Advances in the Study of Behavior, Volume 36 (Advances in the Study of Behavior)

Advances in

THE STUDY OF BEHAVIOR VOLUME 36

Advances in THE STUDY OF BEHAVIOR Edited by

H. Jane Brockmann Peter J. B. Slater Charles T. Snowdon Timothy J. Roper Marc Naguib Katherine E. Wynne-Edwards

Advances in THE STUDY OF BEHAVIOR Edited by H. Jane Brockmann Department of Zoology University of Florida Gainesville, Florida

Peter J. B. Slater

Charles T. Snowdon

School of Biology University of St. Andrews Fife, United Kingdom

Department of Psychology University of Wisconsin Madison, Wisconsin

Timothy J. Roper

Marc Naguib

Department of Biology and Environmental Science University of Sussex Sussex, United Kingdom

Department of Animal Behavior University of Bielefeld Bielefeld, Germany

Katherine E. Wynne-Edwards Department of Biology Queen’s University Kingston, Canada

VOLUME 36

AMSTERDAM • BOSTON • HEIDELBERG • LONDON NEW YORK • OXFORD • PARIS • SAN DIEGO SAN FRANCISCO • SINGAPORE • SYDNEY • TOKYO Academic Press is an imprint of Elsevier

Academic Press is an imprint of Elsevier 525 B Street, Suite 1900, San Diego, California 92101-4495, USA 84 Theobald’s Road, London WC1X 8RR, UK

This book is printed on acid-free paper. Copyright ß 2006, Elsevier Inc. 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. (www.copyright.com), 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-2006 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. 0065-3454/2006 $35.00 Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK: phone: (+44) 1865 843830, fax: (+44) 1865 853333, E-mail: [email protected]. You may also complete your request on-line via the Elsevier homepage (http://elsevier.com), by selecting ‘‘Support & Contact’’ then ‘‘Copyright and Permission’’ and then ‘‘Obtaining Permissions.’’ For information on all Elsevier Academic Press publications visit our Web site at www.books.elsevier.com ISBN-13: 978-0-12-004536-5 ISBN-10: 0-12-004536-2 PRINTED IN THE UNITED STATES OF AMERICA 06 07 08 09 9 8 7 6 5 4 3 2 1

Contents

Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

ix xi

Suckling, Milk, and the Development of Preferences Toward Maternal Cues by Neonates: From Early Learning to Filial Attachment? RAYMOND NOWAK I. II. III. IV. V.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Udder, the Milk, and the Neonate . . . . . . . . . . . . . . . . Milk: An Astonishingly Complex Fluid . . . . . . . . . . . . . . . . Suckling and Early Learning . . . . . . . . . . . . . . . . . . . . . . . . Suckling and the Development of a Preference for the Mother in Sheep . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI. The First Hours After Birth . . . . . . . . . . . . . . . . . . . . . . . . . VII. Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VIII. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 4 17 23 28 40 43 45 47

A Neuroethological Approach to Song Behavior and Perception in European Starlings: Interrelationships Among Testosterone, Neuroanatomy, Immediate Early Gene Expression, and Immune Function GREGORY F. BALL, KEITH W. SOCKMAN, DEBORAH L. DUFFY, AND TIMOTHY Q. GENTNER I. Introduction: Song, European Starlings, and the Neuroethological Approach . . . . . . . . . . . . . . . . . . . . . . . . . II. Description of European Starling Song and Its Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III. Song Control Circuit and the Neuroendocrine Control of Song . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Perception of Song in Starlings . . . . . . . . . . . . . . . . . . . . . . V. Physiological Responses to Song in Starlings . . . . . . . . . . . VI. Functional Basis of Song Preferences in European Starlings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v

59 61 64 71 74 90

vi

CONTENTS

VII. Putting It All Together: Song Production/Perception and Hormones. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VIII. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

104 106 107

Navigational Memories in Ants and Bees: Memory Retrieval When Selecting and Following Routes THOMAS S. COLLETT, PAUL GRAHAM, ROBERT A. HARRIS, AND NATALIE HEMPEL-DE-IBARRA I. II. III. IV. V. VI.

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Foraging Routes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Navigational Memories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Retrieval of Memories Along a Route . . . . . . . . . . . . . Choice of Route and Destination . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

123 124 127 148 155 165 167

Functional Genomics Requires Ecology LARA S. CARROLL AND WAYNE K. POTTS I. The Problem: Many Genes Seem to Be Unnecessary . . . . II. Genes Lacking Phenotypes: Explanations and Experimental Approaches for Their Elucidation . . . . . . . . III. Gene Function Studies Demand Integrative Approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

173 177 203 207 209

Signal Detection and Animal Communication R. HAVEN WILEY I. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Essential Features of Signal Detection . . . . . . . . . . . . . . . . . III. Application of Signal Detection Theory in Experimental Psychophysics . . . . . . . . . . . . . . . . . . . . . . . . . IV. General Assumptions of Signal Detection Theory . . . . . . . V. Specific Assumptions of Signal Detection Theory: Measuring Detectability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI. Properties of Signals That Affect a Receiver’s Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

217 218 221 224 227 229

CONTENTS

VII. Classification of Signals in Addition to Detection . . . . . . . VIII. Complex Patterns: Extension of the Concept of Channels. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IX. Evolution of Signaling and Reception . . . . . . . . . . . . . . . . . X. Interpretation of Playback Experiments in Terms of Signal Detection Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . XI. Practicalities of Experiments in Natural Situations . . . . . . XII. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

vii 234 238 239 240 241 243 244

Preexisting Male Traits Are Important in the Evolution of Elaborated Male Sexual Display GERALD BORGIA I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Alternative Models of Display Trait Evolution . . . . . . . . . III. Problems with Current Models of Elaborate Display Trait Evolution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Evaluating Genetic Correlation Models . . . . . . . . . . . . . . . V. Evaluating the Preexisting Preference Model . . . . . . . . . . . VI. Evidence for the Co-option of Preexisting Traits . . . . . . . . VII. Implications and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . VIII. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

249 250 253 260 264 271 284 287 288

Adaptation, Genetic Drift, Pleiotropy, and History in the Evolution of Bee Foraging Behavior NIGEL E. RAINE, THOMAS C. INGS, ANNA DORNHAUS, NEHAL SALEH, AND LARS CHITTKA I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Comparison Between Species: Flower Constancy . . . . . . . III. Comparison Between Species: Floral Color Preference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Comparison Between Populations: Floral Color Preferences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Variation Within Populations: Color Preference and Foraging Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI. Variation Within Populations: Learning Behavior . . . . . . . VII. Reciprocal Population Transplant Experiments: A Test of Local Adaptation . . . . . . . . . . . . . . . . . . . . . . . .

305 307 311 313 317 320 323

viii

CONTENTS

VIII. Manipulation of the Foraging Environment: Scent Marking and Traplining . . . . . . . . . . . . . . . . . . . . . . . . IX. Manipulating Foraging Phenotypes: The Honeybee Dance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X. Genetic Basis of Foraging Behavior . . . . . . . . . . . . . . . . . . . XI. Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XII. Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XIII. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

326 330 334 335 339 342 344

Kin Selection, Constraints, and the Evolution of Cooperative Breeding in Long-Tailed Tits BEN J. HATCHWELL AND STUART P. SHARP I. II. III. IV. V. VI. VII. VIII.

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Study Species, Study Sites, and General Methods. . . . . . . . Kin Discrimination by Helpers . . . . . . . . . . . . . . . . . . . . . . . Kin Recognition Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . Fitness Consequences of Cooperation . . . . . . . . . . . . . . . . . Ecological Basis for Cooperative Breeding . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

355 358 360 363 370 381 386 389 390

How Do Little Blue Penguins ‘‘Validate’’ Information Contained in Their Agonistic Displays? JOSEPH R. WAAS I. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Natural History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III. Validations for Information Contained in Agonistic Displays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Investment Strategies Validating Signals and Signal Synergy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

397 399

438 441 443

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

449

Contents of Previous Volumes . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

463

402

Contributors

Numbers in parentheses indicate the pages on which the authors’ contributions begin.

GREGORY F. BALL (59), Department of Psychological and Brain Sciences, Johns Hopkins University, Baltimore, Maryland 21218, USA GERALD BORGIA (249), Department of Biology, University of Maryland, College Park, Maryland 20742, USA LARA S. CARROLL (173), Howard Hughes Medical Institute, University of Utah, Utah 84112, USA LARS CHITTKA (305), School of Biological and Chemical Sciences, Queen Mary University of London, London E1 4NS, United Kingdom THOMAS S. COLLETT (123), School of Life Sciences, University of Sussex, Falmer, Brighton BN1 9QG, United Kingdom ANNA DORNHAUS (305), Department of Ecology and Evolutionary Biology, University of Arizona, Tucson, Arizona 85721, USA DEBORAH L. DUFFY (59), Center for the Interaction of Animals and Society, School of Veterinary Medicine, University of Pennsylvania, Philadelphia, Pennsylvania 19104, USA TIMOTHY Q. GENTNER (59), Department of Psychology, University of California, San Diego, La Jolla, California 92093, USA PAUL GRAHAM (123), School of Life Sciences, University of Sussex, Falmer, Brighton BN1 9QG, United Kingdom ROBERT A. HARRIS (123), School of Life Sciences, University of Sussex, Falmer, Brighton BN1 9QG, United Kingdom BEN J. HATCHWELL (355), Department of Animal and Plant Sciences, University of Sheffield, Sheffield S10 2TN, United Kingdom NATALIE HEMPEL-DE-IBARRA (123), School of Life Sciences, University of Sussex, Falmer, Brighton BN1 9QG, United Kingdom THOMAS C. INGS (305), School of Biological and Chemical Sciences, Queen Mary University of London, London E1 4NS, United Kingdom ix

x

CONTRIBUTORS

RAYMOND NOWAK (1), Equipe Comportement, Unite de Physiologie de la Reproduction, et des Comportements, UMR 6175 CNRS-INRAUniversite de Tours-Haras Nationaux, 37380 Nouzilly, France WAYNE K. POTTS (173), Department of Biology, University of Utah, Utah 84112, USA NIGEL E. RAINE (305), School of Biological and Chemical Sciences, Queen Mary University of London, London E1 4NS, United Kingdom NEHAL SALEH (305), School of Biological and Chemical Sciences, Queen Mary University of London, London E1 4NS, United Kingdom STUART P. SHARP (355), Department of Animal and Plant Sciences, University of Sheffield, Sheffield S10 2TN, United Kingdom KEITH W. SOCKMAN (59), Department of Biology, University of North Carolina, Chapel Hill, North Carolina 27599, USA JOSEPH R. WAAS (397), Department of Biological Sciences, University of Waikato, Hamilton, New Zealand R. HAVEN WILEY (217), Department of Biology, University of North Carolina, Chapel Hill, North Carolina 27599, USA

Preface

The aim of Advances in the Study of Behavior 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 the field,’’ as its goal was phrased in the Preface to the first volume in 1965. Since that time, traditional areas of animal behavior have achieved new vigor by developing links with related fields and by forging closer relationships between those studying animal and human subjects. The links with other fields that are now so much a part of animal behavior are clearly apparent in the present volume: Lara Carroll and Wayne Potts argue for the importance of understanding the behavior and ecology of animals when studying the function of genes; Raymond Nowak describes the early neonatal life in mammals when they are completely dependent on milk for nutrition and the development of infant suckling; and Tom Collett, Paul Graham, Robert Harris, and Natalie Hempel-de-Ibarra examine the navigational abilities of ants and bees and how they retrieve memories when following routes. The editors and publishers of Advances in the Study of Behavior are committed to continuing to provide a means for publishing multidisciplinary and integrative studies, which contribute to our understanding of behavior. This volume also reflects current themes in animal behavior. Several chapters are directed toward understanding communication from various perspectives: Haven Wiley argues that the use of signal detection theory results in new insights about communication; Gerry Borgia evaluates current theories about the evolution of exaggerated male sexual display traits and offers a new explanation; Greg Ball, Keith W. Sockman, Deborah Duffy, and Timothy Gentner describe the mechanisms and function of song in European starlings; and Joe Waas identifies four types of signals through a review of his research on little blue penguins. Two studies focus on behavioral adaptations: Ben Hatchwell evaluates the ecological factors that promote cooperation through a detailed case study on the long-tailed tit; and Nigel Raine, Thomas C. Ings, Anna Dornhaus, Nehal Saleh, and Lars Chittka review their research on social bees to understand whether particular behavioral traits represent foraging adaptations. The chapters in this volume cover a diversity of animal taxa, including birds, mammals, and insects; they examine behavior from both proximate and ultimate perspectives; and include both lab and field studies xi

xii

PREFACE

and theoretical and empirical approaches. By inviting extended presentations of significant research programs, by encouraging theoretical syntheses and reformulations of persistent problems, and by highlighting particularly penetrating research programs that introduce important new concepts, Advances in the Study of Behavior hopes to continue its ‘‘contribution to the development of the field.’’ With this volume, we want to welcome Dr. Kathy Wynne-Edwards to our team of editors. Her integrative approach to the study of behavior will contribute greatly to the diversity, expertise, and quality of future volumes. Also, with this volume, Dr. Peter Slater is stepping down as Executive Editor. Peter first joined the editorial team in 1984 with Volume 14 and he took over as Executive Editor with Volume 19. Peter’s extraordinary knowledge of animal behavior, his expertise and breadth of interests, and his high standards and diplomatic approach to editing has maintained the high quality that characterizes these volumes. Peter has left a permanent mark on the field that is deeply appreciated. I will take over as Executive Editor and Peter will continue as an editor of the series, along with Chuck Snowdon, Tim Roper, Marc Naguib, and Kathy Wynne-Edwards. Together, this diverse group of editors will help to ensure the intellectual diversity that has characterized this series from its inception. H. JANE BROCKMANN PETER J. B. SLATER CHARLES T. SNOWDON TIMOTHY J. ROPER MARC NAGUIB KATHERINE E. WYNNE-EDWARDS

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 36

Suckling, Milk, and the Development of Preferences Toward Maternal Cues by Neonates: From Early Learning to Filial Attachment? Raymond Nowak equipe comportement, unite de physiologie de la reproduction et des comportements, umr 6175 cnrs‐inra‐universite de tours‐haras nationaux, 37380 nouzilly, france

I. INTRODUCTION Immediately after birth, mammalian mother and young interact intimately, often in a context of social isolation from other adult conspecifics. The most striking change in behavior of the female is undoubtedly the strong interest that she shows toward the neonate which is concomitantly associated with reduced interest toward other partners or even increased aggression toward them. This change of behavior is believed to favor early parent–offspring contact by reducing interference from other adults, ensuring that maternal care is provided to the mother’s litter, and in some species, to facilitate recognition of the young and attachment (reviewed in Alexander, 1988). Although the process may not always be fully adequate, especially in inexperienced mothers, behaviors of mammalian females have been selected under evolutionary pressure to complement the needs and capabilities of their neonate. Two types of behavior that fulfill fairly specific functions are commonly observed in postparturient females: these include licking or stroking, expressed as soon as the young is born, and then nursing. Several functions of postparturient licking or stroking have been postulated (reviewed in Trevathan, 1987). Licking the neonate serves to remove fetal fluids and membranes or even the entire amniotic sac with the placenta. Ingestion of the birth fluids helps dry the young and limits heat loss, and removal of the membranes from its face prevents suffocation. Licking may help the mother orient her young toward her mammary region as she adopts a specific nursing posture which makes the teats easily accessible. In altricial 0065-3454/06 $35.00 DOI: 10.1016/S0065-3454(06)36001-9

1

Copyright 2006, Elsevier Inc. All rights reserved.

2

RAYMOND NOWAK

mammals licking triggers urination and defecation, and it is also believed to stimulate breathing. Finally, licking also arouses the neonate and stimulates teat‐seeking activity as well as early olfactory learning. However, licking is far from widespread among mammals and therefore is not an absolute necessity for the survival and development of all mammalian neonates (reviewed in Nowak et al., 2000; Trevathan, 1987). It is completely lacking in aquatic species, both in those giving birth in the water (whales, dolphins, sea cows) and in those delivering on land (seals, phocids, walrus). Even some terrestrial species do not lick their young (elephants, camels, suids) while others display such activity around parturition but rarely thereafter (sheep, rabbits). Another well‐known exception is the primates. Instead of licking their young, mothers make extensive use of their hands in immediate interactions with their infant, stroking, holding, and cuddling it. Like licking, tactile stimulation of the skin of the human newborn is seen as a positive sign as it serves to rub off the vernix caseosa, a waxy substance present all over the skin at birth to prevent it from drying out. Furthermore, skin to skin contact and stroking also serves as a means of keeping the infant warm, has a calming effect, and may stimulate early learning and growth (Christensson et al., 1992; Scafidi et al., 1990; Sullivan et al., 1991; Weller and Feldman, 2003). Nursing, unlike licking, is the behavior that characterizes all mammalian females. Of course nursing patterns are extremely diverse across species both in duration (from 1 week in the hooded seal to several years in humans) and frequency (from once every 2 days in tree shrews to permanent attachment to the nipple in marsupials), but nursing is the feature that links them all. While the most obvious function of nursing is to feed the infant, nursing has other important biological functions. In some marsupials, mothers exploit their pups’ firm grip on the nipple to transport them when threatened by potential danger (Hunsaker, 1963, cited by Blass and Teicher, 1980). Nursing also protects the young from certain diseases since colostrum, the early milk produced around parturition, contains antibodies that compensate for the immature neonatal immunological defense system (reviewed in Korhonen et al., 2000a,b), drinking colostrum is absolutely vital for newborn ungulates (reviewed in Levieux, 1982; Patt, 1977). Also as a source of maternal contact, suckling soothes and comforts the infant (reviewed in Blass, 1994, 1996; Blass et al., 1995; Weller and Feldman, 2003). It is in the course of the very first nursing episodes that the most dramatic behavioral changes are observed in the neonate. Soon after parturition, mothers of all species commonly orient their bodies in such a way that the young can find the mammary zone even though mothers may not actively assist their infant in establishing nursing. For a newborn, locating the mammary region and grasping the nipple is of

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

3

vital importance because the mother is initially the only source of nutrients and fluids. Whatever their stage of development, neonates are born with functional sensory systems, memorized information from their prenatal experience, limited though highly efficient learning abilities, and locomotor skills that are sufficient to achieve approach or avoidance responses. This makes the neonate fully adapted to its new ecological niche. A typical behavioral sequence involves motor activities that bring the newborn in contact first with its mother and then more specifically with the mammary zone, rhythmic head movements and oral activity until the teat or the nipple is found, withdrawing milk, disengagement, and general quiescence or even deep sleep after milk consumption. The suckling–nursing relationship between the young and its mother is at the center of their behavioral and physiological interaction, although other characteristics of the mother (thermal, tactile, olfactory, visual, and auditory) play a role in the attraction that she holds for her young. First of all, it is through suckling that the infant exerts a direct influence on maternal physiology stimulating lactation and the release of milk, while estrous cycle may be suspended for several weeks or months. Second, suckling promotes episodic contact between the mother and the young, which maintains maternal care and at reunion arouses the infant eager to find the nipple and ingest milk. Sensory experience with milk influences several neurochemical systems including the endogenous opioidergic and cholecystokininergic systems (reviewed in Blass, 1996). Third, exposure to milk also modulates the attention state of the young which can directly or indirectly influence early sensory responsiveness. In a variety of mammals, experience with the sensory cues provided by the mother during suckling has been documented to influence both kin and species recognition, adult mate choice and reproductive behavior, and the development of dietary preferences. For these reasons, the mammary gland is recognized as an environmentally relevant sensory stimulus in the neonate’s ecological niche and should be viewed as more than just a source of nutrition in early development. The aim of this chapter is to show how repeated sensory information provided by the mother concomitantly with the occurrence of suckling gains signal value for the neonate. This sensory information then becomes selectively sought for by the neonate and elicits preferential behaviors. This point is illustrated by considering several contrasting species that have been studied in detail: the rat and the rabbit pup, as examples of altricial mammals, the lamb as a precocial mammal, and the human baby as an intermediate type. Information on other species will only be included to emphasize converging (or diverging) points or to provide information that is not available for the species cited above. This chapter is organized into five major sections. Section II provides general information on the

4

RAYMOND NOWAK

mammary region and describes how maternal cues guide the neonate to the source of milk. The composition of the lacteal fluids is presented in Section III. It emphasizes the variation occurring in the early stage of lactation as well as on the differences between species. Section IV focuses on the rewarding effect of suckling and its pre‐ and postingestive elements in the establishment of early learning. The suckling act is placed in the biological context of the species in which the events surrounding the act become motivationally charged and can sustain and direct future action. The last two sections concern the cascade of physiological events linking ingestive behavior and the development of early filial attachment, focusing mainly on the gut–brain axis.

II. THE UDDER,

THE

MILK,

AND THE

NEONATE

A. THE SOURCE OF MILK 1. Getting Ready to Nurse Nipples usually develop in both sexes (reviewed in Raynaud, 1969), but in a few species (e.g., rat, mouse) sex differences occur in the pattern of fetal mammary growth such that no nipples are formed in the male. Sub sequent enlargement of the mammary gland is controlled by ovarian hormones. During pregnancy, the mammary gland epithelium experiences its greatest and most rapid phase of cell proliferation. This occurs in response to hormones initially from the corpus luteum (estrogen and progesterone), followed by placental hormones (estrogen, progesterone, and somatotropin), pituitary hormones (prolactin), and adrenocorticoids from the adrenal gland (reviewed in Imagawa et al., 1994; Lascelles, 1976). In some species the increased growth of the mammary gland during pregnancy is stimulated in part by the female’s own behavior. In the rat, females lick their nipple lines as pregnancy evolves, while licking of other parts of the body decreases. Self‐licking triggers the development of the mammary gland (Roth and Rosenblatt, 1968). While preventing self‐licking by fitting the rat with collars impaired growth of the gland, mechanical stimulation with a hair brush induced mammary development in rats wearing collars (Herrenkohl and Campbell, 1976). For species with a short duration of gestation, such as lagomorphs and most rodents, the mammary glands develop during all (rat) or part (rabbit) of lactation. For species having a long gestation period (primates, ruminants, and suids), the mammary glands usually increase in volume during the second half of pregnancy and their development is nearly complete at parturition. In the woman, the

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

5

area around the nipples, the areola, becomes darker and the Montgomery tubercles become more pronounced (Lawrence, 1985). As the glandular tissue increases, the breasts become heavier and develop a looser curve. As a result, the nipples, instead of being centered, are in the lower part of the breast lobes. These morphological changes are believed to be an adaptive response to nursing since the nipples are hence slightly oriented downward making it easier for the newborn baby to take them in its mouth (Kintzinger, 1989). 2. Anatomy of the Mammary Glands The evolutionary beginning of mammary glands is believed to be in the development of a specialized area in the skin of the abdomen for incubating the young, probably arising in the endothermic therapsid (Blackburn et al., 1989; Long, 1969). Sweat glands may have become specialized and enlarged. There is considerable diversity in the structure of mammary glands of current mammals. Nonetheless, the anatomic association with the skin has been maintained and has the sharing of the cutaneous supply of nerves and blood vessels. Figure 1 illustrates schematically the mammary glands of the rat, the rabbit doe, the ewe, and the woman. In the monotremes (the platypus and the echidna), the mother has no nipples. Milk is exuded from 100 to 150 separate gland tubes that each open at the base of a stiff hair. The glands are paired laterally and there is no internal storage of milk. Milk is secreted onto the hairs from which it is lapped by the young (Lascelles, 1976; Raynaud, 1969). In all Metatherian and Eutherian mammals, the mammary gland has become modified by the appearance of a prehensile nipple varying in shape and size, and the development of a complex branching duct system. The glands are arranged in lobes, each being drained by its own duct system. The arrangement of the ducts varies according to whether they join together before the opening onto the surface of the nipple or remain as separate galactophores (reviewed in Cowie, 1982; Raynaud, 1969). Each mammary gland is functionally independent and, in the course of a single feeding episode, can feed either one or several young. For instance piglets, soon after farrowing, select one teat which tends to be suckled exclusively throughout lactation (De Passille´ et al., 1988). In contrast, in rabbit pups the whole litter keeps shifting from one nipple to another during a single 3‐min‐nursing episode (Drewett et al., 1982; Hudson and Distel, 1983). Primates, which by and large give birth to one young, usually feed their offspring on both breasts. The variation in the structure and position of the mammary gland is the result of evolutionary pressure related to the number of young, the maturity of the offspring, optimal attachment of the young to the nipple, and the need

6

RAYMOND NOWAK

Fig. 1. Anatomy of mammary glands in the rat, rabbit doe, ewe, and woman. (Drawings by R. Nowak.)

to cause the least impairment to the mother’s locomotion. Mammary glands are invariably located ventrally and lateral to the midline on all species (Raynaud, 1969). However, there is considerable diversity in the number of glands and location along the midline among mammalian species (from 2 in primates and most ungulates to 24 in the tenrec). Generally, species can be grouped for mammary gland location as having: (1) anterior glands (primates, elephants, seacows, bats); (2) posterior glands (ungulates, whales); and (3) glands extending from the anterior to the posterior (rodents, lagomorphs, suids). Where the nipple contains a storage system, it is usually referred to as a teat. The nipple consists of a specialized layer of hairless skin and a core of connective tissue in which are embedded one or more milk ducts as well as blood vessels and nerves. a. The rat Females have six mammary complexes, three of which are located along the thorax, one on the abdomen, and two in the inguinal region (Hebel and Stromberg, 1986). Before parturition, a considerable increase in the glandular tissue takes place. At the base of the teat, a wide ring‐shaped zone at the transition of the hairless to hairy skin contains large

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

7

sebaceous glands. The lactiferous duct enters the nipple through only one‐ teat canal (or papillary duct) which leads to the teat orifice (or papillary ostium). In the rat, the skin around the base of the nipple is adapted to form a small pouch that retracts between nursing episodes. b. The rabbit doe Rabbit does have four or five pairs of mammary complexes: two on the thorax, one or two on the abdomen, and one in the inguinal area (Barone, 1978). The global structure of the rabbit mammary gland is rather similar to that of the female rat except that the lactiferous duct enters the nipple through several teat canals and leads to 8–10 papillary ostiums. Odor cues on the skin elicit nipple‐search behavior in the newborn pup, and appear to increase in strength toward the nipples (Hudson and Distel, 1983). These cues, identified as a pheromone (Schaal et al., 2003), are produced by pregnant as well as lactating does (Hudson and Distel, 1984). c. The ewe The udder of the sheep consists of two separate glands located in the inguinal region of the ewe. Each gland has one teat and each teat has one opening. In the ewe, as in all ruminants, the lactiferous ducts converge to form larger ducts which eventually empty into a lactiferous sinus (Barone, 1978). The lactiferous sinus is further divided into a large cavity, the gland cistern, and a smaller cavity within the teat called the teat cistern. The latter is continuous with the exterior of the teat through a narrow opening, the streak canal. The udder is usually covered with fine hair, except for some breeds, such as the Merino, where it is partly covered with wool. The right and left halves are entirely separate; externally this is indicated by the intermammary groove seen at the underside of the udder and each teat drains one gland. No sweat glands or sebaceous glands are found on the teats, but the inguinal pockets are endowed with apocrine glands secreting a waxy and odorous substance (Barone, 1978). d. The woman The breasts are a mass of glandular, fatty, and fibrous tissues positioned over the pectoral muscles of the chest wall and attached to it by fibrous strands called Cooper’s ligaments (Lawrence, 1985; Robinson Baker, 1998). A layer of fatty tissue surrounds the breast glands and extends throughout the breast. The fatty tissue gives the breast a soft consistency. The glandular tissue of the breast houses the lobules and the ducts (Ramsay et al., 2005; Russo and Russo, 2004). Toward the nipple, each duct widens to form a sac (the ampulla). The substance of the nipple consists of smooth muscle and connective tissue permitting erection of the nipples and helping the infant to attach to it. The nipple‐areolar region is densely supplied with varied skin glands (reviewed in Schaal and Porter, 1991).

8

RAYMOND NOWAK

The nipple is surrounded by an area of pigmented skin containing raised tubercles called the Montgomery glands. These are known to contain the openings of galactophores and are associated with sebaceous glands. There is no obvious sphincter for retention of milk in the woman and therefore milk can leak from an unsuckled nipple adding intrinsic olfactory cues.

B. FULLY EQUIPPED

TO

FIND

THE

NIPPLE

Birth is the most dramatic change in the life of mammals. In the uterus, the fetus lives in a warm, aquatic, protected environment, and all its nutritional needs are regularly provided by the mother. As soon as it is born, the young faces hostile and unstable environments (climatic hazards, diseases) and has to adapt to the irregularity of the mother’s presence while it is still fully dependent on her for nutritional needs. Profound morphological and physiological changes occur during the transition from a fetal to the neonatal form: aerial breathing, change in blood circulation, and oral ingestion of food. During the immediate postnatal period, the neonate is highly aroused as a consequence of the general stimulation caused by the birth process (Lagercrantz, 1996; Lagercrantz and Slotkin, 1986). In the rat, this sustained arousal promotes exploratory movements of the mother’s body and brings the newborn into contact with sensory cues which facilitate location of the nipples (Ronca et al., 1996). The behavior of the young contributing to the procurement of milk from a teat or a nipple is referred to as suckling (Hall et al., 1988) and is displayed in a stereotyped manner in each species. Even in marsupials that give birth to a small larva‐ like offspring, the neonate crawls from the vaginal opening to the pouch through the mother’s fur guided by the track of moist hair due to the mother’s self‐licking (Sharman and Calaby, 1964) but also by negative geotropism (Cannon et al., 1976). Once inside the pouch, the larva firmly attaches to the nipple for weeks (Renfree et al., 1989). The latency to first reach the nipple or the teat varies between species, but generally suckling occurs shortly after the young are born. As the body energy reserves are very limited in the newborn, any delay in the ingestion of milk may be fatal as it puts the neonate in danger of hypothermia. In addition, some newborn must obtain immunoglobulins from their mother; failure to suckle within hours of birth results in decreased absorption of immunoglobulins and reduces protection against neonatal infection (Korhonen et al., 2000a,b; Patt, 1977). In polytocous species, giving birth to large litters, the first suckling episode often occurs shortly after the entire litter is born and in some cases even before (pig: Fraser, 1990; rabbit: Hudson et al., 1999).

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

9

Fig. 2. Sensory cues used by rat and rabbit pups, newborn lambs, and human babies in their initial search for the teat. (Drawing by R. Nowak.)

A variety of maternal cues aid the newborn in its initial search for the mammary zone (Fig. 2). In altricial species giving birth to immature young which are blind and deaf at birth, localization of the mammary zone is mainly, if not uniquely, dependent on olfactory cues. Among the species that have been studied, the cat appears as an exception, since for kittens the major determinants in nipple attachment have their roots in tactile inputs (Blass et al., 1988; Larson and Stein, 1984). Olfactory disruption never impairs nipple localization when the kittens are in contact with their mother’s fur; however, it does interfere with their ability to find the mother when separated. a. The rat pup The thermotactile characteristics of the mother’s ventrum are of minor importance (Blass et al., 1977). In contrast, elimination of the sense of smell in rat pups either via olfactory bulbectomy or destruction

10

RAYMOND NOWAK

of the olfactory epithelium with ZnSO4 (Hudson and Distel, 1986; Singh and Tobach, 1975; Singh et al., 1976) or washing the mother’s ventrum greatly reduced attachment to the nipples (Hofer et al., 1976; Teicher and Blass, 1976). Contact with the fur elicits a rooting response through which the neonates explore the mother’s body, moving their head laterally until a nipple is found. The pups then stop moving and probe the nipple area, licking it until it becomes erect. The olfactory cues which attract the pups in their initial search are nonetheless not produced by the nipple glandular system and rat milk is not an effective stimulus for nipple attachment (Blass and Teicher, 1980). Rather, it is the amniotic fluid and the saliva deposited by the mother around parturition as she licks herself that orient the pups to the nipple (Teicher and Blass, 1976). Following their initial successful suckling, subsequent localization of the nipple is mediated by the odor of the pup’s own saliva that was deposited in the previous feeding episodes. Pedersen and Blass (1981) discovered that an important salivary component for eliciting nipple attachment was dimethyl disulfide although it has only half the potency of intact saliva suggesting that other olfactory cues are involved. Because dimethyl disulfide was not detectable in the amniotic fluid, its behavioral salience is unlikely to depend on prenatal exposure. While nursing, the mother rat adopts a characteristic posture, consisting of a high arching of the back (kyphosis) which is triggered by the pups actually attaching to the nipples (Stern, 1989, 1996). b. The rabbit pup It takes the pups only a few seconds to attach to the nipple (Distel and Hudson, 1985; Hudson and Distel, 1983, 1986). After making contact with the doe’s fur, rabbit pups push their muzzle into it and display probing and rapid lateral head movements until a nipple is reached. They do not remain attached to a single nipple but change frequently repeating the whole search sequence several times during a suckling bout (Drewett et al., 1982; Hudson and Distel, 1983). Fur helps pups to encounter the nipple and stimulate the lateral head movements. By investigating the cues governing this behavior, it has been shown that an odor produced by the nipples, the nipple‐search pheromone, is essential for the onset and the maintenance of searching behavior and for nipple attachment. The pups are very sensitive to the volatile cues which are present not only on the doe’s ventrum but also in the milk (Coureaud et al., 2001). When testing the reaction of pups to fresh milk presented on a glass rod, it was found that even milk diluted 10,000‐fold elicited searching and grasping (Keil et al., 1990). Pups remained inactive when blood from either pregnant or lactating females or amniotic fluid was presented; therefore the reactivity of the pup to olfactory cues present in the milk is not programed by prenatal

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

11

exposure. A unique volatile substance extracted from rabbit milk has been shown to elicit the typical head searching and grasping response, and Schaal et al. (2003) have named this compound, 2‐methylbut‐2‐enal, the mammary pheromone. Production of the pheromone is under hormonal control and increases in late pregnancy (Hudson and Distel, 1984). Sequential administration of estradiol, progesterone, and prolactin, mimicking the physiological state of pregnancy in nonpregnant does, stimulates secretion of the pheromone within a few days (Gonzalez‐Mariscal et al., 1994; Hudson et al., 1990). It is notable that even pups delivered by caesarian section 1 day before term respond to a lactating doe (Hudson, 1985) and to the mammary pheromone (Schaal et al., 2003) with normal teat‐searching activity and grasping. The fact that the pheromone is such a powerful releaser of neonatal teat‐searching behavior does not exclude the involvement of other olfactory cues, in particular those from the uterine environment. Aromas perceived from the mother’s diet may provide additional information (Coureaud et al., 2002) and prenatal learning of odors present in the uterus could help in the neonatal search for the nipple by providing additional facilitating cues. c. The lamb In contrast to rat and rabbit pups, lambs are precocious mammals. Lambs are born with fully functional sensory modalities at birth and therefore are expected to use multisensory cues in their search for the teat (reviewed in Nowak et al., 2000; Vince, 1993). After parturition, the ewe emits mostly low‐pitched bleats, licks her lamb, and moves around her lamb as it attempts to stand up. The neonate approaches its mother guided by visual and auditory cues, and the first contact is made with the chest or flank of the ewe. The lamb spends time nosing the angles of the body until it finally finds the teat. Tactile stimulation on the face strongly activates oral exploration and orientation movements of the head, but the intensity of the response depends on the characteristics of the stimulus: lambs respond preferentially to smooth, nonwoolly, intermediately yielding surfaces (Vince et al., 1984). Thermal cues are also important and may direct the lamb toward the udder region (Vince, 1993). Measurement of the body surface temperature of the ewe showed that the highest readings were obtained from the udder and the inguinal region (areas free of wool: 35–37  C) whereas the lowest readings were obtained from the flank and the neck (area covered with wool: 25–28  C). Lambs can discriminate between differences in temperatures of 4  C, the surface eliciting the greatest response being 36  C, the temperature of the udder (Vince, 1984). Finally, lambs also use olfactory cues. Vince et al. (1987) have demonstrated that when lambs were made anosmic by spraying lidocaine into their nostrils, localization of the teat was delayed. The inguinal pockets of ewes contain glands secreting a waxy, strong‐smelling

12

RAYMOND NOWAK

substance which could activate the udder‐seeking behavior in combination with tactile stimuli. Records of respiration and heart rates showed that unsuckled lambs responded to the smell of inguinal wax (Vince and Ward, 1984) and they could also discriminate between the smell of their mother and that of an alien ewe. During the exploratory activity of the newborn lamb, the mother may even help it to find the udder by focusing her licking activity on the anal region and pushing the neonate toward her inguinal area. Experienced ewes tend to arch the back and spread the hind legs, or lift one leg as their lamb approaches the udder to facilitate access to the teat (Vince, 1993). d. The human baby The mother usually assists her newborn in finding the nipple by holding it in her arms and putting her baby to the breast. However, when left quietly on the mother’s abdomen after birth, human neonates are able to crawl gradually up to her breast, find the nipple, and start to suckle without any maternal assistance. If the mother has not received any pain‐killing medication during labor and delivery and the infant is dried thoroughly, placed on her abdomen, kept warm with the heat of the maternal body and a towel, the baby usually begins a five‐part sequence that ends with proper attachment to the mother’s nipple (Widstrom et al., 1987). For the first 20 min, the newborn rests and looks up periodically at the mother. At 30–45 min, mouthing and lip‐smacking movements are displayed, and the infant begins hand–mouth movements. The baby then begins to move forward slowly, starts to turn its head from side to side, and opens its mouth widely on nearing the nipple. After several attempts, the lips latch onto the areola. Odors produced by the breast of lactating women attract the neonate. When babies were placed skin‐to‐skin on their mother’s chest, free to move, they displayed a preference for the natural smelling breast in comparison to the alternative breast that has been thoroughly washed (Varendi et al., 1994). Attraction to the smell of the breast is further supported by the fact that when newborns were placed on their stomach, they displayed increased crawling toward a pad situated 17 cm in front of their nose when it contained maternal breast odors (Varendi and Porter, 2001). The nature of the olfactory cues that attract the newborn infant is unknown, but infants are strongly attracted to amniotic fluid‐coated breasts indicating that, like nonprimate mammals, the human newborn responds to cues to which they were exposed prenatally (Varendi et al., 1996). There is some evidence suggesting that the olfactory cues emanating from the breast may have some degree of similarity with the odor of amniotic fluid (Mennella et al., 2001). Thus, the neonate may be initially attracted to breast odors because of their overlap with familiar scent of amniotic fluid. According to Varendi

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

13

et al. (1996), attraction to amniotic fluid might also be a product of the human evolutionary processes. She hypothesizes that in the past women commonly handled their wet baby during and after delivery and soiled their breasts with birth fluids. This transfer of prenatal cues would have facilitated nipple location. Although odors are not as crucial as in nonprimate species, attraction to the smell of the mother’s breast is a characteristic that is present across species and its importance persists over the days following birth (Varendi et al., 1997). e. Conclusions The rapidity in the location of the source of milk shows that neonates are prepared to face new challenges encountered during early postnatal existence. Rat and rabbit pups find the nipple within minutes after birth (Hudson et al., 1999), most newborn lambs suckle between 1 and 2 hr of age (Alexander et al., 1990; Slee and Springbett, 1986), and even the human baby can find the mother’s breast at such an early age (Widstrom et al., 1988). To succeed in their new behavioral activity, the young are guided by two sets of information. One consists of various stimuli that have a high degree of similarity with those found in utero, if not complete similarity (e.g., amniotic fluid). This transnatal sensory continuity evokes orientation responses that bring the neonate into contact with the mammary area and includes chemosensory signals (amniotic fluids, odorants present both in the amniotic fluid and mother’s secretions: reviewed in Porter et al., 2005; Schaal, 2005), and also probably soft tactile stimulation (uterine wall–mammary gland) and thermal cues (warm maternal womb–warm mammary zone). The second set of information is unlearned (i.e., not dependent on prenatal experience) and induces instantaneous positive responses (e.g., mammary pheromone in rabbit pups). C. PATTERNS

OF

NURSING BEHAVIOR

It is the distinguishing characteristic of mammals that mothers feed their young by means of the secretion of milk until they are weaned. There is, however, an amazing variation in the frequency and the duration of nursing between different species, even when they are closely related. In primates, weaning age ranges from 1 to more than 100 months (Harvey and Clutton‐ Brock, 1985) and among pinnipeds it varies from over 1 year in some fur seals to an incredible 4 days in the hooded seal (Bowen et al., 1985; Oftedal et al., 1987). The tree shrew suckles its mother for about 10 min on alternate days (Martin, 1966) while marsupials have continuous contact with their mother’s nipple until they become mature enough to leave the pouch. In a number of altricial and semiprecocial species, the nursing–suckling relationship may be divided into three developmental phases: (1) a

14

RAYMOND NOWAK

‘‘neonatal phase’’ during which the mother plays the most active role in the relationship and meets the young’s requirements without any restriction, (2) a second phase during which nursing–suckling interactions are initiated by both partners but where the young plays a more active role in the procurement of milk while the mother limits feeding duration, (3) and a third phase where suckling is exclusively initiated by the young while the mother evades the feeding approaches as weaning proceeds. Only the first phase of total maternal dependence will be described in this section. a. The rat The female plays a major role in initiating feeding by approaching the young in the nest. Over this nursing phase, which progressively comes to an end between 12 and 14 days postpartum, the pups show improvement in their ability to react to the approach of the dam and to attach to a nipple (Rosenblatt and Lehrman, 1963). Nursing may occur one to eight times per 2‐hr period, and it is not unusual for mothers to spend at least 12 hr of each day with the pups attached to her nipples (Lincoln et al., 1973; Shair et al., 1984); not all the nursing bouts are necessarily followed by milk ejection. The mother stimulates the young by licking them or carrying them and on arousal, the whole litter begins to nuzzle the dam’s fur and attach to the nipples. She facilitates suckling by hovering over the pups with her mammary region easily accessible. As they crawl to her nipples she remains in the nursing position throughout the entire nursing episode, the milk ejection reflex only occurring when the mother is in a somnolent state (Lincoln et al., 1980; Voloschin and Tramezanni, 1979). In this species milk letdown occurs periodically inducing a rhythmic sleep/ awake state pattern in the litter. Milk elicits an immediate arousal state in pups, consisting of a stretching phase where all pups pull strongly against the nipple with their legs outstretched and their backs arched, and a phase in which the young detach themselves from the nipple, treadle, and move away before coming back to another nipple (Hall, 1979; Lincoln et al., 1973). This is followed by profound quiescence. During a nursing bout pups can spend two‐thirds of their time asleep. b. The rabbit The nursing behavior of the rabbit contrasts markedly with the typical model of frequent nursing described in other altricial species. First of all, the mother is fully alert (Neve et al., 1982). Second, the rabbit doe leaves her nest almost immediately after giving birth and returns to feed the young once a day for only 3–5 min with an extraordinary circadian periodicity (Hudson and Distel, 1982, 1989). This pattern is observed until weaning. During the short nursing episode, the young must locate the nipples and suckle enough milk to sustain them for the next 24 hr. In view of the brevity of these episodes, rabbit pups have developed

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

15

an incredible efficiency at drawing milk out of the nipple. They can increase their body weight from 60 g on day 1 to almost 100 g on day 6 (Caba et al., 2003). The amount of milk they ingest represents an increasingly higher percentage of the pups’ body weight so that by day 7 they consume more than 30% of their body weight. Behavioral observations have shown that mother and pups display perfectly synchronized activities: shortly before the doe enters the nest, the pups huddle tightly together, uncover themselves from nest materials, and become agitated (Hudson and Distel, 1989). Thus the rabbit pups are fully prepared by the time the mother arrives, which in most cases occurs during the night. This pattern of the young is clearly associated with the occurrence of feeding and can be divided into four periods: suckling itself, a postsuckling period of approximately 15 min, then 22 hr of ‘‘rest,’’ and finally a presuckling period of 1–2 hr. Following suckling, the pups urinate on each other, dig vigorously, push themselves under the nest material, and disperse within the nest. This digging and burrowing activity helps them dry out. The pups then reaggregate and remain covered until the next suckling episode. The anticipatory uncovering of the pups may represent a circadian rhythm itself, even in neonates, as this activity is displayed regardless whether the pups were fed during the previous nursing bout or not. The behavioral pattern of the rabbit pups remains unchanged until they are 2 weeks old after which they start to leave the nest (Hudson and Distel, 1982). c. The sheep In contrast to the consistency of mother–young interactions in rabbits, the nursing pattern changes rapidly over time in sheep. Nursing frequency is very high on the first day, then declines to about once an hour by the end of the first week, and decreases even more thereafter. The large volume of colostrum produced by ewes around parturition (between 1.2 and 3.6 kg in the first 48 hr) (Shubber and Doxey, 1979) combined with the high suckling frequency of the neonate means that lambs have the opportunity of ingesting a substantial volume of fluids. It is estimated that within 48 hr following birth, single‐ and twinborn lambs consume 37% and 33% of their body weight, respectively (Shubber and Doxey, 1979). However, because of the high water content of ovine colostrum, the increase in weight only averages 10%. During the first week of life, lambs are allowed to suckle at any time and for as long as they wish (Bareham, 1976; Gordon and Siegmann, 1991). Newly born lambs spend most of their time resting or sleeping once they have suckled, but after 2 or 3 days, they are more alert and occasionally leave their mother to play with age mates. After the first week or two, ewes may prevent their lambs from suckling by walking away or lying down when they attempt to push their head into the mothers’ inguinal region (Ewbank, 1967). Lambs usually suckle in a parallel‐inverse position after passing in front of their mother.

16

RAYMOND NOWAK

This pattern allows identification of the young by the ewe: she smells the infant’s anal region and if it is not hers, she rejects it (Poindron, 1976). Recognition of twins by the mother leads to synchronized nursing. Thus, if from birth, twinborn lambs suckle at will, after the first week they are allowed to feed only if they are together (Ewbank, 1964). d. The human Detailed information on the early suckling pattern in humans is scarce; however, anthropologists highlight important differences in the nursing behavior of women across cultures. For the first 3 days, the breast secretes colostrum (Kulski and Hartmann, 1981) which is regarded as nonfood in many societies or even dangerous, thus in these societies the baby is not nursed until the third day. It is now known, however, that colostrum is beneficial for the baby (Fetherston et al., 2001; Thoman et al., 1972), and mothers are encouraged to breast‐feed their infant in the first hours following birth. The volume produced is rather low during the first few days after parturition, usually less than 100 g on day 1, approximately 200–400 g on day 3, and up to 700 g on day 5. However, there is marked interindividual variation (Casey et al., 1986; Saint et al., 1984). It is estimated that the average colostrum intake at each breast‐feed during the first day of life is only 6 g. This low intake does not cover the high demand for energy at this time, and suckling may serve other functions in addition to providing immunological protection. Early contact between the mother and her newborn is encouraged in many countries and seems to influence positively nursing behavior in the following days and increases the duration of breast‐feeding (Carlsson et al., 1977; Thoman et al., 1972). During nursing, mothers hold their baby in their arms while rubbing and petting the child, stroking the infant’s bare skin and clothes in a calm manner. Women often report feeling calm and sleepy suggesting that the neuroendocrinological changes during feeding have a sedative effect (Uvna¨s‐Moberg, 1996). Babies usually suckle in bursts separated by intervals of pauses (Kintzinger, 1989). During a feed, bursts of 3–14 sucking acts may be observed, interspersed by pauses lasting about half the time of the sucking burst. It is usually during these pauses that the mother talks to her baby, mainly with a high‐pitched voice. As the feed draws to the end, the baby goes into deep sleep. In western societies, it is often ‘‘recommended’’ that women nurse their babies approximately six times per day, each nursing episode lasting 10–25 min, and night time feedings commonly cease within a matter of months. Such nursing patterns, however, are not necessarily natural. Gambian mothers are observed to nurse their infants up to 15 times daily (Prentice et al., 1986). In the !Kung tribe of Botswana and Namibia, women nurse their babies about four times an hour in bouts of 2 min (Konner and

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

17

Worthman, 1980). Even in western societies, many mothers nowadays choose to feed their baby on demand. While it is the unique form of providing milk to an infant, nursing is obviously associated with other functions such as comforting and intimacy with the mother.

