 Co/lamia grandiflora 75 Colobopsis tz;ppotliaiS 22J colonies. cusocial and parasorial 224-5 colonization changes in selection pressures 370-1 of new territory 355 progressive 353 repeated. and role of selenion 371-2 colony fission 223 Columba livia (pigeon) 52.60 Columba palambus (woodpigeon) 60 Co{wnbina i'ca (dovt,. inca) 47 communication signals 149 competition 254 among signallers during communication 177-8 compelitive ability. in foraging 382--;; competitive relationships 254-65 ahernatives to dominance 263-4 benefits and costs 01 dominance 262-3 dominance hierarchies 2S8-62 pairwise contests 255-8 computational capabilities, constraims sct by 30-40 conniet and cooperation 7. 229 in family groupings 9. 237-41
intra genomic 28S. 290 organellar/nuclear genes 299300 over who reproduces 241-6, 252 parem-offspring l02 sexual 142-4 conflidS of interest evolutionary 240 genetic 229 conjugation 291. 292
COl1ocephalus nigropleunun
(ka'ydid) 78 conservation 308, 373 consortship 142 conspicuousness 156-7. 158. 159.193 constrainlS 313-14 cooperation 254. 265-83, 284 coerced 273-4. 275 and conrIiet 7. 229 mlllualism in 152 short· and long· term 271-2 producers vs scroungers 2723. 275 reciprocily 265-70 cooperative breeding 234. 249. 251 changing scope of research 228-30 preferential assistance to closest relatives 235 coordination 284 Cope's law 341 copula duration. male dungflics 132-4 copulation. females wilh muJliplt.' males. benefits 125, 125-6
coreplicons 290. 303-4 rorticosteron~
119
Corvus monedu/a (jackdaw) 66
courtship, additive genetic variance in 368-9 courtship trembling 16 t
Crmicich/a alta 160 Crocma crocUla (hyena. spolled)
61 CrolopJraga sulciroslris (an is. gruove-billed) 245 cryptic female choice 141 cryptic prey. deleCtion of 51-2 Cryptocerclis (wood roach) 214 cub.rearing. communal 277-8 Cuculus canorus (cuckoo) 4. 85 cues 71 chcmical 74 kin recognition 74-7 optimal cue systC'm 77 phenotypic and nun· phenUlypic 88 predictive. shun· and l(ln~· term 101
INDEX recognition cues 72-4 temporal 88 cuhural diffusion, regulation of 65-6 cultural transmission 63-6, 67 population approach 63.64 Cyrlodiopsis dolmanni (stalk-eyed fiy) 80 cytoplasmic commons. management of 290 cytoplasmic conflict, nuclearenforced suppression of 300 D-present cues 72.74.75.81 daily energy expenditure (DEE) 322-3 Daphnia (water flea) 384 dawn chorus I 16 dt'ceit. in signalling 166-71 decision rules 313 defence mechanisms 5 Delichon urbiea (house martin) 93 Delphinium IItlsonii (larkspur) 71. 367 deme< 351 Of'droica coro'ala (warbler) 357 density dependence behavioural approach to 37981 dctcrmining strength of 37890 importanCl' of 374-7. 378 and macroptery 365 measurement of in lhe breeding season 38990 in the non-breeding season 381-8 determinism 10 development 4
DictyoSlt/ium cal/tatum 294 Dietyos/t1ium discoideum 294-5 dimensionality theories 321-2 dioedous species 121 disassonative mating 185. 196 discrimination, absent. effects of 93 dispersal. delayed 230-1.243 parental influence 243 dispersal variation 364-6 display 156-8.169-70.180 and the physical environment 158-60 visual 159. 162 dizygotic twins. chimerism between 295 DNA, compartmemalized in vertebratt.~s 296-7 DNA markers 362-3 00110' law 341 dominance benefits and costs of 262-3
carpclller bee pairs 212-13 and mate rcplacemellt conflict 242 and models of reproductive skew 264-5 ownership an alternative 263-4 and susceptibility to interference 383-4. 385-6 Set also soda I dominance dominance hierarchies 258--62. 273 dominan e relalionships 255--6 dominance variance 369 Drosophilo (fruil fly) 30. 144. 351.353.359.369 mating in 144
Drosophila bi/urea 140 Drosophila dif!trtns 370 Drosophila htteroneura 370 Drosophila melanogaster 371 Drosophila planitibia 370 Drosophila pseudoobscura 357 Drosophila silws/ris 370 Drosophila willis/oni 357 dungfly, yellow stt Scoloph'go stereoraria dynamic programming 17.114, 115-16. 118
