Advances in
THE STUDY OF BEHAVIOR VOLUME 26
This Page Intentionally Left Blank
Advances in
THE STUDY OF BEHAVIOR ...
14 downloads
987 Views
26MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Advances in
THE STUDY OF BEHAVIOR VOLUME 26
This Page Intentionally Left Blank
Advances in
THE STUDY OF BEHAVIOR Edited by
PETERJ. B. SLATER School of Biological and Medical Sciences University of St. Andrews Fife, United Kingdom
JAY S. ROSENBLATT Institute of Animal Behavior Rutgers University Newark, New Jersey
CHARLES T. SNOWDON Department of Psychology University of Wisconsin-Madison Madison. Wisconsin
MANFRED MILINSKI Zoologisches Institwt Abteilung Verhaltensokologie Universitat Bern Hinterkappelen, Switzerland
VOLUME 26
ACADEMIC PRESS San Diego London Boston New York Sydney Tokyo Toronto
/
This book is printed on acid-free paper.
@
Copyright 0 1997 by ACADEMIC PRESS All Rights Reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the Publisher. The appearance of the code at the bottom of the first page of a chapter in this book indicates the Publisher’s consent that copies of the chapter may be made for personal or internal use of specific clients. This consent is given on the condition, however, that the copier pay the stated per copy fee through the Copyright Clearance Center, Inc. (222 Rosewood Drive, Danvers. Massachusetts 0 1923). for copying beyond that permitted by Sections 107 or 108 of the U.S. Copyright Law. This consent does not extend to other kinds of copying, such as copying for general distribution, for advertising or promotional purposes, for creating new collective works, or for resale. Copy fees for pre-1997 chapters are as shown on the title pages, if no fee code appears on the title page, the copy fee is the same as for current chapters. 0065-3454f97 $25.00
Academic Press a division of Harcourt Brace & Company 525 B Street, Suite 1900, San Diego, California 92101-4495, USA http://www.apnet.com
Academic Press Limited 24-28 Oval Road, London NWI 7DX, UK http://www.hbuk.co.uk/ap/ International Standard Book Number: 0- 12-004526-5 PRINTED IN THE UNITED STATES OF AMERICA 97 9 8 9 9 00 01 0 2 Q W 9 8 7 6
5
4
3 2 I
Contents
Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Preface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
ix xi
Sexual Selection in Seaweed Flies
THOMAS H. DAY AND ANDRE s. GILBURN I. 11. 111. IV.
V. VI. VII. VIII. IX. X.
Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . General Biology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Population Genetics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mating Behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sexual Selection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Variation in Female Mating Behavior. . . . . . . . . . . . . . . . . The Preferred Trait: Male Size. . . . . . . . . . . . . . . . . . . . . . Evolution of Female Mate Preferences. . . . . . . . . . . . . . . . Discussion ...................................... Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1 3 6 7 12 19 25 32 44 48 49
Vocal Learning in Mammals VINCENT M. JANIK AND PETER J. B. SLATER I. 11. 111. IV. V. VI.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Evidence for Vocal Learning ....................... Survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Functional Significance and Origin . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
59 61 63 75 85 90 91
Behavioral Ecology and Conservation Biology of Primates and Other Animals KAREN B. STRIER
I. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Comparative Paradigms. . . . . . . . . . . . . . . . . . . . . . . . . . . . V
101 103
vi
CONTENTS
I11. Genetics and Ecology ............................. IV. Methodological Bridges to Disciplinary Convergence . . . . V. Summary ....................................... References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
126 137 143 145
How to Avoid Seven Deadly Sins in the Study of Behavior MANFRED MILINSKI I . Unjustified Conclusions from Observational Data . . . . . . . I1. Data Are Not Independent: “Pseudoreplication” . . . . . . . I11. Treatments Are Confounded by Time and Sequence Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. No Efforts to Avoid Observer Bias . . . . . . . . . . . . . . . . . . V. Potential Artifacts When Animals Are Not Accustomed to Experimental Procedures ........................ VI . Unsuitable Controls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII . “Proving” the Null Hypothesis with Small Samples . . . . . . VIII . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IX . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
160 165 168 169 174 175 177 178 179 179
Sexually Dimorphic Dispersal in Mammals: Patterns. Causes. and Consequences LAURA SMALE. SCOTT NUNES. AND KAY E . HOLEKAMP I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1. Biology and Study Methods for Spotted Hyenas and Belding’s Ground Squirrels . . . . . . . . . . . . . . . . . . . . . . . . . I11. Sex Differences in Mammalian Dispersal: Patterns and Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Sex Differences in the Proximal Causes of Natal Dispersal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Consequences of Sex Differences in Dispersal Behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI . Directions for Future Research . . . . . . . . . . . . . . . . . . . . . . VII . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
181 184 191 208 228 234 237 239
vii
CONTENTS
Infantile Amnesia: Using Animal Models to Understand Forgetting H. MOORE ARNOLD AND NORMAN E. SPEAR 1. Orientation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11. 111. IV. V. VI. VII.
252 254 257 263 267 275 278 279
Ontogeny of Non-Associative Memory. . . . . . . . . . . . . . . . Ontogeny of Short-Term Retention . . . . . . . . . . . . . . . . . . Ontogeny of Long-Term Retention: Infantile Amnesia. . . Perspectives on Infantile Amnesia . . . . . . . . . . . . . . . . . . . Infantile Amnesia as an “Ontogenetic Adaptation” . . . . . Summary and Comment ........................... References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Regulation of Age Polyethism in Bees and Wasps by Juvenile Hormone SUSAN E. FAHRBACH I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
285
11. Age Polyethism and Juvenile Hormone in the European
Honey Bee, Apis mellifera. .........................
288
111. Effects of Experimental Treatment with Juvenile
IV.
V.
VI. VII. VIII.
Hormone, Mimics, and Analogs on Behavioral Maturation in the Honey B e e . . ..................... Correlational Data Indicating Juvenile Hormone Titers Increase during Behavioral Maturation and Are At Their Highest in Foraging Bees .......................... Colony Manipulations that Induce Altered Juvenile Hormone Titers and Altered Behavior; Seasonal Changes in Juvenile Hormone Titers and Behavior.. . . . . . . . . . . . Brain Changes Correlated with Behavioral Maturation in the Honey Bee: A Proposed Mechanism for Juvenile Hormone Action on Age Polyethism.. . . . . . . . . . . . . . . . Comparative Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions and Significance. . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
293 297 300 301 306 309 310
Acoustic Signals and Speciation: The Roles of Natural and Sexual Selection in the Evolution of Cryptic Species GARETH JONES
I. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
317
11. A Case Study: Britain’s Most Common Bat is Two
Species . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
320
viii
CONTENTS
111. Acoustic Resouce Partitioning by Echolocation . . . . . . . . . IV . Cryptic Speices of Echolocating Bats . . . . . . . . . . . . . . . . . V . Acoustic Signals and Cryptic Species in Nonecholocating Animals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI . Speciation in Cryptic Species That Use Acoustic Signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII . Cryptic Species. Genetic Divergence. and Hidden Biodiversity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VIII . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
330 335 337 339 344 346 348
Understanding the Complex Song of the European Starling: An Integrated Ethological Approach MARCEL EENS
I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1. The Study Species and Its Song ..................... 111. Song Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV . Relating Song Behavior to the Underlying Neutral Substrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V . Functions of Song . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI . Functions and Evolution of the Song Repertoire . . . . . . . . VII. Conclusions and Future Directions . . . . . . . . . . . . . . . . . . . VIII . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
355 357 362 380 391 406 417 422 423
Representation of Quantities by Apes SARAH T. BOYSEN
I. Representation of Quantities by Apes . . . . . . . . . . . . . . . . I1. Symbolic Facilitation of Quantity Judgments . . . . . . . . . . . I11. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Index
...........................................
435 447 458 460
463
Contributors
Numbers in parentheses indicate the pages on which the authors’ contributions begin.
H. MOORE ARNOLD (251), Department of Psychology, Duke University, Durham. North Carolina 27706 SARAH T. BOYSEN (439, Comparative Cognition Project, Department of Psychology, Ohio State University, Columbus, Ohio 43210 THOMAS H. DAY (l),Genetics Department, Queens Medical Centre, University of Nottingham, Nottingham NH7 2UH, United Kingdom MARCEL EENS (355). Department of Biology, University of Antwerp, UIA, B2610 Wilrijk, Belgium SUSAN E. FAHRBACH (285), Department of Entomology, University of Illinois at Urbana-Champaign, Urbana, Illinois 61801 ANDRE S. GILBURN (l), Genetics Department, Queens Medical Centre, University of Nottingham, Nottingham NH7 2UH, United Kingdom KAY E. HOLEKAMP (181), Department of Zoology, Michigan State University, East Lansing, Michigan 48824 VINCENT M. JANIK (59), School of Biological and Medical Sciences, University of St. Andrews, St. Andrews, Fife, KY16 9TS, United Kingdom GARETH JONES (317), School of Biological Sciences, University of Bristol, Bristol BS8 I UG, United Kingdom MANFRED MILINSKI (159), Abteilung Verhaltensokologie, Zoologisches Institut, Universitat Bern, CH-3032 HinterKappelen, Switzerland SCOTT NUNES (181), Department of Zoology, Michigan State University, East Lansing, Michigan 48824 PETER J. B. SLATER (59), School of Biological and Medical Sciences, University of St. Andrews, St. Andrews, Fife, KY16 9TS, United Kingdom LAURA SMALE (181), Department of Psychology, Michigan State University, East Lansing, Michigan 48824 ix
X
CONTRIBUTORS
NORMAN E. SPEAR (251), Centerfor Developmental Psychobiology, Department of Psychology, Binghamton University, Binghamton, New York 13902 KAREN B. STRIER (101),Department ofAnthropology, University of WisconsonMadison, Madison, Wisconsin 53706
Preface
The aim of the Advances in the Study of Behavior series remains as it has been since the series began: to serve the increasing number of scientists who are engaged in the study of animal behavior by presenting their theoretical ideas and research to their colleagues and to those in neighboring fields. We hope that the series will continue its “contribution to the development of cooperation and communication among scientists in our field,” as its intended role was phrased in the Preface to the first volume in 1965. Since that time the traditional areas of animal behavior research have achieved new vigor by the links they have formed with related fields and by the closer relationship that now exists between those studying animal and human subjects. The range of scientists studying behavior today is wider than ever before: from ecologists and evolutionary biologists to geneticists, endocrinologists, pharmacologists, neurobiologists, and developmental psychobiologists, not forgetting the ethologists and comparative psychologists for whom the subject is the prime domain. It is our intention not to focus narrowly on one or a few of these fields but to publish articles covering a broad spectrum of the best behavioral work. The skills and concepts of scientists in such diverse fields necessarily differ, making the task of developing cooperation and communication among them a difficult one. However, it is a task of great importance, and one to which the editors and publisher of Advances in the Study of Behavior are committed. We will continue to provide the means to this end by publishing critical reviews, inviting extended presentations of significant research programs, encouraging the writing of theoretical syntheses and reformulations of persistent problems, and highlighting especially penetrating research that introduces important new concepts. The wide range of topics dealt with in the present volume illustrates these aims well, with a good mixture of psychological and biological approaches, as well as laboratory and field studies. In this volume several chapters are on mammals, two are on insects, and one is on birds. In addition, we have something of a departure with one chapter dealing with problems of experimental design in the study of behavior. What they all have in common is that each tackles an important topic and provides insights of wide significance to those interested in the study of behavior. Volume 25 had as a special topic “Parental Care.” It is planned that Volume 27 will focus on the subject of “Stress and Behavior.” By contrast, the present volume features wide-ranging topics in line with the norm for the series. It is our intention to continue the series with this mixture of volumes of eclectic interest and ones focusing on particular themes that appear timely to us. xi
This Page Intentionally Left Blank
ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 26
Sexual Selection in Seaweed Flies THOMAS H. DAYAND ANDRBS. GILBURN GENETICS DEPARTMENT UNIVERSITY OF NOTTINGHAM QUEENS MEDICAL CENTRE NO’ITINGHAM, UNITED KINGDOM
I. INTRODUCTION Darwin is credited with the idea that organisms can change in response to both natural selection, mediated by differences in survival, and sexual selection involving differential reproduction (Darwin, 1859,1871). He proposed the existence of sexual selection because in many species the males possess characters, such as the antlers of deer and the peacock’s tail, that seem to be maladaptive in terms of survival. Such structures are supposed to confer an advantage in reproduction either by increasing a male’s ability to compete for mates, or by making males more attractive to females. These ideas were almost totally ignored for about sixty years, possibly because in the prudish atmosphere of Victorian England, sex was not a subject that could be openly discussed. However, even the scientific community was unconvinced by Darwin’s suggestion that it is a sense of esthetics that renders females reluctant to mate. In the century following the publication of The Descent of Man, sexual selection was mainly considered in the context of speciation. For example, Mayr thought that “many phenomena that have been recorded as furthering intra-specific sexual selection, are actually specific recognition marks” (Mayr, 1963). The contributions of Fisher (1930), which are to this day being actively interpreted, Bateman (1948), Maynard-Smith (1958), and O’Donald (1962) are among the few studies carried out during this period. It was Trivers (1972) who offered the first explanation of why females are usually the more choosy sex; he suggested that the greater investment in terms of energy and parental care would lead females to be more reluctant to mate, and to be discriminating in their choice of mates. The current renaissance of interest in sexual selection occurred in the early 1980s and developed in two largely discrete directions of research. 1
Copyright Q 1997 by Academic Press All rights of reproduction in any form reserved. 0065-3454/Y7525.00
2
THOMAS H. DAY AND
ANDRB s. GILBURN
The first was primarily concerned with understanding the evolution of exaggerated male characters, and of the female preferences that were thought to be involved in generating them. This approach relied heavily on theoretical models to explore possible mechanisms. The models of Lande, Kirkpatrick, Arnold, Barton, Turelli, Pomiankowski, and many others have been reviewed by Anderson (1994). The second area of research involved ethologists and experimental biologists who were trying to understand mating behavior in a particular species, or group of species. While there were many who used both approaches, it is generally true that in the 1980s the theoreticians rarely modeled actual experimental systems, and few experiments were designed to test the assumptions and predictions of models. At that time the prospects of the two groups interacting were far from bright. It seemed the advice of Grafen was being heeded: To theoreticians I would say: resist the temptation to make your models more realistic. Your function now is to provide the material which will determine what field workers will find interesting in five years time and beyond. Particular species are in themselves boring and deserve modelling only to make general points. To field workers I would say: whatever else you do, don’t go out and measure a parameter in somebody else’s model. Your role is not to be a theoretician’s technician. A good field worker is nobody’s poodle (Grafen, 1987, pp. 231-232).
Fortunately, the 1990s have seen increasing dialogue between experimenters and theoreticians, and, in consequence, our understanding of sexual selection has deepened. This review is unashamedly from the experimentalists’ camp (if, indeed, it still exists), and concerns a single species. The Nottingham group has been studying seaweed flies for over 20 years and has taken a broad approach, being interested in their genetics, ecology, mating behavior, and evolution. Because many results have remained in the form of unpublished theses, or have been published in ecological or genetic journals, the seaweed fly story is not familiar to many behavioral scientists. The study of genetics has benefited immensely from there being several species that are amenable to experimental analysis (including Escherichia coli, Drosophila melanogaster, Caenorhabditis elegans, and Mus domesticus). No single species has contributed in such a dominant way to our understanding of behavior, or of ecology. The wide diversity of species studied have suggested that there are fewer fundamental principles in these disciplines, and that arguably every species is potentially informative. Nevertheless, some species provide greater scope for experimental analysis than do others, and in our view Coelopa frigidu is among them. Seaweed flies do, of course, have limitations, but there are four qualities that make them particularly attractive material for the study of mating behavior:
SEXUAL SELECTION IN SEAWEED FLIES
3
1. The mating process is straightforward and easily studied under controlled laboratory conditions.
2. In the wild they occupy a well-defined habitat and natural populations are accessible to study. 3. They can be cultured in the laboratory and have a rapid life cycle. Controlled crosses can be performed so allowing genetic analysis. 4. A large proportion of their genome is contained within polymorphic chromosomal inversions. This means there are very large gene complexes each of which has a readily detectable effect on the phenotype, yet is inherited in a monogenic, rather than polygenic, fashion. Here we shall first review the general biology, ecology, and genetics of seaweed flies before focusing on their mating behavior. Although we are still a long way from identifying the genes determining mating behavior, it is clear that individuals do carry alternative alleles at the relevant loci. It is also evident that mating behavior, and in particular female choice, is subject to multiple selective forces-sexual selection is a pluralistic process. Finally, the combination of laboratory and field work has given us the impression that the mate preferences seen in natural populations are fluid in the sense that gene frequencies are responsive to various ecological factors. 11. GENERAL BIOLOGY
Seaweed flies constitute a small group of Acalypterate Dipterans; worldwide there are about 30 species of Coelopidae, all of them found associated solely with piles of decomposing seaweed beached above high-water mark (Egglishaw, 1960;Dobson, 1974a;McAlpine, 1991). Because of their dependence on marine algae (Dobson, 1976) they are found only on rocky coasts that support sufficient stands of large, brown algae to generate fairly deep deposits on the shore. The distributions of natural populations of Coelopa frigida and C. pilipes, which are restricted to coasts around the North Atlantic, are shown in Fig. 1. Coelopa frigida is aptly named; it is found in Iceland, the Faroes, northern Norway, and Spitzbergen. For much of its rangc in Europe it coexists with C. pilipes (Dobson, 1973,1974b);the larvae of the two species compete (Leggett, 1993; Wilcockson, Day, Phillips, and Arthur, 1996), and adults coexist and even mount animals of the “wrong” species (Leggett, 1993). Whether they can properly be described as sibling species is not clear, but copulation between the two has never been observed, and no evidence for interspecific hybridization has been obtained. However, there is no doubt that within the species C. frigida, adults from
4
THOMAS H. DAY A N D ANDRB S. GILBURN
FIG.1. The distribution of seaweed flies in Northern Europe. The left side of each circle indicates the presence or absence of C. frigida, the right side C. pilipes. Redrawn from Phillips er al. (1995).
SEXUAL SELECTION IN SEAWEED FLIES
5
north American, Icelandic, British, and Swedish populations can mate and produce viable offspring in the laboratory. Two physical factors, winds and tides, profoundly influence the biology of seaweed flies. The primary effect is on the availability of their food (predominantly species of Laminaria and Fucus), but it is clear that certain aspects of mating behavior also correlate with tidal variation. The principal agent that detaches large brown algae from their rocky substrate are storms. Once the seaweed is detached, onshore winds can deposit it on the shore and occasionally remove it from above the high-water line. The more regular removal and deposition of stranded weed is a consequence of the tidal movement of seawater, particularly the spring tides that occur once each month. The presence or absence of tides results in there being two extreme types of habitat in which seaweed flies live. In the Kattegat (between Denmark and Sweden) and in the Baltic Sea, occasional storms deposit weed, but it remains beached for long periods of time because of the almost complete absence of tidal movement. In consequence, populations of seaweed flies are effectively living in continuous culture, often with several batches of food present in varying stages of decomposition. Populations living in the Channel Islands in the Gulf of St. Malo (off the northwestern coast of France) experience a distinctly different regime of food availability. The very high tidal fluctuation-among the highest in the world-results in seaweed rarely being present for more than one lunar cycle. This means that during the winter when storms commonly occur, seaweed is regularly replaced, but during the calmer summers, beaches may be almost completely clear of weed. The extreme pattern of food availability to which Baltic flies are exposed results in very stable populations, whereas Channel Island populations are subject to greater fluctuations; when food is abundant, population numbers may reach 109-10'0 animals, whereas at other times there can be no food at all, and the populations are considerably smaller (Butlin, Collins, and Day, 1984). The habitats in the Baltic Sea and the Channel Islands are extremes. Around northern Europe and the east coast of the United States and Canada there is a complete range of variation in tidal movement, and different populations of Coelopa experience varying levels of stability of their food source, and in consequence, of their population sizes. We believe that the observed patterns of mating are a consequence of exposure for countless generations to these different conditions. The life cycle of seaweed flies is typical of Dipterans. Adult C. frigid0 mate deep in the seaweed deposit usually near hotspots of active microbial decomposition. Eggs are laid in clutches of 40-100, which hatch and pass through three larval instars. The larvae ingest the decomposing slime and are nourished by the diverse microbial flora, as well as by the contents of
6
THOMAS H. DAY AND ANDR!~ s. GILBURN
ruptured algal cells that release high concentrations of the sugar mannitol (Cullen, Young, and Day, 1987).The survival of larvae depends on colonization of the gut by microbial flora, the presence of fairly high concentrations of NaCl (at least 6L), and on the presence of some constituent-possibly a secondary metabolite-of algal cells (Cullen el ul., 1987). These nutritional requirements account for the distribution of C. frigidu around the North Atlantic. Pupation occurs in the drier parts of the deposit, and the eclosing adults become sexually mature after about 18 hours. The complete life cycle takes 1-6 weeks and is extremely temperature dependent; at 26°C under controlled laboratory conditions a complete generation takes roughly 10 days. 111. POPULATION GENETICS
The amount of genetic variation in C. frigidu is typical of Diptera and of invertebrates in general; approximately one third of enzyme-determining loci are polymorphic (Collins, 1978). However, a rather more novel feature of C. frigidu is that roughly one quarter of its genome is contained within polymorphic chromosomal inversions (Philip, 1958,1966).Eight such inversion systems have been described on the five pairs of chromosomes, but one of them, the ap system on chromosome I, is extremely large and includes about 10%of all polytene bands in the genome (Aziz, 1975; Day, Dobson, Hillier, Parkin, and Clarke, 1982). Although this inversion is a complex of three overlapping inversions, natural populations segregate for only two forms-the a and p sequences. The absence of intermediates must be a consequence of the seriously unbalanced nature of crossover products. The significance of the ap inversion system is that thousands of loci cosegregate as a single Mendelian entity. This means the diverse and profound effects that the inversion has on the biology and behavior of this fly are inherited in a straightforward fashion. All natural populations sampled, however isolated, are polymorphic for the inversion; individuals are either aa or pp homokaryotypes, or are ap heterokaryotypes (Butlin, Collins, Skevington, and Day, 1982a; Day, Dawe, Dobson, and Hillier, 1983). The inversion appears to be of considerable antiquity; dating the original chromosomal mutations by measuring nucleotide sequence divergence has not been attempted. The ap inversion is associated with differences in the rate of development from egg to adult, with pps being the fastest, am the slowest, and aps intermediate (Day, Dobson, Hillier, Parkin, and Clarke, 1980). Whether this is a result of differential rates of larval feeding or of utilization of nutrients is unknown, but the consequence is that aa adults are significantly
SEXUAL SELECTION IN SEAWEED FLIES
7
larger than pp individuals, again with aps being intermediate. The genotypes differ in another important respect. Under controlled laboratory conditions there are large differences in egg-to-adult survival, with aps surviving best and aas exhibiting the lowest survival (Butlin et al., 1984; Butlin and Day, 1985a). Furthermore, this heterokaryotypic advantage is strongly density dependent, That such heterosis also occurs in natural populations can be inferred from the nearly ubiquitous heterokaryotypic excess over Hardy-Weinberg expectations (Butlin et al., 1982a; Day et al., 1983; Butlin and Day, 1989), and from the positive relationship that exists between heterokaryotypic excess and larval density (Butlin, 1983; Phillips, Leggett, Wilcockson, Day, and Arthur, 1995). Heterosis is a routine feature of chromosomal inversion polymorphisms (Dubinin and Tiniakov, 1946; Dobzhansky, 1947; Birch, 1955; da Cunha, 1955; Lewontin and White, 1960; Budnik, Brncic, and Koref-Santibanez, 1971; Anderson, Dobzhansky, Pavlovsky, Powell, and Yardley, 1975). Its relevance in the context of sexual selection is that there exist very substantial differences in fitness between individuals, and we might expect nonrandom mating behavior to have evolved that results in the production of offspring of the fittest genotypes. However, it should be noted that heterotic systems do not involve directional selection on additive genetic effects, but rather stabilizing selection and overdominance. In nongenetic parlance, the quality of a male’s genes depends on the genes carried by the interested female-it is complementarity of sperm and eggs, rather than absolute quality, that is important. inversion system in C. Much of this review is concerned with the frigida. In practice the karyotypes of individual animals have not been determined from preparations of salivary gland polytene chromosomes. It has proved much faster and more reliable to type animals on the basis of their genotype at the alcohol dehydrogenase locus (Adh).The two common alleles at the Adh locus are in complete linkage disequilibrium with the ap inversion system such that the Adh-B allele is always associated with the a sequence and Adh-D with the p sequence (Day et al., 1982). With the technique of starch gel electrophoresis a third allele, Adh-C, common in some populations, cannot be used to infer karyotypes unambiguously, since it is associated with both a and p inversions. The Adh-A, -E, and -F alleles are extremely rare in natural populations and have been routinely excluded from analyses. IV. MATINGBEHAVIOR
There have been countless descriptions of weird and elaborate courtship behaviors in animals. These have provoked speculation concerning the
8
THOMAS H. DAY AND
ANDRB s. GILBURN
adaptive significance of courtship that may involve mate attraction, species recognition, the induction of receptivity, and assessment of mate quality. However, some species such as the dung fly, Scutophugu stercoruriu (Parker, 1970a), and the house fly, Muscu domesticu (Tobin and Stoffolano, 1973), copulate with no obvious behavioral preliminaries. At least to human senses, courtship behavior in seaweed flies is particularly unelaborate (Thompson, 1951). The first impression is that the flies bump into each other, mount, and then mate. However, further study has shown that the process naturally terminates at various stages, and that there is abundant scope for mate recognition and choice. It has been routine practice in studies of the mating behavior of seaweed flies to allow a minimum of 18 hours recovery period following anesthesia with carbon dioxide. Ether is never used, particularly since etherization of Drosophilu melunogusfer has been shown to have lasting effects on their behavior (Joachim and Curtsinger, 1990). Consider first by what mechanism males and females encounter each other. Rager (1985) and Soffe (1986) conducted maze tests in which animals were given the choice of proceeding along one of two alternative routes through which flowed a countercurrent of air that had passed over either males or females. Neither males nor females showed any attraction to the opposite sex. However, both sexes were strongly attracted to decomposing seaweed. In particular, they moved against an air current that had passed M solution of 3,5 M solution of D-mannitol, or a through a dibromo, 4 hydroxybenzoic acid, both of which are present in brown algae (Rashid and Prakash, 1972). The implication is that, like many phytophagous insects (see Fraenkel, 1969;Berenbaum, 1990),both sexes are attracted to the larval food source. Mating behavior has been studied by observing and video-recording animals in small arenas and then analyzing the behavioral components during slow playback of the tapes (Day, Foster, and Engelhard, 1989). These data must obviously be interpreted with caution because mating normally takes place deep within seaweed deposits where conditions must differ in many respects. The animals move haphazardly until they are within about 2 cm of each other. The male then moves toward the other animal and mounts. If the male has mounted another male, he invariably dismounts immediately. However, the probability of a female being mounted is significantly higher than a male, suggesting that males do exercise some recognition at a distance. Furthermore, a large female is more likely to be mounted than a small one. By whatever sensory modality such recognition is mediated (perhaps vision or olfaction), it seems probable that some form of passive male choice is occurring; larger females would be more readily sensed than small ones. Females have never been seen to mount another animal.
SEXUAL SELECTION IN SEAWEED FLIES
9
Within a few seconds of a female being mounted she may curl her abdomen downward making it virtually impossible for the male to engage genitalia (Fig. 2). In addition, the female may kick with her metathoracic legs, flick her wings, and with particularly persistent males, roll completely over. This series of movements has been interpreted as the female rejection response (Day et al., 1989), and it almost invariably results in the male being unable to copulate. After varying lengths of time he usually dismounts. Very shortly after mounting, but prior to any female rejection response, the male moves forward on top of the female so that his prothoracic legs are located in the vicinity of the female’s antennae. The legs are moved up and down several times in such a way that it is probable that his sex combs-serrated, spoon-shaped structures on the inner side of the first metatarsi-rub against the female’s antennae. This action provides the opportunity for stimulatory exchange between the two animals, but its precise function has yet to be determined. Either animal could be gaining information about the other upon which the acceptance/rejection decision is based. Alternatively, the female may be informed that she is mounted by a potential mate rather than being in contact with a piece of seaweed;
FIG.2. The female rejection response in C. frigidu. The female is curling her abdomen away from the mounted male. Reproduced, with permission, from Foster (1989).
10
THOMAS H. DAY AND ANDRB S. GILBURN
these animals inhabit small crevices in compressed deposits of seaweed. Experiments involving masking of the female aristae (the terminal part of the antennae) or of the males’ sex combs indicate that females are normally in the rejection mode, and require positive stimulation to accept a male (Foster, 1989; Day et al., 1989). Whether such stimulation is involved in habitat discrimination, species recognition, or the quality assessment of conspecifics is not known. If the female does not reject, the male may nevertheless dismount. The circumstances of such dismounting have been studied (see later discussion), and it is clear that males exercise active mate choice. In the absence of both female and male rejection, the genitalia engage and they remain in copula for 30 s to about 30 min. In spite of this very variable duration of copulation, the number of eggs fertilized is not associated with copulation duration. J. J. Hewitt (unpublished results) allowed pairs a single copulation and then counted the number of fertile eggs subsequently laid. He found a bimodal distribution. For about 93% of males from a population at St. Mary’s Island, the mean number was 89 (SE = 4.6), but the remaining males fertilized distinctly fewer eggs (mean = 23, SE = 4.8). Both males and females mate repeatedly in quick succession (Pitafi, Simpson, Stephen, and Day, 1990; Pitafi, Hewitt, Gilbert, and Day, 1994; Pitafi, 1991). The mating behavior of individual males when given single females in quick succession was observed over a 90-min period. During these short bouts of mating, males copulated with a mean of 17 different females (range 9-35). Overnight each male was allowed access to 10 females, which in no case were all fertilized. This regime of observed pairings during the daytime and an excess of females available overnight was continued until the male died. Pitafi calculated that during the male’s lifetime, the mean number of females fertilized by single males was 63 (range 11108). The total number of fertilized eggs laid by each female was also counted, and from this the mean lifetime fecundity of males, as measured by the number of eggs fertilized, was estimated to be 2270 (range 13832987). Those males that apparently transferred fewer sperm at fertilization inseminated significantly more females, but exhibited similar lifetime fecundities. These results suggest there may be alternative insemination strategies among males. The majority of wild-caught males appear to transfer sufficient sperm to fertilize an average clutch of eggs, but probably not many more. About 5 1 0 % of males seem to transfer distinctly fewer sperm, but to many more females. One possible advantage of the latter strategy is that larvae from multipaternal clutches experience weaker competition. J. J. Hewitt tested this possibility by measuring the survival of mono- and multipaternal clutches, but failed to find any difference (unpublished results).
11
SEXUAL SELECTION IN SEAWEED FLIES
Whether there are genuinely alternative strategies, and if so, the nature of the benefits, are clearly questions that deserve further study. Returning to the mating behavior of Coelopu, the probability of female rejection (at least in an inbred laboratory strain) dramatically changes following a prolonged mount; 80% of mounts immediately prior to a prolonged mount ended in rejection, whereas only 20%of males were rejected if the female had accepted the previous male (Fig. 3). Copulation not only changes mating receptivity but various other aspects of behavior. The female’s rate of feeding increases, as does the rate of egg production. Virgin females oviposit a mean total of 340 ( S E = 34) eggs (obviously unfertilized), whereas females with males continuously available have a mean lifetime fecundity of 560 eggs ( S E = 49). Mated females produce significantly more clutches. These changes cannot be attributed to the transfer of substantial resources from the male (see later discussion), and so must be a consequence either of trace quantities of some substance in the seminal fluid (see Eberhard and Cordero, 1995), or of the act of copulation itself. Other species in which lifetime fecundities have been determined for both sexes have yielded similar estimates. The once commonly held belief that males have substantially greater fecundity than females-certainly
I
2
3
4
5
6
7
8
9
10
Sequence of copulations FIG. 3. Change in the rejection rate of females with increasing sexual experience. The data are based on video recordings of 68 mating trials involving two males and two females. Only the first 10 copulations (prolonged mounts) are considered. The solid bars refer to the outcome of mounts immediately preceding the lst, 2nd. 3rd. etc. copulation. The shaded bars refer to mounts immediately following copulations. Data from Pitafi (1991).
12
THOMAS H. DAY AND A N D R 6 S. GILBURN
much more than a fourfold difference-is not supported by these data. There is no doubt that in C.frigidu, as in other species (Lefevre and Jonsson, 1962; Jones, 1973; Nadel and Luck, 1985; Simmons, 1990), males can and do become exhausted of effective sperm. Dewsbury (1982) was evidently correct in pointing out that parental investment in gametes is a misleading way of comparing the reproductive potential of males and females. There is a final point concerning female fecundity. Lifetime fecundities in excess of 500 eggs were measured using a strain that had been maintained in laboratory culture for several years. Butlin and Day (1985b) obtained estimates of about 400 eggs per female for two strains that had been collected less than two years previously. In contrast, the lifetime fecundity of females recently collected from natural populations rarely exceed 200 (Shuker, unpublished results). These data suggest that the regime of laboratory culture imposes directional selection for greater female fecundity, and that substantial additive genetic variance may exist in wild populations for this character. If this is true, it would be worthwhile to identify the nature of trade-offs maintaining such variation in a character that is obviously an important component of fitness. V. SEXUAL SELECTION
For many years students of sexual selection have valued Huxley’s distinction between “intrasexual” selection, usually referring to competition between males, and “intersexual” selection, usually mediated by female mate preferences (Huxley, 1938a,b). In his excellent review, Anderson (1994) pointed out that the competition for mates is similar to competition for any other type of resource, and that the distinctions made by ecologists should be applied in the study of sexual selection. Since sexual selection is now widely recognized as involving many diverse yet interacting processes, it is useful to categorize these processes. We shall follow Anderson’s categorization and consider in turn the effects of contests, scrambles, endurance rivalry, sperm competition, and finally, and in some detail, mate choice. A. CONTESTS
Perhaps the archetypal contest is the fighting between males for access to, and control of, a harem of females such as that occurring in red deer, Cervus elephus (Clutton-Brock et ul., 1979), and elephant seals, Miroungu sp. (Deutsch, Haley, and Le Boeuf, 1990). A long-term consequence of such contests is the evolution of heavy weaponry and pronounced sexual dimorphism. However, elements of contest competition exist in species
SEXUAL SELECTION IN SEAWEED FLIES
13
with much more modest weaponry. In C.frigidu, interactions between males have not been observed in the absence of females. Nevertheless, video recordings involving one female and several males clearly indicate that contests do take place, but only after a male has mounted a female. A second male may then mount the pair and the outcome is dependent on the relative sizes of the “resident” and “challenging” male (Day et al., 1989). A large male is rarely displaced yet can readily displace a smaller male without necessarily himself remaining mounted on the female. No particular structures are involved in this contest, the outcome of which appears to be determined by general strength and wrestling ability. The role of these contests as a selective force in the field is difficult to assess. On occasions when hundreds or even thousands of adults collect at hotspots in seaweed deposits, it seems probable that mounted males are challenged in this way.
B. ENDURANCE RIVALRY AND SCRAMBLES In some species there is a discrete polymorphism in male morphology and behavior. Male Salmonid fish (Gross, 1982,1984,1985) and the bluegill sunfish, Lepomis mucrochirus (Gross, 1984), either become sexually mature early and remain small, or take longer to develop into larger males. The two types of male stably coexist in populations and neither is consistently advantaged over the other. Similar ESSs, but known to have a genetic basis, have been described in the cricket, Gryllus integer (Cade, 1979), the isopod, Purucerceis sculptu (Shuster and Wade, 1991), and in swordtail fish of the genus Xiphophorus (Ryan and Causey, 1989). Although the idea of alternative mating tactics usually refers to distinctly different morphs, it can readily be applied to continuously variable characters, as has been done in Scutophugu stercoruriu (Parker, 1970a), Centris pallida (Alcock, Jones, and Buchmann, 1977), and in the genus Punorpu (Thornhill, 1981). In each case size is the relevant character. At one extreme of the size distribution, large animals win contests and perhaps are more enduring rivals, to use Andersson’s terminology (Anderson, 1994). At the other extreme, small males win scramble competitions by being more mobile and available to mate earlier. In such a situation there is obviously the potential for the coexistence of types subject to very different selection pressures. This situation may well exist in C. frigidu. act individuals develop slowly and eclose as large adults, whereas @3s develop fast and are small. There are clear mating advantages to being large: not only do large males win contests but they also exhibit significantly greater longevity as adults (Foster, 1989). They are capable of inseminating more females under optimal conditions (Pitafi, 1991), and at times of stress when food is absent or in
14
THOMAS H. DAY AND ANDRI? S. GILBURN
short supply, they are more likely to survive to father the next generation. The observation that the frequency of the a inversion increases in tidal populations during calm summers is consistent with them being superior endurance rivals (Butlin, 1983; Butlin and Day, 1989). On the other hand, the highest frequencies of small p/3 individuals are seen in Baltic populations in which food is continuously available and in which scramble competition is more likely to be operating. In the laboratory small males appear to have a much higher mount rate than large males (Gilburn, 1992). Further support for the existence of scrambles comes from experiments conducted under relatively natural conditions (Day, Miles, Pilkington, and Butlin, 1987). Larvae were collected (from St. Mary’s Island and Whitburn) in the seaweed in which they were living and allowed to develop through to adulthood in the laboratory. Eclosing adults could mate as and when they wished, and at intervals females were removed and left to lay eggs. The Adh genotypes of the offspring and the mother were determined, and used to infer the genotype of the father. Genotypes of inferred fathers were then compared with the males available to be fathers. A complication was that the genotype frequencies of inferred fathers were modified to take into account multiple paternity and the finite number of offspring typed. The results confirmed that multiple paternity is common and that it increased with age; between the second and eighth day after eclosion started, the proportion of multiply inseminated females increased from 20 to 65%. More surprisingly, a negative relationship was found between male size and reproductive success (rs = -.570, p = -001). Under these conditions small males exhibited greater mating success than large ones. In spite of all the advantages accruing in terms of contest competition and of female choice (see later discussion), small males appeared to be sexually selected. C . SPERM COMPETITION
Sperm competition refers to the contest between gametes from different males. Its power as an evolutionary force is attested by the extraordinarily bizarre behaviors of males that can best be understood in terms of males increasing the chances of their own sperm being successful at fertilization (Parker, 1970b; Smith, 1984; Eberhard, 1985; Birkhead and Moller, 1992). These behaviors include the removal of sperm from previously mated females (see Birkhead and Hunter, 1990), plugging the female’s genital aperture, and spraying the female with “anti-aphrodisiacs” (Gilbert, 1976). The more mundane, and more common, behaviors of mate guarding and repeated inseminations can also be interpreted in terms of sperm competition (Alcock, 1994). Walker (1980) suggested that the mechanics of sperm storage within the female may influence the evolution of mating behavior.
SEXUAL SELECTION IN SEAWEED FLIES
15
Tubular spermathecae in which sperm are “stacked” and then used in reversed sequence, favor males that mate last; in contrast, spermathecae that allow free mixing of sperm favor large ejaculates and repeated inseminations. In seaweed flies the spermathecae are spherical and sperm enter and exit by the same route (Foster, 1989). The somewhat complex pattern of sperm precedence can be understood in terms of this anatomy. If two males inseminate in quick succession there is sperm mixing and the progeny are of mixed paternity. Furthermore, the paternity is associated with the relative sizes of the males (Butlin, 1983). However, if inseminations are separated by a longer time interval, there is precedence to the sperm from the later mating male (Thompson, 1951; Burnet, 1961; Butlin, 1983). From the females’ viewpoint, every clutch is likely to be of mixed paternity, but subsequent clutches may be sired by a different set of males. Nevertheless, since stored sperm remain viable throughout her life, later clutches can if necessary be fertilized by sperm acquired from the first males to have mated. From the males’ standpoint, there is selection for multiple mating and for ejaculates containing many sperm. There appears to have been a response to the former, but not the latter. There is increasing evidence that postcopulatory sperm competition may result from females exercising choice in their use of sperm from different males. In several insect species, following insemination by both conspecific and heterospecific males, sperm from conspecifics is preferentially used (Hewitt, Mason, and Nichols, 1989; Bella, Butlin, Ferris, and Hewitt, 1992; Howard and Gregory, 1993; Wade, Patterson, Chang, and Johnson, 1994), and there may even be discrimination between different conspecific males (LaMunyon and Eisner, 1993; Ward, 1993; see also Eberhard and Cordero, 1995). There is no evidence that “cryptic” female choice of this kind occurs in seaweed flies. The considerable variance in the duration of copulation may be interpreted as mate guarding. Insemination usually occurs within 60 s (Pitafi, 1991), but some males remain in copula for much longer. Whether long-copulating males gain in terms of sperm competition has not been tested. Although sperm competition usually refers to competition between males, under some circumstances we might expect it to operate within ejaculates. Heterotic systems involving chromosomal inversions provide the ideal opportunity to study intraejaculate competition. Consider, for example, an aa female inseminated by an CXPmale. The zygotes are expected to consist equally of the fittest a$ heterokaryotypes and the least fit aa homokaryotypes. Any mutation that resulted in the eggs being preferentially fertilized by P-carrying sperm would surely be positively selected. J. J. Hewitt (unpublished results) examined the progeny of such crosses
16
THOMAS H. DAY AND ANDR6 S. GILBURN
and failed to find any evidence for segregation distortion. This negative result is perhaps surprising, bearing in mind the large fitness differentials involved. However, because very few loci are expressed in sperm cells, the scope for recognition by eggs of genetic differences between sperm may be limited. D. MATECHOICE The traditional view (Darwin, 1871) that females are more choosy and males more aggressive is supported by Andersson’s review (1994) of 232 studies. In 167 of these, evidence was obtained that females exercise mate choice, whereas in only 30 were the males choosing. For many years there has been active debate over the reasons for this difference in behavior. Bateman (1948) argued that the variance in reproductive success was crucial, but parental investment (Trivers, 1972), the operational sex ratio (Emlen and Oring, 1977), and the potential rate of offspring production (Clutton-Brock and Vincent, 1991) have also been suggested as important factors. While this discussion continues, it is becoming clear that, although females very rarely compete with each other for mates, it is not uncommon for males to discriminate between females. In several species of fish, mate choice is seen in both sexes (Downhower and Brown, 1980; Berglund, Rosenqvist, and Svensson, 1986; Foote and Larkin, 1988; C8te and Hunte, 1989), but the phenomenon is certainly not limited to fish (Burley and Moran, 1979; Burley, 1986; Verrell, 1989; Brown, 1990; Houck and Reagan, 1990). In C. frigida both sexes exhibit mate choice. Because male choice presents fewer interpretational problems, it will be considered first. 1.
Male Choice
The probability that a female is mounted depends on her size, as measured by wing length (Day et d., 1989). As already discussed, this could be passive choice by the male in that large females are likely to be more readily perceived than small ones. Once mounted, the male may dismount with no apparent rejection by the female. This is also dependent on female size; dismounting from large females occurs at a significantly lower frequency. It is possible that female size is sensed during the interaction between the sex combs and the female’s antennae. Alternatively, the length of the female’s body could be assessed using some sort of calliper action involving the male’s pro- and metathoracic legs. By whatever mechanism size is measured, it would seem that the male is actively reaching a decision concerning the female’s acceptability. Whether males exhibit weaker discrimination when the variance in female size is small, and whether there
SEXUAL SELECTION IN SEAWEED FLIES
17
is any involvement of inversion karyotype of either sex (in addition to its effect on size), has not been determined. Male dismounting is influenced by a second attribute of the female, the distension of the abdomen (Pitafi et al., 1990; Pitafi, 1991). Females that have yet to lay any eggs possess a strongly convex abdomen. As females lay their first, second, and sometimes third and even fourth clutch of eggs, their abdomen becomes progressively more concave until the ventral surface is closely adpressed to the dorsal surface. While controlling for wing length, Pitafi showed that males are more likely to dismount from females with collapsed abdomens than they are from fully gravid ones. Shortly after mounting, the male’s mesothoracic legs grasp the female’s abdomen. These legs are in exactly the right position to assess the extent of ventral distention of the abdomen, perhaps using proprioreceptors located in the leg joints. It is not difficult to guess the evolutionary raison d’etre of the males’ behavior. As in many other insects (see Anderson, 1994), female size is strongly correlated with lifetime fecundity (Butlin and Day, 1985b). However, from the males’ perspective it is the size of the next clutch to be laid that is of more consequence. Clutch size is also correlated with wing length, although the strength of the association declines in successive clutches. In addition, clutch size declines with clutch order. By combining assessments of both body size and abdominal distention, the male could be gaining an accurate estimate of the number of eggs available to be fertilized. It would seem that the force driving the evolution of the males’ mating behavior is a direct advantage to the discriminating male in terms of fertility. Of course, this interpretation assumes that males do not possess an inexhaustible supply of sperm. That males can and do become impotent following repeated inseminations has been demonstrated (Pitafi et al., 1994). 2. Female Choice
When mounted by a male, females respond in one of two ways; they either accept the male and allow copulation to take place, or exhibit a rejection response and attempt to prevent copulation from occurring. One possible reason for the existence of such a rejection response is that females are expressing a mating preference. If this is true, what is the nature of the preference? Are females preferring to mate with males possessing particular characters, and, if so, what is the nature of these characters? The experiments carried out in the 1980s were aimed at identifying the subject of any preference, and yielded confusing and apparently contradictory results. Some of these problems have now been resolved but others are still not understood. At that time there were many examples of female choice in insects being based on the body size of males (McCauley and Wade, 1978; Borgia, 1981; Gwynne, 1982; Johnson, 1982; Burk and Webb,
18
THOMAS H. DAY AND ANDRE S. GILBURN
1983; Hughes and Hughes, 1985; Simmons, 1986). It has been an empirical observation that sexually selected characters are more exaggerated and more variable in the preferred sex. In a study of the variation in tail length in birds (Alatalo, Hogland, and Lundberg, 1988), sexually selected characters were usually found to have coefficients of phenotypic variation of greater than 6, while those maintained solely through the action of viability selection had coefficients of variation less than 6. In C. frigida mean male size is larger than mean female size in most, though not all, populations (Gilburn and Day, 1994b), and male size is consistently more variable than female size. The coefficient of variation for male size ranged from 8.6 to 19.4 in a study of 30 populations. Body size was from the start a strong candidate to be the preferred character, and indeed the first experiments supported this idea. Behavioral trials were conducted in which a female was given the choice of two males (Butlin, Read, and Day, 1982b: Butlin, 1983). One of the males was distinctly larger than the other; in half the trials the larger male was an aa homokaryotype, and in the others the male was a @. The inversion karyotype of the father (or in some cases, fathers) was determined by typing the offspring. The females significantly preferred the larger male, rather than either of the karyotypes. A similar experiment conducted on a mass mating basis produced the same results; females chose their mate on the basis of size rather than inversion karyotype. Butlin and Day attempted to demonstrate that such a preference also occurred in natural populations (Butlin, 1983; Day and Butlin, 1987). They collected wild adult females from St. Mary’s Island and Whitburn, and allowed them to lay eggs. The Adh genotypes of the progeny were then determined, which, knowing the mother’s genotype, allowed inference of the father’s genotype. The genotype frequencies among inferred fathers were then compared with a sample of wild collected males that were presumably available to be fathers. There was a significant excess of Adh-CD fathers over the number available to be fathers, and a deficiency of BB and BC fathers. Because CD males are relatively small, and BB and BC males relatively large, this seemed to conflict with the results from laboratory experiments. The experiment was repeated on other natural populations and also using mass mating cultures in the laboratory. Foster (1989) found that BD as well as CD males gained higher than expected mating success. He pointed out that this type of experiment in which paternity is inferred from the Adh genotypes of the mother and her offspring is flawed if females multiply mate. Foster calculated the minimum remating rate to be over 20%, and the true rate was probably much higher. (For example, a female that mated with a BB and a D D male would be scored as having mated with a single BD male.) A technique that at first promised to be a
SEXUAL SELECTION IN SEAWEED FLIES
19
powerful one to study mating in natural populations could clearly yield misleading results. A very different approach was adopted by Engelhard, Foster, and Day (1989) who used mating arenas to record on videotape the behavior of females in the presence of three males (all animals originating from St. Mary’s Island). They concluded first, that females carrying an Adh-C allele (i.e., BC, CC, or CD) exhibited a higher rate of acceptance than non-Cbearing females, and second, that it was only non-C females that showed any mating preference. The C allele appeared to confer a lack of mate discrimination. However, Gilburn (1992) reanalyzed their results, and also repeated the experiment, and showed that the non-C grouping of animals was heterogeneous. It is the DD females that have the lowest acceptance rate. A further complication in the studies on natural populations and in most of the early laboratory experiments arose from the effects of male-male competition and male mate choice. It was not until so-called “single-mount” experiments were conducted that a clearer picture of female choice was obtained (Gilburn, 1992; Gilburn, Foster, and Day, 1992). Pairs of virgin adults of standardized age were placed together in mating chambers, and the reaction of the female to the very first mount by the male observed and recorded as acceptance or rejection. The Adh genotype and size of the flies were then determined. Multiple regression analysis showed that this initial acceptance-rejection response was based on the male’s size and not at all on his Adh genotype or karyotype. Many subsequent results have confirmed this conclusion. This long saga of early experiments serves as a warning that the various components of mating behavior including female choice, male choice, and male-male competition can interact with each other in complex ways. Multiple mating is a further complication. Indeed, it remains a puzzle why some females exhibit such strong discrimination in their very first copulation, yet thereafter mate almost randomly. This would make sense if they were monogamous, but they are not. In laboratory trials females mate with several males in quick succession, and wild-caught females are often, indeed usually, inseminated by more than one male (Butlin, 1983; Day and Butlin, 1987; Foster, 1989). VI. VARIATION I N FEMALE MATING BEHAVIOR Two distinct aspects of female mating behavior can be distinguished: acceptance rate and preference. The overall acceptance rate of a group of females refers to the proportion of all mount attempts by males that result
20
THOMAS H. DAY AND
ANDRB s. GILBURN
in female acceptance. This completely disregards the sizes of males. The strength of female mating preference has been measured using logistic regression of acceptance rate on male size. The regression coefficient estimates the strength of female mating preference with respect to male size; it indicates how strongly females discriminate between males of differing size. In theory both acceptance rate and preference are characteristic of individual females: under any given circumstances a female exhibits a certain willingness to mate, and there is a distribution of probabilities that she will accept males of varying size. However, in practice they are features of groups of females because single females cannot be tested against a panel of males. The reason for this is that the probability of acceptance appears to change depending on previous mating experience. Both aspects of female behavior have been studied in natural populations widely distributed throughout the range of C. frigida. Measurements of female acceptance rate will be described before considering female preference. A. ACCERANCE RATESIN NATURAL POPULATIONS
Single-mount experiments were performed on adults collected as larvae from natural populations (Gilburn et al., 1992; Gilburn, Foster, and Day, 1993; Gilburn, Crean, and Day, 1996; Gilburn and Day, 1994a). The inversion karyotype of each female was inferred from its Adh genotype, and individuals whose karyotype could not be determined unambiguously were excluded from the analyses. Consistent differences in acceptance rate existed between inversion karyotypes (Table I). In every population the acceptance rates of act females were higher than those of aps (Wilcoxon signed rank test: z = -3.52, p < .001), which in turn were usually higher than those of pps ( z = -2.76, p < .01). These data provide a test of the Foster and Engelhard hypothesis (Foster, 1989; Engelhard el al., 1989) that Adh-C-bearing females have a higher acceptance than non-Cs (Table 11). It was found that, on average, AdhBCfemales have a similar acceptance rate to BDs, and CDs are intermediate between BDs and DDs. However, there is considerable heterogeneity between populations. Because the Adh-C allele (as defined by electrophoretic mobility in starch gels) can be associated with either form of the inversion, these estimates of acceptance rates do not distinguish between C, and C, alleles, and clearly depend on the relative frequencies of the two types in each sample. If it is true that C, is more common than C, (i.e., BCs are usually cups, and CDs often pps), these results are consistent with the acceptance rates being determined by the inversion karyotype, rather than
21
SEXUAL SELECTION IN SEAWEED FLIES
TABLE I FEMALE ACCEFTANCE RATESI N SAMPLES FROM NATURAL POPULATIONS" Female inversion karyotype Population Kampinge, Sweden Traslovslage (A), Sweden' Traslovslage (B), Sweden' Lerkil, Sweden Giant's Causeway, N. Ireland Cushendun, N. Ireland Turnley's Port, N. Ireland Portland Bill, England Wicklow, Eire Helgoland, Germany St. Mary's Island, England Shag Harbour. Canada Niarbyl, Isle of Man Petit Port, Channel Islands L'Etacq (A), Channel Islands' L'Etacq (B), Channel Islands'
CYCY
4
PP
Tidal range (m)b
0.63 0.88 0.73 0.56 0.80 0.71 0.61 0.67 0.67 0.79 0.81 0.46 0.39 0.83 0.94 0.72
0.49 0.67 0.59 0.34 0.58 0.50 0.47 0.61 0.53 0.78 0.77 0.45 0.31 0.72 0.82 0.71
0.43 0.58 0.45 0.36 0.65 0.50 0.36 0.61 0.44 0.56 0.61 0.44 0.22 0.70 0.79 0.56
0.1 0.3 0.3 0.3 1.5 1.6 1.6 1.9 2.0 2.7 4.2 4.2 4.8 9.8 9.8 9.8
' Data taken from Gilburn et al., 1992; Gilburn, Foster, and Day, 1993; Gilburn, Crean, and Day, 1996, Gilburn and Day, 1994a; and Crean and Gilburn, unpublished results. "Tidal range was calculated as the difference between mean high water and mean low water at spring tides for the nearest datum point to each sample site. The data were taken from the Admiralty Tide Tables. The populations at Traslovslage and L'Etacq were sampled on two separate occasions.
Adh genotype. They certainly do not lend support to the idea that Adh-Cbearing females exhibit higher acceptance than non-Cs.
B. STRENGTH OF PREFERENCE IN NATURAL POPULATIONS The regression coefficients of acceptance rate on male size estimate the strength of the preference, with positive values indicating a preference for large males and negative values a preference for small males (Table 111). In every population ap and pp females preferred large males, although the preference was not always statistically significant. It is noteworthy that aa females were more variable in their strengths of preference, and in four populations actually preferred small males. Furthermore, their preference is associated with a feature of the habitat from which they were collected, namely tidal range. This association was discussed by Gilburn and Day (1994a) and is confirmed by the larger data set now available. The prefer-
22
THOMAS H. DAY AND ANDR6 S. GILBURN
TABLE I1 THEACCEFTANCE RATESOF FEMALES CATEGORIZED BY ADHGENOTYPE" Adh genotype
Population
BE
EC
ED
CD
DD
Helgoland, Germany Petit Port, Channel Islands L'Etacq (A), Channel Islands L'Etacq (B), Channel Islands Portland Bill, England St. Mary's Island, England Niarbyl, Isle of Man Giant's Causeway, N. Ireland Cushendun, N. Ireland Turnley's Port, N. Ireland Wicklow, Eire Shag Harbour, Canada Mean acceptance rate
0.79 0.83 0.94 0.72 0.67 0.81 0.39 0.80 0.71 0.61 0.67 0.46
0.67 0.75 0.75 0.71
0.78 0.72 0.82 0.71 0.61 0.77 0.31
0.56 0.70 0.79 0.56 0.61 0.61 0.22 0.65
0.47 0.53 0.45
0.72 0.67 0.65 0.65 0.62 0.70 0.37 0.56 0.50 0.52 0.46 0.3 1
0.70
0.60
0.60
0.56
0.54
0.55 0.70 0.41 0.50 0.50 0.47 0.70 0.55
0.58 0.50
0.50 0.36 0.44 0.44
" Data from sources in Table 1. Swedish populations have very low frequencies of the AdhC allele and are not included.
ences of aa and c.P females are stronger in nontidal compared to tidal populations for both karyotypes (Mann-Whitney I/ test, z = -2.67, p < .Ol), whereas no significant association exists for /3P females (z = -1.33, p = .18). The interpretation of this association is far from easy because tidal range correlates with several other habitat features of seaweed flies (including the stability of the food supply and geographical location), as well as with other variables likely to be relevant to mating behavior (such as adult density and the mean and variance of male size). We shall return to this problem later.
C. GENETIC DETERMINATION OF FEMALE PREFERENCE The association between mating behavior and inversion karyotype suggests there may be genetic variation in female mating behavior. This genetic variation might reside at loci present within or near the ap inversion system, or at loci completely unlinked to the inversion. Nonrandom mating could result in linkage disequilibrium between the inversion system carrying genes determining the male trait, and those loci determining the preference. In order to understand the evolution of female mating behavior it is crucial to determine whether the loci determining preference and trait are physically linked.
23
SEXUAL SELECTION IN SEAWEED FLIES
TABLE 111 OF PREFERENCE EXHIBITED BY FEMALES IN SAMPLES FROM NATURAL STRENGTHS POPULATIONS" Female inversion karyotype Population _____
~
KLmpinge, Sweden Traslovslage (A), Sweden Trtislovslage (B), Sweden Lerkil, Sweden Helgoland, Germany Petit Port, Channel Islands L'Etacq (A), Channel Islands L'Etacq (B). Channel Islands Portland Bill, England St. Mary's Island, England Niarbyl, Isle of Man Giant's Causeway, N. Ireland Cushendun, N. Ireland Turnley's Port, N. Ireland Wicklow, Eire Shag Harbour, Canada Coefficient of variation in strength of preference
ffa
ffP
PP
+1.76 +1.81 +1.16 +3.24 +0.50 +0.38 +0.96 +0.13 +1.13 0.00 +1.71 +0.44 +0.28 +0.62 +1.26
+0.74 +0.46 +0.69 + 1.86 +0.63 +0.19 +0.40 +0.03 +1.58 +0.39 +1.74 +0.78 +0.29 +0.05 +0.41 + 1.64 0.83
~_______
+2.34
+ 13.14 +5.91 +1.31 -0.11 - 1.46 +0.36 -1.09 + 1.06 -0.19 +1.28 +0.51 +0.11 +1.33 +2.21 + 0.24 2.07
+0.78
0.82
Data from sources in Table I. Preferences are expressed as logistic regression coefficients of acceptance rate on male size. (1
Gilburn and Day (1994~)studied the inheritance of female mating behavior. Animals were collected from a Swedish population (Traslovslage) in which both acceptance rate and female preference had been found to differ significantly between the two inversion homokaryotypes. A laboratory line was established in which only two alleles were segregating: a B allele associated with the a inversion, and a C allele almost certainly associated with the p form of the inversion (as Adh-CC males were of the small size characteristic of /3p homokaryotypes). Males from this " B + C" line were paired with virgin females from a broad-based population (also from Traslovslage) and the behavior of the progeny from any BC X BD crosses typed in mating trials. Because the parents were inversion heterokaryotypes, and there is no recombination at all in males (Day et al., 1982) and none in females in the region of the cup inversion, a finding that the four inversion haplotypes are inherited intact would indicate the relevant loci are located near the inversion. Strong associations were observed between the daughters' Adh genotype and both their acceptance rate and their mating prefer-
24
THOMAS H. DAY AND ANDR6 S. GILBURN
ence (Table IV). When an (Y haplotype was inherited from one parent, daughters exhibited a stronger preference than when a P haplotype was inherited, although this was only the case when an (Y haplotype was also inherited from the other parent. These results indicate a relatively simple genetic basis to female mating behavior: the relevant genes appear to be located in the region of the CUPinversion system on chromosome I. The experiment was repeated after the B + C line and the broadbased laboratory stock had been maintained in culture for seven generations. The pattern of inheritance of acceptance rate was unchanged; act females showed the highest acceptance, PPs the lowest, and aps were intermediate, although none of the differences was significant (Table IV). However, in contrast to the earlier tests, no category of female exhibited a significant preference for large males. Either the genes for strong preference had been lost, or they were no longer being expressed. This result is strikingly similar to those obtained in the ladybird beetle, Aduliu bipunctutu. An initial female preference, apparently inherited in a simple fashion (Majerus, O'Donald, and Weir, 1982; Majerus, O'Donald, Kearns, and Ireland, 1986), was lost after laboratory culture for several generations (Kearns, Tomlinson, Veltman, and O'Donald, 1992). The reasons for this apparent instability of preferences in both Aduliu and Coelopu remain to be determined.
TABLE IV ACCEFTANCE RATESA N D PREFERENCES OF FEMALES FROM THE FIRST A N D SEVENTH OF LABORATORY CULTURE^ GENERATIONS Acceptance rates
Preference
1st gen"
7th gen"
1st gen"
7th gen"
(Adh-BE)
0.76
0.56
+ 1.09 (0.82)
( ~ / 3(Adh-BC)
0.48
0.46
(Adh-BD)
0.42
0.45
/3p (Adh-CD)
0.28
0.37
+3.46 (1.08)b p < 0.01 +0.82 (0.63) p = 0.19 +0.71 (0.48) p = 0.14 + 1.93 (0.66) p < 0.01
Karyotype (Y(Y
p = 0.23 +1.23 (0.66) p = 0.09 + 1.23 (0.61) p = 0.07 +0.80 (0.64) p = 0.28
Preferences are given as logistic regression coefficients of acceptance rate on male size (data from Gilburn and Day, 1994~). Standard errors of regression coefficients are given in parentheses. The p values indicate whether the coefficients differ significantly from zero.
SEXUAL SELECTION IN SEAWEED FLIES
25
VII. THEPREFERRED TRAIT: MALESIZE DETERMINATION OF MALESIZE A. GENETIC In a few species the preferred character in the male is known to have a relatively straightforward genetic determination. For example, most of the genetic variation in male size in the swordtail fish, Xiphophorus nigrensis, is determined by allelic variation at a single locus on the Y chromosome (Ryan and Causey, 1989). There are three classes of male size, small, intermediate, and large, which appear to have different mating strategies (Ryan and Wagner, 1987). In the marine isopod, Purucerceis sculpru, there are also three genetically distinct male morphs probably determined by allelic variation at a single autosomal locus (Shuster and Wade, 1991). The largest morph appears to gain through female choice, the intermediate males mimic females, and the smallest males mature most rapidly and attempt sneaky copulations. In spite of these examples, in most species the male trait is likely to be polygenically determined. It seems improbable that characters as complex as adult body size, color patterns, or mating calls could be predominantly encoded by a single locus in most species. The task of mapping loci that determine a metric character is a formidable one, even in species amenable to genetic analysis (Spickett, 1963; Wolstenholme and Thoday, 1963;Caligari and Mather, 1975;Shrimpton and Robertson, 1988a,b). The prospect that it will be achieved, or even attempted, in the near future for any sexually selected trait must be remote. In C. frigidu it has been clear for some time that the male trait is at least partly determined by the chromosome I inversion system. A study of three British populations revealed that male size was associated with inversion karyotype (Butlin et al., 1982b); act homokaryotypes were the largest, pps smallest, and Cvps were intermediate. A much more extensive survey of 30 populations (Gilburn and Day, 1994b) confirmed the relationship between inversion karyotype and size in all 30 populations. The influence of the inversion was quantified by Wilcockson, Crean, and Day (1995), who cultured the progeny of pairs of adults under conditions of excess food, so that the full genetic potential would be realized. About 80% of the variance in the size of fathers was explained by the father’s karyotype, and the equivalent value for sons was about 77%. This suggested that the ap inversion system is the major determinant of male size. Although the inversion is inherited in a simple Mendelian fashion, there may still be many loci located within it that influence size. If this is so, there must be strong linkage disequilibrium between the relevant alleles such that most, if not all, of the alleles for large size are located on the a form of the inversion and the “small alleles” on the p form. The number of loci involved would then be
26
THOMAS H. DAY AND
ANDRB s. GILBURN
of little consequence because the haplotypes, or coadapted gene complexes, would be subject to the same selection and evolve as if they were a single locus. B. SEXUAL DIMORPHISM IN ADULT SIZEIN NATURAL POPULATIONS For many students of sexual selection the principal challenge has been to understand the evolution of exaggerated secondary sexual characters in males. The responses to this challenge have been reviewed by Anderson (1994). In many cases the preferred trait is expressed only in males, but the very nature of some characters, including, for instance, body size, means they are present in both sexes. For such characters expressed in both sexes, but subject to sexual selection only in males, theory predicts first the evolution of differential expression in the two sexes, and second, that over considerable evolutionary time, the genetic correlation between the sexes could become negligible (Lande, 1981;Lande and Arnold, 1985). The latter prediction is exceedingly difficult to test, but there have been many studies exploring the former. A review of these studies by Pomiankowski and Mgller (1995) showed that additive genetic variation is usually higher in sexually selected traits than in traits not subject to sexual selection. Seaweed flies have provided substantial evidence supporting at least one of Lande and Arnold’s predictions. Coefficients of additive genetic variation have been estimated in samples from five natural populations ( Wilcockson et al., 1995; Day, Crean, Gilburn, Shuker, and Wilcockson, 1996). Traditionally, the additive genetic contribution to the phenotypic variance of a character has been measured by its heritability (Falconer, 1989). However, it has been pointed out by Charlesworth (1984) and Houle (1992) that a better measure, certainly for comparing different traits, is the coefficient of additive genetic variance (the square root of the additive genetic variance divided by the mean). This coefficient excludes nonadditive and environmental effects, and standardizes the variance by the mean. In recognition of the greater familiarity of heritabilities, both CVA and h2 values have been calculated. The values deriving from parent-offspring regressions measured under conditions of excess food are given in Table V. Although coefficients vary considerably between populations, there is a consistent difference between the sexes. The expression of the relevant genes is relatively strong in males, but negligible, and statistically nonsignificant, in females. There does appear to have been an uncoupling of their expression in the two sexes. If it is true that this uncoupling is a slow process (Lande, 1981), we can speculate that male size has been subject to sexual selection for a considerable time.
27
SEXUAL SELECTION IN SEAWEED FLIES
TABLE V HERITABILITIES AND COEFFICIENTS OF ADDITIVE GENETIC VARIATION FOR BODYSIZE MEASURED UNDER CONDITIONS OF EXCESS FOOD"
h2 ( S E )
CVA
Population
Location
Males
Females
Males
Females
Trasl(ivs1lge Torekov Svarte St. Mary's Isl. Niarbyl
W coast of Sweden W coast of Sweden S coast of Sweden NE coast of England Isle of Man
70.4 (7.6) 30.4 (10.9) 56.0 (15.3) 41.8 (11.8) 36.8 (12.4)
17.9 (8.1) 8.0 (10.9) -1.6 (15.7) 8.2 (14.9) 10.0 (12.1)
9.8 7.7 10.5 7.7 7.3
2.2 1.6 0.8 1.1 1.8
Data from Wilcockson ef al. (1995) and Day et al. (1996). Both h2 and CVA are expressed as percentages.
Sexual dimorphism in adult size is also observed in natural populations, though the nature of the dimorphism depends on karyotype (Gilburn and Day, 1994b). aa males were larger than aa females in all but one population (,y2 = 23.2, p < .001), whereas /3/3 males were smaller than /3/3 females in 24 of the 30 populations (x2 = 10.8, p = .001). One possible explanation for these observations is that the male trait, enlarged body size, has evolved only on the a form of the inversion, and that body size on the /3 form of the inversion is at or near its optimum as determined by natural selection alone. However, the observation that in @/3sfemales are larger than males does not necessarily mean the male trait is at its naturally selected optimum on the /3 form of the inversion. The presence of hyper- or hypoallometry also provides evidence that a character is sexually selected (Fairbairn and Preziosi, 1994). Indeed, it is probably a better indicator than sexual dimorphism when the naturally selected optima differ in the two sexes, yet the sex with the lower optimum is subject to sexual selection of an intensity that restores monomorphism. Hyperallometry for sexual size dimorphism occurs if males are larger than females and the extent of the dimorphism increases as mean size increases, assuming that males are the selected sex. Hypoallometry is observed when females are larger than males, and the extent of the dimorphism decreases as mean size increases. The two phenomena are caused by the same mechanism. Sexual selection favoring large male size in a species at its naturally selected optimum should lead t o an increase in body size in males and a correlated response in females. Female size may then move back toward its optimum as determined by natural selection, thus reducing the genetic correlation in size between the sexes (Lande and Arnold, 1985). If the male trait is produced relatively late in development, which is often the case,
28
THOMAS H. DAY AND ANDRk S. GILBURN
the expression of the trait depends on resources available toward the end of development. In other words, males are predicted to develop first to a minimum size necessary for survival and then, if conditions permit, develop further. When conditions are harsh, females and males remain at their naturally selected optima or smaller, but in more favorable conditions, the males may approach their optimum as determined by a trade-off between natural and sexual selection. In C. frigida hyperallometry is observed, but only for the aas and aps (Fig. 4). At the smallest male sizes the dimorphism is not significantly different to unity (in other words, males and females are the same size), but the dimorphism increases as size increases. Because pp homokaryotypic males are generally smaller than females, if they were responding to sexual selection, hypoallometry would be predicted. The fact that this is not seen suggests that pp males are genuinely at their optimum as determined by natural selection. A further complication in understanding the dimorphism is the existence of male mate choice. Because males prefer to mate with large females (Pitafi et al., 1990 Pitafi, 1991), the mean size of females is expected to be displaced from their viability optimum. If both sexes are subject to sexual selection for increased size, the extent of dimorphism would depend on the relative intensities of selection in the two sexes. Furthermore, the reversed dimorphism observed among pps at high larval densities may in part be a consequence of male mate preferences. C. GENETIC VARIATION I N MALESIZEI N NATURAL POPULATIONS Any indirect mechanism of sexual selection requires that the preferred character has a strong genetic determination, or more precisely, has a high coefficient of additive genetic variation (CV,). Furthermore, any good genes mechanism could operate only if the preferred character is a reliable indicator of the relevant fitness genes. In C. frigida adult size could potentially be used by females as an indicator of inversion karyotype, which is known to have a major effect on fitness. The results from laboratory crosses (see Table V) suggest that male size has a high CVA.Because virtually all of this genetic variance is attributable to the aj3 inversion (Wilcockson et al., 1995; Day et al., 1996), male size could serve as a reliable indicator of karyotype. However, these experiments were intentionally carried out at low larval densities in order that environmental influences would be minimized. The crucial issue as far as evolution is concerned is whether there are high CVAsin natural populations, or whether the genetic contribution to male size is effectively swamped by nongenetic factors.
SEXUAL SELECTION IN SEAWEED FLIES
E
1t
1.3
29
A *.
. ./
c
I
.8
6
5
1.4
7
9
8
1 0 1 1 1 2
1B
.8
6
5
I .4
2 e 0 E
3 a
Q
I .3 1.2 1.1 1
lc
8
9
1 0 1 1 1 2
C
I
5
6
.. . . *
.*
.
.9 .8
7
c
7
8
9
. 1 0 1 1 1 2
Mean size of aol males (mm) FIG. 4. Relationship between the strength of sexual dimorphism and mean size of aa males. Dimorphism values were calculated separately for (A) aa homokaryotypes, (B) ap heterokaryotypes,and (C) pp homokaryotypes.A value of LOcorresponds to sexual monomorphism; values less than 1.0 indicate reverse dimorphism, with females larger than males. Redrawn from Gilburn and Day (1994b).
30
THOMAS H. DAY AND
ANDRB s. GILBURN
CVAshave been calculated using samples collected from natural populations (Day et al., 1996). The method essentially assesses the difference in size between Adh-BB and DD homozygous males (Falconer, 1989). Unlike the coefficients deriving from parent-offspring correlation, which refer to the total additive genetic variation, these coefficients refer to the additive genetic variation attributable to the afi inversion system. However, because the a@system accounts for almost all the genetic variation, the two estimates are comparable. CVA values ranged from 0 to almost 14. It is expected that as larval density increases, competition would increase, and the effect of genetic factors would decline (Falconer, 1989). If the mean size of aa males is taken as a measure of larval density experienced by the adults in the wild, a strong association is observed between CVA and density (Fig. 5). At low densities (at which aa males are large) high values of CVA are seen, but as density increases, the CVAs fall to 0. This means that in low-density populations (which commonly experience high tidal variation), size would be a very good indicator of karyotype, but it would be completely unreliable in high-density populations.
8
9
'E:
urn O P - u
e
2
*&
4% g$! U%
3
4
Mean size of au males (mm)
Larval density
-b
FIG. 5. Relationship between additive genetic variance attributable to the ap inversion and the mean size of aa males in samples from natural populations. Data from Day er al. (1996).
31
SEXUAL SELECTION IN SEAWEED FLIES
A further insight into indicator reliability is gained by focusing on those populations from which repeat samples were collected at different times (Fig. 6). CVAvalues are strongly associated with larval density (as measured by the size of am males) and are not at all characteristic of individual populations. There is also a significant difference in the relationship between British and Swedish populations. Swedish animals are generally smaller than British ones, and the regression line (not shown) of CVAvalues against size is correspondingly displaced. In spite of males being smaller, Swedish populations exhibit the full range of CVAs. In the nine populations sampled more than once, the reliability of size as an indicator was highly variable. This means that the intensity of indirect sexual selection is far from constant, and on occasion it may be completely inoperative. In particular, at times of stress when heterosis is strongest and the fitnesses of individuals differ most, good genes sexual selection is unlikely to be a significant force of change. The efficacy of indirect sexual
0 Boul
0 .%ah
A
A 0
w
0
+ & a
+ A 0
O
SMI Whit
.
I Kamp 0
D 0 0
A-
I A A 0
A 0
Lerk
W
Sten
A 0
Tras
Yst
+
w 0
6
5
4
3
+Mean size of clcl males (mm) Larval density
-'b
FIG.6 Relationship between additive genetic variance attributable to the ap inversion and the mean size of aa males. Only those populations sampled on more than one occasion are included. The samples were collected from: Boulmer, Seahouses, St. Mary's Island, and Whitburn (all from Britain, and assigned open symbols), Bua, KBmpinge, Lerkil, Steninge, Traslovslage, and Ystad (all from Sweden, and assigned closed symbols). Data from Day et al. (1996).
32
THOMAS H. DAY AND ANDRk S. GILBURN
selection as an evolutionary force must depend on larval densities averaged over many generations. D. EVOLUTION OF THE MALEPREFERRED TRAIT Why should the male preferred trait evolve only on one form of the inversion? The presence of additive genetic variation in the male trait within the region of the inversion and the lack of recombination between the two inversion sequences enable the male trait to evolve independently on each form. This explains how male size could vary between karyotypes, but not why this has happened. One of two other conditions is required. First, the cost of carrying the male trait might differ between the two inversion homokaryotypes. In a model, small differences resulted in the fixation of a trait on one form of an inversion and its loss on the other (A. S. Gilburn, unpublished results). Alternatively, if the two sets of homokaryotypic females express different mating preferences, the male trait could evolve independently on each form of the inversion by maximizing the chances of being chosen by females of the opposite homokaryotype. By indicating their karyotype, males could father fitter offspring. The former mechanism is the simpler, and perhaps the more likely, explanation for the evolution of the association between inversion karyotype and male size in C. frigida. VIII. EVOLUTION OF FEMALE MATEPREFERENCES In many species the force driving the evolution of female mating preferences is uncontroversial; these are species in which there is male parental investment. Females are expected to evolve mating preferences for those males willing to invest most in their offspring. The types of male parental investment that have been shown to be the target of female mating preferences vary from the size of nuptial gifts presented by hanging-flies and scorpion flies (Thornhill, 1976,1979,1980,1981; Alcock, 1979) and the size of spermatophores in katydids (Gwynne, 1982) to the quality of a male’s territory, as seen in green frogs (Wells, 1977), red-winged blackbirds (Searcy, 1979), and pied flycatchers (Read, 1986; Alatalo, Lundberg, and Glynn, 1986). In many other species the selective advantage to mate choice is much less obvious. These species have no male parental investment, and the females apparently gain nothing other than genetic material from the male. The evolution of female preferences in such species has been the subject of much controversy, with numerous alternative mechanisms being suggested
SEXUAL SELECTION IN SEAWEED FLIES
33
(reviewed by Kirkpatrick and Ryan, 1991), and has been termed the “paradox of the lek” (Borgia, 1979),although the problem is certainly not limited to lekking species. The mechanisms by which the mating preference for large male size in C. frigidu may have evolved are considered in the next section. A. POSSIBLE MECHANISMS The expression of a female mating preference in non-resource-based mating systems could provide a direct benefit to the choosing female (direct sexual selection), or it could increase the fitness of the offspring (indirect sexual selection). Many models of direct sexual selection have been developed, including avoidance of hybridization with heterospecific males (Fisher, 1930; Dobzhansky, 1951)’ avoidance of inbreeding (Parker, 1983) or of males with low sperm fertility (Grafen, 1990), lowering of search costs for mates (Kirkpatrick, 1987), increased fecundity, avoidance of disease or parasite transmission (Borgia and Collis, 1990), and pleiotropic effects of other characters involving innate biases in the females’ sensory system (Parker, 1983; Ryan and Rand, 1990; Enquist and Arak, 1993). There is empirical evidence for the operation of some of these mechanisms of direct sexual selection. There are at least two examples of females gaining an increase in fecundity through the expression of a female mating preference in species with non-resource-based mating systems. Females of the cockroach, Nuuphoetu cinereu, prefer to mate with males that produce offspring with fast development (Moore, 1994);females mating with such males gain an advantage, as they shorten the time between clutches. In the Neotropical frog, Ololygon rubru, the ratio of male to female size is associated with the fertilization rate of eggs (Bourne, 1993). Females prefer to mate with males of an optimum size and thereby maximize their fecundity. One possible example of a female mating preference that has evolved in response to the avoidance of disease transmission is in the guppy, Poecilia reticuluta. Females prefer to mate with males with the brightest carotenoid spots, which are known to become paler when the males are infected with a monogenean parasite (Houde and Torio, 1992). However, it is as yet unclear whether females are avoiding infection themselves, or whether the benefit accrues as a result of producing offspring more likely to be resistant to parasites. A similar situation exists in the satin bowerbird, Ptilonorhynchus violuceus, in which females prefer to mate with disease-free males having brighter plumage coloration (Borgia and Collis, 1989). Perhaps the best example of a preference evolving through pleiotropy, as a side effect of another trait, is the preference for males with lowfrequency calls in the Tdngara frog, Physuluemus pusfulosus. In a cross-
34
THOMAS H. DAY AND ANDRB S. GILBURN
species study Ryan, Fox, Wilcynski, and Rand (1990) showed that females of a closely related species, P. colorudorum, can exhibit the preference, yet males of this species do not express the mating call. They inferred that the preference predates the male trait. Females expressing the preference have a bias in their basilar papilla tuning, which is present in both species, and it therefore appears that the preference may have evolved as a side effect of the evolution of the basilar papillae. Alleles for female mating preference may evolve through indirect sexual selection by becoming genetically correlated with another trait subject to positive selection. Models of indirect sexual selection are of two types, depending on the nature of this other trait. In the Fisherian process (Fisher, 1930;Lande, 1981;Kirkpatrick, 1982;Pomiankowski, lwasa, and Nee, 1991), the other trait is the male preferred character itself. The two characters become genetically correlated as a result of nonrandom mating; females carrying a mating preference preferentially mate with males carrying the preferred trait. In the “good genes” or viability indicator mechanism (Zahavi, 1975; Bell, 1978; Hamilton and Zuk, 1982; Andersson, 1982, 1986; Kirkpatrick, 1986; Pomiankowski, 1987; Iwasa, Pomiankowski, and Nee, 1991), the other trait is a viability trait that is assumed to affect the expression of the male trait. Nonrandom mating with respect to the male preferred trait also results in nonrandom mating with respect to the viability trait. The preference and the viability trait consequently become genetically correlated. There is considerable empirical evidence in favor of the operation of the viability indicator mechanism (reviewed by Johnstone, 1995). These empirical tests have sought correlations between the male preferred trait and male viability. In the three-spined stickleback (Gasterosteus aculeatus) femalesprefer to mate with males with the highest intensity of red coloration (Milinski and Bakker, 1990). The males that appear in best condition, as estimated from a standard fisheries measure, also possess the highest intensity of red coloration. Infection with parasites reduced the red coloration, and it was suggested that the extent of redness was being used by females to indicate the presence of parasite-resistance genes in the males. Similarly, Mgller (1990) found a correlation between the level of mite infestation and tail length in male barn swallows, Hirundo rusrica. Cross-fostering studies demonstrated that the offspring of some pairs show much higher mite resistance than the offspring of other pairs present in the same nest. Males with long-tailed fathers have been shown to have a higher survival rate (Mgller, 1994a). Females prefer to mate with long-tailed males and also prefer males with the most symmetrical tails (Mgller, 1994b). Tail symmetry was found to correlate positively with survival in males. Norris (1993) showed in the great tit (Purus major) that the viability of male offspring
SEXUAL SELECTION IN SEAWEED FLIES
35
is positively correlated with the male preferred trait carried by their father. Females prefer to mate with males possessing large black breast stripes. Norris performed a cross-fostering experiment to obtain an estimate of the unbiased heritability of breast strip size, and found a substantial heritable component to the variation in offspring breast stripe size. In addition, male offspring survival was positively correlated with the size of the breast stripe of their fathers. A similar correlation between offspring viability and male trait size has been found in pheasants (Phasianus colchicus) in which females prefer to mate with males carrying long spurs. Using DNA fingerprinting to determine paternity, von Schantz, Grahn, and Goransson (1994) revealed a positive correlation between the survival rate to adulthood of offspring and the spur length of the father. Female cockroaches (Nauphoeta cinerea) prefer to mate with males that produce offspring with shorter development (Moore, 1994). Development time was found to have a heritable component and to be positively genetically correlated with male attractiveness. Offspring of preferred fathers gained an advantage because fast development leads to earlier sexual maturity. Empirical evidence consistent with the Fisherian process is less abundant. Female mating preference and the preferred male trait are by their very nature sex limited, which makes direct identification of the relevant alleles in the same individual extremely difficult. Consequently, studies of the Fisher process seeking positive genetic correlations have used less direct methods (reviewed by Bakker and Pomiankowski, 1995). One has been to measure the preferred trait in fathers, and the preferences of their daughters. For example, Moore (1989) found that socially dominant males of the cockroach, Nauphoeta cinerea, produced female offspring with mating preferences for socially dominant males. Social dominance in N. cinerea is determined by olfactory differences in the males. In a similar study of the three-spined stickleback, Gasterosteus aculeatus, Bakker (1993) crossed females with males possessing varying amounts of red coloration (the preferred trait), and found a positive correlation between the strength of preference for redness in daughters and the extent of redness in their fathers. A positive correlation between daughters’ preference and sons’ redness was also observed. A second approach has been to look for correlated responses to selection. Wilkinson and Reillo (1994) selected for long and short eye span in the stalk-eyed fly, Cyrtodiopsis dalmanni, in which wild females prefer to mate with males having long eye spans. Female mating preference showed a correlated response to selection in the short eye span line, although preference did not increase in the line selected for long eye span. In a similar experiment using the guppy, Poecilia reticulata, selection to increase or reduce the extent of orange coloration resulted in a correlated response in female preference (Houde, 1994). Females from
36
THOMAS H. DAY AND A N D R 6 S. GILBURN
the lines in which orange coloration in males had increased showed stronger preferences for orange males than did females from the line in which orange coloration had reduced. These studies demonstrate that indirect selection involving good genes or the Fisher process could account for the evolution of female mating preferences. However, it is important to note that these mechanisms, as well as those involving direct selection, are not mutually exclusive. This is to be expected because in any species in which female choice maximizes offspring viability (in other words, good genes sexual selection occurs), nonrandom mating is likely to lead to the development of a genetic correlation between the preference and the preferred trait, assuming, of course, that there is additive genetic variation in both characters. The successful operation of a good genes mechanism also requires additive genetic variation in both characters, so the Fisher process is likely to accompany good genes sexual selection. Distinguishing the relative importance of two mechanisms will surely require a knowledge of the genetics of the preferred trait. If it is determined by genes that affect the trait alone, the Fisher process is likely to be operating, whereas if all the additive genetic variation in the trait is attributable to pleiotropic effects of viability genes, then good genes sexual selection is likely to predominate. B. INDIRECT SEXUAL SELECTION IN COELOPA FRIGIDA
Both forms of indirect sexual selection, the Fisher process and the viability indicator mechanism, could be playing a role in the maintenance of the female mating preference for large male size in C. frigida. Both require additive genetic variation to be present in the preference and the preferred trait, and the establishment of a genetic correlation between them. It has been shown that there is substantial additive genetic variation in the male trait due to the inversion system (Wilcockson et al., 1995; Day er al., 1996), and that differences in female mating behavior are associated with the same inversion system (Gilburn et al., 1993, 1996; Gilburn and Day, 1994a). At least some of these differences appear to be heritable (Gilburn and Day, 1994c;A. S. Gilburn, unpublished results). In most models of the Fisher process the loci determining the preference and the trait are assumed to be unlinked (Lande, 1981; Kirkpatrick, 1982; Pomiankowski et al., 1991), and the genetic correlation between the two characters is assumed to develop as a result of nonrandom mating. Because in seaweed flies both characters are associated with the Olp inversion system, physical linkage rather than nonrandom mating could generate genetic correlations. Theoretical support for this idea was provided by Trickett and Butlin (1994) who added an inversion system to the Kirkpatrick model (1982) of the
SEXUAL SELECTION IN SEAWEED FLIES
37
Fisher process. The absence of recombination allowed much stronger genetic correlations to develop than when the loci are unlinked. The presence of an inversion system in C.frigida, carrying additive genetic variation in both the preference and the trait, could generate very strong positive genetic correlations that could fuel the Fisher process. Just as the prerequisite conditions for the operation of the Fisher process exist in C. frigidu, so do they for “good genes” sexual selection. The existence of the ap inversion system allows the evolution of a slightly unusual kind of viability indicator mechanism. Given the very large viability differences between individuals of different inversion karyotypes, homokaryotypic females could maximize their offspring viability by mating disassortatively (aa X pp crosses would produce progeny with very high survival). This differs from traditional ideas of good genes sexual selection in that female seaweed flies are expected to mate not with males of the highest fitness (aps), but with males carrying genes that complement their own. Females could potentially increase their chances of mating with a male of suitable karyotype by exercising a mating preference based on size. Such an indicator system would work because the crp inversion is the major genetic determinant of male body size. Trickett and Butlin’s (1994) model of sexual selection incorporating an inversion system was based on a haploid organism. Although the effect of physical linkage between the preference and preferred character could be determined, any effects of heterosis could not be studied. Their model was limited to the Fisher process alone. In order to assess the potential good genes effects of heterosis on the evolution of preferences, a diploid model was constructed (Gilburn and Day, 1996a). This suggested that when indirect sexual selection alone operates the only type of mating preferences likely to evolve are disassortative. Under most conditions simulated, the advantages accruing from good genes sexual selection are greater than those from Fisherian sexual selection. If the strength of heterosis is weak, a strong assortative mating preference can evolve through the Fisher process, but this results in the rapid loss of the inversion polymorphism. Consequently, it seems unlikely that Fisherian selection acting alone can be responsible for the evolution of mating preferences in C. frigida because the inversion is polymorphic in all populations sampled. If indirect sexual selection alone has shaped the evolution of mating behavior, all homokaryotype females are predicted to mate disassortatively with respect to the ap inversion system. The pattern of mating behavior has now been studied in samples from many wild populations (Gilburn et al., 1992, 1993, 1996; Gilburn and Day, 1994a). Fully disassortative mating was found in only 4 of them; in an additional 12 samples all females preferred large males (see Table 111). In
38
THOMAS H. DAY AND ANDRJ? S. GILBURN
all populations the pp females preferred large males and consequently tended to mate with aa males; they were mating disassortatively. These females presumably gain a good genes advantage through the production of heterokaryotype offspring, and a Fisherian advantage through the production of large male offspring (which, overall, is the preferred type in all populations). In contrast, aa females varied considerably between populations in their preferences. In some they preferred small males, although significantly so in only one, and in others they exhibited extremely strong preferences for large males. In one of these populations, Trasl6vslage (Gilburn et d.,1993), the aa females expressed a significantly stronger preference for large males than the @ females. As the a form of the inversion is associated with large male size, it appears that a positive genetic correlation exists in this population between the female mating preference and the male preferred trait. The preferences at Traslovslage were measured on two occasions separated by six months and both samples revealed a positive genetic correlation between preference and trait; the Fisherian process may well have been operating in this population throughout this period. In summary, Fisherian and good genes sexual selection could account for the preferences exhibited by p/3 females; the two processes would operate in concert. However, Fisherian selection on aa females is predicted to render the inversion monomorphic, which has occurred in none of the many populations studied, and under good genes selection they are expected to prefer small males, which they do (at least significantly so) in only one population. There appear to be serious difficulties understanding how indirect sexual selection operating in isolation could account for the observed patterns of mating of aa females. These difficulties have led us to consider the role of other mechanisms promoting the evolution of female mate preferences. In particular, a model was constructed that explored whether some form of direct selection could account for the assortative mating usually exhibited by aa females (Gilburn and Day, 1996a). The model incorporated a cost to mating with any male, and a cost to rejecting-a cost that increases as male size increases. Not surprising, /3p females gained an advantage through both indirect and direct sexual selection by expressing a preference for large male size. In contrast, direct and indirect selection would act in opposition for aa females. If aa females mate only with the largest males they would be limiting their mating costs (direct sexual selection), but at the expense of producing offspring of low viability (indirect sexual selection). Likewise, if they prefer small males they produce highly fit offspring, but would increase their own costs, as the rejection of large males carries a higher cost than does rejection of small ones.
SEXUAL SELECTION IN SEAWEED FLIES
39
The relative strengths of indirect and direct sexual selection determine what mate preferences are expressed by aa females. If the relative costs and benefits of direct and indirect selection differ widely between populations, the preferences of aa females could also vary considerably; PP females are predicted to express a more uniform preference because they are advantaged through both direct and indirect selection. This is consistent with the observation that aa females are more variable in their preference. C. DIRECT SEXUAL SELECTION
What possible direct advantages might accrue from preferring large males or, alternatively, from rejecting small ones? There are, of course, diverse ways in which females might benefit, but no formal rationale exists for identifying the mechanism, or mechanisms, operating in any particular species. In C. frigidu several types of potential benefits have been studied. The first potential direct advantage studied was of a resource-based system comparable to that described in mecopterans (Thornhill, 1976), lepidopterans (Boggs and Watt, 1981), neuropterans (Hayashi, 1993)’ and many orthopteran species (Bowen, Codd, and Gwynne, 1984; Gwynne, 1984,1990; Gwynne and Simmons, 1990; Pardo, Lopez-Leon, Hewitt, and Camacho, 1995). Male seaweed flies do not appear to transfer a large spermatophore or any type of nuptial gift during copulation (Pitafi, 1991). The possibility that the semen contains nutritious components was tested by comparing the weight loss of virgin males with that of males allowed repeated matings (Pitafi et al., 1994). No difference was observed. The only potential cost to males was a modest (13%) reduction in the longevity of mating males compared with virgins, an observation consistent with the findings from many other insect species (Ridley, 1988). Furthermore, analyses of the longevities, weight changes, and fecundities of females yielded no evidence for the transfer of nutrients during copulation (Pitafi, 1991). Clearly, our confidence in excluding a resource-based system in Coelopa is limited by the sensitivity of the weighing machine used, which could detect changes of about 1% of the adults’ weight. A second possibility is that females, by choosing, maximize the cost effectiveness of searching for a mate. Video recordings of mate trials gave no indication that females do any searching at all, and even males recognize potential mates only when they are extremely close to them (Day et al., 1989). The avoidance of mating with small males could potentially benefit females if a copulation reduces the chances of subsequent copulations being successful and small males transfer fewer sperm than large ones. However, females have been observed to remate many times in quick succession
40
THOMAS H. DAY AND ANDRB S . GILBURN
(Pitafi, 1991). Furthermore, the pattern of sperm utilization indicates sperm mixing, unless inseminations are well separated temporally, in which case there is precedence to those sperm in the most recent insemination. This does not suggest that male behavior has evolved in response to competition from earlier or later mating males. Perhaps females avoid mating with small males because they are more often parasitized or infected with pathogens. In natural populations adult seaweed flies commonly carry a mite, Thinoseiusfucicolu, which is detrimental to cultures maintained in the laboratory (Butlin, 1983). However, these mites are phoretic and very quickly detach themselves from flies when decomposing seaweed is available. Because females mate only in the presence of seaweed, it seems improbable (and has never been observed) that mites transfer to the female during mating. Sexual transmission of pathogenic microorganisms certainly cannot be excluded, but would be an effective force of sexual selection only if small males pose a greater risk to females than do large ones. The suggestion that courtship behavior and female choice have evolved as a species-isolating mechanism is an old one (Huxley, 1942; Mayr, 1942). It is possible that females, by avoiding matings with small males, reduce the chances of producing sterile hybrid offspring? C.frigidu and C. pilipes occur sympatrically throughout much of their range (see Fig. l), and males do occasionally mount females of the “wrong” species (Leggett, 1993). Nevertheless, there are several observations that argue against the preference in C. frigidu having evolved through reinforcement. First, when males do mount females of the wrong species they invariably dismount almost immediately; there would appear to be no selection favoring a speciesspecific female mating preference. Second, the distributions of male sizes are very similar in the two species (although the variance in C. pilipes is smaller), yet both types of female exhibit strikingly similar preferences for large males (A. S . Gilburn and T. H. Day, unpublished results). Finally, the expectation that preferences might be strongest where the species occur together is not supported. Swedish and North American populations in which C. pifipes has never been found express preferences just as strong as, if not stronger than, sympatric populations. If C. frigidu populations are categorized according to whether they occur sympatrically with C. pifipes or not, there is no difference in the strength of preference in the two groups. The close coexistence of the two species over much of their range does not seem to have influenced the evolution of female choice. Let us consider one further mechanism by which female choice could directly benefit the female, one that concerns female fecundity. It is possible to propose many ways in which the exercise of a preference results in increased fecundity, even in non-resource-based systems. For example, the
SEXUAL SELECTION IN SEAWEED FLIES
41
interclutch interval in ovoviviparous cockroaches is reduced by females preferring certain males (Moore, 1994). In the frog species Physaluemus pustulosus and Ololygon rubra, a preference for optimally sized males results in greater female fertility (Ryan, 1985; Bourne, 1993). In C. frigida Pitafi (1991) showed that mated females of a highly inbred strain produce significantly more eggs than do unmated ones. This admits the possibility that the acceleration in egg production, apparently triggered by mating, is dependent on male size. This question has yet to be answered.
D. PLEIOTROPY Most models of intersexual selection assume that the evolution of female preference is driven by an advantage either directly or indirectly associated with the male character. A distinctly different idea is that preferences could have originated as a side effect of the evolution of some other character unconnected with mate choice. The preference is supposed to have evolved in response to natural selection, perhaps in the context of feeding or predator avoidance, and only later to have become subject to sexual selection as a pleiotropic effect. Two forms of pleiotropy have been proposed: sexual selection for sensory exploitation (Ryan and Rand, 1990; Shaw, 1995) and sexual selection as a side effect of natural selection on female mating behavior (Rowe, Arnqvist, Sih, and Kruppa, 1994).The sensory exploitation hypothesis envisions that biases in the females’ sensory system result in particular male phenotypes being preferentially selected as mates. Frogs of the genus Physalaemus (Ryan et al., 1990; Ryan and Rand, 1990) and swordtail fish of the Xiphophorus species group (Ryan and Wagner, 1987; Ryan and Causey, 1989) provide the most thoroughly studied examples of this type of selection. The crucial feature of pleiotropy is that the female preference predates the evolution of the preferred trait (Basolo, 1990, 1995; Ryan and Rand, 1990). This is not an easy prediction to test without suitable material to construct phylogenetic relationships, and even then the evidence relies on historical inference of evolutionary pathways. In seaweed flies, both C. frigida and C. pilipes express a preference for large males and this preference is mediated through the same mechanism, a physical rejection response during which males attempting to mount are forcibly removed by the female. These observations are consistent with the preference having originated before the species diverged, but with data on only two species available, convergent evolution cannot be excluded. Sexual selection as a side effect of natural selection on female mating behavior is certainly a possible explanation for the evolution of the female preference in C. frigida. Rowe et al. (1994) have suggested that female
42
THOMAS H. DAY AND ANDRG S. GILBURN
preferences for large males in water striders have evolved as an adaptive reluctance to mate. In this species mating is known to be costly, and nonrandom expression of the rejection response results in an apparent preference for large males. The water strider mating system is similar to that of C. frigida. Female seaweed flies perform a rejection response that results in selection for large males and there also appears to be a cost to mating that manifests itself in terms of reduced longevity (Pitafi, 1991; A. S. Gilburn and T. H. Day, unpublished results). A similar cost has been described in Drosophila melanogaster (Partridge, Green, and Fowler, 1987; Fowler and Partridge, 1989). If the cost of rejecting small males in Coelopa is lower than the cost of mating with them, nonrandom expression of the rejection response would create an apparent mating preference for large males, as is seen in water striders. OF PREFERENCES I N DIFFERENT POPULATIONS E. EVOLUTION
Why does the strength and direction of the preferences of cra females vary between different populations, and why is this apparently associated with tidal range? Variation in the strength of direct or indirect sexual selection between populations could be reflected in the preferences exhibited by crcr females. Nothing is known about variation in the strength of direct sexual selection between different populations; however, the benefit from good genes selection will be limited through the strength of the association between male size and inversion karyotype. Although in the laboratory most of the variation in male size can be attributed to inversion karyotype (Wilcockson et al., 1995; Day et al., 1996), this association becomes weaker in natural populations because of an increase in environmental variation in male size. If the coefficient of additive genetic variation (CV,) in size varies between populations, then the potential benefits of good genes sexual selection should also vary. In populations in which virtually all the variance in male size is environmental, mating disassortatively with respect to size is unlikely to produce fitter offspring. As the proportion of the variance in male size explained by the a@ inversion system increases, so the advantage gained from good genes selection increases. Samples from natural populations were used to calculate both the CVA in male size and the strength of female mating preferences (Gilburn and Day, 1996b). A significant association exists (Fig. 7) between the preference of aa females and the CVA for male size attributable to the inversion (F = 6.6; df = 1, 11; p = .026, using log[preference + 21). In populations in which the CVA is high, aa females are likely to exhibit a preference for small male size. In other words, when the association between male size and inversion karyotype is the strongest, aa females actually avoid mating
43
SEXUAL SELECTION IN SEAWEED FLIES
$
1.2
’
1.0
.
0
0
0
-0.2 . 0
-0.4 3
5
7
9
11
13
Coefficient of additive genetic variation of male size FIG.7. Relationship between the mean preference of aa females and the coefficient of additive genetic variance (CV,) in 14 natural populations. Preferences were transformed as log(preference + 2). CV, values refer to the variance attributable to the cxp inversion system. Data from Gilburn and Day (1996b).
with large males. This may be because under these circumstances a preference for large males would probably result in mating with ( ~males, a and so result in low viability offspring. In these populations the good genes cost of mating with the “wrong” type of male may outweigh any advantage gained through direct sexual selection. The good genes cost to any (Y(Y females preferring large males may account for the increased acceptance rates generally exhibited by (YQ females. The greater willingness to mate associated with the (Y form of the inversion may therefore be a response to selection reducing preference for large male size. The association seen between the preference of aa females and CVAin male size could also offer an explanation for the relationship between their preference and tidal range (Gilburn and Day, 1994a). Using an enlarged data set, a very strong association was seen when the preference of ( ~ a females was regressed on tidal range ( F = 20.8; df = 1, 12; p < . 0 1 , using logtpreference + 21 and log[tidal range + 11); CVAin male size attributable to the inversion was also positively correlated with tidal range ( r = .67; p = .011, using log[tidal range + 11). When the preference of aa females was regressed on tidal range after variation associated with CVA in male
44
THOMAS H. DAY AND ANDRC S. GILBURN
size had first been removed, the association remained significant (F = 8.3; df = 1, 10; p = .016). This suggests that some, though not all, of the variation in aa preferences is due to the CVAin male size, but that additional variation due to tidal range does exist. A further complication is that tidal range and geographical location are themselves highly correlated: all nontidal populations studied were Swedish, and all tidal populations were located outside Sweden. With the data set available it was not possible to separate the confounded effects of these two variables; when either was removed (having first removed the effects of CVA in male size), aa preference was no longer associated with the other variable. However, suggestive evidence was found of an association within the non-Swedish samples (F = 5.5; df = 1, 7; p = .052). It appears that tidal range may have some effect on the preference of aa females independently of variation in the CVA in the male preferred trait, at least in western Europe. How might tidal range influence female mating preferences? Clearly, tides influence the supply of food. Nontidal populations are effectively continuous cultures, while those subject to tidal conditions usually experience regular replacement of seaweed but also periods when there is none at all. The presence or absence of seaweed must also affect oviposition sites; tidal populations sometimes lack suitable sites, which may in turn affect mating behavior. How the availability of seaweed influences the forces of sexual selection remains a problem to be solved. IX. DISCUSSION We are familiar with the idea that natural selection is a very heterogeneous process. Parasites, predators, inability to withstand climatic conditions, and shortage of essential resources are just some of the causes of selective mortality. Likewise, organisms may exhibit differential reproduction for a multitude of reasons. While it has long been recognized that the forces of sexual selection are diverse, the work on seaweed flies illustrates well that they may operate together and interact in complex ways. It is to be hoped that experimental and field studies, like those reviewed here, will stimulate theoreticians to explore the interactions between different forms of sexual selection. It was Fisher (1930) who first explicitly suggested that natural (viability) selection may act as a brake on sexual selection. This has usually been considered in the context of long-term evolutionary change, with viability selection constraining the evolution of exaggerated male traits. However, animals do not live in habitats that are constant, either spatially or temporally. We should therefore not expect every individual to be subject to the
SEXUAL SELECTION IN SEAWEED FLIES
45
same, constant forces of sexual selection. The demonstration that seaweed flies in British and Scandinavian populations may be subject to very different types of sexual selection reinforces this expectation. Furthermore, successive generations of flies living in the same place may experience selection of very different intensities. Such variation in selection will make the task of modeling what is going on even more difficult. We are still some way from understanding the interaction between habitat differences in C. frigidu and differences in female preferences. The problem is likely to be a general one. Multiple associations are observed between, on one hand, the strength of female preference, the mean value of the preferred character (male size), and the coefficients of additive genetic variation in male size (i.e., “reliability of the indicator”), and on the other, larval density, food stability, synchrony of generations, and physical factors such as tidal range. The existence of such a maze of associated variables is a problem familiar to ecologists, but it does make the job of working out a chain of causality extremely difficult. Our best guess at the moment (and it is little more than that) is that tidal range is the principal determinant of food availability. This in turn determines the synchrony of generations, the profile of adults available to mate, and the additive genetic variation in the male trait. How mate availability and genetic variation influence the type and intensity of sexual selection is not yet clear. It is perhaps significant that those laboratory experiments in which the available adults were determined by the natural eclosion sequence yielded results apparently contrary to all other controlled experiments; small males exhibited high mating success. Clearly, more work is needed to clarify the influence of population synchrony. What does appear to be true is that the additive genetic variation in the male trait is associated with the pattern of female mating preferences observed in a population. Perhaps the Holy Grail in the field of sexual selection is the identification of the genes responsible for female mate choice. In at least one respect seaweed flies would seem to be good experimental animals. Differences in female preferences exist that are associated with a chromosomal inversion system. Although such large blocks of genes have a substantial effect on the phenotype and segregate in a straightforward Mendelian fashion, the lack of recombination between alternative inversion sequences will surely hamper detailed genetic analyses. However, there is a more immediate problem to solve. Why are preferences unstable during prolonged laboratory culture? This may not be some esoteric genetical problem restricted to C. frigidu, as a strikingly similar phenomenon has been described in the ladybird, Aduliu bipuncruru (Majerus et ul., 1982,1986; Kearns et ul., 1992). One category of explanation involves changes in allele frequencies. Perhaps culture under unnatural conditions results in mate discrimination being
46
THOMAS H. DAY AND A N D R ~s. GILBURN
disfavored. If this is true, then identifying the mechanism of selection occurring in the laboratory is likely to shed light on the nature of selection in wild populations. Another possibility is that the loci determining female mating behavior are closely linked to those determining the preferred trait. In laboratory culture recombinants may not be selected out, or may even be positively favored. Again, identifying the nature of the selection will surely be informative. The implication of both these mechanisms is that the relevant alleles (or haplotypes) are, generation by generation, being actively maintained by selection, and that laboratory culture somehow alters the selection regime. This would mean that allele frequencies in natural populations are fluid and able to respond fairly rapidly to changes in the environment. An alternative explanation for the apparent instability of mating preferences involves changes in the additive genetic variation in the male trait. It is facile to point out that if all males were identical there could be no scope for the expression of any female preference. In general, the effects of female discrimination necessarily depend on the variation in the males available. However, there may be more subtle interactions between preferences and male variation. As the additive genetic variance in male size increases, so does the scope for disassortment in mating, and the consequent advantages to be gained from good genes selection. A further possibility is that preferences depend on the physiological condition of females. Age and sexual experience are known to be important, but developmental history and nutritional status may also be relevant. It is a biological inevitability that larval densities affect the additive genetic variation among males, but they may also influence female preferences. The observed associations between preference, preferred trait, and environmental factors may be a consequence of nongenetic effects on the expression of female preferences. Over the last 20 years or so of work on sexual selection there has been a shift in emphasis. The peacock’s tail was the paradigm-by what mechanism have male traits become so extravagantly exaggerated? More recently the focus of interest has moved to female preferences, and numerous models have been proposed describing how they may have evolved. Usually each model has explored the features of one particular mechanism, and it has been easy to get the impression that our task, as students of sexual selection, is to decide which mechanism is the correct one. The principal lesson learned from the seaweed fly work is that many types of sexual selection operate simultaneously and that female choice has evolved under the influence of several forces. On reflection this conclusion is hardly surprising. Because abundant additive genetic variation in the preferred trait is now recognized as the norm (Pomiankowski and Mdller, 1995), choosy females are expected to produce
SEXUAL SELECTION IN SEAWEED FLIES
47
sexy sons, and Fisherian sexual selection (though not necessarily the runaway process) is inevitable. When we tested for a positive genetic correlation between the preference and the preferred trait, it was present in the population at Traslovslage, suggesting that the Fisher process can operate in natural populations. However, the Fisher process probably plays only a relatively minor role in shaping the evolution of preferences in C. frigidu because the effects of the Fisher process on overall offspring fitness are likely to be swamped by the heterotic effects of the inversion (Gilburn and Day, 1996a). If the Fisher process was unchecked by good genes selection, it would result in the eventual loss of the inversion polymorphism, which evidently has not occurred in any of the populations studied. If additive genetic variation in the male character results from the pleiotropic effects of viability genes, then good genes selection could also operate. As many sexually selected traits are exaggerated and therefore costly to produce, variation in viability genes may well influence the ability of males to successfully produce the trait. This particularly applies to species in which the preference is based on size because genes affecting viability are especially likely to have pleiotropic effects on body size. The existence of the heterotic system in C. frigida means that the potential benefits of good genes selection are extremely large in comparison with other species not polymorphic for chromosomal inversions. In spite of the near certainty that indirect sexual selection occurs, models of the C. frigidu system suggest that offspring benefits are not sufficient to account for the evolution of female discrimination seen in natural populations. Some form of direct selection must operate as well. The problem in identifying forces of direct selection is that there are diverse ways in which the female may benefit by the exercise of choice. Females may be avoiding matings with males of the wrong species or males carrying pathogens or parasites. One form of direct selection for which there is growing support is sexual selection for sensory exploitation (reviewed by Shaw, 1995). Ryan and Rand (1990) suggested that some male secondary sexual characters have evolved because of biases in the females’ sensory system. This is essentially an extension of Parker’s (1983) theory of passive choice; females simply move toward the most intense source of the conspecific cue, rather than actively rejecting particular males. Alternatively, sexual selection could be a side effect of natural selection on female mating behavior. In various species of water striders Rowe et ul. (1994) showed that adaptive female reluctance to mate incidentally selects for certain male phenotypes. Thus, preferences may evolve not only through biases in the females’ sensory system, but through biases in the rejection response. The striking similarities in the mating behavior of sea-
48
THOMAS H. DAY AND ANDRfi S. GILBURN
weed flies and water striders suggest this type of direct selection also occurs in Coelopa. It is now very clear that the mating behavior of seaweed flies is subject to a cocktail of selective forces. The relative strengths of Fisherian, good genes, and direct sexual selection may well vary between populations, generating the differences in female preferences between natural populations and between karyotypes within populations. In addition, sexual selection very much depends on environmental factors, and so is far from constant temporally. We are familiar with this in the context of viability selection; we should no longer think of sexual selection as a single constant force.
X. SUMMARY Seaweed flies (Coelopa frigida) exhibit no dramatic visual or acoustic displays during courtship, yet are subject to many types of sexual selection. Intrasexual selection occurs both in the form of contests between males and in terms of endurance rivalry, and both sexes exercise mate discrimination on the basis of body size. Male mate choice is unproblematical; it has almost certainly evolved in response to selection for increased fertility, as males prefer to mate with large, gravid females. In contrast, studies on female mate choice have revealed a complex interaction between preference and preferred character (male size), both of which are genetically polymorphic, and several environmental variables. The genetic determination of male size is associated with the a@ inversion system on chromosome I; aa homokaryotypes are large, pps small, and aps intermediate. Laboratory crosses suggest the genes affecting female preference are also associated with this inversion, and in at least some natural populations, a correlation exists between the two sets of genes. A further feature of the ap system is that it is heterotic in natural populations--cup heterokaryotypes exhibit higher egg-to-adult viability than do (YQ or pp homokaryotypes. Mating trials indicate that pp females usually prefer to mate with large a-cu males, and by doing so produce offspring of high viability and increased attractiveness; “good genes” selection and the Fisher process are operating in concert. In contrast, the preferences of aa females vary from population to population. In Scandinavian populations subject to no tidal variation and consistently high larval densities, aa females behave in the same way as pps and prefer large males. They gain a Fisherian advantage but at the expense of producing low viability progeny. In the strongly tidal conditions common in Britain, a-cufemales either show
SEXUAL SELECTION IN SEAWEED FLIES
49
little preference, or mate disassortatively with small /3/3 males. They are mating in a good genes manner, yet incur a Fisherian disadvantage. Females may also benefit themselves by exercising choice. Several possible types of direct selection have been studied. There does not appear to be any transfer of gifts or nutrients from the male during copulation; mating in seaweed flies is not a resource-based system. Neither does the female rejection response appear to have evolved as a species-isolatingmechanism. However, preferences may have evolved as an adaptive reluctance to mate. Females rejecting rather than accepting small males exhibit increased longevity, and so, by exercising a preference, they avoid or reduce a cost to mating. It would therefore appear that the Fisher process, good genes selection, as well as direct selection all play a role in the maintenance of female preferences. Furthermore, these forces of selection do not operate equally on all females in a population, and they vary in intensity, depending on environmental factors such as tidal conditions, food availability, and larval density. We discuss the mechanisms by which genes and the environment have molded the mating behavior of seaweed flies.
Acknowledgments Most of the work on seaweed flies reviewed here was supported by grants and studentships from the Natural Environment and Science and Engineering Research Councils. Their help is gratefully acknowledged. We are also grateful to The0 Bakker, Caroline Crean, Margaret Leggett, Michael Ritchie. and David Shuker for valuable comments on the manuscript.
References Alatalo, R. V., Hogland, J., and Lundberg, A. (1988). Patterns of variation in tail ornament size in birds. Biol. J. Linn. SOC.34, 363-374. Alatalo, R. V., Lundberg, A., and Glynn, C. (1986). Female pied flycatchers choose territory quality and not male characteristics. Nature (London) 323, 152-153. Alcock, J. (1979). Selective mate choice by females of Harpobittacus australis. Psyche 86, 213-217. Alcock, J. (1994). Postinsemination associations between males and females in insects-The mate-guarding hypothesis. Ann. Rev. Entomol. 39, 1-21. Alcock, J., Jones, C. E., and Buchmann, S. L. (1977). Male mating strategies in the bee Centris pallida Fox (Hymenoptera: Anthophoridae). Am. Nut. 111, 145-155. Anderson, W., Dobzhansky, T., Pavlovsky, O., Powell, J., and Yardley, D. (1975). Genetics of natural populations. XLII. Three decades of genetic change in Drosophifa pseudoobscura. Evolution 29, 24-36. Anderson, M. B. (1982). Sexual selection, natural selection and quality advertisement. Biol. J. Linn. Soc. 17, 375-393.
50
THOMAS H. DAY AND ANDRB S. GILBURN
Anderson, M. B. (1986). Evolution of condition-dependent sex ornaments and mating preferences: Sexual selection on viability differences. Evolution 40, 804-816. Anderson, M. (1994). “Sexual Selection.” Princeton University Press, Princeton, New Jersey. Aziz, J. B. (1975). Investigations into chromosomes I , 2 and 3 of Coelopa frigida (Fab.). Unpublished doctoral dissertation, University of Newcastle upon Tyne, United Kingdom. Bakker, T. C. M. (1993). Positive genetic correlation between female preference and preferred male ornament in sticklebacks. Nature (London) 363,255-257. Bakker, T . C. M., and Pomiankowski, A. (1995). The genetic basis of female mate preferences. J. Evol. Biol. 8, 129-171. Basolo, A. L. (1990). Female preference predates the evolution of the sword in swordtail fish. Science 250,808-809. Basolo, A. L. (1995). Phylogenetic evidence for the role of a pre-existing bias in sexual selection. Proc. R. SOC. Lond. B 259,307-311. Bateman, A. J. (1948). Intra-sexual selection in Drosophila. Heredity 2, 349-368. Bell, G. (1978). The handicap principle in sexual selection. Evolution 32, 872-885. Bella, J. L., Butlin, R. K., Ferris, C., and Hewitt, G. M. (1992). Asymmetrical homogamy and unequal sex ratio from reciprocal mating-order crosses between Chorthippus parallelus subspecies. Heredity 68, 345-352. Berenbaum, M. R. (1990). Coevolution between herbivorous insects and plants: tempo and orchestration. In “Insect Life Cycles. Genetics, Evolution and Co-ordination” (F. Gilbert, ed.), pp. 87-99. Springer-Verlag, London. Berglund, A,, Rosenqvist, G., and Svensson, I. (1986). Mate choice, fecundity and sexual dimorphism in two pipefish species (Syngnathidae). Behav. Ecol. Sociobiol. 19,301 -307. Birch, L. C. (1955). Selection in Drosophila pseudoobscura in relation to crowding. Evolution 9,389-399. Birkhead, T. R., and Hunter, F. M. (1990). Mechanisms of sperm competition. Trends Ecol. E d 5,48-52. Birkhead, T. R., and MBller, A. P. (1992). ‘Sperm Competition in Birds: Evolutionary Causes and Consequences.” Academic Press, London. Boggs, C. L., and Watt, W. B. (1981). Population structure of pierid butterflies. IV. Genetic and physiological investment in offspring by male Colias. Oecologia 50, 320-324. Borgia, G. (1979). Sexual selection and the evolution of mating systems. In “Sexual Selection and Reproductive Competition in Insects” (M. S. Blum and N. A. Blum, eds.), pp. 19-80. Academic Press, New York. Borgia, G. (1981). Mate selection in the fly Scatophaga stercoraria: Female choice in a malecontrolled system. Anim. Behav. 29, 71-80. Borgia, G.. and Collis, K. (1989). Female choice for parasite-free male satin bowerbirds and the evolution of bright male plumage. Behav. Ecol. Sociobiol. 25, 44-453. Borgia, G., and Collis, K. (1990). Parasites and bright male plumage in the satin bowerbird (Ptilonorhynchus violaceus). Am. Zool. 30,279-285. Bourne, G. R. (1993). Proximate costs and benefits of mate acquisition at leks of the frog, Ololygon rubra. Anim. Behav. 45, 1051-1059. Bowen, B. J., Codd, C. G., and Gwynne, D. T. (1984). The katydid spermatophore (Orthoptera: Tettigoniidae): Male nutritional investment and its fate in the mated female. Austral. J. ZOO^. 32,23-31. Brown, W. D. (1990). Size-assortative mating in the blister beetle (Lytta magisrer) (Coleoptera: Meloidae) is due to male and female preference for large mates. Anirn. Behav. 40,901-909. Budnik, M., Brncic, D., and Koref-Santibanez, S. (1971). The effects of crowding on chromosomal polymorphism of Drosophila pavani. Evolution 25,410-419.
SEXUAL SELECTION IN SEAWEED FLIES
51
Burk, T., and Webb, J. C. (1983). Effect of male size on calling propensity, song parameters, and mating success in Caribbean fruit flies, Anastrepha suspensa (Loew) (Diptera: Tephritidae). Ann. Entomol. SOC. Am. 76, 678-682. Burley, N. (1986). Sexual selection for aesthetic traits in species with biparental care. Am. Nat. 127,415-445. Burley, N., and Moran, N. (1979). The significance of age and reproductive experience in the mate preferences of feral pigeons, Columba livia. Anim. Behav. 27,686-698. Burnet, B. (1961). On the distribution of recessive embryonic lethals in a natural population of Coelopa frigida. Genet. Res., Camh. 2, 249-271. Butlin, R. K. (1983). The maintenance of an inversion polymorphism in Coelopa frigida. Unpublished doctoral dissertation, University of Nottingham, United Kingdom. Butlin, R. K., Collins, P. M., and Day, T. H. (1984). The effect of larval density on an inversion polymorphism in the seaweed fly, Coelopa frigida. Heredity 52, 415-423. Butlin. R. K., Collins, P. M., Skevington, S. J., and Day, T. H. (1982a). Genetic variation at the alcohol dehydrogenase locus in natural populations of the seaweed fly, Coelopa frigida. Heredity 48, 45-55. Butlin, R. K., and Day, T. H. (1985a). Genic and karyotypic selection on an inversion polymorphism in the seaweed fly, Coelopa frigida. Heredity 54, 267-214. Butlin, R. K., and Day, T. H. (1985b). Adult size, longevity and fecundity in the seaweed fly, Coelopa frigida. Heredity 54, 107-1 10. Butlin, R. K., and Day, T. H. (1989). Environmental correlates of inversion frequencies in natural populations of seaweed flies (Coelopa frigida). Heredity 62,223-232. Butlin, R. K., Read, I. L., and Day, T. H. (1982b). The effects of a chromosomal inversion on adult size and male mating success in the seaweed fly, Coelopa frigida. Heredity 49,51-62. Cade. W. (1979). The evolution of alternative male reproductive strategies in field crickets. In “Sexual Selection and Reproductive Competition in Insects” (M. S. Blum and N. A. Blum, eds.), pp. 343-380. Academic Press, New York. Caligari, P. D. S., and Mather, K. (1975). Genotype-environment interaction. 111. Interactions in Drosophila melanogaster. Proc. R. SOC.Lond. E 208, 163-187. Charlesworth, B. (1984).The evolutionary genetics of life histories. In “Evolutionary Ecology” (B. Shorrocks, ed.), pp. 117-133. Blackwell, Oxford. Clutton-Brock, T. H., Albon, S. D., Gibson, R. M., and Guiness, F. E. (1979). The logical stag: Adaptive aspects of fighting in red deer (Cervus elephas L.). Anim. Behav. 27,211-225. Clutton-Brock, T. H., and Vincent, A. C. J. (1991). Sexual selection and the potential reproductive rates of males and females. Nature (London) 351, 58-60. Collins, P. M. (1978). Studies on geneticpolymorphism in Coelopa frigida. Unpublished doctoral dissertation, University of Nottingham, United Kingdom. CBte, 1. M., and Hunte, W. (1989). Male and female mate choice in the redlip blenny: Why bigger is better. Anim. Behav. 38, 78-88. Cullen, S. J., Young, A. M., and Day, T. H. (1987). Dietary requirements of seaweed flies (Coelopa frigida). Estur. Coast. Shelf Sci. 24, 701-710. da Cunha, A. B. (1955). Chromosomal polymorphism in the Diptera. Adv. Genet. 7,93-138. Darwin, C. (1859). “On the Origin of Species by Means of Natural Selection.” Murray, London. Darwin, C. (1871). “The Descent of Man, and Selection in Relation to Sex.” Murray, London. Day, T. H., and Butlin, R. K. (1987). Non-random mating in natural populations of the seaweed fly, Coelopa frigida. Heredity 58, 213-220. Day, T. H., Crean, C. S., Gilburn, A. S., Shuker, D. M., and Wilcockson, R. W. (1996). Sexual selection in seaweed flies: Genetic variation in male size and its reliability as an indicator in natural populations. Proc. R. SOC.Lond. B. 263, 1127-1134.
52
THOMAS H. DAY AND ANDRB S. GlLBURN
Day, T. H., Dawe, C., Dobson, T., and Hillier, P. C. (1983). A chromosomal inversion polymorphism in Scandinavian populations of the seaweed fly, Coelopa frigida. Hereditas 99, 135-145. Day, T. H., Dobson, T., Hillier, P. C., Parkin, D. T., and Clarke, B. (1980). Different rates of development associated with the alcohol dehydrogenase locus in the seaweed fly, Coelopa frigida. Heredity 44, 321-326. Day, T. H., Dobson, T., Hillier, P. C., Parkin, D. T., and Clarke, B. (1982). Non-random association of enzyme and chromosomal polymorphisms in the seaweed fly, Coelopa frigida. Heredity 48, 35-44. Day, T. H., Foster, S., and Engelhard, G. (1989). Mating behavior in seaweed flies (Coelopa frigida). J. Insect Eehav. 3, 105-120. Day, T. H., Miles, S., Pilkington, M. D., and Butlin, R. K. (1987). Differential mating success in populations of seaweed flies (Coelopa frigida). Heredity 58, 203-212. Deutsch, C. J., Haley, M. P., and Le Boeuf, B. J. (1990). Reproductive effort of male northern elephant seals: Estimates from male loss. Can. J. Zool. 68, 2580-2593. Dewsbury, D. A. (1982). Ejaculate cost and male choice. Am. Nut. 119,601-610. Dobson, T. (1973). Coelopa frigida (Fabricius) and Coelopa pilipes (Halliday) (Diptera, Coelopidae): A study of two sympatric species. Unpublished master’s thesis, University of Newcastle upon Tyne, United Kingdom. Dobson, T. (1974a). Studies on the biology of the kelp fly Coelopa in Great Britain. J . Nut. Hist. 8, 155-177. Dobson, T. (1974b). Mechanisms controlling species composition in natural populations of the seaweed fly Coelopa. J. Nut. Hist. 8, 653-673. Dobson, T. (1976). Seaweed flies (Diptera: Coelopidae, etc.). In “Marine Insects” (L. Cheng, ed.), pp. 447-463. North-Holland Publishing Co., Amsterdam. Dobzhansky, T. (1947). Genetics of natural populations XIV. A response of certain gene arrangements in the third chromosome of Drosophila pseudoobscura to natural selection. Genetics 32, 142-160. Dobzhansky, T. (1951). “Genetics and the Origin of Species.” Columbia University Press, New York. Downhower, J. F., and Brown, L. (1980). Mate preferences of female mottled sculpins, Cotrus bairdi. Anim. Eehav. 28,728-734. Dubinin, N. P., and Tiniakov, G. G. (1946). Structural chromosome variability in urban and rural populations of Drosophila funebris. Am. Nut. 80, 393-396. Eberhard, W. G. (1985). “Sexual Selection and Animal Genitalia.” Harvard University Press, Cambridge, Massachusetts. Eberhard, W. G., and Cordero, C. (1995). Sexual selection by cryptic female choice on male seminal products-a new bridge between sexual selection and reproductive physiology. Trends Ecol. Evol. 10,493-496. Egglishaw, H. J. (1960). Studies on the family Coelopidae (Diptera). Trans. R. En!. Soc. Lond. 112, 109-140. E d e n , S. T., and Oring, L. W. (1977). Ecology, sexual selection, and the evolution of mating systems. Science 197,215-223. Engelhard, G., Foster, S. P.. and Day, T. H. (1989). Genetic differences in mating success and female choice in seaweed flies (Coelopa frigida). Heredity 62, 123-131. Enquist, M., and Arak, A. (1993). Selection of exaggerated male traits by female aesthetic senses. Nature (London) 361,446-448. Fairbairn, D. J., and Preziosi, R. F. (1994). Sexual selection and the evolution of allometry for sexual size dimorphism in the water strider. Aquarius remigis. Am. Nut. 144,101-118. Falconer, D. S. (1989). “Introduction to Quantitative Genetics.” Longman Scientific, Harlow, United Kingdom.
SEXUAL SELECTION IN SEAWEED FLIES
53
Fisher, R. A. (1930). “The Genetical Theory of Natural Selection.” Clarendon Press, Oxford. Foote, C. J., and Larkin, P. A. (1988). The role of male choice in the assortative mating of anadromous and non-anadromous sockeye salmon (Oncorhynchus nerka). Behaviour 106,43-62. Foster, S. P. (1989). Selection concerning reproduction in the seaweed fly, Coelopa frigida. Unpublished doctoral dissertation, University of Nottingham, United Kingdom. Fowler, K., and Partridge, L. (1989). A cost to mating in female fruitflies. Nature (London) 338,760-761. Fraenkel, G. (1969). Evaluation of our thoughts on secondary plant substances. Entomol. EXP.Appl. 12,473-486. Gilbert, L. E. (1976). Postmating female odor in Heliconius butterflies: A male-contributed antiaphrodisiac? Science 193,419-420. Gilburn, A. S. (1992). Sexual selection by female choice in the seaweed fry, Coelopa frigida. Unpublished doctoral dissertation, University of Nottingham, United Kingdom. Gilburn, A. S., Crean, C. S., and Day, T. H. (1996). Sexual selection in natural populations of seaweed flies: variation in the offspring fitness of females carrying different inversion karyotypes. Proc. R. SOC. Lond. B 263,249-256. Gilburn, A. S., and Day, T. H. (1994a). Evolution of female choice in seaweed flies: Fisherian and good genes mechanisms operate in different populations. Proc. R. Soc. Lond. B 255, 159-165. Gilburn, A. S.. and Day, T. H. (1994b). Sexual dimorphism, sexual selection and the ap chromosomal inversion polymorphism in the seaweed fly, Coelopa frigida. Proc. R. SOC. Lond. B 257,303-309. Gilburn, A. S., and Day, T. H. (1994~).The inheritance of female mating behaviour in the seaweed fly, Coelopa frigida. Genet. Res., Camb. 64, 19-25. Gilburn, A. S., and Day, T. H. (1996a). The evolution of female choice when the preference and the preferred trait are linked to the same inversion system. Heredity 76, 19-27. Gilburn, A. S., and Day, T. H. (submitted). The evolution of female mating preferences and male preferred traits in seaweed flies. Proc. R. SOC. Lond. B. Gilburn, A. S., Foster, S. P., and Day, T. H. (1992). Female mating preference for large size in Coelopa frigida. Heredity 69, 209-216. Gilburn. A. S., Foster, S. P., and Day, T. H. (1993). Genetic correlation between a female mating preference and the male preferred character in the seaweed fly, Coelopa frigida. Evolution 47, 1788-1795. Grafen, A. (1987). Measuring sexual selection: why bother? I n “Sexual Selection: Testing the Alternatives” ( J . W. Bradbury and M. B. Anderson, eds.), pp. 221-233. Wiley, Chichester, United Kingdom. Grafen, A. (1990). Sexual selection unhandicapped by the Fisher process. J. Theoret. Biol. 144,473-516. Gross, M. R. (1982). Sneakers, satellites and parentals: Polymorphic mating strategies in North American sunfishes. Z. Tierpsychol. 60, 1-26. Gross, M. R. (1984). Sunfish, salmon, and the evolution of alternative reproductive strategies and tactics in fishes. I n “Fish Reproduction: Strategies and Tactics” (G. W. Potts and R. J. Wooton, eds.), pp. 55-75. Academic Press, London. Gross, M. R. (1985). Disruptive selection for alternative life histories in salmon. Nature (London) 313,47-48. Gwynne, D. T. (1982). Mate selection by female katydids (Orthoptera: Tettigoniidae, Conocephalus nigropleurum). Anim. Behav. 30,734-738. Gwynne, D. T. (1984). Courtship feeding increases female reproductive success in bushcrickets. Nature (London) 307,361-363.
54
THOMAS H. DAY AND ANDR6 S. GILBURN
Gwynne, D. T. (1990). The katydid spermatophore: evolution of a parental investment. In “The Tettigoniidae: Biology, Systematics and Evolution” (W. J. Bailey and D. C. F. Rentz, eds.), pp. 27-40. Springer-Verlag, Berlin. Gwynne, D. T., and Simmons, L. W. (1990). Experimental reversal of courtship roles in an insect. Nature (London) 346, 171-174. Hamilton, W. D., and Zuk, M. (1982). Heritable true fitness and bright birds: A role for parasites. Science 218, 384-387. Hayashi, F. (1993). Male mating costs in two insect species (Protohermes, Megaloptera) that produce large spermatophores. Anim. Behav. 45,343-349. Hewitt, G. M.. Mason, P., and Nichols, R. A. (1989). Sperm precedence and homogamy across a hybrid zone in the alpine grasshopper, Podisma pedestris. Heredity 62, 343-353. Houck, L. D., and Reagan, N. L. (1990). Male courtship pheromones increase female receptivity in a plethodontid salamander. Anim. Behav. 39, 729-734. Houde, A. E. (1994). Effect of artificial selection on male colour patterns on mating preferences of female guppies. Proc. R. SOC.Lond. B 256, 125-130. Houde, A. E., and Torio, A. J. (1992). Effect of parasitic infection on male color pattern and female choice in guppies. Behav. Ecol. 3,346-351. Houle, D. (1992). Comparing evolvability and variability of quantitative traits. Generics 130, 195-204. Howard. D. J., and Gregory, P. G. (1993). Post-insemination signalling systems and reinforcement. Phil. Trans. R. SOC. Lond. B 340,231-236. Hughes, A. L., and Hughes, M. K. (1985). Female choice of mates in a polygynous insect, the whitespotted sawyer Monochamus scutellatus. Behav. Ecol. Sociohiol. 17, 385-387. Huxley, J. S. (1938a). Darwin’s theory of sexual selection and the data subsumed by it, in the light of recent research. Am. Nut. 72,416-433. Huxley, J. S. (1938b). The present standing of the theory of sexual selection. In “Evolution (Essays on Aspects of Evolutionary Biology)” (G. R. De Beer, ed.), pp. 11-41. Oxford University Press, Oxford. Huxley, J. S. (1942). “Evolution, the Modern Synthesis.” Allen and Unwin. London. Iwasa, Y.,Pomiankowski, A,, and Nee, S. (1991). The evolution of costly mate preferences 11. The “handicap” principle. Evolution 45, 1431-1442. Joachim, D., and Curtsinger, J. W. (1990). Genotype and anesthetic determine mate choice in Drosophila melanogasrer. Eehav. Genet. 20, 73-79. Johnson, L. K. (1982). Sexual selection in a brentid weevil. Evolution 36, 251-262. Johnstone, R. A. (1995). Sexual selection, honest advertisement and the handicap principle: Reviewing the evidence. Biol. Rev. 70, 1-65. Jones, J. C. (1973). A study of the fecundity of male Aedes aegypti. J. Insect Physiol. 19,435-439. Kearns, P. W. E., Tomlinson, I. P. M., Veltman, C. J.. and O’Donald, P. (1992). Non-random mating in Adalia hipunctata. 11. Further tests for female mating preference. Heredity 68,385-389. Kirkpatrick, M. (1982). Sexual selection and the evolution of female choice. Evolution 36,l-12. Kirkpatrick, M. (1986). Sexual selection and cycling parasites: a simulation of Hamilton’s hypothesis. J. Theoret. Biol. 119,263-271. Kirkpatrick, M. (1987). The evolutionary forces acting on female mating preferences in polygynous animals. In “Sexual Selection: Testing the Alternatives” (J. W. Bradbury and M. B. Anderson, eds.), pp. 41-53. Wiley, Chichester, United Kingdom. Kirkpatrick, M., and Ryan, M. J. (1991). The evolution of mating preferences and the paradox of the lek. Nature (London) 350,33-38. LaMunyon, C . W., and Eisner, T. (1993). Postcopulatory sexual selection in an arctiid moth (Utetheisa ornatrix). Proc. Natl. Acad. Sci. U.S.A. 90,4689-4692.
SEXUAL SELECTION IN SEAWEED FLIES
55
Lande, R. (1981). Models of speciation by sexual selection on polygenic traits. Proc. Natl. Acad. Sci. U.S.A. 78, 3721-3725. Lande, R., and Arnold, S. J. (1985). Evolution of mating preference and sexual dimorphism. J. Theoret. Eiol. 117,651 -664. Lefevre, G. L. Jr., and Jonsson, V. B. (1962). Sperm transfer, storage, displacement, and utilization in Drosophila melanogaster. Genetics 47, 1719-1736. Leggett, M. C. (1993). Competitive interactions between seaweed flies. Unpublished doctoral dissertation, University of Nottingham, United Kingdom. Leggett, M. C., Wilcockson, R. W., Day, T. H., Phillips, D. S., and Arthur, W. (1996). The genetic effects of competition in seaweed flies. Eiol. J. Linn. SOC.57, 1-11. Lewontin, R. C., and White, M. J. D. (1960). Interaction between inversion polymorphism of two chromosome pairs in the grasshopper, Moraba scurra. Evolution 14, 116-129. Majerus, M. E. N., O’Donald, P., Kearns, P. W. E., and Ireland, N. (1986). Genetics and evolution of female choice. Nature (London) 321, 164-167. Majerus, M. E. N., O’Donald, P., and Weir, J. (1982). Female mating preference is genetic. Nature (London) 350, 521-523. Maynard-Smith, J. (1958). Sexual selection. In “A Century of Darwin” (S. A. Barnett, ed.), pp. 231-244. Heinemann, London. Mayr, E. (1942). “Systematics and the Origin of Species.” Columbia University Press, New York. Mayr, E. (1963). “Animal Species and Evolution.” Harvard University Press, Cambridge, Massachusetts. McAlpine, D. K. (1991). Review of the Australian kelp flies (Diptera: Coelopidae). Systemat. Entomol. 16, 29-84. McCauley, D. E., and Wade, M. J. (1978). Female choice and the mating structure of a natural population of the soldier beetle Chauliognarhus pennsylvanicus. Evolution 32, 771-775. Milinski, M., and Bakker, T. C. M. (1990). Female sticklebacks use male coloration in mate choice and hence avoid parasitized males. Nature (London) 344,330-333. Mbller, A. P. (1990). Effects of a haematophagous mite on the barn swallow (Hirundo rusrica): A test of the Hamilton and Zuk hypothesis. Evolution 44,771-784. Moller, A. P. (1994a). Male ornament size as a reliable cue to enhanced offspring viability in the barn swallow. Proc. Nafl. Acad. Sci. U.S.A. 91,6929-6932. Moiler, A. P. (1994b). Sexual selection in the barn swallow (Hirundo rusrica). IV. Patterns of fluctuating asymmetry and selection against asymmetry. Evolution 48,658-670. Moore, A. J. (1989). Sexual selection in Nauphoeta cinerea: Inherited mating preference? Eehav. Genet. 19,717-724. Moore, A. J . (1994). Genetic evidence for the good genes process of sexual selection. Eehav. Ecol. Sociobiol. 35, 235-241. Nadel, H., and Luck, R. F. (1985). Span of female emergence and male sperm depletion in the female-biased, quasi-gregarious parasitoid, Pachycrepoideus vindemiae (Hymenoptera: Pteromalidae). Ann. Entomol. SOC. Am. 78,410-414. Norris, K. (1993). Heritable variation in a plumage indicator of viability in male great tits, Parus major. Nature (London) 362,537-539. O’Donald, P. (1962). The theory of sexual selection. Heredity 22,541-552. Pardo, M. C., Lopez-Leon, M. D., Hewitt, G. M., and Camacho, J. P. M. (1995). Female fitness is increased by frequent mating in grasshoppers. Heredity 74, 654-660. Parker, G. A. (1970a). The reproductive behaviour and the nature of sexual selection in Scatophaga stercoraria L. IV. Epigamic recognition and competition between males for the possession of females. Behaviour 37, 113-139.
56
THOMAS H. DAY AND
ANDRB s. GILBURN
Parker, G. A. (1970b). Sperm competition and its evolutionary consequences in the insects. Eiol. Rev. 45,525-567. Parker, G. A. (1983). Mate quality and mating decisions. In “Mate Choice” (P. Bateson, ed.), pp. 141-164. Cambridge University Press, Cambridge, United Kingdom. Partridge, L., Green, A., and Fowler, K. (1987). Effects of egg production and of exposure to males on female survival in Drosophila rnelanogaster. J. Insect Physiol. 33,745-749. Philip, U. (1958). Genetics and cytology of the seaweed fly, Coelopa frigida. Proc. 10th Int. Congr. Genet., Montreal 2, 217. Philip, U. (1966). Chromosomal polymorphism in Coelopa frigida. Heredity 21, 167-168. Phillips, D. S., Leggett, M. C., Wilcockson, R. W., Day, T. H., and Arthur, W. (1995). Coexistence of competing species of seaweed flies: the role of temperature. Ecol. Entomol. 20365-74. Pitafi, K. D. (1991). Male mate choice in seaweedflies, Coelopa frigida. Unpublished doctoral dissertation, University of Nottingham, United Kingdom. Pitafi, K. D., Hewitt, J. J., Gilbert, F. S., and Day, T. H. (1994). Male reproductive incompetence, fertility, and the cost of mating in male seaweed flies. Genet. (Life Sci. A d v . ) 13,9-18. Pitafi, K. D., Simpson, R., Stephen, J. J., and Day, T. H. (1990). Adult size and mate choice in seaweed flies. Heredity 65, 191-197. Pomiankowski, A. (1987). The handicap principle does work-Sometimes. Proc. R. SOC.Lond. E 231,123-145. Pomiankowski, A., Iwasa, Y.,and Nee, S. (1991). The evolution of costly preferences I. Fisher and biased mutation. Evolution 45, 1422-1430. Pomiankowski, A., and M~ller,A. P. (1995). A resolution of the lek paradox. Proc. R. Soc. Lond. E 260,21-29. Rager, F. J. (1985). Investigations on chemical attraction in seaweed flies (Coelopa frigida). Unpublished bachelor’s thesis, University of Nottingham, United Kingdom. Rashid, M. A., and Prakash, A. (1972). Chemical characteristics of humic compounds isolated from some decomposed marine algae. J. Fish Res. Board Can. 29,55-60. Read, R. F. (1986). Female mate choice in pied flycatchers: An answer and a problem. Trends Ecol. Evol. 1, 85. Ridley, M. (1988). Mating frequency and fecundity in insects. Biol. Rev. 63, 509-549. Rowe, L., Arnqvist, G.,Sih, A., and Kruppa, J. J. (1994). Sexual conflict and the evolutionary ecology of mating patterns: Water striders as a model system. Trends Ecol. Evol. 9, 289-293. Ryan, M. J. (1985). “The Tlingara Frog.” Chicago University Press, Chicago. Ryan, M. J., and Causey, B. A. (1989). “Alternative” mating behavior in the swordtails Xiphophorus nigrensis and Xiphophorus pygmaeus. Eehav. Ecol. Sociobiol. 24,341-348. Ryan, M. J., Fox, J. H., Wilcynski, W., and Rand, A. S. (1990). Sexual selection for sensory exploitation in the frog Physalaemus pustulosus. Nature (London) 343, 66-67. Ryan, M. J., and Rand, A. S. (1990). The sensory basis of sexual selection for complex calls in the tlingara frog, Physalaemus pustulosus (sexual selection for sensory exploitation). Evolution 44, 305-314. Ryan, M. J., and Wagner, W. E. (1987). Asymmetries in mating preferences between species: Female swordtails prefer heterospecific males. Science 236,595-597. Searcy, W. A. (1979). Female choice of mates: a general model for birds and its application to red-winged blackbirds (Agelaius phoeniceus). Am. Nat. 114,77-100. Shaw, K. (1995). Phylogenetic tests of the sensory exploitation model of sexual selection. Trends Ecol. Evol. 10, 117-120. Shrimpton, A. E., and Robertson, A. (1988a). The isolation of polygenic factors controlling bristle score in Drosophila melanogaster. I. Allocation of third chromosome sternopleural bristle effects to chromosome sections. Genetics 118, 437-443.
SEXUAL SELECTION IN SEAWEED FLIES
57
Shrimpton, A. E., and Robertson, A. (1988b). The isolation of polygenic factors controlling bristle score in Drosophila melanogaster. 11. Distribution of third chromosome bristle effects within chromosome sections. Genetics 118, 445-459. Shuster, S. M., and Wade, M. J. (1991). Equal mating success among male reproductive mating strategies in a marine isopod. Nature (London) 350, 608-610. Simmons, L. W. (1986). Female choice in the field cricket Gryllus bimaculatus (De Geer). Anim. Behav. 28, 906-912. Simmons, L. W. (1990). Nuptual feeding in tettigoniidae: Male costs and the rates of fecundity increase. Behav. Ecol. Sociobiol. 27, 43-47. Smith, R. L. (Ed.). (1984). “Sperm Competition and the Evolution of Animal Mating Systems.” Academic Press, Orlando, Florida. Soffe, C. R. (1986). Chemical attraction in the seaweed fly Coelopa frigida. Unpublished bachelor’s thesis, University of Nottingham, United Kingdom. Spickett, S. G. (1963). Genetic and developmental studies of a quantitative character. Nature (London) 199,870-873. Thompson, U. (1951). Studieson thegeneticsandecologyof Coelopa frigida (Fab).Unpublished doctoral dissertation, University of Durham, United Kingdom. Thornhill, R. (1976). Sexual selection and nuptial feeding behaviour in Bitfacus apicalis. Am. Nat. 110, 529-548. Thornhill, R. (1979). Male and female sexual selection and the evolution of mating strategies in insects. I n “Sexual Selection and Reproduction Competition in Insects” (M. S. Blum and N. A. Blum, eds.), pp. 81-121. Academic Press, New York. Thornhill, R. (1980). Mate choice in Hylobittacus apicalis and its relations to some models of female choice. Evolution 34, 519-538. Thornhill, R. (1981). Panorpa (Mecoptera: Panorpidae) Scorpionflies: Systems for understanding resource-defense polygyny and alternative male reproductive efforts. Ann. Rev. Ecol. Sysr. 12, 355-386. Tobin. E. N., and Stoffolano, J. G . (1973). The courtship of Musca species found in North America. I. The house fly Musca domestica. Ann. Entomol. SOC. Am. 66, 1249-1257. Trickett, A. J., and Butlin, R. K. (1994). Recombination suppressors and the evolution of new species. Heredity 73, 339-345. Trivers, R. L. (1972). Parental investment and sexual selection. I n “Sexual Selection and the Descent of Man, 1871-1971” (B. Campbell, ed.), pp. 136-179. Heinemann, London. Verrell, P. A. (1989). Male mate choice for fecund females in a plethodontid salamander. Anim. Behav. 38, 1086-1088. von Schantz, T., Grahn, M., and Goransson, G. (1994). Intersexual selection and reproductive success in the pheasant, Phasianus colchicus. Am. Nat. 144,510-527. Wade, M. J., Patterson, H., Chang, N. W., and Johnson, N. A. (1994). Post-copulatory, prezygotic isolation in flour beetles. Heredity 72, 163-167. Walker, W. F. (1980). Sperm utilization strategies in non-social insects. Am. Nat. 115,780-799. Ward, P. (1993). Females influence sperm storage and use in the yellow dung fly, Scatophaga stercoraria (L.). Behav. Ecol. Sociobiol. 32, 313-319. Wells, K. (1977). Territoriality and male mating success in the green frog Rana clamitans. Ecology 58,750-762. Wilcockson, R. W., Crean, C . S., and Day, T. H. (1995). Heritability of a sexually selected character expressed in both sexes. Nature (London) 374, 158-159. Wilkinson, G. S., and Reillo, P. R. (1994). Female choice responds to selection on an exaggerated male trait in a stalk-eyed fly. Proc. R. SOC. Lond. B 255, 1-6. Wolstenholme, D. R., and Thoday, J. M. (1963). Effects of disruptive selection. VII. A third chromosome polymorphism. Heredity 18,413-432. Zahavi, A. (1975). Mate selection-a selection for a handicap. J. Theoret. Biol. 53,205-214.