III. MILK: AN ASTONISHINGLY COMPLEX FLUID A. FROM COLOSTRUM

TO

MILK

Colostrum is the first milk available to newborn mammals at birth. It is produced for a few days before and after parturition and contains dense nutrients as well as high levels of immunoglobulin, enzymes, hormones, growth factors, and neuroendocrine peptides. No doubt the best known function of colostrum is to transfer antibodies from the mother to the young before its own immunological protection becomes fully functional (reviewed in Korhonen et al., 2000a; Levieux, 1982; Thapa, 2005; Tizard, 2001). Transport of maternal immunoglobulins into colostrum probably occurs in all mammals to varying extents, but the significance of these immunoglobulins depends on the species. For species having an epitheliochorial placenta (ruminants, suids), the process of transfer of immunoglobulins from the mother to the neonate is of paramount importance to survival. After ingestion of colostrum, the immunoglobulins are absorbed intact into the neonate’s blood stream. This process of immunoglobulin absorption in the intestine stops shortly after birth. The timing of this halt, referred to as closure, depends on the species (Patt, 1977). Humans and other primates transport immunoglobulins to the fetus through the placenta via a receptor‐mediated, intraepithelial mechanism similar to that in the mammary gland. Therefore, when the infant is born the maternal immunoglobulins protect it from infections until its own immune system is fully functional (Goldman, 1993). The composition of the mammary secretions changes rapidly over the first few days after parturition with a continuous transition from colostrum to mature milk. All components of the mammary secretion fluctuate during this transition period although in a dissimilar manner according to the species (rat: Nicholas and Hartmann, 1991; rabbit: Peaker and Taylor, 1975; sheep: Hadjipanayiotou, 1995; Williams et al., 1976; human: Arnold et al., 1987; Kulski and Hartmann, 1981). The major compositional changes in bovine and ovine milk during the first week of lactation are given in Fig. 3 to provide an illustration. Protein concentration is highest in colostrum then it declines rapidly over the next day or two. The major proportion of this change in protein concentration is accounted for by

18

RAYMOND NOWAK

Fig. 3. Evolution from colostrum to milk in the days following parturition in the cow and the ewe.

the immunoglobulins. Milk fat percentage generally increases from colostrum to milk but declines in bovine milk after parturition. Lactose concentration in colostrum is generally low at delivery in both species, then increases sharply over the next few hours postpartum, and continues to increase slightly later on. In women, protein and lactose decrease during the first 5 days postpartum while fat increases slightly (Kulski and Hartmann, 1981; Saint et al., 1984). In contrast to ruminants and humans, lactational changes are not as pronounced in the postpartum period in polytocous species but tend to appear once lactation is established so as to sustain the growth of their large litter. In rats, while fat content declines threefold

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

19

within 5 days after parturition, the concentrations of protein and lactose increase gradually throughout lactation (Keen et al., 1981; Nicholas and Hartmann, 1991; Peaker and Taylor, 1975). In rabbits, biochemical changes are not significant during the first 4–5 days postpartum but once lactation is fully established protein and fat concentrations increase sharply (Peaker and Taylor, 1975). Lactose concentrations increase in rabbit milk during the second week of lactation and then decrease. B. BIOCHEMICAL COMPOSITION Milk contains all the nutrients required for the growth and the development of the neonate. Milks of all species are invariably composed of water, carbohydrate (mainly lactose), fat, protein, minerals, and vitamins (reviewed in Jenness, 1985; Jensen, 1995). While each component can be described separately, it is important to remember that milk is secreted as a complex mixture which varies over time and across species (reviewed in Oftedal and Iverson, 1995). Table I gives the milk composition for several contrasting species. 1. Water On birth, the mammalian neonate is not able to seek out its own water supply and would dehydrate rapidly without the water component of milk. Moreover, without some water, milk would be a viscous secretion composed mostly of lipid and protein and would be extremely difficult to extract from the gland. In sheep, colostrum viscosity is correlated with the suckling activity of the young: the thicker the colostrum at birth, the longer the lambs suckle during the first postnatal hours (Holst et al., 1996). The water content of milk can range from a low level in marine mammals to a high content in human milk and that of ungulates (Table I). 2. Milk Protein The combined protein components of milk fall into two categories, those which are milk specific and exclusively synthesized by mammary cells and those which come from maternal plasma. The main milk‐specific proteins are caseins which represent 80%–90% of total protein content (reviewed in Lo¨nnerdal and Atkinson, 1995; Swaisgood, 1995). Major a, b, and k caseins have an amino acid composition which is important for growth and development of the nursing young. Caseins are composed of several similar proteins which form a multimolecular, granular structure called casein micelle. In the stomach of the young of many species, the enzyme rennin specifically hydrolyzes part of the micelle, resulting in the formation of a curd. There are many whey proteins in milk and the set of whey proteins

20

RAYMOND NOWAK

VARIATION

IN

TABLE I MILK COMPOSITION ACROSS SPECIESa

Species

Water (%)

Fat (%)

Crude protein (%)

Lactose and sugars (%)

Ash (%)

Camel Red kangaroo Donkey Pig Horse Goat Human Cow Rhesus macaque African elephant Sable antelope Sheep European hare Red deer Brown rat Tammar wallaby Dog Lion Brush‐tail possum Reindeer Cat European rabbit Polar bear Bottle‐nose dolphin Humpback whale Blue whale Hooded seal Gray seal

90.5 86.7 88.5 80.6 89.5 87.8 88.0 87.8 85.2 85.0 83.1 81.3 79.2 78.5 78.1 77.5 77.1 76.3 76.1 74.9 73.9 70.9 57.1 55.9 52.9 44.5 33.0 31.0

4.3 6.1 0.6 8.2 1.3 3.8 3.8 3.7 4.6 5.0 5.0 7.4 15.6 8.5 8.8 4.0 9.5 8.7 4.4 10.9 10.8 15.2 31.0 29.4 33.0 40.9 61.1 59.8

4.3 7.2 1.4 5.8 1.9 2.9 1.0 3.2 2.3 4.0 6.2 5.5 10.0 7.1 8.1 6.0 7.5 11.8 7.0 9.5 10.6 10.3 10.2 12.2 12.5 11.9 4.9 9.2

– – 6.1 4.8 6.9 4.7 7.0 4.6 7.9 5.3 5.3 4.8 1.5 4.5 3.8 12.5 3.8 3.2 11.0 3.4 3.7 1.8 0.5 2.5 – 1.3 1.0 –

0.9 – 0.4 0.6 0.4 0.8 0.2 0.7 – 0.7 0.4 1.0 – 1.4 1.2 – 1.1 – 1.5 1.3 1.0 1.8 1.2 – 1.6 1.4 – –

a The percentage of water is an approximation as it was deducted from the values obtained for the other components (data collected from various sources cited in the text).

found in mammary secretions varies with the species and the stage of lactation. The major whey proteins in cow milk are b‐lactoglobulin and a‐lactalbumin. a‐Lactalbumin is an important protein in the synthesis of lactose and its presence is central to the process of milk synthesis. The function of b‐lactoglobulin is not known. Other whey proteins include immunoglobulins (antibodies; especially high in colostrum) and serum albumin, as well as a long list of enzymes, hormones, growth factors, nutrient transporters, and defense agents. Several peptides present in milk

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

21

have been demonstrated to influence various aspects of the neonate’s metabolism and development while others are suspected to play a role in its behavior (reviewed in Grosvenor et al., 1992; Meisel and Fitzgerald, 2000; Peaker and Neville, 1991). 3. Milk Fat Most mammalian neonates are born with little body adipose that might be used for insulation or as a source of stored energy. Milk fat is used by neonates for accumulating body adipose in the initial days after birth after which they metabolize it as a source of energy (Mellor, 1993). Milk fat is composed of a heterogeneous group of substances, of which triglycerides are the major type (reviewed in Jensen and Newburg, 1995; Jensen et al., 1995). It ranges from a little over 1–2% in the donkey and the horse to more than 50% in the blue whale, the gray seal, and the hooded seal (Table I). Mammals living in cold or wet ecosystems typically have high milk fat percentages (reviewed in Oftedal and Iverson, 1995). Milk also provides essential lipids which serve specific biological functions in the neonate. Long‐chain polyunsaturated fatty acids are indispensable structural components of cellular membranes that are deposited to a considerable extent in the growing brain and retina during perinatal development (Koletzko et al., 2001). 4. Carbohydrates Lactose is the major carbohydrate of most terrestrial eutherian milks. However, in marsupials, gray and flying squirrels, brown and black bears, and the Western hedgehog, oligosaccharides are predominant (reviewed in Oftedal and Iverson, 1995). Lactose is not as sweet as other sugars present in milk, even to newborn babies. While intraoral infusions of sucrose (and other sweet carbohydrates) induce rapid and sustained calming in crying newborns through a sweetness effect (Barr et al., 1999), lactose does not (Blass and Smith, 1992). Lactose is cleaved to glucose and galactose in the intestine of the neonate by the activity of an enzyme called lactase. Lactose is a major, readily digestible source of glucose which provides energy for the human neonate, the lamb, and the rat pup (Mellor, 1993). Carbohydrates other than lactose are found in milk but at lower concentrations. More than 50 oligosaccharides have been identified in human milk (Jenness, 1979). Carbohydrates found free in milk include amino sugars, sugar phosphates, neutral and acidic oligosaccharides, and nucleotide sugars. Some of these may act as growth factors for lactobacillus which populates the gastrointestinal tract of the infant, or as protective factors against certain potential pathogens (reviewed in Goldman and

22

RAYMOND NOWAK

Goldblum, 1995; Lawrence, 1985; Newburg and Neubauer, 1995). However, their real function is not fully understood. 5. Minerals The major minerals in milk consist of the monovalent ions sodium, potassium, and chloride and the divalent ions calcium, magnesium, citrate, phosphate, and sulfate (reviewed in Atkinson et al., 1995). The concentrations of phosphate and calcium are proportional to the concentration of casein. The high concentration of this protein ensures that the milk ingested by the rapidly growing neonate covers the mineral requirements for bone growth and development of soft tissues. Milk also contains most microminerals found in the body. 6. Other Components in Milk All the major vitamins are found in milk. Milk also contains leukocyte cells, the most prominent being neutrophils and macrophages in humans. The concentration of leukocytes in milk varies with the species (human milk has relatively high somatic cell counts while bovine milk has low cell counts), infection status of the mammary gland, and stage of lactation. Milk has numerous other components, many of which are grouped under the major biochemical components listed earlier (reviewed in Goldman and Goldblum, 1995). These may include bioactive factors, such as hormones and growth factors, many of which exist in concentrations that exceed those found in maternal plasma. C. SPECIES DIFFERENCES The biochemical analysis of milk does not reveal any clear pattern according to the taxonomic relatedness between species (reviewed in Oftedal and Iverson, 1995). On the other hand there is a strong relationship between milk composition, nursing behavior, and ecology of the species. Ben Shaul (1962) has described five patterns. Group I mothers are in permanent contact with their young, for example, marsupials or species giving birth during hibernation (bears). Nursing is either continuous or extremely frequent and the nutrient concentration of the milk is not high. These species have milk that is low in fat and protein. Group II includes females that also remain constantly with their young; however, whereas nursing is frequent, it is not continuous. Lactation is usually fairly long. Most primates including humans fall in this group; their milk has low concentrations of fat and protein, and the young can survive on diluted milk because of frequent nursing. On the other hand, the carbohydrate content exceeds that of almost every other mammal. Milk with relatively

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

23

large amounts of carbohydrates is found in species whose young experience fairly rapid postnatal brain growth. Group III comprises mammals that leave their young in a nest, burrow, or den. They reunite with their young for nursing at fairly widely spaced intervals. Females of this group produce milk that is rich in fat and protein allowing the young to withstand the long periods of food deprivation. Predator species, such as the lion, of which the female spends several hours hunting, and ungulates of which the young hide during the postpartum period while their mother goes foraging, nurse their young at 6‐ to 8‐hr intervals. Rabbit does nurse their young once a day and a nursing episode never exceeds 5 min. Group IV, which includes most rodents and carnivores, is very similar to the previous group, but the intervals between nursing bouts are not as long. The milk is not as high in fat. Finally, mammals in group V live in cold or wet ecosystems and require a large amount of fat to maintain body temperature. This is the case for pinnipeds and whales for which fat content is as high as 50%.

IV. SUCKLING

AND

EARLY LEARNING

A. NEONATAL REWARDS Although the neonate is not completely naı¨ve, it still has a lot to learn about its postnatal environment. Above all, the neonate will need to ensure a regular supply of milk. It is therefore not surprising that the very first forms of appetitive learning pivot around the nipple or the teat and their secretions. Because suckling is the most intimate form of contact with the maternal body, it is at the root of the relationship with the mother. Sensory cues emanating from the maternal body, concomitantly reinforced by suckling, rapidly acquire an attractive value for the neonate. For instance, newborn lambs need only a few successful suckling attempts to go straight to the udder without showing the awkward exploratory behavior characteristic of inexperienced young. Newborn kids have been shown to rapidly adjust their suckling behavior when born to mothers whose udder had been transplanted to the neck (Stephens and Linzell, 1974). In parallel, neonatal lambs, no longer reacted to facial touch provided by an experimenter once they had suckled their mother (Vince and Stanier, 1991). This is not linked to feeding alone since the response persists in lambs that were artificially fed. The change in the lambs’ behavioral response reflects the way they quickly learn to respond to natural maternal cues after only a few suckling episodes. The amount, quality, and intensity of contact vary across species. During the suckling bouts, the mother provides the infant with novel olfactory

24

RAYMOND NOWAK

(body odors, milk) and somatosensory cues (the texture of the mammary gland) and transfers heat to her infant through conduction. She may also stimulate it through licking (e.g., rodents), sooth it via stroking and cuddling (e.g., humans), and communicate with it vocally and visually (e.g., humans, ruminants). The hedonist tendency of the mammalian neonate is clearly visible in its propensity to approach stimuli that provide comfort and satiation (those that enhance energy gain and reduce energy loss) and to avoid novel stimuli that are potentially threatening to its biological integrity. This is reflected by the striking rapidity of the infant at establishing preferences for stimuli that satisfy or predict its vital needs. The most powerful rewards (or reinforcements) causing long‐term behavioral changes are met during suckling and are found in three types of maternal stimuli (Blass, 1990). The first type is mainly observed in rats and concerns the behavioral activation induced by the mother’s tactile stimulation as she enters the nest, treads on the pups, moves them around, and licks their anogenital area (Wilson and Sullivan, 1994). On arousal suckling is imminent. The second type is the tactile stimulation provided by the nipple and the motor pattern of nonnutritive sucking. Events surrounding milk ingestion constitute a third type of rewards. These three types of rewards act through nutritional as well as nonnutritional pathways, the taste of milk, the release of endogenous peptides by milk, or even through components specifically found in milk (Blass, 1994, 1996). Through them, the neonate will progressively learn about its postnatal environment, be specifically attracted to maternal cues, and form multimodal representations of the mother. In some species the infant will organize its behavior in a privileged manner by establishing specific contact with its dam. Over the past 30 years there have been numerous convincing demonstrations of learning through suckling in immature neonates. The attraction expressed by the infant for its mother, or for sensory cues provided by her, runs from simple olfactory or auditory conditioning as in rats, rabbits, and humans, to the development of early filial attachment as in sheep. Figure 4 reviews some data published on these species. Attachment refers to a reciprocal emotional bond between two individuals and is inferred from physiological and behavioral measures which result in proximity to a specific figure. Attachment relies on mechanisms by which individuals learn to identify their partners and applies to a broad range of individuals, including attachment of parents to their offspring, offspring to their mother or father, between siblings, and to related and unrelated adults (reviewed in Ainsworth, 1979; Gubernick, 1981; Mason and Mendoza, 1998). Attachment to a specific figure takes time to develop, varies with age and experience, and may eventually decline. In monoparental species, such as the sheep, the mother will be the primary attachment figure.

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

25

Fig. 4. Conditioned responses in rat pups, rabbit pups, and human babies induced while suckling. See text in Section IV.B for details on experimental procedures.

26

RAYMOND NOWAK

The clearest indices of its existence are spatial: the lamb approaches, follows, chooses to be near her rather than another ewe. It is in this strict sense that the word attachment will be used. B. DEVELOPMENT

OF

CONDITIONED RESPONSES

Much of the evidence concerning the role of milk in regulating infant behavior comes from studies of altricial mammals. From a historical point of view, research was aimed at determining the existence of conditioning in mammalian neonates and determining the extent to which their learning abilities varied from that of adults. It is therefore not surprising that, as a result of Pavlov’s findings, the very first experimental work was carried out on dog puppies. For the young of species born without functional visual and auditory systems, the sense of smell, together with somatosensory input, is of particular importance (Rosenblatt, 1976, 1983). As early as 1959, Volokhov (cited by Stanley et al., 1963) presented evidence of food‐ reinforced conditioning in 1‐day‐old pups. His findings were later confirmed by Stanley who showed that 2‐week‐old puppies could learn to discriminate between two types of surface (cloth vs. wire) with milk as a reward (Bacon and Stanley, 1970; Stanley et al., 1970) and refuted the then current idea that neonatal learning was limited. But it was not until the early 1980s that early learning was placed in a more biologically relevant situation: the relationship with the mother. a. The rat pup Relying on Freud’s belief (1940) that infants form a bond with their mother through the milk delivered at the breast, Brake (1981) demonstrated that when presented with a novel, aversive, olfactory stimulus, such as orange, while suckling an anesthetized female, 11‐ to 14‐day‐old rat pups acquired a suppression of aversion for that odor. The magnitude of the olfactory response was greater if the rat pups receive milk while suckling than if they did not receive any, suggesting that nutritive sucking is more rewarding than nonnutritive sucking (Fig. 4A). Nonlactating nipples have been shown to provide strong incentive for initiating and maintaining contact with the mother in an operant conditioning paradigm: 10‐day‐old rats quickly learned to approach when the reward was an opportunity to attach to a nipple of an anesthetized dam for a brief interval (Amsel et al., 1976; Kenny and Blass, 1977). However, these particular studies relied on the infants’ performance in a runway or a maze, tasks that are not suitable for younger pups. The discovery that newborn rats could be fed with pulses of milk via an intraoral cannula has provided opportunities to study very early learning in the context of suckling. Using small infusions of milk as reinforcers,

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

27

Johanson and Hall (1979) were able to show operant conditioning in pups as young as 1 day of age. Newborn pups could learn to probe into a paddle above their heads to receive small infusions of milk. In this operant setting, pups could discriminate between two paddles on the basis of odor and learned to respond selectively to the paddle that delivered milk. Oral milk infusions per se act as a reinforcer in classical conditioning paradigms as well. Several experiments demonstrated that if a novel odor was paired several times with an infusion of milk, neonatal pups came to prefer that odor and oriented themselves toward it in a two‐choice chamber (Johanson and Hall, 1982; Johanson and Teicher, 1980; Johanson et al., 1984; Sullivan and Hall, 1988). The appetitive learning capabilities were not triggered by intraoral infusion of water; however, pairing a new odor with sucrose or corn oil changed the odor’s hedonic value (Shide and Blass, 1991). Altered preferences were not explained by taste novelty, texture (mineral oil is ineffective), infusion rate, or the mere experience of fluid passing through the oropharynx. Although the findings by Shide and Blass (1991) do not imply that milk fat and sugars are the provoking factors when the fluid delivered through the cannula is milk, they demonstrate that olfactory preferences can be triggered through gustatory stimulation. b. The rabbit pup Newborn rabbit pups can rapidly learn to associate new odors with nipple‐search behavior, although they do not appear to depend on this ability under normal nursing conditions. The learned response, however, appears to be very similar to inborn responses. Hudson (1985) scented the doe’s ventrum with one of four odorous substances (cologne Chypre, Chanel No. 5, oil of camphor, citral) 15 min before the first nursing episode. When tested the next day on a mature female cat with well‐developed nipples (chosen because it is of a similar size to the rabbit while being completely free of rabbit‐specific odors), 2‐day‐old pups showed strong conditional responses specific to the odor experienced during their first suckling episode. Ninety percent of the pups responded with nipple‐ search behavior, and 50% attached to a nipple. In contrast to rat pups, the magnitude of the conditioned response declined rapidly with age making it very difficult to establish by day 5 (Kindermann et al., 1994). In further work, Hudson et al. (2002) showed that nonnutritive sucking was the major reinforcer while obtaining milk appeared unimportant. Rabbit pups that could suck nipples without obtaining milk gave as clear evidence of conditioning as pups that did obtain milk (Fig. 4B). The apparent lack of importance of milk as a reinforcer might seem surprising given its obvious biological relevance and its effectiveness in infant rats. However, Hudson et al. (2002) did not rule out completely the potential rewarding role of milk. It appears that in such young pups, it is the mammary pheromone and not the suckling act itself that acts mainly as an unconditioned stimulus (Coureaud et al., 2005).

28

RAYMOND NOWAK

c. The human baby Newborn babies display asymmetric mouthing, consisting of a combination of mouth opening and strong twisting of the upper lip toward the left or the right, just before seizing the nipple. Asymmetrical mouthing is associated with head turning in the same direction and is an oral response to the tactile stimuli which indicate the commencement of feeding. Noirot and Alegria (1983) studied asymmetrical mouthing in newborn infants who were given the opportunity to hear a recorded human voice during suckling. Infants were exposed to the human voice in the first week of life and were tested at the age of 7 hr (naı¨ve babies), 7–25 hr (after being exposed to the voice between one and three times), and 1–7 days (after being exposed to the voice at least four times). During the test, the human voice was heard from loudspeakers positioned on either side of the infant’s head. Noirot and Alegria (1983) found that not only did the percentage of infants responding to the voice with asymmetrical mouthing increase with the number of training sessions (Fig. 4C) but also the expression of this response depended on the feeding situation. Breast‐fed babies generally mouthed in the direction of the voice whether it came from the right or the left loudspeaker while bottle‐fed infants usually made asymmetrical mouthing toward the left, where the bottle had been presented during the feeding sessions (Alegria and Noirot, 1978; Noirot and Alegria, 1983). Even after only one to three previous associations, the rapid enhancement of mouthing in response to the human voice resulted from the fact that this asymmetrical mouthing was followed by suckling when the voice was heard on those prior occasions. Breast‐fed babies had to twist their mouth toward the right to find the nipple when in the mother’s left arm, and toward the left when in her right arm, and therefore were conditioned to orient differentially to the source of food in these two contexts. The experience for bottle‐fed babies was quite different. Because mothers were predominantly right handed, they systematically hold their baby in their left arm and the bottle in the right arm. As a consequence, the baby had to mouth toward the left in order to find the teat, and never experienced the bottle from the other side. Their oral response was therefore shaped by their early feeding experience. V. SUCKLING

A. KEY ROLE

AND THE

OF THE

DEVELOPMENT OF A PREFERENCE MOTHER IN SHEEP

FOR THE

FIRST SUCKLING EPISODES

Rosenblatt (1983) proposed that olfactory cues, particularly those from the nest and the mother, became established as incentives during early development because they had been associated with maternal stimulation.

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

29

Based on the literature cited earlier, a clear implication of Rosenblatt’s proposal is that in rats and rabbits maternal odors associated with milk or nonnutritive sucking become preferred, and in humans the mother’s voice associated with suckling become attractive. However, it must be stressed that the substantial amount of work on early learning in rat or rabbit pups has always involved testing either the development of a preference toward a simple stimulus (such an artificial odor) or the motivation to make contact with an anesthetized female which was not the mother. Although some authors positioned their findings in the context of filial attachment (reviewed in Roth et al., 2004), no study had truly investigated the respective roles of nonnutritive sucking and milk intake in the establishment of an individual relationship with the mother. The lamb provides an excellent model to tackle this problem. In ruminants, the survival of the young depends on the maintenance of proximal contact with its own dam since maternal selectivity eliminates the possibility of alloparental care (Nowak et al., 2000). This behavioral constraint is probably the basis of the development of an early preference for the mother by the neonate. Most lambs can discriminate between their own and an alien maternal ewe by 24 hr after birth and display a preference for their dam (Nowak, 1991, 1994; Nowak and Lindsay, 1990; Nowak et al., 1989; Shillito and Alexander, 1975). Their preferential orientation improves markedly in the first few days of life. While recognition of the mother at 24 hr is based primarily on cues that lambs can perceive at close quarters (<50 cm), they can clearly discriminate their mother from a distance of several meters when 3 days old (Nowak, 1990, 1991). Although lambs respond specifically to some maternal cues immediately after birth (Vince and Ward, 1984), the development of a preference for the mother is based on postnatal learning facilitated by suckling. Goursaud and Nowak (1999) showed that when suckling was prevented during the first 6 hr after birth by covering the ewe’s udder, the lambs’ discriminative ability was impaired at the age of 24 hr even though they had access to the maternal udder after the period of food deprivation. Moreover, the development of the preference for the mother was directly linked to the length of the delay of the first suckling episode. In contrast to lambs suckled immediately after birth, those that had been deprived of suckling for 2, 4, or 6 hr did not display any preference for their mother at 24 hr (Fig. 5). By 48 hr, lambs prevented from suckling for 6 hr still did not display a preference for their dam, while those deprived for 4 hr tended to spend more time near their mother. When tested for a third time at 4 days of age, lambs from all groups had restored normal relationships with their dam. The poor performance in mother recognition at 24 hr was not a reflection of physical weakness in food‐ deprived animals or disturbed maternal behavior in ewes temporarily

30

RAYMOND NOWAK

Fig. 5. Time spent near the mother or alien maternal ewe in a two‐choice test by lambs at 24, 48, and 96 hr of age (mean  SEM). Control lambs had unrestricted access to their mother’s udder. Lambs from the other groups were left with their dam but deprived of suckling from birth for either 2, 4, or 6 hr after which they were allowed to suckle.

prevented from nursing. Rather, our results point out to the key role of the first suckling episodes as rewards and to the initiation of a biological state in the lamb facilitating early learning of maternal cues. The importance of suckling in developing a preference is unique to the neonatal phase as preventing lambs from suckling at a later stage is void of effect. Preventing 3‐day‐old lambs from accessing the udder for 6 hr did not

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

31

affect the preferential relationship previously established between dam and offspring as measured 24 hr later (Nowak et al., 1997b). In a related study, Napolitano et al. (2003) have shown that lambs that had been permitted to suckle only during the first day after birth and then remained with their nonnutritive dam while being artificially fed, still displayed a clear preference for her at 1 month of age. Obviously, the association between the positive reinforcement of suckling and the mother occurs only within the first hours of life and consequently leads to a long‐lasting preferential relationship with her. Once the bond is formed the influence of sensory factors provided by the mother becomes predominant in the maintenance of the lamb’s relationship with her while the nurturing role of the dam is unimportant. B. SENSORY STIMULI EMBEDDED

IN

SUCKLING

Suckling is a complex behavior composed of oral and gastrointestinal stimulation. The work described in an earlier section on olfactory conditioning in rat pups clearly demonstrates the involvement of various sensory stimuli along the oro‐gastrointestinal sphere. The same applies to the development of a preference for the mother in sheep: pre‐ and postingestive processes can trigger the development of filial preference independently from each other. In our experimental paradigms, ewes and lambs were kept in individual pens for the first 6 hr after birth to favor early interactions and facilitate handling of the neonates by the experimenter. Lambs were temporarily prevented from suckling by covering the mother’s udder at parturition and suckling was replaced by various human interventions during which oral or gastrointestinal stimulation always occurred in the presence of the mother. Ewes and lambs were then transferred into a collective yard so that they could interact with other individuals until the preference for the mother was tested. 1. Gastrointestinal Stimulation The effects of gastrointestinal stimulation were studied with the use of a tube‐feeding system to control the nature and the volume of the liquids given to the lambs. The cannulae, identical to those used in pediatric units to feed premature human babies, were designed to avoid any injury or irritation of the mucosa. The experimenter entered the lambing pen calmly so as not to scare the ewe and her young, caught the lamb gently, and inserted the cannula slowly into one of its nostrils. Intranasal tube feeding was preferred to intraoral tube feeding because it ensured that the lambs did not display any oral activity during feeding, nor did they perceive the taste of the various fluids infused. The treatment consisted of four to seven human interventions, each lasting only a few minutes, and covered

32

RAYMOND NOWAK

respectively a period of 6 or 12 hr after birth. After these treatments, the udder of the mother was uncovered and lambs could suckle freely. It must be stressed that the behavioral outcome was exactly the same when lambs were tested for mother recognition at the age of 24 hr, whether the treatments lasted for 6 or 12 hr. Our results systematically demonstrated that tube‐feeding lambs with colostrum for the first few hours after birth was sufficient to trigger the development of a preference for the mother within the first day and lambs behaved as if they had suckled their dam (Goursaud and Nowak, 1999; Nowak et al., submitted for publication; Val‐Laillet et al., 2004b). When volumes of diverse infused liquids were small (2.5%–5% birth body weight spread over the four or seven tube‐feeding sessions), only lambs receiving colostrum developed a preference for their mother (Fig. 6A). Those receiving either saline or water behaved like unsuckled lambs: in the two‐choice test, they selected a ewe at random or shared their time between the two animals (Nowak et al., submitted for publication; Val‐Laillet et al., 2004b). Even an isocaloric solution of lactose, the main sugar found in ovine colostrum, was not followed by any specific behavioral output, suggesting that this nutrient is not involved in the development of early filial attachment. Lambs tube‐fed with either ovine or bovine colostrum, both collected within hours after parturition, developed a clear preference for their mother, while those receiving mature ovine milk, obtained at 3 weeks of lactation, displayed a weaker response (Nowak et al., submitted for publication). This implies that the behavioral effect of colostrum is not specific to the species, bovine and ovine colostrums being rather similar in their gross composition (Fig. 3) but more likely to the biochemical composition of ‘‘early’’ milk. However, the development of early filial preferences in sheep is not uniquely influenced by milk constituents. When the volumes infused over 12 hr were increased up to 10% birth body weight, mimicking the amounts ingested when lambs suckle freely, both colostrum and saline facilitated the development of a preference for the mother (Fig. 6C). The behavioral effect was even visible as early as 12 hr after birth, after the completion of the intubation treatments (Val‐Laillet et al., 2004b). Thus, a nonnutritive liquid, such as saline, has a behavioral impact only if a significant volume is infused into the stomach of the lamb, which demonstrates that beyond a threshold, internal stimuli can also trigger mother preference through noncaloric, nonnutritional factors. By linking directly the infusion of colostrum to the development of a social preference, our experimental paradigm mimicked those commonly used with rat pups to study olfactory conditioning. In rats, ingestion of milk through an intraoral cannula in association with a novel, initially aversive

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

33

Fig. 6. Time spent near the mother or alien maternal ewe in a two‐choice test by lambs at 24 hr of age (mean  SEM). (A) Control lambs had unrestricted access to their mother’s udder (suckling), the others were left with their dam but deprived of suckling for 6 hr by covering the udder of the ewe; they were tube‐fed small amounts of colostrum, lactose, or water (5% body weight). (B) Lambs were left with their dam but deprived of suckling for 6 hr by covering the udder of the ewe; they were tube‐fed small amounts of ovine colostrum, bovine colostrum, ovine mature milk, or sham fed. (C) Lambs were left with their dam but deprived of suckling for 12 hr by covering the udder of the ewe; they were tube‐fed small volumes (5% body weight, left histogram), or large volumes (10% body weight, right histogram) of either ovine colostrum or saline.

34

RAYMOND NOWAK

odor leads to either a preference for that odor or a suppression of aversion (Brake, 1981; Johanson and Hall, 1982; Johanson et al., 1984; Sullivan and Hall, 1988). The odor acquires a positive hedonic value only if it is presented before milk infusion (forward pairing). Neither mere exposure to the new odor, backward presentation of milk, nor random pairing is effective. The development of mother preference in sheep also relies on an association between a reward and new sensory signals emanating from the mother’s body. However, our experimental situation did not follow one major rule of classical conditioning, that is, the temporal association between a conditioned signal and the reward in a specific time window. The lambs were left constantly with their mother while an experimenter intervened briefly for the tube‐feeding procedure. While in rat pups the odor served as a signal reliably predicting the delivery of milk (Brake, 1981; Johanson and Hall, 1982; Sullivan and Hall, 1988), this was not the case for our lambs since tube feeding was not associated with any specific preceding event. Nonetheless, it was true associative learning as the lambs associated a reward, infusions of colostrum or of nonnutritive fluids, with an ‘‘unconditioned stimulus,’’ the mother’s characteristics and behavior. 2. Oral Stimuli To investigate the role of the oral phase of suckling, we conducted a series of experiments with lambs that were bottle‐fed small amounts of water by an experimenter in the presence of their mother. Small volumes of water (10 ml) were added to the bottle to maintain nonnutritive sucking activity over several minutes. Lambs, unlike rat pups or human babies, do not suck a dry teat for long. Their oral activity declines rapidly unless they receive fluids into their mouth. As in the experiments described previously, the treatment consisted of four to seven human interventions, each lasting only a few minutes and covering respectively a period of either 6 or 12 hr after birth. Nonnutritive sucking induced a preference for the mother as tested at the age of 24 hr (Shayit et al., submitted for publication; Val‐Laillet et al., 2006). In contrast, infusing similar amounts of water via a nasogastric cannula was without any effect (Val‐Laillet et al., 2006). Drinking water from a bottle exerts several nonnutritive actions at the level of the digestive tract: sucking, swallowing, gastric distension and emptying, and hydration. Whereas the rewarding effects of a large volume of nonnutritive fluids infused into the stomach are modulated postprandially, at the level of the gastrointestinal tract, the effects due to bottle feeding reflect an activation of the oral sphere. In our experiments, care was taken to provide newborn lambs with tiny amounts of water, which in most cases represented between 0.5% and 1% of their body weight. These volumes were not likely to

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

35

trigger gastrointestinal distension nor could they provide sufficient hydration to modulate the behavior of the lamb. Rather, our results suggest that it is the combination of nonnutritive sucking and swallowing that is the main reinforcer. Lambs that suckled a feeding bottle containing water expressed as clear a preference for their dam as those that sucked a bottle containing the same volume of colostrum. The fact that colostrum did not lead to better behavioral performance compared to water demonstrates that nonnutritive sucking per se is as powerful a reward as colostrum in facilitating the development of mother preference. C. BIOLOGICAL MECHANISMS 1. Immediate Effects Nutritive and nonnutritive gastrointestinal stimuli could act through several mechanisms, the elucidation of which still remains hypothetical for the lamb. In rats, arousal may be an important factor determining the pup’s subsequent response to an odor associated with milk infusion. Diverse manipulations, such as stroking the ventral and perineal areas of pups, tail pinching, amphetamine or morphine injections, prolonged isolation, huddling with conspecifics, or a warm environment all lead to the development of odor preferences (Leon et al., 1987). These experimental situations mimic in some way the stimuli provided by the dam during nursing and the pups’ response to them. In lambs, general arousal in itself is not sufficient to explain the development of mother preference. Human intervention played a role in arousing the lamb and hence enhanced its motivation to interact with the mother, explore her body, and seek the teat. However, the enhancement of exploratory behavior immediately after intubation was observed equally in lambs which did not receive any fluids (Val‐Laillet et al., 2004b). The lower level of vocal activity noticed in lambs receiving colostrum in comparison to saline or sham manipulation is more likely to provide an explanation (Val‐Laillet et al., 2004b). Garcia Gonzalez and Goddard (1998) have also reported that lambs receiving supplementary colostrum immediately after birth were less active and vocal in the presence of their dam than their nonsupplemented siblings. Milk has been clearly demonstrated to have calming effects in isolated rat pups and human infants (Blass, 1997; Blass and Fitzgerald, 1988). This effect has been shown to be controlled by endogenous opioids and cholecystokinin (CCK), two neuropeptidic systems also involved in early learning and social attachment (Blass, 1996; Nelson and Panksepp, 1998; Weller and Blass, 1988a; Weller and Feldman, 2003). Colostrum ingestion by newborn lambs may well trigger physiological processes involved in quieting in a similar manner to

36

RAYMOND NOWAK

those described in rat pups. The ‘‘comforting state’’ induced by the ingestion of colostrum could well participate in the development of the preference for the mother by providing physiological conditions facilitating early learning. In addition, our results show that lambs infused with saline were globally less vocal than those receiving no fluids (Val‐Laillet et al., 2004b) suggesting that beyond a volumetric threshold the effects of nutritive and nonnutritive gastrointestinal stimulation are quite alike. In rat pups, gastric saline infusion reduces ultrasonic vocalizations (Nelson and Alberts, 2002) and increases paradoxical sleep (Lorenz et al., 1998), just like milk. Therefore the infusion of large volumes of saline in newborn lambs may also facilitate the development of mother preference through a ‘‘comforting state’’ induced by noncaloric, nonnutritional factors. The mechanisms of action triggered by nonnutritive sucking differ from those elicited by gastrointestinal stimulation. One immediate unexpected consequence of nonnutritive sucking is that lambs developed an interest in the human (Val‐Laillet et al., 2006). During human interventions, bottle‐fed but not tube‐fed animals made gradual contact with the experimenter when the person entered the pen and displayed typical, neonatal teat‐seeking behavior toward the person’s body. After 12 hr of this treatment, lambs bottle‐fed with water displayed an ambivalent attraction to the human and their dam (Val‐Laillet et al., 2006). Further tests performed in an open‐field arena, in the presence/absence of a human, demonstrated that lambs bottle‐ fed water calmed down and sought contact with the human as soon as the person entered the pen. Once the human left, bottle‐fed lambs were the only ones to display increased distress behavior, this being similar to that described by Ainsworth as the attachment response for human infants (Ainsworth, 1982; Ainsworth and Bell, 1970). Despite this initial hypothetical ‘‘affective’’ relationship with the human care giver, bottle‐fed lambs never preferred the person over their mother in a two‐choice test, demonstrating that the immediate reinforcing properties of nonnutritive sucking was associated both with the dam and the human. That the preference for the mother is displayed at 24 but not 12 hr suggests that the human interventions interfered with the normal filial attachment process. The fact that the rewarding effects of nonnutritive sucking are directly linked to the partner providing the stimulus suggests that activation of the mechanisms involved in this associative learning is immediate and does not last long once the stimulus has ceased. An illustration of it is found in the calming effect of pacifiers in crying babies. Calming is only obtained as long as the pacifier remains in the babies’ mouth. Once it is removed, crying and distress resume (Blass, 1994). Because the human intervention was predictive for the occurrence of nonnutritive sucking, the relationship that lambs developed with the person could have been established through a classical

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

37

conditioning procedure. Initially neutral, the human was rapidly associated with the bottle‐feeding and the satisfaction of fulfilling basic neonatal needs, in this case sucking (the reward). The same applies when lambs suck their mother’s teat. Of course, under natural conditions the reward provided by suckling is far more diverse. Oral and gastrointestinal stimuli whether nutritive or not act in complement of each other. They are each sufficient on their own to facilitate the development of early filial preferences suggesting that the neurophysiological pathways involved are either similar or redundant. 2. Opioids and Cholecystokinin Among the array of postabsorptive physiological mechanisms activated by milk ingestion that modulate infantile behavior, two neurochemical systems have been examined for their relation with filial attachment. At the gastrointestinal level, the gut peptide CCK, which is released after suckling, not only induces sedation and sleep but is also involved in learning (reviewed in Blass, 1996; Dauge´ and Le´na, 1998). Uvna¨s‐Moberg et al. (1987) were the first to point out that suckling induces simultaneous changes in CCK levels in the mother and young, and regarded this as ‘‘an expression of physiological symbiosis between mother and child.’’ In rat pups, physiological studies have shown that plasma CCK increases significantly after reunion with the dam (Weller and Blass, 1988b) which is presumably due to the combined effects of food ingestion and nonnutritive interactions. Administering exogenous CCK decreases ultrasonic vocalizations of isolated rat pups like milk (Weller et al., 2001) while devazepide, a CCK‐1 receptor antagonist, reduced the quieting effects of milk (Blass and Shide, 1993). Moreover, administration of CCK leads to the formation of conditioned olfactory preferences (Weller and Blass, 1990) whereas such conditioned responses are blocked by devazepide (Weller and Blass, 1988b). From the data obtained in rats, it is proposed that CCK, peripheral CCK‐1 receptors, and vagal afferent fibers are all involved in food‐rewarded learning (Morley et al., 1992). Centrally, the brain endorphin reward system is well known for its role in social affiliative behaviors, including infant–mother attachment (reviewed in Nelson and Panksepp, 1998, Panksepp et al., 1994), and numerous studies have shown that endogenous opioids are released in response to physical contact and social stimuli. Reduction of the negative affective response to social isolation, measured by isolation‐induced vocalization, is reported following administration of morphine and opioid agonists in puppies (Panksepp et al., 1978a),infant guinea pigs (Herman and Panksepp, 1978), domestic chicks (Panksepp et al., 1978b), primates (Kalin et al., 1988, 1995), and preweaning rat pups (Kehoe and Blass, 1986; Weller and Blass, 1988a).

38

RAYMOND NOWAK

In addition, milk‐induced analgesia in neonatal rat pups is blocked by pretreatment with an opioid antagonist (Blass and Fitzgerald, 1988), and injections of morphine, like milk ingestion, lead to the formation of conditioned olfactory preferences (Shide and Blass, 1991). Like in the species cited earlier, the use of selective receptor antagonists revealed a physiological role of endogenous CCK and opioids in the lamb. In both cases, when antagonists were injected immediately after birth, the expression of mother discrimination by lambs was impaired subsequently at 24 hr and 48 hr (CCK: Goursaud and Nowak, 2000; Nowak et al., 1997a,c; opioids: Shayit et al., 2003). Behavioral disturbances were produced by CCK‐1 receptor antagonists, even when the agent was known to act only at the peripheral level. The effect was less durable than that induced by a CCK‐1 receptor antagonist crossing the blood–brain barrier, and suggests that (1) an action of CCK at the peripheral level is highly probable and (2) the vagus nerve relays the information from the periphery to the brain, like in rats (Morley et al., 1992). Another interesting fact is that the inhibitory effect of a CCK‐1 receptor antagonist declined when postnatal injections were delayed (Nowak et al., 2001). This fits well with our data showing that the detrimental effect of food deprivation on filial preferences declines the longer after birth it is imposed (Nowak et al., 1997b). In support of our pharmacological studies, two other findings corroborate the involvement of CCK as one of the first biological steps in the cascade of events leading to the development of a preference for the mother. First, plasma CCK levels were low at birth, increased after suckling, and only lambs showing an increase in CCK over the first 6 hr after birth displayed a preference for their mother at the age of 24 hr (Nowak et al., 1997b,c; Nowak et al., submitted for publication). Second, there was a strong parallel between the type of fluids ingested by lambs, their plasma CCK level at 6 hr, and their behavior in the two‐choice test. Lambs tube‐fed with colostrum, but not with water, lactose, or mature milk, showed an increase in plasma CCK and developed a preference for their dam (Nowak et al., submitted for publication). Neonatal suckling, colostrum ingestion, and release of CCK in the hours after birth are all implicated in the early development of mother preference. Once this relationship is established, the action of these three factors is no longer required. The existence of neonatal time window for the mediation of both suckling and CCK shows that their mechanisms of action are intimately linked. This time window may reflect the transition from an initial memory trace under the influence of both suckling and CCK to a more permanent memory which is under the influence of other factors. We do not know if the opioidergic system follows the same pattern of activation. Likewise, it is not known if in lambs postingestive stimuli act mainly via the cholecystokininergic system while oral stimuli activate the opioidergic

39

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

Learning of the mother’s characteristics

Amygdala

Maternal behavior, nursing

O pi oi ds

Nucleus of the solitary tract

Central compartment

Hypothalamus

Trigeminal, facial, glossopharagial, vagal nerves Vagal nerve

Suckling Gastrointestinal phase

Colostrum, gastric distension

Visceral compartment

Nonnutritive sucking, taste

CC

Oral phase

K

Birth

Fig. 7. Hypothetical activation of the opioidergic and cholecystokininergic systems by suckling and their involvement in the development of early filial attachment.

network, as in rats (Shide and Blass, 1991). There is evidence that nonnutritive sucking might act through the opioidergic system (Shayit et al., submitted for publication) and our neurophysiological studies on the nucleus of the solitary tract which receives afferent fibers are consistent with the hypothesis that CCK acts on vagal mechanoreceptive endings (Guevara‐Guzman et al., 2005). This neural activation may reflect the pathway by which colostrum ingestion specifically activate brain structures involved in early learning of maternal cues, satiation, and even appeasement (Val‐Laillet, 2004; Val‐Laillet et al., 2004a) (Fig. 7).

40

RAYMOND NOWAK

As for colostrum, the endogenous opioids and CCK may facilitate the early development of filial attachment through a ‘‘calming effect.’’ In other species, modification of opioid or CCK activity not only alters markedly the infant’s learning ability but also affects the comfort obtained from social interactions (reviewed in Blass, 1996; Nelson and Panksepp, 1998; Panksepp et al., 1994). In our two‐choice test, unlike controls, lambs treated at birth with CCK‐1 receptor antagonists or naltrexone did not exhibit any preference for their mother suggesting a possible disruption in learning of maternal characteristics. In both instances, we have also found that treated lambs were more vocal than controls (Goursaud and Nowak, 2000; Shayit et al., 2003). Consequently, the disturbance observed in the vocal activity of the lambs following the injection of the antagonist supports the idea of an unsatisfied need for comforting maternal contact, and this may have affected the development of filial attachment.