E-vector patterns 34. 35 early experience (dominance), importance of 256 ecological constraints 248. 249 and formation of family groups 230-4 and reproductive sharing 245 severe. effect of 247 and skew theory 215-18 ecological conslraints theory 228-9 ecological diversification 349 ecological models. dilemma of 394 ecology, and behaviour 5-6 ecomorphs, a classificatory 1001 348 ecosystems. latitudinal shift in 392 Ectatomma midum (ant, neotropical) 73 effective population size 362-3 egg size. and post-hatching survival 329-3 ejaculation, mulliple 140 energetic efficiency 19-20 energy. constraints on acquisition and spending 99-100 energy allocation 322-4 plasticity of 324 energy efficiency 99 energy reserves I 19 and avian song 115-16
449
bene filS of 100-4 costs of I04-1 I environmental change 308, 392 environmental constraillls, adaptations to 20-1 Eopsaltria georgiana (robin, whilebreasted) 242
Epimyrma krallSSti 223 epislasis 369 escape performance, and increased body mass 109.... 10
Escherichia coli 295
eukaryotic cell cycle, replicatil1n in 296 Eulampms tympa'um (water skink) 110
Ellmomota superci/iosa 79 Eupltaes jacksoni (Jackson's widowbird) 158. 175 eusocial insecls 204, 205-6 ('usodality 151. 203 Hymenopteran. and the haplodiploidy hYPOlhe
(ESS') 7-8 cooperation as 274 stable coexistence of male and female 123 eVOlutionary change. models of 343-4 evolutionary history 4 evolutionary inenia. rod pigments 21-2 eVOlutionary rates. m('asurem~nt of 339-40 evolutionary social theory, unified 249-51 eXlrapair copulation 137-8, 1....0, 143-4 sought by male< 141-2 exrrapair panners, in family groups 236 eyt'S. compound and single-lens 15.27-9
Falco columbarius (merlin) J 70 Falco tinnultculus (kestrel) 98. J22. 326 familieslfamily group(ing)s 251. 350-1 changing composition, cauSing conflict 237-41 connicts in 9 delinitions 229 extrapair partners 236 formation of and dissolution 231-4,24; economic model of 231 multigencrational 230-1, 215
450
INDEX
S1able, myth of 246-9 unslClble 231 su also replacemC'nt families; sodal groups family resources. inheritance of 234 fasling. enforced 100-1 fat storage 119 avian, and feeding interruptions 101-3 costso[ 106-11 dnd social dominance 104 ferundity effeds. consequence of mate choice 184-6. 201
f«dback 367 negative 380 ret'ding acquisition/storage COStS t 05--6 dnd energy stofilge I 19-20 patterns 97 feeding areas, evaluation of easier 380-1 feeding interruptions 101-3 female mating preferences 17982, 282 direct selection on 184-90 null model for 182-4 females masculinization 263 and offspring paternity 141 Ficedula hYPO/fUca (flycatcher, pied) 77 fighting ability and skew 218 and sodal·dominancc 246 Fisherian runaway evolution 175 fitness 184.312-13. J/4 consequences of crosses 10 male parents 367 COSIS and benefits 333 direct and indirect 9. 150 and foraging efficiency 384-5 night and increased body mass 107-8 speed of. COslS and benefits 98-9 fluctuating asymmelry (FA) 199 indicating genetic Quality 195 food acquisition see feeding; foraging food allocation. and chick begging 167-9. 171. 172, 177,178 food caching coSIS of storing energy 105-6 use of spatial memory for 53-9 food choice. and obj(ct copying 60-1 food st~aling (kl~ptoparasilism) 383-4 foraging foraging options 112-14
models 98-9 pr~dation risk 105 social. use of public information 50-l foraging behaviour. and food availabilily 101 foraging decisions. energetic gain. rate of (t) 98 foraging efficiency 134. 382, 383, 387-8 individual efficiency in 384-5 foraging theory 42-3 ability to measure and remember time 43-4
Formica pOdzoliea 222 Formica truncorum (am. wood)
22J. 225 founder events 355. 370 role of 368, 369 frequency filtering 24 frog. tungara su Physalatmus
pustulosus Fulica atra (COOl. European) 326.
J28 function 4. II
Gallinaso mtdia (snipe. great) 90 Gallus gallus 79 Gallus gallus domtstieus (chicken. domestiC) 66
Gallus gallus spadiceus Ounglefowl.