This Page Intentionally Left Blank
ADVANCES IN THE STUDY OF BEHAVIOR. VOL. 26
Vocal Learning in Mammals VINCENT M. JANIK A N D PETERJ. B. SLATER SCHOOL OF BIOLOGICAL A N D MEDICAL SCIENCES
UNIVERSITY OF ST. ANDREWS
ST. ANDREWS, FIFE, UNITED
KINGDOM
1. INTRODUCTION In this chapter we survey the occurrence of vocal learning in mammals and discuss possible reasons it has evolved. But first it is important to be clear about what we mean by “vocal learning.” The term has been used to describe the influence of learning on a variety of different aspects of vocal communication. Learning can affect the generation of sounds, their usage, and their comprehension. While modifications in sound generation as a result of experience can be described as learning sounds, those in comprehension and usage are rather different phenomena, which are perhaps better described as learning about sounds. Vocal learning, as we discuss it here, refers only to learning sounds, that is, to instances where the vocalizations themselves are modified in form as a result of experience with those of other individuals. Learning that affects usage and comprehension of sounds will be referred to as contextual learning as opposed to vocal learning. Contextual learning in relation to vocal communication is relatively common among mammals. The list of animals in which the utterance of a vocal signal has been brought under conditional control, i.e., which have learned to change the context in which they are using sounds, comprises rats (Lal, 1967), guinea pigs (Burnstein and Wolff, 1967), dogs (Salzinger and Waller, 1962), cats (Molliver, 1963), sea lions (Schusterman and Feinstein, 1965), primates (Myers, Horel, and Pennypacker, 1965; Randolph and Brooks, 1967; Wilson, 1975; Aitken and Wilson, 1979), and dolphins (Lilly, 1965). Other forms of contextual learning in vocal communication involve learning to recognize particular sounds, or learning to react to sounds differently as a result of experience. These are important ways in which learning may influence vocal communication, and the behavior associated with it, but they are not examples of vocal learning in the strict sense in which we use the term here. 59
Copyright 0 1997 by Academic Press All rights of reproduction in any form reserved. w65-3454f97525.00
60
VINCENT M. JANIK AND PETER J. B. SLATER
Some types of modification through learning are likely to be easier to achieve than others because different sound parameters are controlled by different mechanisms. Overall duration and amplitude of a sound can be altered by simple modifications of exhalation alone. A longer exhalation phase or higher air pressure while producing the sound is all that is required to cause such changes. For learning to occur here, only the activity of respiratory muscles need be modified by experience. Such changes in duration or amplitude do not affect the overall form of a call. O n the other hand, the detailed matching of a sound pattern requires much more complicated alterations. The fundamental frequency of a sound can be altered only if the activity of muscles controlling the vocal apparatus itself can be modified by experience. In tonal signals this is required to achieve change in the frequency contour of a call. Of course, other more complex processes can be involved in sound production. Rapid amplitude modulation, for example, can cause additional frequency bands in a call. Seemingly subtle changes that require a high degree of coordination between respiratory, laryngeal, and articulatory muscles, like those leading to differences in voice-onset time, can also be involved. Where possible we look at modifications in duration and amplitude and those in frequency parameters separately, and point to these different levels of motor control as they occur in the vocal learning of different groups. Vocal learning has been described only in birds and mammals, and even among these the evidence is patchy. It has been found in all songbirds (Oscines) studied to date (Kroodsma and Baylis, 1982), but appears not to occur in the closely related Suboscines (Kroodsma, 1984,1989). Convincing evidence comes from only two of the twenty or so other orders of birds: the hummingbirds (Apodiformes) (Baptista and Schuchmann, 1990; Gaunt, Baptista, Sanchez, and Hernandez, 1994) and the parrots (Psittaciformes) (Todt, 1975; Pepperberg, 1981). The three groups showing vocal learning are only distantly related to each other, suggesting that it has evolved among birds on at least three separate occasions. In mammals, the importance of vocal learning in our own species contrasts remarkably with the scarcity of evidence elsewhere. Part of the reason for this may be the lack of relevant studies. Absence of evidence for vocal learning in a particular species is certainly not evidence for its absence. In our review of the literature on vocal learning in mammals we attempt to determine the extent to which it occurs in species other than our own, and whether it is widespread or patchily distributed as in birds. This survey enables us to compare and contrast birds and mammals, and to consider the possible functional significance of vocal learning. It may perhaps also shed some light on why it occurs in humans. But first we discuss the methods
VOCAL LEARNING IN MAMMALS
61
that have been used in various studies and the extent to which these can give unequivocal evidence for or against vocal learning.
11. EVIDENCE FOR VOCAL LEARNING There are several pitfalls in trying to show vocal learning in a particular species. A fundamental problem is the question of whether a particular call was in an animal’s repertoire before it was first noted. Apparent changes in the call repertoire of an animal often relate to other changes in its environment. Examples are alterations in social context, because of the introduction of foreign animals or a change in status, changes in the habitat that alter its sound transmission characteristics, or seasonal events that influence the diurnal behavior patterns of the animal. If a new call arises at the time of such an event, it could be because of a change in the frequency of occurrence of calls that were already present in the repertoire rather than vocal learning. But even a truly new call could arise for different reasons. Maturational processes or improvisation could be responsible, rather than copying from other individuals. Maturational processes lead to changes in vocal tract morphology that can influence sound characteristics. Thus, simple observations of changes in the call repertoire during ontogeny are difficult to interpret. Vocal learning may or may not be involved. Improvisation is another process that leads to the production of new calls. Various different mechanisms can be used to achieve improvisation, and vocal learning, again, may or may not be involved. One possibility is the production of completely new sounds through random sound generation. This form of improvisation would be an interesting case of vocal flexibility. According to our definition it does not involve vocal learning, however, since experience is not required. As we see in our survey, completely random sound production has so far never been the only possible explanation for an observed change in call structure. But there are other forms of improvisation that do involve learning. If an animal produces a completely new call that avoids overlap with calls of other individuals, experience might be used to achieve this avoidance. This would be a case of vocal learning according to our definition. A more restricted form of improvisation might involve a recombination of given subunits of a call. If these units can be produced on their own, this form of improvisation represents a special case of contextual learning. It is simply a matter of calls that are already present in an individual’s repertoire being produced in a new context. Finally, an animal could learn different parts of other individuals’
62
VINCENT M. JANIK AND PETER J. 0. SLATER
calls and combine them t o form a new call. This sort of improvisation would be a clear case of vocal learning. Vocal learning is obviously difficult to investigate if changes in calls do not result in matching those of other individuals or model sounds. The most clinching evidence for vocal learning comes from experiments in which animals are trained to imitate sounds that have not been in their repertoire before. If animals are able to copy sounds that are very different from sounds in their natural repertoire, like human speech or computergenerated sounds that were designed t o be different, vocal learning has clearly been demonstrated. In animals that are not capable of imitating such sounds, vocal learning within the natural repertoire can nevertheless be shown by rearing experiments, if infants that were raised with different acoustic stimuli are found to match the sounds they heard in detail. 'It is unlikely that selective reinforcement of randomly produced sounds could result in detailed matching of sounds produced by other animals. Such rearing experiments have commonly been carried out on birds, but few have been attempted on mammals. There are probably two main reasons for this. First, many of the mammalian species involved, such as whales and dolphins, are difficult to keep in the highly controlled acoustic environments necessary for such studies. Second, the species involved are highly sociable and subjecting individuals to experimental treatments involving deprivation likely to lead to suffering is not easy to justify. Because of these difficulties, for many mammals the evidence for or against vocal learning is more circumstantial. The main source of such less direct evidence comes from geographic variation in vocal signals, Where neighboring animals, or those in a social group, share sounds that differ from more distant individuals or those in other groups, vocal learning is a probable reason. This is most obvious where groups are not geographically isolated from each other. However, one has to be careful in interpreting such observations. Geographic variation in vocalizations can also arise because of ecological differences. Transmission characteristics of the environment may influence the extent to which a particular sound in an animal's repertoire is used. Differences in social structure in different locations could have the same effect on the usage of different call types that are present in the repertoire of all members of the species. Even different preferences for certain call versions can be the reason for geographic variation. In what he called action-based learning Marler (1991) suggested that an animal may produce only those sounds from its repertoire that are selectively reinforced by social stimulation. If there are different preferences for particular call types in different populations, action-based learning may result in geographic variation of their call repertoires. This would be a case of contextual learning, but not vocal
VOCAL LEARNING IN MAMMALS
63
learning. Finally, if animals tend to stay in their maternal groups, genetic transmission is a possible reason for differences in call repertoires. Further research is needed in such cases to show vocal learning unequivocally.
111. SURVEY A. BATS In many bat species the isolation calls of pups exhibit vocal “signatures” (significant individual differences that remain constant over time) and these are used by mothers to find their offspring when they return to the colony to suckle them. Variation in the isolation calls of young big brown bats (Epresicus fuscus) is as great within as between colonies, and between twins as between less close relatives (Rasmuson and Barclay, 1992). The production of a completely new call could be achieved either by producing a call at random, which need not involve learning, by active avoidance of matching the calls of neighboring individuals, or by composing a call mixing different parts of heard calls. Although the last two mechanisms would involve learning, there is no conclusive evidence that they are involved here. However, it is interesting that there are closer similarities in echolocation calls within families than between them in this species (Masters, Raver, and Kazial, 1995), suggesting that echolocation and isolation call development are controlled by different mechanisms. In the lesser spear-nosed bat (Phyllostornus discolor) mothers and infants exchange calls when they reunite, and Esser and Schmidt (1989) suggest that the infant’s isolation call becomes progressively more like the mother’s call over the first few weeks of life. Esser (1994) argues that this is due to learning: he found that the calls of isolated pups that were played a taperecorded call over the first 50 days of life tended to become somewhat similar to it, while those of unstirnulated controls remained highly variable. However, one problem with isolation experiments is the general lack of stimulation. Bats might simply need auditory input of some nonspecific sort to develop normal calling behavior. An experiment with two groups hearing different calls would clarify whether learning is involved. The most convincing evidence for vocal learning in bats comes from the development of echolocation calls in greater horseshoe bats (Rhinolophus ferrurnequinurn) (Jones and Ransome, 1993). This call is an almost pure tone of around 83 kHz. Its sound frequency is higher in summer than in winter, but it also rises in the first year or two of life, and later falls off in old age. When young bats first start to hunt for themselves at a few weeks of age the frequency of the calls they adopt is strongly correlated with that
64
VINCENT M. JANIK AND PETER J. B. SLATER
of their mother. In both males and females young bats with mothers over five years old have significantly lower echolocation calls than those born to younger mothers. This correlation with maternal age strongly suggests that learning is involved. B.
PINNIPEDS
Most authors place the pinnipeds in the order carnivora. They are classified into the true seals (phocids), the sea lions and fur seals (otariids), and the walruses (odobenids). Evidence for vocal learning has so far been found only in phocids. The most conclusive evidence that seals can learn new sounds comes from the imitation of human speech by captive animals. Two male harbor seals (Phoca vitulina) at the New England Aquarium were able to mimic speech sounds (Ralls, Fiorelli, and Gish, 1985). The more impressive of the two, Hoover, spontaneously developed imitations of a variety of phrases typically produced by visitors to the aquarium, such as “hello there” and “come over here,” as well as his own name and an imitation of laughter. In a controlled conditioning experiment, the second seal was trained to imitate its own name. Evidence in the wild is not so easily obtained. Geographical variation has been described in several species, but authors vary in whether they attribute this to learning. Thomas and Stirling (1983) examined Weddell seal (Leptonychotes weddelli) calls at Palmer Peninsula and McMurdo Sound, two sites on the edge of the Antarctic continent some 4000 km apart. Weddell seals have a large repertoire of vocalizations that they use extensively in the mating season (Thomas and Kuechle, 1982). Although some call types were found at both sites, each colony also had several unique ones. Marked spectral and temporal differences were found in some of the shared calls. Furthermore, seals at McMurdo had more call types in their repertoire than those at Palmer Peninsula. The authors suggest that the combination of fidelity to breeding sites and learning may account for these differences. The problem is to differentiate between these ideas: over a long period of time geographical isolation may lead to genetic differences affecting calls just as learning may give rise to call differences over shorter periods. Morrice, Burton, and Green (1994) compared vocal repertoires of two Weddell seal colonies located in adjacent fjords only 20 km apart. Despite the close proximity of these two colonies they had only 5 of 44 described vocalization types in common. Of the five shared call types two showed significant differences in their start or end frequency between colonies. Furthermore, a song type recorded in 1984 in one of the fjords (Green and Burton, 1988) could not be found again in 1989/1990 (Morrice et al.,
VOCAL LEARNING IN MAMMALS
65
1994). Such strong dialects between adjacent colonies are rare in mammals and make vocal learning a likely explanation for their occurrence. All other studies on geographic variation in seal calls have concentrated on very distant sites. Thomas and Golladay (1995) studied geographic variation in leopard seal (Hydrurga leptonyx) underwater vocalizations at Palmer Peninsula and McMurdo Sound, the same sites as in the Weddell seal study by Thomas and Stirling mentioned earlier. They found that seals at Palmer Peninsula had more call types in their repertoire and that shared calls varied in frequency and time parameters between the two sites. Bearded seals (Erignathus barbatus) at six distant sites north of Canada and Alaska also differed in various temporal and spectral features of their underwater “trill” vocalizations (Cleator, Stirling, and Smith, 1989). Again, the authors suggest that this may be because of strong fidelity to specific breeding areas, but they do not discuss learning. Terhune (1994), in a study of harp seal (Phoca groenlandica) underwater calls in the Gulf of St. Lawrence and on Jan Mayen Island, sites some 3500 km apart, found spectral and temporal differences in shared call types. Each colony also had one unique call type not found at the other study site. Terhune attributes differences between sites to reproductive isolation. Given the possible role of learning, this seems somewhat sweeping. It is interesting, though, that different samples at each of the two sites were collected on occasions 18-20 years apart and only slight differences, probably attributable to sampling error, were found between the recordings at the same place (Terhune, 1994). This suggests that call repertoires at a particular place are conservative in a way that would be less likely if learning were involved. However, harp seals can live up to 35 years (Reeves, Stewart, and Leatherwood, 1992), so that 20 years might not be enough time to pick up changes caused by copying errors in a learning process. Differences in the pattern of northern elephant seal (Miroungu angustirostris) threat vocalizations between several islands off California were interpreted by Le Boeuf and Peterson (1969a) as evidence for learning. There were marked differences in the pulse rate of threat calls between the colonies, one of which, with its distinct dialect, had been in existence only for a few years. However, later studies showed that, even though there were still clear individual differences in pulse rates, colony differences had disappeared (Le Boeuf and Petrinovich, 1974; Shipley, Hines, and Buchwald, 1981). Note that to change pulse rate the animal has to produce the same sound in quicker or slower succession. It does not change the form of the call. Therefore, this would not qualify as an example of vocal learning according to our definition, unless pulse sequences represent minimum units of call production and single pulses cannot be produced on their own. However, the disappearance of geographic differences in calling behavior
66
VINCENT M. JANIK AND PETER J. B. SLATER
is an interesting phenomenon that is relevant for our interpretation of geographic variation in other cases. By the end of the 19th century, northern elephant seals were reduced to less than 100 animals all living on Isla de Guadalupe 243 km west of Baja California (Bartholomew and Hubbs, 1960). The population was then protected and started to grow again and to recolonize old breeding sites (Le Boeuf and Petrinovich, 1974). There are two ways in which differences between colonies could have arisen in the first phase of recolonization. The first is a founder effect. The small number of males that emigrated to recolonize a particular breeding site could have had pulse rates on average different from those of the original colony. Once an elephant seal has chosen a breeding site in the first year of its reproductive life, it tends to return to that same site each year (Le Boeuf, Ainley, and Lewis, 1974). A difference in pulse rates between colonies would therefore be likely to persist as long as no further migration occurred. The second possibility is that a few males monopolized most of the females and spread their pulse rate, either through being imitated by younger animals or through genetic transmission. If young males return to their natal rookery to breed, a particular pulse rate could thus establish itself and persist. In small colonies only a few males do indeed monopolize all the females and are the only ones that breed (Le Boeuf, 1974), but there is no clear evidence on whether males return to their natal rookeries to breed. Dispersion from the island of birth after 1968 was considerable (Le Boeuf et ul., 1974), but of 400 males tagged as juveniles on Aiio Nuevo Island between 1964 and 1969 none were sighted at any of the other breeding sites in their first breeding season at the age of 5-6 years (Le Boeuf, 1974). If there is continuing immigration from other colonies, pulse rate in a recolonized site would change and eventually the differences in pulse rate between colonies would disappear. The subsequent studies during rapid population expansion showed this to be the case. Such a process would occur as long as there was no isolating mechanism between populations, whether or not learning has a role in pulse rate development. The longevity of elephant seals made it possible to witness these changes occurring. The observed differences cannot therefore be regarded as a result of learning. C. CETACEANS Toothed Whales (Odontocetes) As with seals, the smaller cetaceans can be kept in captivity, and observations there provide some evidence of vocal imitation. There is anecdotal evidence, largely from keepers, that these animals can modify the broadband frequency squeaks that they produce so as to imitate human speech (Tursiops truncufus, Caldwell and Caldwell, 1972; Delphinupterus leucas,
1.
VOCAL LEARNING IN MAMMALS
67
Eaton, 1979). The most extensive experimental evidence comes from the bottlenose dolphin (Tursiops truncutus), as this is the cetacean most commonly kept in captivity. They produce whistles, clicks, and burst-pulsed sounds that are all modifiable by experience. Most of the evidence of vocal learning concerns their whistles. Caldwell and Caldwell (1972) recorded a case of spontaneous whistle mimicry when their study animal matched a whistle that was being used to test its sound localization abilities. Reiss and McCowan (1993) also found bottlenose dolphins mimicking whistle sounds spontaneously. Two young males were given control over stimuli by means of an underwater keyboard. Pressing a symbol on this released a sound linked to it and also the presentation to the dolphin of an object or activity, such as a ball or being rubbed. The animals learned to mimic the sounds and later produced them frequently without having pressed the key to hear the appropriate stimulus, but often while playing with the object or engaged in the activity with which it had been linked. When it comes to training, there is no doubt that dolphin whistles can be modified by experience. Richards, Wolz, and Herman (1984) trained a dolphin to imitate computer-generated sounds using food or petting by the trainer as rewards. The animal matched a variety of sounds that were quite different from those it produced before the training. In some cases it produced passable copies the first time a sound was introduced. In further training the animal was rewarded for producing particular whistles when specific objects were shown to it, and would do so with a high degree of reliability. Thus, the dolphin effectively learned vocal labels for those objects. In a subsequent series of experiments, Sigurdson (1993) also successfully trained two dolphins to match computer-generated whistles. A few studies have described the changes in whistle repertoires of infant bottlenose dolphins over time (Caldwell and Caldwell, 1979; Sayigh, 1992; McCowan and Reiss, 1995). However, even though the species is clearly capable of vocal learning, the role of vocal learning in whistle development of infants has not been demonstrated so far. Wang and his colleagues (Wang, Wiirsig, and Evans, 1995) found marked geographic variation in spectral and temporal features of bottlenose dolphin whistles in the wild at sites only a few hundred kilometers apart. Each site had its own resident population with some individuals moving between them. Since dolphins produce individually specific whistle contours, the study did not look at different call types separately, but simply compared general parameters like start and end frequency in all recorded whistles at each site. Experimental work by Moore and Pawloski (1990) provided evidence that vocal flexibility applies to click sounds used in echolocation as well. They succeeded in training a bottlenose dolphin to shift the peak frequency
68
VINCENT M. JANIK AND PETER J. B. SLATER
of its broad-band ultrasonic echolocation clicks. The anecdotal evidence of dolphins mimicking speech sounds also suggests that clicks can be modified, as they use very rapid click trains called burst-pulsed sounds to produce these imitations. Turning to larger odontocetes, there is also evidence in favor of vocal learning. Pods of killer whales (Orcinus orca) off British Columbia have repertoires of 7-17 call types, which appear to be shared by most individuals in the pod (Ford and Fisher, 1983; Ford, 1991). Groups of pods can be placed in “clans,” which share some call types, although shared types still have pod-specific features; there is no sharing of calls between clans. New pods appear to form by splitting of preexisting ones, with slow divergence of calls thereafter, though there is evidence that pod repertoires can persist for 25 years with little change. Since all animals within a pod are closely related (Bigg, Olesiuk, Ellis, Ford, and Balcomb, 1990), such differences could be caused by genetic differences. However, Ford (1991) reports that individuals occasionally seem to mimic the calls of other pods, suggesting that learning is involved. Studies on captive killer whales have not addressed the question of vocal learning directly. However, Bain (1986) gave an anecdotal report that a female killer whale from Iceland started to mimic calls of a female from British Columbia, Canada, after they had been housed together for a few years. Differences in the vocal repertoire of killer whales between such distant sites are larger than those between sympatric pods (Awbrey, Evans, Jehl, Thomas, and Leatherwood, 1982). However, no spectrographic analysis was provided in Bain’s study. Bowles, Young, and Asper (1988) described the vocal development of a captive killer whale calf, but the role of vocal learning in call development could not be addressed. As discussed earlier, changes in the vocal repertoire during ontogeny can be maturational, genetic, or learned. Van Heel, Kamminga, and van der Toorn (1982) reported an experiment in which a killer whale seemed to spontaneously imitate computer-generated tonal signals that were used in training the animal to perform different tasks. However, these signals were designed to resemble the whale’s own vocalizations, so that vocal learning was not necessary to produce them. 2. Baleen Whales (Mysticetes)
Perhaps the best known example of vocal learning among whales in the wild is that of male humpback whales (Megapreru novaeangliae) in which individuals produce long and elaborate songs lasting up to 20 minutes before repetition (Payne and McVay, 1971). Recordings of the songs of this species off Bermuda over a period of 18 years showed that they changed with time but that at any one time the songs of different individuals were
VOCAL LEARNING IN MAMMALS
69
similar (Payne and Payne, 1985). Detailed study of songs recorded around Hawaii in two singing seasons (mid-November to mid-May) demonstrated progressive change from month to month through the season, with little difference between the end of one season and the start of the next, the period when the whales are away on their feeding grounds and sing little (Payne, Tyack, and Payne, 1983). This suggests that change is an active process rather than one of forgetting sounds with the passage of time. Nor can the changes be attributed to changing group membership, as Guinee, Chu, and Dorsey (1983) obtained two recordings from each of three identified individuals and found both that they changed and that they did so in parallel with each other. These detailed short-term changes, both within and between individuals, can be accounted for only if animals are imitating one another. However, songs of different humpback whale populations differ completely (Winn, Thompson, Cummings, Hain, Hudnall, Hays, and Steiner, 1981). But even though this species is clearly capable of vocal learning, geographic variation could have evolved because of genetic isolation between populations in different oceans. Bowhead whales (Balaena mysticetus) studied during their spring migration have also been found to sing (Cummings and Holliday, 1987). Their song consists of repeated phrases that can be categorized into one to three themes. A song lasts about one minute but is usually repeated several times. As with humpback whales, all animals in the study population off Alaska sang the same song but the song was different in each singing season (Wursig and Clark, 1993). Cummings and Holliday (1987) always heard one animal singing at a time but another one often started as soon as the first stopped. It is not clear yet whether individuals change their song over the season, but, if so, it would be strong evidence for vocal learning. D. PRIMATES As we get closer to humans, one might imagine that the evidence for vocal learning would become more and more impressive. As we shall see, this is far from being the case. 1.
Monkeys
“There is no conclusive evidence of vocal learning in monkeys” (Snowdon, 1990, p. 225). Perhaps the strongest evidence against vocal learning is in squirrel monkeys (Saimiri sciureus), where isolation-reared animals (Winter, Handley, Ploog, and Schott, 1973), and even deafened ones (Talmage-Riggs, Winter, Ploog, and Mayer, 1972), show normal vocal development. In the squirrel monkey there is also a good example of a dialect that is not based on learning. Two distinct phenotypes of squirrel monkeys,
70
VINCENT M. JANIK AND PETER J. B. SLATER
the Roman Arch and the Gothic Arch population, named after characteristics of their supraorbital color pattern, show differences in the structure of their isolation calls (Winter, 1969). The unique call structure of each population was present in infants from the first day of their lives (Lieblich, Symmes, Newman, and Shapiro, 1980). Given the normal vocal development of both deafened and isolated individuals, these differences are clearly not based on vocal learning. Furthermore, Struhsaker (1970), who investigated geographic variation in the call structure of vervet monkeys (Cercopithecus uethiops) in various parts of Africa, could not find any differences between populations. Marmosets and tamarins have been studied extensively, mainly by Snowdon and his colleagues. In the field, Hodun, Snowdon, and Soini (1981) measured four parameters of the long call of saddle-backed tamarins (Suguinus fuscicollis). They found differences in the long calls of different subspecies, with that of one individual having features both of its own subspecies, S. f nigrifrons, and of an adjacent one, S. f illigeri, suggesting that learning might be involved. To find out whether this animal could have been a hybrid between the two subspecies, the same four call parameters were measured from two known hybrids of these subspecies in captivity. The hybrids developed a long call with one of the sound frequency parameters being closest to that of a third subspecies, which they had been able to hear in the room where they were housed, while the other parameters resembled those of nigrifrons. The wild animal, on the other hand, resembled the illigeri subspecies in this frequency parameter and in its call duration. The data were not given in detail, and concern up to only 12 calls of each of the three different animals, so it is difficult to assess whether differences in long-call structure reflected vocal learning, genetic differences, or different motivational states during recording sessions. Maeda and Masataka (1987) found variation in long calls of red-chested moustached tamarins (Suguinus lubiutus Zubiutus), that had been caught at two sites 27 km apart. In a subsequent study, a third group, which was caught only 15 km from one of these sites, produced a third call variant (Masataka, 1988). Each particular variant of the call was used by animals living within 6 to 15 km of each of the original catching sites, while animals from further away used different variants. Masataka (1988) reported that there were no geographical barriers between sites where different call variants were found, but there are no data on whether interbreeding between these sites occurs. It is also possible that tamarins adjust their calls according to the habitat by using different forms that are already present in their repertoire or that the variations found belong to particular matrilines. In either case learning need not be involved.
VOCAL LEARNING IN MAMMALS
71
In a more recent study, Elowson and Snowdon (1994) showed that members of pygmy marmoset (Cebuella pygmaea) colonies modified the structure of their individually distinctive trill contact calls after being placed in a cage adjacent to another colony. The modifications did not result in converging or diverging call structure. This, again, could be evidence for learning, but introducing new animals, even if contact is only acoustic, might shift social relationships within and between colonies. Thus, the modification of calling behavior over time might not be due to learning but could be related to a change in the role or dominance structure of each colony. In this example it is not clear whether call variants that were found after the introduction of the other colony were in the call repertoire before. They simply could have occurred less often because of the different social status of the individual before the introduction. Whether or not macaque food calls are learned has been the cause of some controversy. Field observations by Green (1975), in which he found differences between troops in Japanese macaques (Macaca fuscafa), suggested that food calls might be learned. However, in all three locations studied by Green, extensive food provisioning might have conditioned monkeys to produce only a particular version of their food call from a range present in their repertoire before provisioning started. Subsequently, Masataka and Fujita (1989) carried out cross-fostering experiments between Japanese and rhesus monkeys (Macaca mulafta),and found that the young developed calls more typical of their foster species. However, their result was based on only three animals that might have been atypical (Snowdon, 1990). Owren, Dieter, Seyfarth, and Cheney (1992) failed to replicate Masataka and Fujita’s result. They argue that the calls of adult females of the two species vary a great deal and that there was no significant difference between species in any of the measures used. Even though infants of both species differed significantly in several parameters, calls developed by the cross-fostered infants fell within the distribution range for normally raised members of their own species in most cases. Despite this, it seemed that cross-fostered Japanese macaques did become more similar to rhesus monkeys in several frequency measures after 2 years of age. Given the variability of calls within species and the overlap between them, this is not an easy system in which to test for vocal learning. In training experiments it has been shown that rhesus monkeys can be conditioned to increase overall call duration and amplitude (Sutton, Larson, Taylor, and Lindeman, 1973). Therefore, primates seem to have control over these basic call parameters. However, as pointed out earlier, such shifts do not require any changes in the setting of the sound production organ, but only a longer or stronger expiration phase. In another report about these experiments, Larson, Sutton, Taylor, and Lindeman (1973)
72
VINCENT M. JANIK AND PETER J. B. SLATER
described changes in the spectral components of the trained calls in the 58 kHz range. It is important to note that these changes were not a requirement of the conditioning procedure but occurred independently of the selective rewards used in the experimental setup. Therefore, they do not represent evidence for vocal learning in the frequency domain. Larson et al. argue that they reflect a decrease in stress over the period of the experiment. Hauser (1992) found differences in the “coo” social contact vocalization of rhesus monkeys between different matrilines on Cay0 Santiago. Again, it is not clear whether these are caused by learning, genetic predisposition, or differences in parental behavior that might affect the extent to which a particular call version in an animal’s repertoire is used. Hauser mentioned that some animals changed their coo calls after migrating into another group, but did not provide further details. As in other studies, such changes could also be the result of a change in usage of calls that were present in the animal’s repertoire before. 2.
Gibbons
Among the primates, gibbons are undoubtedly the most elaborate singers. Both male and female sing, and they often do so in the form of coordinated duets in which the female, with her “great call,” may take the leading role. Gibbons are monogamous and territorial, and the song duets are assumed to function in territory maintenance and pair bonding. In playback experiments, Mitani (1985) showed that females do not approach singing males, but males and females react strongly toward new duets in or close to their territory. Despite their complexity and intricacy, gibbon songs are highly stereotyped with little geographic variation (Marshall and Marshall, 1976). The major role of experience in the development of gibbon vocalizations lies not in copying but in coordination between members of a pair. A newly introduced pair of siamangs (Hylobates syndactylus) completed their great call sequences on only 24% of occasions, the songs terminating because one or the other animal produced a call that was inappropriately placed or timed. However, in recordings made after they had been together for 18 weeks, some 79% of sequences were completed (Maples, Haraway, and Hutto, 1989). There is also evidence, some anecdotal (Marshall and Marshall, 1976) some better documented (Geissmann, 1983; Srikosamat, 1982), that females deprived of a male or with one that does not sing can produce the male contribution to the duet as well as their own. When it comes to development in gibbons, there is no evidence that individuals copy from each other, but there is strong evidence for a genetic influence. This comes from hybrids in nature and in captivity. White-handed gibbons (Hylobates lar) and pileated gibbons ( H . pileafus) hybridize in a contact zone in central Thailand. Brockelman and Schilling (1984) found
VOCAL LEARNING IN MAMMALS
73
that daughters of mixed parentage develop great calls unlike those of their mothers, but intermediate between the two species. This occurs despite the fact that the calls develop during mutual singing sessions between mothers and daughters. A similar conclusion has been reached from studies of hybrids between the two species in captivity (Geissmann, 1984,1987). Tenaza (1985) also found that the song of two young, a male and a female, from a cross between H. lar and H. muelleri included characteristics typical of their own sex in both species despite having heard those elements from only one of these, that of their own sex parent. Although these hybrid studies appear to argue strongly against a role of learning in gibbon song development, Marler and Mitani (1988) press for caution. If, as has been suggested in birds, young animals are born with a species-specific template that constrains what they will learn, the template in a hybrid might not be well matched to the sounds it experiences, so that it fails to learn parental calls. Mitani (1987) also presented evidence that agile gibbons (Hylobates agilis) show geographic variation in their calls. The populations involved were on separate islands very distant from one another. Therefore, genetic differences could be responsible. Further studies focusing on possible dialects in the wild are needed to evaluate whether some learning could be involved in gibbon call development.
3. Great Apes Of the great apes, the vocalizations of chimpanzees have been the most extensively studied (e.g., Marler and Tenaza, 1977). It was natural that those interested in whether apes could master language first turned to humans’ closest relatives. Although subsequent efforts with various media, ranging from sign language to computer keyboards, met with considerable success, early attempts to “teach chimps to speak” were almost fruitless. Vicki, the common chimpanzee (Pan troglodytes) trained by Hayes (1951), eventually seemed to produce four English words after 7 years of language training. However, no data on the similarity of these sounds with actual words were presented. The bonobo (P. paniscus) called Kanzi, studied by Hopkins and Savage-Rumbaugh (1991), developed a variety of speciestypical vocalizations but, despite extensive interactions with humans, only four calls not shared with control animals. However, the different rearing of Kanzi could have delayed the development of certain bonobo calls, so that these four sounds could have been left over from his call repertoire as an infant. Therefore, they do not represent evidence for vocal learning in bonobos either. When it comes to imitation of conspecific sounds, there is little evidence. Mitani, Hasegawa, Gros-Louis, Marler, and Byrne (1992) describe differences in the pant hoot of chimps between two sites in Tanzania, Mahale
74
VINCENT M. JANIK AND PETER J. B. SLATER
and Gombe. In a subsequent analysis, Mitani and Brandt (1994) demonstrated that population differences in these calls were largely encoded in only one parameter, namely frequency range. The differences seem, however, quite subtle, which may explain why there are no other reports of geographical variation in chimp sounds. Mitani et al. (1992) argue that these differences could result, not from copying, but from contextual learning stemming from the selective reinforcement within each population if an animal produces sounds that match the population norm. This would be what Marler (1991) called action-based learning, in which an individual produces only those sounds from its repertoire that have been selectively reinforced by social stimulation. In this case the difference in call structure between recording sites would have been caused by different preferences for certain call versions by each population. Alternatively, food provisioning by humans at both sites may have had a conditioning effect that resulted in the observed differences, if individual humans were more responsive to certain call variants. Again, this may have led the animals to use another call variant that was present in their repertoire before without vocal learning being involved. Mitani and Brandt (1994) presented interesting evidence that male chimpanzees match the acoustic characteristics of each other’s pant hoots when calling together. This was statistically significant for the one individual for which a sufficient sample size was available. Depending on its chorusing partner this animal changed the spectral structure of its calls to match those of the other caller. Further research is needed to investigate the flexibility of pant hoot matching and assess whether these changes are learned or a side effect of subtle differences in calling contexts. E. OTHER MAMMALIAN ORDERS A few studies of other mammalian orders have shown similar phenomena to those discussed for primates, although they have not so often been interpreted as evidence for vocal learning. Romand and Ehret (1984) studied the development of sound production in the domestic cat (Felis catus). To investigate the influence of auditory feedback, motivation, and ontogenetic changes in the vocal tract, they compared calls of normally raised, deafened, and isolated kittens. The three groups differed in certain call parameters but individual variability was great. The authors interpret these differences as resulting from the lack of auditory feedback in deafened kittens and from different motivational states in isolated ones. Call parameters of isolated individuals indicated a delayed development of the vocal apparatus and a higher stress level than in normally raised kittens. Romand and Ehret come to the conclusion that call development in cats “follows
VOCAL LEARNING IN MAMMALS
75
a ‘self-centered strategy’ rather than an ‘open and environmentally dependent’ one” (Romand and Ehret, 1984, p. 648). Molliver (1963) successfully trained cats to increase the duration of their calls in an operant conditioning procedure. As discussed earlier, this does not require any changes in the setting of the sound production organ, but only a longer expiration phase. It is a demonstration of a limited form of vocal learning that has also been found in primates. Slobodchikoff and Coast (1980) found geographic variation in various time parameters of alarm calls given by groups of Gunnison’s prairie dogs (Cynomys gunnisoni) living 13 km apart. An alarm call in this species is a series of several barks. Groups differed in total call length, number of barks per call, duration of each bark, and the time intervals between barks in a series. Again, these parameters describe the timing and duration of a call but not its frequency structure. Pikas (Ochotona princeps) in the United States also show geographic variation in duration and in the fundamental frequency of their short calls (Somers, 1973; Conner, 1982). The two populations studied by Somers are geographically separated by the Colorado River. However, there is an overlap zone at the source of the river where he found individuals with calls intermediate between the two variants. Conner (1982) only found differences between widely separated populations. He argued that variations do not represent vocal dialects but are the result of independent evolutionary histories. As with other cases of geographic variation, further studies are needed to clarify the origin of these differences. Finally, a recent study on banner-tailed kangaroo rats (Dipodomys spectabilis) has shown that they adjust their footdrumming signatures to differ from new neighbors after they change their territory (Randall, 1995). This is an interesting case of plasticity in sound production, although it concerns signals that are not produced by the vocal apparatus and concerns only the timing of signals, so it is not strictly relevant to our purpose.
IV. FUNCTIONAL SIGNIFICANCE AND ORIGIN
In this section we will consider possible reasons why vocal learning evolved in mammals. In discussing this it is important to bear in mind two separate problems. First is the question of why vocal learning arose in the first place, the answer to which may be far back in time and within a different functional context to that in which it occurs today. Second is the more accessible question of why vocal learning persists in certain species. Similar hypotheses may be relevant to both contexts, so that a discussion
76
VINCENT M. JANIK AND PETER J. 9. SLATER
of all possible reasons for vocal learning may shed light on both its origin and its current advantages for a particular species. We will consider each of the main hypotheses that have been put forward to account for vocal learning. Earlier discussions have been largely in relation to birds, as vocal learning in this group has been extensively studied. Before discussing the possible relevance of each hypothesis to mammals, we will make brief mention of how these hypotheses measure up to the bird evidence.
A. INTERSEXUALSELECTION In many animals male songs attract females. If a longer or more elaborate song is more effective in this respect, then sexual selection could result in the evolution of very complex displays. Different mechanisms could account for an increased response in the female. Complex auditory input might lead to greater stimulation of the reproductive system of a female or it might simply prevent habituation by females. From a functional perspective, song complexity might be a cue used by females to assess male fitness, if it is costly for males to achieve complexity. Whatever its causal and functional basis, however, if complexity is advantageous, vocal learning may be favored as a means of achieving it. The copying of song elements from conspecifics allows an animal to incorporate new elements into its song, while still retaining the species-specific pattern. The development of completely new sounds without such copying would risk the loss of species recognition. Furthermore, the generation of vocal complexity through an increase in genetic information would undoubtedly be a slow and costly process. Thus, vocal learning could well have been favored as a means of acquiring an elaborate song. Many songbirds have complex song repertoires built up by learning, and sexual selection has frequently been proposed as a reason for this complexity. In line with this suggestion there is evidence that female birds of various species solicit more to repertoires of songs than to single song types (e.g., red-winged blackbird, Agelaius phoeniceus, Searcy, 1988; starling, Sturnus vulgaris, Eens, Pinxten, and Verheyen, 1991), and that the reproductive system of female canaries (Serinus canaria) may be more stimulated by complex than by simple songs (Kroodsma, 1976). In the field, male sedge warblers (Acrocephafusschoenobaenus) with large syllable repertoires have been found to attract females earlier than those with smaller ones (Catchpole, 1980). Thus, several lines of evidence point to the importance of sexual selection in the generation of complex song repertoires in birds.
VOCAL LEARNING IN MAMMALS
77
Turning to mammals, we know little about the functions of many male vocal displays. In phocids they are often highly varied, and several observations suggest that some of them may be sexual displays. During the mating season male harbor seals produce repetitive vocal displays underwater for up to several hours (Hanggi and Schusterman 1994; Van Parijs, Thompson, Tollit, and MacKay, in press). Male Weddell seals defend small underwater territories against other males; they sing on these territories and stop doing so if a female enters the territory to mate (Bartsh, Johnston, and Siniff, 1992). A large part of the species’ repertoire is produced only by males (Thomas and Kuechle, 1982). Bearded seals and leopard seals, the vocalizations of which vary geographically, also produce underwater songs in the breeding season (Ray, Watkins, and Burns, 1969; Stirling and Siniff, 1979). Further studies of these and other “singing” seals, like the crabeater seal (Lobodon carcinophagus) (Stirling and Siniff, 1979), the Ross seal (Ommatophoca rossi) (Watkins and Ray, 1985), and the walrus (Odobenus rosmarus) (Stirling, Calvert, and Spencer, 1987), should clarify the role of vocal learning in the acquisition of song. Behavioral observations of humpback whales, and the fact that it seems to be only males that sing (Tyack and Whitehead, 1983) and then mainly during the breeding season, suggest that song is a reproductive advertisement display and/or a territorial one. However, even though humpback whale and bowhead whale songs are very complex and change over the singing season, all males in a population sing the same song a t any particular time. We can only speculate on the origin of this phenomenon but, if songs are used to attract females, sexual selection holds a possible explanation. If a maximal response in the female was elicited by presenting the same acoustic pattern repetitively, the singing behavior of humpback whales could represent a communal vocal display that increases the responsiveness of females. The synchronized changes in song over time combined with song complexity could then be a result of conflicting pressures on individual males to make their own song more attractive by introducing new variations, but to maintain a communal display at the same time. Although whales are not easy to study, there is a clear need for more information about how their songs relate to other aspects of their behavior if we are to understand their functional significance. Other baleen whales that produce their sounds in repetitive songlike sequences are fin whales (Buluenopteru physulus) (Watkins, Tyack, Moore, and Bird, 1987) and blue whales (Balaenopteru musculus) (Cummings and Thompson, 1971). However, mysticetes also produce a variety of other social sounds. Repetitive sequences of the same sound could also be used to coordinate group movements. Since low-frequency sounds of whales can travel long distances (e.g., Cummings and Thompson, 1971), it is hard to
78
VINCENT M. JANIK AND PETER J. B. SLATER
determine whether widely spaced animals belong to one social group or are solitary individuals that happen to be within hearing range. The calls of some bats, like the hammer-headed bat (Hypsignachus monstrosus) (Bradbury, 1977) and the African false vampire bat (Cardioderma COT) (McWilliam, 1987), are used to attract females or act in territorial defense. Studies of whether and to what extent these species learn their calls would be interesting given the small amount of information on vocal learning in bats. B. INTRASEXUAL SELECTION A N D RESOURCE DEFENSE
Vocal learning could also have evolved in relation to territory maintenance and defense. In birds, small song repertoires are often used in matched countersinging between neighboring males on their territories, and the precise matching of songs that learning allows may confer a reproductive advantage. Payne, Payne, and Doehlert (1988) showed that male indigo buntings (Passerina cyanea) that share songs with neighbors do better in various measures of reproduction, including fledging more young. Song learning might enable birds to mimic established and successful individuals (Payne, 1981). This could discourage intruders and result in improved intruder detection if shared songs cannot be reproduced by foreign birds (see also Section IV,D on group recognition). Furthermore, the development of varied song repertoires, which seems to depend on learning, as they have been recorded only in species where vocal learning occurs, can also have a deterrent effect as far as intruders looking for territories are concerned. This has been shown most clearly in the speaker replacement experiments carried out by Krebs, Ashcroft, and Webber (1978) on great tits (Parus major). Krebs (1976) argued that this might be because repertoires give the impression that there are several birds present rather than just one. If repertoire size is a measure of fitness, it could also be used to assess a male’s fighting ability in territorial conflicts. In animals that defend territories it is often unclear whether females choose a particular male because of his vocal display or because of territory quality. In contrast to many terrestrial animals, underwater territories of marine mammals do not hold useful resources for the raising of offspring. The mating system of most singing marine mammals resembles more that of a lek. Therefore, if singing territories are positioned on the migration routes of females, the choice of a territory might have a considerable influence on breeding success. In that case vocal learning might have evolved through intrasexual selection. In marine mammals, there is some evidence that song might be used in male spacing on the breeding grounds. Weddell seals defend small underwater territories next to female haul-out
VOCAL LEARNING IN MAMMALS
79
sites while they are singing (Watkins and Schevill, 1968; Bartsh et al., 1992). Distances between singing humpback whales are about 2.4 times bigger than between nonsinging whales, suggesting a spacing function of song (Frankel, Clark, Herman, and Gabriele, 1995). No evidence for vocal learning has been found in mammals that defend females or resource-based territories against other males. Examples are elephant seals and gibbons. Elephant seals defend breeding territories around their harems on land and produce threat vocalizations to defend these territories against other males (Le Boeuf and Peterson, 1969b). In gibbons song is thought to function in territory maintenance and pair bonding but not in mate attraction (Mitani, 1985, 1988). In social groups vocal learning could be used in a variety of ways to gain access to resources. In highly social animals the ability to match sound features of other individuals might help to establish and maintain social relationships, or even raise the possibility of deception through the mimicry of sounds made by dominant animals. In bottlenose dolphins the formation of alliances between males has been observed (Connor, Smolker, and Richards, 1992). Imitation of their signature whistles could be used to maintain social bonds or deceive other individuals alike. In primates increasing sociality and the formation of complex social relationships could provide reasons for vocal learning to arise. Chimpanzees are known to form alliances against other group members (Harcourt, 1988) and the behavior of monkeys in social interactions sometimes involves what appears to be deception (reviewed in Cheney and Seyfarth, 1990). However, they do not seem to be capable of vocal learning. There are two possible explanations for this difference between primates and cetaceans. It could mean that vocal learning did not evolve because of advantages in social interactions. Alternatively, it may be because primates use other modes of communication (e.g., facial/gestural displays) to achieve similar results. Poor underwater visibility and the limited ability to use gestures and facial expressions in cetaceans could have favored the use of vocal communication more than in primates. C . INDIVIDUAL RECOGNITION
Nottebohm (1972), in a classic paper on the origins of vocal learning, argued against individual recognition being an important factor in the evolution of vocal learning in birds. Learning, as he points out, often leads to precise similarities between animals, and these would hinder rather than aid individual recognition. Furthermore, slight individual differences in the morphology of the sound production apparatus within a population, if consistent over time, usually introduce enough variability in unlearned sounds to allow individual recognition (Beer, 1969). Individual recognition
80
VINCENT M. JANIK AND PETER J. B. SLATER
on the basis of simple unlearned calls has been demonstrated to occur in many bird species, such as auks (Tschanz, 1968) and gulls (Beer, 1969), in which vocal learning is unknown. However, set against these arguments, there are ways in which vocal learning may enhance differences between individuals and, especially in high-background-noise environments, this may be important as an aid to discrimination. Even though birds learn their songs from conspecifics, the frequently learn different songs from different individuals so that their repertoires differ. New elements may also appear in their songs as a result of copying errors, or a bird may combine elements from different songs to form a new song (Slater and Ince, 1979). Thus, learning can lead both to new repertoires and to new songs within a bird’s repertoire, so that it is not correct to assume that learning necessarily leads to greater call similarity between individuals. Indeed, one route to individually specific calls may be by combining features learned from several other animals. All groups of mammals that show evidence for vocal learning do live in high-background-noise environments. Bat roosts, in which thousands of animals vocalize in the same frequency range, are one example, sea mammals provide another (Spiesberger and Fristrup, 1990). Additionally, in diving animals air-filled cavities that are involved in sound production get distorted with changing pressure. This could change voice characteristics considerably and mask individually specific cues that could otherwise be used for individual recognition (Tyack, 1991). If individual recognition requires more variability in calls than arises from differences in vocal tract morphology, or if such cues become unreliable during diving, improvisation rather than vocal learning might provide the answer. But there are theoretical reasons for thinking that this way of producing a completely new behavior pattern is a very difficult task. Every pattern that is generated reflects in some way the mechanism by which it is produced. For this reason, computer programs can create only pseudo-random numbers (Morgan, 1984). Even though some birds seem to produce completely new songs in their ontogeny (Marler, Mundinger, Waser, and Lutjen, 1972), improvisation has rarely been reported, suggesting that composing a new sound out of parts of others is an easier way to produce a unique individually specific call. It also ensures that the new call is not accidentally similar to that of other group members. Vocal learning may thus lead to individual distinctiveness by allowing the individual to produce a distinctive new call type that has not been present in the repertoire of its social group before. Examples of mammals that seem to make their signals different from those of other group members are dolphins (Caldwell and Caldwell, 1965,1968), big brown bats (Rasmuson and Barclay, 1992) and banner-tailed kangaroo rats (Randall, 1995). However, if individual recognition is why vocal learn-
VOCAL LEARNING IN MAMMALS
81
ing evolved, it must be supposed that there are benefits to the individual in allowing itself to be identified. Investigations of vocal learning in bats have concentrated on call similarities between mothers and pups. All species mentioned in Section II1,A are colonial. Mothers leave the roost every night to feed and have to find their offspring on return. Here, the gain in being identified is clear because the offspring will die if not fed frequently. Interestingly, infant greater horseshoe bats of 5 to 7 weeks of age that are still suckled do not have specific isolation calls (Matsumura, 1979), but show learned modifications in their echolocation calls. Perhaps their ability to match the main frequency of the mother’s echolocation call is sufficient for mother-pup recognition so that specific isolation calls are not needed. The need for effective individual recognition does seem to be the most likely reason why vocal learning has evolved in these bats. However, more studies, especially on species that do not breed in big roosts, are needed to assess whether this is the only possible explanation. Information on functions of odontocete sounds is still scarce, but the evidence we have points toward their use in individual recognition and group cohesion. These animals use most of their calls in social interactions, and have never been observed to show singing behavior like that of mysticetes. Even though evidence for vocal learning has been found in every cetacean that has been examined, mysticetes and odontocetes have very different social systems. While mysticetes seem to live alone most of the time, odontocetes live in relatively stable social groups (Tyack, 1986a). Bottlenose dolphins have a large repertoire of whistles, but most of those produced by an isolated individual are of a particular form more or less peculiar to itself and thus termed its “signature whistle” (Caldwell and Caldwell, 1965,1968; Caldwell, Caldwell, and Tyack, 1990).These signature whistles can remain stable for at least 12 years (Sayigh, Tyack, Wells, and Scott, 1990). This stability of signature whistles and their frequent production by isolated individuals supports the idea that they have a role in individual recognition. Tyack (1986b) found that 77% of all whistles of two interacting dolphins fell into two categories. One whistle was largely produced by one animal and the second mainly by the other. Tyack suggested that these were the signature whistles of the two animals, and that the fact that both individuals could produce both whistles may represent mimicry. Bottlenose dolphins have been shown to hunt cooperatively (Hoese, 1971) and support each other if injured (Lilly, 1963). In these situations it is certainly of advantage to a vocalizing animal to be individually identified by its allies. In a playback experiment Sayigh (1992) showed that mother bottlenose dolphins were also more likely to turn toward the signature
82
VINCENT M. JANlK AND PETER J. B. SLATER
whistles of their independent offspring (and vice versa) than toward those of other dolphins. Smolker, Mann, and Smuts (1993) presented evidence for use of signature whistles in mother-calf reunions. Individually specific signals are very common in odontocetes (e.g., Delphinus delphis, Caldwell and Caldwell, 1968; Lagenorhynchus obliquidens, Caldwell and Caldwell, 1971; Physeter macrocephalus, Watkins and Schevill, 1977; Stenella plagiodon, Caldwell, Caldwell, and Miller, 1973) and add to the evidence that vocal learning might have evolved in this group because of benefits it brought to individual recognition and group cohesion.
D. FAMILIAL OR GROUP RECOGNITION In familial or group recognition we encounter the same problem as in individual recognition. There must be some benefit to each individual to be recognized by its group or family members for vocal learning to evolve. Possible reasons are avoidance of inbreeding, cooperation between group members, or the identification and exclusion of strangers. All these arguments have been put foward for birds. Inbreeding avoidance is unlikely to be important, as male birds do not often sing the same songs as their fathers, as would be necessary if females were to use song as a cue. In Darwin’s finches (Geospiza spp.), which include some of the small number of species where fathers and sons are known to sing the same songs, females have been found to mate randomly in relation to song type (Millington and Price, 1985). Group recognition is perhaps more likely as birds usually respond more aggressively to alien songs than to those of neighbors, and, in colonies and in groups of territories, sharing of vocalizations through learning is common. This led Feekes (1977, 1982) to put forward the idea that shared songs might act as a “password” in the colonies of caciques (Cacicus cela) that she studied. It is striking that most odontocetes show individually specific calls, while killer whales have group-specific ones. The social organization of killer whales, with their stable family groups, is very different from that of bottlenose dolphins. Most dolphins live in fission-fusion societies with few stable associations between individuals (reviewed in Norris and Dohl, 1980). Killer whales tend to stay in their parental group throughout their lives (Bigg et al., 1990). They hunt cooperatively on a variety of different prey, ranging from herring to other marine mammals (Smith, Siniff, Reichle, and Stone, 1981; Simila and Urgate, 1993), and food sharing within pods has been observed (Hoelzel, 1991). Here, family-specific calls could help to avoid inbreeding or be used to maintain social bonds between group members and to exclude foreign individuals.
VOCAL LEARNING IN MAMMALS
83
As with individual recognition, vocal learning could have been a good solution for marine mammals to the problems imposed on group recognition calls by the environment. High background noise could easily mask the subtle effect of differences in vocal tract morphology between different matrilines, making them useless for group recognition. Learning would allow animals to produce completely new sounds that are different enough to be recognized even in noisy conditions. The production of new, very distinctive sounds could also compensate for changing voice characteristics caused by the effects of water pressure on the air-filled vocal tract at different diving depths. In the more fluid social system of bottlenose dolphins, individual recognition to maintain social relationships is likely to be more important. These animals associate in temporary groups of variable size and composition. However, certain long-term associations have been discovered between related females (Wells, Scott, and Irvine, 1987). Small subgroups of two to three males also form relatively stable alliances (Connor et al., 1992; Smolker, Richards, Connor, and Pepper, 1992). Vocal learning could be of advantage in alliance formation and maintenance in this species. Males have been shown to cooperate with other males in aggressive interactions and in herding of females (Connor et al., 1992). Male alliances could use signature whistle mimicry or learned alliance specific calls to maintain their bonding and exclude strangers. Off Sarasota, Florida, approximately half the male bottlenose dolphins develop signature whistles that are very similar to those of their mother, while most females produce signature whistles highly distinct from those of their mothers (Sayigh et al., 1990; Sayigh, Tyack, Wells, Scott, and Irvine, 1995). It seems unlikely that only males develop family-specific calls as a kinship label that facilitates inbreeding avoidance because, while females tend to associate in matrilines later in life, males do not associate closely with their female relatives, though remaining in the same general area. Females may need to develop a signature whistle as different as possible from their mother’s to avoid misidentification. The higher degree of similarity of signature whistles between mothers and sons could stem from the lack of this requirement or it could benefit males if matriline affects dominance in dolphin societies. There is some evidence that it is the sons of only certain females that produce signature whistles like those of their mothers (Sayigh et al., 1995), but further investigation is needed.
E. POPULATION IDENTITY The development of dialects between neighboring populations of potentially interbreeding individuals could lead to assortative mating; this in turn
84
VINCENT M. JANIK AND PETER J. B. SLATER
might benefit individuals if there are local genetic adaptations that can thereby persist. Nottebohm (e.g., 1972) suggested that dialects that are common in birds might have this effect. Although the idea has received some subsequent support, particularly in the white-crowned sparrow (Zonotrichia leucophrys, Baker and Cunningham, 1985), the weight of evidence is against it. For example, white-crown dialect boundaries do not seem to limit dispersal, song learning in males may occur after dispersal, and females often mate with males singing a different dialect from their natal one (see Catchpole and Slater, 1995, pp. 205-209 for a more detailed discussion). Even though this idea is now generally discounted in birds, it could still be true for marine mammals and bats that possess a similar potential for quick dispersal. If these animals return to their home area to mate, dialects might help to maintain local adaptations. Even in the relatively homogeneous marine environment, differences in local adaptations could exist between coastal and pelagic populations or between areas with different prey species. In many of the examples of geographic variation in mammal calls the actual extent of each population is unknown. Even though some of the locations where seals have been found to differ in their vocalizations are several thousand kilometers apart, these species are mobile as well as widely distributed and could easily cover such distances in their migrations. Humpback whale populations in different oceans on the other hand may well be truly isolated and their dialects are therefore unlikely to be adaptive in maintaining population identity. Variations in call structure of different primate populations have been interpreted as possible evidence for vocal learning by several authors. If call variations between neighboring populations are actually learned, a function in population recognition might be a reason. However, there is no clear evidence for learned differences yet. F. INTENSE SPECIATION Nottebohm (1972) discussed intense speciation as one possible factor in the development of vocal learning in birds. He argued that, with large numbers of species in small geographic areas, vocal learning might have evolved among passerine birds because of the need for rapid change in signals. Subsequently, it was suggested that vocal learning might itself have enhanced speciation in passerines, if learned signals acted to restrict gene flow. However, Baptista and Trail (1992) argue that there is little evidence that differences in bird vocal signals act as a barrier to interbreeding. In mammals, vocal learning does not appear to correlate with rapid speciation, making either of these ideas unlikely to apply. Marine mammals could have gone through a phase of intense speciation when they returned
VOCAL LEARNING IN MAMMALS
85
from land to sea. However, this seems unlikely to have put any pressure on the development of vocal learning for species recognition, as the available space for dispersal was so great and the habitat was relatively homogeneous.
G. HABITAT MATCHING Different habitats can have very different sound transmission characteristics (Marten and Marler, 1977; Marten, Quine, and Marler, 1977; Wiley and Richards, 1978; Waser and Brown, 1986). If a species lives in various different habitats, or if the transmission characteristics of its habitat change frequently, vocal learning could help to optimize its vocal signals. This idea was originally put forward by Hansen (1979) for birds, many species of which disperse widely and live in a variety of habitats. Data in support of it have been obtained on two species: the great tit (Parus major, Hunter and Krebs, 1979) and the chingolo sparrow (Zonorrichiu cupensis, King, 1972; Nottebohm, 1975). Particularly detailed studies by Handford (1981, 1988; Handford and Lougheed, 1991) suggest that trill rates in this last species in agricultural areas match the habitat that was present before cultivation began, thus pointing to a very slow rate of change. Studies on primates have shown that some species do have calls that are matched to the transmission qualities of their habitat (Brown, Gomez, and Waser, 1995). However, there is no clear evidence for vocal learning in primates. In marine mammals it is unlikely that vocal learning has evolved because of advantages in habitat matching because, compared to most terrestrial environments, the sea has very stable sound transmission characteristics (Spiesberger and Fristrup, 1990). In bats all evidence for vocal learning involves mother-pup interactions. However, it could be possible that vocal learning enables them to match their echolocation calls to particular habitats. Further studies are needed to investigate this possibility.
V. CONCLUSIONS A. FLEXIBILITY I N VOCAL LEARNING We have identified several different levels of complexity in vocal learning. The most simple, in which animals can be trained to alter the overall duration and amplitude of a sound, seems to be a relatively common feature of mammalian communication systems (e.g., bottlenose dolphin, Lilly, 1965; domestic cat, Molliver, 1963; rhesus monkey, Sutton et al., 1973). It is likely that this form of sound alteration is closely linked to contextual learning
86
VINCENT M. JANIK AND PETER J.
B. SLATER
of sounds. If an animal has learned to withhold or produce a sound in different contexts, it has a certain degree of learned control over the onset and offset of calling. An alteration of temporal parameters, like intercall intervals in sound sequences, in relation to auditory experience would thus become possible. This control over the muscles of the respiratory system and their coordination with the vocal apparatus might also be sufficient for learning how to alter total amplitude and duration of single calls. A more elaborate form of this sort of learning might involve significant changes in voice-onset time in a call or complex amplitude modulations that superimpose additional frequency bands on a call. To our knowledge there are as yet no studies of the role of learning in modifying these parameters. The second stage, in which an animal is able to learn how to alter certain frequency parameters to match another individual’s calls, has been demonstrated only in greater horseshoe bats, marine mammals and humans. Here, the activity of muscles of the vocal apparatus itself needs to be modifiable by experience. This needs a different level of neural control and is a significant next step in the evolution of vocal learning. While there are other reasons why an animal might gain control over the respiratory system to make it modifiable by experience (e.g., contextual learning in relation to vocal communication, diving), the vocal apparatus is used only in call production. However, not many studies have looked at this relatively limited form of vocal learning yet. It may be more common than it seems at this point. The ability to copy completely new sounds seems to be rare among mammals. Our review has shown that it is known to occur only in marine mammals and humans. However, we do not know whether the distinction between the ability to imitate new sounds and that to change only certain parameters in a limited way is a real one. More studies focusing on the extent to which calls are modifiable through vocal learning are needed. Bats, for example, might be capable of more drastic changes than the ones we know of so far. B. FACTORS AFFECTING THE EVOLUTION OF VOCAL LEARNING: THEROLEOF MOBILITY Many bird species in three different orders (Passeriformes, Apodiformes, Psittaciformes) show vocal learning, suggesting that it evolved separately in each of these groups. In mammals it is known to occur in humans, cetaceans, phocid seals, and bats, taxa that do not share a unique common ancestor. Here, too, it seems vocal learning has evolved independently in each group. Given this lack of a direct phylogenetic connection, it is interesting to ask whether there is a common factor that could have caused a convergent evolution of vocal learning in these groups.
VOCAL LEARNING IN MAMMALS
87
We have identified several possible reasons why vocal learning could have evolved in different mammalian groups. If we examine features that these groups have in common with each other and with birds, an obvious similarity is their high mobility. With the exception of humans, they all spend at least part of their lives swimming or flying in three-dimensional environments. The high mobility of animals living in air or water is a fundamental common factor that has influenced various different aspects of their lives. It is more difficult for males to monopolize females. Members of a social group are more likely to lose contact in three-dimensional space. Vocal signals are likely to be used more because animals can disperse more quickly and visual contact is difficult to maintain. The increased use of vocal communication instead of visual signals increases background noise in the acoustic frequency bands used for communication and adds pressure toward signal diversity. Thus, living in these environments could make the development of vocal learning in a species more likely. There are, of course, other mammals, like arboreal or nocturnal ones, that face similar problems to a lesser extent. However, the impact of such problems is certainly highest in flying and swimming animals, as they disperse much more quickly and are not limited to moving only along solid structures in their environment, as are tree-living species, for example.
C. IMPLICATIONS FOR THE EVOLUTION OF HUMAN LANGUAGE In primates, researchers have been looking for evidence of vocal learning for a long time. Even though they clearly have considerable control over their vocal utterances, including the duration and amplitude of their calls, vocal learning has not been unequivocally established in primates other than humans. The fact that dolphins, seals, and many birds spontaneously start to imitate sounds from their acoustic environment in captivity suggests that imitation plays an important role in their lives. Such imitation of new sounds has never been observed in nonhuman primates. It seems strange that our closest relatives show no evidence for vocal imitation, yet humans are so adept at it. The suggested environment of early humans does not seem to have been strikingly different from those of other primates. Thus, all the possible advantages of vocal learning mentioned in Section IV would have been present for other primates, too. Why then did the ability to imitate sounds evolve in humans and not in nonhuman primates? Jakobson (1941) claimed that all the sounds of the world’s languages occur in infant vocalizations, suggesting that the learning of new sounds is not involved in language acquisition. However, today most researchers accept that humans are capable of vocal learning. More recent discussions on natural predisposition for language learning have concen-
88
VINCENT M. JANlK AND PETER J. B. SLATER
trated on the acquisition of language as a communication system (Bickerton, 1990; Pinker and Bloom, 1990), not on whether vocal learning is involved in the acquisition of the sound repertoire. A convincing explanation for the difference between nonhuman and human primates relates to the evolution of language. It might seem difficult to imagine how imitating sounds could have been of any advantage to the first individual that did so. This is certainly true if we assume that the evolution of language relied on vocal learning. However, several authors have argued that language developed in gestures before it affected vocal behavior (Hewes, 1973; Parker and Gibson, 1979). Once a gestural communication system with learned signals had become established, vocal learning would have been greatly favored because it makes it possible to use language even where visual contact is absent. If this sequence of events is correct, the lack of evidence for vocal learning in nonhuman primates would not be so surprising. However, more research, especially on the possibility of a more limited form of vocal learning in nonhuman primates, is needed.
D. FUTURE RESEARCH We have suggested that sexual selection, defense of resources, and individual recognition are the most likely reasons why vocal learning evolved in mammals, assuming that its functional significance has not changed since then. However, further studies are needed to find out how flexible the vocal system of each group is and what they are using this flexibility for. The body of evidence for vocal learning in mammals is still very small compared with that in birds. In many very vocal mammals, such as elephants, vocal learning has not been studied at all. We have already mentioned various other species that would be very interesting to study in this context. The most powerful experimental approaches to the study of vocal learning are certainly conditioning experiments and studying the effects of keeping experimental animals in controlled acoustic environments. If an animal can be trained to imitate a new sound that has not been in its repertoire before or if infants that have been exposed to different stimuli match what they have been hearing in detail, we have found unequivocal evidence for vocal learning. As our survey has shown, many authors interpret less convincing results as evidence for vocal learning. Geographic variation, differences in vocalizations between different matrilines or between normally raised and isolated individuals, changes in the vocal repertoire during ontogeny, and changes in vocal behavior after being housed with new individuals have all been interpreted as indicating vocal learning. All these examples could involve vocal learning, but they do not represent unequivo-
VOCAL LEARNING IN MAMMALS
89
cal evidence for it. However, they give valuable information on species that could be interesting subjects in which to investigate its existence. Recent investigations of primate communication systems have revealed a greater flexibility than previously thought. Elowson and Snowdon (1994), for example, found significant alterations in call parameters of pygmy marmosets in relation to changes in their social environment. Because individuals did not match conspecifics in this study, trying to show that vocal learning was responsible would be difficult, but it deserves further investigation. In chimpanzees, in which individuals seem to match the pant hoots of their chorusing partners (Mitani and Brandt, 1994), this might be easier. Crossfostering experiments within this species might be able t o give clearer results on whether a limited form of vocal learning exists in nonhuman primates. Concurrent investigations of gestural communication in wild great apes could give information on what role learned gestures play in their natural communication system. Such studies could indicate how likely it is that gestural or vocal communication was the basis for language evolution. Even in those groups where evidence for vocal learning has been presented, information on its significance and flexibility is often lacking. The evidence in bats, for example, is still sparse, though it seems clear that their learning is not as versatile as that in marine mammals. Training bats with different stimuli would help to find out to what extent learning can influence the development of their communication and echolocation calls. More experiments on vocal learning in those marine mammals that show singing behavior are also needed. We still do not know how widespread vocal learning is in these groups and how they learn their songs. In seals, phocids seem to be capable of vocal learning, but there is no evidence for it in otariids. This could simply reflect a lack of studies, but there are also marked differences in the mating strategies of otariids and phocids that could explain the apparent discrepancy. Apart from the elephant seal, all the phocids that show geographic variation in their vocalizations breed at least partly on ice and copulate in the water (Stirling, 1975). This makes it difficult for a male to monopolize several females: they are either scattered on available haul-out sites or else in the water where there is poor visibility combined with increased mobility in all directions. Otariids, on the other hand, breed on land and males often defend harems against other males. The fact that singing behavior has so far been found only in phocids suggests that vocal learning may have evolved in relation to their mating strategy. However, the only clear evidence for vocal learning in seals comes from a harbor seal that imitated human speech spontaneously. There are as yet no further studies of vocal learning in seals, and there is a clear need for them. Comparative studies of how background noise influences call development and usage might also shed light on the origin of vocal learning. While
90
VINCENT M. JANIK AND PETER J. 0. SLATER
odontocetes seem to use vocal learning to improve individual recognition, mysticetes apparently use it only in their singing behavior. We d o not know whether vocal learning evolved once or twice in the lineage of cetaceans. Studies on how sexual selection on one hand and individual recognition on the other influence vocal behavior are needed to clarify how vocal learning evolved and persisted in these animals. Further studies on the vocal communication systems of toothed whales would be especially valuable. Even though individual recognition seems to be a likely reason for vocal learning to have evolved in dolphins, vocal learning could be used extensively in their complex social systems, including the possibility to use it in vocal deception. A study of how dolphins use imitation in the wild might open up a new perspective on why their vocal learning evolved. Given that its function might have changed in a particular species, and that dolphins show remarkable cognitive capacities (review in Herman, Pack, and Morrel-Samuels, 1993), including the ability to process syntactical information in signal sequences (Herman, Richards, and Wolz, 1984), it is likely that, once evolved, vocal learning had profound effects on various aspects of their natural communication.
VI. SUMMARY Vocal learning, as we discuss it in this review, refers to instances where vocalizations are modified in form as a result of experience with those of other individuals. While many birds are capable of vocal learning, unequivocal evidence for it is rare in mammals. The most versatile mammalian vocal learners are cetaceans, harbor seals, and humans, all of which are able to imitate new sounds. Greater horseshoe bats learn the main frequency of their echolocation calls from their mothers and are the only other mammals shown so far to be capable of learning t o change frequency parameters in their calls. Nonhuman primates have been conditioned to alter the duration and amplitude of their calls but not their frequency parameters. We suggest that sexual selection, defense of resources, and individual recognition are the most likely reasons why vocal learning has evolved in mammals. However, we know little about the functional significance of vocal learning for these animals and more studies are badly needed.
Acknowledgments We would like to thank Karen McComb, Laela S. Sayigh, Robert M. Seyfarth, and Charles T. Snowdon for helpful comments on the manuscript. V. M. Janik is a fellow of the Studienstif-
VOCAL LEARNING IN MAMMALS
91
tung des Deutschen Volkes and was supported by a DAAD-Doktorandenstipendium aus Mitteln des zweiten Hochschulsonderprogramms.
References Aitken, P. G., and Wilson, W. A. (1979). Discriminative vocal conditioning in Rhesus monkeys: Evidence for volitional control? Brain and Language 8,227-240. Awbrey, F., Evans, W., Jehl, J., Thomas, J., and Leatherwood, S. (1982). Comparison of Ross Sea and Pacific Northwest killer whale vocalizations. Rep. Int. Whal. Commn. 32,667-670. Bain, D. E. (1986). Acoustic behavior of Orcinus: Sequences, periodicity, behavioral correlates and an automated technique for call classification. In “Behavioral Biology of Killer Whales” (B. C. Krikevold and J. S. Lockard, eds.), pp. 335-371. Alan R. Liss, New York. Baker, M. C., and Cunningham, M. A. (1985). The biology of bird-song dialects. Behav. Brain Sci. 8, 85-133. Baptista. L. F.. and Schuchmann, K. (1990). Song learning in the Anna hummingbird (Calypfe anna). Ethology 84, 15-26. Baptista, L. F., and Trail, P. W. (1992).The role of song in the evolution of passerine diversity. Syst. Biol. 41, 242-247. Bartholomew. G . A., and Hubbs, C. L. (1960). Population growth and seasonal movements of the northern elephant seal, Mirounga angusrirosfris. Mammalia 24, 313-324. Bartsh. S. S.. Johnston, S. D., and Siniff, D. B. (1992). Territorial behaviour and breeding frequency of male Weddell seals (Lepfonychotes weddelli) in relation to age, size, and concentration of serum testosterone and cortisol. Can. J . Zool. 70,680-692. Beer, C. G. (1969). Laughing gull chicks: Recognition of their parents’ voices. Science 166, 1030-1032. Bickerton, D. (1990). “Language and Species.” University of Chicago Press, Chicago. Bigg, M. A,, Olesiuk, P. F., Ellis. G. M., Ford, J. K. B., and Balcomb, K. C. (1990). Social organization and genealogy of resident killer whales (Orcinus orca) in the coastal waters of British Columbia and Washington State. Rep. fnr. Whal. Commn. (Special Issue) 12, 383-405. Bowles, A. E., Young, W. G., and Asper, E. D. (1988). Ontogeny of stereotyped calling of a killer whale calf, Orcinus orca, during her first year. Rir Fisk. 11, 251-275. Bradbury, J. W. (1977). Lek mating behavior in the hammer-headed bat. Z . Tierpsychol. 45,225-255. Brockelman. W. Y., and Schilling. D. (1984). Inheritance of stereotyped gibbon calls. Nature (London) 312,634-636. Brown, C. H.. Gomez, R.. and Waser, P. M. (1995). Old world monkey vocalizations: Adaptation to the local habitat? Anim. Behav. 50,945-961. Burnstein. D. D.. and Wolff, P. C. (1967). Vocal conditioning in the guinea pig. Psychon. Sci. 8, 39-40. Caldwell. M. C.. and Caldwell, D. K. (1965). Individualized whistle contours in bottlenose dolphins (Tursiops rruncatus). Nature (London) 207, 434-435. Caldwell, M. C., and Caldwell, D. K. (1968). Vocalization of naive captive dolphins in small groups. Science 159, 1121-1123. Caldwell, M. C., and Caldwell, D. K. (1971). Statistical evidence for individual signature whistles in Pacific whitesided dolphins, Lagenorhynchus obliquidens. Cefology 3, 1-9. Caldwell, M. C., and Caldwell, D. K. (1972). Vocal mimicry in the whistle mode by an Atlantic bottlenosed dolphin. Cerology 9, 1-8.
92
VINCENT M. JANIK AND PETER J. B. SLATER
Caldwell, M. C., and Caldwell, D. K. (1979). The whistle of the Atlantic bottlenosed dolphin (Tursiops truncatus)-ontogeny. In “Behavior of Marine Animals: Current Perspectives in Research. Vol. 3. Cetaceans” (H. E. Winn and B. L. Olla, eds.), pp. 369-401. Plenum Press, New York. Caldwell, M. C., Caldwell, D. K., and Miller, J. F. (1973). Statistical evidence for individual signature whistles in the spotted dolphin, Stenella plagiodon. Cetology 16, 1-21. Caldwell, M. C., Caldwell, D. K., and Tyack, P. L. (1990). Review of the signature-whistlehypothesis for the Atlantic bottlenose dolphin. In “The Bottlenose Dolphin” (S. Leatherwood and R. R. Reeves, eds.), pp. 199-234. Academic Press, New York. Catchpole, C. K. (1980). Sexual selection and the evolution of complex songs among warblers of the genus Acrocephalus. Behaviour 74,149-166. Catchpole, C. K., and Slater, P. .I. B. (1995). “Bird Song: Biological Themes and Variations.’’ Cambridge University Press, Cambridge. Cheney, D. L., and Seyfarth, R. M. (1990). “How Monkeys See the World: Inside the Mind of Another Species.” University of Chicago Press, Chicago. Cleator, H. J., Stirling, I., and Smith, T. G. (1989). Underwater vocalizations of the bearded seal (Erignarhus barbatus). Can. J. Zool. 67, 1900-1910. Conner, D. A. (1982). Geographic variation in short calls of pikas (Ochotonu princeps). J. Mamm. 63,48-52. Connor, R. C., Smolker, R., and Richards, A. F. (1992). Two levels of alliance formation among male bottlenose dolphins (Tursiops sp.). Proc. Nail. Acud. Sci. U.S.A.89,987-990. Cummings, W. C., and Holliday, D. V. (1987). Sounds and source levels from bowhead whales off Pt. Barrow, Alaska. J. Acoust. SOC.Am. 82, 814-821. Cummings, W. C., and Thompson, P. 0. (1971). Underwater sounds from the blue whale, Balaenoprera musculus. J. Acousr. SOC.Am. 50, 1193-1 198. Eaton, R. L. (1979). A beluga whale imitates human speech. Carnivore 2,22-23. Eens, M., Pinxten, R., and Verheyen. R. F. (1991). Male song as a cue for mate choice in the European starling. Behaviour 116,210-238. Elowson, A. M., and Snowdon, C. T. (1994). Pygmy marmosets, Cebuella pygmueu, modify vocal structure in response to changed social environment. Anim. Behav. 47,1267-1277. Esser, K. H. (1994). Audio-vocal learning in a non-human mammal: The lesser spear-nosed bat Phyllostomus discolor. NeuroReport 5, 1718-1720. Esser, K. H., and Schmidt, U. (1989). Mother-infant communication in the lesser spear-nosed bat Phyllostomus discolor (Chiroptera, Phy1ostomidae)-Evidence for acoustic learning. Ethology 82, 156-168. Feekes, F. (1977). Colony specific song in Cacicus cela (Icteridae, Aves): The pass-word hypothesis. Ardea 65, 197-202. Feekes, F. (1982). Song mimesis within colonies of Cacicus c. cela (Icteridae, Aves). A colonial password? Z. Tierpsychol. 58, 119-152. Ford, J. K. B. (1991). Vocal traditions among resident killer whales (Orcinus orca) in coastal waters of British Columbia. Can. J. Zool. 69, 1454-1483. Ford, J. K. B., and Fisher, H. D. (1983). Group-specific dialects of killer whales (Orcinus orca) in British Columbia. In “Communication and Behavior of Whales” (R. Payne, ed.), pp. 129-161. Westview Press, Boulder, Colorado. Frankel, A. S., Clark, C. W., Herman, L. M., and Gabriele, C. M. (1995). Spatial distribution, habitat utilization, and social interactions of humpback whales, Megaptera novaeungliue, off Hawai’i, determined using acoustic and visual techniques. Can. J. Zool. 73,1134-1 146. Gaunt, S. L. L., Baptista, L. F., Sanchez, J. E., and Hernandez, D. (1994). Song learning as evidenced from song sharing in two hummingbird species (Colibri coruscans and C. thalussinus). Auk 111, 87-103. Geissmann, T. (1983). Female capped gibbon (Hylobates pileatus Gray 1891) sings male song. J. Hum. Evol. 12,667-671.
VOCAL LEARNING IN MAMMALS
93
Geissmann, T. (1984). Inheritance of song parameters in the gibbon song, analysed in 2 hybrid gibbons (Hylobates pileatus X H. lar). Folia Primatol. 42, 216-235. Geissmann, T. (1987). Songs of hybrid gibbons (Hylobates pileatus X H. lar). Int. J. Primatol. 8, 540. Green, K., and Burton, H. R. (1988). Do Weddell seals sing? Polar Biol. 8, 165-166. Green, S. (1975). Dialects in Japanese monkeys: Vocal learning and cultural transmission of locale-specific vocal behavior? Z. Tierpsychol. 38, 304-314. Guinee, L. N., Chu, K., and Dorsey, E. M. (1983). Changes over time in the song of known individual humpback whales (Megaptera novaeangliae).In: “Communication and Behavior of Whales” (R. Payne, ed.), pp. 59-80. Westview Press, Boulder, Colorado. Handford, P. (1981). Vegetational correlates of variation in the song of Zonotrichia capensis. Behav. Ecol. Sociobiol. 8,203-206. Handford, P. (1988). Trill rate dialects in the rufous-collared sparrow, Zonotrichia capensis, in northwestern Argentina. Can. J. Zool. 66,2658-2670. Handford, P., and Lougheed, S. C. (1991). Variation in duration and frequency characters in the song of the rufous-collared sparrow, Zonotrichia capensis, with respect to habitat, trill dialects and body size. Condor 93, 644-658. Hanggi, E. B., and Schusterman, R. J. (1994). Underwater acoustic displays and individual variation in male harbour seals. Phoca vitulina. Anim. Behav. 48, 1275-1283. Hansen, P. (1979). Vocal learning: Its role in adapting sound structures to long-distance propagation and a hypothesis on its evolution. Anim. Behav. 27, 1270-1271. Harcourt, A. H. (1988). Alliances in contests and social intelligence. In “Machiavellian Intelligence: Social Expertise and the Evolution of Intellect in Monkeys, Apes, and Humans” (R. W. Byrne and A. Whiten, eds.), pp. 132-152. Oxford University Press, Oxford. Hauser, M. D. (1992). Articulatory and social factors influence the acoustic structure of rhesus monkey vocalizations: A learned mode of production? J. Acoust. SOC.Am. 91,2175-2179. Hayes, K. J. (1951). “The Ape in Our House.” Harper, New York. Herman. L. M., Pack, A. A., and Morrel-Samuels, P. (1,993).Representational and conceptual skills of dolphins. in “Language and Communication: Comparative Perspectives” (H. L. Roitblat, L. M. Herman, and P. E. Nachtigall, eds.), pp. 403-442. Lawrence Erlbaum Associates, Hillsdale, New Jersey. Herman, L. M., Richards, D. G., and Wolz, J. P. (1984). Comprehension of sentences by bottlenosed dolphins. Cognition 16, 129-219. Hewes, G. (1973). Primate communication and the gestural origin of language. Cur. Anthropol. 14, 5-24. Hodun, A., Snowdon, C. T., and Soini, P. (1981). Subspecific variation in the long calls of tamarin. Saguinus jiucicollis. Z. Tierpsychol. 57, 97-110. Hoelzel, A. R. (1991). Killer whale predation on marine mammals at Punta Norte, Argentina; food sharing, provisioning and foraging strategy. Behav. Ecol. Sociobiol. 29, 197-204. Hoese, H. D. (1971). Dolphin feeding out of water in a salt marsh. J. Mamm. 52, 222-223. Hopkins, W. D.. and Savage-Rumbaugh, E. S. (1991). Vocal communication as a function of differential rearing experiences in Pan paniscus: A preliminary report. Int. J. Primatol. 12,559-583. Hunter, M. L., and Krebs, J. R. (1979). Geographical variation in the song of the great tit (Parus major) in relation to ecological factors. J. Anim. Ecol. 48,759-785. Jakobson, R. (1941). “Kindersprache, Aphasie, und allgemeine Lautgesetze.” Almqvist and Wiksell, Uppsala. Jones, G., and Ransome, R. D. (1993). Echolocation calls of bats are influenced by maternal effects and change over a lifetime. Proc. R. SOC.Lond. B 252, 125-128.
94
VINCENT M. JANIK AND PETER J. B. SLATER
King, J. R. (1972). Variation in the song of the rufous collared sparrow, Zonotrichia capensis, in northwestern Argentina. Z . Tierpsychol. 30, 344-373. Krebs, J. R. (1976). Habituation and song repertoires in the great tit. Eehav. Ecol. Sociobiol. 1,215-227. Krebs, J. R.,Ashcroft, R., and Webber, M. (1978). Song repertoires and territory defence in the great tit. Nature (London) 271, 539-542. Kroodsma, D. E. (1976). Reproductive development in a female songbird: Differential stimulation by quality of male song. Science 192, 574-575. Kroodsma, D. E. (1984). Songs of the alder flycatcher (Empidonax alnorum) and willow flycatcher (Empidonax traillii) are innate. Auk 101, 13-24. Kroodsma, D. E. (1989). Male eastern phoebes (Sayornis phoebe; Tyrannidae, Passeriformes) fail to imitate song. J. Comp. Psychol. 103,227-232. Kroodsma, D. E., and Baylis, J. R. (1982). A world survey of evidence for vocal learning in birds. I n “Acoustic Communication in Birds” (D. E. Kroodsma and E. H. Miller, eds.), pp. 311-337. Academic Press, New York. Lal, H. (1967). Operant control of vocal responding in rats. Psychon. Sci. 8,35-36. Larson, C. R., Sutton, D., Taylor, E. M., and Lindeman, R. (1973). Sound spectral properties of conditioned vocalization in monkeys. Phonetica 27, 100-110. Le Boeuf, B. J. (1974). Male-male competition and reproductive success in elephant seals. Am. Zool. 14,163-176. Le Boeuf, B. J., Ainley, D. G.. and Lewis, T. J. (1974). Elephant seals on the Farallones: population structure of an incipient breeding colony. J . Mamm. 55, 370-385. Le Boeuf, B. J., and Peterson, R. S. (1969a). Dialects in elephant seals. Science 166,1654-1656. Le Boeuf, B. J., and Peterson, R. S. (1969b). Social status and mating activity in elephant seals. Science 163, 91-93. Le Boeuf, B. J., and Petrinovich, L. F. (1974). Dialects of northern elephant seals, Mirounga angustirostris: origin and reliability. Anim. Eehav. 22, 565-663. Lieblich, A. K., Symmes, D., Newman, J. D., and Shapiro, M. (1980). Development of the isolation peep in laboratory-bred squirrel monkeys. Anim. Eehav. 28, 1-9. Lilly, J. C. (1963). Distress call of the bottlenosed dolphin: Stimuli and evoked behavioral responses. Science 139, 116-118. Lilly, J. C. (1965). Vocal mimicry in Tursiops: Ability to match numbers and durations of human vocal bursts. Science 147, 300-301. Maeda, T., and Masataka, N. (1987). Locale-specific vocal behavior of the tamarin (Sanguinus labiains). Ethology 75, 25-30. Maples, E. G., Haraway, M. M., and Hutto, C. W. (1989). Development of coordinated singing in a newly formed siamang pair (Hylobares syndactylus). Zoo Eiol. 8, 367-378. Marler, P. (1991). Song-learning behavior: The interface with neuroethology. Trends Neurosci. 14,199-205. Marler, P., and Mitani, J. (1988). Vocal communication in primates and birds: Parallels and contrasts. In “Primate Vocal Communication” (D. Todt, P. Goedeking. and D. Symmes, eds.), pp. 3-14. Springer-Verlag, Berlin. Marler, P., Mundinger, P., Waser, M. S., and Lutjen, A. (1972). Effects of acoustical stimulation and deprivation on song development in red-winged blackbirds (Agelaius phoeniceus). Anim. Eehav. 20,586-606. Marler, P., and Tenaza, R. (1977). Signalling behavior of apes, with special reference to vocalization. In “How Animals Communicate” (T. Sebeok, ed.), pp. 965-1033. Indiana University Press, Bloomington. Marshall, J. T., Jr., and Marshall, E. R. (1976). Gibbons and their territorial songs. Science 193,235-237.
VOCAL LEARNING IN MAMMALS
95
Marten, K., and Marler, P. (1977). Sound transmission and its significance for animal vocalization. I: Temperate habitats. Eehav. Ecol. Sociobiol. 2, 271-290. Marten, K., Quine, D., and Marler, P. (1977). Sound transmission and its significance for animal vocalization. 11: Tropical forest habitats. Behav. Ecol. Sociobiol. 2, 291-302. Masataka, N. (1988). The response of red-chested moustached tamarins to long calls from their natal and alien populations. Anim. Eehav. 36,55-61. Masataka, N., and Fujita, K. (1989). Vocal learning of Japanese and rhesus monkeys. Eehaviour 109, 191-199. Masters, W. M.. Raver, K. A. S., and Kazial. K. A. (1995). Sonar signals of big brown bats, Eptesicus fuscus, contain information about individual identity, age and family affiliation. Anim. Behav. 50, 1243-1260. Matsumura, S. (1979). Mother-infant communication in a horseshoe bat (Rhinolophus ferrumequinum nippon): Development of vocalizations. J. Mamm. 60,76-84. McCowan, B.. and Reiss, D. (1995). Whistle contour development in captive-born infant bottlenose dolphins (Tursiops truncatus): role of learning. J. Comp. Psychol. 109,242-260. McWilliam, A. (1987). Territorial and pair behavior of the African false vampire bat, Cardioderma cor (Chiroptera: Megadermatidae), in coastal Kenya. J . Zool. 213,243-252. Millington, S . J., and Price, T. D. (1985). Song inheritance and mating patterns in Darwin’s finches. Auk 102, 342-346. Mitani, J. C. (1985). Responses of gibbons (Hylobates muelleri) to self, neighbor, and stranger song duets. Int. J. Primatol. 6, 193-200. Mitani, J. C. (1987). Species discrimination of male song in gibbons. Am. J. Primatol. 13, 413-423. Mitani, J. C. (1988). Male gibbon (Hylobates agilis) singing behavior-Natural-history, song variations and function. Ethology 79, 177-194. Mitani, J. C., and Brandt, K. L. (1994). Social factors influence the acoustic variability in the long-distance calls of male chimpanzees. Ethology 96,233-252. Mitani, J. C., Hasegawa, T., Gros-Louis, J., Marler, P., and Byrne, R. (1992). Dialects in wild chimpanzees? Am. J. Primatol. 27,233-243. Molliver, M. E. (1963). Operant control of vocal behavior in the cat. J. Exp. Anal. Eehav. 6, 197-202. Moore, P. W. B., and Pawloski, D. A. (1990). Investigations on the control of echolocation pulses in the dolphin (Tursiops truncatus).In “Sensory Abilities of Cetaceans” (J. Thomas and R. A. Kastelein, eds.), pp. 305-316. Plenum Press, New York. Morgan, B. J. T. (1984). “Elements of Simulation.” Chapman and Hall, New York. Morrice. M. G., Burton, H. R., and Green, K. (1994). Microgeographic variation and songs in the underwater vocalization repertoire of the Weddell seal (Leptonychotes weddelli) from the Vestfold Hills, Antarctica. Polar Eiol. 14,441-446. Myers, S . A., Horel, J. A., and Pennypacker, H. S. (1965). Operant control of vocal behavior in the monkey Cebus albifrons. Psychon. Sci. 3, 389-390. Norris, K. S., and Dohl, T. P. (1980). The structure and function of cetacean schools. In “Cetacean Behavior: Mechanisms and Functions” (L. M. Herman, ed.), pp. 21 1-261. John Wiley and Sons, New York. Nottebohm, F. (1972). The origins of vocal learning. Am. Nat. 106, 116-140. Nottebohm, F. (1975). Continental patterns of song variability in Zonotrichia capensis: Some possible ecological correlates. Am. Nut. 109, 605-624. Owren, M. J., Dieter, J. A., Seyfarth, R. M., and Cheney, D. L. (1992). “Food” calls produced by adult female rhesus (Macaca mulatta) and Japanese (M. fuscata) macaques, their normally-raised offspring, and offspring cross-fostered between species. Behaviour UO, 218-231.
96
VINCENT M. JANlK AND PETER J. B. SLATER
Parker, S., and Gibson, K. R. (1979). A developmental model for the evolution of language and intelligence in early hominids. Behav. Brain Sci. 2, 367-408. Payne, K., and Payne, R. (1985). Large scale changes over 19 years in songs of humpback whales in Bermuda. Z. Tierpsychol. 68,89-114. Payne, K., Tyack, P., and Payne, R. (1983). Progressive changes in the songs of humpback whales (Megapferanovaeangliae):A detailed analysis of two seasons in Hawaii. In “Communication and Behavior of Whales” (R. Payne, ed.), pp. 9-57. Westview Press, Boulder, Colorado. Payne, R. B. (1981). Song learning and social interaction in indigo buntings. Anim. Behav. 29,688-697. Payne, R. B., Payne, L. L., and Doehlert, S. M. (1988). Biological and cultural success of song memes in indigo buntings. Ecology 69, 104-1 17. Payne, R. S., and McVay, S. (1971). Songs of humpback whales. Science 173,585-597. Pepperberg, I. M. (1981). Functional vocalizations by an African grey parrot. Z. Tierpsychol. 55, 139-160. Pinker, S.. and Bloom, P. (1990). Natural language and natural selection. Behav. Brain Sci. 13,707-784. Ralls, K., Fiorelli, P., and Gish, S. (1985). Vocalizations and vocal mimicry in captive harbor seals, Phoca vifulina. Can. J. 2001.63, 1050-1056. Randall, J. A. (1995). Modification of footdrumming signatures by kangaroo rats: Changing territories and gaining new neighbours. Anim. Behav. 49, 1227-1237. Randolph, M. C., and Brooks, B. B. (1967). Conditioning of a vocal response in a chimpanzee through social reinforcement. Folia Primafol. 5,70-79. Rasmuson, T. M., and Barclay, M. R. (1992). Individual variation in the isolation calls of newborn big brown bats (Epfesicusfuscus):Is variation genetic? Can. J. Zooi. 70,698-702. Ray, C., Watkins, W. A., and Burns, J. J. (1969). The underwater song of Erignathus (bearded seal). Zoologica 54,7943. Reeves, R. R., Stewart, B. S., and Leatherwood, S. (1992). “The Sierra Club Handbook of Seals and Sirenians.” Sierra Club Books, San Francisco. Reiss, D., and McCowan, B. (1993). Spontaneous vocal mimicry and production by bottlenose dolphins (Tursiops rruncafus): Evidence for vocal learning. J. Comp. Psychol. 107, 30 1-3 12. Richards, D. G., Wolz, J. P., and Herman, L. M. (1984). Vocal mimicry of computer-generated sounds and vocal labeling of objects by a bottlenosed dolphin, Tursiops truncatus. J. Comp. Psychol. 98, 10-28. Romand, R., and Ehret, G. (1984). Development of sound production in normal, isolated, and deafened kittens during the first postnatal months. Dev. Psychobiol. 17, 629-649. Salzinger, K., and Waller, M. B. (1962). The operant control of vocalization in the dog. J . Exp. Anal. Behav. 5,383-389. Sayigh, L. S. (1992). Development and functions of signature whistles of free-ranging bottlenose dolphins, Tursiops truncafus.Unpublished doctoral dissertation, The Massachusetts Institute of Technology and Woods Hole Oceanographic Institution Joint Program, WHO1 92-37. Sayigh, L. S., Tyack, P. L., Wells, R. S., and Scott, M. D. (1990). Signature whistles of freeranging bottlenose dolphins, Tursiops truncafus: Mother-offspring comparisons. Behav. Ecol. Sociobiol. 26, 247-260. Sayigh, L. S., Tyack, P. L., Wells, R. S., Scott, M. D., and Irvine, A. B. (1995). Sex differences in signature whistle production of free-ranging bottlenose dolphins, Tursiops rruncarus. Behav. Ecol. Sociobiol. 36, 171-177. Schusterman, R. J., and Feinstein, S. H. (1965). Shapingand discriminative control of underwater click vocalizations in a California sea lion. Science 150, 1743-1744.
VOCAL LEARNING IN MAMMALS
97
Searcy, W. A. (1988). Dual intersexual and intrasexual functions of song in red-winged blackbirds. In “Acta XIX Congressus Internationalis Omithologici” (H. Ouellet, ed.), pp. 1373-1381. University of Ottawa Press, Ottawa. Shipley, C., Hines, M., and Buchwald, J. S. (1981). Individual differences in the threat calls of northern elephant seal bulls (Mirounga angustirostris).Anim. Behav. 29, 12-19. Sigurdson, J. (1993). Frequency-modulated whistles as a medium for communication with the bottlenose dolphin (Tursiops truncarus). In “Language and Communication: Comparative Perspectives” (H. L. Roitblat, L. M. Herman, and P. E. Nachtigall. eds.), pp. 153-173. Lawrence Erlbaum Associates, Hillsdale, New Jersey. Simila, T., and Urgate, F. (1993). Surface and underwater observations of cooperatively feeding killer whales in northern Norway. Can. J. Zool. 71, 1494-1499. Slater, P. J. B., and Ince, S. A. (1979). Cultural evolution in chaffinch song. Behaviour 71, 146-166. Slobodchikoff, C. N., and Coast, R. (1980). Dialects in the alarm calls of prairie dogs. Behav. Ecol. Sociobiol. 7, 49-53. Smith, T. G., Siniff, D. B., Reichle, R., and Stone, S. (1981). Coordinated behaviour of killer whales (Orcinus orca) hunting a crabeater seal (Lobodon carcinophagus). Can. J. Zool. 59, 1185-1189. Smolker, R. A., Mann,J., and Smuts, B. B. (1993). Use of signature whistles during separations and reunions by wild bottlenose dolphin mothers and infants. Behav. Ecol. Sociobiol. 33,393-402. Smolker, R. A,, Richards, A. F., Connor, R. C., and Pepper, J. W. (1992). Sex differences in patterns of association among Indian Ocean bottlenose dolphins. Eehaviour l23,38-69. Snowdon, C . T. (1990). Language capacities of nonhuman animals. Yearb. Phys. Anthropol. 33, 215-243. Somers, P. (1973). Dialects in southern Rocky Mountain pikas, Ochotona princeps (Lagomorpha). Anim. Behav. 21, 124-137. Spiesberger, J. L., and Fristrup, K. M. (1990). Passive localization of calling animals and sensing their acoustic environment using acoustic tomography. Am. Nut. 135, 107-153. Srikosamat, S. (1982). Imitation of vocal duet by a widow of the pileated gibbon (Hylobares pileatus) in Southeast Thailand. Int. J. Primatol. 3,336. Stirling, I. (1975). Factors affecting the evolution of social behaviour in the pinnipedia. Rapp. P.-v. Reun. Cons. Int. Explor. Mer 169,205-212. Stirling, I., Calvert, W., and Spencer, C. (1987). Evidence of stereotyped underwater vocalizations of male Atlantic walruses (Odobenus rosmarus rosrnarus). Can. J. Zool. 65,23112321. Stirling, I., and Siniff, D. B. (1979). Underwater vocalizations of leopard seals (Hydrurga leptonyx) and crabeater seals (Lobodon carcinophagus) near the South Shetland Islands, Antarctica. Can. J. Zool. 57, 1244-1248. Struhsaker, T. T. (1970). Phylogenetic implications of some vocalizations of Cercopithecus monkeys. In “Old World Monkeys: Evolution, Systematics, and Behavior” (J. R. Napier and P. H. Napier, eds.), pp. 365-444. Academic Press, New York. Sutton, D., Larson, C., Taylor, E. M., and Lindeman, R. C. (1973). Vocalization in rhesus monkeys: Conditionability. Brain Res. 52, 225-231. Talmage-Riggs, G., Winter, P., Ploog, D., and Mayer, W. (1972). Effect of deafening on the vocal behaviour of the squirrel monkey (Saimiri sciureus). Folia Primatol. 17,404-420. Tenaza, R. (1985). Songs of hybrid gibbons (Hylobates lar X H. muelleri). Am. J. Primatol. 8,249-253. Terhune, J. M. (1994). Geographical variation of harp seal underwater vocalizations. Can. J. ZOO^. 72, 892-897.
98
VINCENT M. JANlK AND PETER J. B. SLATER
Thomas, J. A,, and Golladay, C. L. (1995). Geographic variation in leopard seal (Hydrurga lepronyx) underwater vocalizations. In “Sensory Systems of Aquatic Mammals” (R. A. Kastelein, J. A. Thomas, and P. E. Nachtigall, eds.), pp. 201-221. De Spil Publishers, Woerden. Thomas, J. A., and Kuechle, V. B. (1982). Quantitative analysis of Weddell seal (Lepronychores weddelfi)underwater vocalizations at McMurdo Sound, Antarctica. J. Acoust. Soc. Am. 72, 1730-1738. Thomas, J. A., and Stirling, I. (1983). Geographic variation in the underwater vocalizations of Weddell seals (Lepronychotes weddelli) from Palmer Peninsula and McMurdo Sound, Antarctica. Can. J. Zool. 61, 2203-2212. Todt, D. (1975). Social learning of vocal patterns and modes of their application in grey parrots (Psiftacus erithacus). Z. Tierpsychol. 39, 178- 188. Tschanz, 9. (1968). Trottellummen. 2. Tierpsychol. Suppl. 4. 1-100. Tyack, P. (1986a). Population biology, social behavior, and communication in whales and dolphins. Trends Ecol. Evol. 1, 144-150. Tyack, P. (1986b).Whistle repertoires of two bottlenosed dolphins, Tursiops truncafus:mimicry of signature whistles? Behav. Ecol. Sociobiol. 18, 251 -257. Tyack, P. (1991). If you need me. whistle. Nar. Hisr. 8/91, 60-61. Tyack. P., and Whitehead, H. (1983). Male competition in large groups of wintering humpback whales. Behaviour 83, 132-154. Van Heel, W. H. D., Kamminga, C., and van der Toorn, J. D. (1982). An experiment in twoway communication in Orcinus orca L. Aqitar. Mamm. 9, 69-82. Van Parijs. S., Thompson, P. M.. Tollit, D. J., and MacKay, A. (in press). Changes in the distribution and activity of male harbour seals during the mating season. Anim. Behav. Wang, D., WUrsig, B., and Evans, W. E. (1995). Whistles of bottlenose dolphins: comparisons among populations. Aquar. Mamm. 21, 65-77. Waser, P. M., and Brown, C. H. (1986). Habitat acoustics and primate communication. Am. J. Primatol. 10, 135-156. Watkins, W. A., and Ray, G. C. (1985). In-air and underwater sounds of the Ross seal, Ommatophoca rossi. J. Acoust. SOC. Am. 77, 1598-1600. Watkins, W. A., and Schevill, W. E. (1968). Underwater playback of their own sounds to Lepronychotes (Weddell seals). J . Mamm. 49, 287-296. Watkins, W. A., and Schevill, W. E. (1977). Sperm whale codas. J. Acousr. SOC.A m . 62,14851490. Watkins, W. A., Tyack, P., Moore, K. E., and Bird, J. E. (1987). The 20-Hz signals of finback whales (Balaenoptera physalus). J. Acousr. SOC.Am. 82, 1901-1912. Wells, R. S., Scott, M. D., and Irvine, A. B. (1987). The social structure of free-ranging bottlenose dolphins. In “Current Mammalogy, Vol. 1” (H. H. Genoways, ed.), pp. 247305. Plenum Press, New York. Wiley, R. H., and Richards, D. G. (1978). Physical constraints on acoustic communication in the atmosphere: implications for the evolution of animal vocalizations. Behav. Ecol. Sociobiol. 3, 69-94. Wilson, W. A. (1975). Discriminative conditioning of vocalizations in Lemur carra. Anirn. Behav. 23,432-436. Winn, H. E., Thompson, T. J., Cummings, W. C., Hain, J., Hudnall, J., Hays, H., and Steiner, W. W. (1981). Song of the humpback whale-Population comparisons. Behav. Ecol. Sociobiol. 8, 41-46. Winter, P. (1969). Dialects in squirrel monkeys: Vocalization of the Roman arch type. Folia Primatol. 10. 216-229.
VOCAL LEARNING IN MAMMALS
99
Winter, P., Handley, P., Ploog, D., and Schott, D. (1973). Ontogeny of squirrel monkey calls under normal conditions and under acoustic isolation. Behaviour 47, 230-239. WUrsig, B., and Clark, C. (1993). Behavior. In “The Bowhead Whale” (J. J. Burns, J. J. Montague, and C. J. Cowles, eds), pp. 157-199. Society for Marine Mamrnalogy, Lawrence, Kansas.
This Page Intentionally Left Blank
ADVANCES IN THE STUDY OF BEHAVIOR. VOL. 26
Behavioral Ecology and Conservation Biology of Primates and Other Animals KARENB. STRIER DEPARTMENT OF ANTHROPOLOGY
WISCONSIN-MADISON MADISON, WISCONSIN
UNIVERSITY OF
I. INTRODUCTION Continuing pressures on the world’s tropical ecosystems have significantly affected the status of primates and their habitats. Of some 300 species recognized today, more than half are classified as threatened or endangered based on the small sizes and fragmented distributions of their populations (Mace and Lande, 1991; IUCN, 1994; Pope, 1996). Evaluating the viability of remaining populations is the first step in establishing conservation tools in conservation priorities, and population viability analyses, or PVAs, have become critical tools in conservation biology (Boyce, 1992; Lacy, 1993; Caughley, 1994; Akgakaya and Burgman, 1995; Walsh, 1995). Conservation efforts on behalf of endangered primates ultimately depend on behavioral and ecological data to develop informed management plans and policies, and behavioral ecologists are eager to contribute their knowledge. Yet, even the most comprehensive long-term field studies on primates rarely supply the demographic data required to assess what constitutes a minimum viable population. Part of the difficulty of translating insights from behavioral ecology to conservation biology can be attributed to the different levels of analyses employed by each discipline. Research in primate behavioral ecology has tended to focus on individuals and groups rather than populations, whereas PVAs are generally insensitive t o the ways in which the behavior of individual primates interacts with ecological processes to affect their populations. Without deliberate efforts to reconcile these complementary approaches, the viability of primate populations will be defined by arbitrary criteria (Boyce, 1992), and confusion about the relative importance of genetic, demographic, or ecological variables that affect the populations of primates and other animals will persist (Caughley, 1994; Harcourt, 1995). 101
Copyright 0 1997 by Academic Press All rights of reproductionin any form reserved. 0065-3454/97 $25.00
102
KAREN B. STRIER
The importance of considering behavioral and ecological data in conservation programs has been demonstrated in a variety of animals. For example, Walters (1991) found that incorporating knowledge of the social structure of endangered red-cockaded woodpeckers (Picoides borealis) in the southeastern United States led to different conservation management strategies than those based solely on population models. In these birds, the number of breeding groups is determined by the availability of cavity tree clusters, whereas group size is determined by the number of juveniles that delay dispersal and act as helpers. Management efforts to improve survival and reproduction would increase group sizes, but have little effect on stimulating population recovery unless the number of suitable cavity tree clusters available to potential dispersers to establish breeding groups also increased. Car0 and Durant (1995) provide similar evidence for the ways in which behavioral ecology can contribute to conservation strategies for East African carnivores. They note, for example, that reproductive suppression in wild dogs and dwarf mongooses reduces the number of breeding individuals to the equivalent of the number of groups, and therefore makes these species more vulnerable to demographic fluctuations and loss of genetic diversity than other carnivores in which a larger proportion of adults breed. Similarly, knowledge of differences in grouping and ranging patterns is important to understanding the effects of disease and habitat fragmentation on carnivore population processes. Even concerns about the deleterious effects of reduced genetic variability on cheetah survival have been challenged by the recent field observations that predation by lions accounts for the high mortality of cheetah cubs (only 5% reach independence; Laurenson, 1994; Car0 and Laurenson, 1994; see also Lindburg, Durrant, Millard, and Oosterhuis, 1993). The integration of behavioral ecology and conservation biology has been slower for primates than for other animals, in part because traditional anthropocentric perceptions have treated primates as more like humans than like other animals (Richard, 1981). These perceptions, together with the logistical difficulties of accumulating data on population ecology and demography in long-lived, socially complex animals with slow rates of growth, have restricted the questions and methods in primate behavioral ecology to “bottom-up” approaches, which combine correlations between individual and group characteristics with ecological variables to construct interspecific comparative models. The intermediate, population-level data required for realistic models in conservation biology have been largely neglected. Like many other animals (e.g., fishes, amphibians, birds, and other mammals), the ability of primates to modify their behavior, including their mating and reproductive strategies, in response to environmental stimuli
PRIMATE BEHAVIORAL ECOLOGY A N D CONSERVATION
103
can profoundly alter the genetic and demographic models on which viability analyses of their populations are based. The focus on population processes in conservation biology has led to corresponding gaps in understanding how individual primates interact with their physical and social environments to affect the genetic and demographic structures of their populations. Incorporating insights about these interactions from primate behavioral ecology into population models would contribute to the reliability of the models that are used to determine conservation strategies for endangered primates. More than a decade ago, May and Seger (1986, p. 267) identified “a need for behavioral ecologists to work up toward the population consequences of behavior, and for population biologists to work down toward the individual behavior that determines population parameters.” Disciplinary boundaries between individual approaches in behavioral ecology and population approaches in conservation biology may be more rigidly defined for primates than for most other animals. Recent methodological advances in noninvasive DNA and steroid assay techniques have dual applications in both primate behavioral ecology and conservation biology because they provide ways of evaluating the genetic consequences and ecological and life history corollaries of primate social and reproductive behavior. These new techniques have already begun to play pivotal roles in uniting primate behavioral ecology and conservation, and are considered briefly in the final section of this review. First, however, it is appropriate to examine some of the principal distinctions underlying the complementary paradigms that have been employed in primate behavioral ecology and conservation biology, and the ways in which divergent assumptions about the importance of behavioral, genetic, ecological, and life history variables have influenced these fields. Suggestions for areas where greater interdisciplinary exchange would be most productive are provided by considering some persisting questions about the behavioral ecology and conservation of muriqui monkeys (Bruchyteles uruchnoides),one of the many endangered primates that would directly benefit from greater convergence between theory and practice (Strier, 1992a). 11. COMPARATIVE PARADIGMS
Models in behavioral ecology and Conservation biology share the common goal of identifying general principles that can explain and predict individual behavior and population processes, respectively (Clutton-Brock and Harvey, 1984; Krebs and Davies, 1991; Caughley, 1994). They differ, however, in the degree to which they emphasize precision and realism (Levins, 1966). The thoretical orientations of behavioral ecology have led
104
KAREN B. STRIER
to a greater concern with precision, whereas the practical applications of conservation biology have led to an emphasis on realism. As a result, models in both disciplines have been limited in their ability to infer causation from correlations between behavioral and ecological variables (in the case of primate behavioral ecology) and between demographic and genetic variables (in the case of primate conservation biology). Merging individual and population approaches can broaden theoretical perspectives in conservation biology (Caughley, 1994) and make behavioral ecology more applicable to conservation (Caro and Durant, 1995; Pope, 1996). However, the development of predictive models in both disciplines has also been limited by their reliance on species-level comparisons. Increasing evidence of marked behavioral and genetic variation among populations of the same or closely related species suggests that contemporary classifications of species may not be as discrete or as objective as their treatment in models of both primate behavioral ecology and conservation biology has assumed them to be. Understanding the ways in which assumptions about species, populations, and individuals have defined primate behavioral ecology and conservation biology is essential to understanding the interacting effects of phylogeny, ecology, and demography on individual behavior and the dynamic processes that affect primate population viability (Fig. 1). A. SPECIES-LEVEL ANALYSES 1. Historical Perspectives
a. Anthropocentric Influences in Primate Behavioral Ecology. The assumption that the species is an appropriate unit for comparison has been implicit in the anthropocentric tradition underlying comparative primate studies (Richard, 1981; Fedigan, 1982; Loy and Peters, 1991). Comparisons with humans provided the original justifications for many primate studies, influencing which species were studied and which results were emphasized (Richard, 1981;Fedigan, 1982; Strier, 1994a). Early efforts to identify particular species as referential models for human social evolution shifted to strategic models based on general evolutionary and ecological principles (Tooby and DeVore, 1987). Nevertheless, whether the focus has been on continuities and continua, or discontinuities and dichotomies, primates continue to be used as the comparative out-group for biologically oriented analyses of human behavioral evolution. The known diversity of human social and cultural practices established a range of species-specific variation against which other primates (and other animals) have been compared. The degree of intraspecific behavioral variation in nonhuman primates could be expected to fall well within the human range because no nonhuman primate matches anatomically modern
PRIMATE BEHAVIORAL ECOLOGY AND CONSERVATION
105
Individuals SYSTEMATIC
+
METHODS
EVOLUTIONARY THEORY
Groups Precision
Generality
”
Typical primate”
LONG-TERM DATA
TAXONOMIC DIVERSITY
COMPARATIVE MODELS
LREALISTIC
4
Populations MULTIPLE LEVELS OF EXPLANATION
POPULATION THEORIES OF CONSERVATION BIOLOGY
FIG.1. Comparative paradigms in primate behavioral ecology. Individuals and groups have been the traditional levels of analyses for precise systematic behavioral sampling and testing general evolutionary principles. Long-term data on diverse species have challenged assumptions about “typical” primates (Strier, 1994a). Population-level analyses from new methods of examining genetics and reproduction and conservation biology contribute to the development of more realistic comparative models of primate behavior.
Homo supiens in its geographic distribution or technological adaptations. However, it is also the case that no nonhuman primate species matches modern humans in their mobility, and therefore in their ability to maintain genetic and cultural flow between populations. The consequences of this fundamental difference have only recently received critical attention in genetic and behavioral comparisons between primate populations (e.g., Morin, 1993; Morin, Moore, Chakraborty, Jin, Goodall, and Woodruff, 1994). The species level of analysis in comparative models of primate behavioral diversity has been intransigent ( Jolly, 1993). Species-level analyses have withstood direct evidence of high degrees of intraspecific variation, such as that between infanticidal and noninfanticidal langur populations, because behavioral differences also correlated with ecological and demographic differences (Hrdy, 1977; Bartlett, Sussman, and Cheverud, 1993). As accu-
106
KAREN B. STRIER
mulating data demonstrated “differences between species allocated to the same ecological categories were in many cases as pronounced as differences between categories” (Clutton-Brock and Harvey, 1984, p. 14), primatologists refined their ecological variables and fine-tuned their ecological models rather than question whether the species was the appropriate unit for comparison. Despite methodological attention to the population variation advocated by evolutionary biology (Mayr, 1966), primatology has been steadfast in its typological approach to species comparisons. Species-level comparisons have also persisted despite concern over the lack of independence between closely related species due to phylogenetically conservative traits, or homologies (Clutton-Brock and Harvey, 1984; Harvey and Purvis, 1991; Garland and Adolph, 1994; Rendall and Di Fiore, 1996). Comparative interspecific models of primate behavior have been constrained by uneven taxonomic representations (Strier, 1990,1994a),and even models depicting general principles about the behavioral corollaries of group size and composition (e.g., Wrangham, 1980; van Schaik, 1989) are confounded by the absence of independent samples of these variables from discrete populations. Expanding comparative perspectives to include intraspecific variation is as critical to the development of more comprehensive models of primate behavioral ecology as it is to the interpretation of population models from conservation biology. Indeed, the utility of specieslevel comparisons in primate behavioral ecology is limited without an underlying understanding of the relationships between intra- and interspecific variation. b. Sign8cance of Species Classifications in Conservation Biology. The species concept in conservation biology dates back to the historical influence of models of island biogeography, which focused on species diversity and area relationships (Simberloff, 1988). The subsequent inclusion of genetics in conservation biology formally established the link between populations and species, which Mayr’s (1942) Biological Species Concept defined as “groups of actually or potentially interbreeding populations that are reproductively isolated from other such groups.” The Biological Species Concept emphasizes the primacy of reproductive isolation in providing “an effective protective device against genetic disintegration of the species genotype” (O’Brien and Mayr, 1991a, p. 1188) as well as “the existence of appreciable genetic diversity within species that is often partitioned geographically (or temporally) by population subdivision into subspecies” (O’Brien and Mayr, 1991b, p. 1187). Nonetheless, disagreement over the relative importance of reproductive isolation between species and genetic diversity within species has led to controversies over taxonomic classifications (Kimbel and Martin, 1993) and over how
PRIMATE BEHAVIORAL ECOLOGY A N D CONSERVATION
107
conservation policies to protect endangered species should be applied (O’Brien and Mayr, 1991b). Both controversies have broad implications for primate behavioral ecology as well as conservation biology. For example, persistent examples of hybrids produced by interbreeding between olive baboons (Papio anubis) and both yellow baboons (P. cynocephafus) and hamadryas baboons (P. hamadryas) have led to the suggestion that they be reclassified into reproductively compatible subspecies ( Jolly, 1993). Yet, in their social behavior and organization, hamadryas baboon societies differ strikingly from those shared by olive and yellow baboons (Phillips-Conroy, Jolly, Nystrom, and Memmalin, 1992). Compressing such diversity into a single taxonomic species would exaggerate existing difficulties in comparative models of behavior (Gittleman and Hang-Kwang, 1992; see Section II,A,I,a). Primate conservationists generally split rather than lump potential species or subspecies based on known or perceived morphological and genetic diversity. Part of this tendency may be related to the practical motives of attracting attention and funds to address conservation concerns. Arguments about the importance of preserving relic populations representing two possible species may be more effective than those advocating the protection of a single species whose distribution has been severely fragmented into multiple populations. As Rylands, Mittermeier, and Luna, (1995, p. 114) explain in their recent revision of the conservation status of New World primates, “our aim is to provide an estimate of the diversity. . . and while there is still discussion as to the validity or otherwise of numerous forms, we prefer to maintain them.” Although such splitting may be justified for practical motives, it reflects widespread confusion about the relative importance of reproductive isolation and genetic variation. The problem is compounded when patterns of variation in morphological and behavioral traits are included with patterns of genetic variation. Thus, in less than a decade, the taxonomic classification of muriquis has shifted from “a monotypic genus, Brachyteles arachnoides E. Geoffroy, 1806” (Aguirre, 1971), to the morphologically based distinction of southern and northern subspecies, B. a. arachnoides and B. a. hypoxanthus, respectively (Lemos de SA and Glander, 1993), to the most recent separation into northern and southern species, B. arachnoides and B. hypoxanthus (Rylands et al., 1995). Analyses of the morphological data supporting two muriqui subspecies have been challenged (Leigh and Jungers, 1994), and it is difficult to evaluate whether the genetic differences between one northern and one southern population (Lemos de S i , Pope, Glander, Struhsaker, and da Fonseca, 1990) are greater than those among isolated populations within either of the two regions, particularly when the unusually high level of genetic variation within populations is considered
108
KAREN 8 . STRIER
(Pope, 1996). Although present day distributions of the genus would preclude genetic exchange in the wild, southern males and northern females have successfully interbred in captivity (Coimbra-Filho, Pissinatti, and Rylands, 1993). Similar confusion exists over the taxonomic classifications of great apes, which are generally divided into four species: gorillas (Gorilla gorilla) with three subspecies, orangutans (Pongo pygmaeus) with two subspecies, common chimpanzees (Pan troglodytes) with three subspecies, and bonobos (Pan paniscus). Recent studies have found the genetic differences between the two subspecies of orangutans to be greater than those among gorilla subspecies, and genetic differences between western lowland gorillas and the two eastern gorilla subspecies to be greater than those between the two species of chimpanzees (reviewed in Uchida, 1996). As with muriquis, separating subspecies that are known to be capable of interbreeding in captivity into distinct species violates the reproductive isolation component of the Biological Species Concept. However, as Uchida (1996, p. 164) suggests, “only a few genetic variations may be necessary to achieve reproductive isolation in nature.” Comparative studies on the behavioral ecology of different populations of apes, muriquis, and other primates are clearly needed to reconcile genetic, morphological, and behavioral variation in models of behavioral ecology, and to translate extant variation into informed conservation programs.
2. Applications of Species Comparisons in Primate Behavioral Ecology a. Phylogenetic Analyses of Behavior. Recent applications of phylogenetic approaches to primate behavior provide unique opportunities to distinguish evolutionarily conservative traits from those that are more labile, and thus more responsive to ecological and demographic conditions (Di Fiore and Rendall, 1994; Rendall and Di Fiore. 1996). Nonetheless, even the most cautious applications of phylogenetic analyses, like other comparative models in primate behavioral ecology, are only as accurate as the data they utilize. Many primates are still known only from studies on single study groups or populations, and it is impossible to evaluate how representative of their species they may be. Indeed, long-term data on single study groups frequently demonstrate fluctuations in variables such as group size, sex ratios, and rates of aggression. Yet, it has been common practice to compress long-term data from single study groups into single average values for comparative analyses. Even when multiple populations of the same species have been studied, intraspecific variation has been compressed into single, species-specific averages. The resulting average species values often have little or unclear biological meaning (Rowell, 1979; DeRousseau, 1990), and
PRIMATE BEHAVIORAL ECOLOGY A N D CONSERVATION
109
can introduce unknown error into comparative analyses (Clutton-Brock and Harvey, 1984). For most primates, intraspecific variation is too extreme to justify using any one population to represent the species, and the determinants of this variation are still too poorly understood to justify averaging across populations. For example, muriqui grouping patterns described from four study sites range from cohesive, multimale, multifemale groups (Lemos de S6, 1991), to fluid, mixed-sex groups (Strier, Mendes, Rimoli, and Rimoli, 1993), to fluid, mixed- and single-sex groups (Carvalho, Strier, and Ferrari, 1995), to single-sex groups that associate with one another exclusively during periods of sexual activity (Milton, 1984). These differences in muriqui grouping patterns do not correspond to differences in population densities, which range from an estimated 0.01 individuals/ha in the largest (> 37,000 ha), most pristine forest to 0.30 individualslha in the smallest (40 ha), most disturbed forest (Stuart, Strier, and Pierberg, 1993). It is still unclear how the presence o r absence of natural predators and food competitors, differences in available food resources provided by primary versus secondary and disturbed forest growth, and the reduction of available habitat and increasing fragmentation contribute to differences in either grouping patterns or population densities (Stallings and Robinson, 1991; Pinto, Costa, Strier, and Fonseca, 1993). Within-population averaging of annual fluctuations in group size and sex ratios presents similar problems. During a 14-year study of muriquis in an 800-ha forest at the EstaqSio Biologica de Caratinga, in Minas Gerais, Brazil, the number of distinct groups has increased from two to three, and the size of one study group, which has been systematically monitored, has increased from 22 to 56 individuals (Strier, in press, a). The increase in group size may be related to the fact that female muriqui age at first reproduction has been less than the average muriqui interbirth interval of 3 years (Strier, 1991), which is consistent with Dobson and Lyles’s (1989) survey of expanding primate populations. Low female mortality rates and female-biased infant sex ratios have also contributed to an increasingly skewed adult muriqui sex ratio, which may or may not be responsible for other behavioral changes observed over the years (Strier, 1991,1994b; Strier et aL, 1993). Comparing interspecific or phylogenetic differences in social behavior or demography without first understanding how ecological variables, such as habitat size, predator and interspecific competitor communities, and the original population size and composition, affect intraspecific variation is particularly problematic for endangered primates, whose behavior and demography are likely to reflect responses to unique environmental alterations as well as evolutionary adaptations. Because so many primates have been subjected to historical and more recent habitat alterations, species averaging
110
KAREN B. STRIER
for comparative phylogenetic analyses of primate behavior may be more vulnerable to error than for other animals. b. Behavioral Variation in Phylogenetic Analyses. Other recognized sources of error in phylogenetic analyses include the difficulties of dividing continuous behavioral variables into discrete categories (Rendall and Di Fiore, 1996), but the broader implications of such practices are rarely considered. For example, two of the behavioral variables that Di Fiore and Rendall’s analysis (1994) identified as phylogenetically conservative traits among primates, female dispersal and dominance hierarchies, cannot be reduced to discrete categories without prior assumptions about what proportion of females must transfer between groups and how overt a dominance hierarchy must be to have biological meaning. Assumptions about these variables differ between disciplines. Thus, in conservation biology, one migration event per generation is considered to be sufficient to maintain gene flow between populations (Ralls, Harvey, and Lyles, 1986). Yet, such minimal female dispersal would rarely justify a species being characterized by female-biased dispersal in behavioral ecology (see Moore, 1984). Similarly, dominance hierarchies may be visibly maintained by distinct and consistent behavioral interactions between dominant and subordinate individuals, and may correspond to measurable differences in access to resources, as appears to be the case in many cercopithecine primate females (Fedigan, 1983). However, dominance relationships that are imposed through subtle behavioral or pheromonal regulation of reproductive physiology during the conception season may have even more dramatic consequences for female reproductive success than overt hierarchies, as is the case among many callitrichids (Abbott, 1989; Garber, 1994) and possibly other New World primates (Strier, 1996a). Assumptions about the biological significance of such variation in the expression of dominance hierarchies will influence how species are classified in phylogenetic models. Basic social traits, such as whether groups of common marmosets (Callithrix jacchus) contain a single or multiple breeding females, appear to be strongly influenced by population densities (Digby and Ferrari, 1994),which may vary between study sites as well as within a population over time. Local population densities can fluctuate dramatically because of reproductive failure or high rates of migration in response to rare events, such as droughts that affect food availability (Sussman, 1992;Savage, Giraldo, Soto, and Snowdon, 1996). The duration of studies will affect whether rare events are documented and how their effects are interpreted (Weatherhead, 1986), and therefore influence how social traits, such as whether mating systems are polyandrous, polygynous, or polygamous, are ultimately characterized. Without long-term observations, social and demographic instability could
PRIMATE BEHAVIORAL ECOLOGY AND CONSERVATION
111
not be correlated or causally associated with ecological events, and the values considered in phylogenetic analyses cannot be evaluated. c. The Role of Demography in Behavioral Phylogenies. Classifications of species-specific behavior will have limited utility in behavioral ecology or conservation biology whenever behavioral variables shift in response t o population variations. Comparisons among the New World atelins, which include muriquis, woolly monkeys (Lagothrix), and spider monkeys (Ateles), illustrate some of the ways in which demographic variables can affect both interspecific and intraspecific behavioral comparisons (Strier, 1994b). All three genera have been described by female-biased dispersal and male philopatry, but they differ from one another in the types of relationships and strength of associations among males. Hierarchical relationships occur when males can individually dominate females, as in the cases of woolly monkeys, which are sexually dimorphic in both body and canine size, and spider monkeys, which are sexually dimorphic in canine size. Egalitarian relationships occur when males are codominant with females, as in the case of sexually monomorphic muriquis. The strength of male affiliative associations appears to covary with group socionomic sex ratios (ratio of breeding females to males) and whether female grouping patterns are cohesive or fluid. Thus, low socionomic sex ratios ranging from 1.1 to 1.6 correspond to cohesive groups and weak affiliative associations among males in eight woolly monkey groups sampled, whereas high socionomic sex ratios ranging from 1.5 to 4.2 corresponded to fluid groups and strong affiliative associations among males in five spider monkey groups representing three species (Nishimura, 1990,1994). Comparable data on sex ratios, grouping patterns, and male affiliative associations come from a single muriqui group that was monitored over a 10-year period (Strier et al., 1993). During this period, female grouping patterns became increasingly fluid as both the size and socionomic sex ratio (1.1 to 1.9) of the group increased. Although male affiliative associations were uniformly strong, the success of males that cooperated in preventing incursions from extragroup males declined (Strier, 1994b). Differences in the types of male relationships appear to be related to the degree of sexual dimorphism in these taxa, and are not likely to be affected by current group or population demographic conditions. However, differences in the strength of male affiliative associations are clearly associated with the socioeconomic sex ratios in atelin groups, which may reflect random demographic processes or behavioral adjustments through migration. Long-term monitoring of population demography is necessary to distinguish between these mechanisms as well as to evaluate whether the pronounced shifts in sex ratios and female grouping patterns detected in the muriqui group described here occur in woolly monkeys, spider monkeys,
112
KAREN B. STRIER
or other muriqui populations, and whether the behavior of males responds to changes in sex ratios within groups and populations. In one of the first demographic feedback models of primate behavior, Altmann and Altmann (1979, p. 60) considered the possibility of “a degree of species-typical stereotype that is much narrower than one would expect from their experimentally demonstrable capacity for learning and for other environmentally induced variability.” Similar assumptions about speciestypical traits are made in phylogenetic analyses of primate behavior. However, as Dunbar’s (1979, p. 84) more populationist perspective suggests, “we need to know not only why a species has the mean value it does, but also why individual populations or groups differ from the species’ mean in the way they do.” 3. Applications of Species-SpeciJic Life Histories in Conservation Biology
In conservation biology, populations are assumed to differ from one another based on size, composition, and the degree and type of ecological disturbance that affects demographic variables. Individual life history tactics affect population natality, mortality, and migration rates, which are measured by sampling variation among individuals in a population. Random variations in whether individuals survive and how many offspring they produce are modeled in conservation biology as demographic stochasticity, and the effects of environmental variation on population parameters are modeled as environmental stochasticity (Simberloff, 1988). However, as DeRousseau (1990, p. 6) notes, “most comparative population studies gloss over the individual as an entity that changes through its lifetime and in different environments, in favor of statistical treatments that describe the population as a whole. This is particularly the case when species are being compared to answer evolutionary questions.” Both individual life history tactics and the effects of environmental variation on populations are strongly influenced by a species’ biology, and conservation biologists recognize that “recovery plans for individual species must be tailored to their behaviour” (Reed and Dobson, 1993, p. 255). In this respect, conservation biology makes similar assumptions about speciestypical behavior to those of phylogenetic and other comparative analyses in primate behavioral ecology. Larger animals with longer generation times are likely to be less vulnerable to the effects of environmental variation on rates of population growth, whereas small animals, which are capable of more rapid recovery because of intrinsically higher reproductive rates, are less vulnerable to the effects of environmental variation on population size (Belovsky, 1987). Yet, primates have slower rates of growth compared to other mammals of similar body
PRIMATE BEHAVIORAL ECOLOGY AND CONSERVATION
113
size (Charnov and Berrigan, 1993). Delayed female age at first reproduction and extended interbirth intervals also contribute to the comparatively lower rates of potential population expansion in primates (Ross, 1988,1991), and therefore minimum viable population sizes for primates may differ from those estimated for other animals. Because life history variables such as interbirth intervals affect reproductive rates, it is important to understand how they are shaped by environmental conditions as well as evolutionary history (Partridge and Harvey, 1988). For example, a recent comparison among sympatric capuchin monkeys (Cebus capucinus), spider monkeys (Atefes geoffroyi), and howler monkeys (Afouattapaffiatu)monitored during a 10-year study at Santa Rosa National Park, Costa Rica demonstrated significant differences in median interbirth intervals corresponding to relative brain, but not body, weights (Fedigan and Rose, 1995). Gestation length differences, which accounted for roughly 20% of the capuchin monkeys’ 26.4-month interbirth interval, 22% of the spider monkeys’ 34.7-month interbirth interval, and 31% of the howler monkeys’ 19.9-month interbirth interval, were only partially responsible for interspecific differences in interbirth intervals. In addition, the length of postpartum amenorrhea and the number of cycles before conception were associated with environmental, nutritional, and social factors that affected the different species in different ways. In addition to species differences, environmental and social factors can have different effects on reproductive parameters among individuals. In many primates, dominant females with priority of access to high-quality foods begin reproducing at earlier age, produce healthier offspring, have shorter interbirth intervals, and live longer to reproduce (Dittus, 1979; Mori, 1979; Small, 1981; Whitten, 1984). However, high-ranking baboon females (Papio cyncocephafus anubis) at Gombe National Park, Tanzania, also suffer from higher incidences of miscarriage, and some experience reduced fertility (Packer, Collins, Sindlimwo, and Goodall, 1995). Stress associated with attaining or maintaining high rank may offset some of the nutritional advantages associated with dominance. Among wild chimpanzees at Gombe National Park, Tanzania, variation in gestation length from 203 to 234 days has been attributed to seasonal effects of diet on female nutrition and reproductive physiology (Wallis, 1995). Similarly, muriqui females in captivity resumed postpartum cycling and conceived more than a year earlier than wild females (Coimbra-Filho et af., 1993; Strier, in press, a), presumably because captive females expended less energy and were provided with a stable and highly nutritious diet year-round. However, although year-to-year variation in the timing of births corresponded to fluctuations in rainfall patterns and the availability of food resources, at least some of the annual variation in the timing of
114
KAREN B. STRIER
reproductive events among wild female muriquis was related to individual social histories, which appeared to stimulate reproductive synchrony among close associates. Individual responses to ecological and social conditions that affect the timing of reproduction also influence the size, composition, and density of populations. Thus, mountain gorillas (Gorilla gorilla) form single- or multimale breeding groups depending on the number of males in the population (Robbins, 1995), and small groups of Barbary macaques occupying marginal habitats have higher rates of increase than those in prime habitat because of lower intergroup competition (Mehlman, 1989). Indeed, comparisons among marmosets (Callithrix) revealed a positive relationship between species with high population densities and those with multiple breeding females in their groups (Digby and Ferrari, 1994). In stable marmoset groups under high population densities, the alternatives of postponing reproduction to inherit a breeding position or dispersing to form a new breeding group are precluded. Fluctuations in environmental conditions that affect population size and stability may cause female primates to shift their reproductive strategies. Thus, “what happens in the individual’s life is crucial to the full understanding of the evolutionary picture, and therefore, individuals and populations together are the fundamental units of study” (DeRousseau, 1990, p. 6). B. POPULATION-LEVEL ANALYSES Individual, group, and species levels of analysis have been invoked to various degrees and in various combinations in an increasingly sophisticated array of demographic (e.g., Collias and Southwick, 1952; Altmann & Altmann, 1979; Dunbar, 1979, 1987; Newton, 1988; Stephen, 1989; Sussman, 1992; Strier et al., 1993), ecological (e.g., Wrangham, 1980; Popp, 1983; van Schaik and van Hooff, 1983; Robinson and Janson, 1987; Symington, 1990; O’Brien, 1991;Fedigan, 1993), and phylogenetic models of primate behavior (Struhsaker, 1969; Sussman, 1977; Wrangham, 1987; Rosenberger and Strier, 1989; Di Fiore and Rendall, 1994; Rendall and Di Fiore, 1996; see Section 11,A,2). Long-term studies have provided insights into the life histories of individuals and groups as ecological and demographic conditions change, and an increasing number of studies on a wide diversity of species have contributed comparative perspectives on the range of primate behavioral diversity (Strier, 1994a). Comparative data exist for many species that have been studied at multiple sites, but these data usually pertain to single or adjacent study groups rather than to entire populations. Consequently, the application of these data to PVAs requires extrapolating from individual and group life histories
PRIMATE BEHAVIORAL ECOLOGY AND CONSERVATION
115
to populations and species, with little consideration to the potential significance of intraspecific variation. Instead, primate behavioral ecologists have accepted intraspecific variation either as evidence of “cultural traditions,” an anticipated product of smart, adaptable individuals whose long socialization is known to involve a large component of learning (Peters, 1991), or as evidence of the effects of altered habitats or unnatural ecological and demographic conditions on behavior (Asquith, 1989; Yeager, 1995). However, the interacting effects of ecological and demographic variation on behavior have not yet been established, in part because the populationlevel data necessary to calibrate behavioral differences among groups versus those across populations have been lacking. 1. Ecological Variation in Primate Behavioral Ecology
By the 1980s, comparative data on a broad taxonomic composite of primates with divergent ecological adaptations had begun t o accumulate. The range of interspecific variation in primate behavioral ecology overshadowed both reported intraspecific variation and the incorporation of life history data from long-term studies of a few well-known species. Comparative models became increasingly oriented toward explaining behavioral diversity in ecological terms, and greater significance was typically attributed to the more naturalistic studies whenever data from wild, unprovisioned primates challenged data on the same species from captive or provisioned populations. Paradoxically, it has been difficult to identify characteristics of “natural” populations, even among some of the best-studied species, such as baboons (Papio), macaques (Macaca), and chimpanzees (Pan), which have been monitored intensively at multiple sites, because these and most wild primates studied over the last century survive either in or adjacent to habitats that have been directly or indirectly altered by human activities. Some recent human conservation practices, including bans on exporting primates such as rhesus macaques (Macaca mulatta) for biomedical research, have helped declining populations to begin to recover (Southwick and Siddiqi, 1988), but in most cases humans were responsible for the original population declines. Habitat alterations affect primate behavior and primate populations, and thus figure prominently in both behavioral ecology and conservation biology. Distinguishing between the effects of historical events and more recent disturbances on primate behavior and populations requires comparative data from multiple sites. a. Interpreting Historical Events. Some primates have been especially successful at adapting to historical or evolutionary alterations in their habitats. For example, Samuels and Altmann (1991) found that the population
116
KAREN B. STRIER
of yellow baboons (Papio cynocephalus) in the Amboseli Basin, Kenya, remained stable from 1963 to 1987 despite loss of habitat due to human encroachment that affected other sympatric fauna. The baboons shifted their home ranges, and exploited new areas and new food resources, including human refuse dumps. Females that fed at dumpsites had higher feeding efficiency than females that fed only on wild foods (Muruthi, Altmann, and Altmann, 1991). Faster maturation and reproductive rates also characterized the females that fed at dumpsites (Samuels and Altmann, 1991), resembling responses to provisioning observed in other primates (Asquith, 1989). Historical alterations in primate habitats may be responsible for population and species differences in geographic distributions as well as life history traits and behavior. Thus, Richard, Goldstein, and Dewar’s (1989) reclassification of Fooden’s (1980) four phyletic species-groups of macaques (Macaca fasicularis group, M . silenus-sylvanus group, M . sinica group, and M . arctoides group) into “weed” and “non-weed” species was based on differences in their ability to “establish themselves in disturbed sites and in their reliance upon human activities to provide food” (p. 571). Similar distinctions may be applicable to other primate species, such as the widespread distribution of tufted capuchin monkeys (Cebus apella) relative to other capuchin species (Freese and Oppenheimer, 1981) and to intraspecific, population differences, such as those between northern and southern Indian langur monkeys (Presbytis entellus, Hrdy, 1977). According to Richard et al.3 (1989) hypothesis, the “weed” macaques effectively expanded their ecological niche to include habitats altered by human agricultural practices during the last 10,OOO years. The widespread biogeographic distributions of some macaque species have also been attributed to their ability to adapt to altered habitats during Pleistocene glacial episodes (Eudey, 1980; see also Nozawa et al., 1991). Indeed, the effects of Pleistocene climatic changes on tropical forests are probably responsible for the contemporary distributions of primates other than macaques. The expansion and contraction of the Brazilian Atlantic forest during the Pleistocene glacial episodes created refuges that fragmented populations of muriquis and other endemic primates (Kinzey, 1982). The geographic separation of muriquis into a northern refuge, centered in the Rio Doce valley of the state of Minas Gerais, and a southern refuge, centered in the Rio TiCte valley of Silo Paulo state, is thought to have limited gene flow between populations even after continuity in forest cover was reestablished (see Section II,A,Z,b). b. Interpreting Recent Habitat Fragmentation. It is difficult to distinguish between the effects of historical events and recent habitat fragmentation on the behavior of individuals and the viability of their populations. Muriquis may have already survived multiple population bottlenecks and re-
PRIMATE BEHAVIORAL ECOLOGY AND CONSERVATION
117
peated bouts of inbreeding and drift that slowly eliminated deleterious alleles in their evolutionary past (see Section 111,BJ). Consequently, muriquis, and other primates with similar evolutionary histories of population isolation, may be less susceptible to the genetic consequences of recent habitat disturbances and population fragmentation than is typically assumed in population genetic models (Soul6 and Kohn, 1989; Lande, 1988 Boyce, 1992;Caughley, 1994;but see Balmford, 1996 and Brooks and Balmford, 19%). Both behavioral ecology and conservation biology recognize that obvious sources of local disturbance, such as the elimination or reduction of primate predators, irrigation systems that divert water from protected habitats to surrounding farms and pastures, selective logging, and the introduction of infectious diseases through human and livestock vectors, can affect habitat carry capacities and population densities in ways that influence behavior and population processes. Yet, although the effects of fluctuations in carrying capacity and population density are critical variables in population viability analyses and in ecological models of behavior, the effects of these variables on survival and reproduction are still limited to descriptive correlations (Johns and Skorupa, 1987; Fimbel, 1994; Ganzhorn, 1995). The size and composition of populations that experience sudden isolation will affect demographic and genetic processes, and ultimately population viability (see Section III,A,2) as well as behavioral responses. Comparative population studies are still necessary to assess how the behavior of individual responses to such changes interacts with a species’ evolutionary history. 2.
Behavioral Variation in Conservation Biology
Models in conservation biology are explicitly sensitive to population variation in ecology and demography. Indeed, the number of populations being modeled is the first variable identified in VORTEX, the PVA most widely employed to evaluate extinction probabilities for endangered primates (Lacy, 1993). Nonetheless, because few population data are available for most primates, single-group values are often extrapolated to represent a population (Dunbar, 1987; Kinnaird and O’Brien, 1991). The population viability estimates obtained in this way may be no more realistic than ecological and behavioral models of primate group sizes or sex ratios based on species or group averages (Strier, 1996b). Unlike models in behavioral ecology, PVAs are intended to assess the effects of fluctuations in ecological and demographic variables on population processes. PVA users are required to specify the rate and effects of environmental catastrophes, such as drought or fire, on reproduction and survival,and changes in the specified carrying capacity due to environmental variance or trends in habitat destruction or recovery (e.g., Simberloff, 1988; Boyce, 1992; Caughley, 1994). However, as with phylogenetic models in
118
KAREN B. STRIER
primate behavioral ecology (see Section II,A,2), the values entered are only as reliable as the available data. The sensitivity of population viability analyses to the relationship between environmental and demographic variables does not extend to the relationships among these variables and behavior. In fact, only two behavior variables, type of breeding system and whether all males breed, are included in VORTEX (Lacy, 1993). Both variables will influence the genetic composition of populations over time, yet neither have straightforward values for most primates. Many other aspects of behavior also affect the variables of import to PVAs (Caro and Durant, 1995; Pope, 1996). For example, dispersal and mating patterns influence the demographic and genetic structures of populations, and are important components of models in primate behavioral ecology. The absence of comparative population data for many primates has required that PVAs not only rely on mating data to infer genetics, but also that they extrapolate from single groups to populations. The error that such extrapolations may introduce into population viability estimates will depend on dispersal and mating patterns, which are inextricably linked to population-specific demographic and ecological conditions. a. Dispersal. The absence of paternity data for most primates has meant that mating data have had to substitute for breeding. And, because most primates copulate much more frequently and with a greater number of partners than is necessary for fertilization (Hrdy, 1981; Small, 1989), mating data provide only limited insights into the actual genetic composition of offspring produced, and consequently, into the genetic variability of a population. Behavioral ecology has grappled with the parallel dilemma of inferring male reproductive success from mating success. Abundant examples from a diversity of primates demonstrate that mating patterns are unreliable predictors of reproductive success (Fedigan, 1983; Gray, 1985). The probability of fertilization varies markedly across a female’s ovulatory cycle, such that although related male golden-lion tamarins (Leontopithecus rosalia) both copulate with the same female, only the dominant male of each pair was observed to copulate during the periovulatory period when conception occurs (Baker, Dietz, and Kleiman, 1993). Furthermore, female strategies t o confuse paternity or use sex t o manipulate their social relationships can confound extrapolations from mating t o reproductive success (see Section IV,A). In the absence of genetic data, patterns of dispersal and philopatry have been used in behavioral ecology to infer whether primate societies are based on genetic matrilines, patrilines, or aggregations of unrelated individuals. The prevalence of male-biased dispersal among Malagasy lemurs and
PRIMATE BEHAVIORAL ECOLOGY A N D CONSERVATION
119
Old World cercopithecine monkeys has long been assumed to extend across most of the Primate order (e.g., Pusey and Packer, 1987), despite comparative data indicating that females or both sexes routinely disperse in many Old World colobine monkeys, most New World monkeys, and all of the apes (Moore, 1984,1992; Strier, 1990, 1994a). Although dispersal patterns appear to be phylogenetically conservative traits (Strier, 1990; Di Fiore and Rendall, 1994), they may vary intraspecifically or over time in a single population depending on local demographic and ecological conditions that affect reproductive opportunities (Sussman, 1992; Strier, in press, b). The genetic consequences of different types of breeding systems or the numbers of different males that breed will be strongly affected by the genetic relatedness among individuals. In fact, the genetic consequences of a multimale breeding system in which males are closely related to one another may be more similar to those of a single-male breeding system than to those of a multimale breeding system in which males are unrelated or distantly related (Pope, 1990; Martin, Dixson, and Wickings, 1992; Durant and Mace, 1994). In general, kinship among males is associated with greater cooperation in between-group competition, but not necessarily with tolerance in within-group competition over access to mates (van Hooff and van Schaik, 1992; Strier, 1994b). In muriquis, male philopatry may facilitate cooperation among male kin and contribute to the low levels of overt competition among males over sexual access to females (Strier, 1992b, in press, c). At the two sites where observed copulations have been reported, females routinely mate with multiple partners (Milton, 1985; Strier, 1987). Significant variance in male muriqui mating success was detected during a 5-year study period at the Estaqao Biologica de Caratinga, Minas Gerais, Brazil (Strier, in press, b). However, the genetic consequences of potentially high variance in individual male reproductive success cannot be distinguished from low variance in male inclusive fitness from behavioral observations alone. Behavioral and demographic data are also inadequate to evaluate the genetic consequences resulting from whether migrants disperse together or independently (Altmann and Altmann, 1979). During a 16-month study at Beza Mahafaly, Madagascar, male ring-tailed lemurs (Lemur cutfa) routinely migrated in cohorts of two or three individuals in response to the size and sex ratios of their new groups (Sussman, 1992). In wild vervet monkeys (Cercopithecus aethiops) at Amboseli National Park, Kenya, young males were found to disperse into the same groups that their older maternal brothers had previously joined (Cheney and Seyfarth, 1983). Migrating together may permit males to cooperate in detecting predators and in defending themselves against attacks from resident males in the groups they attempt to join, and may also affect their reproductive opportunities.