VI. THE FIRST HOURS AFTER BIRTH A. VITALITY A neonatal period of food deprivation even as short as 2–6 hr delays the development of a preference for the mother by up to 2 days in sheep (Goursaud and Nowak, 1999; Nowak et al., 1997b; Nowak et al., submitted for publication). Although food deprivation affects the neonatal metabolism of the lamb (Eales and Small, 1981), none of the indices collected during our work could attribute the delay in filial attachment to poorer vitality in unfed lambs. Body temperatures decreased slightly in lambs receiving small amounts of colostrum, nonnutritive fluids, or no liquid at all, reflecting the thermogenic effect of colostrum (Val‐Laillet et al., 2004b). Unsurprisingly, lambs receiving the smallest loads of liquid had lost the most weight at the end of the treatment. Nonetheless, these neonatal variations in body weight and temperature were never correlated with the performance of the lambs in the choice test. Moreover, after having access to their mothers’ udder at the end of the period of food deprivation, lambs fully recovered from food deprivation within a few hours so that by the time of the test all differences had vanished (Val‐Laillet et al., 2004b). Compared to lambs that were suckled or received colostrum immediately after birth, food‐deprived lambs were just as vigorous and eager to reach the ewes in the choice test. Only the preference for the mother was affected. Depriving newborn lambs of suckling has obviously more dramatic and persistent effects on their affiliative behavior than on their general metabolism since it defers the establishment of a

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

41

preferential relationship with the mother to beyond the period of suckling deprivation. B. A SENSITIVE PERIOD

FOR THE

ACTION

OF

SUCKLING?

The development of a preference for the mother was systematically delayed in lambs that did not experience gastrointestinal stimulation around birth (Goursaud and Nowak, 1999; Nowak et al., 1997b; Val‐Laillet et al., 2004b). No matter what amounts of colostrum were ingested after the period of food deprivation, suckling never had the same facilitatory effects as in lambs which had the opportunity to suckle immediately after birth. The first postnatal hours appear to be a sensitive period during which the stimulatory effect of suckling on learning is optimal. It suggests that the positive association between neonatal visceral reinforcers and the mother could be facilitated by a specific neurophysiological state the infant is in by the time of delivery. In rats and humans, labor contractions have been demonstrated to arouse and prepare the fetus for adaptation to the postnatal environment. Human neonates delivered vaginally are awake, appear fully alert, and are highly responsive immediately after birth (Lagercrantz, 1996; Lagercrantz and Slotkin, 1986). In contrast caesarian‐delivered infants seem less active, a behavioral difference that has also been reported in rhesus monkeys (Meier, 1964). Even if the effects of maternal medication prior to caesarian cannot always be ruled out, data obtained in rat pups demonstrate that the mechanical action of the birth process has beneficial effects on the early behavior of the newborn. Rat pups born via caesarian section are not as vigorous as pups born by vaginal delivery. Simulating uterine contractions by repeated periods of mechanical compression subsequently induces normal breathing patterns compared to control fetuses, facilitates nipple attachment, and reduces mortality (Abel et al., 1998; Ronca and Alberts, 1995). Data obtained on caesarian‐delivered human babies even indicate that a brief period of mere exposure to an odorant immediately after birth is sufficient for the development of a learned preference for that odor, provided that the fetus has been subjected to uterine contractions (Varendi et al., 2002). Thus, the neurophysiological changes occurring during the final stage of the birth process prepare the fetus to become rapidly familiar with some sensory cues in its postnatal environment. Neonatal arousal may increase the lambs’ responsiveness to external stimuli, favor early mother–young interactions, and in association with suckling, may enhance learning efficiency so that the filial bond develops rapidly. C. SPECIFICITY

OF

COLOSTRUM

The similar effects of ovine or bovine colostrums on the development of a preference for the mother in sheep suggest that their stimulating

42

RAYMOND NOWAK

properties are related to common characteristics in their biochemical composition. Both colostrums have also comparable effects on the release of plasma CCK (Nowak et al., submitted for publication). The nature of the major colostral components is rather similar for the ewe and the cow even though their ratios vary between the two species (Hadjipanayiotou, 1995; Tyran and Lisowski, 1979) and would explain why bovine colostrum is biologically sound for lambs. In contrast, the effects of mature ovine milk, the composition of which differs markedly from colostrum, are rather different: the preference for the mother is less pronounced, the behavioral outcome more variable, and the release of CCK nonexistent. The greater biological impact of colostrum compared to other fluids, other than in the nutritional and immunological sphere, has been described in lambs in at least two other experimental situations. The first one, based on electrophysiological studies by Robinson et al. (1995), showed that fetal lambs express motor and autonomic responses to gustatory stimuli infused over the surface of their tongue. About 134‐ to 137‐day‐old fetuses respond differently to infusions of bovine milk, mature ovine milk obtained at 1–2 weeks postpartum, or colostrum. Colostrum resulted in a significant decrease in masseter and geniohyoid activity (respectively jaw adductor and tongue elevator muscles involved in swallowing) several minutes after infusion. On the other hand, ovine milk induced increased activity and bovine milk no change in activity. The fine chemosensory discrimination between colostrum and mature milk by fetal lambs prepare them to respond more positively to the organoleptic properties of early milk. That fetal lambs react differently to various lacteal fluids suggests that their brain has the potential to process information from the oral cavity prior to any gustatory experience. The second demonstration of a specific biological action of early milk comes from Sevi et al. (1999). They found that a gradual transition from maternal milk to reconstituted milk minimized stress when very young lambs were switched from maternal suckling to artificial rearing. Lambs benefiting from a gradual passage from maternal to reconstituted milk had similar behavioral, endocrine, and immune responses to suckled lambs. On the other hand, lambs subjected to an abrupt change in their diet exhibited higher stress responses and poorer growth rate. These results along with the unique effectiveness of colostrum in triggering the development of a preference for the mother, support the idea that the newborn lamb’s gut–brain axis is physiologically adapted to respond to early milk constituents. A specificity of the action of colostrum has been demonstrated in neonates of other species, in particular in the development of organs and their functions. For instance brain, liver, kidney, spleen, skeletal muscle, and heart protein synthesis rates are greater in colostrum‐fed piglets than in

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

43

piglets fed either milk or formula during the first 6 hr after birth only (Burrin et al., 1992, 1995, 1997). Metabolic and endocrine variables in calves are also strongly influenced by the amount of colostrum intake before feeding with milk substitutes. Feeding newborn calves with only trace amounts of colostrum impairs absorption capacity as well as protein and fat metabolism, and has deleterious effects on the gastrointestinal endocrine system (Hammon and Blum, 1998; Rauprich et al., 2000a,b). Thus, even if mature milk is a nourishing liquid for growing infants, it is not adapted to the specific physiological needs of the newborn and logically it will not be as effective as colostrum in triggering the biological processes involved in early development. In addition to the effects of colostrum cited earlier, numerous studies have identified lacteal bioactive substances (reviewed in Grosvenor et al., 1992; Meisel, 2004; Peaker and Neville, 1991; Strbak, 1992). Among these are the opioid‐like peptides which have been suggested to have an effect on neonatal behavior, an interesting point given that milk ingestion exerts its effects partly through the opioidergic system (Blass and Fitzgerald, 1988). The possibility of an action of these biopeptides is suggested by the detection of b‐casomorphin immunoreactive material in the plasma of calves and lambs after the ingestion of milk (Read et al., 1990; Umbach et al., 1985). Though it is still unclear whether they are released in vivo in amounts that are physiologically significant, intraperitoneal injections of b‐casomorphins induce sleep in infant rats (Taira et al., 1990). Another protein, colostrinin, isolated from ovine colostrum (Janusz and Lisowski, 1993; Janusz et al., 1981) has been shown to have promnesic properties in adult rats (Popik et al., 1999) and to facilitate cognitive function in patients with Alzheimer’s disease (Leszek et al., 1999). When newborn lambs suckle their mother they ingest not only nutrients but also a wide array of hormones and biopeptides that affect their development and growth. The involvement of these potential candidates in the development of the colostrum‐induced mother preference is worth investigating.

VII. CONCLUSIONS Nursing–suckling interactions are the result of physiological, morphological, and behavioral changes that ensure extrauterine survival at a time when the young is entirely dependent on its mother for the provision of food. The main functions of early maternal care are providing the neonate with milk to satisfy its needs of hunger and thirst, protecting it from aggression from the postnatal environment, and in some species providing intimate bodily contact. Although the facilitating role of suckling in the

44

RAYMOND NOWAK

development of filial attachment has only been demonstrated in sheep to date (Goursaud and Nowak, 1999; Nowak et al., 1997b; Val‐Laillet et al., 2004b), it is quite conceivable that this phenomenon does not constitute an exception in mammals. Data presented in this chapter show that suckling plays a role in facilitating early learning in species as diverse as rat pups (Brake, 1981; Johanson and Hall, 1979; Johanson et al., 1984; Sullivan and Hall, 1988), rabbit pups (Hudson, 1985; Hudson et al., 2002), puppies (Bacon and Stanley, 1970; Stanley et al., 1970), and human neonates (Alegria and Noirot, 1978; Noirot and Alegria, 1983). Even if some of these authors have extrapolated the data obtained on conditioned preferences to the context of mother–young interaction and filial attachment, they have never demonstrated a true relationship between suckling and the development of a preference for the mother. This does not mean that, under natural conditions, neonates of these four species do not learn individual maternal cues while suckling, and there is some evidence that human babies can learn their mother’s odor at the breast (Cernoch and Porter, 1985). Part of the explanation for the lack of evidence for the role of suckling in the establishment of filial preferences, especially in altricial mammals, lies in the fact that attachment is not a general process (Gubernick, 1981) and that despite their early learning ability, neonates of these species do not develop a preference for their own mother. A second possibility is that attachment takes time to develop and the dynamics of its development varies across species. In altricial young, the preference for the mother may not establish in early age. In the ground squirrel for instance, this preference does not develop until the young start leaving their burrow, when 2 weeks old, and interact with other individuals than their mother (Holmes, 1990). Natural selection favors the evolution of mechanisms to ensure that the caregiver’s offsprings are the beneficiaries of parental investment. When a mother and her neonate are likely to intermingle with other unrelated adults and young, as commonly occurs in gregarious mammals, mother and young rapidly develop the ability to recognize each other (Gubernick, 1981). In this context, it is not surprising that in sheep, a highly gregarious species in which mothers develop selective maternal care within hours after parturition, the neonate has the capacity to form rapidly preferential relationship with its dam. Freud (1940) was the first to claim that it is through suckling at the breast (and its hedonic value) that the infant progressively develops a bond with its mother. It was on the basis of Freud’s psychoanalytical interpretation that Brake (1981) built his line of argument to explain the facilitating effect of suckling and milk ingestion on the development of olfactory preferences in rat pups. It was on the basis of the same Freudian arguments that Harlow

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

45

(1959) refuted the importance of the nurturing mother in the development of filial attachment in primates. His work, now considered classic, showed that the attraction of contact comfort provided by a nonnursing cloth surrogate prevailed entirely over that of the activities associated with nursing or hunger appeasement. Infant macaques spent almost all their time clinging to the cloth surrogate, and little time on the wire surrogate regardless as to which provided milk. Although contact comfort provided by the surrogate mother was important for the development of the rhesus macaques (Harlow and Harlow, 1962), it did not imply that suckling was meaningless. Harlow himself (1961) involuntarily provided a demonstration of this fact. When infant macaques were raised in the presence of two cloth‐surrogate mothers, only one of which provided milk, the infants initially formed a preference for the feeding mother. Moreover, when tested in an open field in the presence of a new and potentially frightening stimulus, infant macaques preferentially went to the feeding mother. This initial preference vanished at about 100 days and suggests that, as for lambs, suckling may only be a reinforcing agent in the early stage of the relationship with the mother. However, even though it is noticeable that results on the effect of suckling on early learning from different species tend to fit into a similar pattern, the importance of suckling must be set in the neonatal ecology of each species. In primates characterized by biparental or communal care (titi monkeys, tamarins, marmoset) the primary attachment figures are the father or older siblings (Hoffman et al., 1995, Kostan and Snowdon, 2002; Mendoza and Mason, 1986), in spite of the mother’s major contribution to the infants’ nutritional needs. In cotton‐top tamarins for instance, infant carrying is shared among several individuals. The attachment object is the family member that invests the most effort in carrying, but also in transferring solid food in complement to the milk provided by the mother (Kostan and Snowdon, 2002). Suckling is one stimulation provided by parental care among several others. Its relevance must be cautiously balanced with regards to the effects of other stimulations such as licking, stroking, clinging, arousing, providing warmth, which have been demonstrated in some species to be just as rewarding as suckling (Leon et al., 1987; Roth et al., 2004; Sullivan and Hall, 1988; Sullivan et al., 1991; Wilson and Sullivan, 1994).

VIII. SUMMARY In mammals, the newborn is entirely dependent on its mother for the provision of food. Complementing specific nursing postures, a variety of maternal cues aid the newborn in its initial search for the teat allowing

46

RAYMOND NOWAK

suckling to occur shortly after birth. Milk is a complex water‐based solution, consisting of proteins, lactose as the main sugar, lipids, and minerals, which provides all the elements required for the neonate to survive, except for the oxygen drawn from the air it breathes. However, suckling has vital functions other than just supplying nutrients and fluids to the growing infant. Establishing contact with the maternal body, locating and attaching to a nipple or a teat, and suckling constitute the most intimate form of contact existing between the young and its mother. Stimuli associated with suckling shape the behavior of the neonate in diverse manners: in particular they induce calming, facilitate learning, and are common to a wide range of species including rats, rabbits, dogs, sheep, and humans. In altricial species (rat, rabbit), the neonate establishes a preference for initially neutral or aversive olfactory stimuli when they are associated with suckling. In the sheep, these neonatal learning processes result in an acquired recognition and a preference for the individual characteristics of the mother. Even human babies may develop preferential responses toward some of their mother’s sensory cues through suckling. The mechanisms of rewards appear to be rather similar across species. Suckling facilitates learning through pre‐ and postprandial stimulation: nonnutritive sucking, gustatory stimuli, oral ingestion of milk or colostrum can rapidly trigger olfactory conditioning in rat and rabbit pups as well as a preference for the mother in lambs. Some of these behavioral effects can even be attributed to the biochemical composition of early milk. The pathways of these reinforcing agents have been partly established and point to a major role of the opioidergic and cholecystokininergic systems as they exert calming effects and facilitate early learning (olfactory conditioning in the rat pup and development of a preference for the mother in the lamb). During feeding interactions, the maternal body provides new sensory cues which, being concomitant with endogenous opioid release after oral stimulation and CCK release via gut stimulation, will acquire hedonic properties and will be specifically sought after in subsequent separation–reunion situations. Over the past 40 years, suckling behavior has provided researchers with a rich source of theories and debates concerning the role of nature versus nurture in filial attachment. Our studies on lambs provide clear evidence of the rewarding role of suckling in the development, but not the maintenance, of a specific preference toward the mother. Acknowledgments None of the work carried out on the behavior of newborn lambs would have been possible without the precious help from numerous collaborators and friends. It is impossible to name them all but I wish to thank them for having always been willing to spend days and nights

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

47

collecting data, without counting their working hours, even when the environmental conditions were not very pleasant when lambing was in winter, whether in the Australian bush or in a French shed. This chapter would have never reached its final stage without the participation of two key contributors. First, I really appreciate the constructive comments of Dick Porter on the draft of this paper. Not only is he a man of great knowledge but his kindness and availability to help me at any time, as a next door neighbor in the lab, are irreplaceable. Second and most of all, I would like to thank Denise, my Australian wife. Not only did she accept to leave her country and come to live in France, but with the birth of our three boys, she involved me in the extraordinary beauty of breast‐feeding at a time when it was considered as ridiculous to breast‐feed a baby in the first hours after birth, and almost perverse to nurse it beyond 3 months. She also took great care in reading this chapter and improving my English.

REFERENCES Abel, R. A., Ronca, A. E., and Alberts, J. R. (1998). Perinatal stimulation facilitates suckling onset in newborn rats. Dev. Psychobiol. 32, 91–99. Ainsworth, M. D. S. (1979). Attachment as related to mother‐infant interaction. Adv. Study Behav. 9, 1–51. Ainsworth, M. D. S. (1982). Attachment: Retrospect and prospect. In ‘‘The Place of Attachment in Human Behavior’’ (C. M. Parkes and J. Stevenson‐Hinde, Eds.), pp. 3–30. Basic Books, New York. Ainsworth, M. D. S., and Bell, S. M. (1970). Attachment, exploration, and separation: Illustrated by the behavior of one‐year‐olds in a strange situation. Child Dev. 41, 49–67. Alegria, J., and Noirot, E. (1978). Neonate orientation behavior towards human voice. Int. J. Behav. Dev. 1, 291–312. Alexander, G. (1988). What makes a good mother? Components and comparative aspects of maternal behaviour in ungulates. Proc. Aust. Soc. Anim. Prod. 17, 25–41. Alexander, G., Stevens, D., Bradley, L. R., and Barwick, S. A. (1990). Maternal behaviour in Border Leicester, Glen Vale (Border Leicester derived) and Merino sheep. Aust. J. Exp. Agric. 30, 27–38. Amsel, A., Burdette, D. R., and Letz, R. (1976). Appetitive learning, patterned alternation, and extinction in 10‐d‐old rats with non‐lactating suckling as reward. Nature 262, 816–818. Arnold, J., Leslie, G., and Chen, S. (1987). Protein, lactose and fat concentration of breast milk of mothers of term and premature neonates. Aust. Paediatr. J. 23, 299–300. Atkinson, S., Alston‐Mills, B., Lo¨nnerdal, B., and Neville, M. C. (1995). Major minerals and ionic constituents of human and bovine milks. In ‘‘Handbook of Milk Composition’’ (R. G. Jensen, Ed.), pp. 593–622. Academic Press, San Diego. Bacon, W. E., and Stanley, W. C. (1970). Reversal learning in neonatal dogs. J. Comp. Physiol. Psychol. 70, 344–350. Bareham, J. R. (1976). The behaviour of lambs on the first day after birth. Br. Vet. J. 132, 152–161. Barone, R. (1978). Anatomie Compare´e des Mammife`res Domestiques. Tome 3: Splanchnologie, Foetus et ses Annexes. Fascicule II: Appareil Uro‐Ge´nital, Foetus et ses Anexes, Pe´ritoine et Topographie Abdominale Ecole National Ve´te´rinaire, Lyon.

48

RAYMOND NOWAK

Barr, R. G., Pantel, M. S., Young, S. N., Wright, J. H., Hendricks, L. A., and Gravel, R. (1999). The response of crying newborns to sucrose: Is it a ‘‘sweetness’’ effect? Physiol. Behav. 66, 409–417. Ben Shaul, D. M. (1962). The composition of milk of wild animals. Int. Zoo Yearbook 4, 333–342. Blackburn, D. G., Hayssen, V., and Murphy, C. J. (1989). The origins of lactation and the evolution of milk: A review with new hypothesis. Mammal. Rev. 19, 1–26. Blass, E. M. (1990). Suckling: Determinants, changes, mechanisms, and lasting impressions. Dev. Psychol. 26, 520–533. Blass, E. M. (1994). Behavioral and physiological consequences of suckling in rat and human newborns. Acta Paediatr. 397(Suppl.), 71–76. Blass, E. M. (1996). Mothers and their infants: Peptide‐mediated physiological, behavioral and affective changes during suckling. Regul. Pept. 66, 109–112. Blass, E. M. (1997). Infant formula quiets crying human newborns. J. Dev. Behav. Pediatr. 18, 162–165. Blass, E. M., and Fitzgerald, E. (1988). Milk‐induced analgesia and comforting in 10‐day‐old rats: Opioid mediation. Pharmacol. Biochem. Behav. 29, 9–13. Blass, E. M., and Shide, D. J. (1993). Endogenous cholecystokinin reduces vocalization in isolated 10‐day‐old rats. Behav. Neurosci. 107, 488–492. Blass, E. M., and Smith, B. A. (1992). Differential effects of sucrose, fructose, glucose, and lactose on crying in 1 to 3‐day‐old human infants: Qualitative and quantitative considerations. Dev. Psych. 28, 804–810. Blass, E. M., and Teicher, M. H. (1980). Suckling. Science 210, 15–22. Blass, E. M., Teicher, M. H., Cramer, C. P., Bruno, J. P., and Hall, W. G. (1977). Olfactory, thermal, and tactile controls of suckling in preauditory and previsual rats. J. Comp. Physiol. Psychol. 91, 1248–1260. Blass, E. M., Shuleikina‐Turpaeva, K., and Luschekin, V. (1988). Sensory determinants of nipple‐attachment behavior in 2–4‐day‐old kittens. Dev. Psychobiol. 21, 365–370. Blass, E. M., Shide, D. J., Zaw‐Mon, C., and Sorrentino, J. (1995). Mother as shield: Differential effects of contact and nursing on pain responsivity in infant rats—evidence for nonopioid mediation. Behav. Neurosci. 109, 342–353. Bowen, W. D., Oftebal, O. T., and Boness, D. J. (1985). Birth to weaning in four days: Remarkable growth in the hooded seal. Can. J. Zool. 63, 2841–2846. Brake, S. C. (1981). Suckling infant rats learn a preference for a novel olfactory stimulus paired with milk delivery. Science 211, 506–508. Burrin, D., Shulman, R. J., Reeds, P. J., Davis, T. A., and Gravitt, K. R. (1992). Porcine colostrum and milk stimulate visceral organ and skeletal muscle protein synthesis in neonatal piglets. J. Nutr. 122, 1205–1213. Burrin, D. G., Davis, T. A., Ebner, S., Schoknecht, P. A., Fiorotto, M. L., Reeds, P. J., and McAvoy, S. (1995). Nutrient‐independent and nutrient‐dependent factors stimulate protein synthesis in colostrum‐fed newborn pigs. Pediatr. Res. 37, 593–599. Burrin, D. G., Davis, T. A., Ebner, S., Schoknecht, P. A., Fiorotto, M. L., and Reeds, P. J. (1997). Colostrum enhances the nutritional stimulation of vital organ protein synthesis in neonatal pigs. J. Nutr. 127, 1284–1289. Caba, M., Rovirosa, M. J., and Silver, R. (2003). Suckling and genital stroking induces Fos expression in hypothalamic oxytocinergic neurons of rabbit pups. Brain Res. Dev. Brain. Res. 143, 119–128. Cannon, J. R., Bakker, H. R., Bradshaw, S. D., and McDonald, I. R. (1976). Gravity as the sole navigational aid to the newborn quokka. Nature 259, 42.

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

49

Carlsson, S. G., Fagerberg, H., Horneman, G., Hwanf, C. P., Larsson, K., Ro¨dholm, M., Schaller, J., Danielsson, B., and Gundewall, C. (1977). Effects of amount of contact between mother and child on the mother’s nursing behavior. Dev. Psychobiol. 11, 143–150. Casey, C. E., Neifert, M. R., Seacat, J. M., and Neville, M. C. (1986). Nutrient intake by breast‐ fed infants during the first five days after birth. Am. J. Dis. Child. 140, 933–936. Cernoch, J. M., and Porter, R. H. (1985). Recognition of maternal axillary odors by infants. Child Dev. 56, 1593–1598. Christensson, K., Siles, C., Moreno, L., Belaustequi, A., De La Fuente, P., Puyol, P., and Windberg, J. (1992). Temperature, metabolic adaptation and crying in healthy full‐term newborns care for skin‐to‐skin or in a cot. Acta Paediatr. 81, 488–493. Coureaud, G., Schaal, B., Langlois, D., and Perrier, G. (2001). Orientation response of newborn rabbits to odours of lactating females: Relative effectiveness of surface and milk cues. Anim. Behav. 61, 153–162. Coureaud, G., Schaal, B., Hudson, R., Orgeur, P., and Coudert, P. (2002). Transnatal olfactory continuity in the rabbit: Behavioral evidence and short‐term consequence of its disruption. Dev. Psychobiol. 40, 372–390. Coureaud, G., Montigny, D., Moncomble, A. S., Patris, B., Perrier, G., and Schaal, B. (2005). Apprentissages olfactifs du lapereau sous la me`re. 11e Journe´es de la Recherche Cunicole, November 29–30, 2005, Paris. Cowie, A. T. (1982). Lactation. In ‘‘Reproduction in Mammals: Hormonal Control of Reproduction’’ (C. R. Austin and R. V. Short, Eds.), pp. 195–231. Cambridge University Press, Cambridge. Dauge´, V., and Le´na, I. (1998). CCK in anxiety and cognitive processes. Neurosci. Biobehav. Rev. 22, 815–825. De Passille´, A. M. B., Rushen, J., and Hartsock, T. G. (1988). Ontogeny of teat fidelity in pigs and its relation to competition at suckling. Can. J. Anim. Sci. 68, 325–338. Distel, H., and Hudson, R. (1985). The contribution of olfactory and tactile modalities to the performance of nipple‐search behavior in newborn rabbits. J. Comp. Physiol. A 157, 599–605. Drewett, R. F., Kendrick, K. M., Sanders, D. J., and Trew, A. M. (1982). A quantitative analysis of the feeding behavior of suckling rabbits. Dev. Psychobiol. 15, 25–32. Eales, F. A., and Small, J. (1981). Effects of colostrum on summit metabolic rate in Scottish blackface lambs at five hours old. Res. Vet. Sci. 30, 266–269. Ewbank, R. (1964). Observations on the suckling habits of twin lambs. Anim. Behav. 12, 34–37. Ewbank, R. (1967). Nursing and suckling behaviour amongst Clun Forest ewes and lambs. Anim. Behav. 15, 251–258. Fetherston, C. M., Cregan, M. D., and Hartmann, P. E. (2001). Mammary gland defense: The role of colostrum, milk and involution secretion. Adv. Nutr. Res. 10, 167–198. Fraser, D. (1990). Behavioural perspectives on piglet survival. J. Reprod. Fertil. 40(Suppl.), 335–370. Freud, S. (1940). ‘‘An Outline of Psychoanalysis.’’ Norton, New York. Garcia Gonzalez, S. G., and Goddard, P. J. (1998). The provision of supplementary colostrum to newborn lambs: Effects on post‐natal lamb and ewe behaviour. Appl. Anim. Behav. Sci. 61, 41–50. Goldman, A. S. (1993). The immune system of human milk: Antimicrobial, antiinflamatory and immunomodulating properties. Pediatr. Infect. Dis. J. 12, 664–671. Goldman, A. S., and Goldblum, R. M. (1995). Defense agents in human milk. In ‘‘Handbook of Milk Composition’’ (R. G. Jensen, Ed.), pp. 727–745. Academic Press, San Diego.

50

RAYMOND NOWAK

Gonzalez‐Mariscal, G., Chirino, R., and Hudson, R. (1994). Prolactin stimulates emission of nipple pheromone in ovariectomized New Zealand white rabbits. Biol. Reprod. 50, 373–376. Gordon, K., and Siegmann, M. (1991). Suckling behavior of ewes in early lactation. Physiol. Behav. 50, 1079–1081. Goursaud, A. P., and Nowak, R. (1999). Colostrum mediates the development of mother preference by newborn lambs. Physiol. Behav. 67, 49–56. Goursaud, A. P., and Nowak, R. (2000). 2‐NAP, a peripheral CCK‐A receptor antagonist, modulates the development of a preference for the mother by the newborn lamb. Pharmacol. Biochem. Behav. 67, 603–611. Grosvenor, C. E., Picciano, M. F., and Baumrucker, C. R. (1992). Hormones and growth factors in milk. Endocr. Rev. 14, 710–728. Gubernick, D. J. (1981). Parent infant attachment in mammals. In ‘‘Parental Care in Mammals’’ (D. J. Gubernick and P. H. Klopfer, Eds.), pp. 243–305. Plenum Press, New York. Guevara‐Guzman, R., Le´vy, F., Jean, A., and Nowak, R. (2005). Electrophysiological responses of nucleus tractus solitarius neurons to CCK and gastric distension in newborn lambs. Cell. Mol. Neurobiol. 25, 393–406. Hadjipanayiotou, M. (1995). Composition of ewe, goat and cow milk and of colostrum of ewes and goats. Small Rum. Res. 18, 255–262. Hall, W. G. (1979). Feeding and behavioral activation in infant rats. Science 205, 206–209. Hall, W. G., Hudson, R., and Brake, S. C. (1988). Terminology for use in investigations of nursing and suckling. Dev. Psychobiol. 21, 89–91. Hammon, H. M., and Blum, J. W. (1998). Metabolic and endocrine traits of neonatal calves are influenced by feeding colostrum for different durations or only milk replacer. J. Nutr. 128, 624–632. Harlow, H. F. (1959). Love in infant monkeys. Sci. Am. 200, 68–74. Harlow, H. F. (1961). The development of affectional patterns in infant monkeys. In ‘‘Determinants of Infant Behavior’’ (B. M. Foss, Ed.). John Wiley and Sons Inc., New York. Harlow, H. F., and Harlow, M. K. (1962). Social deprivation in monkeys. Sci. Am. 473, 2–10. Harvey, P. H., and Clutton‐Brock, T. H. (1985). Life history variation in primates. Evolution 39, 559–581. Hebel, R., and Stromberg, M. W. (1986). ‘‘Anatomy and Embryology of the Laboratory Rat.’’ BioMed Verlag, Wo¨rthsee. Herman, B. H., and Panksepp, J. (1978). Effects of morphine and naloxone on separation distress and approach attachment: Evidence for opiate mediation of social affect. Pharmacol. Biochem. Behav. 9, 213–220. Herrenkohl, L. R., and Campbell, C. (1976). Mechanical stimulation of the mammary gland development in virgin and pregnant rats. Horm. Behav. 7, 183–197. Hofer, M. A., Shair, H., and Singh, P. (1976). Evidence that maternal ventral skin substances promote suckling in infant rats. Physiol. Behav. 17, 131–136. Hoffman, K. A., Mendoza, S. P., Hennessy, M. B., and Mason, W. S. (1995). Responses of infant titi monkeys, Callicebus moloch, to removal of one or both parents: Evidence for paternal attachment. Dev. Psychobiol. 28, 399–407. Holmes, W. G. (1990). Parent‐offspring recognition in mammals: A proximate and ultimate perspective. In ‘‘Mammalian Parenting Biochemical, Neurobiological, and Behavioral Determinants’’ (N. A. Krasnegor and R. S. Bridges, Eds.), pp. 441–460. Oxford University Press, New York.

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

51

Holst, P. J., Hall, D. G., and Allan, C. J. (1996). Ewe colostrum and subsequent lamb suckling behaviour. Aust. J. Exp. Agr. 36, 637–640. Hudson, R. (1985). Do newborn rabbits learn the odor stimuli releasing nipple‐search behavior? Dev. Psychobiol. 18, 575–585. Hudson, R., and Distel, H. (1982). The pattern of behaviour of rabbit pups in the nest. Behaviour 79, 255–271. Hudson, R., and Distel, H. (1983). Nipple location by newborn rabbits: Behavioural evidence for pheromonal guidance. Behaviour 85, 260–275. Hudson, R., and Distel, H. (1984). Nipple‐search pheromone in rabbits: Dependence on season and reproductive state. J. Comp. Physiol. A 155, 13–17. Hudson, R., and Distel, H. (1986). Pheromonal release of suckling in rabbits does not depend on the vomeronasal organ. Physiol. Behav. 37, 123–128. Hudson, R., and Distel, H. (1989). The temporal pattern of suckling in rabbit pup: A model of circadian synchrony between mother and young. In ‘‘Development of Circadian Rhythmicity and Photoperiodism in Mammals’’ (S. M. Reppert, Ed.), pp. 83–102. Perinatalogy Press, Boston. Hudson, R., Gonza´les‐Mariscal, G., and Beyer, C. (1990). Chin‐marking behavior, sexual receptivity, and pheromone emission in steroid‐treated, ovariectomized rabbits. Horm. Behav. 24, 1–13. Hudson, R., Cruz, Y., Lucio, R. A., Ninomiya, J., and Martı´nez‐Go´mez, M. (1999). Temporal and behavioral patterning of parturition in rabbits and rats. Physiol. Behav. 66, 599–604. Hudson, R., Labra‐Cardero, D., and Mendoza‐Soylovna, A. (2002). Sucking, not milk, is important for the rapid learning of nipple‐search odors in newborn rabbits. Dev. Psychobiol. 41, 226–235. Imagawa, W., Yang, J., Guzman, R., and Satyabrata, N. (1994). Control of mammary gland development. In ‘‘The Physiology of Reproduction’’ (E. Knobil and J. D. Neill, Eds.), pp. 1033–1063. Raven Press, New York. Janusz, M., and Lisowski, K. (1993). Proline‐rich polypeptide (PRP), an immunomodulatory peptide from ovine colostrum. Arch. Immunol. Ther. Exp. 41, 275–279. Janusz, M., Staroscik, K., Zimecki, M., Wieczorek, Z., and Lisowski, J. (1981). Chemical and physical characterization of proline‐rich polypeptide from sheep colostrum. Biochem. J. 199, 9–15. Jenness, R. (1979). The composition of human milk. Semin. Perinatal 3, 225–239. Jenness, R. (1985). Biochemical and nutritionan aspects of milk. In ‘‘Lactation’’ (B. L. Larson, Ed.), pp. 164–197. The Iowa State University Press, Iowa. Jensen, R. G. (1995). ‘‘Handbook of Milk Composition.’’ Academic Press, San Diego. Jensen, R. G., and Newburg, D. S. (1995). Bovine milk lipids. In ‘‘Handbook of Milk Composition’’ (R. G. Jensen, Ed.), pp. 543–575. Academic Press, San Diego. Jensen, R. G., Bitman, J., Carlson, S. E., Couch, S. C., Hamosh, M., and Newburg, D. S. (1995). Human milk lipids. In ‘‘Handbook of Milk Composition’’ (R. G. Jensen, Ed.), pp. 495–542. Academic Press, San Diego. Johanson, I. B., and Hall, W. G. (1979). Appetitive learning in 1‐day‐old rat pups. Science 205, 419–421. Johanson, I. B., and Hall, W. G. (1982). Appetitive conditioning in neonatal rats: Conditioned orientation to a novel odor. Dev. Psychobiol. 15, 379–397. Johanson, I. B., and Teicher, M. H. (1980). Classical conditioning of an odor preference in 3‐day‐old rats. Behav. Neural Biol. 29, 132–136. Johanson, I. B., Hall, W. G., and Polefrone, J. M. (1984). Appetitive conditioning in neonatal rats: Conditioned ingestive responding to stimuli paired with oral infusions of milk. Dev. Psychobiol. 17, 357–381.

52

RAYMOND NOWAK

Kalin, N. H., Shelton, S. E., and Barksdale, C. M. (1988). Opiate modulation of separation‐ induced distress in non‐human primates. Brain Res. 440, 285–292. Kalin, N. H., Shelton, S. E., and Lynn, D. E. (1995). Opiate systems in mother and infant primates coordinate intimate contact during reunion. Psychoneuroendocrinology 20, 735–742. Keen, C. L., Lonnerdal, B., Clegg, M., and Hurley, L. S. (1981). Developmental changes in composition of rat milk: Trace elements, minerals, protein, carbohydrate and fat. J. Nutr. 111, 226–236. Kehoe, P., and Blass, E. M. (1986). Opioid‐mediation of separation distress in 10‐day‐old rats: Reversal of stress with maternal stimuli. Dev. Psychobiol. 19, 385–398. Keil, W., von Stralendorff, F., and Hudson, R. (1990). A behavioral bioassay for analysis of rabbit nipple‐search pheromone. Physiol. Behav. 47, 525–529. Kenny, J. T., and Blass, E. M. (1977). Suckling as incentive to instrumental learning in preweanling rats. Science 196, 898–899. Kindermann, U., Hudson, R., and Distel, H. (1994). Learning of suckling odors by newborn rabbits declines with age and suckling experience. Dev. Psychobiol. 27, 111–122. Kintzinger, S. (1989). ‘‘Breastfeeding Your Baby.’’ Boubleday, Sydney. Koletzko, B., Rodriguez‐Palmero, M., Demmelmair, H., Fidler, N., Jensen, R., and Sauerwald, T. (2001). Physiological aspects of human milk lipids. Early Hum. Dev. 65, S3–S18. Konner, M., and Worthman, C. (1980). Nursing frequency, gonadal function, and birth spacing among !Kung hunter‐gatherers. Science 207, 788–791. Korhonen, H., Marnila, P., and Gill, H. S. (2000a). Milk immunoglobulins and complement factors. Br. J. Nutr. 84, S75–S80. Korhonen, H., Marnila, P., and Gill, H. S. (2000b). Bovine milk antibodies for health. Br. J. Nutr. 84, S135–S146. Kostan, K. M., and Snowdon, C. T. (2002). Attachment and social preferences in cooperatively‐reared cotton‐top tamarins. Am. J. Primatol. 57, 131–139. Kulski, J. K., and Hartmann, P. E. (1981). Changes in human milk composition during the initiation of lactation. Aust. J. Exp. Biol. Med. Sci. 59, 101–114. Lagercrantz, H. (1996). Stress, arousal, and gene activation at birth. News Physiol. Sci. 11, 214–218. Lagercrantz, H., and Slotkin, T. A. (1986). The ‘‘stress’’ of being born. Sci. Am. 254, 100–107. Larson, M. A., and Stein, B. E. (1984). The use of tactile and olfactory cues in neonatal orientation and localization of the nipple. Dev. Psychobiol. 17, 423–436. Lascelles, A. K. (1976). Structure, development, and evolution of the mammary gland. In ‘‘Veterinary Physiology’’ (J. W. Phillis, Ed.), pp. 763–782. Wright‐Scientechnica, Bristol. Lawrence, R. A. (1985). ‘‘Breastfeeding. A Guide for Medical Profession.’’ The C. V. Mosby Company, St. Louis. Leon, M., Coopersmith, R., Lee, S., Sullivan, R. M., Wilson, D. A., and Woo, C. C. (1987). Neural and behavioral plasticity induced by early olfactory learning. In ‘‘Perinatal Development: A Psychobiological Perspective’’ (N. A. Krasnegor, E. M. Blass, M. A. Hofer, and W. P. Smotherman, Eds.), pp. 145–167. Academic Press, Orlando. Leszek, J., Inglot, A. D., Janusz, M., Lisowski, J., Krukowska, K., and Georgiades, J. A. (1999). Colostrinin: A proline‐rich polypeptide complex isolated from ovine colostrum for treatment of Alzjeimer desease. A double blind, placebo‐controlled study. Arch. Immunol. Ther. Exp. 47, 377–385.

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

53

Levieux, D. (1982). Transmission de l’immunite´ passive colostrale: Le point des connaissances. In ‘‘Physiologie et Pathologie Pe´rinatales chez les Animaux de Ferme’’ (R. Jarrige, Ed.), pp. 345–369. Institut National de la Recherche Agronomique, Paris. Lincoln, D. W., Hill, A., and Wakerley, J. B. (1973). The milk‐ejection reflex of the rat: An intermittent function not abolished by surgical levels of anaesthesia. J. Endocrinol. 57, 459–476. Lincoln, D. W., Hentzen, K., Hin, T., Van der Scoot, P., Clarke, G. D., and Summerlee, A. J. S. (1980). Sleep: A prerequisite for milk‐ejection in the rat. Exp. Brain Res. 38, 151–162. Long, C. A. (1969). The origin and evolution of the mammary gland. Bioscience 19, 519–523. Lo¨nnerdal, B. O., and Atkinson, S. (1995). Nitrogenous components of milk. Human milk. In ‘‘Handbook of Milk Composition’’ (R. G. Jensen, Ed.), pp. 351–368. Academic Press, San Diego. Lorenz, D. N., Poppe, C. J., Quail, C., Seipel, K., Stordeur, S. A., and Johnson, E. (1998). Filling the gut activates paradoxical sleep in suckling rats. Dev. Psychobiol. 32, 1–12. Martin, R. D. (1966). Tree shrews: Unique reproduction mechanism of systemic importance. Science 151, 1402–1404. Mason, W. A., and Mendoza, S. P. (1998). Generic aspects of primate attachments: Parents, offspring and mates. Psychoneuroendocrinology 23, 765–778. Meier, G. W. (1964). Behavior in infant monkeys: Differences attributable to mode of birth. Science 143, 968–970. Meisel, H. (2004). Multifunctional peptides encrypted in milk proteins. Biofactors 21, 55–61. Meisel, H., and Fitzgerald, R. J. (2000). Opioid peptides encrypted in intact milk protein sequences. Br. J. Nutr. 84, S27–S31. Mellor, D. J. (1993). Some aspects of perinatal maturation and adaptation. Equine Vet. J. Suppl. 14, 17–22. Mendoza, S. P., and Mason, W. A. (1986). Parental division of labour and differenciation of attachments in monogamous primates (Callicebus moloch). Anim. Behav. 34, 1336–1347. Mennella, J. A., Jagnow, C. P., and Beauchamp, G. K. (2001). Prenatal and postnatal flavor learning by human infants. Pediatrics 107, E88. Morley, J. E., Flood, J., and Silver, A. J. (1992). Effects of peripheral hormones on memory and ingestive behaviors. Psychoneuroendocrinology 17, 391–399. Napolitano, F., Annicchiarico, G., Caroprese, M., De Rosa, G., Taibi, L., and Sevi, A. (2003). Lambs prevented from suckling their mothers display behavioral, immune and endocrine disturbances. Physiol. Behav. 78, 81–89. Nelson, E. E., and Alberts, J. R. (2002). Gastric saline infusion reduces ultrasonic vocalizations and brown fat activity in suckling rat pups. Dev. Psychobiol. 40, 160–167. Nelson, E. E., and Panksepp, J. (1998). Brain substrates of infant‐mother attachment: Contributions of opioids, oxytocin, and norepinephrine. Neurosci. Biobehav. Rev. 22, 437–452. Neve, H. A., Paisley, A. C., and Summerlee, A. J. S. (1982). Arousal: A prerequisite for suckling in the rabbit? Physiol. Behav. 28, 151–162. Newburg, D. S., and Neubauer, S. H. (1995). Carbohydrates in milks: Analysis, Quantities, and Significance. In ‘‘Handbook of Milk Composition’’ (R. G. Jensen, Ed.), pp. 273–349. Academic Press, San Diego. Nicholas, K. R., and Hartmann, P. E. (1991). Milk secretion in the rat: Progressive changes in milk composition during lactation and weaning and the effect of diet. Comp. Biochem. Physiol. A 98, 535–542. Noirot, E., and Alegria, J. (1983). Neonate orientation towards human voice differs with type of feeding. Behav. Proc. 8, 65–71. Nowak, R. (1990). Mother and sibling discrimination at a distance by three‐ to seven‐day‐old lambs. Dev. Psychobiol. 23, 285–295.

54

RAYMOND NOWAK

Nowak, R. (1991). Senses involved in discrimination of Merino ewes at close contact and from a distance by their newborn lambs. Anim. Behav. 42, 357–366. Nowak, R. (1994). Mother location by newborn lambs in repetitive testing: Influence of first successful reunion. Appl. Anim. Behav. Sci. 41, 75–86. Nowak, R., and Lindsay, D. R. (1990). Effect of genotype and litter size on discrimination of mothers by their twelve‐hour‐old lambs. Behaviour 115, 1–13. Nowak, R., Poindron, P., and Putu, I. G. (1989). Development of mother discrimination by single and multiple newborn lambs. Dev. Psychobiol. 22, 833–845. Nowak, R., Goursaud, A. P., Levy, F., Orgeur, P., Schaal, B., Belzung, C., Picard, M., Meunier‐ Salaun, M. C., Alster, P., and Uvnas‐Moberg, K. (1997a). Cholecystokinin receptors mediate the development of a preference for the mother by newly born lambs. Behav. Neurosci. 111, 1375–1382. Nowak, R., Murphy, T. M., Lindsay, D. R., Alster, P., Andersson, R., and Uvnas‐Moberg, K. (1997b). Development of a preferential relationship with the mother by the newborn lamb: Importance of the sucking activity. Physiol. Behav. 62, 681–688. Nowak, R., Orgeur, P., Piketty, V., Alster, P., Andersson, R., and Uvnas‐Moberg, K. (1997c). Plasma cholecystokinin concentrations in 3‐day‐old lambs: Effect of the duration of fasting preceding a sucking bout. Reprod. Nutr. Dev. 37, 551–558. Nowak, R., Porter, R. H., Levy, F., Orgeur, P., and Schaal, B. (2000). Role of mother‐ young interactions in the survival of offspring in domestic mammals. Rev. Reprod. 5, 153–163. Nowak, R., Breton, G., and Mellot, E. (2001). CCK and development of mother preference in sheep: A neonatal time course study. Peptides 22, 1309–1316. Nowak, R., Goursaud, A. P., Orgeur, P., Le´vy, F., Boivin, X., and Uvnas‐Moberg, K. Development of mother preference by newborn lambs: Colostral factors and CCK. (Submitted for publication.) Oftedal, O. T., and Iverson, S. J. (1995). Phylogenetic variation in the gross composition of milks. In ‘‘Handbook of Milk Composition’’ (R. G. Jensen, Ed.), pp. 749–789. Academic Press, San Diego. Oftedal, O. T., Boness, D. J., and Tedman, R. A. (1987). The behavior, physiology and anatomy of lactation in the Pinnipedia. Curr. Mammal. 1, 175–245. Panksepp, J., Herman, B., Conner, R., Bishop, P., and Scott, J. P. (1978a). The biology of social attachments: Opiates alleviate separation distress. Biol. Psychiatry 13, 607–618. Panksepp, J., Vilberg, T., Bean, N. J., Coy, D. H., and Kastin, A. J. (1978b). Reduction of distress vocalization in chicks by opiate‐like peptides. Brain Res. Bull. 3, 663–667. Panksepp, J., Nelson, E., and Siviy, S. (1994). Brain opioids and mother‐infant social motivation. Acta. Paediatr. 397(Suppl.), 40–46. Patt, J. A. J. (1977). Factors affecting the duration of intestinal permeability. Biol. Rev. 52, 411–429. Peaker, M., and Neville, M. C. (1991). Hormones in milk: Chemical signals to the offspring? J. Endocrinol. 131, 1–3. Peaker, M., and Taylor, J. C. (1975). Milk secretion in the rabbit: Changes during lactation and the mechanism of ion transport. J. Physiol. (Lond.) 253, 527–545. Pedersen, C. A., and Blass, E. M. (1981). Olfactory control over suckling in albino rats. In ‘‘The Development of Perception: Psychobiological Perspectves’’ (J. R. Aslin, J. R. Alberts, and M. R. Petersen, Eds.), pp. 359–381. Academy Press, New York. Poindron, P. (1976). Mother‐young relationships in intact or anosmic ewes at the time of suckling. Biol. Behav. 2, 161–177.