Burmese) 59 gamete competition 122
Gammarus /awrendanus 47 Gonllnarus pu/tX 142 Gasterosreus aeu/talUS 372 Gastrophryne carolilltnsis (toad.
narrow·mouthed) J 62-3
Gasrrophryne olivacta 162-3
gene connicts 152 gene flow 332, 333, 358-9 disrupted 347 limit on 361 non-random 363-4 and population structure 361-2 and population subdivision 362-3 genes 126 autosomal 30 I chromosomal 291 dominant. for sibling care 21 I high-viability 78 material and informational 286,287 nuclear and organellar 299300 sell ish 82, 152 social gene 284-304 strategic 286. 287-9 as strategists 285 successful. phenotypic effects of 286
variant 286-7 su also good g~nes genetic correlation and good genes hypothesis 193-5 trait and preference 191-2. 201-2 genetic diversity. loss of 355 gene'ic drift 189.356,358-9 genetic endogenous labels 74-5, 77 genetic interaction 288-9 genelic markers 351 genetic variation 35 I. 371. 374
Chorthippus paral/elus 354-5
genomes, relugial 354-5 genomic imprinting 152. 302-3 genotypes 350 genotypic variation 350 glacials and interglacials 353-4 good genes 126.180,198,199 hypOlheses '92-7 selection for 197.201 good genes-parasite hypothesis 185 gr~en beard effeas 288. 292. 294,298 group hunting 276-7 group selection 8-9 group territoriality 278-9 group-living 244. 254, 272 promoting socialleaming 61 growlh. and maturation 314-17 Grusgrns (crane) 47 Gryllus binraeu/atus (cricket. field) 82, 131 guarding 131.142.213 guppies Set Potdlia rttiCll/ata guppies courtship display 176-7 display design and sensory capacities 160-1 Gymnorhinus cyanoaphallls (jay. pinyon) 57, 58, 235 Gyrodaely/us tumbulli 177 habital fidelity. and genetic divergence 363-4 habitat loss consequences of 374-7.378 effects on population size 390-2 flexibility and constraint in response to 392-3 habi'alS 20, 372 variable quality of 386. 388 habituation 186.197 Haematopus osrraltgus (oysfercalcher) 386,39/ Ha~matopusostra/tgus
ostra/tgus
390 Hamilton's rule 9.204-8.212-14 handicap principle 8. 167-9
INDEX cnndilion·depcndent handicap 192 haplodiploidy 15()-J and Hymenoplcran eusuciality 209-14 Htlagale parv'la (mongoose. dwarf) 235, 248-9 'helping' behaviour 213, 228-9, 236,239, 243 HemilfpislUS rtmmrtri (woodlouse. desert) 73 h(Tlllaphrodiles 121-2 hidden preferences 161-2. 180. 189 hierarchies linear 258.261-2 matrilineal 259-62 in populalion SlruCturc 350-l hippocampu 54-5. 56. 68
Hipposideros 24 Hinmdo dallrica 347 Hirundo rUSlica (swallow. barn)
79,169, /73, 193-4,346-7
Homo sapirtls 296 honesty 149 advenising quality/need
167-9 in display /68 maintenance of 166. 169 in signalling 165 honey bees. dance language 164 hunting. flight or perch 98 Hyaena brmmea (hyena. brown) 235 hybrid zones 351. 355, 368 Hy/a (hrys
Jurya purchasi (scale insect) 295 ideal despotic distribution model 380, 389 ideal free-distribution models 379-80, 386 deriving density-dependent functions from 38~ imitation set behaviour copying imprinting Set template learning inbreeding 70 avoidance 83-4. 236 reduction of 82 inbreeding depression 364, 366. 367, 369 incest avoidance 236-7.247 inclusive fitness (theory) 204. 234, 285 independent contrasts. method
451
of 308, 341-5 indeterminale growers 316 information. outdated. discarded or devalued 48-9 information processing ability, importance of 67 information sharing su an:a copying inhibitory resetting 163 insect navigation 33-8 insects. eusocial 205-6 interception courses. computation of 38-40 interference. susceptibility to 383 individual variation in 385-6 inversions. duomosomal 289 'island models' 357 isogamy 122. 123 isolation asymmetric 370.371 by distance 358, 362 reproductive 371
Lasius niger 222,223 last male precedence. models and mechanisms of 130-8 bird models 134-8 insect models 130-4 learning non-soda I 63 via self-inspection 82. 85 lek, paradox of 126,179,185 upomis macrochirus (sunfish. bluegill) 47, 384 leptothoracine ants. skew evolution in 219
Junco hytma/is (junco, dark-eyed)
111-18 linear operator models 48-9 linear regression models 344 linkage disequilibrium 184.190, 289, 297-8, 300 lions 276-9 cooperation in 276-9 local mate competition. and female bias in sex ratios 222-3 local resource competition, effects of 22~ locomolion and body size 26 costs of fat slorage 107 effects of mass increase 108 loxodrome (constant-angle) rOutes 32-3