120
KAREN B. STRIER
The dispersal of female kin also affects the genetic structure of primate populations. For example, although adolescent female muriquis routinely disperse from their natal groups, the limited number of groups (2-3) in the only population where long-term dispersal patterns have been monitored means that granddaughters are likely to disperse into groups where their grandmothers reside and are still reproductively active (Strier, 1992a). Dispersal by female cohorts may include paternal sisters, and the muriquis’ 3year interbirth interval is short enough relative to their reproductive career to result in maternal sisters ending up in the same group. Similarly, female red colobus monkeys (Colobus badius) monitored over a 5 1/2-year period at the small, isolated Abuko Forest in The Gambia routinely dispersed into troops that included at least one other member of their natal troop (Starin, 1994). It is unclear, however, whether females disperse with kin in larger populations with greater dispersal options. b. Extragroup Matings. Patterns of mate choice, like dispersal patterns, can affect the genetic structure of primate groups. When mating opportunities extend to include extragroup copulations, as they do in a number of primates, mate choices can also affect the genetic structure of populations. Extragroup males accounted for 35% of ring-tailed lemur matings observed in one year (Sussman, 1992), and up to 13% of muriqui matings over a 5year period (Strier, in press, b). Significant levels of extragroup copulations have also been observed in polygamous patas monkeys (Harding and Olsen, 1986; Rowell and Chism, 1986), forest guenons (Cords, 1987), Barbary macaques (Mehlman, 1986; Small, 1990), Japanese macaques (Sprague, 1991,1992), and a variety of lemurs (Richard and Dewar, 1991). Extrapair copulations have also been documented in three groups of wild pair-bonded gibbons in Thailand, indicating that even monogamous social groups may be genetically polygamous (Reichard, 1995). The prevalence of extragroup copulations raises the question of whether primate social groups are legitimate units for analyses of reproductive or genetic processes (Cords, 1987; Moore, 1992; Morin and Woodruff, 1992; Rowell, 1993; Strier, 1994a). Groups may still provide the social contexts for individual primates in their day-to-day activities on an ecological time scale, and therefore may still be relevant in primate behavioral ecology. However, populations may be more appropriate units for analyzing primate reproduction on the evolutionary time scale of interest to behavioral ecologists, just as they are for evaluating population genetics in conservation biology. Even if appropriate data on population-wide breeding parameters were available, PVAs would still be vulnerable to error due to unaccountable variation in the behavior of individuals. For example, female muriquis who were known to have given birth to sons that had reached maturity and
PRIMATE BEHAVIORAL ECOLOGY AND CONSERVATION
121
become sexually active in the main study group mated significantly more frequently with extragroup males than did females with no known male kin in the study group (Strier, in press, b). By extending their copulations to include extragroup males, these females may have been able to avoid close inbreeding without compromising the number of different sexual and potential reproductive partners available to them. Conversely, females that had recently immigrated into the study group avoided (or did not seek out) extragroup copulations, and therefore the possibility of conceiving with closely related males from their natal groups. Avoidance of deleterious inbreeding has been one of the primary explanations for dispersal patterns in primates and other animals, and will be considered in more detail later (see Section II,C,2,b). Here, it is worth emphasizing that behavioral mechanisms to avoid close inbreeding evidently differ among individuals depending on their genetic relationships with potential mates, and among species depending on their ability or opportunities to recognize close kin. For example, among ring-tailed lemurs, male migration patterns have been attributed to female preferences for extragroup or peripheral males and expulsion of older, more familiar and related males (Sussman, 1992). Captive female cotton-top tamarins generally fail to ovulate when housed with related males, even in the absence of another female, suggesting that reproductive suppression may be a mechanism that promotes female emigration and prevents inbreeding when female dispersal opportunities are limited (Widowski, Ziegler, Elowson, and Snowdon, 1990). Demographic and life history processes, including the sex and survivorship of offspring produced, dispersal rates, and maturation rates, affect and are affected by population size, composition, and density. Thus, the goal of understanding individual behavioral variability from an evolutionary perspective in behavioral ecology requires a corresponding commitment to understanding population demography, just as PVAs in conservation biology require greater attention to individual variability. C. INDIVIDUAL-LEVEL ANALYSES 1. Sampling Methods in Primate Behavioral Ecology and Conservation
The comparative paradigm in primatology requires that data obtained by different researchers from different studies be collected in similar ways. The widespread acceptance of systematic methods of behavioral sampling in the 1970s helped define the parameters of data collection to reduce observer biases (Altmann, 1974). Standardized methods emphasized the individual as the basic unit of behavioral analysis, and permitted the applica-
122
KAREN B. STRIER
tion of quantitative statistical analyses to describe behavior and evaluate hypotheses about behavioral variation. The use of statistics to attain mean and deviant values assumes variability in the behavior being sampled. Indeed, the concept of sampling is itself an attribute of statistics that distinguishes primatology (and other areas of ethology and behavioral ecology) from most ethnographic research (Strier, 1993). Behavioral variation across individuals or across the lifetime of an individual can be documented and compared against other presumably more independent, ecological variables. Probabilistic analyses used in estimating minimum viable populations in conservation biology have been limited by the lack of detailed knowledge of how variable individual behavior affects population norms. Population approaches employed in conservation biology, unlike the individual approaches in behavioral ecology, are explicit in their reliance on normative data and central tendency statistics (DeRousseau, 1990). Yet, as in phylogenetic analyses of primate behavior (see Section II,A,2,a), such population averaging can lead to misleading predictions about genetic and demographic processes.
2. Individual Strategies and Demographic Stochasticity Female fertility is determined by a combination of age at first reproduction, interbirth interval, and longevity, whereas female reproductive success includes fertility and the survival of offspring to reproductive age (Richard, 1985; Richard, Rakotomanga, and Schwarz, 1991). Both maturation and survival rates are influenced by metabolic rates associated with body size and brain size (Martin, 1981; Partridge and Harvey, 1988; Foley and Lee, 1991), which may be species specific. Environmental constraints related to age, nutritional condition, vulnerability to predation or disease, and social history are attributes of individuals that may change over an individual’s lifetime, and may vary among individuals within a population and across populations of the same species. Individual strategies to maximize lifetime reproductive success must therefore be considered within the context of biological constraints and the interactions between environmental variables affecting reproduction, maturation, and survivorship. For example, when reproduction is not tightly seasonal, as in capuchin monkeys, spider monkeys, howler monkeys, and muriquis, infant mortality may lead to shorter interbirth intervals (Fedigan and Rose, 1995; Strier, in press, a). By contrast, in primates with tight reproductive seasonality, such as sifakas (Propithecus verreauxi), infant mortality may have little or no effect on annual interbirth intervals (Richard et al., 1991). Indeed, male reproductive strategies such as infanticide, which interrupts postpartum lactation amenorrhea, or high levels of infant care,
PRIMATE BEHAVIORAL ECOLOGY A N D CONSERVATION
123
which help females to recover their energy reserves, have been associated with the degree to which males can alter female reproductive rates (Strier, 1996a). Population models applied to primates have failed to consider the effects of individual behavioral strategies on reproductive or survival rates (Asquith, 1989). Instead, individual variation in fertility is treated as demographic stochasticity, which will have more severe consequences in small populations than in large ones (Caughley, 1994). By contrast, the effects of environmental stochasticity on population rates of increase are similar for large and small populations. However, distinctions between demographic and environmental stochasticity may have limited utility because both can alter the individual demographic traits, such as the probability of survival, that affect populations (Simberloff, 1988; Boyce, 1992; Caughley, 1994). Examples of how individual strategies that affect offspring sex ratios and dispersal and mating patterns affect the demographic structure of populations emphasize the importance of individual-level analyses in conservation biology. a. Sex Ratios. Population sex ratios are affected by age-specific mortality, maturation rates, and dispersal events, and by the interactions between individual life history strategies and demography. For example, individual differences in the social and reproductive conditions of mothers affect offspring survival and reproductive success, and may select for different reproductive strategies (Trivers and Willard, 1973; Gomendio, CluttonBrock, Albon, Guiness, and Simpson, 1990). In yellow baboons, daughters remain in their natal groups and inherit the dominance ranks of their mothers, whereas males generally disperse from their natal groups and must establish their dominance ranks and breeding positions on their own (Altmann, Hausfater, and Altmann, 1985). The female-biased infant sex ratios of dominant mothers and male-biased infant sex ratios of subordinates correspond to the higher rank and presumably reproductive success that daughters of dominant females can attain relative to daughters of subordinate females (Altmann, 1980). Conversely, in spider monkeys (Ateles paniscus), female-biased dispersal and male philopatry have been associated with the tendency for dominant females to produce more sons than daughters (Symington, 1987). However, if female rank is related to age in spider monkeys, as it is in howler monkeys (Glander, 1980; Jones, 1980) and some of the colobine monkeys (e.g., Presbytis entellus, Hrdy, 1977), then it is possible that infant sex and material investment strategies may change over a female spider monkey’s lifetime as she matures and rises in the social hierarchy (Strier, in press, a). Female primates may bias the sex of their offspring in response to social conditions at different stages in their life histories (Gomendio, 1995). For
124
KAREN E. STRIER
example, the higher proportion of sons produced by primiparous female muriquis may reflect the benefits to immigrants of producing a son early in their reproductive careers and thus assuring the presence of a related male ally in their group for life. However, producing too many sons might limit the number of potential mating partners available to a female, necessitating extragroup copulations or dispersal to avoid inbreeding (see Section II,B,2,6). Similar shifts in the ratio of sons and daughters produced by captive cotton-top tamarins have also been suggested (C. Snowdon, personal communication). Investment by dominant, healthy females in sons may be greater because of the higher variance in male reproductive potential compared to that of females (Trivers and Willard, 1973). However, evidence for such differential investment in offspring is mixed. Longer interbirth intervals following the birth of daughters have been reported in a colony of long-tailed macaques (van Schaik er al., 1989) and in provisioned Japanese macaques (Takahata, Koyama, Huffman, Norikoshi, and Suzuki, 1995), but interbirth intervals were independent of offspring sex in Barbary macaques (Paul and Kuester, 1990), olive baboons (Smuts and Nicolson, 1989), and chimpanzees (Nishida, Takasaki, and Takahata, 1990). A comparative survey of spider monkeys found no difference in interbirth intervals following the births of sons or daughters (Chapman and Chapman, 1990; but see Symington, 1987). Although muriqui mothers wean sons 2-4 months earlier than daughters (Rimoli, 1992) and resume postpartum cycling sooner after the births of sons than daughters, interbirth intervals following sons and daughters do not vary (Strier, in press, a). Biased birth and adult sex ratios may be a product of the interacting effects of stochastic variation on reproductive strategies. For example, in most Old World primates, sex ratios shift from being male biased among infants to being female biased among adults because of greater male mortality. More balanced adult sex ratios can occur if survivorship among males approaches that of females, as appears to have been the case in one population of sifakas monitored at Beza Mahafaly, Madagascar (Richard et al., 1991). Skewed sex ratios due to individual reproductive strategies and stochastic variation will affect the demographic structure, and thus viability, of populations (Boyce, 1992; Caughley, 1994). In primates, male-biased sex ratios increase the probability of extinction, whereas female-biased sex ratios increase the probability of population expansion (Gabriel, Burger, and Lynch, 1991). In a preliminary simulation of muriqui population viability, even slight deviations from a balanced infant sex ratio had dramatic consequences on estimated extinction probabilities (Strier, 1996b). However, a shift in male-biased infant sex ratios could lead to a corresponding shift in
PRIMATE BEHAVIORAL ECOLOGY AND CONSERVATION
125
male survivorship due to aggression and other sources of mortality resulting from increased competition, or to other behavioral adjustments, such as dispersal and group fissioning (Sussman, 1992). Indeed, a reversal from male-biased to female-biased sex ratios was observed in spotted hyenas (Crocutu crocutu) following clan fission due to the manipulation of cub survivorship by high-ranking females (Holekamp and Smale, 1995). Such behavioral responses may offset the effects of stochastic processes that produce unfavorable sex ratios, and must be considered in population viability analyses as well as in sex allocation theories in behavioral ecology. b. Dispersal and Inbreeding. Although dispersal events involve individuals, their effect on population structure “appears to be important in smoothing inequalities in sex ratios or in numbers of animals within groups, as well as minimizing inbreeding and increasing gene flow” (Sussman, 1992, p. 408). In this respect, the consequences of individual dispersal strategies on population structure may be similar to those of individual investment strategies in sons and daughters (Trivers and Willard, 1973). Predation risks, population density, and the size and composition of groups will influence an individual’s decision to disperse at any particular time (Alberts and Altmann, 1995). Thus, male mantled howler monkeys may spend up to 4 years as solitary individuals because opportunities to breed in their natal groups, join existing groups, or establish their own groups are limited in some habitats (Glander, 1992). Alternatively, female buffy-headed marmosets (Cullithrix flaviceps) may remain in their natal groups as sexually mature nonbreeders until they can inherit their mother’s reproductive position because the stability of social groups substantially limits their opportunities to disperse in search of breeding opportunites elsewhere (Ferrari and Diego, 1992; Digby and Ferrari, 1994). While the act of dispersing is a discrete property of an individual, who either disperse or doesn’t, behavioral ecologists have tended to classify primates by whether males, females, or both sexes disperse from their natal groups (e.g., Pusey, 1987; Pusey and Packer, 1987). The pattern of dispersal is defined by some arbitrary proportion of observed or suspected events, but the proportion of individuals that deviate from this pattern need not be high to have significant biological consequences for group size, composition, and social behavior (Moore, 1984; Moore and Ali, 1984;Packer, 1985). Indeed, simulations employed in conservation biology estimate that only minimal migration, on the order of one migration event per generation, may be sufficient to maintain genetic variation and reduce extinction probabilities for small isolated populations of endangered species (Ralls et ul., 1986; Harcourt, 1995). Population viability models in conservation biology do not consider the role of individual decisions or strategies in dispersal events. Rather, these
126
KAREN B. STRIER
models focus on the effects of dispersal on the genetic composition of populations. Dispersal between populations tends to increase gene flow, and thus is usually regarded as a way of introducing genetic variation into populations and maintaining the genetic integrity of a species (Gilpin and Soul& 1986; Chepko-Sade et al., 1987; Boyce, 1992). Yet, many primates appear to avoid close inbreeding and even increase gene flow by mechanisms other than dispersal (see Section II,BZ,b). Incorporating these alternatives to dispersal into demographic and genetic models of populations may lead to more reliable predictions about population viabilities (Caro and Durant, 1995).
111. GENETICS AND ECOLOGY
Behavioral ecology and conservation biology share an evolutionary and ecological approach. However, disciplinary differences in the underlying assumptions about genetic and environmental variation are as extreme as the assumptions affecting their respective levels of analyses (Table I). Understanding how some of these divergent assumptions have influenced
TABLE I COMPLEMENTARY ASSUMPTIONS IN PRIMATE BEHAVIORAL ECOLOGY AND CONSERVATION BIOLOGY Variable
Primate behavioral ecology
Conservation biology
Orientation Variability in Species:
Theoretical. long-term studies
Applied, rapid assessments
Homologies versus convergence in behavioral traits Ecological or “cultural” conditions usually based on intergroup comparisons
Genetics and preservation of biodiversity Habitat fragmentation: effects on population genetic and demographic structure and size determine viabilities Demographic stochasticity: averaged for effects on populations Population genetics; concern with loss of genetic variability (genc frequencies) Stochastic, catastrophes affect populations Effects on populations (e.g., dispersal, mate choices)
Populations:
Individuals:
Behavioral responses and strategies affect fitness
Genetics:
Behavior is heritable: selection on fitness of individuals (genotypic) Deterministic, individuals respond Individual adaptations to ecology and demography
Environment: Behavior:
PRIMATE BEHAVIORAL ECOLOGY AND CONSERVATION
127
primate behavioral ecology and conservation biology can provide insights into areas of past confusion and potential intersection between the fields. One of the fundamental distinctions between conservation biology and behavioral ecology involves the interactions between individuals and their environment. This distinction is analogous t o DeRousseau’s (1990, p. 7) characterization of the difference between population and life history studies: “Where population study assumes that the organism is a passive manifestation of its genes, in life history the organism is active and constantly negotiating with its environment to survive and reproduce.” In behavioral ecology, as in DeRousseau’s definition of life history studies, individual behavioral variation, “is not ‘noise’ in the system, but instead relevant to our understanding of evolution.” In conservation biology, by contrast, it is the “detection of a signal (such as declining population size) over the noise (or random fluctuations) inherent in most systems” (Soul6 and Kohn, 1989, p. 31) that is of primary concern. A.
ASSUMITIONS ABOUT GENETICS
1. Properties of Genes in Primate Behavioral Ecology
The application of evolutionary theory to primate behavior has been consistent with primatology’s long-standing comparative paradigm and its methods of sampling individual behavior and ecological variables systematically (see Section 11). Evolutionary theory has resonated with traditional comparative approaches in primate behavior because of the continuities it assumes between species and in the principles of natural selection that operate on the behavior of individuals. Similarly, evolutionary theory’s prediction of a relationship between the variance in individual behavior and fitness has been concordant with systematic methods of behavioral sampling because reproductive success and inclusive fitness provide a (theoretically) measurable currency against which individual behavioral variation can be evaluated and correlated with other individual attributes, such as age-sex class, reproductive history, maternal kinship, and dominance status. Thus, variation in individual behavior should reflect evolutionary selection pressures operating on individuals to maximize their fitness or genetic contribution to future generations. Carriers of these genes, in the form of surviving offspring and kin, should behave in similarly genetically successful ways. a. Effects of Selection. Inherent in this simplistic depiction of the application of evolutionary theory to primate behavior is the assumption that behavior has a strong genetic, or heritable, component. Indeed, genetic variation is the foundation of evolutionary theories of behavioral adaptation (Krebs and Davies, 1991), and many primates behave in ways that are
128
KAREN B. STRIER
consistent with predictions from evolutionary theory. For example, individuals known to be genetic kin behave differently toward one another than they do toward unrelated individuals. In general, affiliative interactions such as grooming and cooperative coalitions occur more frequently among kin than among nonkin, whereas agonistic interactions and overt forms of competition occur more rarely and tend to be less intense among kin than among nonkin (Kurland, 1977; Gouzoules, 1984; Gray 1985). The application of evolutionary theory to primate behavior also assumes that the strength of selection pressures and the rate of influx of new genetic variation (through mutation, sexual recombination, and migration) will determine the distribution of a particular behavioral strategy and the rate of fixation of the associated genes responsible in a population over time. Under strong directional selection pressures, less successful behavioral strategies will be selected against in successive generations until eventually the only behavioral variation expressed in a population is a consequence of mutation rates, recombination, and gene flow. Strong stabilizing pressures should also tend to reduce genetic variation at the loci under selection, and thus reduce behavioral variation in the population. When averaged over all loci, the implications of homozygosity for population viability may be significant (see Section 111,AJ). b. Interpreting Behavioral Variation. There are a number of explanations for why individual behavior may be more variable, and less uniformly adapted, than what would be predicted based on the effects of selection on behavioral phenotypes and the underlying genetic variation. Fluctuations in environmental conditions, including features of the physical, demographic, and social environment, differences in the optimal behavioral strategies at different stages of an individual’s life history, and frequencydependent behavioral responses are some of the processes that make behavioral adaptations as difficult to identify as morphological adaptations (e.g., Lewontin, 1978). Like some morphotypes, behavioral variants, often described as alternative behavioral strategies, may be functionally equivalent or selectively neutral in terms of their effects on fitness. For example, male baboons that establish affiliative relationships with females may become preferred mates, and therefore be as reproductively successful as males that compete aggressively with one another for dominance rank (Strum, 1982, 1983; Smuts, 1985). Hausfater’s (1975) suggestion that dominance rank in male baboons might not correlate as expected with lifetime variance in reproductive success because males change ranks throughout their lives has been supported by a number of other studies on male and female rank and reproductive success (Gray, 1985). Indeed, in a long-term demographic study of one marked population of mantled howler monkeys, Glander (1992) found that
PRIMATE BEHAVIORAL ECOLOGY A N D CONSERVATION
129
females that had recently immigrated into a group had the highest levels of infant mortality while they were attaining their dominant position. As these same individuals dropped in rank with age, the survival of their infants, and thus their reproductive success, improved. Ontogenetic processes, including learning, socialization, and experimentation, may also interact with evolutionary forces that affect behavior, particularly in long-lived primates with delayed maturation. Presumably, the ability to learn and experiment with novel materials has some basis in genetics, but abundant evidence has demonstrated that innate behavioral tendencies can be enhanced or inhibited during socialization (e.g., Harlow and Zimmermann, 1959). Consequently, it is difficult to evaluate whether the unexpected rise in rank by Mike, the male chimpanzee at Gombe National Park that startled his competitors with his innovative can-smashing display (Goodall, 1986, p. 7 9 , was a consequence of a genotype that led him to manipulate features of his environment in a novel way, a deprived socialization period that led him to prefer inanimate objects over social interactions, or an enriched socialization period that gave him confidence to experiment, or a chance event. Whatever the cause, to the extent that his rise in rank was accompanied by greater opportunities to fertilize females, his behavior had evolutionary consequences.
2. Population Genetics in Conservation Biology Genetic variation is as fundamental to conservation biology as it is to the applications of evolutionary theory in behavioral ecology. As is the case in behavioral ecology, genetic variation provides the raw material for natural selection pressures to operate on individuals and for populations to evolve. However, in conservation biology, population gene frequencies are more important than the individual genotype, and it is the processes that introduce new genetic variation into populations rather than the behaviors of individuals that carry these genes that are emphasized in population viability models. Genetically heterozygous individuals are often assumed to have greater fitness than homozygotes, but there is little evidence to support this assumption (Soul6 and Kohn, 1989; Boyce, 1992). Indeed, it is not clear whether the greater vigor in fertility, survival, or growth reflects high overall heterozygosity or simply low homozygosity for deleterious recessive alleles (Allendorf and Leary, 1986). Conservation biologists have emphasized the importance of preserving genetic diversity, but how much or what type of genetic variation is necessary for population viability is not known (Boyce, 1992; Caughley, 1994). Species and populations are known to vary in their genetic loads, and some, such as cheetahs, have persisted with very low levels of genetic variation for many generations without evident deleterious
130
KAREN 8 . STRIER
effects (Ralls, Ballou, and Templeton, 1988; Lindburg et al., 1993; Car0 and Laurenson, 1994; Merola, 1994; O’Brien, 1994; Frankham, 1995; see Section I). a. Evolution and Population Size. Genetic mutations occur spontaneously at roughly constant rates (but see Lenski and Sniegowski, 1995 for discussion of directed and adaptive mutations). Mutation rates tend to be lower for single genes with large effects than they are for those with minor effects (Lande and Barrowclough, 1987). In general, mutations with large effects are almost always eliminated by natural selection, but minor mutations can accumulate over generations, and ultimately reduce the genetic viability of a population (Lande, 1995). On the other hand, the accumulation of small, quasi-neutral mutations can also maintain genetic variability in a population, and thus its adaptive potential. The accumulation of quasi-neutral mutations is sensitive to both time and population size. The variation introduced by quasi-neutral mutations with small effects on fitness can be lost due to inbreeding or genetic drift in small, isolated populations where opportunities for gene flow with other genetically different populations are rare or nonexistent. Thus, small population size and high levels of inbreeding can interfere with the viability of a population through the gradual depletion of genetic variation necessary for adaptation and through the accumulation of minor mutations with additive deleterious effects (Lande, 1995). Slow inbreeding and selection can purge a population of deleterious alleles with the paradoxical result that the population exhibits little genetic variation without suffering from inbreeding depression (Simberloff, 1988; Boyce, 1992; Caughley, 1994; Lande, 1995). These processes may explain the loss of heterozygosity without corresponding loss in survival among chimpanzees at Gombe National Park, Tanzania (Morin et al., 1994). However, inbreeding depression may have a severe impact on a population if a previously outbred species is suddenly forced to inbreed due to rapid fragmentation of its habitat because selection may not have time to eliminate deleterious alleles as they are expressed (Simberloff, 1988). The genetic consequences of past inbreeding can also accumulate before selection can eliminate the deleterious alleles, making it difficult to distinguish between population extinctions caused by past or recent inbreeding depression (Frankham, 1995). In conservation biology, effective population size, N,, represents only the number of breeding individuals. Early estimates of minimum viable population sizes were based on a balance between how small a population could be without deleterious inbreeding depression and how large it must be for selection forces to counteract the random fixation of alleles through genetic drift. Lande’s (1995) more recent analyses of mutation and evolu-
PRIMATE BEHAVIORAL ECOLOGY AND CONSERVATION
131
tionary processes have challenged Franklin’s (1980) values for minimal effective population size. However, the interacting effects of inbreeding and drift make it difficult to generalize minimum viable population sizes across species. b. Confounding Variables. Species differences in reproductive rates, which affect generation lengths, will alter the speed and severity with which inbreeding affects population viability. Large, iteroparous species that maintain small populations may be at higher risk for inbreeding depression than small bodied species with high reproductive rates (Gilpin and SoulC, 1986). By contrast, species with rapid generations may be more vulnerable to environmental and demographic stochasticity, which affects population age and sex structures (Caughley, 1994; Mills and Smouse, 1994). The actual effects of genetic drift may be similarly confounded by both a population’s history and its mating system. Genetic drift is a random process in which alleles become fixed in a population independent of their selective value (Caughley, 1994). As with other random processes, the smaller the sample size, the higher the probability of skewed outcomes. Small effective population sizes may arise in large populations when high variance in male (or female) reproductive success or high inbreeding occurs (Smith, Rolfs, and Lorenz, 1992). The flexible mating strategies employed by many primates can reduce the occurrence of inbreeding between close relatives, even in small, isolated populations (see Section II,A,2,b). The behavioral responses of individuals can thus influence the genetic structure of populations in ways that are not generally considered in population models.
B. ASSUMPTIONS ABOUT ENVIRONMENTAL VARIABLES 1. Ecological Determinism in Primate Behavioral Ecology
Primatologists have always been sensitive to the ways in which environmental variables shape behavior. Such ecological perspectives extend from and are consistent with models of human behavioral evolution, which considered shifts such as those from arboreal to terrestrial lifestyles and from vegetarian to omnivorous diets to be prime movers of human morphological and behavioral adaptations (Tooby and DeVore, 1987). The early comparative ecological models divided primates into adaptive grades based on gross ecological distinctions that appeared to correspond to broadly defined social phenomena such as group composition and mating systems (e.g., DeVore, 1963; Crook and Gartlan, 1966; Eisenberg, Muckenhirn, and Rudran, 1972). Better data on a more diverse array of species have led to increasingly fine-grained analyses of both ecological and behavioral variables, and comparative ecological models have, to some
132
KAREN 9. STRIER
extent, been replaced with ecological principles that transcend simple ecological categories (Garber, 1987). Thus, early ecological classifications of primates based on the relative proportions of leaves and fruits in the diet (e.g., Milton and May, 1976; Clutton-Brock and Harvey, 1977; Hladik, 1981; Kay, 1984) have been replaced by classifications based on the chemical properties of primate foods (Glander, 1978, 1982; Davies, Bennett, and Waterman, 1988; Oftedal, 1992) and the spatial and temporal distribution and availability of foods ( Wrangham, 1980). These ecological distinctions stimulated a number of empirical studies on the relationships between food patch size and feeding group size in primates with flexible grouping patterns (e.g., Leighton and Leighton, 1982; White and Wrangham, 1988; Chapman, 1988; Symington, 1988; Strier, 1989). Contemporary classifications further distinguish annual proportions of foods with different chemical and nutritional properties included in primate diets from foods that are utilized during periods of overall food scarcity (Koshimizu, Ohigashi, Huffman, Nishida, and Takasaki, 1993; Leighton, 1993; Waterman and Kool, 1994; Mowry, Decker, and Shure, 1996). Primates appear to shift their behavior in response to seasonal fluctuations in food availability and other environmental resources (Wrangham, 1981). Increasing specificity in how ecological variables have been defined and measured has been accompanied by correspondingly fine-grained analyses of behavioral variables. Insights into the relationships between primate ecology and behavior have come from detailed analyses of the compromises between avoiding feeding competition among related individuals and cooperating with reliable allies in competition with unrelated individuals ( Wrangham, 1980). These compromises require distinguishing between direct, or contest, competition and indirect, or scramble, competition (Wrangham, 1982;van Schaik, 1989) and between strategies for avoiding predation and maximizing feeding efficiency (van Schaik and van Hooff, 1983; Terborgh, 1985; Terborgh and Janson, 1986). In most of these analyses, the relationship is assumed to be unidirectional, with primates adjusting their behavior in response to, but with little effect on, their ecology. There are notable exceptions to the general perception of ecology as the independent variable affecting primate behavior. Among these exceptions are studies demonstrating the effects of food patch depletion on foraging and ranging strategies (Chapman, 1988) and mixed-species associations (see review in Chapman and Chapman, 1996), and the roles of primates as pollinators (Sussman and Raven, 1978) and seed dispersers or seed predators (Davies, 1991; Kinzey and Norconk, 1993;Peters, 1993;Chapman, 1996). However, unlike conservation biology, which has historically been interested in community interactions (Simberloff, 1988), primate behavioral ecology has tended to focus on how ecology affects behavior rather than
PRIMATE BEHAVIORAL ECOLOGY AND CONSERVATION
133
on the interactions between them. Consequently, despite the importance of density-dependent effects on population viability in conservation biology (Ginzburg, Ferson, and Akqakaya, 1990; Lacy, 1993), primate behavioral ecology has contributed few insights into how primates may alter their population densities and carrying capacities.
2. Environmental Stochasticity in Conservation Biology Environmental variables, including carrying capacity, rainfall patterns, and catastrophes such as drought and fire, figure prominently in population models in conservation biology. Population responses to these environmental variables are reflected in survival, mortality, and reproductive rates, and are modeled as stochastic environmental processes (Simberloff, 1988). Because natural catastrophes occur at unpredictable rates, they can lead to indiscriminate mortality across age-sex classes and disruptions in reproduction. While demographic stochasticity can have large impacts on small populations (see Section III,A,2), environmental stochasticity affects populations independent of their size (Boyce, 1992). Inadequate knowledge of how individuals modify their life history and behavioral strategies in response to changing conditions in their physical environments has limited the ability of population models to distinguish the effects of environmental variables on population size from those on population age and sex composition (Caughley, 1994). Demographic characteristics of primate groups and populations may also affect and be affected by environmental variables. For example, intestinal parasites may be more prevalent among primates living in large groups o r at high population densities where opportunities for infectious transmission are greater (e.g., baboons, Appleton, Henzi, Whiten, and Byrne, 1986; chimpanzees, McGrew et al., 1989). Alternatively, primate group sizes and population densities may be limited by the risks of infectious parasites, which depend on environmental conditions affecting the life cycles of parasites (Freeland, 1976; Hart, 1990; Stuart and Strier, 1995). Moist conditions may promote a higher prevalence of parasitic infections because they are conducive to parasite survival in excreted feces (Hausfater and Meade, 1982). In a comparative study of parasitic infections in four muriqui populations, Stuart et a!. (1993) found a higher prevalence of parasites at the wettest, most humid site. However, the lowest prevalence of parasitic infection also occurred in the largest, most cohesive muriqui group, making it difficult to generalize about the interactions between climate, muriqui group sizes, and the risks of parasite infections. Changes in parasitic infections over time may reflect the interaction between behavioral and environmental variables. For example, the absence
134
KAREN E. STRIER
of parasites in muriqui feces collected at the Estaq3o Biologica de Caratinga may have reflected relatively dry conditions, small group sizes, or recent habitat disturbances that had disrupted parasitic life cycles (Stuart et al., 1993). Whether parasitic infections in these muriquis will affect the fitness of individuals or the viability of their population is not yet clear. Indeed, conservation strategies to establish corridors between isolated populations to facilitate dispersal and reduce inbreeding have been debated, in part, because of the risks of transmitting infectious parasites or disease (Boyce, 1992). It is evident that environmental factors cannot be decoupled from the behavior of individuals and the dynamics of populations (Thomas, 1990).
C. INTERACTIONS BETWEEN GENES AND ENVIRONMENTS 1. Behavioral Responses of Individuals and Groups
Evolutionary models of behavior assume that selection pressures lead individuals to adjust their behavior, including their grouping and mating patterns, in response to environmental stimuli (Richard, 1981; Asquith, 1989). Through these behavioral adjustments to demographic conditions, individuals also influence their social environment by altering the size and composition of their groups. For example, expanding primate populations may have higher levels of kinship among group members, and therefore greater opportunities for the expression of nepotistic behavior (Altmann and Altmann, 1979). Social structure, or group size and composition, may therefore be regarded as both the outcome of individual behavioral decisions, or strategies, and a determinant of them, as appears to be the case in primates ranging from mountain gorillas (Robbins, 1995) to common marmosets (Digby and Ferrari, 1994). While individual behavior, including life history tactics and social strategies, may be inextricably related to social structures, it tends to be too variable to permit more than imperfect inferences about group organizational properties (Rowell, 1993). Attempts to understand societies from their individual and dyadic components are misleadingly simplistic because the most complex, and therefore most interesting, features of primate societies may be greater than the sum of their parts. The difficulty of translating individual strategies or dyadic social relationships into group-level phenomena that embody manipulation of resources has contributed to a widening gap among social scientists who study human and nonhuman primates (Schubert and Masters, 1991). If complex primate societies cannot be understood from their individual component parts, it may be necessary instead to examine the larger social and biological environment in which aggregations of individuals, or groups, function.
PRIMATE BEHAVIORAL ECOLOGY A N D CONSERVATION
135
Ecological variables, such as carrying capacity, and demographic variables, such as population density, comprise this larger context. Thus, whether an individual disperses o r remains in her natal group may depend as much on perceived or actual opportunities to find essential resources elsewhere as it does on the dynamics of intragroup relationships and individual life histories (Vehrencamp, 1983). While intra- and intergroup dynamics have become a major topic in primate behavioral ecology (e.g., Wrangham, 1980,1982; van Schaik, 1989), virtually nothing is known about the behavioral correlates of habitat saturation or stable populations. Marginal habitats that support smaller groups of black and white colobus monkeys (Colobus guereza) or Barbary macaques may even be beneficial to individual reproductive success because individuals are released from stressful competition within and between groups (Dunbar, 1987; Mehlman, 1989). If primate populations tend to be unstable, as Dunbar (1979) has suggested based on the variable life history strategies and behavioral responses of individuals, then evolutionary theory, which assumes a heritable component of behavior, may offer only approximate explanations of the emergent properties of primate groups (Rowell, 1993). Consequences of Behavior for Populations The loss of genetic variation in populations was a focus of concern in the early PVAs of conservation biology (Franklin, 1980 Gilpin, 1987). However, the roles of environmental stochasticity and habitat fragmentation are now regarded as more immediate threats to population viability (Gilpin, 1987; Shaffer, 1987), and the interactions rather than distinctions between genetic, demographic, and environmental models of extinction are becoming more widely considered (Ewens, Brockwell, Gani, and Resnick, 1987; Lynch, 1996). Genetic and environmental effects interact at the level of differential survival and reproduction of individuals over time. For example, in a genetic analysis of a song sparrow population (Molospiza rnelodia) on Mandarte Island, British Columbia, Keller, Arcese, Smith, Hochachka, and Stearns (1994) demonstrated that survivors of a population crash caused by severe winter weather were less inbred than the average precrash population. In a study of laboratory mice subjected to intensive inbreeding over multiple generations, smaller litter “populations” had higher survivorship than populations of average or greater than average size (Bowman and Falconer, 1960). Thus, the genetic structure of populations may affect a population’s response to environmental catastrophes (in the case of song sparrows), and demography may affect a population’s response to loss of genetic variation (in the case of inbred mice). 2.
136
KAREN B. STRIER
Comparable evidence for the interacting effects of genetics and environmental processes in wild primate populations does not yet exist. Instead, viability analyses of primate populations use starting population size and age-sex structure to model the rate at which genetic variation is expected to be lost. However, primates have longer generations than song sparrows and mice, and the effects of inbreeding depression on population viability may take longer than the 100 years that are usually simulated. Indeed, in primates such as muriquis, in which the earliest reported age at first reproduction is 7 years (Strier, 1991), a century represents only about 10 generations. The fact that inbreeding did not alter extinction probabilities or population size at 100 years in one PVA simulation of a muriqui population with a starting population of 52 individuals suggests that 10 generations may be insufficient time for deleterious effects of inbreeding to accumulate (Strier, 1996b). Most PVAs of endangered primates must also rely on extrapolations from data on size, composition, life history variables, and mating behavior based on single-study groups. However, the exchange of genes within populations is strongly affected by the ways in which individual primates are partitioned within groups. Thus, documenting the genetic consequences of intergroup dispersal and extragroup fertilizations (see Section II,B,2) are as important to conservation biology as they are to primate behavioral ecology. For example, Izawa (1994) has recently proposed that differentiated patterns in dispersal and intergroup encounters in capuchin monkeys may reflect a superstructural organization that transcends single groups. Population-level analyses may reveal higher order organizational principles, such as those found in hamadryas baboons (Rodseth, Wrangham, Harrigan, and Smuts, 1991), in other primates as well. The genetic consequences of such superstructural organizations on primate populations may be pronounced. For example, Sussman (1992) suggests that group fission in ring-tailed lemurs is more important than dispersal in adjusting population size over larger areas. In the matrilineal societies of rhesus macaques (Mucucu rnulutfu),“Fission accelerates subpopulational differentiation beyond the rate expected by drift alone. . . because average degree of maternal relatedness is always higher in fission groups than in preexisting parent groups” (Melnick and Kidd, 1983). In captive primates with limited opportunities for gene flow, the extinction of entire matrilines has been attributed to loss of genetic diversity (Smith etul., 1992). However, the genetic effects of group fission on populations will depend on the genetic composition of groups, whether they split along kinship lines (maternal or paternal), and whether individuals in splinter groups disperse or reproduce with individuals in other groups in the population. Indeed, if relatives
PRIMATE BEHAVIORAL ECOLOGY AND CONSERVATION
137
routinely disperse together, then at a genetic level, group fission may be no more than an extreme example of dispersal. Splinter groups usually move into less desirable habitats (Sussman, 1992), where differences in food availability and quality may alter individual life history variables such as survivorship and reproductive rates. Differences in vulnerability to disease, predator pressure, and interspecific competition will also affect both the size and composition of primate groups as well as populations. Groups, like populations, tend to expand when they are released from predator pressures or interspecific competition. Characteristics of expanding populations, such as higher birth rates, lower mortality rates, and female-biased infant sex ratios (Caughley, 1994) may also characterize expanding groups in ways that directly impact the larger population. Yet, virtually nothing is known about the behavioral determinants or consequences of variations in primate habitat capacities (e.g., Yeager, 1995).
IV. METHODOLOGICAL BRIDGES TO DISCIPLINARY CONVERGENCE New techniques for analyzing DNA and assessing reproductive condition in wild primates have made genetic and reproductive data on individuals and populations accessible to fieldworkers, where demographic and reproductive variables can be correlated with ecological conditions. The use of these methods has significantly influenced contemporary theory and research in behavioral ecology, and provides valuable tools for assessing the consequences of individual reproductive behavior on the genetic and demographic structure of populations important to conservation biologists. For example, the revival of reformulations of sexual selection and mate choice theories in behavioral ecology may be attributed to the development of more precise methods of genetic paternity analyses (Krebs and Davies, 1991), which can also be applied to understanding genetic variation within and between primate populations. Similarly, recent advances in noninvasive fecal steroid assays have begun to provide insights into the social and reproductive functions of sex and the interactions between ecological and life history variables. Knowledge of reproductive physiology in wild primates is directly applicable to monitoring and managing endangered primate populations and captive breeding programs. New capabilities to correlate genetic variability and reproductive physiology with behavioral data on mating systems, competition and cooperation, and diet and life histories facilitate the extension of primate behavioral ecology to include population-level analyses and of conservation biology to include individuals. While most research in primate behavioral ecology and conservation biology has, and will continue to be, theory and need
138
KAREN 8. STRIER
driven, new techniques to understand the genetic and reproductive bases of individual behavior and population viability provide opportunities for the fields to explore questions of mutual concern simultaneously at multiple levels. The result will be more informed comparative models of primate behavioral ecology that provide the data needed for effective conservation models (Fig. 2). A. DNA
AND
SEXI N PRIMATE POPULATIONS
I. Historical Approaches Prior to the development of reliable genetic paternity analyses, understanding of primate kinship has been limited to matrilineal genealogies derived from long-term studies of recognized individuals. Extrapolations from behavioral data to biological paternity in primates are limited and highly unreliable for several reasons. First, it is virtually impossible to observe all copulations in wild primates, especially if copulations occur at night or in dense vegetation. Second, most female primates copulate with multiple partners, and few species exhibit external signs of ovulation that can be used to infer which males have the highest probabilities of fertilization (Hrdy, 1981; Dixson, 1983). Indeed, genetic analyses revealing a preponderance of full siblings in gray seals (Halichoerus grypus) indicate a surprising level of mate fidelity (Amos, Twiss, Pomeroy, and Anderson, 1995) that was not predicted from behavioral data. By contrast, paternity exclusion studies on successive infants in captive Japanese macaques demonstrate that females conceived with different males in successive years (Inoue et al., 1992). In at least some primates, female mate choices are evidently not static over time (see Section II,B,2,b). 2. New Insights
Evaluations of predictions about males based on kin selection, inbreeding avoidance, and the relationships between dominance rank, mating success, reproductive success, and infant care are limited without genetic paternity data (Fedigan, 1983; Gray, 1985; Cowlishaw and Dunbar, 1991; Bartlett et al., 1993). However, recent methods of DNA analyses have succeeded in obtaining paternity exclusions for a large proportion of individuals sampled in some species, and have begun to demonstrate much greater diversity in the genetic consequences of behavior within and across populations. Thus, genetic analyses of wild Barbary macaques have supported evolutionary predictions that variance in reproductive success is higher among males than among females (Kuester, Paul, and Arnemann, 1995), and paternity analyses of chimpanzees (Pan troglodytes) at Gombe National Park, Tanzania, have demonstrated that philopatric males are more closely related
Steroid Assays
of Individuals’ Life Histories
of Population
Wlthin and Between PopulationComparisons
Ecol ical and Social Torrelates
Ecological and DensityDependent Variables on Reproductive Rates
Paternity Studies
of Population
of Individual Reproductive
Population Comparisons
Taxonomic Revisions for Conservation Strategies
Taxonomic Revisions for Phylogenetic
FIG. 2. Contributions to behavioral ecology and conservation biology from DNA and steroid analyses.
140
KAREN B. STRIER
to one another than to females and supported hypotheses linking male cooperation to kin selection theory (Morin et al., 1994). In the matrilineal societies of macaques, male rank and reproductive success were correlated in wild long-tailed macaques (Macaca fasicularis, de Ruiter et al., 1992; de Ruiter, van Hoof, and Scheffrahn, 1994) and in three out of four breeding seasons in a captive colony of Barbary macaques (M. sylvanus, Paul, Kuester, Timme, and Arnemann, 1993). However, only weak, if any, relationships between male rank and paternity were found in wild and semi-free-ranging Barbary macaques (MCnard, Scheffrahn, Vallet, Zidane, and Reber, 1992, and Kuester, Paul, and Arnemann, 1992, respectively) and captive Japanese macaques (M. fuscata, Inoue et al., 1992). Inbreeding was avoided among maternal kin in captive Japanese macaques, despite the fact that maternal kin were observed to copulate (Inoue et al., 1992), and in a large population of semi-free-ranging Barbary macaques (Kuester et al., 1992). Knowledge of paternity is also essential to understanding the reproductive significance of extragroup copulations and the relationship between social and reproductive units (Morin and Woodruff, 1992; see Section II,B,,?,b). Indeed, because genetic data suggest that variation is greater within than between many primate groups (Turner, Weiss, and Pereira, 1992), extragroup fertilizations, together with dispersal between groups and kinship among group members, may contribute to the genetic homogeneity of primate populations (Morin et al., 1994). 3. New Challenges
Genetic data obtained from paternity exclusion analyses can also be extended beyond groups and populations to understand the evolutionary relationships between geographically isolated populations or subspecies and between closely related species. Nozawa et al. (1991) suggest that the lower genetic variation found among southern populations of Japanese macaques compared with central and northern populations may be indicative of historical genetic isolation between regions. Genetic similarities among eastern long-haired chimpanzees (Pan troglodytes schweinfurthii) and central black-faced chimpanzees (P. t. rroglodytes) are believed to reflect substantial gene flow in their evolutionary past despite present-day isolation (Morin et al., 1994). By contrast, the significant genetic differences that distinguish western or pale-faced chimpanzees ( P . t. werus) from both central and eastern populations led Morin et al. (1994) to suggest the importance of reexamining comparative morphological and behavioral data to determine whether chimpanzee taxonomy warrants revision. The consequences of such taxonomic distinctions are clearly important for conservation strategies, including habitat preservation and reserve design, captive breeding programs and housing facilities, and the use of pri-
PRIMATE BEHAVIORAL ECOLOGY AND CONSERVATION
141
mates in biomedical research, as well as for comparative phylogenetic analyses of primate behavioral ecology (see Section II,A,2). Melnick and Hoelzer (1992) emphasize that different kinds of DNA are more appropriately analyzed for different questions. Thus, the maternal inheritance pattern of mitochondria1 DNA makes it suitable for phylogenetic studies, whereas nuclear DNA is more appropriate for paternity and population genetic analyses. Furthermore, not all primates are equally good candidates for paternity exclusion studies. Paternity exclusion methods require high levels of genetic variability within a population or species to work (Rogers, 1992), and may therefore have limited applications in small, isolated populations composed of genetically similar individuals (Martin, 1992). Genetic variability appears to be low in colobus monkeys compared to baboons and macaques (Turner er ul., 1992), though whether these differences reflect their divergent evolutionary histories or behavioral strategies remains unclear. The common pattern of producing twins and triplets in callitrichid primates results in higher genetic homogeneity than in species that give birth to single offspring (Anzenberger, 1992; Dixson, Anzenberger, Monteiro da Cruz, Patel, and Jeffreys, 1992), and is compounded by blood chimerism due to a shared placenta in callitrichid twins. The dispersal of siblings or half siblings into the same groups (e.g., vervet monkeys, Cheney and Seyfarth, 1983; muriquis, Strier, 1991; ring-tailed lemurs, Sussman, 1992) may also lower the genetic variability within populations. Paternity exclusion studies also require that mother-offspring relationships can be confidently established, and that all potential fathers can be sampled. Samples on all relevant individuals can be more consistently obtained in captive populations than in wild primate populations (Rogers, 1992), where the prevalence of extragroup copulations and dispersal in primates requires sampling at the population rather than group level. Nonetheless, the potential insights into the effects of behavior on genetic processes in wild primate populations justify the greater investment in obtaining appropriate samples. Recent advances in techniques to obtain DNA through noninvasive means, such as from hair samples extracted from sleeping nests constructed by chimpanzees and other apes, buccal cells attached to food wadges, or feces, facilitate the possibilities for genetic sampling of wild primates (Takasaki and Takenaka, 1991; Inagaki and Tsukahara, 1993; Takenaka, Takasaki, Kawamoto, Arakawa, and Takenaka, 1993; Morin and Woodruff, 1992; Morin ef al., 1994). Fecal samples may prove to have the greatest utility for obtaining genetic samples from endangered primates, but the problems of low genetic variability in small populations will require more sensitive analytical techniques than are currently available (Martin et al., 1992). As developments in techniques for DNA analyses become increas-
142
KAREN 0. STRIER
ingly applicable to genetic studies of primates, behavioral ecologists and conservation biologists will find greater opportunities to integrate individual and population approaches. SEX, A N D PRIMATE REPRODUCTIVE ECOLOGY B. HORMONES,
1. Background Insights into the relationships between sexual, behavioral, and reproductive ecology have been limited by the difficulties of obtaining accurate measures of female reproductive condition in wild primates. High levels of intraspecific variation in ovarian cycle length and gestation in most primates (Martin, 1992; see Section II,BJ), as well as the fact that most female primates copulate with many more partners than are necessary for conception (Hrdy, 1981), reduce the reliability of extrapolations from behavioral observations to the reproductive consequences of sex. For example, it has been difficult to distinguish whether female promiscuity and mate choices reflect strategies to confuse male paternity certainty and thus increase male investment in future offspring (Hrdy, 1981; Harcourt, 1981; Hamilton, 1984; Wallis and Englander-Golder, 1992), strategies to regulate tension and reduce the risks of male aggression (deWaal, 1987; Smuts and Smuts, 1993), or strategies to assess male quality and incite male-male competition (Clutton-Brock and Harvey, 1976). Reproductive seasonality and the relationship between diet, nutrition, and female reproductive condition have also been limited to inferences from behavioral data, such as when females mate, and observations of timing of births. Such data are inadequate to discriminate among a female’s failure to ovulate, conceive, or carry a fetus to term, or between general life history traits, such as adolescence or senescence, and individual behavioral strategies that affect fertility (Paul et al., 1993). These discriminations are especially important for assessing the viability of small populations of endangered primates, and for monitoring the success of reintroductions. Yet, until recently, knowledge of primate reproductive biology was based on laboratory studies, and the effects of ecological and social variables on reproduction in the wild have been poorly understood. 2. New Insights and Future Potential Efforts to understand primate reproduction in the wild must rely on noninvasive sampling techniques, such as urinary (Andelman, 1986) or fecal assays (Wasser, Risler, and Steiner, 1988; Wasser, Monfort, and Wildt, 1991; Strier and Ziegler, 1994; Brockman, Whitten, Russell, Richard, and Izard, 1995; Brockman and Whitten, 1996) because repetitive samples of female hormonal levels cannot be obtained through invasive capture tech-
PRIMATE BEHAVIORAL ECOLOGY A N D CONSERVATION
143
niques without severely disturbing the animals. Fecal samples are typically easier to collect and preserve in the wild than are urine samples, and offer opportunities to understand the relationships between behavior and reproduction among individuals within a group or population as well as between populations and species. The impact of differential fertility on the intrinsic rate of increase in wild populations can have important consequences for viability models in conservation biology as well as for evolutionary models in behavioral ecology. For example, understanding whether reproductive failure is due to a failure to ovulate, conceive, or support a pregnancy can be correlated with female behavior, diet, and social and reproductive history in assessments of population viability and individual life history strategies. Obtaining comparative data on reproductive parameters in captive and wild primates can lead to improvements in captive breeding protocols and to insights into the ways in which nutritional and energetic stress may affect reproductive potential. Although most noninvasive steroid studies on wild primates have concentrated on ovarian hormones, current efforts are extending the techniques to include monitoring testosterone in males and measuring cortisol as an indicator of stress. Differences in chronic and acute levels of stress can affect survival and reproduction in both wild and captive breeding populations, and can be correlated with social and ecological variables. Fecal steroid assays, like DNA analyses, provide immediate ways of linking individual-level analyses of behavior and reproduction with populationlevel analyses of demographic and genetic processes.
V. SUMMARY Convergence between primate behavioral ecology and conservation biology has increased in recent years as both fields have recognized their mutual reliance on one another and the mutual benefits of greater disciplinary exchanges. The limits of understanding primate behavioral diversity from traditional analyses of individual strategies and group dynamics have stimulated population-level approaches in behavioral ecology. Similar limitations on understanding population-level processes that affect population viability have stimulated interest in individuals and groups in conservation biology. Historical influences on primate behavioral ecology and conservation biology are responsible for their independent origins and evolution as distinct disciplines. Yet, as theoretical and methodological approaches converge, there may be little justification for their continuation as separate but overlapping fields. Considerations of primate mating systems, dispersal
144
KAREN B. STRIER
lPb 1
Long-term data from multiple populations of dlvers;specles
I Populatlon theories of
of DNA & hormones
I
1 COMPARATIVE POPULATION MODELS
1
FIG. 3. Integration of basic behavioral and ecological studies, population theories from conservation biology, and noninvasive methods for DNA and steroid studies for comparative population models.
systems, and life history variables illustrate the interactions between individual and population-level approaches. Individual behavioral and life history strategies respond to and are influenced by population size, composition, and density, whereas population processes reflect and affect individual strategies. Assumptions about the deterministic or stochastic properties of environmental variables on individuals and populations remain vague in both disciplines, and will also require additional long-term data from a diversity of populations to be reconciled. Noninvasive methods for sampling genetic and reproductive variation in wild primate populations provide access to data that address common theoretical and practical concerns. The highly specific and dynamic interactions between individual and population attributes require both behavioral ecologists and conservation biologists to incorporate data on intraspecific variation into their models, and to develop increasingly efficient noninvasive techniques to sample variation within and between populations. Declines in funding for basic research and the accelerating destruction of primate habitats make a unified agenda for primate behavioral' ecology and conservation biology an urgent priority (Fig. 3).
Acknowledgments This review benefited from support from NSF Grant BNS 8958298, and discussions with many people over many years. I am especially grateful to Manfred Milinski, Jim Moore,
PRIMATE BEHAVIORAL ECOLOGY A N D CONSERVATION
145
Martha Robbins, and Charles T. Snowdon for their valuable comments and suggestions on an earlier draft of this manuscript.
References Abbott, D. H. (1989). Social suppression of reproduction in primates. In “Comparative Socioecology: The Behavioural Ecology of Humans and Other Mammals” (V. Standen and R. A. Foley, eds.), pp. 285-304. Blackwell Scientific, Oxford. Aguirre, A. C. (1971). 0 mono Brachyteles arachnoides (E. Geoffroy): Situaqlo atual da espkcie no Brasil. Academia Brasileira de CiEncias, Rio de Janeiro. Akqakaya, H. R., and Burgman, M. (1995). Letters on PVA in theory and practice. Conservation Biol. 9, 705-706. Alberts, S. C., and Altmann, J. (1995). Balancing costs and opportunities: dispersal in male baboons. Am. Nat. 145,279-306. Allendorf, F. W., and Leary, R. F. (1986). Heterozygosity and fitness in natural populations of animals. In “Conservation Biology: The Science of Scarcity and Diversity” (M. E. Soult, ed.), pp. 57-76. Sinauer Associates, Sunderland, Massachusetts. Altmann, J. (1974). Observational study of behavior: Sampling methods. Behaviour 49, 227-267. Altmann. J. (1980). “Baboon Mothers and Infants.” Harvard University Press, Cambridge. Altmann, J.. Hausfater, G., and Altmann, S. A. (1985). Demography of Amboseli baboons. Am. J . Primatol. 8, 114-125. Altmann, S. A,, and Altmann, J. (1979). Demographic constraints on behavior and social organization. In “Primate Ecology and Human Origins” (I. S. Bernstein and E. 0.Smith, eds.), pp. 47-64. Garland Press, New York. Amos, B., Twiss, S., Pomeroy, P., and Anderson, S. (1995). Evidence for mate fidelity in the gray seal. Science 268,1897-1899. Andelman, S . J. (1986). Ecological and social determinants of cercopithecine mating patterns. In “Ecological Aspects of Social Evolution” (D. I. Rubenstein and R. W. Wrangham, eds.). pp. 201-216. Princeton University Press, Princeton. Anzenberger, G. (1992). Monogamous social systems and paternity in primates. In “Paternity in Primates: Genetic Tests and Theories” (R. D. Martin, A. F. Dixson, and E. J. Wickings, eds.), pp. 203-224. Karger, Basel. Appleton, C. C., Henzi, S. P., Whiten, A., and Byme, R. (1986). The gastrointestinal parasites of Papio ursinus from the Drakensberg Mountains, Republic of South Africa. Int. J. Primatol. I , 449-456. Asquith, P. J. (1989). Provision and the study of free-ranging primates: history, effects, and prospects. Yrbook Phys. Anthropol. 32, 129-158. Baker, A. J., Dietz, J. M., and Kleiman, D. G. (1993). Behavioural evidence for monopolization of paternity in multi-male groups of golden lion tamarins. Anim. Behav. 46, 1091-1103. Balmford. A. (1996). Extinction filters and current resilience: The significance of past selection pressures for conservation biology. Trends in Ecology and Evolution 11, 193-196. Bartlett, T. Q., Sussman, R. W., and Cheverud, J. M. (1993). Infant killing in primates: A review of observed cases with specific reference to the sexual selection hypothesis. Am. Anthropol. 95,958-990. Belovsky, G. E. (1987). Extinction models and mammalian persistence. In “Viable Populations for Conservation” (M. E. Soult, ed.), pp. 35-57. Cambridge University Press, Cambridge. Bowman, J. C., and Falconer, D. S. (1960). Inbreeding depression and heterosis of litter size in mice. Genet. Res., Cambridge 1, 262-274.
146
KAREN B. STRIER
Boyce, M. S. (1992). Population viability analysis. Ann. Rev. Ecol. Syst. 23,481-506. Brockman, D. K., and Whitten, P. L. (1996). Reproduction in free-ranging Propithecus verreauxi: Estrus and the relationship between multiple partner matings and fertilization. Am. J. Phys. Anthropol. 100, 57-69. Brockman, D. K., Whitten, P. L., Russell, E., Richard, A. R., and Izard, M. K. (1995). Application of fecal steroid techniques to the reproductive endocrinology of female Verreaux’s sifaka, Propithecus verreauxi. Am. J. Primatol. 36, 313-325. Brooks, T., and Balmford, A. (1996). Atlantic forest extinctions. Nature (London) 380, 115. Caro, T. M., and Durant, S. M. (1995). The importance of behavioral ecology for conservation biology: examples from Serengeti carnivores. In “Serengeti 11: Dynamics, Management, and Conservation of an Ecosystem” (A. R. E. Sinclair and P. Arcese, eds.), pp. 451-472. University of Chicago Press, Chicago. Caro, T., and Laurenson, M. K. (1994). Ecological and genetic factors in conservation: a cautionary tale. Science 263, 485-486. Carvalho, Jr., O., Strier, K. B., and Ferrari, S. F. (1995). Tamanho e composiq8o de agrupamento de mono-carvoeiros Brachyteles arachnoides no Parque Estadual Carlos Botelho (PECB). Programma e Resumos (p. 106), VII Congress0 Brasileiro de Primatologia, Natal. Caughley, G. (1994). Directions in conservation biology. J. Anim. Ecol. 63,215-244. Chapman, C. A. (1988). Patch use and patch depletion by the spider and howling monkeys of Santa Rosa National Park, Costa Rica. Behaviour 105, 99-116. Chapman, C. A. (1996). Primate seed dispersal: coevolution and conservation implications. Evol. Anthropol. 4,74-82. Chapman, C. A., and Chapman, L. J. (1990). Reproductive biology of captive and free-ranging spider monkeys. Zoo Biol. 9, 1-9. Chapman, C. A., and Chapman, L. J. (1996). Mixed-species primate groups in the Kibale forest: Ecological constraints on association. Internarl. J. Primatol. 17, 31-50. Charnov, E. L., and Berrigan, D. (1993). Why do female primates have such long lifespans and so few babies? or life in the slow lane. Evol. Anthropol. 1, 191-194. Cheney, D. L.. and Seyfarth, R. M. (1983). Non-random dispersal in free-ranging vervet monkeys: Social and genetic consequences. Am. Nut. 122,392-412. Chepko-Sade, B. D., Shields, W. M., Berger, J., Halpin, 2. T., Jones, W. T., Rogers, L. L., Rood, J. P., and Smith, A. T. (1987). The effects of dispersal and social structure on effective population size. I n “Mammalian Dispersal Patterns: The Effects of Social Structure on Population Genetics” (B. D. Chepko-Sade and 2. T. Halpin, eds.), pp. 287-321. University of Chicago Press, Chicago. Clutton-Brock, T. H., and Harvey, P. H. (1976). Evolutionary rules and primate societies. In “Growing Points in Ethology” (P. P. G. Bateson and R. A. Hinde, eds.), pp. 195-237. Cambridge University Press, Cambridge. Clutton-Brock, T. H., and Harvey, P. H. (1977). Primate ecology and social organization. J. Zool. (London) 183, 1-39. Clutton-Brock, T. H., and Harvey, P. H. (1984). Comparative approaches to investigating adaptation. I n “Behavioural Ecology: An Evolutionary Approach” (J. R. Krebs and N. B. Davies, eds.), pp. 7-29. Blackwell Scientific Publications, Oxford. Coimbra-Filho, A. F., Pissinatti, A., and Rylands, A. B. (1993). Breeding muriquis Brachyteles arachnoides in captivity: the experience of the Rio de Janeiro Primate Centre (CPRJFEEMA). Dodo, Journal of the Wildlife Preservation Trusts 29,66-77. Collias, N., and Southwick, C. (1952). A field study of population density and social organization in howling monkeys. Proc. Am. Phil. SOC.96,143-156. Cords, M. (1987). Forest guenons and patas monkeys: male-male competition in one-male groups. I n “Primate Societies” (B. B. Smuts, D. L. Cheney. R. M. Seyfarth, R. W. Wrangham, and T. T. Struhsaker, eds.), pp. 98-111. University of Chicago Press, Chicago.
PRIMATE BEHAVIORAL ECOLOGY A N D CONSERVATION
147
Cowlishaw, G., and Dunbar, R. I. M. (1991). Dominance rank and mating success in male primates. Anim. Behav. 41, 1045-1056. Crook, J. H., and Gartlan, J. S. (1966). Evolution of primate societies. Nature (London) 210, 1200-1203. Davies. G. (1991). Seed-eating by red leaf monkeys (Presbytis rubicunda) in dipterocarp forest of northern Borneo. Internatl. J. Primatol. 12, 119-144. Davies, A. G., Bennett, E. L., and Waterman, P. G. (1988). Food selection by two south-east Asian colobine monkeys (Presbyris rubicunda and P. melalophos) in relation to plant chemistry. Biol. J. Linn. SOC.34, 34-56. DeRousseau, C. J. (1990). Life-history thinking in perspective. In “Primate Life History and Evolution” (C. J. DeRousseau, ed.), pp. 1-13. Wiley-Liss, Inc., New York. de Ruiter, J. R., Scheffrahn, W., Trommelen, G. J. J. M., Uitterlinden, A. G., Martin, R. D., and van Hooff, J. A. R. A. M. (1992). Male social rank and reproductive success in wild long-tailed macaques. In “Paternity in Primates: Genetic Tests and Theories” (R. D. Martin, A. F. Dixson, and E. J. Wickings, eds.), pp. 175-191. Karger, Basel. de Ruiter. J. R., van Hooff. J. A. R. A. M., and Scheffrahn, W. (1994). Social and genetic aspeces of paternity in wild long-tailed macaques (Macaca fasciculuris). Behaviour U9,203-224. DeVore, I. (1963). A comparison of the ecology and behavior of monkeys and apes. In “Classification and Human Evolution” (S. L. Washburn. ed.), pp. 301-319. Aldine Publishing Company, Chicago. de Waal, F. B. M. (1987). Tension regulation and nonreproductive functions of sex in captive bonobos (Pan paniscris). Natl. Geogr. Res. 3,318-335. Di Fiore, A,. and Rendall, D. (1904). Evolution of social organization: A reappraisal for primates by using phylogenetic methods. Proc. Natl. Acad. Sci. U.S.A. 91,9941-9945. Digby, L. J., and Ferrari. S. F. (1994). Multiple breeding females in free-ranging groups of Callithrix jaccus. Internatl. J. Primatol. IS, 389-397. Dittus. W. (1979). The evolution of behaviors regulating density and age-specific sex ratios in a primate population. Behaviour 69,265-301. Dixson. A. F. (1983). Observations on the evolution and behavioral significance of “sexual skin” in female primates. Adv. Study Beh. 13, 63-106. Dixson. A. F., Anzenberger, G., Monteiro da Cruz, M. A. 0.. Patel, I., and Jeffreys, A. J. (1992). In “Paternity in Primates: Genetic Tests and Theories” (R. D. Martin, A. F. Dixson. and E. J. Wickings, eds.), pp. 192-202. Karger, Basel. Dobson, A. P., and Lyles, A. M. (1989). The population dynamics and conservation of primate populations. Conserv. Bi d. 3, 362-380. Dunbar, R. I. M. (1979). Population demography, social organization, and mating strategies. In “Primate Ecology and Human Origins” (I. S . Bernstein and E. 0. Smith, eds.), pp. 67-88. Garland Press, New York. Dunbar. R. I. M. (1987). Habitat quality, population dynamics, and group composition in colobus monkeys (Colobus guereza). Internatl. J. Primatol. 8, 299-321. Durant, S. M., and Mace, G. M. (1994). Species differences and population structures in population viability analyses. In “Creative Conservation: Interactive Management of Wild and Captive Animals” (P. J. S. Olney, G. M. Mace, and A. T. C. Feistner, eds.), pp. 67-91. Chapman and Hall, London. Eisenberg, J. F., Muckenhirn, N. A., and Rudran, R. (1972). The relation between ecology and social structure in primates. Science 176, 863-874. Eudey, A. A. (1980). Pleistocene glacial phenomena and the evolution of Asian macaques. I n “The Macaques: Studies in Ecology, Behavior and Evolution” (D. G. Lindburg, ed.), pp. 52-83. Van Nostrand Reinhold, New York.
148
KAREN 0. STRIER
Ewens, W. J., Brockwell, P. J., Gani, J. M., and Resnick. S. I. (1987). Minimum viable population size in the presence of catastrophes. In “Viable Populations for Conservation” (M. E. SoulC, ed.), pp. 59-68. Cambridge University Press, New York. Fedigan, L. M. (1982). “Primate Paradigms: Sex Roles and Social Bonds.” Eden Press, Montreal. Fedigan, L. (1983). Dominance and reproductive success in primates. Yrbook Phys. Anthropol. 26,91-129. Fedigan, L. (1993). Sex differences and intersexual relations in adult white-faced capuchins (Cebus capucinus). Infernatl. J . Primatol. 14, 853-878. Fedigan, L.M., and Rose, L. M. (1995). Interbirth interval variation in three sympatric species of neotropical monkey. Am. J . Primatol. 37, 9-24. Ferrari, S. F., and Diego, V. H. (1992). Long-term changes in a wild marmoset group. Folia Primarol. 58,215-218. Fimbel, C. (1994). Ecological correlates of species success in modified habitats may be disturbance- and site-specific: the primates of Tiwai Island. Conserv. Biol. 8, 106-113. Foley, R. A., and Lee, P. C. (1991). Ecology and energetics of encephalization in hominid evolution. Phil. Trans. R. S. Lond. B 334, 223-232. Fooden, J. (1980). Classification and distribution of living macaques (Macaca LacCpede. 1799). In “The Macaques: Studies in Ecology, Behavior and Evolution” (D. G. Lindburg, ed.), pp. 1-9. Van Nostrand Reinhold, New York. Fankham, R. (1995). Inbreeding and extinction: a threshold effect. Conserv. Biol. 9,792-799. Franklin, 1. R. (1980). Evolutionary change in small populations. In “Conservation Biology, an Evolutionary-Ecological Perspective” (M. E. Soul6 and B. A. Wilcox, eds.), pp. 135-149). Sinauer Associates, Sunderland, Massachusetts. Freeland, W. J. (1976). Pathogens and the evolution of primate sociality. Biofropica 8,12-24. Freese, C., and Oppenheimer, J. R. (1981). The capuchin monkeys, genus Cebus. In “Ecology and Behavior of Neotropical Primates” (A. F. Coimbra-Filho and R. A. Mittermeier, eds.), pp. 331-390. Academia Brasileira de CiCncias, Rio de Janeiro. Gabriel, W., Biirger, R., and Lynch, M. (1991). Population extinction by mutational load and demographic stochasticity. In “Species Conservation: A Population-Biological Approach” (A. Seitz and V. Loeschke, eds.), pp. 49-59. Birkhaiiser Verlage, Basil, Switzerland. Ganzhorn, J. U. (1995). Low-level forest disturbance effects on primary production, leaf chemistry, and lemur populations. Ecology 76, 2084-2096. Garber, P. A. (1987). Foraging strategies among living primates. Ann. Rev. Anthropol. 16, 339-364. Garber, P. A. (1994). Phylogenetic approach to the study of tamarin and marmoset social systems. Am. J. Primatol. 34, 199-219. Garland, T. Jr., and Adolph, S. C. (1994). Why not to do two-species comparative studies: limitations on inferring adaptation. Phys. Zool. 67, 797-828. Gilpin, M. E. (1987). Metapopulation structure and population vulnerability. In “Viable Populations for Conservation” (M. E. SoulC, ed.), pp. 125-139. Cambridge University Press, Cambridge. Gilpin, M. E., and SoulC, M. E. (1986). Minimum viable populations: processes of species extinction. In “Conservation Biology: The Science of Scarcity and Diversity” (M. E. Soule, ed.), pp. 19-34. Sinauer Associates, Sunderland, Massachusetts. Ginzburg, L. R., Ferson, S., and Akcakaya, H. R. (1990). Reconstructibility of density dependence and the conservative assessment of extinction risks. Conserv. Biol. 4, 63-70. Gittleman, J. L., and Hang-Kwang, L. (1992). On comparing comparative methods. Ann. Rev. EcoI. Syst. 23, 383-404.
PRIMATE BEHAVIORAL ECOLOGY A N D CONSERVATION
149
Glander, K. (1978). Howling monkey feeding behavior and plant secondary compounds: a study of strategy. In “The Ecology of Arboreal Folivores” (G. G. Montgomery, ed.), pp. 561-574. Smithsonian Institution Press, Washington, D.C. Glander, K. E. (1980). Reproduction and population growth in free-ranging mantled howling monkeys. Am. J. Phys. Anfhropol. 53,25-36. Glander, K. E. (1982). The impact of plant secondary compounds on primate feeding behavior. Yrbook Phys. Anfhropol. 25, 1-19. Glander, K. E. (1992). Dispersal patterns in Costa Rican mantled howling monkeys. Infernall. J . Primatol. 13, 415-436. Gomendio. M. (1995). Maternal styles in Old World primates: their adaptive significance. I n “Motherhood in Human and Nonhuman Primates” (C. R. Pryce, R. D. Martin, and D. Skuse, eds.), pp. 59-68. Karger, Basel. Gomendio. M., Clutton-Brock, T. H., Albon. S. D.. Guiness, F. E., and Simpson, M. J. (1990). Mammalian sex ratios and variation in the costs of rearing sons and daughters. Nature (London) 343,260-263. Goodall, J. (1986). “The Chimpanzees of Gombe: Patterns of Behavior.” Harvard University Press, Cambridge. Gouzoules. S. (1984). Primate mating systems, kin associations, and cooperative behavior: evidence for kin recognition? Yrbook Phys. Anfhropol. 27,99-134. Gray, J. P. (1985). “Primate Sociobiology.” HRAF Press, New Haven. Hamilton, W. D. (1984). Significance of paternal investment by primates to the evolution of male-female associations. I n “Primate Paternalism” (D. M. Taub, ed.), pp. 309-335. Van Nostrand Reinhold, New York. Harcourt, A. H. (1981). Inter-male competition and the reproductive behavior of the great apes. In “Reproductive Biology of the Great Apes: Comparative Biomedical Perspectives” (C. E. Graham, ed.), pp. 301-318. Academic Press, New York. Harcourt, A. H. (1995). Population viability estimates: theory and practice for a wild gorilla population. Conserv. Biol. 9, 134-142. Harding, R. S. 0.. and Olson, D. (1986). Patterns of mating among male patas monkeys (Eryfhrocebus patas) in Kenya. Am. J. Primafol. 11,343-358. Harlow, H. F., and Zimmermann, R. R. (1959). Affectional responses in the infant monkey. Science 130,421 -432. Hart, B. L. (1990). Behavioral adaptation to pathogens and parasites: five strategies. Neurosci. Biobehav. Rev. 14,273-294. Harvey, P. H., and Purvis, A. (1991). Comparative methods for explaining adaptations. Nature (London) 351,619-624. Hausfater, G. (1975). “Dominance and Reproduction in Baboons (Papio cynocephalus).” Karger, Basel. Hausfater, G., and Meade, B. J. (1982). Alteration of sleeping groves by yellow baboons (Papio cynocephalus) as a strategy for parasite avoidance. Primates 23,287-297. Hladik, C. M. (1981). Diet and the evolution of feeding strategies among forest primates. In “Omnivorous Primates” (R. S. 0. Harding and G. Teleki, eds.), pp. 215-254. Columbia University Press, New York. Holekamp, K. E., and Smale, L. (1995). Rapid change in offspring sex ratios after clan fission in the spotted hyena. Am. Naf. 145,261-278. Hrdy, S . B. (1977). “The Langurs of Abu.” Harvard University Press, Cambridge. Hrdy, S. B. (1979). Infanticide among animals: a review, classification, and examination of the implications for reproductive strategies. Erhol. Sociobiol. 1, 13-40. Hrdy, S. B. (1981). “The Woman That Never Evolved.” Harvard University Press, Cambridge.
150
KAREN B. STRIER
Inagaki, H., and Tsukahara, T. (1993). A method of identifying chimpanzee hairs in lion feces. Primates 34, 109-112. Inoue, M., Mitsunaga, R., Ohsawa, H., Takenaka, A., Sugiyama. Y., Gaspard Soumah, A,, and Takenaka, 0.(1992). Paternity testing in captive Japanese macaques (Macacafuscata) using DNA fingerprinting. In “Paternity in Primates: Genetic Tests and Theories” (R. D. Martin, A. F. Dixson, and E. J. Wickings, eds.), pp. 131-140. Karger, Basel. IUCN. (1994). “IUCN Red List Categories.” IUCN Species Survival Commission, Gland, Switzerland. Izawa, K. (1994). Group division of wild black-capped capuchins. Field Studies ofNew World Monkeys, La Macarena Colombia 9, 5-14. Johns, A. D., and Skorupa, J. P. (1987). Responses of rain-forest primates to habitat disturbance: a review. Internatl. J. Primatol. 8, 157-191. Jolly, C. J. (1993). Species, subspecies, and baboon systematics. In “Species, Species Concepts, and Primate Evolution” (W. H. Kimbel and L. B. Martin, eds.), pp. 67-107. Plenum Press, New York. Jones, C. B. (1980). The functions of status in the mantled howler monkey, Alouatta palliata Gray: Intraspecific competition for group membership in a folivorous neotropical primate. Primates 21, 389-405. Kay, R. F. (1984). On the use of anatomical features to infer foraging behavior in extinct primates. In “Adaptations for Foraging in Nonhuman Primates” (P. S. Rodman and J. G. H. Cant, eds.). pp. 21-53. Columbia University Press, New York. Keller, L. F., Arcese, P., Smith, J. N. M.. Hochachka, W. M., and Steams, S. C. (1994). Selection against inbred song sparrows during a natural population bottleneck. Nature (London) 372,356-357. Kimbel, W. H., and Martin, L. B. (Eds.) (1993). “Species, Species Concepts, and Primate Evolution.” Plenum Press, New York. Kinnaird, M. F., and O’Brien, T. G. (1991). Viable populations for an endangered forest primate, the Tana River crested mangabey (Cercocebus galeritus galeritus). Conserv. B i d . 5,203-214. Kinzey, W. G. (1982). Distribution of primates and forest refuges. I n “Biological Diversification in the Tropics” (G. T. Prance, ed.), pp. 455-482. New York: Columbia University Press. Kinzey, W. G., and Norconk, M. A. (1993). Physical and chemical properties of fruit and seeds eaten by Pithecia and Chiropotes in Surinam and Venezuela. Internatl. J. Primatol. 14,207-227. Koshimizu, K., Ohigashi, H., Huffman, M. A,, Nishida, T., and Takasaki, H. (1993). Physiological activities and the active constituents of potentially medicinal plants used by wild chimpanzees of the Mahale Mountains, Tanzania. Internatl. J. Primatol. 14,345-356. Krebs, J. R., and Davies, N. B. (1991). Introduction to part 3: Sexual selection and reproductive strategies. In “Behavioural Ecology” (J. R. Krebs and N. B. Davies. eds.). pp. 199-201. Blackwell Scientific Publications, Oxford. Kuester, J., Paul, A,, and Arnemann, J. (1992). Paternity determination by oligonucleotide DNA fingerprinting in Barbary macaques (Macaca sylvanus). In “Paternity in Primates: Genetic Tests and Theories” (R. D. Martin, A. F. Dixson, and E. J. Wickings, eds.). pp. 141-154. Karger, Basel. Kuester, J., Paul, A., and Arnemann, J. (1995). Age-related and individual differences of reproductive success in male and female Barbary macaques (Macaca sylvanus). Primates 36,461-476. Kurland, J. A. (1977). “Kin Selection in the Japanese Monkey.” Karger, Basel. Lacy, R. C. (1993). VORTEX: A computer simulation model for population viability analysis. Wildl. Res. 20.45-65.
PRIMATE BEHAVIORAL ECOLOGY A N D CONSERVATION
151
Lande. R. (1988). Genetics and demography in biological conservation. Science 241,1455-1460. Lande, R. (1995). Mutation and conservation. Conserv. Biol. 9, 782-791. Lande, R., and Barrowclough, G. F. (1987). Effective population size, genetic variation, and their use in population management. In “Viable Populations for Conservation” (M. E. Soule, ed.), pp. 87-123. Cambridge University Press, Cambridge. Laurenson. M. K. (1994). High juvenile mortality in cheetahs (Acinonyx jubatus) and its consequences for maternal care. J. Zool. (London) 234,387-408. Leigh, S. R., and Jungers, W. L. (1994). A re-evaluation of subspecific variation and canine I Phys. . Anthropol. dimorphism in woolly spider monkeys (Brachyieles arachnoides). Am. . 95,435-442. Leighton, M. (1993). Modeling dietary selectivity by Bornean orangutans: Evidence for integration of multiple criteria in fruit selection. Internail. J. Primatol. 14, 257-313. Leighton, M., and Leighton, D. R. (1982). The relationship of size of feeding aggregate to size of food patch: Howler monkeys (Alouatta palfiata) feeding in Trichilia cipo fruit trees on Barro Colorado Island. Biotropica 14, 81-90. Lemos de Sa, R. M. (1991). PopulaqZo de Brachyieles arachnoides (Primates, Cebidae) da Fazenda Esmerald, Rio Casca, Minas Gerais. In “A Primatologia No Brasil-3” (A. B. Rylands and A. T. Bernardes, eds.), pp. 235-238. Fundaclo Biodiversitas, Belo Horizonte, Brasil. Lemos de Sa, R. M., and Glandcr, K. E. (1993). Capture techniques and morphometrics for the woolly spider monkey, or muriqui (Brachyieles arachnoides, E. Geoffroy 1806). Am. J . Primatol. 29, 145-153. Lemos de Sa, R. M., Pope, T. R., Glander, K. E., Struhsaker, T. T., and da Fonseca, G. A. B. (1990).A pilot study of genetic and morphological variation in the muriqui (Brachyteles arachnoides). Primate Conserv. 11, 26-30. Lenski, R. E., and Sniegowski, P. D. (1995). “Adaptive mutation”: The debate goes on. Science 269,285-286. Levins, R. (1966). The strategy of model building in population biology. Am. Sci. 54,421-431, Lewontin, R. D. (1978). Fitness, survival and optimality. In “Analysis of Ecological Systems” (D. H. Horn, R. Mitchell, and G. R. Stairs, eds.), pp. 3-21. Ohio State University Press, Columbus. Lindburg, D. G., Durrant, B. S.. Millard, S. E., and Oosterhuis, J. E. (1993). Fertility assessment of cheetah males with poor quality semen. Zoo Biol. l2,97-103. Loy, J. D., and Peters, C. B. (1991). Mortifying reflections: primatology and the human disciplines. In “Understanding Behavior: What Primate Studies Tell Us about Human Behavior” ( J . D. Loy and C. B. Peters,eds.), pp. 3-16. Oxford University Press.New York. Lynch, M. (1996). A quantitative-genetic perspective on conservation issues. In “Conservation Genetics” ( J . C. Avise and J. L. Hamrick, eds.), pp. 471-501. Chapman and Hall, New York. Mace, G. M.. and Lande. R. (1991). Assessing extinction threats: Toward a re-evaluation of IUCN threatened species categories. Conserv. Biol. 5, 148-157. Martin, R. D. (1981). Relative brain size and basal metabolic rate in terrestrial vertebrates. Nature (London) 293,57-60. Martin, R. D. (1992). Female cycles in relation to paternity in primate societies. In “Paternity in Primates: Genetic Tests and Theories” (R. D. Martin, A. F. Dixson, and E. J. Wickings, eds.), pp. 238-274. Karger, Basel. Martin, R. D., Dixson, A. F., and Wickings, E. J. (Eds.). (1992). “Paternity in Primates: Genetic Tests and Theories.” Karger, Basel. May, R. M., and Seger, J. (1986). Ideas in ecology. Am. Sci. 74,256-267.
152
KAREN B. STRIER
Mayr, E. (1942). “Systematics and the Origin of Species.” Columbia University Press. New York. Mayr, E. (1966). Introduction. In C. Darwin, “On the Origin of Species” (pp. vii-xxvii). Harvard University Press, Cambridge. McGrew, W. C., Tutin, C. E. G., Collins, D. A., and Files, S. K. (1989). Intestinal parasites of sympatric Pan troglodytes and Papio spp. at two sites: Gombe (Tanzania) and Mt. Asserik (Senegal). Am. J. Primatol. 17, 147-155. Mehlman, P. (1986). Male intergroup mobility in a wild population of the Barbary macaque (Macaca sylvanus), Ghomaran Rif Mountains, Morocco. Am. J. Primatol. 10,67-81. Mehlman, P. T. (1989). Comparative density, demography, and ranging behavior of Barbary macaques (Macaca sylvanus) in marginal and prime conifer habitats. lnternatl. J . Primatol. 10,269-292. Melnick, D. J., and Hoelzer, G. A. (1992). Differences in male and female macaque dispersal lead to contrasting distributions of nuclear and mitochondria1 DNA variation. Internail. J. Primatol. 13, 379-393. Melnick, D. J., and Kidd, K. K. (1983). The genetic consequences of social group fission in a wild population of rhesus monkeys (Macaca mulatta). Behav. E d . Sociobiol. 12,229-236. MCnard, N., Scheffrahn, W., Vallet, D., Zidane, C., and Reber, Ch. (1992). Application of blood protein electrophoresis and DNA fingerprinting to the analysis of paternity and social characteristics of wild Barbary macaques. In “Paternity in Primates: Genetic Tests and Theories” (R. D. Martin, A. F. Dixson, and E. J. Wickings. eds.), pp. 155-174. Karger, Basel. Merola, M. (1994). A reassessment of homozygosity and the case for inbreeding depression in the cheetah, Acinonyx jubarus: Implications for conservation. Conserv. Biol. 8,961-971. Mills, L. S., and Smouse, P. E. (1994). Demographic consequences of inbreeding in small remnant populations. Am. Nut. 144, 412-431. Milton, K. (1984). Habitat, diet, and activity patterns of free-ranging woolly spider monkeys (Brachyteles arachnoides, E. Geoffroy, 1806). lnternatl. J. Primatol. 5, 491-514. Milton, K. (1985). Mating patterns of woolly spider monkeys, Brachyteles arachnoides: implications for female choice. Behav. Ecol. Sociubiol. 17, 53-59. Milton, K., and May, M. L. (1976). Body weight, diet, and home range area in primates. Nature (London) 259,459-462. Moore, J. (1984). Female transfer in primates. Internatl. J. Primatol. 5, 537-589. Moore, J. (1992). Dispersal, nepotism, and primate social behavior. lnternatl. J . Prirnarol. 13,361-378. Moore, J., and Ali, R. (1984). Are dispersal and inbreeding avoidance related? Anim. Behav. 32,94-112. Mori, A. (1979). Analysis of population changes by measurement of body weight in the Koshima troop of Japanese monkeys. Primates 20,371-397. Morin, P. A. (1993). Reproductive strategies in chimpanzees. Yrbook Phys. Anrhropol. 36, 179-212. Morin, P. A,, Moore, J. J., Chakraborty, R., Jin, L., Goodall, J., and Woodruff, D. S. (1994). Kin selection, social structure. gene flow, and the evolution of chimpanzees. Science 265,1193-1201. Morin, P. A., and Woodruff, D. S. (1992). Paternity exclusion using multiple hypervariable microsatellite loci amplified from nuclear DNA of hair cells. In “Paternity in Primates: Genetic Tests and Theories” (R. D. Martin, A. F. Dixson, and E. J. Wickings, eds.), pp. 63-81. Karger, Basel. Mowry, C. B.. Decker, 8 . S., and Shure, D. J. (1996). The role of phytochemistry in dietary choices of Tana River red colobus monkeys (Procolobus badius rufornitrarus). Internatl. J. Primatol. 17, 63-84.
PRIMATE BEHAVIORAL ECOLOGY A N D CONSERVATION
153
Muruthi, P., Altmann, J., and Altmann, S. (1991). Resource base, parity, and reproductive condition affect females’ feeding time and nutrient intake within and between groups of a baboon population. Oecologia 87,467-472. Newton, P. N. (1988). The variable social organization of hanuman langurs (Presbytis enfellus), infanticide, and the monopolization of females. Internafl. J. Primatol. 9, 59-77. Nishida, T., Takasaki, H., and Takahata, Y. (1990). Demography and reproductive profiles. In “The Chimpanzees of the Mahale Mountains: Sexual and Life History Strategies” (T. Nishida, ed.), pp. 63-97. University of Tokyo Press, Tokyo. Nishimura. A. (1990). A sociological and behavioral study of woolly monkeys, Lagothrix lagotricha, in the upper Amazon. The Science and Engineering Review of Doshisha University 31, 87-121. Nishimura, A. (1994). Social interaction patterns of woolly monkeys (Lagofhrix lagotricha): a comparisons among the atelins. The Science and Engineering Review of Doshisha University 35, 91-254. Nozawa, K., Shotake, T., Minezawa, M., Kawamoto, Y.,Hayasaka, K., Kawamoto, S., and Ito, S. (1991). Population genetics of Japanese monkeys: 111. Ancestry and differentiation of local populations. Primates 32, 411-435. O’Brien, S. J. (1994). The cheetah’s conservation controversy. Conserv. Biol. 8,1153-1155. O’Brien, S . J., and Mayr, E. (1991a). Bureaucratic mischief: Recognizing endangered species and subspecies. Science 251, 1187-1188. O’Brien, S. J., and Mayr, E. (1991b). Response to “Species Hybridization and Protection of Endangered Animals.” Science 253, 251-252. O’Brien, T. G. (1991). Female-male social interactions in wedge-capped capuchin monkeys: Benefits and costs of group living. Anim. Behav. 41,555-568. Oftedal, 0. T. (1992). The nutritional consequences of foraging in primates: The relationship of nutrient intakes to nutrient requirements. In “Foraging Strategies and Natural Diet of Monkeys, Apes, and Humans” (A. Whiten and E. M. Widdowson, eds.), pp. 1-10, Oxford University Press, Oxford. Packer, C. (1985). Dispersal and inbreeding avoidance. Anim. Behav. 33,676-678. Packer, C., Collins, D. A,, Sindlimwo, A., and Goodall, J. (1995). Reproductive constraints on aggressive competition in female baboons. Nature (London) 373,60-63. Partridge, L.. and Harvey, P. H. (1988). The ecological context of life history evolution. Science 241, 149-1455, Paul, A,, and Kuester, J. (1990). Adaptive significance of sex ratio adjustment in semifreeranging Barbary macaques (Macaca sylvanus) at Salem. Behav. Ecol. Sociobiol. 27, 287-293. Paul, A,, Kuester, J., Timme. A.. and Arnemann, J. (1993). The association between rank, mating effort, and reproductive success in male Barbary macaques (Macaca sylvanus). Primates 34,491-502. Peters, C. B. (1991). Apes, humans, and culture: what primatological discourse tells us about ourselves. In “Understanding Behavior: What Primate Studies Tell Us about Human Behavior” ( J . D. Loy and C. B. Peters, eds.), pp. 242-255. Oxford University Press, New York. Peters, C. R. (1993). Shell strength and primate seed predation of nontoxic species in eastern and southern Africa. Internatl. J. Primatol. 14,315-344. Phillips-Conroy, J. E., Jolly, C. J., Nystrom, P., and Memmalin, H. A. (1992). Migration of male hamadryas baboons into anubis groups in the Awash National Park, Ethiopia. Internatl. J . Primatol. 13, 455-476. Pinto, L. P. S., Costa, C. M. R., Strier, K. B.. and da Fonseca, G. A. B. (1993). Habitats, density, and group size of primates in the Reserva Biologica August0 Ruschi (Nova Lombardia), Santa Teresa, Brazil. Folia Primatol. 61, 135-143.
154
KAREN B. STRIER
Pope, T. (1990). The reproductive consequences of male cooperation in the red howler monkey: Paternity exclusion in multi-male and single-male troops using genetic markers. Eehav. Ecol. Sociobiol. 27, 439-446. Pope, T. R. (1996). Socioecology, population fragmentation, and patterns of genetic loss in endangered primates. In “Conservation Genetics” ( J . A. Avise and J. L. Hamrick, eds.), pp. 119-159. Chapman and Hall, New York. Popp, J. L. (1983). Ecological determinism in the life histories of baboons. Primates 24,198-210. Pusey, A. E. (1987). Sex-biased dispersal and inbreeding avoidance in birds and mammals. Trends Ecol. Evol. 2,295-299. Pusey, A. E., and Packer, C. (1987). Dispersal and philopatry. In “Primate Societies” (B. B. Smuts, D. L. Cheney, R. M. Seyfarth, R. W. Wrangham, and T. T. Struhsaker, eds.), pp. 250-266. University of Chicago Press, Chicago. Ralls, K., Ballou, J. D., and Templeton, A. (1988). Estimates of lethal equivalents and the cost of inbreeding in mammals. Conserv. Eiol. 2, 185-193. Ralls, K., Harvey, P. H., and Lyles, A. M. (1986). Inbreeding in natural populations of birds and mammals. I n “Conservation Biology: The Science of Scarcity and Diversity” (M. E. SoulC, ed.), pp. 35-56. Sinauer Associates, Sunderland, Massachusetts. Reed, J. M., and Dobson, A. P. (1993). Behavioral constraints and conservation biology: Conspecific attraction and recruitment. Trends Ecol. Evol. 8, 253-256. Reichard, U. (1995). Extra-pair copulations in a monogamous gibbon (Hylobates lar). Ethology 100,99-112. Rendall, D., and Di Fiorc, A. (1996). The road less traveled: Phylogenetic perspectives in primatology. Evol. Anfhropol. 4,43-52. Richard, A. F. (1981). Changing assumptions in primate ecology. Am. Anthropol. 83,517-533. Richard, A. F. (1985). “Primates in Nature.” W.H. Freeman, New York. Richard, A. F., and Dewar, R. E. (1991). Lemur ecology. Ann. Rev. Ecol. Syst. 22, 145-175. Richard, A. F., Goldstein, S. J., and Dewar, R. E. (1989). Weed macaques: The evolutionary implications of macaque feeding ecology. Internntl. J. Primafol. 10, 569-594. Richard, A. F., Rakotomanga, P., and Schwarz, M. (1991). Demography of Propithecus verreauxi at Beza Mahafaly, Madagascar: Sex ratio, survival, and fertility, 1984-1988. Am. J. Phys. Anthropol. 84,307-322. Rimoli, A. 0. (1992). 0 filhote muriqui (Erachyteles arachnoides): Urn estudo do desenvolvimenfo da independ2ncia. Unpublished master’s thesis, Universidade de SBo Paulo, S2o Paulo. Robbins, M. M. (1995). A demographic analysis of male life history and social structure of mountain gorillas. Eehaviour 132, 21-47. Robinson, J. G., and Janson, C. H. (1987). Capuchins, squirrel monkeys, and atelines: Socioecological convergence. In “Primate Societies” (B. B. Smuts, D. L. Cheney, R. M. Seyfarth, R. W. Wrangham, and T. T. Struhsaker, eds.), pp. 69-82. University of Chicago Press, Chicago. Rodseth, L., Wrangham, R. W., Harrigan, A. M., and Smuts, B. B. (1991). The human community as a primate society. Current Anthropol. 32, 221-254. Rogers, J. (1992). Nuclear DNA polymorphisms in hominoids and cercopithecoids: Applications to paternity testing. In “Paternity in Primates: Genetic Tests and Theories” (R. D. Martin, A. F. Dixson, and E. J. Wickings, eds.), pp. 82-95. Karger, Basel. Rosenberger, A. L., and Strier, K. B. (1989). Adaptive radiation of the ateline primates. J. Hum. Evol. 18,717-750. Ross, C. (1988). The intrinsic rate of natural increase and reproductive effort in primates. J. Zool., London 214,199-219.
PRIMATE BEHAVIORAL ECOLOGY AND CONSERVATION
155
Ross, C. (1991). Life history patterns of New World monkeys. Internatl. J. Primatol. 12, 481-502. Rowell, T. E. (1979). How would we know if social organization were not adaptive? I n “Primate Ecology and Human Origins” (I. Bernstein and E. 0. Smith, eds.), pp. 1-22. Garland Press, New York. Rowell, T. E. (1993). Reification of social systems. Evol. Anthropol. 2, 135-137. Rowell, T. E., and Chism, J. (1986). Mating and residence patterns of male patas monkeys. Ethology 72,31-39. Rylands, A. B., Mittermeier, R. A., and Luna, E. R. (1995). A species list for the New World primates (Platyrrhini): Distribution by country, endemism, and conservation status according to the Mace-Land system. Neotrop. Primates 3 (suppl.), 113-160. Samuels, A., and Altmann, J. (1991). Baboons of the Amboseli Basin: Demographic stability and change. Internatl. J. Primatol. U,1-19. Savage, A., Giraldo, L. H., Soto, L. H., and Snowdon, C. T. (1996). Demography, group composition, and dispersal in wild cotton-top tamarin (Saguinus oedipus) groups. Am. J. Primatol. 38, 85-100. Schubert, G., and Masters, R. D. (Eds.). (1991). “Primate Politics.” Southern Illinois University Press, Carbondale, Illinois. Shaffer, M. L. (1987). Minimum viable populations: Coping with uncertainty. I n “Viable Populations for Conservation” (M. E. Soule, ed.), pp. 69-86. Cambridge University Press, Cambridge. Simberloff, D. (1988). The contribution of population and community biology to conservation science. Ann. Rev. Ecol. Syst. 19,473-511. Small, M. F. (1981). Body fat, rank, and nutritional status in a captive group of rhesus macaques. Internatl. J. Primatol. 2, 91-95. Small, M. F. (1989). Female choice in nonhuman primates. Yrbook Phys. Anthropol. 32, 103-127. Small, M. F. (1990). Alloparental behaviour in Barbary macaques, Macaca sylvanus. Anim. Behav. 39,297-306. Smith, D. G., Rolfs, B., and Lorenz, J. (1992). A comparison of the success of electrophoretic methods and DNA fingerprinting for paternity testing in captive groups of rhesus macaques. I n “Paternity in Primates: Genetic Tests and Theories” (R. D. Martin, A. F. Dixson, and E. J. Wickings, eds.), pp. 32-52. Karger, Basel. Smuts, B. B. (1985). “Sex and Friendship in Baboons.” Aldine, New York. Smuts, B. B., and Nicolson, N. (1989). Reproduction in wild female olive baboons. Am. J. Primatol. 19, 229-246. Smuts, B. B., and Smuts, R. W. (1993). Male aggression and sexual coercion of females in nonhuman primates and other mammals: Evidence and theoretical implications. Adv. Study Behav. 22, 1-63. Soule, M. E. (1987). Introduction. In “Viable Populations for Conservation” (M. E. Soult5, ed.), pp. 1-10. Cambridge University Press, Cambridge. SoulC, M. E., and Kohn, K. A. (Eds.). (1989). “Research Priorities for Conservation Biology.” Island Press, Washington, D.C. Southwick, C. H., and Siddiqi, M. F. (1988). Partial recovery and a new population estimate of rhesus monkey populations in India. Am. J. Primatol. 16, 187-197. Sprague, D. S. (1991). Mating by non-troop males among the Japanese macaques of Yakushima. Folia Primatol. 57, 156-158. Sprague, D. S. (1992). Life history and male inter-troop mobility among Japanese macaques (Macaca fuscata). Internatl. J. Primatol. 13,437454.
156
KAREN B. STRIER
Stallings, J. R., and Robinson, J. G. (1991). Disturbance, forest heterogeneity and primate communities in a Brazilian Atlantic forest park. I n “A Primatologia No Brasil-3” (A. B. Rylands and A. T. Bernardes, eds.), pp. 357-368. Fundapo Biodiversitas, Belo Horizonte, Brazil. Starin, E. D. (1994). Philopatry and affiliation among red colobus. Eehaviorcr 130, 253-270. Stephen, M. E. (1989). Tamarins: alpha female, queen bee or white rat. Internail. J. Primatol. 10, 115-121. Strier, K. B. (1987). Reproducao de Brachyteles arachnoides. In “A Primatologia No Brasil2” (M. Thiago de Mello, ed.), pp. 163-175. Sociedade Brasileira de Primatologia, Brasilia. Strier, K. B. (1989). Effects of patch size on feeding associations in muriquis (Erachyteles arachnoides). Folia Primatol. 52,70-77. Strier, K. B. (1990). New World primates, new frontiers: Insights from the woolly spider monkey, or muriqui (Brachyteles arachnoides). Internad J . Primatol. 11, 7-19. Strier, K. B. (1991). Demography and conservation in an endangered primate, Brachyteles arachnoides. Conserv. Eiol. 5, 214-218. Strier, K. B. (1992a). “Faces in the Forest: The Endangered Muriqui Monkeys of Brazil.” Oxford University Press, New York. Strier, K. B. (1992b). Causes and consequences of nonaggression in woolly spider monkeys. In “Aggression and Peacefulness in Humans and Other Primates” (J. Silverberg and J. P. Gray, eds.), pp. 100-116. Oxford University Press, New York. Strier, K. B. (1993). The animal boundaries of anthropology, or why don’t all anthropologists consider themselves primatologists? Rev. Anthropol. 22, 165-174. Strier, K. B. (1994a). Myth of the typical primate. Yrbook Phys. Anihropol. 37,233-271. Strier, K. B. (1994b). Brotherhoods among atelins. Behaviour 130, 151-167. Strier, K. B. (1996a). Male reproductive strategies in New World primates. Human Nature. 7, 105-123. Strier, K. B. (1996b). Viability analyses of an isolated population of muriqui monkeys (Erachyides arachnoides): Implications for primate conservation and demography. Primate Conserv. 14-15 (1993-1994). 43-52. Strier, K. B. (in press, a). Reproductive ecology of female muriquis. In “Adaptive Radiations of Neotropical Primates” (M. Norconk, A. Rosenberger, and P. Garber, eds.). Plenum Press, New York. Strier, K. B. (in press, b). Mate preferences of wild muriqui monkeys (Brachyteles arachnoides): reproductive and social correlates. Folia Primatol. Strier, K. B. (in press, c). Subtle cures of social relations in male muriqui monkeys (Erachyieles arachnoides).In “New World Primates: Evolution, Ecology and Behavior” (W. G. Kinzey, ed.). Adline de Gruyter, New York. Strier, K. B., Mendes, F. D. C., Rimoli, J., and Rimoli, A. 0. (1993). Demography and social structure in one group of muriquis (Erachyieles arachnoides). Internatl. J. Primatol. 14,513-526. Strier, K. B., and Ziegler, T. E. (1994). Insights into ovarian function in wild muriqui monkeys (Erachyteles arachnoides). Am. J. Primatol. 32, 31-40. Struhsaker, T. T. (1969). Correlates of ecology and social organization among African cercopithecines. Folia Primatol. 11, 80-118. Strum, S. C. (1982). Agonistic dominance in male baboons: An alternative view. Iniernatl. J. Primatol. 3, 175-202. Strum, S . C. (1983). Use of females by male olive baboons (Papio anubis). Am. J. Primaiol. 5, 93-109. Stuart, M. D., and Strier, K. B. (1995). Primates and parasites. Internail. J. Primatol. 16,577-593.
PRIMATE BEHAVIORAL ECOLOGY AND CONSERVATION
157
Stuart, M. D., Strier, K. B., and Pierberg, S. M. (1993). A coprological survey of parasites of wild woolly spider monkeys. Brachyteles arachnoides, and brown howling monkeys, Alouatta fitsca. J. Helmintholog. Soc. 60, 11 1-1 15. Sussman, R. W. (1977). Feeding behavior of Lemur catta and Lemur fulvus. In “Primate Ecology” (T. H. Clutton-Brock, ed.), pp. 1-36. Academic Press, New York. Sussman, R. W. (1992). Male life history and intergroup mobility among ringtailed lemurs (Lemur catta). Internatl. J . Primatol. 13, 395-413. Sussman. R. W., and Raven, P. H. (1978). Pollination by lemurs and marsupials: An archaic coevolutionary system. Science 200, 731-736. Symington, M. M. (1987). Sex ratio and maternal rank in wild spider monkeys: When daughters disperse. Beh. Ecol. Sociobiol. 20, 333-335. Symington, M. M. (1988). Food competition and foraging party size in the black spider monkey (Ateles paniscits chamek). Behaviour 105, 117-134. Symington, M. M. (1990). Fission-fusion social organization in Ateles and Pan. Internatl. J. Primatol. 11, 47-61. Takahata, Y., Koyama, N., Huffman, M. A., Norikoshi, K., and Suzuki, H. (1995). Are daughters more costly to produce for Japanese macaque mothers?: Sex of the offspring and subsequent interbirth intervals. Primates 36, 571-574. Takasaki, H., and Takenaka, 0.(1991). Paternity testing in chimpanzees with DNA amplification from hairs and buccal cells in wadges: A preliminary note. In “Primatology Today” (A. Ehara, T. Kimura, 0. Takenaka, and M. Iwamoto, eds.), pp. 613-616. Elsevier, Amsterdam. Takenaka, O., Takasaki, H., Kawamoto, S., Arakawa, M., and Takenaka, A. (1993). Polymorphic microsatellite DNA amplification customized for chimpanzee paternity testing. Primates 34, 27-35. Terborgh, J. (1985). “Five New World Primates.” Princeton University Press, Princeton. Terborgh, J., and Janson, C. H. (1986). The socioecology of primate groups. Ann. Rev. Ecol. S y ~ t 17, . 111-135. Thomas. C. D. (1990). What do real population dynamics tell us about minimally viable population sizes? Conserv. Eiol. 4, 324-327. Tooby, J., and DeVore, I. (1987). The reconstruction of hominid behavioral evolution through strategic modelling. In “The Evolution of Human Behavior: Primate Models” (W. G. Kinzey, ed.), pp. 183-237. SUNY Press, Albany. Trivers. R. L., and Willard, D. E. (1973). Natural selection of parental ability to vary the sex ratio of offspring. Science 191, 249-263. Turner, T. R., Weiss, M. L., and Pereira, M. E. (1992). DNA fingerprinting and paternity assessment in Old World monkeys and ringtailed lemurs. In “Paternity in Primates: Genetic Tests and Theories” (R. D. Martin, A. F. Dixson, and E. J. Wickings, eds.), pp. 96-112. Karger, Basel. Uchida, A. (1996). What we don’t know about great ape variation. Trends Ecol. Evol. 11, 163- 168. van Hooff, J. A. R. A. M., and van Schaik, C. P. (1992). Cooperation in competition: the ecology of primate bonds. In “Coalitions and Alliances in Humans and Other Animals” (A. H. Harcourt and F. B. M. de Waal, eds.), pp. 357-389. Oxford Scientific Publications, Oxford. van Schaik, C. P. (1989). The ecology of social relationships amongst primate females. In “Comparative Socioecology: The Behavioural Ecology of Humans and Other Mammals” (V. Standen and R. A. Foley, eds.), pp. 195-218. Blackwell Scientific, Oxford. van Schaik, C. P., Netto, W. J., van Amerongen, A. J. J., and Westland, H. (1989). Social rank and sex ratio of captive long-tailed macaque females (Macaca fascicularis). Am. J. Primatol. 19, 147-161.
158
KAREN 0. STRIER
van Schaik, C. P., and van Hooff, J. A. R. A. M. (1983). On the ultimate causes of primate social systems. Behaviour 85, 91-117. Vehrencamp, S. L. (1983). A model for the evolution of despotic versus egalitarian societies. Anim. Behav. 31,667-682. Wallis, J. (1995). Seasonal influence on reproduction in chimpanzees of Gombe National Park. Internatl. J. Primatol. 16,435-551. Wallis, J., and Englander-Colder, P. (1992). A comparative review of primate sexuality across the menstrual cycle. I n “Menstrual Health in Women’s Lives” (A. J. Dan and L. L. Lewis, eds.), pp. 111-130. University of Chicago Press, Chicago. Walsh, P. D. (1995). Letters on PVA in theory and practice. Conserv. Biol. 9, 704-705. Walters, J. R. (1991). Application of ecological principles to the management of endangered species: The case of the red-cockaded woodpecker. Ann. Rev. Ecol. Syst. 22,505-523. Wasser, S . K., Risler, L., and Steiner, R. A. (1988). Excreted steroids in primate feces over the menstrual cycle and pregnancy. Biol. Reprod. 39,862-872. Wasser, S. K., Monfort, S. L., and Wildt, D. E. (1991). Rapid extraction of faecal steroids for measuring reproductive cyclicity and early pregnancy in free-ranging yellow baboons (Papio cynocephalus cynocephalus). J. Reprod. Fertil. 92,415-423. Waterman, P. G., and Kool, K. M. (1994). Colobine food selection and plant chemistry. I n “Colobine Monkeys: Their Evolutionary Ecology” (A. G. Davies and J. F. Oates, eds.), pp. 251-284. Cambridge University Press. Cambridge. Weatherhead, P. J. (1986). How unusual are unusual events? Am. Nut. 128, 150-154. White, F. J., and Wrangham, R. W. (1988). Feeding competition and patch size in the chimpanzee species Pan paniscus and Pan troglodytes. Behaviour 105,48-164. Whitten, P. L. (1984). Competition among female vervet monkeys. I n “Female Primates: Studies by Women Primatologists” (M. F. Small, ed.), pp. 127-140. Alan R. Liss, New York. Widowski, T. M., Ziegler, T. E.. Elowson, A. M., and Snowdon, C. T. (1990). The role of males in the stimulation of reproductive function in female cotton-top tamarins. Saguinus o. oedipus. Anim. Behav. 40,731-741. Wrangham, R. W. (1980). An ecological model of female-bonded primate groups. Behaviour 75,262-299. Wrangham, R. W. (1981). Drinking competition in vervet monkeys. Anim. Behav. 29,904-910. Wrangham, R. W. (1982). Mutualism, kinship and social evolution. I n “Current Problems in Sociobiology” (Kings College Study Group, eds.), pp. 269-289. Cambridge University Press, Cambridge. Wrangham, R. W. (1987). The significance of African apes for reconstructing human social evolution. In “The Evolution of Human Behavior: Primate Models” (W. G. Kinzey, ed.), pp. 51-71. SUNY Press, Albany. Yeager, C. P. (1995). Does intraspecific variation in social systems explain reported differences in the social structure of the proboscis monkey (Nasalis larvatus)? Primates 36,575-582.
ADVANCES IN THE STUDY OF BEHAVIOR. VOL. 26
How To Avoid Seven Deadly Sins in the Study of Behavior MANFRED MILINSKI ABTEILUNG VERHALTENS~KOLOGIE
ZOOLOGISCHES INSTITUT UNIVERSITAT BERN
HINTERKAPPELEN, SWITZERLAND
Scientific journals have become both thicker and more numerous during the last twenty years. An enormous flood of papers has driven the traditional refereeing system to its limits and scientists to increasing specialization. For someone of my generation who originally analyzed his data with pencil and paper, drew his figures with ink, and wrote his first publications on a mechanical typewriter, producing papers in our software age has become extremely easy and many times faster. Two steps in the paper milling process must still consume much time and effort, namely designing and performing experiments. The consequence of speeding these steps up is a high rate of flaws and mistakes in papers, even those that are published in respected journals: obviously, referees cannot manage (or are not trained) to find all the procedural mistakes when they d o their altruistic job in their limited time. I should not be too pessimistic. I am happy to acknowledge that the application of statistical procedures has been improved on average in behavior research and that there are exceptional scientists who do high quality research despite a high publication rate. The following is a list of mistakes that I have often found in published research: 1. Unjustified conclusions are made from observational (i.e., correlational) data.
2. Data are not independent (“pseudoreplication”). 3. Treatments are confounded by time and sequence effects.
4. No efforts are made to avoid observer bias. 5. Potential artifacts arise when animals are not accustomed to experi-
mental procedures. 159
Copyright 0 I997 by Academic Press All rights of reproduction in any form reserved. m5-3454/97s2s.w
160
MANFRED MILINSKI
6. Unsuitable controls are used. 7. An attempt is made to “prove” the null hypothesis with small samples.
This list is not complete but I think future publications would be improved to a large extent if these mistakes were avoided. There are excellent reviews on experimental design (e.g., Hurlbert, 1984; Martin and Bateson, 1986, 1993) that also discuss these mistakes and give advice on how to avoid them. Why has this advice been ignored so often? Reading a very detailed review or book might be regarded as too time consuming (although performing a flawed experiment is a complete waste of time); the advice might be too complex to be easily remembered; the mistakes might be spotted but they are difficult a n d o r expensive to avoid (“making the best of a bad job”); the mistakes might be regarded as “accepted” because they are often made; “habituation” of both researchers and referees might be occurring. It is worthwhile to write another advice paper only if this advice is more likely to be accepted and used. To achieve this I concentrate on only a few, but important, mistakes. I avoid giving detailed information and present only a few theoretical or technical explanations. Instead, I concentrate on simple but telling examples and incorrect and correct conclusions. I hope to be successful with this procedure because empiricists such as myself can handle examples better than we can handle theory, and because our intuition is trained by understanding examples. I do not cite specific negative examples from the literature, not even from my own research, although I have to admit that I have made one or another of these mistakes myself. Instead, I invent simple cases or cite positive examples and justified conclusions from the literature, and particularly from my own work because I know its weaknesses best and can discuss them. One of my colleagues is convinced that there is no such thing as a perfect experiment: let us try to find a close approximation to it. I. UNJUSTIFIED CONCLUSIONS FROM OBSERVATIONAL DATA
To observe animals and to draw conclusions about the function and the mechanisms of the observed behavior has a long tradition, particularly in ethology. What is wrong with this approach? It would be acceptable if observed correlations were only described and no cause-effect relationship conclusion was drawn from them. Let us take three examples. A. CORRECT DESCRIPTION 1. Conspicuous individuals have a higher predation risk. 2. Smokers have a higher risk of lung cancer.
3. Ornamented males are more attractive to females.
HOW TO AVOID SEVEN DEADLY SINS IN THE STUDY OF BEHAVIOR
161
B. UNJUSTIFIED CONCLUSION
1. Conspicuous individuals have a higher predation risk because they are conspicuous. 2. Smokers have a higher risk of lung cancer because of smoking. 3. Ornamented males are more attractive to females because they are ornamented.
CONCLUSIONS UNJUSTIFIED? C. WHYARETHESE 1. Conspicuous individuals are, for example, larger (or have some other trait by which they differ from cryptic individuals); they may have a higher predation risk because they are larger and not because they are conspicuous.
2. People who smoke may do so because they have, for example, a gene that makes them like tobacco; the same gene may predispose them to lung cancer; they may develop lung cancer because they have that gene and not because they smoke. This hypothetical example demonstrates the weakness of epidemiological evidence. 3. Ornamented males may be stronger, and females may like these males because they are stronger and not because they are ornamented.
D. Do A N EXPERIMENT Example I : What can be done if one wants to determine whether an observed trait A (e.g., high predation risk) is caused by an observed trait B (e.g., being conspicuous)? One must do an experiment, in which one removes the possibility that a difference in predation risk between conspicuous and cryptic individuals is caused by any other (undetected) trait that conspicuous individuals usually possess (e.g., larger size). One determines randomly which of, for example, 48 experimental animals has to be made conspicuous and which cryptic. Procedure: Take two animals, toss a coin t o determine which one is to be made conspicuous. Take another two animals, toss a coin, and so on, until there are 24 animals that are to be made conspicuous and 24 that are to be made cryptic. Present a predator with a simultaneous choice between an individual that has been made conspicuous and an individual that has been made cryptic. Repeat this procedure with each pair separately (independently). If conspicuous individuals are preferred over cryptic individuals (sign test), one may conclude that the specific conspicuousness that has been tested increases an individual’s risk of predation. I should, however, emphasize that a statistical test provides only a nonzero probability, and
162
MANFRED MILINSKI
that one can be only reasonably confident in a result, but never certain (many of today’s computer programs round off p at the third decimal place, tempting users to uncorrectly report a p equal to zero). By this procedure, all other traits that may affect an individual’s predation risk (e.g., size) were assigned randomly to the experimental (conspicuous) and control (cryptic) groups. Randomization is therefore indispensable. We assume that the experiment has been performed properly in all other respects. There are, however, other possibilities for making mistakes that I will discuss later. I will use this example throughout this review. It turns out that this apparently simple experiment is extremely difficult to do properly.
E. AN EXPERIMENT MAYBE UNETHICAL Example 2: To determine experimentally whether smoking causes lung cancer one would have to assign young adult people randomly (indispensable condition!) to the group who has to smoke, say, for the next twenty years and the group who must refrain from smoking for the same period. If the smokers have developed lung cancer significantly more often than the nonsmokers, one may conclude that smoking causes lung cancer. Of course, this experiment cannot be done for ethical reasons. What can be done instead? Should one believe in the correlational (epidemiological) evidence because it is the best we have? I do not smoke because I “believe” that smoking causes lung cancer. However, nobody is forced to believe such a conclusion. Correlational evidence can never prove a cause-effect relationship. Other problems may lead an experiment to be considered either unethical, or too difficult or too expensive to do. All this is no excuse for accepting correlational evidence as proof of a cause-effect relationship.
F. ANOTHER KINDOF EXPERIMENT Example 3: One can do the same kind of experiment as proposed for Example 1 to test whether females prefer ornamented males because they are ornamented: males are selected randomly and supplied with either a small or a large ornament. This has been done successfully by elongating or shortening the tail feathers of widowbirds (Eupfectesprogne) by Anderson (1982) and of barn swallows (Hirundo rusrica) by Moller (1988). It may, however, be difficult to supply male sticklebacks (Gasterosteus acufeatus) with a red belly without producing artifacts. The solution would be to establish that females prefer males that are naturally redder. This is a staged observation (not an experiment) and provides only correlational evidence. The next step would be to prevent the females from seeing the natural differences in male red coloration by repeating the previous staged observa-
HOW TO AVOID SEVEN DEADLY SINS IN THE STUDY OF BEHAVIOR
163
tion with other females under filtered (green) light (Milinski and Bakker, 1990), where differences in red cannot be detected. If one finds a significant decrease of the earlier preference, one may conclude that females prefer males more when they are redder. This is experimental evidence. The problem is that one has to be lucky that females do not use other male traits that correlate with red color and that are still detectable under green light. We were lucky. G. THREE FURTHER EXAMPLES TO TRAIN ONE’SINTUITION
Example 4: If someone does not know the price of either a Rolls Royce or a Volkswagen Rabbit and looks instead at what is left in the bank accounts of people who just bought either a Rolls Royce or a Rabbit, the person would conclude that a Rabbit is much more expensive than a Rolls. The person is mistaken because rich people can buy expensive things and still have more remaining. The correlational evidence is misleading. The necessary experiment would be to force randomly chosen people to buy either a Rolls Royce or a Rabbit. I bet that only the new Rabbit owners would have on average a positive bank account after the purchase. Example 5: Male secondary sexual ornaments are regarded as handicaps for survival. Moller (1990a) observed that male barn swallows that had naturally longer tail feathers arrived earlier at their breeding sites (from their overwintering sites) than short-tailed males. Does this correlational evidence mean that elongated tails improve flying rather than handicap it? No, it needs an experiment for such a conclusion. Moller (1989) found that the males with experimentally elongated tail feathers (compared to males with experimentally shortened tail feathers) had viability costs, such as impaired foraging efficiency, which proved that this ornament was indeed a handicap. Why did males with naturally longer tails have an improved flying ability? As natural Rolls Royce buyers, male barn swallows that grew a longer tail had more reserves than others (Moller, 1989). Stronger males should invest an amount of reserves in larger ornaments so that they are, nevertheless, still stronger than weaker males that invested in the smaller ornament (Grafen, 1990). Example 6: Small passerine birds that have detected a predator, for example, an owl or a kestrel, approach it closely at which time they call and flick their wings. The function of this “mobbing” behavior may be the tradition of knowledge of predators passed on to offspring (Curio, Ernst, and Vieth, 1978) and/or making the predator “move on” to another site where it has not yet been detected (Pettifor, 1990). Is approaching the predator risky? It has almost never been observed that a mobbing bird is attacked by the predator that is being mobbed. However, it may well be
164
MANFRED MILINSKI
that the birds take their own fleeing ability into account when they decide on their mobbing distance. Curio (1983) found that birds of species that have a high maneuverability have a closer minimum mobbing distance than birds of species that can fly less well. Birds that approach a predator more closely may less often be taken by the predator than birds that approach the predator less closely (analogous to Rolls Royce buyers and naturally long-tailed male barn swallows). To test whether predation risk increases with shorter distance to the predator, one has to assign randomly chosen individuals to different approach distances and measure their probability of being caught. This has been done with dead sticklebacks that we made by remote control to “inspect” a predatory pike (Esox lucius) up to predetermined distances; analogous to experimentally elongated barn swallows, experimentally inspecting sticklebacks have increasing costs when they approach a pike more closely: they are more likely to be caught. H. Is CORRELATIONAL EVIDENCE OF ANYVALUE? Yes, correlational evidence is useful whenever it is important to know a relationship between traits. We expect both richer people and richer barn swallows to invest more in handicaps and remain still richer than poorer individuals with smaller handicaps. To establish this correlation provides some support for the ESS prediction. Knowing that predator approach distance correlates negatively with a mobbing bird’s fleeing ability helps to predict differential mortality in a given situation. We regard the correlation as a matter of fact but refrain from concluding any cause-effect relationship. Furthermore, correlational evidence may be enough if a visible trait can be used as an indicator of another trait that is invisible. If female barn swallows benefit from mating with males that are genetically resistant to mites, the females may find these males by selecting long-tailed ones, because the mite load of a male’s offspring correlates negatively with their father’s tail length (MZller, 1990b). Similarly, the intensity of male sticklebacks’ red breeding coloration correlates positively with their physical condition (Milinski and Bakker, 1990). Females can use the existence of this correlation for selecting males of superior condition. It is not easy to find more applications for using correlational evidence correctly. Another example might be that an interesting correlation that is found unexpectedly might help to generate hypotheses for a decisive experiment; for example, Wedekind, Seebeck, Bettens, and Paepke, (1995) found that women prefer the smell of men to whom they are dissimilar in their MHC-alleles, which may be adaptive because the offspring would resist a broad range of infectious diseases with their large number of different MHC-alleles; however, women who take the contraceptive pill prefer
HOW TO AVOID SEVEN DEADLY SINS IN THE STUDY OF BEHAVIOR
165
the smell of men with similar MHC genetics. Because this is only correlational evidence, people would like to know whether the pill actually causes this switch of preference. If so, there should be a warning on the package. It is sometimes suggested that if the direction (e.g., positive) of a correlation is predicted under the assumption of a specific cause-effect relationship, then this cause-effect relationship is supported when that correlation is found. This is mistaken. Imagine the following scenario: some hypothetical biologists are asked to investigate whether animals that live close to motorways are negatively affected by the traffic. If anything, one expects a reduction of condition near the motorway. The biologists measure the reproductive success of rabbits that live either close to a motorway or a hundred meters away from it. They find that rabbits that breed next to the motorway have indeed significantly reduced reproductive success. They erronously conclude that motorways cause a reduction of reproductive success in rabbits. Why are they mistaken? Imagine that rabbits do not prefer territories next to the motorway (perhaps because foxes prefer those places for some reason). This has the consequence that the stronger rabbits win the preferred places farther away from the motorway and the weaker rabbits must settle close to the traffic. Weaker rabbits may have a lower reproductive success anyway, and this continues when they settle close to the motorway. Our biologists would have been quite surprised if they had found that rabbits close to the motorway have an increased reproductive success (e.g., if foxes hated noise and therefore the stronger rabbits fight successfully for places near the traffic). The only way to solve the problem is to do a suitable experiment. 11. DATAARENOTINDEPENDENT:“PSEUDOREPLICATION”
The term pseudoreplication was proposed by Hurlbert (1984). He found that pseudoreplication occurred in almost 50% of publications in respected ecological journals. People have become aware of this problem in the meantime and Hurlbert’s paper has become a “citation classic” (see also Kroodsma, 1986, 1989). However, pseudoreplication is still found in many recent publications, perhaps because it is sometimes difficult to detect: “it can appear in different guises” (Hurlbert, 1984). Pseudoreplication occurs when replicates are not statistically independent. A. PSEUDOREPLICATION IN EXPERIMENTS
When the experiment I proposed for Example 1 is performed, there are several opportunities for pseudoreplication. When pairs of fish are taken
166
MANFRED MILINSKI
and a coin is tossed to decide which fish has to be made conspicuous and which cryptic, both sorts of fish must be housed until the predation experiment. We have two separate tanks, one for the conspicuous fish and one for the cryptic fish. Of course (hopefully) the tanks are the same size, have the same water level, light, equipment, and so on. Unfortunately, we are not aware that something happened to one of the tanks before the experiment (e.g., someone unwittingly hit the table with the tank that contained the fish that will be made cryptic). Because of this event all the cryptic fish have been frightened and will, therefore, be more cautious in the predation experiment and thus less frequently preyed upon than the conspicuous fish. We conclude erronously that a prey’s conspicuousness increases its risk of predation. This kind of pseudoreplication can be avoided by maintaining each fish in a separate tank. Of course tanks of conspicuous fish have to be interspersed with tanks of cryptic fish, otherwise hitting one table would affect one group more than the other. One wants to investigate whether conspicuousness increases a prey’s risk of predation. Another opportunity for pseudoreplication: 24 pairs each consisting of a conspicuous and a cryptic fish and 6 pike are available. Each pike is tested with four pairs of fish, one pair per day on four consecutive days. How large is n , 24 or 6? Whether conspicuousness increases a prey’s risk of predation depends on the preference of the predator. Therefore, actually the behaviour of the pike is being investigated, and each pike has its individuality. The four choices of each pike are not independent because they are made by the same pike. To avoid pseudoreplication, each pike has to be treated as a statistical unit and the choices of each pike must be entered as one data point in the analysis (e.g., percentage of conspicuous fish chosen). Thus, n is only 6; an n of 24 would be pseudoreplication. What if there is only one pair of a conspicuous and a cryptic fish that are confined in bottles and presented to all pike sequentially? Our n is 1 in this case, and anything else would be pseudoreplication (the cryptic fish could have been frightened, more satiated, etc.). One still wants to investigate whether conspicuousness increases a prey’s risk of predation. What if 24 pairs of prey fish are available but only 1 pike, named Fisher. This case is not as bad as the previous one. One will find out a lot about this individual pike. One cannot, however, conclude that pike prefer conspicuous prey (or even that predators prefer conspicuous prey). One may conclude that Fisher prefers conspicuous prey, which does not help much t o understand pike-prey fish relationships in general; perhaps Fisher prefers conspicuous prey because of a terrifying experience he had with a cryptic bait from a fisherman. Results from a single animal can be very valuable when it is tested for its abilities. If one chimpanzee can be taught a sign language that she uses afterward to communicate in a
HOW TO AVOID SEVEN DEADLY SINS IN THE STUDY OF BEHAVIOR
167
sensible way with people (Gardner and Gardner, 1969), one knows what chimpanzees could do in principle. Similarly, I would be impressed if a single fish would solve a very complex optimal foraging task-at the risk that I have tested the R. A. Fisher among all fish by chance.
B. PSEUDOREPLICATION IN CORRELATIONAL STUDIES: NESSYAND LOCHNESSYLESS It is common knowledge that “Nessy” lives in Loch Ness and it is known for sure that there is no such thing in Loch Nessyless 50 km away. One wants to know whether the fish in Loch Ness have developed some sort of anti-Nessy behavior. If this is the case, this knowledge would be of enormous importance for tourism because one can test the fish of all the other 200 lochs and tell whether they contain a Nessy. Fish of the same species, sex, size, and age are collected from both Loch Ness and Loch Nessyless and maintained in individual interspersed tanks until testing with a sophisticated model of Nessy. One does not need to test lab-bred offspring of those fish because it does not matter whether a Nessy recognition is learned or genetic. In the test the fright reaction of individual fish is quantified with the result that fish from Loch Ness are more readily frightened by a dummy Nessy than fish from Loch Nessyless. What can be concluded? The problem is similar to that of maintaining conspicuous fish in one tank and cryptic fish in another tank (see previous discussion). Many things might have happened in Loch Ness that did not happen in Loch Nessyless, which make fish from Loch Ness more cautious than fish from Loch Nessyless. It would be pseudoreplication to compare fish from only two populations that differ both in the trait of interest and in many other traits that one is not aware of. Would it help if six different lochs that each have a Nessy can be compared with six other lochs without a Nessy? Yes, this would be much better. Now, a loch would be the statistical unit and n is 12. Pseudoreplication has been avoided, but the results would still suffer from being only correlational evidence. The lochs that have a Nessy were probably more suitable for Nessies than the lochs without a Nessy. That means that Nessy lochs have a number of traits in common by which they differ from Nessyless lochs. One can hardly rule out that one or several of these traits are responsible for the more cautious behavior of the fish from Nessy lochs and not the presence of Nessies themselves. Are population comparisons completely useless? No, there is a way to improve them. Suppose the dummy Nessy has less pronounced effects on the cautious behavior of the fish of Nessy lochs when the dummy is made step-by-step less realistic. One, therefore, has the impression that the com-
168
MANFRED MILINSKI
plex Gestalt of Nessy matters for the fish. Moreover, if the measures of anti-Nessy behavior include specific behaviors that are suitable to avoid Nessy but not other large objects, then even sceptics might accept the possibility that anti-Nessy behavior has been developed by fish from Nessy Lochs. There are some examples of the latter kind of population comparison, Magurran (1990a) found that offspring of minnows (Phoxinus phoxinus) from a population with pike developed a sophisticated antipike behavior when presented with a realistic dummy pike; this behavior was much less pronounced and sophisticated in minnow offspring from pike-free waters. 111. TREATMENTS ARECONFOUNDED BY TIME AND SEQUENCE EFFECTS
A. SELECTION OF SUBJECTS Imagine that we had not caught pairs of fish from our storage tank and had not decided, by tossing a coin, which should become conspicuous and which cryptic of each pair. Instead, the 24 fish that were to become conspicuous were caught first, and second were caught the 24 fish that were to become cryptic. By this procedure the two experimental groups would be biased by two different sequence effects: the first fish that are caught would be the easiest to catch, that is, the least cautious fish, whereas the last fish would be the most difficult to catch, that is, the most cautious fish. Consistent temperament differences, shy versus bold, among individual fish have been found in many species of fish (e.g., Huntingford, 1976; Milinski, 1987; Magurran, 1993) and other animal species including man (Wilson, Clark, Coleman, and Dearstyne, 1994). A second sequence effect would inevitably come about by disturbing the fish each time one is caught. Fish that are caught early in the sequence are less often disturbed than fish that are caught late in the sequence. Both sequence effects combine to place the more cautious and the more often disturbed fish in the group that are made cryptic and the less cautious and less often disturbed fish in the group that are made conspicuous. It would thus come as no surprise if the conspicuous fish were preyed upon more frequently by the pike, not necessarily because they were conspicuous but possibly because they were the less wary fish because of sequence effects. The elegant solution to this problem is to catch two fish (even sequentially) and determine by tossing a coin which becomes a treatment fish and which a control fish. In this way all time and sequence effects will affect treatment and control groups similarly. B. SEQUENCE OF TREATMENT A N D CONTROL
Again, we already have the best sequence of treatment and control subjects by presenting both the conspicuous fish and the cryptic fish simulta-
HOW TO AVOID SEVEN DEADLY SINS IN THE STUDY OF BEHAVIOR
169
neously to a pike; one has to arrange only that the side on which the conspicuous fish is presented is randomized between trials, Now, assume that a simultaneous choice situation is not possible for technical reasons and one has to present each prey fish singly in a sequential choice situation and measure the latency of the pike’s attack. If one started with all the conspicuous fish followed by all the cryptic fish for each pike, one’s results would again suffer from time and sequence effects: the pike may become faster (or slower) during their time in the experiment because they learn about the prey, become more accustomed to the procedure, become hungrier, simply become older, and so on. The results would be of little value. How should one proceed instead? R. A. Fisher would have suggested strict randomization of the sequence of treatment and control (see Hurlbert, 1984). I did this when I performed my first experiment for my Master’s Thesis 22 years ago and obtained the following sequence of eight trials: T, T, T, C, T, C, C, C. Although I had randomized, this was almost no better than having all treatments before all controls. The best solution is probably, again, to have pairs of treatment and control and determine each time by tossing a coin which is first and which is second. The way in which animals were assigned should be described in detail in the Methods section so that readers can judge for themselves whether the method that was used was suitable to achieve randomization. One should not say “animals were assigned randomly to treatment and control,” but say “animals were assigned to groups applying the following constraints: within each pair selected a random choice was made of which would be the treatment and which the control animal.” IV. No EFFORTS To AVOIDOBSERVER BIAS
I do not assume that you and I want to manipulate results deliberately. It is, however, known from psychological studies that people tend to interpret unconsciously what they see in a way that fits their expectations better. This phenomenon is called observer bias (e.g., Martin and Bateson, 1993). How does it occur? In the early days of ethology, researchers often just observed a subject’s behavior and described what they had seen. This method allows observer bias, as can be shown with Example 1. Conspicuous prey may have a higher predation risk because they were more easily detected than cryptic prey. In Experiment 1 this hypothesis can be tested by determining the prey type that the pike approaches first. Approach means “moving toward something.” There are two opportunities for interpretation: the length of a move can vary between long (e.g., 20 cm; see Fig. 1A) and just recognizable (a few mm; see Fib. 1B). A just recognizable move might be interpreted as a “move” when toward the conspicuous prey
170
MANFRED MILINSKI
El
El OH”
El
El
El
FIG. 1. A pike is observed to approach either the conspicuous or the cryptic prey. (A) A move over a long distance will always be regarded as a “move.” (B) A move over a very short distance may be regarded as a “move” only when it is directed toward the conspicuous prey. (C) The direction of a move may be noted correctly when it is on a straight line toward a prey. (D) When the direction of a move is between the conspicuous and the cryptic prey. one may be inclined to regard it as being directed toward the conspicuous prey (or what is expected).
and as “no move” when toward the cryptic prey. “Toward” may vary between “in a straight line to the cryptic prey” (Fig. 1C) and “more toward the cryptic prey than toward the conspicuous prey” (Fig. 1D). In the latter case one might be inclined to see the direction of a move as being toward the conspicuous prey when it is actually toward the cryptic prey. How can we solve this problem? A. UNBIASED OBSERVATIONS IN THE FIELD One could ask someone to do the observation who does not know one’s hypothesis. The problem here is that because the two prey types are obviously different, the naive observer can unconsciously invent a hypothesis about the type of prey that the pike should prefer and bias his observations accordingly. Another problem is that the observer may miss detecting the
HOW TO AVOID SEVEN DEADLY SINS IN THE STUDY OF BEHAVIOR
171
cryptic prey more often than the conspicuous prey. So this does not work. Instead, one could make a video recording that is shown to naive persons so that the difference in conspicuousness between the two prey types is not detectable (e.g., color turned off, if the difference is in color). With this procedure observer bias is avoided, but the results will suffer from a large variance because the observer may often make mistakes in detecting just recognizable approaches.
B.
UNBIASED OBSERVAITONS UNDER
EXPERIMENTAL CONDITIONS
In an experiment it is much easier, but technically demanding, to avoid observer bias by defining exactly “approach toward.” A naive observer is not needed. We use again Example 1. A tank is divided into halves by an opaque partition that has a sliding door in the middle. The two prey fish are each confined in a Plexiglas container in one side of the tank at the same distance v i s - h i s the sliding door. The pike lives in the other half of the tank. One such tank is needed for each pike. When the sliding door is lifted, by remote control, at a preset time (otherwise one may be inclined to lift it when the conspicuous fish is moving!), a video camera records the pike’s behavior from above. Two thin lines are drawn on the video screen: one parallel to the opaque partition at a distance from the door where one finds it useful to determine the pike’s decision, another line that cuts that half of the tank into a left and a right part as seen by a pike that enters by the door (Fig. 2). Now, one has almost no opportunity for subjective bias. One determines whether the tip of the pike’s snout is in the right or the left part when it passes the decision line. Unfortunately, this rather complicated design is necessary for obtaining unbiased observations.
C. WHENBEHAVIOR Is DIFFICULT To CLASSIFY What can be done if the subject’s behavior is not precisely classifiable so that there is an opportunity for subjective observer bias? One wants to determine the pike’s preference after it has detected both prey types. We did this kind of experiment with sticklebacks that had the choice between parasitized and unparasitized copepods that were confined in Plexiglas cells (Wedekind and Milinski, 1996). We wanted to count the fish’s bites (snout contacts with the Plexiglas) toward each prey type for one minute. The problem was that the fish could bite at a high rate so that two bites could be regarded as a single bite on the video record taken from the front wall (Fig. 3). Because the prey types could be easily recognized by their behavior, a naive observer would not have solved the problem. We used a second video camera aimed exactly along the front wall of the cells that contained
172
MANFRED MILINSKI
FIG. 2. Video screen showing the experimental setup from above. A pike has entered the experimental side of its tank by a door; when the tip of the pike’s snout passes the vertical decision line that is drawn on the screen, it is regarded as having chosen either the conspicuous or the cryptic prey dependent on where the tip of its snout is with respect to the horizontal decision line.
FIG. 3. Experimental setup with which behaviors of the fish can be recorded as bites without the risk of observer bias; see text for further explanations. Drawing by C. Wedekind.
HOW TO AVOID SEVEN DEADLY SINS I N THE STUDY OF BEHAVIOR
173
the copepods so that only the bites toward the Plexiglas could be seen and not the type of prey that was attacked. From this view I decided when behaviors were bites and spoke these events into the audio channel of the video record of the fish’s choice of prey type (which I could not see). From the latter video and audio record we could determine how often each prey type (as defined by the cell that it was confined in) was attacked by the fish. Sometimes a very elaborate technical design is necessary to remove the opportunity for observer bias. The solution must be tailored specifically to the actual problem so that no cookbook recipe can be offered that fits every situation. It needs creativity to find the solution for specific problems after the researcher has become aware of all opportunities for observer bias in the design.
D. EXPERIMENTER BIAS If one does not use a video camera but observes the pike’s predation attempts directly, one’s presence might affect the behavior of both the pike and the prey. There is a further more subtle opportunity for biasing the results: because one expects the pike to prefer the conspicuous prey one might “hold one’s breath” and “freeze” whenever the conspicuous prey is more likely to be attacked. In this way the pike may be less often disturbed when it tries to attack the conspicuous prey, which might therefore inevitably be attacked more frequently. This is similar to the “clever Hans effect”: the clever horse appears to be able to count because his master provides him with subtle signs when the correct number is achieved. The use of a video camera with the observer sitting in another room would help to prevent this bias. A blind with a peep hole for the observer would be the second best solution because it still allows for interaction between observer and subject, for example, by vibrations when the observer becomes nervous. The results may also be biased even when one observes the pike’s behavior from another room: each of the two prey types had to be transferred to its Plexiglas tank; each fish had to be caught from its individual tank (with a net), carried to the experimental tank and released into its Plexiglas container. Here, there is ample opportunity for experimenter bias. One may be more careful with the cryptic fish so that they are less frightened than the conspicuous fish in the experiment. The pike may prefer the conspicuous fish because they are more nervous. This problem can be solved by asking another person (who must, of course, be trained to handle live fish carefully) to transfer the fish. However, the naive helper may like the cryptic fish more than the conspicuous fish and may thus treat them differently. Now we are stuck. It is tempting to suggest that one can always find a simple and elegant
174
MANFRED MILINSKI
solution. One can toss a coin to decide whether a fish becomes conspicuous or cryptic after it has been transferred to its tank! One can position plates behind each Plexiglas tank that either match the conspicuous fish’s color (and contrast with the cryptic fish’s color) or contrast with the conspicuous fish’s color (and match the cryptic fish’s color). This kind of technique has been used, for example, by Dawkins (1971). Why not use only one type of fish and affect its conspicuousness only by matching or contrasting backgrounds? Because the results would be ambiguous, since the pike may like one kind of background better.
ARTIFACTS WHENANIMALS ARENOTACCUSTOMED TO V. POTENTIAL
EXPERIMENTAL PROCEDURES Whenever an animal is handled (caught, marked, kept under restricted conditions, etc.), this experience will probably affect its future actions. For example, if an animal is frightened by being caught and transferred t o a new artificial environment where it is expected to solve a foraging task, it may give priority to avoiding any potential danger instead of to foraging. When it finally starts to forage, it will certainly make a compromise between avoiding potential predation and efficient foraging (e.g., Milinski, 1993). If the experimenter aims to test the predictions of an optimal diet model, he or she might conclude that this species does not select the optimal diet that is predicted. Although this example might appear to be an extreme one, I am convinced that this kind of mistake is made very often. Neither the referees nor the readers of a publication can judge from the Methods section whether the results are flawed by severe handling artifacts. I would like to propose that a detailed description of the methods that were used to habituate the animal to the experimental procedures is required by the “Instructions for Authors” of all behavioral journals. What kind of method should be used to avoid artifacts from handling? Experimenters could find this method if they tried to put themselves in the animal’s place to see what happens to it from its point of view. What information is actually available to the animal? What might it mean functionally for the animal? The only general rule that I can propose is to take the animal through all the experimental steps many times, during the course of several days, except for presenting it with the actual test. For example, being caught and transferred should become a neutral or even a positive event with respect to the task that the animal will be presented with. To illustrate this: I hung the net used to catch my sticklebacks for an experiment in their holding tank every day. Occasionally I offered some food in the net and moved the net with the fish, at first only a bit but later I lifted it out of the tank
HOW TO AVOID SEVEN DEADLY SINS I N THE STUDY OF BEHAVIOR
175
for a few seconds (e.g., Milinski, 1985). After several such training sessions the fish swam into the net even when no food was offered. If the experimental environment is different from its usual environment, the animal should become accustomed to it by being transferred back and forth often enough for it to perceive the experimental environment as familiar with a neutral or positive value. A sentence such as, “after being transferred to the experimental cage, tank, etc., the animal was allowed to become accustomed to it for five minutes before the experiment started,” makes me very sceptical about the value of the results. Such a procedure would be justified only if the animal’s housing conditions are almost identical to the experimental conditions. The experimenter needs experience with the animal, the necessary skills and intuition to provide the animal with only the test stimulus and not with many strange influences in addition to the experimental ones. In our Example 1 the pike lives in the experimental tank and may be trained to attack fish that are confined in Plexiglas containers: it should receive some other food after each attack, otherwise it would cease to attack unreachable prey. The prey fish will, however, behave abnormally within their containers especially during the pike’s approach. They would not only need to be accustomed to this environment but also to be prevented from seeing the pike, for example, by one-way mirrors that allow the pike to see the prey fish but the latter not to see the pike (Magurran. 1990b).