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

55

Popik, P., Bobula, B., Janusz, M., Lisowski, J., and Vetulani, J. (1999). Colostrinin, a polypeptide isolated from early milk, facilitates learning and memory in rats. Pharmacol. Biochem. Behav. 64, 183–189. Porter, R. H., Winberg, J., and Varendi, H. (2005). Prenatal preparation for ealy postnatal olfactory learning. In ‘‘Prenatal Development of Postnatal Functions’’ (B. Hopkins and S. P. Johnson, Eds.), pp. 103–129. Praeger, Westport. Prentice, A. M., Paul, A. R., Black, A., Cole, T., and Whitehead, R. (1986). Cross‐cultural differences in lactational performance. In ‘‘Maternal Environmental Factors in Human Lactation’’ (M. Hamosh and A. S. Goldman, Eds.), pp. 13–44. Plenum Press, New York. Ramsay, D. T., Kent, J. C., Hartmann, R. A., and Hartmann, P. E. (2005). Anatomy of the lactating human breast redefined with ultrasound imaging. J. Anat. 206, 525–534. Rauprich, A. B., Hammon, H. M., and Blum, J. W. (2000a). Effects of feeding colostrum and a formula with nutrient contents as colostrum on metabolic and endocrine traits in neonatal calves. Biol. Neonate 78, 53–64. Rauprich, A. B., Hammon, H. M., and Blum, J. W. (2000b). Influence of feeding different amounts of first colostrum on metabolic, endocrine, and health status and on growth performance in neonatal calves. J. Anim. Sci. 78, 896–908. Raynaud, A. (1969). Mamelles. In ‘‘Traite´ de Zoologie. Anatomie‐Syste´matique‐Biologie. Mamelles, Appareil Ge´nital, Game´togene`se, Fe´condation, Gestation’’ (P. P. Grasse´, Ed.), Tome XVI, Fascicule VI, pp. 1–147. Masson, Paris. Read, L. C., Lord, A. P., Brantl, V., and Koch, G. (1990). Absorption of beta‐casomorphins from autoperfused lamb and piglet small intestine. Am. J. Physiol. 259, G443–G452. Renfree, M. B., Fletcher, T. P., Balnden, D. R., Lewis, P. R., Shaw, G., Gordon, K., Short, R. V., Parer‐Cook, E., and Parer, D. (1989). Physiological and behavioural events around time of birth in macropodid parsupials. In ‘‘Kangaroos, Wallabies and Rat Kangaroos’’ (P. Jarman, I. D. Hume, and G. Grigg, Eds.), pp. 323–337. Surrey Beatty and Sons, Sydney. Robinson, S. R., Wong, C. H., Robertson, S. S., Nathanielsz, P. W., and Smotherman, W. P. (1995). Behavioral responses of the chronically instrumented sheep fetus to chemosensory stimuli presented in utero. Behav. Neurosci. 109, 551–562. Robinson Baker, R. (1998). The breast. In ‘‘The John Hopkins Atlas of the Human Functional Anatomy’’ (G. D. Zuidema and L. Schlossberg, Eds.), 4th ed., pp. 102–109. John Hopkins University Press, London. Ronca, A. E., and Alberts, J. R. (1995). Simulated uterine contractions facilitate fetal and newborn respiratory behavior in rats. Physiol. Behav. 58, 1035–1041. Ronca, A. E., Abel, R. A., and Alberts, J. R. (1996). Perinatal stimulation and adaptation of the neonate. Acta Paediatr. 416(Suppl.), 8–15. Rosenblatt, J. S. (1976). Stages in the early behavioural development of altricial young of selected species of non‐primate mammals. In ‘‘Growing Points in Ethology’’ (P. Bateson and R. A. Hinde, Eds.), pp. 345–383. Cambridge University Press, Cambridge. Rosenblatt, J. S. (1983). Olfaction mediates developmental transition in the altricial newborn of selected species of mammals. Dev. Psychobiol. 16, 347–375. Rosenblatt, J. S., and Lehrman, D. S. (1963). Maternal behavior of the laboratory rat. In ‘‘Maternal Behavior in Mammals’’ (H. L. Rheingold, Ed.), pp. 8–57. John Wiley and Sons, Inc., New York. Roth, L. L., and Rosenblatt, J. S. (1968). Self‐licking and mammary development during pregnancy in the rat. J. Endocrinol. 42, 263–378. Roth, T. L., Wilson, D. A., and Sullivan, R. M. (2004). Neurobiological development of infant learning and memory: Implications for infant attachment. Adv. Study Behav. 34, 103–133.

56

RAYMOND NOWAK

Russo, J., and Russo, I. H. (2004). Development of the human breast. Maturitas 49, 2–15. Saint, L., Smith, M., and Hartmann, P. E. (1984). The yield and nutrient content of colostrum and milk of women from giving birth to 1 month post‐partum. Br. J. Nutr. 52, 87–95. Scafidi, F. A., Field, T. M., Schanberg, S. M., Bauer, C. R., Tucci, K., Roberts, J., Morrow, C., and Kuhn, C. M. (1990). Massage stimulates growth in preterm infants: A replication. Infant Behav. Dev. 13, 167–188. Schaal, B. (2005). From amnion to colostrum: Odor bridging in early development transistions. In ‘‘Prenatal Development of Postnatal Functions’’ (B. Hopkins and S. P. Johnson, Eds.), pp. 51–102. Praeger Publisher, Westport. Schaal, B., and Porter, R. H. (1991). ‘‘Microsmatic humans’’ revisited: The generation and perception of chemical signals. Adv. Study Behav. 20, 135–199. Schaal, B., Coureaud, G., Langlois, D., Ginies, C., Semon, E., and Perrier, G. (2003). Chemical and behavioural characterization of the rabbit mammary pheromone. Nature 424, 68–72. Sevi, A., Napolitano, F., Casamassina, D., Annicchiarico, G., Quarantelli, T., and De Paola, R. (1999). Effect of gradual transition from maternal to reconstituted milk on behavioural, endocrine and immune response of lambs. Appl. Anim. Behav. Sci. 64, 249–259. Shair, H., Brake, S., and Hofer, M. (1984). Suckling in the rat: Evidence for patterned behavior during sleep. Behav. Neurosci. 98, 366–370. Sharman, G., and Calaby, J. H. (1964). Reproductive behaviour in the red kangaroo, Megaleia rufa, in captivity. CSIRO Wildlife Res. 85, 58–85. Shayit, M., Nowak, R., Keller, M., and Weller, A (2003). Establishment of a preference by the newborn lamb for its mother: The role of opioids. Behav. Neurosci. 117, 446–454. Shayit, M., Goursaud, A. P., Weller, A., Schaal, B., Uvnas‐Moberg, K., and Nowak, R. Development of mother preference by newborn lambs: Non nutritive sucking, CCK, and opioids. (Submitted for publication.) Shide, D. J., and Blass, E. M. (1991). Opioid mediation of odor preferences induced by sugar and fat in 6‐day‐old rats. Physiol. Behav. 50, 961–966. Shillito, E. E., and Alexander, G. (1975). Mutual recognition amongst ewes and lambs of four breeds of sheep (Ovis aries). Appl. Anim. Ethol. 1, 151–165. Shubber, A. H., and Doxey, D. L. (1979). Colostrum production by ewes and amount ingested by lambs. Res. Vet. Sci. 27, 280–282. Singh, P. J., and Tobach, E. (1975). Olfactory bulbectomy and nursing behavior in rat pups (Wistar DAB). Dev. Psychobiol. 8, 151–164. Singh, P. J., Tucker, A. M., and Hofer, M. A. (1976). Effect of nasal ZnSO4 irrigation and olfactory bulbectomy on rat pups. Physiol. Behav. 17, 373–382. Slee, J., and Springbett, A. (1986). Early post‐natal behaviour in lambs of ten breeds. Appl. Anim. Behav. Sci. 15, 229–240. Stanley, W. C., Cornwell, C. A., Poggiani, C., and Trattner, A. (1963). Conditioning in the neonatal puppy. J. Comp. Physiol. Psychol. 56, 211–214. Stanley, W. C., Bacon, W. E., and Fehr, C. (1970). Discriminated instrumental learning in neonatal dogs. J. Comp. Physiol. Psychol. 70, 335–343. Stephens, D. B., and Linzell, J. L. (1974). The development of sucking behaviour in the newborn goat. Anim. Behav. 22, 628–633. Stern, J. M. (1989). Maternal behavior: Sensory, hormonal, and neural determinants. In ‘‘Psychoendocrinology’’ (F. R. Brush and S. Levine, Eds.), pp. 105–196. Academic Press, New York. Stern, J. M. (1996). Somatosensation and maternal care in Norway rats. Adv. Study Behav. 25, 243–294. Strbak, V. (1992). Hormones and bioactive substances in milk: A rudiment or a message? Mater. Med. Pol. 24, 209–214.

SUCKLING, MILK, AND DEVELOPMENT OF PREFERENCES BY NEONATES

57

Sullivan, R. M., and Hall, W. G. (1988). Reinforcers in infancy: Classical conditioning using stroking or intra‐oral infusions of milk as UCS. Dev. Psychobiol. 21, 215–223. Sullivan, R. M., Taborsky‐Barba, S., Mendoza, R., Itano, A., Leon, M., Cotman, C. W., Payne, T. F., and Lott, I. (1991). Olfactory classical conditioning in neonates. Pediatrics 87, 511–518. Swaisgood, H. E. (1995). Protein and amino acid composition of bovine milk. In ‘‘Handbook of Milk Composition’’ (R. G. Jensen, Ed.), pp. 464–468. Academic Press, San Diego. Taira, T., Hilakivi, L. A., Aalto, J., and Hilakivi, I. (1990). Effect of beta‐casomorphin on neonatal sleep in rats. Peptides 11, 1–4. Teicher, M. H., and Blass, E. M. (1976). Suckling in newborn rats: Eliminated by nipple lavage, reinstated by pup saliva. Science 193, 422–425. Thapa, B. R. (2005). Health factors in colostrum. Indian J. Pediatr. 72, 579–581. Thoman, E. B., Leiderman, P. H., and Olson, J. (1972). Neonate‐mother interaction during breast‐feeding. Dev. Psychol. 6, 110–118. Tizard, I. (2001). The protective properties of milk and colostrum in non‐human species. Adv. Nutr. Res. 10, 139–166. Trevathan, W. R. (1987). ‘‘Human Birth: An Evolutionary Perspective.’’ Aldine de Gruyter, New York. Tyran, B., and Lisowski, J. (1979). Immunoglobulins of colostrum. V. Comparative studies of ovine and bovine serum and colostral IgM. Arch. Immunol. Ther. Exp. 27, 487–498. Umbach, M., Teschemacher, H., Praetorious, K., Hirschha¨user, R., and Bostedt, H. (1985). Demonstration of a beta‐casomorphin immunoreactive material in the plasma of newborn calves after milk intake. Regul. Pept. 12, 223–230. Uvna¨s‐Moberg, K. (1996). Neuroendocrinology of the mother‐young interaction. Trends Endocrinol. Metabol. 7, 126–131. Uvna¨s‐Moberg, K., Widstrom, A. M., Marchini, G., and Winberg, J. (1987). Release of GI hormones in mother and infant by sensory stimulation. Acta Paediatr. Scand. 76, 851–860. Val‐Laillet, D. (2004). Etablissement d’un lien pre´fe´rentiel avec la me`re chez l’agneau nouveau‐ne´. Facteurs renforc¸ateurs ingestifs et environnementaux. Me´canismes comportementaux et neurobiologiques The`se, Universite´ Franc¸ois Rabelais. Val‐Laillet, D., Meurisse, M., Tillet, Y., and Nowak, R. (2004a). Differential c‐Fos expression in the newborn lamb nucleus tractus solitarius and area postrema following ingestion of colostrum or saline. Brain Res. 1028, 203–212. Val‐Laillet, D., Simon, M., and Nowak, R. (2004b). A full belly and colostrum: Two major determinants of filial love. Dev. Psychobiol. 45, 163–173. Val‐Laillet, D., Giraud, S., Tallet, C., Boivin, X., and Nowak, R. (2006). Non nutritive sucking: One of the major determinants of filial love. Dev. Psychobiol. 48, 220–232. Varendi, H., and Porter, R. H. (2001). Breast odour as the only maternal stimulus elicits crawling towards the odour source. Acta Paediatr. 90, 372–375. Varendi, H., Porter, R. H., and Winberg, J. (1994). Does the newborn baby find the nipple by smell? The Lancet 344, 989–990. Varendi, H., Porter, R. H., and Winberg, J. (1996). Attractiveness of amniotic fluid odor: Evidence of prenatal olfactory learning? Acta Paediatr. 85, 1223–1227. Varendi, H., Porter, R. H., and Windberg, J. (1997). Natural odor preferences of newborn change over time. Acta Paediatr. 86, 985–990. Varendi, H., Porter, R. H., and Winberg, J. (2002). The effect of labor on olfactory exposure learning within the first postnatal hour. Behav. Neurosci. 116, 206–211. Vince, M. A. (1984). Teat seeking behaviour in newly born lambs: Possible effects of maternal skin temperature. Anim. Behav. 32, 249–254.

58

RAYMOND NOWAK

Vince, M. A. (1993). Newborn lambs and their dams: The interaction that leads to sucking. Adv. Study Behav. 22, 239–268. Vince, M. A., and Stanier, M. W. (1991). The effect of food intake on young Soay and Clun Forest lambs’ response to touch on the face. Appl. Anim. Behav. Sci. 30, 87–96. Vince, M. A., and Ward, T. M. (1984). The responsiveness of newly born Clun Forest lambs to odour sources in the ewe. Behaviour 89, 117–127. Vince, M. A., Ward, T. M., and Reader, M. (1984). The part played by tactile stimulation in newly born lambs. Anim. Behav. 32, 1179–1184. Vince, M. A., Lynch, J. J., Mottershead, B. E., Green, G. C., and Elwin, R. L. (1987). Interactions between normal ewes and newly born lambs deprived of visual, olfactory and tactile sensory information. Appl. Anim. Behav. Sci. 19, 119–136. Voloschin, L. M., and Tramezanni, N. (1979). Milk ejection reflex linked to slow‐wave sleep in nursing rats. Endocrinology 105, 1202–1207. Weller, A., and Blass, E. M. (1988a). Behavioral evidence for cholecystokinin‐opiate interactions in neonatal rats. Am. J. Physiol. 255, R901–R907. Weller, A., and Blass, E. M. (1988b). Cholecystokinin‐induced conditioned odor‐preference is blocked by the selective antagonist L‐364,718. Proc. Am. Soc. Neurosci. 14, 199. Weller, A., and Blass, E. M. (1990). Cholecystokinin conditioning in rats: Ontogenetic determinants. Behav. Neurosci. 104, 199–206. Weller, A., and Feldman, R. (2003). Emotion regulation and touch in infants: The role of cholecystokinin and opioids. Peptides 24, 779–788. Weller, A., Gispan, I. H., Ritter, R. C., Brenner, L., and Smith, G. P. (2001). Release of endogenous cholecystokinin in response to gastric preloads in rats on postnatal days 9–12. Physiol. Behav. 72, 1–4. Widstrom, A. M., Ransjo‐Arvidson, A. B., Christensson, K., Matthiesen, A. S., Winberg, J., and Uvnas‐Moberg, K. (1987). Gastric suction in healthy newborn infants. Effects on circulation and developing feeding behaviour. Acta Paediatr. Scand. 76, 566–572. Widstrom, A. M., Marchini, G., Matthiesen, A. S., Werner, S., Winberg, J., and Uvnas‐ Moberg, K. (1988). Nonnutritive sucking in tube‐fed preterm infants: Effects on gastric motility and gastric contents of somatostatin. J. Pediatr. Gastr. Nutr. 7, 517–523. Williams, A. P., Bishop, D. R., Cockburn, J. E., and Scott, K. J. (1976). Composition of ewe’s milk. J. Dairy Res. 43, 325–329. Wilson, D. A., and Sullivan, R. M. (1994). Neurobiology of associative learning in the neonate: Early olfactory learning. Behav. Neural Biol. 61, 1–18.

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 36

A Neuroethological Approach to Song Behavior and Perception in European Starlings: Interrelationships Among Testosterone, Neuroanatomy, Immediate Early Gene Expression, and Immune Function Gregory F. Ball,* Keith W. Sockman,{ Deborah L. Duffy,{ and Timothy Q. Gentner} *department of psychological and brain sciences johns hopkins university, baltimore, maryland 21218, usa { department of biology, university of north carolina chapel hill, north carolina 27599, usa { center for the interaction of animals and society school of veterinary medicine, university of pennsylvania philadelphia, pennsylvania 19104, usa } department of psychology, university of california, san diego la jolla, california 92093, usa

I. INTRODUCTION: SONG, EUROPEAN STARLINGS, NEUROETHOLOGICAL APPROACH

AND THE

It is not unusual to hear commentators on research trends in animal behavior lament current intellectual divisions in the field. During the past 25 years it can be argued that investigators in this field can be divided into those who focus on the adaptive significance and evolution of behavior (sometimes called ultimate causation) and those who focus on the study of sensory and physiological mechanisms controlling the production and ontogeny of behavior (sometimes called proximate causation; Dewsbury, 1992, 1999; Sherman, 1988). The concern of course is that the field is splitting into operationally specialized camps with little interaction between them. However, a more encouraging development that is also emerging is the cross‐fertilization between these complementary approaches and the resultant new integrative views of the causes of behavior (Drickamer, 1998). The goal of this chapter is to review studies on the 0065-3454/06 $35.00 DOI: 10.1016/S0065-3454(06)36002-0

59

Copyright 2006, Elsevier Inc. All rights reserved.

60

GREGORY F. BALL ET AL.

interrelationships among endocrine state, brain mechanisms for song perception and production, and immunocompetence and relate these to the question of female choice based on male song in European starlings (Sturnus vulgaris). Our desire is to illustrate how by studying a single species from multiple viewpoints one can start to make connections among the major physiological communications systems, namely the nervous system, the endocrine system, and the immune system. This approach is needed in order to address how behavior is controlled and concomitantly how these physiological systems that regulate behavioral output have themselves been shaped by natural and sexual selection. We adopt an explicitly neuroethological perspective in this research program. What do we mean exactly by this label? Neuroethology is of course a variant of ethology. An ethological research program starts with a view that animals are best understood when studied in their natural context and it encompasses the four questions about animal behavior that Tinbergen observed from the core of ethology (Hinde, 1982; Tinbergen, 1963). These questions consist of two related to proximate causes, that is, immediate causation and development and two questions related to ultimate causes, that is, adaptive significance and evolution (Hinde, 1982; Tinbergen, 1963; but see also Dewsbury, 1999). A neuroethological approach stresses an investigation of neural and physiological mechanisms that might control naturally occurring behavior. This approach can be distinguished from more general neuroscience approaches because it sets as its goal the understanding of the causes of behavior when produced under natural conditions and embraces the study of a wide range of species (Camhi, 1984; Gentner and Ball, 2005). Instead of starting with the goal to identify a model system of a human disease process, one starts with the premise that understanding the causes of complex naturally occurring behavior is interesting by itself. Adopting a neuroethological approach while being aware of questions of evolutionary function is not without potential pitfalls. Bolhuis and MacPhail (Bolhuis, 2005; Bolhuis and MacPhail, 2001) in particular have argued that confounding ultimate and proximate causes can lead one to erroneous notions about behavioral mechanisms. In this chapter we will try to illustrate that knowledge about the adaptive significance of behavior can be an important aide in guiding neuroethological investigations. When considering a complex learned behavior, such as birdsong, it is challenging to decide what aspects of the stimulus are most salient to the birds and are therefore the ones that investigators should concentrate on in sensory and neural studies. As will be illustrated in this chapter, work on aspects of song important in mate choice in starlings has been very helpful in guiding our studies of sensory responses to song and even for an analysis of the neural correlates of song production.

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

61

Another goal of this chapter is to highlight the range of methods that can be applied to the study of behavior. Although many neuroethological studies rely primarily on electrophysiological methods to investigate the structure and function of the nervous system, we argue that the entire range of neuroscience methods can and should be marshaled with this approach. The study of song in a species, such as the European starling, is particularly amenable to such a multimethod analysis. European (or Common) starlings have a huge native distribution in Eurasia but they have also been introduced in many areas of the world including North America, South Africa, and New Zealand (Feare, 1984). Thus, many scientists worldwide have access to this species and over time a substantial number of questions have been investigated about the control of song in these birds based on the use of a wide range of methods. We will begin by considering our some basic facts about song behavior.

II. DESCRIPTION A. FUNDAMENTALS

OF

OF

EUROPEAN STARLING SONG

AND ITS

FUNCTION

STARLING SONG

Song in European starlings is long and complex and can include imitations of sounds from other species (Eens, 1997 for a review). It is often produced at a low amplitude, and field naturalists in the past have typically had the impression that it is rather unorganized with little overt structure (Feare, 1984). Detailed acoustic analyses in the 1980s and 1990s, however, revealed a clear underlying structure to starling song (Adret‐Hausberger and Jenkins, 1988; Chaiken et al., 1993; Eens et al., 1989, 1991a; Mountjoy and Lemon, 1995) albeit one that is harder to discern than that of other well‐studied songbirds. The nomenclature used to describe starling song in this paper follows the guidelines advocated by Eens (1997). Starling song is usually organized into long bouts that may be a minute or longer in duration. These bouts contain shorter phrases or motifs that can be repeated and are generally 0.5–1.0 sec in length. A complete song bout has four acoustically distinct sections (Fig. 1). The first section consists of relatively pure‐toned whistles. The second section includes complex ‘‘warble’’ motifs of low amplitude and heterospecific motifs if the individual has copied any. Motifs in the third section are characterized by the presence of rapid, biphonated, click trains or ‘‘rattles.’’ Finally, motifs in the last section are characterized by high‐frequency and high‐amplitude components, typically the loudest part of the song (Eens, 1997; Fig. 1). Song production is often accompanied by wing movements, the rattle motifs with wing flicks, and the

62 Fig. 1. Sonogram of a single song from one male starling showing the patterning of frequency spectrum power as a function of time. For clarity the sonogram is broken into separate rows. One motif in the last row is outlined in the square. Divisions between whistle, warble, rattle, and high‐frequency motif types are denoted by the solid black bars in rows one, three, and four.

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

63

high‐frequency motifs at the end of the song by wing waving (Bohner and Veit, 1993). B. FUNCTION

OF

SONG

IN

EUROPEAN STARLINGS

Several lines of evidence support the contention that male starling song is important for female mate choice. Male starlings paired with a female will increase their rate of singing just prior to copulation (Eens and Pinxten, 1990) and stop singing only after their mate’s clutch is complete, whereas unpaired males continue singing throughout the breeding season (Kluyver, 1935 in Eens, 1997). In fact, male song output is closely related to different stages in the female breeding cycle. There is a rapid decrease in male singing activity after pairing (Eens et al., 1994; Hindmarsh, 1984), and then an increase 2–4 days prior to egg laying, after which song rates remain elevated until the end of the laying period before nearly ceasing altogether (Eens et al., 1994). Likewise, male song rate is negatively correlated with the date of clutch initiation (Mountjoy and Lemon, 1996; Wright and Cuthill, 1992). Moreover, the postpairing period of high song output coincides with the presumed fertile period for female starlings (Birkhead et al., 1987) and (albeit roughly) with the male’s most rigorous period of mate guarding (Pinxten et al., 1987). In the field, copulations between starlings are almost always preceded by bouts of male song (Eens and Pinxten, 1995; Eens et al., 1989; Mountjoy and Lemon, 1996) and when presented with a conspecific female, unmated captive male starlings sing many more song bouts, than when confronted with a conspecific male (Eens et al., 1993). The number of songs sung in the nest box also increases significantly with the introduction of a female, both for captive male starlings (Eens et al., 1993) and those in the field (Eens et al., 1991b). Average song bout length is positively correlated with the number of young per male (Eens et al., 1991). In the field, there is a significant negative correlation between repertoire size and the delay between male nestbox occupation and clutch initiation, even when nest‐site preference is controlled (Mountjoy and Lemon, 1996). Moreover, although males that mate earlier tend to be older, measures of male condition and gross morphology do not reveal affects on the timing of mating. Male body mass and tarsus lengths are not significantly correlated with initial pairing dates (Eens et al., 1991b). Similarly, male body mass, tarsus length, culmen length, wing length, and the length of the iridescence on a male’s hackle feathers are not significantly correlated with the timing of clutch initiation (Mountjoy and Lemon, 1996). Among males, both repertoire size and song bout length are directly correlated with age and mating success (Eens et al., 1991b). Finally, in the laboratory, female behavioral preferences can be directly controlled by varying the mean

64

GREGORY F. BALL ET AL.

length of male song bouts. That is, females will spend more time listening to long male song bouts than to shorter ones, and will preferentially track the position of the longer bouts coming from different locations (Gentner and Hulse, 2000a).

III. SONG CONTROL CIRCUIT AND THE NEUROENDOCRINE CONTROL OF SONG A. NEURAL CONTROL

OF

SONG BEHAVIOR

European starlings and other members of the songbirds (suborder Passeres or Oscines) have evolved a suite of neural specializations in association with their sophisticated vocal abilities, which facilitate the learning, production, and perception of song (Brenowitz et al., 1997; Farries, 2004; Jarvis, 2004). The best studied of these specializations is the song control system, an interconnected circuit of telencephalic, diencephalic, mesencephalic, and myencephalic nuclei that regulate the learning and production of song. In this section we will provide a succinct review of the structure and function of the song system and related auditory pathways. Most of the work on the song system is based on studies of zebra finches (Taeniopygia guttata) and canaries (Serinus canaria). Although some species‐specific specializations may exist, starlings share much of the same cytoarchitecture and neurochemistry described in these other species (Ball, 1990; Ball et al., 1988; Bernard et al., 1993) consistent with common neuroanatomical principles of song system organization. The song control circuit can be divided into two main parts: the more caudal motor pathway and the more rostral anteriori forebrain pathway (AFP; Fig. 2). The primary motor pathway, in order of descending projection, is made up of the nucleus HVC (used as a proper name), the nucleus robustus arcopallialis (RA), the dorsomedial portion of the intercollicularis, medullary nuclei that modulate respiratory motor neurons (Wild, 1994, 2004), and the tracheosyringal portion of the hypoglossal nucleus (nXIIts) that controls muscles of the syrinx, the avian vocal organ (Nottebohm et al., 1976, 1982; Reinke and Wild, 1998; Wild, 1993a,b; see Reiner et al., 2004 for the current, recently revised nomenclature of the avian brain). HVC appears to be unique to songbirds (Ball, 1994; Brenowitz, 1997; Kroodsma and Konishi, 1991; Nottebohm, 1980). Immediate early gene (IEG) studies (Jarvis and Nottebohm, 1997; Kimpo and Doupe, 1997), lesion data (Nottebohm et al., 1976; Simpson and Vicario, 1990), and electrophysiological recordings (Yu and Margoliash,

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

65

HVC lMAN

Shelf NCM RA

CMM

L

Nif

Cup

Area X

DLM nXIIts

Ov

RAm/rVRG

ICo

Auditory inputs

Syrinx

Respiration

Fig. 2. Schematic representation of a sagittal view of the song control system of songbirds. It consists of at least two basic pathways. One pathway, essential for song production, involves a projection from nucleus HVC (initially misnamed the hyperstriatum ventrale, pars caudale and now known as the acronym only) to the nucleus robustus arcopallialis (RA) that in turn projects to both the nucleus intercollicularis (ICo) and the tracheosyringeal division of the nucleus of the XIIth cranial nerve (nXIIts). Efferent projections from motor neurons in this brainstem nucleus innervate the vocal production organ, the syrinx. ICo and RA also innervate medullary structures that coordinate song production with respiration. HVC also connects with RA through a more circuitous route. This anterior forebrain pathway consists of a projection from HVC to area X of the medial striatum that in turn projects to the medial portion of the dorsolateral nucleus of the anterior thalamus (DLM). DLM projects to the lateral portion of the nucleus magnocellularis of the anterior nidopallium (lMAN) that in turn projects to RA. In contrast to the more posterior pathway that is needed for song production, the anterior forebrain pathway is involved in song learning, maintenance, and various forms of sensory feedback on song production. Some of the auditory inputs to the song system are also illustrated. Nucleus ovoidalis (Ov) of the thalamus projects to telencephalic auditory areas such as field L (L) and the caudal and medial nidopallium (NCM). These in turn project to other auditory areas adjacent and connected to the song system such as the caudal ventral hyperstriatum, Nif, and the shelf near HVC and the RA cup. See text for further details.

1996) consistently implicate the primary motor pathway (HVC!RA! ICo!nXIIts) in song production. Neurons from HVC also innervate the anterior forebrain pathway. HVC projects to area X of the medial striatum (homologue of the caudate/ putamen), which in turn projects to the medial dorsolateral thalamic

66

GREGORY F. BALL ET AL.

nucleus (DLM). DLM projects to the lateral magnocellular nucleus of the anterior nidopallium (lMAN), and lMAN projects (predominantly) to RA. This pathway is organized as follows HVC!X!DLM!lMAN!RA (Fig. 2; see Doupe et al., 2005 for a review). Thus there are two pathways from HVC to RA. The caudal pathway is essential for song production, and the more indirect AFP appears to function in song learning (see Bottjer and Johnson, 1997; Doupe et al., 2004 for reviews) and in the maintenance of stereotypic adult song (Benton et al., 1998). Lesions to nuclei within the AFP do not immediately effect adult song production (Bottjer et al., 1984; Scharff and Nottebohm, 1991; Sohrabji et al., 1990). Explicit functions of the AFP have been slower to yield to investigators than those of the primary motor pathway. One hypothesis is that the AFP carries (or computes) an error signal between the song that the bird is trying to produce and that which it actually does produce (Brainard, 2004). Blocking auditory feedback by deafening (Nordeen and Nordeen, 1992) or distorting it via delayed feedback (Leonardo and Konishi, 1999) results in a decline in the quality of song production. Moreover, the negative effects of deafening are largely reversed by subsequent lesions to lMAN (Brainard and Doupe, 2000), perhaps because the effects of a putative error correction signal provided by the AFP are blocked by the lesion. Consistent with this idea, song‐evoked microstimulation of lMAN disrupts ongoing song, again suggesting that the AFP mediates the ongoing song maintenance (Kao et al., 2005), perhaps in a context‐dependent fashion (Hessler and Doupe, 1999a,b; Jarvis et al., 1998). During ontogeny and in adulthood song is profoundly regulated by hearing. Thus, it is important to understand the anatomy of the auditory inputs to the song system. The basic plan of the passerine auditory system follows a general reptile–bird pattern of connection (Carr, 1992; Carr and Code, 2000; Ulinski and Margoliash, 1990). The auditory nerve projects to the cochlear nuclei magnocellularis and angularis. These nuclei project in turn to second order olivary nuclei, to the lemniscal nuclei, and contralaterally to the central nucleus of the nucleus mesencephalicus lateralis dorsalis (MLd), the avian analogue of the inferior colliculus. Efferent fibers from the central nucleus of the MLd target primarily the medial portion of the dorsal thalamus, the nucleus ovoidalis (Ov; Karten, 1968). The caudal medial portion of the avian telencephalon is composed of five cytoarchitectonic subregions—L1, L2a, L2b, L3, and L—called the field L complex (Fortune and Margoliash, 1992). The field L complex is the primary telencephalic target for auditory information arriving via several parallel pathways from the Ov complex in the thalamus. The subregions of field L are densely interconnected and project to the caudal and medial nidopallium (NCM) and reciprocally to the lateral portions of the caudal mesopallium

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

67

(CLM). The NCM and CLM share reciprocal connections with the caudal medial mesopallium (CMM; Fig. 2). In European starlings, neurons throughout the Ov and the auditory telencephalon show tonotopic organization (Bigalke‐Kunz et al., 1987; Leppelsack and Schwartzkopff, 1972; Rubsamen and Dorrscheidt, 1986). In starlings, roughly 11 different regions can be identified on the basis of the direction of the tonotopic gradient and tuning curve bandwidth (Capsius and Lepplesack, 1999; Hau¨sler, 1996) with similar patterns observed in zebra finches (Gehr et al., 2000). Neurons in L1 and L3 have lower response rates to tone bursts than those in L2 and show greater selectivity to species‐specific vocalizations (Bonke et al., 1979; Leppelsack and Voigt, 1976; Mu¨ller and Leppelsack, 1985; Theunissen and Doupe, 1998). This selectivity is borne out by the complexity of the spectrotemporal receptive fields (STRFs) for many neurons within field L. More reliable estimates of the STRF are derived from responses to conspecific vocalizations than tone pips (Scha¨fer et al., 1992; Theunissen et al., 2000). This general pattern of increasing response selectivity from field L2 to the higher order areas continues into NCM and CM (Grace et al., 2003; Mu¨ller and Leppelsack, 1985; Sen et al., 2001), suggesting that these regions are involved in the extraction of complex features. Early data from white‐crowned sparrows are consistent with this in showing a small subset of neurons in the NCM that are selective for specific directions of frequency modulation (FM) in a common trill element of conspecific song (Leppelsack, 1983). Single neuron recordings from operantly trained starlings (Gentner and Margoliash, 2003) implicate CMM in the representation of complex acoustic features in behaviorally relevant conspecific songs. Neurons in NCM are broadly responsive to conspecific stimuli and respond to the repeated presentation of conspecific song in a stimulus‐ specific manner (Chew et al., 1995; Stripling et al., 1997). The repeated presentation of a single conspecific song elicits a rapid modulation in the initial firing rate of NCM neurons (Stripling et al., 1997). If the same song is repeated on the order of 200 times, this initial modulation of the firing rate is no longer observed when that same song is presented on subsequent trials. This is true even though the initial response modulation can still be observed for other conspecific songs (Chew et al., 1995; Stripling et al., 1997). These stimulus‐specific changes in the response properties of NCM neurons have led to the hypothesis that NCM may contribute to individual vocal recognition (Chew et al., 1996). Consistent with this idea, many neurons in NCM (and CM) show a rapid upregulation of the IEG zenk in response to the presentation of conspecific songs (Mello et al., 1992) that is tuned to the acoustics of particular conspecific song syllables (Ribeiro et al., 1998). The genomic response also habituates to the repeated presentation

68

GREGORY F. BALL ET AL.

of the same conspecific song (Mello et al., 1995) and is elevated during specific components of the vocal‐recognition task described earlier in starlings (Gentner et al., 2004). The mammalian homologue to zenk is required for expression of late long‐term potentiation (LTP) and long‐term memories in mice (Jones et al., 2001). These results suggest that zenk (the avian homologue of and an acronym for zif‐268, egr‐1, NGFI‐A, and Krox‐24) expression in NCM and cHV may be related to learning about conspecific songs and implicates these structures in concomitant processes. B. HORMONAL CONTROL

OF

SONG BEHAVIOR

It is well known that there is a link of some sort between sex steroid hormone secretion and song behavior (see Ball, 1999; Harding, 2004; Schlinger, 1997 for reviews). This link was first posited based on field studies correlating seasonal changes in gonadal size and other aspects of endocrine physiology with changes in song behavior. These descriptive studies have been reviewed in some detail (Ball, 1999; Catchpole and Slater, 1995; Tramontin and Brenowitz, 2000). Therefore, only major features of these findings will be presented here in order to place the hormonal control of song in starlings in a broader context. Starlings are highly seasonal breeders (Ball and Bentley, 2000). Many north temperate zone male song birds sing at high rates in the spring as compared to other seasons (Cox, 1944; Slagsvold, 1977; see Catchpole and Slater, 1995 for a review). In these species seasonal differences in male song are correlated with dramatic seasonal increases and decreases in aspects of reproductive physiology such as gonadal size and plasma hormone concentrations (Dawson et al., 2001; Wingfield and Farner, 1993). However, among these temperate zone birds, there is interspecific variability in the degree to which maximal rates of singing are observed outside the breeding period. For example, robins (Erithacus rubecula) living in northern Europe sing at relatively high rates throughout the year (Hoelzel, 1986) only pausing in July (Cox, 1944) while most other songbirds living in the same region do not. Starling song is not limited to their breeding season (Eens, 1997); for example, song can be heard quite commonly in the fall. Although it has not been quantified properly, song seems to be the least common in the late summer and early fall when the birds become photorefractory and molt (Feare, 1984). Birds, such as starlings, that sing outside the breeding season exhibit seasonal cycles in gonad size and endocrine secretions that are similar to other temperate zone species. Studies of select species that exhibit territorial song production in the autumn, such as song sparrows (Melospiza melodia) in the western United States, mockingbirds, and the European robin clearly suggest that song behavior in the fall can be elicited

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

69

by the appropriate stimulus in the absence of substantial concentrations of testosterone (T) of gonadal origin (Logan and Wingfield, 1990; Schwabl and Kriner, 1991; Wingfield, 1994; Wingfield and Hahn, 1994). In the case of the song sparrow, there is evidence that this autumnal singing involves estrogen acting in the brain even though the gonad is inactive (Soma et al., 2000). The source of this neuroactive estrogen in this case does not appear to be from T of gonadal origin that is then locally metabolized (Soma et al., 2000). It could be derived from the neural aromatization of a substrate (potentially dehydroepiandrosterone, DHEA) produced by the adrenals or synthesized de novo from cholesterol in the brain (i.e., a neurosteroid, Soma and Wingfield, 2001; Soma et al., 2002, 2004). Overall, these data indicate that there is not necessarily a tight correlation between endocrine activity and song rate in all temperate zone species. However, seasonal changes in reproductive physiology in these species may relate to changes in other aspects of song such as repertoire size or stereotypy. For example, seasonal changes in song repertoires have been observed in European starlings (Eens, 1997) and canaries (Nottebohm et al., 1986) where the number of song types and other measures of song complexity may change. Additionally, seasonal changes in other measures of song, such as stereotypy, have been described in white‐crowned sparrows (Zonotrichia leucophrys, Smith et al., 1995) and song sparrows (Smith et al., 1997). A careful consideration of behavioral data from temperate zone songbirds suggests that although song output is positively correlated with various measures of reproductive physiology, including hormone concentration in the plasma, there is not necessarily a strong causal relationship between the two as is the case for sex steroids and certain reproductive behaviors such as lordosis in rats (Pfaff et al., 1994), the bow coo display in male ring doves (Lehrman, 1965), or male‐typical copulatory behaviors in Japanese quail (Balthazart et al., 2004). Experimental studies on the effects of exogenous hormone administration or castration with hormone replacement on song have been performed on a relatively small number of species but these studies confirm this view of the relationship between steroids and song behavior. Administering exogenous T can clearly increase song rate (Hunt et al., 1997; Nowicki and Ball, 1989). Several independent studies of zebra finches have shown that castration greatly reduces but does not eliminate male‐typical song (Arnold, 1975; Harding et al., 1983), whereas in red‐winged blackbirds castration was reported to eliminate adult song production (Harding et al., 1988). In the case of song sparrows in the western United States, castrated males were able to maintain fall territories and sang at high rates in response to territorial challenge in a manner that was indistinguishable from intact controls (Wingfield, 1994). The hormonal

70

GREGORY F. BALL ET AL.

control of song behavior therefore appears to be a clear case of a hormone‐ enhanced, rather than hormone‐dependent behavior. Species‐typical stimulus factors (the presence of a conspecific male and/or female as well as a nest site or a favorable environment) promote song production in males. The presence of gonadal steroids in the plasma can increase the probability and intensity of these behavioral responses to the appropriate stimulus but this presence is not essential for behavioral activation (Wingfield, 1994). One should therefore not be surprised by reports of substantial song production being observed in association with low steroid hormone concentrations in some cases. The stimulus factors releasing song can be so strong in some cases that high gonadal steroid concentrations are not necessary for song production to be observed. In European starlings, T effects on song appear to be limited to certain social contexts. For example, castrated European starlings continue to express high basal rates of singing but fail to exhibit an increase in singing rate when presented with a female (Pinxten et al., 2002). A female‐induced increase in singing rate was, however, observed in castrates treated with T (Pinxten et al., 2002). Breeding season song in starlings usually declines after mating. If male starlings are treated with T during the incubation period, there is a robust song rate increase while a similar treatment during the nestling‐feeding period has little effect (De Ridder et al., 2000). The authors interpret these findings as follows. In the population they studied in Belgium, during the incubation period, there are still a large number of reproductively active females for the males to direct their song at, while by the time the nestling feeding stage starts there are few receptive females available (De Ridder et al., 2000). Thus, T was only effective in inducing an increase in song in breeding starlings when females available for mating were present. Finally, comparing the effects of a female on male song rate in starlings in the spring when T concentrations are high and the fall when they are low reveal an enhancing effect of a female only in the spring (Riters et al., 2000). These data are all consistent with the notion that in starlings T is effective in enhancing song produced in response to the presence of a receptive female. Although T does not appear to be necessary for the initiation of song production in all cases, it does appear to influence aspects of song quality such as stereotypy. For example, castration prevents the onset of crystallized (i.e., stereotyped) adult song in 1‐year‐old song and swamp sparrows singing for the first time in the spring (Marler et al., 1988). On receiving T, the song rapidly crystallizes (Marler et al., 1988). As mentioned previously, in both white‐crowned sparrows and song sparrows, fall song in the presence of low concentrations of T is less stereotypic than spring song

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

71

produced in the presence of high concentrations of T (Smith et al., 1995, 1997). Overall the data indicate that song can be produced (though at a low rate) in the presence of low concentrations of T, but that the stereotypic quality of the song is also regulated by the presence of T. Finally, some steroid hormone replacement studies indicate that both androgenic and estrogenic metabolites of T are needed to fully restore high rates of singing (Harding et al., 1983, 1988). In zebra finches it has been suggested that estrogenic metabolites selectively promote female‐directed song (Walters et al., 1991). Similarly, in canaries there are data suggesting that estradiol selectively activates syllables that are particularly attractive to females while other aspects of song are activated by androgenic metabolites (Fusani and Gahr, 2005). These studies indicate that there may be selective actions of the two primary metabolites of T on song behavior.

IV. PERCEPTION A. BEHAVIORAL EXPERIMENTS

ON

OF

SONG

IN

STARLINGS

SONG RECOGNITION

As noted previously, one can consider starling song as a sequence of phrases or motifs, where each motif is an acoustically complex event. The number of unique motifs that a male starling can sing (i.e., his repertoire size) can be quite large, and consequently different song bouts from the same male are not necessarily composed of the same set of motifs. This broad acoustical variation in their song provides several potential cues that starlings might use when learning to recognize the songs of an individual conspecific and while maintaining that recognition over time. One straightforward recognition mechanism is the association of specific motifs with specific singers. Although some sharing of motifs does occur among captive males (Hausberger, 1997; Hausberger and Cousillas, 1995), the motif repertoires of different males living in the wild are generally unique (Chaiken et al., 1993; Eens et al., 1989). Thus, learning which males sing which motifs can provide discriminative cues for song classification. As shown in Fig. 3, data from operant studies in starlings indicate that song recognition is based at the level of the motif. Starlings trained by operant conditioning procedures recognize individual conspecifics by one set of songs and can readily generalize correct recognition to novel songs from the same singers (Gentner and Hulse, 1998). However, recognition falls to chance when these novel song bouts have no motifs in common with the training songs (Gentner and Hulse, 2000b). Likewise, starlings trained to discriminate among pairs of motifs will reverse the discrimination when

72

GREGORY F. BALL ET AL.

0.8

0.9 0.8 0.7 Upper boundary for chance

0.6

Proportion of male A responses

Proportion of correct responses

1

0.7 0.6 0.5 0.4 0.3 0.2

0.5 Baseline

Novel motifs

9:1 8:2 7:3 6:4 5:5 4:6 3:7 2:8 1:9 Ratio of familiar motifs (male A:male B)

Fig. 3. Song recognition behavior in starlings based on an operant task. The bar graph on the left illustrates the proportion of correct responses made to familiar (baseline) and novel motifs. In the left panel data are shown consisting of response to chimeric songs composed of familiar motifs from two different singers.

transferred to the same motif sung by the opposite individual and perform at chance when transferred to novel motifs sung by the training singers. This failure to generalize correct recognition to songs composed of novel motifs, or to single novel motifs, is inconsistent with the use of particular voice characteristics for vocal recognition. Instead, the data suggest that starlings learn to recognize the songs of individual conspecifics by attending to information contained at (or below) the level of the motif. They appear to associate distinct sets of motifs (or variant motif features) with individual singers. If starlings learn to recognize individuals by the sets of unique motifs that they sing, then once learned, it should be possible to control recognition systematically by varying the proportions of motifs in a given bout that come from two ‘‘vocally familiar’’ males. That is, recognition behavior ought to follow the proportional distribution of motifs from two vocally familiar males rather than the presence or absence of single diagnostic motifs from either male. The behavioral data confirm this prediction by showing that when starlings are compelled to classify conspecific songs, they do so by memorizing large numbers of unique song components (i.e., motifs) and then by organizing subsets of these motifs into separate classes (Gentner and Hulse, 2000b). As a cognitive recognition strategy, classifying songs according to their component (motif) structure represents a straightforward method of dealing with these complex acoustic signals.

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

73

Because individual starlings tend to possess unique motif repertoires, disjoint sets of motifs will generally correspond to individual identity. Therefore, attending to the motif structure captures a significant portion of the individual variation in the signal, albeit at the expense of a large memory. The behavioral data suggest several hypotheses regarding the neural mechanisms underlying the recognition of natural (i.e., high‐dimensional) acoustic events. First, the functionality of motifs as auditory objects in recognition behavior implies their explicit representation in the central nervous system. That is, the response functions of single neurons or of populations of neurons in appropriate forebrain auditory regions should reflect the segmentation of song at the level of the motif. Second, because recognition behavior requires the learned association between sets of motifs and singers, motif representations (or the representations of submotif features that correspond to unique motifs) should reflect the behavioral relevance of specific motifs. That is, there should be a bias for representations of familiar motifs. Third, the representational mechanisms and capacity (i.e., memory) of the system should permit the acquisition of very large numbers of acoustically complex, natural objects (motifs).

B. FEMALE SONG PREFERENCES

IN A

MATE‐CHOICE CONTEXT

In a mate‐choice context, female European starlings prefer male song organized into long bouts over male song organized into short bouts, even when the total duration of song does not differ between the long‐ and short‐bout exposure (Gentner and Hulse, 2000a). This preference for long‐ bout over short‐bout song is apparently independent of the length of the song itself, but instead is related to some other song component associated with song length, such as its motif repetition rate or stereotypy. Thus, females are able to parse differences between long‐ and short‐bout songs independently of the song length and probably relative to highly complex spectrotemporal components of the song associated with its bout length. Complex behavioral preferences such as these require a nervous system with variable sensitivities to these complex stimuli and capable of precipitating motor programs that give rise to a mate‐choice decision (Tinbergen, 1950). Nonetheless, despite the powerful selective forces likely driving mate‐choice preferences, we still know relatively little about the neural processing of cues that release such preference behavior (Wilczynski and Ryan, 1992).

74

GREGORY F. BALL ET AL.