119
'Kaneshiro' hypothesis 370-1 key innovation. and phylogenetiC tree structure 335-6 kin. forming revolutionary alliances 260 kin conflict 203. 208 over worker reproduction 226-7 kin recognition action component 89-90 failures of 92-3 fitness consequences of 69-70 rnisunderslandings concerning 94-5 perception component 81-3 production component 74-7 vertebrate 75-7 kin selection and Hamilton's rule 204-8 necessity of 208-10 and sodality 203-27 and strategic genes 288 theory 150, 204, 234, 249 and skew theory 151.216 kinship 245-6 and tendency to cooperate 234-7 labels set cues 'Lack dale' 325, 326 LAgopus lagopus (grouse. red) 389 landmarks/landmark panoramas, in insect navigation 37-8 Lanius collurio (shrike. redbacked) 393 LArus ridibu'dus (gull. blackheaded) 6
Leplotltorax 219 Uplolhorax IOllgupinoslls 222 uplorhorax lIIbt'rum 22J lire hislory 311,351 studies 311. 332 Iheory 311 concepts in 312-14 lifetime reproductive success (LRS) 313,315,316,331 and short-term behaviour
Macaca tuscala (macaque. Japanese) 63, 64 Macrobrachimn crmulalum (prawn. freshwater) 160 macroplcry. and habitat persistence 365 major hislOcompatibility complex (MHC) 196, 294 male display 150 and resistance to parasites ISJ male pursuits 38-9 male sterility. CYlOplasmic 300 male trailS. variation in and female mate choice 180-2 Malurus cyantw (fairy-wren) 233 Manon'na melanoceplrala (miner, noisy) 235 Manorina melanophrys (miner. bell) 235 Marmola marmOla (marmot. alpine) 101
452
INDEX
matching-tn-memory 37. 38 male choice 138 copying 92 lor correct species 187 decision rules 90-1 and sexual selt'oion 179-202 and signalling 149-50 and speciation, in passerine
birds 336 male copying 200-1 male quality. decision rules 90 male Qualily recognition 78
mate quality recugnition templates 83-4. 86 karned 83-4 mate n:cognition 70 action component 90-2 perception component 83-6 production component 77-9 mate recognition rues. and benefits of discrimination 78-9 mate replacement 237-8 mate sharing 247 malt'switching J 28. 137 male-acceptance thresholds. optimal 91-2 mate·guarding 142. 143 male-resource recognition 78 mating conflicts 142-3 incestuous. rare in family groups 236-7 sl!e also multiple mating mating prderences conspeciric 197
illrJuence of social cues 200 pleiotropic eHects on 186-7 mating systems 9 lek (lek-Iike) 179 polygynandrous 144 r~source-based 184-5 and sexual dimorphism 56 and sperm competition 121-
45
maturation.andgrowlh )14-17 Mf9adfrma lyra (vampire. fals(') 23.24 meiolic drive 299. 300. 30 I MtI,merpes formicivorus (woodpecker. acorn) 88.231-2 MeJitat!Q a'nxia (fritillary. Glanville) 364
MeMpsittacus undlilatus
(budgerigar) 47. 62
Mt/'spiza me/odia
J 7)
memes 300 memory parameters 67-8 memory window models 48. 68 MtrrJps bulJockoidtS (bee-eater. while-fronled) 235.216. 237-8.239.248 Messor adculatus (ant. harvester)
223
Meraohidippus ameoills (spider.
jumping) 29 mt'tapopulatiun dynamics 358 microgametes 122 Microtus ochrogaster (vole. prairie) 56 Microtlls penl1sylvanicus (vole. meadow) 56. 103 Microtus piurlorum (vole. pine) 56 migrallis. mating success of 366-8 migratioll. selective 363 migralion routes evolution in 392-3 see also navigation migratory species changes in migratory behaviour 392 consequences (If hahitat loss 375-7 molecular interactions 294 Molothrus attr (cowbird. brownheaded) 85. 17) monocular ~xclllsion experiments 54. 58-9 mortality and brood size manipulalions 323-4 and mean age at malllrilY 320 multicdlular organisms. somatic security systems 293-4 multiple mating 9. l8. 144 by queens 226-7 see also sperm com}'k:'tilion Mus muscu!lu (mouse. house) 75. 357 M'sCtl domestira (houserJy) 369 mutations 285. 358 rept'at·induced point (RIP) 296 mutualism 152. 208. 276-7. 279. 281 shun· and long·tcnn 271-28 mutualistic advantages 274. 278 Myrm;ca tahomsis 225 Mytilus tdufis (mussel. edible) 357.386 MyXOfOCCUS xanthum 293. 294 natural selection. againsltraits 189 navigation by insects '33-8 dis~cclion into subroutines 33. 40-1 navigarional errors 35-6 neighbourhoods. size and Siructure 360-1 Neoconortphalus spiza (katydid) 164
nepotism 70 nervous systems 19. 20. 36-8 Ntjom;mlls parvulus (mockingbird.