VI. UNSUITABLE CONTROLS Ideally, treatment and control differ only in the trait the effect of which is to be tested. In Example 1 different methods can be used to make one prey fish conspicuous and the other one cryptic. If these fish are cryptic anyway it might seem obvious that one should change the appearance of only the fish that are to become conspicuous. One could apply some red dye to the fish’s skin. The treatment fish are thus conspicuous and the control fish are cryptic. However, the treatment fish differ not only in conspicuousness from the control fish but also by having been handled and the skin treated by a chemical substance. The control fish would have to be treated in the same way, for example, with green dye, so that any preference must be due to the color only. Now the cryptic fish are an excellent control for the treatment fish. A further problem might be that both groups behave abnormally because their skin is continuously irritated by the chemical substance during the test. The pike may thus not discriminate between them as it would if the fish behaved normally. So this does not work. The solution may be to use dead fish as prey. Their behavior
176
MANFRED MILINSKI
will not be changed by being handled and treated with dye. They would mimic fish that “freeze” after having detected a predator. Another solution to our problem with Example 1 would be to change the treatment fish’s color through its diet; an example here is male guppies that received supplementary carotenoids with their food so that they became redder than control fish (Kodric-Brown, 1989). How do we treat the control fish? As far as I know, there is nothing comparable to carotenoids that induces a cryptic color in fish. Even if such a substance were available, it would not help much to create a proper control because its effects on the fish’s behavior might differ from that of carotenoids. It is not obvious that carotenoids change the behavior of a fish. Can one use just untreated fish as controls? Yes, if another experiment shows that the behavior of fish that were treated with carotenoids does not differ from the behavior of untreated fish. One could determine the antipredator behavior of 10 fish that had been treated with carotenoids and of 10 untreated fish, each fish tested singly after a dummy predator attack. Imagine that we do not find a significant difference (at the level of a = .05), but the sample size is much too small for the null hypothesis to be accepted (at the level of p = .20 see next section). This kind of control experiment is not any help. The sample size would have to be increased enormously, which may not be feasable because it would need more than 10 times the effort needed for the main experiment. This is a real dilemma, which has not yet been appreciated by many researchers. I propose the following procedure to circumvent the problem of “proving” that a treatment has no undesirable secondary effect. One increases the magnitude of the treatment (or the trait that cannot be excluded) stepwise until there is a significant secondary effect in the treatment group as compared to the control group; one thus tries to demonstrate an effect instead of “proving” that no effect exists, and this needs only small sample sizes. Then one can do a regression analysis to estimate the magnitude of the secondary effect for the treatment in the main experiment. For example, if the basic amount of carotenoids that is used to induce the red color in the conspicuous fish does not obviously induce a change in their antipredator behavior, another group is fed a higher amount and a third group an even higher amount of carotenoids. From this experimentally established relationship between amount of carotenoids consumed and change in antipredator behavior, the behavioral change that is induced by the basic amount of carotenoids can be estimated. With this information it can be discussed whether the potential secondary effect of the basic treatment is of a biologically important size. A problem can arise if even the largest amount of carotenoids that the fish consumed does not induce a recognizable behavioral change. In this case I would be confident that the much
HOW TO AVOID SEVEN DEADLY SINS IN THE STUDY OF BEHAVIOR
177
smaller basic amount had no effect, but I cannot really prove it. It would be good if this scenario were to be investigated by statisticians and advice provided for behavioral scientists. It cannot be guaranteed that a given treatment differs from a given control only in the trait that is under investigation. This ideal difference can be approached only by spending a lot of effort thinking about other potential differences. These should be discussed in the published paper so that the readers can form their own judgment of the suitability of the control. VII. “PROVING” THE NULLHYPOTHESIS WITH SMALL SAMPLES Sometimes one needs to support the null hypothesis. For example, if treatment and control differ not only in the trait that is under investigation but also in some other trait, we have to “prove” (i.e., reach a threshold of reliability that is set by a convention) that this other trait has no effect (or at most a negligible effect) on the behavior that is measured as a response to the trait under investigation. In our example from the previous section, no significant effect of carotenoids was found on the antipredator behavior of 10 fish in comparison with 10 untreated fish. Can we conclude that the pike prefers fish that were treated with carotenoids because they are redder and not because their antipredator behavior is changed? Did we “prove” the null hypothesis that assumes no effect of carotenoids on antipredator behavior? Is there any convention for accepting the null hypothesis? THE NULLHYPOTHESIS “PROVING”
Cohen writes in his excellent book on power statistics (1988, p. 16): Research reports in the literature are frequently flawed by conclusions that state or imply that the null hypothesis is true. For example, following the finding that the difference between two sample means is not statistically significant, instead of properly concluding from this failure to reject the null hypothesis that the data do not warrant the conclusion that the population means differ, the writer concludes, at least implicitly, that there is NO difference. The latter conclusion is always strictly invalid, and is functionally invalid as well unless power is high. The high frequency of this invalid interpretation can be laid squarely at the doorstep of the general neglect of attention to statistical power in the training of behavioral scientists.
Scientists are usually concerned with having a very low significance level a (Type I error), the probability of falsely rejecting the null hypothesis;
for example, “the pike significantly (p < .05) prefers conspicuous over cryptic prey.” However, taking a very small a results in power values being
178
MANFRED MILINSKI
very small. The complement of power (1 - power) is the p error (Type I1 error); represents the error rate of failing to reject a false null hypothesis. So we need a convention about p if we want to accept the null hypothesis as we have with a for rejecting the null hypothesis. Since p is the complement of power, we obtain a smaller p when we increase power. For example, for power = .95, the p error equals .05. How can we increase power? Power is a function of a,effect size, and n (sample size) (Cohen, 1988). Effect size is the difference between treatment and control, which is negligible or 0 if we assume “no difference.” If we aim at p < .05, we have to increase power to .95. For a zero or negligible effect size we have to determine the sample size with which we can achieve a power of .95. Cohen (1988, p. 17) gives an example. For a correlation analysis, the null hypothesis would be r = 0, or if the effect size is negligible r = .lo. If, for example, one considers a population r = 0.10 as negligible (hence, i), and plans a test of the null hypothesis (at a = 0.05) for power = 0.95 (p = 0.05) to detect i, one discovers that the n required is 1308; for power = 0.90 (p = 0.10), the required n = 1046; and for power = 0.80 (p = 0.20), n = 783.
This example shows that, even if we want effectively to “prove” (assuming a small “negligible” effect size instead of zero) the null hypothesis, we need enormous sample sizes for this proof, which are usually not feasible. Cohen (p. 56) proposes as a convention that one sets the power at .80 ( p = .20). This assumes that Type I errors (falsely rejecting the null hypothesis) are of the order of four times as serious as Type I1 errors (falsely accepting the null hypothesis). Even with this relaxed convention we need sample sizes that are so large that a “proof” of the null hypothesis is probably most often impossible. Everybody who needs to conclude that “no difference between treatment and control existed” should perform a statistical power analysis and include the necessary information in the results section. Any published conclusion of “no effect” or “no difference” without this information should be regarded as unproven. A potential solution to this dilemma is the procedure that I proposed in the previous section. Of course, much smaller sample sizes are needed to reach the level of /3 = .20 if no significant effect can be found and the expected effect size is much larger than “negligible.” “All or nothing” responses that appear when the trait has reached a certain threshold are of this kind. VIII.
CONCLUSIONS
I hope that researchers will become vigilant about the need to avoid the kind of mistakes that I have discussed in this article. Referees and editors
HOW TO AVOID SEVEN DEADLY SINS IN THE STUDY OF BEHAVIOR
179
should ask authors to discuss in detail the methods with which they have avoided these mistakes. As behaviorists we have to work sometimes with limited samples and we cannot have complete control over all conditions. Thus, we have to make compromises. In this case one should be aware of these limitations and be cautious about the conclusions that the study might still allow. I would be glad if readers have become convinced that studying a more detailed review or book on experimental design is worth the investment.
IX. SUMMARY Seven common mistakes in designing and performing studies of behavior and in interpreting their results are discussed with simple examples: (1) unjustified conclusions are made from observational (i.e., correlational) data; (2) data are not independent (“pseudoreplication”); (3) treatments are confounded by time and sequence effects; (4)no efforts are made to avoid observer bias; ( 5 ) potential artifacts arise when animals are not accustomed to experimental procedures; ( 6 ) unsuitable controls are used; (7) an attempt is made to “prove” the null hypothesis with small samples. Acknowledgments I thank Jay Rosenblatt, Peter Salter, Charles T. Snowdon, and Claus Wedekind for very helpful comments.
References Anderson, M. (1982). Female choice selects for extreme tail length in a widowbird. Nature (London) 299,818-820. Cohen. J. (1988). “Statistical Power Analysis for the Behavioral Sciences,” 2nd ed. Lawrence Erlbaum Associates, Hillsdale. New Jersey. Curio, E. (1983). Time-energy budgets and optimization. Experienfia 39,25-34. Curio, E., Ernst, U., and Vieth, W. (1978). Cultural transmission of enemy recognition: One function of mobbing. Science u)2, 899-901. Dawkins, M. (1971). Shifts of “attention” in chicks during feeding. Anim. Behav. 19,575-582. Gardner, B. T., and Gardner, R. A. (1969). Teaching sign language to a chimpanzee. Science 165,664-672. Grafen, A. (1990). Biological signals as handicaps. J. Theor. B i d . 144, 517-546. Huntingford, F. A. (1976). The relationship between anti-predator behaviour and aggression among conspecifics in the three-spined stickleback, Gasferosteusaculeatus. Anim. Behav. 24,245-260. Hurlbert, S. H. (1984). Pseudoreplication and the design of ecological field experiments. Ecol. Monogr. 54, 187-211.
180
MANFRED MILINSKI
Kodric-Brown, A. (1989). Dietary carotenoids and male mating success in the guppy: An environmental component to female choice. Behav. Ecol. Sociobiol. 25,393-401. Kroodsma, D. E. (1986). Design of song playback experiments. Auk 103,640-642. Kroodsma, D. E. (1989). Suggested experimental designs for song playbacks. Anim. Eehav. 37,600-609. Magurran, A. E. (1990a). The inheritance and development of minnow anti-predator behaviour. Anim. Eehav. 39,834-842. Magurran, A. E. (1990b). The adaptive significance of schooling in an anti-predator defence in fish. Ann. Zool. Fennici 27,Sl-66. Magurran, A. E. (1993). Individual differences and alternative behaviours. I n “Behaviour of Teleost Fishes” (T. J. Pitcher, ed), pp. 441-477, 2nd ed. Chapman and Hall, London. Martin, P., and Bateson, P. (1986). “Measuring Behaviour,” 1st ed. Cambridge University Press, Cambridge. Martin, P., and Bateson, P. (1993). “Measuring Behaviour,” 2nd ed. Cambridge University Press, Cambridge. Milinski, M. (1985). Risk of predation of parasitized sticklebacks (Gasterosteus aculeatus L.) under competition for food. Eehuviour 93,203-216. Milinski, M. (1987). TIT FOR TAT in sticklebacks and the evolution of cooperation. Nature (London) 325,433-435. Milinski, M. (1993). Predation risk and feeding behaviour. In “Behaviour of Teleost Fishes” (T. J. Pitcher, ed), pp. 285-305.2nd ed. Chapman and Hall, London. Milinski, M., and Bakker, T. C. M. (1990). Female sticklebacks use male coloration in mate choice and hence avoid parasitized males. Nature (London) 344, 330-333. Meller, A. P. (1988). Female choice selects for male sexual tail ornaments in the monogamous swallow. Nature (London) 332, 640-642. Mgiller, A. P. (1989). Viability costs of male tail ornaments in a swallow. Nature (London) 339, 132-135. Mgiller, A . P. (1990a). Male tail length and female mate choice in the monogamous swallow Hirundo rustica. Anim. Behav. 39,458-465. Mgiller, A. P. (1990b). Effects of a haematophagous mite on the barn swallow (Hirundo rustica): A test of the Hamilton and Zuk hypothesis. Evolution 44, 771-784. Pettifor, R. A. (1990). The effect of avian mobbing on a potential predator, the European kestrel, Falco tinnunculus. Anim. Behav. 39,821-827. Wedekind, C., and Milinski, M. (1996). Do three-spined sticklebacks avoid to consume copepods, the first intermediate host of Schistocephalus solidus?-An experimental analysis of behavioural resistance. Parasitology. 112, 371-383. Wedekind, C., Seebeck, T., Bettens, F., and Paepke, A . J. (1995). MHC-dependent mate preferences in humans. Proc. R. SOC. London E 260,245-249. Wilson, D. S., Clark, A. B., Coleman, K., and Dearstyne, T. (1994). Shyness and boldness in humans and other animals. Trends Ecol. Evol. 9,442-446.
ADVANCES IN THE STUDY OF BEHAVIOR,VOL. 26
Sexually Dimorphic Dispersal in Mammals: Patterns, Causes, and Consequences LAURASMALE DEPARTMENT OF PSYCHOLOGY MICHIGAN STATE UNIVERSITY EAST LANSING, MICHIGAN
SCOTTNUNESAND KAYE. HOLEKAMP DEPARTMENT OF ZOOLOGY MICHIGAN STATE UNIVERSITY EAST LANSING, MICHIGAN
I. INTRODUCTION Many members of most mammalian species permanently leave home partway through their ontogenetic development, and attempt to settle in a new area or social group. The permanent and complete departure of an individual from its birthplace is called natal dispersal (Lidicker, 1975; Greenwood, 1980). Natal dispersal represents the most profound shift in environmental conditions an individual is likely to experience in its lifetime. A dispersing mammal leaves a familiar world for a place in which the whereabouts of such essentail resources as food and shelter must be discovered anew, as must escape routes from predators and other dangers. Often more challenging than these novel aspects of the physical environment are changes in the disperser’s social world. Gregarious mammals that do not disperse usually spend their lives within groups of conspecifics in which membership changes only very slowly and gradually. By contrast, at dispersal an individual may experience a sudden and dramatic transformation of its social life. Virtually the entire suite of conspecifics with which a young mammal interacts changes at dispersal. The new immigrant must immediately begin sorting out the identities of its new associates, identifying patterns in their relationships, and fitting itself into their existing social order in a manner that maximizes its own reproductive success (Mason, 1979). Dispersal thus profoundly alters the social environment in which all subsequent behavioral development occurs. 181
Copyright 0 1997 by Academic Press All rights of reproduction in any form reserved. W-3454i97525.00
182
LAURA SMALE ET AL
Dispersal behavior in many mammals is strongly sexually dimorphic. That is, members of one sex usually disperse, whereas members of the opposite sex tend to be philopatric, spending their entire lives in their natal groups or areas. Sexually dimorphic dispersal behavior often appears to have been favored by natural selection as a mechanism to facilitate inbreeding avoidance or optimal outbreeding (Greenwood, 1980; Shields, 1982). Although determinants of which sex disperses remain somewhat more obscure, they are likely to include differential importance of familiarity with the natal home range (Greenwood, 1980; Pusey and Packer, 1987a; Waser and Jones, 1983), differential ability to acquire competitive advantages by remaining near same-sex kin (Alberts and Altmann, 1995; Andelman, 1986; Wrangham, 1980), and constraints imposed by demands of parental behavior. In general, when the reproductive success of both males and females is strongly influenced by the same socioecological variables, members of both sexes emigrate. Thus, both males and females often disperse among solitary mammals and among monogamous or cooperatively breeding mammals. Solitary species in which both sexes disperse include most marsupials and insectivores, blind mole rats (Spalax ehrenbergi), kangaroo rats (Dipodomys spp.), tree squirrels (Sciurus spp.), chipmunks (Tamius and Eutamius spp.), various cervids, bears, and most cats and mustelids (reviewed in Greenwood, 1980; Rado, Wollberg, and Terkel, 1991; Sinclair, 1992; Waser and Jones, 1983). Monogamous mammals in which both sexes disperse include small bovid antelope, various rodents, most canids, social mongooses, and callitrichid primates (reviewed in Dobson, 1982; Greenwood, 1980; Pusey and Packer, 1987a; Sinclair, 1992; Strier, 1994; Waser, 1996). Interestingly, philopatric individuals sometimes occur in virtually all of these species, and these are more likely to be females than males (Waser and Jones, 1983). Sex differences in dispersal behavior are most striking among long-lived, gregarious mammals exhibiting polygynous mating systems. In these species, dispersal may be biased toward either males or females, although the former is significantly more common (Dobson, 1982; Greenwood, 1980; Holekamp, 1984a; Pusey and Packer, 1987a; Sinclair, 1992; Waser, 1996). Dispersal is strongly male biased in group-living marsupials, most cercopithecine primates, lemurs, ground-dwelling sciurid rodents, lions (Puntheru Leo), African elephants (Loxodonra ufricuna), various pinnipeds, and bovid antelope living in permanent herds (Cockburn, Scott, and Scotts, 1985; Dawson, 1995; Douglas-Hamilton and Douglas-Hamilton, 1975; Holekamp, 1984a;Jones, 1983; Murray, 1982; Pusey and Packer, 1987a,b; Sinclair, 1977; Sussman, 1991; Vick and Pereira, 1989; Warneke, 1975). Female-biased dispersal has been documented in most great apes, some New-World primates in the family Cebidae, red colobus monkeys (Piliocolobus badius),
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
183
hamadryas baboons (Papio hamadryas), and white-lined bats, Saccopyeryx bilineata (Bradbury and Vehrencamp, 1977; Clutton-Brock, 1989; Pusey and Packer, 1987a; Sigg, Stolba, Abegglen, and Dasser, 1982; Strier, 1994). Interestingly, in many mammals exhibiting strongly dimorphic dispersal behavior, some members of the philopatric sex also occasionally disperse. For example, among baboons (Papio spp.), virtually all males disperse but females also occasionally emigrate (Smith, 1992). Early analyses of causes and consequences of dispersal often treated males and females together, and sought unitary explanations for dispersal by all individuals. Results from such studies filled the dispersal literature with apparent contradictions, and failed to answer the question of why animals leave home. Here we shall highlight how the patterns of dispersal behavior often differ between the sexes in ways that reflect different causes and lead to different consequences. The evolutionary consequences of dispersal behavior have received considerable attention during the past two decades, (Chepko-Sade and Halpin, 1987; Dobson, 1982; Lidicker, 1975; Moore and Ali, 1984; Pusey, 1987; Stenseth and Lidicker, 1992), but its proximal causes and consequences remain poorly understood (Holekamp, Smale, Simpson, and Holekamp, 1984;Nunes and Holekamp, 1996).Moreover, whereas trappers and woodsmen of centuries past noted strong sex differences in mammalian dispersal behavior, the immediate causes of this dimorphism and its developmental consequences have yet to be fully elucidated. Most researchers who focus on proximal mechanisms underlying the development of sex differences work in laboratory settings where dispersal poses a far less tractable problem than do many other sexually dimorphic behavior patterns. In the absence of a viable laboratory model of mammalian dispersal behavior, we have investigated the proximal causes and consequences of dispersal in two long-lived mammals in the wild. Our work with spotted hyenas (Crocuta crocuta), and Belding’s ground squirrels (Spermophilus beldingi) has focused on the patterns, causes, and consequences of sexually dimorphic dispersal behavior. Our studies of social development in spotted hyenas made us aware of the importance of sexually dimorphic dispersal behavior in setting males and females onto different ontogenetic trajectories (Wiley, 1981), and taking them into contrasting social worlds demanding different types of behavioral decision making. Thus, our hyena work has focused on the social consequences of dispersal behavior. Our work on hyenas has not focused on the proximate causes of dispersal behavior, because of the difficulties of conducting the relevant experimental manipulations needed to address this issue in this species. However, we have complemented this work with an investigation of the proximate causes of dispersal in S. beldingi. This work has included experi-
184
LAURA SMALE ET AL
mental manipulations in the field aimed at elucidating the physiological mediation of sexually dimorphic dispersal. In the present chapter we discuss sex differences in dispersal behavior in long-lived polygynous mammals. We focus on the socioecological variables influencing dispersal, the physiological mediation of dispersal behavior, and the developmental consequences of sexually dimorphic dispersal. Our goal is not to provide a comprehensive review of mammalian dispersal patterns. Rather, we emphasize studies of mammalian dispersal in which data exist for both sexes, and we use our own findings from Crocutu and S. beldingi to illustrate salient points that we believe might broadly apply to many other gregarious mammals. Because both of our study species exhibit strongly male-biased dispersal, we will deal here mainly with other mammals exhibiting similar dispersal patterns. With this in mind, we first present background information on the social biology of each of our study animals, and on the methods we use to study dispersal in these two species. Second, we compare patterns of dispersal behavior in male and female mammals. We examine, for example, sex differences in the ages at which dispersal occurs, and in tendencies to leave home alone or in groups. Third, we evaluate proximal causes of sex differences in dispersal. That is, we discuss factors promoting dispersal in members of the sex that ordinarily disperses, and we contrast these with the variables stimulating dispersal among members of the sex that disperses only occasionally. Fourth, we examine the ontogenetic consequences of sexually dimorphic dispersal patterns. Specifically, we explore how dispersal behavior differentially affects the social development and cognitive abilities of males and females. Finally, we identify important questions for future research.
11. RELEVANT BIOLOGY A N D STUDY METHODS FOR SPOITED HYENAS AND BELDING’S GROUND SQUIRRELS A. BIOLOGY OF THE SPOITED HYENA
Spotted hyenas, which occur throughout sub-Saharan Africa, are members of the carnivore family Hyaenidae. This family also contains the striped hyena (Hyaena hyaena), the brown hyaena (H. brunnea), and the aardwolf (Proteles cristutus). Crocutu differ from the other members of this family in a variety of ways, many of which are related to the fact that only Crocutu regularly hunt ungulates larger than themselves. In contrast to other hyenas, Crocutu are generalists that exploit not only carrion or invertebrates, but also live ungulate prey (Kruuk, 1972). Compared to other Hyaenidae, Crocutu can thus live at extremely high densities, and the size of their social
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
185
groups, called clans, can exceed 100 individuals in the prey-rich savannahs of eastern Africa. Crocuta live in fission-fusion societies in which all clan members recognize each other, defend a common territory against members of neighboring clans, and rear their cubs at a single communal den (Kruuk, 1972). Although individual hyenas spend much of their time alone or in small groups, particularly when foraging, they join together during territorial defense, interactions with lions and other competitors, and at kills. The carcasses of large ungulates represent extremely rich but ephemeral food patches that occur unpredictably in space and time. In dealing with these precious food resources within their social groups, Crocufa confront a problem not faced by other hyena species: intense competition over access to carcasses. The extraordinary intensity of Crocufa’sfeeding competition is undoubtedly responsible for many of the unique features of the biology of this species. Each Crocuta clan is structured by a rigid linear dominance hierarchy (Kruuk, 1972; Tilson and Hamilton, 1984; Frank, 1986b). An individual’s position in the clan’s hierarchy influences virtually every facet of its existence, including access to food, association patterns, space-use patterns, and many aspects of reproductive performance (Frank, Holekamp, and Smale, 1995; Holekamp, et al., in press; Holekamp, Smale, and Szykman, in press; Mills, 1990). For example, high-ranking females bear their first litters at younger ages, experience shorter interlitter intervals, and enjoy longer reproductive life-spans than do lower ranking females. Female Crocuta are more aggressive than are males in a variety of contexts, particularly those associated with the acquisition or maintenance of social rank. Females are also approximately 10% larger than males, and their external genitalia are heavily masculinized, such that the clitoris is modified to form a fully erectile pseudo-penis, and the vaginal labia form structures closely resembling the male’s scrotum (Matthews, 1939). This syndrome of behavioral and morphological masculinization of female Crocufu is associated with a unique pattern of exposure to androgens, both before and after birth (Glickman, Frank, Pavgi, and Licht, 1992; Licht ef al., 1992). Males reach reproductive maturity at approximately 24 months, and most females first conceive in their fourth year. Life-span among free-living hyenas may exceed 18 years (Frank et al., 1995). Crocufaclans contain two classes of individuals: natal animals and immigrants. Natal animals include adult resident females and their offspring, whereas immigrants are exclusively adult males. All males disperse from their natal clans, but female dispersal is relatively rare (Henschel and Skinner, 1987). Within a clan all natal animals are ranked above all immigrants (Smale, Frank, and Holekamp, 1993). Among natal animals, dominance is determined by maternal rank such that young animals are ranked
186
LAURA SMALE ET AL
just below their mothers (Holekamp and Smale, 1993; Smale et ul., 1993). By the time they are 12-18 months old, youngsters’ ranks, in relation to both peers and larger hyenas, are isomorphic with those of their mothers. A strict pattern of youngest ascendancy exists within matrilines, and young Crocutu outrank their older siblings by the time they are 9 months old (Holekamp and Smale, 1993). Both within and between matrilines, rank relations are importantly influenced by mothers’ interventions on behalf of their cubs. Social rank among immigrant males is highly correlated with immigrants’ tenure in their new clans, such that those arriving first dominate those arriving later (Fig. 1). The maintenance of rank relationships within a clan is associated with high rates of low-intensity, ritualized aggression. More intense aggressive behavior is seen less frequently, and its occurrence is restricted to a few specific contexts. Unrestrained aggression occurs during the process of intralitter rank acquisition (Frank, Glickman, and Licht, 1991; Smale, Holekamp, Weldele, Frank, and Glickman, 1995), and on rare occasions when members of a low-ranking matriline join together to challenge members of a higher ranking matriline (eg., Mills, 1990). Rank reversals among immigrant males may involve similarly extreme aggression. Finally, unrestrained, high-intensity aggression often occurs during interactions between members of different clans (Kruuk, 1972; Holekamp, Ogutu, Frank, Dublin,
Arrival order of immigrants FIG. 1. Rank ordering of immigrant male hyenas present in Talek in 1993, as a function of their tenure in the clan. With few exceptions, those males that arrived earliest outranked later arrivals. Rank relations between two pairs of males were unknown, those ranked fourth and sixth. Results from 1993 were typical of results from all other years.
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
187
and Smale, 1993). Clan wars can be deadly, and intense aggression directed toward outsiders may represent an important barrier to dispersal. USEDTo STUDYSPOITED HYENAS B. METHODS
In May, 1988, we began our ongoing study of Crocuta in the Masai Mara National Reserve, Kenya, located in the northern part of the Serengeti ecosystem. Some of this work and our study methods are described more fully elsewhere (e.g., Holekamp and Smale, 1993; Smale et al., 1993; Holekamp et d.,1993; Holekamp and Smale, 1995; Smale et al., 1995). Our research focused on the members of one large social group, the Talek clan (described in Frank, 1986a,b), and in recent years we concentrated on individuals that emigrated from or immigrated into this clan. Since 1988, the number of hyenas in the clan has ranged from 47 to 79 (mean = 63 3). The Talek clan defends a group territory of approximately 60 km’. Observers were continuously present in the Talek clan’s territory between June 15, 1988, and February 15, 1996, except during April, 1990. During the study period we observed and tracked Talek hyenas for over 25,000 hr. Individual hyenas were recognized by their unique spot patterns, and males were distinguished from females on the basis of penile morphology (Frank, Glickman, and Powch, 1990). During the study period we monitored a total of 36 adult females, 154 juveniles, and 136 immigrant males. Immigrant males were present in the Talek clan’s home range for periods ranging from a few days to several years, but most (51%) were transients who remained in the Talek area for less than 1 month (Fig. 2). We observed Talek hyenas from a vehicle parked 1-30 m from our subjects, using binoculars or night vision goggles. Most observations were recorded between 0600 and 0900 hr, and between 1700 and 2000 hr, but we supplemented these with additional observations at other times of day. We regularly searched for hyenas by driving slowly through all parts of the Talek clan’s home range. Each time a hyena was sighted, we recorded its identity, activity, and location. In addition, we routinely collected behavioral data using a critical incident sampling procedure (Altmann, 1974). Critical incident data were used to assess dominance relationships on the basis of the direction of aggressive and submissive behaviors in 10,582 dyadic interactions between hyenas. All occurrences of courtship behavior ( N = 581; as described by Kruuk, 1972, and Mills, 1990) and copulation ( N = 16) were also recorded as critical incidents. Prey availability within the Talek clan’s home range was estimated throughout the study period at 2week intervals via ungulate censuses along two 4-km transects (Holekamp el al., 1993). We measured temporal variation in intensity of feeding competition among Talek hyenas by counting the number of hyenas present at
*
188
LAURA SMALE ET AL
Male tenure (months) FIG.2. Frequency distribution showing lengths of tenure in the Talek clan of 136 immigrant male hyenas. Over 50%stayed for less than one month, and only 23% stayed for more than one year.
every kill, and also by calculating the number of aggressive interactions observed per hour per hyena feeding at each kill (Holekamp et al., 1993). Specific aspects of Crocuta’s dispersal behavior were examined either by following natal animals when they left Talek, or by monitoring the behavior of immigrants when they arrived in the Talek area. The dispersal of Talek animals was monitored primarily with the help of radio telemetry. Ten natal Talek males were fitted with radio collars when they were 26-36 months of age. These males were tracked from both the ground and the air, according to a schedule that changed at different phases of the study. Between January 1991 and August 1993, potential dispersers were located three times per week. If one of these animals was not found in the home range during a given week, intensive efforts were initiated to locate that animal outside the home range, by car and by balloon, until the animal was located. After August 1993, the tracking schedule became less systematic. However, all 10 natal Talek males that had been collared were relocated after they dispersed from Talek. To relocate natal Talek males who dispersed before being radio-collared, we visited 31 locations throughout the Reserve and environs in 1991-1992, and broadcast through loudspeakers the recorded sounds of hyenas feeding (Holekamp et al., 1993). These broadcasts attracted local hyenas from distances of up to 4 km. By examining the spot patterns of all hyenas (approximately 450) approaching our vehicle in response to these playbacks, we were able to relocate two uncol-
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
189
lared natal males that had dispersed from Talek, as well as a group of Talek females and their offspring that had dispersed en masse when the study clan fissioned in 1989-1990. Two additional males that had dispersed from Talek were relocated in chance encounters outside the Talek home range. C . BIOLOGY OF BELDING’S GROUND SQUIRRELS
Spermophilus beldingi are diurnal, semifossorial sciurid rodents, inhabiting alpine and subalpine meadows of the western United States (Jenkins and Eshelman, 1984). Although S. beldingi populations d o not contain distinct social subdivisions, as do those of black-tailed prairie dogs (Cynomys ludoviciunus; King, 1955; Hoogland, 1995) and Olympic marmots ( M . olympus; Barash, 1973), S. beldingi are gregarious animals that engage in daily social interactions with conspecifics (Sherman, 1977). Populations of S. beldingi typically contain several clusters of related females (Michener, 1983). Within such clusters, each breeding female utilizes a separate burrow system, but interacts more frequently and more amicably with female kin than with unrelated animals (Sherman, 1977, 1981). Female S. beldingi are generally philopatric, whereas all males emigrate from their natal areas before reaching reproductive maturity (Sherman, 1977; Holekamp, 1984b). The life history of S. beldingi is characterized by 4-month active seasons interspersed with 8-month hibernation periods (Sherman and Morton, 1984).The annual cycle of activity is rigidly constrained by extreme seasonal changes in climate (Morton and Sherman, 1978), and ontogenetic development is accelerated compared to that of nonhibernating rodents of similar body size. Between weaning and entry into hibernation 7-12 weeks later, the mean body mass of juveniles increases four- to sixfold. This rapid development presumably permits juveniles to attain body mass and fat reserves necessary for overwinter survival (Morton and Tung, 1971;Morton, Maxwell and Wade, 1974; Maxwell and Morton, 1975). S. beldingi’s breeding season begins 1 or 2 weeks after emergence from hibernation in the spring, and lasts 3 to 4 weeks (Sherman and Morton, 1984). Males begin breeding at 2 years of age. Variance in male mating success is high, and only a small proportion of males successfully copulate during the breeding season (Sherman, 1976). Body mass of male S. beldingi is positively correlated with success in winning fights against other males during the breeding season, and also with mating success (Sherman, 1976). Female S. beldingi first breed as yearlings, and exhibit estrous behavior on only one day per year. Females appear to exhibit mate choice behavior, accepting the mating attempts of some males, while rejecting those of others. Litters are multiply sired, typically by two or three males (Hanken and Sherman, 1981). Gestation lasts 24-25 days, and the lactation interval
190
LAURA SMALE ET AL
is 25-28 days (Holekamp et al., 1984; S. Nunes and K. E. Holekamp, unpublished observations). Young remain underground in their natal burrows until weaning. Females maintain maternal territories during gestation and lactation, and often act cooperatively with female kin in defending these territories (Sherman, 1977, 1981). Female S. beldingi are able to discriminate between littermate sisters, nonlittermate sisters, and other female squirrels, and this capacity to recognize kin appears to facilitate nepotistic interactions among females (Holmes and Sherman, 1982; Holmes, 1986a,b). D. METHODS USEDTo STUDY BELDING’S GROUND SQUIRRELS We studied S. beldingi in Mono County, California, from 1979 to 1982 at Lee Vining Canyon and the Harvey Monroe Hall Natural Area, and from 1993 to 1996 at Tioga Pass. We regularly captured subject animals in Tomahawk live traps. All animals ( N = 2622) were marked with ear tags and fur dye. At each capture, we weighed squirrels, counted their ectoparasites, and recorded capture location and reproductive status (Morton and Gallup, 1974). We observed adult females daily during lactation, and trapped their young within four days of their first appearance above ground. Birth dates were assigned to juveniles by subtracting 27 days, the mean duration of lactation (Holekamp et al., 1984), from the date of their first emergence from the natal burrow. We defined a squirrel’s natal area as a circle with radius of 80 m, centered at its natal burrow. To determine dispersal status, we regularly observed each marked litter in the study area from atop rocks, ladders, and observation tripods. At dawn we noted whether marked juveniles emerged from burrows in their natal areas, and at dusk we noted whether they returned to their natal areas for the night. We determined ages at which juveniles disappeared from their natal areas by subtracting birth dates from the dates on which we ceased to observe them at the natal site. We classified juveniles that had disappeared as dispersers only if we later recovered them outside their natal areas. To relocate juveniles that had disappeared, we systematically trapped and visually searched our study sites and surrounding areas, tracked squirrels with radio telemetry, and examined road kills on nearby highways. We estimated dispersal distances using aerial photographs obtained from the United States Geological Survey, as well as direct measurements on the ground. We observed the behavior of squirrels for 2337 hours from 1979 to 1982, and for 143 hours in 1995, using Altmann’s (1974) focal animal survey technique to monitor behaviors described in Holekamp’s (1983) ethogram for S. beldingi. To estimate population density, we conducted regular scans
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
191
of each study area, during which we counted all squirrels active above ground. We also monitored availability of essential resources such as nest sites and food (Holekamp, 1986). Using these methods, and those described for spotted hyenas, we sought to elucidate the patterns, causes, and consequences of dispersal in our subject organisms. However, before describing what we have learned from these studies, we will review some of what is known more generally about mammalian dispersal patterns from studies of other group-living species.
111. SEXDIFFERENCES IN MAMMALIAN DISPERSAL: PAITERNS AND
PROCESSES
OF DISPERSAL IN SPACE AND TIME A. PAITERNS
Successful mammalian dispersal typically has three components: departure from the original home group or area, travel in search of a new home, and settlement in a new group or area (Lidicker and Stenseth, 1992). However, emigrants may also become solitary, or establish an entirely new social group. Mammalian dispersal patterns are variable in both space and time. In some cases, spatial displacement may be entirely accidental. For example, a mouse foraging in a creek bed might inadvertently be transported several kilometers from its original home area by waters of a flash flood. In such cases, the displaced individual is equally as likely to be male as female, and the displacement might occur at virtually any time during the life-span. In most cases, however, mammals disperse in a fashion that varies predictably both between species and between male and female conspecifics. Three aspects of the dispersal process often vary with sex. First, dispersers may emigrate alone or with conspecifics. Second, after emigrating, individuals may transfer directly into an existing breeding unit, join a group of same-sex peers, become solitary before acquiring mates, or immediately establish a new permanent social unit of their own (e.g., Harcourt, 1978; Gese and Mech, 1991; Rutberg and Keiper, 1993). Third, the timing of dispersal may vary considerably. That is, it may be restricted in its occurrence to a brief age interval, or occur unpredictably at virtually any point in the life-span. 1. Do Individuals Disperse Alone or with Conspec$cs?
Among solitary mammals, members of both sexes usually emigrate independently, as lone individuals. This has been documented, for example, in moles (Talpa europu, Godfrey, 1957); sloths (Brudypus tridactylus, Montgomery and Sunquist, 1978), opossums (Didelphis virginiunus, Gillette,
192
LAURA SMALE ET A L
1980; Trichosurus vulpecula, Dunnett, 1964), blind mole rats (Rado et af., 1991), moose (Alces afces, Cederlund and Sand, 1992; Geist, 1971), and black bears (Ursus americanus,Rogers, 1987). Among gregarious mammals, dispersal may occur by group fission. This is the process by which a parent group divides into two distinct daughter groups. Female dispersal usually occurs via group fission among bighorn sheep (Ovis canadensis),elk (Cervus efephas), bison (Bison bison), muskox (Ovibos moschatus, McCullough, 1985), lions (Panthera leo), coatis (Nasua narica), and most cercopithecine primates (Chepko-Sade and Sade, 1979; Dittus, 1988; Gompper, 1994; Hanby and Bygott, 1987; Prud’Homme, 1991; Pusey and Packer, 1987a,b; Russell, 1982). In contrast, most male elk, bison, coatis, and cercopithecine primates emigrate alone. Dispersal may also occur as the emigration of small groups of same-sex peers or siblings. Male lions and cheetahs (Acinonyx jubatus), for example, often disperse in coalitions composed of brothers or other close kin (Caro, 1994; Hanby and Bygott, 1987; Pusey and Packer, 1987b). Female cheetahs disperse alone (Caro, 1994). Male cercopithecine primates occasionally emigrate with one or two male companions (e.g., Cheney and Seyfarth, 1983; Matsumura, 1993; Pusey and Packer, 1987a). Emigration in small groups occurs commonly in wolves (Canis lupus) and dwarf mongooses (Hefogaleparvula) of both sexes (Ballard, Whitman, and Gardner, 1987; Rood, 1987). Dispersal may even occur as a group fusion event, particularly under conditions of declining group size. This has been reported for female vervet monkeys (Cercopithecus aethiops, Hauser, Cheney, and Seyfarth, 1986; Isbell, Cheney, and Seyfarth, 1991) and female Japanese macaques (Macacafuscata, Takahata, Suzuki, Okayasu, and Hill, 1994). Similarly, all female plains vizcachas (Logostomus maximus) in a small group may emigrate and join another social unit, whereas male vizcachss usually emigrate alone (Branch, Villareal, and Fowler, 1993). Thus, in many mammals, and particularly among gregarious polygynous species, males and females are differentially likely to disperse with conspecifics.
2. Where Do Emigrants Go? Among solitary mammals in which both sexes usually disperse, males generally travel longer distances before settling than do females (e.g., moose, deer, bears, various mustelids: reviewed in McCullough, 1985;Waser and Jones, 1983; Waser, 1996). However, there are some species in which long-distance dispersal is more common among females than among males (e.g., kangaroo rats, Jones, 1987; dwarf mongooses, Waser et al., 1994; European badgers, Meles meles, Waser, 1996). Thus, even when probabilities of dispersal by males and females are similar, dispersal distances are often sexually dimorphic. After emigrating, one option for gregarious mammals is to transfer directly to an existing breeding unit, as d o most male ground squirrels and cercopithecine primates (Holekamp, 1984a; Matsu-
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
193
mura, 1993; Pusey and Packer, 1987a). A second strategy, employed by many male mammals, is to enter an intermediate phase of solitary nomadism or membership in a group of same-sex conspecifics. For example, whereas female impala (Aepyceros melampus) are philopatric, males join bachelor herds after leaving their natal groups (Murray, 1982). Male mountain gorillas (Gorilla gorilla beringei) usually become solitary after departing from the natal group, before joining or establishing breeding units (Harcourt, Stewart, and Fossey, 1976; Harcourt, 1978). Male ponies (Equus caballus) become solitary or join bachelor herds after leaving the natal band, and form new bands by attracting females (Berger, 1986, 1987). In contrast, female mountain gorillas and ponies transfer directly to new groups or to lone males. In cooperative breeders, such as wolves and dwarf mongooses, the opposite pattern appears to be common, and males are generally more likely to directly enter a breeding unit after leaving the original home area (Gese and Mech, 1991; Rood, 1987). Male dwarf mongooses typically transfer between breeding groups, whereas females may become solitary and form new breeding units by attracting males (Rood, 1987;Waser, 1996). Although coalitions of male lions often lead nomadic lives for prolonged periods after leaving the natal pride, male dispersal is almost always followed by eventual immigration into a new social group (Pusey and Packer, 1987b). This is not the case when female lions disperse. Male coatis become solitary after dispersing, but later join other groups temporarily to breed (Russell, 1982). Whereas male lions, coatis, and cercopithecine primates regularly join existing groups at dispersal, females of these species never do so. Instead, females establish a new social unit, and thereby emigrate without immigrating (Waser, 1996). Secondary dispersal, which is any dispersal movement occurring after natal dispersal, may occur among both male and female mammals, but patterns of secondary dispersal are often sexually dimorphic. For example, the target destinations generally differ between males and females when both engage in secondary dispersal. During secondary dispersal in lions, ground squirrels, and red kangaroos (Macropusr u f u ) ,males typically move farther away from the natal area. In contrast, secondary dispersal by females of these species often entails moving back toward the natal area, or actually reestablishing themselves there (Dawson, 1995; Oliver, 1986; Hanby and Bygott, 1987; Holekamp 1984b; S. Nunes, unpublished observation). Thus, during both primary and secondary dispersal, male and female dispersers in many mammalian species differ with respect to where they end up after leaving home. 3. When Do Dispersers Emigrate? Among gregarious polygynous mammals, temporal patterns of natal dispersal by females tend to be more variable than are patterns of male
194
LAURA SMALE ET A L
dispersal. Dispersal from the natal group generally occurs during a relatively narrow window of time within the life histories of many male mammals. For example, recent data by Packer, Collins, Sindimwo, and Goodall (1995) for olive baboons Papio cynocephalus reveal that over 80% of surviving males disperse when they are 8 to 9 years of age. Similarly, except when they are ousted during pride takeovers, male lions emigrate from their natal areas within fairly brief age intervals (Hanby and Bygott, 1987; Packer et al., 1995). Males of some group-living mammals, including many grounddwelling sciurids (Holekamp, 1984a), disperse long before puberty. Dispersal by males of the dasyurid marsupials Antechinus stuartii and A . swainsonii is highly synchronized and temporally predictable, occurring 5 to 6 months before a similarly synchronized and predictable period of breeding (Cockburn et al., 1985). In most gregarious mammals, males disperse shortly after attainment of reproductive maturity. Male dwarf mongooses emigrate from the natal pack with male cohort members at 1 to 3 years of age, soon after achieving sexual maturity (Rood, 1987,1990). Dispersal from the natal group likewise occurs predictably after puberty in male lions, baboons, verbets, and Japanese macaques (Cheney and Seyfarth, 1983; Hanby and Bygott, 1987; Matsumura, 1993; Packer et al., 1995; Pusey and Packer, 1987b). Thus, dispersal is commonly a predictably timed occurrence in the ontogenetic development of many male mammals, as are other important life history events such as puberty and first reproduction (Delemarre-van de Waal, Plant, van Rees, and Schoemaker, 1989;Packer etal., 1995; Wilson, 1992). Among female members of species in which males normally disperse, age of natal dispersal is less often predictable. For example, the ages at which female cercopithecine primates and ungulates engage in dispersal range from immediately postpubertal to quite elderly (Chepko-Sade and Sade, 1979; McCullough, 1985; Prud’Homme, 1991). In species in which both sexes regularly disperse, sex differences in the timing of dispersal can be striking. For example, although both male and female mountain gorillas typically emigrate from the natal group shortly after puberty, this occurs between 10 and 14 years of age in males, but between 6 and 11 years of age in females (Harcourt et al., 1976; Harcourt, 1978). Similarly, male and female ponies disperse from their natal herds at different ages (Rutberg and Keiper, 1993). Even among solitary mammals, the timing of dispersal is often dimorphic. Males and females often emigrate at different ages in solitary opossums (Gillette, 1980), palm civets (Nandinia binotata, Charles-Dominique, 1978), moose (Cederlund and Sand, 1992; Geist, 1971), deer (Nelson, 1993; reviewed in McCullough, 1985), black bears (Rogers, 1987), and various mustelids (King, 1989). To summarize, sex differences in the processes and spatiotemporal patterns of mammalian dispersal behavior are often profound, particularly
195
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
among gregarious, polygynous mammals. In many species, males and females are differentially likely to leave their natal groups, leave at different ages via different processes, and end up in different places with dissimilar companions. These sex differences in patterns and processes may reflect underlying differences in the proximal causation of dispersal behavior, as well as in its adaptive functions for males and females. Next we describe two specific examples of species in which we have documented sex differences in the patterns and causes of dispersal. I N PATTERNS OF HYENA DISPERSAL B. SEXDIFFERENCES
1. Male Hyenas
a. Do Males Leave Gradually or Abruptly? The process by which male hyenas disperse was initially suggested to us by patterns of male disappearances from the natal clan (Fig. 3). When they were 18-24 months of age, males began to vanish from the clan’s home range for periods lasting several days to several weeks. As males grew older, these absences were more prolonged, lasting up to 8 months in extreme cases. In contrast, we rarely observed disappearances among females. Because observers were continu-
I
1
2
1
5
Age (yeas) FIG.3. Disappearances of natal hyenas from the Talek home range as a function of age. If a hyena was not seen during a 2-week observation period it was said to have “disappeared” during that interval. These disappearances apparently represent periods when animals are on excursion, exploring habitat beyond the boundaries of the Talek area. Males begin to go on these excursions at 1.5 years of age and differ significantly from females by 2.5 years of age with respect to their sightings in the natal area (r = -4.72; df = 41; p < .001). Females continue to be regularly seen in the natal home range through 5 years of age. Data points represent mean values in each interval for 21 males and 22 females.
196
LAURA SMALE ET AL
ously present in the study clan’s home range for seven years, watching hyenas approximately 5 hr daily, we were highly likely to see every clan member present in the home range at least once during each 2-week observation interval. Thus, we believe it is reasonable to assume each disappearance lasting at least 2 weeks represented an excursion out of the Talek home range. Analysis of disappearance patterns in 22 male and 24 female hyenas remaining in the home range until they were at least 30 months old revealed the dimorphic pattern apparent in Fig. 3. Assuming each prolonged disappearance indeed represented an excursion out of the Talek home range, young males took such excursions twice as often as did their female peers. The rates at which young males took excursions did not vary significantly with their mothers’ social ranks (Fig. 4A). In contrast, young females who vanished most frequently from the study clan’s home range were the daughters of low-ranking females (Fig. 4B). This dimorphic pattern suggests that males investigated neighboring territories regardless of their competitive ability in the natal clan, whereas females were more likely to explore surrounding territories if their competitive ability in the natal clan was poor. In 1991we began to radio-collar natal males that were at least 26 months of age, in order to track them during the process of dispersal. Following collared individuals enabled us to examine dispersal patterns more carefully. We found that, several months before they dispersed, some males peripheralized themselves such that they spent most of their time near the edges of the clan’s home range. This can be seen in the telemetry data describing sightings of one natal male, MU, during a 1-month period when he was 37 months old (Fig. 5A). Radiotracking also confirmed that animals often travel widely outside of the home range before dispersal. During this exploratory period, males appeared to use the natal clan’s territory as a secure base from which to explore neighboring areas, much like a young primate uses its mother’s body as a base while exploring its surroundings. Figure 5B depicts the pattern of excursions taken by the natal male, SF, during a 1-month period when he was 37 months old. During this time, SF traveled extensively, returning to his natal territory for periods of one to several days between excursions. Other individuals made relatively few excursions before dispersing. The sightings of one such male, Q, are plotted in Fig. 5C.Although he was the same age as MU and SF, Q had already dispersed by the time he was 37 months old, and he was never subsequently tracked to the Talek home range. Q dispersed relatively rapidly, with only 1 month elapsing between the onset of his exploratory excursions and settlement in his new home range. b. When Do Males Disperse? Fourteen natal males that permanently disappeared from Talek were subsequently relocated elsewhere, and were thus known to have dispersed. Fifteen additional natal males disappeared
197
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
2.0-
A a
-
1.5-
d
a
2
*
0.50
v)
5
8
a
E
-
1
\o
2
i
I
B
v)
'28
1.5-
a u x
e,
!3
P
E 1.0-
00
T -
5
I
-
I
10 15 Maternal rank
I
20
I
25
FIG. 4. Relationship between maternal rank and the rate at which juvenile (A) male and (B) female hyenas take excursions from the Talek area. An animal is assumed to be taking an excursion away from the Talek home range if it is not seen there during a 2-week observation period. On average, males take more excursions per 6-month interval than do females. Among females, but not males. excursion rates are significantly correlated with maternal rank. This pattern suggests that young males, all of whom disperse, explore areas beyond the Talek area regardless of their maternal ranks. In contrast, only females of low maternal rank tend to explore potential sites for dispersal.
from Talek in a fashion suggesting they were also engaging in dispersal. That is, they were in good health when last seen, and vanished permanently only after engaging in a series of preliminary excursions out of the Talek
198
LAURA SMALE ET A L
A
FIG. 5 . Sightings of three Talek male hyenas tracked repeatedly via radiotelemetry during November 1991. These males were all 37 or 38 months old at this time, but they were in different stages of dispersal in November 1991. (A) MU had not yet begun to disperse; (B) SF was in the process of dispersing, and explored the home ranges of at least four different neighboring clans during this 1-month interval; (C) Q had already dispcrsed and settled in a clan two home ranges away from Talek. Width of the Talek home range is approximately 10 km. The exact locations of territorial boundaries not abutting the Talek home range are not all known with certainty. The shaded area represents the Talek clan’s home range, and non-shaded areas those of neighboring clans.
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
199
100
Y 80 M
.-
-5
60
-
7
7
-
e,
g
a
20
Age (months) FIG. 6. Frequency distribution representing the ages at which Talek male hyenas dispersed from the natal clan. Filled bars represent males (N = 14) that were relocated after they dispersed. and empty bars represent males (N = 15) that were believed to have dispersed but were not relocated after leaving Talek (see text).
70-
.
6050-
40-
.
. 0
.
* .
.. .
308
20I 5
10
15
20
25
Maternal rank FIG. 7. Ages at which 29 natal male hyenas dispersed from the Talek clan, expressed as a function of their mothers’ social ranks.
200
LAURA SMALE ET AL
area, as described in Fig. 3. The mean age at which these 29 males dispersed was 37.2 2 1.7 months (range 24-62 months; Fig. 6). We found no significant relationship between dispersal age and maternal rank (Fig. 7). Freeliving male Crocuta have viable sperm in their testes at 24 months of age (Matthews, 1939), and captive males complete pubertal development by this time (Glickman et al., 1992; Licht et al., 1992). Thus, free-living males disperse, on average, approximately one year after attainment of reproductive maturity. However, males may remain in the natal area until up to 62 months of age, or 38 months after puberty. We observed no cases in which dispersal occurred prior to attainment of reproductive maturity. c. How Do Males Behave When They Venture into a Foreign Territory? Natal male hyenas on excursion in neighboring territories appear to be keenly interested in resident animals, their odors, and their vocalizations. Data collected by East and Hofer (1991) suggest that dispersing males may attend carefully to long-distance vocalizations because they use these to assess queue length for male reproductive opportunities in neighboring clans. We have observed excursioning Talek males spend hours sniffing grass stalks in neighboring territories, as well as latrines (Kruuk, 1972) and other sites where resident animals were likely to have deposited scent marks. To date we have observed no Talek females engaging in similar behavior. Although Talek males appear to be intensely interested in unfamiliar hyenas, they generally exhibit spontaneous appeasement behavior to all conspecifics they encounter in neighboring territories. In fact, no other hyenas actually need be present in the immediate environment of an excursioning male for him to engage in submissive behavior. For example, we once found GM, a 38-month-old male, loping along through the territory of a neighboring clan. Although he was alone, G M was exhibiting extreme submissive behavior, with his tail lowered between the legs, his ears flattened back against his head, and his lips curled in an appeasement grin. GM maintained this attitude as he loped across the empty plain, stopping periodically to listen, and after about 30 min he went into a bush to sleep, over 4 km from his natal home range. His behavior suggested that he was maintaining a submissive posture in case any unfamiliar hyenas were monitoring his activity in this alien territory. We have observed excursioning natal males interact with animals outside of the Talek home range on numerous occasions. We have seen several excursioning natal Talek males approach, appease, and interact amicably with adult and subadult females. Twice we have even observed excursioning males engage in what appeared to be provisioning behavior toward resident adult females, depositing a food item at their feet. Unless excursioning males attempt to closely approach a den or ungulate carcass on which unfamiliar hyenas are feeding, adult females tend to ignore them. By contrast, although
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
20 1
adult males in neighboring territories sometimes sniff and tolerate excursioning Talek males, they may attack them with extremely intense aggression. We once followed the natal Talek male SF for several hours as he traveled from the Talek territory into that of a neighboring clan. Eventually he entered a bush, apparently to sleep, 2 km into the neighboring clan’s territory. After several minutes, an adult male loped to the bush and entered, a high pitched squeal was heard, and SF emerged running at top speed with the alien male on his heels. Twice during the chase the alien male tackled SF, biting him in an exceptionally intense attack. SF was chased for 2 km, squealing the whole time, until he was back in Talek territory. Within the Talek home range, we have observed numerous interactions between resident hyenas and alien males making excursions into Talek territory. These observations suggestthat prospective immigrantsperceive resident immigrant males to represent a more serious potential threat than that posed by other resident animals. Excursioning alien males and recent immigrants are considerably more likely to be threatened or attacked by previous immigrant males than by resident adult females or their offspring. The intensity of attacks directed at excursioning males is often high, and the appeasement behavior these males exhibit to previous immigrants frequently involves the extreme submissive posturing Kruuk (1972) described as “carpal crawling.” These anecdotal observations suggest that established immigrant males in a clan represent a potentially serious threat to a dispersing male, but that resident females do not. The fact that males exploring new territories often direct extreme submission toward immigrant males suggests that they are aware of this potential threat. However, other observations of males on excursion who interact with alien hyenas without intense agonistic behavior are more difficult to explain. Perhaps the lack of extreme submission in these cases indicates that, prior to our observations of them, the excursioning males have previously interacted with resident clan members, and have already been accepted by them. We know nothing yet about what determines whether a dispersing male is accepted or rejected by members of neighboring clans. d. Where D o Males Go When They Disperse? To date we have recovered 14 Talek natal males after they have settled for several months in neighboring territories. Assuming that neighboring home ranges are of similar size to the Talek clan’s territory, the mean number of home range diameters moved at dispersal was 1.8, and the median was 2.0 (range = 1-5; N = 14). Males moved variable distances during dispersal, ranging from 1 to 30 km from Talek. Although Crocutu are physically capable of traveling distances of over 50 km at intervals as short as 48 hr (Hofer and East, 1993), the distances traveled by dispersing Talek males before settling in new clans averaged less than 10 km. Thus males did not go very far when they dispersed. Although sons of high-ranking females tended to settle in clans located closer to Talek than did sons of lower ranking females, this
202
LAURA SMALE ET AL
relationship was not statistically significant. Some dispersing Talek natal males clearly became integrated into new clans, and remained there for several years, while others were more transient, and soon engaged in secondary and tertiary dispersal. Of 11 dispersing natal Talek males whose fates were known at least one year after dispersal, 7 were alive and 4 had died. By contrast, of 14 natal Talek females born since 1987 whose fates were known when they were at least 48 months old, 13 were alive and 1 had died. All of these females were philopatric. Although this sex difference is not statistically significant (,y2 = 3.29; df = 1; .05 < p < .lo), it suggests that dispersers may suffer higher mortality than do philopatric individuals, as occurs in other mammals (e.g., Van Vuren and Armitage, 1994). 2. Female Hyenas
Although most female Crocutu spend their entire lives in their natal clans (Frank, 1986b; Henschel and Skinner, 1987; Mills, 1990), dispersal by females has been documented at three different African study sites (Henschel and Skinner, 1987; Holekamp et al., 1993; Mills, 1990). The process of female dispersal is strikingly different from that seen when males disperse. Most examples of female dispersal occur in the context of clan fission events, in which several animals leave their natal clan together, and establish a new clan elsewhere (Mills 1990; Holekamp et al., 1993). Mills (1990) described the fission of one small South African Crocuta clan, which split along genetically related lines, with one subordinate matriline breaking away and settling in vacant habitat. Fission of Mills’s (1990) study clan thus resembled troop fission in cercopithecine primates (e.g., Chepko-Sade and Sade, 1979; Prud’Homme, 1991). In our study area in Kenya, we similarly observed a clan fission in which a Talek subgroup containing all members of three relatively low-ranking matrilines dispersed and settled together in vacant habitat adjacent to the parent clan’s territory (Holekamp et al., 1993). Thus, Talek females dispersed as a group, and all moved to the same site. As in Mills’s (1990) study, Talek clan fission occurred along kin lines. However, departing animals included middle- as well as low-ranking matrilines from the Talek clan. Some low-ranking Talek females without living adult female kin did not participate in clan fission. The new clan formed at fission, which we called the Embryo clan, was clearly distinct from the Talek clan. The Embryo clan occupied a region outside of the normal Talek home range, and we observed a clan war in which members of the Embryo and Talek clans fought each other over a kill at the territorial boundary. Embryo clan females were found in the company of adult males, but there was no evidence that this new clan contained any adult females other than those that had originated in the Talek clan. The clan fission observed by Mills (1990) occurred overnight, following
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
203
a major battle between clan matrilines. In contrast, Talek clan fission took place over a period of 7 months, with some individuals leaving abruptly, and others more gradually (Fig. 8). Departures from Talek by adult females and their young were usually preceded by shorter absences, possibly exploratory excursions into new areas. Our data showing a relationship between female excursion rate and maternal rank (Fig. 4B) suggest that some adult females, particularly low-ranking ones, might regularly explore areas outside the Talek clan’s home range, perhaps in search of better feeding opportunities or habitat suitable for dispersal. Whereas all Talek males disperse between 2 and 5 years of age, the seven adult females who emigrated during Talek clan fission ranged in age from 5 to 15 yr. The reproductive condition of dispersing females was highly variable: dispersal occurred in nulliparous as well as multiparous females, and females dispersed when they were lactating as well as when they were not. However, no females dispersed when they had small cubs residing at the Talek clan’s communal den. In fact, two adult females who participated late in the clan fission remained in the Talek area until their cubs were over 6 months old (Fig. 8). Emigration by adult female hyenas, and perhaps other female mammals, might generally be constrained by the presence of dependent offspring too small to move long distances. Some evidence exists that dispersal by female hyenas does not occur exclusively in the context of a clan fission. For example, Tilson and Henschel (1986) reported that a single adult female hyena dispersed and joined 3 adults of unknown sex in southern Africa. In our own study, one old, radiocollared, low-ranking Talek female disappeared from the Talek home range, and was subsequently relocated approximately 25 km from the Talek clan’s territory. This elderly female was never observed to interact with any conspecifics in her new home area. Although we have observed 136 alien males visiting Talek since 1988, we have observed only eight alien females within the Talek clan’s home range boundaries. Most were sighted only once or twice, but one female remained in Talek for several days. Two of the eight alien females were darted and radio-collared in Talek. One was never subsequently relocated, and the other was found repeatedly in the home range of a clan whose territory abuts the Talek range. This latter female has subsequently raised at least two litters there, and to our knowledge she has never returned to Talek. Thus, it appears that she was on a brief excursion to Talek when she was darted. Interestingly, our recent observations of her interactions with other hyenas in her current clan’s home range suggest she is quite low ranking in that clan’s hierarchy. To summarize, many characteristics of dispersal appear to differ for male and female Crocufu.Whereas all males eventually disperse from their natal
204
LAURA SMALE ET AL
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
l
l
l
l
l
1
l
,
1
l
l
l
1
l
1
l
1
l
l
l
~
1 ERSH
2 KB
3 DJ
4 COCH
6 CSL 5MW 7 03
8 OT
9 79 10 UA
E-
1
I
r 4
L
I
-4 I
12 SR 14 Q W 18 BAM 20 EF
21 23s
22 DPS 23 02
[
7 l 1 ES 13 LO
16 BRAD
17 STR 19 17
24 xs 25 BO
I 1w9
1990
loo1
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
205
areas, female dispersal is rare. Males venture forth on excursions alone, and disperse alone (Henschel and Skinner, 1987). In contrast, when females engage in these activities, they are likely to do so with close kin (Mills, 1990; Holekamp et al., 1993). Most dispersing females appear to settle promptly near their natal areas (Mills, 1990 Holekamp et al., 1993),whereas males may roam extensively before settling, and immigrate into clans located at highly variable distances from their natal areas (Mills, 1990; Henschel and Skinner, 1987; K. E. Holekamp and L. Smale, unpublished observations). Female dispersal may require vacant habitat that can sustain a group of breeding females, whereas males usually disperse alone and join existing clans (Henschel and Skinner, 1987; Mills, 1990). There are no unambiguous cases in the literature of an adult female from one clan dispersing and successfully joining another group containing adult females.
C. SEXDIFFERENCES I N PAITERNS OF GROUND SQUIRREL DISPERSAL Male Ground Squirrels Male S. beldingi always disperse from the natal area before mating. Most males depart from their natal areas as juveniles, between 7 and 12 weeks of age, almost 2 years before reaching reproductive maturity. Thus, most male dispersal occurs within a narrow window of time in this species. The proportion of males that disperse as juveniles varies between populations, ranging from 60% to more than 90% (Holekamp, 1984b; Nunes, Zugger, Engh, Reinhart, and Holekamp, in press). All surviving males disperse before the end of the yearling year. Dispersal within cohorts of juvenile males tends to be synchronized with respect to age and calendar date (Holekamp, 1984b; Nunes et al., in press), a tendency that has also been observed among young males of some other mammalian species exhibiting male-biased dispersal (e.g., Cockburn et al., 1985; Hanby and Bygott, 1987; Packer et al., 1995). After natal dispersal, many male S. beldingi engage in postbreeding dispersal movements. Males that have copulated most fre1.
FIG. 8. A depiction of periods during which natal Talek hyenas were present (solid lines) in the Talek home range between July 1988 and May 1991. The social ranks and names of adult females are presented on the left. Related animals (mothers and their offspring and siblings) are connected by vertical lines. Breaks in the horizontal lines represent known deaths (0)or absences from the Talek home range (no symbol). Animals represented in the shaded area in the bottom of the figure disappeared during the period of clan fission, and all except female 27 and her son appeared to disperse as a group at this time. We believe female 27 died during the period of clan fission. Two females (ES and LG) and their offspring returned to Talek 6-8 months after clan fission. Adapted from Holekamp et al. (1993).
206
LAURA SMALE ET AL
quently during the mating season emigrate farthest, whereas adult males unsuccessful at acquiring mates usually do not disperse (Sherman, 1976). Dispersal by juvenile male S. beldingi is accompanied by a suite of changes in nonsocial behaviors (Holekamp, 1986). Juvenile males exhibit increased locomotor activity and exploration near the time that they emigrate. This begins about 4 weeks prior to dispersal and peaks about 2 weeks after the dispersal movement. In addition, around the time of dispersal, juveniles exhibit less fearfulness, as measured in standardized tests based on the “looming silhouette” tests of early ethologists (Tinbergen, 1948). Departure from the natal area by juvenile males is occasionally abrupt, occurring with little apparent exploration of potential new home sites (Holekamp, 1983). However, male dispersal is more commonly preceded by a period of home range expansion and exploration of distant habitat (Fig. 9). Over a period of days or weeks, juvenile males typically make several exploratory forays from the natal area into surrounding habitat. After finding a suitable new home site, males divide their time between the natal area and the prospective new home site, returning to the natal area each night to sleep. Eventually, males stop returning to the natal site. Distances at which juvenile males settle from their natal burrows vary between local habitats (Holekamp, 1984b). For example, the mean minimum distances at which juvenile male dispersers were observed from their natal burrows in 1979-1982 was about 240 m at one of our study sites, but about 330 m at a second site. After dispersing, males have new home areas that do not overlap with those of their brothers (Sherman, 1976; Nunes et al., in press). We never observed males to disperse to areas unoccupied by conspecifics. 2. Female Ground Squirrels
More than 90% of female S. beldingi remain in their natal areas throughout their lives (Holekamp, 1984b;Sherman, 1981).However, a small proportion (< 10%)of each cohort ordinarily disperses (Holekamp, 1984b; Nunes et al., in press; Sherman, 1977). Most young females that disperse leave their natal areas as juveniles, but at slightly later ages than do males (Holein press). For example, in 1993-1994, the median kamp, 1984b;Nunes et d., age of dispersal was 8.0 weeks for males, but 9.5 weeks for females (Nunes et af., in press). Young females appear to make exploratory forays from the natal site prior to dispersing, as do males, and females are sometimes found 100-300 m from their natal burrows as early as 2 weeks after weaning (S. Nunes, unpublished observations). Dispersal distances do not differ appreciably between the sexes (Holekamp, 1984b). The decision to disperse appears not to be as irrevocable in females as it is in males, and young
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
Days after
first emergence
above ground
<
207
mother's home range
&-natal
burrow
7
14
21
28
35
FIG.9. The process by which a typical male Belding's ground squirrel undergoes natal dispersal. The male's home range is shaded. When a male first emerges from his natal burrow, his movements are restricted to the immediate vicinity of the burrow. His home range expands gradually, until by approximately 28 days after emergence, he spends large portions of the day far from the natal burrow, to which he still returns at night. Approximately one week later, the male ceases to return to the natal area in the evening, and the dispersal process is complete. Adapted from Holekamp (1984a).
females sometimes reestablish themselves in their natal areas after briefly (> 1 week) settling at new home sites (Nunes etal., in press). After leaving the natal area, females sometimes utilize home areas overlapping those of littermate sisters that have also dispersed. Of 24 dispersed juvenile females recovered in 1993-1995,lO had sisters that also emigrated as juveniles, and
208
LAURA SMALE ET AL
we observed 8 of these within 5 m of a sister that had also dispersed. Females may thus often maintain the benefits of nepotistic cooperation when they disperse (Nunes et ul., in press).
IV. SEXDIFFERENCES IN THE PROXIMAL CAUSES OF NATAL DISPERSAL Dispersal behavior requires proximal explanation with respect to both the socioecological variables and the physiological events associated with its occurrence. Ecological and social factors experienced in the day-to-day life of the individual animal may influence its decision regarding whether or not to emigrate, and its probability of successful immigration into a new area or social unit. Physiological factors include neuroendocrine, metabolic, and energetic variables that influence the probability or timing of dispersal behavior. The proximal causes of mammalian dispersal have received substantial attention in recent years, but its physiological mediation remains largely obscure. In light of the widespread sex differences observed in patterns of natal dispersal behavior, it seems reasonable t o expect that the proximal mechanisms underlying expression of this behavior should differ for males and females. In this section we review hypotheses suggesting socioecological and physiological causes of mammalian dispersal behavior, and compare these for males and females. We then specifically consider the proximal causes of dispersal in Crocutu and S. beldingi. We emphasize that many of the hypotheses considered here are not mutually exclusive, and that more than one causal agent might concurrently promote dispersal. A. SOCIOECOLOGICAL VARIABLES As PROXIMAL CAUSES OF DISPERSAL Several different ecological variables have been proposed to stimulate mammalian dispersal. In the face of habitat deterioration, an individual may disperse in search of better living conditions (Lidicker and Stenseth, 1992). For example, female bonnet macaques ( M . rudiutu, Ali, 1981), vervets (Isbell et al., 1991), and red kangaroos (Dawson, 1995) were observed to emigrate following massive damage to their home ranges. However, even in stable habitats, individuals may disperse to improve access to such basic environmental resources as food, water, or shelter. Resource abundance may influence dispersal behavior directly, or indirectly through changes in local population density, intensity of resource competition, or both. That is, individuals may base their decisions about whether or not to emigrate on assessment of resource abundance, numbers of conspecifics encountered
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
209
per unit time, or the quantity or quality of their own interactions with conspecifics. In species for which a critical resource required by both males and females is limited in the natal area, dispersal is often monomorphic. For example, in banner-tailed kangaroo rats (Dipodomysspactubilis), dispersal apparently occurs in response to the need for mounds in which individuals can store their seed caches and escape from the elements and from predators (Jones, 1984,1987). In this species both males and females disperse to the nearest available nest mound. Similarly, red squirrels (Tamiasciuris hudsonicus) disperse in response to the need to establish a territory on which a midden is located. Middens contain caches of food necessary for survival during winter. Both male and female red squirrels disperse to the nearest site on which an established midden is located, or on which a new midden can be constructed (Larsen and Boutin, 1994). However, the resource requirements of kangaroo rats and red squirrels are rather unusual, as is the finding that dispersal in both males and females is stimulated by the same proximal factors. Among most solitary mammals studied to date, females are more likely than males to disperse in response to a demand for environmental resources. For example, dispersal patterns in solitary stoats (Mustela erminea) are sexually dimorphic, with females moving significantly shorter distances. Male movements are largely determined by attraction to distant females, whereas females move in response to local densities of other females and competitive interactions with them over food (Sandell and Erlinge, 1989; King, 1989). Similarly, among many gregarious mammals, female dispersal appears to be prompted by a demand for resources or associated competitive interactions. For example, dispersal among cercopithecine primates in the context of troop fission (Dittus, 1988; Prud’Homme, 1991) generally improves the social status of females that were low-ranking animals in their natal troops, and thus improves their ability to access resources (Cheney, 1987). Among gregarious mammals exhibiting male-biased dispersal, males generally do not disperse in search of environmental resources. Perhaps the clearest demonstrations that resource scarcity cannot account for male dispersal in these species come from field experiments in which key resources are artificially made superabundant, yet male dispersal continues to occur at the same rates as in control areas (e.g., Dobson, 1979; Dobson and Kjelgaard, 1985; Nunes and Holekamp, 1996). By contrast, female dispersal in these species can be induced by manipulation of resource abundance (Dobson, 1979; Dobson and Kjelgaard, 1985; Nunes et al., in press). Myriad social variables have been suggested to promote mammalian dispersal behavior. The hypothesis that emigration occurs in response to
210
LAURA SMALE ET AL
conspecific aggression is the longest standing and most popular proximal explanation in the dispersal literature (Charnov and Finerty, 1980; Chitty, 1967,1987;Christian, 1970; Howard, 1960;Stenseth, 1983; Wynne-Edwards, 1962). A related notion proposed by Gauthreaux (1978) is that dominant animals force subordinates to emigrate. Other social hypotheses suggest that individuals disperse because of inadequate formation of social bonds (Bekoff, 1977; Harris and White, 1992), a tendency to avoid members of the natal group (e.g., Armitage, 1962; Downhower and Armitage, 1981), or attraction to members of neighboring groups (reviewed in Pusey and Packer, 1987a). Finally, some observers have suggested that dispersal is promoted by social facilitation, in which the movements of one individual increase the probability that others will disperse (e.g., Sugiyama, 1976). This last hypothesis, however, begs the question of what stimulates the first animal’s dispersal behavior. Dispersers of both sexes may be aggressively driven from their natal areas in species exhibiting a wide variety of social organizations and mating systems. There are three general conditions under which conspecific aggression may cause dispersal. In the first, an animal in its natal group is forced to disperse by the aggressive behavior of a member of a different group. This process, which has been referred to as “abduction,” has been reported for females of some species, but never for males. Female plains zebras (Equus quagga, Klingel, 1974) and hamadryas baboons (Sigg et al., 1982; Stammbach, 1987) are regularly herded away from their natal groups and forced to join the harem of an adult male. The second condition involves aggression associated with a shortage of critical resources (as in pikas, Ochotona princeps, Smith and Ivins, 1983; Whitworth and Southwick, 1984; pocket gophers, Geomys attwaferi, Williams and Cameron, 1984; South American camelids, Franklin, 1983; and various primates, Carpenter, 1940; Sugiyama, 1965; Poirier, 1969). This includes not only species experiencing occasional local population explosions (e.g., deer mice, Peromyscus manicul a m , Fairbairn, 1978), but also species in which population density regularly varies in a cyclical fashion. Thus, for example, aggressive explusion of dispersers has been observed, or inferred from wounding data, at peak population densities of snowshoe hares (Lepus americunus) and various microtine rodents (Boutin, Gilbert, Krebs, Sinclair, and Smith, 1985; Christian, 1970 Gaines and McCleneghan, 1980; Krebs, 1978; O’Donoghue and Bergman, 1992; Windberg and Keith, 1976). Habitat saturation occurs in some species when philopatric behavior by older offspring brings local population density above the carrying capacity of the home range. In such cases, parents may aggressively expel members of a previous litter when a new litter is born (e.g., moose, Alces alces, Cederlund and Sand, 1992; guanacos, Lama guanicoe, Franklin, 1983; deer mice, Savidge, 1974;Chinese
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
21 1
hamsters, Cricetufus griseus, Dasser, 1981). The third general condition under which conspecific aggression regularly promotes dispersal involves turnover in male possession of female groups. Thus, for example, following immigration of a new breeding male into one-male primate groups, natal males may be forcibly driven from the group by the new immigrant (reviewed by Pusey and Packer, 1987a). Similarly, natal lions of both sexes may be aggressively evicted when a new male coalition takes over a pride (Hanby and Bygott, 1987; Pusey and Packer, 1987b). The proximal factors influencing dispersal often differ for males and females, and these differences may be related to sex differences in the selection pressures shaping the evolution of dispersal behavior. For example, the age at which male ponies disperse is positively correlated with the number of peers in the natal band available as social partners (Rutberg and Keiper, 1993). Interactions among young males are thought to improve fighting skills necessary for acquiring and defending mates, and may therefore ultimately enhance male reproductive success. Female ponies, by contrast, typically disperse at the time of first estrus, and consequently avoid mating with their fathers, who are likely to be the stallions of their natal bands (Rutberg and Keiper, 1993). Among white-footed mice (Peromyscus feucopus), females disperse only as far as forced to by competition for vacant home ranges, whereas males leave their natal ranges even in the absence of such competition (Keane, 1990a). Among dwarf mongooses, females are likeliest to disperse when the dominant male in their pack is a close relative, whereas males emigrate at random with respect to the relatedness of the dominant female (Keane, Creel, and Waser, in press). Among polygynous carnivores, interpopulation variance in proportions of females that emigrate often vastly exceeds that observed among males (Waser, 1996), suggesting that female dispersal in these species is more labile, and that it varies in response to local conditions. Among gregarious mammals exhibiting dimorphic dispersal behavior, transfer between groups by members of the dispersing sex often appears to be stimulated by a growing attraction to members of neighboring groups. Emigrants in some species exhibit attraction to same-sexed conspecifics in foreign groups. This has been documented in patas monkeys (Erythrocebus p a m , Gartlan, 1975), gelades (Theropithecus gefada, Dunbar and Dunbar, 1975), and Japanese macaques (Matsumura, 1993). More commonly, however, dispersers express a growing attraction to unfamiliar conspecifics of the opposite sex (e.g., Japanese macaques, Takahata, 1982; olive baboons, Packer, 1979; blue monkeys, Cercopithecus mitus, Henzi and Lawes, 1987; Barbary macaques, M. sylvanus, Paul and Kuester, 1985; sifakas, Propithecus verreauxi, Richard, Rakotomanga, and Schwartz, 1991; ringtailed lemurs, Lemus catta, Sussman, 1992; chimpanzees, Pan troglodytes, Pusey,
212
LAURA SMALE ET AL
1980). Male baboons tend to transfer to groups containing relatively high numbers of cycling females (Packer, 1979; Manzanillo, 1986), and male Barbary macaques that have recently immigrated focus their attention on estrous females in their new groups (Mehlman, 1986). Among lemurs and cercopithecine primates, growing male attraction to unfamiliar females sometimes occurs in conjunction with preferences exhibited by estrous females for consortships and copulations with immigrant over natal males (Cheney, 1983; Henzi and Lucas, 1980; Manson and Perry, 1993; Packer, 1979; Paul and Kuester, 1985; Pereira and Weiss, 1991; van Noordwijk and van Schaik, 1985). Similar findings have been obtained in studies of rodents (e.g., Hoogland, 1982; Hill, 1974) and carnivores (reviewed by Waser, 1996). Male cercopithecine primates sometimes immigrate to groups containing greater numbers of females than did their natal groups (e.g., rhesus macaques, M. mufatfa,Drickamer and Vessey, 1973). More often, however, males move to groups containing numbers of females similar t o those in their natal groups (reviewed in Pusey and Packer, 1987a). This is also true of male sciurid rodents (e.g., prairie dogs, Cynomys ludovicianus, Hoogland, 1995; S. befdingi, Holekamp, 1986) and lions (Pusey and Packer, 1987b). These observations suggest that, in many species, the proximal cause of sexually dimorphic dispersal may be closely associated with mate choice. Among gregarious mammals in which dispersal is strongly male biased, the male’s decision about whether or not to emigrate might be related to mate choice in any of three different ways. First, males might find females in their natal groups unattractive as potential mates. In this case, males should express no sexual interest in resident females and all males should disperse voluntarily to groups containing unfamiliar females. Second, natal males might find resident females attractive, without these females being sexually receptive to them. Under these circumstances, males should attempt to court and mate with resident females, but disperse when their sexual overtures are rebuffed. This sequence of events has been documented in ringtailed lemurs (Pereira and Weiss, 1991; Jones, 1983), and indicates that female choice mediates male dispersal in this species. Third, if matings with relatives do not occur, then regardless of the mechanism by which this happens, natal males will be at a significant disadvantage relative to immigrants with respect to the number of potential mates available to them in their natal group. In this case, some males with few or no surviving female relatives might exhibit philopatric behavior, but most males would be expected to disperse. This scenario has been described for Barbary macaques (Kuester, Paul, and Arnemann, 1994; Paul and Kuester, 1985) and yellow baboons (Papio cynocephalus, Alberts and Altmann, 1995). Half of the natal male yellow baboons observed over a 20-year period engaged in moderate to extensive sexual activity with resident females before they
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
213
emigrated (Alberts and Altmann, 1995). Natal males garnered 20% of the available consort days of females that were not their maternal kin, and only 2% of consort days of females that were their maternal kin. These data suggest that fewer females were available to natal than immigrant males as potential mates, because intrafamilial matings were extremely rare. Furthermore, both natal and other males were likelier to disperse when numbers of adult males in their groups were high, suggesting that males were sensitive to the presence of male competitors for mates. Finally, males successful at obtaining consortships dispersed later than did less successful males, a pattern consistent with the hypothesis that both female availability and receptivity influence male dispersal patterns in this species. B. PHYSIOLOGICAL MEDIATION OF DISPERSAL BEHAVIOR 1.
Endocrine Mechanisms
Little is known about the physiological processes that underlie mammalian dispersal. Hormones have been implicated in the control of longdistance movements in invertebrates, fish, amphibians, and birds (Caldwell, 1974; Caldwell and Rankin, 1972; Johnson, 1969; Meier and Fivizzani, 1980; Dingle, 1996), and might also mediate mammalian dispersal. The possibility that mammalian dispersal might be influenced by gonadal steroid hormones in particular is suggested by the sexual dimorphism in dispersal, and by its common occurrence around the time of puberty. Gonadal steroids can promote sex differences in behavior in two general ways. First, hormones can act during fetal or neonatal life to shape the development of behaviors that are not expressed until later in life. Second, hormones can act directly to induce more immediate behavioral changes in the mature organism. Phoenix, Goy, Gerall, and Young (1959) chose the terms “organizational” and “activational,” respectively, to describe these two modes of hormone action. Either or both might promote sexually dimorphic mammalian dispersal. Temporal coincidence of dispersal with puberty is consistent with the hypothesis that rising levels of gonadal steroids might activate dispersal behavior. Puberty is characterized by dramatic increases in circulating levels of gonadal steroid hormones that stimulate a variety of behavioral changes (Wilson, 1992). A close temporal association between dispersal and puberty has been observed in male and female mountain gorillas, female ponies, male and female dwarf mongooses, male muskrats (Ondatra zibethicus), female chimpanzees, and males of many cercopithecine primate species (Berger, 1986, 1987; Errington, 1963; Harcourt et al., 1976; Harcourt, 1978; Matsumura, 1993; Pusey, 1980; Pusey and Packer, 1987a;Rood, 1987,1990).
214
LAURA SMALE ET A L
However, the activational hypothesis has yet to be empirically tested in any of these species. When dispersal occurs well before puberty (as in many sciurid rodents and marsupial mice, Holekamp, 1984a; Cockburn et al., 1985), the behavior is unlikely to be mediated by activational effects of gonadal hormones. In some of these cases, androgens might act during the perinatal period to promote dispersal later in development. Many other nonsexual, dimorphic behaviors that are predictably exhibited by all males of a species during specific stages of behavioral development are organized by androgens (Beatty, 1979, 1992). For example, perinatal androgen exposure promotes masculine patterns of play behavior in many mammals (e.g., in Norway rats, Rattus norvegicus, Thor and Holloway, 1984, 1986). Theoretically, androgens might similarly organize dispersal behavior in species in which all males disperse, including Crocuta, S. beldingi, lions, ponies, various cercopithecine primates, and Antechinus species. To date, this hypothesis has been experimentally tested only in S. beldingi (see later discussion). Androgens probably do not exert an organizational influence on the more labile dispersal behavior of species such as mountain gorillas and Columbian ground squirrels ( S . columbianus), in which some males disperse while others remain in their natal areas. In species exhibiting seasonal reproduction, dispersal frequently occurs during the breeding period (Berger, 1986, 1987; Matsumura, 1993; Pusey and Packer, 1987a; Sprague, 1992). Reproduction in seasonal breeders is often regulated by circannual or photoperiodic timing mechanisms (Bartness, Powers, Hastings, Bittman, and Goldman, 1993;Zucker, 1988;Zucker, Lee, and Dark, 1991). Thus, it is conceivable that the timing of dispersal, puberty, and reproduction in some seasonal breeders might be regulated physiologically by a common circannual or seasonal timing mechanism. For example, a circannual oscillator that regulates the timing of changes in reproductive status might also regulate temporal variation in behaviors such as exploration and locomotion, or in attraction to unfamiliar conspecifics. Seasonal occurrence of “dispersal restlessness” has been documented in birds (Ritchison, Belthoff, and Sparks, 1992), and might also occur in mammals. 2. Metabolic and Energetic Events
Dispersal to a new home area and immigration into a new social unit may entail both a substantial energy expenditure and a decrease in energy intake. In this section, we explore the possibility that the availability of energy might regulate the timing of dispersal behavior in some mammals (Nunes and Holekamp, 1996).Animals possess physiological and behavioral mechanisms to partition available energy according to priorities that ensure
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
215
survival and optimize long-term reproductive success (Wade and Schneider, 1992). Dispersal is typically characterized by increased locomotor and exploratory behavior (Holekamp, 1986; Messier, 1985a; Mills, 1990; Rood, 1987),so dispersal may be energetically costly. Dispersal can also be dangerous. Dispersers tend to be more vulnerable to predation and other sources of mortality as they explore unfamiliar areas (Messier, 1985b; Metzgar, 1967; Packer et al., 1991; Snyder, Jenson, and Cheney, 1976; Van Vuren and Armitage, 1994). Moreover, dispersers often encounter high levels of conspecific aggression when attempting to establish themselves in new areas (Ballard, et al., 1987; Holekamp, 1986; Pusey and Packer, 1987a; Rood, 1987). Thus, individuals may need to reallocate time spent foraging to vigilant and defensive behaviors during the dispersal period, resulting in decreased energy intake. Body mass, body fat reserves, and physical condition might therefore represent important determinants of when dispersal occurs, particularly in mammals large enough to be able to store energy as fat without compromising locomotor ability. Effects of manipulations of variables like fat reserves should be especially clear in hibernating species in which growth and dispersal must occur within a narrow window of time each year. In these and other species, individuals may delay departure from the natal area until their energy reserves are adequate to support the potential energetic costs of dispersal. This hypothesis is supported by the observation in many sciurid rodents that smaller males tend to disperse at later ages than do their larger peers (Balph and Stokes, 1963; Downhower and Armitage, 1981; Dunford, 1977; Holekamp, 1984a; Rayor, 1985). C. PROXIMAL CAUSES OF DISPERSAL IN BELDING’S GROUND SQUIRRELS I.
Physiological Mediation
In our attempts to elucidate the physiological mechanisms underlying dispersal behavior in S. beldingi, we have focused on the possibility that hormones and metabolic fuels might both play important, perhaps interactive, roles. The possibility that testosterone might promote dispersal via an organizational effect was tested by injecting females on the day of birth with testosterone (T) and monitoring their subsequent behavior. To do this, we trapped 11 pregnant females in areas peripheral to our study sites, and returned them with their newborn litters to our study areas after their female pups received injections of either T or oil vehicle alone (Holekamp et al., 1984). We later determined whether T-treated females exhibited male-typical dispersal behavior. To evaluate the possibility that T might promote dispersal through an activational effect, we repeatedly measured plasma T titers in a captive group of juvenile males throughout their first summer, and evaluated androgen levels in relation to the time juvenile
216
LAURA SMALE ET AL
males ordinarily disperse in nature. We also performed castrations or sham operations on 78 free-living 4-week-old males to determine whether removal of the primary source of androgens would inhibit dispersal behavior. Data obtained from these manipulations indicated that the natal dispersal behavior of male S. beldingi may be organized, but not activated, by androgens (Holekamp et al., 1984). Levels of circulating T in S. beldingi were undetectable throughout the juvenile summer (Holekamp et al., 1984), which is the period during which most male dispersal occurs. Furthermore, castration of juvenile males at weaning delayed, but did not alter the probability of, dispersal behavior. This indicates that elevated levels of circulating androgens are not required to activate dispersal behavior. However, females treated at birth with testosterone later exhibited masculine patterns of dispersal (Fig. lo), suggesting that perinatal androgen exposure promotes the development of male dispersal behavior in this species (Holekamp et a/., 1984; S . Nunes, unpublished observations). We do not know if this effect is mediated directly by testosterone acting on androgen receptors, or through its aromatization to estradiol, which then acts on estrogen receptors. To evaluate the hypothesis that body mass or body fat influences the timing of dispersal behavior, we conducted food provisioning experiments in the wild from 1993 to 1995. We gave extra food (peanuts, sunflower seeds, and peanut butter) daily to adult female S. be/dingi in provisioned areas during gestation and lactation, and to their young from weaning to 10 weeks of age. We provided supplemental food in plywood feeding boxes placed near maternal or natal burrows. Provisioned areas were separated from control sites by at least 100 m to ensure that maternal burrows of control mothers were located more than one home range from the nearest feeding box. We alternated provisioned and control areas between years to control for local environmental effects. Along with these provisioning manipulations, we evaluated body fat content of juveniles using a device that analyzes body composition of small animals (EM-SCAN, Springfield, IL). This device uses total body electrical conductivity to nondestructively estimate lean (fat-free) mass of live animals (Walsberg, 1988). We formulated a species-specificequation (Scott, Grant, and Evans, 1991) for S. beldingi to predict lean mass from readings of total body electrical conductivity, and calculated body fat content as the difference between lean mass and total body mass (Nunes and Holekamp, 1996). For analysis of body fat composition, we livetrapped squirrels, anesthetized them, measured body composition, and then released them at their capture sites after they recovered from anesthesia. The results of this series of experiments suggest that early T exposure determines whether S. beldingi emigrate, while body mass and body fat
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
h
E v
0
---
c
0 T.@eatedlernales
Hrasldsnt males Mbrothers 01 T-healetl
5
6
7
8
9
1 0 1 1
217
12
Age (weeks) Fic. 10. Mean minimum distances from their natal burrows at which testosterone-treated female ground squirrels ( N = 12), their brothers ( N = 5), resident females ( N = 158) and resident males ( N = 93) were relocated during their first summer. Reprinted from Holekamp el al. (1984).
influence when juvenile males disperse (Nunes and Holekamp, 1996). Juvenile male squirrels provided with extra food were heavier than unprovisioned males throughout the juvenile summer, and dispersed at younger ages (Fig. 11). Dispersing males tended to have greater percentages of their bodies composed of fat than did same-aged males remaining in their natal areas (Fig. 12). Most dispersal behavior by juvenile males occurred when body fat content was rapidly increasing, which suggests that natal dispersal may be constrained by the competing energetic demands of prehibernation fat deposition in this species. Furthermore, abbreviation of the juvenile active season by heavy snowfall late in the spring accelerated mass and fat acquisition in juvenile males, and suppressed their dispersal behavior until
218
LAURA SMALE ET AL
10
-
.4 8!2 F B 6-
A .
m
E * 0
j
4-
I
\
2-
z Y
I,
6
j
A
d
lb
Age (weeks) FIG.11. Ages at which provisioned and control male ground squirrels dispersed. Provisioned males dispersed significantly earlier than control males ( x 2 = 13.44,df = 4, p = .009). Reprinted from Nunes and Holekamp (1996).
the following year (S. Nunes, unpublished observations). This reinforces the notion that there is an energetic conflict between dispersal and hibernation, and suggests that a circannual timing mechanism might also be involved in the regulation of dispersal behavior in this species. Energetic constraints on behavior might be more pronounced in S. beldingi than in other mam-
16
1
1
1
i a 4
38
7
8 9 Age (weeks)
10
FIG. 12. Body far content during the 7-to 10-week age intervals for juvenile males dispersing during each interval and males still residing in their natal areas. Animals were more likely to disperse if more of their body mass was composed of fat. Asterisk indicatesp < .05. Samples include both provisioned and control males. Reprinted from Nunes and Holekamp (1996).
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
219
mals because they live at high elevations, and experience one of the longest hibernation periods known among mammals (Sherman and Morton, 1984). Nevertheless, because animals need to judiciously partition energy among a number of life history events, body mass and fat stores might influence the timing of dispersal behavior in a diverse array of mammals. The physiological mechanism through which fat stores influence dispersal in S. beldingi is not known. However, the mechanism might respond directly to sugars or fats, the molecules that ordinarily fuel metabolic processes, or to a specific signal correlated with fat reserves. Metabolic fuel availability has been shown to regulate timing of other events that are potentially energetically costly to mammals, such as estrus in golden hamsters (Mesocricetus auratus). Female hamsters treated with metabolic fuel inhibitors become anestrous regardless of their body fat content or access to food (Schneider and Wade, 1989, 1990a,b; Wade, Schneider, and Friedman, 1991). Energetic regulation of estrus in hamsters appears to be mediated by neural mechanisms designed to detect levels of circulating metabolic fuels (Berriman, Wade, and Blaustein, 1992; Schneider, 1992; Wade and Schneider, 1992). The neural pathway through which energy stores are monitored appears to include the area postrema of the hindbrain (Schneider and Zhu, 1994). A mechanism similar to that mediating effects of metabolic fuel availability on hamster estrous cycles might regulate dispersal behavior in mammals in which energy reserves influence the likelihood that dispersal will occur. Dispersal behavior might be initiated when the energetic demands of growth and fat deposition decrease to the point where metabolic fuels become available to support the dispersal process. The physiological trigger for the initiation of dispersal might be metabolic fuel availability. Another intriguing possibility is that a chemical signal from adipose tissue might inform the CNS about the status of body fat reserves (Campfield, Smith, Guisez, Devos, and Burn, 1995), and thereby influence dispersal behavior. A peptide, leptin, which regulates body fat stores by altering feeding behavior and activity, has recently been isolated in laboratory mice (Halaas et al., 1995; Pellymounter et al., 1995). This chemical signal, or a similar one, might influence dispersal behavior in some mammals. Circulating levels of such a substance might vary with adipose tissue mass, and dispersal behavior might be triggered when titers reach a seasonally appropriate threshold. 2. A n Integrative Model of the Physiological Mediation of Dispersal in S. beldingi Here we present a model suggesting interactions between endocrine and metabolic events in the regulation of natal dispersal behavior in S. beldingi
220
LAURA SMALE ET AL
PERINATAL ANDROGENS
STORES
Reproduction
Feeding
Exploration/
.) Locomotion
4
‘ear
Q
Attraction to ~~~~~~~B
DISPERSAL FIG. 13. A model of endocrine and energetic regulation of dispersal in S. heldingi. According to this model, perinatal androgens promote the development of a neural system mediating dispersal. Perinatal androgens also make this system responsive to physiological signals indicating levels of available metabolic fuels or body fat content. Consequently, when these signals exceed a seasonally appropriate threshold, changes in behavior occur that lead to dispersal. A variant of this model might also describe regulation of dispersal in other species in which all males disperse. In species in which dispersal occurs shortly after puberty, the system promoting dispersal might be activated by pubertal increases in circulating androgens. Finally, information about energy availability should be conveyed, not only to the system regulating dispersal behavior, but also to systems regulating feeding behavior and, in adult animals, reproduction.
(Fig. 13). This model offers testable hypotheses regarding the physiological mediation of dispersal in a hibernating mammal. However, the energetic problems posed by dispersal are not unique to hibernators, and many components of the model might therefore apply to other mammals as well. Nevertheless, we anticipate that the model will not be applicable to many species of small mammals in which body size prohibits significant storage of energy in the form of fat. Because ground squirrels depend almost exclusively on body fat for energy while hibernating, acquisition of adequate fat reserves prior to hibernation is critical for survival. Hibernation in ground squirrels is generally preceded by a period of rapid mass and fat acquisition, and accompanied by a phase of gradual loss of mass and fat. This pattern of change in mass and fat occurs in predictable circannual cycles (Pengelley and Fisher, 1963). Body mass
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
221
and body fat in ground squirrels are defended at seasonally appropriate levels that are continuously adjusted over the course of the annual cycle (Mrosovsky, 1976; Zucker and Boshes, 1982;Bachman, 1993). The potential energetic demands of dispersal are considerable, and may conflict with those of prehibernation fattening. In our model, dispersal behavior is inhibited when body fat reserves are below seasonally appropriate thresholds, and dispersal occurs only when fat reserves exceed these levels. This mechanism ensures that dispersal will not jeopardize an individual’s ability to fatten sufficiently for hibernation. In mammals whose body mass and body fat content do not exhibit predictable seasonal fluctuations, dispersal might be triggered when body fat content exceeds an ontogenetically appropriate level (e.g., to optimize growth), ensuring that dispersers have adequate energy reserves to accommodate the potential costs of emigration. We suggest that perinatal androgen exposure might organize (1) a maletypical predisposition to disperse, (2) patterns of dispersal-related locomotor and exploratory behavior, and (3) a tendency to disperse in response to changes in metabolic fuel levels or other chemical signals associated with energy availability. Thus, dispersal behavior should be initiated in response to a physiological signal correlated with the size of body fat depots (Fig. 13). Furthermore, the response threshold to this signal should change as a function of an interaction between season and age. This model, in which dispersal behavior depends on perinatal androgenic organization of its key physiological components, might usefully predict not only the dispersal behavior of S. beldingi, but also that of other species, such as California ground squirrels ( S . beecheyi), vervets, Japanese macaques, spotted hyenas, and lions, in which all males disperse from their natal areas. Our model attempts to explain the physiological bases of natal dispersal behavior in juvenile male S. beldingi. However, those few juvenile females that disperse from the natal area should face the same energetic problems that confront males. Thus, the timing of dispersal by young females might be influenced by body fat content, as it appears to be in males. Females might also delay departure from the natal area when fat reserves are below a seasonally appropriate threshold. Nevertheless, the patterns of natal dispersal and key life history events differ between males and females, and the specific manner in which fat-related variables regulate dispersal behavior might also differ. For example, female S. beldingi disperse at later ages and greater body masses than do males. Moreover, females begin reproducing as yearlings, whereas males do not reproduce until 2 years of age (Sherman and Morton, 1984). It is possible that the energetic demands of dispersal might impose greater potential costs on the future reproductive success of juvenile females than that of juvenile males, and consequently the putative body fat threshold for dispersal might be higher for juvenile
222
LAURA SMALE ET AL
females than juvenile males. Thus, we suspect that the energetic regulation of dispersal behavior might be fundamentally similar for juvenile male and female S. beldingi, but differ in some of its finer details.
3. Socioecological Variables Influencing Dispersal in Ground Squirrels Since 1979, we have tested hypotheses suggesting a suite of social and ecological variables as proximal causes of natal dispersal behavior in S. beldingi (Holekamp, 1986; Nunes and Holekamp, 1996). The ecological variables we examined were ectoparasite load, food availability, and availability of nest sites. We also tested hypotheses suggesting that social variables promote dispersal in this species. For example, we examined the possibilities that dispersers were aggressively expelled from their natal areas, that animals emigrated in efforts to avoid conspecifics, and that individuals dispersed as a result of social facilitation. Three basic patterns provided evidence that violated the predictions of most hypotheses tested. First, immigration equaled emigration in our study areas, arguing against resource demand, ectoparasite load, and juvenile avoidance of conspecifics as proximal explanations for dispersal. Second, despite differences between years and between study sites with respect to population density, sex ratios, ectoparasite loads, food availability, and weather conditions, all surviving males always dispersed. Third, although juvenile males and females experienced similar social and ecological stimuli, their emigration rates differed greatly. For example, males and females experienced similar rates of conspecific aggression, but only males dispersed, suggesting that they were not driven out of the natal area. It should be noted, however, that males and females may respond differently to a particular socioecological variable. For example, males might be attracted to unfamiliar conspecifics, which could lead them to disperse, whereas females might avoid unfamiliar animals, a tendency that could promote philopatry. This hypothesis received some support from our observation that males settled only in areas inhabited by unfamiliar squirrels, and often rejected apparently suitable habitat containing abundant food and nest sites, but n o conspecifics. The striking sexual dimorphisms in S. befdingi's dispersal behavior suggest that the proximal causes of dispersal may differ for males and females (Dobson, 1979; Holekamp, 1986). Therefore, after concluding that male dispersal was not promoted by demand for environmental resources, we tested hypotheses suggesting that female dispersal was promoted by shortage of food, nest sites, or space for territorial defense. Only about 8% of female S. beldingi dispersed regardless of nest-site abundance in different natal areas (Holekamp, 1986). In our food provisioning experiments (Nunes and Holekamp, 1996) we found that the presence of supplemental food actually increased the probability of emigration from the natal area by
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
223
juvenile females. This result came as a complete surprise, and was not seen in juvenile males (Nunes et al., in press). That is, although provisioning advanced the time of male dispersal, it had no effect on the probability that males would disperse. All surviving male S. beldingi in both provisioned and control areas emigrated from the natal area by the end of their yearling summer. By contrast, young females in provisioned areas were significantly more likely to emigrate than were control females. Only about 8%of control females dispersed by the end of the juvenile and yearling summers, whereas over 30% of provisioned females dispersed by the end of the juvenile active season, and over 40% emigrated by the end of the yearling summer. Of the 17 juvenile females that dispersed from provisioned areas in 1993-1995, none settled in provisioned areas. By contrast, 21% (8 of 38) of provisioned males immigrated into other provisioned areas. Females dispersed from areas containing superabundant food to areas containing less food. Thus, food demand was clearly not the proximal cause of female dispersal. However, density of S. beldingi was approximately two times higher in provisioned than in control areas, and adult females exhibited significantly higher rates of vigilant and aggressive behavior in provisioned areas. Thus, food provisioning resulted in crowding, and increased local competition for space, and it appeared that females dispersed in response to these conditions. Apparently females emigrated to increase access to space in which to defend maternal territories during the following year. Similarly, house mice (Mus musculus) disperse in response to changes in social structure induced by food provisioning (Maly, Knuth, and Barrett, 1985). M . musculus given extra food at a central feeding station exhibit increased population density, more rigid social hierarchies, and increased rates of dispersal (Noyes, Barrett, and Taylor, 1982; Maly et al., 1985). D. PROXIMAL CAUSES OF DISPERSAL IN SPOITED HYENAS 1. Physiological Mediation
We have made no efforts to experimentally investigate the physiological mediation of dispersal in Crocuta. However, as is true in other species, the temporal characteristics of hyena dispersal are consistent with some physiological hypotheses but not others. Only reproductively mature hyenas of either sex were known to disperse from Talek and survive elsewhere. This observation raises the possibility that changing hormone levels associated with puberty might promote dispersal. Dispersal by adult females occurred at ages ranging from 5 to 15 years, and occurred when females were in a variety of reproductive conditions (pregnant, lactating, and neither pregnant nor lactating). Since plasma levels of circulating steroid hormones vary significantly with reproductive condition in this species (Frank, David-
224
LAURA SMALE ET AL
son, and Smith, 1985; Gombe, 1985; Licht et al., 1992), activation of female dispersal by one or more steroids seems unlikely. However, for male Crocuta, both activational and organizational effects of steroid hormones remain viable possibilities as physiological causes of dispersal. Crocuta are exposed to peculiar endocrine environments during early development (Glickman et al., 1992; Licht ef al., 1992), and many sex differences are reversed in this species from mammalian norms. Nevertheless, some behavioral sex differences are quite ordinary in Crocuta. For example, young males exhibit significantly higher rates of play-mounting than do females (L. Smale and K. E. Holekamp, unpublished observations). In other mammals, higher rates of play-mounting by males are promoted by perinatal T exposure (e.g., Beatty, 1979). The presence of this and some other relatively common mammalian sex differences in Crocuta raises the possibility that male hyenas might be exposed to more T than females at a very early stage of development. It is therefore conceivable that early androgen exposure might organize male dispersal behavior in Crocuta as it appears to do in S. beldingi. It also remains possible that rising T levels at puberty activate dispersal in male hyenas, as dispersal only occurs among postpubertal males. The possibility that the timing of dispersal among male hyenas is influenced by stores of available energy is suggested by the observation (Fig. 7) that variance in dispersal ages is least among sons of high-ranking females, and greatest among sons of low-ranking females. In our studies of variables influencing fertility among adult female Crocuta, we have found that members of the alpha matriline are far less vulnerable to fluctuations in the food supply than are lower ranking females (Holekamp et al., 1996). A similar vulnerability may promote variance in dispersal ages among sons of low-ranking females. The importance of energy availability to dispersing males is also suggested by our observations of natal males returning briefly to Talek during times of energetic stress. Natal males in poor physical condition (wounded, ill, or emaciated) have returned briefly to Talek after absences of up to 8 months (K. E. Holekamp and L. Smale, unpublished observations). When they return to the natal area these males reassume ranks associated with those of their mothers and thus enjoy better access to food resources than would be possible in neighboring territories. This suggests that perhaps natal males energetically compromised in their new territories return to Talek specifically t o take renewed, temporary advantage of their maternal rank in feeding competitions with members of the natal clan.
2. Socioecological Variables Influencing Dispersal in Spotted Hyenas a. Female Hyenas. Although dispersal by female Crocuta has been described in the literature (Mills, 1990; Tilson and Henschel, 1986; Holekamp
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
225
et al., 1993). it seldom occurs, and appears to represent an exception to the general hyena rule of female philopatry. Among spotted hyenas, the proximal causes of dispersal behavior by females include aggression and competition related to food. The number of animals in a spotted hyena clan appears to be limited by food availability (Mills, 1990). Feeding is highly competitive in this species, and hierarchical rank relationships result in unequal distribution of food resources among clan members (Kruuk, 1972; Tilson and Hamilton, 1984; Frank, 1986a,b; Mills, 1990). The emigration of female members of one low-ranking matriline from a small South African Crocura clan (Mills, 1990) was preceded by a steady increase in the number of adult females in the population over a period of years, and more immediately by an intense battle between members of two resident matrilines. Thus, conspecific aggression appeared to be the most immediate cause of female dispersal in Mills’s study population, and that aggression, in turn, was probably caused by increasing population density without concurrent increase in the clan’s food supply. In contrast to Mills’s observations, we saw no major conflict preceding the Talek clan fission observed in 1989-1990, nor any unusual occurrence of fresh wounds on Talek females prior to their departure from the clan (Holekamp er al., 1993). Talek clan fission began when prey availability was at its low point for 1989, and ended with the annual increase in prey availability in 1990. Although this suggested that low prey availability influenced clan fission, the same paucity of game occurs each year in Talek, at approximately the same time, yet the fission event we observed was unique. However, the overall size of the Talek clan rose progressively from May 1988 until the start of clan fission in late 1989. As the population increased, so did intensity of intraspecific feeding competition at ungulate kills (Holekamp et al., 1993). Overall, the mean number of individuals able to dominate those Talek females that dispersed increased in the year preceding fission from 20 t o 33. Thus, a substantial increase in the number of higher ranking competitors in the parent clan, occurring in conjunction with the annual period of low prey availability and an opening in nearby habitat, was the likeliest proximal explanation for the female emigration we observed (Holekamp et al., 1993). Among spotted hyenas, members of low-ranking matrilines represent the segment of the population most susceptible to fluctuations in food availability (Frank, 1986a,b; Hofer and East, 1989; Holekamp et al., 1996). Because low-ranking females have poorest access to food for themselves and their offspring, they should be most likely to respond to increasing competitive pressures by emigrating to vacant habitat, when this is possible. However, some of the Talek clan’s lowest ranking females did not emigrate at clan fission. Low-ranking females who failed to disperse had no living adult female kin, whereas all dispersing females emigrated with at least one
226
LAURA SMALE ET AL
adult female relative. Perhaps accompaniment by adult relatives enhances a female’s probability of survival in new habitat, or favorably influences her social status in the new clan. b. Mule Hyenas. In contrast to emigration by female hyenas, male dispersal does not appear to be caused by either aggression or demand for food (Mills, 1990; Henschel and Skinner, 1987). Males are not aggressively ousted from their natal clans, nor do they fall in rank before dispersal (Mills, 1990; Henschel and Skinner, 1987; Holekamp and Smale, 1993). As long as male Crocutu remain in their natal clans, they retain social ranks associated with those of their mothers (Holekamp and Smale, 1991; Holekamp and Smale, 1993; Smale et d., 1993), and can thus dominate all other lower ranking hyenas, including those of much larger body size. All dispersing natal males are socially dominant to the immigrant males that remain behind. What then is the proximal cause of dispersal by male spotted hyenas? Two hypotheses seem most plausible. Male Crocutu might emigrate because they are attracted to conspecifics in neighboring clans, or because estrous females in the natal clan are unwilling to accept them as mates. Like male cercopithecine primates and sciurid rodents, male Crocutu immigrate to neighboring groups containing adult females that are not necessarily more numerous than in the natal clan, but to groups containing females that are unfamiliar. Although some natal males have courted Talek females very intensively, most seemed less strongly interested in resident females than did immigrant males. From 1988 through 1993, we observed 581 courtship interactions among Talek hyenas. During this time, 19 natal Talek males reached reproductive maturity and subsequently remained in Talek for at least 6 months, and 13 of these (68%)participated in courtship interactions with adult females. We have observed as many as 10 males concurrently follow, bait (Kruuk, 1972), and court Talek females, and such an entourage of suitors often contained both natal and immigrant males. Nevertheless, whereas at any given time 25-40% of the adult males in the clan were natal males, these accounted for only 12%of the observed courtship interactions. Thus, although adult natal Talek males court resident females, and sometimes attempt to mount estrous females, their sexual ardor often appears to be somewhat less than that of immigrant males. Of the 16 complete copulations we have observed to date, none were achieved by natal adult males. Similarly, Mills (1990) reported that natal male Crocutu in his study population never mated, regardless of their maternal rank. It may be that many natal males are not adequately motivated to carry their courtship efforts through to complete copulation with resident females. However, our observations in Talek also suggest that female choice plays an important role.
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
227
Natal males are socially dominant to all immigrants, and they can displace immigrants from close spatial proximity to estrous females. Furthermore, courted females appear to respond with less intense aggressive behavior to sexual overtures from natal than from immigrant suitors. Nevertheless, female behavior suggests they perceive the sexual overtures of immigrants as courtship, but those of natal males as nuisance behaviors to which they respond as they do when brushing off a fly. Although we have not yet quantified female responses to male sexual overtures, our impressions of courtship and copulation are that females prefer immigrants over natal males as mates. If this is true, then female choice overrides outcomes of male-male competition in this species, and gives the relatively low-ranking immigrant males a substantial reproductive advantage over natal males. Since 1990 we have collected DNA samples from all Talek mothers, cubs, and potential fathers, including both adult natal and immigrant males. Paternity analysis is currently under way with these samples. If females prefer immigrants over natal males as mates, then natal males should rarely, if ever, father cubs, and natal males should be motivated to disperse in order to access sexually receptive females. This would suggest that dispersal by male Crocuta is mediated by female mate choice, as it is in ringtailed lemurs (Pereira and Weiss, 1991; Jones, 1983).
E. SUMMARY Dispersal may be a highly predictable and relatively inflexible ontogenetic event, or it may be labile, and vary in response to social and ecological factors (e.g., Hanby and Bygott, 1987; Henschel and Skinner, 1987; Holekamp, 1984a; Holekamp et al., 1993; Nunes et al., in press). Spotted hyenas and Belding’s ground squirrels differ conspicuously in body size, diet, social organization, habitat, and life history, and yet these two species exhibit striking parallels in their patterns of dispersal. All surviving males of both species emigrate from the natal area, male dispersal occurs at predictable stages of ontogenetic development, and appears to be motivated by attraction to unfamiliar conspecifics in neighboring areas. Dispersal behavior might be necessary in order for males to come into contact with attractive females willing to accept them as mates. Female dispersal is uncommon in both species, is less predictable in its temporal occurrence, and occurs in response to intense competition for environmental resources. Thus, female dispersal tends to be far more labile, and more responsive to variation in local environmental conditions. We suggest the dimorphic dispersal patterns observed in Crocuta and S. beldingi may be representative of a phenomenon widespread among polygynous, group-living mammals.
228
LAURA SMALE ET A L
V.
CONSEQUENCES OF SEXDIFFERENCES I N DISPERSAL BEHAVIOR
A. CONSEQUENCES OF DIMORPHIC MAMMALIAN DISPERSAL Dispersal has important consequences for both individual dispersers and the populations into which they immigrate, and these effects may express themselves either during the lifetime of the disperser or over long periods of evolutionary time. Several authors have previously considered the effects of dimorphic dispersal behavior on the evolution of social behavior (e.g., Brandt, 1992; Michener, 1983; Sherman, 1977; Trivers, 1971; Waser and Jones, 1983; Wrangham, 1980). Sexually dimorphic dispersal behavior results in differences in the degree to which adult males and females interact with close kin, and maintain long-term relationships with nonkin members of the natal group. This in turn leads to the evolution of sex differences in social behavior via differential operation of kin selection and reciprocal altruism among dispersers and philopatric individuals. The ancestral condition within the class Mammalia was a solitary existence (Eisenberg, 198l), and dispersal occurred among both males and females. Social groups more complex than male-female pairs or motherinfant associations could form only when some individuals remained in the natal area through adulthood. Thus, philopatry is a central characteristic of complex social groups (Waser and Jones, 1983), and the initial behavioral change underlying the evolutionary transition from a solitary to a social way of life is the inhibition of dispersal. Philopatry results in animals sharing a home range, which sets the stage for the evolution of social behaviors leading to both nepotistic interactions among kin and reciprocal altruism. Female ground squirrels and cercopithicene primates, for example, do not ordinarily disperse, but rather remain in the natal area where they share a common home range. In these species, distinct groups have formed, which contain matrilines of adult females and their offspring (Sherman, 1977; Pusey and Packer, 1987a). Among chimpanzees, dispersal is female biased and males are philopatric (Pusey, 1979), so male group members are consequently more closely related than are males in most gregarious mammals (Goodall, 1986). This has set the evolutionary stage for the formation of unusually strong social bonds between male chimpanzees, and the complex patterns of shifting alliances seen in this species (de Waal, 1982). Dispersal may also have important consequences for the evolution of nonsocial behavior. One of the challenges confronting every disperser is the need for orientation in an unfamiliar landscape. Dispersers must often navigate through vastly larger areas than do philopatric individuals. For example, dispersing ground squirrels may travel as far as 10 km, whereas philopatric conspecifics rarely travel farther than 0.5 km from the burrow
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
229
in which they were born (Quanstrom, 1971; Sherman, 1977; Holekamp, 1984b). As they traverse unfamiliar territory, dispersers must quickly learn enough about their new environments to forage effectively, and to identify routes of escape from predators. Dispersers must learn key geographical features in both their natal areas and in the areas to which they immigrate, whereas philopatric individuals need only learn the former. Thus members of the dispersing sex must possess the capacity to form considerably more numerous and more elaborate spatial memories than those required of philopatric animals. This may place demands on spatial memory that far surpass those imposed on philopatric individuals. Sex differences in spatial memory have been documented in a variety of rodents in which dispersal is male biased, particularly among polygynous members of the genus Microtus (Gaulin and Fitzgerald, 1986, 1989). In species in which this sex difference in spatial memory has been documented, the hippocampus, a forebrain structure that plays a critical role in spatial memory, is larger in males than in females (Jacobs, Gaulin, Sherry, and Hoffman, 1990). These sex differences in hippocampal volume and spatial memory have been attributed to sex differences in home range size (Jacobs et al., 1990; Sherry, Jacobs, and Gaulin, 1992), but the demands associated with dispersal may also contribute to the evolution of this sex difference. The more immediate, short-term consequences of dispersal for members of populations into which male mammals immigrate include increased probability of infanticide and spontaneous abortion, and social upheaval associated with incorporation of new individuals into an existing social group (reviewed by Brandt, 1992). Alberts, Sapolsky, and Altmann (1992) recently documented the relationship between circulating hormone levels and social stress induced by the immigration into a free-living baboon troop of one unusually aggressive male, Hobbes. The social upheaval that followed Hobbes’s arrival was associated with a near doubling of plasma cortisol levels in members of the troop. This dispersal incident was also associated with a striking change in immune system function, with many troop members showing significantly decreased lymphocyte counts after Hobbes arrived (Alberts etal., 1992). Thus, immigration by males may have significant physiological consequences for resident animals. It is currently unknown whether immigration by females engenders similar physiological changes. Dispersal can result in profound physiological and behavioral changes for immigrants as well as for members of the populations into which they move. For example, when Hobbes’s blood was sampled shortly after his immigration into his new troop, he was found to have extremely high levels of circulating glucocorticoids, as well as a very low lymphocyte count. Interestingly, Hobbes’s blood contained a testosterone concentration substantially higher than that of any other male in the troop, a pattern thought
230
LAURA SMALE ET AL
to be the result of his extreme aggression and his rise to the top of the male hierarchy. Male dwarf mongooses attempting to immigrate may face intense aggression from resident males for more than a month before being accepted into a new group (Rood, 1987), and recent immigrant males often exhibit wounding, poor body condition, and low testosterone levels (Creel, Wildt, and Montfort, 1993). An extreme example of the physiological consequences of dispersal can be seen in members of the marsupial genus Anfichinus, in which males all mate shortly after dispersing, and die shortly after mating (Cockburn et al., 1985). This mortality apparently results from prolonged gluconeogenic mobilization of body protein, a process that sustains males through dispersal, male-male competition, and mating (Lee and Cockburn, 1985), but which also results in high levels of plasma glucocorticoids, gastrointestinal ulcers, and immune system suppression (Bradley, McDonald, and Lee, 1980). These elevations in glucocorticoid levels are triggered by increases in circulating androgens, and are exacerbated by the stress of aggressive interactions between males that occur during the mating season. In the examples described above, dispersal is associated with unusually high levels of aggression, but it is not unreasonable to expect that such stressors and associated physiological events might be common consequences of dispersal. In addition to its effects on physiological processes, dispersal may also have profound social consequences for individual emigrants. At dispersal, an individual moves into a new environment inhabited by a new cast of characters with whom it may have to learn t o interact according to an entirely new set of rules. The transition to this new social environment can have important effects on many aspects of the animal’s subsequent behavioral development. For example, as young blind mole rats disperse, their preferred mode of intraspecific communication shifts from vocalization to seismic signaling, which then remains the preferred mode throughout the remainder of the animal’s life (Rado et al., 1991). When dispersal is sexually dimorphic it can have a significant impact on the development of behavioral differences between the sexes. Prospective immigrants may face extreme hostility, particularly from same-sexed individuals, and they may be disadvantaged in contests with resident animals fighting on their own territories (reviewed in Brandt, 1992). In such cases, cooperation among dispersers may enhance chances of successful immigration, as occurs among male cheetahs, lions, dwarf mongooses, and vervet monkeys (Caro, 1994; Cheney and Seyfarth, 1983; Waser, in press). For example, among the lions of the Serengeti ecosystem, all males disperse, and spend a period of time without a territory (Hanby and Bygott, 1987; Packer, 1986; Pusey and Packer, 1987b; Schaller, 1972). A t dispersal, males maintain alliances with brothers if they have them, or initiate alliances with unrelated males. These
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
231
young male lions must then assess the chances of effectively challenging other alliances in a bid to take over a pride of females and the territory they occupy. In the event of failure, males must be capable of adopting a completely different strategy, that of nomadism, and seek out receptive females while avoiding alliances of potentially dangerous territorial males. Most female lions are philopatric, and even those females that do disperse never immigrate into new social units. Therefore, females face none of the social consequences of dispersal that are routinely confronted by males (Pusey and Packer, 1987b). The social consequences of dispersal have been particularly well described in group-living primates, in which aggressive interactions with conspec,ific males are commonly an immediate consequence of male dispersal (reviewed in Pusey and Packer, 1987a). At dispersal, male monkeys break social bonds with individuals in the natal group, and must often overcome hostility from conspecifics elsewhere in order to be accepted as a member of a new group. Moreover, immigrants often must adopt new strategies for acquiring critical resources, and assume new rank positions in hierarchically organized societies. This has been described, for example, in olive baboons in which males disperse from their natal troop directly into a new troop at 8-9 years of age (Packer, 1979; Packer el af., 1995). One predictable consequence of dispersal for a young male olive baboon is that he will be attacked by more established immigrant male members of a troop (Packer, 1979). Resident males also frequently chase females away from a male newcomer, blocking him from having contact with them. When a young male baboon disperses, he is socially dominant to all of the females he encounters, presumably because of his larger body size. However, in general, the young immigrant assumes a low social rank relative to other males; as an immigrant male gets older, his social rank typically increases and then decreases (Packer, 1979). Although this is one common pattern, we emphasize that there is a great deal of variability. Some male baboons rise to the top of the male hierarchy immediately after joining a new troop, whereas others may take months merely to be accepted as relatively lowranking troop members (Alberts er af., 1992; Altmann, Hausfater, and Altmann, 1988; Hamilton and Bulger, 1990; Smuts, 1985; Strum, 1987). Another important consequence of dispersal can be an increase in mating behavior. Prior to dispersal, many male baboons engage in disproportionately low levels of mating behavior, apparently due to both their own relative lack of initiative in this regard, and females’ lack of interest in natal males (Alberts and Altmann, 1995; Packer, 1979). A female mating preference for immigrant males over natal males has also been observed in Japanese macaques (Enomoto, 1974) and ringtailed lemurs (Pereira and Weiss, 1991). Although the mechanisms involved have not been as well
232
LAURA SMALE ET AL
studied in carnivores as in primates (but see Keane er al., in press), it is nevertheless generally true that social carnivores emigrate from groups offering few reproductive opportunities into groups presenting many (Waser, 1996). B. CONSEQUENCES OF DISPERSAL IN SPOITEDHYENAS 1. Consequences of Male Dispersal
Undoubtedly the most dramatic and immediate consequence of dispersal for the male spotted hyena is an abrupt decline in social status. Whereas natal males maintain social ranks just below those of their mothers, immigrant males initially occupy social ranks at the bottom of the new clan’s hierarchy (Frank, 1986b; Frank, Glickman, and Zabel, 1989; Henschel and Skinner, 1987; Holekamp and Smale, 1991; Kruuk, 1972; Mills, 1990; Smale et al., 1993). In aggressive interactions, immigrant males are completely subordinate to natal animals, and can even be dominated by small cubs (Smale et al., 1993). This is associated with the sudden isolation of the disperser from his sources of social support. When a male hyena leaves his natal territory, he leaves a social network that supports him in a rank position from which he can dominate all animals from matrilines lower ranking than his own. The male attains this position through a process involving maternal interventions and coalitionary support from both kin and nonkin (Holekamp and Smale, 1991,1993; Jenks, Weldele, Frank, and Glickman, 1995).This is a process in which the male learns which individuals he can dominate, and his subordinates learn to appease him. When the male leaves the natal clan, he leaves all of this behind. He is alone when he disperses, without allies, while the new hyenas he encounters are likely to be on their own turf, inhabiting a social world in which they are embedded in a network of potential allies. At dispersal, the male hyena assumes a social rank that is not only below those of the natal members of his new clan, but also below those of all the immigrant males that arrived before him. Among immigrant males, the primary determinant of social status is the relative length of tenure in the new clan (Fig. 1). That is, the time of a male’s arrival in a new clan is the best predictor of his dominance status relative to that of other immigrant males. Two Talek males that both dispersed to the same neighboring clan provided us with a unique opportunity to evaluate the impact of an immigrant’s arrival time on his social rank. Prior to dispersal, SH was subordinate to PB, as was SH’s mother to PB’s. However, SH dispersed 4 months before PB. Upon PB’s arrival in the new clan, SH was able to dominate him, and has consistently dominated PB for several years. Thus, arrival order in the new clan appears to be a more critical determinant of rank relations among
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
233
immigrant males than are previously held rank relationships in the natal clan. The observed rank reversal between SH and PB was presumably caused by social factors that systematically favor established immigrants over newcomers. Established immigrants appear to enjoy an initial territorial advantage over newcomers, and this may be reinforced by a pattern of coalitionary support that favors animals on the winning side during agonistic interactions (Zabel, Glickman, Frank, Woodmansee, and Keppel, 1992). T o illustrate the dire social consequences of dispersal for male Crocuta, let us consider a “worst case” scenario. This would involve the son of the Talek alpha female dispersing to a neighboring clan of comparable size, roughly 65 hyenas. Before dispersal, the alpha son would be able to displace from food all members of the natal clan except his mother and younger siblings. Thus, his rank in the overall hierarchy of the natal clan would most likely be third or fourth. Upon arrival in his new clan after dispersal, his rank would be lower, not only than those of all natal animals, but also than those of all previous immigrants. Therefore, at dispersal the overall rank of the alpha son would immediately plummet from fourth to sixtyfifth. Ranks of sons of lower ranking females would not decline quite so severely at dispersal, but because all resident animals dominate immigrants, even sons of the lowest ranking female would experience a substantial decline in rank, as well as reduced access to food. In spite of this seemingly disastrous change in social status, all male hyenas disperse. We believe that the negative consequences of dispersal with respect to feeding competition and social rank are outweighed by benefits accruing from two other likely consequences of dispersal among male Crocuta. First, if females choose t o mate with immigrants over natal males, then male mating opportunities should increase as a consequence of dispersal. Second, if males are most attracted to unfamiliar females, then their access to attractive females should increase at dispersal. Thus, the proximal cause of male dispersal might involve both the refusal of local females to select natal males as mates, and a preference on the part of young males for the less familiar females residing in other clans. 2. Consequences of Female Dispersal In contrast to male emigrants, dispersing female Crocuta either become nomadic or form a new social unit in which they are subordinate to fewer adult females than in their former clan. Among cercopithecine primates, participation in troop fission also often appears to improve ranks of previously subordinate females (Cheney, 1987). In the case of the hyena clan fission we observed at Talek, the lowest ranking of the adult females in the Embryo clan was dominated by only 6 other adult females, whereas
234
LAURA SMALE ET AL
she had previously been dominated by 21 other adult females in the parent clan. Thus, with respect to feeding competition, a female’s situation improves at dispersal, whereas a male’s deteriorates. In general, low-ranking female hyenas may face very grim competitive prospects in their natal clan (Frank et ul., 1995), and most likely remain philopatric only because other options do not exist. When dispersal options become available, low-ranking females may take them, as we observed during the 1989-1990 clan fission. C. CONSEQUENCES OF DISPERSAL IN BELDING’S GROUND SQUIRRELS As occurs in Crocuru, the consequences of dispersal behavior appear to differ for male and female S. beldingi. Perhaps most important, dispersal by female squirrels results in improved access to space in which to establish maternal territories, whereas male dispersal results in mating with unrelated females. When males reach reproductive maturity, they are usually isolated from their close female kin. This spatial separation between male S. beldingi and their sisters and mothers reduces the likelihood of incestuous matings, and most likely represents the selective force responsible for the evolution of male-biased dispersal in this species. Sexually dimorphic patterns of dispersal also have important consequences for social development in S. beldingi. Because males immigrate into areas where they are surrounded by unrelated animals, their opportunities for nepotism are limited relative to those available to females. Most females remain in their natal areas, surrounded by close kin, and when they do disperse, females are more likely to settle near sisters that have also dispersed. This difference sets the stage for the evolution of nepotistic interactions among females but not males. In fact, after dispersal, male S. beldingi rarely give alarm calls, whereas adult females commonly call in the presence of close female kin (Sherman, 1977). In addition, female relatives sometimes cooperatively defend territories during periods of gestation and lactation (Sherman, 1977). In contrast to females, male S. beldingi are isolated from relatives after they disperse (Sherman, 1976; Nunes et al., in press), and confront a relatively competitive social milieu. Young males that have dispersed are attacked and chased more frequently, and over greater distances, than are males still residing in their natal areas (Holekamp, 1986). Among ground squirrels, and possibly hyenas, attacks on immigrant males may be related to the observation that recent immigrant males sometimes commit infanticide in their new areas (Sherman, 1981). VI. DIRECTIONS FOR FUTURE RESEARCH
Dispersal results in an abrupt and dramatic discontinuity in the behavioral development of many mammals. Dimorphic dispersal behavior places males
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
235
and females onto divergent developmental trajectories. For those interested in understanding mechanisms underlying the development of sex differences, dispersal behavior represents a vast, unexplored tract of fertile ground. To date, researchers studying mammalian dispersal have generally been field workers trained in behavioral ecology and evolutionary biology. The sex differences revealed by these workers now raise important questions in the areas of neuroendocrinology, physiology, and developmental psychology. Since Frank Beach (1948,1975) formalized the study of behavioral endocrinology, workers in this field have made impressive progress in the laboratory toward elucidating the neural and endocrine mechanisms promoting sex differences in behavior. A focus for many years on simple behaviors in rats revealed how various hormones influence nervous system areas specialized to control behaviors such as the lordosis reflex (Pfaff, 1980) and reflexive components of male copulatory behavior (reviewed in Nelson, 1995). More recently, behavioral endocrinologists have expanded their scope of enquiry to include a wider variety of animal species, more complex components of sexual behavior, and various nonsexual behaviors (e.g., Becker, Breedlove, and Crews, 1992;Nelson, 1995). This expanded research focus has revealed new patterns and new principles of hormone-behavior interaction. We believe the time might now be ripe for a further expansion of focus to include questions about the causes and consequences of dispersal behavior. The pervasive sex differences in the patterns and probability of dispersal suggest that hormones may play an important role in its mediation. This hypothesis can be evaluated with hormonal manipulations at different developmental stages (e.g., Holekamp er al., 1984; S. Nunes and K. E. Holekamp, unpublished observations). One obvious and important question is whether dispersal is mediated by the pubertal rise in T in species in which males leave home around the time of puberty. If a role for T is implicated, this would raise new questions about precisely how the hormone influences dispersal. Our work with Belding’s ground squirrels raises questions regarding how perinatal hormones interact with lipid stores to influence male dispersal behavior, as well as about the nature of the pathway through which energy stores influence the behavior. It is tempting t o speculate, for example, about a role for the newly discovered peptide, leptin, in this process. It is also possible that glucocorticoids, hormones that play a fundamental role in energy metabolism, might influence the timing of dispersal behavior. Finally, an important but completely unexplored issue is the role that hormones might play in the mediation of dispersal in species in which females are more likely to disperse than males.
236
LAURA SMALE ET AL
The development of a viable laboratory model of mammalian dispersal might permit the investigation of its physiological substrates in a manner that could complement field investigations. Under the controlled conditions of the laboratory, both the external environment and the physiology of individual dispersers could be manipulated to evaluate effects of single variables, as well as interactions among them. For example, one could scrutinize relationships among dispersal, hormones, and variables associated with lipid stores, and one could begin to elucidate the brain mechanisms underlying such relationships. One intriguing area for future research involves study of the physiological consequences of dispersal. Anecdotal accounts from the field suggest that dispersal may result in elevated glucocorticoid levels and depressed immune system function among both emigrants and members of social groups into which they immigrate (e.g., Alberts et af., 1992). It is currently unknown whether dispersal is generally stressful and, if so, what socioecological or physiological variables might attenuate this stress. Here, too, availability of a laboratory model of dispersal would be extremely useful. Many important unanswered questions remain regarding the decision rules and other cognitive processes involved in dispersal. For example, we know virtually nothing about how animals monitor resource availability outside their natal areas, nor about how great resource differentials must be before individuals decide to disperse. We still know little about what information is gathered during exploratory forays outside the natal area, and how this might differ between the sexes. Males and females might attend to different features of the new environment. Among gregarious mammals, we know little about what determines which new group a particular individual is likely to join. It also remains a mystery as to why one individual visits a new group only briefly whereas another individual arriving at the same time spends the remainder of its life in this same new group. It would be interesting to examine effects of early experience, such as a male’s social rank in his natal group, on the latency and probability of his becoming socially integrated into a neighboring group, as these may profoundly influence his reproductive success. Another set of unanswered questions pertains to the relationship between mate choice and dispersal behavior. Among many polygynous mammals, males may disperse either because they prefer unrelated females as mates, because familiar females will not mate with them, or both. Tests of these hypotheses demand study of mate choice behavior of both sexes in conjunction with careful monitoring of dispersal. Kin-recognition abilities have been well documented in many mammals (e.g., Holmes, 1986a,b; Holmes and Sherman, 1982), and several recent studies have focused on the relationship between mate choice, kinship, and dispersal (e.g., Alberts and Altmann,
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
237
1995; Keane, 1990a,b; McGuire and Getz, 1981; Pereira and Weiss, 1991). However, no direct link has yet been established between outcomes of male-female sexual interactions and the probability of male dispersal. In addition, it is likely that male traits other than relatedness render them differentially attractive to females as potential mates, and these should also be considered in relation to male dispersal behavior. For example, Meikle, Kruper, and Browning (1995) recently found that adult male house mice (Mus musculus) born to undernourished mothers are unattractive to estrous females. Future studies addressing the question of whether, and how, mate choice influences probability and timing of dispersal have the potential t o relate proximal and ultimate explanations for dispersal behavior in productive new ways. Finally, study of the consequences of dispersal should lead to new insights concerning the mechanisms promoting the development of sex differences in behavior. The hyena example reminds us that the development of different sexually dimorphic behaviors may be interrelated in surprising and complicated ways. In this case, whatever mechanism promotes sex differences in dispersal behavior also appears to have important consequences for the development of sex differences in aggressive behavior. This example serves as a warning, reminding us not to think of dimorphic behaviors in isolation, as though each was mediated by its own discrete control mechanism. It might be quite common that a mechanism evolving in response to selection for one sexually dimorphic behavior pattern functions importantly in the control of another. VII. SUMMARY In this chapter we examine the patterns, causes, and consequences of sex differences in mammalian dispersal. We focus on polygynous groupliving mammals, with particular attention to spotted hyenas and Belding’s ground squirrels, with which we have worked for several years. Although sex differences in dispersal patterns are most pronounced among gregarious mammals, these have also been documented in species in which both sexes regularly disperse. In most polygynous mammals males and females differ with respect both to the probability that they will disperse, and to their patterns of dispersal in space and time. In most species, males are likelier to disperse than are females, males travel farther, and male dispersal often occurs within a fairly narrow window of time during ontogenetic development. This is true in both hyenas and ground squirrels. When male hyenas disperse they join new social groups, whereas females usually emigrate without immigrating into an existing social unit. Among ground squirrels,
238
LAURA SMALE ET AL
too, when females disperse they often settle near kin, whereas males never do. Differences in the patterns of male and female dispersal reflect sex differences in the underlying causes of dispersal behavior. Female dispersal is often promoted by factors related to resource availability. Low-ranking female hyenas disperse when the natal group is unusually large, prey availability is low, and there is vacant habitat available to sustain a new social group. Female ground squirrels similarly disperse when the density of animals in the natal area is unusually high, and resource competition unusually intense. All male hyenas and ground squirrels disperse, regardless of resource availability. Male dispersal in these species appears to be motivated instead by a growing attraction to unfamiliar conspecifics, and in spotted hyenas, by lack of success in courtship interactions with resident females. Both hormones and availability of fat or other metabolic fuels might function in the physiological mediation of dispersal. Correlational data suggest that activational effects of androgens might promote male dispersal in some species. No male hyenas leave home prior to the age at which testicular hormones begin their pubertal rise. For male ground squirrels, field experiments provide evidence that perinatal exposure to testicular hormones promotes dispersal later during ontogeny. In addition, the accumulation of fat stores necessary to sustain animals through hibernation has been shown to influence the timing of dispersal in this species. Experimental manipulations that increase body fat content advance the time at which male dispersal occurs. We propose a model in which perinatal testosterone promotes the development of a mechanism whereby a physiological signal associated with fat availability triggers dispersal behavior in Belding’s ground squirrels. Like the patterns and causes of dispersal behavior, the consequences of dispersal also differ for males and females. Dimorphic dispersal abruptly sets males and females on divergent ontogenetic trajectories and exposes them to entirely different social worlds. This is most striking in spotted hyenas, in which female dispersal leads to an increase in social rank, while male dispersal leads to a precipitous decline in social rank. Dispersal by males places them in proximity to unrelated females, while female dispersal improves their access to environmental resources. In general, the study of dispersal should enhance our understanding not only of the evolution of sex differences in behavior, but also of the neuroendocrine and developmental mechanisms mediating behavioral sex differences. Acknowledgments We thank the Office of the President of Kenya for permission to conduct our hyena research. We also thank the Kenya Wildlife Service, the Narok County Council, and the Senior Warden
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
239
of the Masai Mara National Reserve for their cooperation. We thank the following individuals for their excellent assistance in the field: Kenya: S. M. Cooper, C. 1. Katona, N. E. Berry, K. Weibel, M. Durham, J. Friedman, G. Ording, T. H. Harty, and P. Garrett; California: A. Engh, P. Zugger, K. Reinhart, R. Ojerio, D. Kent, Y. Toda, A. Sheikh, L. Bissett, and V. Dorm This work was supported by NSF Grants BNS8706939, BNS9021461, IBN9296051, and IBN9309805, by a fellowship to L. Smale from the American Association of University Women, and by fellowships to K. E. Holekamp from the Searle Scholars P r o g r a m h e Chicago Community Trust and the David and Lucille Packard Foundation. Finally, we are very grateful to Peter Waser for his comments on an earlier draft of the manuscript.
References
Alberts, S. C.. and Altmann. J. (1995). Balancing costs and opportunities: Dispersal in male baboons. Am. Nat. 145, 279-306. Alberts, S. C., Sapolsky, R. M., and Altrnann, J. (1992). Behavioral, endocrine, and innumological correlates of immigration by an aggressive male into a natural primate group. Horm. Behuv. 26, 167-178. Ali, R. (1981). The ecology and social behavior of the Agastyamali bonnet macaque (Macaca rudiuru dihcru). Unpublished doctoral dissertation, University of Bristol. Altmann, J. (1974). Observational study of behavior: Sampling methods. Behaviour 48, 227-267. Altmann, J., Hausfater, G., and Altmann, S. A. (1988). Determinants of reproductive success in savannah baboons, Papio cynocephalus. In “Reproductive Success” (T. H. CluttonBrock. ed.). pp. 403-418. University of Chicago Press, Chicago. Andelman. S. (1986). Ecological and social determinants of cercopithecine mating patterns. In “Ecology and Social Evolution: Birds and Mammals” (D. I. Rubenstein and R. W. Wrangham, eds.), pp. 201-216. Princeton University Press, Princeton, New Jersey. Armitage, K. B. (1962). Social behavior of a colony of yellow-bellied marmots (Marmora flaviventris). Anim. Behuv. 10,319-331. Bachman, G. C. (1993). The effect of body condition on the trade-off between vigilance and foraging in Belding’s ground squirrels. Anim. Behav. 46,233-244. Ballard, W. B., Whitman, J. S., and Gardner, C. L. (1987). Ecology of an exploited wolf population in south-central Alaska. Wddl. Monogr. 98, 1-53. Balph, D. F., and Stokes, A. W. (1963). On the ethology of a population of Uinta ground squirrels. Amer. Midl. Narur. 69, 106-126. Barash, D. P. (1973). The social biology of the Olympic marmot. Anim. Behav. Monogr. 6, 173-245. Bartness, T. J., Powers, J. B., Hastings, M. H., Bittman, E. L., and Goldman, B. D. (1993). The timed infusion paradigm for melatonin delivery: What has it taught us about the melatonin signal, its reception, and the photoperiodic control of seasonal responses. J . Pineal Res. 15, 161-190. Beach, F. A. (1948). “Hormones and Behavior.” Paul B. Hoeber, New York. Beach, F. A. (1975). Behavioral endocrinology: An emerging discipline. Amer. Sci. 63,178-187. Beatty, W. W. (1979). Gonadal hormones and sex differences in nonreproductive behavior in rodents: Organizational and activational influences. Horm. Behav. 12, 112-163. Beatty. W. W. (1992). Gonadal hormones and sex differences in nonreproductive behaviors. I n “Handbook of Behavioral Neurobiology, Volume 11: Sexual Differentiation” (A. A. Gerall, H. Moltz, and I. L. Ward, eds.), pp. 85-128. Plenum Press, New York.
240
LAURA SMALE ET AL
Becker, J. B., Breedlove, S. M., and Crews, D. (1992). “Behavioral Endocrinology.” MIT Press, Cambridge, Massachusetts. Bekoff, M. (1977). Mammalian dispersal and the ontogeny of individual behavioral phenotypes. Am. Nut. 111,715-732. Berger, J. (1986). “Wild Horses of the Great Basin.” University of Chicago Press, Chicago. Berger, J. (1987). Reproductive fates of dispersers in a harem-dwelling ungulate: The wild horse. In “Mammalian Dispersal Patterns: The effects of Social Structure on Population Genetics” (D. B. Chepko-Sade and Z. T. Halpin, eds.), pp. 41-54. University of Chicago Press, Chicago. Berriman, S. J., Wade, G. N., and Blaustein, J. D. (1992). Expression of Fos-like proteins in gonadotropin-releasing hormone neurons of Syrian hamsters: Effects of estrous cycles and metabolic fuels. Endocrinology 131,2222-2228. Boutin, S., Gilbert, B. S., Krebs, C. J., Sinclair, A. R. E., and Smith, J. N. M. (1985). The role of dispersal in the population dynamics of snowshoe hares. Canad. J. Zool. 63,106-115. Bradbury, J. W., and Vehrencamp, S. L. (1977). Social organization and foraging in emballonurid bats: 11. Parental investment patterns. Behav. Ecol. Sociobiol. 2, 19-29. Bradley, A. J., McDonald, I. R., and Lee, A. K. (1980). Stress and mortality in a small Marsupial (Antechinus stuartii, Macleay). Gen. Comp. Endoc. 40, 188-200. Branch, L. C., Villareal, D., and Fowler, G. S. (1993). Recruitment, dispersal, and group fusion in a declining population of plains vizchacha (Lagostomus maximus; Chinchillidae). J. Mammal. 74, 9-20. Brandt, C. A. (1992). Social factors in immigration and emigration. In “Animal Dispersal: Small Mammals as a Model” (N. C. Stenseth and W. Z. Lidicker, eds.), pp. 96-141. Chapman and Hall, New York. Caldwell, R. L. (1974). A comparison of the migratory strategies of two milkweed bugs. In “Experimental Analysis of Insect Behavior” (L. Barton-Brown, ed.), pp. 304-316. Springer-Verlag, The Hague. Caldwell, R. L., and Rankin, M. A. (1972). The effect of juvenile hormone mimic on flight in the milkweed bug, Oncopelfusfasciatus. Gen. Comp. Endocrinol. 19, 60-605. Campfield, L. A., Smith, F. J., Guisez. Y.,Devos, R., and Burn, P. (1995). Recombinant mouse OB protein: Evidence for a peripheral signal linking adiposity and central neural networks. Science 269,546-549. Caro, T. M. (1994). “Cheetahs of the Serengeti Plains.” University of Chicago Press, Chicago. Carpenter, C. R. (1940). A field study in Siam of the behavior and social relations of the gibbon (Hylobates lar). Comp. Psych. Monogr. 16, 1-212. Cederlund, G. N., and Sand, H. K. G. (1992). Dispersal of subadult moose (Alces alces) in a nonmigratory population. Canad. J. Zool. 70, 1309-1314. Charles-Dominique, P. (1978). Ecologie et vie sociale de Nandinia binotata (Carnivora: viverrides): comparaison avec les prosimiens sympatriques du Gabon. Terre Vie 32,477-528. Charnov. E. L., and Finerty, J. P. (1980). Vole population cycles: A case for kin selection? Oecologia 45, 1-2. Cheney, D. L. (1983). Proximate and ultimate factors related to the distribution of male migration. In “Primate Social Relationships: An Integrated Approach” (R. A. Hinde, ed.), pp. 241-249. Blackwell, Oxford. Cheney, D. L. (1987). Interactions and relations between groups. In “Primate Societies” (B. B. Smuts, D. L. Cheney, R. M. Seyfarth, R. W. Wrangham, and T. T. Strusaker, eds.), pp. 267-281. University of Chicago Press, Chicago. Cheney, D. L., and Seyfarth, R. M. (1983). Nonrandom dispersal in free-ranging vervet monkeys: Social and genetic consequences. Am. Nut. 122,392-412.
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
24 1
Chepko-Sade, B. D., and Halpin, Z. T. (Eds.). (1987). “Mammalian Dispersal Patterns: The Effects of Social Structure on Population Genetics.” University of Chicago Press, Chicago. Chepko-Sade, B. D., and Sade, D. S. (1979). Patterns of group splitting within matrilineal kinship groups. Behav. Ecol. Sociobiol. 5, 67-86. Chitty, D. (1967). The natural selection of self-regulatory behavior in natural populations. Proc. Ecol. SOC. Australia 2, 51-78. Chitty, D. (1987). Social and local environments of the vole Microtus rownsendii. Cunad. J. 2001.65,2555-2566. Christian, J. J. (1970). Social subordination, population density, and mammalian evolution. Science 168,84-90. Clutton-Brock, T. H. (1989). Female transfer and inbreeding avoidance in social mammals. Nature (London) 337,70-72. Cockburn, A., Scott, M. P., and Scotts, D. J. (1985). Inbreeding avoidance and male-biased dispersal in Antechinus supp. (Marsupiala: Dasyuridae). Anim. Behav. 33, 908-915. Creel, S. R., Wildt, D. E., and Montfort, S. L. (1993). Aggression, reproduction and androgens in wild dwarf mongooses: A test of the challenge hypothesis. Am. Nat. 141, 816-823. Dasser, V. (1981). Dispersal tendencies and social behavior of young Chinese hamsters (Criceletus griseus). Behaviour 78, 1-20, Dawson, T. J. (1995). “Kangaroos: Biology of the Largest Marsupials.” Comstock, Ithaca, New York. Delemarre-van de Waal, H. A., Plant, T. M., van Rees, G. P., and Schoemaker, J. (Eds.). (1989). “Control of the Onset of Puberty 111.” Excerpta Medica, New York. de Waal, F. (1982). “Chimpanzee Politics.” Harper and Row, New York. Dingle, H. (1996). “Migration: The Biology of Life on the Move.” Oxford University Press, Oxford. Dittus, W. P. J. (1988). Group fission among wild toque macaques as a consequence of female resource competition and environmental stress. Anim. Behav. 36, 1626-1645. Dobson, F. S. (1979). An experimental study of dispersal in the California ground squirrel. Ecology 60, 1103-1109. Dobson, F. S. (1982). Competition for mates and predominant juvenile dispersal in mammals. Anim. Behav. 30, 1183-1192. Dobson, F. S., and Kjelgaard, J. D. (1985). The influences of food resources on population dynamics in Columbian ground squirrels. Canad. J. Zool. 63, 2095-2104. Douglas-Hamilton, I., and Douglas-Hamilton, 0. (1975). “Among the Elephants.” Collins, London. Downhower, J. F., and Armitage, K. B. (1981). Dispersal of yearling yellow-bellied marmots (Murmora flaviventris). Anim. Behav. 29, 1064-1069. Drickamer, L. C., and Vessey, S. H. (1973). Group changing in free-ranging male rhesus monkeys. Primates. 14, 359-368. Dunbar, R. 1. M., and Dunbar, E. P. (1975). “Social Dynamics of Gelada Baboons.” S . Karger, Basel. Dunford, C. (1977). Behavioral limitation of round-tailed ground squirrel density. Ecology 58,1254-1268. Dunnett, C. C. (1964). A field study of local populations of the brush-tailed opossum Trichosurus vulpeculu in eastern Australia. Proc. Zool. SOC.Lond. 142,665-695. East, M. L., and Hofer, H. (1991). Loud calling in a female-dominated mammalian society: 11. Behavioral contexts and functions of whooping of spotted hyaenas. Anim. Behav. 42,651-670. Eisenberg, J. (1981). “The Mammalian Radiations.” University of Chicago Press, Chicago. Enomoto, T. (1974). The sexual behavior of Japanese monkeys. J. Hum. Evol. 3,351-372.