V. PHYSIOLOGICAL RESPONSES

TO

SONG

IN

STARLINGS

A. IMMEDIATE EARLY GENE (IEG) EXPRESSION 1. Neural Induction of IEGs Developments in molecular neuroscience have provided methods that are useful in understanding the neural underpinnings of complex behavior. One such advance has been the identification and characterization of specific genes encoding transcription factors and other types of proteins which are rapidly induced in the brain in response to a variety of stimuli (Goelet et al., 1986; Hoffman et al., 1993; Morgan and Curran, 1989). These so‐called IEGs are by definition the earliest or first genomic response to an inducing stimulus and they therefore can be operationally defined as those that are induced in the presence of protein synthesis inhibitors because they do not require any other gene to be transcribed in order to be active (Clayton, 2000). The mapping of IEG expression in the brain associated with the occurrence of a behavior or in response to behavioral stimuli has been a particularly powerful method for mapping functional neural circuits (Clayton, 2000; Hoffman et al., 1993; Mello, 2002). Several lines of evidence suggest close associations between IEG activation and activity at the presynaptic neuron responding to the behaviorally relevant stimulus (see Guzowski, 2002 for a brief but detailed review of the signal transduction events leading up to IEG activation). Exactly what sort of activity drives IEG activation is not always known for many of the genes in this category. The induction of an action potential in the IEG‐expressing neuron is one event that can trigger gene activation (Mello, 2004) but it is not essential for gene expression to occur (Clayton, 2000; Jarvis, 2004). If the neuron depolarizes or if neurotransmitter binding results in a graded potential above some presumed threshold, intracellular second messengers, such as cyclic AMP or calcium ions either entering the cell or released from intracellular stores, activate cellular phosphatases, as well as kinases, such as protein kinases C and A, a‐calmodulin protein kinase II, and the tyrosine kinases (Bozon et al., 2003). Among their wide range of functions, these enzymes then activate various constitutive regulatory transcription factors, such as cyclic AMP response element‐binding protein (CREB), which initiate a cascade of neuronal transcription events, some of which include transcription of IEGs. Importantly, the degree to which this intracellular signal transduction cascade is capable of ultimately giving rise to an IEG response is highly sensitive to neuronal experience and can be fine‐tuned by noradrenergic (Cirelli and Tononi, 2004; Yamada et al., 1999), dopaminergic, cholinergic, and cannabinoid (Whitney et al., 2003) neuromodulatory input.

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

75

IEGs fall into two categories. One is the effector IEGs, which include Arc, narp, and homer, among others. The other is the inducible‐ or regulatory‐transcription‐factor IEGs such as c‐fos and zenk (Mello et al., 1992). Lanahan and Worley (1998) determined (with the use of subtractive hybridization techniques) that approximately 30–40 genes make up the total IEG response in the hippocampus of rats, approximately 10–15 of which encode inducible transcription factors and the rest of which encode effector IEGs. A reasonable hypothesis is that a similar suite of genes makes up the total IEG response in other areas of the brain and in other species, but this has not yet been determined. The effects of these newly transcribed IEGs have been among the more vexing problems in the study of IEGs, and, to be sure, these effects are likely to be varied, multifaceted, and heavily dependent on context, condition, and experience. For example, the stimulus‐evoked expression of a particular inducible transcription factor does not necessarily result in the transcription of a particular gene. This is the case in part because of the extraordinary vagaries of eukaryotic gene transcription and its dependence on a suite of potentially interacting proteins present in the cell at any given time, which itself may depend on the cell’s recent synaptic experience. Moreover, the mechanism by which an IEG exerts its function depends on whether it is an effector or inducible transcription factor. Nonetheless, some generalities are emerging about IEG function. Effector genes function in diverse ways, from regulating cellular growth, intracellular signaling, and metabolism to synaptic remodeling and other cell‐structure changes. Inducible transcription factors may largely regulate the transcription of delayed effector genes having similar functions to those of immediate effector genes but possibly playing more of a role in modulating the potential of synapses to express experience‐dependent plasticity, such as LTP or depression, sometimes referred to as metaplasticity (Fischer et al., 1997). Induction of IEGs has been associated with a variety of cellular processes linked to the formation, consolidation, and retrieval of memories as well as to other cognitive processes thought to be mediated by long‐term neuronal plasticity (Bozon et al., 2002, 2003; Clayton, 2000; Guzowski, 2002; Tischmeyer and Grimm, 1999). For example, disruption of the Zif268 IEG transcription factor targeted at the dentate gyrus of the mouse hippocampus prevents late LTP and impairs performance in tasks requiring long‐term memory but does not impair early LTP or performance in tasks requiring short‐term memory (Jones et al., 2001). Lack of central nervous system c‐fos in adult mice impairs hippocampus‐dependent spatial and associative learning tasks, and this impairment is likely due to impairment of NMDA receptor‐dependent LTP formation (Fleischmann et al., 2003). Still, despite these broad general effects of some IEGs,

76

GREGORY F. BALL ET AL.

a detailed understanding of the mechanisms giving rise to their function has yet to emerge. Moreover, although most research on IEG function has focused on synaptic plasticity, IEGs might serve a spectrum of other functions not closely related to neural plasticity. Notwithstanding these gaps in knowledge, the role of IEG quantification in response to behaviorally relevant stimuli in neuroethological research should not be underestimated. Specifically, with very little knowledge of the downstream effects of IEG induction, quantification of IEG induction has been instrumental in identifying and localizing brain regions sensitive to the stimulus and thus for functional mapping of neural circuitry involved in precipitating complex, experience‐dependent behavior (Mello, 2002). This role has been particularly visible in studies of the songbird auditory forebrain and its responses to behaviorally relevant acoustic stimuli sometimes directly related to the task of mate choice in females. 2. IEG Induction in Songbirds Across multiple taxa, a variety of nuclei within the avian brain specifically expresses IEGs in association with a number of behavioral states, from appetitive and consummatory components of sexual behavior in Japanese quail (Tlemc¸ani et al., 2000) to onset of incubation behavior in ring doves (Sharp et al., 1996) and broody behavior in Japanese quail (Ruscio and Adkins‐Regan, 2004) to homing behavior in pigeons (Shimizu et al., 2004). Perhaps in no other avian system has the analysis of IEG expression provided such insight on brain–behavior relationships as it has in songbirds (see Ball and Balthazart, 2001; Ball and Gentner, 1998; Clayton, 2000; Mello, 2002, 2004; Mello et al., 2004; Ribeiro and Mello, 2000 for reviews). Quantification of IEG expression in songbird brains has played an important role in identifying brain areas in songbirds that function in processing complex acoustic stimuli related to behavior. As noted previously, the importance of auditory areas, such as CMM and NCM, for the processing of conspecific song was discovered by virtue of the massive zenk induction described in these areas specifically in response to conspecific song exposure (Mello and Clayton, 1994; Mello et al., 1992). Electrophysiological and tract‐tracing studies have confirmed the role of these areas as an acoustic processing hierarchy fundamental in the behavioral responses to species‐specific song (Chew et al., 1995; Stripling et al., 1997; Vates et al., 1996). The work on IEG expression and song perception has led to important new studies about the locus of the memory needed for song learning (Bolhuis et al., 2000). Two IEGs, zenk and fos, are expressed in nuclei in the motor production pathway for song, including HVC and RA, specifically in association with song production, even in deafened birds that sing (Jarvis and Nottebohm, 1997; Kimpo and Doupe, 1997).

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

77

This latter observation indicates that these gene responses in the song system associated with song production are not related to hearing song. The number of songs produced tends to be positively correlated with the number of cells expressing IEGs (Jarvis and Nottebohm, 1997; Kimpo and Doupe, 1997), although European starlings also exhibit a positive correlation between the protein product of the IEG c‐fos and singing rate in HVC and RA (Heimovics and Riters, 2005). Such positive correlations between fos expression and singing were also observed in brain areas related to the motivation to sing such as the preoptic medial nucleus (POM) and the ventral tegmental area (VTA). However, in this study singing during the nonbreeding season was assessed and compared with singing in a breeding context (i.e., in the spring in the presence of a female) and positive correlations between fos expression and song rate were only apparent in the breeding season but not in the nonbreeding season (Heimovics and Riters, 2005). Based on all these findings related to song production and perception, the application of IEG quantification has since extended beyond the songbirds and resulted in the discovery of homologous song nuclei in other avian orders such as parrots (Jarvis and Mello, 2000) and hummingbirds (Jarvis et al., 2000). IEG expression in the songbird CMM and NCM is very sensitive to conspecific song, with song‐induced expression evident after as little as a 2‐sec duration of song (the length of a single song in the zebra finch; Kruse et al., 2000) and in the presence of substantial levels of background noise (Vignal et al., 2004). The IEG response to conspecific song requires experience with song during the young bird’s primary period of song acquisition, between 20 and 30 days of age in a zebra finch (Jin and Clayton, 1997). One interpretation of this finding is that IEG induction during the sensitive period for song learning may therefore play an important role in this learning. Studies in zebra finches have found that songs that were most accurately copied during the sensitive period for song learning were the most effective in inducing zenk expression in NCM (Bolhuis et al., 2000, 2001; Terpstra et al., 2004). These data have been used to argue that zenk induction in NCM is an important step in the formation of memories of tutor song and that NCM may be an important site for the localization of these memories (Bolhuis et al., 2000, 2001; Terpstra et al., 2004). Although the songbird auditory forebrain expresses ZENK and FOS after an individual is exposed to any of a number of acoustic stimuli, including those which would seemingly be of little relevance to the individual’s natural history, one hallmark of the CMM and NCM is the stimulus specificity that yields a differential IEG response (Bailey and Wade, 2003; Bolhuis and Eda‐Fujiwara, 2003; Clayton, 2000; Mello, 2002, 2004). For example, the type of conspecific vocalization and the sex of the individual exposed to the

78

GREGORY F. BALL ET AL.

vocalization affect distributions of IEG expression patterns. In canaries, IEG induction patterns in NCM are tonotopically organized with the NCM areas of greatest sensitivity corresponding to the frequency of the whistle in the male’s song (Ribeiro et al., 1998). In zebra finches, some degree of hemispheric lateralization emerges in the NCM ZENK response to song (Lieshoff et al., 2004). In black‐capped chickadees, both the fee‐bee song and chick‐a‐dee call elevate ZENK induction in the CMM and NCM, but dorsal NCM is more sensitive to song‐induced than to call‐induced ZENK induction, whereas caudal portions of the auditory forebrain show equal sensitivity (Phillmore et al., 2003). Males exhibited greater vocalization‐induced ZENK expression than females in this species, although no sex differences in song‐induced NCM ZENK induction have been found in European starlings (Duffy et al., 1999). The strength of this IEG expression increases with increasing relevance or novelty of the stimulus and itself can depend on prior experience. For example, ZENK and FOS induction in the CMM and NCM is selective for conspecific over heterospecific song (Bailey et al., 2002; Mello et al., 1992a) and as noted previously for songs better learned during early developmental stages over those not as well learned (Bolhuis et al., 2000, 2001; Terpstra et al., 2004). Isolation from song early in life results in reduced song‐ induced ZENK induction during the adult phase (Hernandez and MacDougall‐Shackleton, 2004), whereas repeated exposure to the same song results in habituation of the ZENK response in the NCM, which is rescued with exposure to a novel song (Chew et al., 1995; Jarvis et al., 1995; Mello et al., 1995). Due to the participation of the IEGs ZENK and FOS in synaptic remodeling in other systems, it is likely these areas of the songbird auditory forebrain, specifically CMM and NCM, undergo the experience‐ dependent plasticity that is necessary for long‐term memory formation, memory consolidation, and memory retrieval (Mello, 2002). Because male songbirds produce song, in part, to attract mates, females are behaviorally sensitive to song exposure, and one might therefore hypothesize that quantification of differential IEG induction in the female brain would reveal neural systems involved in female perception of song. Therefore, of particular interest was the discovery of ZENK and FOS induction in the CMM and NCM of females that was greater after exposure to male conspecific song than after male heterospecific song (Bailey et al., 2002; Duffy et al., 1999; Mello et al., 1992a), suggesting that IEG induction in the auditory forebrain might be involved in song‐mediated recognition of a mate of the appropriate species. Females of a diversity of animal species show behavioral responses to males that suggest a degree of perceptual sensitivity substantially greater than what would be required for mere species recognition. Females of

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

79

many species discern high‐quality from low‐quality males based on phenotypic variation among males that presumably serves as an honest signal of their quality. In birds, vocal phenotypic features are thought to be one such signal. Females tend to choose mates based on variation in song type (Catchpole and Slater, 1995; Searcy, 1992; Searcy and Yasukawa, 1996), which may serve as an honest signal of male quality (Gil and Gahr, 2002). A large body of research on songbird species suggest that at least one dimension of song variability highly salient to choosy females involves song complexity as measured, for example, by repertoire size or by motif stereotypy (Searcy, 1992) which can be limited by nutritional constraints during development (Buchanan et al., 2003b; Nowicki et al., 2002a,b). In the field of mate choice, which concerns the processes whereby individuals choose between different prospective mates based on their phenotypic characters, the study of bird song and related IEG induction has also played a useful role. In fact, avian IEG studies have revealed a level of perceptual discrimination in the female auditory forebrain that is sufficient in some cases to resolve quality differences between males of a single species. For example, in the non‐Oscine budgerigar (Melopsittacus undulatus), NCM ZENK induction increases with the complexity of male song to which they are exposed (Eda‐Fujiwara et al., 2003). In an oscine, the mountain white‐crowned sparrow (Zonotrichia leucophrys oriantha), females prefer in a mate‐choice context song of their local, natal dialect to song of a foreign dialect (MacDougall‐Shackleton et al., 2001), and CMM and NCM ZENK expression is greater in females exposed to local than in those exposed to a foreign dialect (Maney et al., 2003). In another oscine, the canary, females prefer male songs that contain a particular syllable type, and, when females are exposed to songs with this syllable type, ZENK induction in their auditory forebrain increases relative to that in females exposed to songs without this syllable type (Leitner et al., 2005). But, is this auditory forebrain IEG induction part of a signaling pathway that ultimately gives rise to the mate‐choice decision, or is this association between the IEG response and mate choice merely correlational? It is possible that the induction is related to sensory processing but that there is not a causal connection between the IEG induction and the behavioral result. Until we can manipulate IEG induction in a controlled fashion, perhaps via antisense oligonucleotide methods or small interfering RNA we may not be able to determine the answer to this question definitively. However, evidence in white‐crowned sparrows provides further support of the tight association between ZENK induction and mate preference. Specifically, ZENK induction in both the CMM and NCM of females correlates positively with the level of sexual receptivity to a particular song type, as measured by the female’s vocal behavior and copulatory solicitation

80

GREGORY F. BALL ET AL.

behavior (Maney et al., 2003). Thus, there is a strong association between the strength of the preference and the level of ZENK expression, suggesting the presence of a functional tie. 3. IEG Induction and Mate Choice in European Starlings The first demonstration that conspecific song variation directly relevant to mate choice affects IEG induction was found in the European starling (Gentner et al., 2001). Female starlings, which prefer long‐bout over short‐bout song, have greater ventral NCM (NCMv) ZENK induction in response to long‐bout than short‐bout songs, even when the total song exposure does not vary (Fig. 4). Because the NCM and CMM share reciprocal projections (Vates et al., 1996), it was not surprising to find from a second study that this ZENK response bias toward long‐bout song also occurs in the CMM and dorsal NCM (NCMd) (Sockman et al., 2002). However, there is evidence that CMM and NCM process different aspects of song recognition (Gentner et al., 2004) and may release different female responses to song (Maney et al., 2003). These results demonstrated a neural‐response correlate to a mate‐choice preference and raised the question as to how this neural response bias toward the behaviorally favored cue arises. There could well be a complex interplay between genetically influenced predispositions to respond more robustly to particular songs and experience with song that might shape these IEG response biases. There is still much we do not know, but S insight concerning this problem comes from two recently published studies on the role of recent adult song type experience on the IEG response bias in the female starling auditory forebrain. We detail these studies later, but first provide some background on some fundamental concepts in mate choice and the potential role of the prevalent song culture in shaping behavioral preferences and neural response biases that presumably give rise to such preferences. When discussing mate choice, one can often make generalizations about preferences as if they are relatively fixed in populations and therefore might not be modulated by experience (Andersson, 1994). However, the strong directional selection for trait expression predicted by such a perspective is difficult to reconcile with the many observations of substantial between‐population, between‐individual, and even within‐individual variation in expression of sexually selected traits (Badyaev and Qvarnstro¨m, 2002). For a sender’s traits to communicate to the receiver an honest signal of the sender’s quality, expression of the traits must be subject to constraint and therefore incur some cost in order for this device to remain stable as an effective communication mechanism (Grafen, 1990). As a sexually selected

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

81

Fig. 4. Patterns of expression of the protein product of the IEG zenk in response to song organized in long bouts (panels in row 1), the same amount of song organized in short bouts (panels in row 2), and a short amount of song (panels in row 3). Note that expression is highest in response to long‐bout song (row 1).

82

GREGORY F. BALL ET AL.

trait in some species, song production by songbirds can entail a variety of costs, including those associated with increased energy requirements (Eberhardt, 1994; Oberweger and Goller, 2001; Thomas et al., 2003; Ward et al., 2003), increased predation risk (Catchpole and Slater, 1995), and a reallocation of time also required for other activities, such as foraging (Gil and Gahr, 2002). It is not unreasonable to predict that the degree to which each of these costs can constrain song behavior might itself vary. For instance, increased energy requirements of relatively costly forms of song production may constrain expression of such forms relatively little when food is plentiful, resulting in, a greater proportion of individuals expressing the trait than otherwise. Similarly, under heavy predation risk, the probable costs of some song types (i.e., those most likely to attract a predator) increase and may drive a greater proportion of individuals to express the less costly trait than expected under lighter predation pressure. Therefore, within a population, the proportion of individuals singing one song type may vary with environmental context, and females will be faced with variation in the proportion of preferred traits being expressed. Through mate sampling experience and variable population densities of sexually selected traits, individuals should therefore express some flexibility in the phenotypic threshold they set for their mate choice (Badyaev and Qvarnstro¨m, 2002; Jennions and Petrie, 1997; Wiegmann et al., 1996). Otherwise, they pass some years without mating due to the inability of many males to meet high song quality demands. Several studies provide evidence for such frequency‐dependent choice behavior. In damselflies, males choose females of the more prevalent color morph (van Gossum et al., 2001); and in wolf spiders, females likewise choose the male morph which is more familiar (i.e., prevalent; Hebets, 2003). Importantly, this process appears to occur in songbirds, as well. Female white‐crowned sparrows show behavioral preferences for their natal song dialect unless, during preceding months, they experienced a more prevalent foreign dialect (MacDougall‐Shackleton et al., 2001). Female cowbirds (Molothrus ater) also exhibit mating preferences for the male song environment most recently experienced by the female, and it is likely that song was the salient component of the social environment releasing such preferences (Freeberg et al., 1999). Similar findings have been reported in canaries (Nagle and Kreutzer, 1997a,b). In sum, because not every female can mate with a particular male phenotype when the phenotype is in short supply, females seem to adjust the threshold or criteria for mate preference, mate choice, or both. A physiological mechanism for experience‐dependent modulation of mate choice has not been determined. However, evidence was accumulated in female European starlings of two complementary neural systems, each

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

83

mediated by forebrain induction of separate IEGs that are made sensitive by experience with one of the ends of the behaviorally relevant axis of song variation (Sockman et al., 2002, 2005). In these studies, pairs of photosensitive female starlings were exposed to 1 week of male song played 5.5 hr/day while maintained on a photoperiod of 11L:13D. Female pairs differed based on whether the week of song experience to which they were randomly assigned was composed entirely of long‐bout or short‐bout songs. This was our attempt to mimic and manipulate the song environment females might experience when sampling mates before making mate‐choice decisions early in the spring. Following this manipulation of the perceived song experience, we isolated each female of each pair individually for approximately 40 hr and then, while they were still isolated, exposed them to 30 min of song stimulus, sacrificed them 60 min later, and then collected their brains for sectioning and subsequent IEG quantification by immunocytochemistry. Specifically, we exposed one female of each pair to a 30‐min long‐bout stimulus (mean song length ¼ 55.6 sec) and the other to a 30‐min short‐ bout stimulus (mean song length ¼ 25.4 sec). Importantly, the 1 week of experience songs and the 30 min of stimulus songs had been recorded from different males, meaning that the stimulus songs were always novel, even when the experience and stimulus categories were the same (e.g., long‐bout experience followed by long‐bout stimulus; Sockman et al., 2005). In this manipulation, there were four groups of females: (1) long‐bout experience long‐bout stimulus, (2) long‐bout experience short‐bout stimulus, (3) short‐ bout experience long‐bout stimulus, and (4) short‐bout experience short‐bout stimulus. As previously demonstrated (Gentner et al., 2001), we found that ZENK induction in the auditory forebrain of female starlings was greater after exposure to the 30‐min long‐bout than after exposure to the 30‐min short‐ bout song stimulus (Sockman et al., 2002, 2005). However, this ZENK response bias toward the long‐bout stimulus was modulated by experience. That is, the 1‐week long‐bout experience enhanced this bias (i.e., greater ZENK in response to the long‐bout than in response to the short‐bout stimulus), whereas the 1‐week short‐bout experience attenuated it. So, based on these results, it appears that recent song‐sampling experience during adulthood influences the neural response bias toward a preferred male trait important in female mate choice. The presence of experience‐ dependent mate‐choice decisions suggests that such a system must exist, but this is the first identification of such a system in any species. More importantly, these findings demonstrate a surprising level of adult neuroplasticity, whereby responses to a recent adult cultural experience are modulated not by experience with the stimulus itself, but rather by

84

GREGORY F. BALL ET AL.

experience with the category of stimulus to which they are sensitive. That is, because the song sets comprising the experience and stimulus treatments were taken from different males, the songs varied across treatments in their spectrotemporal features while they retained their categorical distinctions as being long bout or short bout. In light of this treatment design, our data reveal an experience‐dependent response plasticity spanning the category into which the stimulus is organised. The auditory forebrain regions of CMM and NCM that are so sensitive to recent adult experience in their ZENK expression levels also express IEGs other than ZENK. For example, the auditory forebrain of zebra finches and canaries upregulates both of the heterodimerizing transcription factors FOS and JUN in response to conspecific song (Bailey and Wade, 2003; Bailey et al., 2002; Bolhuis et al., 2000, 2001; Nastiuk et al., 1994). Moreover, song‐induced expression of both ZENK and FOS correlates with how well an individual learned the stimulus song during development and varies with sex (Bailey and Wade, 2003; Bolhuis et al., 2001). Such associations suggest a dynamic sensitivity of these IEG systems to modulation by experience. An environment conducive to song learning in young birds may modulate their future sensitivity to song stimuli, as reflected by the song‐induced IEG induction in the auditory forebrain. However, explanations that do not require such individual plasticity are also possible. An elevated sensitivity to a particular song during development (as reflected in the magnitude of the IEG response to that song) may predispose an individual to learning that song better than individuals without an elevated sensitivity to the song. Our study on experience‐ dependent modulation of song‐induced ZENK induction in the auditory forebrain of female starlings indicates, in fact, that within an individual, auditory neural responses to song are plastic and shaped by the individual’s recent adult experience. But are other IEGs similar to ZENK in their song‐ induced sensitivity to recent adult experiences? To further characterize the experience‐dependent properties of the CMM and NCM in female starlings, we also examined the FOS response bias in these females that had recent experience with long‐bout or short‐ bout songs. Consistent with the ZENK response, FOS induction in response to long songs is greater than that in response to short songs (Sockman et al., 2005). However, whereas this ZENK response to differences in song length is made sensitive by experience with the long‐song category, the FOS response to differences in song length is made sensitive by experience with the short‐song category. That is, FOS expression was greater in response to novel long than to novel short songs following a 1‐week experience with short but not long songs. Thus, the ZENK‐ and FOS‐ signaling pathways are sensitized to variation in song‐length categories by

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

85

experiences with songs at opposite ends of the starling song‐variation continuum. This is likely because of the separate actions of songs in the long category on the ZENK‐signaling pathway and songs in the short category on the FOS‐signaling pathway. Experience sampling long songs apparently elevates subsequent ZENK expression in response to songs of a category that has common long‐song features, whereas experience sampling short songs apparently suppresses subsequent FOS expression in response to songs of a category that has common short‐song features. In other words, the ZENK pathway is made sensitive to long songs by experience sampling songs of the long category, and the FOS pathway is made less sensitive to short songs by experience sampling songs of the short category. These findings suggest the presence of complementary neural systems made sensitive in register with the natural axis of phenotypic variation, song length, that is fundamental to the female’s mate choice. Complementary ZENK‐ and FOS‐signaling systems in the auditory forebrain could help establish the attractiveness of a given song within the context of the idiosyncratic distribution of song phenotypes sampled by the female. Potentially, the attractiveness of that song can vary with the song experienced locally, as suggested by studies on white‐crowned sparrows, cowbirds, and canaries (see citations earlier). Whether starling mate choice varies with recent song‐sampling experience awaits further research. Of considerable interest would be to determine the functional significance of such complementary neural systems. The use in birds of conditional and localized ZENK and FOS knockouts, which have been developed in rodent systems (Bozon et al., 2002; Fleischmann et al., 2003; Jones et al., 2001), might be helpful in this regard. However, at present there is no easy way to implement knockout technologies in birds so other methods, such as antisense oligonucleotide methods or small interfering RNA, is a more plausible approach to consider (Charlier et al., 2005). But even in the absence of this technology, which would likely reveal the effects specific to the forebrain ZENK‐ and FOS‐signaling systems, the data thus far strongly implicate an important role for the CMM and NCM in experience‐dependent mate choice in the starling. How does this experience dependence arise, and what are the downstream neural and behavioral targets? Experience‐dependent representational plasticity of cortical neurons occurs in most sensory systems (Calford, 2002; Gilbert et al., 2001). In the mammalian auditory system, it is tied to coincident activation of both cholinergic (Bakin and Weinberger, 1996; Kilgard and Merzenich, 1998) and catecholaminergic (Bao et al., 2001) neurotransmitter systems. In rats, noradrenergic activity selectively modulates light‐induced FOS but not

86

GREGORY F. BALL ET AL.

ZENK expression in the visual cortex (Yamada et al., 1999). Thus, one might hypothesize that in the female starling and other songbirds, separate neurotransmitter or neuromodulator systems would mediate the IEGs ZENK and FOS and shape the real‐time responses of single cells to stimuli as a function of recent experience. The noradrenergic system is one obvious candidate for shaping responses in the vertebrate forebrain. Among songbird species, noradrenergic inputs mediate state‐dependent auditory responses in the song system forebrain nucleus interfacials (Cardin and Schmidt, 2004), and evidence for catecholaminergic innervation of the auditory forebrain has been described in canaries (Appeltants et al., 2001) and zebra finches (Mello et al., 1998) and is probably a general phenomenon among songbirds. Lesions to noradrenergic projections through the administration of the noradrenergic‐selective neurotoxin DSP‐4 disrupts auditory processing in female canaries (Appeltants et al., 2002), implicating an important role for this neurotransmitter system in modulating song‐ specific behavioral responses in female songbirds. Finally, social context effects on song‐induced ZENK expression in area X of male zebra finches are also blocked by the administration of DSP‐4, indicating that at least in some brain regions norepinephrine can modulate plasticity in the expression of IEGs (Castelino and Ball, 2005). Some evidence exists for a role from other systems as well, such as those mediated by gamma‐aminobutyric acid (GABA) and cannabinoids. GABA and ZENK colocalize in neurons of the CMM and NCM (Pinaud et al., 2004), and activation of the CB1 cannabinoid receptor inhibits song‐ induced ZENK expression and habituation of ZENK expression in the NCM but has no effect in regions of the field L complex that show song‐ induced ZENK expression (Whitney et al., 2003). Clearly a role of various neuromodulatory systems in song‐induced IEG expression looks probable and merits further study. How any neuromodulatory factor might influence IEG responses to conspecific song is not well understood, but some advances in this area have been made. In zebra finches, within the auditory forebrain exclusively, the phosphorylation of the extracellular signal‐regulated kinase (ERK) shows remarkable similarity to ZENK in its response to conspecific song (Cheng and Clayton, 2004). Specifically, initial exposure to novel song but not tones or noise upregulates phosphorylation of this protein, which then habituates after repeated exposure to the same song. Presentation of a new novel song again elevates phosphorylation of ERK without affecting its habituation to the familiar song. Infusion into the auditory forebrain of the enzyme responsible for ERK activation blocks induction of ZENK. It seems conceivable therefore, that ERK might help to mediate any modulatory action of song‐induced ZENK induction in the female

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

87

auditory forebrain by norepinephrine or other factors. Modulation of FOS may be similar, although two lines of evidence suggest that different IEGs may mediate the actions of different neuromodulatory systems: tuning properties of the forebrain FOS system in starlings that are opposite those of the forebrain ZENK system (Sockman et al., 2005) and evidence from mammalian studies (Yamada et al., 1999) suggesting that different IEGs may mediate the actions of different neuromodulatory systems. In addition to the need for more research on mechanisms mediating the experience‐based modulation of forebrain IEG tuning, there is also a need for studies on the behavioral outcome of this differential IEG expression in the songbird forebrain. The opposite tuning properties of the ZENK‐ and FOS‐signaling systems raise intriguing questions. Do they both participate in systems that ultimately promote preference behavior? Does one mediate preference and the other aversion behavior? Evidence from behavioral and IEG studies on white‐crowned sparrows suggests that ZENK induction is at least positively associated with strength of preference (Maney et al., 2003), as indicated earlier. However, little is known about how FOS might participate in such a system. If the forebrain FOS‐signaling system participates in mechanisms mediating aversion behavior, then the complementary properties demonstrated for ZENK and FOS might form a portion of a neural system for the mediation of mate preferences based on the prevalent song culture or the female’s recent mate‐sampling experience. Regardless of the outcome of these prospective studies, it is clear that female European starlings have evolved a highly plastic nervous system with properties that would seem capable, in part, of mediating mate‐choice decisions based on the female’s recent experience with a dynamic culture of male phenotypes.

B. ELECTROPHYSIOLOGICAL RESPONSES

TO

SONG

There is a relatively long tradition of studying electrophysiological responses to song in auditory areas and in the song control system to assess their potential significance for song perception (Katz and Gurney, 1981; Leppelsack and Voigt, 1976). This is a large area of research that has been reviewed in some detail elsewhere (Mooney, 2004; Theunissen et al., 2004). We will only cover certain aspects relevant to our discussion of starling song. One hallmark of cells within many of the song system nuclei is their selective response to a ‘‘bird’s own song’’ (BOS; Margoliash, 1987). That is, one readily finds neurons throughout the song system whose firing rates and/or temporal response properties are ‘‘tuned’’ to the acoustics of the

88

GREGORY F. BALL ET AL.

song that the bird sings. In the nucleus HVC, the stimulus specificity observed for BOS has both spectral and temporal components, with responses contingent on the presence (and absence) of acoustic energy in specific frequency bands or on specific temporal combinations of sounds (Margoliash, 1983; Margoliash and Fortune, 1992). In addition, these so‐ called ‘‘BOS responses’’ are strongly modulated by behavioral state (Dave et al., 1998; Schmidt and Konishi, 1998) and are, in at least some cases, observed in cells that also show strongly coupled premotor activity during song production (Dave and Margoliash, 2000; Yu and Margoliash, 1996). The sensorimotor integration at both the cellular and system level, which gives rise to the BOS response, is an active area of research among birdsong neuroethologists. Understanding these physiological mechanisms will likely impact the broader study of sensorimotor learning in other systems, and may be of value in understanding the perceptual role of self‐ generated sounds in human speech processing (Margoliash, 2003). Despite early suggestions that BOS selective responses in song system nuclei, specifically the vocal motor pathway (VMP), might reflect a ‘‘motor theory’’ of song perception (sensu Liberman et al., 1967), recent data suggest a somewhat different interpretation. The state‐dependent nature of these responses and the absence of auditory responses altogether in many VMP neurons in the awake animal (Dave et al., 1998) argue against the notion that the pathways controlling vocal output also contribute to sensory representations of song. Instead, BOS selectivity more likely reflects the involvement of acoustic feedback in ongoing regulation of song production mechanisms. It now appears that only very specific sorts of auditory information, namely BOS, are admitted to the song system so that the bird can detect, and thus correct, any deviations between the intended song and that actually produced. Nonetheless, there are data that suggest a production‐independent role for both HVC and the AFP in adult song perception. Lesions to lMAN in canaries affect auditory but not visual discrimination (Burt et al., 2000); lesions to HVC in female canaries abolish female behavioral preferences for conspecific over heterospecific song (Brenowitz, 1991) and for sexually attractive song phrases over other phrases of conspecific song (Del Negro et al., 1998). However, this is not the case for female zebra finches where HVC is quite small and not connected with nucleus RA (MacDougall‐ Shackleton et al., 1998). In both male and female starlings, HVC lesions affect the ability to form new associations with familiar songs while leaving retention of learned conspecific song discriminations intact (Gentner et al., 2000). Clearly the role of these structures, and by extension the general role of vocal–motor systems, in sensory perception and cognition requires further study—ideally through electrophysiology in awake animals.

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

89

As noted previously, auditory areas, such as NCM and CMM, have received attention as key sites for the processing of auditory information related to song. The extensive amount of data on IEG expression just reviewed is certainly consistent with this new direction. In European starlings, electrophysiological data demonstrate directly a role for CMM in learned recognition of song. After training starlings to recognize two sets of conspecific songs, Gentner and Margoliash (2003) observed that single neurons and populations of neurons in the medial CMM respond selectively to acoustic features contained in those songs that the birds had learned to recognize. In contrast, no neurons were selective for similar features in songs that were novel to the birds. This argues very strongly that the response functions of CMM, at both the single unit and the population level, are a direct product of each bird’s unique sensory experience. Mechanisms of experience‐dependent plasticity act to modify the responses of CMM neurons based on the functional demands of song recognition. Several additional results from this study are consistent with this notion. First, the spectrotemporal tuning properties of CMM cells correspond closely to song features correlated with individual motifs. That is, the same auditory objects that control recognition behavior also predict the responses of selective cells in CMM. Second, the variation in neuronal response strength among the familiar songs was dependent on the reinforcement contingencies used for recognition training. For animals trained with a go/no‐go procedure to discriminate between two sets of songs, the Sþ songs (i.e., the songs that were reinforced after a go response) elicited the significantly stronger responses than S
songs (i.e., the songs that were not reinforced after a go response), which in turn elicited significantly stronger responses than novel songs (Getner and Margoliash, 2003). When positive reinforcement is available for both sets of songs, the response strengths associated with each set of familiar songs are similar, but still greater than those associated with novel songs. Thus, the response profiles of neurons in CMM are shaped not only by task relevant acoustic features of conspecific songs in a ‘‘bottom‐up’’ fashion, but also by so‐ called ‘‘top‐down’’ mechanisms presumably through reward systems (Gentner and Margoliash, 2003). Studies of ZENK expression in starlings have extended this view of the different roles played by these two areas (the CMM and NCM) in song recognition behavior (Gentner et al., 2004). In NCM, ZENK expression is elevated in association with an operant song recognition task while the starlings are acquiring novel song discriminations (Gentner et al., 2004). In CMM, there is elevation of ZENK expression during recognition of familiar songs during the acquisition of novel associations with familiar songs and when acquiring novel song discriminations (Gentner et al., 2004).

90 VI. FUNCTIONAL BASIS

GREGORY F. BALL ET AL.

OF

SONG PREFERENCES

IN

EUROPEAN STARLINGS

As discussed previously in this chapter, the primary adaptive function of male starling song is in attracting and obtaining mates (Cuthill and Hindmarsh, 1985; Eens, 1997). Males significantly increase their rate of singing in the presence of a female, much more so than in the presence of another male (Eens et al., 1993). Females choose their mates on the basis of this song, demonstrating clear preferences for males that sing in longer bouts or possess large repertoires (Eens et al., 1991b; Gentner and Hulse, 2000a; Mountjoy and Lemon, 1996). Earlier in this chapter, we discussed the physiological correlates of this behavioral response bias among females. Now we turn our attention to the evolutionary function of female song preferences. What reproductive advantage(s), if any, do choosy females gain by exhibiting song‐based mating preferences? What information, if any, do males provide about themselves by singing long complex songs? Much attention has been given to the hypothesis that male starling song functions as an indicator of some aspect of male quality. For instance, song might provide information about the quality of a male’s territory, his propensity and ability to provide parental care, his age, or even some underlying genetic quality. The first two possibilities are examples of traits that provide direct benefits to females while the latter are examples of indirect benefits. A. DIRECT BENEFITS Male starlings, unlike some other species of songbirds, do not defend large territories that include food resources (Feare, 1984). Rather, starlings defend only a few square meters immediately surrounding the nest (Kessel, 1957) and forage in pairs or flocks away from the nest (Feare, 1984; Kessel, 1957). Starlings are cavity nesters and, thus, the availability of suitable nest sites is limited. Therefore, the quality of the nest cavity location will likely vary among males, which may lead to differential reproductive consequences for the females that mate with them (Eens, 1997; Mountjoy and Lemon, 1996). While it is true for a number of species that song performance and territory quality are correlated (Catchpole and Slater, 1995), such is not the case with starlings. Repertoire size, which is highly and positively correlated with song bout length, did not correlate with nest box location preferences in a controlled field experiment (Mountjoy and Lemon, 1996). However, repertoire size is correlated with a measure of female preference (delay between the claiming of a nest box by a male and the laying of the first egg), with males possessing the

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

91

largest repertoires being most preferred by females (Mountjoy and Lemon, 1996). Thus, in starlings, territory quality (or the location of the nest cavity) does not appear to be a factor in mate‐choice decisions nor is it signaled by male song. Another possible way in which choosy females could receive direct fitness benefits is if a male’s song performance indicates his ability and propensity to help care for offspring. In starlings, both sexes incubate the eggs and feed the nestlings, with the females performing most (70%) of the incubation (Feare, 1984). Reports regarding the division of labor between the sexes for nestling provisioning vary, with females making more feeding trips to the nest than males or both sexes provisioning at equal rates (Feare, 1984). Thus, males provide a substantial amount of parental care and females would have a selective advantage if they could predict which males would provide the best care for their young. However, the evidence to date indicates that male song does not provide such a cue for females. Repertoire size was not correlated with time spent incubating eggs or the rate of nestling provisioning among males in a Canadian population of starlings (Mountjoy and Lemon, 1997). Furthermore, in a Belgian population of starlings in which polygyny is common, males with the most complex song are more likely to obtain a second mate, and polygynous males provide less parental care than their monogamous counterparts (Eens, 1997). Taken together, these data indicate that male starling song probably does not function as an indicator to females regarding which males can offer the most direct material benefits. Thus, starlings are good candidates for investigating other forms of indicator mechanisms, often referred to as ‘‘good genes’’ models, in which females benefit indirectly by mating with males of superior genetic quality. B. INDIRECT BENEFITS Much attention has been given to the hypothesis that song performance indicates potential indirect benefits for females via transmission of good genes to offspring. Good genes models refer to any male phenotype that is heritable and provides some selective advantage to offspring that inherit the trait, thus providing the female with increased reproductive success (Johnstone, 1995; Pomiankowski, 1988; Williams, 1966; Zahavi, 1975). The phenotype in question can be body condition in general, the ability to resist disease, or any aspect of genetic quality that increases the probability of survival or mating success of the female’s offspring (Fisher, 1930; Folstad and Karter, 1992; O’Donald, 1967, 1980). Male starlings with larger repertoires are, on average, older (Eens, 1997). Song also has been correlated with body condition in starlings (Mountjoy

92

GREGORY F. BALL ET AL.

and Lemon, 1996), with males possessing larger repertoires scoring higher on an index of body condition (measured as the residuals of a regression of body weight on tarsus length). Furthermore, some evidence indicates that males with larger repertoires are also more dominant (Eens, 1997; Spencer et al., 2004), suggesting that these males may have better access to food resources. Thus, male song could be an indicator of some underlying genetic quality that increases survival ability. 1. Immunocompetence Handicap Hypothesis One way in which survival ability could manifest itself is in an individual’s capacity to resist disease (i.e., immunocompetence). The immunocompetence handicap hypothesis (ICHH) proposes a mechanism through which an elaborate trait (such as complex song) could be an honest and reliable signal of immunocompetence (Folstad and Karter, 1992). The ICHH proposes that some factor, such as T, which enhances the development of secondary sexual characteristics, also suppresses immunity. According to the ICHH, only those males that possess superior immunocompetence will be able to tolerate the immunosuppressive effects of T and thus be able to maintain the high concentrations of T associated with song production during the breeding season. Thus, females who choose to mate with males exhibiting the most elaborate or most complex song would then benefit, either directly or indirectly, by mating with the most immunocompetent males. A possible direct benefit for choosy females would be that forming pair bonds with males possessing superior disease resistance may reduce their exposure to pathogens. Indirectly, females could enhance their reproductive success if immunocompetence is a heritable trait that is passed on to offspring, increasing their probability of survival (Folstad and Karter, 1992). The literature on the ICHH is vast (see Garamszegi, 2005; Roberts et al., 2004 for reviews) and much of it is beyond the scope of this chapter, thus we will limit our review of the ICHH to how it relates to starlings. In support of one component of the ICHH, recent reports provide strong evidence that male song is an indicator of immunocompetence in starlings (Buchanan et al., 2003b; Duffy and Ball, 2002; Spencer et al., 2004). The first such study revealed that, among adult males, two measures of song performance were positively correlated with two components of the adaptive immune response (Fig. 5; Duffy and Ball, 2002). Song rate was strongly correlated with cell‐mediated immunity, measured as the skin‐swelling response to injection of phytohemagglutinin (PHA). PHA is a common protein that is known to elicit proliferation of cells involved in cell‐ mediated immunity (T cells). When injected under the skin, it induces the trafficking of T cells to the site of injection, resulting in swelling. Song bout

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

93

Fig. 5. (A) Wing web swelling (mm) 48 hr postinjection of PHA in relation to song rate. The linear regression line has the equation y ¼ 0.77 þ 0.02x, r2 ¼ 0.719. (B) Anti‐KLH antibody titers (mean absorbance values relative to negative controls), averaged across days 10 and 15, in relation to song bout length (s). The linear regression line has the equation y ¼ 1.18 þ 0.18x, r2 ¼ 0.388. [Adapted from Duffy D. L., and Ball, G. F. (2002). Song predicts immunocompetence in male European starlings. Proc. Roy. Soc. Lond. Ser. B‐Biol. Sci. 269(1493), p. 849, Fig. 1.]

length was correlated with humoral immunity, measured as the antibody response to inoculation with another inert foreign protein (keyhole limpet hemocyanin or KLH; Fig. 5; Duffy and Ball, 2002). Similarly, a later report demonstrated a positive relationship between the PHA response

94

GREGORY F. BALL ET AL.

as fledglings and song bout length assessed 7 months later (Buchanan et al., 2003b). In that same study the opposite relationship was found between PHA response and song bout length among birds that had received an unpredictable food supply as fledglings, suggesting that resource reallocation occurs during times of nutritional stress (Buchanan et al., 2003b). Finally, the latest evidence indicates a positive relationship between PHA response and repertoire size among male starlings (Spencer et al., 2004). Importantly, the birds in each of the aforementioned studies were the same age; thus age alone cannot account for the variation in song or immunocompetence. Taken together, these experimental studies in starlings provide some of the strongest evidence in support of the hypothesis that song features can function as indicators of immune capability. The open question that remains to be established, however, is whether these relationships between male song performance and immune responsiveness translate into increased fitness among females. Mathematical modeling suggests that the advantage for females in choosing more immunocompetent males will vary depending on the dynamics of the pathogens prevalent in a specific population (Adamo and Spiteri, 2005). Furthermore, as researchers in the field have recently noted, the degree to which more robust responses to immune challenge reflect greater disease and/or parasite resistance remains an empirical question that is likely to be context and pathogen specific (Adamo, 2004; Viney et al., 2005). Finally, there is evidence suggesting that males that are more disease resistant are superior in other ways, some of which may be heritable (Garamszegi, 2005; Gleeson et al., 2005). Thus, females may benefit not from males’ immunocompetence per se, but from an overall superior genetic quality that manifests itself in increased resistance to disease among other things (Adamo and Spiteri, 2005). Song parameters are positively correlated with both age and body condition in starlings (Eens, 1997; Mountjoy and Lemon, 1996), lending support to this idea. Further study assessing various facets of immune function, resistance to disease and, most importantly, measures of fitness of choosy versus nonchoosy females is needed (Adamo, 2004; Adamo and Spiteri, 2005; Viney et al., 2005). Despite these gaps in our knowledge, the evidence linking song to immune responsiveness in starlings is consistent and makes a compelling case in favor of one of the major tenets of the ICHH. The primary feature that sets the ICHH apart from Hamilton and Zuk’s original hypothesis (1982) linking secondary sexual traits and health is the double‐edged‐sword effect of some hormone (e.g., T) proposed by Folstad and Karter (1992) as the mechanism mediating the relationship. According to the ICHH, T should enhance song while suppressing immune function.

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

95

The presence of T is known to facilitate singing behavior in male birds of many species, including starlings (Arnold, 1975; Ball et al., 2002; Catchpole and Slater, 1995; DeVoogd, 1991; Eens, 1997; Marler et al., 1987). Similar to other songbirds, starlings increase their song output during the breeding season when T concentrations are elevated. However, starlings also sing during the fall when secretion of T from the testes is minimal at best (Eens, 1997). Thus, while song may be enhanced by high T concentrations, it does not appear to be dependent on high gonadal T secretion (Ball et al., 2002). One report suggests that song bout length and the ability of certain social factors to induce singing by male starlings increase in the spring when T concentrations are high compared with autumn when T is undetectable (Riters et al., 2000). Furthermore, singing in the absence of a female is not affected by castration and/or implantation with T; whereas castration results in a significant reduction of female‐directed song by male starlings that is restored with subsequent T treatment (Pinxten et al., 2002). However, at least three studies have failed to find any correlation between plasma T concentration and song bout length or song rate in starlings (Buchanan et al., 2003b; Duffy and Ball, 2002; Sartor and Ball, 2005). Therefore, the relationship between T and singing in starlings appears to be context dependent and nonlinear. The third main prediction of the ICHH that T should act as a handicap mechanism via immunosuppression remains its most controversial facet. Across various taxa, the evidence regarding whether T is immunosuppressive remains inconclusive. A meta‐analysis that included birds, mammals, and reptiles found a suppressive effect of T on immunity; however, this effect disappeared when the analyses were corrected for multiple studies on the same species (Roberts et al., 2004). While some evidence exists that T suppresses immune function in starlings (De Ridder et al., 2002; Duffy and Ball, 2002; Duffy et al., 2000), alternative explanations cannot yet be ruled out. As predicted by the ICHH, physiological doses of exogenous T in male starlings in nonbreeding condition (photorefractory) resulted in a reduction of both cell‐mediated and humoral immune responses to PHA and KLH, respectively, while a similar dose in females significantly decreased humoral immune responses (Duffy et al., 2000; Fig. 6). Furthermore, the effect of exogenous T on antibody responses to KLH was dose dependent (Duffy et al., 2000). Similarly, a negative correlation was found between natural variation in endogenous T and antibody responses to KLH among male starlings (Fig. 7; Duffy and Ball, 2002). Furthermore, treatment of female starlings with male‐like doses of exogenous T resulted in increased bacterial infection compared to controls (De Ridder et al., 2002). It is notable that a consistent negative relationship between T and immunity has been observed in starlings, regardless of the source of

96

GREGORY F. BALL ET AL.

Fig. 6. (A) Anti‐KLH antibody titers relative to negative controls at 10 and 15 days postinjection. Error bars represent 95% confidence intervals. Asterisk ¼ p < 0.05 relative to controls. Open bars ¼ blank implants and filled bars ¼ T implants. (B) Log transformed values of the mean ( SE) change in web swelling (% baseline) 24 hr after injection with PHA. Asterisk ¼ p < 0.05 relative to controls (MB ¼ males with blank implants, MT ¼ males with T implants, FB ¼ females with blank implants, and FT ¼ females with T implants). Numbers below the x‐axis indicate the sample size for each group. [Adapted from Duffy, D. L., Bentley, G. E., Drazen, D. L., and Ball, G. F. (2000). Effects of testosterone on cell‐mediated and humoral immunity in non‐breeding adult European starlings. Behav. Ecol. 11(6), p. 658, Fig. 4, by permission of Oxford University Press.]