Galapagos) 235
Neumann;a papiIJator (water mite) 161 neural mechanisms. and female mate choice 181 nl:urogenesis. seasollal 56 Nturospora crassa (brc..·ad mould) 296 niche panitioning. in adaptiyt' radiations 348-9 Ilun·breeding season. measurement of density dependence in 381-8 Nuci!raga columbiana (lllllcracker. Clark's) 54. 57. 58 ubjf'C1 copying (stimulus enhancement) 60-1 OmamheOfllamlJt (wheatear) 393 offspring. size and numbers 32932 offspring viability 193-4
Ololygon mbra
185
optimal behaviour rule I 16 optimality models 6-7.97 optimization 313. 333 of age and size at maturilY 3/5. 315-16 individual 324.326.330-1 origin recognition complex (ORC) 296 unhodrome (greal·circ!c) routt'S 32
oscines 159-60 olltbreeding 70 optimal 367-8 uUlbreeding depression 366.367 outlaw genes 82 Ovis aries (sheep. Soay) 385 OxybeJis aentus (snake. vine) 185 P2 values 129-30. 131. 134 pairbonds 230 pairwise comparisons 308. 346-7 painvise contestS. compelition in 255-8 Pall troglodytes (chimpanzee) 62. 64
panglossianism 10-11 panoramic vision 27-9 Pantl1era leo (lion) 235 paper wasps 17.89. 89. 94 parasite load. influencing remale mating preferences 194-5 parasitism 4-5 brood parasitism 93. 17 I parent-offspring associations 246 parent-oHspring communication 170-1 parental care conflict over 143 evolution o~ 337-8 parental eHon and investment
INDEX 322-4 parental manipulation 208
parcntal repairing 252
change in sexual dynamics
237-8 Partllia surf/ralis (snail. land) 368 Pams aler (lit, coal) 57
Pams atricapillus (chickadee. black-capped) 54
Pams major (tit. great) 82. 101.
102. 108. 110. 322. 323-4. 330. JJI Pams momanus (tit, willow) 104
Parus pa/1I5m's (til. marsh) 54. 58-9
Passermlus sandwichmsis (sparrow. savanna) 322
Passrrcufus sandwichmsis princeps
(sparrow. Ipswich) 116
Passerina
cyallea
173
passive sperm loss model 135. 136. 137. 138 patch quality 16
and Bayesian estimation 45-8 ',ches 351 patchiness and g~netic drift 356-7 and population structure 364
paternity
extra pair 127-8. 128-9 l1lulliple 17-18. 128
and paternal care 127 IlatlernS of 129-38
paternity assignment 128 patcrnity guards 142 paternity insurance 346-7 path integration ~t insect navigation Pavlov strategy 269.270. 277 pay-ofrs. shon-term. role or
PBS
281-3 122-3
phYlOphagy. and insect diversification 336
Pica pica I 71 Picoides borealis (woodIJecker. redcock.ded) 233 Pipistrflilis 24
plasm ids 291 protection rackets 291-2 Pltcrrophenax nivalis (snow buming) 385 pleiotropy, affecting female male preferences 186-7 Pltlhodo' dfltrtus (salamander) 357
Pltlhodo' ouachitat (salamander) 357
plumage brighlness/patLerns 194-5
and signalling
peak shih learning 189
perception. recognition componem 7 L 79-86 Peromyscus cali/omicus (mouse) 357
phenolype matching. selfreferent 96 phenotypic plaslicily 307. 314 pheromones J 85 queen. and honesl signal 226 photoperiod. and body mass 102-3
J 03
phylogenetic conslraint adaplive or gr:nclic 340 phylogenetic trees 6. 334 mapping character change
and dispersal behaviour 364-6 interactions with behaviour 363-8
shaped by the Pleistocene periud 353-6 shaped by present interactions 356-9
lesting predictions 36()'-3 gene now 361-2 neighbourhood size and structure 360-1 population subdivision and gene now 362-3 populations genetic history, and pairwiSt' comparison 347 genetic structure 308 parapatric 355 small, isolated, evolution of behaviour in 368-72 Porphyrio porphyrio (pukeko) 245 predation risk. mass-dependent 107-8. 110. III. 114 predator inspection 279
Priaptlla 189
prisoner's dilemma 152, 266-70, 280. 281-3
producrion. recognition component 71. 72-9 protected invasion hypothesis 211