242
LAURA SMALE ET AL
Errington, P. L. (1963). “Muskrat Populations.” Iowa State University Press, Ames, Iowa. Evans, F. C., and Holdenreid, R. (1943). A population study of the Beechey ground squirrel in central California. J. Mammal. 24, 231-260. Fairbairn, D. J. (1978). Behavior of dispersing deermice (Peromyscus maniculutis). Behav. Ecol. Sociobiol. 3, 265-282. Frank, L. G. (1986a). Social organisation of the spotted hyaena (Crocuta crocufa) I. Demography. Anim. Behav. 35, 1500-1509. Frank, L. G. (1986b). Social organisation of the spotted hyaena (Crocufacrocuta) 11. Dominance and Reproduction. Anim. Behav. 35, 1510-1527. Frank, L. G., Davidson. J. M., and Smith, E. R. (1985). Androgen levels in the spotted hyaena: The influence of social factors. J. Zool., Lond. A 206, 525-531. Frank, L. G., Glickman, S. E., and Licht, P. (1991). Fatal sibling aggression, precocial development, and androgens in neonatal spotted hyenas. Science 252, 702-704. Frank, L. G., Glickman, S. E., and Powch, I. (1990). Sexual dimorphism in the spotted hyaena. J . Zool., Lond. 221,308-313. Frank, L. G., Glickman, S. E.. and Zabel, C. (1989). Ontogeny of female dominance in the spotted hyaena: Perspectives from nature and captivity. Symp. Zool. Soc. Lond. 61, 127-146. Frank, L. G., Holekamp, K. E. and Smale, L. (1995). Dominance, demography, and reproductive success of female spotted hyenas. In “Serengeti 11: Dynamics, Management, and Conservation of an Ecosystem” (A. R. E. Sinclair and P. Arcece, eds.), pp. 364-384. University of Chicago Press, Chicago. Franklin, W. L. (1983). Contrasting socioecologies of South America’s wild camelids: The vicuna and the guanaco. I n “Advances in the Study of Mammalian Behavior” (J. F. Eisenberg and D. G. Kleiman, eds.), pp. 573-629. Special Publication #7 of the American Society of Mammalogists. Gaines, M. S., and McCleneghan, L. R. (1980). Dispersal in small mammals. Ann. Rev. Ecol. Sysr. 11, 163-196. Gartlan, J. S. (1975). Adaptive aspects of social structure in Eryfhrocebus paras. In “Proceedings from the Symposium of the Fifth Congress of the International Primatological Society” (S. Kondo, M. Kawai, A. Ehara, and S. Kawamura, eds.). Japan Science Press, Tokyo. Gaulin, S. J. C.. and Fitzgerald, R. W. (1986). Sex differences in spatial ability: An evolutionary hypothesis and test. Am. Nut. 127,74-88. Gaulin, S. J. C., and Fitzgerald, R. W. (1989). Sexual selection for spatial-learning ability. Anim. Behav. 37,322-331. Gauthreaux, S. A. (1978). The ecological significance of behavioral dominance. In “Perspectives in Ethology” (P. P. G. Bateson and P. H. Klopfer, eds.), pp. 17-45. Plenum Press, London. Geist, V. (1971). “Mountain Sheep: A Study in Behavior and Evolution.” University of Chicago Press, Chicago. Gese, E. M., and Mech, L. D. (1991). Dispersal of wolves (Canis lupus) in northeastern Minnesota, 1969-1989. Can. J . Zool. 69,2946-2955. Gillette, L. N. (1980). Movement patterns of radio-tagged opossums in Wisconsin. Amer. Midl. Nafur. 104, 1-12. Glickrnan,S. E., Frank,L. G., Pavgi, S.,and Licht, P. (1992). Hormonalcorrelates of “masculinization” in female spotted hyaenas (Crocuta crocuta): 1. Infancy to sexual maturation. J. Reprod. Fert. 95,451-462. Godfrey, G . K. (1957). Observations on the movements of moles (Tafpaeuropa) after weaning. Proc. Zool. SOC. Lond. 12s, 287-295.
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
243
Gombe, S. (1985). Short-term fluctuations in progesterone, oestradiol. and testosterone in pregnant and non-pregnant spotted hyaena (Crocuta crocuta Erxleben). Afric. J. Ecol. 23, 269-271. Gompper, M. E. (1994). Behavioral ecology and genetics of a coati (Nasua narica) population in Panama. Unpublished doctoral dissertation, University of Tennessee, Knoxville. Goodall, J. (1986). “The Chimpanzees of Gombe: Patterns of Behavior.” The Belknap Press of Harvard University Press, Cambridge, Massachusetts, and London. Greenwood, P. J. (1980). Mating systems, philopatry, and dispersal in birds and mammals. Anim. Behav. 28, 1140-1162. Halaas, J. L., Gajiwala, K. S., Maffei, M., Cohen, S . L., Chait, B. T., Rabinowits, D., Lallone, R. L., Burley, S. K., and Friedman, J. M. (1995). Weight-reducing effects of the plasma protein encoded by the obese gene. Science 269,543-546. Hamilton, W. J., and Bulger, J. B. (1990). Natal male baboon rank rises and successful challenges to resident alpha males. Behav. Ecol. Sociobiol. 26, 357-362. Hanby, J. P., and Bygott, J. D. (1987). Emigration of subadult lions. Anim. Behav. 35,161-169. Hanken, J., and Sherman, P. W. (1981). Multiple paternity in Belding’s ground squirrel litters. Science 212, 351 -353. Harcourt, A. H. (1978). Strategies of emigration and transfer by primates, with particular reference to gorillas. Z . Tierpsychol. 48,401-420. Harcourt, A. H., Stewart, K. S., and Fossey, D. (1976). Male emigration and female transfer in wild mountain gorilla. Nature (London) 263, 226-227. Harris, S.. and White, P. C. L. (1992). Is reduced affiliative rather than increased agonistic behavior associated with dispersal in foxes? Anim. Behav. 44, 1085-1089. Hauser. M. D., Cheney. D. L., and Seyfarth, R. M. (1986). Group extinction and fusion i n free ranging vervet monkeys. Amer. J. Primatol. 11,63-77. Henschel, J. R., and Skinner, J. D. (1987). Social relationships and dispersal patterns in a clan of spotted hyaenas Crocuta crocuta in the Kruger National Park. S. Afr. Tyskr. Dierl. 22, 18-23. Henzi, S . P.. and Lawes, M. (1987). Breeding season influxes and the behavior of adult male samango monkeys (Cercopithecus mitis albogularis). Folia Primatol. 48, 125-136. Henzi, S. P., and Lucas, J. W. (1980). Observations on the intertroop movements of adult vervet monkeys (Cercopithecus aethiops). Folia Primatol. 33, 220-235. Hill, J. L. (1974). Peromyscus: effects of early pairing on reproduction. Science 186,1042-1044. Hofer, H., and East, M. L. (1993). The commuting system of Serengeti spotted hyaenas: How a predator copes with migratory prey. 11. Intrusion pressure and commuters’ space use. Anim. Behav. 46, 559-574. Holekamp, K. E. (1983). Proximal mechanisms of natal dispersal in Belding’s ground squirrel (Spermophilus beldingi beldingi). Unpublished doctoral dissertation, University of California, Berkeley. Holekamp, K. E. (1984a). Dispersal in ground-dwelling sciurids. In “The Biology of GroundDwelling Sciurids” (J. 0. Murie and G . R. Michener, eds.), pp. 297-320. University of Nebraska Press, Lincoln. Holekamp. K. E. (1984b). Natal dispersal in Belding’s ground squirrels (Spermophilus beldingi). Behav. Ecol. Sociobiol. 16,21-30. Holekamp, K. E. (1986). Proximal causes of natal dispersal in Belding’s ground squirrels (Spermophilus beldingi). Ecol. Monogr. 56, 365-391. Holekamp, K. E., Cooper, S. M., Katona, C. I., Berry, N. A,, Frank, L. G., and Smale, L. (1997). Patterns of association among female spotted hyenas (Crocuta crocuta). J. Mammal. 78, 55-64.
244
LAURA SMALE ET A f
Holekamp, K. E., Ogutu, J., Frank, L. G., Dublin, H. T., and Smale, L. (1993). Fission of a spotted hyena clan: Consequences of female absenteeism and causes of female emigration. Ethology 93,285-299. Holekamp, K. E.. and Smale, L. (1991). Rank acquisition during mammalian social development: the “inheritance” of maternal rank. Amer. Zool. 31, 306-317. Holekamp, K. E., and Smale, L. (1993). Ontogeny of dominance in free-living spotted hyaenas: juvenile rank relations with other immature individuals. Anim. Behav. 46,451 -466. Holekamp, K. E., and Smale, L. (1995). Rapid change in offspring sex ratios after clan fission in the spotted hyaena. Am. Nut. 145,261-277. Holekamp, K. E., Smale, L., Simpson, H. B., and Holekamp, N. A. (1984). Hormonal influences on natal dispersal in free-living Belding’s ground squirrels (Spermophilus beldingi). Horm. Eehav. 18,465-483. Holekamp, K. E., Smale, L., and Szykman. M. (1996). Rank and reproduction in the female spotted hyaena. J. Reprod. Fert. In press. Holmes, W. G. (1986a). Identification of paternal half-siblings by captive Belding’s ground squirrels. Anim. Behav. 34,321-327. Holmes, W. G. (1986b). Kin recognition by phenotype matching in female Belding’s ground squirrels. Anim. Eehav. 34,38-47. Holmes, W. G., and Sherman, P. W. (1982). The ontogeny of kin recognition in two species of ground squirrels. Am. Zool. 22,491-517. Hoogland, J. L. (1982). Prairie dogs avoid extreme inbreeding. Science 215, 1639-1641. Hoogland, J. L. (1995). “The black-tailed prairie dog: Social life of a burrowing mammal.” University of Chicago Press, Chicago. Howard, W. E. (1960). Innate and environmental dispersal in individual vertebrates. Amer. Midl. Natur. 63, 152-161. Isbell, L. A., Cheney, D. L., and Seyfarth, R. M. (1991). Group fusions and minimum group sizes in vervet monkeys (Cercopirhecus aethiops). Am. J. Primatol. 25, 57-65. Jacobs, L. F., Gaulin, S. J. C., Sherry, D. F., and Hoffman, G. E. (1990). Evolution of spatial cognition: Sex-specific patterns of spatial behavior predict hippocampal size. Proc. Narl. Acad. Sci. U.S.A. 87, 6349-6352. Jenkins, S. H., and Eshelman, B. D. (1984). Spermophilus beldingi. Mammalian Species 221, 1-8. Jenks, S. M., Weldele, M. L., Frank, L. G., and Glickman, S. E. (1995). Acquisition of matrilineal rank in captive spotted hyaenas: emergence of a natural social system in peerreared animals and their offspring. Anim. Behav. 50,893-904. Johnson, C. G. (1969). “Migration and Dispersal of Insects by Flight.” Methuen, London. Jones, K. C. (1983). Inter-troop transfer of Lemur c a m males at Berenty, Madagascar. Folia Primatol. 40, 145-160. Jones, W. T. (1984). Natal philopatry in banner-tailed kangaroo rats. Eehav. Ecol. Sociobiol. 15, 151-155. Jones, W. T. (1987). Dispersal patterns in kangaroo rats (Dipodomys spectabilk). In “Mammalian Dispersal Patterns: The Effects of Social Structure on Population Genetics” (D. B. Chepko-Sade and Z. T. Halpin, eds.), pp. 119-127. University of Chicago Press, Chicago. Keane, B. (1990a). Dispersal and inbreeding avoidance in the white-footed mouse, Peromyscus leucopus. Anim. Behav. 40, 143-152. Keane, B. (1990b). The effects of relatedness on reproductive success and mate choice in the white-footed mouse, Perornyscus leucopus. Anim. Behav. 39,264-273. Keane, B.. Creel, S. R., and Waser, P. M. (in press). No evidence of inbreeding avoidance or inbreeding depression in a social carnivore. Behav. Ecol. 7,480-489.
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
245
King, C. (1989). “The Natural History of Weasels and Stoats.” Comstock Publishing, Ithaca, New York. King, J. A. (1955). Social behavior, social organization, and population dynamics of a blacktailed prairie dog town in the Black Hills of South Dakota. Contrib. Lab. Vertebr. Biol. Univ. Mich. 67, 1-123. Klingel, H. (1974). A comparison of the social behaviour of the Equidae. In “Behaviour of Ungulates and Its Relation to Management” (V. Geist and F. Walther, eds.), pp. 124-132. International Union for Conservation of Nature and Natural Resources, Morges, Switzerland. Krebs, C. J. (1978). Aggression, dispersal, and cyclic changes in populations of small mammals. In “Aggression, Dominance, and Individual Spacing” (L. Krames, L. P. Pliner, and T. Alloway, eds.), pp. 49-60. Plenum, New York. Kruuk, H. (1972). “The Spotted Hyena.” University of Chicago Press, Chicago. Kuester, J., Paul, A., and Arnemann, J. (1994). Kinship, familiarity and mating avoidance in Barbary macaques, Macaca sylvanus. Anim. Behav. 48, 1183-1194. Larsen, K. W., and Boutin, S. (1994). Movements, survival, and settlement of red squirrel (Tamiasciurus hudsonicus) offspring. Ecology 75, 214-223. Lee, A. K., and Cockburn, A. (1985). “The Evolutionary Ecology of Marsupials.” Cambridge University Press, Cambridge. Licht, P., Frank, L. G., Pavgi, S., Yalcinkaya, T. M., Siiteri, P. K., and Glickman, S. E. (1992). Hormonal correlates of “masculinization” in female spotted hyaenas (Crocuta crocuta): 2. maternal and fetal steroids. J. Reprod. Fert. 95,463-474. Lidicker, W. Z. (1975). The role of dispersal in the demography of small mammals. In “Small Mammals: Their Productivity and Population Dynamics” (F. B. Golley, K. Petrusewicz, and L. Ryszkowski, eds.), pp. 103-128. Cambridge University Press, London. Lidicker, W. Z., and Stenseth, N. C. (1992). To disperse or not to disperse: Who does it and why? In “Animal Dispersal: Small Mammals as a Model” (N. C. Stenseth and W. Z. Lidicker, eds.), pp. 21-36. Chapman and Hall, London. Maly, M. S., Knuth, B. A., and Barrett, G. W. (1985). Effects of resource partitioning on dispersal behavior of feral house mice. J. Mammal. 66, 148-153. Manson, J. H., and Perry, S. E. (1993). Inbreeding avoidance in Rhesus macaques: Whose choice? Amer. J. Phys. Anthro. 90, 335-344. Mason, W. A. (1979). Ontogeny of social behavior. In “Handbook of Behavioral Neurobiology” (P. Marler and J. G. Vandenbergh, eds.), pp. 1-28. Plenum, New York. Mathews, L. H. (1939). Reproduction of the spotted hyaena, Crocura crocuta (Erxleben). Phil. Trans. Ser. B. 230, 1-78. Matsumura, S. (1993). Intergroup affiliative interactions and intergroup transfer of young male Japanese macaques (Macaca fuscata). Primates 34, 1-10. Maxwell, C. S., and Morton, M. L. (1975). Comparative thermoregulatory capabilities of neonatal ground squirrels. J. Mammal. 56, 821-828. McCullough, D. R. (1985). Long-range movements of large terrestrial mammals. In “Migration: Mechanisms and Adaptive Significance” (M. A. Rankin, ed.), pp. 369-385. Contributions in Marine Science, Suppl. No. 27. McGuire, M. R., and Getz, L. L. (1981). Incest taboo between sibling Microtus ochrogaster. J. Mammal. 62,213-215. Mehlman, P. (1986). Male intergroup mobility in a wild population of the Barbary macaque (Macaca sylvanus), Ghomaran Rif Mountains, Morocco. Am. J. Primatol. 10, 67-81, Meier, A. H., and Fivizzani, A. J. (1980). Physiology of migration. In “Animal Migration, Orientation, and Navigation” (S. A. Gauthreaux, ed.), pp. 225-282. Academic Press, New York.
246
LAURA SMALE ET A L
Meikle, D. B., Kruper, J. H., and Browning, C. R. (1995). Adult male house mice born to undernourished mothers are unattractive to oestrous females. Anim. Behav. 50,753-758. Messier, F. (1985a). Solitary living and extraterritorial movements of wolves in relation to social status and prey abundance. Can. J. Zool. 63,239-245. Messier, F. (1985b). Social organization, spatial distribution and population density of wolves in relation to moose density. Can. J. Zool. 63, 239-245. Metzgar, L. M. (1967). An experimental comparison of owl predation on resident and transient white-footed mice (Peromyscus leucopus). J. Mammal. 48, 387-391. Michener, G. R. (1983). Kin identification, matriarchies, and the evolution of sociality in ground-dwelling sciurids. I n “Advances in the Study of Mammalian Behavior” (J. F. Eisenberg and D. G. Kleiman, eds.), pp. 528-572. American Society of Mammalogists Special Publication No. 7. Mills, M. G. L. (1990). “Kalahari Hyaenas.” Unwin Hyman, London. Montgomery, G. G., and Sunquist, M. E. (1978). Habitat selection and use by two-toed sloths. In “The Ecology of Arboreal Foliviores” (G. G. Montgomery, ed.), pp. 329-360. Smithsonian Institution Press, Washington, D.C. Moore, J., and Ali, R. (1984). Are dispersal and inbreeding avoidance related? Anim. Behav. 32,94-112. Morton, M. L., Maxwell, C. S., and Wade, C. E. (1974). Body size, body composition, and behavior of juvenile Belding ground squirrels. Great Basin Nat. 34, 121-134. Morton, M. L., and Gallup, J. S. (1975). Reproductive cycle of the Belding ground squirrel (Spermophilus beldingi beldingi): seasonal and age differences. Great Basin Nat. 35, 427-433. Morton, M. L., and Sherman, P. W. (1978). Effects of a spring snowstorm on behavior, reproduction, and survival of Belding’s ground squirrels. Can. J. Zool. 56, 2578-2590. Morton, M. L., and Tung, H. L. (1971). Growth and development in the Belding’s ground squirrel (Spermophilus beldingi beldingi). J. Mammal. 52,611-616. Mrosovsky, N. (1976). Lipid programmes and life strategies in hibernators. Am. Zool. 16, 685-697. Murray, M. G. (1982). Home range, dispersal, and the clan system of the impala. Afric. J . EcoI. 20,253-269. Nelson, M. E. (1993). Natal dispersal and gene flow in white-tailed deer in northeastern Minnesota. J. Mammal. 74,316-322. Nelson, R. J. (1995). “An Introduction to Behavioral Endocrinology.” Sinauer Associates, Sunderland, Massachusetts. Noyes, R. F.. Barrett, G. W., and Taylor, D. H. (1982). Social structure of feral house mouse (Mus musculus) populations: Effects of resource partioning. Behav. Ecol. Sociobiol. 10, 157-163. Nunes, S., and Holekamp, K. E. (1996). Mass and fat influence the timing of natal dispersal in Belding’s ground squirrels J. Mammal. 77,807-817. Nunes, S., Zugger, P. A., Engh. A. L., Reinhart, K. 0..and Holekamp, K. E. (in press). Why do female Belding’sground squirrels disperse away from food resources?Behav. E d . Sociobiol. O’Donoghue, M., and Bergman. C. M. (1992). Early movements and dispersal of juvenile snowshoe hares. Canad. J . Zool. 70, 1787-1791. Oliver, A. J. (1986). Social organization and dispersal in the red kangaroo. Unpublished doctoral dissertation, Murdoch University, Perth, Australia. Packer, C. (1979). Inter-troop transfer and inbreeding avoidance in Papio anubis. Anim. Behav. 27,37-45. Packer, C. (1986). The ecology of felid sociality. I n “Ecological Aspects of Social Evolution in Birds and Mammals” (D. I. Rubenstein and R. W. Wrangham, eds.). pp. 429-451. Princeton University Press, Princeton, New Jersey.
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
247
Packer, C., Collins, D. A,, Sindimwo, A,, and Goodall, J. (1995). Reproductive constraints on aggressive competition in female baboons. Nature (London) 373, 60-63. Paul, A,, and Kuester, J. (1985). Inter-group transfer and incest avoidance in semi-free-ranging Barbary macaques (Macaca sylvanus) at Salem (FRG). Am. J. Primatol. 8,317-322. Pellymounter, M. A,, Cullen, M. J., Baker, M. B., Hecht, R., Winters, D., Boone, T., and Collins, F. (1995). Effects of the obese gene product on body weight regulation in o b h b mice. Science 269, 540-543. Pengelley. E. T., and Fisher, K. C. (1963). The effect of temperature and photoperiod on the yearly hibernating behaviour of captive golden-mantled ground squirrels (Citellus lateralis tescorum). Can. J . Zool. 41, 1103-1120. Pereira, M. E., and Weiss, M. L. (1991). Female mate choice, male migration, and the threat of infanticide in ringtailed lemurs. Behav. Ecol. Sociobiol. 28, 141-152. Pfaff, D. W. (1980). “Estrogens and Brain Function.” Springer-Verlag, New York. Phoenix, C. H., Goy, R. W., Gerrall, A. A,, and Young, W. C. (1959). Prenatally administered testosterone proprionate on the tissues mediating mating behavior in the female guinea pig. Endocrinol. 65, 369-382. Pokier, F. E. (1969). The Nilgiri langur troop: its composition, function, and change. Folia Primatol. 10, 20-47. Prud’Homme, J. (199 1). Group fission in a semi-free-ranging population of Barbary macaques (Macaca sylvanus). Primates 32, 9-22. Pusey, A. E. (1979). Intercommunity transfer of chimpanzees in Gombe National Park. I n “The Great Apes” (D. A. Hamburg and E. R. McCown. eds.), pp. 465-480. Benjamin/ Cummings, Menlo Park, California. Pusey, A. E. (1980). Inbreeding avoidance in chimpanzees. Anim. Behav. 28,543-582. Pusey, A. E. (1987). Sex-biased dispersal and inbreeding avoidance in mammals. Trends Ecol. E v o ~ .2, 295-299. Pusey, A. E., and Packer, C. (1987a). Dispersal and philopatry. In “Primate Societies” (B. B. Smuts, D. L. Cheney, R. M. Seyfarth, R. W. Wrangham, and T. T. Struhsaker, eds.), pp. 250-266. University of Chicago Press, Chicago. Pusey, A. E., and Packer, C. (1987b). The evolution of sex-biased dispersal in lions. Eehaviour 101,275-310. Quanstrom, W. R. (1 971). Behaviour of Richardson’s ground squirrel Spermophilus richardsonii richardsonii. Anim. Eehav. 19, 646-652. Rado, R., Wollberg, Z., and Terkel, J. (1991). The ontogeny of seismic communication during dispersal in the blind mole rat. Anim. Eehav. 42, 15-21. Rayor. L. S. (1985). Effects of habitat quality on growth, age of first reproduction, and dispersal in Gunnison’s prairie dogs (Cynomys gunnisoni). Can. J. Zool. 63, 2835-2840. Richard, A. F., Rakotomanga, P., and Schwartz, M. (1991). Dispersal by Propithecus verreauxi at Beza Mahafaly, Madagascar: 1984-1989. Am. J. Phys. Anthro. 84, 307-322. Ritchison. G., Belthoff, J. R., and Sparks, E. J. (1992). Dispersal restlessness: Evidence for innate dispersal by juvenile eastern screech owls? Anim. Eehav. 43, 57-65. Rogers, L. L. (1987). Factors influencing dispersal in the black bear. In “Mammalian Dispersal Patterns: The Effects of Social Structure on Population Genetics” (B. D. Chepko-Sade and Z . T. Halpin, eds.), pp. 75-84. University of Chicago Press, Chicago. Rood, J. P. (1987). Dispersal and intergroup transfer in the dwarf mongoose. In “Mammalian Dispersal Patterns: The Effects of Social Structure on Population Genetics” (B. D. Chepko-Sade and Z . T. Halpin, eds.), pp. 85-103. University of Chicago Press, Chicago. Rood, J. P. (1990). Group size, survival, reproduction, and routes to breeding in dwarf mongooses. Anim. Behav. 39,566-572.
248
LAURA SMALE ET AL
Russell, J. K. (1982).Timing of reproduction by coatis (Nasua narica) in relation to fluctuations in food resources. In “The Ecology of aTropical Rain Forest” (E. G. Leigh, Jr., A. S. Rand, and D. M. Windsor, eds.), pp. 468-479. Smithsonian Institute Press, Washington, D.C. Rutberg, A. T., and Keiper, R. R. (1993). Proximate causes of natal dispersal in feral ponies: Some sex differences. Anim. Behav. 46,969-975. Sandell, M., and Erlinge, S. (1989). Dispersal patterns in the stoat (Mustela erminea). Paper presented at the meeting of the Theriological Society, Rome, August, 1989. Savidge, I. R. (1974). Social factors in dispersal of deermice (Peromyscus maniculatus) from their natal site. Amer. Midl. Natur. 91,395-405. Schaller, G . 9. (1972). “The Serengeti Lion: A Study of Predator-Prey Relations.” University of Chicago Press, Chicago. Schneider, J. E. (1992). Metabolic sense. Horm. Behav. 26,l-6. Schneider, J. E., and Wade, G. N. (1989). Availability of metabolic fuels controls estrous cyclicity in Syrian hamsters. Science 244, 1326-1328. Schneider, J. E., and Wade, G. E. (1990a). Decreased availability of metabolic fuels induces anestrus in golden hamsters. Am. J. Physiol. 259, R750-R755. Schneider, J. E., and Wade, G. E. (1990b). Effects of diet and body fat content on coldinduced anestrus in Syrian hamsters. Am. J. Physiol. 259, R1198-Rl2W. Schneider, J. E., and Zhu, Y. (1994). Caudal brain stem plays a role in metabolic control of estrous cycles in Syrian hamsters. Brain Res. 661, 70-74. Scott, I., Grant, M., and Evans, P. R. (1991). Estimation of fat-free mass of live birds: Use of total body electrical conductivity (TOBEC) measurements in studies of single species in the field. Funct. Ecol. 5, 314-320. Scott, M. P. (1987). The effect of mating and agonistic experience on adrenal function and mortality of male Antechinus stuartii (Marsupialia). J. Mammal. 68,479-486. Sherman, P. W. (1976). Natural selection among some group-living organisms. Unpublished doctoral dissertation, University of Michigan, Ann Arbor. Sherman, P. W. (1977). Nepotism and the evolution of alarm calls. Science 197, 1246-1253. Sherman, P. W. (1981). Kinship, demography, and Belding’s ground squirrel nepotism. Behav. Ecol. Sociobiol. 8, 251 -259. Sherman, P. W.. and Morton, M. L. (1984). Demography of Belding’s ground squirrels. Ecology 65,1617-1628. Sherry, D. F., Jacobs, L. F., and Gaulin, S. J. C. (1992). Spatial memory and adaptive specialization of the hippocampus. Trends Neurosci. 15,298-303. Shields, W. M. (1982). “Philopatry, Inbreeding, and the Evolution of Sex.” State University of New York Press, Albany. Sigg, H., Stolba, A., Abegglen, J. J., and Dasser, V. (1982). Life history of hamadryas baboons: Physical development, infant mortality. reproductive parameters, and family relationships. Primates 23,473-487. Sinclair, A. R. E. (1977). “The African Buffalo.” University of Chicago Press, Chicago. Sinclair. A. R. E. (1992). Do large mammals disperse like small mammals? In “Animal Dispersal: Small Mammals as a Model” (N. C. Stenseth and W. Z. Lidicker, eds.), pp. 225-242. Chapman and Hall, London. Smale, L., Frank, L. G., and Holekamp, K. E. (1993). Ontogeny of dominance in free-living spotted hyaenas: Juvenile rank relations with adults. Anim. Behav. 46,467-477. Smale,L., Holekamp, K. E., Weldele, M., Frank, L. G.,andGlickman, S. E. (1995). Competition and cooperation between littermates in the spotted hyaena. Anim. Behav. 50,671-682. Smith, A. T., and Ivins, 9.L. (1983). Colonization in a pika population: Dispersal vs. philopatry. Behav. Ecol. Sociobiol. 13, 37-47. Smith, E. 0. (1992). Dispersal in sub-saharan baboons. Folia Primatol. 59, 177-185.
SEXUALLY DIMORPHIC DISPERSAL IN MAMMALS
249
Smuts, B. B. (1985). “Sex and Friendship in Baboons.” Aldine, Hawthorne, New York. Snyder, R. L., Jenson, W., and Cheney, C. D. (1976). Environmental familiarity and activity: Aspects of prey selection for a ferrunginous hawk. Condor 78,138-139. Sprague, D. S. (1992). Life history and male intertroop mobility among Japanese macaques (Macaca fuscata). Inr. J. f rimatol. 13,437-454. Stammbach, E. (1987). Desert, forest and montane baboons: Multilevel-societies. In “Primate Societies” (B. B. Smuts, D. L. Cheney, R. M. Seyfarth, R. W. Wrangham, and T. T. Struhsaker, eds.), pp. 112-120. University of Chicago Press, Chicago. Stenseth, N. C. (1983). Causes and consequences of dispersal in small mammals. In “The Ecology of Animal Movement” (I. R. Swingland and P. J. Greenwood, eds.), pp. 63-101. Clarendon Press, Oxford. Stenseth, N. C., and Lidicker, W. Z. (Eds.). (1992). “Animal dispersal: Small Mammals as a Model.” Chapman and Hall, New York. Strier, K. B. (1991). Demography and conservation of an endangered primate, Brachyteles arachnoides. Conserv. Biol. 5,214-218. Strier, K. B. (1994). Myth of the typical primate. Yrbook fhys. Anthropol. 37,233-271. Strum, S. (1987). “Almost Human: A Journey into the World of Baboons.” Random House, New York. Sugiyama, Y. (1965). On the social change of Hanuman langurs (fresbyris entellis) in their natural condition. Primates 6,381-418. Sugiyama, Y. (1976). Life history of male Japanese macaques. Adv. Study Behav. 7,255-284. Sussman, R. W. (1991). Demography and social organization of free-ranging Lemur catta in the Beza Mahafaly Reserve, Madagascar. Am. J. fhys. Anthropol. 84,43-58. Takahata, Y. (1982). The socio-sexual behavior of Japanese monkeys. Z. Tierpsychol. 59, 89-108. Takahata, Y.. Suzuki, S., Okayasu, N., and Hill, D. (1994). Troop extinction and fusion in wild Japanese macaques of Yakushima Island, Japan. Am. J. frimatol. 33,317-322. Tilson, R. T.. and Hamilton, W. J. (1984). Social dominance and feeding patterns of spotted hyaenas. Anim. Behav. 32,715-724. Tilson, R. T., and Henschel, J. R. (1986). Spatial arrangement of spotted hyaena groups in a desert environment, Namibia. Afr. J. Ecol. 24, 173-180. Tinbergen, N. (1948). Social releasers and the experimental method required for their study. Wilson Bull. 60,6-52. Thor, D. H.. and Holloway, W. R., Jr. (1984). Sex and social play in juvenile rats (RUIIUS norvegicus). J. Comp. Psych. 98, 276-284. Thor, D. H., and Holloway, W. R., Jr. (1986). Social play soliciting by male and female juvenile rats: Effects of neonatal androgenization and sex of cagemates. Behav. Neurosci. 100,275-279. Trivers, R. L. (1971). The evolution of reciprocal altruism. Q. Rev. Bid. 46, 35-57. van Noordwijk, M. A,, and van Schaik, C. P. (1985). Male migration and rank acquisition in wild long-tailed macaques (Macaca fmcicularis). Anim. Behav. 33, 849-861. Van Vuren. D.. and Armitage, K. B. (1994). Survival of dispersing and philopatric yellowbellied marmots: what is the cost of dispersal? Oikos 69, 179-181. Vick, L. G., and Pereira, M. E. (1989). Episodic targeting aggression and the histories of Lemur social groups. Behav. Ecol. Sociobiol. 25,3-12. Wade, G. N., and Schneider, J. E. (1992). Metabolic fuels and reproduction in female mammals. Neurosci. Biobehav. Rev. 16,235-272. Wade, G. N., Schneider, J. E., and Friedman, M. I. (1991). Insulin-induced anestrus in Syrian hamsters. Am. J. fhysiol. 260, R148-Rl52.
250
LAURA SMALE ET AL
Walsberg, G. E. (1988). Evaluation of a nondestructive method for determining fat stores in small birds and mammals. Physiol. Zoo/. 61, 153-159. Warneke, R. M. (1975). Dispersal and mortality of juvenile fur seals, Arcrocephalus pusillus doriferus in Bass Strait, Southeastern Australia. Rapp. P. V. Reun. Cons. Int. Explor. Mer. 169,296-302. Waser, P. M. (1996). Patterns and consequences of dispersal in gregarious carnivores. In “Carnivore Biology, Ecology, and Evolution: Volume 11” (J. Gittleman, ed.), pp. 267-295. Cornell University Press, Ithaca. Waser, P. M., and Jones, W. T. (1983). Natal philopatry among solitary mammals. Q. Rev. Biol. 58, 355-390. Whitworth, M. R., and Southwick, C. H. (1984). Sex differences in the ontogeny of social behavior in pikas: possible relationships to dispersal and territoriality. Behav. Ecol. Sociobiol. 15, 175-182. Wiley, R. H. (1981). Social structure and individual ontogenies: Problems of description, mechanism and evolution. In “Perspectives in Ethology” (P. P. G. Bateson and P. H. Klopfer, eds.), pp. 105-133. Plenum, New York. Williams, L. R., and Cameron, G. N. (1984). Demography and dispersal in Attwater’s pocket gopher (Geomys artwateri). J. Mamm. 65, 67-75. Wilson, M. E. (1992). Factors determining the onset of puberty. In “Handbook of Behavioral Neurobiology, Sexual Differentiation” (A. A. Gerall, H. Moltz, and 1. L. Ward, eds.), pp. 275-312. Plenum, New York. Windberg, L. A., and Keith. L. B. (1976). Experimental analysis of dispersal in snowshoe hare populations. Canad. J. Zoo/. 54,2061-2081. Wrangham, R. W. (1980). An ecological model of female-bonded primate groups. Behavioctr 75, 262-300. Wynne-Edwards, V. C . (1962). “Animal Dispersion in Relation to Social Behavior.” Oliver and Boyd, Edinburgh. Zabel, C. J., Glickman, S. E., Frank, L. G., Woodmansee, K. B., and Keppel, G. (1992). Coalition formation in a colony of prepubertal spotted hyenas. I n “Coalitions and Alliances in Humans and Other Animals” (A. H. Harcourt and F. B. M. de Waal, eds.), pp. 112-135. Oxford Science Publications, Oxford. Zucker, 1. (1988). Neuroendocrine substrates of circannual rhythms. In “Biological Rhythms and Mental Disorders” (D. J. Kupfer, T. H. Monk, and J. D. Barchas, eds.), pp. 219-251. Guilford Press, New York. Zucker, I., and Boshes, M. (1982). Circannual body weight rhythms of ground squirrels: Role of gonadal hormones. A m . J . Physiol. 243, R546-R551. Zucker, I., Lee, T. M., and Dark, J. (1991). The suprachiasmatic nucleus and annual rhythms of mammals. In “Suprachiasmatic Nucleus: The Mind’s Clock” (D. C. Klein, R. Y. Moore, and S. M. Reppert, eds.), pp. 246-259. Oxford University Press, New York.
ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 26
Infantile Amnesia: Using Animal Models to Understand Forgetting -~
~
H. MOOREARNOLD A N D NORMAN E. SPEAR
What I would like to prophesy . . . is that future research will extend and confirm the generalization that memory is never better at early ages, provided that the degree of original and interpolated learning is held constant. B. A. Campbell
Incited by increased awareness of the abuse of children, both the general public and the scientific community have become increasingly interested in memory for the events of one’s childhood. Memory is an issue because detection of childhood abuse has often been retrospective, sometimes many years after the event. In some cases, memory for events may serve as key evidence in criminal prosecution of an alleged abuser. Memory is also believed to serve therapeutic purposes of a substantial number of clinical psychologists and psychiatrists who believe that recall and conscious comprehension of past traumatic events can alleviate not only the symptoms but the basis of severe psychological problems such as neuroses. However, limitations on memory for events early in life complicate these issues. Major contemporary concerns about the poor retention of early events are, (1) the effectiveness of psychotherapists, attorneys, or law-enforcement personnel in alleviating one’s forgetting of childhood traumas, and (2) the degree of veracity for memories “recalled” as a consequence (Ceci and Bruck, 1993; Lindsay and Read, 1994). In the history of psychology this issue is hardly new. Modern concerns for understanding the limitations on memory of childhood events are really no different than those that inspired the theories of Sigmund Freud (e.g., Freud, 1953), aside from current questioning of the veracity of early memories-a clearly critical point for the courts in prosecution of alleged abusers. The fundamental conceptual error this time around is also the same as Freud’s: the implicit assumption that our difficulties in remembering the events of our early childhood are due to the social structure, linguistic behavior, and consciousness that are unique to humans. These attributes may well be unique to humans, but the phenomenon that many theorists try to explain with them-infantile amnesia (Freud’s term for the severe 25 1
Copyright 0 19Y7 by Academic Press All rights of reproduction in any form reserved. 0065-3454/97 $25.00
252
H. MOORE ARNOLD AND NORMAN E. SPEAR
forgetting of events of infancy)-is not at all unique to humans. This phenomenon has in fact been observed in every altricial species in which it has been tested. The purpose of this chapter is to illustrate the generality of infantile amnesia (defined for our purposes as rapid forgetting of events during infancy relative to that of events occurring later in life). Our thesis is that features of childhood memories currently given so much attention-their ephemeral character, susceptibility to distortion, and potential for reactivation-represent basic biological processes mandating their study from a biological orientation. This orientation is needed for a variety of reasons: the likely involvement of brain development that is correlated tightly with the severity of infantile amnesia and the reality that experimental manipulations of brain development can be achieved only with animals; the need for experimental rather than correlational evidence associated with early traumas and the reality that such tests can be achieved only with animals; and the likelihood that many of the basic social and regulatory functions that contribute to infantile amnesia are an extreme variety that can be studied experimentally only with animals. This is not to deny that human attributes such as consciousness and related personality features (such as the development of “self,” e.g., Howe and Courage, 1993), as well as susceptibility to the linguistic forces of suggestion associated with social pressures, are relevant to evaluating one’s retention of alleged abuse during childhood. The point is merely that determinants of the ontogeny of memory are too basic to consider them solely with tests of humans. It is possible that these more basic processes have more relevance, even for legal issues, than is conventionally assumed. I. ORIENTATION Historically, reviews of the memory capabilities of developing animals have been concerned with the observation that infant organisms forget more rapidly than adults of the same species over the same retention interval. That younger animals forget more quickly than adults may seem obvious to the current student of memory development, but this was not the understanding in the field thirty years ago. It was not until an influential chapter by B. A. Campbell in 1967 that this view began to be accepted. Campbell pointed out that many researchers actually believed the opposite to be true-that effectiveness in memory processing declined with advancing age, even between infancy and early adulthood. The prophesy cited at the beginning of our chapter concluded Campbell’s paper and marked the beginning of a significant interest in developmental psychobiology-the
INFANTILE AMNESIA
253
rapid forgetting of events learned during early development of the brain referred to here as infantile amnesia. Prior to the paper by Campbell (1967), studies on the development of retention were confusing and often in disagreement as to whether retention improved or declined with age. Campbell pointed out that research prior to 1967 was not always designed or conducted in a manner that would allow one to make firm conclusions about the development of memory with age. Campbell (1967) stated four main reasons for conflict and confusion in the literature prior to that time. First, researchers often failed to control the level of motivation and reinforcement in their subjects. In one example cited by Campbell (Liu, 1928), rats were trained under food deprivation conditions, yet the researcher failed to weigh any of the animals or keep track of the amount of food each animal was fed during any given day. In addition, the diet was occasionally changed in the middle of the experiment. These failures to control deprivation level and motivation were accompanied by confounding of these variables with age, making firm conclusions about infantile amnesia impossible. The second problem with the early studies was an inadequate appreciation of the behavioral limitations of developing animals. It is impossible to measure learning or retention of a behavior that an animal is unable to express at the particular age of testing. Third, failure to control for nonassociative consequences or transfer of learning from earlier experiences limited the value of many early experiments. This was a problem because animals were often used in multiple experiments. Finally, the techniques for measuring learning and memory in young animals were just not well developed during these early years of investigation, at least not to the extent that they are now (Spear and Rudy, 1991), and the degree of learning was rarely controlled. (For valuable elaboration the reader is encouraged to refer to Campbell’s 1967 paper.) There now exists compelling evidence that capacity for retention of acquired memories does in fact improve over the early life span in both humans (e.g., Rovee-Collier, Sullivan, Enright, Lucas, and Fagan, 1980) and other animals (Campbell and Spear, 1972). The purpose of this chapter is to outline the current understanding of the ontogeny of retention. Much of the research to date has focused on determining the age at which the memory process begins to function, probably not the most fruitful approach. It is now known that retention is possible even for learning that occurs prenatally (e.g., Fifer, 1987; Stickrod, Kimble, and Smotherman, 1982). The preferred alternative approach is to determine the age at which memory processing begins to function like that of the adult animal. This analysis recognizes that younger animals are quite able to process memories at a very early age, albeit less effectively, or differently, than adults.
254
H. MOORE ARNOLD AND NORMAN E. SPEAR
We begin this chapter by discussing the ontogeny of nonassociative memory. Although relatively little attention has been paid to the retention of nonassociative memories, studies of this type provide a basic link toward understanding how associative memory develops. This is followed by reviewing retention of associative learning, considered in two categories, the ontogeny of short-term and long-term retention. Our distinction between “short-term” and “long-term” is strictly operational, implying no difference in memory processes and used only for ease of presentation. The experiments discussed in relation to short-term retention all used retention intervals on the order of seconds or minutes, whereas those studying long-term retention used retention intervals ranging from minutes to days or weeks. 11. ONTOGENY OF NONASSOCIATIVE MEMORY
Nonassociative memory refers to the retention of those learning processes that are not thought to involve a specific association between an environmental event and a specific reinforcer or outcome. Probably the best studied example of nonassociative learning is habituation, typically described as a reduction in responding over the course of repeated presentations of a stimulus that cannot be attributed to fatigue or sensory adaptation (Groves and Thompson, 1970). The other major nonassociative learning process is “sensitization,” characterized by an increase in response to a stimulus as a result of previous exposure to the same or a similar stimulus. Research on the ontogeny of memory in animals has been concerned primarily with associative memories. This research has focused on learning processes that promote the formation and retention of associations (Spear, Miller, and Jagielo, 1990). Relatively little attention has been given to retention of nonassociative learning. Retention has been studied in habituation paradigms with developing humans (Cohen and Gelber. 1975; Fagan, 1974) and adult invertebrates (Carew, Pinsker, and Kandel, 1972), but retention of habituation has been studied infrequently in developing laboratory mammals. Richardson and Campbell (1991) cited two reasons why the ontogeny of nonassociative memory is of theoretical interest. The first is to consider the sequential emergence of learning capacity from simpler tasks to more complex ones. For example, Rudy, Vogt, and Hyson (1984) observed that rat pups were capable of learning to attenuate a neophobic response t o a flavor, a simple form of nonassociative learning, prior to the age at which they were able to show a simple form of associative learning, conditioned taste aversion. This result can be interpreted as evidence that the ability to exhibit more complicated forms of learning emerges sequentially during
INFANTILE AMNESIA
255
the animal’s development. If one can extend this line of reasoning to include retention in addition to learning ability, then retention of nonassociative learning should emerge earlier in development than retention of associative learning. Development of nonassociative memory is important also because of the widespread use of nonassociative memory procedures in research on the neurobiology of learning. Several scientists have made extensive use of nonassociative procedures to study the neurobiological basis of learning, most notably Kandel and his colleagues (Pinsker, Kupferman, Castellucci, and Kandel, 1970) and Thompson and his co-workers (e.g., Groves and Thompson, 1970). Some exciting recent work in this area has involved a developmental analysis of habituation, sensitization, and dishabituation in Aplysiu (Marcus, Nolen, Rankin, and Carew, 1988; Nolen and Carew, 1994; Rankin and Carew, 1988). With a growing understanding of the learning processes that underlie habituation and sensitization comes a growing need to understand postacquisition processes that occur after such learning and determine retention. In one of the first ontogenetic comparisons of nonassociative memory, Parsons, Fagen, and Spear (1973) examined short-term retention of habituation of nosepoke behavior. Rats in different age groups (the youngest were 15 days old and the oldest were 1.5 years old) were placed in an open field and the number of nosepokes emitted into a hole in the wall were recorded. It was observed that although infants habituated more slowly than the adults, all ages demonstrated equivalent performance on an immediate test of retention. At longer intervals (1 and 24 hr), infants demonstrated poorer retention of the habituated response than the older animals. In studies of habituation of head turning to an air puff, File and Plotkin (1974) found no evidence in the neonatal rat of 1-hr retention of habituation, up to postnatal day 6. In a subsequent study, pups 1-6 days old were found to retain habituation of the head-turning response for 30 min and pups 8-15 days old for at least 1 hr (File and Scott, 1976). In neither of the studies by File and her colleagues was there any evidence that rat pups 1-16 days of age were capable of retaining habituation for 24 hr. Campbell and Stehouwer (1980) examined retention of habituation and sensitization of the forelimb withdrawal response to shock. With their procedures habituation was a “transient phenomenon” in pups 3-15 days old, retained for only a few minutes following the end of training; sensitization was more durable, emerging in 6-day-old pups and lasting for at least 24 hr in pups 10-15 days old. Based on these data, Campbell and Stehouwer proposed that sensitization may be more closely linked with associative learning than is habituation. Like associative learning, sensitization involves the production of a response and is retained in a fashion consistent with
256
H. MOORE ARNOLD A N D NORMAN E. SPEAR
that observed in conventional learning paradigms. In contrast, habituation involves the cessation of a response, is transient, emerges very early (postnatal day 3), and remains, according to these studies, essentially unchanged during ontogenesis. More recent evidence has shown that associative learning, like habituation, can involve the cessation of a response and can emerge prenatally (e.g., Arnold, Robinson, Spear, and Smotherman, 1993). In addition, some forms of habituation d o seem to change during ontogeny (Richardson and Campbell, 1991). That sensitization is more closely linked to associative learning than habituation during development therefore remains an open, albeit intriguing, hypothesis. Richardson and Campbell (1991) measured both an autonomic index of habituation (decrease in the cardiac response to a tone) and an overt behavior (head jerk) to examine the ontogeny of long-term retention of habituation in developing rats 16, 30, or 75 days old. No differences in the rate of habituation were observed, permitting conclusions free from confounds in either learning rate or terminal response level. These experiments demonstrated that 16-day-old rat pups are not able to retain habituation longer than 4 hr, whereas adult rats retained the habituation for at least one week after the final training episode (Fig. 1). Richardson and Campbell (1991) suggest that these data support the notion that nonassociative learning emerges before associative learning. This conclusion is most clearly observed with the autonomic measure used in their studies, heart rate. There were no age differences in rate of habituation to a novel auditory stimulus between pups 16 and 75 days of age, yet conditioned cardiac response to an auditory stimulus, readily acquired by adults, cannot be established until the rat is at least 21 days of age (Campbell and Ampuero, 1985).This is consistent with the general principle that simple forms of learning precede more complex forms of learning in development. These data do not address directly whether nonassociative memory can be dissociated from associative memory, but they are consistent with this possibility.
SUMMARY Although the study of nonassociative learning has received increasing attention during the last twenty years and is coming to be understood even at the neurobiological level, our current understanding of the development of nonassociative memory lags behind our knowledge of the development of nonassociative learning. The few studies reported here summarize the last two decades of research on nonassociative memory in developing animals. What seems clear from the existing literature is that infant animals retain nonassociative information for shorter durations than do their more
257
INFANTILE AMNESIA
r
ma, I
= &
1
-10
*
)----
I
.-
-20
\
4
\
c m
\
-30
Y \
m
\
\ \
-40
-50
\
-1
!
rn
start
16-day-olds 75-day-olds
rn [
rn
~
-
end
Habituation
,
-
~
- ~~ ~~~~~
~
~
r- -~--1
~- 7
20 min I hour 4 hours 1 day 3days 7 days
Retention
FIG. 1. Mean changes in cardiac orienting response to a novel auditory stimulus in 16- and 75-day-old rats during the “start” and “end” of training and the retention of habituation during various retention intervals. Adapted from Richardson and Campbell (1991).
mature conspecifics. What remains unresolved is not whether there is a developmentally sequential progression in memory processing efficacy from simple to more complex tasks, but what are the essential differences between simple and complex tasks. It seems likely that the associational requirement is a critical dimension. Clearly, we need to learn more about nonassociative memory during development.
111. ONTOGENY OF SHORT-TERM RETENTION As with nonassociative memory, the ontogeny of short-term retention and the hypothetical processes that could be considered short-term or “working” memory have received comparatively little attention relative to studies of long-term retention. This section on short-term retention considers experiments that have employed retention intervals within the range of seconds or minutes. A common paradigm used to study short-term retention is the trace conditioning procedure. In this procedure there is a delay between the
258
H. MOORE ARNOLD AND NORMAN E. SPEAR
offset of the conditioned stimulus (CS) and the onset of an unconditioned stimulus (US). For an association to be formed it is often assumed that the subject must maintain a memory representation of the CS until the US is presented (although other interpretations are possible; cf. Marlin, 1981; McKinzie and Spear, 1995). This procedure has been used to investigate the ontogeny of short-term retention in olfactory (Rudy and Cheatle, 1979), gustatory (Steinert, Infurna, Jardula, and Spear, 1979; Vogt and Rudy, 1984), and auditory and visual systems (Moye and Rudy, 1987a,b). Perhaps the clearest early demonstration of ontogenetic differences in short-term retention examined the ability of rat pups to learn a conditioned odor aversion (Rudy and Cheatle, 1979; also see Klein and Spear, 1969, for a different early procedure). Rat pups ranging from 2 to 14 days of age were conditioned to avoid an odor paired with illness induced by LiCl injection. Pups at all ages learned the aversion and expressed it on a test 6 days later. However, if as little as 15 s intervened between the odor and the injection of LiCI, the 2-day-old was apparently unable to form the association. The 8-day-old pup acquired the aversion when as long as 90 min intervened between the exposure to the odor and the LiCl injection, but failed to acquire an aversion when the delay was extended to 240 min. Rudy and Cheatle (1979) concluded from their experiments that the capacity to learn about stimuli separated by short delays is not present at birth, but develops between 4 and 6 days of age (see Fig. 2). No other studies examining short-term retention have tested animals as young as those investigated by Rudy and Cheatle (1979), but several experiments have examined short-term retention in slightly older (but still quite young) rats. Extending the work of Rudy and Cheatle (1979), Moye and Rudy (1987a) suggested that the ability to associate events separated in time emerges later during ontogenesis than the ability to learn about temporally contiguous events. They found that 17-day-old rats would condition to a visual CS that terminated with the US, but they failed to condition if CS offset was separated from the US onset by either 10 or 30 s. However, 28-day-old pups demonstrated strong conditioning even when the CS was terminated 30 s prior to US onset. Similar results have been found using long-delay taste conditioning with rat pups 12 and 15 days old (Gregg, Kittrell, Domjan, and Amsel, 1978; Rudy et al., 1984). These experiments showed that the 12-day-old rat pup was not capable of learning a tasteillness association when the delay between the taste and the reinforcer exceeded 30 min, a duration that did not disrupt conditioning in the 15day-old rat pup. There are limits to the usefulness of the trace conditioning procedure for the investigation of short-term retention. For example, this procedure cannot be used to investigate memory for elements of the training episode
259
INFANTILE AMNESIA
6ol 401
[ I Z Z 0 Delay 15’ Delay
E 50 ! ! I E iO = -J
m
c
C
8al
301
20
a
l lo 0
~
-1
I
2
4
6
a
Age of Pup on Conditioning Day FIG.2. Mean percentage of time spent over lemon-scented shavings by rat pups trained at either 2, 4, 6 , or 8 days of age. The pups were exposed to the shavings and then injected with LiCl after no delay (0 delay) or a 15-min delay (15’ delay). Adapted from Rudy and Cheatle (1979).
that are not predictive of the US. Trace conditioning also involves variations in the delay of reinforcement following the target CS that may have, with multiple conditioning pairings, motivational as well as associative consequences (D’Amato, Safarjan, and Salmon, 1981) and the potential for associative competition between context and CSs with longer trace intervals (Marlin, 1981). Each of these factors is likely to vary with ontogeny, and this has been verified specifically for the latter (McKinzie and Spear, 1995). A Pavlovian conditioning procedure that minimizes these problems has been used to investigate the ontogeny of short-term retention. This procedure involves nonreinforced presentation of a stimulus (CS-) just prior to presentation of a second reinforced stimulus (CS+). Presentation of a CShas been shown to enhance conditioning to the target stimulus (CS+) in experiments using visual stimuli (Kucharski, Richter, and Spear, 1985; Miller and Spear, 1989) and olfactory stimuli (Kucharski and Spear, 1984; Miller, Jagielo, and Spear, 1989), and robust one-trial conditioning occurs in quite young rat pups if the CS- immediately precedes the CS+. In one application of this procedure, to condition an aversion to a black compartment (CS-t), Kucharski et al. (1985) demonstrated that the 16-day-
260
H. MOORE ARNOLD AND NORMAN E. SPEAR
old preweanling rat does not benefit from the CS- (exposure to a white compartment) if it occurs more than 40 min prior to the presentation of CS+. Adults tolerate delays up to 12 hr before retention of the CS- no longer facilitates conditioning. Miller and Spear (1989) extended these results in a systematic study comparing retention of CS- occurrence in 16and 21-day-old rat pups. In both age groups it was observed that the strength of conditioning to the black compartment was dependent on the delay between the CS- and the CS+, indicating, as Miller and Spear (1989) suggested, that an active representation of CS- is required at the time of conditioning to CS+ in order for an association to be formed. These data also demonstrated that whereas the 21-day-old rat tolerated delays of about 40 min between CS- and CS+, the 16-day-old pup did not learn with delays longer than 30 min. Further experiments using olfactory stimuli as the CS- and CS+ have extended this analysis to pups 8 and 12 days of age (Miller et al., 1989). In these experiments it was observed that the 12day-old pup tolerated a CS-/CS+ interval of 20 min, but the 8-day-old pup tolerated a CS-/CS+ interval of only about 10 min. Common to all of these experiments using the CS-/CS+ conditioning procedure is the observation that the younger the animal, the shorter the delay between the CS- and CS+ needed for conditioning to the CS+. This finding is consistent with other observations of short-term retention in the developing animal. The clearest of these can be found in another series of experiments by Miller and his colleagues (Miller and Spear, 1989; Miller, Jagielo, and Spear, 1989, 1990, 1991). The prototypic experiment tested retention of an aversion to a CS+ (odor) that had been immediately preceded by a CS- (alternative odor) and paired on a single occasion with footshock. Thereafter, the rat pups were placed in isolation (which turns out to be very important, as will be discussed later) for intervals of varying length, after which a test of odor preference for the CS+ was conducted (e.g., Miller et al., 1989). The relative retention observed among 8-, 12-, and 18-day-olds after short intervals is shown in Fig. 3. It was clear that more rapid forgetting occurred the younger the animal. If we define the performance of control animals, given unpaired presentations of the CS + and footshock, as “complete forgetting” (in that no evidence of learning was present), 8-day-olds reached this level within 45 min, 12-day-olds within 75 min, and 18-day-olds did not reach this level of forgetting until 150 min after conditioning. Collectively this evidence confirms that infantile amnesia is also evident after short-term intervals and is not limited to long-term retention over intervals that constitute a significant portion of an animal’s life-span (and the accompanying systemic changes, such as brain development associated with long-term growth). This is an important point to which we shall return.
261
INFANTILE AMNESIA
Better 20
9
Worse o
4
- 7---1-
10
--r- I
1 - - i -
20
30
40
60
T
90
-
-
120
150
Retention Interval (min.) FIG.3. The retention index (calculated by subtracting the mean of the paired condition from the mean of the unpaired condition) for 8-, 12-, or 18-day-old rat pups after various retention intervals between training and testing. Adapted from Miller et al. (1989).
Clearly, infantile amnesia is more general than simply not being able t o remember one’s childhood memories as an adult. Short-term retention paradigms suggest that those factors responsible for infantile amnesia must be active as an event is being learned or within a very short time following acquisition. Tests of instrumental learning have confirmed that short-term retention is less effective the younger the animal during early ontogeny. Green and Stanton (1989) tested delayed alternation of location within a T maze apparatus. In their experiment, pups 15, 21-23, and 27-29 days old were trained on a discrete-trials alternation task. In this task animals were forced to one side of the T maze on the first run, followed immediately by a free choice run in which both arms of the maze were open. Animals were rewarded for choosing the arm not visited on the previous run. In this task each trial is independent of each preceding or following trial. Although older animals were able to master this task, 15-day-old rat pups did not learn even when there was only a brief (5-10 s) delay between the forced choice and the free run, suggesting that short-term or working memory is less developed in this youngest set of animals.
262
H. MOORE ARNOLD AND NORMAN E. SPEAR
Castro. Paylor, and Rudy (1987) compared 21-, 28-, and 38-day-old rats on a similar version of the discrete trials alternation task using a waterfilled T maze. They observed that the number of trials required to solve the problem was directly related to the subject’s age, with younger subjects requiring more trials. Although all age groups performed at about 90 percent correct when the delay interval separating the forced run and choice run was less than 30 s, performance for the youngest animals fell to chance when the interval was increased to 60 s. The 28-day-old rat pups were able to perform above chance until the interval was increased to 180 s, but in 38-day-olds learning occurred even with this interval. Castro, Paylor, Moye, and Rudy (1990), in an investigation of the central mechanisms that may underlie differences in retention between young and adult animals, demonstrated that the cholinergic agonist physostigmine enhances short-term memory in the developing rat using the same procedures as Castro et al. (1987). After administration of a 0.02 mg/kg dose of physostigmine, the 24- to 25-day-old performed as well with a 60-s delay as with a 10-s delay between the forced run and the choice run, implying that maturation of central cholinergic systems might mediate the ontogeny of retention. Further experiments supporting the notion that maturational differences in central cholinergic systems may contribute to age-related differences in short-term retention were conducted by Moye and Rudy (1987b) using a trace conditioning procedure. These authors had found that the 21-day-old rat’s ability to condition to a visual CS was impaired relative to 25-day-old pups when the trace interval was 10 s. Pretreatment with physostigmine improved the performance of the 21-day-old such that conditioning with a 10-s trace interval was possible. In addition, pretreating 25-day-old pups with the cholinergic antagonist scopolamine disrupted the otherwise strong conditioning performance exhibited by pups of this age. Finally, Moye and Vanderynn (1988) found that physostigmine enhanced the 15-day-old rat’s ability to condition to an auditory CS separated from the US by a 10-s trace, a duration that results in poor conditioning in nontreated 15-day-old pups. These experiments taken together suggest that the younger animals’ relatively poor performance in short-term retention tests is related to their less developed cholinergic systems. Kraemer, Miller, Jagielo, and Spear (1992) found that the short-term retention in a long delay flavor-conditioning paradigm may be enhanced by prior learning experience with the to-be-conditioned stimuli. In their experiments, 19-day-old preweanling rat pups showed better conditioning to a taste or odor paired with LiCl if they had received similar training 24 hr prior to conditioning with a different taste or odor. This suggests that
INFANTILE AMNESIA
263
the young animals’ difficulty in conditioning over long delays may be due to experiential as well as physiological factors. SUMMARY In a variety of circumstances younger developing animals tend to forget more quickly than older animals over retention intervals of a few seconds or minutes. That the younger animals forget more rapidly is consistent with findings from nonassociative memory experiments and, as will be discussed later, studies of long-term retention. The implication of this observation is particularly relevant to theories of infantile amnesia that attribute the forgetting of infantile experiences to neural development between learning in infancy and testing in adulthood. It seems unlikely that neural growth over seconds or minutes is sufficient to cause significant forgetting but, even if it is, the growth over longer periods would be expected to yield much greater age-related differences in forgetting. Perhaps this will eventually be found, but comparison across the experiments completed so far gives no such indication. The infantile deficiency in short-term retention seems, for now, to limit the theoretical significance of general brain growth for understanding infantile amnesia (Spear et al., 1990).
IV. ONTOGENY OF LONG-TERM RETENTION: INFANTILE AMNESIA When asked about the events of early childhood most people fail to recall the earliest occurrences. Over half a century ago, Dudycha and Dudycha (1941) reviewed several studies that had asked humans to state their early memories. From this they concluded that failure to recall early events is most severe for those that occurred during the first 3-4 years of life, an estimate frequently confirmed by subsequent studies using more objective procedures (e.g., Sheingold and Tenney, 1982). The study of long-term retention over intervals sufficiently long to represent a substantial portion of one’s life-span is extremely difficult in the developing human, so most empirical studies of long-term retention have been done with animal models, primarily rodents. Campbell and Campbell (1962) reported perhaps the first systematic investigation of long-term memory in developing animals that did not confound age by motivational conditions. In their study 18-, 23-, 38-, 54-, and 100-day-old rats were conditioned to fear a distinctive compartment of a double-compartment shuttle box. All groups were then tested for retention of fear 0, 7, 21, or 42 days after original training. Retention, measured by avoidance of the fear side, increased dramatically with age. Eighteen-day-
264
H. MOORE ARNOLD AND NORMAN E. SPEAR
old rats showed nearly 100% avoidance of the fear side on an immediate test after training, but only chance avoidance at the 21- and 42-day retention intervals. In contrast, 54- and 100-day-old rats showed little or no forgetting at any of the retention intervals. In a second experiment, retention of conditioned punishment was investigated over 0-, 21-, and 42-day intervals. Rats were 23 and 100 days of age when conditioned and, again, juveniles were clearly inferior to adults at the longer retention intervals. Similar results have been observed frequently, for many learning tests (for reviews, see Campbell and Spear, 1972; Spear and Campbell, 1979; Spear and Rudy, 1991). In the case of instrumental aversive conditioning, for instance, Kirby (1963) trained rats to traverse a 30-in. runway at 25, 50, and 100 days of age. When tested for retention 1, 25, or 50 days after training, the 25-dayold animals were not impaired on the 1-day retention test, but had poorer retention on the 25- and 50-day retention tests (Fig. 4). It has been suggested that animals cannot remember events that occurred during their infancy because of dramatic changes in context caused by changes in the animal’s size since infancy. To compensate for possible changes in the perceived size of the environment brought about by growth
9
1
I
I
25
50
25DAY 50 DAY 100 DAY
Retention Interval (days) FIG.4. Retention of the avoidance response as measured by a relearning score (determined by subtracting the number of avoidance responses on the second day of training from the number of responses during reconditioning). From Kirby (1963).
INFANTILE AMNESIA
265
of the animal, Thompson, Koenigsberg, and Tennison (1965) and Feigley and Spear (1970) increased the size of the apparatus for the youngest age groups. Thompson et af. (1965), using the same ages and retention intervals as Kirby (1963), did not observe any age differences in retention. However, in their experiment there was no indication of retention at any age, so it was difficult to determine whether changing the size of the apparatus influenced retention in the younger animals. Feigley and Spear (1970) trained 23- and 100-day-old rats and then tested them 25 days later. Subjects 23 days old during training were conditioned in a relatively small apparatus, then tested in a larger, “age-appropriate” apparatus. Despite holding constant the size of the apparatus relative to size of the animal, the younger animals still forgot more than the older animals. Spear and Parsons (1976) also found that young animals continued to forget more quickly than adults even when the size of the apparatus was adjusted for the animal’s growth during the retention interval. Coulter, Collier, and Campbell (1976) exposed infant rats to a Pavlovian fear conditioning task (tone-shock pairings) and then tested the tone’s ability to disrupt ongoing bar-pressing behavior at different retention intervals. The authors concluded that although the animal’s physical size might conceivably be important for events that occur later in development, neurological immaturity may be a more important component of forgetting early in development. That younger animals forget more quickly than adults has also been demonstrated using escape training. Smith (1968) trained rats 25 and 100 days of age to escape shock and then tested their retention 75 days later. Rats trained at 25 days of age took longer to relearn the task than rats trained at 100 days of age. In an extension of this experiment, Campbell, Misanin, White, and Lytle (1974) trained rats 15, 17, 20, 25, and 35 days of age to escape shock. They observed that after a 7- or 14-day retention interval the three youngest groups retained less than the older animals. This conclusion holds for still younger animals as well. Misanin, Nagy, and their colleagues (Misanin, Nagy, Keiser, and Bowen, 1971; Nagy, Misanin, Newman, Olsen, and Hinderliter, 1972) trained neonatal rats and mice to perform an instrumental response in order to escape a mild footshock in a straight runway. The dependent measure was number of competing behaviors emitted by the animal in the runway, which decreases progressively with learning over successive trials. Such learning was seen in one strain of mice (C3H) by the age of 3 days, and in a relatively mixed strain of mice (Swiss-Webster) and rats (Wistar) by about 5 days of age. Despite being able to learn the task at these early ages, significant savings during retention tests were obtained only for animals that learned when 9 days of age or older. Nagy and Mueller (1973) investigated the possibility that the better retention expressed by 9-day-old mice was due to a higher degree
266
H. MOORE ARNOLD AND NORMAN E. SPEAR
of learning by these older mice, rather than to postacquisition processes more directly responsible for retention. Groups of mice trained at 7 or 9 days of age revealed that even in those cases in which degree of original learning was equal to or stronger for 7-day-old mice than for 9-day-old mice, retention of the 7-day-old mice was poorer than that of the 9-dayold mice. There were potential problems with these experiments. First, learning was seen only in the decrease in competing behaviors; time to reach the end of the maze did not decrease over training trials. Second, the decrease in competing behaviors was just as likely when animals were trained with noncontingent footshock, implying ontogenetic changes in reactivity to shock rather than learning (Campbell, Riccio, and Rohrbaugh, 1971). Despite these potential problems for interpretation, this series of experiments was important in expanding and elaborating the evidence that the ontogeny of retention capacity may be dissociated from the ontogeny of learning capacity, and the general results have been replicated. Retention of passive avoidance also improves with age (Feigley and Spear, 1970; Fox, 1971; Schulenberg, Riccio, and Stikes, 1970). Comparing 25-day-old and adult rats in a task in which subjects were punished with footshock for stepping from one compartment to another, Feigley and Spear (1970) observed that younger animals remembered less than adults 28 days after training. Schulenberg et al. (1970) examined retention of rats trained at 15, 21, and 100 days of age with the same task. Again, it was observed that the two younger age groups showed significantly less retention of passive-avoidance training than adults 24 days after training. In a study with dogs, Fox (1971) examined retention of puppies trained not to approach a human observer. He observed that 5-week-old puppies were not able to retain the task for 7 or 14 days, whereas older puppies (13 weeks old) were able to remember the task at both retention intervals. Campbell et al. (1974) examined several groups of rats ranging from 16 to 100 days old for their retention of a step-through passive-avoidance task. Performance of pups in the younger age groups was consistently worse than that of adults, and more so the longer the retention interval. After 21 days the youngest pups showed no memory of the original training task, whereas the adults showed virtually no forgetting. Finally, in experiments by Stehouwer and Campbell (1980), rat pups 10-15 days old were given passive-avoidance training in which the response was to avoid touching the wall in a circular chamber. It was observed that retention improved with age; passive avoidance was seen for no more than 6 hr after training in the 10-day-old rat pup but was evident at least 5 days later in the 15-day-old pups. Stehouwer and Campbell (1980) suggested that maturation of storage and/or retrieval mechanisms, rather than perceptual or experiential changes induced by
INFANTILE AMNESIA
267
development, was responsible for improved retention during ontogeny. This is similar to a view expressed by Spear (1979a), discussed later. Faster forgetting by younger animals has also been observed in appetitive as well as aversive conditioning. For instance Campbell, Jaynes, and Misanin (1968) trained rat pups 23-26 days of age on a discriminated bar-press task with food as the reinforcer. Tests for retention 38, 75, and 150 days later revealed that the pups did not retain the discrimination or the basic response to the same degree as animals trained as adults and tested after the same delays. Studies examining conditioned taste aversions have also supported the general finding that younger animals forget faster than do more mature animals. Although there has been some disagreement as to the exact age at which retention for a conditioned taste aversion becomes adultlike (cf. Schweitzer and Green, 1982), there is agreement that young rat pups forget taste-illness associations more quickly than d o adults (Campbell and Alberts, 1979; Schweitzer and Green, 1982;Steinert, Infurna, and Spear, 1980). Steinert et al. (1980) observed poorer retention of a conditioned taste aversion in 18-day-old rat pups 60 days after training compared to that of animals trained as adults, after careful equation of retention after a short interval. However, other studies have observed “adultlike” retention in 18- to 20-day-old pups and have not noted deficits in retention among rats older than 10-12 days (Campbell and Alberts, 1979; Schweitzer and Green, 1982).
V. PERSPECTIVES ON INFANTILE AMNESIA The study of long-term retention has been guided by two underlying theoretical questions. Each has served to generate valuable research toward an understanding of the general phenomenon of infantile amnesia. The first of these questions is something like this: “At what age does the ability to retain information develop?” For example, Schweitzer and Green (1982) state that “It is important . . . that the age of maturation of extended retention be established. Only then can physiological and/or psychological changes occurring before or after this age be manipulated to determine the nature of these changes” (p. 793). Although this seems to be a reasonable position, it has become increasingly evident over the past two decades that there is no one specific time point during ontogeny identifiable as the moment at which long-term memory develops. Careful examination of the review of experiments described above will demonstrate that the age at which retention “develops” is not consistent across experiments or tasks. Whereas some experiments do not reveal retention until the pup is quite
268
H. MOORE ARNOLD AND NORMAN E. SPEAR
old (e.g., Campbell and Campbell, 1962),other studies have shown retention for events occurring during gestation (Stickrod et al., 1982). Perhaps a more appropriate way to ask this question is “At what age does retention become adultlike?” This question should be tempered with the understanding that is may appear empirically to be a task-specific issue, dependent at least on what is learned and the circumstances or context of the learning episode, with the ultimate determinants found in age-related differences in encoding (e.g., Spear, 1984). The second question is aimed at understanding why younger animals forget more quickly than more mature animals. Several different explanations have been offered to account for infantile amnesia. For instance, changes in the central nervous system between infancy and adulthood may be responsible for the poor retention of infancy (Campbell and Spear, 1972). An important point can be remade here. If one believes there is not sufficient brain growth during a period of a few seconds or minutes to promote memory loss, then the infantile deficit in short-term retention tasks described earlier restricts the importance of such growth in fully accounting for infantile amnesia. Another explanation assumes that infants and adults differ in terms of what is selected for storage in memory (Gordon, 1979; Spear, 1979b; Spear and Kucharski, 1984). There is a limit to the usefulness of repeatedly demonstrating that younger animals do, in fact, forget more quickly than more mature animals. With the generality of infantile amnesia firmly established (Campbell and Spear, 1972; Campbell and Coulter, 1976), researchers have begun to spend more time investigating potential causes of infantile amnesia. Although the problem of infantile amnesia is often formulated as a dichotomy between physiological factors and psychological factors, it has been pointed out that such a dichotomy cannot truly exist (Spear, 1979b). A more fruitful approach to the study of infantile amnesia suggests that researchers focus on the consequences of immaturity during original memory storage and the effects of growth, in a general sense, interceding between storage and retention. A. EMPHASIS ON ENCODING DURING MEMORY ACQUISITION It has been proposed by Spear and his colleagues that younger animals may be less or differently selective than adults in storing events in memory (Mellon, Kraemer, and Spear, 1991; Spear and Kucharski, 1984; Spear and Molina, 1987). An extension of this viewpoint is that infant animals may select attributes of the training episode that happen to be forgotden more quickly than other, more stable attributes (Gordon, 1979; Spear, 1979b). For example, Riccio, Richardson, and Ebner (1984) suggested that contextual
INFANTILE AMNESIA
269
stimuli may be forgotten more quickly than CS attributes, and some recent evidence suggests that infants may have a high priority for processing contextual information (Lariviere, Chen, and Spear, 1990; McKinzie and Spear, 1995; Solheim, Hensler, and Spear, 1980). This would suggest that infants do not necessarily forget all events of an episode faster than adults. Differences in retention may be an artifact of differences in the selection of attributes for encoding. What appears to be faster forgetting in infants, then, could be a result of the infant’s greater dependence on attributes that are forgotten more quickly by both immature and mature animals. The study of infantile amnesia has moved from experiments demonstrating poorer retention in young animals and trying to determine the age at which memory “develops” to a closer analysis of the processes that occur during initial memory storage, or between this point and retrieval, that might contribute to poorer retention in younger animals (Arnold and Spear, 1995; Parsons et al., 1973; Smith and Spear, 1981). An example is a series of experiments comparing stimulus selection in infants and adults (Kucharski and Spear, 1985). Preweanling and adult rats were given pairings of one or two solutions (coffee and/or sucrose) followed by illness. Subsequent tests of the preference for the separate solutions revealed age-related differences in the conditioned aversions. Whereas adult animals were more likely to show overshadowing-reduced learning of one element of a compound due to the presence of a second element-the infant animals behaved quite differently. Rather than overshadowing, infant rats more readily demonstrated potentiation-enhanced learning of one element due to the presence of a second element. The infants acted as if they had learned more about sucrose when it was presented at the same time as the coffee solution, whereas adults acted as if they had learned less. Infant animals also show less stimulus selection than adults in processing redundant stimuli that are more contextual in nature. Solheim et al. (1980) demonstrated that in some circumstances of instrumental learning pups were more sensitive to changes in the context than were adult animals. Lariviere et al. (1990) found that a contextual odorant present during visual/ location Pavlovian conditioning affected this conditioning in infants but not in adults and tended to be learned more effectively by preweanlings than by adults. McKinzie and Spear (1995) have shown that in some circumstances, infant pups given auditory conditioning in a highly salient conditioning context learn more about the context than do adults and are more adept at auditory conditioning the more salient the context. In contrast, adults show less auditory conditioning when context is more salient and learn less about the context when auditory conditioning occurs than when it does not. The general observation is less stimulus selection in infants than in adults.
270
H. MOORE ARNOLD AND NORMAN E. SPEAR
B. EMPHASIS O N POSTACQUISITION PROCESSES We (Arnold and Spear, 1995) have published a series of experiments that examine factors that can intercede between storage and retrieval and influence the rate of forgetting in younger animals (and perhaps older animals). These experiments also illustrate the importance of context during the retention interval and complement the sensitivity of preweanling rats to context described earlier. We described earlier a series of experiments by Miller et al. (1989, 1990, 1991) testing age-related differences in short-term retention (shown in Fig. 3). Note that for 18-day-old rats forgetting was complete by about 150 min following training. In that study rat pups spent the entire retention interval in isolation. The question in the following series of experiments was “How does this context influence retention during the retention interval?” Our first experiment was straightforward. Eighteen-day-old rat pups were trained in the same fashion used by Miller et af. in their studies (e.g., Miller et af., 1991). Briefly, pups were first trained in an odor aversion task; an odor (CS+) was paired on a single occasion with footshock. All pups were tested 24 hr following the learning episode for their preference for the target odor (CS+) relative to a novel odor. Prior to the retention test separate groups of pups were removed from the home cage and placed in isolation for 0, 15, 60, or 180 min. As can be seen in Fig. 5, retention in pups tested directly out of the home cage (0 min in isolation) was quite strong. Retention dropped off for pups tested after isolation and was quite poor at the longest interval. Thus, despite an identical retention interval, the pretest environmental context of the rat pups (placement in isolation) had a profound impact on retention. Why such a brief period in isolation should disrupt retention is not at all obvious. Our next step in this investigation was to systematically eliminate potential explanations of this isolation-induced disruption of retention. One possibility was that the rat pups were expressing state-dependent retention of the odor aversion. Hofer (1987) has shown that separation of the rat pup from its mother can result in many physiological and behavioral changes in the young rat. We examined whether these changes were profound enough following a 3-hr isolation to change the pup’s state during the retention test. Using the exact conditioning procedure described previously, in a 2 X 2 design we isolated pups or left them in their home cage with their mother and siblings for 3 hr before conditioning andlor before testing. The results (shown in Fig. 6) revealed that there was no evidence in support of state-dependent retention. That is, training and testing in similar states (derived from isolation vs. remaining at home) did not improve retention. What was clear in this experiment, as in the previous study,
271
INFANTILE AMNESIA
50
1
+
r
/
30
i
0 20-
E
i=
T
6 Unpaired -DPaired
10-
6 Paired
0 ‘
I
0
I
I
I
60 180 Time in Isolation (min) 15
FIG. 5 . Mean seconds spent on the CS+ side of the odor preference chamber after 0, 15, 60, or 180 min in isolation. All pups were tested 24 hr after conditioning. Solid circles represent subjects given paired training and open squares unpaired training. All groups were tested 24 hr after conditioning. Vertical bars represent SEMs. From Arnold and Spear (1995).
40
1
Home-Home
T
T
Home-lso
Iso-Home
Iso-Is0
Treatment Condition FIG. 6. Mean seconds spent on the CS+ side of the odor preference chamber. “Home” represents pups that were in the home cage and “Iso” represents pups that were in isolation either 3 hr before conditioning (in front of the dash “-”) and/or 3 hr before the odor preference test (following the dash). Testing occurred 24 hr after conditioning. Vertical bars represent SEMs. From Arnold and Spear (1995).
272
H. MOORE ARNOLD A N D NORMAN E. SPEAR
was that 3 hr of isolation prior to the odor preference test impaired retention substantially, regardless of whether the pups were isolated or not prior to conditioning. However, we still did not have an explanation for this impairment. We then considered whether being moved to the new (isolation) environment, which might have included olfactory stimuli somewhat more novel than those present in the home cage, may have resulted in associative interference with retention of the acquired odor aversion. This experiment also permitted a preliminary test of whether the critical component of the pretest context removed by isolation was interaction, particularly nutritional interaction, between the rat pup and its mother during the few hours prior to the retention test. To examine these possibilities each pup in one group spent the last 3 hr of the 24-hr retention interval with its mother and siblings in a new cage with a distinct odor (e.g., vanilla) that was changed to a different odor (e.g., banana) every half hour, resulting in exposure to 6 novel odors. Other pups remained in the home cage with their mother and siblings, but the mother was anesthetized, which blocks milk letdown; hence the pups were unable to acquire milk. Other pups remained in their home with a nonanesthetized mother. Although the novel olfactory environment did not disrupt retention, pups that remained in their home cage with their anesthetized mother were significantly poorer in their retention of the odor aversion than pups whose mother remained awake for the entire retention interval (see Fig. 7). We were surprised that pups left with an anesthetized dam would show a disruption in retention similar to that of isolated pups. Many studies have shown that an anesthetized dam alleviates many of the reactions to isolation such as corticoid response to stress (Stanton and Levine, 1990), cardiac measures of fear of a novel environment (Richardson, Siegel, and Campbell, 1988), and some of the other physiological reactions to separation (Hofer, 1987). Rather than coming closer to a resolution of the mechanism underlying the contextual manipulation disrupting retention, we seemed to be generating even more questions. However, it was becoming clear that retention can be markedly influenced by contextual manipulation during the retention interval. In our next experiment we tested further the potential importance of the nutritional consequences of 3 hr of isolation or being with an anesthetized (nonlactating) mother and explored the conditions necessary for pups to maintain retention of the conditioned odor aversion. Four groups were conditioned as described in the previous experiments, pairing an odor with footshock, and tested 24 hr later. During the final three hours of the retention interval pups were placed with either their own mother, an anesthetized dam, a lactating foster dam, or a nonlactating foster dam. However, as can
-
273
INFANTILE AMNESIA
Paired
0Unpaired 50
h
0 40 P)
v)
Y
+ 30
v)
0
i
0 20
i? c- 10 0
31 I Anesth
Home
Isolation
Novel
Treatment Condition
FIG.7. Mean seconds spent on the CS+ side of the preference chamber 24 hr after paired (solid bars) or unpaired (open bars) training. Three hours prior to the test, pups were left in their home cage (Home), placed in isolation, given six novel odors every half hour for 3 hr (Novel), or placed with an anesthetized dam (Anesth). Vertical bars represent SEMs. From Arnold and Spear (1995).
be seen in Fig. 8, retention was quite strong for all conditions with the exception of those pups with the anesthetized dam. Subsequent studies revealed that retention remained intact when pups spent the final 3 hr of a 24-hr retention interval with their father or simply in a group of other pups. Pups separated from the dam in groups of 4 were not disrupted in their retention. In addition, tactile stimulation presented to singly isolated pups did not alleviate the retention deficit produced by the brief isolation. The last experiment we consider in this section focused on whether returning pups to their homes would reinstate retention, once retention had been disrupted by isolation. In this experiment, using the same conditioning procedure as our previous experiments, six separate groups of rats were conditioned to avoid an odor by pairing it with a mild shock. The next day five of these groups were isolated for 3 hr to produce a disruption in retention of the odor aversion. That this treatment was effective in disrupting retention was confirmed by testing one of these groups immediately following 3 hr of isolation. The other four isolated groups were then returned to their home cage with their mother and siblings for 30 s, 10 min,
274
H. MOORE ARNOLD A N D NORMAN E. SPEAR
50
1 T
Mother
Lad
Non lact
Anesth
Maternal Treatment FIG.8. Mean seconds spent on the CS+ side of the preference chamber 24 hr after conditioning. For 3 hr prior to the odor preference test pups were placed with either their own mother (Mother), a lactating foster dam (Lact), a nonlactatingfoster dam (Non lact), or an anesthetized dam (Anesth). The figure shows the results for pups that had their temperature measured only once following the odor preference test. Vertical bars represent SEMs. From Arnold and Spear (1995).
1 hr, or 3 hr. The final group remained in the home cage for the entire retention interval, which was about 25 hr in length. Although pups returned to their home cage following isolation for shorter durations continued to exhibit impaired retention, those returned to the home cage for 3 hr following a 3-hr isolation expressed the odor aversion as effectively as pups that spent the entire retention interval in their home cage (see Fig. 9). This series of experiments demonstrates the importance of the environmental context during the retention interval. Removing pups from their home cage and placing them in isolation for as little as 3 hr results in relatively poor performance on a subsequent test of retention. Other manipulations of context between conditioning and testing also can disrupt retention, as seen in those pups who experienced an anesthetized dam during the retention interval. Finally, returning rat pups to their home cage following isolation-induced forgetting can alleviate this forgetting. Although this particular series of experiments tested pups at only one age, other studies have demonstrated that contextual stimuli associated with the pup’s home cage have influences on learning in developing animals that are not present in older animals (see Spear, Kucharski, and Hoffmann,
275
INFANTILE AMNESIA
40
1
T
Is0
30Sec 10Min
1 Hr
3Hr
Home
Treatment Condition FIG.9. Mean time spent over the CS+ 24 h r after conditioning. Pups were removed from the home cage for 3 hr and were then tested (10) or returned to the home cage for 30 s, 10 min. 1 hr. or 3 hr. One group remained in the home cage for the entire retention interval (Home). Vertical bars represent SEMs. From Arnold and Spear (1995).
1985, for a review). If future studies confirm that these effects are unique to younger animals, their importance for understanding infantile amnesia could be significant.
VI. INFANTILEAMNESIA As
AN
“ONTOGENETIC ADAPTATION”
It has been suggested frequently that the failure to recall details from events early in childhood may be adaptive (e.g., Bjorklund and Green, 1992; Oppenheim, 1981; Spear, 1988,1990). A similar suggestion was made by Freud (1905/1953), that early memories are traumatic due to their sexual nature and are actively repressed to accomplish social adaptation. However, this is quite different from the view expressed by Bjorklund and Green (1992). These authors propose, as have others (e.g., Lennenberg, 1967; Oppenheim, 198l), that some aspects of the early developing cognitive system are qualitatively different from those later in life and are well suited to attain, in humans, for example, important cognitive-social milestones such as attachment and language. Bjorklund and Green go on to suggest that “many of the experiences of infancy and childhood are relevant only to that time in development. Having vivid memories of early experiences
276
H. MOORE ARNOLD AND NORMAN E. SPEAR
. . . may interfere with and contradict the knowledge that is needed to function in later environments” (p. 50). Oppenheim (1981) described the presence of neurobehavioral characteristics of immature animals that have a specific role in survival during infancy or youth but disappear when they are no longer necessary. These ontogenetic adaptations are not simply incomplete versions of adult characteristics but serve specific adaptive functions for the developing animals. Oppenheim suggested that “perhaps even the absence of adult capabilities may be developmentally adaptive . . . and should be considered in any comprehensive theory of ontogeny” (p. 92). The following passage from Oppenheim (1981) expresses how this author would have us view the processes of development in relation to ontogenetic adaptations: development is not always progressive and constructive. For example, immature animals frequently inhabit environments that are markedly different from those of the adult. Consequently, each of these stages may have required the evolution of specific morphological, biochemical, physiological, and behavioral mechanisms which are different from the adult, and which may require modification, suppression or even destruction before the adult stage can be obtained. (p. 74)
€ all and Oppenheim (1987) describe an ontogenetic adaptation as an “eai y
behavior [that] may serve some immediate adaptive role for the embryo, fetus, or infant” (p. 116). Like Oppenheim (1981), Turkewitz and Kenny (1982) suggested that the lack of adult characteristics may be an adaptive feature of ontogenetic development. They proposed that early limitations in sensory and motor systems may play adaptive roles in ontogeny by facilitating perceptual organization. For example, the limited motor capacities of young animals prevents them from wandering from their mother, enhancing their chances of surviving to reproductive age. The sensory limitations of many newborn and juvenile animals could be adaptive by reducing the amount of information infants must deal with, aiding in constructing a simplified comprehensible world. Turkewitz and Kenny (1982) noted that developmental rates of different sensory systems are unequal and onset of their functioning is sequential. They proposed that such differential onset results in relative independence among emerging systems, thereby reducing competition and promoting regulation of subsequent neurogenesis and functioning. Substantial evidence indicates that there may be a gap during ontogeny between the onset of function in a given sensory system for some purposes and the time at which that system can take part in the processes of learning and memory (e.g., Rudy et af., 1984). It might be argued that the sequence of development of detection and learning processes within a sensory system is also particularly adaptive for the immature animal in a manner similar
INFANTILE AMNESIA
277
to that described by Turkewitz and Kenny (1982) for the sequential development of different sensory systems. The connection is not obvious, however-why should detection and learning be competitive in the way that audition and vision are competitive?-and it is correspondingly less obvious why the earlier emergence of learning than memory (retention) should be adaptive. It is easier on other bases to generate reasons why long-term retention of the events of infancy might be ineffective, but less easy t o understand why short-term retention for these events also should be relatively ineffective. It seems clear that in order for a behavior to be considered ontogenetically adaptive it must serve some benefit for the infant animal that would be less so for an adult (Hall and Oppenheim, 1987; Oppenheim, 1981). Thus, in order to consider infantile amnesia on ontogenetic adaptation, one must be able to demonstrate that it is beneficial to the developing animal t o forget the events of its infancy, despite the lack of benefit for forgetting events of adulthood. It could be argued, for instance, that the infant moves from one environment to another during development and that information acquired in one environment is not appropriate for later environments. Thus, animals that are moving relatively rapidly from fetus, to neonate, t o juvenile, would tend to forget more quickly than the adult, which, having reached maturity, is in a more stable environment so that the accelerated rate of forgetting is no longer beneficial. However, it is not at all clear that forgetting of all early events is, in fact, beneficial to the animal. Another possibility is that infantile amnesia is a consequence of how infant animals process sensory information (Spear, 1979b; Turkewitz and Kenny, 1982). That is, forgetting of infancy may be the result of infantile processes for the selection and encoding of sensory information. For example, the sensory development of the young animal can dictate in very obvious ways which stimuli from the environment are selected for further processing. It has been suggested (Turkewitz and Kenny, 1982) that the limitations on infant sensory functioning may produce adaptive advantages for infants by facilitating the organization of perceptual information. By expanding this idea to encompass the encoding and storage processes involved in infant memory, it becomes reasonable to suggest that although some incidental benefit may result from instances of infantile amnesia, fast forgetting for all of infancy might not be adaptive. The learning and memory processes used by the infant might instead produce their adaptive advantage, with fast forgetting as a mere by-product or consequence of these processes. This is consistent with the suggestion that many early events organized for memory by infantile encoding might profitably be forgotten (Spear, 1988,1990), but this forgetting has been assumed to occur through mechanisms associated with the development of stimulus selection. There
278
H. MOORE ARNOLD AND NORMAN E. SPEAR
may be no need to postulate that the forgetting of infancy in general, that is, forgetting of all events of infancy, is adaptive, and it seems unlikely that it would be so.
VII. SUMMARY A N D COMMENT This review confirms that for memories acquired early in development, prior to attainment of adult levels of most indices of brain growth, forgetting is more rapid than for memories acquired later in life. This was illustrated for nonassociative memories as well as for associative memories, and for tests given a few minutes or hours after their acquisition or a few days or weeks after their acquisition. The data base for this conclusion is quite broad and ranges from tests involving habituation of cardiac responding to a variety of instances of Pavlovian conditioning and instrumental avoidance learning. Although this review covered primarily experiments with rodents, the basic principles described are true for all altricial mammals tested so far, including humans. The review confirmed the prediction made thirty years ago by Campbell (1967) and stated at the beginning of this chapter: there is no evidence that retention for memories acquired early in development is ever better than that for memories acquired later on. The invariant confirmation of this prediction becomes more striking in view of the substantial variance in degree of infantile amnesia observed in animals. Infantile amnesia can be at least partially prevented or alleviated if animals are given a sufficiently high degree of original learning, wide distribution of learning episodes, or reactivation treatments presented between training and testing to maintain or reinstate the accessibility of the memory (e.g., Potash and Ferguson, 1977; Spear, 1979b; Campbell and Jaynes, 1966). Yet the effect of such treatment on infantile amnesia per se is somewhat problematic because, although these circumstances clearly reduce forgetting for the memories of infancy, it is not clear that the reduction is different from that observed for memories acquired later in development. Implicit in this review is the value of a comparative analysis of infantile amnesia. There are a few questions surrounding infantile amnesia that may apply only to humans, such as the role of the infant’s identification of self and the conscious processing of memories. Yet there are many more basic questions that can only be answered by studying infantile amnesia in animals. Given the historic implications for the role of trauma associated with early memories prone to infantile amnesia, it is quite clear that experimental variation in the emotional content of early memories is critical to its understanding and this can be done to any reasonable degree only with animals.
INFANTILE AMNESIA
279
The existence of infantile amnesia over intervals of a few minutes suggests that postlearning changes in the central nervous system are not likely to be the primary cause of infantile amnesia. This does not rule out the possibility that age-related differences in the structure or physiology of the brain might still provide significant variance in infantile amnesia at the time of learning, at the time of retrieval for retention tests early in development, or between training and a test given long after development is complete. It is obvious that the experiments necessary to test these factors can be accomplished only in animals. A final example is the environmental control that can be exerted in animals between a critical episode of infancy and the test for retention of the memory of that episode. Control over these events can be obtained in animals to a degree quite impossible in developing humans. This is important in view of the wide variety of postacquisition events that can drastically alter the expression of memories acquired at any age (Spear et al., 1990; Spear and Riccio, 1994). A pertinent instance from the current legal and social controversy over the veracity of memories for early traumas is the profound role of suggestion, and other modes of questioning introduced between the episode and a test for its retention, on what is in fact expressed about that episode (Ceci and Loftus, 1994; Ceci, Loftus, Leichtman, and Bruck, 1994). Yet the effects of mere isolation during infancy on retention described in the present chapter (Arnold and Spear, 1995) imply that more fundamental processes might be responsible for the forgetting of early traumas. Perhaps these are endocrine in origin and central in effect and yield nonassociative interference in retention of early memories, or perhaps the circumstances and consequences of nurturing are required to make early memories retrievable. Whether such effects are unique to infancy, and whether they form part of an ontogenetic adaptation in which forgetting is profitable to the developing animal, are issues that can be decided only with further tests using animals.
References Arnold, H. M., Robinson, S. R., Spear, N. E., and Smotherman, W. P. (1993). Conditioned opioid activity in the rat fetus. Behuv. Neurosci. 107, 963-969. Arnold, H. M., and Spear, N. E. (1995). Isolation disrupts retention in preweanling rat pups. Behuv. Neurosci. 109, 744-758. Bjorklund, D. F., and Green, B. L. (1992). The adaptive nature of cognitive immaturity. Am. Psychol. 47,46-54. Campbell, B. A. (1967). Developmental studies of learning and motivation in infra-primate mammals. In “Early Behavior: Comparative and Developmental Approaches” (H. W. Stevens, E. H. Hess, and H. R. Rheingold, eds.). Wiley, New York.
280
H. MOORE ARNOLD AND NORMAN E. SPEAR
Campbell, B. A., and Alberts, J. R. (1979). Ontogeny of long-term memory for learned taste aversions. Behav. Neur. Biol. 25, 139-156. Campbell, B. A., and Ampuero, M. (1985). Dissociation of autonomic and behavioral components of conditioned fear during development in the rat. Behav. Neurosci. 99, 10891102. Campbell, B. A., and Campbell, E. H. (1962). Retention and extinction of learned fear in infant and adult rats. J. Comp. Physiol. Psychol. 55, 1-8. Campbell, B. A., and Coulter, X. (1976). The ontogenesis of learning and memory. In “Neural Mechanisms of Learning and Memory” (M. R. Rosenzweig and E. L. Bennett, eds.), pp. 209-235. MIT Press, Cambridge, Massachusetts. Campbell, B. A., and Jaynes, J. (1966). Reinstatement. Psychol. Rev. 73,478-480. Campbell, B. A., Jaynes, J., and Misanin, J. R. (1968). Retention of light-dark discrimination in rats of different ages. J. Comp. Physiol. Psych. 66,467-472. Campbell, B. A., Misanin, J. R., White, B. C., and Lytle, L. D. (1974). Species differences in ontogeny of memory. Support for neural maturation as a determinant of forgetting. J. Comp. Physiol. Psych. 87, 193-202. Campbell, B. A., Riccio, D. C., and Rohrbaugh, M. (1971). Ontogenesis of learning and memory: Research and Theory. In “Second Western Symposium on Learning: Early Learning” (M. E. Meyer, ed.), pp. 76-109. Western Washington State College, Bellingham, Washington. Campbell, B. A., and Spear, N. E. (1972). Ontogeny of memory. Psychol. Rev. 79,215-236. Campbell, 8 . A., and Stehouwer, D. J. (1980). Retention of habituation and sensitization in neonatal rats. Behav. N e w . Biol. 29, 190-202. Carew, T. J., Pinsker, H. M., and Kandel, E. R. (1972). Long-term habituation of a defensive withdrawal reflex in Aplysia. Science 175, 451-454. Castro. C. A., Paylor, R., Moye, T. B., and Rudy, J. W. (1990). The cholinergic agent physostigmine enhances short-term-memory-based performance in the developing rat. Behav. Neurosci. 104,390-393. Castro, C. A., Paylor, R., and Rudy, J. W. (1987). A developmental analysis of the learning and short-term-memory processes mediating performance in conditional-spatial discrimination problems. Psychobiol. 15, 308-316. Ceci, S.J., and Bruck, M. (1993). Suggestibility of the child witness: A historical review and synthesis. Psych. Bull. 113,403-439. Ceci, S. J., and Loftus, E. F. (1994). “Memory work”: A royal road to false memories? Special Issue: Recovery of memories of childhood sexual abuse. Appl. Cog. Psych. 8, 351-364. Ceci, S. J., Loftus, E. F., Leichtman, M. D., and Bruck, M. (1994). The possible role of source masattributions in the creation hypnosis and delayed recall: I. Internatl. J. Clin. Exp. Hyp. 42,304-320. Cohen, L. B., and Gelber, E. R. (1975). Infant visual memory. In “Infant Perception from Sensation to Cognition” (L. Cohen and P. Salapatek, eds.), pp. 347-403. Basic Visual Processes, Vol. I . Academic Press, New York. Coulter, X.,Collier, A. C., and Campbell. B. A. (1976). Long-term retention of early Pavlovian fear conditioning in infant rats. J. Exp. Psych. Anim. Behav. Process 2, 48-56. D’Amato, M. R., Safarjan, W. R., and Salmon, D. (1981). Long-delay conditioning and instrumental learning: Some new findings. In “Information Processing in Animals: Memory Mechanisms,” (N. E. Spear and R. R. Miller, eds.), pp. 113-142. Erlbaum, Hillsdale, New Jersey. Dudycha, G. J., and Dudycha, M. M. (1941). Childhood memories: A review of the literature. Psych. Bull. 38,668-682.
INFANTILE AMNESIA
281
Fagan, J. F. (1974). Infant recognition memory: The effects of length of familiarization and type of discrimination task. Child Develop. 45, 351-356. Feigley, D. A., and Spear, N. E. (1970). Effect of age and punishment condition on longterm retention by that rat of active- and passive-avoidance learning. J. Comp. Physiol. Psych. 73,515-526. Fifer, W. P. (1987). Neonatal preference for mother’s voice. In “Perinatal Development: A Psychobiology Perspective” (N. A. Krasnegor, E. M. Blass, M. A. Hofer, and W. P. Smotherman, eds.), pp. 111-124. Academic Press, New York. File, S. E., and Plotkin, H. C. (1974). Habituation in the neonatal rat. Develop. Psych. 7, 121-127. File, S. E.. and Scott, E. M. (1976). Acquisition and retention of habituation in the preweanling rat. Develop. Psych. 9,97-107. Fox, M. (1971). “Integrative Development of Brain and Behavior in the Dog.” The University of Chicago Press, Chicago. Freud, S. (1953). Three essays on the theory of sexuality. In “The Standard Edition of the Complete Psychological Works of Sigmund Freud, Vol. 7” (J. Strachey, ed. and trans.), pp. 125-145. Hogarth Press, London. (Original work published in 1905). Gordon, W. C. (1979). Age: Is it a constraint on memory content? In “The Ontogeny of Learning and Memory” (N. E. Spear and B. A. Campbell, eds.), pp. 271-287. Erlbaum, Hillsdale, New Jersey. Green, R. J., and Stanton, M. E. (1989). Differential ontogeny of working memory and reference memory in the rat. Eehav. Neurosci. 103, 98-105. Gregg, B., Kittrell, E. M. W.. Domjan, M., and Amsel, A. (1978). Ingestional aversion learning in preweanling rats. J . Comp. Physiol. Psych. 92, 785-795. Groves, P. M., and Thompson, R. F. (1970). Habituation: A dual process theory. Psych. Rev. 77,419-450. Hall, W. G., and Oppenheim, R. W. (1987). Developmental psychobiology: Prenatal, Perinatal, and early postnatal aspects of behavioral development. Ann. Rev. Psych. 38, 91-128. Hofer, M. A. (1987). Shaping forces within early social relationships. I n “Perinatal Development: A Psychobiological Perspective” (N. A. Krasnegor, E. M. Blass, M. A. Hofer, and W. P. Smotherman, eds.), pp. 83-110. Academic Press, New York. Howe, M. L.. and Courage, M. L. (1993). On resolving the enigma of infantile amnesia. Psych. Bull. 113, 305-326. Kirby, R. H. (1963). Acquisition, extinction, and retention of an avoidance response as a function of age. J. Comp. Physiol. Psych. 56, 158-162. Klein, S. B., and Spear, N. E. (1969). Influence of age on short-term retention of activeavoidance learning. J. Comp. Physiol. Psych. 69, 583-589. Kraemer, P. J., Miller, J. S., Jagielo, J. A., and Spear, N. E. (1992). Facilitation of long-delay conditioned taste aversion in preweanling and adult rats. Psychobiol. u),223-228. Kucharski, D., Richter, N. G., and Spear, N. E. (1985). Conditioned aversion is promoted by memory of CS-. Anim. Learn. Eehav. 13, 143-151. Kucharski, D.. and Spear, N. E. (1984). Conditioning of aversion to an odor paired with peripheral shock in the developing rat. Develop. Psych. 17,465-479. Kucharski, D., and Spear, N. E. (1985). Potentiation and overshadowing in preweanling and adult rats. J . Exp. Psychol.: Anim. Eehav. Processes 11, 15-34. Lariviere, N. A., Chen, W. J., and Spear, N. E. (1990). The influence of olfactory context on Pavlovian conditioning and its expression in preweanling (16-day-old) and adult rats. Anim. Learn. Behav. 18, 179-190. Lennenberg, E. H. (1967). “Biological Foundations of Language. Wiley, New York.
282
H. MOORE ARNOLD AND NORMAN E. SPEAR
Lindsay, D. S., and Read, J. D. (1994). Psychotherapy and memories of childhood sexual abuse: A cognitive perspective. Special Issue: Recovery of memories of childhood sexual abuse. Appl. Cogn. Psych. 8,281-338. Liu, S. X. (1928). The relation of age to learning ability of the white rat. J. Comp. Physiol. Psych. 8,75-85. Marcus, E. A,, Nolen, T. G . , Rankin, C. H., and Carew, T. J. (1988). Behavioral dissociation of dishabituation, sensitization, and inhibition in Aplysia. Science 241,210-213. Marlin, N. A. (1981). Contextual associations in trace conditioning. Anim. Learn. Behav. 9,519-523. Mellon, R. C., Kraemer, P. J., and Spear, N. E. (1991). Intersensory development and Pavlovian conditioning: Stimulus selection and encoding of lights and tones in the preweanling rat. J . Exp. Psychol.: Anim. Behav. Processes 17,448-464. McKinzie, D. L., and Spear, N. E. (1995). Ontogenetic differences in the expression of context and CS conditioning as a function of context saliency and CS-US interval. Miller, J. S., Jagielo, J. A,, and Spear, N. E. (1989). Age-related differences in short-term retention of separable elements of an odor aversion. J. Exp. Psych.: Anim. Behav. Processes 15, 194-201. Miller, J. S., Jagielo, J. A,, and Spear, N. E. (1990). Alleviation of short-term forgetting: Effects of the CS- and other conditioning elements in prior cueing or as context during test. Learn. Motiv. 21, 96-109. Miller, S., Jagielo, J. A., and Spear, N. E. (1991). Differential effectiveness of various prior cueing treatments on the reactivation and maintenance of memory. J. Exp. Psych.: Anim. Behav. Processes 17,249-258. Miller, J. S.. and Spear, N. E. (1989). Ontogenetic differences in short-term retention of Pavlovian conditioning. Develop. Psychol. 22, 377-387. Misanin, J. R., Nagy, Z . M., Keiser, E. F., and Bowen, W. (1971). Emergence of long-term memory in the neonatal rat. J . Comp. Physiol. Psych. 77,188-199. Moye, T. B., and Rudy, J. W. (1987a). Ontogenesis of trace conditioning in young rats: Dissociation of associative and memory processes. Develop. Psych. 20, 405-414. Moye, T. B., and Rudy, J. W. (1987b). Visually mediated trace conditioning in young rats: Evidence for cholinergic involvement in the development of associative memory. Psychobiol. 15, 128-136. Moye, T. B., and Vanderynn, J. (1988). Physostigmine accelerated the development of associative memory processes in the infant rat. Psychopharmacol. 95,401-406. Nagy, Z . M., Misanin, J. R., Newman, J. A., Olsen, P. L., and Hinderliter, C. F. (1972). Ontogeny of memory in the neonatal mouse. J . Comp. Physiol. Psych. 81,380-393. Nagy, Z . M., and Mueller, P. M. (1973). Effects of amount of original training upon onset of a 24-hour memory capacity. J. Comp. Physiol. Psychol. 85, 151-159. Nolen, T. G . , and Carew, T. J. (1994). Ontogeny of serotonin-immunoreactive neurons in juvenile Aplysia California: Implications for the development of learning. Behav. N e w . Biol. 61, 282-295. Oppenheim, R. W. (1981). Ontogenetic adaptations and retrogressive processes in the development of the nervous system and behavior: A neuroembryological perspective. In “Maturation and Development: Biological and Psychological Perspectives” (K. J. Connolly and H. F. R. Prechtl, eds.), pp. 73-109. J. P. Lippencott, Philadelphia. Parsons, P. J., Fagen, T., and Spear, N. E. (1973). Short-term retention of habituation by infant, weanling, and adult rats. J. Comp. Physiol. Psych. 84, 543-545. Pinsker, H., Kupferman, I., Castellucci, V., and Kandel, E. R. (1970). Habituation and dishabituation of the gill-withdraw reflex in Aplysia. Science 167, 1740-1742.
INFANTILE AMNESIA
283
Potash, M., and Ferguson, H. B. (1977). The effect of criterion level on the acquisition and retention of a one-way avoidance response in young and old rats. Develop. Psychobiol. 10,347-354. Rankin, C. H., and Carew, T. J. (1988). Dishabituation and sensitization emerge as separate processes during development in Aplysia. J. Neurosci. 8, 197-211. Riccio, D. C., Richardson, R., and Ebner, D. L. (1984). Memory retrieval deficits based upon altered contextual cues: A paradox. Psychol. Bull. 96,152-165. Richardson, R., and Campbell, B. A. (1991). Ontogeny of long-term, nonassociative memory in the rat. Anim. Learn. Behav. 19, 1-10, Richardson, R., Siegel, M. A., and Campbell, B. A. (1988). Effect of maternal presence on the fear response to an unfamiliar environment as measured by heart rate in rats as a function of age. Develop. Psychobiol. 21, 613-633. Rovec-Collier, C. K., Sullivan, M. W., Enright, M. K.. Lucas, D., and Fagan, J. W. (1980). Reactivation of infant memory. Science 208, 1159-1 161. Rudy, J. W., and Cheatle, M. D. (1979). Ontogeny of associative learning: Acqu odor aversions by neonatal rats. In “The Ontogeny of Learning and Memory” (N. E. Spear and B. A. Campbell, eds.), Erlbaum. Hillsdale, New Jersey. Rudy, J. W., Vogt, M. B., and Hyson, R. L. (1984). A developmental analysis of the rat’s learned reactions to gustatory and auditory stimulation. I n “Comparative Perspectives on the Development of Memory” (R. Kail and N. E. Spear, eds.), pp. 181-208. Erlbaum. Hillsdale, New Jersey. Schulenberg, C. J.. Riccio. D. C., and Stikes, E. R. (1971). Acquisition and retention of a passive avoidance response as a function of age in rats. J. Comp. Physiol. Psych. 74,7543. Schweitzer. L., and Green, L. (1982). Acquisition and extended retention of a conditioned taste aversion in preweanling rats. J. Comp. Physiol. Psych. 3, 791-806. Sheingold, K.. and Tenney, Y. (1982). Memory for a salient childhood event. I n “Memory Observed: Remembering in Natural Contexts.” (U. Neisser, ed.), Freeman, San Francisco. Smith, G. J., and Spear, N. E. (1981). Role of proactive interference in infantile forgetting. Anim. Learn. Behav. 9,371-380. Smith, N. (1968). Effects of interpolated learning on the retention of an escape response in rats as a function of age. J. Comp. Physiol. Psych. 3,422-426. Solheim, G. S., Hensler, J. G., and Spear, N. E. (1980). Age-dependent contextual effects on short-term active avoidance retention in rats. Behav. N e w . Biol. 30,250-259. Spear, N. E. (1973). Retrieval of memory in animals. Psych. Rev. 80, 163-194. Spear, N. E. (1979a). Memory storage factors in infantile amnesia. In “The Psychology of Learning and Motivation, Vol. 13” (G. Bower, ed.), pp. 91-154. Academic Press, New York. Spear, N. E. (1979b). Experimental analysis of infantile amnesia. In “Functional Disorders of Memory” (J. F. Kihlstrom and F. J. Evans, eds.), pp. 75-101. Erlbaum, Hillsdale, New Jersey. Spear, N. E. (1984). Ecologically determined dispositions control the ontogeny of learning and memory. I n “Comparative Perspectives on the Development of Memory” (R. Kail and N. E. Spear, eds.), pp. 325-357. Erlbaum, Hillsdale, New Jersey. Spear, N. E. (1988). Infantile learning and memory in the rat. Presidential Address, presented at meetings of the International Society for Developmental Psychobiology. November, Toronto, ON, Canada. Spear, N. E. (1990). Infantile encoding, learning and memory. Presidential Address presented at meetings of the Midwestern Psychological Association, May, Chicago, IL. Spear, N. E., and Campbell, B. A. (Eds.). (1979). “The Ontogeny of Learning and Memory.” Erlbaum, Hillsdale, New Jersey.
284
n. MOORE ARNOLD
AND NORMAN E. SPEAR
Spear, N. E., and Kucharski, D. (1984). Ontogenetic differences in stimulus selection during conditioning. In “Comparative Perspectives on the Development of Memory” (R. Kail and N. E. Spear, eds.), pp. 227-252. Erlbaum, Hillsdale, New Jersey. Spear, N. E., Kucharski, D., and Hoffmann, H. L. (1985). Contextual influences on conditioned taste aversions in the developing rat. Ann. New York Acad. Sci. 443, 42-53. Spear, N. E., Miller, J. S., and Jagielo, J. A. (1990). Animal memory and learning. Ann. Rev. Psych. 41, 169-211. Spear, N. E., and Molina, J. C. (1987). The role of sensory modality in the ontogeny of stimulus selection. In “Perinatal Development: A Psychobiological Perspective” (N. A. Krasnegor, E. M.Blass, M. A. Hofer, and W. P. Smotherman, eds.), pp. 83-110. Academic Press, New York. Spear, N. E., and Parsons, P. (1976). Analysis of a reactivation treatment: Ontogenetic deterrninants of alleviated forgetting. In “Processes of Animal Memory” (D. Medin, W. Roberts, and R. Davis, eds.), pp. 135-165. Erlbaum, Hillsdale, New Jersey. Spear, N. E., and Riccio, D. C. (1994). “Memory: Phenomena and Principals.” Allyn & Bacon, Needham Heights, Massachusetts. Spear, N. E., and Rudy, J. W. (1991). Tests of the ontogeny of learning and memory: Issues, methods, and results. In “Developmental Psychobiology: Current Methodological and Conceptual Issues” (H. N. Shair. G. A. Barr, and M. A. Hofer, eds.). Oxford University Press, Oxford. Spear, N. E., and Smith, G. (1978). Alleviation of forgetting in neonatal rats. Dev. Psychobiol. 11,513-529. Stanton, M. E., and Levine, S. (1990). Inhibition of infant glucocorticoid stress response: Specific role of maternal cues. Develop. Psychobiol. 5, 411-426. Stehouwer, D. J., and Campbell, B. A. (1980). Ontogeny of passive avoidance: Role of task demands and development of species-typical behaviors. Develop. Psychobiol. 13,385-398. Steinert, P. A., Infurna. R. N., Jardula, M. F., and Spear, N. E. (1979). Effects of CS concentration on long-delay taste aversion learning in preweanling and adult rats. Behav. New. Biol. 27,487-502. Steinert, P. A., Infurna, R. N., and Spear, N. E. (1980). Long-term retention of conditioned taste aversion in preweanling and adult rats. Anim. Learn. Behav. 8,375-381. Stickrod, G.,Kimble, D., and Smotherman, W. P. (1982). In utero taste/odor aversion conditioning in the rat. Physiol. Behav. 28, 5-7. Thompson, R. W., Koenigsberg, L. A., and Tennison, J. C. (1965). Effects of age on learning and retention of an avoidance response in rats. J. Comp. Physiol. Psych. 60,457-459. Turkewitz, G.,and Kenny, P. A. (1982). Limitations on input as a basis for neural organization and perceptual development: A preliminary theoretical statement. Develop. Psychobiol. 15,357-368. Vogt, M. B., and Rudy, J. W. (1984). Ontogenesis of learning: I. Variation in the rat’s reflexive and learned responses to gustatory stimulation. Develop. Psychobiol. 17, 11-33.
ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 26
Regulation of Age Polyethism in Bees and Wasps by Juvenile Hormone SUSANE. FAHRBACH DEPARTMENT OF ENTOMOLOGY UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN URBANA, ILLINOIS
The function of JH in the sterile female caste of honeybees cannot be compared with the function in fertile female insects. Its function consists apparently in the control of a number of physiological processes which are probably in close relation to behavioural changes associated with the division of labour. It seems likely that the increasing JH titre is responsible for inducing the transformation of hive bees into field bees. (Rutz, Gerig, Wille, and LUscher, 1976, p. 1490) The hypothesis of polyphenism of honeybees being controlled by juvenile hormone has still a deficient base. (Fluri, Liischer, Wille, and Gerig, 1982, p. 65) As a bee ages, she may undergo a programmed change in central nervous system (CNS) response thresholds to task-associated stimuli, mediated by changes in JH titre. (Page and Robinson, 1991, p. 133)
I. INTRODUCTION A N D THE DIVISION OF LABOR A. EUSOCIALITY
A long human appreciation of the division of labor characteristic of insect societies has in the twentieth century matured into an understanding rooted in evolutionary biology. The “advantage” of social life is recognized to lie in the efficiencies of task specialization and cooperative defense of the nest. The ecological success of the termites, ants, bees, and wasps results in large part from the division of labor that is at the heart of eusociality (Wilson, 1971; Oster and Wilson, 1978). Among the social insects, the primary division of labor is reproductive versus nonreproductive. Societies comprise reproductive and nonreproductive castes, with the reproductives being accorded “royal” status. Nonrepro285
Copyright Q 1997 by Academic Press All rights of reproduction in any form reserved. 0065-3454/97 $25.00
286
SUSAN E. FAHRBACH
ductives (workers) are in some species also divided into distinct physical castes, such as the major and minor workers of some ants and termites. The mechanisms of physical caste determination have been investigated, and are, depending on the group studied, possibly under genetic control or susceptible to extrinsic factors, such as larval nutritional status and larval endocrine profiles (Wilde and Beetsma, 1982; Nijhout and Wheeler, 1982; Hardie and Lees, 1985). For the social bees and wasps that form the subject of this review, caste polymorphism is restricted to females. Depending on the species, differences between reproductives (queens) and nonreproductives (workers) may be based on differences in physiology, size, external morphology, behavior, or a combination of these features (reviewed by Hardie and Lees, 1985). In some groups, queens may be distinguished from workers primarily on the basis of their reproductive function, and may be as competent as workers to build a nest, forage, and to rear offspring. Such a description characterizes the queens of many of the primitively eusocial bees, such as the bumble bees, as well as those of temperate zone social wasps that found colonies in the spring after solitary hibernation. Some halictine species exhibit varying degrees of behavioral and morphological differentiation between the queens and the worker castes (Wilson, 1971). By contrast, the queens of the honey bees (Apini) and stingless honey bees (Meliponini) are unable to function as workers under any circumstances. B. AGEPOLYETHISM Even where differences between queens and workers are notable, workers are often very nearly monomorphic. This is particularly true for workers of the well-studied European honey bee, Apis mellifera (Wilson, 1971). This means that division of labor is often not associated with distinctive external morphologies. Age polyethism, one of the most common forms of behavioral polymorphism found in bee and wasp societies, is of this type (Wilson, 1971; Hardie and Lees, 1985; Robinson, 1992). Age polyethism refers simply to division of labor on the basis of worker age. Phrases such as “temporal castes” and “temporal polyethism” are also used to indicate this phenomenon. Scholars of the social insects have recognized that division of labor on the basis of behavioral characteristics alone is not well described by the term “polymorphism,” and many writers have followed the proposal of Michener (1961) that age polyethism be described instead as a form of “polyphenism.” Although Liischer (1976) subsequently used “polyphenism” to include division of labor on the basis of morphological differences, the sense intended by Michener remains highly useful. More recently, polyphenism has been defined as “the occurrence of two or more
REGULATION OF AGE POLYETHISM IN BEES AND WASPS
287
distinct phenotypes which can be induced in individuals of the same genotype by extrinsic factors” (Hardie and Lees, 1985, p. 443). This definition is highly appropriate for a consideration of age polyethism. Age polyethism is itself a neutral term indicating no fixed sequence of tasks, but almost without exception younger workers perform tasks in the nest, while older workers forage (Wilson, 1971; Page and Robinson, 1991; Robinson, 1992). Although few other species have been studied in as much detail as the European honey bee, all of the existing evidence demonstrates that this pattern is the rule among species of highly evolved social insects. Consistent and predictable differences in behavior must reflect underlying or “covert” differences in physiology or anatomy. Particularly in the social bees, age polyethism refers to a set of characters rather than a single feature (Hardie and Lees, 1985; Robinson, 1992). Hive bees not only spend almost all of their time within the nest, but also have better developed brood food glands (hypopharyngeal glands) than their foraging sisters (Gast, 1967; Imboden and Liischer, 1975; Rutz et af., 1976; Wilde and Beetsma, 1982; Fluri el al., 1982). Hive bees also have significantly lower titers of the sesquiterpenoid juvenile hormone than field bees. As I discuss later in this review, it is juvenile hormone that holds the key to understanding the mechanisms of strong age polyethism in the bees and wasps. As the quotations that open this chapter indicate, this is not a particularly new idea. Yet, despite significant recent improvements in juvenile hormone measurements that addressed the “deficient base” of Fluri ef al. (1982), this literature still has, if not a deficient base, a hollow core. This is because of our utter lack of understanding of how juvenile hormone regulates behavior in adult insects. However, recent demonstrations of striking structural changes in the brains of adult honey bees correlated with age polyethism suggest that it may be possible to link juvenile hormone and neuroanatomical plasticity in the bee brain. This provides a focus for future investigations of juvenile hormone regulation of behavioral plasticity that can be conducted at the cellular and molecular as well as the ethological and ecological levels. OF AGEPOLYETHISM C. OTHERASPECTS
A comprehensive review of the details of age polyethism, its function in insect societies, and likely evolutionary scenarios are not provided here. There is a large, excellent, and highly accessible literature available on this topic (e.g., Wilson, 1971; Michener, 1974; Alford, 1975; Seeley, 1985; Winston, 1987; Ross and Matthews, 1991; Robinson, 1992). Honey bees have also been put to excellent use in investigations of insect visual and olfactory learning; these studies are not reviewed here, as they have by and large emphasized the capabilities of the forager and have ignored the
288
SUSAN E. FAHRBACH
fact of age polyethism (Hammer and Menzel, 1995). This review is instead from the perspective of a neurobiologist interested directly in the neuroendocrine mechanisms that control the behavior of individual animals. 11. AGEPOLYETHISM A N D JUVENILE HORMONE IN THE EUROPEAN HONEY BEE,APISMELLIFERA
A. AGEPOLYETHISM The division of labor prevailing in honey bee colonies has been remarkably well described. Both marked cohorts of bees and individually tagged bees have been studied. A single worker bee in an undisturbed colony will perform a stereotyped sequence of tasks as she ages. As outlined by Seeley (1982), the major honey bee age castes are (in order of occurrence within the life of a single worker bee) cell cleaning, brood and queen care, food storagehest maintenance, and foraging. Tasks within the hive are typically performed by workers 1-3 weeks of age, while bees older than 3 weeks forage outside the hive or engage in colony defense. Careful studies have revealed that the transitions between age castes prior to the switch to foraging (e.g., the change from care of larvae to food storage) involve age-correlated changes in the relative frequencies with which different categories of tasks are performed (Robinson, 1992). By contrast, the final shift to foraging is more emphatic. The end of adult behavioral development is a behavioral phase devoted to specific tasks associated with foraging (alternatively, a subset of bees this age may become soldiers, engaged mainly in colony defense rather than resource collection). The world of the forager is quite unlike that of the younger hive bees. Younger bees take only short defecation and orientation flights; foragers take long-distance flights covering hundreds of meters and lasting as long as an hour (Winston, 1987). It has also been widely recognized that foraging bees display capacities for rapid learning, sun compass navigation, longterm memory formation, and symbolic communication remarkable for an invertebrate (Frisch, 1967; Winston, 1987; Menzel, 1985, 1990). Only rare idiosyncratic individuals fall outside of this analysis by exhibiting extreme task specialization, such as dedication to water carrying or social grooming (Robinson, Underwood, and Henderson, 1984; Moore, Angel, Cheeseman, Robinson, and Fahrbach, 1995). Two influences have been identified that regulate age polyethism in the honey bee. The first is a set of yet-undescribed genetic factors (Calderone and Page, 1991; Robinson and Page, 1988, 1989; Robinson, Page, Strambi, and Strambi, 1989; Page and Robinson, 1991; Robinson, 1992). The second is change in the titers of juvenile hormone during adult behavioral development.
REGULATION OF AGE POLYETHISM IN BEES AND WASPS
289
INFLUENCES ON AGEPOLYETHISM B. GENETIC The genetic basis of the division of labor in honey bees is being actively investigated (Page, Waddington, Hunt, and Fondrk, 1995; Hunt, Page, Fondrk, and Dullum, 1995; Dreller, Fondrk, and Page, 1995; Robinson and Page, 1995). A honey bee colony consists of different “subfamilies” of workers because diploid queens mate with multiple haploid drones (reviewed by Page and Robinson, 1991). The sperm obtained at these matings (which typically occur during a small number of “nuptial” flights taken during the first 10 days of the queen’s life) are then stored in the spermatheca, a sperm storage organ adjacent to the reproductive tract (reviewed by Winston, 1987). The sperm from different drones are mixed in this storage organ and used by the queen throughout the remainder of her life to fertilize eggs to produce female offspring (in the Hymenoptera, haploid eggs become males, while diploid eggs become females). Females that share both a mother and a father are referred to as super sisters (see Page and Laidlaw, 1988, for a full discussion). Assuming random mating, super sisters will have on average 75% of their genes in common. Individuals within a colony whose fathers are unrelated drones are called half sisters. Members of different subfamilies have on average only 25% of their genes in common. It has been shown that, within individual colonies, honey bee subfamilies show genetic variation in the probability of performing different tasks (Robinson and Page, 1988, 1989). Subfamilies vary in their propensity to perform tasks commonly done by older workers, such as nectar foraging and pollen foraging, as well as the more specialized tasks of nest site scouting, guarding the hive entrance, and undertaking. There also appear to be significant genotypic differences in the rate of development of foraging behavior (Giray and Robinson, 1994). INFLUENCES ON AGEPOLYETHISM C . HORMONAL The second factor, developmental changes in titers of juvenile hormone, is the topic of this review. This factor is related not only to the normal progression of an individual worker through the age castes but also to the behavioral flexibility that is a normal part of the age polyethism displayed by honey bee colonies. It has long been appreciated, despite the inevitable march from within-hive tasks to field tasks, that the age of the transitions differs among colonies (reviewed by Michener, 1974; Seeley, 1985;Winston, 1987; Robinson, 1992). This plasticity in the behavior of adult workers permits colonies to make rapid adjustments to new situations. This plasticity ranges from accelerated behavioral development (premature foraging in response to a shortage of older workers) and temporary slowdowns in the
290
SUSAN E. FAHRBACH
progress of age polyethism (often seen in times of bad weather), to delayed behavioral development (older bees rearing brood) in response to a shortage of younger workers. In extreme cases, it is even possible for foragers to revert to rearing brood in response to an extreme shortage of younger bees. Although this plasticity is often revealed by experimental manipulations of colony population demographics, there is no doubt that it exists in nature and can be elicited by natural situations including weather catastrophes, severe predation, and reproductive swarming. It is now well established that changes in juvenile hormone titer accompany all of these transitions, including accelerated development (Robinson ef al., 1989) and behavioral reversions (Robinson, Page, Strambi, and Strambi, 1992; Farris, Robinson, and Fahrbach, submitted). D. JUVENILE HORMONE A N D AGEPOLYETHISM
Juvenile hormone (Fig. 1) is an insect sesquiterpenoid hormone produced by the paired corpora allata glands located in the head on either side of the esophagus (Snodgrass, 1984). The Hymenoptera produce the form of this hormone referred to as JHIII (Hagenguth and Rembold, 1978), and hemolymph titers of juvenile hormone in the honey bee are closely related to rate of synthesis, rather than to variations in rate of metabolism (see Fahrbach and Robinson, 1996, for a recent review of juvenile hormone in honey bees). Juvenile hormone was first identified on the basis of its central role in the regulation of metamorphosis and maturation of the ovaries in adult females, but is now known to influence many aspects of insect life histories including diapause, migration, coordination of reproduction with environmental cues, and caste determination (Wigglesworth, 1934, 1936; Riddiford, 1994). It may even act as a pheromone in some species of
A
OCH3
B
FIG.1. The structures of compounds with juvenoid activity in the honey bee. ( A ) JHIII, the native form of juvenile hormone in the honey bee. (B) Methoprene, a potent juvenile hormone analogue widely used in behavioral studies of honey bee behavioral development.
REGULATION OF AGE POLYETHISM IN BEES AND WASPS
291
parasitic wasp (Holler, Bargen, Vinson, and Witt, 1994). The production of juvenile hormone by the corpora allata is regulated both by circulating factors and by direct innervation by neurosecretory cells located in the brain. These peptidergic signals are referred to as allatotropins and allatostatins. They are produced by both the central nervous system and peripheral tissues such as the ovaries, and are currently best described in terms of the reproductive function of the cockroach, Diploptera punctatu (Woodhead, Stay, Seidel, Tobe, and Khan, 1989; Yu, Stay, Joshi, and Tobe, 1993; Stay ef al., 1994). The molecules regulating juvenile hormone production in the honey bee are likely to be similar, but are currently undescribed. Additionally, little is known about the cellular and molecular mechanisms of juvenile hormone actions. Progress in this field has been hampered by a lack of information on the juvenile hormone receptor as well as by the difficulties of measuring juvenile hormone titers in the hemolymph of individual insects (reviewed by Fahrbach and Robinson, 1996; Jones, 1995; Riddiford, 1994). For example, the heroic microsurgeries required to remove the glandular source of juvenile hormone, the corpora allata, have until very recently been compromised by an inability to assess the results of the surgery in terms of juvenile hormone titers. On the other hand, progress has been greatly assisted by academic and commercial interest in developing juvenile hormone analogues to serve as insect growth regulators for pest management (Williams, 1967).This has resulted in the development of potent compounds that mimic the effects of juvenile hormones (juvenoids) such as methoprene, which was developed by the Zoecon Corporation (the development of artificial insect growth regulators is described by Djerassi, 1992), and has proved extremely useful in research on polyethism among the social Hymenoptera. The idea that juvenile hormone is associated with the transition of hive bees to field bees can be traced in the literature back to the late 1960s. The idea reflected accumulating evidence for the association of juvenile hormone with polymorphisms and polyphenisms (see, for example, Liischer’s work on the differentiation of termite castes: Liischer, 1960, 1972). Also influential were observations that the size of the corpora allata changes as worker bees age, with greatest corpora allata volumes attained by foragers (see references given by Jaycox, Skrwronek, and Gwynn, 1974). Gast (1967) demonstrated that implantation of corpora allata from foraging bees or the injection of farnesyl methyl ester, a juvenile hormone mimic, induced a physiological correlate of this behavioral transition, the degeneration of the hypopharyngeal glands. It was subsequently shown that a similar inhibitory effect on the hypopharyngeal glands could be obtained by application of JHIII, the native hormone, as well as by application of the juvenile hormone analogue, triprene (Rutz et al., 1976). Additionally, allatectomy
292
SUSAN E. FAHRBACH
(surgical removal of the corpora allata) prevented hypopharyngeal gland degeneration in aging worker bees (Imboden & Liischer, 1975). Jaycox and colleagues (1974) performed the first published studies in which behavioral endpoints in worker honey bees were assessed subsequent to treatment with a juvenile hormone mimic, the so-called Law-Williams mixture, hydrochlorinated methyl farnesenate (Law, Yuan, and Williams, 1966). Worker bees were treated on the day of emergence with the mimic, establishing a paradigm that has been followed in every subsequent juvenile hormone treatment study. The results included inhibitory effects on hypopharyngeal gland development, but also demonstrated that bees treated with the mimic appeared to attain behavioral maturity earlier than controls housed in the same colony. This result has since been confirmed by other investigators, with the landmark study by Robinson (1987b) and subsequent reports by this investigator and his colleagues forming much of the basis for our current understanding of the relationship of juvenile hormone to age polyethism in honey bees. Three lines of evidence are essential for demonstrating naturally occurring hormonal regulation of a specific behavior. First, the behavior in question should be reliably induced by application of that hormone at physiological levels in animals at the appropriate stage of development and, in the case of social insects, housed in the appropriate social setting. It should be noted that a technical barrier to demonstrating some of the effects of juvenile hormone on age polyethism is the inability of workers of many species to survive as isolated individuals. Second, naturally occurring changes in the titer of the suspect hormone or in the sensitivity of the target tissues should be correlated with the changes in behavior. Until recently data of this type have been highly suggestive but not entirely persuasive because of a lack of juvenile hormone titer determinations on individual bees. As will be discussed later, this difficulty has now been forcefully addressed by the development of a sensitive radioimmunoassay for juvenile hormone in honey bee hemolymph. Finally, it should be possible to disrupt the performance of the behavior in question by removal of the source of the hormone, or by blocking its action at its receptors. The definitive experiments of this type awaited the development of the ability to assay juvenile hormone titers in the p1 hemolymph samples that can be obtained from individual bees. An additional form of powerful correlational data can be obtained from the use of colony manipulations to induce alterations in juvenile hormone titers and behavior. Such manipulations are entirely feasible when honey bees are reared in experimental apiaries using techniques developed for commercial apiculture.
REGULATION OF AGE POLYETHISM IN BEES AND WASPS
293
The next sections will summarize the current evidence for the relationship of juvenile hormone to age polyethism in the European honey bee. This will serve as the basis for a discussion of structural plasticity in the adult bee brain, and its relationhsip to behavioral maturation. 111. EFFECTS OF EXPERIMENTAL TREATMENT WITH JUVENILE HORMONE, A N D ANALOGUE ON BEHAVIORAL MATURATION IN THE HONEY BEE MIMICS, A. EFFECTS OF EXPERIMENTAL TREATMENTS
Table I summarizes the literature on the behavioral effects of treatment of worker honey bees with juvenile hormone, juvenile hormone mimics, and juvenile hormone analogue such as methoprene. Several points are clear. First, results are consistent across experiments. Treatment of newly emerged adult worker bees (0-24 hr postemergence) with an active juvenoid (at a time when juvenile hormone levels are naturally quite low) results in accelerated behavioral maturation. Dose-dependent effects can be seen regardless of whether marked treatment groups or marked individual bees are studied, regardless of whether the experimental bees are housed in large or small colonies (or in glass-sided observation hives), and regardless of the specific behavior sampling strategy used. Although the native hormone JHIII is most active when injected (dissolved in oil) into the hemocoel, the analogue methoprene is extremely active when applied topically (dissolved in acetone), and can even be fed to bees dissolved in sugar syrup and still be effective. The report of Robinson is noteworthy in that it is based on the study of individually number-tagged bees housed in a large observation hive (Robinson, 1987b). This permitted the effects of topical application of methoprene to be examined on within-hive behavior as well as on activity monitored at the hive entrance. In this study it was established clearly that not only is the mean age at first foraging reduced by treatment with methoprene, but also that methoprene treatment inhibits brood and queen care and reduces the food-storing phase. The impression is not so much one of an abnormal trajectory of development as it is one of a “compression of normal development” (Robinson, 1987b). Another important finding was that worker bees treated with methoprene on the first day of adult life displayed precocious but otherwise apparently normal foraging behavior (Robinson, 1985). B. QUESTIONS RAISED BY TREATMENT STUDIES These accumulated data raise the following questions. First, is methoprene a suitable replacement for juvenile hormone in behavioral studies
TABLE I
EFFECTS OF TREATMENT WITH JUVENILE HORMONE, ANALOGUES. AND MIMICS ON Subjects"
Treatment
Method
THE
BEHAVIOR OF ADULT WORKER HONEYBEES,MIS
Observed behavior
Results of treatment leave brood nest earlier; precocious guarding; precocious pollen foraging; increase in total number of Bights no response to topical JHI; weak enhancement of flight activity with topical Law- Williams Mixture flight away from the hive begins earlier; more bees fly in treated groups; increase in total number of flights highest dose lethal; 250-pg dose results in early shift to food storage, earlier orientation flights, earlier foraging
0-24 hr adults, small observation hives
Law-Williams Mixtureb, injected in oil 10-200 p g
distribution in hive, activity at entrance
0-24 hr adults, single frame observation hives
JHI, 0.1-10 pg Law-Williams Mixture, 100 pg
topical application in 95 :5 acetone : olive oil
activity at entrance
0-24 hr adults, single frame observation hives
JHI, 10 p g Law-Williams Mixture, 100 pg
injected in oil
activity at entrance
0-24 hr adults, typical colony with modified entrance
methoprene, 2.52500 Pg
topical application in acetone
number-tagged bees; daily 1 hr entrance observations
N
P
MELLIFERA
Reference Jaycox
et
al., 1974
Jaycox, 1976
Jaycox, 1976
Robinson, 1985
0-24 hr adults, typical colony
methoprene. 50 or 200 1Lg
topical application in acetone
0-24 hr adults. typical colony with modified entrance
hydroprene, 20 or 200 pg methoprene. 200 p g or 1 mJ.3
orally, dissolved in sugar syrup
0-24 hr adults, large observation hives
methoprene, 25-250 pg
topical application in acetone
response to alarm pheromone, tests at 1-23 days posttreatment entrance observations at 11. 12. 13, 15. 22, 26 days 8 hr observatiordday. frequency of 30
behaviors; duration of task phase; daily entrance observations
g vI
0-24 hr adults, typical colony
JHIII, 1 pg methoprene. 0.1-200 p g hydroprene, 2 or 20 pg
topical application in acetone O R injected in oil
activity at entrance; latency to forage
0-24 hr adults, singlecohort colony
methoprene, 200 p g
topical application in acetone
number-tagged bees; daily 1 hr entrance observations
a
Age of worker honey bee at time of treatment. Hydrochlorinated methyl farnesenate (Law er al., 1966).
dose-dependent strengthening of all responses
Robinson, 1987a
dose-dependent increase in number of flights during week 2 of study dose-dependent reduction in mean age at foraging; dosedependent inhibition of brood and queen care; reduction in food-storing phase precocious guarding, pollen foraging up to 5 days earlier than controls; weaker effects with topical applications 83% treated bees forage by 10 days of age, compared with 16% controls
Robinson and Ratnieks. 1987
Robinson, 1987b
Sasagawa, Sasaki, and Okada, 1989
Robinson er al., 1989
296
SUSAN E. FAHRBACH
of honey bees? Second, are the effects of early treatment with juvenoid agents a result of effects exerted at the time of application (the first 24 hr of life), or do they reflect instead the persistence of unmetabolized hormone analogue, which then acts at a later time? Third, what is (are) the target(s) of juvenile hormone action? In response to the first question, methoprene has been widely used in studies of insect endocrinology, and is a potent juvenile hormone analogue, with activity in all insect species in which it has been tested (Staal, 1975). It has the advantage of being more active than JHIII when topically applied, which is an alternative to injection that minimizes handling of treated bees. There are no discrepancies in the behavioral actions of juvenile hormone and methoprene (see Table I). A comparison of physiological and anatomical effects of juvenile hormone and methoprene treatment of adult worker honey bees (such as effects on hypopharyngeal gland development) also reveals few discrepancies. It has been reported that wax secretion, which is higher in hive bees than in foragers, may not be under inhibitory control by JHIII, but may be sensitive to methoprene (Muller and Hepburn, 1994). This report is difficult to interpret in light of the lack of information concerning juvenile hormone .titers in manipulated bees and the lack of any other behavioral or physiological measures validating the effectiveness of the treatment. Discrepancies in this study between the effects of JHIII and methoprene may reflect the greater potency of methoprene at the doses selected to accelerate behavioral development, which led in turn to the aforementioned “compression” of the wax secreting period. The doses of JHIII may have been less effective in producing accelerated development. Therefore, at the present time we can be confident that data based on methoprene treatment are useful and reliable. Second, because studies have not systematically varied the time of juvenile hormone treatment, it is not possible to determine the time of action on the basis of these experiments alone. Methoprene is likely to persist longer in the body than JHIII, and may be absorbed through the cuticle at a different rate than JHIII. Whether or not this treatment is the honey bee equivalent of implantation of a hormone-filled Silastic capsule or osmotic minipump is not known. This issue is taken up again in the following discussion of the naturally occurring fluctuations in juvenile hormone titers that occur during adult life in the honey bee. What are the targets of juvenile hormone action with regard to behavior? Sites within the central nervous system are obvious candidates. Only one study has investigated potential peripheral sites of action. Application of methoprene on the first day of adult life was shown to induce a precocious behavioral response (wing flickering) to alarm pheromone under controlled conditions in the laboratory (Robinson, 1987a). This same dose had no
REGULATION OF AGE POLYETHISM IN BEES AND WASPS
297
effect on the electroantennogram response to these compounds, suggesting that the hormone-mediated changes in behavior occurred centrally rather than at the level of the olfactory receptors.
IV. CORRELATIONAL DATAINDICATING JUVENILE HORMONE TITERS INCREASE DURING BEHAVIORAL MATURATION A N D AREAT THEIR HIGHEST I N FORAGING BEES DEVELOPMENTAL CHANGES IN JUVENILE HORMONE PRODUCTION As discussed in the preceding section, premature exposure to high levels of juvenile hormone leads to accelerated behavioral development, with an earlier onset of foraging. These results gain significance when taken together with measurements showing that older, foraging bees typically have higher levels of juvenile hormone than younger hive bees. As Table I1 suggests, it has been appreciated for some time that foraging bees are characterized by high levels of juvenile hormone relative to hive bees. A major problem, however, has been to develop methods sensitive enough to measure titers in individual bees. The most widely used methods of juvenile hormone determination have been bioassay (the Galleria wax-wound test, in which a wound made in the pupal cuticle of a moth is sealed with wax containing juvenile hormone: see description in Nijhout, 1994), gas chromatography combined with mass spectrometry (Mauchamp, Lafont, and Krien, 1981), and radioimmunoassay (Granger and Goodman, 1988). The first radioimmunoassay used for juvenile hormone determinations in the honey bee required pooled hemolymph samples in order to achieve detectable amounts of hormone in the sample (Strambi, Strambi, deReggi, Hira, and Delaage, 1981). Researchers in this field, however, have dealt systematically with the problems of quantification of this small, lipophilic molecule, and recently new, more sensitive assays that can be used with simple but effective extraction procedures have been developed (Goodman et al., 1995). The development of a new radioimmunoassay for juvenile hormone that permits determinations on hemolymph samples from individual bees has added clarity to this field, and will be widely used in future behavioral studies (Hunnicutt, Toong, and Borst, 1989; Huang, Robinson, and Borst, 1994). This methodological development also permits the rigorous assessment of the effects of allatectomy. An alternative to radioimmunoassay that permits insight into the endocrine function of individual bees is the biosynthesis assay. In this assay, cultured corpora allata glands are provided with a radiolabeled precursor and the amount of juvenile hormone produced is determined by scintillation
TABLE I1 MEASUREMENTS OF JUVENILE HORMONE TITERS IN ADULT WORKER HONEYBEES.APlS Subjects h)
8
Method
Pooled or individual
MELLlFERA
Values"
Reference
0-24 hr adults (i) 12 day hive bees (ii) 24 day field bees (iii)
bioassay using Galleria wax test
pooled: 40 bees/sample
328 GUh/ml (i) 1807 GU/ml (ii) 3820 Gdml (iii)
Rutz. er al., 1976
0-24 hr adults (i) 3-13 day adults (ii)
gas chromatography with electron capture detection
pooled: 1-4 g bees
0.7 ng/g (i) 10 nglg (ii)
Hagenguth and Rembold, 1978
0-24 hr adults (i) 10-15 day adults (ii) field bees (iii)
bioassay using Galleria wax test
pooled: 40 bees/sample
very low (i) 1000-1500 GUlml (ii) >2000 GU/ml (iii)
Fluri ef al., 1982
hive bees guard bees foragers
HPLC-determination
pooled: 10 beeslsample: and individual
low until day 14; H. Sasagawa, 1988. intermediate until day 28: unpublished observations 2-3 ndp1 in older bees
swarm nurses (i) foragers (ii)
RIA (Strambi et af., 1984)
pooled: 4-16 beedgroup
t 5 pmo1/100 p1 (i) 20-65 pmo1/100 pI (ii)
workers from a single-cohort RIA (Strambi er al.. 1984) colony of known age and behavioral status
pooled: 4-16 bees/group
2-10pmo1/100pl, nurse bees Robinson, et al., 1989 20-75 pmo1/100 @I, foragers
Robinson, ef al., 1989
nurse bees (i) foragers of known age (ii)
radiochemical analysis of JHIII biosynthesis by cultured CC-CA complex
individual
nurse bees ( i ) foragers (ii)
RIA (Strambi er al., 1984)
pooled 5-9 bees/age group <10 pmoUl00 pl (i) 40-45 pmoUlOO pl (ii)
Huang et al., 1991
workers of known age: wax producers, food storers, guards, undertakers. soldiers, foragers
radiochemical analysis of JHIII biosynthesis by cultured CC-CA complex
individual
1-2 pmol/hr: queen attendants, nurses, wax producers, food storers 3-5 pmolhr: foragers. guards, undertakers
Huang, Robinson, and Borst. 1994
0-24 hr adults (i) foragers (ii)
radiochemical analysis of JHIII biosynthesis by cultured CC-CA complex
individual
approx. 2 pmoYhr (i) approx. 6 pmol/hr (ii)
Huang and Robinson, 1995
0-24 hr adults (i) foragers (ii)
RIA (Hunnicutt, Toong, and individual Borst, 1989)
(100 nglml (i)
Huang and Robinson, 1995
nurse bees (i)
RIA (Hunnicutt era[., 1989) individual
foragers (ii) Expressed as concentration per hemolymph volume where possible. Galleria units.