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

97

T (exogenous or endogenous) and using different measures of immunity (response to antigenic challenge or bacterial infection). However, given that two of the three aforementioned studies with starlings involved administration of T to birds that normally would not experience high T concentrations

Fig. 7. (A) Anti‐KLH antibody titers relative to negative controls (day 15 postinjection) of T‐treated birds as a function of exogenous T concentrations (ng/ml). The linear regression line has the equation y ¼ 2.450
.206x; r2 ¼ .318, p < .01; (B) Anti-KLH antibody titers relative to negative controls (averaged across days 10 and 15 postinjection) in relation to endogenous plasma T concentrations (ng/ml). The linear regression line has the equation y ¼ 3.92
1.40x; r2 ¼ 0.350. [Part A adapted from Duffy, D. L. Bentley, G. E., Drazen, D. L., and Ball, G. F. (2000). Effects of testosterone on cell‐mediated and humoral immunity in non‐breeding adult European starlings. Behav. Ecol. 11(6), p. 657, Fig. 3, by permission of Oxford University Press; part B: adapted from Duffy, D. L., and Ball, G. F. (2002). Song predicts immunocompetence in male European starlings, Proc. Roy. Soc. Lond. Ser. B‐Biol. Sci. 269(1493), p. 850, Fig. 2.]

98

GREGORY F. BALL ET AL.

(nonbreeding males and females), further study is required to corroborate the immunosuppressive effect of T under natural conditions. While the data from starlings thus far appear to support the predictions of ICHH, it remains possible that a more complex mechanism may be involved. For instance, artificial elevations of T have been demonstrated to induce concomitant increases in circulating corticosterone (CORT) in a number of bird species, including dark‐eyed juncos (Junco hyemalis, Casto et al., 2001; Klukowski et al., 1997), house sparrows (Passer domesticus, Evans et al., 2000; Poiani et al., 2000), song sparrows (Owen‐Ashley et al., 2004), and starlings (Duffy et al., 2000). Chronic elevation of CORT (e.g., days or weeks) generally has been demonstrated to be immunosuppressive (Buchanan, 2000). Thus, the immunosuppressive effects of T treatment observed in the aforementioned studies could in fact be due to the concomitant T‐induced rise in CORT. One study in house sparrows indicates that when the immunosuppressive effects of CORT are statistically controlled for, T has an enhancing effect on immune responses (Evans et al., 2000). However, the relationship between naturally occurring fluctuations in T and CORT remains poorly understood. Most studies have found an inverse relationship between T and CORT, primarily via a stress‐induced rise in CORT simultaneous with a decrease in T (Knol, 1991; Silverin, 1998). For example, following mate‐choice trials in which males competed for females, chosen males were found to have elevated CORT and decreased T concentrations (Sorenson et al., 1997). In contrast, a positive correlation between naturally occurring seasonal fluctuations in T and baseline CORT has been reported in dark‐eyed juncos (Deviche et al., 2000). And yet, evidence in house sparrows indicates that, while exogenous T increased CORT concentrations in the postbreeding season, a negative correlation was seen with endogenous T and baseline CORT concentrations in intact males during the breeding season (Buchanan et al., 2003a). Thus, various factors must be considered when assessing physiological and behavioral correlates of one or both of these steroids, such as the timing of measurement (e.g., time of year, baseline or stress induced), and whether the observed relationship involves naturally occurring endocrine responses to environmental stimuli or is the result of manipulation of either T or CORT. The issue is further complicated by the observation that T also can cause an increase in corticosteroid‐binding globulin (CGB), which binds to both T and CORT (although CORT has a greater affinity; Deviche et al., 2001; Schoech et al., 1999). The complexity of the interactions between T and CORT remains a challenge to understanding the relationship between the endocrine and immune systems. Additional studies are needed in which factors, such as dose and duration of hormone treatment, social milieu, and

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

99

the limitations of various immune measures, are considered (Braude et al., 1999; Poiani et al., 2000). Furthermore, new hypotheses that have been proposed to explain the relationships among hormones, immunocompetence, and sexually selected signals cannot be ruled out by the existing data and deserve closer attention (Braude et al., 1999; Buchanan, 2000; Poiani et al., 2000). In sum, these studies provide strong evidence in support of the hypothesis that features of male starling song that are important to females during mate selection are indicators of immune function. At this point, however, the mechanism underlying this relationship remains unclear. It has become increasingly evident that the mechanism as originally proposed by the ICHH is too simplistic to explain the relationship between song and immune responsiveness. Though T is likely involved to some degree, more recent efforts have indicated that CORT probably plays a pivotal role. Accordingly, new hypotheses have been proposed in which the stress response is a key factor underlying the development or expression of condition‐dependent signals (Braude et al., 1999; Buchanan, 2000; Nowicki et al., 1998, 2002a; Poiani et al., 2000). One such hypothesis is the ‘‘developmental stress hypothesis’’ (Nowicki et al., 1998, 2002a). 2. Developmental Stress Hypothesis Brain structures important for song learning and production, and their interconnections, develop early in life within weeks after hatching (Catchpole and Slater, 1995). It is during this time that young birds are most likely to experience physiological stress, for example, undernourishment (Nowicki et al., 1998). According to the developmental stress hypothesis (originally termed the ‘‘nutritional stress hypothesis’’), stressors, such as inadequate nutrition during the period of development of the song control system, result in a delayed or decreased ability to learn song, for which birds may not be able to fully compensate (Nowicki et al., 1998, 2002a). Thus, in adulthood, learned song features may reliably reflect the extent of or the response to stressors experienced early in life. By choosing males based on learned song features, such as repertoire size, females are able to select those males who were better able to cope with early developmental stress and, thus, are presumably of superior genotypic quality (Nowicki et al., 1998, 2002a). Recent reports in starlings add to a growing body of literature in support of the developmental stress hypothesis (Buchanan et al., 2003b; Spencer et al., 2004). Starlings that experienced unpredictable short‐term food deprivation on a daily basis for 3 months after fledging showed a decrease in the quality and quantity of song produced the following spring compared to controls that received an ad libitum food supply (Buchanan et al., 2003b;

100

GREGORY F. BALL ET AL.

Spencer et al., 2004). Birds that received the unpredictable food supply spent less time singing, showed a longer latency to begin singing, sang fewer and shorter song bouts, and had smaller repertoires compared to controls (Buchanan et al., 2003b; Spencer et al., 2004; Fig. 8). Furthermore, birds in the unpredictable food supply group tended to have higher peak CORT concentrations than controls (Buchanan et al., 2003b). This again suggests that this stress hormone could be acting as a physiological constraint on the development or expression of sexually selected song features. Conversely, peak CORT during development was positively correlated with repertoire size as adults (Spencer et al., 2004). While the role of CORT in song development in starlings remains unclear, a study of zebra finches demonstrated that experimental manipulation of

Fig. 8. Song production of males in two treatment groups: unpredictable food supply and ad libitum‐fed controls. (A) Mean total time spent singing (s/hr); (B) mean latency to start singing; (C) mean number of songs bouts/hr; and (D) mean song bout duration. [Adapted from Buchanan, K. L., Spencer, K. A., Goldsmith, A. R., and Catchpole, C. K. (2003b). Song as an honest signal of past developmental stress in the European starling (Sturnus vulgaris). Proc. Roy. Soc. Lond. Ser. B‐Biol. Sci. 270, p. 1154, Fig. 4, by permission of The Royal Society.]

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

101

CORT in nestlings affected adult song output in a similar manner as food restriction (Spencer et al., 2003). The core of the developmental stress hypothesis is that variation in exposure or response to stressors early in life results in variation in neural development, specifically within the song control system, which is reflected in variation of learned song features as an adult (Nowicki and Searcy, 2004; Nowicki et al., 2002). One assumption of this hypothesis is that the morphology of the song control system correlates with production of learned song features. In agreement with this assumption is the observation that the volumes of both RA and HVC are correlated with song bout length and HVC is correlated with song rate in male starlings (Bernard et al., 1996; Sartor and Ball, 2005), as mentioned earlier in this chapter. However, because this evidence is correlational, the direction of causality cannot be determined. Although Buchanan et al. (2003b) did not examine the song control system in their study, it would be interesting to learn whether manipulation of developmental stressors early in life result in subsequent differential development of the song control system. In swamp sparrows (Melospiza georgiana), Nowicki et al. (2002) demonstrated that a restricted diet early in development led to smaller volumes of HVC and RA in adulthood; however, when differences in telencephalon volume were controlled only the difference in RA volume remained significant. At the time of this writing, only a handful of studies have tested the developmental stress hypothesis but thus far the evidence is promising. Importantly, the ICHH and the developmental stress hypothesis are not mutually exclusive. In addition to neural development, stress experienced early in development also can affect immune function. In fact, a study of starling nestlings indicates that the effect of early developmental stress on immune responses is sex specific and resource dependent (Chin et al., 2005). Studying natural variation in a wild population of starlings, Chin et al. (2005) found that when resource availability was low, thus lowering parental provisioning rate and chick growth rate, larger brood size negatively affected cell‐mediated immune responsiveness to PHA in male nestlings, whereas no effect was observed in female nestlings. In a resource‐rich environment, however, brood size had no effect on immune responses of nestlings of either sex, suggesting that the trade‐off responsible for higher sensitivity to developmental stress among males compared to females may be compensated for under resource‐abundant conditions (Chin et al., 2005). This resource‐dependent effect of developmental stress echoes that found by Buchanan et al. (2003b) in which unpredictable food resources resulted in a negative relationship between immune responsiveness as nestlings and later song performance, while abundant food resources revealed a positive relationship. Furthermore, the sex‐specific nature of these relationships would be

102

GREGORY F. BALL ET AL.

Fig. 9 Schematic diagram illustrating interrelationships among song production in male starlings, the associated neuroanatomy of the song system that is positively correlated with song bout length, song preferences in female starlings that are related to biases in zenk expression in their auditory forebrain, and immunocompetence. Plasmas T is one physiological variable that is involved in coordinating these different traits. See text for more details.

expected if there is a trade‐off between the development of sexually‐selected characteristics and other energetically expensive processes (e.g., somatic growth, immune function). A hypothesis has recently been put forward that links immunity to neural development (Moller et al., 2005). Specifically, it has been hypothesized that parasitic infection can have negative consequences for neural development and learning. In turn, higher susceptibility to infection in males may result in greater investment in immune defense compared to females. The prediction that follows is that males that have evolved strong immune systems will be least likely to suffer from infection and therefore should be capable of developing larger brains (Moller et al., 2005). When Moller et al. (2005) compared the relative size of various body organs to brain size across different bird species, they found that relative size of organs involved in immune function (bursa of Fabricius and spleen) covaried positively with relative brain size. Furthermore, there was a significant positive correlation between the relative mass of immune defense organs and the degree of sexual dimorphism in brain size, indicating stronger selection pressure for males compared to females. In contrast, scant evidence indicated such covariation among the sizes of immune defense organs or brain size, and with the sizes of the heart or liver (Moller et al., 2005). 3. Sexy Son Hypothesis Given what we know about the relationships between multiple aspects of male quality (e.g., age, condition, immunocompetence, developmental stress, brain morphology) and song features that are important during mate

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

103

choice, it seems reasonable to conclude that song functions as a condition‐ dependent indicator in starlings. However, it is notable that one recent study offers support for a hypothesis that does not require a male’s attractiveness to be dependent on his condition. The ‘‘sexy son hypothesis’’ states that females benefit by mating with highly attractive (polygynous) males at the cost of reduced paternal care when their sons inherit those attractive qualities and thus have better mating success, providing their mothers with more grandchildren (Weatherhead, 1994; Weatherhead and Robertson, 1979, 1981). Gwinner and Schwabl (2005) reported that sons of polygynous starlings possessed superior competitive ability in contests for nest boxes and performed more female‐directed song compared to sons of monogamous males (Gwinner and Schwabl, 2005). However, there was only a nonsignificant tendency for sons of polygynous males to attract more females than sons of monogamous males in an aviary setting. The authors suggest that their hand‐rearing and housing conditions may have hindered adequate testing of female choice, leading to the lack of a significant effect. Importantly, the competitive advantages gained by sons of polygynous males could not be attributed to differences in maternal deposition of hormones into the eggs, though other early environmental effects during the nestling stage could not be ruled out. It is worth noting that although the sexy son hypothesis does not require attractiveness to be condition dependent, it does not preclude it either. In fact, it has been recently proposed that the Fisherian sexy son hypothesis and the quality‐indicator good genes’ models are complementary to one another and their relative weights depend on the costs associated with being choosy (Kokko, 2001; Kokko et al., 2002; Radwan, 2002). Taken together, the studies discussed herein strongly suggest that, by selecting males based on singing performance, choosy females are likely to obtain a variety of indirect fitness benefits. 4. Future Directions The next logical step in exploring the functional basis of starling song is to empirically establish the fitness benefits gained by choosy females that mate with males that display superior song performance. To accomplish this, it is necessary to turn more focus toward the offspring. For instance, do females paired with robust singers fledge more young (Forstmeier and Leisler, 2004; Gil and Slater, 2000; Reid et al., 2005; Sheldon et al., 1997)? Do their offspring experience mating or reproductive advantages compared to the offspring of nonchoosy females (Head et al., 2005)? Do the offspring of highly immunocompetent males exhibit superior disease resistance (Johnsen et al., 2000; Kleven and Lifjeld, 2004)?

104

GREGORY F. BALL ET AL.

In addition, more focus needs to be directed toward the mechanisms and evolution of female mate preferences. For example, how much variation exists among (and within) females in their preferences and what maintains that variation (Forstmeier and Birkhead, 2004)? Earlier in this chapter, we discussed how recent exposure to short song bouts versus long song bouts can alter neural responses to novel songs of different bout lengths (Sockman et al., 2002). How does this neural plasticity translate into behavioral responses to song? What factors influence plasticity in mate choice and what are the fitness advantages for various alternative strategies? Furthermore, why have particular mechanisms evolved as opposed to some other alternative processes? For example, a few different hypotheses have been put forward to explain the function of steroid‐induced immunosuppression (Besedovsky and DelRey, 1996; Derting and Virk, 2005; McEwen et al., 1997; Muehlenbein and Bribiescas, 2005; Raberg et al., 1998; Wedekind and Folstad, 1994). Could evolution has co‐opted these processes and applied them to sexually selected traits? The answers to these questions would broaden what we know about the functional significance of song in starlings.

VII. PUTTING IT ALL TOGETHER: SONG PRODUCTION/PERCEPTION AND HORMONES In this chapter, we have tried to tie together a number of different types of findings about the mechanism and function of song in European starlings. In this concluding section we make a few general points. First, one theme that emerges from this work is how a basic behavioral observation, namely that male starling song of a particular type (i.e., long‐bout song) is attractive to female starlings and is used for mate choice, can guide an entire research program in valuable ways. This fact along with the related observation that male song sung in the breeding season is responsive to the presence of a female, while song produced outside the breeding season is not has guided an entire series of mechanistic studies in valuable ways. For example, the regulation of male starling song by the steroid hormone T can only be explained if one realizes that female‐directed song is enhanced by T but not song produced for other reasons outside the breeding context (Pinxten et al., 2002). The importance of long‐bout song for female choice has also guided studies on song production and perception. Variation in the volume of key song nuclei HVC and RA in starlings positively correlates with song bout length in males older than 1 year (Bernard et al., 1996).

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

105

The expression of the IEG zenk in the auditory telencephalon of female starlings is best explained if one considers variation in song bout length (Gentner et al., 2001; Sockman et al., 2002, 2005). Thus, aspects of both male neuroanatomy and female neural responses in starlings can be explained by taking account of the functional significance of behavior. Finally, the potential functional significance of song as a predictor of male immunocompetence was again revealed when simple immune measures were correlated with the appropriate measures of song performance (Duffy and Ball, 2002). The challenge now is to understand the web of causality that explains these intriguing correlations between song behavior and various aspects of brain and physiology. The gonadal hormone T is involved in the interplay between female choice and various aspects of male behavior, neuroanatomy, and physiology in that it regulates to some degree all these aspects of the male phenotype in starlings. As just noted, T stimulates male song behavior during the breeding season (Pinxten et al., 2002) and it promotes the growth of song nuclei such as HVC and RA (Bernard and Ball, 1995, 1997). It also inhibits cell‐mediated and humoral immune measures in male starlings, although this may be through increased CORT secretion (Duffy et al., 2000). What is needed now is to take neuroendocrine studies to the next level of sophistication. The arrow of causation needs to be elucidated to explain these correlations. Is variation in the volume of song nuclei, such as HVC and RA, permissive for long‐bout starling song or a consequence of engaging in such song (Sartor and Ball, 2005). How and where does T act to enhance only male‐directed song? Is that an effect of its action in the preoptic area and other areas related to song motivation (Riters and Ball, 1999; Riters et al., 2000)? Is the fact that female‐directed song only occurs during the breeding season related merely to seasonal variation in T concentrations or does the central control of song as it relates to stimulus context also change in a seasonal fashion? For example, this seasonal change in the stimulus that elicits male song could be related to seasonal changes in the volume of the preoptic nucleus or perhaps to dramatic seasonal changes in hormone receptor types that have been described in the song system of starlings (Bentley and Ball, 2000). We know in zebra finches that song directed at females exhibits a very different pattern of IEG expression than song that is undirected (Jarvis et al., 1998). This context effect is regulated at least in part by ascending noradrenergic projections to song nuclei such as area X (Castelino and Ball, 2005). Are these modulatory inputs to the song system changing seasonally and resetting the way in which song is used? It is clear that an entirely new generation of mechanistic studies is now in order.

106

GREGORY F. BALL ET AL.

If these continue to be guided by a sensitivity to behavior and its functional context, the future for the neuroethological approach is promising indeed.

VIII. SUMMARY Male European starlings produce a long complex song that is used by females to choose mates in a breeding context. Females tend to prefer males that produce songs organized into long bouts. The production of song is controlled by a specialized neural circuit called the song control system that is present in all oscine songbirds. Using long‐bout song as a tool allowed us to explicate several aspects of the neural control of song perception and production in starlings. Variation in the volume of key song control nuclei, such as HVC and RA, correlates positively with variation in song bout length in adult starlings. IEG expression in the auditory forebrain of females is enhanced in response to long‐bout song as compared to conspecific song organized into short bouts. Long‐bout song as well as high rates of singing also predicts variation in measures of cellular and humoral immunity in male starlings. The cause and effect relationships among these variables still need to be elucidated. But T enhances the size of song control nuclei in starlings that correlate with the production of long‐bout song preferred by females. Female choice therefore drives aspects of male physiology that results in long‐bout song. In the female auditory forebrain, there is evidence for physiological responses to song as measured by IEG expression and electrophysiology that are tuned to aspects of male song that they prefer. However, these neural biases exhibit a plasticity that allows the females to modify their neural responses and memorize male songs as a function of the types of songs they encounter in their local social environment. These studies illustrate how behavioral investigations of functional significance can provide tools to implement a neuroethological investigation of behavioral mechanisms (see Fig. 9).

Acknowledgments We thank the NSF (IBN 9028803) and the NIH (NS R01 35467) for support over the past 15 years of our research program on starling song. We also thank Stew Hulse for his many contributions to perceptual aspects of this project and Randy Nelson for his help and advice on aspects of this project related to immune function.

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

107

REFERENCES Adamo, S. A. (2004). How should behavioural ecologists interpret measurements of immunity? Anim. Behav. 68, 1443–1449. Adamo, S. A., and Spiteri, R. J. (2005). Female choice for male immunocompetence: When is it worth it? Behav. Ecol. 16(5), 871–879. Adret‐Hausberger, M., and Jenkins, P. F. (1988). Complex organization of the warbling song in starlings. Behaviour 107, 138–156. Andersson, M. (1994). ‘‘Sexual Selection.’’ Princeton University Press, Princeton, NJ. Appeltants, D., Ball, G. F., and Balthazart, J. (2001). The distribution of tyrosine hydroxylase in the canary brain: Demonstration of a specific and sexually dimorphic catecholaminergic innervation of the telencephalic song control nuclei. Cell Tissue Res. 304, 237–259. Appeltants, D., Del Negro, C., and Balthazart, J. (2002). Noradrenergic control of auditory information processing in female canaries. Behav. Brain Res. 133, 221–235. Arnold, A. P. (1975). The effects of castration and androgen replacement on song, courtship, and aggression in zebra finches. J. Exp. Zool. 191, 309–326. Badyaev, A. V., and Qvarnstro¨m, A. (2002). Putting sexual traits into the context of an organism: A life‐history perspective in studies of sexual selection. Auk 119, 301–310. Bailey, D. J., and Wade, J. (2003). Differential expression of the immediate early genes FOS and ZENK following auditory stimulation in the juvenile male and female zebra finch. Mol. Brain Res. 116, 147–154. Bailey, D. J., Rosebusch, J. C., and Wade, J. (2002). The hippocampus and caudomedial neostriatum show selective responsiveness to conspecific song in the female zebra finch. J. Neurobiol. 52, 43–51. Bakin, J. S., and Weinberger, N. M. (1996). Induction of a physiological memory in the cerebral cortex by stimulation of the nucleus basalis. Proc. Natl. Acad. Sci. USA 93, 11219–11224. Ball, G. F. (1990). Chemical neuroanatomical studies of the steroid‐sensitive songbird vocal control system: A comparative approach. In ‘‘Hormones, Brain and Behaviour in Vertebrates. 1. Sexual Differentiation, Neuroanatomical Aspects, Neurotransmitters and Neuropeptides’’ (J. Balthazart, Ed.), Vol. 8, pp. 148–167. Karger, Basel. 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. (1999). Neuroendocrine basis of seasonal changes in vocal behavior among songbirds. In ‘‘The Design of Animal Communication’’ (M. Hauser and M. Konishi, Eds.), pp. 213–253. MIT Press, Cambridge, MA. Ball, G. F., and Balthazart, J. (2001). Ethological concepts revisited: Immediate early gene induction in response to sexual stimuli in birds. Brain Behav. Evol. 57, 252–270. Ball, G. F., and Bentley, G. E. (2000). Neuroendocrine mechanisms mediating the photoperiodic and social regulation of seasonal reproduction in birds. In ‘‘Reproduction in Context: Social and Environmental Influences on Reproductive Physiology and Behavior’’ (K. Wallen and J. E. Schneider, Eds.), pp. 129–158. MIT Press, Cambridge. Ball, G. F., and Gentner, T. Q. (1998). They’re playing our song: Gene expression and birdsong perception. Neuron 21, 271–274. Ball, G. F., Faris, P. L., Hartman, B. K., and Wingfield, J. C. (1988). Immunohistochemical localization of neuropeptides in the vocal control regions of two songbird species. J. Comp. Neurol. 268, 171–180. Ball, G. F., Riters, L. V., and Balthazart, J. (2002). Neuroendocrinology of song behavior and avian brain plasticity: Multiple sites of action of sex steroid hormones. Front. Neuroendocrinol. 23, 137–178.

108

GREGORY F. BALL ET AL.

Balthazart, J., Baillien, M., Cornil, C. A., and Ball, G. F. (2004). Preoptic aromatase modulates male sexual behavior: Slow and fast mechanisms of action. Physiol. Behav. 83, 247–270. Bao, S., Chan, V. T., and Merzenich, M. M. (2001). Cortical remodelling induced by activity of ventral tegmental dopamine neurons. Nature 412, 79–83. Bentley, G. E., and Ball, G. F. (2000). Photoperiod‐dependent and ‐independent regulation of melatonin receptors in the forebrain of songbirds. J. Neuroendocrinol. 12, 745–752. Benton, S., Nelson, D. A., Marler, P., and DeVoogd, T. J. (1998). Anterior forebrain pathway is needed for stable song expression in adult male white‐crowned sparrows (Zonotrichia leucophrys). Behav. Brain Res. 96, 135–150. Bernard, D. J., and Ball, G. F. (1995). 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., and Ball, G. F. (1997). Photoperiodic condition modulates the effects of testosterone on song control nuclei volumes in male European starlings. Gen. Comp. Endocrinol. 105, 276–283. 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 volumes of song control nuclei in male European starlings. J. Neurobiol. 30, 329–339. Besedovsky, H. O., and Del Rey, A. (1996). Immune‐neuro‐endocrine interactions: Facts and hypotheses. Endocr. Rev. 17, 64–102. Bigalke‐Kunz, B., Rubsamen, R., and Dorrscheidt, G. (1987). Tonotopic organization and functional characterization of the auditory thalamus in a songbird, the European starling. J. Comp. Physiol. A 161, 255–265. Birkhead, T. R., Atkin, L., and Moller, A. P. (1987). Copulation behaviour in birds. Behaviour 101, 101–138. Bohner, J., and Veit, F. (1993). Song structure and patterns of wing movement in the European starling (Sturnus‐vulgaris). J. fu¨r Ornithologie 134, 309–315. Bolhuis, J. J. (2005). Function and mechanism in neuroecology: Looking for clues. Anim. Biol. 55, 457–490. Bolhuis, J. J., and MacPhail, E. M. (2001). A critique of the neurecology of learning and memory. Trends Cogn. Sci. 5, 426–433. Bolhuis, J. J., and Eda‐Fujiwara, H. (2003). Bird brains and songs: Neural mechanisms of birdsong perception and memory. Anim. Biol. 53, 129–145. Bolhuis, J. J., Zijlstra, G. G. O., Den Boer‐Visser, A. M., and Van der Zee, E. A. (2000). Localized neuronal activation in the zebra finch brain is related to the strength of song learning. Proc. Natl. Acad. Sci. USA 97, 2282–2285. Bolhuis, J. J., Hetebrij, E., Den Boer‐Visser, A. M., DeGroot, J. H., and Zijlstra, G. G. O. (2001). Localized immediate early gene expression related to the strength of song learning in socially reared zebra finches. Eur. J. Neurosci. 13, 2165–2170. Bonke, D., Scheich, H., and Langner, G. (1979). Responsiveness of units in the auditory neostriatum of the Guinea fowl (Numida meleagris) to species‐specific calls and synthetic stimuli. I. Tonotopy and functional zones. J. Comp. Physiol. 132, 243–255. Bottjer, S. W., and Johnson, F. (1997). Circuits, hormones, and learning: Vocal behavior in songbirds. J. Neurobiol. 33, 602–618. Bottjer, S. W., Miesner, E. A., and Arnold, A. P. (1984). Forebrain lesions disrupt development but not maintenance of song in passerine birds. Science 224, 901–903.

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

109

Bozon, B., Davis, S., and Laroche, S. (2002). Regulated transcription of the immediate‐early gene Zif268: Mechanisms and gene dosage‐dependent function in synaptic plasticity and memory formation. Hippocampus 12, 570–577. Bozon, B., Davis, S., and Laroche, S. (2003). A requirement for the immediate early gene zif 268 in reconsolidation of recognition memory after retrieval. Neuron 40, 695–701. Brainard, M. S. (2004). Contributions of the anterior forebrain pathway to vocal plasticity. In ‘‘Behavioral Neurobiology of Birdsong’’ (H. P. Zeigler and P. Marler, Eds.), Vol. 1016, pp. 377–394. The New York Academy of Sciences, New York, NY. Brainard, M. S., and Doupe, A. J. (2000). Interruption of a basal ganglia‐forebrain circuit prevents plasticity of learned vocalizations. Nature 404, 762–766. Braude, A., Tang‐Martinez, Z., and Taylor, G. T. (1999). Stress, testosterone, and the immunoredistribution hypothesis. Behav. Ecol. 10, 345–350. Brenowitz, E. A. (1991). Altered perception of species‐specific song by female birds after lesions of a forebrain nucleus. Science 251, 303–305. Brenowitz, E. A. (1997). Comparative approaches to the avian song system. J. Neurobiol. 33, 517–531. Brenowitz, E. A., Margoliash, D., and Nordeen, K. W. (1997). An introduction to birdsong and the avian song system. J. Neurobiol. 33, 495–500. Buchanan, K. L. (2000). Stress and the evolution of condition‐dependent signals. Trends Ecol. Evol. 15, 156–160. Buchanan, K. L., Evans, M. R., and Goldsmith, A. R. (2003a). Testosterone, dominance signalling and immunosuppression in the house sparrow, Passer domesticus. Behav. Ecol. Sociobiol. 55, 50–59. Buchanan, K. L., Spencer, K. A., Goldsmith, A. R., and Catchpole, C. K. (2003b). Song as an honest signal of past developmental stress in the European starling (Sturnus vulgaris). Proc. R. Soc. Lond. B 270, 1149–1156. Burt, J. M., Lent, K. L., Beecher, M. D., and Brenowitz, E. A. (2000). Lesions of the anteriori forebrain song control pathway in female canaries affect song perception in an operant task. J. Neurobiol. 42, 1–13. Calford, M. B. (2002). Dynamic representational plasticity in sensory cortex. Neuroscience 111, 709–738. Camhi, J. M. (1984). ‘‘Neuroethology: Nerve Cells and the Natural Behavior of Animals.’’ Sinauer Associates, Sunderland, MA. Capsius, B., and Leppelsack, H. (1999). Reponse patterns and their relationship to frequency analysis in auditory forebrain centers of a songbird. Hear Res. 136, 91–99. Cardin, J. A., and Schmidt, M. F. (2004). Noradrenergic inputs mediate state dependence of auditory responses in avian song system. J. Neurosci. 24, 7745–7753. Carr, C. E. (1992). Evolution of the central auditory system in reptiles and birds. In ‘‘The Evolutionary Biology of Hearing’’ (D. B. Webster, R. R. Fay, and A. N. Popper, Eds.), pp. 511–544. Springer-Verlag, New York. Carr, C. E., and Code, R. A. (2000). The central auditory system of reptiles and birds. In ‘‘Comparative Hearing: Birds and Reptiles’’ (R. J. Dooling, R. R. Fay, and A. N. Popper, Eds.), pp. 197–248. Springer-Verlag, New York. Castelino, C. B., and Ball, G. F. (2005). A role for norepinephrine in the regulation of context‐ dependent ZENK expression in male zebra finches (Taeniopygia guttata). Eur. J. Neurosci. 21, 1962–1972. Casto, J. M., Nolan, V., and Ketterson, E. D. (2001). Steroid hormones and immune function: Experimental studies in wild and captive dark‐eyed juncos (Junco hyemalis). Am. Nat. 157, 408–420.

110

GREGORY F. BALL ET AL.

Catchpole, C. K., and Slater, P. J. B. (1995). ‘‘Bird Song: Biological Themes and Variations.’’ Cambridge University Press, Cambridge. Chaiken, M., Bo¨hner, J., and Marler, P. (1993). Song acquisition in European starlings, Sturnus vulgaris: A comparison of the songs of live‐tutored, tape‐tutored, untutored, and wild‐ caught males. Anim. Behav. 46, 1079–1090. Charlier, T. D., Ball, G. F., and Balthazart, J. (2005). Inhibition of steroid receptor coactivator‐ 1 blocks estrogen and androgen action on male sexual behavior and associated brain plasticity. J. Neurosci. 25, 906–913. Cheng, H. Y., and Clayton, D. F. (2004). Activation of habituation of extracellular signal‐ regulated kinase phosphorylation in zebra finch auditory forebrain during song presentation. J. Neurosci. 24, 7503–7513. Chew, S. J., Mello, C., Nottebohm, F., Jarvis, E. D., and Vicario, D. S. (1995). Decrements in auditory responses to a repeated conspecific song are long‐lasting and require two periods of protein syntheis in the songbird forebrain. Proc. Natl. Acad. Sci. USA 92, 3406–3410. Chin, E. H., Love, O. P., Clark, A. M., and Williams, T. D. (2005). Brood size and environmental conditions sex‐specifically affect nestling immune response in the European starling Sturnus vulgaris. J. Avian Biol. 36, 549–554. Cirelli, C., and Tononi, G. (2004). Locus ceruleus control of state‐dependent gene expression. J. Neurosci. 24, 5410–5419. Clayton, D. F. (2000). The genomic action potential. Neurobiol. Learn. Mem. 74, 185–216. Cox, P. R. (1944). A statistical investigation into bird‐song. Brit. Birds 38, 3–9. Cuthill, I., and Hindmarsh, A. (1985). Increase in starling song activity with the removal of mate. Anim. Behav. 33, 326–327. Dave, A. S., and Margoliash, D. (2000). Song replay during sleep and computational rules for sensorimotor vocal learning. Science 290, 812–816. Dave, A. S., Yu, A. C., and Margoliash, D. (1998). Behavioral state modulation of auditory activity in a vocal motor system. Science 282, 2250–2254. Dawson, A., King, V. M., Bentley, G. E., and Ball, G. F. (2001). Photoperiodic control of seasonality in birds. J. Biol. Rhythms 16, 365–380. De Ridder, E., Pinxten, R., and Eens, M. (2000). Experimental evidence of a testosterone‐ induced shift from paternal to mating behavior in a facultatively polygynous songbird. Behav. Ecol. Sociobiol. 49, 24–30. De Ridder, E., Pinxten, R., Mees, V., and Eens, M. (2002). Short‐ and long‐term effects of male‐like concentrations of testosterone on female European starlings (Sturnus vulgaris). Auk 119, 487–497. Del Negro, C., Gahr, M., Leboucher, G., and Kreutzer, M. (1998). The selectivity of sexual responses to song displays: Effects of partial chemical lesion of the HVC in female canaries. Behav. Brain Res. 96, 151–159. Derting, T. L., and Virk, M. K. (2005). Positive effects of testosterone and immunochallenge on energy allocation to reproductive organs. J. Comp. Physiol. B Biochem. Syst. Environ. Physiol. 175, 543–556. Deviche, P., Wingfield, J. C., and Sharp, P. J. (2000). Year‐class differences in the reproductive system, plasma prolactin and corticosterone concentrations, and onset of prebasic molt in male dark‐eyed juncos (Junco hyemalis) during the breeding period. Gen. Comp. Endocrinol. 118, 425–435. Deviche, P., Breuner, C., and Orchinik, M. (2001). Testosterone, corticosterone, and photoperiod interact to regulate plasma levels of binding globulin and free steroid hormone in dark‐eyed juncos, Junco hyemalis. Gen. Comp. Endocrinol. 122, 67–77. DeVoogd, T. J. (1991). Endocrine modulation of the development and adult function of the avian song system. Psychoneuroendocrinology 16, 41–66.

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

111

Dewsbury, D. A. (1992). On the problems studied in ethology, comparative psychology, and animal behavior. Ethology 92, 89–107. Dewsbury, D. A. (1999). The proximate and the ultimate: Past, present, and future. Behav. Process. 46, 189–199. Doupe, A. J., Solis, M. M., Kimpo, R., and Boettiger, C. A. (2004). Cellular, circuit, and synaptic mechanisms in song learning. In ‘‘Behavioral Neurobiology of Birdsong’’ (H. P. Zeigler and P. Marler, Eds.), Vol. 1016, pp. 495–523. The New York Academy of Sciences, New York, NY. Doupe, A. S., Perkel, D. J., Reiner, A., and Stern, E. A. (2005). Birdbrains could teach basal ganglia research a new song. Trends Neurosci. 28, 353–363. Drickamer, L. C. (1998). Vertebrate behavior: Integration of proximate and ultimate causation. Amer. Zool. 38, 39–42. Duffy, D. L., and Ball, G. F. (2002). Song predicts immunocompetence in male European starlings (Sturnus vulagaris). Proc. R. Soc. Lond. B 269, 847–852. Duffy, D. L., Bentley, G. E., and Ball, G. F. (1999). Does sex or photoperiodic condition influence ZENK induction in response to song in European starlings? Brain Res. 844, 78–82. Duffy, D. L., Bentley, G. E., Drazen, D. L., and Ball, G. F. (2000). Effects of testosterone on cell‐mediated and humoral immunity in non‐breeding adult European starlings. Behav. Ecol. 11, 654–662. Eberhardt, L. S. (1994). Oxygen consumption during singing by male Carolina Wrens (Thryothorus ludovicianus). Auk 111, 124–130. Eda‐Fujiwara, H., Satoh, R., Bolhuis, J. J., and Kimura, T. (2003). Neuronal activation in female budgerigars is localized and related to male song complexity. Eur. J. Neurosci. 17, 149–154. Eens, M. (1997). Understanding the complex song of the European starling: An integrated ethological approach. Adv. Study Anim. Behav. 26, 355–434. 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., Pinxten, M., and Verheyen, R. F. (1989). Temporal and sequential organization of song bouts in the European starling. Ardea 77, 75–86. Eens, M., Pinxten, M., 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. (1993). Function of the song and song repertoire in the European starling (Sturnus vulgaris): An aviary experiment. Behaviour 125, 51–66. Eens, M., Pinxten, R., and Verheyen, F. R. (1994). Variation in singing activity during the breeding cycle of the European starling Sturnus vulgaris. Belg. J. Zool. 124, 167–174. Evans, M. R., Goldsmith, A. R., and Norris, S. R. A. (2000). The effects of testosterone on antibody production and plumage coloration in male house sparrows (Passer domesticus). Behav. Ecol. Sociobiol. 47, 156–163. Farries, M. A. (2004). The avian song system in a comparative perspective. In ‘‘Behavioral Neurobiology of Birdsong’’ (H. P. Zeigler and P. Marler, Eds.), Vol. 1016, pp. 61–76. The New York Academy of Sciences, New York, NY. Feare, C. J. (1984). ‘‘The Starling.’’ Oxford University Press, Oxford.

112

GREGORY F. BALL ET AL.

Fisher, R. A. (1930). ‘‘The Genetical Theory of Natural Selection.’’ Oxford University Press, Oxford. Fischer, T. M., Blazis, D. E., Priver, N. A., and Carew, T. J. (1997). Metaplasticity at identified inhibitory synapses in Aplysia. Nature 389, 860–865. Fleischmann, A., Havalby, O., Jensen, V., Strekalova, T., Zacher, C., Layer, L. E., Kvello, A., Reschke, M., Spanagel, R., Sprengel, R., Wagner, E. F., and Gass, P. (2003). Imparied long‐term memory and NR2A‐type NMDA receptor‐dependent synaptic plasticity in mice lacking c‐FOS in the CNS. J. Neurosci. 23, 9116–9122. Folstad, I., and Karter, A. J. (1992). Parasites, bright males, and the immunocompetence handicap. Am. Nat 139, 603–622. Forstmeier, W., and Birkhead, T. R. (2004). Repeatability of mate choice in the zebra finch: Consistency within and between females. Anim. Behav. 68, 1017–1028. Forstmeier, W., and Leisler, B. (2004). Repertoire size, sexual selection, and offspring viability in the great reed warbler: Changing patterns in space and time. Behav. Ecol. 15, 555–563. Fortune, E. S., and Margoliash, D. (1992). Cytoarchitectonic organization and morphology of cells in the field L complex in male zebra finches (Taeniopygia guttata). J. Comp. Neurol. 325, 388–404. Freeberg, T. M., Duncan, S. D., Kast, T. L., and Enstrom, D. A. (1999). Cultural influences on female mate choice: An experimental test in cowbirds, Molothrus ater. Anim. Behav. 57, 421–426. Fusani, L., and Gahr, M. (2005). Hormonal influence on song structure and organization: The role of estrogen. Neuroscience 138, 939–946. Garamszegi, L. Z. (2005). Bird song and parasites. Behav. Ecol. Sociobiol. 59, 167–180. Gehr, D. D., Komiya, H., and Eggermont, J. J. (2000). Neuronal responses in cat primary auditory cortex to natural and altered species‐specific calls. Hear Res. 150, 27–42. Gentner, T. Q., and Ball, G. F. (2005). A neuroethological perspective on perception of vocal communication signals. In ‘‘The Handbook of Speech Perception’’ (D. B. Pison and R. E. Remez, Eds.). Blackwell Publishing, Malden, MA. Gentner, T. Q., and Hulse, S. H. (1998). Perceptual mechanisms for individual vocal recognition in European starlings, Sturnus vulgaris. Anim. Behav. 56, 579–594. Gentner, T. Q., and Hulse, S. H. (2000a). Female European starling preference and choice for variation in conspecific male song. Anim. Behav. 59, 443–458. Gentner, T. Q., and Hulse, S. H. (2000b). Perceptual classification based on the component structure of song in European starlings. J. Acoust. Soc. Am. 107, 3369–3381. Gentner, T. Q., and Margoliash, D. (2003). Neuronal populations and single cells representing learned auditory objects. Nature 424, 669–674. Gentner, T. Q., Hulse, S. H., Bentley, G. E., and Ball, G. F. (2000). Individual vocal recognition and the effect of partial lesions to HVc on discrimination, learning, and categorization of conspecific song in adult songbirds. J. Neurobiol. 42, 117–133. Gentner, T. Q., Hulse, S. H., Duffy, D., and Ball, G. F. (2001). Response biases in auditory forebrain regions of female songbirds following exposure to sexually relevant variation in male song. J. Neurobiol. 46, 48–58. Gentner, T. Q., Hulse, S. H., and Ball, G. F. (2004). Functional differences in forebrain auditory regions during learned vocal recognition in songbirds. J. Comp. Physiol. A 190, 1001–1010. Gil, D., and Gahr, M. (2002). The honesty of bird song: Multiple constraints for multiple traits. Trends Ecol. Evol. 17, 133–141. Gil, D., and Slater, P. J. B. (2000). Multiple song repertoire characteristics in the willow warbler (Phylloscopus trochilus): Correlations with female choice and offspring viability. Behav. Ecol. Sociobiol. 47, 319–326.

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

113

Gilbert, C. D., Sigman, M., and Crist, R. E. (2001). The neural basis of perceptual learning. Neuron 31, 681–697. Gleeson, D. J., Blows, M. W., and Owens, I. P. F. (2005). Genetic covariance between indices of body condition and immunocompetence in a passerine bird. BMC Evol. Biol. 5, 6. Goelet, P., Castelucci, V., Schachter, S., and Kandel, E. (1986). The long and the short of long‐ term memory: A molecular framwork. Nature 332, 419–422. Grace, J. A., Amin, N., Singh, N. C., and Theunissen, F. E. (2003). Selectivity for conspecific song in the zebra finch auditory forebrain. J. Neurophysiol. 89, 472–487. Grafen, A. (1990). Sexual selection unhandicapped by the Fisher process. J. Theor. Biol. 144, 473–516. Guzowski, J. F. (2002). Insights into immediate‐early gene function in hippocampal memory consolidation using anitsense oligonucleotide and fluorescent imaging approaches. Hippocampus 12, 86–104. Gwinner, H., and Schwabl, H. (2005). Evidence for sexy sons in European starlings (Sturnus vulgaris). Behav. Ecol. Sociobiol. 58(4), 375–382. Hamilton, W. D., and Zuk, M. (1982). Heritable true fitness and bright birds: A role for parasites? Science 218, 384–387. Harding, C. F. (2004). Hormonal modulation of singing: Hormonal modulation of the songbird brain and singing behavior. In ‘‘Behavioral Neurobiology of Birdsong’’ (H. P. Zeigler and P. Marler, Eds.), Vol. 1016, pp. 524–539. New York Academy of Sciences, New York, NY. Harding, C. F., Sheridan, K., and Walters, M. J. (1983). Hormonal specificity and activation of sexual behavior in male zebra finches. Horm. Behav. 17, 111–133. Harding, C. F., Walters, M. J., Collado, D., and Sheridan, K. (1988). Hormonal specificity and activation of social behavior in male red‐winged blackbirds. Horm. Behav. 22, 402–418. Hausberger, M. (1997). Social influences on song acquisition and sharing in the European starling (Sturnus vulgaris). In ‘‘Social Influences on Vocal Development’’ (C. Snowden and M. Hausberger, Eds.), pp. 128–156. Cambridge University Press, Cambridge, UK. Hausberger, M., and Cousillas, H. (1995). Categorization in birdsong: From behavioural to neuronal responses. Behav. Process. 35, 83–91. Hau¨sler, U. (1996). Measurement of short‐time spatial activity patterns during auditory stimulation in the starling. In ‘‘Acoustical Signal Processing in the Central Auditory System’’ (J. Syka, Ed.). Plenum Press, New York. Head, M. L., Hunt, J., Jennions, M. D., and Brooks, R. (2005). The indirect benefits of mating with attractive males outweigh the direct costs. PLoS. Biol. 3, 289–294. Hebets, E. A. (2003). Subadult experience influences adult mate choice in an arthropod: Exposed female wolf spiders prefer males of a familiar phenotype. Proc. Natl. Acad. Sci. USA 100, 13390–13395. Heimovics, S. A., and Riters, L. V. (2005). Immediate early gene activity in song control nuclei and brain areas regulating movtivation relates positively to singing behavior during, but not outside of, a breeding context. J. Neurobiol. 65, 207–224. Hernandez, A. M., and MacDougall‐Shackleton, S. A. (2004). Effects of early song experience on song preferences and song control and auditory brain regions in female house finches, (Carpodacus mexicanus). J. Neurobiol. 59, 247–258. Hessler, N. A., and Doupe, A. J. (1999a). Singing‐related neural activity in a dorsal forebrain‐ basal ganglia circuit of adult zebra finches. J. Neurosci. 19, 10461–10481. Hessler, N. A., and Doupe, A. J. (1999b). Social context modulates singing‐related neural activity in the songbird forebrain. Nat. Neurosci. 2, 209–211.

114

GREGORY F. BALL ET AL.