173-4 Podiceps cristattls (grebe. great crested) 393 Podisma pedtSlris (grasshopper. brown mountain) 351. J 59,
J52. 357. 359
Poedlia formosa (molly, Amazun)
Prunella collaris (accentor, alpine) 93. 144
Prunella modularis (dunnock) 93. 17J
pseudopregnancy 24S-9 public information 50-1. 59 pursuit deterrence 279
200-1
PoedUa fatipinna (molly. sailfin) PoedUa reticulala (guppy) 160, 161.176
poison-al1lidote Illechanisms
123-4
Pltylloscoplls 174
Physalaeltlilscoloradorum 187.188 PltysalatnIlIs petersi 187 Physalatmlls pustulosus (frog, tunga... ) 174.187.188
200-1
depends un sperm competition
pholUrect=pturs 21 pholOrerraCioriness
onW 337 and phylogeographies 359-60 structure of 335-6 as working hypolheses 335 phylogenies 307-8 bifurcating, independent conlrasts 342 can mislead 344-5 phylogeographies 355. 359--60
453
qua lily estimation 42-51 in social environments 50-I in unchanging environmenh 45-8
Qurrws fntvis (oak. Thrkey) 360
291-2. 299
PoUsus dominulm (paper wasp) 219
PoUstts fuscattls (polisline wasp) 218
polistille wasps. skew evolution in 218-19 polymerascs. DNA and RNA 290 polymorphism 8 and dispersal cosls/benefits 364-5 genetiC. balanced 333 Pongo pygmat!us (orangutan) 62 populalicm genetic models 357-8 populalion size. and habitat loss 37fr-7. 390-2 population slructure 350-3 current vs historical effects 359-60
rarnes. fair and loaded 130. 131. 135. 139-40
Raila cascadat 75 Rana sylvarica (frog. wood) 75. 76 Rana Itmporan·a (frog. common) 75
range changes due to djrnatic/cnvironrnemal fluctualion 353-6 range expaniion 360 rank determination experiments 259-61
high, and reduced reproductive success 262-3 maintenance of 260 rank inheritance, mat£'rnal 259. 259--60
454
INDEX
'rate of living'. allometry of J 17. 318
Raltus norvegicus (rae Norway) 61. 67 reaction norms 313. 332 rearing, communal 249
receivers
influencing display design 160-3
selective pressure on 155. 164-72 redprocily 265-70.275-83 basic problem 266-7 blue jays. cooperative key pecking 281-3 cooperation in lions 276-9 predator approach behaviour 279-81.28/
recognition
forms and functions of 69-70 parem-offspring 75 research 92-5 recognition systems 16-17,6996 components of 70-92 evolution of 73 recombination 289. 290-1. 298 bacterial 297. 304 chromosomal 292 creating relatives 304 eukaryotic 304 meiotic 297
reduced search ratc hypothesis 52 reduction principle 298
redundancy. and signal detection 158
rerrainers 276
refugia (kipuka) 353 ice-age 354. 355 rejection errors 72. 92-3 relatedness 204, 215 alfeetingskew 217-18 genetic, and skew models 245 levels of. Hymenoptera 20910
costs of 322. 323-4 and ecological constraints 252-3
independent, probable success 244-5
seasonal timing of 324-9 reproductive sharing 242-4, 245, 246. 249. 250. 252-3
reproductive skew and dominance 264-5 stable 203.215-19 theory 151.229.242-4.247-9 reproductive success 381 future 98 reduced, with high rank 262-3 see also lifetime reproductive success reproductive value 323 Requena vertiealfs (cricket. bush) 185
resource allocation 307.316-17 resource holding power (RHP), and competition 255. 256. 258
resource-management strategies 308. 373
resources, high-quality 251 Rhagoletjs pomollella (fly. apple maggot) 363 Rhinopoma 14
Riparia riparia (swallow. bank) 8.94. 108
risk-spreading theorem 114 Risso tridactyla (gull. kittiwake) 5 ritualization see signal evolution
Rivullis hartii 160 RNA editing 292
runaway sexual sel~ction (Fisher) 180. /83. 190-2. 197.199.201
Rupicola rupieola (cock-or-therock) 90
Sa/rna salar (salmon) 113 Salve/inus alpinus (charr. Aretic) 3/6
in a sodal insect colony 207 relatedness asymmetry 220-1.