1-2 pmol/hr (i) 4-6 pmol/hr (ii)
Huang et al., 1991
400 ng/ml (ii) 200 ng/ml (i) 600-800 ng/ml (ii)
Farris, Robinson, and Fahrbach, submitted
300
SUSAN E. FAHRBACH
counting of an extract of the medium. A biosynthesis assay developed for the honey bee (based on the successful Diploptera punctata assay developed by Tobe and co-workers: Tobe and Pratt, 1974; Pratt and Tobe, 1974) produced data in good agreement with subsequent radioimmunoassaydetermined titers, and confirmed that rate of synthesis is the primary determinant of juvenile hormone titer in the honey bee (Huang etaf.,1991,1994). Because it is labor intensive relative to radioimmunoassay, the usefulness of the biosynthesis assay has been superseded for large-scale behavioral studies, although it will be vital for the exploration of factors regulating juvenile hormone synthesis. As in the case of the effects of treatment with juvenile hormone and its analogue, there is striking agreement as to the overall profile of juvenile hormone titers during behavioral development. The correlations between low levels of juvenile hormone and nursing and high levels of juvenile hormone and foraging are very, very strong. One surprising discovery has been the finding that a few tasks not directly related to foraging are typically accompanied by foragerlike high levels of juvenile hormone. Examples of such activities are guarding the hive entrance and undertaking (Huang et al., 1994). An area for future investigation is the filling in of the day-byday details of the profile. It is possible that transient peaks in juvenile hormone titers are hidden in the current profiles, particularly early in adult life. This information will be vital to our interpretation of the timing of juvenile hormone action during development.
V. COLONY MANIPULATIONS THATINDUCE ALTERED JUVENILE HORMONE TITERS A N D ALTERED BEHAVIOR; SEASONAL CHANGES I N JUVENILE TITERS A N D BEHAVIOR HORMONE
A. SINGLE-COHORT COLONIES The predictable association of high levels of juvenile hormone with foraging in honey bees is no less strong when behavior is manipulated by alterations in colony population demography. Although the oldest workers are typically a colony’s foragers, age alone determines neither behavior nor endocrine status. An experimental single-cohort colony of honey bees can be founded with a queen and 1000-2000 day-old workers. In these experimental colonies, a subset of workers will begin to forage between 7 and 10 days of age, rather than at the more typical 3 weeks. It has been shown that precocious foragers in a single-cohort colony have levels of juvenile hormone equal to those of normal-aged foragers (Robinson et al., 1989). While these precocious foragers have high levels of juvenile hormone, the
REGULATION OF AGE POLYETHISM IN BEES AND WASPS
301
same-aged nurses from the same colony have the lower levels characteristic of hive bees. In the same way, delayed behavioral development (“overage” nursing) is associated with maintenance of relatively low juvenile hormone titers (Robinson et al., 1989). Behavioral reversion from foraging to nursing is accompanied by a drop in juvenile hormone titers (Robinson et af., 1992; Farris et al., submitted; Huang and Robinson, 1996).
B. SWARMING Natural scenarios can also lead to changes in behavior and changes in juvenile hormone production that are independent of age. For example, when a new colony is founded by a reproductive swarm, bees of the same age may perform age-inappropriate tasks. These bees have juvenile hormone titers that are predictable on the basis of their observed behavior, not their age (Robinson et al., 1989). C. SEASONAL CHANGES It has also been noted that, in temperate zones, those workers that overwinter have lower levels of juvenile hormone than do summer foragers (Fluri, Wille, Gerig, and Luscher, 1977; Fluri er al., 1982). Recently, both juvenile hormone titers and rates of biosynthesis were shown to be decreased in workers at times of the year when colony activity is reduced (Huang and Robinson, 1995). These responses strengthen the possibility that changes in juvenile hormone titer may mediate plasticity in age polyethism required by changing environmental conditions.
VI. BRAIN CHANGES CORRELATED WITH BEHAVIORAL MATURATION HONEY BEE:A PROPOSED MECHANISM FOR JUVENILE HORMONE ACTIONON AGEPOLYETHISM
I N THE
A. PLASTICITY OF THE MUSHROOM BODIES
The capacity for behavioral change in adult worker bees is correlated with anatomical plasticity in the honey bee brain. The corpora pedunculata, or mushroom bodies, of the protocerebrum undergo significant volume changes during adult behavioral development. In this structure, a decrease in the volume occupied by the somata of the intrinsic neuronal population (the Kenyon cells) is accompanied by an increase in the volume of the associated neuropil (Withers, Fahrbach, and Robinson, 1993). One-dayold workers have a neuropil-to-neuronal somata ratio for the mushroom
302
SUSAN E. FAHRBACH
bodies of approximately 1: 1; in foragers, this ratio is approximately 2: 1. Equivalent volume changes are seen in normal-aged and precocious foragers (Withers et al., 1993). Volume changes are detected by application of the Cavalieri method to camera lucida drawings of the mushroom bodies. The drawings are typically based on 10-pm-thick Paraplast sections, which are stained with Luxol Fast Blue and cresyl violet. (For a detailed description of the methods, see Fahrbach, Giray, and Robinson, 1995.) The changes observed within the mushroom bodies are highly selective, as the overall volume of the brain, and the volumes of most other brain regions, remain stable during the period of behavioral maturation (Withers et ul., 1993). BODIES A N D LEARNING B. THEMUSHROOM The mushroom bodies are the insect brain center primarily responsible for the integration of olfactory information; in the Hymenoptera there is also a substantial input from visual centers (Mobbs, 1982, 1985). In all insects studied, the mushroom bodies have been shown to be critical for learning and memory (Erber, Masuhr, and Menzel, 1980; Menzel, 1990; Davis, 1993; Strausfeld, Buschbeck, and Gomex, 1995; Davis and Han, 1996). In addition to a well-established role in olfactory learning, the mushroom bodies may also function in spatial learning (Mizunami, Weibrecht, and Strausfeld, 1993), which is of signal importance to insects maintaining a fixed nest site. The anatomy of the mushroom bodies of the insect brain has been extensively described in honey bees and in several other insects, although with rare exceptions all studies have been conducted outside of a developmental context. Mutant Drosophilu melanogaster deficient in olfactory learning have both anatomical and molecular abnormalities of the mushroom bodies (reviewed by Davis, 1993), a finding that has heightened interest in this insect brain structure. Neuroanatomical analyses of enhancer trap lines of Drosophila have recently demonstrated the heterogeneity of the population of Kenyon cells, the intrinsic neurons of the mushroom bodies, in terms of patterns of gene expression (Yang, Armstrong, Vilinsky, Strausfeld, and Kaiser, 1995).
C. SUMMARY OF THE STRUCTURE OF THE MUSHROOM BODIES The mushroom bodies are formed by the Kenyon cells and their processes and afferents (Fig. 2). As is typical in arthropods, the neuronal somata of the Kenyon cells are completely segregated from all regions of synaptic contact. The neuropil of the hymenopteran mushroom bodies can be divided into distinct regions: the medial and lateral calyces, the peduncle,
REGULATlON OF AGE POLYETHISM IN BEES AND WASPS
303
'afferent axon
FIG.2. A highly simplified diagram of a single calyx of the mushroom bodies of the honey bee. The arborizations of the intrinsic neurons (Kenyon cells) form the walls of the cup that contains them. while the axons travel via the peduncle to form the alpha and beta lobes (not shown). The afferents to the Kenyon cells are segregated according to modality. The lip receives olfactory input from the antennal lobes of the brain: the collar receives visual input from the optic lobes; and the basal ring afferents arise in both the antennal and optic lobes (Mobbs. 1982. 1985). There are separate medial and lateral calyces in each hemisphere of the honey bee brain. The Kenyon cells are small and tightly packed: the total number of Kenyon cells in the brain of a worker honey bee is approximately 340,000 (Witthlift, 1967).
and the alpha and beta lobes. The calycal neuropils are the input regions of the mushroom bodies, receiving both olfactory and visual information from the antennal lobe and optic lobe, respectively. The dendrites of the Kenyon cells arborize extensively in the calyces. Kenyon cell axons form the peduncle. The Kenyon cell axons divide at the base of the peduncle so that each cell sends one branch into the alpha lobe and one branch into the beta lobe. The extrinsic neurons of the mushroom bodies are located in the lateral protocerebrum. A complex set of projections forms circuits within the mushroom bodies in addition to these connections with extrinsic neurons. In the honey bee, there are between 100,000 and 200,000 densely packed Kenyon cells in each brain hemisphere. D.
FURTHER
STUDIES OF NEUROANATOMICAL PLASTICITY
IN
WORKERS
Subsequent studies have confirmed the first reports that the mushroom bodies are reorganized in foragers, regardless of age (Durst, Eichmiiller, and Menzel, 1994), and have shown that the largest changes in neuropil volume may occur within the subregion of neuropil that receives visual input, the collar. It has been shown that these changes also occur in bees that have not had foraging experience (Withers, Fahrbach, and Robinson, 1995). In this experiment, a single-cohort colony was established. A subset
304
SUSAN E. FAHRBACH
of bees was treated with methoprene on the day of emergence, but were not allowed to fly because a plastic disk glued onto their dorsal thorax made the bee too large to squeeze through a modified hive exit. The structure of the mushroom bodies of these bees did not differ from that of methoprene-treated bees allowed to forage. More recently, the brain changes associated with foraging have been shown to be stable, being maintained in foragers that had reverted to nursing, even in the face of reduced levels of juvenile hormone (Farris et al., submitted). OF NEUROANATOMICAL PLASTICITY IN E. STUDIES
QUEENS
It appeared possible to test the association of foraging, changes in the structure of the brain, and juvenile hormone by studying brain development in honey bee queens. Mature queens (more than 2 months of age) have very low levels of juvenile hormone (Fluri, Sabatini, Vecchi, and Wille, 1981; Robinson, Strambi, Strambi, and Feldlaufer, 1992), and they do not take flights away from the hive except on the rare occasion of a reproductive swarm. Queen bees, however, were shown to undergo reorganization of the mushroom bodies, but the changes were seen earlier than in worker bees (during the first week of adult life). This result appeared to contradict the association of high levels of juvenile hormone with altered volumes in the mushroom bodies until radioimmunoassay analysis of juvenile hormone titers in queens (including the queens taken for brain analysis) revealed the surprising finding that 1-day-old queens, in dramatic contrast to 1-dayold workers, have high levels of juvenile hormone that are equal to those of foragers (Fahrbach et al., 1995).
F. ALLATECTOMY STUDIES There are currently no data inconsistent with the hypothesis that exposure to raised levels of juvenile hormone is responsible for the structural changes observed in the bee brain during the transition from hive bee to forager. Several key questions, however, remain unanswered. What is the effect of allatectomy on behavior and brain development? When must the brain be exposed to juvenile hormone before these changes occur? And what are the cellular mechanisms that drive these changes? Allatectomy verified by subsequent radioimmunoassay has recently been performed in collaboration with G. E. Robinson (J. P. Sullivan, G. E. Robinson, and S. E. Fahrbach, unpublished results). Workers allatectomized on the first day of adult life were returned to a large observation colony for several weeks until unmanipulated and sham-operated controls had completed normal development. Allatectomy produced a striking and
REGULATION OF AGE POLYETHISM IN BEES AND WASPS
305
selective change in worker behavior. Allatectomized bees appeared normal and were able to take orientation flights near the hive. They did not, however, make the transition to foraging that is a normal part of the age polyethism pattern in a honey bee colony. The brains of bees undergoing truncated behavioral development in the absence of juvenile hormone are currently being subjected to volume estimation. This study explicitly confirms the interpretation of earlier studies as indicating an activational effect for juvenile hormone on foraging behavior. We predict that these bees will not exhibit significant reorganization of the mushroom bodies. The critical time when exposure to juvenile hormone can have behavioral impact remains undetermined, in part because of the methodological uncertainties detailed earlier. It has recently been shown, however, that some of the volume changes seen within the honey bee mushroom bodies are detectable before workers begin foraging, suggesting that the differences seen between foragers and younger bees represent the culmination of a process begun earlier rather than an abrupt transition (Fahrbach, Farris, and Robinson, 1995). Such a temporal dissociation suggests that the high levels of juvenile hormone seen later in foragers could not have the sole or primary organizing effect on the mushroom bodies. One possibility is that the volume changes seen in the mushroom bodies are the result of both hormone-dependent and hormone-independent processes, with the hormone-dependent processes taking over after the onset of foraging. Another possibility is suggested by the study of juvenile hormone profiles in queens. Might there be an earlier period of hormone exposure critical for activating structural changes in the brain? This question will in part be answered by allatectomy studies. It can also be addressed by studies of juvenile hormone titers in workers with a finer temporal resolution than those previously published. Such studies are currently in progress in the Robinson laboratory, motivated in part by the detection of early changes in the structure of the mushroom bodies. One previous report suggests the possibility of a small peak around the second day of adult life. Radioimmunoassay determinations have suggested that this peak may be a consistent feature of worker endocrine development (0.Jassim, Z.-Y. Huang, and G. E. Robinson, unpublished results).
G. CELLULAR MECHANISMS The cellular mechanisms that may account for these changes are currently under active investigation, and have been recently reviewed (Fahrbach and Robinson, in press). They do not result from hormone-induced neurogenesis in the honey bee (Fahrbach, Strande, and Robinson, 1995), although there is evidence that juvenile hormone may regulate the birth of adult
306
SUSAN E. FAHRBACH
specific neurons in the brain of other insects such as crickets (Cayre, Strambi, and Strambi, 1994). The anatomical studies described here suggest, however, that it will be necessary and highly profitable to focus studies of juvenile hormone action on the brain on the mushroom bodies. It is in this brain region alone that the structural plasticity correlated with behavioral maturation and juvenile hormone titers converges with the neural circuitry essential for insect learning and memory.
VII. COMPARATIVE STUDIES A. EUSOCIAL BEESA N D WASPS Within the order Hymenoptera, there are more than 20,000 identified species in the superfamily, Apoidea (Michener, 1974). Although primitive eusocial or quasi-social behavior may be found in some members of the families Xylocopinae (stem- or wood-boring bees) and Halictinae (sweat bees), advanced social behavior is restricted to the three tribes of the family Apidae: the Bombini (bumble bees), the Meliponini (stingless honey bees), and the Apini (true honey bees of the genus Apis). Since the seventeenth century, there have been excellent descriptions available of the temporal polyethism characteristic of honey bee colonies (Butler, 1609, cited in Wilson, 1971); the study of the division of labor in other advanced insect societies is more recent. It has been noted that the social bees and wasps (discussed later) differ from the ants and termites in their sole reliance on age polyethism to provide an efficient division of labor, rather than on differentiation of morphologically discrete worker castes (Wilson, 1971). This means that there should be identifiable other species of bees and wasps that would be predicted to show similar links between juvenile hormone, behavior, and brain structure. It also means that there are many closely related species available in which these relationships ought not to hold, given significant differences in worker life histories. Five members of the genus Apis, of which the European honey bee Apis mellifera is by far the best studied, have been defined, although the specieslevel taxonomy of Apis is currently subject to revision (Michener, 1974; Winston, 1987; Sheppard and Berlocher, 1984). Other Apis spp. are found in Asia: Apis florea, Apis dorsata, Apis cerana, and Apis laboriosa. Little is known of the endocrinology and detailed neuroanatomy of the species other than A. mellifera. Like A . mellifera, A. cerana is a medium-sized bee that builds comb within enclosed cavities, while A . florea, A . dorsata, and A . laboriosa all build single combs in the open (Michener, 1974; Winston, 1987). A. florea is a dwarf honey bee that builds its nest amid dense vegeta-
REGULATION OF AGE POLYETHISM IN BEES AND WASPS
307
tion, while A . dorsata and A . laboriosa are large and aggressive. These species provide testing grounds for hypotheses generated by the study of A . mellifera. Strong age polyethism is well documented in the Meliponini, the stingless bees, and is similar to that seen in Apis mellifera, although the life-span is longer and foraging typically begins at a later age (Kerr and Netos, 1956; Hebling, Kerr, and Kerr, 1964, cited in Wilson, 1971). A recent study based on observations of paint-dot-marked individuals of the species Plebeiu rernotu, a stingless bee that nests in cavities in tree trunks in South America, documented that building activities occurred between 6 and 30 days of age, activities on the royal chambers were performed between 23 and 85 days of age, and foraging was performed by bees thirty days of age or older (Benthem, Imperatriz-Fonseca, and Velthuis, 1995). Little is known of the endocrinology and neuroanatomy of the several hundred species of social stingless bees. The stingless bees constitute a group in which patterns of juvenile hormone production and plasticity of the mushroom bodies are predicted to parallel relationships described for honey bees. The bumble bees (genus Bombus) exhibit a weak version of typical age polyethism, with worker size a confounding factor, as smaller workers are less likely to forage outside the nest than the largest workers (Wilson, 1971). Their brains and patterns of juvenile hormone production in the adults will also be of great interest, given the apparent variability of their behavior relative to Apis spp. and the stingless bees. There seems to be little published information available on the neuroanatomy of bumble bees, with only general descriptions of gross brain anatomy in the literature (Alford, 1975). The eusocial wasps are members of a different superfamily within the Hymenoptera, superfamily Vespoidea (Wilson, 1971). Many degrees of social behavior can be found within the family Vespidae, which includes the subfamilies Stenogastrinae (hover wasps), Polistinae (paper wasps), and Vespinae (hornets and yellowjackets). Examples of eusocial behavior are also found within the family Sphecidae (superfamily Sphecidea), although most sphecid wasps are solitary. Interestingly, the Sphecidae and the Vespidae are not phylogenetically closely related, with the sphecid wasps more closely allied to the bees than to all other social wasps. The phylogenetic relationships of the Vespidae have been discussed in light of the origins of social behavior by Carpenter (1991), in a notable volume devoted to the social biology of wasps (Ross and Matthews, 1991). A comprehensive review of age polyethism among workers in social wasps has been provided by Jeanne (1991). For example, in the paper wasp Polistes dominulus, there is a juvenile phase of 1-5 days, during which the young female works exclusively on the nest. This is followed by a “construction
308
SUSAN E. FAHRBACH
phase,” which lasts from 6 to 20 days of age. During this phase, nest construction activity peaks and foraging begins. At approximately 20 days of age a final phase begins, during which construction activity declines and foraging predominates. (Another Polistes species, Polistes fadwigae, provides the only reported example of reverse age polyethism, in which older workers have a greater tendency to participate in nest work: Yoshikawa, 1963.) Examples of age polyethism are also clear in the advanced swarm-founding wasps. In the vespid Polybia occidentalis, there are at least two clear-cut age castes: nest workers, which engage in building, nest maintenance, defense of the colony, and brood care, and foragers, which collect pulp, prey, water, or nectar. Although the ages at which individual workers switched to foraging ranged from 6 to 40 days, the typical individual made the transition relatively abruptly (Jeanne, 1991). OF JUVENILE HORMONE TREATMENT IN SPECIES OTHER THAN B. EFFECTS
A P I S MELLIFERA
Juvenile hormone, or more typically, methoprene, treatment has been tested on several representatives of non-Apis groups. Juvenile hormone and its analogues were found to have no effect on the division of labor when applied to primitively eusocial bumble bees: Bombus terrestris (Roseler and Roseler, 1978), Bombus impatiens (Cameron and Robinson, 1990), and Bombus bimaculatus (Cameron and Robinson, 1990). Juvenile hormone and/or methoprene also had no effect on age polyethism when applied to several paper wasps: Polistes metricus (Bohm, 1972), Polistes annularis (Barth, Lester, Sroka, Kessler, and Hearn, 1975), and Polistes gallicus (Roseler, 1985). Unlike the paper wasps just described, in which reproductive females found new colonies independently, the advanced social wasps are characterized by swarm founding of new colonies, as in Apis mellifera. As described above, the advanced social wasps also have well-developed temporal castes. In Polybia occidentalis, methoprene accelerates age polyethism with effects highly reminiscent of methoprene effects on honey bees (O’Donnell and Jeanne, 1993). It is obviously a question of great interest regarding the juvenile hormone profiles and configuration of the mushroom bodies of these wasps. A comparative neuroendocrinology of the Hymenoptera does not currently exist, but its outlines can be glimpsed in present studies of the honey bee. C. SUMMARY To summarize, the social bees and wasps provide numerous examples of age polyethism, ranging from strong expression in honey bees, some
REGULATION OF AGE POLYETHISM IN BEES AND WASPS
309
stingless bees, and the advanced social wasps to more variable expression in bumble bees and primitively eusocial wasps. It is a testable hypothesis that all cases of strong age polyethism will be associated with high levels of juvenile hormone in individuals working outside the hive and in associated changes in the brain space devoted to the mushroom bodies. Conversely, such changes in the structure of the adult nervous system are unlikely to be seen where age polyethism is weak and the youngest workers can both work in the nest and forage.
VIII. CONCLUSIONS AND SIGNIFICANCE ROLESFOR A SINGLE A. MULTIPLE HORMONE Pener points out in his consideration of the role of hormones in the control of insect flight and migration that preexisting chemical signals were very likely “captured” during arthropod evolution for new functions (Pener, 1985). The same endocrine factor may therefore have different roles in different species with different adaptive strategies. He gives the relationship of juvenile hormone and flight muscle as an illustrative example: in most insects juvenile hormone induces degeneration of the flight muscles, but in the Colorado potato beetle juvenile hormone induces flight muscle growth or regeneration (Kort, 1969). Developmental hormones may similarly have been “captured” for use by the nervous system (Truman and Riddiford, 1974). In its regulation of age polyethism in social insects, juvenile hormone appears to have been “captured” by advanced social insects more than once, with a possible independent evolution of this strategy in Apidae and Vespidae (O’Donnell and Jeanne, 1993). Juvenile hormone is predicted to share the same role in these advanced social insects: a mediator of joint brain and behavior development. It is likely to be limited to such a role in those insects in which age polyethism is so strong that there is a transition of consequence between the hive and the field. The cognitive demands of navigation and learning to handle floral resources are assumed to require adult development of the mushroom bodies. This development is coordinated with behavior by juvenile hormone. Bees without these endocrine-mediated changes in the structure of the mushroom bodies are predicted to perform less adeptly on tasks that reflect the cognitive demands of orientation to the nest site and foraging. This hypothesis can also be readily tested. Robinson initially hypothesized that changes in responses to environmental cues were a result of juvenile hormone action on the nervous system (Robinson, 1987a,b). More recent neuroanatomical investigations suggest that the effects of juvenile hormone may be more accurately described as
310
SUSAN E. FAHRBACH
leading to more brain space dedicated to complex information processing. Such changes are intriguingly analogous to changes in hippocampal volume detected in food-storing birds such as the marsh tit (Krebs, Sherry, Healey, Perry, and Vaccarino, 1989; Krebs, 1990), or in the song centers of the avian forebrain in many songbirds (Nottebohm, 1981; Konishi, 1994). By study of the relationships of mushroom body structure, juvenile hormone, and the ontogeny of foraging in the honey bee, we are able to study the control of behavior in individual bees. By future studies of the incidence of juvenile hormone regulation of age polyethism in the social insects, we can gain insight into the evolution of mechanisms for neural and behavioral plasticity. B. SUMMARY The regulation of the division of labor by juvenile hormone has been well established for honey bees. Recent studies suggest that juvenile hormone may control the volume of specific regions of the honey bee brain. Comparative neuroanatomical and neuroendocrinological studies of the eusocial and primitively social Hymenoptera are needed to reveal the relationships between juvenile hormone, brain structure, and behavioral maturation. Acknowledgments G . E. Robinson introduced me to the topic of juvenile hormone and age polyethism in the social Hymenoptera, and I thank him for many fruitful discussions of his intellectual and experimental contributions to this field. I also thank my graduate students past and present (G. S. Withers, S. M. Farris, and J. P. Sullivan) for their vigorous investigations of the juvenile hormone-brain-behavior relationship in honey bees. Recent research described here was supported by the National Science Foundation.
References Alford, D. V. (1975). “Bumblebees.” Davis-Poynter Ltd., London. Barth, R. H., Lester, L. J., Sroka, P., Kessler, T., and Hearn, R. (1975). Juvenile hormone promotes dominance behavior and ovarian development in social wasps (Polistes anniilaris). Experientia 31, 691 -692. Benthem, F. D. J. von, Imperatriz-Fonseca, V. L.. and Velthuis, H. H. W. (199s). Biology of the stingless bees Plebeia remofa (Holmberg): Observations and evolutionary implications. Insectes Sociaux 42, 71-87. Bohm, M. K. (1972). Effccts of environment and juvcnile hormone on ovaries of the wasp. Polistes metricus. J . Insect Physiol. 18, 1875- 1883. Calderone, N. W., and Page, R. E. (1991). The evolutionary genetics of division of labor in colonies of the honey bee, Apis mellifera. Am. Nat. 138,69-92.
REGULATION OF AGE POLYETHISM IN BEES A N D WASPS
311
Cameron, S. A., and Robinson, G . E. (1990). Juvenile hormone does not affect division of labor in bumble bee colonies (Hymenoptera: Apidae). Ann. Enfomol. Soc. Am. 83, 626-631. Carpenter, J. M. (1991). Phylogenetic relationships and the origin of the social behavior in the Vespidae. In “The Social Biology of Wasps” (K. G . Ross and R. W. Matthews, eds.), pp. 7-32. Comstock Publishing, Ithaca, New York. Cayre. M.. Strambi. C., and Strambi, A. (1994). Neurogenesis in an adult insect brain and its hormonal control. Nature (London) 368,57-59. Davis, R. L. (1993). Mushroom bodies and Drosophila learning. Neuron 11, 1-14. Davis, R. L., and Han, K-A. (1996). Mushrooming mushroom bodies. Curr. Biol. 6, 146-148. Djerassi, C. (1992). “The Pill, Pygmy Chimps, and Degas’ Horse.” HarperCollins. New York. Dreller. C., Fondrk, M. K.. and Page, R. E. (1995). Genetic variability affects the behavior of foragers in a feral honeybee colony. Naturwissenschafi. 82,243-245. Durst, C.. Eichmiiller. S..and Menzel, R. (1994). Development and experience lead to increased volume of subcompartments of the honeybee mushroom body. Behav. Neural. Biol. 62, 259-263. Erber, J., Masuhr, T., and Menzel, R. (1980). Localization of short-term memory in the brain of the bee, Apis mellifera. Physiol. Enromol. 5, 343-358. Fahrbach, S. E., Farris. S. M., and Robinson, G . E. (1995). Coincident maturation of flight behavior and the mushroom bodies in the honey bee. Soc. Neurosci. Absrr. 21,458. Fahrbach, S. E., Giray. T.. and Robinson, G . E. (1995). Volume changes in the mushroom bodies of adult honey bee queens. Neurobiol. Learn. Memory 63, 181-191. Fahrbach, S. E., and Robinson, G. E. (1996). Juvenile hormone, behavioral maturation, and brain structure in the honey bee. Dev. Neurosci. 18, 102-114. Fahrbach, S. E., Strande, J. L., and Robinson, G . E. (1995). Neurogenesis is absent in the brains of adult honey bees and does not explain behavioral neuroplasticity. Neurosci. Lrrr. 197, 145-148. Farris. S. M., Robinson, G . E., and Fahrbach, S. E. (submitted). Behavioral reversion affects juvenile hormone titers but not brain structure in the worker honey bee. J . Neurobiol. Fluri. P.. Liischer, M., Wille, H., and Gerig. L. (1982). Changes in the weight of the pharyngeal gland and haemolymph titres of juvenile hormone, protein and vitellogenin in worker honey bees. J . Insect Physiol. 28, 61-69. Fluri. P.. Sabatini, A. G . , Vecchi, M. A., and Wille, H. (1981). Blood juvenile hormone, protein, and vitellogenin titres in laying and non-laying queen honeybees. J. Apiculrural Res. 20, 221-225. Fluri. P.. Wille, H., Gerig, L., and Liischer, M. (1977). Juvenile hormone, vitellogenin, and haemocyte composition in winter worker honeybees (Apis mellifera). Experienria 33, 1240- 1241. Frisch, K. von (1967). “The Dance Language and Orientation of Bees.” Harvard University Press, Cambridge, Massachusetts. Cast. R. (1967). Untersuchungen iiber den Einfluss der Koniginnensubstanz auf den Entwicklung dcr endokrinen Driisen bei der Arbeiterin der Honigbiene (Apis mellifica). Insectes Sociaux 14, 1-12. Giray, T., and Robinson, G . E. (1994). Effects of intracolony variability in behavioral development on plasticity of division of labor in honey bee colonies. Behav. Ecol. Sociobiol. 35, 13-20. Goodman, W. G., Orth. A. P., Toong, Y. C., Ebersohl, R., Hiruma, K., and Granger, N. A. (1 995). Recent advances in radioimmunoassay technology for the juvenile hormones. Arch. Insect Biochem. Physiol. 30, 295-306. Granger, N. A., and Goodman, W. G . (1988). Radioimmunoassays: Juvenile hormones. In “Immunological Techniques in Insect Biology” (L. 1. Gilbert and T. A. Miller, eds.), pp. 215-251. Springer-Verlag. New York.
312
SUSAN E. FAHRBACH
Hagenguth, H., and Rembold, H. (1978). Identification of juvenile hormone 3 as the only J H homolog in all developmental stages of the honey bee. Z . Naturforschung. 33C, 847-850. Hammer, M., and Menzel, R. (1995). Learning and memory in the honeybee. J . Neurosci. 15, 1617-1630. Hardie, J., and Lees, A. D. (1985). Endocrine control of polyphenism and polymorphism. In “Comprehensive Insect Physiology, Biochemistry, and Pharmacology” (G. A. Kerkut and L. I. Gilbert, eds.). Pergamon Press, Oxford, United Kingdom. Holler, C., Bargen, H., Vinson, S. B., and Witt, D. (1994). Evidence for the external use of juvenile hormone for host marking and regulation in a parasitic wasp, Dendrocerus carpenteri. J. Insect Physiol. 40,317-322. Huang, Z.-Y., and Robinson, G. E. (1995). Seasonal changes in juvenile hormone titers and rates of biosynthesis in honey bees. J. Comp. Physiol. B 165, 18-28. Huang, 2.-Y., and Robinson, G. E. (1996). Regulation of division of labor in honey bee colonies by an activator-inhibitor mechanism. Behav. Ecol. Sociobiol. 39, 147-158. Huang, Z.-Y.. Robinson, G. E., and Borst, D. W. (1994). Physiological correlates of division of labor among similarly aged honey bees. J. Comp. Physiol. A 174,731-739. Huang, 2.-Y., Robinson, G. E., Tobe, S. S., Yaki, K. J., Strambi, C.. Strambi, A., and Stay, B. (1991). Hormonal regulation of behavioural development in the honey bee is based on changes in the rate of juvenile hormone biosynthesis. J . Insect Physiol. 37,733-741. Hunnicutt, D., Toong, Y. C., and Borst, D. W. (1989). A chiral specific antiserum for juvenile hormone. Am. Zool. 29,48a. Hunt, G. J., Page, R. E., Fondrk, M. K., and Dullum, C. J. (1995). Major quantitative trait loci affecting honey bee foraging behavior. Genetics 141, 1537-1545. Imboden, H.. and LUscher, M. (1975). Allatektomie bei adulten Bienen-Arbeiterinnen (Apis mellifera). Rev. Suisse Zool. 82,694-698. Jaycox, E. R. (1976). Behavioral changes in worker honey bees (Apis mellifera L.) after injection with synthetic juvenile hormone (Hymenoptera: Apidae). J . Kansas Entomol. SOC. 49,165-170. Jaycox, E. R., Skrwronek, W., and Gwynn, G. (1974). Behavioral changes in worker honey bees (Apis mellifera) induced by injections of a juvenile hormone mimic. Ann. Entomol. Soc. Am. 67,529-534. Jeanne, R. L. (1991). Polyethism. In “The Social Biology of Wasps” (K. G. Ross and R. W. Matthews, eds.). Comstock Publishing, Ithaca, New York. Jones, G. (1995). Molecular mechanisms of action of juvenile hormone. Annu. Rev. Entomol. 40, 147-169. Kerr, W. E., and Netos, G. R. d. S. (1956). Contribuocao para o conhecimento da binomia dos Meliponin. 5 . Divisao de trabalho entre as operarias de Melipona quadrifasciata quadrifasciata Lep. Insectes Sociaux 3,423-430. Konishi, M. (1994). An outline of recent advances in birdsong neurobiology. Brain Behav. EvoI. 44,279-285. Kort, C. D. (1969). Hormones and the structural and biochemical properties of the flight muscles in the Colorado beetle. Medelingen Landbouwhogeschool Wageningen Ned. 69, 1-63. Krebs, J. R. (1990). Food-storing birds: Adaptive specialisation in brain and behaviour? Phil. Trans. R. SOC.Lond. B 329,153-160. Krebs, J. R., Sherry, D. F., Healey, S. D., Perry, V. H., and Vaccarino, A. L. (1989). Hippocampal specialization of food-storing birds. Proc. Natl. Acad. Sci. U.S.A. 86, 1388-1392. Law, J. H., Yuan, C., and Williams, C. M. (1966). Synthesis of a material with high juvenile hormone activity. Proc. Natl. Acad. Sci. U.S.A. 55, 576-578.
REGULATION OF AGE POLYETHISM. IN BEES A N D WASPS
313
Liischer. M. (1960). Hormonal control of caste differentiation in termites. Ann. New York Acad Sci. 89, 549-563. Liischer, M. (1972). Environmental control of juvenile hormone (JH) secretion and caste differentiation in termites. Gen. Comp. Endocrinol. (Suppl. 3). 509-514. Liischer, M. (1976). Introduction. I n “Phase and Caste Determination in Insects-Endocrine Aspects” (M. Liischer, eds.), pp. 1-4. Pergamon Press, Oxford. Mauchamp, B., Lafont, R., and Krien. P. (1981). Analysis of juvenile hormones by highperformance liquid chromatography coupled with mass spectrometry. In “Juvenile Hormone Biochemistry” (G. E. Pratt and G. T. Brooks, eds.), pp. 21-31. ElseviedNorthHolland Biomedical Press, Amsterdam. Menzel, R. (1985). Learning in honey bees in an ecological and behavioral context. I n “Experimental Behavioral Ecology and Sociobiology” (B. Halldobler and M. Lindauer, eds.), pp. 55-74. Gustav Fischer Verlag, New York. Menzel. R. (1990). Learning, memory, and “cognition” in honey bees. I n “Neurobiology of Comparative Cognition” (R. P. Kesner and D. S. Olton, eds.), pp. 237-292. Lawrence Erlbaum. Hillsdale, New Jersey. Michener. C. D. (1961). Social polymorphism in the Hymenoptera. Symp. R. Enl. Soc. Lond. 1, 43-56. Michener, C. D. (1974). “The Social Behavior of the Bees. A Comparative Study.” Harvard University Press, Cambridge, Massachusetts. Mizunami, M., Weibrecht, J. M., and Strausfeld, N. J. (1993). A new role for the insect mushroom bodies: Place memory and motor control. I n “Biological Neural Networks in Invertebrate Neuroethology and Robotics” (R. D. Beer, R. E. Ritzman, and T. McKenna. eds.), pp. 199-225. Academic Press, New York. Mobbs, P. G. (1982). The brain of the honeybee Apis mellifera L. The connections and spatial organization of the mushroom bodies. Phil. Trans. R. SOC. Lond. B 298,309-354. Mobbs, P. G . (1985). Brain structure. I n “Comprehensive Insect Physiology, Biochemistry and Pharmacology” (G. A. Kerkut and L. Gilbert, eds.), pp. 299-370. Pergamon, Oxford. Moore, D.. Angel, J. E., Cheeseman, I. M., Robinson, G. E., and Fahrbach, S . E. (1995). A highly specialized social grooming honey bee (Hymenoptera: Apidae). J . Insect Behav. 8, 855-861. Muller, W. J., and Hepburn, H. R. (1994). Juvenile hormone I11 and waxsecretion in honeybees. J. Insect Physiol. 40,873-883. Nijhout, H. F. (1994). “Insect Hormones.” Princeton University Press, Princeton, New Jersey. Nijhout. H. F., and Wheeler, D. E. (1982). Juvenile hormone and the physiological basis of insect polymorphisms. Quart. Rev. Biol. 57, 109-133. Nottebohm, F. (1981). A brain for all seasons: Cyclical anatomical changes in song control nuclei of the canary brain. Science 214, 1368-1370. O’Donnell, S.. and Jeanne, R. L. (1993). Methoprene accelerates age polyethism in workers of a social wasp (Polybia occidentalis). Physiol. Entomol. 18, 189-194. Oster, G. F., and Wilson, E. 0. (1978). “Caste and Ecology in the Social Insects.” Princeton University Press, Princeton, New Jersey. Page, R. E., and Laidlaw, H. H. (1988). Full sisters and super sisters: A terminological paradigm. Anim. Behav. 36, 944-945. Page, R. E., and Robinson, G. E. (1991). The genetics of division of labour in honey bee colonies. Adv. Insect Physiol. 23, 117-169. Page. R. E., Waddington, K. D.. Hunt, G. J., and Fondrk, M. K. (1995). Genetic determinants of honey bee foraging behaviour. Anim. Behav. 50,1617-1625. Pener, M. P. (1985). Hormonal effects on flight and migration. I n “Comprehensive Insect Physiology, Biochemistry, and Pharmacology” (G. A. Kerkut and L. I. Gilbert, eds.), pp. 44-490. Pergamon Press, Oxford.
3 14
SUSAN E. FAHRBACH
Pratt, G. E., and Tobe, S. S. (1974). Juvenile hormones radiobiosynthesized by corpora allata of adult female locusts in vitro. Life Sci. 14, 575-586. Riddiford, L. M. (1994). Cellular and molecular actions of juvenile hormone. 1. General considerations and premetamorphic actions. Adv. Insect Physiol. 24, 213-274. Robinson. G. E. (1985). Effects of a juvenile hormone analogue on honey bee foraging behavior and alarm pheromone production. J. Insect Physiol. 34, 277-282. Robinson. G. E. (1987a). Modulation of alarm pheromone perception in the honey bee: evidence for division of labor based on hormonally regulated response thresholds. J. Comp. fhysiol. A 160, 613-619. Robinson, G. E. (1987b). Regulation of honey bee age polyethism by juvenile hormone. Behav. Ecol. Sociobiol. 20, 329-338. Robinson, G . E. (1992). Regulation of division of labor in insect societies. Annil. Rev. Entomol. 37,637-665. Robinson, G. E., and Page, R. E. (1988). Genetic determination of guarding and undertaking in honey-bee colonies. Nature (London) 333, 356-358. Robinson, G . E., and Page, R. E. (1989). Genetic determination of nectar foraging, pollen foraging, and nest-site scouting in honey bee colonies. Behav. Ecol. Sociobiol. 24,317-323. Robinson, G . E., and Page, R. E. (1995). Genotypic constraints on plasticity for corpbe removal in honey bee colonies. Anim. Behav. 49,867-876. Robinson, G . E., Page, R. E., Strambi, C., and Strambi, A. (1989). Hormonal and genetic control of behavioral integration in honey bee colonies. Science 246, 109-1 12. Robinson, G. E., Page, R. E., Strambi, C.. and Strambi, A. (1992). Colony integration in honey bees: Mechanisms of behavioral reversion. Ethology 92,336-348. Robinson, G. E., and Ratnieks. F. L. (1987). Induction of premature honey bee (Hymenoptera: Apidae) flight by juvenile hormone analogs administered orally or topically. J. Econ. Entomol. 80, 784-787. Robinson, G. E., Strambi, C.. Strambi, A., and Feldlaufer, M. F. (1992). Comparison of juvenile hormone and ecdysteroid titres in adult honey bee workers and queens. J . Insect Physiol. 37, 929-935. Robinson, G. E., Underwood, B. A., and Henderson, C. E. (1984). A highly specialized watercollecting honey bee. Apidologie 15, 355-358. Roselcr, P. F. (1985). Endocrine basis of dominance and reproduction in polistine paper wasps. In “Experimental Behavioral Ecology and Sociobiology” (B. Holldobler and M. Lindauer, eds.), pp. 259-272. Sinauer Associates, Sunderland, Massachusetts. Roseler, P. F., and RL)seler, I. (1978). Studies on the regulation of the juvenile hormone titer in bumblebee workers, Bombus terrestris. J. Insect fhysiol. 24, 707-713. Ross, K . G.. and Matthews. R. W. (1991).“The Social Biology of Wasps.” Comstock Publishing, Ithaca, New York. Rutz, W., Gerig, L., Wille, H., and Liischer, M. (1976). The function of juvenile hormone in adult worker honeybees, Apis mellifera. J . Insect Physiol. 22, 1485-1491. Sasagawa. H. (1988). Hormonal control of division of labor in worker honeybees. Apis mellifera L.-with special reference to the implication of juvenile hormone ( J H ) and quantitative determination of J H titer by H f LC. Unpublished doctoral dissertation, University of Tsukuba, Japan. Sasagawa, H., Sasaki, M., and Okada, I. (1989). Hormonal control of the division of labor in adult honeybees (Apis mellifera L.)I. Effect of methoprene on corpora allata and hypopharyngeal gland, and its alpha-glucosidase activity. Appl. Entomol. Zool. 24,66-77. Seeley, T. D. (1982). Adaptive significance of the age polyethism schedule in honeybee colonies. Behav. Ecol. Sociobiol. 11, 287-293. Seeley, T. D. (1985). “Honeybee Ecology.” Princeton University Press, Princeton, New Jersey.
REGULATION OF AGE POLYETHISM IN BEES A N D WASPS
315
Sheppard. W. S., and Berlocher, S. H. (1984). Electrophoretic phylogeny of Apis species. Apidologic 201, 115-123. Snodgrass. R. E. (1984). “Anatomy of the Honey Bce.” Cornell University Press, Ithaca, New York. Staal, G . B. (1975). Insect growth regulators with juvenile hormone activity. Annu. Rev. Enromol. 20,417-460. Stay. B., Bachmann, J. A. S., Stoltzman, C. A., Fairbairn, S. E., Yu, C . G . , and Tobe, S. S . (1994). Factors affecting allatostatin release in a cockroach (Diplopferupuncruru): Nerve section, juvenile hormone analog and ovary. J. Insect Physiol. 40,365-372. Strambi. C.. Strambi, A,, deReggi, M. L. and Delaage. M. A. (1984). Radioimmunoassays of juvenile hormone. State of the methods and recent data on validation. I n “Biosynthesis, Metabolism, and Mode of Action of Invertebrate Hormones” (J. Hoffman and M. Porhet, eds.). pp. 355-362. Springer-Verlag, Berlin. Stramhi, C., Strambi, A,, deReggi, M. L., Hira, M. H., and Delaage, M. A. (1981). Radioimmunoassay of insect juvenile hormones and their diol derivatives. Eur. J. Biochem. 118, 40 1-406. Strausfeld. N. J., Buschbeck. E. K., and Gomez, R. S. (1995). The arthropod mushroom body: Its functional roles, evolutionary enigmas and mistaken identities. In “The Nervous System of Invertebrates: An Evolutionary and Comparative Approach” (0.Breidhach and W. Kutsch. eds.). pp. 349-381. Birkhauser Verlag. Basel, Switzerland. Tohe. S. S., and Pratt. G . E. (1974). The influence of substrate concentrations on the rate of insect juvenile hormone hiosynthesis by corpora allata of the desert locust in vifro. Biochem. J . 144,417-422. Truman, J. W.. and Riddiford. L. M. (1974). Hormonal mechanisms underlying insect behaviour. Adv. Insect Physiol. 10, 297-352. Wigglesworth, V. B. (1934). The physiology of ecdysis in Rhodnius prolixus. 11. Factors controlling molting and metamorphosis. Q. J. Microscop. Sci. 77, 191-222. Wigglesworth, V. B. (1 936). The function of the corpora allatum in the growth and reproduction of Rhorfnius prolixus (Hemiptera). Q. J . Microscop. Sci. 79, 91-121. Wildc. J . de. and Beetsma, J. (1982). The physiology of caste development in social insects. Adv. lnsecr Physiol. 16, 167-246. Williams. C. M. (1967). Third generation pesticides. Sci. Amer. 217, 13-17. Wilson. E. 0.(1971). “The Insect Societies.” Harvard University Press, Cambridge, Massachusetts. Winston, M. L. (1987). “The Biology of the Honey Bee.” Harvard University Press, Cambridge, Massachusetts. Withers, G . S., Fahrbach, S. E., and Robinson, G . E. (1993). Selective neuroanatomical plasticity and division of labour in the honeybee. Nature (London) 364,238-240. Withers. G . S.. Fahrbach. S. E., and Robinson, G. E. (1995). Effects of experience and juvenile hormonc on the organization of the mushroom bodies of honey bees. J. Neurohiol. 26, 130-144. Witthiift. W. (1967). Absolute Anzahl und Verteilung der Zellen im Hirn der Honig biene. Z . Morphol. Tiere 61, 160-164. Woodhead. A. P.. Stay, B.. Seidel. S. L., Tobe, S. S., and Khan, M. A. (1989). Primarystructure of four allatostatins: neuropeptide inhibitors of juvenile hormone synthesis. Proc. Narl. Acad. Sci. U.S.A. 86, 5997-6001. Yang. M. Y., Armstrong, J . D., Vilinsky, I., Strausfeld, N. J., and Kaiser, K. (1995). Subdivision of the Drosophilu mushroom bodies by enhancer-trap expression patterns. Neuron 15, 45-54.
316
SUSAN E. FAHRBACH
Yoshikawa, K. (1963). Introductory studies on the life economy of polistine wasps. 11. Superindividual stage. 2. Division of labor among workers. Japanese J. Ecol. 13,53-57. Yu, C. G., Stay, B., Joshi, S., and Tobe, S. S. (1993). Allatostatin content of brain, corpora allata, and haemolymph at different developmental stages of the cockroach, Diplopteru punctatu: quantitation by ELISA and bioassay. J. Insect Physiol. 39, 111-122.
ADVANCES IN THE STUDY OF BEHAVIOR. VOL. 26
Acoustic Signals and Speciation: The Roles of Natural and Sexual Selection in the Evolution of Cryptic Species GARETH JONES SCIENCES, BRISTOL,
SCHOOL OF BIOLOGICAL
UNIVERSITY OF
BRISTOL, UNITED
KINGDOM
I. INTRODUCTION In 1768, Gilbert White, a pioneer of natural history studies, wrote to fellow naturalist Thomas Pennant about warblers singing near his home at Selborne: “I have now, past dispute, made out three distinct species of the willow-wrens. . . . which constantly and invariably use distinct notes” (White, 1813). White was referring to three species of warblers in the genus Phylloscopus-the chiffchaff, P. collybita, the willow warbler, P. trochilus, and the wood warbler, P. sibilutrix. The three warblers look so similar superficially, that it was not until 30 years after White’s acute study of their different songs that they were described as different species. In Europe, the marsh tit, Parus palustris, and willow tit, P. rnontunus, are similar in appearance, but are separated easily by voice. The treecreeper, Certhia familiaris, and the short-toed treecreeper, C. brachydactylu, are almost identical to our eyes, yet their calls are quite different. Many animal species may remain undescribed because, although they seem indistinguishable from described species by their appearance, they maintain reproductive isolation by cues that are not so obvious to our senses. To humans, many animal species look similar, yet sound different. Humans have considerable visual abilities over a wide range of light wavelengths, and animals can be classified with ease according to visual criteria. Although humans also have considerable auditory skills, especially between 500 Hz and 3 kHz, differences in acoustic signals are more difficult to quantify than differences in appearance. Vocalizations must be studied in living animals, while visual differences can be assessed by examination of museum specimens. Acoustic signals can be difficult to locate, especially 317
Copyright 0 1W7 by Academic Press All rights of reproduction in any form reserved. m5-3454197 $U.Oo
318
GARETH JONES
as they are often used by animals living in niches where vision is of limited use (at night, or in dense cover, for instance, see Endler, 1992). Many animals call at frequencies beyond our range of hearing, and descriptions of vocal differences often require sophisticated bioacoustical analyses. For these reasons, acoustic differences among animals may have been neglected in comparison with visual differences. Moreover, for reasons that I shall explore later, morphology may often be conserved during evolution, while behavioral traits, such as acoustic signaling, may be more flexible. Behavioral traits often reflect the presence of species that might remain cryptic on morphological grounds alone. Given that the rates of morphological, behavioral, and genetic divergence during and after speciation are only very loosely coupled (e.g., Bruna, Fisher, and Case, 1996), it is interesting to explore under what selection pressures behavioral divergence occurs. Species that seem similar are called cryptic species. A cryptic species is “a species the diagnostic features of which are not easily perceived: a sibling species” (Mayr, 1977). Cryptic species are often thought of as synonymous with sibling species (see Mayr’s definition), though Mayr defines sibling species separately as “morphologically similar or identical populations that are reproductively isolated” (Mayr, 1977). What is not clear from the definition of a cryptic species is what is meant by the perceiver. Animals often have perception that is very different from humans, and two species that seem identical to our eyes may appear very different t o the animals concerned. Birds, for example, are sensitive to ultraviolet (UV) light and may perceive colors differently from humans. Indeed, UV vision is probably important in signaling in birds (Bennett, Cuthill, and Norris, 1994), and humans are probably atypical among animals in not having sensitivity to UV (Bennett and Cuthill, 1994). Color, after all, is not absolute, and is in part the consequence of an animal’s neural processing abilities. In this chapter, I will use the term cryptic in terms of human vision. The important point here is that some cryptic species probably do look similar to other, closely related species, while others may not. Without fully understanding the visual physiology of the animals concerned, crypticity can be defined only according to our visual performance. Throughout this chapter I define “appearance” as apparent form, look, or aspect (Oxford English Dictionary, 1991). “Appearance” is clearly a subjective description of an item. The quantitative description of shape is termed “morphometrics” (Rohlf, 1990). Cryptic species are often similar in morphometrics also; although morphology can be assessed relatively objectively, comparisons based on appearance may be more difficult for us to quantify. Morphology, that is, structure, is only one component of appearance. Why do some animals look so alike to our eyes, yet are so different in their acoustic behavior? Of course, vocal differences d o not always imply
ACOUSTIC SIGNALS AND CRYPTIC SPECIES
319
speciation. In songbirds, for example, often only males sing, and regional dialects occur within species that may have little relevance to speciation (Catchpole and Slater, 1995). Nevertheless, I believe that the study of animals that use acoustic signals may shed light on speciation mechanisms, and on how natural selection and sexual selection may play quite different roles in shaping acoustic divergence in cryptic species. I also believe that the limited sensory performance of humans may result in a bias in descriptions of biological diversity. Species that are so obviously different to our eyes are more likely to get described. Many cryptic species probably remain undescribed, yet these may have considerable genetic divergence from their closest living relatives. Relationships between genetic divergence and crypticity are important in that they may allow determination of whether cryptic species evolved recently, or whether “appearance” may be conserved despite long-term separation from ancestral species. The study of cryptic species is clearly important for understanding biodiversity. Cryptic species that differ in olfactory characters also occur, though little research seems to have been conducted on such animals, probably because humans have a relatively poor sense of smell. Two species of Australian scorpionflies are almost identical in morphology, but fail to interbreed in the laboratory because the male sex pheromones differ between the cryptic species (Bornemissza, 1966). I predict that many cryptic species of animals with different olfactory characters that are used as secondary sexual signals will be described in the future, but this chapter will concentrate on cryptic species that differ in vocalizations. Animals that communicate largely by acoustic signals include bats, anurans, and many insects and birds. Vision may not be an effective means of signaling in low levels of ambient light (e.g., Martin, 1990; Endler, 1992; Romer, 1993), and I expect acousticcryptic species to be more prevalent in nocturnal taxa, and in species that live in dense foliage or turbid water, where vision is of restricted use. Vocal signals are also important for orientation and prey capture by echolocation in bats (Griffin, 1958), and in this chapter I use signals in a broader definition than that normally used by behavioral ecologists (e.g., Dawkins and Krebs, 1978), where signaling is synonymous with communication. Bat echolocation calls are traditionally referred to as signals, even though their use is primarily in orientation and prey capture, rather than in communication (though see Fenton, 1985, for the role of echolocation calls in communication). I will use a case study on pipistrelle bats to illustrate how much hidden biodiversity may remain unrecognized. It was long assumed that the pipistrelle, Pipistrellus pipistrelfus, was one species. The pipistrelle is believed to be the most widespread European bat, and is probably the most common (Stebbings and Griffith, 1986; Stebbings, 1988). Studies on echolocation calls showed that pipistrelles belong to one of two “phonic types,” however
320
GARETH JONES
(Jones and Parijs, 1993). Subsequent work, summarized later, confirmed that vocal differences were not the consequence of jamming avoidance or habitat-specific echolocation, but that the phonic types are cryptic species with considerable genetic divergence. The pipistrelle example shows how natural selection may shape vocal divergence in echolocating bats. Why the differences in echolocation calls, yet the remarkable similarities in appearance and morphology? I argue that there may be strong selective pressures for acoustic divergence in echolocating animals where call frequency may influence echo strength from acoustic targets of different sizes, and hence prey selection. In this respect, echolocating bats may differ from most insects, anurans and birds, where female choice for elaborate vocalizations may be a stronger force promoting acoustic divergence. I consider whether divergence in the echolocation calls of cryptic bat species can evolve by sympatric speciation. I discuss whether intersexual selection for acoustic characters may by itself result in sympatric speciation in nonecholocating animals. The bird group containing the largest number of species is the passerines, where acoustic signals have reached their greatest elaboration, so the importance of acoustic signaling in speciation certainly deserves closer attention. I argue that sexual selection, rather than natural selection, is the prime force in shaping acoustic divergence in nonecholocating animals. I consider whether cryptic species are recently evolved species, or whether they are as genetically distinct as “conventional” species, and conclude by discussing the relevance of the genetic constitution of cryptic species to our understanding of biodiversity.
BRITAIN’S MOSTCOMMON BATIS Two SPECIES 11. A CASESTUDY: A. ECHOLOCATION CALLS
The pipistrelle has long been known to have a diverse range of echolocation calls. Calls may change within individuals in relation to foraging situation, reflecting behavioral adaptation (e.g., Kalko and Schnitzler, 1993; Kalko, 1995), but substantial variation in call frequency also exists between individuals. Miller and Degn (1981) interpreted individual variation in echolocation calls as having an antijamming function. They hypothesized that, when bats flew in groups, they altered the dominant frequency of their calls so that their own echoes had less chance of being confused with echoes from the calls of conspecifics. This hypothesis was based on observational data, yet pipistrelles flying in the laboratory do not appear to change the frequency of their calls when flying in groups compared with when flying
321
ACOUSTIC SIGNALS AND CRYPTIC SPECIES
alone (G. Jones, unpublished data). Thus, jamming avoidance cannot explain between-individual variation in the echolocation calls of pipistrelles. Jamming is avoided not by frequency shifting, but perhaps by each bat having a distinctive call and associated personalized frequency response (see Suga, Niwa, Taniguchi, and Margoliash, 1987), and by neural filters that limit the bats’ responses to echoes that arrive only within particular time windows that correspond to expected return times from the individual’s own calls (Pollak and Casseday, 1989). The echolocation calls emitted by pipistrelles searching for prey are described as “FM/CF.” Thus, they have a broadband frequency-modulated (FM) sweep terminating in an almost (or quasi-) constant-frequency (CF) tail (Jones and Parijs, 1993; Fig. 1). Most of the energy in the call is concentrated in the CF tail, so the peaks of power spectra (frequencies containing most energy) from sound analyses correspond to the frequency of the CF tail. The frequency-modulated part of the call is probably adapted to range determination, while the CF tail may be effective for long-distance echolocation. or for the detection of “glints” from insect wingbeats (e.g.,
b
a 120
n
As
80
I
I1 I
I
L,
6a
3
;
w
40
10 ms 0
20
60
100
140
Frequency (kHz) FIG. 1. Search phase echolocation calls of the two phonic types of pipistrelle bats. (A) Sonagrams. (B) Power spectra. Note how most energy in the calls is in the terminal CF tail. Call I had a frequency containing most energy (FMAXE) of 46 kHz, and was emitted by a “45-kHz” pipistrelle. Call I1 had a FMAXE of 54 kHz, and was emitted by a “55-kHz” pipistrelle. Redrawn from Jones and Parijs (1993) with permission from The Royal Society.
322
GARETH JONES
Neuweiler, 1990). When the bats are searching for insects in open spaces, they emit “search-phase’’ echolocation calls (Griffin, Webster, and Michael, 1960), like those described previously. If the bats fly in more cluttered situations, call duration is shortened, the CF tail is less pronounced, and the calls become more broadband (Kalko and Schnitzler, 1993). Broadband calls will give greater spatial resolution in clutter (Simmons, Fenton, and O’Farrell, 1979), and the shortening of the calls allows the bats to avoid overlap between outgoing pulse and incoming echo as the bats approach targets (Kalko and Schnitzler, 1993; Kalko, 1995). Pipistrelles in Switzerland emit search phase calls with C F tails between 41 and 62 kHz that form two separate clusters in a multivariate analysis of call parameters (Zingg, 1990). The two groups of pipistrelles showed very little overlap in the terminal frequencies of their echolocation calls: one group showed a median end frequency of 44.8 kHz, the other 57.5 kHz. Zingg interpreted this difference in terms of habitat selection by pipistrelles-when the bats flew in one type of habitat they used one sort or signal, switching to another call type in a different situation. Jones and Parijs (1993) showed that pipistrelle calls fell into two frequency bands in Britain (Fig. 1). One call type had frequencies of most energy averaging 46 kHz, the other close to 55 kHz. The bats are referred to as the “45-kHz” and “55-kHz” phonic types for simplicity hereafter. In summer, female bats usually form maternity colonies where they give birth and rear their young. Every maternity colony studied contained bats of only one phonic types (Fig. 2). Because volant juveniles of both sexes were recorded leaving maternity roosts, it was clear that the call differences of the magnitude observed were not caused by sex or age effects, as seen in some other bat species (e.g., Jones and Ransome, 1993; Jones and Kokurewicz, 1994). Males, the smaller sex, emit calls that have peak frequencies on average about 2 kHz higher than females in both phonic types of pipistrelle (Park, Altringham, and Jones, 1996), but between-sex differences were small compared with differences between phonic types. Habitat specificity of echolocation calls, as suggested by Zingg (1990), could be rejected, since a bimodal distribution of call frequencies was maintained when bats of both phonic types were released into similar habitats. The released bats always used the same frequencies as those of roostmates flying from their maternity colonies (Jones and Parijs, 1993; Fig. 3). The phonic types exist in sympatry over much of Britain and Europe, so the possibility of them being geographical races or subspecies was dismissed (Jones and Parijs, 1993). Interpulse intervals of 45-kHz bats are longer (82.25 ? 10.48 (SD)ms, N = 125) than those of 55-kHz pipistrelles (76.60 2 10.49 ms, N = 229) on average (Jones and Parijs, 1993). This difference is expected if 45-kHz bats were larger (see later discussion), since repetition rate (which is linked
ACOUSTIC SIGNALS AND CRYPTIC SPECIES
I-
n
10
-
20
-
3
0 10
-
nl
20
3
cd P
323
cw
0 L
E
0
0 5
:e0
10
40
45
50
55
60
Frequency (kHz) FIG.2. Maternity colonies of pipistrelle bats contain only one phonic type of bats. Data are taken from bats exiting 12 maternity colonies, and represent frequency distributions of the frequencies of most energy of cells emitted by individual bats. Colonies are arranged in order of increasing mean call frequency for the sample. Means (solid squares) 2 SDs of calls from each roost are illustrated.RedrawnfromJones andParijs(l993) withpermissionfromThe RoyalSociety.
with wingbeat frequency 1: 1 in the search phase for bats that hunt by aerial hawking) scales negatively with body mass in a cross-species comparison (Jones, 1994). In Europe, 45-kHz bats have been recorded in France, The Netherlands (Jones and Parijs, 1993), Germany (Kalko and Schnitzler, 1993), Luxembourg (C. Harbusch, unpublished data) and in Poland, Slovakia, and The
324
GARETH JONES
I T
T
I
30
31
32
33
34
Forearm length (mm) FIG.3. Pipistrelle bats maintain phonic type when released in the same habitat. The relation between call frequency and forearm length is shown for 48 pregnant or lactating female pipistrelles from seven roosts. All bats were released in open habitats, away from clutter. Solid symbols are 55-kHz pipistrelles, open symbols are 45-kHz bats, and each symbol represents membership of a different maternity colony. Means f SDs of five calls are illustrated. Redrawn from Jones and Parijs (1993) with permission from The Royal Society.
Czech Republic (J. Rydell and P. A. Racey, unpublished data; Fig. 4). However, most Scandinavian (Jones and Parijs, 1993; Denmark: Miller and Degn, 1981; Norway: T. Stormark, unpublished data) and Mediterranean (e.g., Greece: Weid and Helversen, 1987; Portugal and Spain: T. Guillen, A. Rainho, and N. Vaughan, unpublished data) records are of 55-kHz pipistrelles. Forty-five-kHz bats occur in south Denmark (H. Baagge, unpublished data), and this phonic type has recently been discovered in northern Germany (H. J. G. A. Limpens, unpublished data) and northeastern Poland (A. Rachwald, unpublished data). Both phonic types are widespread and sympatric in Britain (Jones and Parijs, 1993), Northern Ireland (J. M. Russ, unpublished data) and Switzerland (Zingg, 1990). Despite extensive searching in the middle latitudes of Europe (e.g., The Netherlands: Kapteyn, 1993), 55-kHz pipistrelles appear to be absent over much of this area.
ACOUSTIC SIGNALS AND CRYPTIC SPECIES
325
FIG.4. The distribution of phonic types of pipistrelle bat in Europe. Open squares represent 55-kHz pipistrelles; solid squares represent 45-kHz bats. The source of data is given in the text. Note how 55-kHz bats may be absent from much of central Europe, though they are widespread in Scandinavia, Britain, and the Mediterranean. The map does not show the relative abundance of the two phonic types in different parts of their range (see text for details).
The overall impression is that the 55-kHz phonic type is more abundant in northern and southern Europe, with the 45-kHzphonic type being found mainly in middle latitudes. Thus, 55-kHz bats may be pushed to the edge of the range of pipistrelles, on either side of a core distribution of 45-kHz bats. B. SOCIAL CALLS
On foraging grounds, pipistrelles often emit “social calls.” These are relatively low frequency calls (frequency of most energy typically between
326
GARETH JONES
16 and 23 kHz), and are more complex in structure and longer in duration (approximately 35 ms) than echolocation signals. Social calls are very similar in structure to the songflight calls illustrated in Fig. 5, but are repeated at a lower rate (Barlow and Jones, in press, b). Social calls probably travel further than echolocation calls, because the relatively low frequencies used in social calls suffer less from excess atmospheric attenuation (Lawrence and Simmons, 1982), and they are thus well designed for relatively long range communication. Social calls were believed to serve an agonistic function, as they were often associated with chasing behavior (Miller and Degn, 1981; Racey and Swift, 1985). Moreover, most chases (Racey and Swift, 1985) and social calls (Barlow and Jones, in press, c) occur at low insect densities, so social calls may be associated with defense of feeding sites. The social calls of the two phonic types of pipistrelle differ substantially (Barlow and Jones, in press, a,b). Social calls of the 45-kHz bats tend to be slightly lower in frequency than those of 55-kHz bats, and usually consist of four, rather than three components (Barlow and Jones, in press, b). The function of the calls as agonistic signals was confirmed in playback experiments: bats of the 45-kHz phonic type were less active (fewer bat passes were recorded) during playbacks of 45-kHz social calls than during control (tape noise) playbacks, and 55-kHz bats showed a similar response
80 loo
I
45 kHz phonic type
55 kHz phonic type
20
Time (ms) FIG.5. Sonagrams of typical songflight calls of the 45-kHz and 55-kHz phonic types of pipistrelle.
ACOUSTIC SIGNALS AND CRYPTIC SPECIES
327
to playbacks of calls from their own phonic type. A reduction in activity suggests an aversive response to playbacks because it implies that bats left the vicinity of the speaker. Interestingly, neither phonic type responded to playback of calls from the other phonic type, suggesting that social calls do not function in interspecific communication (Barlow and Jones, in press, c). AND REPRODUCTIVE ISOLATION C . MALESONGFLIGHT
The mating strategy of pipistrelles can be described as resource defense polygyny. Males occupy roosts in autumn, and call repeatedly during an advertisement flight (songflight) close to the roost, presumably to attract females. Mating groups of females with single males can be found in bat boxes during the autumn (Gerell and Lundberg, 1985;Lundberg and Gerell, 1986; Gerell-Lundberg and Gerell, 1994). The songflight calls (Fig. 5) are similar in structure to the social calls described earlier (Barlow and Jones, in press, b), but are repeated at a higher rate, often over several hours in the night, and over several weeks in the mating season. As with social calls, bats of the 45-kHz phonic type tend to emit songflight calls with fewer components and of relatively low frequency compared to calls of 55-kHz bats. Because the songflight calls of the two phonic types differ significantly in structure (all individuals can be assigned to the correct phonic type from their calls in multivariate analysis; Barlow and Jones, in press, b), female pipistrelles may use the songflight calls as cues for species recognition during the mating period. The composition of mating groups of pipistrelles in bat boxes has been analyzed in relation to phonic type by Park ef al. (1996). For 16 ringed males and 34 females, all of 27 mating groups investigated contained females of the same phonic type as the male. Where genetic data are available, females associate with males of the same genetic clade (discussed later) (E. M. Barratt, G. Jones, P. A. Racey, T. M. Burland, R. K. Wayne, M. W. Bruford, and R. Deaville, unpublished data). Assortative associations occurred even in areas where the phonic types were sympatric, and hence reproductive isolation between phonic types occurs. D. MORPHOLOGICAL DIFFERENCES Although the two phonic types of pipistrelle appear very similar superficially, subtle differences in appearance are obvious when the bats are examined closely. The 45-kHz bats have fur that is darker brown in color than that of 55-kHz bats. They also usually have a black face mask, while the eyes of 55-kHz bats are surrounded by flesh. It must be emphasized
328
GARETH JONES
that the two phonic types are very variable in fur color and in appearance, so characters such as these may be of little use in field identification. The 45-kHz pipistrelles tended to be larger in the study of Jones and Parijs (1993): 45-kHz pipistrelles had a longer average forearm length (32.37 2 0.59 (SD)mm, N = 18) than 55-kHz bats (31.91 t 0.82 mm, N = 30) and a lower wing aspect ratio (45-kHz bats, 5.46 ? 0.42 (SD), N = 18; 55-kHz bats, 5.82 5 0.72, N = 30). The results from this relatively small sample must be treated with caution, however. The difference in aspect ratio was not confirmed in a larger sample of bats (K. E. Barlow, unpublished data), and no significant differences were found between phonic types in univariate standard measurements of wing morphology (Norberg and Rayner, 1987) or in forearm length. Moreover, the two phonic types could not be separated by a multivariate analysis of wing-shape characters (Jones and Parijs, 1993). Thus, an interesting question arises: if the two phonic types are so similar in flight morphology, how do they partition resources? Presumably resource partitioning does occur, as both phonic types fail to respond to agonistic calls of the other phonic type at feeding sites and a response would be expected if interspecific competition was strong (Barlow and Jones, in press, c). Resource partitioning by diet occurs in sympatric populations of Myotis lucijkgus and M . volans in Canada, despite the two species having similar morphology and echolocation calls (Saunders and Barclay, 1992). Hence resource partitioning between the phonic types of pipistrelle is expected given the radical differences in echolocation, even considering their similar flight morphologies. Morphometric analysis of skulls revealed significant differences between the phonic types. In a multivariate analysis of 7 skull parameters from 57 bats, 79% were classified correctly, with cross-validation included in the model. The length of the mandibular tooth row between the upper canine and M3 was the most important parameter separating the two phonic types (K. E. Barlow and G. Jones, unpublished data). Morphological differences in the baculum (penis bone, or 0s penis) exist among several cryptic bat species (Lanza, 1960; Baagae, 1973). If a difference in baculum shape results in a change in penile morphology, reproductive isolation between cryptic species may occur (Patterson and Thaeler, 1982). The bacula of the two phonic types of pipistrelle are very similar. Bacula have a paired basal flange, an elongate shaft, and a bifid tip (see Hill and Harrison, 1987). The bacula do not show the obvious differences seen between cryptic species in the genera Plecofus (Lanza, 1960) or Myofis (Baagae, 1973). Slight differences do occur in the angle between basal flanges and the shaft between cryptic species, however. The bacula of the 45-kHz bats have flanges that rise gradually from the base, while the bacula of 55-kHz bats show a more acute angle between flange and shaft when
ACOUSTIC SIGNALS AND CRYPTIC SPECIES
329
viewed in profile (K. E. Barlow and G. Jones, unpublished data). Overall, skull and bacular morphometrics emphasize the morphological similarity between the two cryptic pipistrelle species.
DIFFERENCES AND IMPLICATIONS FOR E. BEHAVIORAL RESOURCE PARTITIONING Can differences in echolocation pulses be used to predict differences in microhabitat use by the two phonic types? Bats that emit relatively short duration FM/CF signals are believed to process echoes that do not overlap with the outgoing pulse (Kalko and Schnitzler, 1993), and that return before the next search phase pulse is emitted (Roverud and Grinnell, 1985). For two calls of similar frequency structure but different duration, the shorter call may not mask an echo from a nearby target, while the longer one may. Since pulse duration does not differ between phonic types (approximately 5.3 ms; Jones and Parijs, 1993), a minimum detection distance (Kalko and Schnitzler, 1993) for insects of about 90 cm is expected for both phonic types. Kalko (1995) measured average distances for prey detection of 1.68 m for P. pipistrellus (presumed 45-kHz phonic type), and 1.63 m for P.p. mediterruneus (bats presumed to be of the 55-kHz phonic type). Pipistrelles often use calls of longer duration when searching for prey in open habitats (6-10 ms; Kalko, 1995), and this increased duration probably accounts for the discrepancy between measured detection distances and those predicted by the call durations provided by Jones and Parijs (1993). The longer interpulse interval of the 45-kHz type may give it a greater maximum detection distance (Fenton, 1990), though most echoes from distant objects are probably attenuated severely at this distance (approximately 14 m). Although the 45-kHz phonic type may potentially have a greater overlap-free window (Kalko and Schnitzler, 1993), whether differences in interpulse interval have any implication for influencing habitat use by the phonic types is unlikely. Assuming that the P. pipistrellus mediterraneus recorded by Kalko (1995) in Spain are bats of the 55-kHz phonic type of P. pipistrellus, then the 55-kHz bats have a lower flight speed than the 45-kHz pipistrelles. In a survey of habitat use by bats in southwestern England, Vaughan (1996) found 45-kHz bats over a wide range of habitats, while 55-kHz pipistrelles showed stronger selection for waterside habitats. The 55-kHz pipistrelle may form larger roosts than the 45-kHz phonic type in Britain. The maximum size of four maternity roosts of 45-kHz bats studied by Jones and Parijs (1993) was 160 bats, whereas four of seven roosts of 55-kHz pipistrelles contained more than 250 bats.
330
GARETH JONES
F. GENETIC DIFFERENCES
Phylogenetic relationships of the genus Pipistreflushave been determined using a 400-base-pair sequence of the cytochrome-b gene of mitochondria1 DNA (mtDNA) amplified by the polymerase chain reaction (Barratt et af., 1995). Mitochondria1 DNA was used because it evolves rapidly at the sequence level, and cytochrome-b is a gene sequenced routinely in phylogenetic analyses, thus allowing comparisons to be made with other studies (Avise, 1994). A consensus of the 100 most parsimonious trees showed two distinct clades for the Pipistreflus species investigated, each clade differing by about 12% in cytochrome-b sequence (Fig. 6). Bats classified as P. pipistreflus appeared in both clades, implying that this species may have diphyletic origins. The phylogeny presented in Fig. 6 implies that the speciation event that separated the two clades of P. pipistreflus occurred a long time ago, and may have been followed by other speciation events within the same lineage. Importantly, the two clades of P. pipistreflushave recently been found to correspond completely with the two phonic types described by Jones and Parijs (1993) (E. M. Barratt, G. Jones, P. A. Racey, T. M. Burland, R. K. Wayne, R. Deaville, and M. W. Bruford, unpublished data). A 12% sequence divergence in mtDNA (Barratt et af., 1995) confirms that the two phonic types differ substantially in genetic constitution, despite their morphological similarity. The magnitude of the sequence divergence between the phonic types of pipistrelle described by Barratt et af. (1995) is similar to that found among noncryptic species of Phyffostomus (Van Den Bussche and Baker, 1993), in other genera of phyllostomid bats (Van Den Bussche, Baker, Wichman, and Hamilton, 1993; Baker, Taddei, Hudgeons, and Van Den Bussche, 1994), and among species in the molossid genus Nyctinomops (Sudman, Barkley, and Hafner, 1994) (Fig. 7). In summary, the two phonic types of pipistrelles are classic cryptic species. Despite superficial similarities in appearance, they differ radically in their echolocation calls and in mtDNA sequence. Differences are also apparent in skull morphology, social calls, habitat use, and roost size. Differences in male songflight calls may facilitate reproductive isolation of the cryptic species, and assortative associations between males and females are found in bat boxes during the mating season. The nomenclature of Pipistrelfus pipistreflusis now being revised in light of these findings (G. Jones, in preparation). RESOURCE PARTITIONING BY ECHOLOCATION 111. ACOUSTIC Why might differences in echolocation call frequency promote resource partitioning in echolocating bats? The target strength of insects is likely to
ACOUSTIC SIGNALS AND CRYPTIC SPECIES
331
P. pipistmllus P. ariel
P. bodmhcimm' P. rueppelli
P. pipistreflus
79
P.p . mcditenanrus P. somalicur
99
P. tmuipinnis
P.javanicus P. stmoptcnrs P. mimus p. nRnUS NyctRluS
tlOChllR
BRrbRSteffR barbbastcfhs
FIG. 6. Phylogeny of bats in the genus Pipistrellus based on parsimony analysis of cytochrome-b gene sequences of mitochondria1 DNA. A consensus tree from 100 trees was generated by using heuristic search in PAUP (phylogenetic analysis using parsimony) version 3.0 (Swofford. 1989).with Nyctahs noctula and Barbastella barbastellus as outgroups. Numbers on the tree show bootstrap node confidence values from 100 replications. P. p . mediterruneus is almost certainly the 55-kHz phonic type of pipistrelle. Hence, P. pipistrellus in the lower clade is likely to be the 55-kHz phonic type, that in the upper clade the 45-kHz phonic type. From Barratt et al. (1995). and reproduced with permission from The Zoological Society of London.
depend largely on call frequency (Pye, 1993). For spheres, echo strength falls rapidly when wavelength (the inverse of frequency) exceeds target circumference (Fig. 8). Effectively this relation means that if insects reflect sound in a way similar to spheres, frequencies lower than (i.e., with wavelengths greater than) some measure of target circumference will return
332
GARETH JONES
0 1
2 3
* Chirodema +
4
+
trinitahun - C. doriae Dermanura (Dermanura) azreca - D. phaentis & Chiroderma frinitaturn - C. villn.vum & m u m oueca - D. tolteca & m u m phaeotis D. tolteca
.
5 6
N y ~ t i t w t aurirpitwsus ~p~ - N. laricaudntus Arfibeusfuliginosus - A . ~irumtus
7
Phyllostomus latifolious
0
+ D e m u m phaeotis - D. cinema
9
+ Phyllostomus hastarus - R elongatus
10 11 -b
-
- P discolor
+ Chimdenna
* *
salvini - C. rriniranun Derrnanura cinerea . D . rnlteca (Inlteca)
Phyllos~omurlatifolious - P hasratus Phyllostomus latifolious P elongatus
-
Phyllostomus latifolious
- P stenops
12
Pipistellus pipismllus
13
+ Phyllosrornus slennpf - P. husratus
*
Phyllosrornus elongatus
- P. stennps
14
FIG.7. Magnitude of sequence divergence (%) in cytochrome-b gene sequences (400-1 140 base pairs) for a range of congeneric microchiropteran bats. Data from Van Den Bussche and Baker (1993). Van Den Bussche el al. (1993), Baker et al. (1994). Sudman ef al. (1994), and Barratt ef al. (1995).
little energy in echoes. When wavelength is considerably less than target circumference, then echo intensity remains relatively high and constant. In a simplified way, then, frequencies whose wavelength exceeds sphere circumference will reflect echoes poorly, and high frequencies give stronger echoes from smaller targets. The frequency when echo intensity begins to decline can be predicted with accuracy for spheres, but do such predictions hold for more complex targets such as insects? The relation between target strength and frequency
333
ACOUSTIC SIGNALS AND CRYPTIC SPECIES
-20
4 j 0.2
0.5
1
2
5
10
20
Sphere circumference/wavelength FIG.8. The relation between echo intensity reflected from a sphere in relation to sound frequency. Where sphere circumference/wavelengthis greater than 1, echo intensity is highest. When wavelength is longer than sphere circumference, echo intensity falls off rapidly as wavelength increases. The x-axis represents the scale in frequency for a sphere of 10.8 cm diameter. Adapted from Pye (1993).
has received little empirical study: Waters, Rydell, and Jones (1995) found no obvious relation between target strength and frequency for moths and small flies between 20 and 100 kHz, and argued that the target strength predictions for spheres may not hold for insects. Nevertheless, recent empirical studies on discs (A. P. Norman, unpublished data) and spheres (R. D. Houston, unpublished data) supported Pye’s predicted relationship between target strength and call frequency. If, therefore, target strength from insects did depend on frequency in the way measured for spheres and discs, I predict that the 55-kHz pipistrelles would receive stronger echoes from smaller flying insects than would 45-kHz bats. I would then predict that 55-kHz pipistrelles should eat smaller prey than 45-kHz bats, if the 1.2-mm difference in wavelength between the two phonic types is biologically significant. This hypothesis is now being tested (K. E. Barlow, in preparation). If the sphere model of target strength applies to insect targets, then the echolocation call frequency used by a bat should influence the bat’s ability to detect prey of a given size. Call frequency could be thought of as the ecological equivalent of bill size in birds, with bats that use higher frequencies specializing in eating smaller prey. Hence, there may be strong selective pressure for divergence in call frequency between species of echolocating bats to minimize interspecific competition for insect prey. Intraspecific competition for different sizes of prey may indeed have been the driving force for speciation in pipistrelles.
334
GARETH JONES
Perhaps the most likely cases of acoustic resource partitioning in bats can be seen in rhinolophid and hipposiderid bats. These bats use CF calls that are either short (hipposiderids) or long (rhinolophids) in duration. The high duty cycles (proportion of time filled with sound) of these bats mean that echoes returning to the bats have potential to contain many glints (small deviations in amplitude or frequency of echoes) from insect wingbeats (Schnitzler, 1987). Bats that use long-duration CF calls, like the greater horseshoe bat Rhinolophus ferrumequinum, select prey in ways predicted by optimal foraging theory. Unprofitable prey, such as small Diptera and ichneumonids, were rejected except when more profitable prey such as moths were scarce (Jones, 1990). Prey size selection in relation to prey availability by R. ferrumequinum showed remarkable parallels with that seen in visually hunting insectivorous birds, such as swallows, Hirundo rustica (Turner, 1982), and spotted flycatchers, Muscicapa striatu (Davies. 1977). Bats that use long CF calls may therefore obtain considerable information about prey characteristics in echoes. The echolocation calls of rhinolophoid bats are relatively simple in structure (Fig. 9) and are dominated by a relatively intense CF portion. Frequency differences in this CF portion are therefore likely to have the most impact on any acoustic resource partitioning in these bats. In the Krau Game Reserve, an area of primary rainforest and high biodiversity in Peninsular Malaysia, at least 12 rhinolophoid species coexist. The CFs of these species range between 40 kHz and 200 kHz. but the distribution of call frequencies is not random: frequencies lying close together are avoided, suggesting that resource partitioning may occur with respect to prey size as a mechanism for reducing interspecific competition for insect food (Heller and Helversen, 1989). Large bat species use lower frequencies than small species (Heller and Helversen, 1989; Barclay and Brigham, 1991; Jones, 1996), so separating the effects of body size and call frequency on prey size selection can be problematic because of the confounding effects of body size and call frequency. Three sympatric Indian hipposiderids use CFs filling the spectrum between 130 kHz and 170 kHz. There is, however, virtually no overlap in CF between species, and in one species, Hipposideros speoris, males emitted higher frequencies than females (Jones et al., 1994; Fig. 9). Sexual differences in CF occur in several rhinolophoid bats (Jones, 1995), so acoustic resource partitioning even between the sexes is possible. If acoustic resource partitioning by echolocation does occur, then I predict that acoustic character displacement will exist (see Grant, 1986, for character displacement in bill sizes of Darwin’s finches). In the case of pipistrelles, for example, I predict that 55-kHz bats will use lower frequencies and 45-kHz bats will use higher frequencies in areas of allopatry than when in
335
ACOUSTIC SIGNALS AND CRYPTIC SPECIES
130.5
135.5
140.5
LII) 145.5
150.5
155.5
160.5
165.5
Resting frequency (IcHz) FIG.9. Frequency distribution of resting frequencies of three sympatric species of hipposiderid bats in southern India. Open bars, Hipposideros speoris females; upwards cross-hatching, H . speoris males; solid bars, H . fulvus; downwards cross-hatching, H . ater. The smaller species use highcr frequency calls, and overlap between species in call frequency is generally avoided. The inset shows sonagrams of typical calls from the three species: s, H. speoris; f, H. fiilvus; and a, H . ater. Modified from Jones, Sripathi, Waters, and Marimuthu (1994), reproduced with permission from The Academy of Sciences of the Czech Republic.
sympatry. In allopatry, the absence of interspecific competition may allow bats of one phonic type to occupy some of the niche space normally occupied by the other phonic type during sympatry.
IV. CRYPTIC SPECIES OF ECHOLOCATING BATS How widespread are cryptic bat species? Are cryptic species of bats relatively more frequent than cryptic species in other animals? The second question is difficult to answer, but nevertheless I predict that many bat species remain undescribed because of the likely importance of echolocation in determining resource use, because selection for acoustic divergence may be stronger than for differences in appearance, and because differences in echolocation calls are difficult for us to study and appreciate.
336
GARETH JONES
Several other cryptic bat species appear to use different echolocation call frequencies. Pye (1972) found two groups of putative Hipposideros commersoni using CFs at 56 kHz and 66 kHz, respectively, inhabiting one cave in Kenya. He interpreted the difference as an intraspecific variation to prevent interference from band spreading due to beating wings when several bats were flying together. The high-frequency bats were significantly smaller, however, and I predict that the two echolocating types are indeed cryptic species. The cryptic species Hipposideros caffer and H . ruber are widespread in Africa, and are difficult to key to species on external characters (Fenton, 1986). Bats keyed as H . cuffer used CFs between 128 and 153 kHz, while those keyed as H . ruber used 121-136 kHz (Jones, Morton, Hughes, and Budden, 1993). It seems likely that more bats currently believed to belong to a single species will be found to comprise several species from analysis of their echolocation calls, and acoustic resource partitioning in bats certainly merits close attention. Cryptic species are known from several bat faunas. In Britain, for example, the long-eared bats, Plecotus auritus and P. auriculus, and the whiskered and Brandt's bats, Myotis mystacinus and M . brandtii, are considered as pairs of sibling species, with the new cryptic species discovered only in the past 40 years (Schober and Grimmberger, 1987). Hence, in the well-studied British bat fauna, 6 of 15 species (40%) form sibling species pairs. Baker (1984) discovered the cryptic species Rhogeesa genowaysi in Central America by karyotyping, and the bat was not distinct in morphology from other species in its genus. In continental Europe Myotis myotis and Myoris blythii are extremely similar in morphology, yet show clear genetic differences and trophic resource partitioning (Arlettaz, 1995, 1996; Arlettaz and Perrin, 1995). Myotis myotis forages mainly in meadows, orchards, and forests without undergrowth where it feeds on ground-dwelling prey, especially carabid beetles. Myotis blythii is associated more with grassland habitats, where it catches large numbers of bushcrickets (Arlettaz, 1995). Arlettaz (1995) believes that allopatric mechanisms, perhaps occurring after glaciations, were responsible for speciation in M . myotis and M . blythii. A new, cryptic specis in the M . mystacinus group was discovered in Greece by analysis of mtDNA (Nemeth and Helversen, 1994). The genus Myotis also contains several cryptic species in North America; the longeared species M . evotis, M . keenii, and M . septentrionalis overlap in distribution in some areas, and their field identification is problematic (Zyll de Jong and Nagorsen, 1994). In British Columbia, Myotis lucifugus and M . yumanensis coexist. These species are extremely similar in morphology, and some individuals appear intermediate in morphology between the two. Nevertheless, no evidence of hybridization was found from electrophoretic
ACOUSTIC SIGNALS AND CRYPTIC SPECIES
337
studies, and resource partitioning by habitat use and diet was discovered (Herd and Fenton, 1983). Obstacle negotiation experiments confirmed that M . yurnanensis was the more maneuverable species, partly as a consequence of its lower wing loading and wingspan (Aldridge, 1986). Myotis nigricans may be a composite of sibling species based on bacular morphology (LaVal, 1973). Cryptic species of bats are clearly widespread, perhaps especially in the genus Myotis, and they occur frequently even in well-documented faunas. What is not known at this stage is whether many species that resemble one another in morphology differ in their echolocation calls in the way that pipistrelles do. V. ACOUSTIC SIGNALS A N D CRYPTIC SPECIES IN NONECHOLOCATING ANIMALS Swarms of cryptic species are known in several insect groups that produce near-field, substrate-borne signals. Although the physical nature of these songs differs from far-field acoustic signals, their songs are difficult for humans to detect, and a discussion of such species seems relevant in a consideration of cryptic species. Henry (1994) has argued that species that use such signals are conducive to speciation via acoustic recognition and assortative mating according to song type. Cryptic species of planthoppers (order Hemiptera) and lacewings (order Neuroptera) are described (review in Henry, 1994). At least five “song morphs” of the lacewing species Chrysoperla carnea exist in North America, with a further five in Europe. The morphs are virtually identical in morphology (Henry, 1985a; 1994), yet each has a distinctive song and responds strongly only to the song of its own morph (Wells and Henry, 1992). Henry (1985b) argued that the simple genetic transmission of song type could be modified easily by mutations, and sexual selection for elaborate male songs by females could drive speciation. It is not clear whether speciation in these lacewings occurs sympatrically or whether it is facilitated by geographical isolation (Henry, 1985a,b). There are numerous examples of cryptic insect species that use far-field acoustic signals. Advances in bioacoustics led to a proliferation in the numbers of species of singing orthopterans that were described. About a quarter of North America’s ensiferan Orthoptera remained undescribed until their songs were studied in detail (Walker, 1964). Further examples of cryptic species of insects that use acoustic signals are reviewed by Ewing (1989) and Bailey (1991). Cryptic species are widespread in anuran taxa where males use advertisement calls. For example, the leopard frog Rana pipiens was found to be a complex of four species after studying variation in its vocalizations (Little-
338
GARETH JONES
john and Oldham, 1968). Many Australian frogs are best separated by vocal differences (review in Littlejohn and Watson, 1985). The frog Leptodactylus mystaceus was believed to be a single species from morphological investigations, but analysis of advertisement calls suggests that a t least two sibling species are present (Heyer, Garcia-Lopez, and Cardoso, 1996). Species number in the anuran genus Xenopus has trebled in the last 20 years (review in Kobel, Loumont, and Tinsley, 1996). Many cryptic species exist in the genus, some of which are allopolyploid species. Differences in male advertisement calls exist among many of the cryptic species (Vigny, 1979), and may be useful clues in understanding the taxonomy and phylogeny of Xenopus. First, the gross structure of advertisement calls is often similar in apparently related species with different ploidy levels. Second, some taxa currently described as subspecies have different advertisement calls and show genetic differences that suggest they may be better considered as cryptic species (Kobel er af., 1996). Most Xenopus species are found in cloudy water conditions, where they locate prey by using a lateral line system (Elepfandt, 1996), and where acoustic advertisement would be favored over visual communication for mate attraction. This acoustic diversity, rather than variability in morphology or color, contributes to the relatively high number of cryptic species in this genus. Allopolyploidy makes Xenopus unusual among vertebrates, and it is possible that new species arise by hybridization when previously allopatric species reestablish contact. This could occur, for example, when a rise in water level allows species previously separated by mountain ranges to establish sympatry (J. Measey, personal communication). Among European birds, sibling species are found in the warbler genera Phylfoscopus and Acrocephalus, in the titmice and treecreepers described in the Introduction, and in nightingales in the genus Luscinia. Mayr (1963) believed that sibling species were relatively uncommon among birds, constituting about 5% of the class. He ascribed the relative paucity of avian sibling species to the importance of vision in the recognition of secondary sexual characters in bird behavior. Mayr believed that sibling species were more widespread in taxa where chemical senses were more highly developed than vision. He made little reference to the possible importance of acoustic differences between sibling species. It is notable that sibling species are frequent in taxa of marine invertebrates. Knowlton (1993) argues that the number of marine species may increase by an order of magnitude if sibling species are considered, and that our current lack of knowledge about sibling species in the sea arises from our reliance on auditory and visual senses, whereas chemical recognition is widespread in marine species.
ACOUSTIC SIGNALS AND CRYPTIC SPECIES
339
VI. SPECIATION IN CRYP~IC SPECIES THAT USE ACOUSTIC SIGNALS
In this section, I review possible speciation mechanisms that may account for the diversity of cryptic species that use acoustic signals. I discuss mechanisms that may apply to echolocating bats, t o animals other than bats that use acoustic signals for mate attraction, and to all species. In any discussion of speciation in Pipistrellus pipistrellus, it is, of course, important to realize that the speciation event separating the two clades was probably ancient, and that differences observed today in, for example, acoustic behavior, may not have been associated with the speciation event. A. ALLOPATRIC SPECIATION Allopatric speciation is widely believed to be the most likely mechanism of speciation in most animal taxa (Mayr, 1963,1977). Several recent studies of cryptic species favor allopatric explanations of speciation (sticklebacks, Schluter and McPhail, 1992; mouse-eared bats, Arlettaz, 1995). Allopatric speciation involves geographical isolation of a population from other populations of the parental species, and acquisition of characters that promote or ensure reproductive isolation once sympatry is reestablished. A plausible scenario for pipistrelle speciation may be as follows. Imagine that bats in the parent population echolocate at 45 kHz. A small population becomes isolated by, for example, mountain barriers. Perhaps a glaciation event pushed this isolated population into a refuge that ensured its isolation. Bats in the isolated population changed echolocation call frequency to 55 kHz, social call structure altered, but morphological conformity with the parent population was maintained. Perhaps echolocation call frequency changed to exploit a new insect resource encountered by the isolates. As conditions became warmer, barriers between the populations broke down; however, the changes that occurred during isolation caused the previous parent and isolated populations to remain reproductively isolated. Both nascent species then spread over a wide geographic range and avoided competition because they used different call frequencies. This scenario may be testable if the date of divergence of the species could be ascertained by application of a molecular clock. Conditions at the time of divergence could be explored, to determine if geographic separation of populations would have been facilitated by climatic conditions at that time.
B. SYMPATRIC SPECIATION BY DISRUP~IVE SELECTION As described earlier, differences in call frequencies used in echolocation may allow resource partitioning in bats. Call frequency can therefore be
340
GARETH JONES
viewed as a “resource acquisition character” that may be favored to diverge if a new niche becomes available, hence allowing some individuals to exploit the new niche and minimize competition with other animals. Models involving disruptive selection on resource acquisition characters have been described as “competitive speciation” models (e.g., Maynard Smith 1966; Rosenzweig, 1978; Pimm, 1979; Wilson and Turelli, 1986; Wilson, 1989). Disruptive selection favoring extreme resource acquisition characters (e.g., beak size, echolocation call frequency) may promote sympatric speciation if the distribution of resources is not normal (Seger, 1985). What may initially begin as a polymorphism in resource use (Sktilason and Smith, 1995) may result in sympatric speciation if reduced gene flow occurs between sympatric morphs (Bush, 1994). In pipistrelles, it is easy to understand how a shift in call frequency by some bats may reduce competition with conspecifics if a new niche becomes available. However, one difficulty sometimes raised about competitive speciation models is how the bimodality in the resource acquisition character is maintained. If random breeding occurs, recombination will result in intermediate forms remaining in the population (Felsenstein, 1981). This difficulty has been circumvented by several authors. Pimm (1979) and Wilson and Turelli (1986) modeled scenarios where new niches were invaded by poorly adapted heterozygotes. The heterozygote is later eliminated by homozygotes for the new character, with the result that unfit heterozygotes are replaced. Disruptive selection may result in character divergence if assortative mating occurs between the divergent phenotypes. The evolution of character divergence by disruptive selection and assortative mating of divergent phenotypes might be unlikely (Felsenstein, 1981). A more plausible scenario is that disruptive selection favors divergence in mating traits (such as timing of breeding), and resource acquisition characters are pleiotropic with the mating trait (Rice and Hostert, 1993). If new morphs that arise live in novel habitats away from the parental population, assortative mating among morphs may be promoted via habitat segregation (e.g., Johnson, Hoppensteadt, Smith, and Bush, 1996). Competitive speciation models may be realistic scenarios for the evolution of echolocation call divergence in pipistrelle bats if some of the above assumptions are met. C. SYMPATRIC SPECIATION BY SEXUAL SELECTION Advertisement calls cannot be viewed as resource acquisition characters, and so are unlikely to evolve by competitive speciation. Natural selection imposes tight constraints on the design of echolocation signals in bats: bats that forage in similar habitats show convergence in the design of
ACOUSTIC SIGNALS AND CRYPTIC SPECIES
341
echolocation signals, even though the species may be phylogenetically distant (Norberg and Rayner, 1987; Neuweiler, 1989; Fenton, 1990). Although natural selection will shape the design of acoustic signals in other animals, there is likely to be more scope for signal flexibility than in bats, where echolocation call design is likely to be tightly shaped for orientation and especially for the detection of insect prey. Songflight and social calls of bats should show less stereotypy than echolocation calls, and this has indeed been shown by Fenton (1994), who argued that echolocation calls were less likely to be used as dishonest signals than were social calls. The ways in which habitat influences the design of relatively low frequency acoustic signals have been studied for temperate (Marten and Marler, 1977) and tropical (Marten, Quine, and Marler, 1977) habitats. How habitat features influence the design of bird song was described by Morton (1975). Although features such as vegetation structure are likely to select for the best frequencies to, say, maximize the range of sound transmission, there will still be scope for flexibility in signal design within a particular frequency window. For example, repetition rate of pulses, pulse duration, and components that describe elaboration of bird song may be little affected by habitat structure; the constraints may be less tight than those shaping the design of bat echolocation pulses. Consequently I predict that scope for variation in the vocalizations of birds, anurans, and insects should be greater than that in bat echolocation. Bat echolocation calls will be shaped strongly by natural selection, making the signals designed optimally for the detection of, for example, different types of insects. The songs of insects, anurans, and birds will probably be shaped more by sexual selection, with female choice for particular signal designs being a potentially strong force shaping divergence in acoustic signals of these groups. Consequently the potential role of sexual selection in the evolution of animals that use acoustic signals in mate advertisement deserves scrutiny. Can divergence in advertisement calls occur within a population, and hence promote speciation? Several models have been proposed to show how sympatric speciation may occur as the consequence of female choice for elaborate male traits. Most models involve Fisherian runaway selection, with correlated selection between the degrees of male trait elaboration and female response (e.g., Lande, 1981; West-Eberhard, 1983). Lande and Kirkpatrick (1988) proposed a mechanism of sympatric speciation that could occur if female choice of mates was based on ecologically important characters and if more than one ecological niche was available for the species. Lande and Kirkpatrick’s model relied on the trait that was being sexually selected also being important for niche differentiation by natural selection, and considered body size as being a trait that would be important both in female choice and in niche differentiation. Wu (1985) developed
342
GARETH JONES
a model of sympatric speciation by runaway sexual selection, and concluded that “one would expect to find occasionally in natural populations the existence of two or more reproductive units. These units may have diverged very little in their morphological and ecological characters” (p. 79). Divergence in female preference and male sexual ornaments has been modeled under non-Fisherian conditions by Schluter and Price (1993), when male traits are viewed as handicaps. Although many models of speciation by sexual selection usually involve disruptive natural selection as being important, a recent model (Turner and Burrows, 1995) claimed that sympatric speciation may occur in the absence of disruptive natural selection. If sexual selection has been important in speciation, then secondary sexual traits should show the most remarkable differences in taxonomic groups containing many species (Anderson, 1994). However, female preferences are predicted to vary less than male traits, and the most plausible examples of sexual selection driving speciation are when closely related species show divergent mate preferences but little post mating isolation (Schluter and Price, 1993). Some authors have argued that acoustic signals will diverge @er populations that underwent allopatric speciation meet again in sympatry. Signal divergence may be the consequence of reinforcement (the unlikely situation where gene flow between populations still occurs), or more probably reproductive character displacement (no gene flow) when, rather than promoting speciation, acoustic divergence occurs rapidly after speciation (Butlin, 1987). Divergence between song types before reproductive isolation may require assortative mating between females that prefer a particular song type, and males who sing it. Currently, however, there is no convincing evidence for a genetic coupling between signal production and reception in animals. It remains possible that coevolution of characters affecting signal production and reception has occurred, rather than signal and reception characters having a common genetical basis (Butlin and Ritchie, 1989; Boake, 1991). It remains difficult to reject the view that female preferences evolved after speciation. For species in which sexual selection by female choice on vocalizations is important (insects, anurans, birds), song probably evolves in a less conservative fashion than morphology, and may be subject to faster evolutionary change. Small mutations in song structure may impose less of a cost to the bearer than would small changes in appearance. The groups of passerine birds containing the greatest number of species are those with the most advanced capacity for sound production (Raikow, 1986), suggesting that female choice for song may have contributed to speciation in these taxa. Mistakes during vocal learning may also have contributed to species diversity in passerines, though its importance has been disputed (Fitzpatrick, 1988; Vermeij, 1988; Baptista and Trail, 1992). In a similar way, those taxa
ACOUSTIC SIGNALS AND CRYPTIC SPECIES
343
of birds containing the greatest number of species also contain the highest proportion of sexually dichromatic species, a trend expected if sexual selection by female choice increased speciation rate (Barraclough, Harvey, and Nee, 1995). The taxa of frogs that contain large numbers of species appear to be those that detect the widest range of frequencies (Ryan, 1986). Rapid change may occur in song when small populations become isolated, and divergence may occur rapidly in sympatric populations (Lande, 1981; Wu, 1985). Natural selection may constrain an animal’s morphological design within tight limits, but sexual selection may elaborate song rapidly by runaway selection. Plumage or coloration may also be constrained by natural selection, especially if an animal’s main predators hunt visually. Divergence in song would then occur faster than divergence in appearance, and hence the large genetic differences seen in some cryptic species are not surprising if appearance and morphology are constrained tightly by natural selection. A potential speciation mechanism that may apply to pipistrelles links sexual selection by females with resource acquisition in males. Females choose male characters that affect resource exploitation, and if the resource acquisition characters are inherited, sexual selection may drive speciation (see also the model of Lande and Kirkpatrick, 1988, described earlier). The mechanism may not be unique to bats, and the way in which female choice may drive resource acquisition in bats is slightly more complicated than in other animals. In bats, one feature of resource acquisition, echolocation call frequency, may be correlated with the character driven by female choice, songflight call frequency. In other animals, female choice may influence the evolution of the resource acquisition character directly (Lande and Kirkpatrick, 1988), and female choice for echolocation call frequency may have this effect in bats. Hence, female choice for songflight call frequency might drive divergence in echolocation call frequency. The entire population of bats may not be selected to shift to lower echolocation call frequencies because, although some bats would benefit from moving into a new foraging niche (exploitation of larger prey associated with lower call frequencies), others would benefit from reduced interspecific competition by remaining with a higher call frequency and being better able to exploit smaller prey. D. EVALUATION OF THE SPECIATION MODELS Allopatric speciation is widely accepted, and should not be dismissed in pipistrelles. However, one problem with allopatric hypotheses concerns finding a reason why call frequency should diverge in allopatry, with the species later becoming largely sympatric. Competitive events in sympatry
344
GARETH JONES
may be a more plausible driving force for the divergence in call frequency. Although the adaptive advantages of bimodality in call frequency are easy to visualize, how bimodality would be maintained in an interbreeding population is problematic. There is at present no evidence for assortative mating according to call frequency within phonic types of pipistrelle (Park et al., 1996). Thus, disruptive selection models may involve mechanisms other than assortative mating to maintain character divergence. Sympatric speciation by sexual selection in pipistrelles has little empirical support. Within phonic types of pipistrelle there is no correlation between songflight call frequency and echolocation call frequency (Barlow and Jones, in press, b), and such a relation might be expected in sexual selection models that involve genetic correlations between echolocation call and songflight call frequencies. Speciation by sexual selection would be restricted to bat species that choose mates on the basis of acoustic cues from songflights:pipistrelles are some of the few bats for which songflight displays are known. Pipistrelle bats may be animals where mate choice depends on resource choice (or vice versa), and such species are prime candidates for possible sympatric speciation (Bush, 1994). An alternative explanation for the differences in call frequency and songflight call frequency between phonic types is that disruptive natural selection favors divergence in echolocation call frequency, and songflight frequencies change as a correlated response. This speciation mechanism could occur without geographical isolation, but would not involve sexual selection by female choice. A further alternative is that divergence in echolocation and songflight call frequencies evolved independently, with divergence in echolocation calls being favored by disruptive natural selection, and songflight calls evolving largely under the influence of sexual selection. Thus, although speciation by sexual selection seems unlikely in pipistrelles, it remains possible for animal species that use acoustic signals for mate attraction, such as birds, anurans, and orthopterans. As in the competitive speciation models, intermediate song morphs must be selected against if extreme song types are to be favored. Genetic coupling of signal type and reception may promote song divergence, but, as described earlier, there is no evidence for coupling to date. It is also important to realise that sexual selection may promote rapid divergence in advertisement calls in small geographically isolated populations. In conclusion, it is likely that sexual selection facilitates acoustic divergence in cryptic taxa after divergence, but there is no convincing evidence to date that it promotes divergence before populations are isolated. VII. CRYFTIC SPECIES, GENETIC DIVERGENCE, AND HIDDEN BIODIVERSITY
Many cryptic species do in fact show small morphological differences, and such subtle differences may influence niche use (e.g., Marchetti, Price,
ACOUSTIC SIGNALS AND CRYPTIC SPECIES
345
and Richman, 1995). It would be of interest to determine whether cryptic species that are morphologically indistinguishable (e.g., many Chrysoperla lacewings) show any differences in habitat use (especially if habitat divergence promotes assortative mating between morphs), and whether the relative genetic divergence in such species is smaller than that found in cryptic species that show subtle morphological differences (e.g., Phylloscopus warblers). Genetic markers allow the relative evolutionary rates of morphological conservatism and acoustic divergence to be quantified in ways that have hitherto been impossible, and should shed more light on the evolutionary reasons why cryptic species proliferate in certain taxa. Recent advances in molecular biology may allow insights into speciation in cryptic species, and whether cryptic species evolve at similar rates to species that look different in appearance to our eyes. Sequence divergence in mtDNA could be compared between cryptic species and other closely related species that differ in appearance to infer whether cryptic species are more similar genetically, and hence at an earlier stage in their phylogenetic history. If genetic divergence can serve as a molecular clock, then dates for speciation events may be inferred, as may the role of geographical events in driving speciation. As already mentioned, the cytochrome-b data for pipistrelles show a large sequence divergence between phonic types, of a magnitude typical of species differences in taxa that are morphologically more obviously different (Fig. 7). Avise and Zink (1988) studied genetic differences within four pairs of avian sibling species where the merits of specific status were unclear. For three pairs (rails Rallus elegans and R. longirostris;titmice Parus bicolor bicolor and P.b. artricristatus, and grackles Quiscalus major and Q. mexicana) results were equivocal, with mtDNA distances typical of extremely closely related species, but overlapping with maximum values reported for some conspecific pairs. However, distance for the dowitchers Limnodromus scolopaceus and L. griseus was large, and close to maximum values reported for avian congeners (Avise and Zink, 1988). Genetic differences between cryptic species of Phylloscopus warblers have been calculated by Helbig, Seibold, Martens, and Wink (1995). There appear to be parallels with pipistrelle bats in that study. The western Phylloscopus b. bonelli and eastern P. b. orientalis Bonelli’s warblers were conventionally classified as subspecies. The warblers are allopatric, and, although very similar in morphology and apparently in plumage, have very different songs. Genetic differences between the warblers were large, with an 8.3-8.6 sequence divergence in a sequence of cytochrome-b of mtDNA. Thus, the warblers were genetically more different than were the shoveler (Anus clypeata) and teal (A. crecca) ducks, that are obviously different in appearance to humans. Clearly, the two subspecies of Bonelli’s warbler merit specific status because of vocal differences (and species-specific re-
346
GARETH JONES
sponses to song playbacks; Helb, Bergmann, and Martens, 1982) and because of the large genetic differences. It is important to emphasize that the small morphological differences in cryptic taxa such as Phylloscopus may have ecological importance. Phylloscopus b. orientalis has longer and more pointed wings, for example, perhaps as an adaptation to a longer migration distance (Helbig et al., 1995). Subtle ecological differences occur in Phyllosc o p s warblers as a consequence of small morphological differences (Richman and Price, 1992; Marchetti er al., 1995). Species with longer wings hover less and migrate further, while species with more pointed wings migrate further and are more arboreal (Marchetti et al., 1995). Before the revolution in molecular genetics, Mayr (1977) pointed out correctly that sibling species are not necessarily species that have formed in the recent past. As the previous discussion outlines, cryptic species can be as distinct genetically as other species, and may often be cryptic only to human vision. The average rate of sequence divergence in the entire mtDNA genome is about 2% per million years in birds and mammals (Wilson et al., 1985). If the cytochrome-b mutation rate approximates that of the entire mtDNA genome, then the two Bonelli’s warblers may have separated from a common ancestor over 4 million years ago (Helbig et al., 1995). The divergence of the cryptic pipistrelle species may have occurred at a similar time. Even if these extrapolations are incorrect, the point still remains that cryptic species can be as genetically distinct as are more morphologically divergent congeners (see Henry, 1985a,for a discussion of possible exceptions). There is potential for considerable hidden genetic diversity in cryptic species, and some of that may be unrealized as species become extinct through habitat destruction (Wilson, 1992). At least four new species of Phylloscopus were described in 1992-1993 (Alstrom, Olsson, and Colston, 1992; Olsson, Alstrom, and Colston, 1993). The chiffchaff, P. coflybira, is probably at least three cryptic species in Europe, each with a distinct male song (Helbig, Martens, Seibold, Hening, Schottler and Wink, 1996). Our emphasis on morphological rather than behavioral characters, and our sensory biases in determining the extent of biodiversity may mean that many cryptic species remain undiscovered, or may even evade discovery before extinction. Given the potentially large genetic divergence between some cryptic species, the amount of genetic diversity as yet undescribed on earth may be substantially greater than generally assumed. VIII. SUMMARY
Animal taxonomy has often relied on separating species by their appearance or by morphological characters. Sometimes behavioral features such
ACOUSTIC SIGNALS A N D CRYITIC SPECIES
347
as differences in vocalizations have been overlooked as indicators of species differences. Moreover, humans are limited in sensory abilities. Although we primarily use vision for sensing the world, even our visual capacities are restricted when compared with those of other animals. Such restrictions are important in our appreciation of biodiversity. Our appreciation of diversity in nonvisual signals in animals is limited. Even if animals resemble one another in appearance, they may differ in, for example, acoustic or olfactory signals, and such differences may be important isolating mechanisms between animal species that appear similar to our eyes. Moreover, even our visual perception of the world may fail to detect differences in, for example, animal coloration that may distinguish different species through the eyes of the animals concerned. Hence, sensory limitations in humans probably limit our description of species diversity to species that we can most easily distinguish. Species that appear similar to us are termed cryptic species, yet many cryptic species use quite different acoustic signals, and the diversity of such animals has only begun to be appreciated since advances in bioacoustical analysis have made their study possible. In many animals, factors such as predation pressure from visual predators may select for conservation in appearance during speciation, while acoustic signals may be free to diversify more rapidly. The taxa of frogs and birds containing the largest number of species are sometimes those with the most complex auditory morphology or song structure. In singing insects, anurans, and birds, sexual selection may in theory drive acoustic divergence rapidly. Speciation by such divergence should involve assortative mating within morphs, however, and whether sympatric speciation can result from female choice remains controversial. It is more likely that female choice would accelerate acoustic divergence in small, isolated populations. Cryptic species are widespread in bats, and their diversity may be underestimated because of their use of signals above the frequencies of human hearing. Acoustic signals may be important mechanisms for resource partitioning in echolocating bats because, at least in theory, call frequency will determine echo strength from prey of different sizes. Bat acoustic signals are probably more severely constrained in their flexibility by natural selection than are the songs of insects, anurans, and birds, but sympatric speciation may still be possible if the frequency of calls used in mate attraction is coupled to echolocation call frequency, or if differences in echolocation call frequencies are favored by disruptive selection. Assortative mating within morphs may again be a prerequisite for sympatric speciation. Evidence from echolocation calls, social calls, mating associations, morphology, and genetics shows that Britain’s commonest bat, the pipistrelle, is actually two cryptic species, and many more cryptic bat species probably await
348
GARETH JONES
discovery. Techniques from molecular genetics show great promise in revealing the hidden genetic diversity shown by cryptic species. Studies on warblers and pipistrelle bats suggest that cryptic species are often not recently formed species, and that visual appearance may have been conserved during their evolution, while acoustic divergence has been substantial.
Acknowledgments Financial support was received from The Royal Society and NERC. I thank Kate Barlow for permission to quote some of her unpublished findings on the pipistrelle studies, and for her comments on an earlier draft. Comments from Roger Butlin, Peter Slater, and Charles T. Snowdon improved an earlier draft.
References Aldridge, H. D. J. N. (1986). Manoeuverability and ecological segregation in the little brown (Myoris lucifugus) and Yuma ( M . yumanensis) bats (Chiroptera: Vespertilionidae). Can. J. ZOO^. 64,1878-1882. Alstrom, P., Olsson, U., and Colston, P. R. (1992). A new species of Phylloscopus warbler from central China. Ibis 134, 329-334. Anderson, M. B. (1994). “Sexual Selection.” Princeton University Press, Princeton, New Jersey. Arlettaz, R. (1995). “Ecology of the Sibling Mouse-Eared Bats (Myoris myoris and Myotis blyrhii): Zoogeography, Niche, Competition, and Foraging.” Horus Publishers, Martigny, Switzerland. Arlettaz, R. (1996). Feeding behaviour and foraging strategy of free-living mouse-eared bats, Myoris myotis and Myoris blyrhii. Anim. Behav. 51, 1-11. Arlettaz, R., and Perrin, N. (1995). The trophic niches of sympatric sibling Myoris myoris and M . blythii: do mouse-eared bats select prey? Symp. Zool. SOC. Lond. 67,361-376. Avise, J. C. (1994). “Molecular Markers, Natural History and Evolution.” Chapman and Hall, New York. Avise, J. C., and Zink, R. M. (1988). Molecular genetic divergence between avian sibling species: King and clapper rails, long-billed and short-billed dowitchers, boat-tailed and great-crested grackles, and tufted and black-crested titmice. Auk 105, 516-528. Baagoe, H. J. (1973). Taxonomy of two sibling species of bats in Scandinavia Myotis mysracinus and Myotis brandtii (Chiroptera). Vidensk. Meddr dansk narurh. Foren. 136, 191-216. Bailey, W. J. (1991). “Acoustic Behaviour of Insects: An Evolutionary Perspective.” Chapman and Hall, London. Baker, R. J. (1984). A sympatric cryptic species of mammal: A new species of Rhogeesa (Chiroptera: Vespertilionidae). J. Mammal. 33, 178-183. Baker, R. J., Taddei. V. A., Hudgeons, J. L., and Van Den Bussche, R. A. (1994). Systematic relationships within Chiroderma (Chiroptera: Phyllostomidae) based on cytochrome b sequence variation. J . Mammal. 75, 321-327. Baptista, L. F., and Trail, P. W. (1992). The role of song in the evolution of passerine diversity. Sysr. Biol. 41, 242-247.
ACOUSTIC SIGNALS AND CRYPTIC SPECIES
349
Barclay, R. M. R., and Brigham, R. M. (1991). Prey detection, dietary niche breadth and body size in bats: Why are aerial insectivorous bats so small? Am. Nat. 137,693-703. Barlow, K. E., and Jones, G . (in press, a). Pipistrellus nathusii (Chiroptera: Vespertilionidae) in Britain in the mating season. J . Zool. Lond. Barlow, K. E., and Jones, G. (in press, b). Differences in songflight calls and social calls between two phonic types of the vespertilionid bat Pipistrellus pipistrellus. J. Zool. Lond. Barlow, K. E., and Jones, G. (in press, c). Function of Pipistrellus pipbtrellus (Chiroptera) social calls: Field data and a playback experiment. Anim. Behav. Barraclough. T. G..Harvey, P. H., and Nee, S. (1995). Sexual selection and taxonomic diversity in passerine birds. Proc. R. Soc. Lond. 2598,211-215. Barratt, E. M., Bruford, M. W., Burland, T. M., Jones, G., Racey, P. A., and Wayne, R. K. (1995). Characterization of mitochondria1 DNA variability within the microchiropteran genus Pipistrellus: Approaches and applications. Symp. Zool. Soc. Lond. 67,377-386. Bennett, A. T. D., and Cuthill, I. C. (1994). Ultraviolet vision in birds: what is its function? Vision Res. 34, 1471-1478. Bennett, A. T. D., Cuthill, I. C., and Norris, K. J. (1994). Sexual selection and the mismeasure of color. Am. Nar. 144,848-860. Boake, C. R. B. (1991). Coevolution of senders and receivers of sexual signals: Genetic coupling and genetic correlations. Trends Ecol. Evol. 6,225-227. Bornemissza, G . F. (1966). Observations on the hunting and mating behaviour of two species of scorpionflies (Bittacidae: Mecoptera). Aust. J. Zool. 14, 371-382. Bruna, E. M., Fisher, R. N., and Case, T. J. (1996). Morphological and genetic evolution appear decoupled in Pacific skinks (Squamata: Scinidae: Emoia). Proc. R. SOC. Lond. 2638, 681-688. Bush, G. L. (1994). Sympatric speciation in animals: New wine in old bottles. Trends Ecol. Evol. 9,285-288. Butlin, R. K. (1987). Species, speciation and reinforcement. Am. Nat. 130,461-464. Butlin, R. K., and Ritchie, M. G. (1989). Genetic coupling in mate recognition systems: What is the evidence? Biol. J. Linn. Soc. 37, 237-246. Catchpole, C. K., and Slater, P. J. B. (1995). “Bird Song: Biological Themes and Variations.” Cambridge University Press, Cambridge. Davies, N. B. (1977). Prey selection and the search strategy of the spotted flycatcher (Muscicapa striafa):A field study on optimal foraging. Anim. Behav. 25, 1016-1033. Dawkins, R., and Krebs, J. R. (1978). Animal signals: Information or manipulation? In “Behavioural Ecology” (J. R. Krebs and N. B. Davies, eds.), pp. 282-309. Blackwell Scientific Press, Oxford. Elepfandt, A. (1996). Sensory perception and the lateral line system in the clawed frog, Xenopus. Symp. Zool. Soc. Lond. 68,97-120. Endler, J . A. (1992). Signals, signal conditions, and the direction of evolution. Am. Nat. 139, S125-SI53. Ewing, A. W. (1989). “Arthropod Bioacoustics: Neurobiology and Behaviour.” Edinburgh University Press, Edinburgh. Felsenstein, J. (1981). Skepticism towards Santa Rosalia, or why are there so few kinds of animals? Evolurion 35, 124-138. Fenton, M. B. (1985). “Communication in the Chiroptera.” Indiana University Press, Bloomington. Fenton, M. B. (1986). Hipposideros caffer (Chiroptera: Hipposideridae) in Zimbabwe: morphology and echolocation calls. J. Zool. Lond. 210, 347-353. Fenton, M. B. (1990). The foraging behaviour and ecology of animal-eating bats. Can. J. Zool. 68.41 1-422.
350
GARETH JONES
Fenton, M. B. (1994). Assessing signal variability and reliability: ‘To thine ownself be true.’ Anim. Behuv. 41,757-764. Fitzpatrick, J. W. (1988). Why so many passerine birds? A response to Raikow. Syst. Zool. 37, 72-77. Gerell, R., and Lundberg, K. (1985). Social organisation in the bat Pipistrellus pipistrellus. Behav. Ecol. Sociobiol. 16, 177-184. Gerell-Lundberg, K., and Gerell, R. (1994). The mating behaviour of the pipistrelle and the Nathusius’ pipistrelle (Chiroptera)-A comparison. Folia Zool. 43, 315-324. Grant, P. R. (1986). “Ecology and Evolution of Darwin’s Finches.” Princeton University Press, Princeton, New Jersey. Griffin, D. R. (1958). “Listening in the Dark.” Cornell University Press, Ithaca. Griffin, D. R., Webster, F. A., and Michael, C. R. (1960). The echolocation of flying insects by bats. Anim. Behav. 8, 141-154. Helb, H.-W., Bergmann, H.-H., and Martens, J. (1982). Acoustic differences between populations of western and eastern Bonelli’s warbler (fhylloscopus bonelli, Sylvidae). Experienria 38,356-357. Helbig, A. J., Martens, J.,Seibold, I., Henning, F., Schottler, B., and Wink, M. (1996). Phylogeny and species limits in the Palearctic chiffchaff (fhylloscopus collybira) complex: Mitochondrial genetic differentiation and bioacoustic evidence. Ibis. 138, 650-660. Helbig, A. J., Seibold, I., Martens, J., and Wink, M. (1995). Genetic differentiation and phylogenetic relationships of Bonelli’s warbler fhylloscopus bonelli and green warbler f. nitidus. J . Avian Biol. 26, 139-153. Heller, K.-G., and Helversen, 0.von (1989). Resource partitioning of sonar frequency bands in rhinolophoid bats. Oecologia 80, 178-186. Henry, C. S. (1Y85a). Sibling species, call differences, and speciation in green lacewings (Neuroptera: Chrysopidae: Chrysoperla). Evolution 39,965-984. Henry, C. S. (198%). The proliferation of cryptic species in Chrysoperla green lacewings through song divergence. Fla. Entomol. 68, 18-38. Henry, C. S. (1994). Singing and cryptic speciation in insects. Trends Ecol. Evol. 9, 388-392. Herd, R. M., and Fenton, M. B. (1983). An electrophoretic, morphological. and ecological investigation of a putative hybrid zone between Myotis lucifugus and Myotis yumanensis (Chiroptera: Vespertilionidae). Can. J. Zool. 61,2029-2050. Heyer, W. R., Garcia-Lopez, J. M., and Cardoso, A. J. (1996). Advertisement call variation in the Leptodactylus mystaceus species complex (Amphibia: Leptodactylidae) with a description of a new sibling species. Amphibiu-Reptilia 17, 7-31. Hill, J. E., and Harrison, D. L. (1987). The baculum in the Vespertilioninae (Chiroptera: Vespertilionidae) with a systematic review, a synopsis of fipistrelhts and Eptesicus, and the descriptions of a new genus and subgenus. Bull. Br. Mus. Nut. Hist. (Zool.)52,225-305. Johnson, P. A., Hoppensteadt, F. C., Smith, J. J., and Bush, G. L. (1996). Conditions for sympatric speciation: A diploid model incorporating habitat fidelity and non-habitat assortative mating. Evol. Ecol. 10, 187-205. Jones, G. (1990). Prey selection by the greater horseshoe bat (Rhinolophus ferrumequintim): Optimal foraging by echolocation’?J. Anim. Ecol. 59, 587-602. Jones, G. (1994). Scaling of wingbeat and echolocation pulse emission rates in bats: Why are aerial insectivorous bats so small? Funct. Ecol. 8, 450-457. Jones, G. (1995). Variation in bat echolocation: Implicztions for resource partitioning and communication. Le Rhinofophe 11, 53-59. Jones, G. (1996). Does echolocation constrain the evolution of body size in bats? Symp. Zool. SOC.Lond. 69.111-128.
ACOUSTIC SIGNALS AND CRYPTIC SPECIES
351
Jones, G . , and Kokurewicz, T. (1994). Sex and age variation in echolocation calls and flight morphology of Daubenton’s bats Myotis daubentonii. Mammalia 58,41-50. Jones, G.,Morton, M., Hughes, P. M., and Budden, R. M. (1993). Echolocation, flight morphology and foraging strategies of some West African bats. J. Zool. Lond. 230, 385-400. Jones, G.,and Parijs, S. M. van (1993). Bimodal echolocation in pipistrelle bats: Are cryptic species present? Proc. R. SOC.Lond. 251B, 119-125. Jones, G., and Ransome, R. D. (1993). Echolocation calls of bats are influenced by maternal effects and change over a lifetime. froc. R. SOC.Lond. 252B, 125-128. Jones, G., Sripathi, K., Waters, D. A., and Marimuthu, G. (1994). Individual variation in the echolocation calls of three sympatric Indian hipposiderid bats, and an experimental attempt to jam bat echolocation. Folia Zool. 43, 347-362. Kalko, E. K. V. (1995). Insect pursuit, prey capture and echolocation in pipistrelle bats (Microchiroptera). Anim. Behav. 50, 861 -880. Kalko. E. K. V., and Schnitzler, H.-U. (1993). Plasticity in echolocation signals of European pipistrelle bats in search flight: Implications for habitat use and prey detection. Behav. Ecol. Sociobiol. 33,415-428. Kapteyn, K. (1993). lntraspecific variation in echolocation of vespertilionid bats and its implications for identification. In “Proceedings of the First European Bat Detector Workshop” (K. Kapteyn, ed.), pp. 45-57. Netherlands Bat Research Foundation, Amsterdam. Knowlton, N. (1993). Sibling species in the sea. Annu. Rev. Ecol. Syst. 24, 189-216. Kobel. H. R.. Loumont, C., and Tinsley, R. C. (1996). The extant species. Symp. Zool. SOC. Lond. 68, 10-33. Landc, R. (1981). Models of speciation by sexual selection on polygenic traits. froc. Natl. Acad. Sci. U.S.A. 78, 3721-3725. Lande, R., and Kirkpatrick, M. (1988). Ecological speciation by sexual selection. J . Theor. Biol. 133, 85-98. Lanza, B. (1960). Su du specie criptiche di orrechione: flecotus auritus (L.) e f. w a d i Thomas (Mammalia: Chiroptera). Monitore Zoologico Italian0 68, 7-23. LaVal, R. K. (1973). A revision of the Neotropical bats of the genus Myotis. Science Bulletin Natitrul History Museum Los Angeles County 15, 1-54. Lawrence, B. D., and Simmons, J. A. (1982). Measurements of atmospheric attenuation at ultrasonic frequencies and the significance for echolocation by bats. J. Acoust. SOC. America 71,585-590. Littlejohn, M. J., and Oldham, R. S. (1968). Rana pipiens complex: Mating call structure and taxonomy. Science 162, 1003-1005. Littlejohn, M. J., and Watson, G. F. (1985). Hybrid zones and homogamy in Australian frogs. Annu. Rev. Ecol. Syst. 16, 85-112. Lundberg, K.,and Gerell, R. (1986).Territorial advertisement and mate attraction in the bat Pipistrellus pipistrellus. Ethology 71, 11 5-1 24. Marchetti, K., Price, T., and Richman, A. (1995). Correlates of wing morphology with foraging behaviour and migration distances in the genus fhylloscopus. J. Avian Biol. 26,177-181. Marten, K., and Marler, P. (1977). Sound transmission and its significance for animal vocalization. 1. Temperate habitats. Behav. Ecol. Sociobiol. 2, 271-290. Marten, K.. Quine, D., and Marler, P. (1977). Sound transmission and its significance for animal vocalization. 11. Tropical habitats. Behav. Ecol. Sociobiol. 2, 291-302. Martin, G.(1990). “Birds by Night.” Poyser. London. Maynard Smith, J. (1966). Sympatric speciation. Am. Nur. 100,637-650. Mayr, E. (1963). “Animal Species and Evolution.” Harvard University Press, Cambridge, Massachuset ts.
352
GARETH JONES
Mayr, E. (1977). “Populations, Species, and Evolution.” 6th printing. Harvard University Press, Cambridge, Massachusetts. Miller. L. A., and Degn, H. J. (1981). The acoustic behaviour of four species of vespertilionid bats studied in the field. J. Comp. Physiol. 142A, 67-74. Morton, E. S. (1975). Ecological sources of selection on avian sounds. Am. Nat. 108, 17-34. Nemeth, A., and Helversen, 0. von (1994). The phylogeny of the Myotis mysracinus group-a molecular approach. Bar Research News 35,37. Neuweiler, G. (1989). Foraging ecology and audition in echolocating bats. Trends Ecol. Evol. 4, 160-166. Neuweiler, G. (1990). Auditory adaptations for prey capture in echolocating bats. Physiol. Rev. 70,615-641. Norberg, U. M., and Rayner, J. M. V. (1987). Ecological morphology and flight in bats (Mammalia; Chiroptera): Wing adaptations, flight performance, foraging strategy and echolocation. Phil. Trans. R. SOC.Lond. 3168,335-427. Olsson, U., Alstrdm, P., and Colston, P. R. (1993). A new species of Phylloscopics warbler from Hainan Island, China. Ibis 135,3-7. Park, K. J., Altringham, J. D., and Jones, G. (1996). Assortative roosting in the two phonic types of Pipistrellus pipistrellus during the mating season. Proc. R. SOC.Lond. B. Patterson, B. D., and Thaeler, C. S. Jr. (1982). The mammalian baculum: Hypotheses on the nature of bacular variability. J . Mammal. 63, 1-15. Pimm, S. L. (1979). Sympatric speciation: A simulation model. Biol. J. Linn. SOC.11,131-139. Pollak, G. D., and Casseday. J. H. (1989). “The Neural Basis of Echolocation in Bats.” Springer-Verlag, Berlin. Pye, J. D. (1972). Bimodal distribution of constant frequencies in some hipposiderid bats (Mammalia: Hipposideridae). J . Zool. Lond. 166,323-335. Pye, J. D. (1993). Is fidelity futile? The “true” signal is illusory, especially with ultrasound. Bioacoustics 4, 271-286. Racey, P. A., and Swift, S. M. (1985). Feeding ecology of Pipistrellus pipisrrellus (Schreber) (Chiroptera: Vespertilionidae) during pregnancy and lactation. 1. Foraging behaviour. J. Anim. Ecol. 54,205-215. Raikow, R. J. (1986). Why are there so many kinds of passerine birds? Syst. Zool. 35,255-259. Rice, W. R., and Hostert, E. E. (1993). Laboratory experiments on speciation: What have we learned in 40 years? Evolution 47, 1637-1653. Richman, A. D., and Price, T. (1992). Evolution of ecological differences in the Old World leaf warblers. Nature (London) 355, 817-821. Rohlf, F. J. (1990). Morphometrics. Annu. Rev. Ecol. Syst. 21, 299-316. Ri5mer. H. (1993). Environmental and biological constraints for the evolution of long-range signalling and hearing in acoustic insects. Phil. Trans. R. SOC.Lond. MOB, 179-185. Rosenzweig, M. L. (1978). Competitive speciation. Biol. J. Linn. SOC.10, 275-289. Roverud, R. C., and Grinnell, A. D. (1985). Echolocation sound features processed to provide distance information in the CF/FM bat, Noctilio albiventris: Evidence for a gated time window using both CF and FM components. J. Comp. Physiol. 156A, 457-469. Ryan, M. J. (1986). Neuroanatomy influences speciation rates among anurans. Proc. Narl. Acad. Sci. U.S.A.83, 1379-1383. Saunders, M. B., and Barclay, R. M. R. (1992). Ecomorphology of insectivorous bats: A test of predictions using two morphologically similar species. Ecology 73, 1335- 1345. Schluter, D., and McPhail, J. D. (1992). Ecological character displacement and speciation in sticklebacks. Am. Nat. 140, 85-108. Schluter, D., and Price, T. (1993). Honesty, perception and population divergence in sexually selected traits. Proc. R. SOC.Lond. 2328, 117-122.
ACOUSTIC SIGNALS AND CRYITIC SPECIES
353
Schnitzler, H.-U. (1987). Echoes of fluttering insects: Information for echolocating bats. In “Recent Advances in the Study of Bats” (M. B. Fenton, P. A. Racey, and J. M. V. Rayner, eds.), pp. 226-243. Cambridge University Press, Cambridge. Schober, W., and Grimmberger, E. (1987). “A Guide to Bats of Britain and Europe.” Hamlyn, London. Seger, J. (1985). Intraspecific resource competition as a cause of sympatric speciation. In “Evolution” (P. J. Greenwood, P. H.Harvey, and M. Slatkin, eds.), pp. 43-53. Cambridge University Press, Cambridge. Simmons, J. A., Fenton, M. B., and O’Farrell, M. J. (1979). Echolocation and pursuit of prey by bats. Science 203, 16-21. Skulason, S., and Smith, T. B. (1995). Resource polymorphism in vertebrates. Trends Ecol. EvoI. 10,366-370. Stebbings, R. E. (1988). “Conservation of European Bats.” Christopher Helm, Bromley. Stebbings, R. E., and Griffith, F. (1986). “Distribution and Status of Bats in Europe.” Institute of Terrestrial Ecology, Huntingdon. Sudman, P. D., Barkley, L. J., and Hafner, M. S. (1994). Familial affinity of Tomopeas r a w s (Chiroptera) based on protein electrophoretic and cytochrome b sequence data. J. Mammal. 75, 365-377. Suga. N., Niwa, H., Taniguchi, I.. and Margoliash, D. (1987). The personalized auditory cortex of the mustached bat: Adaptation for echolocation. J. Neurophysiol. 58, 643-654. Swofford, D. L. (1989). “PAUP: Phylogenetic analysis using parsimony.” Version 3.0. Illinois Natural History Surv., Illinois. Turner, A. K. (1982). Optimal foraging by the swallow (Hirundo rustica, L.): Prey size selection. Anim. Behav. 30,862-872. Turner, G . F., and Burrows, M. T. (1995). A model of sympatric speciation by sexual selection. Proc. R. SOC.Lond. 2608,287-292. Van Den Bussche, R. A,, and Baker, R. J. (1993). Molecular phylogenetics of the New World bat genus Phyllosromus based on cytochrome b DNA sequence variation. J. Mammal. 74, 793-802. Van Den Bussche, R. A., Baker, R. J., Wichman, H. A., and Hamilton, M. J. (1993). Molecular phylogenetics of Stenodermatini bat genera: Congruence of data from nuclear and mitochondrial DNA. Mol. Biol. Evol. 10,944-959. Vaughan, N. (1996). “Partitioning of Foraging Habitat Resources Among Bats (Chiroptera) in Great Britain.” Unpublished doctoral dissertation, University of Bristol. Vermeij, G. J. (1988). The evolutionary success of passerines: A question of semantics? Sysr. ZOO^., 37, 69-71. Vigny, C. (1979). The mating calls of 12 species and sub-species of the genus Xenopus (Amphibia: Anura). J . Zool. Lond. 188, 103-122. Walker, T. J. (1964). Cryptic species among sound-producing ensiferan Orthoptera (Gryllidae and Tettigonidae). Q. Rev. Biol. 39, 345-355. Waters, D. A., Rydell, J., and Jones, G. (1995). Echolocation call design and limits on prey size: a case study using the aerial-hawking bat Nycralus leisleri. Behav. Ecol. Sociobiol. 37, 321 -328. Weid, R., and Helversen, 0. von (1987). Ortungsrufe europlischer Fledermluse beim Jagflug im Freiland. Myoris 25, 5-27. Wells, M. M., and Henry, C. S. (1992). The role of courtship songs in reproductive isolation among populations of green lacewings of the genus Chrysoperla (Neuroptera: Chrysopidae). Evolution 46,31-42. West-Eberhard, M. J. (1983). Sexual selection, social competition, and speciation. Q. Rev. Biol. 58, 155-183.
354
GARETH JONES
White, G. (1813). “The Natural History and Antiquities of Selborne.” Facsimile edition. The Ray Society, London (1993). Wilson, A. C., Cam, R. L., Carr, S . M., George, M., Gyllensten, U. B., Helm-Bychowski, K. M., Higuchi, R. G., Palumbi, S. R., Prager. E. M., Sage, R. S., and Stoneking, M. (1985). Mitochondria1 DNA and two perspectives on evolutionary genetics. Biol. J. Linn. Soc. 26,375-400. Wilson, D. S. (1989). The diversification of single gene pools by density- and frequencydependent selection. In “Speciation and its Consequences” (D. Otte and .I.A. Endler, eds.), pp. 366-385. Sinauer Associates Inc., Sunderland, Massachusetts. Wilson, D. S . , and Turelli, M. (1986). Stable underdominance and the evolutionary invasion of empty niches. Am. Naf. 127, 835-850. Wilson, E. 0. (1992). “The Diversity of Life.” Allen Lane, London. Wu, L.-I. (1985). A stochastic simulation study on speciation by sexual selection. Evolution 39,66-82. Zingg, P. E. (1990). Akustische Artidentifikation von Fledermausen (Mammalia: Chiroptera) in der Schweiz. Rev. Suisse Zool. 97,263-294. Zyll de Jong, C. G. van, and Nagorsen, D. W. (1994). A review of the distribution and taxonomy of Myofis keenii and Myoris evoris in British Columbia and the adjacent United States. Can. J. Zool. 72, 1069-1078.
ADVANCES IN THE STUDY OF BEHAVIOR. VOL. 26
Understanding the Complex Song of the European Starling: An Integrated Ethological Approach MARCEL EENS DEPARTMENT OF BIOLOGY UNIVERSITY OF ANTWERP, U.I.A. BELGIUM
I. INTRODUCTION The study of bird song has made significant contributions to the development of modern ethology (Konishi, 1985; Konishi, Emlen, Ricklefs, and Wingfield, 1989; Catchpole and Slater, 1995). The impetus of bird song has been important in at least three different areas. First, studies of the development of bird song have made clear that behavior patterns cannot just be labeled as “innate” or “learned” but arise through an intricate interplay between the two (Johnston, 1988; Slater, 1989; Catchpole and Slater, 1995). Second, ever since Darwin (1871) proposed the theory of sexual selection to explain the evolution of elaborate mating displays that differ between the sexes, empirical studies of the role of acoustic signals in the context of reproduction have been important and influential. Bird song and other acoustic signals are probably the sexually selected traits that have been studied the most (Anderson, 1994). Finally, the discovery of neural substrates for song has introduced a new dimension to the study of bird song, making integration of behavioral and neurobiological studies feasible (Nottebohm, Stokes, and Leonard, 1976; Konishi, 1985; Nottebohm, 1993). Research on the vocal centers of the bird brain provides perhaps the best example in which the neural substrates of a complex behavior have been characterized (Catchpole and Slater, 1995; Francis, 1995; Jacobs, 1996). Ideally, to get a good understanding of the function and evolution of the song and song repertoire in a given species, direct interactions and exchange of ideas among those working on these questions are necessary. Although the study of bird song is probably one of the best examples of the ethological approach (Hinde, 1982) involving questions about the four major aspects of animal behavior (causation, ontogeny, evolutionary history, and adaptive 355
Copyright 0 1997 by Academic Press A11 rights of reproduction in any form reserved. 0065-3454lW $25.00
356
MARCEL EENS
significance; Tinbergen, 1963), it seems to me that proximate questions have been addressed mainly in controlled laboratory conditions, whereas ultimate questions dealing with the functions of song and song repertoires have occurred more in natural settings. For instance, most of our knowledge of the timing of vocal learning is based on laboratory studies and these may not always replicate the complex ecological and social interactions experienced by birds in the wild (Kroodsma, 1982; Petrinovich, 1990; O’Loghlen and Rothstein, 1993; West, King, and Freeberg, 1994). Similarly, neurobiological studies on the song system have occurred mainly in laboratory conditions and have largely been limited to a few species such as the domestic canary, Serinus canaria, and the zebra finch, Taeniopygia guttata (DeVoogd, 1994). Since several song learning and neurobiological studies have suggested discrepancies between field and laboratory results (Baptista and Petrinovich, 1984; Kirn, Clower, Kroodsma, and DeVoogd, 1989; Barnea and Nottebohm, 1994; Bernard and Ball, 1995a), a combination of laboratory and field approaches is necessary and may yield complementary information. An integration of proximate and ultimate questions about bird song is likely to be fruitful when studying song in the context of sexual selection. For instance, in most songbird species that have a repertoire of song types (i.e., sing different versions of the species-specific song), there appears to be a large variation among males in the number of song types that they sing, and often males that have a larger repertoire size are more successful in defending a territory or attracting mates (Anderson, 1994; Catchpole and Slater, 1995). However, if we want to really understand the effects of a song repertoire in intra- or intersexual competition in a given species, it is important that we not only look for correlations between repertoire size and components of fitness, but also try to understand why males vary so much in the number of song types they sing. Since song appears to be learned in every songbird species in which this has been examined, it could be that repertoire size is determined in part by the variety of songs that a male hears during his first few months of life (Kroodsma and Byers, 1991). Besides differences in learning opportunity, variation between individuals of the same species within a population might be a reflection of age and learning ability. Also, since both song acquisition and production are influenced by gonadal steroids (Marler, Peters, Ball, Dufty, and Wingfield, 1988), understanding the neuroendocrine control of song is important. Where there is evidence that the development of secondary sexual traits is testosterone dependent, it is also becoming increasingly clear that there are costs of circulating androgens, such as reduced resistance to disease (Folstad and Karter, 1992; Zuk, 1996). There is also evidence that intraspecific variation in song repertoire size is related to variation in certain song nucleus volumes
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
357
in the brain suggesting that there may be a cost in terms of brain space in having a large repertoire size (Nottebohm, Kasparian, and Pandazis, 1981; Canady, Kroodsma, and Nottebohm, 1984; Jacobs, 1996). Quite clearly, to fully understand the costs and benefits of secondary sexual characteristics such as song and song repertoires, ideally, an integration of behavioral, physiological, endocrinological, and neurobiological data is necessary. The complex song of the European starling, Sturnus vulgaris, has recently received much attention. These studies have involved questions at both the proximate and the ultimate level and have employed field observations and experiments as well as controlled laboratory experiments. In this chapter I try to integrate the results of these studies, using an ethological approach. After a brief introduction to the starling and its song, I first concentrate on song development and song learning in starlings. Next, I review recent studies on the song system neurobiology of starlings that have linked song behavior to the underlying neural substrate. Then, I review the current knowledge of the functions of the song of the European starling, and finally I also look at intraspecific variation in song behavior and at the intra- and intersexual effects of the starling’s song repertoire. Special attention is paid to analyzing and interpreting the results of studies on song development and on the function(s) of the song and the song repertoire in relation to the social and life history context of the starling.
11. THESTUDY SPECIES AND ITS SONG A. THEEUROPEAN STARLING Given that the development of song and the function(s) of song vary among species in ways that are likely to be related to other aspects of their breeding biology and life history, I first summarize briefly some relevant data on the way of life of European starlings in the wild. Due to human introductions, the European starling inhabits a large part of the world: where they originally inhabited Europe and Asia Minor, they have been introduced in North America, Australia, New Zealand, and South Africa (Feare, 1984). Since starlings tend to be common wherever they do occur, and since they adapt well to life in captivity, their breeding biology has been studied in detail and is well known. Starlings are highly vocal all year and they are very social birds, breeding in colonies and feeding and roosting in flocks through most of the year (Cramp and Perrins, 1994; Feare, 1984). Even close to egg laying, the members of a colony spend much time together in flocks (Wright and Cotton, 1994a). Starlings usually breed in loose colonies. Being a hole-nesting species, the distance between
358
MARCEL EENS
nests is often determined by the availability of nest holes. Males defend only a nest hole and a small area around it. In migrant populations, males usually arrive in the nesting area before the females, and older males usually arrive before yearlings. Many females breed at 1 year, while most males generally do not until 2 years old (Cramp and Perrins, 1994). Some male starlings defend two or more nest sites and try to become polygynous (Pinxten, Eens, and Verheyen, 1989; Pinxten and Eens, 1990). For males it pays to attract second females, whereas females do better in terms of breeding success if they are monogamous and have exclusive male parental investment (Pinxten, Eens, and Verheyen, 1993; Pinxten and Eens, 1994). Young remain in the nest for about 21 days. The duration of family bonds after fledging lasts between 7 and 16 days, with an average of 10 days (Van Elsacker, 1988). Thereafter, young starlings group into flocks of similaraged birds (Cramp and Perrins, 1994). If second broods occur, mates often do not stay together (Feare and Burnham, 1978). In southern England and Belgium where starlings have been individually marked, a very low natal site fidelity has been recorded (Feare, 1984; Pinxten et al., 1989). The starling is one of the few European songbirds that undergo a complete postjuvenile molt (Bent, 1950; Ginn and Melville, 1983), which takes place between July and September depending on the date of hatching. In autumn, and especially in winter, starlings often roost in huge numbers. Although starlings are highly social birds spending much time together in flocks, they can be extremely aggressive, especially during the breeding season (Feare, 1984; Eens and Pinxten, 1996), and intraspecific killing by both sexes (mainly between starlings of the same sex) has been observed in aviary studies (Risser, 1975; Meijer and Langer, 1995) and in the field (Flux and Flux, 1992). B. STARLING SONG
Starlings are well known for their long and complex songs and for the mimicry of other species and sounds that occurs within them. Starling song has been described as “a lively rambling melody of throaty warbling, chirruping, clicking and gurgling notes interspersed with musical whistles and pervaded by a peculiar creaking quality” ( Witherby, Jourdain, Ticehurst, and Tucker, 1943). Before spectrographic analyses were applied to starling song, it was thought to be a medley of a great many sounds. More recent spectrographic studies have shown that what at first may seem a random outpouring is in fact highly organized (Adret-Hausberger and Jenkins, 1988; Eens, Pinxten, and Verheyen, 1989, 1991a; Chaiken, Bohner, and Marler, 1993; Mountjoy and Lemon, 1995). Although the organization of starling song has been described by different authors, there has not been
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
359
complete agreement on the categories or the terminology used to describe this complex song. I use the following terminology (modified after Eens et al., 1991a; Chaiken et al., 1993; Mountjoy and Lemon, 1995). A starling song or song bout may last up to 1 min and has a complex syntactical and temporal organization. It is composed of many distinct units that are repeated once, twice, or several times before the next unit is introduced. I refer to these units as “phrases.” A phrase is a fixed combination of acoustic elements, with a duration of about 0.5-1.5 s. Phrases are mostly repeated two or more times before the next “phrase type” is introduced. A complete starling song bout usually includes four relatively distinct sections or categories of phrases, which tend to occur at particular points in the song bout (see Fig. 1). 1. A starling song or song bout usually begins with one or several whistles, which are interspersed with pauses of 1 s or more. Although these introductory or initial phrases are mostly simple pure-toned whistles, they sometimes also contain harsh, more structurally complex elements and they may rarely also contain heterospecific imitations. Each male has a repertoire of 2-12 different whistles. 2. In the second section, the song accelerates into a series of complex phrases, which are of relatively low amplitude and have no pauses or only short ones between them. Heterospecific imitations are often incorporated into this part of the song. The phrase types of this section of a given male may differ very much in structure. These complex phrases have been termed “variable phrases.” Each male has a repertoire of 10-35 variable phrase types. 3. The second part of the song is normally followed by phrases characterized by having a rapid series of clicks (“click trains”) running through the phrase at the same time as other sounds are being produced. The clicks, which have their maximum energy below 4 kHz, are delivered at a rate of about 10-15/s. These “rattle phrases” or “click phrases” only rarely include heterospecific imitations, and often have no distinct temporal gaps between them. Each male has between 2 and 14 rattle phrase types in his repertoire.