Hinde, R. A. (1982). ‘‘Ethology. Its Nature and Relations with Other Sciences.’’ Oxford University Press, New York. Hindmarsh, A. (1984). Vocal mimicry in starlings. Behaviour 90, 304–324. Hoelzel, A. R. (1986). Song characteristics and reponse to playback of male and female Robins Erithacus rubecula. Ibis 128, 115–127. Hoffman, G. E., Smith, M. S., and Verbalis, J. G. (1993). c‐Fos and related immediate early gene products as markers for neuronal activity in neuroendocrine systems. Front. Neuroendocrinol. 14, 173–213. Hunt, K. E., Hahn, T. P., and Wingfield, J. C. (1997). Testosterone implants increase song but not aggression in male Lapland longspurs. Anim. Behav. 54, 1177–1192. Jarvis, E. D. (2004). Brains and birdsong. In ‘‘Nature’s Music: The Science of Birdsong’’ (P. Marler and H. Slabbekoorn, Eds.), pp. 226–271. Elsevier, Amsterdam. Jarvis, E. D., and Mello, C. V. (2000). Molecular mapping of brain areas involved in parrot vocal communication. J. Comp. Neurol. 419, 1–31. Jarvis, E. D., and Nottebohm, F. (1997). Motor‐driven gene expression. Proc. Natl. Acad. Sci. USA 94, 4097–4102. Jarvis, E. D., Mello, C. V., and Nottebohm, F. (1995). Associative learning and stimulus novelty influence the song‐induced expression of an immediate early gene in the canary forebrain. Learn. Mem. 2, 62–80. Jarvis, E. D., Scharff, C., Grossman, M. R., Ramos, J. A., and Nottebohm, F. (1998). For whom the bird sings: Context‐dependent gene expression. Neuron 21, 775–788. Jarvis, E. D., Ribeiro, S., Da Silva, M. L., Ventura, D., Vielliard, J., and Mello, C. V. (2000). Behaviourally driven gene expression reveals song nuclei in hummingbird brain. Nature 406, 628–632. Jennions, M. D., and Petrie, M. (1997). Variation in mate choice and mating preferences: A review of causes and consequences. Biol. Rev. 72, 283–327. Jin, H., and Clayton, D. F. (1997). Localized changes in immediate‐early gene regulation during sensory and motor learning in zebra finches. Neuron 19, 1049–1059. Johnsen, A., Andersen, V., Sunding, C., and Lifjeld, J. T. (2000). Female bluethroats enhance offspring immunocompetence through extra‐pair copulations. Nature 406, 296–299. Johnstone, R. A. (1995). Sexual selection, honest advertisement and the handicap priniciple—reviewing the evidence. Biol. Rev. Camb. Philos. Soc. 70, 1–65. Jones, M. W., Errington, M. L., French, P. J., Fine, A., Bliss, T. V., Garel, S., Charnay, P., Bozon, B., Laroche, S., and Davis, S. (2001). A requirement for the immediate early gene Zif268 in the expression of late LTP and long‐term memories. Nat. Neurosci. 4, 289–296. Kao, M. H., Doupe, A. J., and Brainard, M. S. (2005). Contributions of an avian basal ganglia‐ forebrain circuit to real‐time modulation of song. Nature 433, 638–843. Karten, H. J. (1968). The ascending auditory pathway in the pigeon (Columba liva). II. Telencephalic projections of the nucleus ovoidalis thalami. Brain Res. 11, 134–153. Katz, L. C., and Gurney, M. E. (1981). Auditory responses in the zebra finch’s motor system for song. Brain Res. 211, 192–197. Kessel, B. (1957). A study of the breeding biology of the European starling (Sturnus vulgaris L.) in North America. Am. Midl. Nat. 58, 257–321. Kilgard, M. P., and Merzenich, M. M. (1998). Cortical map reorganization enabled by nucleus basalis activity. Science 279, 1714–1718. Kimpo, R. R., and Doupe, A. J. (1997). FOS is induced by singing in distinct neuronal populations in a motor network. Neuron 18, 315–325. Kleven, O., and Lifjeld, J. T. (2004). Extrapair paternity and offspring immunocompetence in the reed bunting, Emberiza schoeniclus. Anim. Behav. 68, 283–289.

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

115

Klukowski, L. A., Cawthorn, J. M., Ketterson, E. D., and Nolan, V., Jr. (1997). Effects of experimentally elevated testosterone on plasma corticosterone and corticosteroid‐binding globulin in dark‐eyed juncos (Junco hyemalis). Gen. Comp. Endocrinol. 108, 141–151. Kluyver, H. N. (1935). Waarnemingen over de levenswijze van den spreeuw (Sturnus v. vulgaris) met behulp van geringde individuen. Ardea 24, 133–166. Knol, B. W. (1991). Stress and the endocrine hypothalamus‐pituitary‐testis system: A review. Vet. Q. 13, 104–114. Kokko, H. (2001). Fisherian and ‘‘good genes’’ benefits of mate choice: How (not) to distinguish between them. Ecol. Lett. 4, 322–326. Kokko, H., Brooks, R., McNamara, J. M., and Houston, A. I. (2002). The sexual selection continuum. Proc. R. Soc. Lond. Ser. B Biol. Sci. 269, 1331–1340. Kroodsma, D., and Konishi, M. (1991). A suboscine bird (Eastern phoebe, Sayornis phoebe) develops normal song without auditory feedback. Anim. Behav. 42, 477–487. Kruse, A. A., Stripling, R., and Clayton, D. F. (2000). Minimal experience required for immediate‐early gene induction in zebra finch neostriatum. Neurobiol. Learn. Mem. 74, 179–184. Lanahan, A., and Worley, P. (1998). Immediate‐early genes and synaptic function. Neurobiol. Learn. Mem. 74, 179–184. Lehrman, D. S. (1965). Interaction between internal and external environments in the regulation of the reproductive cycle of the ring dove. In ‘‘Sex and Behavior’’ (F. A. Beach, Ed.), pp. 355–380. John Wiley & Sons, Inc., New York. Leitner, S., Voigt, C., Metzdorf, R., and Catchpole, C. K. (2005). Immediate early gene (ZENK, Arc) expression in the auditory forebrain of female canaries varies in response to male song quality. J. Neurobiol. 64, 275–284. Leonardo, A., and Konishi, M. (1999). Decrystallization of adult birdsong by perturbation of auditory feedback. Nature 399, 466–470. Leppelsack, H. J. (1983). Analysis of song in the auditory pathway of song birds. In ‘‘Advances in Vertebrate Neuroethology’’ (J. P. Evert, B. R. Capranica, and D. J. Ingle, Eds.), pp. 783–799. Plenum Press, New York. Leppelsack, H. J., and Schwartzkopff, J. (1972). Properties of acoustic neurons in the caudal neostriatum of birds. J. Comp. Physiol. A 80, 137–140. Leppelsack, H. J., and Voigt, M. (1976). Response to auditory neurons in the forebrain of a song bird to stimulation with species‐specific sounds. J. Comp. Physiol. A 107, 263–274. Liberman, A. M., Cooper, F. S., Shankweiler, D. P., and Studdert‐Kennedy, M. (1967). Perception of the speech code. Psychol. Rev. 74, 431–461. Lieshoff, G., Grosse‐Ophoff, J., and Bischof, H.‐J. (2004). Sexual imprinting leads to lateralized and non‐lateralized expression of the immediate early gene zenk in the zebra finch brain. Behav. Brain Res. 148, 145–155. Logan, C. A., and Wingfield, J. C. (1990). Autumnal territorial aggression is independent of plasma testosterone in mockingbirds. Horm. Behav. 24, 568–581. MacDougall‐Shackleton, S. A., Hulse, S. H., and Ball, G. F. (1998). Neural bases of song preferences in female zebra finches (Taeniopygia guttata). Neuroreport 9, 3047–3052. MacDougall‐Shackleton, S. A., MacDougall‐Shackleton, E. A., and Hahn, T. P. (2001). Physiological and behavioural responses of female mountain white‐crowned sparrows to natal‐ and foreign‐dialect songs. Can. J. Zool./Rev. Can. Zool. 79, 325–333. Maney, D. L., MacDougall‐Shackleton, E. A., MacDougall‐Shackleton, S. A., Ball, G. F., and Hahn, T. P. (2003). Immediate early gene response to hearing song correlates with receptive behavior and depends on dialect in a female songbird. J. Comp. Physiol. A 189, 667–674.

116

GREGORY F. BALL ET AL.

Margoliash, D. (1983). Acoustic parameters underlying the responses of song‐specific neurons in the white‐crowned sparrow. J. Neurosci. 3, 1039–1057. Margoliash, D. (1987). Neural plasticity in birdsong learning. In ‘‘Imprinting and Cortical Plasticity’’ (J. P. Rauschecker and P. Marler, Eds.), pp. 23–54. Plenum Press, New York. Margoliash, D. (2003). Offline learning and the role of autogenous speech: New suggestions from birdsong research. Speech Commun. 41, 165–178. Margoliash, D., and Fortune, E. S. (1992). Temporal and harmonic combination‐sensitive neurons in the zebra finch’s HVc. J. Neurosci. 12, 4309–4326. Marler, P., Peters, S., and Wingfield, J. (1987). Correlations between song acquisition, song production, and plasma levels of testosterone and estradiol in sparrows. J. Neurobiol. 18, 531–548. 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 336, 770–772. McEwen, B. S., Biron, C. A., Brunson, K. W., Bulloch, K., Chambers, W. H., Dhabhar, F. S., Goldfarb, R. H., Kitson, R. P., Miller, A. H., Spencer, R. L., and Weiss, J. M. (1997). The role of adrenocorticoids as modulators of immune function in health and disease: Neural, endocrine and immune interactions. Brain Res. Rev. 23(1–2), 79–133. Mello, C. V. (2002). Mapping vocal communication pathways in birds with inducible gene expression. J. Comp. Physiol. A 188, 943–959. Mello, C. V. (2004). Gene regulation by song in the auditory telencephalon of songbirds. Front. Biosci. 9, 63–73. Mello, C. V., and Clayton, D. F. (1994). Song‐induced ZENK gene expression in auditory pathways of songbird brain and its relation to the song control system. J. Neurosci. 14, 6652–6666. Mello, C., Vicario, D. S., and Clayton, D. F. (1992). Song presentation induces gene expression in the songbird forebrain. Proc. Natl. Acad. Sci. USA 89, 6818–6822. Mello, C., Nottebohm, F., and Clayton, D. (1995). Repeated exposure to one song leads to a rapid and persistent decline in an immediate early gene’s response to that song in zebra pinch telencephalon. J. Neurosci. 15, 6919–6925. Mello, C. V., Pinaud, R., and Ribeiro, S. (1998). Noradrenergic system of the zebra finch brain: Immunocytochemical study of dopamine‐beta‐hydroxylase. J. Comp. Neurol. 400, 207–228. Mello, C. V., Velho, T. A. F., and Pinaud, R. (2004). Song‐induced gene expression: A window on song auditory processing and perception. In ‘‘Behavioral Neurobiology of Bird Song’’ (H. P. Ziegler and P. Marler, Eds.), Vol. 1016, pp. 263–281. New York Academy of Sciences, New York, NY. Moller, A. P., Erritzoe, J., and Garamszegi, L. Z. (2005). Covariation between brain size and immunity in birds: Implications for brain size evolution. J. Evol. Biol. 18, 223–237. Mooney, R. (2004). Synaptic mechanisms for auditory‐vocal inteegration and the correction of vocal errors. In ‘‘Behavioral Neurobiology of Birdsong’’ (H. P. Ziegler and P. Marler, Eds.), Vol. 1016, pp. 476–494. The New York Academy of Sciences, New York, NY. Morgan, J. I., and Curran, T. (1989). Stimulus‐transcription coupling in neurons: Role of cellular immediate early genes. Trends Neurosci. 12, 459–462. Mountjoy, D. J., and Lemon, R. E. (1995). Extended song learning in wild European starlings. Anim. Behav. 49, 357–366. Mountjoy, D. J., and Lemon, R. E. (1996). Female choice for complex song in the European starling: A field experiment. Behav. Ecol. Sociobiol. 38, 65–71. Mountjoy, D. J., and Lemon, R. E. (1997). Male song complexity and parental care in the European starling. Behaviour 134, 661–675.

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

117

Muehlenbein, M. P., and Bribiescas, R. G. (2005). Testosterone‐mediated immune functions and male life histories. Am. J. Hum. Biol. 17, 527–558. Mu¨ller, C. M., and Leppelsack, H. J. (1985). Feature extraction and tonotopic organization in the avian forebrain. Exp. Brain Res. 59, 587–599. Nagle, L., and Kreutzer, M. L. (1997a). Song tutoring influences female song preferences in domesticated canaries. Behaviour 134, 89–104. Nagle, L., and Kreutzer, M. L. (1997b). Adult female domesticated canaries can modify their song preferences. Can. J. Zool. 75, 1246–1350. Nastiuk, K. L., Mello, C. V., George, J. M., and Clayton, D. F. (1994). Immediate‐early gene responses in the avian song control system: Cloning and expression analysis of the canary c‐jun cDNA. Mol. Brain Res. 27, 299–309. Nordeen, K. W., and Nordeen, E. J. (1992). Auditory feedback is necessary for the maintenance of stereotyped song in adult zebra finches. Behav. Neural Biol. 57, 58–66. Nottebohm, F. (1980). Brain pathways for vocal learning in birds: A review of the first 10 years. In ‘‘Progress in Psychobiology and Physiological Psychology’’ (J. M. Sprague and A. N. Epstein, Eds.), Vol. 9, pp. 85–214. Academic Press, New York. Nottebohm, F., Kelley, D. B., and Paton, J. A. (1982). Connections of vocal control nuclei in the canary telencephalon. J. Comp. Neurol. 207, 344–357. Nottebohm, F., Stokes, T., and Leonard, M. (1976). Central control of song in the canary, Serinus canarius. J. Comp. Neurol. 165, 457–486. Nottebohm, F., Nottebohm, M. E., and Crane, L. (1986). Developmental and seasonal changes in canary song and their relation to changes in the anatomy of song‐control nuclei. Behav. Neural Biol. 46, 445–471. Nowicki, S., and Ball, G. F. (1989). Testosterone induction of song in photosensitive and photorefractory male sparrows. Horm. Behav. 23, 514–525. Nowicki, S., and Searcy, W. A. (2004). Song function and the evolution of female preferences—why birds sing, why brains matter. In ‘‘Behavioral Neurobiology of Birdsong’’ (H. P. Ziegler and P. Marler, Eds.), Vol. 1016, pp. 704–723. New York Academy of Sciences, New York, NY. Nowicki, S., Peters, S., and Podos, J. (1998). Song learning, early nutrition and sexual selection in songbirds. Am. Zool. 38, 179–190. Nowicki, S., Searcy, W. A., and Peters, S. (2002a). Brain development, song learning and mate choice in birds: A review and experimental test of the ‘‘nutritional stress hypothesis.’’ J. Comp. Physiol. A 188, 1003–1014. Nowicki, S., Searcy, W. A., and Peters, S. (2002b). Quality of song learning affects female response to male bird song. Proc. R. Soc. Lond. B 269, 1949–1954. O’Donald, P. (1967). A general model of sexual and natural selection. Herediy 22, 499–518. O’Donald, P. (1980). ‘‘Genetic Models of Sexual Selection.’’ Cambridge University Press, Cambridge, UK. Oberweger, K., and Goller, F. (2001). The metabolic cost of birdsong production. J. Exp. Biol. 204, 3379–3388. Owen‐Ashley, N. T., Hasselquist, D., and Wingfield, J. C. (2004). Androgens and the immunocompetence handicap hypothesis: Unraveling direct and indirect pathways of immunosuppression in song sparrows. Am. Nat. 164, 490–505. Pfaff, D. W., Schwartz‐Giblin, S., McCarthy, M. M., and Kow, L. M. (1994). Cellular and molecular mechanisms of female reproductive behaviors. In ‘‘The Physiology of Reproduction’’ (E. Knobil and J. D. Neill, Eds.), pp. 107–220. Raven Press, New York. Phillmore, L. S., Bloomfield, L. L., and Weisman, R. G. (2003). Effects of songs and calls on ZENK expression in the auditory telencephalon of field‐ and isolate‐reared black capped chickadees. Behav. Brain Res. 147, 125–134.

118

GREGORY F. BALL ET AL.

Pinaud, R., Velho, T. A., Jeong, J. K., Tremere, L. A., Leao, R. M., von Gersdorff, H., and Mello, C. V. (2004). GABAergic neurons participate in the brain’s response to birdsong auditory stimulation. Eur. J. Neurosci. 20, 1318–1330. Pinxten, R., Van Elsacker, L., and Verheyen, R. (1987). Duration and temporal pattern of mate guarding in the starling. Ardea 75, 263–269. Pinxten, R., DeRidder, E., Balthazart, J., and Eens, M. (2002). Context‐dependent effects of castration and testosterone treatment on song in male European starlings. Horm. Behav. 42, 307–318. Poiani, A., Goldsmith, A. R., and Evans, M. R. (2000). Ectoparasites of house sparrows (Passer domesticus): An experimental test of the immunocompetence handicap hypothesis and a new model. Behav. Ecol. Sociobiol. 47, 230–242. Pomiankowski, A. (1988). The evolution of female mate preferences for male genetic quality. Oxford Surv. Evol. Biol. 5, 136–184. Raberg, L., Grahn, M., Hasselquist, D., and Svensson, E. (1998). On the adaptive significance of stress‐induced immunosuppression. Proc. R. Soc. Lond. Ser. B Biol. Sci. 265, 1637–1641. Radwan, J. (2002). Good genes go fisherian. Trends Ecol. Evol. 17, 539–539. Reid, J. M., Arcese, P., Cassidy, A. L., Hiebert, S. M., Smith, J. N., Stoddard, P. K., Marr, A. B., and Keller, L. F. (2005). Fitness correlates of song repertoire size in free‐living song sparrows (Melospiza melodia). Am. Nat. 165(3), 299–310. Reiner, A., Perkel, D. J., Bruce, L. L., Butler, A. B., Csillag, A., Kuenzel, W., Medina, L., Paxinos, G., Shimizu, T., Striedter, G., Wild, M., Ball, G. F., et al. (2004). Revised nomenclature for avian telencephalon and some related brainstem nuclei. J. Comp. Neurol. 473, 377–414. Reinke, H., and Wild, J. M. (1998). Identification and connections of inspiratory premotor neurons in songbirds and budgerigar. J. Comp. Neurol. 391, 147–163. Ribeiro, S., and Mello, C. V. (2000). Gene expression and synpatic plasticity in the auditory forebrain of songbirds. Learn. Mem. 7, 235–243. Ribeiro, S., Cecchi, G. A., Magnasco, M. O., and Mello, C. V. (1998). Toward a song code: Evidence for a syllabic representation in the canary brain. Neuron 21, 359–371. Riters, L. V., and Ball, G. F. (1999). Lesions to the medial preoptic area affect singing in the male European starling (Sturnus vulgaris). Horm. Behav. 36, 276–286. Riters, L. V., Eens, M., Pinxten, R., Duffy, D. L., Balthazart, J., and Ball, G. F. (2000). Seasonal changes in courtship song and the medial preoptic area in male European starlings (Sturnus vulgaris). Horm. Behav. 38, 250–261. Roberts, M. L., Buchanan, K. L., and Evans, M. R. (2004). Testing the immunocompetence handicap hypothesis: A review of the evidence. Anim. Behav. 68, 227–239. Rubsamen, R., and Dorrscheidt, G. J. (1986). Tonotopic organization of the auditory forebrain in a songbird, the European starling. J. Comp. Physiol. A 158, 639–646. Ruscio, M. G., and Adkins‐Regan, E. (2004). Immediate early gene expression associated with induction of brooding behavior in Japanese quail. Horm. Behav. 46, 19–29. Sartor, J. J., and Ball, G. F. (2005). Social suppression of song is associated with a reduction in volume of a song control nucleus in European starlings (Sturnus vulgaris). Behav. Neurosci. 119, 233–244. Scha¨fer, M., Rubsamen, R., Dorrscheidt, G. J., and Knipschild, M. (1992). Setting complex tasks to single units in the avian auditory forebrain. II. Do we really need natural stimuli to describe neuronal response characteristics? Hear. Res. 57, 231–244. Scharff, C., and Nottebohm, F. (1991). A comparative study of the behavioral deficits following lesions of various parts of the zebra finch song system: Implications for vocal learning. J. Neurosci. 11, 2896–2913.

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

119

Schlinger, B. A. (1997). Sex steroids and their actions on the birdsong system. J. Neurobiol. 33, 619–631. Schmidt, M. F., and Konishi, M. (1998). Gating of auditory response in the vocal control system of awake songibirds. Nat. Neurosci. 6, 513–518. Schoech, S. J., Ketterson, E. D., and Nolan, V., Jr. (1999). Exogenous testosterone and the adrenocortical response in dark‐eyed juncos. Auk 64, 64–72. Schwabl, H., and Kriner, E. (1991). Territorial aggression and song of male European robins (Erithacus rubecula) in autumn and spring: Effects of antiandrogen treatment. Horm. Behav. 25, 180–194. Searcy, W. A. (1992). Song repertoire and mate choice in birds. Am. Zool. 32, 71–80. Searcy, W. A., and Yasukawa, K. (1996). The reproductive success of secondary females relative to that of monogamous and primary females in Red‐winged Blackbirds. J. Avian Biol. 27, 225–230. Sen, K., Theunissen, F. E., and Doupe, A. J. (2001). Feature analysis of natural sounds in the songbird auditory forebrain. J. Neurophysiol. 86, 1445–1458. Sharp, P. J., Li, Q. S., Georgiou, G., and Lea, R. W. (1996). Expression of fos‐like immunoreactivity in the hypothalamus of the ring dove (Streptopelia risoria) at the onset of incubation. J. Neuroendocrinol. 8, 291–298. Sheldon, B. C., Merila, J., Qvarnstrom, A., Gustafsson, L., and Ellegren, H. (1997). Paternal genetic contribution to offspring condition predicted by size of male secondary sexual character. Proc R. Soc. Lond. Ser. B Biol. Sci. 264, 297–302. Sherman, P. W. (1988). The levels of analysis. Anim. Behav. 36, 616–619. Silverin, B. (1998). Stress responses in birds. Poultry Avian Biol. Rev. 9, 153–168. Shimizu, T., Bowers, A. N., Budzynski, C. A., Kahn, M. C., and Bingman, V. P. (2004). What does a pigeon (Columba livia) brain look like during homing? Selective examination of ZENK expression. Behav. Neurosci. 118, 845–851. Simpson, H. B., and Vicario, D. S. (1990). Brain pathways for learned and unlearned vocalizations differ in zebra finches. J. Neurosci. 10, 1541–1556. Slagsvold, T. (1977). Bird song activity in relation to breeding cycle, spring weather, and environmental physiology. Ornis Scand. 8, 197–222. Smith, G. T., Brenowitz, E. A., Wingfield, J. C., and Baptista, L. F. (1995). Seasonal changes in song nuclei and song behavior in Gambels white‐crowned sparrows. J. Neurobiol. 28, 114–125. Smith, G. T., Brenowitz, E. A., Beecher, M. D., and Wingfield, J. C. (1997). Seasonal changes in testosterone, neural attributes of song control nuclei, and song structure in wild songbirds. J. Neurosci. 17, 6001–6010. Sockman, K. W., Gentner, T. Q., and Ball, G. F. (2002). Recent experience modulates forebrain gene‐expression in response to mate‐choice cues in European starlings. Proc. R. Soc. Lond. B 269, 2479–2485. Sockman, K. W., Gentner, T. Q., and Ball, G. F. (2005). Complementary neural systems for the experience‐dependent integration of mate‐choice cues in the European starling. J. Neurobiol. 62, 72–81. Sohrabji, F., Nordeen, E. J., and Nordeen, K. W. (1990). Selective impairment of song learning following lesions of a forebrain nucleus in juvenile zebra finches. Behav. Neural Biol. 53, 51–63. Soma, K. K., and Wingfield, J. C. (2001). Dehydroepiandrosterone in songbird plasma: Seasonal regulation and relationship to territorial aggression. Gen. Comp. Endocrinol. 123, 144–155. Soma, K. K., Tramontin, A. D., and Wingfield, J. C. (2000). Oestrogen regulates male aggression in the non‐breeding season. Proc. R. Soc. Lond. B 267, 1089–1096.

120

GREGORY F. BALL ET AL.

Soma, K. K., Wissman, A. M., Brenowitz, E. A., and Wingfield, J. C. (2002). Dehydroepiandrosterone (DHEA) increases territorial song and the size of an associated brain region in a male songbird. Horm. Behav. 41, 203–212. Soma, K. K., Alday, N. A., Hau, M., and Schlinger, B. A. (2004). Dehydroepiandrosterone metabolism by 3beta‐hydroxysteroid dehydrogenase/Delta5‐Delta4 isomerase in adult zebra finch brain: Sex difference and rapid effect of stress. Endocrinology 145, 1668–1677. Sorenson, L. G., Nolan, P. M., Brown, A. M., Derrickson, S. R., and Monfort, S. L. (1997). Hormonal dynamics during mate choice in the northern pintail: A test of the ‘challenge’ hypothesis. Anim. Behav. 54, 1117–1133. Spencer, K. A., Buchanan, K. L., Goldsmith, A. R., and Catchpole, C. K. (2003). Song as an honest signal of developmental stress in the zebra finch (Taeniopygia guttata). Horm. Behav. 44, 132–139. Spencer, K. A., Buchanan, K. L., Goldsmith, A. R., and Catchpole, C. K. (2004). Developmental stress, social rank and song complexity in the European starling (Sturnus vulgaris). Proc. R. Soc. Lond. B S121–S123. Stripling, R., Volman, S. F., and Clayton, D. F. (1997). Response modulation in the zebra finch neostriatum: Relationship to nuclear gene regulation. J. Neurosci. 17, 3883–3893. Terpstra, N. J., Bolhuis, J. J., and den Boer‐Visser, A. M. (2004). An analysis of the neural representation of birdsong memory. J. Neurosci. 24, 4971–4977. Theunissen, F. E., and Doupe, A. J. (1998). Temporal and spectral sensitivity of complex auditory neurons in the nucleus HVc of male zebra finches. J. Neurosci. 18, 3786–3802. Theunissen, F. E., Sen, K., and Doupe, A. J. (2000). Spectral‐temporal receptive fields of nonlinear auditory neurons obtained using natural sounds. J. Neurosci. 20, 2315–2331. Theunissen, F. E., Amin, N., Shaevitz, S. S., Woolley, S. N., Fremouw, T., and Hauber, M. E. (2004). Song selectivity in the song system and in the auditory forebrain. In ‘‘Behavioral Neurobiology of Birdsong’’ (H. P. Ziegler and P. Marler, Eds.), Vol. 1016, pp. 187–207. New York Academy of Sciences, New York, NY. Thomas, R. J., Cuthill, I. C., Goldsmith, A. R., Cosgrove, D. F., Lidgate, H. C., and Proctor, S. (2003). The trade‐off between singing and mass gain in a daytime‐singing bird, the European robin. Behaviour 140, 387–404. Tinbergen, N. (1950). The hierarchical organization of nervous mechanisms underlying instinctive behaviour. Sym. Soc. Exp. Biol. 4, 305–312. Tinbergen, N. (1963). On aims and methods of ethology. Zeitschrift fu¨r Tierpsychologie 20, 410–433. Tischmeyer, W., and Grimm, R. (1999). Activation of immediate early genes and memory formation. Cell. Mol. Life Sci. 55, 564–574. Tlemc¸ani, O., Ball, G. F., D’Hondt, E., Vandesande, F., Sharp, P. J., and Balthazart, J. (2000). Fos induction in the Japanese quail brain after expression of appetitive and consummatory aspects of male sexual behavior. Brain Res. Bull. 52, 249–262. Tramontin, A. D., and Brenowitz, E. A. (2000). Seasonal plasticity in the adult brain. Trends Neurosci. 23, 251–258. Ulinski, P. S., and Margoliash, D. (1990). Neurobiology of the reptile‐bird transition. In ‘‘Cerebral Cortex’’ (E. G. Jones and A. Peters, Eds.), pp. 217–265. Plenum Press, New York. van Gossum, H., Stoks, R., and De Bruyn, L. (2001). Reversible frequency‐dependent switches in male mate choice. Proc. R. Soc. Lond. B 268, 83–85. Vates, G. E., Broome, B. M., Mello, C. V., and Nottebohm, F. (1996). Auditory pathways of caudal telencephalon and their relation to the song system of adult male zebra finches (Taenopygia guttata). J. Comp. Neurol. 366, 613–642.

A NEUROETHOLOGICAL APPROACH TO SONG BEHAVIOR AND PERCEPTION

121

Vignal, C., Attia, J., Mathevon, N., and Beauchaud, M. (2004). Background noise does not modify song‐induced genic activation in the bird brain. Behav. Brain Res. 153, 241–248. Viney, M. E., Riley, E. M., and Buchanan, K. L. (2005). Optimal immune responses: Immunocompetence revisited. Trends Ecol. Evol. 20, 665–669. Walters, M. J., Collado, D., and Harding, C. F. (1991). Oestrogenic modulation of singing in male zebra finches: Differential effects on directed and undirected songs. Anim. Behav. 42, 445–452. Ward, S., Speakman, J. R., and Slater, P. J. B. (2003). The energy cost of song in the canary, Serinus canaria. Anim. Behav. 66, 893–902. Weatherhead, P. J. (1994). Mixed mating strategies by females may strengthen the sexy son hypothesis. Anim. Behav. 47, 1210–1211. Weatherhead, P. J., and Robertson, R. J. (1979). Offspring quality and the polygyny threshold—sexy son hypothesis. Am. Nat. 113, 201–208. Weatherhead, P. J., and Robertson, R. J. (1981). In defense of the sexy son hypothesis. Am. Nat. 117, 349–356. Wedekind, C., and Folstad, I. (1994). Adaptive or nonadaptive immunosuppression by sex‐ hormones. Am. Nat. 143, 936–938. Whitney, O., Soderstrom, K., and Johnson, F. (2003). CB1 cannabinoid receptor activation inhibits a neural correlate of song recognition in an auditory/perceptual region of the zebra finch telencephalon. J. Neurobiol. 56, 266–274. Wiegmann, D. D., Real, L. A., Capone, T. A., and Ellner, S. (1996). Some distinguishing features of models of search behavior and mate choice. Am. Nat. 147, 188–204. Wilczynski, W., and Ryan, M. J. (1992). Introduction to the symposium: Mechanisms of mate choice. Am. Zool. 32, 15–17. Wild, J. M. (1993a). The avian nucleus retroambigualis: A nucleus for breathing, singing and calling. Brain Res. 606, 319–324. Wild, J. M. (1993b). Descending projections of the songbird nucleus robustus archistriatalis. J. Comp. Neurol. 338, 225–241. Wild, J. M. (1994). The auditory‐vocal‐respiratory axis in birds. Brain Behav. Evol. 44, 192–209. Wild, J. M. (2004). Functional neuroanatomy of the sensorimotor control of singing. In ‘‘Behavioral Neurobiology of Birdsong’’ (H. P. Zeigler and P. Marler, Eds.), Vol. 1016, pp. 438–462. New York Academy of Sciences, New York, NY. Williams, G. C. (1966). ‘‘Adaptation and Natural Selection,’’ p. 307. Princeton University Press, Princeton. Wingfield, J. C. (1994). Regulation of territorial behavior in the sedentary song sparrow, Melospiza melodia morphna. Horm. Behav. 28, 1–15. Wingfield, J., and Farner, D. (1993). Endocrinology of reproduction in wild species. ‘‘In Avian Biology,’’ Vol. 9, pp. 163–327. Academic Press, New York. Wingfield, J. C., and Hahn, T. P. (1994). Testosterone and territorial behaviour in sedentary and migratory sparrows. Anim. Behav. 47, 77–89. Wright, J., and Cuthill, I. (1992). Monogamy in the European starling. Behaviour 120, 262–285. Yamada, Y., Hada, Y., Imamura, K., Mataga, N., Watanabe, Y., and Yamamoto, M. (1999). Differential expression of immediate‐early genes, c‐fos and zif268, in the visual cortex of young rats: Effects of a noradrenergic neurotoxin on their expression. Neuroscience 92, 473–484. Yu, A. C., and Margoliash, D. (1996). Temporal hierarchical control of singing in birds. Science 273, 1871–1875. Zahavi, A. (1975). Mate selection—a selection for a handicap. J. Theoretical Biol. 53, 205–214.

This page intentionally left blank

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 36

Navigational Memories in Ants and Bees: Memory Retrieval When Selecting and Following Routes Thomas S. Collett, Paul Graham, Robert A. Harris, and Natalie Hempel‐de‐Ibarra school of life sciences university of sussex falmer, brighton bn1 9qg united kingdom

I. INTRODUCTION Insect navigation is an intriguing mix of rigidity and flexibility. Rigidity can be seen in the fixed, visually guided routes that individual ants of some species follow when traveling through familiar territory to a particular destination. Flexibility is exhibited in that ants and bees are not constrained to a single route leading to a single goal, but may select one goal out of several that they know, and take the particular route that leads to their chosen destination. At the simplest level, an insect chooses between going to a foraging area and returning to its nest. But insects can also go from their nest along different routes or different branches of a route to reach alternative feeding places or to find different resources. Single routes are underpinned by route‐specific memories of visual landmarks and motor patterns that are retrieved in the correct sequence. We argue here that a crucial component of an insect’s capacity to be flexible in choosing a route, but rigid in following it, is its ability to select which memories it retrieves. A review by Menzel et al. (2001) covers similar topics but from a different perspective. Memorizing and performing a route is an example of a general problem faced by many animals when they learn and retrieve an ordered sequence of actions. But in one important respect routes are special. They are laid out spatially, often within terrain containing rich mnemonic cues, so that much of the sequence resides in the external world through which the insect travels: if an ant walks in the correct direction it will encounter a familiar signpost that tells it what to do next. Orators were trained to internalize this characteristic of routes to help them remember long verbal sequences in the 0065-3454/06 $35.00 DOI: 10.1016/S0065-3454(06)36003-2

123

Copyright 2006, Elsevier Inc. All rights reserved.

124

THOMAS S. COLLETT ET AL.

centuries before external cue cards were available to prompt their memories (Yates, 1966). To remember complex speeches and arguments in the correct order, orators imagined themselves strolling through a long hallway with vivid and distinctive objects distributed in alcoves along it. They associated with each object an item that they wanted to remember. They could then recall the sequence of items by repeating their imaginary walk during which they ‘‘saw’’ each object and with it the associated item. We begin this chapter with a brief description of natural routes and then discuss what landmark memories are and how they are used for guidance. Because we are particularly concerned with what makes the retrieval of navigational memories reliable, we emphasize recent studies on the associative linking or binding that occurs both between the separate components of a landmark memory and between separate memories. Such links allow individual or groups of memories to be primed. Thus, encountering one familiar landmark along a route can prime or facilitate the retrieval of the memory of another landmark (Srinivasan et al., 1998). An insect’s destination and route are chosen according to the insect’s internal state, or the state of the colony, or by its reading of a variety of environmental cues that predict what destination is currently most productive. The second part of the chapter considers evidence that ants and bees prime selectively the navigational memories that are associated with a particular route and describes some of the cues that determine their choice. While details of the cellular basis of learning and memory in insects are becoming clearer through remarkable genetic, molecular, and physiological experiments (e.g., Davis, 2005; Gerber et al., 2004; Hammer, 1997; Liu et al., 2006), a reader should be warned that the terminology of insect memory is little more than convenient shorthand for labeling processes that are not yet well understood. The use of terms like memory, retrieval, and priming should not be taken to mean that insects recall and picture to themselves scenes of familiar surroundings. We only know that an insect on encountering a familiar scene or being in a particular state will perform an appropriate response or attend to one familiar stimulus rather than another. Similarly, the term ‘‘snapshot’’ that we use to describe an insect’s memory of visual landmarks and scenes is helpful in summarizing a set of hypotheses arising from behavioral data, but there is little neurophysiological evidence to tell us whether it accounts well for what is happening in the brain.

II. FORAGING ROUTES Many ants, bees, and wasps are central place foragers that go out from their nest to collect and return with food and other resources. We focus on foraging

NAVIGATIONAL MEMORIES IN ANTS AND BEES

125

ants, because their routes have been described and studied in more detail than those of flying bees and wasps, for the simple reason that the paths of walking insects are easier to monitor. An established visually guided route to a feeding area is the consummation of a learning process that is only possible because insects possess navigational strategies that can operate before they have acquired detailed knowledge of their environment. Some ants find a new site through information supplied by others; for instance, by following a chemical trail (Ho¨lldobler and Wilson, 1990). Others are like the Mediterranean desert ant, Cataglyphis bicolor, and forage individually, keeping track of their location through path integration (PI) (Collett and Collett, 2000; Wehner, 1992; Wehner and Srinivasan, 2003). A naı¨ve Cataglyphis forager leaves its nest and explores the surrounding scrubby desert terrain for dead insects and edible debris. While doing so, it performs global PI: the ant monitors the path that it takes with a celestial compass and some kind of odometer. This path‐related information is used to update an accumulator, the state of which encodes the ant’s current PI coordinates relative to the nest. An ant provided with this knowledge of its current coordinates is always able to specify a direct homewards path over unfamiliar terrain. When the ant encounters food, it stores in memory the PI coordinates of the site relative to the nest and it can use PI to return to the food site as well as to go home (Collett et al., 1999; Schmid‐ Hempel, 1984; Wehner et al., 1983; Wolf and Wehner, 2000). Figure 1 shows the first foraging trips of a new forager (Wehner et al., 2004). After several abortive trips in different directions, this ant discovers an item of food (Fig. 1, run 9). It returns straight home and then on its next outward trip makes directly for this new site guided by PI. Although its second visit is fruitless, it makes a third trip to the same site before giving up and looking elsewhere. Limited persistence after a successful trip allows the ant to take advantage of a patchy distribution of food and to learn visual landmarks along its repeated path. It means that if food is reasonably abundant, the ant will tend to stick to a single familiar sector around its nest, exploring outward from its previous find (Schmid‐Hempel, 1984). This win‐stay and lose‐shift strategy when adopted by the whole foraging force of a colony leads to the efficient exploitation of different spatial distributions of resources while limiting the route learning demanded of an individual (Deneubourg et al., 1987; Wehner, 1987). Cataglyphis routes have been studied several times since Santschi first illustrated them in 1913 (Fig. 2). The examples of Fig. 3A are of ants returning from a fixed feeding place to their nest over terrain that has been marked with artificial landmarks. Different ants may follow slightly different paths, according to whether they skirt to the left or right of the first landmark. But ants going the same way around that landmark follow very similar routes, suggesting that relatively few rules determine what these

126

THOMAS S. COLLETT ET AL.

Fig. 1. The sequential foraging paths of a desert ant (Cataglyphis bicolor) at the beginning of its foraging career. (A) Foraging trips 2–5. These early runs were short, unsuccessful, and their direction varied. (B) Trips 8–12. On run 9, the ant discovers a food item and its subsequent two trips, both unsuccessful, are in the same direction. From trip 12 onward, the disappointed ant changes direction (adapted from Wehner et al., 2004).

ants learn and what they do. One rule followed by both ants and bees is to approach prominent isolated landmarks close to their route (Chittka et al., 1995b; Graham et al., 2003; von Frisch, 1967), learn the appearance of each landmark and attach different responses to them. For instance, if one landmark is placed to the left of the direct homeward path and a second to the right, ants learn to steer to the right of the first landmark and to the left of the second, and they will steer in the correct direction, even when the landmark appears in an unusual position along their route (Fig. 3B). This example also makes the point, to which we will come back several times, that visual landmark memories are at the service of a particular route. They are procedural and tell an animal what to do next rather than where it is on any kind of map (Collett et al., 2002). It is not a tough job to recognize such distinct artificial landmarks. More remarkably, ants can recognize their location when placed among tussocks of grass. Figure 4 illustrates homeward routes of an Australian desert ant, Melophorus bagoti (Kohler and Wehner, 2005). When caught either at its feeding site or close to the nest and placed midway along the route, the ant rejoins the route and runs in the correct direction. The fact that ants

NAVIGATIONAL MEMORIES IN ANTS AND BEES

127

Fig. 2. An early recording of the foraging route of a desert ant (C. bicolor). Traces show two outward journeys and one return journey performed by a single ant. Despite the lack of chemical trails, all three paths are remarkably similar. Idiosyncratic and stereotyped routes provide indirect evidence that ants learn their routes. Stippling shows vegetation (adapted from Santschi, 1913).

continue a route correctly, both when they are placed directly on the path or at a short distance from it, means that they are retrieving and using memories of specific routes and that they are similar to Cataglyphis (Bisch‐ Knaden and Wehner, 2003; Collett et al., 1998; Wehner et al., 1996) in doing so independently of their global PI coordinates. This example raises two questions about reliable memory retrieval that we try to answer in Sections IV and V. First, how does the ant know where it is when placed among confusable tussocks of grass? Second, how does the ant decide whether to move in a homeward direction or toward the food?

III. NAVIGATIONAL MEMORIES Studies of natural routes tell us that insects following routes can be guided by visual and motor memories. What are these memories and how are they used?

128

THOMAS S. COLLETT ET AL.

Fig. 3. (A) Homeward routes of ants (C. bicolor) after several days foraging at a fixed location. Four visually distinct landmarks (1–4) were distributed close to the direct path between food and nest. Left panel: individual trajectories from an ant that went to the left of landmark 1. Middle panel: trajectories from ant that detoured to the right of landmark 1. Right panel: mean trajectories of each of five ants that passed to the left of landmark 1. (B) Direct evidence that C. bicolor recognizes the appearance of familiar landmarks along a route. Left: ants are trained along a foraging path with an oil barrel positioned to the right of the ants’ homeward path about 20 m from the nest and triangles to the left at about 2 m from

NAVIGATIONAL MEMORIES IN ANTS AND BEES

129

Fig. 4. Australian desert ants (Melophorus bagoti) rejoin routes after displacement, showing that visually guided route following is independent of other navigational strategies. Dotted lines show normal homeward trajectories of an ant. Black solid lines show that the ant takes similar paths, when (A) it runs almost to the nest (N) and is returned to a position midway between the feeder (F) and the nest, or (B) it is taken from feeder to the same midway position. Gray areas depict grass tussocks (adapted from Kohler and Wehner, 2005, with permission from Elsevier).

A. VISUAL LANDMARK MEMORIES A variety of evidence from flies, water striders, ants, bees, and wasps suggests that insects remember scenes as two‐dimensional views of their surroundings from particular vantage points, and that they can use such a stored view or snapshot to return to where the snapshot was acquired (Cartwright and Collett, 1983; Collett and Land, 1975; Junger, 1991; Wehner and Ra¨ber, 1979; Zeil, 1993a,b). From most positions within a limited catchment area surrounding the acquisition point, the current image on an insect’s retina will transform smoothly into the snapshot allowing the insect to move incrementally to the acquisition point.

the nest. Right: single ants are released on a test ground with either the barrel or the triangles placed 10 m from the release point in the path of their home vector. Ants detour consistently to the left of the barrel and to the right of the triangles so that the landmark appears on the same side of the visual field as it does in training. Filled circles and error bars give mean
1 SD (adapted from Collett et al., 1992).

130

THOMAS S. COLLETT ET AL.

Fig. 5. Wood ants (Formica rufa) search for an absent feeder using a stored two‐view or snapshot. (A) Ants are trained to find food midway between two cylinders and their search distribution is concentrated on that spot when the feeder is missing. (B) Training cylinders are replaced by one cylinder that is smaller (in height and width) and one that is larger. Ants search where the cylinders look the same as they did in training. (C) Another group of ants is trained to find food midway between two large cylinders. (D) When tested with two small cylinders replacing the training cylinders, the ants search in two locations, one appropriate for each cylinder. (E) Subsets of distribution D to show where ants search when

NAVIGATIONAL MEMORIES IN ANTS AND BEES

131

Figure 5A shows the search distribution of a wood ant trained to feed at a location between two cylinders of the same size. If one cylinder is enlarged and the other made smaller (Fig. 5B), the ant’s search point shifts to where the two cylinders subtend the angular size normally seen from the feeding site (Durier et al., 2003). The insects try to recapture the correct appearance of each cylinder as seen from the goal and care less whether the cylinder is at the correct distance. A further experiment suggests that the ants have acquired two views at the goal, the first when facing one cylinder and the second when facing the other, and that in their search for the goal they try to match their stored view to the cylinder that they currently face (Graham et al., 2004). Ants were again trained with food midway between two cylinders (Fig. 5C). When both cylinders were made smaller, the search distribution was double peaked, one peak close to each cylinder (Fig. 5D). This search distribution was partitioned into two, with one distribution only including data points when ants faced in the rough direction of the left cylinder and the other when they faced the right cylinder. The two distributions each had a single major peak close to the cylinder that the ants faced (Fig. 5E and F). Ants will also acquire snapshot memories of landmarks encountered along a route. Guidance by these memories can be seen in the paths of several species of ants when they are near to an extended landmark, such as a wall (Collett et al., 2001; Graham and Collett, 2002; Pratt et al., 2001). To demonstrate that wood ants could be guided by a wall and nothing else, ants were trained to go from a start point to a feeding site along a line that was parallel to a 20 cm high wall (Fig. 6A). On every training trial, the wall, the starting point, and the feeder were rotated together as a group. In tests, the feeder was removed and the ants’ paths recorded individually. Ants followed the training path relative to the wall, irrespective of the wall’s position and orientation within the training arena (Graham and Collett, 2002). How does the wall fix the ants’ path? Three experimental manipulations on ants trained to take a path parallel to the wall imply that ants have learnt the retinal elevation of some features of the wall (e.g., the top) seen from the trained path, and that they adjust their path, so as to keep this feature at the learnt elevation. First, when the wall was made twice as high as in training and an ant was started at its accustomed distance from the

they fixate within
20 of the center of one or other cylinder, as shown by the schematic ant to the left of each distribution. These distributions indicate that ants are guided primarily by the cylinder that they face. White spots show the predicted positions of the search peaks, assuming that ants search where the apparent size of the cylinder matches the size seen from the feeding site (adapted from Graham et al., 2004).

132

THOMAS S. COLLETT ET AL.

Fig. 6. Snapshots can guide routes. (A) Ants trained along a path 20 cm from and parallel to a 20 cm high wall are started either 20 or 40 cm from the wall. Their path is parallel to the wall with a 20 cm start and turns toward the wall with a 40 cm start. (B) With a 40 cm high wall, the ants’ paths are parallel to the wall if they are started 40 cm away, and they turn away from the wall if they are started 20 cm from the wall. The ants steer their path using the apparent height of the wall (adapted from Graham and Collett, 2002).

wall, its path was directed away from the wall, so bringing the top of the wall toward its usual retinal elevation (Fig. 6B). But when both the height of the wall and the ant’s starting distance were doubled, putting the top of the wall at the usual retinal elevation, the ant’s paths were parallel to the wall. Third, when the height of the wall was as in training and ants were started at double their normal distance from the wall, so that the retinal elevation of the top of the wall was unexpectedly low, the ants approached the wall (Fig. 6A). In the next few sections, we review some of what is known about the details of snapshots and the process of image matching. 1. Retinotopic and Nonretinotopic Encoding of Snapshots Several lines of behavioral evidence from water striders and ants indicate that snapshots are partly encoded in retinal coordinates. Water striders hold station on the surface of moving streams and catch insects, which have been trapped in the surface layer as they float past. Laboratory experiments show that water striders keep in one spot by facing upstream and learning the retinal elevation and horizontal position of landmarks above them (Junger, 1991). In a darkened laboratory, a single, small lightbulb will act as a landmark to stabilize the insect’s position. When the bulb is moved downward or upward on the insect’s retina, the water strider immediately compensates by rowing forward or drifting backward to return this landmark to the desired retinal elevation.