scalar expectancy theory (SET)
222.224. 225. 303 relationships 301-3 ecology 01 254-83
scaling, of time and energy 317-
relative payoff sum (RPS) 48--9 rel)Ctition 157-8 replacement breeders 241-2. 252 replacement families 237,240 less stable 240 reduction in indirect benefits 238--9
replication 295-6 genetic 289-90 reproduction 314 alLernating with growth 316 (oortiet over 241-6
16.43-5 22
scaling relationships 307 Scaphiopus bombifrons (toad, spadefoot) 90 Scatophaga 133. 142
S({Ilophaga Sltrcoraria (dungny.
yellow) 7.18.131 seance-hoarding 105 See/opoms occidentalis 108 Schizocosa ocriata (spider, bushlegged woll) 176 Schizocosa rovner; 176 Sdurus vulgaris (squirrel. red) 389 scrounging, preventing skills
spread 65-6 search costs 185-6 search image hypothesis 51-2 selection direct 181. 183. 184. 197.201 on female mating preferences 184-90, 195 subtle 189-90 and fitness 4 indirect 18 L 184 on female mating preferences L92-7 on mating preferences 190-2
and 'Iying' mutants 166 see also sexual selection self-aggrandizement 290 self-displacement 131 self-inspection 82-3, 96 mediating disassortative mating 85 self-peptide 196 self-recognition 75 cost of using V-absent cues 74 genetic see green beard effects self.rejection 74
Semolilus atromaeulatus 113
sensory bias 186. 196 sensory capacity, and display design 160-2 sensory exploitation (sensory trap) 161-2.174.186. 187-9
sensory systems 15. 20. 186 bias in 181-2 Sen·nusranarills (canary) 177-8 Serinus serinus (serin) 393 sex allocation manipulation of 225-6 Queen-controlled 222 worker control 221, 225 sex ratio theory, social Hymenoptera 220-1 tests of 221-6 sex ratios biased 30 I and PBS 123-4 sex investment ratios. effect of resource levels 222 sex ratio evolution 203 see a/so split sex ratios sexual competition 17 absent in intact families 235-6 sexual conflicts, and mating systems 142-4 sexual dimorphism 18/ and sexual seledion 198-9 sexual display 168 sexualseled.ion 17,78,121. 124 and mate choice 179-202 shoaling 280
SiaUa siaUs 173
sibling competition. and food
INDEX allocation 178 sibling-sibling associations 245 signal detection theory 51 signal evolution 155-78 coop~ration and conflict in 164-7 selective pressure on receivers 155,164-72 on signallers 155. 156-64. 173 signaller interaction effects 162-4 signallers multiple l77-8 selective pressure on ISS. 156-;;4. 173 signalling assessment of immigrant genetic contributions 368 costS of 156 deceit in physical constraints 166-8 possibilily rare 170-1 strategic conSlraints 16770 and mate choice 149-50 and plumage brightnessl patterns 159, 173-4 sexual 165 see also acoustic signalling signalling systems 8, 164-5 signals diversity 01, stralegic explanalions 171-2, 17J erreetive 156-8 efficient and reliable 173-4 multiple 175-7 as source of information 164-5 'silver spoon' effects 332 single nesting vs helping 212-14 Siun europnea (nuthatch) 389 skylight compass 34 skylight pattern. in insect navigation 34-5 slave-makers 221-2 sneak-guarder situations 140 social actions 204 social behaviour 8 evolution of 150-2 social dominance 104. 247. 249. 253 and fighting ability 246 social groups. slability of 152 social insects monogyny or polygyny 21516 sex ralio evolution and kin conflict in 220-7 social learning 16. 59-62 and trait acquisition 66 social predispoSitions, heritable, undt'rstanding of 250
social relationships 254 social tensions 151 social venebrates, predicting family dynamics in 22853 Sodality affording protection 212-14 and kin selection 203-27 societies. subsocial and semisodal 217-18 soma-germ line. physical cohesion 293-4 somatic exploitation. of slime moulds 294 somatic spedalization 293 sonar 23 song and escape ability 170 temporal structure of 159-60 songbirds 197 singing vs foraging 115-16 sound--source detection, and body size 30 spatial memory 16, 68 as an adaptive cognilive specialization 53-9 speciation 347,364 by r('inforcement 367 parallel 371-2 in passerine birds, and mate choice 336 species diverSity, concern over 373 sperm competition 9, 17-18. 121. 123-4 detection and incidence of 126-9 female behaviour 141 sperm displacement with instant mixing 130. 131 mixing after displacemelll 130, 135 sperm expenditure theory 13940 spenn polymorph isms 140 Sptrmophillis beldiugi (sqUirrel, Belding's ground) 75. 80. 82.88.94 Sptnnophilus columbianlls (squirrel. Columbian ground) 384 Spermophilus rridfamlineotus (squirrel, 13-lined ground) 139 spherical geometry, coping with 31-3 split sex ratios in eusocial Hymenoptera 224-6 and origin of eusodality 210II