4. A starling song bout is typically completed by a series of phrases that are characterized by high-frequency song elements (6-10 kHz). These “high-frequency phrases” are the loudest in the typical sequence, and also tend to be repeated more often than phrases of the previous three categories. Males have up to six high-frequency phrase types. A starling song bout thus increases in both tempo and frequency as it progresses from beginning to end. Different categories of phrase types are
:..I:
c-
m
u
s. ..
..
i ;1 LL
_--- ...
a
~
-- . ,.
__
UNDERSTANDINGTHE COMPLEX SONG OF THE EUROPEAN STARLING
361
frequently also accompanied by different patterns of wing movement: the rattle phrase types are often accompanied by wing-flicking (Bohner and Veit, 1993), while the high-frequency phrase types are significantly more often accompanied by wing-waving than other categories of phrase type (Bohner and Veit, 1993; see also Eens et al., 1989). While the song bouts of all starling males consist of a predictable sequence of four different categories of phrases, the song bouts of an individual starling usually also follow a more or less set sequence, with the majority of phrase types being predictably followed by a single other phrase type in most or all cases (Adret-Hausberger and Jenkins, 1988; Eens et af., 1989, 1991a; Chaiken et af., 1993; Mountjoy and Lemon, 1995). Although a complete song bout contains four categories of phrases, a song may end at any point in this sequence, or very rarely a singer may revert to a phrase type already sung and repeat part of the song sequence (Eens et af.,1991a; Chaiken et al., 1993;Bohner, 1993;Mountjoy and Lemon, 1995). Since a phrase type nearly always occurs only once in a given song bout, it follows that the longer a song bout is, the more phrase types it contains (Eens et al., 1989). Since the beginning of a starling song bout may contain long pauses (i.e., between successive whistles), whereas pauses longer than 0.2 s extremely rarely occur later in the sequence, we have operationally defined a song bout as a period of at least 5 s of song with pauses no longer than 1.5 s in duration (Eens et al., 1991a,b). In several “versatile singers” such as the brown thrasher, Toxostoma rufum, and the Sardinian warbler, Sylvia melanocephala, it has been shown that males apparently do not possess a finite repertoire of notes and seem to improvise new notes as they sing (Kroodsma and Parker, 1977; Luschi, 1993). Fortunately, in the European starling, several authors have shown that, when plotting the cumulative number of phrase types encountered against the number of phrases examined (Fig. 2), asymptotes are reached in each case after several hundreds of phrases (Hindmarsh, 1984a,b; AdretHausberger and Jenkins, 1988; Eens et al., 1991a; Chaiken et al., 1993; Mountjoy and Lemon, 1995). Such a plot typically rises steeply at first, because the stereotypy of individual song sequences combined with the low repetition rate of phrase types result in a rapid presentation of new phrase types. Asymptotes are reached more quickly for low-repertoire than for high-repertoire males (Fig. 2). Between 15 and 20 song bouts need to
FIG.1. Spectrogram of a complete song bout of a male starling. The four categories of phrase type that characterize a starling song bout can be distinguished: (W) introductory whistle; ( V ) “variable phrases”; (R) “rattle phrases”; (T) “high-frequency phrases.” This song bout lasted 25 s and contained 30 phrases and 12 phrase types (indicated by a letterfigure combination).
362
MARCEL EENS
70
e
40
A=
n
0
0
200
400
600
800
1000
Number of phrases recorded
FIG.2. Cumulative plots showing how the number of new phrase types found rises as an increasing number of phrases are analyzed in four male starlings. Note that three out of four males still sang new phrase types after they had already sung 300 phrases. After Eens et al. (1991a). Reprinted with permission from the Belgian Journal of Zoology.
be analyzed to obtain the entire song repertoire of phrase types. Although the majority of song is sung in song bouts, some whistles and heterospecific imitations are also sung outside song bouts. Total repertoire sizes of male starlings range from 15 to 70 phrase types (see later discussion). Although the main focus of this chapter is on male song behavior, recent studies of female starlings’ song behavior are also reviewed. Although repertoire sizes are smaller in female starlings, the organization of female song is similar to that found in males except that the high-frequency phrase types typically occurring at the end of song bouts in males appear to be absent in females (Hausberger, Henry, and Richard, 1995a; Hausberger, Richard-Yris, Henry, Lepage, and Schmidt, 1995b). 111. SONG LEARNING The young princess had a starling and also nightingales that were actually trained to talk Greek and Latin, and moreover practiced diligently and spoke new phrases every day, in still longer sentences. Birds are taught to talk in private where no other utterance can interrupt, with the trainer sitting by them to keep on repeating the words he wants retained, and coaxing them with morsels of food. Pliny the Elder
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
363
European starlings have long been renowned for their vocal virtuosity. Both Pliny and Shakespeare referred to the starling largely on account of its ability to mimic the sounds of other birds and animals that it encountered in its surroundings (Pliny, 1940; Feare, 1984; Hindmarsh, 1984a). More than a hundred years ago, Saxby (1874) noted that “the starling imitates the note or cry of almost any bird.” Many other early authors noticed that the starling is one of the best of mimics (Witchell, 1896; Townsend, 1924; Chisholm, 1932; Allard, 1939; Bent, 1950). Such interspecific vocal imitation by free-living birds implies that vocal learning plays a role in vocal development (Kroodsma, 1982). When held in captivity, it has been known since antiquity that starling vocalizations can contain fragments of human speech. Pliny the Elder, writing in first century Rome, reported starlings mimicking Latin and Greek (see quotation at beginning of this section). Townsend (1924) mentions the case of a starling, living in 1582, that sang and spoke in German and Polish. Although these authors probably exaggerated the singing and learning capabilities of starlings, it is surprising that their songlearning ability has only recently been the interest of scientific study. In this section, I first examine how the syntactical and temporal features of starling song are acquired by young males. Then, I describe the time course of song acquisition and production focusing on field and aviary studies as well as controlled laboratory studies. Next, I look at song tutor choice in aviary conditions and at dialectal patterns found in the wild. A.
I N F L U E N C E OF AUDITORY A N D SOCIAL STIMULATION ON THE DEVELOPMENT OF SYNTACTICAL, TEMPORAL, A N D PHONOLOGICAL FEATURES I N STARLING SONG
All songbird species studied so far have to learn the species-typical song pattern, and this is usually through the acquisition of auditory models from adult conspecifics (but see West, King, and Eastzer, 1981; West et al., 1994, for a notable exception). If young birds are deprived of acoustic models or are exposed to heterospecific song or inadequate learning conditions, they will develop song patterns that differ from the species-specific one (Kroodsma, 1982; Slater, 1989). Earlier, I described how starling songs or song bouts have a complex syntactical and temporal organization. As such, starling song is an ideal model system to investigate how syntactical and temporal song features are acquired by young males. Chaiken et al. (1993) reared young starling males in complete acoustical isolation. They found that isolate starlings’ song differed in three important parameters from the normal starling song. First, starlings reared in total acoustical isolation lacked entirely the typical sequential organization of song parts. Second, the song bouts of starling isolates did not increase as much in frequency as they progressed from beginning to end than normal starling song bouts. Third, isolate starling song bouts displayed a completely
364
MARCEL EENS
even distribution of breaks and thus lacked any increase in speed of performance. Starling isolates also had significantly smaller repertoires than livetutored males (Fig. 3). Chaiken et al. (1993) also compared the song of live-tutored, tape-tutored, and acoustically isolated birds to tease apart specific contributions of the social and acoustic environments. Both live-tutored and tape-tutored males developed songs displaying the basic features of the species-specific song organization. The songs of the live-tutored birds largely matched those of their tutors in two respects: they copied a large number of phrase types of their tutors (25.3 on average accounting for 69% of their repertoire; see Fig. 3) and they also acquired the species-typical song organization. The tape-tutored males copied many fewer phrase types from their song models (2.5 on average accounting for 18%of the total repertoire; see Fig. 3), yet tape-tutoring had a clear influence on song organization. Tape-tutored males exhibited the same qualitative characteristics of song organization as the live-tutored males, although they showed quantitative syntactic and phonological deficits. For instance, their songs did not always display the degree of stereotypy in the sequencing of the song sections typical for wild-caught males. Furthermore, although tape-tutored males as a group acquired all categories of phrases, a few males displayed phonological 40 35
T
0Repertoim size
=
Phrasetypes copied fmm tutors
30
t
25
.K
9.
20
u-
0
3 5 z
15
ia 5
a
Untutored
Tape-tutored Live-tutored Males
FIG.3. Repertoire sizes (number of phrase types) and number of phrase types copied from tutors in untutored, tape-tutored, and live-tutored males. Error bars are standard errors. After Table 1 from Chaiken et al. (1993).
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
365
abnormalities such as having no rattle phrases or producing whistles lacking the usual degree of stereotypy. Despite these abnormalities, the tapetutored males achieved a much closer approximation of species-specific syntax than did birds reared in acoustic isolation. Chaiken et al.’s findings thus clearly illustrate that the tape-tutored males made use of the training tapes in acquiring the species-specific features of song syntax despite the fact that they copied only two or three phrase types. Acquisition of the species-typical song organizaton requires exposure to song, but does not depend on the imitation of minimal song units. As such, the data led them to postulate that the acquisition of phonological and syntactical information are two different processes in starlings. In a recent paper, Bohner and Todt (1996) further examined how syntactical and temporal song features are acquired by young male starlings. They investigated how young male starlings were influenced during song learning by acoustic patterns lacking certain syntactical and temporal features of the species-specific song. For that purpose, hand-reared males were exposed to a sequence of heterospecific song patterns of nightingales, Luscinia megarhynchos, which, in contrast to normal starling song, were not reiterated, were evenly separated by breaks throughout the entire tutor sequence, and did not show an increase in frequency from the beginning to the end of the sequence. The nightingale song was arranged on tape and presented by a “social tutor” (i.e., a person who was familiar to the birds). Despite the fact that the experimental males did not acquire many phrase types from the training tape, they were clearly influenced by the exposure to auditory stimuli. They developed songs with aspects of conspecific syntax (sequential order of low- and high-frequency phrase types) and temporal organization (more breaks between phrases at the beginning than at the end of a song bout), which are absent in the songs of starlings reared in complete acoustic isolation (Chaiken et al., 1993). Because the tutor song did not display any of these features, it can be concluded from this experiment that young starlings have a predetermined knowledge of these syntactical and temporal song features (i.e., that no experience with the respective acoustic pattern is necessary), but need an acoustic input from the environment to convert such information into a corresponding vocal output. As such, the acoustic patterns effective in stimulating such a development d o not need to contain specific features of starling song: it seems that some kind of general acoustic input is sufficient for the development of the species-typical song. At the moment, it is not clear yet how unspecific such a stimulus may be. Bohner and Todt (1996) argued that the syntactical organization of the starling song seems to be a rather conservative song feature of which many aspects do not need to be acquired from external models.
366
MARCEL EENS
B. SOCIAL INFLUENCES ON SONGLEARNING
Research on a wide variety of species during the last decade has shown that social factors play an important role during song learning (Payne, 1981; Baptista and Petrinovich, 1984; Clayton, 1987; Catchpole and Slater, 1995), and several studies show that starlings are no exception to this. First, as already briefly mentioned, Chaiken er al. (1993) found that there was a dramatic difference between live-tutored and tape-tutored males in the number of phrase types copied and in repertoire size (Fig. 3). Live-tutored males copied 10 times as many phrase types from their tutors as tapetutored birds did from their tapes. In the tape-tutored males, imitations of the starling songs accounted for only 18% of the total repertoire compared to 69% in the live-tutored males (Fig. 3). Although tape tutoring can be as effective as live tutoring in some species (e.g., song and swamp sparrows, Melospiza melodia and M. georgiana, respectively, Marler and Peters, 1987, 1988), the superiority of live tutoring has been demonstrated for many other species; in some species, such as the zebra finch, conspecific song will not usually be copied even from a tape (ten Cate, 1994). The power of social interactions in shaping the singing behavior of starlings has also been demonstrated by West, Stroud, and King (1983). They investigated how the social context experienced by starlings influenced their mimicry of human vocalizations and sounds (see also West and King, 1990). They found that the occurrence of human mimicry depended on the nature of social interaction between the starling and its mimicked partner. Handreared starlings were adept at mimicking human speech, but they did so only if they were hand-reared in a human household where they were part of the social life of their caregivers. Thus, only starlings in interactive contact with their caregiver mimicked sounds with a clearly human origin. Hand-reared starlings that lacked a social tutor but instead were housed in an aviary with other birds (either cowbirds or other starlings) failed to mimic human sounds. West and King (1990) also tried tutoring nine starlings using tapes of the caregiver’s voice. There was no evidence of mimicry, except that one bird learned the sound of tape hiss, supporting again the superiority of live tutoring. Further evidence for the importance of social interactions on vocal learning in starlings comes from an aviary study of Hausberger er al. (1995b). They examined the social organization in a captive group of adult male and female starlings caught in different localities to try to understand the possible social basis of song sharing. They found that there was a clear relation between social organization and the pattern of song sharing. Females that were socially associated shared phrase types, while this was not the case between females that were not socially associated. Similarly, males
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
367
shared songs with other males to an extent that depended on their degree of social association. The results of this study suggest that familiarity (i.e., “who likes to be with whom”) may be more important than agonistic interactions in the process of song learning in the starling. OF SONGLEARNING I N FIELD A N D LABORATORY CONDITIONS C. TIMING
In most songbird species studied to date, the acquisition of song from adult models occurs principally or exclusively during a restricted time period early in life. Apparently, there is a time window or sensitive phase during which song learning occurs easily and other times when it is difficult or impossible (DeVoogd, 1994; Catchpole and Slater, 1995). Most temperate zone songbirds are thought to learn their species-typical song during their hatching year and to produce fully developed song by the time they are 1 year old and beginning their first breeding season (O’Loghlen, 1995).For most birds, these will be the songs they sing for the rest of their lives. In contrast to these “agelimited” learners, some species apparently retain the ability to learn new songs beyond the first breeding season and perhaps throughout life, although experimental evidence is rare (Catchpole and Slater, 1995). For instance, in the domestic canary, males add some new syllables to their repertoire when 2 years old, modify others, and drop a few of those sung the previous year, although repertoire size does increase with age (Nottebohm and Nottebohm, 1978; Nottebohm, Nottebohm, and Crane, 1986). Later, I review field and aviary studies in which recordings were made of individual male starlings over successive years to document whether repertoire changes occur in this presumed “open-ended” learner (Feare, 1984). I also look at controlled laboratory studies that have exposed hand-reared male starlings to a series of tutors over a period of 18months, and I explain differences in results obtained in laboratory and field conditions. 1. Extended Song Learning in Wild and Captive Starlings
Studying starlings at a nest box colony in Belgium, Eens et al. (1991b) found that yearling males had significantly smaller repertoire sizes and sang significantly shorter song bouts than older males. Although these findings suggest that male starlings are able to learn new song types after their first year of life, two alternative explanations are possible. First, changes in song from year to year do not necessarily imply new memorization. Males may memorize more songs than they produce early in life, and those in their repertoire each year may be a subset of the ones that they learned in their first year (Slater, Jones, and ten Cate, 1993). This can be decisively answered only in controlled experiments where the time of exposure to specific acoustic stimuli is known exactly (see later discussion). Second, it might
368
MARCEL EENS
be that yearlings with smaller repertoires experience higher mortality and that the disappearance of birds with small repertoires occurs sometime between the first and second breeding season. Indeed, both in the great tit, Purus major, and in the song sparrow, where 1-year-old males also have smaller repertoires than older males (Lambrechts and Dhondt, 1986,1987; Hiebert, Stoddart, and Arcese, 1989), this apparent change in repertoire size is due to attrition of males with smaller repertoires rather than to an increase in repertoire size of individuals with age (see also McGregor and Krebs, 1989). This explanation is highly unlikely for the starling because Eens, Pinxten, and Verheyen (1992a) found that the range of repertoire sizes recorded for six yearlings did not overlap the range for 19 older males: repertoire sizes of yearling males ranged from 21 to 31, while that of older males ranged from 35 to 67. In recent years, I have been collecting additional data and I have now information on repertoire size for 21 yearlings and 33 older males. Using this larger sample, it remains clear that there are dramatic differences in repertoire size between yearling and older males (see Fig. 4 and Table I, which summarizes the results of this and other
Yearling males
40 30 20
6 c JU
n
10
nn
o!
. I -
L
c
g
Older males 40
0)
a
30 20 10
llllllu 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80
Repertoire size class
FIG.4. Comparison of the repertoire sizes of yearling (N = 21) and older (N = 33) male European starlings. Repertoire sizes are divided into classes of five (i.e.,35 contains repertoire sizes ranging from 31 to 35). Note that there is only a slight overlap in repertoire size between yearling and older males. After Eens el al. (1992a) and M. Eens, unpublished data.
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
COMPARISON
OF
369
TABLE I REPERTOIRE SIZE BETWEEN YEARLING AND OLDER M A L E EUROPEAN STARLINGS IN FIVEDIFFERENT POPULATIONS Repertoire size
Population
Yearlings
Older
Reference
Belgium
26.5 (14-36; n = 21)
47.6 (29-67; n
Canada
33.9 (24-47; n = 9)
46.6 (35-65; n = 19)
Germany
17.0 ( n
30.6 (21-39; n = 5)
Germany United States
18.7 (13-24; n 36.7 (28-51; n
=
1) = =
3) 6)
73.0 ( n 1) 70.7 (41-90; n
=
33)
7
=
6)
Eens ef al. (1992a) and M. Eens (unpublished data) Mountjoy and Lemon (1995) Adret-Hausberger ef 01. (1990) Bahner and Todt (1996) Chaiken ef al. (1993) and M. Chaiken (unpublished data)
studies), but it can also be seen that there is a slight overlap in repertoire size between them. Mountjoy and Lemon (1995) also found that repertoire size estimates for nine yearlings were significantly smaller than those for older males, although in their population there was a greater overlap in repertoire size. Furthermore, Adret-Hausberger, Guttinger, and Merkel (1990) reported that a yearling male had the smallest repertoire out of six males recorded in the 1987 breeding season of their study. Similarly, Chaiken et al. (1993) found that yearling males that were live-tutored in laboratory conditions had much smaller repertoires than older wild-caught males. A similar pattern has also been found by Bohner and Todt (1996) in a laboratory experiment. These five studies in different populations clearly indicate that yearling males have reduced repertoires and also that adult male starlings are able t o incorporate new phrase types into their crystallized song. I am unaware of any other songbird species where the differences in repertoire size between yearling and older males are so clear, although in brown-headed cowbirds, Molothrus ater, yearlings also have smaller repertoires of the “perched song” than older males (O’Loghlen and Rothstein, 1993; O’Loghlen, 1995). O’Loghlen and Rothstein (1993) argued that in their population yearling cowbirds have little or no opportunity to hear, and thus memorize, the local songs as juveniles. This explanation is unlikely for starlings, since many authors have reported that starlings have an extremely long singing season (Feare, 1984) so that tutors are available for a long time and give juveniles plenty of opportunity to learn songs. It is possible that there might be a constraint on the capacity of song
370
MARCEL EENS
memory that changes with age, perhaps through greater investment in the brain nuclei responsible for song memory as males age (see later discussion). In three different studies, individual male starlings have been recorded in successive breeding seasons t o document whether and how repertoire size and composition changes between years. First, studying a small nest box colony in Germany, Adret-Hausberger et af. (1990) reported that four starling males added new phrase types to their repertoires from one year to the next. All four males increased their repertoire size, and one male even added new phrase types to his repertoire when he was 7 years old. From two out of the four males, however, very few songs were obtained in one of the recording years, making it difficult to exclude the possibility that changes in repertoire size are the result of inadequate sampling (Eens, Pinxten, and Verheyen, 1992b). In an aviary study, Eens et af. (1992b) documented changes in the repertoires and other song characteristics of captive males recorded in successive years. They found that seven out of the nine males studied sang new phrase types in their second year of recording, clearly indicating that starlings go on modifying their repertoires in adulthood. Four out of the five males that were at least 4 years old when first recorded were found to modify their repertoire. Nevertheless, Eens el al. (1992b) found that the amount of change in repertoire size and song bout length occurring between years depended on the age of the males. Three males, which were yearlings when recorded for the first time, increased their repertoire size and song bout length significantly more than six older males. These results obtained in an aviary situation suggested that the ability t o incorporate new phrase types into the song decreases with age. Finally, Mountjoy and Lemon (1995) recorded the songs of wild starlings over successive years and found that repertoire size estimates increased between the first and last year of recording for five out of seven males that were at least 2 years old when first recorded. Overall, there was a significant increase in the repertoire size estimates. Furthermore, all seven males showed extensive changes in the composition of their repertoires: many phrase types were dropped from the repertoire, and others were modified. Most males were found to sing 20 to 30 new phrase types each year. Compared with the older captive males studied by Eens et al. (1992b), the older wild birds studied by Mountjoy and Lemon increased their repertoire size more and also showed a much higher degree of repertoire modification. Although it is possible that the males recorded by Eens et af. were older on average than the birds Mountjoy and Lemon studied, a more likely explanation might be that a major effect of being in captivity was a reduction of the variety of song tutors available. A decreased exposure to novel
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
371
phrase types from year to year in contrast to wild starlings, might explain the observed difference (Mountjoy and Lemon, 1995;see also later discussion). Overall, the above studies indicate that extended learning occurs in adult male starlings and also that the ongoing process of song learning is likely to result in a correlation between repertoire size and the age of the individual, as can be seen in Fig. 5 where the repertoire size of known-age captive and wild males is given between the age of one to four. These results and the field data discussed above suggest that it is likely that starlings retain the ability to change their repertoires throughout life. Recently, Hausberger ef al. (1995a,b) also looked at changes in repertoire size in female starlings. They induced female starlings to sing by injecting them with testosterone. They found that large changes occurred in both repertoire size and composition for all females between the different recording sessions covering one year. Some of the changes in song repertoires corresponded to changes in the social associations between females between the recording sessions, indicating that female starlings can also acquire new songs in adulthood. Apparently, both male and female starlings are ageindependent learners. Hausberger et al. (1995a) did not find age-related
3
50
*O
I
i
1
i
1
FIG. 5. Average repertoire sizes of male starlings in different age categories. Error bars are standard deviations. Figures above the bars represent the number of males observed. From M. Eens, unpublished data.
372
MARCEL EENS
differences in repertoire size or song bout length as has been reported in male starlings. There was, however, strong evidence that, even under testosterone treatment, females did not necessarily sing all the songs that they may have in memory. In alpine accentors, Prunella collaris, song complexity of known-age females increases significantly with age (Langmore, Davies, Hatchwell, and Hartley, 1996). Timing of Song Acquisition in Controlled Laboratory Conditions Although the observed changes in repertoire size and composition reported in the previous studies strongly suggest continued song learning in adulthood in the starling, it remains possible that the newly produced songs had been memorized much earlier. As I have already stressed, this can be decisively answered only in controlled experiments where the time of exposure to specific acoustic stimuli is known exactly. To establish unambiguously whether starlings could memorize new songs in adulthood, Chaiken, Bohner, and Marler (1994) exposed hand-reared males to a series of either live conspecific tutors or tape recordings of conspecific song, and isolated them from any other exposure to conspecific song. The tutoring schedule was identical for the six live-tutored and six tape-tutored birds: each male was exposed to conspecific songs during three 6-week training sessions. The first began when the males were in their second month (37-40 days old), the second when they were 3 months old (94-97 days), and the third when they were 10-11 months old. Three of the six live-tutored males were given an additional 6-week training session when they were 18 months old. The phrase types produced by the liveand tape-tutored males were recorded and analyzed when they were 9 and 13 months old and compared with those of their tutors or training tapes to infer the time of memorization. The song of the three live-tutored males given a fourth tutoring session was recorded when they were 20 months old. This elegant experiment yielded several important results and conclusions. First, it clearly demonstrated that starlings are capable of memorizing as well as producing new phrase types up to at least the age of 18 months: all of the live-tutored males and four of the six tape-tutored males imitated phrase types presented in each of the tutoring sessions. Although extended learning had been strongly implied by earlier studies on wild and aviary starlings and on other species in which males were found to match the songs of new neighboring males during or after the first breeding seasons (Jenkins, 1977; Payne, 1985; McGregor and Krebs, 1989), this study was the first to demonstrate conclusively that memorization of new phrase types occurs in adult songbirds. The results suggest that the sensitive period for song learning is extended, and perhaps even lifelong, in the starling. Whereas in the canary, another open-ended learner, the extended plasticity
2.
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
373
probably occurs during annually recurrent sensitive phases occurring in autumn (Nottebohm, Nottebohm, Crane, and Wingfield, 1987; Marler, 1987), starlings apparently are able to memorize new phrase types both during and outside the breeding season (Chaiken et al., 1993,1994; see also Bohner, Chaiken, and Marler, 1990 and later discussion). It would be interesting to investigate in detail the precise timing of the readiness to acquire new songs throughout life in open-ended learners such as starlings and canaries. When the starlings studied by Chaiken et al. (1994) added new phrase types to their repertoire, they dropped some of the old ones. The rate of repertoire turnover (defined as the percentage of new phrase types in the repertoire that had not been present in the previous sample) was similar for live- and tape-tutored males, with 5 5 9 2 % new phrase types in each repertoire sample. The starlings showed repertoire turnover both within and between breeding seasons. In this respect, there are some marked differences with results obtained in the field and in aviaries. First, whereas there is a high repertoire turnover within a breeding season in laboratory conditions, song repertoires appear to be stable during the breeding season in field and aviary conditions (Adret-Hausberger et al., 1990; M. Eens and R. Pinxten, unpublished data). Second, repertoire turnover between seasons was much higher in laboratory conditions than that reported in the wild (Mountjoy and Lemon, 1995); repertoire turnover was lowest in an aviary study reported by Eens et al. (1992b). While the low repertoire turnover in the latter study may follow from the low number of song tutors available (see previous discussion), the much higher turnover in the laboratory may be induced by prolonged housing with a new tutor in small cages, forcing interaction with him. A similar explanation was suggested by Chaiken et al. (1994) to explain the within-season repertoire turnover in laboratory conditions. Since they had the impression that turnover occurred more quickly during or just after a tutoring session than in the periods between sessions, they argued that within-season turnover may have been precipitated by the introduction of a new tutor or tape. However, it is also possible that the conditions of laboratory housing may explain the observed differences in within-season repertoire turnover between the laboratory and field (and aviary) studies. Several studies have shown that the presence of a female can affect the pattern of testicular growth and regression in male starlings (Burger, 1953; Schwab and Lott, 1969; Gwinner, 1975). Similarly, the presence of nest boxes is known to affect testosterone levels in male starlings (Dittami, Wozniak, Ganshirt, Hall, and Gwinner, 1986; Gwinner, Gwinner, and Dittami, 1987). Since males in the laboratory situation were housed in cages without nest boxes and females, it seems very likely that they had reduced testosterone levels, It has been shown that circulating
374
MARCEL EENS
levels of sex steroid and pituitary hormones are lower in the laboratory than in the field in starlings (Dawson and Goldsmith, 1982,1984; Bernard, 1995) and in other species (Wingfield and Moore, 1987). Testosterone has been implicated in the crystallization of song (Nottebohm, 1969; Marler et al., 1988). Perhaps vocal learning is temporarily arrested during the breeding season in the wild when testosterone has reached a level sufficient to produce crystallization (see O’Loghlen and Rothstein, 1993, for a similar proposal for the cowbird). Nottebohm (1993) also suggested that high levels of testosterone interfere with vocal learning. This could easily be tested by subjecting two groups of males with different testosterone levels to the same tutoring regimes. Although Bohner ef al. (1990) have also shown experimentally that in laboratory conditions starlings are capable of memorizing new phrase types regardless of whether they are in or out of breeding condition, it is best to be cautious when extrapolating such findings to the situation in the wild. T. R. Birkhead (personal communication) recently dissected two male starlings that had been housed in laboratory conditions and that appeared to be in reproductive condition, but found that both of them had regressed testes and no sperm in the seminal glomera. This suggests that androgen levels need to be measured to be sure of breeding condition. In support of my interpretation of the discrepancy between field and laboratory results, it has been found that area X, a song control nucleus implicated in song learning, decreases in volume during the breeding season in wild starlings, but that it is static in volume under laboratory conditions (Bernard, 1995; Bernard and Ball, 1995a; see later discussion). A third important result from Chaiken et al.’s study was that tutoring appeared to be relatively ineffective during the earliest period of exposure to song: at 9 months, both live- and tape-tutored males copied phrase types predominantly from the second tutor (after the juvenile molt) rather than from the first tutor (just after fledging). The live-tutored males copied a mean of 2.7 (? 1.1 S E ) phrase types from the first tutor and a mean of 22.6 t 3.1 from the second; the tape-tutored males copied a mean of 0.3 ? 0.2 phrase types from the first tutoring tape and a mean of 2.2 t 1.1from the second. This finding is in strong contrast with results obtained for other passerines, at least for those that are known to learn their songs before the first breeding season (Chaiken et al., 1994). In such cases, song acquisition peaks early in life, usually within a few days of fledging, and then drops off sharply (song sparrows, Marler and Peters, 1987; swamp sparrows, Marler and Peters, 1988; for reviews, see Kroodsma, 1982; Catchpole and Slater, 1995). It is unclear at the moment why memorization of song begins later in starlings than in other species studied, although several explanations may be suggested. Perhaps, this late onset of song acquisition may be linked to the unusual molting behavior found in the starling. Recall
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
375
that the starling is one of the very few temperate zone songbirds in which juveniles undergo a complete postjuvenile molt. Juvenile starlings may be “programmed” not to invest time and energy in song learning during the molting period with its concomitant demands for energy and nutrients, especially because molting occurs in the summer when starlings experience a period of food shortage (Feare, 1984) and when there appears to be a segregation of juvenile and adult birds (Feare, 1984; Cramp and Perrins, 1994). This could be tested by looking at other species that undergo a complete postjuvenile molt, such as, for example, skylarks, Alauda arwensis, and corn buntings, Miliaria cafandra (Ginn and Melville, 1983). Another possibility, suggested by Chaiken et al. (1994), might be that starlings actually did memorize phrase types that they were exposed to during the first tutoring session, but tended to produce imitations of the songs they had heard most recently. This explanation gained support from the finding that many of the phrase types copied from the second tutor or tape were replaced by phrase types copied from the third. Apparently, there was a consistent preference during the first year for singing phrase types imitated from the most recent tutor. At 20 months, this trend was reversed for the three livetutored males: more phrase types were copied from the third tutor as compared to the fourth. Finally, it is possible that the overlapping of the sensory and motor phases of song learning may have facilitated learning in the second and third tutoring session as compared to the first tutoring period when the males were still in the earliest stages of subsong (Chaiken et al., 1994). Another interesting finding of Chaiken et al. (1994) was that not all of the imitated phrase types in the starlings’ adult repertoires had been acquired from the most recent tutor or tape recording. Some of the new phrase types were imitations of song models to which males had been exposed many months before. In some cases, there was as much as an 18month delay between the time of exposure and the production of the phrase types. This finding, that starlings can add or reinstate phrase types after delays of over one year, indicates that it may be unreliable to infer openended learning on the basis of repertoire changes in adulthood in other species, unless birds change their songs to match those of neighboring males precisely. A final result of Chaiken et al.3 (1994) study worth mentioning is that, although live-tutored males imitated a much greater number of phrase types and developed larger repertoires than tape-tutored males, there were no significant differences between the two groups of males in the timing of learning or in the rate of repertoire turnover. Several studies have suggested that the nature of the tutoring stimulus can affect the time course of song learning. For instance, in the white-crowned sparrow, Zonotrichia
376
MARCEL EENS
leucophrys, which acquires songs from tape recordings and from live tutors, the timing of the sensitive phase seems to be dependent on rearing conditions, with social factors playing a prominent role. While song acquisition from tape recordings is restricted to a short sensitive phase during the first 50 days of life, they do acquire songs well after 50 days from live avian tutors with whom they can interact, even copying the song of another species, although they can hear conspecific song in the background (Marler, 1970; Baptista and Petrinovich, 1984, 1986; Baptista, Bell, and Trail, 1993). In the starling, tape tutoring is apparently sufficient to stimulate song learning after sexual maturity. Baptista et al. (1993) pointed out that social interactions not only extend the sensitive phase, but also accelerate the song development process. Chaiken et al. (1994) found that the songs of tape-tutored male starlings crystallized at the same time as those of livetutored males, suggesting that there was no delay in motor development in the males deprived of social interactions. D. SONGTUTORCHOICE AND SONGDIALECTS 1. Song Tutor Choice in Aviaries
In recent years, many (experimental) studies have dealt with the choice of a song tutor both in controlled laboratory conditions and in aviaries (Slater, Eales, and Clayton, 1988; Williams, 1990; Mann and Slater, 1995). Many such studies have sought links between song tutor selection and social behavior. In the zebra finch, which has been studied in great detail in this respect, it has been shown that tutor choice can be influenced by morphology, song characteristics, and by the quality and quantity of certain social interactions, especially those involving parental care, tutor aggression, and spatial proximity (see Mann and Slater, 1995, for a recent review). In several species, but in particular the zebra finch, there has been much interest in whether the father’s song is most likely to be that copied by young males (Mann and Slater, 1995). Two aviary studies have looked at song tutor selection in captive starlings. First, in a study by Eens et al. (1992b), four male starlings were reared in captivity by four different pairs. Although all four males copied a large proportion of their song from adult males with whom they shared the aviary, they did not necessarily learn their songs from their father: two males copied most phrase types from their father, while the other two copied most of their song from another adult male with whom they shared the aviary during their hatching year. The first breeding season’s song of two out of four males contained phrase types copied from more than one model male, while all three males recorded during their second breeding season copied phrase types from more than one male. One male even
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
377
derived his phrase types from four different males, resulting in a composite song consisting of portions of song from several adult males added to its own improvised one. The second study has been briefly mentioned earlier. Hausberger et al. (1995b) brought together male and female starlings from different localities and examined if, and how, their social organization within the aviary influenced the amount of song sharing. They found that the social organization in the aviary relied on spatial proximities rather than aggressive interactions, which appeared to be extremely rare. Captive starlings tended to form within-sex groups, in particular females. They observed that the pattern of song sharing clearly reflected the social organization. Song sharing was almost completely restricted to birds of the same sex, which suggests that song tradition may occur along sexual lines in starlings as has been shown in Indian hill mynahs, Gracula religiosa, another member of the Sturnidae (Bertram, 1970). Close proximity was a good predictor of song sharing: females that were socially associated shared many of their songs, while this was not the case between females that were not socially associated. Similarly, but less than in females, males shared songs with other males to an extent that depended on their degree of social association. When Hausberger et al. (1995b) changed the group compositions to examine whether changes in the social situation elicited changes in song behavior, both adult females and males seemed to be able to acquire new songs in a social context. Hausberger et a1.k results with respect to the social organization are in strong contrast with earlier studies on starling social behavior that showed a clear peck order in aviary groups (Davis, 1959; van der Mueren, 1977; see later discussion). Furthermore, extreme intraspecific fighting and even intraspecific killing (mainly between starlings of the same sex) have been observed in several aviary studies (Risser, 1975; Meijer and Langer, 1995; Eens and Pinxten, 1996; M. Eens, personal observation) and in the field (Flux and Flux, 1992). It is not clear why almost no aggressive behavior occurred in Hausberger et al.’s study. Perhaps the extremely high bird density may be of relevance. For instance, from November to the end of March, the birds were housed in small indoor aviaries (1.4 X 1.8 X 0.7 m) with a bird density of 0.196 and 0.294 m3/bird (9 and 6 birds, respectively). Hausberger et al. suggested that the forced prolonged contact in a dense situation may increase the need for tolerance, and therefore vocal sharing. 2.
Tutor Choice and Song Dialects in Wild Starlings
In general, starling song is highly individual: the largest part of the repertoire seems to be unique to each individual in the wild; usually only the high-frequency phrase types occurring at the end of a song bout and some of the whistles are shared between neighboring (breeding) starlings
378
MARCEL EENS
(Adret-Hausberger and Jenkins, 1988; Eens et af., 1989, 1991a; Chaiken et al., 1993). This might suggest that the whistles and the high-frequency phrase types, which are usually the loudest parts of the song, are learned or copied at the breeding place, while the other parts of the song are learned at other places. At this stage, it might be interesting to look in more detail at the period from fledging to first breeding in the starling. Starlings usually fledge at an age of 21 days. During the period of further parental care, which lasts 3-12 days, males and females each take care of a number of fledglings. Adult males usually do not sing during this period, making it highly unlikely that juveniles learn their song from their fathers in the wild. After this period, juveniles move away from the natal site and have a postbreeding dispersal. Until the next breeding season, most starlings, both juveniles and adults, tend to gather in large flocks. Since starlings often sing within short range in these flocks, juvenile males come into close contact with many different potential learning sources. Eens et al. (1992b) suggested that free-living starlings learn their song from many different models with whom they come into contact between fledging and first breeding, the more as most starling males breed for their first time only during their second season (i.e., in their third calendar year). This scenario may explain why wild starlings have a highly individual song, whereas in laboratory and aviary conditions males copy a high proportion of their song from conspecifics with whom they interact. Although Hausberger et al. (1995b) presented evidence that captive female starlings share their song with individuals of the same sex, the few field data available are less clear. Studying a polygynous trio in the wild, Hausberger and Black (1991) found that females also shared whistle types with their male and also song-matched with him. Hausberger and co-workers have looked in detail at the whistle repertoire in different populations in France, Germany, and Australia (Hausberger and Guyomarc’h, 1981; Adret-Hausberger, 1983, 1989; Adret-Hausberger and Giittinger, 1984; Hausberger, 1993). Each male has a repertoire of 7-12 whistle types. They found that some whistle types were found in all or most males in all populations studied (the “species-specific themes”), whereas others were unique t o each male in its colony (“individual themes”). Five species-specific themes have been recognized: each theme has specific characteristics and a particular range of variability. The speciesspecific themes are the songs that are mainly used in vocal interactions between males either in the colony or in other social contexts. The speciesspecific themes also give rise to a complex system of dialects. Each of the species-specific themes shows local variations on a scale varying according to the theme, from a few hundred square meters (in the case of the “rhythmic theme”) to more than 900 km2 (“simple theme”). The boundaries between
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
379
areas are not the same for different themes, resulting in a system reminiscent of human language isoglosses (Mundinger, 1982). Playback experiments have shown that males discriminate their own dialect from a strange variant: they respond more often and more quickly t o the familiar dialect (AdretHausberger, 1982). But, even if an unfamiliar variant is broadcast, males do respond by using their own variant of the theme. This suggests that starlings have developed the perceptual ability to recognize the characteristics of each theme. Leppelsack and Vogt (1976) and Miiller and Leppelsack (1985) have shown that some neurons in the auditory field L of starlings, the main receptive area in the brain, are very selective and show a response to only a few species-specific sounds. Hausberger (1993) recently examined how auditory neurons responded to different whistle categories. She found that no neurons in area L responded to a whole whistle, but many responded to particular parts of the whistle. She also found that neighboring neurons responded in similar and often complementary ways. These first results suggest that there are groups of neurons in auditory field L that work together to encode particular sounds. This type of encoding could be a first step in the recognition and categorization process. E. HETEROSPECIFIC VOCALMIMICRY European starlings are renowned as mimics and they will readily reproduce the sounds of other birds and animals and even mechanical sounds (Witchell, 1896; Allard, 1939; Bent, 1950; Hindmarsh, 1984a,b; Hausberger, Jenkins, and Keene, 1991; Eens et al., 1991a). All individuals habitually incorporate alien sounds in their song: Hindmarsh (1984a,b) showed that male starlings sing on average 18 different imitations, although there is a large variation among males, as the range of values is 9 to 31. There is evidence that individual starlings learn their imitations from one another as well as from the original models, as shown by the following observations. First, studying starlings on Fair Isle, Hindmarsh (1984a,b) found that there were local dialects of mimicked sounds. Since this result was not due to an uneven distribution of the model species over the island, it suggests that starlings learn many of their imitations from other starlings. Second, although starlings appear largely to imitate those species commonest in their environment, certain species are represented very much more than their abundance would predict (Hausberger et af., 1991). For instance, in most populations studied the golden oriole, Oriofus oriofus, is one of the most preferred mimicries (Eens, 1984; Hausberger et al., 1991), even in places where this species does not occur and where starlings are sedentary. It is likely that this imitation is learned from other (migratory) starlings rather than from the original model. Finally, studying starlings in aviaries, Eens
380
MARCEL EENS
et al. (1992b) presented direct evidence that starlings can learn hetero-
specific imitations from one another. In their laboratory experiments, Chaiken el al. (1993) also observed that heterospecific imitations in the songs of tutors were copied in the same manner as other song elements. That starlings also copy sounds from the original models follows from the observation that neither in North America nor in Australia or New Zealand d o European mimicries seem to have persisted from the time when starlings were introduced from Britain more than a century ago (Bent, 1950; Hausberger et aL, 1991). Instead, starlings in these introduced populations imitate exclusively sounds from the local environment. Although many hypotheses have been proposed to explain the phenomenon of heterospecific song mimicry, its functional significance remains something of a mystery (Catchpole and Slater, 1995). There is no evidence that starlings preferentially mimic competing species or predators to exclude them from their territories (Hindmarsh, 1986). Hindmarsh (1984b, 1986) proposed that vocal mimicry in starlings may simply result from mistakes in copying and he concluded that there is no additional functional explanation of starling vocal mimicry (see also discussion). TO THE UNDERLYING NEURAL SUBSTRATE IV. RELATING SONGBEHAVIOR
In this section, I adopt an interdisciplinary approach by linking song behavior with its underlying neural substrate. Song learning and production in oscine songbirds are controlled by a discrete network of interconnected forebrain nuclei. This specialized set of brain structures, commonly called the song (control) system, is found only in birds that learn their song with reference to auditory information. In recent years, the European starling has been used as a model to relate behavioral and neurobiological findings in the song system (Ball, 1994; Ball, Bernard, and Casto, 1993; Ball, Casto, and Bernard, 1994; Bernard, Casto, and Ball, 1993; Bernard, Eens, and Ball, 1996; Bernard and Ball, 1995a,b; Casto and Ball, 1994). After a brief description of the neural circuit mediating vocal learning and production in songbirds, I look at the degree of difference in the volume of song control nuclei between male and female starlings. Next, I look at seasonal changes occurring in song control nuclei of male and female starlings in laboratory and natural conditions. Finally, I relate interindividual variation in male song behavior to variation in song nucleus volume. OF VOCAL BEHAVIOR: THESONGSYSTEM A. NEURAL CONTROL
In songbirds, song production and learning are controlled by a discrete network of interconnected forebrain nuclei. Detailed descriptions of this
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
381
song control system can be found elsewhere (Konishi, 1989; Brenowitz, 1991; Doupe, 1993; Doupe and Konishi, 1991); here I only briefly mention the main song control centers and pathways in the brain. The song system is organized into two functional pathways. The main descending pathway that mediates song production includes projections from the higher vocal center (HVc) to the robust nucleus of the archistriatum (RA) in the forebrain, and from R A to the hypoglossal nucleus in the hindbrain, which innervates the muscles of the sound-producing organ, the syrinx. Lesions of these motor nuclei in birds of any age lead to abnormal song production. A second circuit, called the anterior forebrain pathway, is essential for song learning. It consists of area X in the parolfactory lobe of the forebrain and IMAN, which are connected through a dorsolateral nucleus in the thalamus (DLM); this circuit indirectly connects HVc to RA. In contrast to the motor pathway, these song nuclei are not essential for normal song production in adult birds with stable crystallized song (Doupe, 1993). I N THE VOLUME OF SONG CONTROL NUCLEI B. SEXDIFFERENCE
One of the most striking and exciting aspects of the avian song system is the dramatic sexual dimorphism it exhibits and the apparent correlation between the degree of dimorphism in the volume of various song nuclei and the extent of behavioral sex differences in song (Ball et al., 1994). This can be illustrated by looking at the two extremes of sexual dimorphism in behavior. In zebra finches, in which females do not produce any learned vocalizations, several of the song nuclei are approximately four to five times larger in males than females (Nottebohm and Arnold, 1976). Bay wrens, Thryothorus nigricapillus, are a duetting species from the tropics in which male and female both sing in a coordinated fashion and sing in roughly equal amounts, and there sex differences in nuclear volume among the song nuclei are small or nonexistent (Brenowitz, Arnold, and Levin, 1985; Brencwitz and Arnold, 1986). Sex differences in nuclear volume for at least HVc and R A have now been measured in about 10 passerine species (Ball et al., 1994), and there seems to be a reasonable correlation between the extent of behavioral dimorphism and the degree to which the song control nuclei differ in volume. Although male starlings sing more often and have a more complex song than females, recent studies have shown that female starlings sing in a variety of contexts (Hausberger and Black, 1991; Hausberger et al., 1995a,b; Eens and Pinxten, 1996) and probably also with greater frequency than all of the other nonduetting female songbirds in which the dimorphism of the song control system has been studied. Furthermore, female starlings have a complex song in comparison with females of other temperate zone species
382
MARCEL EENS
and they are able to learn songs beyond the first year of life (Hausberger et al., 1995b). Since female starlings appear to exhibit an intermediate level sex difference in singing behavior related to the species that have been studied to date (including the tropical duetting species), Bernard ef al. (1993) expected to find an intermediate level of brain dimorphism in the starling. They found that the volumes of three song control nuclei, as defined by Nissl staining, were all significantly larger in males than in females: HVc, RA, and area X were, respectively, 1.64,1.66,and 1.95 times larger in males than in females (Fig. 6). There was no significant sex difference in the volume of a control nucleus Rt (nucleus rotundus), suggesting that the dimorphisms in the song nuclei reflect a specialized sexually differentiated system and not a general sex difference in brain size. Thus, the extent of dimorphism was not as dramatic as has been reported for zebra finches, but was clearly more pronounced than that found in duetting species (Brenowitz et al., 1985). The intermediate degree of sex difference in behavior therefore seems to be paralleled by an intermediate degree of sex difference
7 ,
6
D =
I
Males Females
T 1
;
$ 3
> 2 -
A:
'M
n
HVc
ires X
FU
Rotundus
Nucleus
FIG. 6. Sexual dimorphism in the volume of song control nuclei in European starlings. Frequency histograms illustrate the means and standard deviations of the volumetric measurements for HVc, area X, RA, and Rotundus in male and female European starlings. The volumes represent the mean of measurements taken from the left and right hemispheres. Asterisks indicate a significant sex difference at a = .05. Reprinted with permission from Wiley-Liss, copyright 1993.
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
383
in the volume of song control nuclei. However, since there is evidence that the song control system is important for the perception of song as well as production, interspecific variability in the volume of song control nuclei may be related to variation in song perception as well as production (Ball et al., 1994). Traditionally, standard histological techniques (i.e., Nissl stains) have been used to define the song nuclei. However, Nissl stains, like other markers of cellular morphology, are limited in the information they provide, and sole reliance on Nissl stains may at times lead to inaccurate, or at least incomplete, conclusions regarding supposedly structurally distinct populations of cells (see Ball et al., 1993, 1994; Bernard et al., 1993, for detailed reviews). Bernard ef al. (1993) and Ball et al. (1994) also measured the volume of area X in starlings as defined via autoradiography for either muscarinic cholinergic receptors or cr2-adrenergic receptors. They found that the densities of both these receptors are higher in several song nuclei (e.g., HVc, area X, and RA) than in surrounding tissues. This differential density allowed them to define clearly the boundaries of the nuclei, and thereby to reconstruct volumes. Nissl staining and receptor autoradiography for both types of receptors indicated the same degree of differences in area X volume between male and female starlings, which demonstrates that receptor autoradiography can be used to assess dimorphisms in nuclear volume. Broad application of this approach to a number of neurotransmitter receptor systems will better characterize the dimorphisms in the song system, and may provide greater insight into the neuroanatomical and neurochemical control of bird song. Using this approach, Ball and colleagues were also able to show that the density of both muscarinic cholinergic receptors and a2-adrenergic receptors in area X did not differ between male and female starlings. Given that the two sexes differ in overall volume of the nucleus, but did not differ in terms of receptor density, male starlings must have more muscarinic and cr2-adrenergic receptors in area X than females. The functional significance of these sex differences is unknown at present; however, given the role of acetylcholine in the formation and maintenance of memories in other vertebrates and the putative role of area X in song learning, it is possible that the sex difference in muscarinic receptor number may modulate the differential ability of acetylcholine to contribute to the acquisition and later production of song in males and females (Ball et al., 1993; Bernard et al., 1993). In contrast to the above studies, a sex difference in D, dopamine receptor density in area X relative to the dorsolateral parolfactory lobe has been detected (Ball et al., 1994; Casto and Ball, 1994). The functional significance of a sex difference in D, receptor density to sexually dimorphic song behavior in starlings is, at
384
MARCEL EENS
present, unknown. Dopamine might act in area X to regulate the sex differences in the propensity of starlings to learn song (Ball et al., 1994).
C. SEASONAL VARIATION IN THE VOLUME OF SONGCONTROL NUCLEI: BETWEEN LABORATORY A N D FIELD STUDIES DISCREPANCIES In temperate zone songbirds, males sing most frequently in spring, both as part of territory acquisition and defense, and for mate attraction and for guarding their mate (Catchpole and Slater, 1995). Males of most species usually sing less frequently at other times of year. Paralleling these behavioral changes are pronounced seasonal changes in the morphology of the neural circuit controlling song learning and production. This relationship was first described in male domestic canaries by Nottebohm (1981). Laboratory-housed males receiving a naturally cycling photoperiod were sacrificed either in spring (long days) when the singing activity was high or in the fall (short days) when the singing activity was low. Using Nissl staining, Nottebohm demonstrated that HVc and RA, two nuclei involved in song production, were substantially larger (respectively, 99 and 77%) in the spring than in the fall in adult male canaries. Subsequent studies in well-controlled laboratory conditions using either a naturally cycling photoperiod or photoperiod manipulations, have shown that there are also “seasonal changes” in song control nuclei in other species (Arai, Taniguchi, and Saito, 1989; Kirn ef al., 1989; Brenowitz, Nalls, Wingfield, and Kroodsma, 1991). Recently, Bernard (1995) and Bernard and Ball (1995a,b) investigated changes in the song control system in male and female European starlings under both laboratory and natural conditions. In a first study, Bernard and Ball (1995b) used photoperiod manipulations in the laboratory to place male starlings into different physiological conditions approximating those of spring and fall to characterize changes in the song control system. Photosensitive male starlings were placed on 11L: 13D or 16L:8D for about 5 months. Birds on 11L: 13D have enlarged gonads and circulating testosterone levels (characteristic of early spring). In contrast, starlings maintained on 16L:8D initially show marked gonadal growth. However, after about 6-8 weeks the birds are photorefractory (i.e., the gonads are regressed and T falls to undetectable levels, a condition characteristic of the late summer and fall). Since recent work has indicated that changes in nucleus volume evident in Nissl-stained tissue are not always apparent when investigated with other histochemical markers (Gahr, 1990), Bernard and Ball used both Nissl-stain and a*-adrenergic receptor autoradiography to characterize changes in the song system. Both markers indicated that the volume of HVc was 44% larger in photosensitive males housed on 11L: 13D than in
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
385
photorefractory males housed on 16L: 8D. Thus, starlings can be added to the growing list of species showing photoperiodic differences in song control nucleus volume. Interestingly, HVc was the only nucleus in which a difference in volume was found. Photoperiodic manipulation did not have any effects on R A and area X volumes. In this regard, starlings are different from all other bird species studied to date, in that all of the major telencephalic song nuclei (thus also R A and area X) appear to increase in volume in response to increased photoperiod (or testosterone) in other species (see Bernard and Ball, 1995b). Furthermore, the percentage change in HVc in starlings was smaller than the amount of change found in other species (e.g., 99% in canaries, Nottebohm, 1981;68% in rufous-sided towhees, Pipilo erythropthalmus, Brenowitz et al., 1991). While this difference may be attributable to species variation in the degree of photoperiodic changes in the song system, it is also possible, however, that the design of the experiment may have led to an underestimate of the full extent of seasonal plasticity of the starling song system. In the 11L: 13D group, the gonads were only about one third to one half the size they reach in April in the field when the birds are breeding (Bernard and Ball, 1995b). The 11L: 13D birds were probably not maximally stimulated, and therefore the effects reported may underestimate the magnitude of effect one might observe under natural conditions. In female starlings, there also appears to be no effect of photoperiodic condition on the volume of area X (Ball et al., 1994). Bernard and Ball (1995a) and Bernard (1995) recently also investigated changes in the song control nuclei of adult male and female starlings under natural conditions. Birds of each sex were trapped each month of the year and information on their reproductive physiology and morphology was obtained. Their study represents the first thorough attempt to characterize seasonal changes in the avian song system under natural conditions. Overall, there were very few significant changes in the volume of the song control nuclei over the course of the year both in male and female starlings. HVc and R A did not change significantly in volume over the course of the year. The only nucleus exhibiting statistically significant plasticity in volume in both male and female starlings was area X. The volume of nucleus pretectalk (a nucleus not involved in song production or auditory information processing) did not change significantly over the year for either males o r females, nor did brain mass vary significantly in either sex. The lack of significant changes in HVc volumes was unexpected. Given the findings that photosensitive males have significantly larger HVc volumes than photostimulated males and that the administration of exogenous testosterone significantly increases the volume of HVc in laboratory conditions (Bernard, 1995), one would predict that HVc would be largest in the field during the breeding season when testosterone levels are highest and the males are
386
MARCEL EENS
photosensitive. Although the values of HVc were largest in April when testosterone levels were highest, overall these values did not differ significantly from other points during the year. There appeared to be two peaks in HVc volume (in April and June) for males. This may reflect the fact that, in double-brooded starlings, testosterone levels increase in males during the egg-laying periods (typically in April and June: Ball and Wingfield, 1987) when males exhibit a peak in singing activity (Eens, Pinxten, and Verheyen, 1994; see later discussion). Just as the seasonal pattern in HVc volume was unexpected, the results of area X were surprising. In both males and females, area X was regressed in volume at two times of year: (1) in January and February in both sexes, and (2) between May and June in females and between May and July in males (Bernard, 1995; Bernard and Ball, 1995a). In general, area X volumes were decreased when circulating testosterone levels were elevated. These results were unexpected in the light of previous laboratory findings for at least two reasons. First, laboratory studies manipulating photoperiodic and hormonal conditions have failed to show significant changes in area X volume (Bernard, 1995). Second, in other species, area X was larger in males on long days with elevated testosterone levels. In the study of Bernard and Ball on free-living starlings, area X was significantly smaller in volume in May than it was at other times of the year even though testosterone levels at this point were higher than at other times of the year. Furthermore, the largest volume of area X was observed in September when day length was decreasing and when testosterone was undetectable. The causes of the discrepancies between the field and the laboratory studies are not known. However, because the song control system has been shown to be modulated by sex steroids both in development and adulthood (see DeVoogd, 1991, for a review), it is very likely that the differences are attributable to differential regulation of hormone secretion in the two contexts (Bernard, 1995). Previous studies have shown that circulating levels of sex steroid and pituitary hormones are lower in laboratory than in field conditions (Dawson and Goldsmith, 1982, 1984; Bernard, 1995). A variety of cues or stimuli in the natural environment that supplement the effects of photoperiod on hormone secretion (e.g.. presence of a nest site, nest mate and competitors, food availability, temperature) are mostly not provided in laboratory studies. Intuitively, one would have expected that in the field, where males are exposed to higher levels of hormones, more pronounced changes would occur (Bernard, 1995). However, for both starlings and redwinged blackbirds (Kirn el af., 1989), significant changes did not occur in the field, but they did in laboratory conditions. A possible explanation suggested by Bernard (1995) is that, in the absence of other salient cues, testosterone can induce changes in the song nuclei. In the field, however,
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
387
where birds are exposed to a variety of cues that can modulate both song and its neural substrate, testosterone plays a less important role in plasticity of song nucleus volume: whereas testosterone is necessary t o increase nucleus volumes in laboratory conditions, large volumes can be maintained in the field in the absence of testosterone. In conclusion, Bernard and Ball's studies clearly demonstrate that laboratory results do not reflect changes as they occur under natural conditions. Their results convincingly indicate the need for more naturalistic studies to put laboratory findings in a realworld context. Bernard and Ball's field data also clearly showed that the degree of sexual dimorphism in the song nuclei is not the same in different seasons. Although all three of the song nuclei measured were larger in males than in females during every month of the year (Bernard and Ball, 1995a), the degree of sex dimorphism in the three song nuclei varied strongly from month to month. For example, in February HVc was 1.88 times larger in males than in females, while in June the ratio was almost twice as large, namely 3.73 (Bernard, 1995). The functional significance of the changes in the volume of the song control nuclei in males and females is still poorly understood. D. AGE-A N D BEHAVIOR-RELATED VARIATION SONGNUCLEI
IN THE
VOLUMES OF
Although sex and physiological condition appear to contribute to differences in the volume of song control nuclei among starlings, these factors alone do not account for all of the interindividual variation. Several studies have attempted to relate differences in song behavior, using mainly repertoire size as a song measure, to variation in song nucleus volume. In male canaries, there appeared to be a strong positive correlation between repertoire size (i.e., number of song syllable types) and the volumes of HVc and RA (Nottebohm et al., 1981). Canady et af. (1984) looked at intraspecific variation in repertoire size and brain space in two distinct populations of long-billed marsh wrens (Cistithorus pafustris) that differ strongly in repertoire size. Marsh wrens on the east coast of North America had significantly smaller repertoire sizes than conspecifics on the west coast. This was also reflected in the volumes of both HVc and RA, which were significantly smaller in the eastern population. Furthermore, there was a strong positive correlation between left HVc volume and repertoire size within each of the two populations. In red-winged blackbirds, however, there was no correlation between repertoire size and HVc and R A volumes (Kim et af., 1989).
388
MARCEL EENS
So far, only one study has attempted to relate the volumes of song nuclei to age-related changes in song repertoire size in adult male songbirds. Repertoire size increases with age in male canaries, but this change in repertoire size was not paralleled by increases in the volumes of HVc and R A (Nottebohm et al., 1981). Recently, Bernard et al. (1996) used male starlings to explore the relationship between song nuclei volumes and (1) age-related differences in song behavior, and (2) interindividual variation in song behavior in adults. In a first study, Bernard et al. (1996) compared the song and the volume of song control nuclei of five yearling and five older males. In agreement with previous studies (Eens et al., 1991b, 1992a), older males had significantly larger repertoire sizes and sang significantly longer song bouts than yearling males. An examination of the volumes of the song control nuclei indicated that HVc and R A were significantly larger in older males than in yearlings (46 and 74% larger, respectively, see Fig. 7). Neither area X nor Pt (nucleus pretectalis, a nucleus not involved in song behavior) differed significantly in volume between the two age groups, although there was a trend for area X to be larger in older than in yearling males (Fig. 7). These results indicate that age-related
1
Yearlings
7
DAdults
1
HVc
RA
AreaX
Pt
FIG.7. Age-related variation in song nuclei in male European starlings. Frequency histograms showing the group mean volumes for HVc, RA, area X, and Pt (nucleus pretectalis) in yearling and older male starlings. Means are based on N = 5 per group, except for HVc for yearlings (N = 4) and area X for older males ( N = 4). Error bars are standard errors. Asterisks indicate age differences at p < .05.
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
389
differences in song behavior are paralleled by neuroanatomical differences. While developmental changes in the volumes of song control regions have been described for other passerines (Nottebohm et al., 1986), these results represent the first demonstration of age-related changes in the song nuclei among adult birds. These results thus indicate that the song nuclei show continued potential for development in adulthood in starlings and these changes may have consequences for behavior. While the above results were obtained using captive males housed in outdoor aviaries, a field study yielded similar results (Bernard, 1995): HVc and R A in June were significantly larger in older males than in yearlings (43 and 41%, respectively), while area X and Pt, again, did not differ between the two age groups. By August, however, the age-related differences in HVc and R A volume were no longer apparent. This transition was attributable to increases in HVc and RA volume in yearlings as well as slight regressions in HVc volumes in older males from June to August. The increases in HVc and R A volumes in yearling males were unlikely to be caused by increases in circulating testosterone, as testosterone levels were undetectable in August (Bernard, 1995). Instead, it seems more likely that these increases reflect increases in “song complexity” or amount of singing in yearlings. Although older males have a larger repertoire and sing longer songs than yearling males, this difference disappears (or becomes smaller) by the second year when yearlings add new songs to their repertoire and increase their song bout length (Eens et al., 1992a,b). Perhaps this process starts to develop between June and August (Bernard, 1995). In a second study, Bernard et al. (1996) examined whether repertoire size and song bout length of male starlings older than one year were correlated with the size of song system nuclei. There were large differences in repertoire size (range: 29-60 phrase types) and average song bout length (range: 18-40 seconds) among the nine males recorded in this study. While repertoire size did not correlate significantly with any of the song nuclei measured, song bout length was significantly positively correlated with HVc and R A volume (Fig. 8). Thus, among older adult starlings, males singing longer song bouts had larger HVc and R A volumes than males singing shorter song bouts. There was also a trend for area X to be positively related to song bout length. After correcting for brain size, HVc and R A volume accounted for 42 and 51%, respectively, of the variance in song bout length. Song bout length is the first song variable other than repertoire size that has been found to correlate with the volumes of song nuclei. This result emphasizes the importance of considering several behavioral parameters when relating behavior to brain morphology. It is likely that in most songbird species, apart from repertoire size, other aspects of song behavior differ between individuals, and these may also contribute to the
390
MARCEL EENS
4-m
JzaE
3-
-f
P >” I
2 -
0
1 10
1.4
0.0
15
20
25
30
35
40
45
1B
j 10
15
20
25
30
35
40
45
Song bout length (s)
FIG.8. Scattergrams showing the positive correlations between song bout length (in s) and the volume of HVc (A) and RA (B) in older male starlings.
observed variability in neuroanatomy. Interestingly, in both studies HVc and RA volume were related to song behavior, but the volume of area X-a nucleus involved in song learning-was not. If the number of songs in memory is related to the amount of brain space dedicated to memory storage as suggested by Nottebohm et al. (1981), then one would expect to find a significant correlation between repertoire size and area X. Such a relationship did not exist for age-related or for interindividual variation in repertoire size. HVc and RA volumes did correlate, however, with differ-
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
391
ences in song bout length in both studies. Since both of these nuclei have been implicated in song production and since there was evidence that males singing longer song bouts spent more time singing (while repertoire size did not correlate significantly with the time spent singing), Bernard et af. (1996) suggested that differences in volume may reflect differences in song output. Clearly, the idea that large volumes of song nuclei may be a result rather than a cause of increased singing behavior needs further investigation.
V. FUNCTIONS OF SONG The functions of male song have been widely studied in a great variety of species and are usually mainly concerned with the defense of territory and/or the attraction of mates (Searcy and Anderson, 1986; Kroodsma and Byers, 1991; Catchpole and Slater, 1995). In recent years, the growing realization that sperm competition plays an important role in the reproductive strategies of birds (Birkhead and Mdler, 1992) has forced researchers to focus also in more detail on the functions of postpairing song (Moller, 1988, 1991). In this section I review the observational and experimental knowledge of the functions of song in the starling. Since male starlings, in contrast to many other temperature zone songbird species, show a remarkably high song output almost all year long except for a brief period during the postbreeding molt (Feare, 1984; Bohner et al., 1990), I first concentrate on the different functions of male song during the breeding season. Then, I briefly discuss possible functions of male song outside the breeding season. Finally, since female starlings have been observed singing both during and outside the breeding season, I focus on possible functions of the female song. OF SONG A. MATE-A~TRACTION FUNCTION
The first indications of the function of song often come from seasonal information or from the pattern of song output during different stages of the breeding cycle (Catchpole, 1982;Johnson and Kermott, 1991; Catchpole and Slater, 1995). For instance, a decrease of male singing activity after pairing has been interpreted as evidence that an important function of song is female attraction (Catchpole, 1973; Espmark and Lampe, 1993). The first qualitative observations on the function of song in the starling came from Kluyver (1933), who noted that unpaired males continue singing throughout the breeding season, whereas paired males stop singing when their female’s clutch is complete. More recently, Eens etaf. (1994) quantified the song production of 10 males before and immediately after pairing.
392
MARCEL EENS
They found a highly significant decrease in singing activity after pairing, suggesting that song is important as a female attractant (see also Fig. 9). The decrease in singing activity of male starlings after pairing is, however, not as dramatic as in male sedge warblers, Acrocephalus schoenobaenus, where males, as soon as they are paired, become silent and do not sing again unless they lose their female (Catchpole, 1973). Male starlings have an increase in singing activity in the period coinciding with the presumed fertile period of their female (Figs. 9 and 11). This clearly shows that mate attraction is not the only function of song in the starling. Paired males that possess more than one nest box and thus have the potential to pair with more than one female decrease their singing activity temporarily after having attracted a female but start singing again at a high song rate at their second nest box after their first female has started laying (Fig. 9C; see also
60
jA
L ' -8
-10
1 a w a
-10
2o0
-4
-2
+m-4 l-*T-o-
0
2
4
6
8
10
12
*-e-e14
16
18
20
22
>4
-8
L
-10
6
L
B
oc-,
Y
'LJ k
-8
I
,
I
,
,
-, s ,' ic-, r--r-
-6
-4
-2
0
2
4
6
8
10
12
14
16
18
6
-4
-2
0
2
4
6
8
10
12
14
16
18
-A
-r
~
-
1
-
3
-20
22
20
22
Day FIG.9. Percentage of time that males spent singing in relation to the onset of egg laying by their female for two monogamous males (A and B) and a polygynous male breeding in a nest box colony in Belgium (C). Day 0 is the day on which the first egg was laid. Open circles indicate that the males are still unmated at that time. The polygynous male (temporarily) deserted his (first) female on day 3 of her laying period and started singing again at a level comparable to that of unpaired males at his second nest box. After Eens et al. (1994). Reprinted with permission from the Belgian Journal of Zoology.
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
393
Pinxten et al., 1989; Eens et al., 1994). This, again, supports the femaleattraction function of song (Temrin, 1986; Johnson and Kermott, 1991). The first experimental evidence on the function of song in starlings came from Cuthill and Hindmarsh (1985), who removed the female from pairs of starlings and observed the effect on male singing behavior. Female removal caused a dramatic increase in the amount of song produced by the male until the original female was returned, again suggesting that female attraction is an important function of song. Cuthill and Hindmarsh also found that already-paired males sang for a much greater proportion of the time when they were alone at the nest than when the female was present. Males that have lost contact with their mate may be singing either t o reattract the mate to the nesthole or to attract a new female. Both possibilities cannot readily be distinguished but in either case represent the use of song in order to attract a mate. In many other species, it has been shown that paired males increase their singing activity when they have lost contact with their female mate (Wasserman, 1977; Slagsvold, Dale, and Saetre, 1994). In agreement with Cuthill and Hindmarsh’s experiments, Henry, Hausberger, Jenkins (1994) also observed that male starlings from which the incubating female is removed start singing again, whereas control males remain silent. Further evidence for a mate-attraction function of song came from aviary experiments of Eens, Pinxten, and Verheyen (1993). They showed that unmated captive male starlings dramatically changed their singing behavior when they were confronted with a female, while presentation of a male stimulus invoked less response. Although resident males increased the time spent close to their nest box both after female and male presentations, they sang significantly more in response to a female than to a male stimulus (Fig. 10). Males also sang more in the nest box during female than during male presentations (Fig. 10). From these experiments, Eens et al. (1993) concluded that the song of male starlings serves an intersexual more than an intrasexual function. The finding that all males sang in their nest box after the introduction of a female into the aviary is an agreement with field observations (Eens et al., 1994). When a female starling approaches an unpaired male, he will direct his displaying activity toward this female by flying into his nest(box) and starting to sing there, apparently trying to entice her toward the nesthole (see also Kluyver, 1933; Cuthill and Hindmarsh, 1985; Pinxten and Eens, 1990). Often males sit and sing in the nest box for minutes as long as the prospecting female remains in the vicinity of the nesthole. The observation that unpaired males sing very vigorously in the presence of prospecting females indicates not only that males sing to be rapidly detected by any potential female passing by (i.e., “passive attraction” sensu Searcy and Anderson, 1986; Slagsvold, Lifjeld, Stenmark,
394
-
MARCEL EENS
2000 -
0control h
; 150u)
0
A
experiment
n CI
I
5m
-d
1m-
c cn
E F
500 -
0
800 -
a
,, 3
u)
-
0control
c m 600 -
B
experiment
E
r
n
400 -
kl
n
E
-92
200 -
0
FIG.10. Mean responses of the “resident” captive males to female and male presentations as compared to controls in: (A) time spent at the nest box (s); (B) total number of phrases sung; (C) number of phrases sung in the nest box; (D) number of phrase types sung. Error bars are standard erros. *, p < .05; **. p < .01. After Eens ef al. (1993). Reprinted with permission from E. J . Brill, copyright 1993.
and Breiehagen, 1988), but also suggests that song characteristics may be used by females as a cue for their mate choice. The most direct evidence that starling song functions in female attraction comes from an experimental study of Mountjoy and Lemon (1991). Using a design similar to that of Eriksson and Wallin (1986), they broadcast natural starling song from speakers attached to nest boxes, while paired boxes with silent speakers served as controls. Mountjoy and Lemon could demonstrate that female starlings were more likely to visit nest boxes from which recorded songs of their species were being broadcast than nest boxes where no sound was played back. The results were very straightforward:
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
395
C
acontrol c m
5
300
-
v) v)
B e
r
200 -
R
L
**
0
100
-
9
5
=
o Female
Male
D
acontrol c
3
w v,
30
Dexperiment
T
Female
Male
FIG. 10. (continued)
not a single female was attracted to the “silent” nest boxes, whereas 12 females were attracted to the “natural song” boxes. Further experimental evidence for a mate-attraction function of starling song comes from Smith (1995). He manipulated the opportunity for males to attract females by presenting some monogamous males with an additional nest box. He found that males having two nest boxes sang more than males with only one nest box, during both the prelaying and the incubation period, and also that these males were visited by prospecting females significantly more. This latter result, however, does not establish the direction of causation: it may be either that males are visited more often by females because they sing more or that males sing more because they are visited or approached more often by females. A proximate explanation for Smith’s result may be that having more nest boxes leads to an increase in testosterone levels (Dittami
396
MARCEL EENS
et af., 1986; Gwinner et af., 1987) and in this way to higher song rates
(Gwinner and Gwinner, 1994). B. TERRITORIAL FUNCTION OF SONG Whereas the mate-attraction function of starling song is very obvious, there is at first sight little evidence that the song has an important function in territory defense. First, song matching between neighboring males occurs only infrequently and it occurs only with some of the whistle song types. Furthermore, there is little evidence that song matching between neighbors has a territorial function, since it is often not preceded by and does not lead to agonistic encounters (Adret-Hausberger, 1982; M. Eens, unpublished data). Second, at the approach of an intruding male, resident males behave aggressively using specific behavioral postures and they eventually attack the opponent, but they only rarely sing. I have never observed in the field a male flying into his nest box to sing at the approach of another male. Hartby (1969) also noted that intruders usually fly off when the owner appears in the nesthole, without any sound being uttered. I have, however, observed several times during daily nest checks in the breeding season that two males were fighting in a nest box and that this fighting was accompanied by singing of at least one of the males. Kluyver (1935) also caught two males that were fighting and singing in a nest box. These findings suggest that singing may also serve an intrasexual function in deterring rivals during close encounters. This is further supported by several observations and by experimental work. First, I observed that captive males that are housed in single-sex groups in the autumn-winter period fight very often during the evening over nest boxes that they use for roosting. These fights, which often take place in the nest boxes, are very often accompanied by singing. Davis (1959, p. 215) suggested that singing is used by male starlings to establish dominance. Second, in autumn and winter, when starlings roost in large flocks, singing occurs very frequently (Bent, 1950; Charman, 1965; Feare, 1984). Within a roost, the distribution of individuals is nonrandom with respect to age and sex: the dominant older males tend to occupy the most preferred positions in the center of the roosts (Summers et al., 1987; Feare, Gill, McKay, and Bishop, 1995). Again, it does not seem unlikely that song is important in establishing dominance hierarchies. Finally, that singing may serve an intrasexual function to deter rivals during close encounters also follows from the aviary experiments of Eens, Pinxten, and Verheyen (1990, 1993). When they confronted unpaired male starlings housed in a large aviary (6 X 4 X 2.5 m; length X width X height, respectively) with a male stimulus, they did not observe a strong increase in singing activity and none of the eight males sang in its nest box (Eens et
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
397
al., 1990). However, when males housed in small aviaries (1.75 X 1.35 X 2 m) were confronted with a male stimulus, they tended to increase their singing activity and 4 out of 11 males sang in their nest boxes (Eens et al., 1993). The more intense intrasexual encounter brought about in the smaller aviary (in which the introduced male was always closer than 1 m to the resident male’s nest box) may account for the observed differences in song response between the two studies.
C . OTHERFUNCTIONS OF SONG Although Hindmarsh (1984a) claimed that male starlings stop singing almost completely when they get paried, Eens et al. (1994) showed that the singing activity of males does not stop entirely at pairing and, furthermore, that there is a strong increase in singing activity in the period coinciding with the presumed fertile period of the female (see Fig. 9). Considering the period after pair formation, male starlings sing at a significantly higher rate during the fertile period of their female (defined as the period from five days before laying to the laying of the penultimate egg; Mgller, 1985) than during the nonfertile period (Fig. 11; R. Pinxten and M. Eens, unpublished data). In agreement with this, Kluyver (1933) also mentioned that paired male starlings sing most during the period of egg laying, which is a rough approximation of the fertile period. At the physiological level, Ball and Wingfield (1987) found that the plasma levels of testosterone of male starlings peak during the egg-laying stage of their female. Several hypotheses, which are not always mutually exclusive, may explain why mated males sing at a high frequency during the fertile period of their female (GreigSmith, 1982; Mace, l986,1987a,b; Mdler, 1988,1991; Catchpole and Slater, 1995). I discuss the most relevant hypotheses in detail here and also comment on other hypotheses that try to explain why birds continue to sing after having acquired a mate. A first hypothesis that tries to explain the high song activity during the fertile period is that song is important in the context of sperm competition, that is, in the competition between the sperm from two or more males to fertilize the eggs of a single female during one reproductive cycle (Birkhead and Mdler, 1992). This hypothesis has been elaborated by M ~ l l e (1988, r 1991) partly on the basis of earlier work by Greig-Smith (1982) and Mace (1986, 1987a,b). Although Meller (1988) considered only two alternatives when he developed the “sperm competition hypotheses,” I think that three alternatives should be considered. First, males may sing at a high rate during their own female’s fertile period to incite her to copulate at a high frequency, since frequent copulation is an efficient paternity guard (“own female incitation to copulate hypothesis”). This alternative was not dis-
398
MARCEL EENS
06
i
05 m E 0 c
i
’4
04d .
C Q)
u) Cl
-
03
4-
0
9
T I
-6
-4
-2
0
2
4
7
6
0
10 12 > I 3
Day in female cycle
FIG.11. Average male singing rate in relation to day of female nesting cycle, where day 0 is the day on which the first egg is laid. Data shown are mean 5 SE. Numbers above bars indicate sample sizes. From R. Pinxten and M. Eens (unpublished data).
cussed by Mgller (1988). Second, males may sing at a high rate during their female’s fertile period as a mate-guarding strategy to prevent other males from invading and obtaining extrapair copulations with their mate (“male deterrence hypothesis”). Third, males may sing to attract neighboring females to obtain extrapair copulations (“female attraction hypothesis”). Although there is strong evidence in favor of the sperm competition hypothesis in several species (Mace 1987a; Mgller 1988,1991; Cuthill and Macdonald, 1990;Welling, Koivula, and Lahti, 1995), it has been difficult to conclude whether a male singing during the fertile period of his female is trying to deter rival males, enticing his own female to copulate, or attracting other females to obtain extrapair copulations. I consider each of these possibilities in detail. In the starling, both extrapair copulation behavior (Eens and Pinxten, 1990,1995; Wright and Cotton, 1994) and extrapair paternity have been reported (Hoffenberg, Power, Romagnano, Lombardo, and McGuire, 1988; Pinxten et al., 1993; Smith and von Schantz, 1993). Several lines of evidence suggest that an important function of the song after pairing in the starling might be to invite and to stimulate the bird’s
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
399
own female to solicit copulations (Eens et ul., 1994). First, nearly all withinpair copulation solicitations (99%; 282 of 286) are preceded by singing of the male mate (Eens and Pinxten, 1995). Furthermore, all within-pair copulation solicitations are female solicited in the starling. Usually the male is singing in the vicinity of his nest box; the female then solicits copulation by flying toward the singing male and adopting a submissive posture close to him. The female often pecks the male while he is still singing (Hartby, 1969; Feare, 1984; Goude, 1988; Eens and Pinxten, 1990, 1995). I have never observed a male flying toward his own female to initiate a copulation. The observation that all copulation solicitations are preceded by singing of the male mate and that it is always the female that flies to the singing male, strongly suggests that inviting and stimulating the female to solicit copulations is an important function of male song. I know of no other songbird species where the relationship between singing of the male and the solicitation of copulations by the female is so obvious and direct in field conditions. Hartby wrote in 1969: in the starling the very complex song of the male has the function of inviting and stimulating the female to copulations, as well as attracting a female to the nest-site in the first place. So the male does not need a special nest-site call or other courtship calls, as they are found in many species of bird.
Indeed, it is interesting to point out that male starlings, in contrast to female starlings and to many other male songbirds, d o not have a copulation call or precopulatory vocalization. In many species, other vocalizations besides the song may have sexual functions. For instance, Baker and Baker (1988) showed that a precopulatory vocalization is necessary in conjunction with song to elicit courtship in female buntings of two species. Similarly, Searcy (1989) demonstrated that mixed bouts of song and precopulatory calls elicit more female solicitation than bouts of song alone in red-winged blackbirds (Ageluiusphoeniceus). That the song of male starlings during the fertile period is mainly directed at their own female also follows from the observation that males sing at a significantly higher rate when their female is inside her nest box than when she is outside the nest box (R. Pinxten and M. Eens, unpublished data). In our opinion, the higher male song rate when the female is in the nest box is related to the fact that males do not know for certain when their female is going to lay her egg in any given day during the fertile period. Indeed, in contrast to most other species, female starlings do not lay their eggs at dawn, but usually between 9:OO and 11:OO in the morning (Feare, Spencer, and Constantine, 1982; Meijer, 1992). As a result, it is much more difficult for males to time their copulations to coincide with the “fertilization
400
MARCEL €ENS
window.” According to the fertilization-window hypothesis (Cheng, Burnes, and McKinney, 1983), inseminations that occur in the hour following egg laying are most likely to be successful in fertilizing the next egg. As all within-pair copulations are female solicited in the starling (i.e., under complete female control), and since females solicit copulations only when their male is singing, the safest strategy for males to maximize the chance that their female will solicit a copulation shortly after having laid her egg is to sing whenever she is in her nest box (since this may be a cue that their female is going to lay). Further strong evidence in favor of this comes from the fact that males sing significantly more between 9:00 and 12:OO A.M. than between 6:OO and 9:OO A.M. (R. Pinxten and M. Eens, unpublished data). I am unaware of any other songbird species where males sing more late in the morning than early. Apparently, male starlings sing most when their female is in her period of peak diurnal fertility. When discussing the role of song in sperm competition, Mace (1986) predicted that, in species that lay at other times of day than at dawn, peaks of song should occur just prior to these times. The results found in the starling seem to support her prediction. Detailed observations in other (songbird) species where egg laying does not occur at dawn are necessary to confirm the generality of this behavioral pattern. In conclusion, there is strong evidence that male starlings sing at a high rate during their female’s fertile period in an attempt to maximize their number of within-pair copulations. Is there evidence that male starlings sing at a high rate during the fertile period of their female to attract nonmate females to copulate? During a 2-year field study of the copulation behavior of individually marked females and males in a nest box colony, we observed 12 extrapair copulation attempts compared to 291 within-pair copulation solicitations (Eens and Pinxten, 1995). All extrapair copulation attempts were initiated by the males. In all cases, paired males approached nonmate neighboring females and started singing very close (< 0.5 m) to them. Only once did the female solicit copulation, and a successful extrapair copulation then occurred. Although song seems necessary to stimulate nonmate females to engage in an extrapair copulation, during hundreds of hours of observation we have never observed females approach nonmate singing males to solicit an extrapair copulation. We can, however, not exclude the possibility that such behavior occurs outside the nest box colony. This finding does suggest, however, that males need to approach potential extrapair females closely and then start singing, rather than that “undirected” song attracts females. The result that males sing more when their female is in the nest box is not predicted by this hypothesis. The observation that monogamous males almost entirely cease singing after their mate’s laying period (Fig. 9) is also not consistent with the hypothesis that males sing to obtain extrapair
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
401
copulations (see also Mace, 1987b; Slagsvold et al., 1994). This is especially true, as there are always fertile females (i.e., later arriving females that mostly become secondary females of polygynous males) present in the colony after the synchronized laying period (Pinxten, Eens, and Verheyen, 1990). Furthermore, Birkhead, Fletcher, Pellat, and Staples (1995) recently showed that the best time for males to perform extrapair copulations is outside their own-pair copulation period, because in this way males maximize both the quality and number of sperm in ejaculates. Overall, the evidence in favor of the “female attraction hypotheses” is not strong. Do observations support the hypothesis that males sing during the fertile period of their female as a mate-guarding strategy to avoid being cuckolded? When trying to answer thisquestion it isimportant torecall that male starlings guard their female intensively during her fertile period: both members of a monogamous pair spend more than 90% of their time together in the colony (Pinxten and Eens, in press) and monogamous males follow their female nearly always when she leaves the colony or flies out of sight (Pinxten, Van Elsacker, and Verheyen, 1987; Smith, 1995). Since starlings defend only a very small territory consisting of a nesthole and a few meters around it, males are always very close to their own female. Therefore, it is almost impossible for a female to solicit an extrapair copulation (either initiated by herself o r by a nonmate male) when her male is present. In agreement with this, all 17 extrapair copulations observed by Wright and Cotton (1994b) and all 12 extrapair copulation solicitations observed by Eens and Pinxten (1995) occurred in the absence of both birds’ mates. Eens et al. (1994) suggested that, in a semicolonially breeding songbird such as the starling, the presence of the male close to the female is more important than his song in deterring other males. Furthermore, the fact that males sing more when their female is in the nest box suggests that the song is directed to their own female rather than to other (neighboring) males. There is, however, some experimental evidence that singing by males can affect the probability of direct challenges from opponents. Mountjoy and Lemon (1991) found that more males visited nest boxes where “simple” songs were broadcast than nest boxes where “complex” starling songs were played. However, in these experiments, no live o r stuffed males were present when song was broadcast. Overall, observations favor the idea that males sing at a high rate during their female’s fertile period to entice their female to copulate rather than to attract nonmate females or to deter other males. If song rate is a reliable cue as regards male quality (Moller, 1991), then it is not unlikely that this information is used by all potential receivers including neighboring males and females, and floaters, even if it is not primarily directed at them. A peak of song activity during the fertile period may also be important to stimulate the last stages of ovarian development and ovulation (Moller,
402
MARCEL EENS
1988). Evidence in support of this female-reproduction hypothesis has been found in several species. For instance, it has been shown that male song is important in stimulating the ovarian cycle and that song may positively affect the nest building activity of the female (Brockway, 1965; Hinde and Steel, 1976; Kroodsma, 1976; Morton, Pereyra, and Baptista, 1985; Logan, Hyatt, and Gregorcyk, 1990). The seasonal and die1 pattern of singing described above is not entirely consistent with this hypothesis because it predicts that the singing activity is high during both the prefertile and fertile period, which is not the case in the starling. Furthermore, the femalereproduction hypothesis does not predict a higher song rate during the late morning nor does it predict a higher song rate when the female is in her nest box. Nevertheless, there is circumstantial evidence that male song has a stimulating or synchronizing function in the starling. First, Feare (1984) suggested that the highly synchronous breeding in females of the same colony (all females producing usually first clutches within one week: Feare 1984; Pinxten et al., 1990) is brought about by social stimulation through singing. This suggestion is now supported by recent aviary experiments of Meijer and Langer (1995). To unravel the mechanisms influencing egg laying in the starling, they examined how the behavior of food-rationed starlings affected the timing of reproduction of ad libitum-fed birds with which they had visual contact. In both years of their study, Meijer and Langer found that ad libitum-fed starlings that could not see food-rationed birds laid earlier than a d libitum-fed birds that had visual contact with food-rationed birds. Since food-rationed males showed almost no courtship (i,e., singing) behavior, this may have delayed the start of egg laying. Note that the stimulation function of song described here is considered at the colony level rather than at the individual (pair) level. Studying a breeding colony in a British resident population for a month prior to egg laying, Wright and Cotton (1994b) also observed that the whole colony was involved in either singing activity around the nest boxes or feeding as one big flock in fields. The second finding in agreement with the female reproduction hypothesis comes from a field study of Wright and Cuthill (1992). They found that males who sang more were paired to females who laid earlier. Although it is mechanistically plausibe that female laying date could be advanced through high song rates per se, alternative explanations are possible. Clearly, the circumstantial evidence in favor of this hypothesis begs for experimental manipulations of song output in controlled laboratory conditions. Johnson and Kermott (1991) suggested that the song of male house wrens, Troglodyres aedon, after pairing functions mainly in informing female mates that there is no immediate threat of predation. Our results are inconsistent with this hypothesis, since monogamous male starlings completely stop
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
403
singing at the end of the laying period. Furthermore, this hypothesis does not predict a higher song rate late in the morning (between 9:OO and 1290 A.M.),since there is no evidence that the risk of being predated is higher at that time than at other periods of the day. Many of the intersexual functions of song described above imply that females recognize their mates’ song. There is circumstantial evidence that female starlings are capable of recognizing and responding to their mates’ song. When studying the copulation behavior of starlings, Eens and Pinxten (1995) observed many times that a primary female left her nest box and flew directly to her male who was singing at his second nest box (5-30 m away from the first nest box) presumably to attract a second female. As the females could not see their male before flying out of the nest box, this strongly suggests that females can recognize their males on the basis of their song. Such female recognition may be very likely, as each starling male has an almost unique song repertoire. In most polygynous bird species including the starling, primary females receive less male help in feeding nestlings than monogamous females (Pinxten and Eens, 1994). Since this is costly, paired females should try to prevent or delay the settlement of other females. A strategy that primary female starlings may use in this respect is to disturb the singing of the male mate (Eens and Pinxten, 1995, 1996; see also Temrin, 1989; Slagsvold and Lifjeld, 1994). In such a conflict situation, it is very likely that selection would favor females that could learn their mates’ song quickly. In the polygynous pied flycatcher, Ficedula hypofeuca, Slagsvold, Amundsen, Dale, and Lampe (1992) also described how primary females are able to locate their mate when he is singing in a distant territory to attract a second female. More recently, Lampe and Slagsvold (1994) carried out playback experiments in the field and found that female pied flycatchers are able to discriminate between the song of their mate and that of foreign males after an extremely short period of song exposure. Wiley, Hatchwell, and Davies (1991) also demonstrated that female dunnocks, Pruneflarnodularis, can discriminate individual males by their songs alone. OF SONGOUTSIDE D. FUNCTIONS
THE
BREEDING SEASON
Although many authors have observed that male starlings sing almost all year round except for a brief period during molt (Feare, 1984; Bohner et al., 1990; Eens, 1992), the seasonal variation in song output has not been quantified in detail and surprisingly little is known on the function(s) of song outside the breeding season. Before dealing with the functions of song, it might be interesting to look briefly at the hormonal control of song. During the breeding season, song is directly influenced by circulating
404
MARCEL EENS
hormones in the starling; castration of a male dramatically reduces the singing activity, while the rate of singing in castrated males is increased after testosterone administration (M. Eens and R. Pinxten, unpublished data). Outside the breeding season, however, testosterone concentrations are low or undetectable (Ball and Wingfield, 1987; Bernard, 1995), suggesting that song is less dependent on androgens. In an interesting experiment, Calhoun, Hulse, Braaten, Page, and Nelson (1993) showed that song perception, as well as song production, may change with the photoperiod in male and female starlings. As mentioned previously, song might be important in establishing dominance relationships in flocks or in roosts (see later discussion). Studying the singing behavior of white-crowned sparrows on their wintering grounds, DeWolfe and Baptista (1995) also noticed that much singing is often heard prior to roosting and may be involved in squabbling for best roosting spots and maintaining individual distances. In many cases, however, song seems to be used in a nonaggressive situation in the starling. Adret-Hausberger (1982) observed foraging flocks of starlings of different sizes in October and found that starlings sing one preferential whistle type depending on the number of individuals present. Starlings sang mostly the “harmonic whistle theme” when they were in groups of 25-50 individuals, “the inflection theme” when in flocks of 50-100 individuals, and the “simple theme” when they were in flocks of more than 150 individuals. Another whistle type, the “rhythmic theme,” was mainly sung in roosts. Hausberger and Black (1991) suggested that frequent song matching between males in flocks might enable them to recognize who belongs to the same dialectal unit, thus adjusting their behavior in social encounters. In single-sex groups of captive males, song can be elicited in autumn by the introduction of a female. Although such a female introduction resulted in male agonistic interactions in the breeding season, this was not the case in autumn: very often males sat close together on the same perch or nest box while singing (M. Eens and R. Pinxten, unpublished data). It is unclear why males sing at this time, but it may not be unlikely that, especially in resident populations where pair formation can take place long before egg laying (see Merkel, 1980), song produced in autumn and winter may influence female mating decisions. Hausberger et al. (1995b) also suggested that the social organization in the nonbreeding context may play a role in mating decisions during the breeding season. Finally, it may be possible that in the months prior to breeding, males sing to self-stimulate hormone production and gonadal recrudescence (see Brockway, 1967; Cheng, 1992).
E. FUNCTIONS OF FEMALE SONG Song in passerine birds has traditionally been seen as an exclusively male activity, especially at temperate latitudes. In recent years, however, female
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
405
song has been reported in many species. The view that female song is a functionless result of abnormally high levels of androgens can no longer hold (Cooney and Cockburn, 1995; Langmore et al., 1996). In the starling, early observations by Freitag (1936,1937,1939) described how female song often occurred in an intrasexual context in the breeding season. These observations have recently been confirmed by field and aviary experiments. First, in a nest box colony in Sweden, Sandell and Smith (in press) examined intraspecific aggressive behavior of breeding female starlings by exposing them to a simulated intrusion of a conspecific bird: a caged male or female starling was placed close to the nest of a breeding pair. They found that female song was addressed almost exclusively toward female intruders. Females sang toward intruding females only during the prelaying and the laying period and not during the incubation and nestling period. From their experiments, Sandell and Smith concluded that female song and aggressive behavior are associated with maintaining monogamy. Second, Eens and Pinxten (1996) recently examined how captive female starlings that had an exclusive male partner responded if their male started courting a newly introduced female at a second nest box. From these experiments, it appeared that female starlings try to prevent prospecting females from mating with their mate in a variety of ways including aggressive behavior, copulation solicitation behavior, and song. Eens and Pinxten found that after the introduction of a second female to the aviary, 5 out of 14 females sang while they were blocking the entrance hole of the second nest box, whereas none of the 14 females was observed singing during the control periods. Both studies indicate that female starling song occurs under conditions of high female-female competition. The results of both experiments are consistent with studies on several other species, which have proposed that territorial defense is an important function of female song (Arcese, Stoddard, and Hiebert, 1988; Hobson and Sealy, 1990; Cooney and Cockburn, 1995). In agreement with this, I have observed several times in the field, when two females were fighting in a nest box, that at least one of them was singing (M. Eens and R. Pinxten, unpublished data). Hausberger and Black (1991) made observations on the two females of a polygynous trio during the breeding season. Both females were observed singing: most of the song occurred in song-matching bouts between the two females or between the females and the male. In this case no aggressive behavior between the females was observed. Several authors have observed female song outside the breeding season (Kluyver, 1933; Witschi and Miller, 1938; Feare, 1984; Hausberger et al., 1995a,b). According to Hausberger et al. (1995a,b), female song occurs more often in autumn and winter than during the breeding season. Nothing is known about the possible functions of song outside the colony, in the flocks or roosts.
406
MARCEL EENS
VI. FUNCTIONS A N D EVOLUTION OF THE SONG REPERTOIRE
In the starling, as in most other songbird species, individual males possess “repertoires” of multiple versions of the species’ song. Explaining the function(s) of repertoires has been an important focus of song research (Krebs and Kroodsma, 1980 Searcy and Anderson, 1986 Kroodsma and Byers, 1991; Searcy, 1992a). A number of hypotheses have been proposed to explain the evolution of song repertoires in songbirds (Krebs and Kroodsma, 1980). In this section, I review the available evidence in the starling and I also discuss variation in other aspects of singing behavior. A. VARIATION I N REPERTOIRE SIZEAND OTHER SONG TRAITS In the starling, like in most other songbird species, song-type classifications typically are arrived at by sorting spectrograms of songs according to relatively subjective criteria of similarity and dissimilarity (Kroodsma, 1982; Searcy, Podos, Peters, and Nowicki, 1995). In a continuously singing species such as the starling, where pauses, if any, between successive renditions of the same phrase type or between successive phrase types are often very small, different researchers may use different classification methods. As a result, it may be difficult to assess whether differences between populations are real or due to different classification methods used. Repertoire sizes of starlings have been determined in five different populations (Table 11). Overall, measurements of repertoire size do not appear to differ very much among them. Adret-Hausberger and co-worker(s) did not include the whistles when determining the repertoire sizes in German and French male starlings, and this may explain why their measures are lower than the others. At the other extreme, Chaiken et al. (1993) were probably more “splitters” than “lumpers”: they counted any detectable different phrase
TABLE I1 COMPARISON OF REPERTOIRE SIZERANGEI N SIXDIFFERENT POPULATIONS OF THE EUROPEAN STARLING Population
Repertoire size
Reference
Belgium Canada France Germany United Kingdom United States
14-68 24-65 11-36 12-55 21-55 28-90
Eens er al. (1991a; unpublished data) Mountjoy and Lemon (1995, 1996) Adret-Hausberger and Jenkins (1988) Adret-Hausberger et al. (1990) Hindmarsh (1984a) Chaiken er al. (1993)
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
407
as a new type (M. Chaiken, personal communication), and this may explain why their repertoire sizes are larger than the others. In all studies, there appear to be very large differences in repertoire size among individuals. In each of the populations considered repertoire sizes can vary almost threefold, or more, between males. I have also looked at differences in song bout length, and found very large differences in average length among males: they range from 12 to 40 s (average = 24.5 s; SD = 6.0; N = 49). There is a highly significant positive correlation between repertoire size and song bout length (Fig. 12). B. DIFFERENT SONGSHAVE DIFFERENT FUNCTIONS Patterns of song use have revealed that in some species different song types convey different messages and that males sing multiple song types so that they can communicate each of the messages. In several species, separate intersexual and intrasexual functions of distinct song types have been suggested (Catchpole, 1980, 1982; Capp and Searcy, 1991; Spector, 1991). Based on qualitative observations of the singing behavior of starlings,
40 45
I
0
“‘“1 4 25
v1
a,
0
4
10
1 . 10
20
I
I
I
I
I
30
40
50
60
70
Repertoire size FIG.12. Scattergram showing the positive correlation between repertoire size and average song bout length (in seconds) in male European starlings ( r = +.81,p < .0001, N = 49). From M. Eens and R. Pinxten (unpublished data).
408
MARCEL EENS
Adret-Hausberger and Jenkins (1988) proposed that the complex song bouts (“warbling songs” in their terminology) of starlings function mainly in mate attraction while the more simple whistles that are sometimes used in male-male interactions have a more intrasexual function. However, on the basis of a quantitative experimental study, Eens et al. (1993) did not find evidence for a specialized intrasexual function of the whistles nor for a specialized intersexual function of the complex song bouts. Individually caged males did not sing more whistles when presented with a male than with a female. Furthermore, the fact that complex song bouts were also sung when males were confronted with a male stimulus argues against the use of song bouts as an exclusive intersexual signal. However, since we have studied mainly migratory starlings while Adret-Hausberger and her co-workers have studied mainly sedentary ones, it is not impossible that the observed differences reflect differences between populations. It is possible that males in resident populations are more likely to share whistles with neighbors than are those from migratory populations, and therefore matched countersinging may occur more frequently in resident populations (see Ewert and Kroodsma, 1994). C. SONGREPERTOIRES A N D MATECHOICE
When Eens et al. (1990,1993) looked at the use of song and of the song repertoire in male-female versus male-male contexts (Fig. lo), they found that males produced more song and sang more phrase types when females (as compared to males) were introduced into the aviary. Male starlings thus seem to display their repertoire primarily to females rather than to males. Since field observations also showed that males sing vigorously in the presence of prospecting females, investigation of the intersexual effects of repertoires seemed necessary. To test whether song characteristics influence female mate choice in starlings, Eens er al. (1991b) studied a colony of starlings breeding in nest boxes, and related female preferences to male repertoire size and average song bout length. The date of initial pair formation was used as a measure of female preferences (see Catchpole, 1980; Searcy, 1984). They found that males that sing longer song bouts and that have larger repertoires attracted females earlier than males with shorter song bouts and smaller repertoires. Choice experiments were then carried out in large outdoor aviaries, containing six identical nest boxes and three resident males. Female choice appeared to be nonrandom with respect to repertoire size and average song bout length. Again, females selected the males with the larger repertoires and the longer song bouts. There was no evidence that female choice was
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
409
influenced by morphological characteristics such as weight and tarsus length (Eens et al., 1991b). In the starling, where males defend only their nesthole and not food resources, territory quality is reduced to the quality of the nesthole. Although Eens et al. (1991b) tried to reduce the variation in nest box attractiveness as much as possible (by using identical nest boxes and putting them at a similar height whenever possible), it remains possible that factors such as the orientation of the nest box might have influenced the pairing decisions of females. Mountjoy and Lemon (1996) recently tested the hypothesis that female starlings choose mates on the basis of the complexity of their songs, rather than on the quality of the “territories” the males defended. In order to determine whether certain “territories” were preferred over others, the first set of males to settle in a series of nest boxes was removed, and a second set was allowed to settle. Despite the fact that all the nest boxes were of the same design, Mountjoy and Lemon found consistent preferences for certain nest boxes as indicated by correlations between the settlement patterns of the first and second sets of males. However, males with the largest repertoire size did not necessarily occupy the most preferred nest sites. In agreement with the results of Eens et al. (1991b), Mountjoy and Lemon found that males with larger repertoires acquired females faster than males with smaller repertoire sizes. Mountjoy and Lemon did not use pairing date as a measure of mate preference, but instead used the number of days between the date when a nest box was claimed and the date when the first egg was laid. They found a highly significant negative correlation between repertoire size and this measure (Fig. 13). This relationship remained significant when nest site preference was statistically controlled for, indicating that female starlings chose males with complex song rather than those that defended preferred nest sites. This is only the second field study to demonstrate a correlation between repertoire size and female preferences after controlling for possible effects of territory quality. Mountjoy and Lemon’s data also indicated that female preference for males with complex song did not result from a correlation with any of several morphological traits: male body weight, the lengths of culmen, tarsus, and wing, and the length of the iridescent portion of the throat hackle feathers were all unrelated to the measure of female preference, or the relationship did not remain significant when repertoire size was controlled in a partial correlation. They did find, however, that repertoire size was significantly positively related to a measure of condition (see Fig. 14) and that the number of days between nest box occupation and first egg date (i.e., the measure of female preference) was negatively correlated with this condition index. Partial correlations did indicate, however, that condition is not as important as repertoire size in explaining female choice. If female starlings
410
MARCEL EENS
30
1
h
21 v)
m 3
\ @
h
@@
10
5
1 20
25
K
I
I
I
30
35
40
v 45
m 50
55
Repertoire size
FIG.13. Size of song phrase repertoire versus the delay (in days) between the date a nest box was claimed and the date when the first egg was laid ( r = -.625. one-tailed p = .006,
N
=
15). Reprinted with permission from Springer-Verlag. copyright 1996.
cannot measure male condition directly, the preference for males with larger repertoires may help them obtain a high-quality mate, since repertoire size is a good predictor of condition. When Mountjoy and Lemon (1991) were able to show that more female starlings were attracted to nest boxes with playback of starling song than to silent controls, they subsequently also tried to test whether females preferred more complex song over simple song. Although more females were captured in boxes with playback of complex song than in boxes where simple song was played (three vs. zero), too few females entered boxes to show a clear preference for either song stimulus. In recent years, the technique of implanting females with estradiol has been used in laboratory conditions to enhance their willingness to give copulation solicitation displays in response to song stimuli (Catchpole, 1987; Searcy, 1992b). Most studies on species having song repertoires have found that females respond more to playback of larger repertoires (Searcy, 1992a,b). The interpretation of these results has been complicated by the findings that estradiol-treated females respond more to repertoires even in two species where field studies were unable to detect any influence of repertoires on female choice (McGregor, Krebs, and Perrins, 1981; Searcy,
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
55 50 45
8
.-
In a, L
1
a
a
1
411
a
I
40
.-
;
35
a
a, K
30
-
i
25 20 -0.06
I
a I
-0.04
I
-0.02
a I
1
0
0.02
I
0.04
I
0.06
Condition
FIG.14. Scattergram showing the positive correlation between condition index (using the residuals from regressing the log of weight against the log of tarsus) and repertoire size in male European starlings ( r = +.455.one-tailed p = .015,N = 23). Reprinted with permission from Springer-Verlag. copyright 1996.
1984), and that implanted females respond more to repertoires even in species where males have only one song type (Searcy, 1992a). Nevertheless, the solicitation display assay is a powerful technique for measuring female responses (Searcy, 1992b). Both Hindmarsh (1984a) and Eens (1992) have attempted to apply this technique to starlings, but in each case this was unsuccessful, as none of the female subjects responded with copulation solicitation postures to playback of conspecific song. As suggested by Searcy (1992b), several factors may account for this failure: perhaps the female subjects were aware of the presence of human observers, or song should have been combined with other stimuli such as a live male. Goude (1988) did find, however, that female starlings injected with estradiol responded with copulation solicitations when confronted with male song. Although she did not give much information on the song stimuli that were used in her experiments, she found that female starlings responded significantly more to “complex” songs than to “less complex” songs. As such, Goude’s results suggest that large repertoires are more effective than smaller repertoires in stimulating females to court and copulate and her results are in agreement with the field and aviary data discussed earlier.
412
MARCEL EENS
Further evidence that females prefer males with a more complex song came from the field study of Eens et af. (1991b). In their population, where polygyny occurs frequently, repertoire size and average song bout length were significantly and positively correlated with the number of females attracted. There also was a strong, positive correlation between both measures of song complexity and male breeding success. Finally, it is not unlikely that song characteristics may be important in mate choice in the context of extrapair copulations. In recent years, it has become apparent that females of many species of birds often copulate with, and produce young sired by, males other than their social mate (Birkhead and Meller, 1992). While in many species males visiting females in search of extrapair copulations do this in a furtive and sneaky way, this does not appear to be the case in male starlings. Instead, they approach nonmate females very closely and then start singing, apparently to invite copulation (Eens and Pinxten, 1990, 1995). It is possible that the “quality of song” might be important for a female to decide whether or not she will copulate with a nonmate male. In the great reed warbler, Acrocephafus arundinaceus, females seek extrapair fertilizations from neighboring males with larger song repertoires than their social mate (Hasselquist, 1994; Hasselquist, Bensch, and von Schantz, 1996). In the starling, monogamous males stop singing when their female’s clutch is completed (Fig. 9) and share incubation duties about equally with their mate (Pinxten and Eens, 1994), while polygynous males (which tend to have larger repertoire sizes) continue singing to attractkourt additional females after the fertile period of their first female (Pinxten and Eens, 1994). As a result, it is also likely that polygynous males engage more in extrapair copulations and this could mean that the variation in breeding success is even larger than is counted by the number of fledglings produced from the male’s own nests. D. SONGREPERTOIRES AND TERRITORIAL DEFENSE In contrast to most other songbird species, the intrasexual effects of the song repertoire have been studied less than the intersexual in the starling. The reason for this is probably that it is difficult to investigate the intrasexual effects of the repertoire in a species where males defend only an extremely small territory, where most of the song seems to be directed to females, and where males do not react (aggressively) to the playback of the complex song bouts (Adret-Hausberger, 1982;Adret-Hausberger and Jenkins, 1988). Mountjoy and Lemon (1991), however, have shown experimentally that repertoire size can affect intraspecific competition for breeding resources. They demonstrated that playback of “complex song,” which includes a larger number of phrase types, is more effective than “simple song” in
UNDERSTANDINGTHE COMPLEX SONG OF THE EUROPEAN STARLING
413
deterring males from entering a nest box. This result suggests that song complexity may function as an indicator of male quality and may be used by male starlings to assess potential competitors. Mountjoy and Lemon’s results have stimulated us to investigate the relationship between song repertoire size and aggressive dominance in male starlings in captivity (M. Eens and R. Pinxten, unpublished data). Aggressive dominance was measured as the ability to defeat other males in encounters in captive birds, and expressed as the percentage of interactions won. To eliminate possible effects of age, only yearling males were used in this experiment. As predicted by the results of Mountjoy and Lemon, we found a significant positive relationship between repertoire size and the percentage of encounters won in a captive group of 11 yearling males (Fig. 15). This strongly suggests that repertoire size reflects a male’s competitive ability. Mountjoy and Lemon’s findings, together with the results of our aviary study, suggest that song repertoire size may reflect some aspect of male quality and that listening birds could use that information. From our aviary study, it is unclear, however, whether song is actually important in the establishment of dominance relationships, especially since
80
c
70
1I
0
0
g 60 a,
4-
c
3
50
w
8
40
2
0
1
I
I
I
I
I
I
I
I
I
I
16 18 20 22 24 26 28 30 32 34 36 38 40
Repertoire size
FIG. 15. Size of song phrase repertoire versus the percentage of encounters won for 1 1 captive male yearling starlings ( r = +.572, one-tailed p = ,032, N = 11). From M. Eens and R. Pinxten (unpublished data).
414
MARCEL EENS
intrasexual encounters are not often accompanied by song. This would require further experimental work such as devocalizing males, as has been done in brown-headed cowbirds, Molothrus ater (Dufty, 1986). Our results do suggest, however, that potential intruders, even at long distances, could use repertoire size to determine the dominance position and to avoid costly interaction or potentially costly escalated disputes. There was also a positive relationship between song bout length and dominance, but this correlation was nonsignificant (r = +.325, df = 9, one-tailed p = .165). It is unclear why song bout length, which was correlated with repertoire size, was not correlated with aggressive dominance. E. SONGA N D SEXUAL SELECTION
In this section and the previous one, I have reviewed observations and experiments that attempted to tease apart the relative importance of the two basic functions of the song and the song repertoire in the starling. Overall, the song and song repertoire appear to be more important for attracting and stimulating females than for territory defense, which suggests that intersexual selection has been more important than intrasexual selection in the evolution of the complex song of the starling. However, although the song seems to be aimed mainly at females and females seem to favor males with larger song repertoires, there is also evidence that large song repertoires give an advantage in repelling other males. Therefore, it cannot be ruled out that intrasexual selection may also have acted in shaping the complex song, especially since we can examine only present function (current utility) and can therefore not be sure about evolutionary origins (Kroodsma and Byers, 1991; Searcy, 1992a; Catchpole and Slater, 1995). When reviewing the intersexual effects of repertoires in different songbird species, Searcy (1992a) concluded that laboratory experiments show consistent preferences for large repertoires in female songbirds, while field observations argue that, in general, repertoire size has at most weak effects on female settlement patterns. In the starling, mating advantages of larger repertoires in laboratory and aviary conditions are supported by field results obtained in two different populations. In this respect, the results in starlings are comparable to those obtained in sedge warblers (Catchpole, 1980; Catchpole, Dittami, and Leisler, 1984). When interpreting this, it may be important to stress that both sedge warblers and starlings (at least those of which settlement patterns were studied in the field) are migratory. Perhaps mate and territory fidelity are more important in (more) resident species, such as, for example, the great tit and the song sparrow, than in migratory species, and this may hinder “optimal” mate choice by females in the field in the two latter species. Furthermore, both sedge warblers and
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
415
starlings have small territories with few resources. Females may therefore base their choice on a high-quality male rather than on a high-quality territory, and one aspect of this may be song structure, leading eventually to increased song complexity (Catchpole and Slater, 1995). In this section I have thus far presented evidence in support of the idea that variation in male repertoire size (and song bout length) affects mating success. Female choice is the most likely mechanism leading to variation in male mating success: males with larger repertoires attracted females earlier, obtained more females, and produced more young. Knowing that female starlings are stimulated more by “complex song” than by “simple song” in laboratory conditions (Goude, 1988), it is not unlikely that males with a complex song also obtain more extrapair copulations and fertilizations, leading to an even higher variance in male mating success. Next, we need to consider why female starlings have developed a preference for males with elaborate song repertoires. Three major hypotheses have been proposed to explain the evolution of female preferences for large repertoires or for exaggerated mating displays in general: direct selection, the runaway process, and selection for “good genes” (Kirkpatrick and Ryan, 1991; Andersson, 1994; Johnstone, 1995). Is there any evidence that female mate choice for large repertoires can be explained by advantages in the form of resources provided by the male (territory, parental care, etc.)? First, the results of Mountjoy and Lemon (1996) indicated that the most preferred nest sites were not occupied by males with more complex song. It is therefore unlikely that the preference for large repertoires may benefit females by helping them obtain males with better “territories.” Second, the preference for large repertoires may benefit females by helping them to obtain males who will provide “superior” parental care. This is not plausible for starlings. Mountjoy and Lemon (1996) found that, in their Canadian population where polygyny is extremely rare, there was no evidence that males with large repertoires provide more parental care. In our Belgian population, where polygyny is frequent (Pinxten et al., 1989; Pinxten and Eens, 1990), males with the most complex song are most likely to become polygynous (Eens et al., 1991b), and these males provide less parental care than monogamous males (Pinxten et af., 1993; Pinxten and Eens, 1994). Third, it is possible that male nest building behavior is an important component of paternal care in starlings. In starlings, only males bring green nest material to the nest. Males select for inclusion in their nest only a small subset of the available plant species in their habitat and these preferred species contain secondary plant compounds that reduce pathogen and ectoparasite populations (Clark and Mason, 1985, 1988). In the presence of a female, males often pick green nesting material and drop it into the nest apparently to entice the female to enter, and males often
416
MARCEL EENS
hold it in their beaks and vocalize while the female inspects the nest (Feare, 1984; Fauth, Krementz, and Hines, 1991; Eens et af., 1990, 1993, 1994). Bringing green nest material to the nest thus is a sexually selected behavior that may be used by females for mate selection. Clark and Mason (1985) mentioned that first-year breeding males are less selective in the choice of green nest materials than older males. Furthermore, starlings probably use chemical cues to differentiate among plants that act as pesticides, and seasonal changes in olfactory sensitivity are probably controlled by gonadal steroids (Clark, 1991). Therefore, it remains possible that females use song complexity as a means of identifying males that are able to select the best plant species. Extreme ornaments such as large repertoires can theoretically arise from female choice alone without the ornament bearing any relation to the viability of the male. According to this Fisherian explanation of sexual selection, the female preference evolves together with the preferred male trait in a process that becomes self-reinforcing (Anderson, 1994). It can therefore bring the male trait to an extreme stage of development until the costs of possessing it counter the benefit of increased mating success. Alternatively, indicator or “good genes” models propose that traits preferred by females are reliable indicators of genetically heritable variation in male quality; indirect selection then acts on female preferences because discriminating females tend to produce more viable offspring (Andersson, 1994). So far, there is little empirical support in favor of Fisherian runaway models, since there is no evidence that song repertoires or female preferences for them are inherited (Catchpole and Slater, 1995). Furthermore, several findings obtained in the starling make it unlikely that a Fisherian process involving female choice for an arbitrary character is responsible for the evolution of the complex song (see also Mountjoy and Lemon, 1996). First, song complexity is not arbitrary with respect to age, since yearling males have smaller repertoires than older males and repertoire size can still increase after the second year of life. Females may benefit from choosing older males because they have demonstrated their survival ability and are likely to be of higher than average genetic quality. Second, the positive correlation between repertoire size and condition obtained by Mountjoy and Lemon (1996) points to song being a condition indicator rather than an arbitrary trait. This positive correlation is of interest with regard to the suggestion that sexual ornaments may serve as indicators of male viability. This correlation also suggests that the development of a larger repertoire may be energetically costly, and that only males in good condition can develop an extensive repertoire. Alternatively, it may be that older males with a higher repertoire size have an increased foraging efficiency. Third, the observed positive relationship between song repertoire
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
417
size and male competitive ability, and the observation that increased song complexity reduces the probability that an intruding male will enter a nest box (Mountjoy and Lemon, 1991), suggest that a Fisherian explanation is not plausible in the starling. It seems unlikely that an arbitrary trait evolved through a Fisherian process would be used to resolve important competitive interactions if real asymmetries exist between the males involved (Mountjoy and Lemon, 1996). Finally, although there may be some underlying genetic variation that influences a male’s repertoire size, the finding that repertoire size increases with age strongly suggests that there is not a strong correlation between the potential genetic variation and the expression of the trait (Mountjoy and Lemon, 1996). This will make it difficult for a strong genetic covariance to be established between the trait and the female preference, as is required in most Fisherian runaway models (Andersson, 1994).Overall, it seems unlikely that the starling’s complex song has evolved by a Fisherian process. Instead, several findings suggest that the evolution of song complexity in the starling is in agreement with indicator models of sexual selection where age and condition are revealed by repertoire size (and/or song bout length). To be a reliable cue of male (genetic) quality, repertoire size should be costly to produce or possess (Andersson, 1994). The results of Mountjoy and Lemon (1996) suggest that repertoire size may be a conditiondependent ornament and develop in proportion to the phenotypic quality of the male (see Andersson, 1994). Furthermore, intraspecific variation in song repertoire size has been found to be related to variation in the size of the higher vocal center (HVc), suggesting that a large repertoire size may be costly in terms of brain space (Nottebohm et al., 1981). In the starling, average song bout length was significantly related to the size of HVc, but repertoire size was not. This might suggest that a long complex song is costly to produce, although alternative explanations are possible (Bernard et al., 1996). Although there are at present no data available for the starling, it seems likely that the ultimate target of the female preference is to confer a heritable component of viability to her offspring. Several studies have demonstrated that male ornamentation may correlate with offspring survival (Norris, 1993; von Schantz, Grahn, and Goransson, 1994; Petrie, 1994), and Hasselquist et al. (1996) has shown that the relative postfledging survival of offspring was positively correlated with their genetical fathers’ song repertoire size in the great reed warbler. AND FUTURE DIRECTIONS VII. CONCLUSIONS
During the last fifteen years, interest in starling song has increased exponentially (Eens, 1992). The cosmopolitan occurrence of European starlings,
418
MARCEL EENS
the fact that they are relatively easy to keep in laboratory and aviary conditions, their complex but nevertheless quantifiable song, and their unusual song learning behavior have probably all contributed to their increased popularity as a model system. Unlike most other songbird species, European starlings modify their songs in adulthood. Furthermore, the European starling is one of the few songbird species for which there is evidence that males increase their song repertoire size after the first breeding season. The other species are the redwinged blackbird (Marler, Mundinger, Waser, and Lutjen, 1972;Yasukawa, Blank, and Patterson, 1980;but see Kroodsma and James, 1994), the domestic canary (Nottebohm and Nottebohm, 1978), the yellow warbler Dendroica petechia (Cosens and Sealy, 1986), the northern mockingbird, Mimus pofygfotros(Derrickson, 1987), the pied flycatcher (Espmark and Lampe, 1993), and the great reed warbler (Hasselquist, 1994). The available evidence indicates that the increase in repertoire size continues even after the second breeding season in the starling (Mountjoy and Lemon, 1995) and probably even later (Fig. 5 ) , while this is less clear in the other species (but see Nottebohm et al., 1986; Hasselquist, 1994). Recent data have shown that female starlings are also age-independent learners that can change their repertoire and acquire new songs in adulthood (Hausberger et af., 1995a,b). Since singing in most songbird species is done predominantly by males, most research on song and its neural basis has focused on males. In particular, there have been few studies on song learning in females, as most of the indices for measuring the timing, nature, and amount of this learning depend on eventual song production (DeVoogd, 1994). Starlings could prove to be a useful model for studying female song learning behavior. Because both male and female starlings can continue to learn song as adults, there is a large temporal window within which mechanisms underlying the regulation of song learning can be studied. Starlings differ from most species mentioned previously in that yearling males have dramatically smaller repertoire sizes than older males: this pattern has been found in all populations that have been studied in detail (see Table I). At present, it remains unclear why such a pattern is found only in starlings. Perhaps the fact that juvenile starlings, in contrast to most other temperate zone songbirds, undergo a complete postjuvenile molt may be of importance. The observation that yearling males tended to have smaller volumes of area X than older males might suggest that there is a constraint on the capacity of song memory that changes with age. Finally, it may be of importance that juvenile and yearling starlings are more often infected with endoparasites than older starlings (Feare, 1984; Bernard, 1987). Maybe males can invest heavily in the development of their repertoire only after they have acquired some kind of immunity (Folstad and
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
419
Karter, 1992). Starlings that have anomalously high parasite loads (Hamilton, 1990, p. 343) would be ideal subjects to test the Hamilton and Zuk (1982) hypothesis that within a species we should find a negative correlation between parasite load and measures of song complexity (see Read and Weary, 1990). The seasonal time course of adult song modification in wild starlings is still largely unknown at present. In adult canaries, most of the changes occur after the end of the breeding season (during summer and early fall) and immediately before the onset of the next breeding season in late winter (Nottebohm et al., 1986). These times correspond roughly to periods when the males’ testosterone levels are low or beginning to rise again after having been low for several months (Nottebohm eta/., 1987). It would be interesting to investigate whether wild or aviary starlings modify their song during late summer and fall when area X volumes are highest and testosterone levels are low (Bernard, 1995; Bernard and Ball, 1995a). The neurobiological data presented in Section IV emphasize that it would be useful to (1) expand the number of species studied in neurobiological studies, (2) examine animals under more naturalistic conditions, and (3) increase the number of song variables when looking for correlations between behavior and brain measures. Although a recent study has attempted to examine the relationship between repertoire and brain nuclei size across 41 songbird species (DeVoogd, Krebs, Healy, and Purvis, 1993), most of the research on the song system has been completed on a limited number of species. As a result, the generality of many findings in the literature has not been well established. Just as there is an enormous variation in singing patterns among species, there might be large interspecific variation in the song system (and in the factors influencing it). For instance, Bernard and Ball (1995b) showed that photoperiodic manipulation in the starling did not have any effects on RA and area X volumes in laboratory conditions, while in all other species studied so far these nuclei increased in volume in response to increased photoperiod (or testosterone). The finding that the form and magnitude of volumetric changes in the song control nuclei of starlings differ between laboratory and natural conditions (Bernard 1995; Bernard and Ball, 1995a) must urge other researchers to look at seasonal changes occurring in the wild to validate laboratory results with field observations. Finally, the finding that variation in the length of song bouts sung by adult male starlings was significantly related to the volume of HVc and RA, while repertoire size was not related to either volume, stresses that song variables other than repertoire size should also be taken into account in neurobiological studies. Since there was evidence that males singing longer song bouts also spent more time singing, it will
420
MARCEL EENS
be intriguing to determine whether large volumes are a cause or consequence of increased singing. Although observations and experiments suggest that the song and the song repertoire of male starlings serve an intersexual more than an intrasexual function, there is also evidence that both of them have an intrasexual component. In this respect, the starling differs from the sedge warbler, another species with a highly complex song. In sedge warblers females clearly prefer males with larger repertoires (Catchpole, 1980; Catchpole et af., 1984), but repertoire size has no effect on measures of approach shown by responsive males during playback (Catchpole, 1989). Although the present evidence indicates that female starlings prefer males singing larger repertoires and having longer song bouts, it may still be possible that other song variables are correlated with repertoire size and/or song bout length, and that these are the characteristics that females actually prefer. For instance, none of the field or aviary studies focusing on female mate choice have controlled for the singing activity of the males, which may be important in starlings (Wright and Cuthill, 1992) and other species (e.g., Alatalo, Glynn, and Lundberg, 1990;Wasserman and Cigliano, 1991). Since the carrying of green nest materials and wing-waving (a visual display often associated with singing) are also sexually selected behaviors (Eens etal., 1993), ideally, these behavioral variables should also be quantified and controlled for in studies that investigate the importance of song in female mate choice. It is likely that female choice of mate depends on the assessment of more than one character. It seems reasonable to assume that repertoire size and song rate may give different kinds of information to prospecting females: while repertoire size probably reflects an individual’s quality at the time song memorization took place, song rate is more likely to reveal a male’s quality at the moment of singing. Both song rate and song bout length may be more sensitive to short-term changes in condition than repertoire size. Field and aviary as well as laboratory experiments clearly indicate that female starlings prefer males having larger repertoire sizes and singing longer song bouts. As discussed earlier, repertoire size in starlings may be a reliable or honest signal of quality since it correlates significantly with age, condition, and dominance position (see Figs. 5,14, and 15). European starlings differ from many other songbird species in this respect because measures of song complexity usually have not been found to correlate with male quality (Searcy, 1984; Catchpole, 1987; but see Hasselquist, 1994, and Lampe and Saetre, 1995). The positive correlation between repertoire size and condition obtained by Mountjoy and Lemon (1996) suggests that repertoires are costly to develop or maintain. Further evidence for this comes from a recent aviary study that investigated whether there was a cost of
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
421
breeding in yearling starlings (M. Eens and R. Pinxten, unpublished data). This found that yearling males that bred in their first breeding season had a smaller increase in repertoire size in the following year than yearling males that did not breed. This strongly suggests that male song repertoire size is a condition-dependent trait that is sensitive to a rough measure of reproductive effort in the previous year, at least in yearlings. In recent years, it has become clear that sperm competition and mate guarding may have played a more important role in the function and evolution of song than previously suspected (Birkhead and Mdler, 1992). After pair formation, male starlings have a clear peak in song activity during their female’s fertile period. Furthermore, males sing most when their female is in her period of peak diurnal fertility between 9:OO and 12:OO A.M. than in the early morning. Observations suggest that males sing mainly to stimulate their own female to solicit copulation at this time rather than to ward off rival males or to attract other females to obtain extrapair copulations. Studies using DNA profiling are needed to investigate the real reproductive success of male starlings with different qualities and quantities of song. While there has been much interest in the functions of song in the starling and in other species, the costs of producing song have been largely neglected (see Anderson and Iwasa, 1996). Very few studies have measured the energy requirements of song display behavior in wild birds. Other possible costs of singing that should be taken into account are the increased conspicuousness of singers to predators, the announcement of the nest site to intraspecific brood parasites, and the attraction of competing males by song (Mountjoy and Lemon, 1991; Stamps, 1994). The pressure to acquire a large repertoire might also account for the starlings’ accomplishments as a heterospecific mimic. Hindmarsh (1986) argued that, if there is strong selection for a complex song, then the penalty for learning song elements of a wrong species is likely to be much less than that for species with relatively simple songs. If diversity is what matters, then the broad pattern of rhythm and syntax will become more important in the process of species recognition rather than the detailed form of the different notes. This evolutionary scenario agrees with the finding that, although starling song is highly individual, the general organization of song seems to be similar all over the world wherever it has been studied. If a song’s effectiveness is less dependent on the detailed form of the individual song components, then the criteria for selecting sounds for song learning (restrictions on the timing of learning, the type of sound that is learned, and on the models learned from) can, and probably must, be relaxed. A combination of relaxed selection criteria and song effectiveness independent of detailed note morphology will then almost inevitably give rise to
422
MARCEL EENS
some mimicked elements being incorporated into the song (Hindmarsh, 1986). Slater er al. (1988) also predicted that song adapted primarily as a mate attractant would be likely, through sexual selection, to become more varied and elaborate, would not necessarily be copied accurately, and might well be modified throughout life.
VIII. SUMMARY Bird song is a unique model system for the integrated study of the development, neurobiology, and functions of learned vocal communication signals. The complex song of the European starling has recently received much attention. In this chapter I review (1) the development of song in this species, (2) neurobiological findings that have linked song behavior to the underlying neural substrate, (3) the different functions of the song, and (4)the functions of the song repertoire. Both field and laboratory studies indicate that extended song learning occurs in male starlings: the sensitive period for song learning is extended, and perhaps even lifelong. The ongoing process of song learning results in a correlation between repertoire size and age in male starlings. Although female starlings can also acquire new songs in adulthood, age-related differences in repertoire size have not been found. Social interactions are shown to play an important role in the acquisition of song in both males and females. Discrepancies between laboratory and field results suggest that the conditions of laboratory housing do not expose birds to the conditions that play instructive roles in nature. Female starlings, which sing less than males and have a less complex song, have significantly smaller song control nuclei than males. Apart from sex and physiological condition, age and song complexity also contribute to differences in the volume of song control nuclei among starlings. Volumetric changes in the song control nuclei occur under natural conditions, but the form and the magnitude of the changes are different from what would have been predicted in light of laboratory results. Observations and experiments suggest that the most important function of male starling song is to attract females. There is, however, also evidence that the song and the song repertoire are used in an intrasexual context. After pair formation, male starlings have a clear peak in song activity during their female’s fertile period, suggesting that song is important in the context of sperm competition. Observations suggest that males sing mainly to stimulate their own female to solicit copulation at this time rather than to ward off rival males or to attract other females to obtain extrapair copulations.
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
423
There are large differences in repertoire size among males, ranging from 15 to 70 phrase types. Field and aviary as well as laboratory experiments indicate that female starlings prefer males with larger repertoire sizes. Repertoire size may be a reliable or honest signal of quality, since it correlates significantly with age, condition, and competitive ability. In many other songbird species measures of song complexity have not been found to correlate with the quality of the singer. By contrast, the complex song of the starling is most probably a sexually selected trait that reliably advertises some aspect of a male’s quality. Acknowledgments
I am grateful to Clive Catchpole, Manfred Milinski, Rianne Pinxten, and Peter Slater for comments on the manuscript. I thank Rianne Pinxten for help with making the figures and for her continuous support. 1 thank Dan Bernard, Jorg Bohner, MarthaLeah Chaiken, Jim Mountjoy, Maria Sandell, and Henrik Smith for providing details of unpublished work. Funding from the Belgian National Fund for Scientific Research, the University of Antwerp, and from the University of Newcastle (by an EC-grant to Marion Petrie) is gratefully acknowledged.
References Adret-Hausberger, M. (1982). Social influences on the whistled songs of starlings. Behav. Ecol. Sociobiol. 11, 241-246. Adret-Hausberger, M. (1983). Variations dialectales des sifflements de I’Etourneau sansonnet (Sturnus vulgaris) sedentaire en Bretagne. Z. Tierpsychol. 62, 55-71. Adret-Hausberger, M. (1989). The species-repertoire of whistled songs in the European starling: species-specific characteristics and variability. Bioacousrics 2, 137-162. Adret-Hausberger, M., and Giittinger, H. R. (1984). Constancy of basic patterns in the songs of two populations of starlings (Sturnus vulgaris). A comparison of song variation between sedentary and migratory populations. 2. Tierpsychol. 66,309-327. Adret-Hausberger, M., and Giittinger, H.-R., and Merkel, F. W. (1990). Individual life history and song repertoire changes in a colony of starlings (Sturnus vulgaris). Ethology 84, 265-280. Adret-Hausberger, M., and Jenkins, P. F. (1988). Complex organization of the warbling song of starlings. Behaviour 107, 138-156. Alatalo, R. V., Glynn, C., and Lundberg, A. (1990). Singing rate and female attraction in the pied flycatcher: an experiment. Anirn. Behav. 39,601-603. Allard. H. A. (1939). Vocal mimicry of the starling and the mockingbird. Science 90,370-371. Anderson. M. (1994). “Sexual Selection.” Princeton University Press, Princeton, New Jersey. Anderson. M., and Iwasa, Y.(1996). Sexual selection. Trends Ecol. Evol. 11, 53-58. Arai, A., Taniguchi, I., and Saito, N. (1989). Correlation between the size of song control nuclei and plumage color change in orange bishop birds. Neurosci. Leu. 98, 144-148. Arcese, P., Stoddard, P. K., and Hiebert, S. M. (1988). The form and function of song in female song sparrows. Condor 90,4450. Baker, M. C., and Baker, A. E. M. (1988). Vocal and visual stimuli enabling copulation behaviour in female buntings. Behav. Ecol. Sociobiol. 23, 105-108.
424
MARCEL EENS
Ball. G. F. (1994). Neurochemical specializations associated with vocal learning and production in songbirds and budgerigars. Brain Behav. Evol. 44,234-246. Ball, G . F., Bernard, D. J., and Casto, J. M. (1993). Neurochemistry of the avian vocal control circuit: hormonal modulation and sex differences. Poultry Sci. Rev. 5, 15-27. Ball, G. F., Casto, J. M., and Bernard, D. J. (1994). Sex differences in the volume of avian song control nuclei: comparative studies and the issue of brain nucleus delineation. Psychoneuroendocr. 19,485-504. Ball, G . F., and Wingfield, J. C. (1987). Changes in plasma levels of luteinizing hormone and sex steroid hormones in relation to multiple-broodedness and nest-site density in male starlings. Physiol. Zool. 60, 191-199. Baptista, L. F., Bell, D. A., and Trail, P. W. (1993). Song learning and production in the white-crowned sparrow: parallels with sexual imprinting. Nerh. J. Zool. 43, 17-33. Baptista, L. F., and Petrinovich, L. (1984). Social interaction, sensitive phases and the song template hypothesis in the white-crowned sparrow. Anim. Behav. 32, 172-181. Baptista, L. F., and Petrinovich, L. (1986). Song development in the white-crowned sparrow: social factors and sex differences. Anim. Behav. 34, 1359-1371. Barnea. A., and Nottebohm, F. (1994). Seasonal recruitment of hippocampal neurons in adult free-ranging black-capped chickadees. Proc. Nail. Acad. Sci. U.S.A. 91, 11217-1 1221. Bent, A. C. (1950). Life histories of North American Birds. U.S. Nut. Mus. Bulls. Smithsonian Inst. Washington. Bernard, D. J. (1995). The effects of testosterone, photoperiod, and season on plasticity in the song control system of European starlings (Sturnus vulgaris). Unpublished doctoral dissertation, Johns Hopkins University. Bernard, D. J., and Ball, G. F. (1995a). Seasonal changes in the volume of song control nuclei in free-living male and female European starlings. SOC.Neurosci. Absrr. 21, 961. Bernard, D. J., and Ball, G. F. (1995b). Two histological markers reveal a similar photoperiodic difference in the volume of the high vocal center in male European starlings. J. Comp. Neurol. 360,726-734. Bernard, D. J.. Casto, J. M., and Ball, G. F. (1993). Sexual dimorphism in the volume of song control nuclei in European starlings: assessment by a Nissl stain and autoradiography for muscarinic cholinergic receptors. J. Comp. Neurol. 334, 559-570. Bernard, D. J., Eens, M., and Ball, G. F. (1996). Age- and behavior-related variation in the volumes of song control nuclei in male European starlings. J. Neurobiol. 30, 329-339. Bernard, J. (1987). Helminthes parasites de I’etourneau sansonnet (Sturnus vulgaris L.) dans son aire de nidification. Archs. Inst. Pasteur Tunis 64,477-499. Bertram, B. (1970). The vocal behaviour of the Indian hill mynah. Anim. Behav. 3,81-192. Birkhead, T. R., Fletcher, F., Pellat, E. J., and Staples, A. (1995). Ejaculate quality and the success of extra-pair copulations in the zebra finch. Nature (London) 377,422-423. Birkhead, T. R., and Meller, A. P. (1992). “Sperm Competition in Birds.” Academic Press, London. Btihner, J. (1993). Der Gesang des Stars (Sturnus vulgaris): Untersuchungen zum Aufbau und Enverb. Sber. Ges. Naturj Freunde Berlin 32, 133-147. Bahner, J., Chaiken, M., and Marler, P. (1990). Song acquisition in photosensitive and photorefractory male European starlings. Hormon. Behav. 24, 582-594. Btihner, J., and Todt, D. (1996). Influence of auditory stimulation on the development of syntactical and temporal features in European starling song. Auk 113,450-456. BBhner, J., and Veit, F. (1993). Gesangsstruktur und Formen der FlUgelbewegung beim Star (Sturnus vulgaris). J . O m . 134,309-315. Brenowitz, E. A. (1991). Evolution of the vocal control system in the avian brain. Semin. Neurosci. 3,399-407.
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
425
Brenowitz, E. A., and Arnold, A. P. (1986). Interspecific comparisons of the size of neural song control regions and song complexity in duetting birds: evolutionary implications. J. Neurosci. 6, 2875-2879. Brenowitz, E. A,, Arnold, A. P.. and Levin, R. N. (1985). Neural correlates of female song in tropical duetting birds. Brain Res. 343, 104-1 12. Brenowitz, E. A,, Nalls, B., Wingfield, J. C., and Kroodsma, D. E. (1991). Seasonal changes in avian song control nuclei without seasonal changes in song repertoire. J. Neurosci. 11,1367-1374. Brockway, B. F. (1965). Stimulation of ovarian development and egg-laying by male courtship vocalizations in budgerigars, Melopsittacus undulatus. Anim. Behav. 13, 575-579. Brockway, B. F. (1967). The influence of vocal behavior on the performer’s testicular activity in budgerigars (Melopsittacus undulatus). Wilson Bull. 79, 328-334. Burger. J. W. (1953). The effect of photic and psychic stimuli on the reproductive cycle of the male starling, Sturnus vulgaris. J. Exp. Zool. W, 227-239. Calhoun, S., Hulse, S . H., Braaten, R. F., Page, S. C., and Nelson, R. J. (1993). Responsiveness to conspecific and alien by canaries (Serinus canaria) and European starlings (Sturnus vulgaris) as a function of photoperiod. J. Comp. Psychol. 107,235-241. Canady, R. A,, Kroodsma, and Nottebohm, F. (1984). Population differences in complexity of a learned skill are correlated with the brain space involved. Proc. Nail. Acad. Sci. U.S.A. 81,6232-6234. Capp, M. S . , and Searcy, W. A. (1991). An experimental study of song type function in the bobolink (Dolichonyx oryzivorus). Behav. Ecol. Sociobiol. 28, 179-186. Casto, J. M., and Ball, G .F. (1994). Characterization and localization of D1 dopamine receptors in the sexually dimorphic vocal control nucleus, area X, and the basal ganglia of European starlings. J. Neurobiol. 25, 767-780. Catchpole, C. K. (1973). The functions of advertising song in the sedge warble and reed warbler. Behaviour 46,300-320. Catchpole, C. K. (1980). Sexual selection and the evolution of complex songs among European warblers of the genus Acrocephalus. Behaviour 74, 149-166. Catchpole, C. K. (1982). The evolution of bird sounds in relation to mating and spacing behaviour. I n “Acoustic Communication in Birds” (D. E. Kroodsma and E. H. Miller, eds.), Vol. 1, pp. 297-319. Academic Press, New York. Catchpole, C. K. (1987). Bird song, sexual selection and female choice. Trends Ecol. Evol. 2, 94-97. Catchpole. C. K. (1989). Responses of male sedge warblers to playback of different repertoire sizes. Anim. Behav. 37, 1046-1047. Catchpole, C. K., Dittami, J., and Leisler, B. (1984). Differential responses to male song repertoires in female songbirds implanted with oestradiol. Nature (London) 3l2,563-564. Catchpole, C. K., and Slater, P. J. B. (1995). “Bird Song: Biological Themes and Variations.” Cambridge University Press, Cambridge. Chaiken, M., Bohner, J., and Marler, P. (1993). Song acquisition in European starlings (Sturnus vulgaris): a comparison of the songs of live-tutored, tape-tutored, untutored and wildcaught males. Anim. Behav. 46, 1079-1090. Bohner, J., and Marler, P. (1994). Repertoire turnover and the timing of song acquisition in European starlings. Behaviour 128,25-39. Charman, K. (1965). Unpublished doctoral dissertation, University of Durham. Cheng, K. M., Burnes, J. T., and McKinney, F. (1983). Forced copulation in captive mallards. 111. Sperm competition. Auk 100, 302-310. Cheng, M. F. (1992). For whom does the female dove coo? A case for the role of vocal selfstimulation. Anim. Behav. 43, 1035-1044.
426
MARCEL EENS
Chisholm, A. H. (1932). Vocal mimicry among Australian birds. Ibis 13, 605-625. Clark, L. (1991). The nest protection hypothesis: the adaptive use of plant secondary compounds by European starlings. In “Bird-Parasite Interactions: Ecology, Evolution and Behaviour” (J. E. Loye and M. Zuk, eds.), Vol. 1, pp. 205-231. Oxford University Press, Oxford. Clark, L., and Mason, J. R. (1985). Use of nest material as insecticidal and anti-pathogenic agents by the European starling. Oecologia 67,169-176. Clark, L., and Mason, J. R. (1988). Effect of biologically active plants used as nest material and the derived benefit to starling nestlings. Oecologia 77, 174-180. Clayton, N. S. (1987). Song tutor choice in zebra finches. Anim. Behav. 35,714-721. Cooney, R., and Cockburn, A. (1995). Territorial defence is the major function of female song in the superb fairy-wren, Malurus cyaneus. Anim. Behav. 49, 1635-1747. Cosens, S. E., and Sealy, S. G. (1986). Age-related variation in song repertoire size and repertoire sharing of yellow warblers (Dendroica perechia). Can. J. Zool. 64, 1926-1929. Cramp, S., and Perrins, C. M. (1994). “The Birds of the Western Palearctic,” Vol VIII. Oxford University Press, Oxford. Cuthill, I., and Hindmarsh, A. M. (1985). Increase in starling song activity with removal of mate. Anim. Behav. 33,326-328. Cuthill, I., and Macdonald. W. A. (1990). Experimental manipulation of the dawn and dusk chorus in the blackbird Titrdus merula. Behav. Ecol. Sociobiol. 26,209-216. Darwin, C. (1871). “The Descent of Man and Selection in Relation to Sex.” Murray, London. Davis, D. E. (1959). Territorial rank in starlings. Anim. Behav. 7, 214-221. Dawson, A., and Goldsmith, A. R. (1982). Prolactin and gonadotropin secretion in wild starlings (Sturnus vulgaris) during the annual cycle, and in relation to nesting, incubation, and rearing young. Gen. Comp. Endocr. 48, 213-221. Dawson, A., and Goldsmith, A. R. (1984). Effects of gonadectomy on seasonal changes in plasma LH and prolactin concentrations in male and female starlings (Srurnus vulgaris). J. Endocr. 100, 213-218. Derrickson, K. C. (1987). Yearly and situational changes in the estimate of repertoire size in northern mockingbirds (Mimrts polyglorros). Auk 104, 198-207. DeVoogd. T. J . (1991). Endocrine modulation of the development and adult function of the avian song system. fsychoneuroendocr. 16,41-66. DeVoogd, T. J. (1994). The neural basis for the acqu on and production of bird song. In “Causal Mechanisms of behavioural development” (J. A. Hogan and J. J. Bolhuis, eds.), pp. 49-81. Cambridge University Press, Cambridge. DeVoogd, T. J., Krebs, J. R., Healy, S. D., and Purvis, A. (1993). Relations between song repertoire size and the volume of brain nuclei related to song: comparative evolutionary analyses amongst oscine birds. Proc. R. SOC. Lond. Ser. B. 254,75-82. DeWolfe, B. B.. and Baptista, L. F. (1995). Singing behavior, song types on their wintering grounds and the question of leap-frog migration in puget sound white-crowned sparrows. Condor 97,376-389. Dittami, J., Wozniak, J., Ganshirt, G.. Hall, M., and Gwinner, E. (1986). Effects of females and nest boxes on the reproductive condition of male European starlings, Sturnus vulgaris. during the breeding cycle. Vogelwarre 33, 226-231. Doupe, A. J. (1993). A neural circuit specialised for vocal learning. Curr. Opin. Neurobiol. 3,104-111. Doupe, A. J., and Konishi, M. (1991). Song-selective auditory circuits in the vocal control system of the zebra finch. Proc. Narl. Acad. Sci. U.S.A. 88, 11339-11343. Dufty, A. M. (1986). Singing and the establishment and maintenance of dominance hierarchies in captive brown-headed cowbirds. Behav. Ecol. Sociobiol. 19, 49-55.
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
427
Eens, M. (1984). Inleidende studie van de zang van de spreeuw Sturnus vulgaris vulgaris L. Licentiaatsverhandeling (M.Sc. thesis), University of Antwerp. Eens, M. (1992). Organization and function of the song in the European starling Sturnus vulgaris. Unpublished doctoral dissertation, University of Antwerp. Eens, M., and Pinxten, R. (1990). Extra-pair courtship in the starling Sturnus vulgaris. Ibis 132,618-619. Eens, M., and Pinxten, R. (1995). Inter-sexual conflicts over copulations in the European starling: evidence for the female mate-guarding hypothesis. Behav. Ecol. Sociobiol. 36, 71-81. Eens, M., and Pinxten, R. (1996). Female European starlings increase their copulation solicitation rate when faced with the risk of polygyny. Anim. Behav. 51,1141-1147. Eens, M., Pinxten, R.. and Verheyen, R. F. (1989). Temporal and sequential organisation of song bouts in the Starling. Ardea 77,75-86. Eens. M., Pinxten. R., and Verheyen, R. F. (1990). On the function of singing and wingwaving in the European starling Sturnus vulgaris. Bird Study 37, 48-52. Eens, M.. Pinxten, R., and Verheyen, R. F. (1991a). Organization of song in the European starling: species-specificity and individual differences. Belg. J. Zool. 121, 257-278. Eens, M.. Pinxten, R., and Verheyen, R. F. (1991b). Male song as a cue for mate choice in the European starling. Behaviour 116, 210-238. Eens. M., Pinxten, R., and Verheyen, R. F. (1992a). No overlap in song repertoire size between yearling and older European starlings. /his 134, 72-76. Eens. M.. Pinxten, R., and Verheyen, R. F. (1992b). Song learning in captive European starlings, Sturnus vulgaris. Anim. Behav. 44, 1131-1143. Eens, M., Pinxten. R., and Verheyen, R. F. (1993). Function of the song and song repertoire in the European starling (Sturnus vulgaris): an aviary experiment. Behaviour 125,51-66. Ems, M., Pinxten, R., and Verheyen, R. F. (1994). Variation in singing activity during the breeding cycle of the European starling Sturnus vulgaris. Belg. J. Zool. 124, 167-174. Eriksson, D., and Wallin, L. (1986). Male bird song attracts females-a field experiment. Behav. Ecol. Sociobiol. 19, 297-299. Espmark. Y. 0..and Lampe, H. M. (1993). Variations in the song of the pied flycatcher within and between breeding seasons. Bioacoustics 5, 33-65. Ewert. D. N.. and Kroodsma, D. E. (1994). Song sharing and repertoires among migratory and residcnt rufous-sided towhees. Condor 96, 190-196. Fauth, P. T., Krementz, D. G.. and Hines, J. E. (1991). Ectoparasitism and the role of green nesting material in the European starling. Oecologia 88,22-29. Feare. C . J . (1984). “The Starling.” Oxford University Press, Oxford. Feare, C. J., and Burnham, S. E. (1978). Lack of nest site tenacity and mate fidelity in the starling. Ibis 25, 189-191. Feare, C. .I. Gill, . E. L., McKay, H. V., and Bishop, J. D. (1995). Is the distribution of starlings Sturnus vulgaris within roosts determined by competition? Ibis 137, 379-382. Feare, C. J., Spencer, P. L.. and Constantine, D. A. T. (1982). Time of egg-laying of starlings Sturnus vulgaris. 124, 174-178. Flux, J. E. C., and Flux, M. M. (1992). Nature red in claw: how and why starlings kill each other. Notornis 39, 293-300. Folstad, I., and Karter, A. J. (1992). Parasites, bright males and the immunocompetence handicap. A m . Nat. 139, 603-622. Francis. R. C. (1995). Evolutionary neurobiology. Trends Ecol. Evol. 10, 276-281. Freitag, F. (1936). Aus dem Leben beringter Stare zur Fortpflanzungszeit. Vogelring 8,8-15. Freitag, F. (1937). Aus dem Leben beringter Stare zur Fortpflanzungszeit. Vogelring 9,43-49. Freitag, F. (1939). Aus dem Leben beringter Stare zur Fortpflanzungszeit. Vogelring 11,l-9.
428
MARCEL EENS
Gahr, M. (1990). Delineation of a brain nucleus: comparisons of cytochemical. hodological, and cytoarchitectural views of the song control necleus HVc of the adult canary. J. Comp. Neurol. 294,30-36. Ginn, H. B., and Melville, D. S. (1983). Moult in birds. British Trust for Ornithology, Tring. Goude, M. (1988). Kommunikation vid partnervalet-om alderns betydelse vid den intersexuella selektionen. In “Biokommunikation. Biologisk kommunikationsforskning en nordisk antologi” (T. Dabelsteen, Y. Espmark, and S. Sjolander, eds.), pp. 239-246. Akademisk Forlag, Kobenhavn. Greig-Smith, P. W. (1982). Seasonal patterns of song production by male stonechats, Saxicola torquata. Ornis Scand. 13, 225-231. Gwinner, E. (1975). Die circannuale Penodik der Fortpflanzungsaktivitat beim star (Sfurnus vulgaris) under dem Einfluss gleich- und andersgeschlechtiger Artgenossen. Z. Tierpsychol. 38934-43. Gwinner, H.. and Gwinner, E. (1994). Effects of testosterone on nest-box occupation and associated behaviours by male European starlings (Sfurnus vulgaris). Eehaviour 129, 141-148. Gwinner, H., Gwinner, E., and Dittami, J. (1987). Effects of nestboxes on LH, testosterone, testicular size, and the reproductive behaviour of male European starlings in spring. Eehaviour 103,68-82. Hamilton, W. D. (1990). Mate choice near or far? Am. Zool. 30,341-352. Hamilton, W. D., and Zuk, M. (1982). Heritable true fitness and bright birds: a role for parasites? Science 218,384-387. Hartby. E. (1969). The calls of the starling (Sfurnus vulgaris). Dansk. O m . Foren. Tidsskr. 62,205-230. Hasselquist, D. (1994). Male attractiveness, mating tactics and realized fitness in the polygynous reed warbler. Unpublished doctoral dissertation, University of Lund. Hasselquist, D., Bensch, S., and von Schantz, T. (1996). Correlation between male song repertoire, extra-pair paternity and offspring survival in the great reed warbler. Nature (London) 11,229-232. Hausberger, M. (1993). How studies on vocal communication in birds contribute to a comparative approach of cognition. Etologiu 3, 171-185. Hausberger, M., and Black, J. M. (1991). Female song in European starlings: the case of noncompetitive song-matching. Ethol. Ecol. Evol. 3, 337-344. Hausberger, M., and Guyomarc’h J. C. (1981). Contributions i l’ttude des vocalisations territoriales sifflkes chez I’ktourneau sansonnet (Srurnus vulgaris) en Bretagne. Eiol. Eehav. 6,79-98. Hausberger, M., Henry, L., and Richard, M. A. (1995a). Testosterone-induced singing in female European starlings (Sturnus vulgaris). Ethology 99, 193-208. Hausberger, M., Jenkins, P. F., and Keene, J. (1991). Species-specificity and mimicry in bird song: are they paradoxes? A reevaluation of song mimicry in the European starling. Eehaviour 117.53-81. Hausberger, M., Richard-Yris, M.-A., Henry, L., Lepage, L.. and Schmidt, I. (1995b). Song sharing reflects the social organization in a captive group of European starlings (Sturnus vulgaris). J. Comp. Psychol. 109,222-241. Henry, L., Hausberger, M., and Jenkins, P. F. (1994). The use of song repertoire changes with pairing status in male European Starling. Eioacousrics 5,261-266. Hiebert, S. M., Stoddart, P. K., and Arcese, P. (1989). Repertoire size, territory acquisition and reproductive success in the song sparrow. Anim. Behav. 37,266-273. Hinde, R. A. (1982). Foreword. I n “Acoustic Communication in Birds” (D. E. Kroodsma and E. H. Miller eds.), Vol. 1, pp. xiii-xvi. Academic Press, New York.
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
429
Hinde, R. A., and Steel, E. (1976). The effect of male song on an oestrogen dependent behaviour in the female canary Serinus canarius. Horm. Eehav. 7,293-304. Hindmarsh, A. M. (1984a).Vocal mimicry in starlings (Sturnus vulgaris). Unpublished doctoral dissertation, University of Oxford. Hindmarsh, A. M. (1984b). Vocal mimicry in starlings. Behaviour 90,302-324. Hindmarsh, A. M. (1986). The functional significance of vocal mimicry in song. Behaviour 99,87-100. Hobson, K. A., and Sealy, S. G. (1990). Female song in the yellow warbler. Condor92,259-261. Hoffenberg, A. S., Power, H. W., Romagnano, L. C., Lombardo, M. P., and McGuire, T. R. (1988). The frequency of cuckoldry in the European starling (Sturnus vulgaris). Wilson BUN. 100,60-69. Jacobs, L. F. (1996). Sexual selection and the brain. Trends Ecol. Evol. 11,82-86. Jenkins, P. F. (1977). Cultural transmission of song patterns and dialect development in a free-living bird population. Anim. Behav. 25, 50-78. Johnson, L. S., and Kermott, L. H. (1991). The functions of song in male house wrens (Troglodytes aedon). Behaviour 116, 190-209. Johnston, T. D. (1988). Developmental explanation and the ontogeny of bird song: nature/ nurture redux. Behav. Brain Sci. 11, 617-663. Johnstone, R. A. (1995). Sexual selection, honest advertisement and the handicap principle: reviewing the evidence. Biol. Rev. 70, 1-65. Kirkpatrick. M., and Ryan, M. J. (1991). The evolution of mating preferences and the paradox of the lek. Nature (London) 350, 33-38. Kim, J. R., Clower, R. P., Kroodsma, D. E., and Devoogd, T. J. (1989). Song-related brain regions in the red-winged blackbird are affected by sex and season but not repertoire size. J. Neurobiol. 20, 139-163. Kluyver, H. N. (1993). Bijdrage tot de biologie en de ecologie van den spreeuw (Sturnus vulgaris vulgaris L.) gedurende zijn voortplantingstijd. Verslagen en Mededeelingen van den Plantenziektenkundigen Dienst te Wageningen 69,l-145. Kluyver, H. N. (1935). Waarnemingen over de levenswijze van den spreeuw (Sturnus v. vulgaris) met behulp van geringde individuen. Ardea 24,133-166. Konishi, M. (1985). Birdsong: from behavior to neuron. Ann. Rev. Neurosci. 8, 125-170. Konishi. M. (1989). Birdsong for neurobiologists. Neuron 3,541-549. Konishi, M., Emlen, S. T., Ricklefs, R. A., and Wingfield, J. C. (1989). Contributions of bird studies to biology. Science 246,465-472. Krebs, J. R., and Kroodsma, D. E. (1980). Repertoires and geographical variation in bird song. Adv. Study Eehav. 11,143-177. Kroodsma, D . E. (1976). Reproductive development in a female songbird differential stimulation by quality of male song. Science 192, 574-575. Kroodsma. D. E. (1982). Learning and the ontogeny of sound signals in birds. In “Acoustic Communication in Birds” (D. E. Kroodsma and E. H. Miller, eds.), Vol. 2, pp. 1-23. Academic Press, New York. Kroodsma, D. E., and Byers, B. E. (1991). The function(s) of bird song. Amer. Zool. 31, 318-328. Kroodsma, D. E., and James, F. C. (1994). Song variation within and among populations of red-winged blackbirds. Wilson Bull. 106, 156-162. Kroodsma, D. E., and Parker, L. D. (1977). Vocal virtuosity in the brown thrasher. Auk 94,783-785. Lambrechts, M., and Dhondt, A. A. (1986). Male quality, reproduction, and survival in the great tit (Parus major). Eehav. Ecol. Sociobiol. 19, 57-64.
430
MARCEL EENS
Lambrechts, M., and Dhondt, A. A. (1987). Differences in singing performance between male great tits. Ardea 75,43-52. Lampe, H. M., and Saetre, G.-P. (1995). Female pied flycatchers prefer males with larger song repertoires. Proc. R. Soc. Lond. E. 262, 163-167. Lampe, H. M., and Slagsvold, T. (1994). Individual recognition based on male song in a female bird. J. Ornithol. 135, 163. Langmore, N. E., Davies, N. B., Hatchwell, B. J., and Hartley, I. R. (1996). Female song attracts males in the alpine accentor Prunella collaris. Proc. R. Soc. Lond. E. 263,141-146. Leppelsack, H. J., and Vogt, M. (1976). Responses of auditory neurons in the forebrain of a songbird to stimulation with species-specific sounds. J. Comp. Physiol. 107, 263-274. Logan, C. A., Hyatt, L. E., and Gregorcyk, L. (1990). Song playback initiates nest building during clutch overlap in mockingbirds, Mimus polygloffos.Anim. Behav. 39,943-953. Luschi, P. (1993). Improvisation of new notes during singing by male Sardinian warblers Sylvia melanocephala. Eioacoustics 4, 235-244. Mace, R. (1986). Importance of female behaviour in the dawn chorus. Anim. Eehav. 34, 621-622. Mace, R. (1987a). The dawn chorus in the great tit Parus major is directly related to female fertility. Nature (London) 330, 745-746. Mace, R. (1987b). Why do birds sing at dawn. Ardea 75, 123-132. Mann, N. I., and Slater, P. J. B. (1995). Song tutor choice by zebra finches in aviaries. Anim. Behav. 49,811-820. Marler, P. (1970). A comparative approach to vocal learning: song development in whitecrowned sparrows. J. Comp. Physiol. Psychol. Monogr. 71, 1-25. Marler, P. (1987). Sensitive periods and the roles of specific and general scnsory stimulation in birdsong learning. In “Imprinting and Cortical Plasticity” (J. P. Rauschecker and P. Marler, eds.), pp. 99-135. Wiley, New York. Marler, P., Mundinger, P., Waser, M. S., and Lutjen, A. (1972). Effects of acoustical stimulation and deprivation on song development in red-winged blackbirds (Agelaius phoeniceus). Anim. Eehav. 20,586-606. Marler, P., and Peters, S. (1987). A sensitive period for song acquisition in the song sparrow. Melospiza melodia: a cage of age-limited learning. Ethology 76, 89-100. Marler, P., and Peters, S. (1988). Sensitive periods for song acquisition from tape recordings and live tutors in the swamp sparrow, Melospiza georgiana. Ethology 77,76-84. Marler, P., Peters, S., Ball, G. F., Dufty, A. M., and Wingfield, J. C. (1988). The role of sex steroids in the acquisition and production of birdsong. Nature (London) 336, 770-772. McGregor, P. K.. and Krebs, J. R. (1989). Song learning in adult great tits (Parus major): effects of neighbours. Eehaviour 108, 139-159. McGregor, P. K., Krebs, J. R., and Perrins, C. M. (1981). Song repertoires and lifetime reproductive S U C C ~ S Sin the great tit (Parus major). Am. Naf. 118, 149-159. Meijer, T. (1992). Egg-laying patterns in captive starlings. Ardea 80,301-310. Meijer, T., and Langer, U. (1995). Food availability and egg-laying of captive European starlings. Condor 97, 718-728. Merkel, F. W. (1980). Sozialverhalten von individuell rnarkierten Staren Stumus vulgaris in einer kleinen Nistkastenkolonie (3. Mittei1ung)-Die Rolle der Polygynie. Luscinia 44, 133-158. Meller, A. P. (1985). Mixed reproductive strategy and mate guarding in a semi-colonial passerine, the swallow Hirundo rustica. Eehav. Ecol. Sociobiol. 17, 401-408. Meller, A. P. (1988). Spatial and temporal distribution of song in the yellowhammer Emberiza citrinella. Ethology 78, 321 -331.
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
431
Mflller, A. P. (1991). Why mated songbirds sing so much: mate guarding and male announcement of mate fertility status. Am. Nut. 138, 994-1014. Morton, M. L., Pereyra, M. E., and Baptista, L. F. (1985). Photoperiodically induced ovarian growth in the White-crowned sparrow (Zonotrichia leucophrys gambelii) and its augmentation by song. Comp. Eiochem. Physiol. 80A, 93-97. Mountjoy, D. J., and Lemon, R. E. (1991). Song as an attractant for male and female European starlings, and the influence of song complexity on their response. Eehav. Ecol. Sociobiol. 28,97-100. Mountjoy, D. J., and Lemon, R. E. (1995). Extended song learning in wild European starlings. Anim. Eehav. 49,357-366. Mountjoy, D. J., and Lemon. R. E. (1996). Female choice for complex song in the European starling: a field experiment. Eehav. Ecol. Sociobiol. 38, 65-71. Miiller, C., and Leppelsack, H. J. (1985). Feature extraction and tonotopic organization in the avian forebrain. Exp. Brain Res. 59, 587-599. Mundinger, P. C. (1982). Microgeographic and macrogeographic variation in the acquired vocalizations of birds. I n “Acoustic Communication in Birds” (D. E. Kroodsma and E. H. Miller. eds.), Vol. 1 , pp. 147-208. Academic Press, New York. Norris, K. (1993). Heritable variation in a plumage indicator of viability in male great tits Parus major. Nature (London) 362,537-539. Nottebohm, F. (1969). The “critical period” for song learning in birds. Ibis 111,386-387. Nottebohm. F. (1981). A brain for all seasons: cyclical anatomical changes in song-control nuclei of the canary brain. Science 214, 1368-1370. Nottebohm, F. (1993). The search for neural mechanisms that define the sensitive period for song learning in birds. Neth. J. Zool. 43, 193-234. Nottebohm, F., and Arnold, A. P. (1976). Sexual dimorphism in vocal control areas of the songbird brain. Science 194,211-213. Nottebohm, F., Kasparian, S., and Pandazis, C. (1981). Brain space for a learned task. Bruin Res. 213, 99-109. Nottebohm, F.. and Nottebohm. M. E. (1978). Relationship between song repertoire and age in the canary Serinus canarius. 2. Tierpsychol. 46, 298-305. Nottebohm, F., Nottebohm, M. E., and Crane, L. A. (1986). Developmental and seasonal changes in canary song and their relation to changes in the anatomy of song-control nuclei. Eehav. Neural Eiol. 46,445-471. Nottebohm, F., Nottebohm, M. E., Crane, L. A., and Wingfield, J. C. (1987). Seasonal changes in gonadal hormone levels of adult male canaries and their relation to song. Eehav. Neural. Eiol. 47, 197-21 1. Nottebohm, F., Stokes, T. M., and Leonard, C. M. (1976). Central control of song in the canary. J . Comp. Neurol. 165,457-486. O’Loghlen, A. L. (1995). Delayed access to local songs prolongs vocal development in dialect populations of brown-headed cowbirds. Condor 97,402-414. O’Loghlen, A. L., and Rothstein, S. I. (1993). An extreme example of delayed vocal development: song learning in a population of wild brown-headed cowbirds. Anim. Eehav. 46, 293-304. Payne, R. B. (1981). Song learning and social interaction in indigo buntings. Anim. Eehav. 29, 688-697. Payne, R. B. (1985). Behavioral continuity and change in local song populations of village indigobirds Vidua chalybeata. Z. Tierpsychol. 70, 1-44. Petrie, M. (1994). Improved growth and survival of offspring of peacocks with more elaborate trains. Nature (London) 371, 598-599.
432
MARCEL EENS
Petrinovich, L. (1990). Avian song development: methodological and conceptual issues. I n “Contemporary Issues in Comparative Psychology” (D. A. Dewsbury, ed.), pp. 340-359. Sinauer Associates, Massachusetts. Pinxten, R., and Eens, M. (1990). Polygyny in the European starling: effect on female reproductive success. Anim. Eehav. 40,1035-1047. Pinxten, R., and Eens, M. (1994). Male feeding of nestlings in the facultatively polygynous European starling: allocation patterns and effect on female reproductive success. Eehaviour 129,113-140. Pinxten, R., and Eens, M. (in press). Copulation and mate-guarding patterns in relation to sperm competition in the facultatively polygynous starling. Anim. Eehav. Pinxten, R., Eens, M., and Verheyen, R. F. (1989). Polygyny in the European starling. Eehaviour 111,234-256. Pinxten, R., Eens, M., and Verheyen, R. F. (1990). Intermediate clutches in the starling (Sfurnusvulgaris):replacement clutches, additional clutches of polygynous males or late first clutches. J. O m . 131, 141-150. Pinxten, R., Eens, M., and Verheyen, R. F. (1993). Male and female nest attendance during incubation in the facultatively polygynous European starling. Ardea 81, 125-133. Pinxten, R., Hanotte, 0.. Eens, M., Verheyen, R. F., Dhondt, A. A., and Burke, T. (1993). Extra-pair paternity and intraspecific brood parasitism in the European starling, Sturnus vulgaris: evidence from DNA-fingerprinting. Anim. Eehav. 45,795-809. Pinxten, R., Van Elsacker, L., and Verheyen, R. F. (1987). Duration and temporal pattern of mate guarding in the starling. Ardea 75, 263-269. Pliny (1940). “Natural History,” Vol. X. William Heinemann Ltd, London. Read, A. F., and Weary, D. M. (1990). Sexual selection and the evolution of bird song: A test of the Hamilton-Zuk hypothesis. Eehav. Ecol. Sociobiol. 26,47-56. Risser, A. C. (1975). Experimental modification of performance by density in captive starlings. Condor 77,125-132. Sandell, M., and Smith, H. G. (in press). Female aggression in the European starling during the breeding season. Anim. Eehav. Saxby, H. L. (1874). “Birds of Shetland.” Maclachlan & Stewart, Edinburgh. Schwab, R. G., and Lott, D. F. (1971). Testes growth and regression in starlings (Sturnus vulgaris) as a function of the presence of females. J. Exp. Zool. 171, 39-42. Searcy, W. A. (1984). Song repertoire size and female preferences in song sparrows. Eehav. Ecol. Sociobiol. 14, 281-286. Searcy, W. A. (1989). Function of male courtship vocalizations in red-winged blackbirds. Eehav. Ecol. Sociobiol. 24, 325-331. Searcy, W. A. (1992a). Song repertoire and mate choice in birds. Am. Zool. 32, 71-80. Searcy, W. A. (1992b). Measuring responses of female birds to male song. In “Playback and Studies of Animal Communication” (P. K. McGregor, ed.), pp. 175-190. Plenum Press, New York. Searcy, W. A., and Anderson, M. (1986). Sexual selection and the evolution of song. Annu. Rev. Ecol. Sysf. 17,507-533. Searcy, W. A., Podos, J., Peters, S., and Nowicki, S. (1995). Discrimination of song types and variants in song sparrows. Anim. Eehav. 49, 1219-1226. Slagsvold, T., Amundsen, T., Dale, S., and Lampe, H. (1992). Female-female aggression explains territoriality in male pied flycatchers. Anim. Eehav. 43, 397-407. Slagsvold, T., Dale, S., and Saetre, G.-P. (1994). Dawn singing in the great tit (Parus major): mate attraction, mate guarding, or territorial defence. Eehaviour 131, 115-138. Slagsvold, T., and Lifjeld, J. T. (1994). Polygyny in birds: the role of competition between females for male parental care. Am. Nut. 143, 59-94.
UNDERSTANDING THE COMPLEX SONG OF THE EUROPEAN STARLING
433
Slagsvold, T., Lifjeld, J. T., Stenmark, G., and Breiehagen, T. (1988). On the cost of searching for a mate in female pied flycatchers Ficedula hypoleuca. Anim. Behav. 36,433-442. Slater, P. J. B. (1989). Bird song learning: causes and consequences. Ethol. Ecol. Evol. 1,19-46. Slater, P. J. B., Eales, L. A., and Clayton, N. F. (1988). Song learning in zebra finches: progress and prospects. Adv. Study Behav. 18, 1-34. Slater, P. J. B., Jones, A., and ten Cate, C. (1993). Can lack of experience delay the end of the sensitive phase for song learning? Neth. J. Zool. 43,80-90. Smith, H. G. (1995). Experimental demonstration of a trade-off between mate attraction and paternal care. Proc. R. SOC.Lond. B. 260, 45-51. Smith, H. G., and von Schantz, T. (1993). Extra-pair paternity in the European starling: the effect of polygyny. Condor 95,1006-1015. Spector, D. A. (1991). The singing behavior of yellow warblers. Behaviour 117,29-52. Stamps, J. (1994).Territorial behavior: testing the assumptions. Adv. Study Behav. 23,173-232. Summers, R. W., Westlake, G. E., and Feare, C. J. (1987). Differences in the ages, sexes and physical condition of starlings Sturnus vulgaris at the centre and periphery of roosts. Ibis 129, 96-102. Temrin, H. (1986). Singing behaviour in relation to polyterritorial polygyny in the wood warbler (Phylloscopus sibilatrix). Anim. Behav. 34, 146-152. Temrin, H. (1989). Female pairing options in polyterritorial wood warblers Phylloscopus sibilatrix: are females deceived? Anim. Behav. 37, 579-586. ten Cate, C. (1994). Perceptual mechanisms in imprinting and song learning. In “Causal Mechanisms of Behavioural Development” (J. A. Hogan and J. J. Bolhuis, eds.), pp. 116-146. Cambridge University Press, Cambridge. Tinbergen, N. (1963). On aims and methods of ethology. Z. Tierpsychol. 20,410-433. Townsend, C. W. (1924). Mimicry of voice in birds. Auk 41,541-552. Van der Mueren, E. (1977). Dominance hierarchy in a group of caged starlings. Gerfuut 67,261-210. Van Elsacker, L. (1988). Bijdrage tot de ethologie van de spreeuw. Studie van de herkenningsproblematiek tussen ouders en jongen. Unpublished doctoral dissertation, University of Antwerp. von Schantz, T., Grahn, M., and Goransson, G. (1994). Intersexual selection and reproductive success in the pheasant Phasianus colchicus. Am. Nut. 144, 510-527. Wasserman, F. E. (1977). Mate attraction function of song in the White-throated sparrow. Condor 79,125-127. Wasserman, F. E., and Cigliano, J. A. (1991). Song output and stimulation of the female in white-throated sparrows. Behav. Ecol. Sociobiol. 29,55-59. Welling, P., Koivula, K., and Lahti, K. (1995). The dawn chorus is linked with female fertility in the Willow Tit Parus montanus. J. Avian Biol. 26,241-246. West, M. J., and King, A. P. (1990). Mozart’s starling. Am. Sci. 78, 106-114. West, M. J., King, A. P., and Eastzer, D. H. (1981). The cowbird: reflections on development from an unlikely source. Am. Sci. 69, 57-66. West, M. J., King, A. P., and Freeberg, T. M. (1994).The nature and nurture of neo-phenotypes: a case history. In “Behavioral Mechanisms in Evolutionary Ecology” (L. A. Real, ed.), pp. 238-251. The University of Chicago Press, Chicago. West, M. J., Stroud, A. N., and King, A. P. (1983). Mimicry of the human voice by European starlings: the role of social interaction. Wilson Bull. 95,635-640. Wiley, R. H., Hatchwell, B. J., and Davies, N. B. (1991). Recognition of individual males’ songs by female dunnocks: a mechanism increasing the number of copulatory partners and reproductive success. Ethology 88, 145-153.
434
MARCEL EENS
Williams, H. (1990). Models for song learning in the zebra finch: fathers or others? Anim. Behav. 39,745-757. Wingfield, J. C., and Moore, M. C. (1987). Hormonal, social, and environmental factors in the reproductive biology of free-living male birds. In “Psychobiology of Reproductive Behavior, an Evolutionary Perspective” (D. Crews, ed.), pp. 148-175. Prentice Hall, New Jersey. Witchell, C. A. (1896). “The Evolution of Bird Song.” Adam & Charles Black, London. Witherby, H. F., Jourdain, F. C. R., Ticehurst, N. F., and Tucker, B. W. (1943). “The Handbook of British Birds.” H. F. & G. Whiterby Ltd., London. Witschi, E., and Miller, R. A. (1938). Ambisexuality in the female starling. J. Exp. Zool. 79,475-487. Wright, J.. and Cotton, P. A. (1994a). Song versus food: trade-off decisions in a breeding colony of European starlings Sfurnus vulgaris. Bird Srudy 41, 95-101. Wright, J., and Cotton, P. A. (1994b). Experimentally induced sex differences in parental care: an effect of certainty of paternity? Anim. Behav. 47, 1311-1322. Wright, J., and Cuthill, I. (1992). Monogamy in the European starling. Behaviour 120,262-285. Yasukawa, K.,Blank, J. L., and Patterson, C. B. (1980). Song repertoires and sexual selection in the red-winged blackbird. Behav. E d . Sociobiol. 7, 233-238. Zuk, M. (1996). Disease. endocrine-immune interactions, and sexual selection. Ecology 77, 1037- 1042.
ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 26
Representation of Quantities by Apes SARAH T. BOYSEN COMPARATIVE COGNITION PROJECT DEPARTMENT OF PSYCHOLOGY THE OHIO STATE UNIVERSITY
COLUMBUS. OHIO
OF QUANTITIES BY APES I. REPRESENTATION
OVERVIEW: DEFINITIONS OF ANIMAL COUNTING A. HISTORICAL
Animal counting studies, including experiments with apes, historically cover an extensive archive of empirical work (Koehler, 1950; also see Rilling, 1993; Yerkes and Coburn, 1915). These include the most infamous animal “counting” case, Clever Hans, the horse that was touted as being capable of, among other complex skills, counting, addition, and subtraction (Rosenthal, 1911/1965). Ultimately, Clever Hans’s limitations were unmasked by psychologist Oskar Pfungst, who discovered that Hans’s owner, von Osten, was unwittingly providing cues to the horse in the form of subtle head and body movements, which could be interpreted by the horse whenever von Osten and Clever Hans arrived at the correct number of counts. The case of Clever Hans certainly contributed a cloud of doubt over the legitimacy of studying countinglike phenomena in nonhuman species in general, and has further served as a warning example of the potential for social cuing (Rilling, 1993; Davis, 1993) in some types of animal studies, particularly those involving sign language or other symbol systems with apes. Most systematic counting work with nonhuman species has been explored using birds, rats, and monkeys (see reviews by Davis and Memmott, 1982; Davis and PerussC, 1988; also see Boysen and Capaldi, 1993), although several studies have employed other mammals (e.g., raccoons, Davis, 1984; pigs, Yerkes and Coburn, 1915). In addition, several important empirical efforts antedate recent approaches to numerical competence in apes, and all have employed chimpanzees, in particular. These include studies by Ferster (1964), Hayes and Nissen (1971), and Woodruff and Premack (1981). 435
Copyright 0 1997 by Academic Press All rights of reproduction in any form reserved. 006.5-3454D7$25.00
436
SARAH T. BOYSEN
Definitions of Counting in Animals It has been the legacy in ethology and comparative psychology that efforts to arrive at some definitions were destined to result in polar points of view, such as past disagreements over the contribution of nature versus nurture, or definitions that were historically derived and maintained so as to exclude all other animal species, like species-centric definitions of language or imitation, which could be ascribed only to Homo sapiens. Counting as a phenomenon under study in nonhuman species has been subjected to similar scrutiny and criticism (Davis, 1993;Davis and PerussC, 1988;Thomas and Lorden, 1993). In their important contribution to the study of the emergence of counting competence in children, Gelman and Gallistel (1978) outlined five principles that provided the framework for, and defined, their view of the counting process. The first three principles govern rules or procedures for how one goes about counting, the fourth principle defines what may be counted, and the fifth principle is a composite of features of the other principles (Gelman and Gallistel, 1978). Of the three how-to-count principles, the first is the one-one principle, which entails the ticking off of items in a collection of things to be counted in such a way that only one distinct tag (or number word, e.g.) is applied to each object in an array. They note that, in order to coordinate these efforts, the individual doing the counting has to simultaneously orchestrate the processes of both partitioning and tagging. Partitioning involves keeping track of the items in two categories: those that have already been counted and those that remain in the array to be counted. The coordinated process of tagging, that is, the application of distinct tags such as verbal count words, has to proceed hand-in-hand with the partitioning process in order for the count sequence to end correctly, with one count word assigned to one and only one item in the collection. Children initially use pointing or similar behaviors to facilitate the coordination of partitioning and tagging, to minimize errors such as tagging the same item twice, or skipping an item altogether (Fuson, 1988). The second how-to-count principle is known as the stable-order principle, and requires that the counting tags be used in a repeatable order. The third principle, the cardinalprinciple, specifies that the final tag used in a counting series has special status since it also represents the total number of items in the array. The fourth principle, which is known as the abstraction principle, dictates that the three how-to-count principles can be applied to anything, with no necessary distinction between concrete objects and nonphysical items, in repeatable order (e.g., one, two, three). Finally, the orderirrelevance principle states that the order in which items are counted does not matter, and that the final cardinal assignment to a collection of items
REPRESENTATION OF QUANTITIES BY APES
437
will be the same, regardless of the starting point in the array. According to Gelman and Gallistel (1978), these principles need not be confined to use by humans, since the possibility exists that animals, too, might use a counting procedure, albeit nonverbal in nature. Their terminology thus allows for the use of nonverbal tag sequences by any individual, human or otherwise. Gallistel (1993) has further proposed that numerical discriminations and arithmetic reasoning in animals may be a primitive and foundational aspect of animal mentation upon which other processes build. These studies may provide insights toward understanding abstract categorization and conceptualization in animals in general. H e draws a distinction between numbers as categories and as concepts, noting that animals may be said to have numerical categories if they can be shown to respond to the absolute quantity of a collection, regardless of any other attributes of the items. Animals may be shown to have numerical concepts if they can demonstrate some ability to manipulate numbers, such as adding, subtracting, or ordering. Gallistel (1993) notes that there is considerable empirical evidence demonstrating that laboratory animals are capable of exhibiting numerical categorization, such as the simple counting of motor responses (e.g., Mechner, 1958; Platt and Johnson, 1971). Other approaches have included presenting items successively, such as the procedures followed by Capaldi and Miller (1988), during which rats were required to make successive runs down an alley, with a predictable series of rewarded and nonrewarded trials. Because the animals ran slower on nonrewarded trials and faster on rewarded trials, Capaldi and Miller argued that the animals were counting both trial types. Simultaneously presented sets of food items or objects have also been presented to several nonhuman species during counting tasks. Davis (1984), for example, presented a raccoon (Procyon lotor) with a cube containing one, two, or three grapes, and the animal was required to select the cube that contained three items. A similar approach was also taken with chimpanzees by Matsuzawa (1985). Previously tutored in a graphic symbol language modeled after the work of Rumbaugh and colleagues (1977), Matsuzawa’s chimpanzee, Ai, responded to collections of simultaneously presented objects, such as five green pencils, by specifying the number, color, and object symbols for each array. A similar approach was used by Pepperberg (1987) with an African Grey parrot (Alex) with an extensive representational vocal repertoire. Alex was required to respond to collections of objects whose attributes differed in shape, material, and number, and was able to respond correctly in response to all three features of the object array. We have used the simultaneous presentation approach with our chimpanzee subjects, as well, beginning with food items presented simultaneously, followed by generalization tests using junk objects (Boysen and Berntson, 1989).
438
SARAH T. BOYSEN
Evidence for arithmetic reasoning also can be derived from studies of animal timing and counting. In a series of studies, Church and colleagues (Meek and Church, 1983; Church and Meck, 1984) tested their model of animal timing and counting, suggesting that the mechanism could operate in an event or a timing mode, and thus accrue either numbers of events or durations for later transfer into memory. In an additional series of studies exploring the decision process in timing, Gibbon and others have provided strong evidence that such processes involve addition, subtraction, and division of temporal intervals (Gibbon, 1981; Gibbon and Church, 1990), thus providing empirical support for the proposition that the neural substrates for manipulating information arithmetically are available in nonhuman species. Further evidence in support of Gallistel’s definition of number conceptualization and reasoning is derived from our own studies with chimpanzees (Boysen and Berntson, 1989; Boysen. 1993). Following the acquisition of both productive labeling of arrays and demonstrated number comprehension, two novel tasks were introduced to a young female chimpanzee (Sheba). Sheba already had extensive experience labeling arrays of candies and junk objects with Arabic numbers between 0 and 8, as well as specific training on number comprehension with the numbers 1-3 (see Boysen and Berntson, 1989; Boysen, 1993). With the new counting tasks, the idea was to provide a somewhat more informal testing context for the introduction of several new concepts related to numerical reasoning, which focused principally on summation, or rudimentary addition. In the first task, which we called functional counting, three sites were designated within the testing area where one to four food items could be hidden. Oranges were used because of their visual appeal, although they were not a preferred food; we were concerned that the use of novel objects might encourage more interest in the items themselves, and less motivation to complete the counting sequence. In this setting (Fig. l ) , the chimpanzee was encouraged to travel unattended to each site, note the number of items available, then return to Point A, where a series of number placards were available. Once the foray had been completed, the total number of food items hidden among the three sites was to be communicated by selecting the Arabic numeral that represented that total. That is, if one and three oranges were hidden among the three locations, Sheba indicated that four oranges were present, by pointing to the number symbol 4. Quite unexpectedly, when given the first few opportunities with the task, Sheba was correct in assigning the cardinal number to the total array available. While we had speculated that we might eventually teach the numerous novel dimensions of the task to our subject and that the conceptual ability to master elementary summation was likely within her capacity, we were unprepared for her
439
REPRESENTATION OF QUANTITIES BY APES
\
I
FIG.1. Testing context for Functional and Symbolic Counting Tasks. Three possible sites were selected at which zero to four oranges (functional task) or numerals (symbolic task) were hidden at two of the sites on each trial. Sheba was required to visit each site, and return to the workstation (A), where she could choose from among the numerals 0-4, and select the numeral that correctly represented either the total number of food items or the sum of the numerals seen among the sites she visited. Arrows indicate Sheba’s travel route as she attended to each site.
essentially immediate mastery of the task. Sheba had no prior training whatsoever on summing of arrays, yet had little difficulty demonstrating her understanding of this new game. Given her exceptional performance, we opted to challenge her further by making the task more difficult, initially reasoning that if she was able to add arrays of objects (Fig. 2A), perhaps she might eventually be taught to sum representations of those arrays, using Arabic numerals. Consequently, a second task was devised, which we called symbolic counting (see Fig. 2B). Sheba was encouraged to explore the three sites, with numbers hidden in two of the three possible locations. Much to our chagrin, Sheba was 80% correct on the very first double-blind trials she completed, and we were at a loss to explain her newly discovered ability to add number symbols (Boysen and Berntson, 1989). Our own review of the children’s counting literature provided some intriguing possibilities for explaining Sheba’s performance with both the food and numeral arrays. Apparently, in part
440
SARAH T. BOYSEN
FIG.2. The Functional Counting Task (A) required that Sheba attend to food items (oranges) hidden at two of three possible sites, while the Symbolic Counting Task (B) consisted of two different Arabic numberals (representing the quantities 0-4).hidden at two sites among the three possible locations.
through the kinds of counting games that children typically play with their parents, during which they learn much about increasing and decreasing numerosity, children invent spontaneous addition algorithms (Groen and Parkman, 1972; Groen and Resnick, 1977; Starkey and Gelman, 1982). The use of such algorithms appears long before any formal training in arithmetic, and has been demonstrated by children as young as 3 years old (Groen and Resnick, 1977). For example, children will spontaneously apply the counting rules that they have acquired when they encounter two or more arrays by procedures such as “counting-on” (Fuson, 1988), during which they count the first array (e.g., “one, two, three . . .”), and then begin the count of a second array with the next number, “four, five, six, seven.” These types of counting algorithms emerge without formal teaching in very young children, and may represent the type of counting process that Sheba used when she first encountered more than one collection of items, separated in both time and space. Such abilities in children are highly reminiscent of the kinds of skills that Sheba was exhibiting in both the functional and symbolic counting tasks, and suggested to us that she had also come to understand similar addition rules, which allowed her to sum multiple arrays. Sheba’s only counting experience has been through the two structured counting tasks, described in more detail later. These included the productive labeling of quantities (selecting the correct Arabic numeral, from among several, when shown a specific quantity of foods or objects), and number comprehension (viewing a single Arabic number, decoding that symbol,
REPRESENTATION OF QUANTITIES BY APES
441
and selecting the correct number of items that corresponds to the numeral, from among several different arrays). This training must have provided her with the conceptual underpinnings for the emergent capabilities that she demonstrated in the summation tasks. The most interesting point, however, was the basic assumption regarding the use of addition algorithms by children. They must be able to count in order to successfully utilize such algorithms. Since we could find no other logical explanation for Sheba’s demonstrated competence in adding arrays other than the invocation of similar algorithm-type rules, we could only conclude that she could also count, and count in a manner that was subserved by similar conceptual processes to those that support children’s counting skills (Boysen and Berntson, 1989). OF PRODUCTIVE COUNTING SKILLS B. ACQUISITION
How does one initiate the conceptual understanding of the counting process in a nonhuman, nonverbal organism? Our work with counting abilities in the chimpanzee began with an initial one-to-one correspondence game. Both Fuson (1988) and Gelman and Gallistel (1978) have discussed the significance of an understanding of one-to-one relationships in children’s early stages of learning to count. Their work suggested that an important stepping stone to establishing counting in a chimpanzee would be specific training with one-to-one relations that could serve as the necessary bridge for teaching the associations between number symbols and their corresponding quantities. The one-to-one task that the chimpanzees learned was designed to teach them to track a small magnet marker, which was affixed to a round placard, from among three possible spatial locations (Fig. 3). Stimulus items (gumdrop candy), which were only used for counting tasks, with the choice stimuli (three round lids with small magnetic disks glued on them) were placed in a row, directly above a tray. The experimenter placed candy arrays composed of one to three items on the tray, then pointed to or touched the items individually. The chimpanzee was required to select the correct placard bearing the corresponding number of markers. Initially, single candies were presented on each trial, and the chimps were required to match it with the correct placard (Fig. 3A). After criterion performance of 85% correct responses for two successive sessions, all trials presented to the chimps consisted of the presentation of two candies, and the subjects had to select the placard bearing two magnets (Fig. 3B). At criterion, trial types now included the presentation of one or two candies, presented in random order during a given session (Fig. 3C). This phase of training entailed an explicit one-to-one matching between the two possible candy arrays and the corresponding marked placards, and thus training
442
A
SARAH T. BOYSEN
C
D FIG.3. Stimulus placards used for all phases of the One-to-one Correspondence Task, with the chimpanzees required to match one gumdrop candy to a placard bearing one round magnet (A), followed by trials in which two gumdrops were matched with a placard required bearing two magnets (B). A third training phase required matching of one or two gumdrops with the corresponding placard (C). In the final phase of the one-to-one correspondence task, the chimpanzees matched candy arrays of one, two, or three items with the respective placard (D). Arrows indicate movement of placards on each trial. to control for spatial bias in responding.
investment was considerably more extensive than in the initial phases. Once criterion was achieved, the third marked placard, bearing three magnets, was introduced, and the animals now encountered trial types in which one, two, or three candies were presented for matching with the corresponding placard (Fig. 3D). Several significant procedural details were likely critical to the animals’ success. Among these was the fact that a correct choice allowed the chimpanzees to eat the stimulus item(s). Clearly, the high incentive value of the candies helped to maintain attention and motivation to the task, the verbal count words were used by the experimenter during training, and the process of eating the different quantities of candies all likely contributed toward a multifeatured representation of quantities. On each counting trial, the teachedexperimenter displayed an array of candies, and once the chimpanzees made their choice of a specific numeral, the experimenter also said the number word aloud (e.g., “Right! One, two, three”). The chimps were then allowed to eat the candy stimuli. Thus, early in their experience with different quantities, direct participation with the concepts of increasing and
REPRESENTATION OF QUANTITIES BY APES
443
decreasing numerosity were an integral part of the training context. As the chimps watched, the experimenter created an increasingly larger array of candies for that trial. As the chimps consumed the candies, the array size grew smaller and smaller, until all the items for the trial were eaten. Through this process, counting may have taken on a more functional quality, with more broadly based conceptual underpinnings than earlier attempts to teach chimps to simply match dots on cards (e.g., Hayes and Nissen, 1971). The one-to-one correspondence task entailed matching quantities of candies between one and three. Once the animals were reliably matching the candy arrays with the correct placards, Arabic numerals were systematically substituted for the placards (Fig. 4).This transition represented a number generalization task, with numerals introduced one at a time, while the chimpanzees’ performance was permitted to stabilize between 85 and 90% correct before a second numeral was introduced. Eventually, they were correctly selecting the Arabic numerals 1-3 in response to candy arrays of the same quantity. Subsequent numerals (zero and four, respectively), as well as all additional numbers (to date, 5-8), were introduced within the framework of this task. Thus, only the quantities 1-3 were used in both the one-to-one correspondence and the numerical generalization paradigms. At the same time that new numbers 0 and 4 were being introduced, an assortment of motor behaviors (pointing to the individual items in the arrays, moving the items around on the tray, or rearranging the entire array) emerged in one chimpanzee during the counting task (Boysen, Berntson, Shreyer, and Hannan, 1995). Sheba, a younger and physically much smaller female chimpanzee, continued to enjoy the flexibility and freedom of unrestrained access within the chimpanzee facility, while her two older male companions, in the full throes of the exponential growth spurt characteristic of adolescent male chimpanzees, were no longer permitted outside their
FIG.4. Stimulus items presented for introducing Arabic numerals to the chimpanzees, which systematically replaced the magnet placards used in the one-to-one correspondence task shown in Fig. 3.
444
SARAH
T. BOYSEN
home cage. This meant that Sheba continued to have direct physical contact with the candy arrays and choice stimuli, while Kermit and Darrell had access only to the choice placards that could be reached with a single digit, because their very large hands could literally no longer fit through the openings between the bars of their homecage area. An elaborate series of touching and pointing behaviors was exhibited by Sheba, such that the arrays were physically monitored and rearranged by her on virtually every trial. These behaviors were highly reminiscent of the descriptions for partitioning behaviors reported by Fuson (1988) and Gelman and Gallistel (1978), among others, by very young children who were just learning to count. Subsequent evaluation of these behaviors revealed significant correlations between the number of candies available for counting on a given trial, the number of times Sheba touched, pointed, or moved each item, and the cardinal number she finally selected to represent the array (Table I). These data were evaluated from a series of counting sessions, which were videotaped over a 3-month period. The resulting correlations suggested that Sheba's spontaneous utilization of these monitoring behaviors helped her to achieve a more accurate count, just as such behaviors support the facilitation of accurate counting of arrays for very young children (Fuson, 1988; Gelman and Gallistel, 1978). There have been several notable references to the possibilities of motoric monitoring of countinglike processes in other animals, including birds and rats, although the data from Sheba's sessions are topographically more similar in both appearance and function to those of children. For example, in Koehler's (1950) classic studies with birds, he reports an instance where a jackdaw (Corvus rnonedulu, a member of the crow family) was required to eat a specific number of baits, and no more. In one case, the bird was observed to bow in a highly ritualized manner before each box before coming to the final fifth box whose contents he was permitted to eat (Koehler, 1950). In another study, a rat was observed to move its paw in
TABLE I INTERCORRELATIONSBETWEEN CANDY ARRAYS, NUMBER OF MOTOR TAGS,A N D ARABIC NUMBER SELECTED~
Trials
n
All 422 Correct 229 Incorrect 193
Number of items/ Cardinal number selected .74 1.oo .32
" p < .01 for all correlations.
Number of items/ Number tagged/ Number tagged Cardinal number selected .67 .68 .67
.52 .68 .31
REPRESENTATION OF QUANTITIES BY APES
445
a semicircle around the operant bar in a Skinner box, and upon completion of this movement pressed for a reinforcer (Mechner, 1958). This collateral behavior may have served as a temporal and/or counting bridge between opportunities to bar-press under the specific reinforcement contingencies of the study.
C. RECEPTIVE COMPREHENSION OF NUMBERS Previous studies with artificial language systems with chimpanzees (e.g., Savage-Rumbaugh, 1986) have suggested that, unlike human children, chimpanzees that were taught to use referential symbols did not spontaneously demonstrate receptive comprehension of those same symbols. That is, if a chimpanzee were able t o use a representational symbol, such as a lexigram or ASL sign, to “name” an object or food item, the same animal was unable to initially respond to the use of that same symbol by another individual. Thus, if an experimenter held up an orange, the chimpanzee readily selected the proper graphic symbol or could form the correct ASL gesture that had been associated with the food item. If, however, the experimenter arranged an assortment of objects or foods, and then she herself used the lexigram symbols or ASL gestures, and asked the chimpanzee to behaviorally indicate which item was the “orange,” the animals were unable to decode the use of the symbols by another person. Receptive understanding of the symbol system had to be taught separately, and did not emerge de novo from a conceptual understanding of the productive manipulation of lexigrams or ASL gestures. The two approaches to symbol manipulation (productive and receptive) are necessary for carrying on a dialogue using our own verbal language system. That is, in order for a conversation to unfold, the speaker produces verbal representational symbols, and the receiving individual decodes them, prior to making her own response. Thus, both directions of symbol use, the productive use of symbols and receptive comprehension, are critical between both members of a conversing dyad. Given these findings from the animal language field, it seemed apparent that if we hoped for number symbols to take on representational status with our subjects, they would need to be able to demonstrate both the productive and receptive use of numbers. The productive tasks had already been established through the one-to-one correspondence game, followed by the number generalization task, such that the chimpanzees were able to “produce” labels or names for arrays ranging in size from zero to four items. It was at this point in their understanding of number concepts that we initiated training for number comprehension. The number comprehension task was initially undertaken with the following format, although very quickly, Darrell, the pilot subject, informed us
446
SARAH T. BOYSEN
that our approach was off the mark. Individual Arabic numerals from 1 to 3 were presented on a black-and-white video monitor, which was situated directly above a tray that had been divided in half by a bold black stripe (Phase I training approach). Our original idea was to present Darrell with two candy arrays of different quantities on the tray, display a number between 1 and 3, and have him select the candy array whose quantity matched the depicted number. Thus, he would be decoding the graphic numerical representation, and indicating behaviorally that he understood the real-world referent of that number. After several weeks of daily training sessions, during which Darrell’s performance continued to hover around chance levels, we reexamined our teaching strategy. It was quite clear how Darrell was interpreting the task, since the majority of his choices were made by selecting the larger of the two arrays presented on each trial. He completely failed to take into account the number symbol appearing on the television monitor and, instead, interpreted the task as an opportunity to choose between a larger and a smaller quantity of candy. Given such a choice, he usually chose the larger array (see Table 11). There is clear precedent for similar judgments by chimpanzees and orangutans from the work of Menzel and his colleagues (Draper and Menzel, 1966; Menzel, 1960,1961; Menzel and Draper, 1965). In his first study with chimpanzee subjects only, Menzel (1961) found the animals reliably chose the largest piece of food (from among five that differed by .lo-in increments), with the mean correlation between order of selection and physical size averaging .73. The animals were able to detect size differences as small as 5%. Menzel (1961) concluded that, in adult animals, the preference was already established, daily performance was highly stable, and individual variability was small, with performance mediated visually. Given our chimpanzees’ performance on the initial number comprehension task, and the related findings from Menzel’s work, we concluded that
TRIALS TO CRITERION
Subject Darrell Sheba Kermit
Phase Phase Phase Phase
I I1
I1 I1
TABLE I1 RESPONSES DURING ESTABIJSHMENT OF NUMBER COMPREHENSION SKILLS
A N D CORRECT
Number of trials to criterion
Mean correct responses (last 3 sessions)
489 20 1 282 315
67%* 96?6 82% 77 %
Failure to achieve criterion; task approach terminated.
REPRESENTATION OF QUANTITIES BY APES
447
the animals were not likely to inhibit responding to the larger edible arrays, and that we should adjust our training accordingly. Therefore, in Phase I1 of receptive training, we replaced the candy arrays with the original training placards used in the one-to-one correspondence task (see Fig. 3) and retained the video presentation of the number symbols. Now the animals were required to attend to the Arabic numeral presented on the video screen, and choose the placard showing the corresponding number of markers. Given this procedural change, the chimpanzees were able to attend t o the critical features of the task, rather than simply respond to the different array sizes, and all began to show a more normal learning curve indicative of their emerging understanding of the number comprehension task (see Table 11). 11. SYMBOLIC FACILITATION OF QUANTITY JUDGMENTS
Had we not attempted to extend the chimpanzees’ use of number symbols to an area of social cognition, notably deception, we might have gone on thinking that there was virtually no end to their ability to demonstrate innovative strategies and emergent understanding in new areas of numerical competence. While there are probably new skills that we can and will tap as we continue to explore these issues, we were nevertheless stopped in our tracks recently by the animals’ overwhelming inability to perform optimally on a quantity-based discrimination task (Boysen and Berntson, 1995). As additional studies were completed with this paradigm, a provocative picture of the ability of the chimpanzee to invoke what might be called a “cognitive shift,” which was dependent on the types of stimuli we used, revealed itself (Boysen, Berntson, Hannan, and Cacioppo, 1996). Our original intent in designing the first quantity judgment study was t o provide an opportunity for evaluating the potential use of deception by our chimpanzees during a food-sharing task. Two chimpanzees were to work as a team, with one chimpanzee (the selector) given access to two food dishes, which contained different amounts of candies, while the other chimpanzee served as a passive observer. The selector chimpanzee was permitted to choose one of the dishes, whereupon the experimenter intervened and provided the observer chimpanzee with the contents of the chosen dish. The selector animal was then given the contents of the second remaining dish. Because of the reverse contingency arrangement, the selector chimpanzee’s optimal response was to pick the dish containing the smaller amount of candy, which would then be given to her partner. The larger remaining quantity in the second dish would then be provided to the selecting chimp. If the animals achieved some reasonable understanding
448
SARAH T. BOYSEN
of the task, we had hoped to set up a testing situation whereby a partition would be erected between the two chimpanzees so that the observer animal would not be able to see the contents of either dish. If given the chance to choose between two dishes, under circumstances whereby the observer chimpanzee would never know how many candies were available, would the selecting chimpanzee always make sure that they themselves received the larger amount of candy? While it was unfortunate, on one hand, that we never got to ask that question, the chimpanzees’ inability to master the food-sharing game in its original format led us ultimately to some very intriguing findings. Our first efforts with the quantity judgment task enlisted two female chimpanzees, Sheba, who was 12 years old at the time, and Sarah, 35 years old, and a seasoned veteran with a long and colorful history of participation in comparative cognition studies (Premack, 1976,1986). Both chimpanzees had extensive backgrounds with a range of conceptual tasks including, in Sarah’s case, early training with an artificial symbol system (e.g., see Premack, 1976). Thus, they brought to the task considerable cognitive sophistication, which made their inability to optimize the discrimination demands of the quantity judgments even more striking. As noted earlier, the rules of the task were quite simple: whichever quantity was chosen was then provided to the second chimp by the experimenter. The chimp who had the opportunity to do the choosing had to be content with receiving the remaining array of candy. Sarah was chosen as the selector chimp initially, but failed to grasp the essential elements of the game. She persisted in choosing the larger of the two arrays, trial after trial, and session after session. The first arrays that she could choose between were composed of either one or two candies. Perhaps the incentive to choose the smaller one was simply not significant enough, given a choice between one or two pieces. We decided to increase the size of one array, and now presented Sarah with a choice between bowls containing one and four candies. She again persisted in choosing the larger array, and continued to receive only the remaining single piece of candy on most trials. Again, in an effort to increase her motivation to respond optimally, we raised the stakes. Now, arrays of one and six items were presented on each trial, and Sarah was permitted to make a choice between the two. Time and time again, she persisted in choosing the bowl containing six chocolates and, given the reverse contingency we had imposed, Sarah netted only a single candy. Sheba, meanwhile, found herself the recipient of what must have seemed like mountains of candy, again and again, with absolutely no effort whatsoever. Sarah simply was not catching on to the basics of the task, and we decided at this point to give her observer partner (Sheba) the chance to move into the role of selector.
REPRESENTATION OF QUANTITIES BY APES
449
Though we were unsure how quickly Sheba might demonstrate an understanding of the task, we anticipated that she would probably benefit in some way from her role as the observer during the 24 sessions she had completed with Sarah. We thus expected that she might show more rapid learning of the reverse contingencies of the task. We had also speculated that perhaps Sarah’s poor performance was idiosyncratic to her, as we had other chimpanzees among our group who had shown highly specific learning difficulties with some types of tasks, particularly those with certain attentional demands (Boysen, 1992b). It was also possible that the quantity judgment task required some conceptual understanding or cognitive grasp that was beyond Sarah’s individual capability, and was specific to her. With Sheba in the role of selector chimp, we soon learned that we were wrong in ascribing the difficulties of the task to Sarah’s apparent intellectual shortcomings. Sheba, too, failed to implement the rules for making an optimal choice between the two quantities, and persisted at selecting the larger array over and over again. Sarah, now the passive observer, garnered the larger quantity of candy on nearly every trial. In utter frustration, we abandoned all hope of using the task to explore social cognition and deception, and instead were intrigued as to why two of the most wellschooled apes appeared to be unable to decipher the simple rules of the task in order to earn the larger reward. In fact, given that their performance was significantly below chance, it appeared that some other mechanism was exerting a powerful bias over their responding. Counting and other number-related tasks besides the quantity judgment paradigm were among several concurrent skills for which the chimpanzees were receiving daily training, and Sheba had already demonstrated a fairly elaborate understanding of numerical concepts using Arabic numerals (e.g., Boysen, 1993; Boysen and Berntson, 1989). At the time, she was proficient in counting arrays of foods or objects composed of zero to eight items; had shown spontaneous emergent abilities with rudimentary addition using both foods and numerals; could appropriately label portions of foods with fraction symbols representing 1/2 or 1/4; and had undergone training in creating arrays of objects that corresponded to displayed numerals from 1 to 3. We decided to use Sheba’s capabilities with symbolic representations for quantities to further explore her understanding (or lack thereof) of the quantity judgment task. Therefore, in the next phase of testing, we presented two different Arabic numerals (1 vs. 2, 1 vs. 4, and 1 vs. 6) within a session, and allowed Sheba to choose one of the two numbers. Once a number was selected, the corresponding number of candies was given to Sheba’s observer partner, Sarah. Sheba thus earned the number of candies represented by the other remaining (nonselected) numeral. With absolutely no hesitation, and given the first opportunity to make such a choice, Sheba
450
SARAH
T. BOYSEN
chose the smaller numeral “1.” She continued to select the smaller of the two numerals on 67%of the trials during the very first session that number symbols were used in place of candy arrays. Two additional sessions were completed in which numerals were used as stimuli, and Sheba’s performance continued to be significantly above chance. During Session 4, candies were reintroduced as stimuli, and her performance plummeted. That is, Sheba chose the larger candy array repeatedly, with her overall response for that session registering a dismal 17%. In other words, with two quantities of real candies once again available as choice stimuli, Sheba chose the larger array 83%of the time, and had to relinquish the larger portion to her partner. Her poor performance was all the more striking, given her consistent and significant performance throughout the previous three sessions when Arabic symbols had been used. During the fourth session, when candies were again introduced, there was no evidence of any transfer or generalization from her successful performance using numeric stimuli. Instead, the powerful bias exerted over her choices between the candy arrays was immediately evident. Subsequent sessions with Sheba, using all possible numerical combinations between 1 and 6, or candy arrays composed of one to six items, in an ABBA design, revealed stable levels of responding, which were entirely dependent on the type of stimuli used (Table 111). When Arabic numerals were presented, Sheba had little difficulty in selecting the smaller of the two, thus reaping the greater reward. But each time that candy arrays were presented, she reverted to her persistent choice of the larger array, and received the smaller portion of the total reinforcers available. Following completion of this phase of testing, several questions came to mind immediately. How critical was the presence of a social partner during the task? Was the immediacy of the high-incentive candy (which was physi-
TABLE 111 PERFORMANCE BY SHEBA WITH NOVEL SYMBOL COMEI~NATIONS OF ARABIC NUMERALS 1-6 CANDY ARRAYS OF ONETO SIX ITEMS
AND
Session
Stimuli
Correctltotal trials
% Correct
1 2 3 4 5 6 7 8
Numerals Candy Candy Numerals Numerals Candy Candy Numerals
11/15 4/15 2/15 10/15 11/15 2/15 5/15 loll5
73 26 13 67 73 13 33 67
REPRESENTATION OF QUANTITIES BY APES
45 1
cally located directly in front of both chimpanzees) the major source of the interference effect? What were the important features of using number symbols, which permitted Sheba to respond to her advantage? And finally, was Sheba’s success with numerals, and both chimpanzees’ difficulties making optimal choices with the candy arrays, demonstrable by other chimpanzees who had comparable training with numbers? Each of these issues was addressed in three subsequent experiments. The same quantity judgment paradigm was used, with two different arrays of items presented, while the chimpanzee subject made a choice between the two arrays. However, there were no other chimpanzees present in any of the subsequent studies. Instead, each of five chimpanzees was tested individually, including Sheba and Sarah, and three additional male chimpanzees (Darrell, 14.5 yr; Kermit, 14 yr; and Bobby, 6 yr). In the first of the two experiments, we tested all of our chimpanzees that had a working numerical repertoire, using the original quantity judgment task, as described previously. By testing the animals individually, we were able to address the possible contribution that conspecific social competition might have played in determining the results of our earlier study. As in the previous task, two dishes were presented with different amounts of candy (zero to six items), and the subject simply pointed to one of the dishes. The contents of that dish were then returned to the supply bowl, in full view of the chimpanzee, who then received the contents of the second, nonchosen bowl. Each session included 20 trials, with a total of 20 sessions completed, for an overall total of 400 trials per animal. The results across all five chimpanzees tested were striking in their similarity to each other, and to our overall findings from the original quantity judgment study (Fig. 5). That is, none of the five animals tested was able to select the smaller of the two arrays with any reliability but, instead, persisted at selecting the larger of the two candy arrays even though this resulted in consistently fewer rewards. And, as seen in the original study, performance was significantly below chance, suggesting that perhaps the incentive properties of the larger array introduced a response bias that interfered with the optimal strategy based on the prevailing reinforcement contingencies. The animals’ overall performances were remarkably similar, falling within a very narrow range of proportion correct for each of the five subjects (between 27 and 31%). There was also a definitive relationship between performance on the task, which was related to the relative size disparity between the two arrays, and the overall array size. That is, there was a greater degree of interference on trials in which there was a larger disparity between the quantity of candies presented in the two arrays, but this incentive disparity effect was also qualified by the overall size of the arrays (Fig. 6). Thus, the interference effect for a particular absolute dispar-
452
SARAH T. BOYSEN
1 .o
0.9 0.8 0.7
0.6 ... ... ... ... ... ...
0.5 0.4
0.3
0.2 0.1
0.0
‘fL
L 6‘ 6
?I
I
0
%%“%%L.%%’,%%.9 overall
I
Sessions FIG.5. Overall performance by all chimpanzees (N= 5 ) whenchoosingbetween twodifferent quantities of candies during the Quantity Judgment task. Selection of either quantity resulted in the chosen candy array being discarded, with the selecting chimp then receiving the remaining array. Bold box indicates mean performance, with standard error: hatched bars indicate session by session performance; dashed line depicts chance performance, with results indicating that all animals performed significantly below chance, and were more likely to consistently choose the larger array.
ity was attenuated with a larger mean array size, reminiscent of a Weberlike function.’ To quantify the task interference effect as a function of arfay disparity, we derived an index in which the absolute numerosity differences of the two choice arrays were scaled by the overall array sizes (disparity index = disparity/total array size). This yields an index that can range from 0, with no disparity between the choice stimuli, to 1.0 with maximal relative disparity (i.e., when one array is null). Linear regression analysis revealed that this disparity index could account for over 93% of the variance in performance across all disparity ratios. This suggests that the slope of the disparity ratio/performance function may provide an index of the incentive processing of the task, and the extent to which the task interfer-
* Weber’s Law stipulates that difference thresholds (AI) increase proportionately with an increase in stimulus magnitude or intensity (I), such that A111 is a constant. This implies that the psychophysical effect of a given stimulus difference is a function of the ratios of the two stimuli. In the present results, over 90% of the variance in performance could be accounted for by the ratio of the sizes of the two choice stimuli.
REPRESENTATION OF QUANTITIES BY APES
453
A
B 1.0 1
-
y = .58 . 5 3 ~
RZ = -93
Disparity Ratio FIG.6. (A) Overall performance expressed as a function of disparity and candy array size. (B) Probability of an optimal response (choosing the smaller array, and thus receiving the larger, remaining array as a reward) as a function of the disparity ratio. Results indicate that the greater the disparity between array sizes, the more likely the subjects were to choose incorrectly, although this effect was attenuated by larger mean array sizes.
454
SARAH T. BOYSEN
ence associated with the incentive-disparity between the candy arrays biased their choices (Boysen ef al., 1996). In the next study, we were interested in examining the contribution of visuoperceptual features of the arrays. In other words, were the chimpanzees drawn to the appearance of a larger perceptual mass, which was correlated with the larger quantity of candy? To examine this question, we alternated the test stimuli between candy arrays and arrays composed of rocks (large limestone gravel pieces measuring about 1.5 in. in diameter). As in the previous studies, the chimpanzees were confronted with a choice between two arrays that differed in quantity, and that were composed of either candy or rocks within a given session. Once a choice between arrays was made, the experimenter returned the selected candy array o r rocks to the supply bowl, and provided the chimpanzee with either the number of reinforcers that corresponded to the number of rocks in the second, remaining bowl, or the candy array itself from the second, unselected bowl. In either case, the same reverse contingencies remained in effect; whichever array was chosen, the contents of that dish were returned to the supply, and the subject received an equivalent number of reinforcers as the remaining array. A total of 12 test sessions, including 6 with rock arrays and 6 with candy arrays, were completed with each of the five chimpanzees. Their overall performance indicated that, like all previous testing using candy arrays, the animals were unable to inhibit the selection of the larger array, and therefore received fewer reinforcers. Their performance with candy arrays was again significantly below chance, and overall responding was similar to their performance in Experiment 1 with candies alone. The introduction of the rock stimuli did not significantly enhance their performance, and overall group results, while below chance, did not differ significantly. There was greater variability across the five animals when rock arrays were used, with two animals (Sheba, Sarah) showing a significantly higher, overall performance with the rock stimuli. These individual differences, however, did not impact appreciably on group performance, and thus the general conclusion that could be drawn from the comparisons between responses to candy arrays and arrays composed of inedible rocks was that they were not different. This further suggested that the high-incentive features of the candy arrays, which contributed to the interference phenomenon that was observed in the animals’ performance in previous studies, was not sufficient to account for the resultant response biases, since arrays composed of rocks produced similar, though more variable, interference across all subjects. These findings led us to compare the animals’ performance in a third experiment during which candy arrays were again presented, along with separate sessions using Arabic numerals. Because all five chimpanzees now
REPRESENTATION OF QUANTITIES BY APES
455
had a working repertoire of numbers and counting experience on various quantity-based tasks, we were interested in determining if the previously demonstrated “symbol effect” might be replicated by the group (Boysen and Berntson, 1995). In our initial study, one chimpanzee (Sheba) had been successful in optimizing her responses during the task if a choice could be made between two different numerical representations, instead of candy arrays (Boysen and Berntson, 1995). Acquired proficiency with number symbols by the other chimpanzees in our colony since the time of Sheba’s testing now permitted us to ask the same question of additional subjects, a critical test of the phenomenon’s reliability. The testing context and procedures were the same as in all previous quantity judgment tasks, with each subject tested alone under the same reversed-contingency reinforcement criteria. This time, all chimpanzees chose between arrays composed of differing quantities of candies, or between all possible number combinations of two different Arabic numerals between 0 and 6. Whichever dish was chosen, the animals were reinforced with candies from the second dish, or with the same number of candies that corresponded to the remaining, nonselected Arabic numeral. Analysis of their performance with six sessions using candy arrays and six sessions with number symbols revealed a significant difference in performance, which was dependent on the type of stimulus employed (candy vs. number symbols). Like previous experiments using candy arrays, the animals’ overall performances were virtually indistinguishable from one another, and also did not differ from their performance with candy arrays observed in all previous studies. This reflected an enduring, highly stable, and persistent interference effect, which appeared to be highly resistant to additional exposure and experience with the candy arrays. In sharp contrast, all five chimpanzees exhibited significantly better performance when number symbols were used instead of candy, with their overall performance significantly better than chance (Fig. 7). That is, whenever candy arrays were presented, the animals continued to select the larger array, thus receiving the smaller number of candies as a reward. If symbolic representations were used, in the form of Arabic numerals, the previous interference effect was immediately and consistently overcome. Thus, whenever numerical representations were presented as choice stimuli, all five chimpanzees were able to optimize their response strategies and reap the larger reward. And, consistent with our previous findings, performance with candy arrays was related to the disparity between the two arrays, with the disparity ratio accounting for the majority of variance in the animals’ overall performance (Fig. 6). Regression analyses of the results when Arabic numerals were presented were distinctively different. Not only was the overall mean level of performance significantly increased, but the slope of
456
SARAH
Candy
T. BOYSEN
Numerals
FIG.7. Overall performance with Quantity Judgment Task in which two different sizes of candy arrays or two different numerals were presented as choices. Results indicate that the animals consistently choose the larger candy arrays (and received a smaller reward), but were able to select the small number symbol (and thus receive the larger reward), across all five animals tested. Bold box indicates mean performance, with standard error; dashed line depicts chance performance.
the disparity ratio function accounted for a minimal proportion of the variance in their performance with symbols. We concluded that the chimpanzees' performance when numbers were used was not affected by the disparity-dependent interference that so dramatically affected their performance when candy arrays were presented. Quantity judgments are likely ubiquitous determinants of both human and animal behavior, but the numerical processing capabilities of animals have not been fully defined, nor have their mechanisms been clearly illuminated. This is especially true for the nature of symbolic representations of quantity, and the impact of numerical processing and quantity judgments on behavior. In our studies, chimpanzees were unable to select the smaller of two arrays of foods in order to obtain a larger reward, but were able to do so if the food stimuli were replaced by number symbols. These results could be explained based on the influence of multiple evaluative dispositions that may be operative in a given context. We have hypothesized that a direct incentive-based disposition toward the larger food arrays interferes with an opposite disposition arising from the reinforcement contingency.
REPRESENTATION OF QUANTITIES BY APES
457
The more optimal performance with Arabic numerals may reflect the ability of symbols to represent selective aspects or features of their referents, such as numerosity, while permitting an organism to disregard others, such as incentive properties. The ability of symbols to encompass selective features of their referents may constitute a critical advantage of symbolic representations. Symbols may permit an organism to efficiently process selected information, and respond adaptively based on that knowledge structure, while at the same time minimizing potential interference from lower level or more primitive evaluative dispositions. EVALUATIVE PROCESSES AND SYMBOLS Of particular interest is the relationship between numerical processing and behavioral choice, where response dispositions may arise from quantity judgments. Our previous findings on the behavioral consequences of quantity judgments may be viewed within a broader framework of evaluative processes (Berntson, Boysen, and Cacioppo, 1993; Cacioppo and Berntson, 1994). Evaluative dispositions, which can be characterized by approach/avoidance tendencies or appetitivelaversive reactions, are widely represented in animal and human behavioral settings. Evaluations of the adaptive significance of objects and events in the environment are so central to survival that all species have biological mechanisms for approaching, acquiring, or ingesting certain classes of stimuli; withdrawing from, avoiding, or rejecting others; and for the establishment of enduring response predispositions toward classes of stimuli. Such dispositions range from simple pain-withdrawal reflexes to conditioned approach/avoidance responses, to higher level attitudinal predispositions toward broad classes of stimuli (Berntson et af., 1993). The significance of this multiplicity of levels in evaluative mechanisms lies in the potential for multiple or even competing dispositions expressed at the same time. Under natural conditions, such dispositions would generally lead to concordant behavioral judgments or actions. In other cases, opposing dispositions could lead to conflicts, as seen in our work with chimpanzees’ quantity judgments using food or symbol arrays. Although the chimpanzees had clearly acquired a food-distribution rule, as evidenced by their immediate improved performance when numeric symbols were used as choice stimuli, the direct incentive features of the candy arrays appeared to have introduced a powerful, conflicting dispositional bias. The relative potency of the interference was no doubt partly related to the perceptual immediacy of the candy arrays. The apparent interference effect on task performance with the chimpanzees was largely eliminated when Arabic numerals were substituted for the
458
SARAH T. BOYSEN
candy arrays, and performance increased to better levels. Over counterbalanced sessions with symbols and candy arrays, the chimpanzees’ performance shifted dramatically and immediately from significantly above chance with Arabic symbols to significantly below chance with candy arrays. These findings are reminiscent of the self-regulation literature, as children who were unable to inhibit approach tendencies to food items were able to respond more adaptively when the food stimuli were presented more abstractly using color slide representations (Mischel, Shoda, and Rodriguez, 1989). The animals’ improved performance with Arabic numbers may be attributable to the fact that the symbolic stimuli failed to evoke the interfering, nonassociative disposition. That is, the Arabic symbols capture the requisite numerical features, without encompassing the incentive or perceptual properties that trigger the interfering response bias. Evidence for this view comes from the fact that the animals’ performance with Arabic symbols was independent of the specific numerical disparity between the choice stimuli. As noted, the advantage of symbols for encompassing attributes of their real-world referents may constitute an important advantage of abstract representations. Symbols may not necessarily show equivalence relationships with their referents, as suggested by others (e.g., Cerutti and Rumbaugh, 1993;D’Amato, Salmon, Loukas, and Tomie, 1985; McIntire, Cleary, and Thompson, 1987; Sidman and Tailby, 1982), since they may represent only selective features of their referents. This would violate the requirement of “symmetry” in the relation between the symbol and its referent. Although equivalence relationships can develop between stimuli and associated reinforcers (e.g., Dube, McIlvane, Mackay, and Stoddard, 1987), they do not necessarily emerge, nor would they be optimal in all cases. The chimpanzees in our studies had extensive training in labeling arrays with numerals across several different contexts, and these numerical symbols were not consistently associated with any specific reinforcer. Rather, the relevant dimension across training contexts was the functional numeric significance of the symbols. Through the apparent facility of symbolic representations of numerosity, the animals were able to maximize reward payoff. The evaluative dispositions theoretical framework speaks precisely to this intersection and modulation of such competing or conflicting patterns of behavior. 111. SUMMARY
The study of comparative cognition, including the investigation of counting and numerical skills in nonhuman animals, continues to emerge as an
REPRESENTATION OF QUANTITIES BY APES
459
area of great interest and potential for students of animal learning and behavior. Since our initial studies of numerical competence, our chimpanzees have demonstrated remarkable facility with complex number concepts that were heretofore unknown for any primate other than humans. These studies began with two simple matching tasks, during which the animals learned to associate a collection of candies with either a pattern of markers on a placard (one-to-one correspondence), o r the corresponding Arabic numeral (number generalization task). In the process of acquiring these two associative tasks, a more fundamental understanding of numbers clearly emerged, as the animals were able to subsequently demonstrate new skills and an understanding of numerous novel, number-related concepts for which no explicit training was completed (e.g., functional and symbolic counting, including rudimentary summation and addition). These newfound skills were remarkably similar in both topography and developmental emergence to those observed in young children in the early stages of learning to count, and thus, more significantly, may reflect similarities in the cognitive demands and resultant capacity in the chimpanzee to support features of time, space, and number in their environment. Undoubtedly, the demands of social living, including cooperative behavior, alliance formation, kin recognition, and reciprocal altruism, among others, required significant changes in behavioral inhibition and regulation in both human populations and among the great apes. With the changes in neural plasticity that have allowed for tremendous learning potential and behavioral malleability, which we share with our ape cousins, has come reorganization of a brain whose overall structure and functional organization overlaps significantly between the two species. With very limited tutoring and enculturation from a human teacher, it is possible to provide the chimpanzee with a new attentional focus and to redirect emphasis to specific features of the environment, such as the specifying of quantity via a representational number system. By the same token, our chimpanzee subjects have reminded us that, despite our best efforts, their attentional capacity and ability to symbolically represent the world are not functionally equivalent to how humans encode events, objects, or features of the environment. In our more recent studies on quantity judgments, in which chimpanzees were exposed to a series of choices between quantities of foods, rocks, or number symbols, the animals based their decisions on differing criteria, depending on the type of stimulus items they encountered. The teasing apart of such similarities and differences between apes and humans represents a tremendous challenge for providing new insights into the critical departures that each species made during their respective changes in cognitive abilities and capacities over the course of parallel evolutionary histories.
460
SARAH T. BQYSEN
References Berntson, G. G., Boysen, S. T., and Cacioppo, J. T. (1993). Neuro-behavioral organization and the cardinal principle of evaluative bivalence. Ann. New York Acad. Sci. 702,75-102. Boysen, S . T. (1992a). Counting as the chimpanzee views it. I n “Cognitive Aspects of Stimulus Control” (W. K. Honig and J. G. Fetterman, eds.), pp. 367-383. Lawrence Erlbaum, Hillsdale, New Jersey. Boysen, S. T. (1992b). Pongid pedagogy: The contribution of humanlchimpanzee interaction to the study of ape cognition. In “The Inevitable Bond: Scientist-Animal Interactions” (H. Davis and D. Balfour, eds.). Cambridge University Press, London. Boysen, S. T. (1993). Counting in chimpanzees: Nonhuman principles and emergent properties of number. In “The Development of Numerical Competence: Animal and Human Models” (S. T. Boysen and E. J. Capaldi, eds.), pp. 39-59. Lawrence Erlbaum, Hillsdale, New Jersey. Boysen, S. T., and Berntson, G. G. (1989). Numerical competence in a chimpanzee (Pun troglodytes). J. Comp. Psych. 103, 23-31. Boysen, S. T.. and Berntson. G. G. (1990). The emergence of numerical competence in the chimpanzee (Pan troglodytes). In “Language and Intelligence in Animals: Developmental Perspectives” (S. T. Parker and K. R. Gibson, eds.), pp. 435-450. Cambridge University Press. Boysen, S. T., and Berntson, G. G. (1995). Quantity judgments: Perceptual vs. cognitive mechanisms in chimpanzees (Pan troglodytes). J. Exp. Psych.: Anim. Behav. Proc. 21, 82-86. Boysen, S. T., Berntson, G. G., Hannan, M. B., and Cacioppo, J. T. (1996). Quantity-based interference and symbolic representations in chimpanzees (Pan troglodytes). J. Exp. Psych.: Anim. Behav. Proc. 22,7646. Boysen, S. T., Berntson, G. G., Shreyer, T. A., and Hannan, M. B. (1995). Indicating acts during counting by a chimpanzee ( f u n rroglodyres).J. Comp. Psych. 109,47-51. Boysen, S . T., Berntson, G. G., Shreyer, T. A., and Quigley, K. S. (1993). Processing of ordinality and transitivity by chimpanzees (Pan troglodytes).J. Comp. Psych. 107,208-21 5 . Boysen, S. T., and Capaldi, E. J. (1993). (Eds.). “The Development of Numerical Competence: Animal and Human Models.” Lawrence Erlbaum, Hillsdale, New Jersey. Cacioppo, J. T., and Berntson, G. G. (1994). Relationships between attitudes and evaluative space: A critical review with emphasis on the separability of positive and negative substrates. Psych. Bull. 115,401-423. Capaldi, E. J., and Miller, D. J. (1988). Counting in rats: Its functional significance and the independent cognitive processes which comprise it. J. Exp. Psych.: Anim. Behav. Proc. 14, 3-17. Cerutti, D. T., and Rumbaugh, D. M. (1993). Stimulus relations in comparative primate perspective. Special issue: Stimulus equivalence. Psych. Record 43,811-821. Church, R. H., and Meek, W. H. (1984). The numerical attributes of stimuli. In “Animal Cognition” (H. L. Roitblat, T. G. Bever, and H. S . Terrace, eds.), pp. 445-464. Erlbaum. Hillsdale, New Jersey. D’Amato, M. R., Salmon, D. P., Loukas, E., and Tomie, A. (1985). Symmetry and transitivity of conditional relations in monkeys, (Cebus apella) and pigeons (Columba livia). J. Exp. Anal. Behav. 44, 35-47. Davis, H. (1984). Discrimination of the number “three” by a raccoon (Procyon lotor). Anirn. Learn. Behav. 12,409-413. Davis, H. (1993). Numerical competence in animals: Life beyond Clever Hans. In “The Development of Numerical Competence: Animal and Human Models” (S. T. Boysen and E. J. Capaldi, eds.), pp. 109-125. Lawrence Erlbaum, Hillsdale, New Jersey.
REPRESENTATION OF QUANTITIES BY APES
461
Davis, H., and Memmott, J. (1982). Counting behavior in animals: A critical evaluation. Psych. Bull. 92, 547-571. Davis, H., and Perusse, R. (1988). Numerical competence in animals: Definitional issues, current evidence and a new research agenda. Behav. Brain Sci. 11,561-579. Draper, W. A., and Menzel, E. W., Jr. (1966). Size and distance of food: Cues influencing the choice behavior of orangutans (Pongo pygmaeus). Folia Primatologica 4,186-190. Dube, W . V., McIlvane, W. J., Mackay, H. A., and Stoddard, L. T. (1987). Stimulus class membership established via stimulus-reinforcer relations. J. Exp. Anal. Behav. 47, 159-175. Fernandes, D. M., and Church, R. H. (1982). Discrimination of the number of sequential events by rats. Anim. Learn. Behav. 10, 171-176. Ferster, C. B. (1964). Arithmetic behavior in chimpanzees. Sci. Amer. 210,98-106. Fuson, K. C. (1988). “Children’s Counting and Concepts of Number.” Springer-Verlag, New York. Gallistel, C. R. (1989). Animal cognition: The representation of space, time and number. In Annu. Rev. Psych. 40 (M. R. Rosenzweig and L. W. Porter, eds.), pp. 155-189. Gallistel, C. R. (1993). A conceptual framework for the study of numerical estimation and arithmetic reasoning in animals. In “The Development of Numerical Competence: Animal and Human Models” (S. T. Boysen and E. J. Capaldi, eds.), pp. 211-223. Lawrence Erlbaum, Hillsdale, New Jersey. Gelman, R., and Gallistel, C. R. (1978). “The Child’s Understanding of Number.” Harvard University Press, Cambridge, Massachusetts. Gibbon, J. (1981). On the form and location of the psychometric bisection function for time. J. Math. Psych. 24,58-87. Gibbon, J., and Church, R. H. (1990). Representation of time. Cognition 37,23-54. Groen, G . J., and Parkman, J. M. (1972). A chronometric analysis of simple addition. Psych. Rev. 79,329-343. Groen, G. J., and Resnick, L. B. (1977). Can preschool children invent addition algorithms? J. Educ. Psych. 69,645-652. Hayes, K. J., and Nissen, C. (1971). Higher mental functions of a home-raised chimpanzee. In “Behavior of Nonhuman Primates, 4” (A. M. Schrier and F. Stollnitz, eds.), pp. 59-115. Academic Press, New York. Koehler, 0. (1950). The ability of birds to “count.” Bull. Anim. Behav. 9,41-45. Matsuzawa, T. (1985). Use of numbers by a chimpanzee. Nature (London) 315, 57-59. McIntire, K. D., Cleary, J. P., and Thompson, T. (1987). Conditional relations by monkeys: Reflexivity, symmetry, and transitivity. J. Exp. Anal. Behav. 47, 279-285. Mechner, F. M. (1958). Probability relations within response sequences under ratioreinforcement. J. Exp. Anal. Behav. 1, 109-121. Meck, W., and Church, R. H. (1983). A mode control model of counting and timing processes. J . Exp. Psych.: Anim. Behav. Proc. 9,320-334. Menzel, E. W. (1960). Selection of food by size in the chimpanzee and comparison with human judgments. Science 131,1527-1528. Menzel, E. W. (1961). Perception of food size in the chimpanzee. J. Comp. Physiol. Psych. 54, 588-591. Menzel, E. W., and Draper, W. A. (1965). Primate selection of food by size: Visible versus invisible rewards. J. Comp. Physiol. Psych. 59,231-239. Mischel, W., Shoda, Y., and Rodriguez, M. L. (1989). Delay of gratification in children. Science 244,933-937. Pepperberg, I. (1987). Evidence for conceptual quantitative ab es in the African grey parrot: Labeling of cardinal sets. Ethology 75, 37-61.
462
SARAH T. BOYSEN
Platt, J. R., and Johnson, D. M. (1971). Localization of position within a homogeneous behavior chain: Effects of error contingencies. Learn. Motivation 2, 386-414. Premack, D. (1976). “Intelligence in Ape and Man.” Lawrence Erlbaum, Hillsdale, New Jersey. Premack, D. (1986). “Gavagai.” Cambridge University Press, London. Rilling, M. (1993). Invisible counting animals. In “The Emergence of Numerical Skills: Animal and Human Models” (S. T. Boysen and E. J. Capaldi, eds.), pp. 3-37. Lawrence Erlbaum, Hillsdale, New Jersey. Rosenthal, R. (Ed.). (1965). “Clever Hans.” (Carl L. Rahn, Trans.). Holt, Rinehard & Winston, New York. (Original work published 1911.) Rumbaugh, D. M. (Ed.). (1977). Language Learning by a Chimpanzee: The LANA Project. New York: Academic Press. Savage-Rumbaugh, E. S . (1986). “Ape Language: From Conditioned Response to Symbol.” Columbia University Press, New York. Sidman, M., and Tailby, W. (1982). Conditional discrimination vs. matching-to-sample: An expansion of the testing paradigm. J. Exp. Anal. Behav. 37, 5-22. Starkey, P., and Gelman, R. (1982). The development of addition and subtraction abilities prior to formal school in arithmetic. In “Addition and Subtraction: A Cognitive Perspective” (T. P. Carpenter, J. M. Moser, and T. A. Romberg, eds.), pp. 99-1 16. Lawrence Erlbaurn, Hillsdale, New Jersey. Thomas, R. K., and Lorden, R. B. (1993). Numerical competence in animals: A conservative view. In “The Development of Numerical Skills: Animal and Human Models” (S. T. Boysen and E. J. Capaldi, eds.), pp. 127-147. Lawrence Erlbaum, Hillsdale, New Jersey. Woodruff, G., and Premack, D. (1981). Primitive mathematical concepts in the chimpanzee: Proportionality and numerosity. Nature (London) 293,568-570. Yerkes, R. M. (1934). Modes of behavioral adaptation in chimpanzees to multiple-choice problems. Comp. Psych. Monogr. 10, 1-108. Yerkes, R. M., and Coburn, C. A. (1915). A study of the behavior of pig (Sus scrofa) by the multiple choice method. J. Anim. Behav. 5, 185-225.
Index
A
Abduction, female, 210 Abstraction counting principle, 436 Acoustic divergence. 319, 339-346 Acoustic signals, see also Birds, song repertoire cryptic species, 319,335-339 female choice, 341-342 habitat structure, 341 natural selection, 340-341 resource partitioning, 330-335, 349 sexual selection, 341-343, 344, 414-417 visual signals versus, 317-318 Acrocephalus spp., see also Sedge warblers cryptic species, 338 song repertoire, 412, 417 Adalia hipicnctata, mate choice, 24, 45 Addition algorithms, 440, 441 Adipose tissue, 219; see also Energy reserves African false vampire bats, vocal calls, 78 African Grey parrots, numerical categorization, 437 Agelaius phoenicncs, see Red-winged blackbirds Age polyethism, 286-288 comparative studies, 306-309 honey bees behavioral development, 286, 288, 289-290, 301-306 genetic influences, 289 juvenile hormone, see Juvenile hormone, regulation of age polyethism neuroanatomical plasticity, 301-302, 303-306 physiological development, 287, 288 Aggression, see also Dominance; Territorial defense birds, intraspecific, 405,413-414
mammals, dispersal, 210-21 1, 225, 230-231, 237 Agile gibbons, vocal learning, 73 Algorithms, addition, 440, 441 AIouatfa palliafa, see Howler monkeys Altruism, reciprocal, 228 Androgens birds costs, 356 females, 371-372 song control nuclei, 374.386-387 vocal learning, 374 dispersal behavior, 214, 215-216, 221, 224, 229-230,235 Animal counting, see Counting, animal; Representation of quantities, apes Animal models evolution, primates, 104-106 infantile amnesia, 251-279 Anrechinus spp., dispersal behavior, 194, 230 Anthropocentrism, 104-106 Apes numerosity, see Representation of quantities, apes taxonomic classification, 108 vocal learning, 73-74 Apis spp., 306-307; see also European honey bees Appearance cryptic species, 318 natural selection, 342 Approach-avoidance tendencies, 457 Area X, 374, 381, 383-384, 385-386 Arithmetic reasoning, 438-441 Artifacts, experimental, 174-175 Ateles geoffroyi reproduction, 113 sex ratios, 123, 124 Auks, individual recognition, 80 463
464
INDEX
B Baboons dispersal behavior, 194, 212, 231 dominance ranks, 128 habitat adaptations, 116 reproduction, 113 sex ratios, 123 taxonomic classification, 107 Balaena mysricerus, vocal learning, 69, 77 Balaenoptera spp., vocal learning, 77 Baleen whales, vocal learning, 68-69, 77, 81-82 Banner-tailed kangaroo rats dispersal behavior, 209 sound production learning, 75, 80 Barbary macaques dispersal behavior, 212 extragroup matings, 120 genetic analyses, 138 Bats, acoustic signals, see also Pipistrelle bats cryptic species, 335-337 flexibility, 340-341 resource partitioning, 330-335 vocal learning, 63-64, 78, 80, 81 Bay wrens, sex differences, 381 Bearded seals, vocal learning, 65, 77 Bees, see Bumble bees; Honey bees Behavioral ecology, primate behavioral variation, 110-111, 126, 128-129, 132-133 comparative models, 103-104, 105 anthropocentric influences, 104-106 demographic feedback, 112 phylogenetic analyses, 108-112 population, 144 conservation biology and, 101-103. 104 complementary assumptions, 126 convergence between, 143-144 methodological bridges, 137-143 environmental variation, 115-117, 126, 131-133 behavioral interactions, 132-133 genetic interactions, 134-135 genetic paternity analyses, 137, 138-142 genetic variation, 126, 127-129, 134-135 individual-level analyses, 122-126, 134-135 sampling methods, 121-122
species comparisons, 104-106, 108-1 12 steroid assays, 137, 142-143 Behavioral endocrinology, 235 Behavioral polymorphism, 286 Behavioral variation conservation biology, 117-121 primate behavioral ecology, 110-111, 126, 128-129, 132-133 Belding’s ground squirrels, 183-184, 189-191 dispersal behavior, 190-191 consequences, 234 females, 206-208 males, 205-206 proximal causes, 215-223 testosterone, 215-216, 221. 229-230 Bias experimenter, 173-1 74 observer, 169-174 response, chimpanzees, 451-454,455-457 Big brown bats, vocal learning, 63, 80 Biodiversity, hidden, 319, 346-348 Biological Species Concept, 106 Birds aggression, intraspecific, 405,413-414 copulation extrapair, 400-401, 412 female solicitation displays, 405, 410-411 vocal solicitations, 398-401, 405, 410-41 1 counting behaviors, 444 cryptic species acoustic signals, 338 genetic differences, 345 habitat structure, 85, 341 sexual selection, 18, 341, 356-357 interspecific, 76, 414-417, 420 intraspecific, 78 social influences, 366-367 song repertoire copulation solicitations, 398-401, 405, 410-411 dominance, 396-397,404,413-414 ethological approach, 355-356 functions, 406-414 hormonal influences, 356,374,386-387 improvisation, 361 migratory versus resident species, 414-415
INDEX
mortality, 368 neurobiological research, 355,380-391, 419-420 sexual selection, 356, 414-417 ultraviolet light, 318 vocal learning, 362-380 distribution, 60 functional significance, 76, 78, 79-80, 82, 84, 85 sensitive period, 367, 372-373 testosterone, 374 Blackbirds repertoire size and brain space, 387 sexual selection, 32 vocal learning, 76 Bluegill sunfish, sexual selection, 13 Blue whales, vocal learning, 77 Body fat, see Energy reserves Bonnet macaques, dispersal behavior, 208 Bonobo, vocal learning, 73 Bottlenose dolphins, vocal learning, 66, 67-68,79,81-82,83 Bowerbirds, sexual selection, 33 Bowhead whales, vocal learning, 69, 77 Brachyteles spp.. see Muriquis Breeding, synchronous, birds, 402 Brown-headed cowbirds, repertoire size, 369 Brown thrasher, song repertoire, 361 Buffy-headed marmosets, sex ratios, 125 Bumble bees, 306 age polyethism, 307 juvenile hormone treatment, 308 queen-worker differentiation, 286 Buntings, vocal learning, 78
C
Callithrix jlaviceps, sex ratios, 125 Canaries, song neural substrates, 384, 385 vocal learning, 76, 372-373 Capuchin monkeys, reproduction, 113 Cardinal counting principle, 436 Cardioderma cor, vocal calls, 78 Caste polymorphism, 286 Cats contextual learning, 59 vocal learning, 74-75
Cause-effect relationships, 160-165 Cebuella pygrnaea, vocal learning, 71, 89 Cebus apella, habitat adaptations, 116 Cebus capucinus, reproduction, 113 Centris pallida, sexual selection, 13 Cercopithecine primates, dispersal behavior, 209 Cercopithecus aethiops dispersal patterns, 119 vocal learning, 70 Certhia spp., acoustic signals, 317 Cervus elephas, sexual selection, 12 Cetaceans, vocal learning, 66-69 Cheetahs, conservation management, 102 Childhood memories, traumatic, 251-252, 279; see also Infantile amnesia Chimpanzees dispersal behavior, 228 genetic analyses, 138, 140 representation of quantities, 458-459 arithmetic reasoning, 438-441 compared to human children, 436-437, 439440,459 motor monitoring, 443-444 productive counting skills, 437, 441-445,449 quantity judgments, 447-457 response interference, 451 -454, 455-458 symbolic facilitation, 457-458 receptive number comprehension, 445-447,449 symbolic language learning, 445 vocal learning, 73-74, 79 Chingolo sparrows, vocal learning, 85 Cholinergic system, rats, memory development, 262 Chromosonal inversion polymorphisms, seaweed flies, 6-7 Cistithorus palustris, repertoire size and brain space, 387 Cockroaches juvenile hormone. 291 sexual selection, 33, 34-35, 41 Coefficient of genetic variance, body size, 28-32.42-44 Coelopa frigida, see Seaweed flies Coelopa pilipes, 3, 40 Colobine monkeys, sex ratios, 123 Colobus badius, dispersal patterns, 120
466
INDEX
Communal displays, vocal learning, 77 Communication systems, see also Acoustic signals; Vocal learning gestural, 88 language, 87-88 mammalian, 85-86, 230 primate, 89 Comparative models, 103-104, 105 individual-level, 121-126 population-level, 114-121, 144 species-level, 104-1 14 Competition in primates, 132 space. 223 sperm, 14-16, 397-401, 421 Competitive speciation models, 340 Conditioned punishment, rats, retention, 264 Confounding effects, experimental design, 168-169 Conservation biology behavioral variation, 117-121, 133-134 comparative models, 104. 144 demographic stochasticity, 112, 122-126, 133 environmental stochasticity, 112-114. 126, 133-134 behavioral interactions, 133-134 genetic interactions, 135-137 genetic paternity analyses, 137, 138-142 life history applications, 112-1 14 population averaging, 122 population genetics, 117, 120, 126, 129-131. 135-137. 144 population viability analyses, 117-118, 120 primate behavioral ecology and. 101-103, 104 complementary assumptions, 126 convergence between, 143-144 methodological bridges, 137-143 sampling methods, 121-122 species classifications, 106-108, 112-114 steroid assays, 137, 142-143 Contests, 12-13, 132 Contextual learning, sounds, 85-86 action-based, 62-63, 74 improvisation, 61 vocal learning versus, 59
Control groups confounding effects, 168-169 unsuitable, 175-177 Convergent evolution, vocal learning, 86-87 Cooperation, dispersal behavior, 230-231 Copulation birds extrapair, 400-401. 412 female solicitation displays, 405, 410-411 vocal solicitations, 398-401, 405, 410-411 transfer of nutrients, 39 Corpora allata glands, honey bees, 290, 291-292 Corpora pedunculata, see Mushroom bodies Correlational evidence versus cause-effect relationships, 160-165 pseudoreplication, 167-168 value, 164-165 Corvus monedula, counting behaviors. 444 Cottontop tamarins, sex ratios, 124 Counting, see also Representation of quantities, apes algorithms, 440, 441 animal categorization, 437 conceptualization, 438-441 definitions, 436-437 productive skills, 44-445 receptive number comprehension, 445-447 timing, 438 motor monitoring, 443-445 principles, 436-441 Courtship behavior, adaptive significance, 7-8 Cowbirds, repertoire size, 369 Crabeater seals, vocal learning, 77 Crickets, sexual selection, 13 Crocuta crocuta, see Spotted hyenas Cryptic species, 318-319 acoustic divergence, 319, 339-346 biodiversity, 319, 346-348 echolocating bats, 335-337 genetic differences, 345-346 morphological differences, 344, 346 nonecholating animals, acoustic signals, 337-339 olfactory signals, 319
467
lNDEX
Cynomys gunnisoni, vocal learning, 75 Cyrtodiopsis dalmanni, sexual selection, 35
D Darwin’s finches, mating behavior, 82 Deception research, chimpanzees, 446-457 Deer, sexual selection, 12 Delphinapterus leucas, vocal learning, 66 Demographic stochasticity, 112 versus environmental stochasticity, 112, 133 individual strategies, 122-126 Development time, sexual selection, 35 Dialects mammals, 83-84 starlings, 378-379 Dimorphism, see Mammals, dimorphic; Sex differences; Sexual selection Dipodomys spectabifis dispersal behavior, 209 sound production learning, 75, 80 Disease transmission avoidance, sexual selection, 33 Disparity interference effect, chimpanzees. 451-454,455-457 Dispersal ground squirrels, 190-191 consequences, 234 females, 206-208 males. 205-206 proximal causes, 183-184, 215-223 testosterone, 215-216, 221, 229-230 hibernation, 218-219 inbreeding, 120-121, 125-126, 182 individual strategies, 125-126 mammals, dimorphic, 181-184, 234-238 consequences. 228-232 models, 193-195,216-219,235 process, 191-192 proximal causes, 208-215 target destination, 191-193 timing, 193-195.216-219 natal (primary), 181 primates, 118-120, 121 secondary, 193, 205 spotted hyenas, 187-189 consequences, 183,232-234
females, 202-205,223-224.225-226, 233-234 males, 195-205,224,226-227,232-233 maternal rank, 196, 198, 202, 203, 232 proximal causes, 223-227 Dispersal restlessness, 214 Disruptive selection, 339-340 Diversity, hidden, 319, 346-348 Division of labor, insects, 285-286 age-related, see Age polyethism DNA analyses, 137, 138-142, 141 Dogs contextual learning, 59 infantile amnesia, 266 Dolphins cognitive capacities, 90 contextual learning, 59 vocal learning, 66, 67-68, 79, 80, 81-82, 83 Domestic cats, vocal learning, 74-75 Dominance, see also Aggression; Territorial defense behavioral variation, 110, 128-129 birds, song repertoires, 396-397.404. 413-414 dispersal, 210 sexual selection, 35 spotted hyenas, 185-187,225-226, 232-233 Dowitchers, genetic differences, 345 Drosophila melanogaster brain abnormalities, 302 sexual selection, 42 Dung flies, mating behavior, 8 Dunnocks, sexual selection, 403 Dwarf mongooses, dispersal behavior, 194. 21 1, 230
E Echolocation calls acoustic resource partitioning, 330-335 natural selection, 340-341 pipistrelle bats, 320-325 Ecological determinism, 131-133 Ecological variation, see Environmental variation Ecology behavioral, see Behavioral ecology, primate
468
INDEX
Ecology (continued) genetics and, 126-137 reproductive, steroid assays, 137, 142-143 Elephant seals sexual selection, 12 vocal learning, 65-66,79 Encoding, rats, 268-269,277-278; see also Memory Endocrinology, behavioral, 235 Energy reserves dispersal behavior, 214-215,216-219, 220,221-222,224 estrus, 219 Environmental stochasticity, 16, 112-114, 133-134 Environmental variation behavioral interactions, 132-134 conservation biology, 112-114, 126, 133-134 dispersal behavior, 208-209, 222-223, 225 genetic interactions, 134-137 life history and, 112-114 primate behavioral ecology, 115-117, 126,131-133 Eptesicus fuscus, vocal learning, 63, 80 Erignathus barbatus, vocal learning, 65, 77 Error, statistical, 177-178 Escape conditioning, rats, retention, 265 Estradiol, 410-411 Estrus, energy reserves, 219 Ethics, experimental, 162 Ethology, bird song in, 355-356 European honey bees age polyethism behavioral development, 286,288, 289-290.301-306 genetic influences, 289 neuroanatomical plasticity, 301-302, 303-306 physiological development, 287, 288 juvenile hormone, 287 allatectomy studies, 304-305 colony manipulations, 300-301 developmental changes, 289-290, 297-300 experimental treatments, 293-297 functions, 290-291 mechanisms of action, 291, 305-306, 309-310 neuroanatomical plasticity, 303-306 seasonal changes, 301 swarming, 301
European starlings, song, 357-358, 417-423 Costs, 420-421 females, 418 mate preference, 394-395, 408-412, 415-417 neural substrates, 381-384 organization, 361-362 song recognition, 403 song sharing, 366, 377,378 territorial defense, 405 testosterone and, 371-372 functions, 391-405 female attraction, 391-396, 400-401 non-breeding season, 403-404 ovarian cycle stimulation. 401-402 relationship to song type, 407-408 sperm competition, 397-401, 421 territorial, 396,404,405,412-414 hormonal influences females, 371-372 neural substrates, 374, 386-387 seasonal variation, 403-404 vocal learning, 373-374 memory, 369-370, 372-373 mimicry, 362-363 heterospecific, 379-380.421 human vocalizations, 363, 366 neural substrates, 380-391 age differences, 388-389 area X, 374,381,383-384,385-386 behavioral variation, 387-391 seasonal variation, 384-387 sex differences, 381-384, 387 song control system, 380-384 species-specific sounds, 379 testosterone and, 374, 386-387 organization, 358-362, 364-365 postjuvenile molt, 358, 374-375 repertoire size, 361, 362, 364, 367-372 age, 371,418 costs of breeding, 420-421 female choice, 408-412, 415-417 health condition, 409 mortality, 368 neural correlates, 387-389 parasite load, 418-419 song bout length, 407 territorial defense, 412-414 variability, 406-407 repertoire turnover, 370-372, 373 seasonal variation, 384-387.403-404, 419
INDEX
sexual selection, 414-417, 420 social influences, 366-367, 371 breeding, 402 dominance relationships, 386-397, 404, 413-414 sensitive phase, 375-376 song sharing, 366-367.404 testosterone level, 373-374 song bout length, 388-389, 407,408 song rate, 420 vocal learning, 76 dialects, 378-379 isolation experiments, 363-365, 372-376 live versus tape tutoring, 366, 375-376 neural substrates, 381, 383-384 song sharing, 366, 377, 378 species-specific, 378-379 testosterone, 373-374 timing, 367-376,418 tutor choice, 375,376-378 Eusocial insects age polyethism, 306-309 division of labor, 285-286 Evaluative processes. symbols and, 457-458 Evolution convergent, 86-87 ontogenetic processes, 129 population size, 130-131 Evolutionary theory, applications to primates, 104-106, 127-129, 134-135 Experimental design, 159-180 confounding effects, 168-169 experimental artifacts, 174-175 nonindependence, 165-168 observer bias. 169-174 proving null hypothesis, 177-178 unjustified conclusions, 160-165 unsuitable controls, 175-177 Extinction probabilities, see Population viability analyses Extragroup matings, primates, 120-121, 140 Extrapair copulation, birds, 400-401, 412
F
Fecundity male versus female, 11-12 sexual selection, 33, 40-41 Felis carus, vocal learning, 74-75
469
Female mate preference, see also Sexual selection birds, 394-395.408-412,415-417 evolution, 32-44 Fisherian process, 34.35-37. 341-342, 416-4 17 good genes selection, 31, 34-35, 36, 37-39,42-44,417 exaggerated male characters, 2, 17-19 male dispersal, 212-213, 226-227 resource-based, 32,343 seaweed flies, see Seaweed flies, female mate preference speciation rate, 342-343 sympatric speciation, 341, 344 Females abduction, 210 dispersal behavior ground squirrels, 206-208 mammals, 192,208-209,210,223 primates, 110, 119, 120, 125 spotted hyenas, 202-205,223-224, 225-226.233-234 fecundity, 11-12 mate choice, see Female mate preference Fertilization window, birds, 399-400 Finches mating behavior, 82 sex differences, 381 vocal learning, 376 Fin whales, vocal learning, 77 Fish, sexual selection, 13, 17, 25, 41 Fisherian process, female choice acoustic signals, 341-342, 416-417 male size, seaweed flies, 34, 35-37 Fission, group mammals, 192,202-203,209, 225 primates, 136-137 Flavor conditioning, rats, retention, 262-263,267 Flies, see also Seaweed flies brain abnormalities, 302 mating behavior, 8 sexual selection, 32, 35, 42 Flycatchers prey selection, 334 sexual selection, 32, 403 Food availability, dispersal behavior, 208-209,210-211,222-223,225-226 Forest guenons, extragroup matings, 120 Forgetting, see Infantile amnesia; Memory
470
INDEX
Frogs, acoustic signals cryptic species, 337-338 sexual selection, 32, 33-34, 41, 341 speciation, 342 Fruit flies brain abnormalities, 302 sexual selection, 42 Fusion, group, mammals, 192 G
Gasterosteus aculeatus, sexual selection, 34,35 Genetic correlations female mate preferences, 34-36 preference and trait, 36, 38 Genetic divergence, cryptic species, 345-346 Genetic drift, 131 Genetic mutations, 130-131 Genetic paternity analyses, 137, 138-142 Genetics and ecology, 126-137 environmental assumptions, 131 -134 environmental-genetic interactions, 134-137 genetics assumptions, 127-131 Genetic variance, coefficient of, 28-32, 42-44 Genetic variation conservation biology, 126, 129-131 migration, 125 primate behavioral ecology, 126, 127-129 versus reproductive isolation, 106-108 Geographic variation. vocal communication, 62, 83-84 Geospiza spp., mating behavior, 82 Gestural communication systems, 88 Gibbons extragroup matings, 120 vocal learning, 72-73, 79 Golden hamsters, estrus. 219 Good genes sexual selection, female choice, 31,34-35,36,37-39,4244,417 Gorillas dispersal behavior, 194 reproductive strategies, 114 Gray seals, genetic analyses, 138 Great apes numerosity, see Representation of quantities, apes
taxonomic classification, 108 vocal learning, 73-74 Greater horseshoe bats prey selection, 334 vocal learning, 63-64 Great reed warblers, song repertoire, 412, 417 Great tits acoustic signals, 317 sexual selection, 34-35 song repertoire, 368 vocal learning, 78, 85 Green frogs, sexual selection, 32 Ground squirrels, see Belding’s ground squirrels Group fission mammals, 192,202-203, 209, 225 primates, 136-137 Group functions, vocal learning coordination, 77-78 recognition, 82-83 Group fusion, mammals, 192 Group-level analyses, generalizability, 134-135 G r y h s integer, sexual selection, 13 Guinea pigs, contextual learning, 59 Gulls, individual recognition, 80 Gunnison’s prairie dogs, vocal learning, 75 Guppies, sexual selection, 33, 35-36
H Habitat acoustic signals, 85. 341 dispersal behavior, 208 inbreeding depression, 130 primate adaptations, 115-117 sexual selection, seaweed flies, 42-44 sympatric speciation, 340 Habituation, 254 experimental procedures, 174-175 rats, retention, 254-255 Halichoerus grypus, genetic analyses, 138 Hammer-headed bats, vocal calls, 78 Hanging-flies, sexual selection, 32 Harbor seals, vocal learning, 64,77 Harp seals, vocal learning, 65 Heterosis, sexual selection and, 7, 31, 37
INDEX
Hibernation, conflict with dispersal, 218-219 Hipposideros spp. acoustic resource partitioning, 334 cryptic species, 336 Hirundo rustica prey selection, 334 sexual selection, 34 Honey bees, see European honey bees; Stingless honey bees Hormonal assays, 137, 142-143 Hormones, see also Juvenile hormone birds females, 371-372 song repertoire, 356, 374, 386-387, 403-404 vocal learning, 373-374 dispersal behavior, 213-214, 215-216. 221,223-224, 235 House flies, mating behavior, 8 House wrens, song repertoire, 402 Howler monkeys dominance ranks, 128-129 reproduction, 113 sex ratios, 125 Humans evolution language, 87-88 primate models, 104-106 infantile amnesia, 251-252, 263 representation of quantities. 436-437, 439-440,459 sensory limitations, 317-319, 346-347 Hummingbirds, vocal learning, 60 Humpback whales, vocal learning, 68-69, 77. 79, x4 Hydrrrrga lepconyx. vocal learning, 65, 77 Hyenas, 184; see a b o Spotted hyenas Hylobates spp., vocal learning, 72-73 Hyperallometry. 27, 28 Hypoallometry, 27-28 Hypopharyngeal glands, honey bees, 287, 291. 292 Hypsignarhirs monsrrosus, vocal calls, 78
I Improvisation contextual learning, 61 vocal learning, 61, 80, 361
47 1
Inbreeding confounding variables, 131 dispersal, 120-121, 125-126, 182 Inbreeding depression, 130-131 Incentive disparity effect, chimpanzees, 451-454,455-457 Independence, statistical, 165- 168 Indian langur monkeys, habitat adaptations, 116 Indigo buntings, vocal learning, 78 Individual-level analyses generalizability, 134-135 in primates, 121-126 sampling methods, 121-122 Individual strategies demographic stochasticity and, 122-126 dispersal, 125-126 sex ratios, 123-125 Infantile amnesia, 251-254, 267-279 adaptive significance, 275-278 dogs, 266 environmental context, 264-265, 270-275 humans, 251-252.263 mice, 265-266 postacquisition processes, 270-275 rats long-term memories, 263-267, 269-275 short-term memories, 260-261 stimulus selection, 268-269, 277-278 Infections, parasite birds, 418-419 primates, 133-134 Inheritance patterns, DNA and, 141 Insects acoustic signals cryptic species, 337 sexual selection, 16, 341 eusocial, 285-286, 306-309 learning and memory, 302 mushroom bodies, 301-303, 305 pest management, 291 Instrumental learning, rats, retention, 261-262,264 Interspecific variation, primates, 106 Intraejaculate competition, 15-16 Intraspecific variation behavioral ecological view, 115 primates, 105, 106, 108-112, 116 Isopods, sexual selection, 13, 25
472
INDEX 1
Jackdaws, counting behaviors, 444 Japanese macaques extragroup matings, 120 genetic analyses, 138 vocal learning, 71 Juvenile hormone analogues, 291 assay techniques, 297-298 association with polymorphisms and polyphenisms, 291-292 multiple roles, 309-310 regulation of age polyethism, 287 allatectomy studies, 304-305 colony manipulations, 300-301 developmental changes, 289-290. 297-300 experimental treatments, 293-297, 308 functions, 290-291 mechanisms of action, 291, 305-306, 309-310 neuroanatomical plasticity, 303-306 seasonal changes, 301 swarming, 301
neurobiological basis, 255, 302 rats, ontogeny. 254-255 retention, see Memory vocal, see Vocal learning Lekking species, paradox of, 33 Lemurs dispersal behavior, 119,212 extragroup matings, 120 group fission, 136 Leopard frogs, cryptic species, 337-338 Leopard seals, vocal learning, 65, 77 Lepomis macrochirus, sexual selection, 13 Leptin, dispersal behavior, 219, 235 Leprodacrylus mystaceus, cryptic species, 338 Lepronychores weddelli, vocal learning, 64-65, 77, 78-79 Lesser spear-nosed bats, vocal learning, 63 Life history variables, environmental variation and, 112-114 Limnodramus spp., genetic differences, 345 Lions, dispersal behavior, 230-231 Lobodon carcinophagus. vocal learning, 77 Luscinia spp., cryptic species, 338 Lymphocyte counts, dispersal, 229
K
M
Kangaroo rats dispersal behavior, 209 sound production learning, 75, 80 Kangaroos, dispersal behavior, 208 Katydids, sexual selection, 32 Killer whales, vocal learning, 68, 82 Kin selection, 228
L Lacewings, cryptic species, 337 Ladybird beetles, mate choice, 24, 45 Language evolution, 87-88 symbolic, 445 Learning action-based, 62-63, 74 contextual, 59, 61,62-63. 74, 85-86 insects, 302 language, 87-88,445
Macaques dispersal behavior, 208,212 extragroup matings, 120 genetic analyses, 138, 140 group fission, 136 habitat adaptations, 116 sex ratios, 124 vocal learning, 71-72 Males affiliative associations, 111-112, 192 dispersal behavior ground squirrels, 205-206 mammals, 192, 211-213 primates, 119-120, 125 spotted hyenas, 195-205,224, 226-227, 232-233 fecundity, 11-12 mate choice, 16-17, 28 parental investment, 32 Mammals communication systems, 85-86. 230
INDEX
dimorphic, dispersal, 181-184, 234-238 consequences, 228-232 models, 193-195, 235 process, 191- 192 proximal causes, 208-215 sex differences, 191-195, 208-215, 234-238 target destination, 191-193 timing, 193-195 group fission, 192, 202-203, 209, 225 marine, vocal learning, 78-79, 80 philopatric, 111-112, 119, 182, 183, 228 reproductive strategies, 114 vocal learning, 88-90 convergent evolution, 86-87 flexibility, 85-86 functional significance, 77-78, 79. 80-82, 83, 84-85 survey, 63-75 Marine invertebrates, cryptic species, 338-339 Marine mammals, vocal learning, 78-79, 80 Marmosets reproductive strategies, 114 sex ratios, 125 vocal learning, 71, 89 Marsh wrens, repertoire size and brain space, 387 Marsupials, dispersal behavior, 194, 230 Mate choice dispersal behavior, 211-213, 226, 236-237 female, see Female mate preference male. 16-17, 226 Mate guarding, 15,401,421 Maternal rank, spotted hyenas, 196, 198, 202, 203, 232 Mating behavior adaptive significance, 7-8 dispersal, 182-183.231-232 seaweed flies, 5, 7-12, 19, 42-44 Mating patterns extragroup, 120-121, 140 reproductive success and, 118 Megaptera novaeangliae, vocal learning, 68-69,77,79, 84 Memory, see also Infantile amnesia birds, 369-370, 372-373 childhood trauma, 251-252, 279 insects, 302
473
ontogeny, 251-279 humans, 251-252,263 rats, 254-267 spatial, sex differences, 228-229 Mesocricetus auratus, estrus, 219 Methoprene, 291,293-297,308 Mice dispersal behavior, 211 inbreeding, 135 infantile amnesia, 265-266 Migration, 125; see also Dispersal Mimicry, 362-363 heterospecific, 379-380,421 human vocalizations, 363,366 Mirounga spp., see Elephant seals Mites, 40 Molospiza melodia genetic analyses, 135 repertoire size, 368 Molothrus ater, repertoire size, 369 Monkeys dispersal patterns, 119, 120 dominance ranks, 128-129 extragroup matings, 120 habitat adaptations, 116 reproduction, 113 sex ratios, 123, 124, 125 vocal learning, 69-70, 69-72, 70, 79 Monogamy maintenance strategies, birds, 405 Morphology cryptic species, 344, 346 natural selection, 343 polymorphisms, 13 Mother-pup recognition, vocal learning, 81-82 Motor monitoring, counting, 443-444 Mountain gorillas dispersal behavior, 194 reproductive strategies, 114 Muriquis dispersal patterns, 120 habitat adaptations, 116-117 males affiliative associations, 111-112, 119 mating success, 119 mate choice, 120-121 parasite infections, 133, 134 reproduction, 113-114 sex ratios, 124
474
INDEX
Muriquis (continued) species averaging problems, 109 taxonomic classification, 107 Musca domesrica, mating behavior, 8 Muscicapa striata, prey selection, 334 Mushroom bodies learning and memory, 302 plasticity, 301-302, 305 structure, 302-303 Mustela erminea, dispersal behavior, 209 Myoris spp. cryptic species, 336-337 morphological differences, 328 resource partitioning, 328 Mysticetes, vocal learning, 68-69, 77, 81-82
Olfactory signals, cryptic species, 319 Olive baboons dispersal behavior, 194, 212, 231 habitat adaptations, 116 sex ratios, 123 Ololygon rubra, sexual selection, 33, 41 Ommatophoca rossi, vocal learning, 77 One-one counting principle, 436 Ontogenetic processes evolutionary consequences, 129 infantile amnesia, 275-278 Orcinus orca, vocal learning, 68, 82 Order-irrelevance counting principle, 436-437 Ovulation, bird song, 401-402
N
P
Natal dispersal, 181 Natural selection acoustic signals, 340-341 appearance, 342 morphology, 343 pleitropy, 41-42 sexual selection and, 41-42, 44-45, 47-48 Neotropical frogs, sexual selection, 33, 41 Nest building, bird song, 401-402, 415-416 Neurobiological research birds, song control center, 355,374, 380-391,419-420 honey bees, age polyethism, 301-302, 303-306 learning, 255 rats. memory development, 262 New World atelins, 111-112 Nightingales. cryptic species, 338 Nonassociative memory, ontogeny, rats, 254-257 Null hypothesis, 177-178 Numerical symbols, see Counting; Representation of quantities, apes
Panorpa, sexual selection, 13 Pan paniscus, vocal learning, 73 Pan troglodytes, see Chimpanzees Paper wasps, age polyethism, 307-308 Papio cynocephalus, dispersal behavior, 194, 231 Paracerceis sculpra, sexual selection, 13, 25 Parasite infections, in primates, 133-134 Parental investment, 1, 32 Parrots numerical categorization, 437 vocal learning, 60 Partitioning, 436 Parus major, see Great tits Passerina cyaneu, vocal learning, 78 Passerine birds sex differences, 381 speciation, 342 vocal learning, 84, 342 Passive avoidance, rats, retention, 265 Patas monkeys, extragroup matings, 120 Paternity analyses behavioral extrapolations, 138 genetic, 137, 138-142 Pavlovian conditioning, rats, retention, 259-260 Peromyscus leucopus, dispersal behavior, 21 1 Pheasants, sexual selection, 34-35 Philopatric mammals dispersal, 182, 183,228 rnuriquis, 111-112, 119
0
Observer bias, 169-174 Ochotona princeps, vocal learning, 75 Odobenus rosmarus, vocal learning, 77 Odontocetes, vocal learning, 66-68, 81, 82 Offspring viability, See Viability indicator mechanisms
475
INDEX
Phocids. vocal learning, 64-65, 77 Phyllostomus discolor, vocal learning, 63 Phylogenetic analyses, 108-112 use of DNA, 141 Physalaemus pustulosus, sexual selection, 33-34. 41 Picoides borealis, conservation management, 102 Pied flycatchers, sexual selection, 32, 403 Pikas, vocal learning, 75 Pileated gibbons, vocal learning, 72-73 Pinnipeds, vocal learning, 64-66 Pipistrelle bats. 319-320 behavioral differences, 329 calls echolocation, 320-325, 321-322 social, 325-327 songflight, 327 genetic differences, 330 mating strategy, 327 morphological differences, 327-329 phonic types, 322-325 resource partitioning, 328, 329 speciation, 343-344, 346 allopatric, 339, 343-344 disruptive selection, 340, 344 sexual selection, 343 Planthoppers, cryptic species, 337 Plecotus spp., morphological differences, 328 Pleitropy, 33-34 sensory exploitation, 41 side effect of natural selection, 41-42 Poecilia reticulata, sexual selection, 33, 35-36 Pofisfesspp.. age polyethism, 307-308 Pofybia occidentalis, juvenile hormone treatment, 308 Polyethism, see Age polyethism Pol ymorphisms behavioral, 286 chromosonal inversion, 6-7 juvenile hormone, 291 -292 male morphology, 13 sympatric speciation, 340 Polyphenism, 286-287; see also Age polyethism Ponies, dispersal behavior, 211 Population identity, vocal learning, 83-84
Population-level analyses, 114-121 appropriateness, 120, 122 behavioral variation, 117-121 ecological variation, 115-117 environmental-genetic interactions, 135-137 genetic variation, 122, 126, 129-131 models, 117, 120, 144 Population size, evolution and, 130-131 Population viability analyses behavioral variables, 118 environmental-genetic interactions, 136 need for behavioral and ecological data, 101-103, 118-121 reliability, 117-1 18 single-group values, 117 Postjuvenile molt, 358, 374-375 Power, statistical, 177-178 Prairie dogs, vocal learning, 75 Presbytis entellus, habitat adaptations, 116 Prey detection, pipistrelle bats, 329, 334 Primate behavioral ecology, see Behavioral ecology, primate Primates communication systems, 79, 89 contextual learning, 59 dispersal patterns, 118-120, 121, 125 genetic variability, 141 habitat adaptations, 115-117 vocal learning, 69-74, 79 Procyon lotor, numerical categorization, 437 Prunella modularis, sexual selection, 403 Pseudoreplication, 165-168 Ptilonorhynchus, sexual selection, 33 Puberty, dispersal behavior, 213-214 PVAs. see Population viability analyses Pygmy marmosets, vocal learning, 71, 89
Q Quantity representations, see Representation of quantities, apes
R Raccoons, numerical categorization, 437 Rana pipiens, cryptic species, 337-338
476
INDEX
Rats contextual learning, 59 infantile amnesia long-term memories, 263-267, 269-275 short-term memories, 260-261 learning, ontogeny, 254-255 memory, ontogeny cholinergic mediation, 262 long-term, 263-267 nonassociative, 254-257 short-term, 257-263 representation of quantities, 437, 444-445 stimulus selection, 268-269, 277-278 Reciprocal altruism, 228 Recognition, vocal learning and, 79-83 Red-chested moustached tamarins, vocal learning, 70 Red-cockaded woodpeckers, conservation management, 102 Red colobus monkeys, dispersal patterns, 120 Red deer, sexual selection, 12 Red kangaroos, dispersal behavior, 208 Red squirrels, dispersal behavior, 209 Red-winged blackbirds repertoire size and brain space, 387 sexual selection, 32 vocal learning, 76 Representation of language, 445 Representation of quantities, apes, 435-462 advantages, 457,458 arithmetic reasoning, 438-441 categorization, 437 compared to human children, 436-437, 439-440,459 conceptualization and reasoning, 438-441 counting principles, 436-441 evaluative processes and, 457-458 motor monitoring, 443-444 productive manipulation, 441-445 quantity judgments, 447-457 response interference, 451-454, 455-458 symbolic facilitation, 457-458 receptive manipulation, 445-447 Reproductive ecology, steroid assays, 137, 142-143 Reproductive strategies individual, 122-125
isolation, 106-108, 327 mammals, 114 Reproductive success, genetic analyses, 138- 140 Reproductive suppression conservation management, 102 versus dispersal, 121 Resource abundance, see Food availability Resource acquisition, female mate preference, 343 Resource partitioning echolocation, 330-335 resource acquisition characters, 339 Response bias, chimpanzees, 451-454, 455-457 Response predispositions, 457 Retention, see Infantile amnesia; Memory Rhesus macaques group fission, 136 vocal learning, 71-72 Rhinolophid bats acoustic resource partitioning, 334 prey selection, 334 vocal learning, 63-64 Rhogeesa spp., cryptic species, 336 Ring-tailed lemurs dispersal behavior, 119, 212 extragroup matings, 120 group fission, 136 Ross seals, vocal learning, 77
S
Saddle-backed tamarins, vocal learning, 70 Saguinus spp., vocal learning, 70 Sairniri sciureus, vocal learning, 69-70 Salmon, sexual selection, 13 Sample size, statistical power and, 177-178 Sampling methods, in primatology, 121-122 Sardinian warblers, song repertoire, 361 Satin bowerbirds, sexual selection, 33 Sca fophaga stercoraria mating behavior, 8 sexual selection, 13 Scorpion flies, sexual selection, 32 Scrambles, 132 Sea lions, contextual learning, 59 Seals genetic analyses, 138
INDEX
sexual selection, 12 vocal learning, 64-66, 77, 78-79 Seaweed flies, 2-6 chromosonal inversion systems, 6-7 female mate preference, 20-24, 36-39 male size, 13, 18-19, 25-26, 28-32, 36 fecundity female, 11, 12, 17, 41 laboratory culture and, 12 male, 10, 17 female mate preference acceptance rates, 18-21, 23-24 genetic basis, 20-24, 36-44 laboratory culture and, 24, 45-46 male size. 17-19, 21-22, 23-24 tidal variation and, 42-44.45 female size, 16-17, 28 habitat differences, 5, 21-22, 30, 42-44 male size female choice, 17-19, 21-22 genetic basis, 25-26, 28-32, 36 mating advantages, 13-14 tidal variation and, 21-22, 30 mating behavior, 7-12 interactions between components, 19 rejection responses, 10, 11, 16-17, 199 tidal variation and, 5, 42-44 sexual selection. 12-19 direct, 39-41 habitat, 42-44 indirect, 36-39 interactions between forms, 36, 44-48, 49 intrasexual, 13-16 pleiotropic, 41-42 Secondary dispersal, 193, 205 Sedge warblers song repertoire, 392, 414, 420 vocal learning, 76 Sensitization, 254, 255-256 Serinus canaria, see Canaries Sesquiterpenoid juvenile hormone, see Juvenile hormone Sex differences, see also Sexual selection dispersal consequences, 228-234 evolution, 211 ground squirrels, 205-208, 215-223, 234 mammals, 191-195.208-215, 234-238
477
spotted hyenas, 195-205.223-227, 232-234 evolution, 211, 228, 235, 237 social behavior, 228, 230-231 spatial memory, 228-229 Sex ratios individual reproductive strategies, 123- 125 in New World atelins, 111-112 Sexual dimorphism, see Mammals, dimorphic; Sex differences; Sexual selection Sexual selection, 1-2, see also Sex differences acoustic signals, 341-343, 344. 414-417 birds, 18,76-78, 341, 356-357, 381, 414-417,420 direct models, 33-34, 39-42, 47-48, 415-416 heterosis, 7, 31, 37 indirect models, 36, 47-48 Fisherian process, 34, 35-36, 37, 341-342,416-417 good genes, 31.34-35, 36, 37-39, 42-44,417 interspecific, 12, 76-78, 410, 414-417 intraspecific, 1, 12, 78-79 contests, 12-13 endurance rivalry and scrambles, 13-14 sperm competition, 14-16,397-401, 421 male size, 25-32 mate choice, 16-19 female, see Female mate preference male, 16-17 natural selection and, 41-42, 44-45, 47-48 seaweed flies, 12-19, 36-49 sympatric speciation, 340-343, 344 vocal learning, 76-79 Siamangs, vocal learning, 72 Sibling species, 318 Size disparity interference, chimpanzees, 451-454,455-457 Social relationships, vocal learning, 79 Social structure dispersal behavior, 209-213, 222, 225-226,227 mammals, 182-183, 191-192 environmental-genetic interactions, 134
478
INDEX
Social traits behavioral variation, 110 sex differences, 228, 230-231 Social wasps age polyethism, 306, 307-308 juvenile hormone treatment, 308 Songbirds, vocal learning, 60, 76, 82 Song control center, birds, 355, 374, 380-391,419-420 Song sparrows genetic analyses, 135 repertoire size, 368 Space competition, female dispersal behavior, 223 Sparrows aggressive singing, 404 genetic analyses, 135 repertoire size, 368 song learning, 376 Spatial memory, sex differences. 228-229 Speciation allopatric, 339, 343-344 sympatric acoustic divergence, 84-85, 339-343, 344 disruptive selection, 339-340 sexual selection. 340-343 Species-level analyses, 104-114 conservation biology. 106-108, 112-114 historical perspectives, 104-108 primate behavioral ecology, 104-106 sources of error, 108-112 Species-specific averages, 108-1 10 Species-specific life histories, 112-114 Sperm competition seaweed flies, 14-16 starlings, 397-401, 421 Sperrnophilus beldingi, see Belding’s ground squirrels Spider monkeys reproduction, 113 sex ratios, 123, 124 Spotted flycatchers, prey selection, 334 Spotted hyenas, 183. 184-189 aggressive behavior, 186-1 87 clan fission, 202-203, 22.5, 233-234 dispersal behavior, 187-189 consequences, 232-234 females, 202-205, 223-224, 225-226. 233-234 males, 195-205,224,226-227.232-233
maternal rank, 196, 198, 202, 203, 232 proximal causes, 223-227 dominance hierarchy, 185-187, 225-226. 232-233 sex ratios, 125 Squirrel monkeys, vocal learning, 69-70 Squirrels, see Belding’s ground squirrels Stable-order counting principle, 436 Stalk-eyed flies, sexual selection, 35 Starlings, see European starlings, song Statistical power, sample size and, 177-178 Steroid assays, 137, 142-143 Steroid hormones, see Hormones Sticklebacks, sexual selection, 34, 35 Stimulus selection. rats, 268-269, 277-278 Stingless honey bees, 306 age polyethism, 306 queen-worker differentiation, 286 Stoats, dispersal behavior, 209 Stochastic variation interactions, 112, 133 reproductive strategies and, 124-125 Stress dispersal-induced, 229 steroid assays, 143 Srumus vulgaris, see European starlings, song Subject selection, 168 Suboscines, vocal learning. 60 Swallows prey selection, 334 sexual selection, 34 Swordtail fish, sexual selection, 13, 25, 41 Sylvia melanocephala, song repertoire, 361 Symbols evaluative processes and, 457-458 manipulation, 445 numerical, see Counting: Representation of quantities, apes Sympatric speciation acoustic divergence, 84-85.339-343, 344 disruptive selection, 339-340 sexual selection, 340-343. 344 Synchronous breeding, bird song, 402
T Tagging, 436 Tamarins sex ratios, 124 vocal learning, 70
479
INDEX
Tamiasciiiris hudsonicus, dispersal behavior, 209 Taste aversion learning, rats, retention, 262-263,267 Taxonomic classifications controversies, 106- 108 genetic analyses. 140-141 human sensory limitations, 317-319, 346-347 Temporal polyethism, see Age polyethism Territorial defense, see also Aggression; Dominance birds, song. 396, 404, 405, 412-414 vocal learning, 78-79 Testosterone birds costs, 356 females, 371-372 song control nuclei, 374, 386-387 vocal learning. 374 dispersal bchavior, 215-216, 221. 224, 229-230, 235 Thinoseiiis fiicicolu. 40 Thrasher, song repertoire, 361 Three-spined sticklebacks, sexual selection, 34.35 Thryothoriis nigricapillus, sex differences, 381 Timing, animal, 438; see ulso Animal counting Toothed whales, vocal learning, 66-68. 81.82 Toxostomcr refim, song repertoire, 361 Trace conditioning, rats, retention, 257-259.262 Traumatic childhood memories, see also Infantile amnesia retention, 251-252 role of suggestion, 279 Treecreepers. acoustic signals, 317 Troglodytes uedon. song repertoire, 402 Tufted capuchin monkeys, habitat adaptations, I16 Tungara frogs, sexual selection, 33-34, 41 Titrsiops trimcatii.7, vocal learning, 66, 67-68, 79, 81-82,83
U Ultraviolet light, birds, 318
V Vervet monkeys dispersal patterns, 119 vocal learning, 70 Vervets, dispersal behavior, 208 Viability indicator mechanisms, female choice and, 31,34-35, 36,37-39, 42-44.417 Visual signals, versus acoustic signals, 3 17-31 8 Vocal communication, contextual learning and, 59,85-86 Vocal learning birds, see Birds, vocal learning versus contextual learning, 59 copying new sounds, 86 duration and amplitude modifications, 60. 85-86 flexibility, 85-86 frequency modifications, 60, 86 functional significance, 75-85 familiar or group recognition, 82-83 group coordination, 77-78 habitat matching, 85 individual recognition, 79-82 population identity, 83-84 sexual selection, 76-79 social relationship maintenance, 79 speciation, 84-85, 342 territorial defense, 78-79 future research, 88-90 humans, 87-88 improvisation, 61, 80, 361 mammals, see Mammals, vocal learning role of mobility, 86-87 testosterone level, 374 types of evidence, 61-63 VORTEX, 117, 118
W Walrus, vocal learning, 77 Warblers acoustic signals. 317 cryptic species, 338 genetic differences, 345-346 song repertoire, 361, 392, 412, 414, 417, 420 vocal learning, 76
480
INDEX
Wasps age polyethism, 306, 307-308 behavioral polymorphisms, 286 juvenile hormone treatment, 308 Water striders, sexual selection, 41-42 Weddell seals, vocal learning, 64-65, 77, 78-79 Whales, vocal learning. 66-69, 77-78, 79, 81-82, 84 White-crowned sparrows aggressive singing, 404 song learning, 376 White-footed mice, dispersal behavior, 21 1 White-handed gibbons, vocal learning, 72-73 Woodpeckers, conservation management, 102 Wrens repertoire size and brain space, 387 sex differences, 381 song repertoire, 402
X Xenopus spp., cryptic species, 338 Xiphophorus spp, sexual selection, 13, 25.41
Y Yellow baboons dispersal behavior, 194, 212, 231 habitat adaptations, 116 sex ratios, 123
Z Zebra finches sex differences, 381 vocal learning, 376 Zonotrichia spp. aggressive singing, 404 vocal learning, 85, 376
Contents of Previous Volumes Volume
IS
Kelationships helwecn Social Organization and Behavioral Endocrinology in ii Monogamous Mammal C. SUE CAK'I'EK. LOWELL L. GE'IZ, AND MAK'I'HA COIIEN-PARSONS
Sex 1)iffcrcnces i n Social Play: 'l'lic Soci;iliiatioii of Sex Roles
MI<'HAEL J. MEANEY. JANE S'I'EWAKI'. A N D WIL.I.IAM W. BEAITY
Lateralization o f Learning Chicks I.. J. R O G E R S
On the Functions o f Play and Its Role in
Behavioral l~evelopment P A l l L MAK'I'IN A N D 'I.M. C'AKO
Circannual Khythnis in the Control of Avian Migrations E B E K H A R D GWINNEII
Sensory Faclors i n thc Hch;ivior;il Ontogciiy 0 1 Altricial Birds S . N. KfIAYlJ'I'IN
The Economics ot Fleeing trom Predators I<. C. YIJENBERG AND L. M. DILL
Food Sloritgc hy Birds and Maniiiials DAVID P. SHEKKY
Social Ecology and Hchavior o f Coyotes MARC HEKOFF A N D MICHAEL C . WELLS
Vocal Affect Signaling: A <-'ompariitivc Approach KLAUS I<. S('HEl<EII
Volume 17
A Respoiisc-Competitioii Mods1 Ilcsigned
to Account for thc Avcraion to Feed on Conspecilic Flesh W. J. CARK AND D A R L E N E F. KENNEDY
Receptive Coiiipetencics of Language. Trained Aniniiils LOlJlS M. HERMAN Self-Generated Experience and the Devclopment ol' Latcralizcd Ncurobchavioral O r -
ganization in Inlmts G E O R G E F. MI('IiEI.
Volume 16 Sensory Organization of Alimentary Dehavior in Ihc Kitten K. V. SHIJLEIKINA-'I'IJKPAEVA Individual Odors among Mammals: Origins and Functions ZLJLEYMA TANG tlALPlN 'I'hc Physiology and Ecology of Puberty Modulation hy Pririizr Pheromones
Bchiivioriil Ecology: Theory into Practice NEIL 13. METCALFE A N D PAT MONAGHAN The Dwarf Mongoose: A Study of Hchavior and Social Structure in Relation to Ecology in a Small, Social Carnivore 0. ANNE E. KASA Ontogenetic Development of Behavior: 'l'hc ('rickel Visual World KAYJOND CAMPAN. GIJY BEIIGNON. A N D MICIiEL LAMBIN
JOllN C;. VANI)ENBERGFI A N D D A V I D M. COPPOLA
481
482
CONTENTS OF PREVIOUS VOLUMES
Volume 18 Song Learning in Zebra Finches (Taeniopygia guttata): Progress and Prospects PETER J. B. SLATER, LUCY A. EALES, AND N. S. CLAYTON
Mode Selection and Mode Switching in Foraging Animals GENE S. HELFMAN Cricket Neuroethology: Neuronal Basis of Intraspecific Acoustic Communication FRANZ HUBER
Behavioral Aspects of Sperm Competition in Birds T. R. BIRDHEAD
Some Cognitive Capacities of an African Grey Parrot (Psirracus erithactcs) IRENE MAXINE PEPPERBERG
Neural Mechanisms of Perception and Motor Control in a Weakly Electric Fish WALTER HEILIGENBERG
Volume 20
Behavioral Adaptations of Aquatic Life in Insects: An Example ANN CLOAREC
Social Behavior and Organization in the Macropodoidea PETER J. JARMAN
The Cicadian Organization of Behavior: Timekeeping in the Tsetse Fly, A Model System JOHN BRADY
The I Complex: A Story of Genes, Behavior, and Population SARAH LENINGTON
Volume 19
The Ergonomics of Worker Behavior in Social Hymenoptera PAUL SCHMID-HEMPEL
Polyterritorial Polygyny in the Pied Flycatcher P. V. ALATALO AND A. LUNDBERG
“Microsmatic Humans” Revisited: The Generation and Perception of Chemical Signals BENOIST SCHAAL AND RICHARD H. PORTER
Kin Recognition: Problems, Prospects. and the Evolution of Discrimination Systems C. J. BARNARD
Lekking in Birds and Mammals: Behavioral and Evolutionary Issues R. HAVEN WILEY
Maternal Responsiveness in Humans: Emotional, Cognitive, and Biological Factors CARL M. CORTER AND ALISON S. FLEMING The Evolution of Courtship Behavior in Newts and Salamanders T. R. HALLIDAY Ethopharmacology: A Biological Approach to the Study of Drug-Induced Changes in Behavior A. K. DIXON, H. U. FISCH, AND K. H. MCALLISTER Additive and Interactive Effects of Genotype and Maternal Environment PIERRE L. ROUBERTOUX, MARIKA NOSTEN-BERTRAND, AND MICHELE CARLIER
Volume 21 Primate Social Relationships: Their Determinants and Consequences ERIC B. KEVERNE The Role of Parasites in Scxual Selection: Current Evidence and Future Directions MARLENE ZUK Conceptual Issues in Cognitive Ethology COLIN BEER Responses in Warning Coloration in Avian Predators W. SCHULER AND T. J. ROPER Analysis and Interpretation of Orb Spider Exploration and Web-Building Behavior FRITZ VOLLRATH
CONTENTS OF PREVIOUS VOLUMES
483
Motor Aspects of Masculine Sexual Behavior in Rats and Rabbits GABRIELA MORAL^ AND CARLOS BEYER
Genetic Correlations and the Control of Behavior, Exemplified by Aggressiveness in Sticklebacks T H E 0 C. M. BAKKER
On the Nature and Evolution of Imitation in the Animal Kingdom: Reappraisal of a Century of Research A. WHITEN AND R. HAM
Territorial Behavior: Testing the Assumptions JUDY STAMPS
Volume 22 Male Aggression and Sexual Coercion of Females in Nonhuman Primates and Other Mammals: Evidence and Theoretical Implications BARBARA B. SMUTS AND ROBERT W. SMUTS Parasites and the Evolution of Host Social Behavior ANDERS PAPE M0LLER. REIJA DUFVA, AND KLAS ALLANDER The Evolution of Behavioral Phenotypes: Lessons Learned from Divergent Spider Populations SUSAN E. RIECHERT Proximate and Developmental Aspects of Antipredator Behavior E. CURIO Newborn Lambs and Their Dams: The Interaction That Leads to Sucking MARGARET A. VINCE The Ontogeny of Social Displays: Form Development, Form Fixation, and Change in Context T. G. GROOTHUIS Volume 23 Sneakers, Satellites, and Helpers: Parasitic and Cooperative Behavior in Fish Reproduction MICHAEL TABORSKY Behavioral Ecology and Levels of Selection: Dissolving the Group Selection Controversy LEE ALAN DUGATKIN AND HUDSON KERN REEVE
Communication Behavior and Sensory Mechanisms in Weakly Electric Fishes BERND KRAMER
Volume 24
Is the Information Center Hypothesis a Flop? HElNZ RICHNER AND PHILIPP HEEB Maternal Contributions to Mammalian Reproductive Development and the Divergence of Males and Females CELIA L. MOORE Cultural Transmission in the Black Rat: Pine Cone Feeding JOSEPH TERKEL The Behavioral Diversity and Evolution of Guppy, Poecilia reticulata, Populations in Trinidad A. E. MAGURRAN, B. H. SEGHERS, P. W. SHAW, AND G . R. CARVALHO Sociality, Group Size, and Reproductive Suppression among Carnivores SCOTT CREEL AND DAVID MACDONALD Development and Relationships: A Dynamic Model of Communication ALAN FOGEL Why Do Females Mate with Multiple Males? The Sexually Selected Sperm Hypothesis LAURENT KELLER AND HUDSON K. REEVE Cognition in Cephalopods JENNIFER A. MATHER
484
CONTENTS OF PREVIOUS VOLUMES
Volume 25 Parental Care in Invertebrates STEPHEN T. TRUMBO Cause and Effect of Parental Care in Fishes: An Epigenetic Perspective STEPHEN S. CRAWFORD AND EUGENE K. BALON Parental Care among the Amphibia MARTHA L. CRUMP An Overview of Parental Care among the Reptilia CARL CANS Neural and Hormonal Control of Parental Behavior in Birds JOHN D. BUNTIN Biochemical Basis of Parental Behavior in the Rat ROBERT S. BRIDGES Somatosensation and Maternal Care in Norway Rats JUDITH M. STERN Experiential Factors in Postpartum Regulation of Maternal Care ALISON S. FLEMING, HYWEL D. MORGAN, AND CAROLYN WALSH Maternal Behavior in Rabbits: A Historical and Multidisciplinary Perspective GABRIELA GONZALEZ-MARISCAL AND JAY S. ROSENBLATT
I S B N O-L2-004526-5
Parental Behavior in Voles ZUOXIN WANG AND THOMAS R. INSEL Physiological, Sensory, and Experiential Factors of Parental Care in Sheep F. LEVY, K. M. KENDRICK, E. B. KEVERNE, R. H. PORTER, AND A. ROMEYER Socialization, Hormones, and the Regulation of Maternal Behavior in Nonhuman Simian Primates CHRISTOPHER R. PRYCE Field Studies of Parental Care in Birds: New Data Focus Questions on Variation among Females PATRICIA ADAIR GOWATY Parental Investment in Pinnipeds FRITZ TRILLMICH Individual Differences in Maternal Style: Causes and Consequences of Mothers and Offspring LYNN A. FAIRBANKS Mother-Infant Communication in Primates DARIO MAESTRIPIERI AND JOSEP CALL Infant Care in Cooperatively Breeding Species CHARLES T. SNOWDEN