squirrelS, red and grey 289 starvation, mortality rate
455
from 387.388. 388 state variables II L 119-20 'stepping stone' models 357-8. 361 strategic genes 286, 287-9 Srurnus vulgaris (starling) 3.44. 50-1,103. 108. 109 subordinates, as breeder replacements 241-2 survival. efficiency ys dominance 387-8 suspidous TFT lSTFTl 268 Sylvia arricapilla (blackcap) 392 symmetry, sensory bias towards 196 Symphodus tinea (wrasse, peacock) 92 synchronization 351 '-complex, and male choice 197 Tochydueta hicolor 17J Tadarida 24 Taeniopygia guttata (finch. zebra) 17.18.66.92.109.135. 137-8.141-2.324.384 temperate zones, increased productivity and metabolism of animals 320-1 template learning 82 timing of 80 lemplates 71,79-81 genetically encoded 80. 83. 85 kin-recognition 82,83 learned 80.81-2.83-5 mate-quality recognition 83-4 mate-recognition 77, 85 neural 35 sex-recognition 84-5 temporal memory 44 temporal variation 351-3 Tenodera australasjat (praying mantis) 29 lerminal reward 111,112-14 termites, origin of eusodalily in 214-15 territorial defence strategies 38990 territorial quality, and family group formation 231-2. 233. 234 territory size. density dependenl 389 Ttltigonia vin-dissjma (bushcrickct) 25 TIJalassoma bifasdatum (wrasse. blue-headed) 78. 141 Thamnophis sirtalis (snake. ganer) 80 Thomonys bottae (gopher) )57 threat diplays 165 time and energy, scaling of 31722
456
INDEX
time and energy management 97-120 time estimation, scalar
V-absent cues 72,74,77.81 Ultethasia 185
expectancy theory (SET) 43-5 Tinbergen's fOUf questions 4-5. 11
underwater hearing 22-3 underwater vision 21-2 updating
tit-for-tat strategy (TFT) 266-7.
267-8. 277 and shoaling 280 weakness of 268-9. 268-70 tit-for-two-tats strategy (TF2T)
268
tits see Porus spp. lrade-offs 17. 23. 97. 307. ~ 13. 394-5
current and future
reproduction 322. 329 dispersal and fecundity 366 fecundity and survival 314-15 traditions, longevity of 66, 67 traits 312. 3J4 cultural, spread of 63, 64
fitness-related. courtship characters 369 manipulation of 332-3 multiple, and their preferences 198-200 and preferences. genetic correlation of 191-2
sexually-selected 126
transduction 292 transfonnation 292-3 transitions. evolutionary 337--8 Trichogramma minulum (wasp.
ichneumonid) JI Thrdoides squamiceps (babbler. Arabian) 242. 247-oS
Turdus merula (blackbird) 116,
389
ultrasound 30
Bayesia n 45-7
and relative payoH sum (RPS) 48-9
of templates 80-1, 92 Upao/ia laevigata 185 Ursus nJaritimus (polar bear) 106 Uta slansburiana (lizard. side·
blotched)
~29
variability in behaviour 7-8 environmental 311-L2. )14,
316.318 19.332 of the experienced S 44-5 genetic 311. 333 vervet monkeys st( Cercopithecus aethiops viability 193-4.300 vibration. substrate·borne 30 Vidua chalybtQftl (indigobird.
village) 80-1 vigilance. cooperative 272 vigilance behaviour 110 Vipera herus (adder) 129
vision and spatial information 21 visual acuity 27-9 visual imitation 62 vocal learning. birds 61
voles 54. 103 VoJucelJa pel/uctfls (hoverfly) 389.39
white·fronled bee·eaterS see
Merops bullockoides
window-shopping teSls 58 wing dimorphism 365 winner-loser effecls 256-8 winter. as a foraging environment 101 winter habilat, consequence of loss of 37(,-7. 390-2 worker policing 226-7 worker reproduction. kin conflict over 226-7 worker--<)ueen conflict 226 sex allocation 221 workers. assessing their relatedness asymmetry 225
XanrJrocephalus xanthocepJralus (blackbird. yellow-headed) 178.381 Xiphophorus macu/ams (swordfish) 188-9 Xiphophorus mu/tilinemus 189 XiphopJrorus nigrensis 189 Xiphophorus pismaeus 189 Xiphophoms variatus (sword· fish) 188-9 Xylocopa sulcatipes (bee. carpenter) 212.213.216 zebra finches see Taeuiopygia
suuata Zenaida aurita (dove. Barbados zenaida) 61
ZOllotrichia Itllcophrys 173 ZOstera spp. 395