NAVIGATIONAL MEMORIES IN ANTS AND BEES

133

Wood ants will learn to approach an upright cone to find sucrose at its base (Fig. 7A; Judd and Collett, 1998). As the ant moves closer to the cone, its bounding edges, seen at ground level, shift peripherally across the retina. Approaches to a test stimulus of a single black–white edge oriented at the same angle as one of the bounding edges of the cone suggest that the ant stores several snapshots, each at a different distance from the cone. During the approach, the edge is held in a number of preferred retinal positions, possibly corresponding to the positions of the edges in the ant’s set of snapshots (Fig. 7B). The preferred retinal positions associated with the cone’s left and right edges are symmetrical about the midline (Fig. 7B and C), suggesting that the ant faced the long axis of the cone while acquiring snapshots. The retinal image of an inverted cone with its apex on the ground transforms differently during an approach. As the ant travels over the ground toward the cone, the bounding edges of the inverted cone extend further upward into the ant’s visual field, while the apex on the ground remains in a constant retinal position. Correspondingly, ants trained to approach inverted cones and tested with a single edge tend to keep the bottom of the edge fixated in their frontal visual field throughout their approach (Fig. 7D and E). A study of visually guided antennal pointing in the cockroach (Kwon et al., 2004) emphasizes that insects can learn both the absolute retinal position of a landmark image and the angular separation between landmarks independently of retinal position. Immobilized cockroaches with their head fixed in space learn that a visual cue (a green spot of light from a light‐emitting diode, LED) flashed in a fixed position on the retina predicts the release of a food odor coming from the same direction (Fig. 8A). Cockroaches, conditioned to this task and tested in the dark, point their antennae at the flashed LED when no odor is presented. The probability of pointing at the LED falls as the LED is shifted away from the training position, indicating that the response to the LED was associated with this fixed retinal position (Fig. 8B). To show that the cockroach could learn more than a retinotopically defined direction, a visual reference cue (a white LED) positioned 90 away is added to the green conditioning LED. The reference cue is illuminated well before the green LED is flashed and remains lit for several seconds after the green LED is extinguished. Its addition causes sharper spatial tuning of the pointing response: the probability of a response now drops more steeply as the green LED is shifted away from the training position (Fig. 8C). The response to the green LED also diminishes if the reference cue is shifted away from the training position, but the green LED is left in place (Fig. 8D). Finally, when both reference cue and green LED are shifted

134

THOMAS S. COLLETT ET AL.

NAVIGATIONAL MEMORIES IN ANTS AND BEES

135

together, maintaining the learnt 90 angle between the two, the cockroach responds over a wider range of retinal positions (Fig. 8E). The cockroach learns both the absolute retinal coordinates of the green LED and the relative 90 separation between the two cues independently of their absolute retinal positions. The reference cue, when properly positioned relative to the green LED, enhances the response to the green LED, and, because the two are linked, can increase the range of retinal positions over which the green LED evokes a pointing response. 2. Features Encoded in a Snapshot Information about the likely contents of snapshots comes mostly from studies of pattern recognition and landmark guidance in honeybees. The studies suggest that bees parse a visual scene for different features and that memories are built up from a small palette of such features, including apparent size, edge orientation (Srinivasan et al., 1994), color (Hempel de Ibarra et al., 2002; Lehrer, 1999; Menzel and Lieke, 1983), and the vertical center of gravity of a shape (Horridge, 1998). The same features are used by Drosophila when discriminating visual patterns (Ernst and Heisenberg, 1999; Tang et al., 2004). Further experiments suggest, but do not prove, that these features may be encoded retinotopically. In one study, bees failed to recognize a single oriented bar shown on a stimulus card at the end of a Y maze when the bar was shifted vertically from its training position on the card (Horridge, 1998). In a more complex situation, honeybees were trained in a Y maze to discriminate between rewarded and unrewarded multicomponent patterns (Stach et al., 2004). Each pattern was divided into four quadrants with

Fig. 7. Snapshot matching in wood ants (F. rufa). Ants are trained to approach the bottom of a cone for food. (A) The track of a single ant walking toward the cone. The sketch on the right illustrates the hypothesis, derived from the multimodal distributions in B and C, that an ant stores several snapshots of the cone, which it has taken while fixating the center of gravity of the cone. (B and C) Approaches to a single black–white edge on a piece of vertical card. The edge is much longer than the cones, but is oriented at the same (15 ) angle to the vertical, corresponding to either the left or the right edge of the cone. The ant places these left and right edges in its left and right visual fields, respectively. The histograms, which show data from several approaches of one ant, plot the horizontal positions of the bottom of each edge on the ant’s retina. The distributions of these horizontal positions are multimodal. Note the corresponding positions of the modes in the ant’s approaches to the two edges. (D and E) Approaches of ants trained with an inverted cone. If snapshots were recorded while ants face an inverted cone, the apex of the cone would fall on the same position at the front of the retina, wherever the snapshot was recorded. (E) During approaches to a single black–white edge, the horizontal position of the bottom of the edge on the retina is mainly at the front of the retina (adapted from Judd and Collett, 1998).

136

THOMAS S. COLLETT ET AL.

Fig. 8. Spatial learning in the cockroach (Periplaneta americana). (A) Experimental arrangement. The restrained cockroach is placed in the center of an arena. White light‐emitting diodes (LEDs; A–D) and green LEDs (1–5) are placed in its left and right visual fields, respectively. (B) Cockroach is trained to expect a food odor when green LED is illuminated at position 1. In this experiment there is no white LED. During tests, only visual cues were provided and antennal pointing at the green LED drops as it is shifted from 1 to 4. (C–E) Reference LED in position A is also lit during training. (C) Probability of response drops faster than in B as the

NAVIGATIONAL MEMORIES IN ANTS AND BEES

137

the edges in each quadrant oriented differently from those in the other quadrants. The rewarded and unrewarded patterns differed only in the arrangement of orientations among the four quadrants (Fig. 9). A variety of tests indicated that the bees had learnt the four orientations of the rewarded pattern and suggested that bees had associated each orientation either with a specific retinal quadrant or with its position relative to the Y maze. A similar uncertainty applies to the interpretation of Horridge’s results (1998). 3. Extent of Snapshots Kwon et al.’s study of cockroach pointing (Fig. 8E) supports the notion that snapshots encompass a large area of retina. Several earlier experi˚ kesson and ments on bees (Cartwright and Collett, 1983) and ants (A Wehner, 1997) also imply that snapshots are extensive. For instance, honeybees were trained to feed at a point defined by three cylinders, which seen from the feeding site were separated from one another by 60 (Fig. 10A). When tested with the cylinders either moved further apart or closer together, the bees searched where the angle between the landmarks was the accustomed 60 , but where the apparent size of each cylinder was smaller or larger than in training. The maintenance of a 60 interlandmark angle indicates that the bees’ snapshot must extend for at least 60 . Analogous experiments on wood ants also suggest that snapshots are large (Durier et al., 2003). Wood ants were trained to find food in the center of an equilateral array of three cylinders (Fig. 10B). The ants’ search peak when food was absent was focused in the center of the triangle. The peak became dispersed when the triangle was slightly distorted from the training arrangement so as to give two possible matching solutions, with one position where the apparent sizes of the cylinders matched those in the snapshot and another where the angles between landmarks were correct (Fig. 10C). If the extent of the ants’ snapshot were limited to a single cylinder and ants reached their goal by facing the three cylinders sequentially and matching the size of each in turn, the ants’ search would peak where the apparent size of each cylinder is correct (Nicholson et al., 1999). The broader search pattern suggests that ants learn both the

green LED is moved from the training position. (D) Probability of response also drops when the green LED remains at 1, but the reference LED is shifted from the training position at A. (E) The response generalizes over a wider area if both reference and green LEDs are shifted together, maintaining the training separation. Cockroach learns both the retinal positions of the LEDs and the separation between them (adapted from Kwon et al., 2004).

138

THOMAS S. COLLETT ET AL.

Fig. 9. Honeybees (Apis mellifera) trained to a multicomponent pattern seem to learn the retinal position of each component. (A) Training patterns. Bees were trained in a Y maze to choose a sequence of six different rewarded (Sþ) patterns over six unrewarded (S–) patterns. The constant feature among the six Sþ patterns was the orientation of the bars in each quadrant. The bars in equivalent quadrants in Sþ and S– were perpendicular to each other. (B) Bees preferred a simplified Sþ pattern over a simplified pattern in which three quadrants had the same orientation as in Sþ and one quadrant was as in S–. Arrows indicate the S– quadrant. (C) Bees prefer patterns with three Sþ quadrants and one S– quadrant over a simplified S– pattern (adapted from Stach et al., 2004).

apparent size of each cylinder and the angles between the cylinders and that their snapshot extends at least 120 from the midline. Extensive snapshots which include widely distributed objects aid the accuracy of both scene recognition and goal localization.

NAVIGATIONAL MEMORIES IN ANTS AND BEES

139

Fig. 10. Honeybees and wood ants learn angles between landmarks. (A) Honeybees are trained to find food at a point defined by three black cylindrical landmarks. Whether the array is in its training arrangement or is expanded or contracted, bees search (gray squares) where the horizontal angles between the landmarks are 60 (adapted from Cartwright and Collett, 1983). (B) Search distribution of wood ants trained to find sucrose in the center of an equilateral triangle formed by three cylinders. During training, triangle is moved within the experimental arena. The ants’ search, in tests with sucrose missing, peaks at the expected feeder location (F). (C) Search during tests in which the triangle is distorted from training so that at one point (S) the apparent size of each cylinder is correct and at the other site (I) the angles between cylinders is 120 . Search is significantly more dispersed in C than in B, indicating that ants store both the apparent size and the spatial arrangement of landmarks (adapted from Durier et al., 2003, with permission from Elsevier).

B. MATCHING SNAPSHOT TO WORLD 1. Orienting Snapshot and World A major problem in using stored, retinotopically defined views for guidance is how to match stored views to the outside world. Different solutions related to an insect’s locomotor style are found in different insects. In honeybees and some wasps, matching appears to be a two‐step process in which the first step is for the insect to rotate so as to align its snapshot with the world and the second is to translate until snapshot and current

140

THOMAS S. COLLETT ET AL.

image become congruent. Because wasps and honeybees can fly sideways as well as forward, they can translate in all directions without changing their body orientation. Once in the vicinity of the goal, they assume a constant body orientation (Fig. 11), which correlates with the one that they adopted when they acquired the snapshot (Collett, 1995; Collett and Lehrer, 1993; Zeil, 1993a,b). To appreciate the advantages of fixing orientation, consider a site specified by just a single cylindrical landmark. If the insect’s orientation were not fixed, the snapshot on its own would not define a single location. An insect keeping only the apparent size and retinal position of the landmark at the correct values would search in a circle at a fixed radius around the landmark. For its search to be restricted to a single location, it must also fix its orientation. Bees can control their body orientation with the aid of a celestial or magnetic compass. Honeybees, which have been trained to find a goal relative to a rotationally symmetrical landmark, use their compass to break symmetry and search predominantly in a single location. Evidence for compass use is that the bees will search in a single position relative to the landmark after other directional cues from the surrounding panorama have been eliminated (Collett and Baron, 1994; Dickinson, 1994) or changed by shifting the bees to a new location, where they must approach the landmark from unfamiliar directions in an unfamiliar landscape (Lindauer, 1960).

Fig. 11. Snapshot matching: wasps adopt a constant orientation when close to a goal. (A and B) Two approach flights by one wasp (Vespula vulgaris) to an inconspicuous sucrose feeder (þ) marked by a cylinder ( ). The wasp’s orientation and position is shown every 20 msec. (C) Circular histogram of wasp’s orientation when it was within 7 cm of the feeder. Data from 28 approaches (adapted from Collett, 1995).



NAVIGATIONAL MEMORIES IN ANTS AND BEES

141

The particular direction in which a wasp or bee takes snapshots, and by which it guides its return can be set by details of the landscape (Zeil, 1993a,b) or the task (Frier et al., 1996). Thus, the solitary wasp, Cerceris rybyensis, on emerging from its inconspicuous nest hole, performs learning flights in which it looks in the general direction of a prominent landmark close to the nest entrance (Zeil, 1993a). It then backs away so that the nest comes to lie between it and the landmark. The process is mirrored on return flights when the wasp approaches its nest so that the landmark is viewed beyond the invisible nest hole (Zeil, 1993b). With this strategy, the wasp can keep the cylinder in its frontal field all the way to the nest. Ants, which have little ability to move sideways, align world and snapshot a little differently. Wood ants following routes indoors face conspicuous landmarks, both when acquiring and using snapshots (Durier et al., 2003; Judd and Collett, 1998; Nicholson et al., 1999). The matching process can then be simplified by limiting it to periods of landmark fixation. Outside, where the ant’s sky compass can be used, compass cues contribute to accurate matching. Ants usually approach a goal along a fixed route. Matching will be safest if ants take snapshots in the compass direction of their habitual approach to a goal and later match snapshots when facing in the same direction. Behavior of this kind is found in desert ants, which like honeybees will employ compass information to break symmetry when locating a goal within a symmetrical ˚ kesson and Wehner, 2002; Fig. 12). Ants were accustomed landmark array (A to find their nest in one corner of a square array of landmarks. They returned there along a fixed path after foraging at a feeding site. In tests on unfamiliar terrain, ants were released from different compass points relative to the landmark array. The ants generated a single search peak when approaching along their habitual path, but produced two search peaks when approaching from other directions. The first peak was in a position consistent with the ants’ learnt compass bearing and the second peak was specified by the ants’ current approach direction. 2. Image Matching With snapshot and world aligned, the insect moves to make the two congruent. The details of this process are uncertain, although a number of successful algorithms have been devised (Cartwright and Collett, 1983; Lambrinos et al., 2000; Vardy and Mo¨ller, 2005; Zeil et al., 2003) that work in simulation or on robots. A plausible heuristic for real insects is to parse a scene for features, which are paired with corresponding features in the snapshot. The insect’s overall direction of motion is derived from differences in the retinal locations of such paired features. Because there is redundancy in the information from multiple paired features, and because the process is repeated many times during the approach, the behavior is robust.

142

THOMAS S. COLLETT ET AL.

Fig. 12. Ants (Cataglyphis fortis) use compass cues to aid snapshot guidance. Ants learn to find their nest in the southeast (SE) corner of an array of four cylinders and are accustomed to approaching it from the SE. If ants ignore compass information, a visual snapshot recorded at the nest would be matched in all four corners. To see whether ants disambiguate the four possible matching sites, a similar array of cylinders is placed on unfamiliar terrain and ants are

NAVIGATIONAL MEMORIES IN ANTS AND BEES

143

To be able to pair a feature between snapshot and current image, the feature in the current image must be identifiable when it lies at some retinal distance from its learnt position in the snapshot. Thus, to some extent the recognition of features must be translationally invariant. Unequivocal demonstration that insects can recognize features at novel retinal locations is recent (cockroaches: Kwon et al., 2004; fruitflies: Tang et al., 2004). To do so required experiments in which a feature’s retinal location was fixed during acquisition and then shifted during recognition tests. An insect limited to a small palette of unique features may find multiple instances of the same feature within a cluttered scene. In such cases, the pairing of individual features over large retinal distances is likely to generate false matches. The likelihood of erroneous pairing can be reduced by restricting matching to a limited retinal area around the learnt position of each feature in the snapshot. We suggest that a matching retinal feature within this area is captured and that the insect moves so as to shift the feature toward its position in the snapshot, while similar features beyond this limit are ignored. Since a given feature may occur in combination with several other features, the likelihood of finding a feature in a particular combination is usually less than the overall likelihood of finding that feature—in a piece of writing the occurrence of a downstroke in an L is less frequent than the occurrence of downstrokes in all letters (L, T, P, and so on). Thus, linkage between separate features in a snapshot can lessen the danger of false matches and allow enlargement of the area within which pairing is permitted. The cockroach in the experiment of Fig. 8 could be said to follow this principle. The probability of its antennal pointing drops less rapidly, if both the white reference and green LEDs are shifted in register away from the training position, than if just the green LED is moved. The storage of the angular separation between features independently of retinal position is also interesting in providing empirical support for an additional component to image matching that has been suggested by simulation studies (Cartwright and Collett, 1983). Snapshot matching can be made more effective by moving toward or away from two linked features to make them transform to the stored separation independently of their absolute retinal positions.

released from points around the array (center panel). (A) Ants taken at the nest with zero home vector and released at ‘‘a,’’ 2 m SE of the array, searched only at the fictive position of the nest (dark cloud). (B) Zero vector ants released at ‘‘b,’’ 2 m SW of the array, generated two search peaks. The first peak was close to the point where the ants entered the array. The second larger peak was at the fictive position of the nest. Both approach direction and compass ˚ kesson and Wehner, 2002). cues aid snapshot matching (adapted from A

144

THOMAS S. COLLETT ET AL.

C. VECTOR MEMORIES 1. Global Vectors In addition to memories of landmarks, routes are also guided by memories of vectors, which come in two distinct forms. The first kind of vector memory contributes to navigation by global PI (see Section II). Since global PI always tells an insect its coordinates with respect to the nest, the insect can store the coordinates of any significant site that it reaches, such as a feeding location and use the information on subsequent trips. Honeybees employ this knowledge in a unique manner that allows bees that are working a rich patch of flowers to transmit its location to other bees in the hive. Bees on returning to the hive broadcast the location of the site by performing a ‘‘waggle’’ dance that encodes the distance and direction of the site from the hive (von Frisch, 1967). The dancing bee communicates information that it has obtained through PI and bees reading the dance perform a flight vector that takes them to the rough position of the signaled food site (Fig. 13; Riley et al., 2005). Desert ants also store the location of a food site in terms of its global PI coordinates (Collett et al., 1999; Schmid‐Hempel, 1984; Wehner et al., 1983; Wolf and Wehner, 2000). One way in which ants display this knowledge is by compensating for enforced detours. At the end of a detour, they turn and travel directly to their food site. In the example of Fig. 14, desert ants ran along a straight channel between their nest and foraging site. The channel gave them a view of the sky, so that they had directional cues, but its walls hid the surrounding landscape (Collett et al., 1999). In tests, the ants followed a rotated channel to an unfamiliar location, where they had to decide what direction to take. The ants knew their current PI coordinates and also those of the feeding site, and they moved in the direction of the feeder as specified by the difference between these two sets of coordinates. Thus, the ants seem not to store a specific movement vector. Instead, they remember the PI coordinates of the food site and compute the movement vector to it by a process analogous to subtraction. We have described PI in terms of coordinates and locations to stress that global PI involves the use of positional or maplike information. An insect knows both its own position and the location of significant sites in terms of their PI coordinates relative to an origin such as its nest. And we have seen that this positional information is employed flexibly to allow novel routes to be performed over unfamiliar terrain. It is significant that in ants, at least, this PI information is isolated from other navigational strategies. Familiar visual landmarks along a route do not acquire PI coordinates and the performance of novel paths, using PI, is only possible if the ant has itself traveled from its nest to the place where it plans a new route to a known

NAVIGATIONAL MEMORIES IN ANTS AND BEES

145

Fig. 13. Foodward and homeward vectors in honeybees recorded with harmonic radar. (A) The bees flying foodward vectors were all new recruits. They were recorded after they had followed a waggle dance. An antenna was attached to an individual bee that was either released at the hive or after displacement (adapted from Riley et al., 2005, with permission from MacMillan). (B) Homeward vectors after feeding. Bees were caught at the feeder and released elsewhere. In all cases (A and B), bees fly the appropriate vector showing that the performance of global vectors is largely independent of landmarks (adapted from Riley et al., 2003, with permission from the Royal Society).

destination. If an insect is passively displaced, by an experimenter or a gust of wind, its path computed by PI will miss the goal by the magnitude of the displacement, even within familiar terrain (Wehner, 1992). Route planning with global PI information seems to be independent of visual landmarks. There is no evidence that ants use landmarks to reset their current global PI coordinates when they are displaced to or unexpectedly

146

THOMAS S. COLLETT ET AL.

Fig. 14. Trajectories to a food site guided by path integration. (A) Ants were trained to find food at the end of a 16‐m channel from which the natural landscape was invisible. (B and C) In tests, the channel was shortened and the trajectories of ants after they left the channel were recorded. After exiting a channel of the same orientation as in training (B), ants head toward the feeder location. When the channel is rotated by 45 (C), ants turn through 90 at the end of the channel and again head toward the feeder location. Since the landscape on exiting the channel was unfamiliar, the ants’ trajectories must have been guided by PI. The underturning is characteristic of this experimental arrangement. There was no feeder in B and C. Grid squares are 1 m across (adapted from Collett et al., 1999, with permission from Elsevier).

find themselves in a location with familiar landmarks (Collett et al., 1998, 2003; Knaden and Wehner, 2005; Sassi and Wehner, 1997). Thus, the accuracy of global PI is not improved by noting landmark features, and global PI is not used to provide landmarks with positional coordinates. Because the distance (Sommer and Wehner, 2004) and directional errors in global PI increase with distance traveled, the insulation of global PI from landmark information makes navigation by global PI less precise than following visually guided routes. In contrast to global PI, the errors accumulating along a visually guided route can be eliminated whenever a

NAVIGATIONAL MEMORIES IN ANTS AND BEES

147

familiar landmark is encountered. The relative weighting that ants give to the two navigational strategies is consistent with this difference in accuracy. When available, ants seem to use information associated with routes and landmarks in preference to global PI (Collett et al., 1998, 2001; Kohler and Wehner, 2005; Wehner et al., 1996). Navigation by global PI is then held in reserve and is only employed to reach a destination in special circumstances such as on bare terrain, or when landmarks are missing, or are too distant to provide positional information, or if an ant performs a detour after being chased far off its route by a predator. 2. Local Vectors Bees and ants also store vectors as a specific part of a route. These local vectors are usually associated with landmarks so that on encountering a landmark an insect retrieves a vector, which makes it depart in the correct direction along the next leg of its journey toward an intermediate or a final goal. In bees, these learnt local vectors have a defined distance and direction (Chittka et al., 1995b; Collett et al., 1993, 1996; Srinivasan et al., 1997). In ants only the directional component of the vector has been established definitively (Bisch‐Knaden and Wehner, 2003; Collett et al., 1998). An example from desert ants is given in Fig. 15. An ant has been trained along an L‐shaped route in which it travels 8 m north from its nest and then turns through a right angle into an open‐topped channel with a view of the sky and a feeder at the channel’s end (Collett et al., 1998). The channel is placed in a trench dug into the sand so that it is invisible to ants once they have left it. The ant, on its return from the feeder, leaves the channel and turns through a right angle to travel south to its nest. If the ant is allowed to return almost to its nest, so that its home vector is nulled, and then caught and released at the end of a similar hidden channel on a test ground, the ant repeats the initial part of the route and heads roughly south on leaving the channel (Fig. 15B). This directional response cannot be due to global PI. Nor is it guided by visual landmarks as the buried channel is invisible to the ant from outside the channel. The dramatic change in view as the ant leaves the channel for open sand provides a landmark to which a local vector can be attached. One sign that insects rely more on local vectors that are associated with routes than on global PI is the outcome of conflicting instructions from local and global vectors. In Fig. 15C and D, the two are put into conflict by taking ants from the feeder on the training ground and releasing them at the end of a channel that is half the training length. The local vector then still points south, but the global vector rotates to point at the virtual position of the nest. The majority of ants head south before adopting their global vector, implying that initially the local vector takes precedence (Fig. 15D).

148

THOMAS S. COLLETT ET AL.

Fig. 15. Local vectors often take precedence over global PI vectors. Ants were trained on an L‐shaped foraging route. The homeward path consisted of 8 m in an open topped channel (thick line) that was hidden in a trench in the sand followed by 8 m over open ground to the nest. (A) The ‘‘homing’’ trajectories of ants taken from the feeder and placed at the end of an 8 m channel similar to the training channel, but set in the unfamiliar landscape of a test ground. The trajectories are oriented southward, as they are in training, toward the fictive position of the nest ( ). (B) The trajectories of ants allowed to return to the nest before being placed in a 4 m channel on the test ground are also oriented southwards, even though their global PI vector is zero. These trajectories are local vectors triggered by leaving the channel. (C and D) Trajectories from ants taken from the feeder to the end of a 4 m channel can be separated into two groups. (C) Some ants follow their home vector from the end of the channel to the fictive nest site. (D) Most ants head due south following a local vector for about 2 m, before taking a global vector to the fictive nest. Performance of the global vector is suppressed while the local vector runs (adapted from Collett et al., 1998).



IV. THE RETRIEVAL

OF

MEMORIES ALONG

A

ROUTE

Figure 4 illustrates an ant continuing a homeward route when displaced to a site partway along it. The remainder of this chapter is concerned with the ramifications of two questions raised by this example. First, how does the ant manage to recognize its location and to retrieve route memories out of sequence? Second, how does the ant know which route to follow out of several possible routes that might pass through that place? That the ant in Fig. 4 chose to finish its homeward route suggests that the set of memories belonging to that route was selectively primed until the ant reached its nest.

NAVIGATIONAL MEMORIES IN ANTS AND BEES

A. RECOGNITION, RETRIEVAL,

AND

149

BINDING

Retrieval of a snapshot is more than recognizing a pattern. In pattern recognition, an animal only has to decide whether a particular example can be accepted as an instance of the remembered pattern. Snapshot retrieval seems to involve an additional step. Once the current visual scene is accepted as an instance of a particular snapshot, the entire snapshot should be retrieved so that all its features can be matched to the world. The problem of snapshot retrieval is especially difficult because a snapshot is only useful for navigation if it is retrieved while there remains a discrepancy between it and the current view. In Section III.A.3, we summarized evidence that landmark memories are extensive and may comprise several spatially separate components. Since different snapshots may have features in common, an insect in a particular landscape may find that several of its snapshots are candidates for retrieval. Choice of the correct snapshot would be easier if the separate features of a snapshot were bound or linked together. The recognition of one feature would then be enhanced by the recognition of another and this mutual facilitation would emphasize the best matching snapshot. Here, we first present evidence that the separate features of a snapshot are bound together. We then consider the roles of panoramic context and sequential priming in aiding snapshot retrieval along a route.

B. BINDING WITHIN A SNAPSHOT It is difficult to obtain firm evidence for the binding of the components of visual stimuli. One experimental approach has been to train bees to two sets of rewarded and unrewarded patterns composed of the same features in different combinations (Menzel et al., 2001). For instance, the two rewarded patterns might be feature A combined with feature B (ABþ) and feature C combined with feature D (CDþ), while the two unrewarded patterns might be AC– and BD–. In this case, the single features A, B, C, and D are rewarded and punished equally often. Only the combinations are informative. Bees can solve problems of this kind (Deisig et al., 2001, 2003; Fauria et al., 2000; Giurfa et al., 2003; Schubert et al., 2002). A potential difficulty with this experimental design, which is not always easy to overcome, is that the ‘‘untrained’’ response of neurons to an individual feature may vary according to the presence or absence of other features, so violating the basic assumption of the experimental design that initially the individual features are independent. A variety of other experimental designs also suggest that separate features of a compound visual pattern can be bound together. One example that we have already discussed (Kwon et al., 2004; Fig. 8) is the linkage the

150

THOMAS S. COLLETT ET AL.

cockroach forms between the two LEDs (Fig. 8). In another example (Graham et al., 2004), the recognition of one snapshot feature helped disambiguate the identification of another. One group of wood ants was trained within a white‐curtained arena to find food midway between a small and a large cylinder, where the two cylinders subtended different angles at the goal (Fig. 16A). Trained ants were tested with the small and large cylinders replaced by two cylinders of the same intermediate size. When the ant reaches the midpoint between the two cylinders, the intermediate cylinders will look the same size. Ants seemed unable to identify which of the two cylinders was a replacement for the large cylinder and which for the small. They searched midway between the cylinders rather than in the predicted position closer to the cylinder that replaced the large one. Ants needed an additional cue to perform correctly. A second group

Fig. 16. Evidence that components of a snapshot are bound together. Ants were trained to a feeder midway between a small and a large cylinder (filled circles). In first experiment (A) the arena was surrounded entirely by white curtains. In a second experiment one wall parallel to the array was patterned (B). Middle row: search distributions with training array. Bottom row: search distributions with two intermediate‐sized cylinders. Open circles represent position of feeder as predicted by a snapshot in which the identity of each cylinder is known. Ants search correctly in both training conditions. They search correctly in tests if they have been trained and tested with the patterned curtain (B). They fail to search correctly in tests when they were trained and tested with a white curtain (A). The patterned curtain helps ants know which intermediate cylinder represents the small cylinder and which the large, but it cannot itself give locational information, since after each trial cylinders and food are shifted along the dashed line (sketch at top) relative to the curtain (adapted from Graham et al., 2004).

NAVIGATIONAL MEMORIES IN ANTS AND BEES

151

of ants was trained with one of the white curtains in the test arena covered with large black shapes (Fig. 16B). These ants searched correctly when tested with the intermediate cylinders. A plausible explanation of the ants’ success with the patterned curtain in place is that the ants’ snapshots bind together features of the patterned curtain and the cylinder. Ants close to their goal and facing one cylinder will not see the other cylinder that lies directly behind them. Their search is driven primarily by the appearance of the cylinder that they are facing (Fig. 5). Because the snapshot of each frontally viewed cylinder includes the patterned curtain, ants can identify the cylinder that they are facing. The snapshot including the small cylinder is retrieved when the patterned curtain is in the ants’ left visual field, and the snapshot including the large cylinder is retrieved when the patterned curtain lies in their right visual field. C. PANORAMIC CONTEXT In the previous example, the patterned wall helped in retrieving the appropriate landmark memory. There is a similar division of function in more natural landscapes between the distant panorama and landmarks that are close to a route or a place. Landmarks near to an inconspicuous goal can pinpoint its location because the landmarks’ appearance on the retina changes appreciably as an insect moves within the immediate neighborhood of the goal (Cheng et al., 1987). While these changes are good for the localization of a goal, they make it difficult to retrieve the correct snapshot. In contrast, the more distant panorama, which has a relatively constant appearance near to the goal, is of little use in precise localization but it can identify the rough area and so provide a unique context for snapshot retrieval. When an insect is close to a particular landmark, the associated panoramic cues induce strong expectations of encountering that landmark so that local evidence for its identification can be weaker. Honeybees will navigate by a landmark set in its usual context, even if the appearance of the landmark is substantially altered from what the bees are used to (Collett and Kelber, 1988). Conversely, ants will ignore a familiar landmark seen in the wrong context (Graham et al., 2003). To demonstrate the importance of panoramic context in memory retrieval, honeybees were trained to collect food in two far apart places with different panoramas (Collett et al., 1997). In place 1, bees flew through a maze in which they learnt to fly left in an arena on seeing a 45 diagonal grating of black and white stripes filling the circular back wall of the arena and to fly right when the wall was yellow (Fig. 17). In place 2, the same bees encountered a similar maze, but in this case they had to fly right on seeing a 135 striped

152

THOMAS S. COLLETT ET AL.

Fig. 17. Pattern recognition and spatial context in honeybees. (A and B) Bees were trained in two places to fly through a cylindrical compartment (radius 70 cm) and to leave the compartment through one of two holes in the back wall. The direction to be taken depended on the stimulus that covered the entire back wall. (C) In place 1, bees flew left if the wall was covered with black‐and‐white stripes oriented at 45 , and to their right when the wall was plain yellow. (D) In place 2, bees flew to the left when the back wall was blue and to the right when the back wall was covered with stripes that were oriented at 135 . Bees were then tested with the two colors and with 45 , 90 , or 135 stripes in both places. Mean trajectories that are plotted in the bottom row show that bees responded correctly to 45 and to 135 gratings in both places, showing that they were sensitive to grating orientation. Bees also responded correctly to the colors (data not shown) in both places. But a vertical (90 ) grating was treated differently in the two contexts. In place 1, bees flew left as though the vertical grating was oriented at 45 , and, in place 2, bees flew right, as if the vertical grating was a 135 grating (adapted from Collett et al., 1997).

grating in the arena and to fly left on seeing blue. After this training, they responded correctly to the two gratings and the two colors in both places, implying that they were sensitive to grating orientation and to color and could recognize the training patterns irrespective of the panorama outside the maze. But they treated a vertical (90 ) grating very differently in the two places showing that the panoramic cue primed recognition. In place 1, bees flew left, as though it were a 45 grating, and in place 2 they flew right, as if it were a 135 grating. Thus, in one place panoramic cues prime recognition

NAVIGATIONAL MEMORIES IN ANTS AND BEES

153

of the 45 grating and broaden what the bees accept as belonging to that category so that it includes a 90 grating. Conversely, in the second place, the 135 category is broadened. D. SEQUENTIAL PRIMING Insect routes are sufficiently stereotyped that what an insect sees during one stage of its route can facilitate or prime the retrieval of a memory of a landmark or action to guide the insect along the next stage of its route. Some evidence for such sequential priming comes from seminatural routes in honeybees (Chittka et al., 1995a). But most experimental work has been conducted in small‐scale Y mazes in which bees are taught two routes, each consisting of a sequence of two or three visual stimuli (Giurfa et al., 2001; Srinivasan et al., 1998; Zhang et al., 1999, 2005). The bees’ task is to follow one or other route. At the entrance to the Y maze, the bee sees a sample stimulus; for instance, a prominent patch of blue or green. The color of this stimulus tells the bee which comparison stimulus she should approach at the choice point. For example, a green stimulus might indicate that the bee should approach the arm of the Y maze displaying a pattern of horizontal black and white stripes, but should avoid the pattern of vertical black and white stripes. If the sample color is blue, the bee should do the reverse: approach the vertical stripes and avoid the horizontal ones. The bee’s correct choice of stimulus in the Y maze shows that it has acquired two sequences of stimuli, symbolized for brevity as A‐B and W‐X and that seeing A primes the retrieval of a memory for B and seeing W primes the retrieval of a memory for X. It is the sequential priming of B by A and of X by W that enables the bee to follow the two routes correctly. Bees will also perform what is called a delayed matching to sample task in which seeing blue as the sample color tells the bee that it must later choose blue over green in the Y maze and seeing green tells the bee to choose green in preference to blue. Here the two sequences are A‐A and X‐X. Zhang et al. (2005) reported data from a delayed matching to sample task demonstrating the kinds of interactions that can occur between spatial and sequential priming. Bees flew along a narrow channel, where they first saw a sample pattern fixed to a baffle. The sample patterns were diagonal blue‐and‐white stripes oriented 45 clockwise from the vertical for one sequence and at 45 anticlockwise for the other. The channel opened into a decision chamber containing both patterns of stripes. Bees learnt to reach a sucrose reward by approaching the stripes that matched the sample that they had viewed previously in the channel. Throughout training, the sample was always placed 120 cm from the decision point. In tests, the other sample pattern was also present. On some test trials this irrelevant sample

154

THOMAS S. COLLETT ET AL.

pattern was at 50 cm from the decision point and in other tests it was at 170 cm. Bees were expected to see the sample at the fixed 120‐cm training distance and their performance was not at all degraded by the addition of an irrelevant sample elsewhere in the channel. When given one sample at 50 cm and the other at 170 cm from the decision point, they chose randomly between the comparison patterns. Bees probably learnt the sample pattern within a spatial context that was defined by external visual landmarks visible from the channel so that sequential priming was strongest when the sample occurred in its usual spatial context at 120 cm from the choice chamber. For convenience of exposition, we have considered the question of associative binding between the components of a snapshot separately from that of binding between a snapshot and the spatial and temporal contexts in which the snapshot is acquired. But, in fact, the distinction between a snapshot and its context is somewhat hazy. There is, for instance, no evidence that distant landmarks with a more contextual role are encoded differently from close ones. Also, it is not always easy to separate panoramic from sequential contexts. An ant negotiating bumpy ground may well see its surroundings one moment and then have them disappear. Similarly, a bee descending to enter a maze will no longer view the surrounding panorama (Collett et al., 1997). There are signs that bees themselves may blur such distinctions by making the effects of any contextual stimulus persist for a while after it has disappeared. Thus, a bumblebee, which has learnt to approach a visual stimulus in one spatial context, continues to be primed by that context when a delay of a few seconds is imposed between presenting the spatial context and the linked stimulus (Fauria et al., 2000). E. INTERFERENCE EFFECTS

WHEN

CONTEXTUAL CUES ARE LACKING

With no distinguishing contextual cues to indicate which memory should be recalled, bees and ants may retrieve several memories simultaneously. Menzel (1969) gives an early example of interference between color memories in reversal learning. Honeybees learn in very few trials to forage at a stimulus of one particular color and to avoid another color. When bees had to reverse their choice after each block of six trials (i.e., a block of six trials of Aþ/B followed by a block of six trials of Bþ/A), their performance deteriorated rapidly over successive blocks. After four blocks, bees chose at chance within each block. In subsequent tests both A and B were preferred over a new color, C, showing that reversal training had not obliterated the bees’ learnt attraction to A and B. Rather bees had no context to indicate which color to retrieve and had not learnt to link the temporal pattern of their choices to the temporal pattern of the presented stimuli. In contrast, bees learn readily such contradictory stimulus–response

NAVIGATIONAL MEMORIES IN ANTS AND BEES

155

associations when Aþ/B is acquired in one temporal or spatial context and Bþ/A in another (Menzel et al., 1996, 2001). Cheng (2005) has compared directly the performance of honeybees in situations with and without differential contextual cues. Bees had to acquire two memories. In task 1, they learnt to find a reward about 10 cm north of a landmark that was shifted from trial to trial, and their correctly positioned search distribution was recorded in an unrewarded test after 10 rewarded acquisition trials. Immediately afterward, they learnt task 2 in which they found food south of a different landmark. Task 2 was either acquired in the same or in a different spatial context from task 1. After 10 acquisition trials on task 2, the bees were tested again on task 1. If tasks 1 and 2 were learnt in the same spatial context, the two tasks interfered with each other and the bees no longer searched correctly. If the spatial contexts of tasks 1 and 2 differed, the acquisition of task 2 did not degrade the ants’ performance when they were retested on task 1. Because an insect’s path can be recorded in detail, interactions between simultaneously retrieved navigational memories are particularly informative. The outcome of retrieving two memories may be guidance to an intermediate location rather than confusion. The example of Fig. 17 shows what happens when bees are trained to fly one vector on seeing 45 stripes in one spatial context and another vector on seeing 135 stripes in the second context. Bees trained this way treat 90 stripes like 45 stripes in the 45 context and like 135 stripes in the 135 context. In contrast, when two vectors are learnt sequentially in a two‐compartment maze so that panoramic cues are the same for the two vectors, 90 stripes evoke an intermediate vector, as though bees had retrieved both vectors and averaged the result (Collett et al., 1996; Fig. 18). Averaging vectors in this way might even be useful in helping bees to reach a destination from an unusual starting point (Collett and Baron, 1995; Menzel et al., 1996, 1998). Suppose a bee acquires one vector memory that will take it to a goal from one starting location and a second vector taking it to the same goal from a second starting location. Suppose also that the two locations are sufficiently close that their associated snapshots have large visual landmarks in common. An intermediate view seen from the two starting locations might evoke both snapshots and their attached vectors, which if averaged would point at the goal.

V. CHOICE

OF

ROUTE

AND

DESTINATION

Since several routes that an insect has learnt may pass through a single place, recognition of that place will not by itself ensure the retrieval of the navigational memories relevant to the insect’s chosen route. We have seen

156

THOMAS S. COLLETT ET AL.

Fig. 18. Trajectories of honeybees trained in a two‐compartment maze in a single place. Bees are trained to fly to their right in the first compartment, which has 135 stripes covering the entire back semicircular wall, and to fly to the left in the second compartment, where 45 stripes cover the back wall. In tests, the responses of bees to different stripe orientations are recorded in the first compartment. (A) Plan view of maze and training conditions. Bees’ route to food (F) at the end of the maze is shown by the dashed line. (B) Average trajectories in the first compartment of bees presented with stripes of different orientations. In contrast to the data of Fig. 17, where the two orientations were acquired in different places, bees’ intermediate orientations elicit vectors in intermediate directions (adapted from Collett et al., 1996).

that sequential priming can help an insect to retrieve memories for one route rather than another. Here we consider cues, which are independent of the route itself, but which can prime the set of memories that underpin a particular route. An insect’s choice of which memories to retrieve is determined by its motivational state, whether, for instance, it is empty and should follow a route to a foraging site or is full and should return to its nest. Its choice can also be set by time of day that predicts which foraging route or which branch of a foraging route is likely to be most profitable. In honeybees, olfactory cues perceived within the hive also inform bees which route they should follow. We will see that both landmark memories

NAVIGATIONAL MEMORIES IN ANTS AND BEES

157

and vector memories can be primed by one or more of these factors, suggesting a general mechanism for route retrieval in which individual memories relating to whole or parts of a route are made ready to be retrieved in sequence as the insect makes its way along the route. A. FEEDING STATE

AND

ROUTE MEMORIES

Ants traveling between their nest and a feeding site guide their outward and homeward routes by different sets of memories. Because the sequence of vectors and views will differ on foodward and homeward trips, separate memories are needed, even when the two routes take the same path, as can happen in the desert ant C. bicolor (Fig. 2). To reiterate, how does an ant placed in the middle of a route (Fig. 4) know whether it should be recalling memories for its homeward trip? Feeding state turns out to control both what vectors a bee or ant performs and what visual memories they retrieve. Wehner et al. (2006) found that the Australian desert ant, M. bagoti, which often follows different outward and inward routes between its nest and foraging site (Fig. 19A; Wehner, 2003) behaves as though lost when it is displaced from a point on its homeward route to a release site on its

Fig. 19. Separate foodward and homeward routes in M. bagoti. (A) Habitual routes of one ant showing the spatial separation between the outbound and inbound trips. Gray areas represent corridors defined by five outbound and five inbound journeys (adapted from Wehner, 2003). (B) Separation between trips is facilitated by the placement of two barriers (thick lines). Ants caught toward the end of the homeward trip are released on their foodward route (R). They then behave as lost and unresponsive to signposts that are familiar when they run their foodward route in the proper motivational state (adapted from Wehner et al., 2006, with permission from Elsevier).

158

THOMAS S. COLLETT ET AL.

Fig. 20. (A and B) Vector switching in honeybees. (A) Bees flew between their hive (H) and a sucrose feeder (F). They were caught at the feeder and held for 3 hr before being released either at the feeder or at an unfamiliar site 1 km away (R). (B) If bees were fed while in captivity, their vanishing bearings were in the feeder‐to‐hive direction whether they were released at F or at R. If bees were unfed during captivity, their vanishing bearings from both release sites were in the hive‐to‐feeder direction. Arrows show mean vanishing bearings. Dashed line shows feeder‐to‐hive direction (adapted from Dyer et al., 2002). (C and D) Food induced directional response in the ant, Gigantiops destructor. (C) Ants beginning a foraging trip were taken from close to their nest to release sites R1–R3 10 m away and supplied with a

NAVIGATIONAL MEMORIES IN ANTS AND BEES

159

foodward route. The ant, when motivated to go back the nest, fails to respond to landmarks that it regularly uses for guidance when it is on the way to the feeder. To encourage good separation between foodward and homeward routes, long low barriers were placed on the ants’ terrain, one close to the feeder and the other close to the nest (Fig. 19B; Wehner et al., 2006). Ants familiar with this one‐way circuit were taken from a position on their homeward route close to the nest, so that no global PI vector would interfere with landmark guidance, to a point on their foodward route (Fig. 19B). Ants typically searched for a long time around the release point and did not follow the foodward route either back to the nest or to the feeder; nor did they head directly for the nest. The only familiarity with the terrain displayed by these homeward bound ants was when by chance they encountered the homeward route. In this case, they joined it immediately and followed it back to the nest. Foraging honeybees on their way to a familiar feeding site perform a flight vector that has roughly the appropriate direction and distance to take them from their hive to the feeding site, and they can find their way home from a feeding site by performing the inverse vector (Riley et al., 2003, 2005; Fig. 13). Dyer et al. (2002) have shown that whether bees perform a homeward or foodward vector depends on their feeding state. If bees are caught after feeding and held in captivity for some hours without further food, they will on release at the feeding site fly in the direction of the hive to the feeder instead of back to the hive (Fig. 20A and B). In contrast, if the bees are fed while in captivity, their flight vector on release is in the direction of the feeder to the hive, even after displacement to an unfamiliar site. A bee’s flight path switches direction according to its state of hunger. Analogous results are reported for the ant, Gigantiops destructor (Beugnon et al., 2005). Individual ants are faithful for long periods to a given foraging site, where they prey on small arthropods and collect nectar (Beugnon et al., 2001). An ant that is taken from the nest to an unfamiliar site will pick up a termite as a prey item and perform a short vector in its habitual nest to feeder direction (Fig. 20C and D). With no termites present, the ant engages in a protracted circular search. For bees and ants, feeding state can control what vector is performed irrespective of the visual surroundings. Visual memories appropriate for guiding foodward or homeward routes are also primed by whether an insect is unfed or has fed. The effect of prey item. They picked up prey and then headed off for a short distance (50 cm) in their usual homeward direction. (D) Mean directions of ant at 30 cm from release sites. Dashed line shows homeward direction (adapted from Beugnon et al., 2005).

160

THOMAS S. COLLETT ET AL.

feeding state on memory retrieval can be seen in two laboratory experiments on wood ants (Harris et al., 2005). In the first experiment (Fig. 21A and B), ants were trained along a short foraging route in which they were taken from their nest and released into a starting pot 20 cm from a black wall. From there they walked 1 m parallel to the wall to a drop of sucrose. After feeding they returned toward the start, from where they were carried back to the nest. The wall was to the ants’ left on the outward trip and to their right on the way home. Ants in such a situation guide themselves by keeping a feature, like the top of the wall, at a fixed retinal elevation (Graham and Collett, 2002). The paths of trained ants when released in the middle of the wall differed according to whether or not they were previously fed. Unfed ants moved so as to place the wall in the correct elevation in their left visual field, as it was in foodward journeys, whereas fed ants placed the wall in their right visual field (Fig. 21A). In another test, ants were started midway between two parallel walls (Fig. 21B). If foodward and homeward visual memories are primed differentially by feeding state, ants should be attracted more by the wall on their left when unfed and to the wall on their right when fed. The paths in Fig. 21 conform to this prediction. In a second series of experiments, ants learnt to run a single Y maze displaying the same two visual stimuli (a white–black edge and a black–white edge), both to reach sucrose and to return to their nest (Fig. 21C). They learnt to approach the white–black edge for food and the black–white edge to return home. For the ants’ outward trip, the arm of the Y maze showing the white–black edge led to a sucrose feeding pot. When an ant reached the feeding pot, the pot with ant inside was moved to the start of the maze. The ant, after feeding, made a second trip through the maze choosing the arm displaying the black–white edge to reach an empty pot, from which it was taken back to the nest. After ants had learnt to choose the correct path on both trips (Fig. 21D), they were given tests in which they made two unrewarded journeys through the maze. Ants taken from the nest were either fed or left unfed before being placed at the start of the maze. On both passages through the maze, unfed ants chose the white–black edge and fed ants chose the black–white edge, showing that the dominant priming cue was feeding state rather than the sequence of events experienced in the maze. B. TIME

OF

DAY

AND

ROUTE MEMORIES

Different flowers deliver nectar at different times causing the relative profitability of different routes to fluctuate with a diurnal rhythm that bees and ants will match by changing their routes during the course of a day

NAVIGATIONAL MEMORIES IN ANTS AND BEES

161

Fig. 21. Visual memories in wood ants are primed by feeding state. (A and B) Ants were trained to go from a start pot (white circle) along a route parallel to a 20‐cm high wall to collect sucrose and then return to the start. The trajectories of trained ants were recorded individually on release from the start pot, both when ants were unfed and after they had been fed. The start was either 20 cm from one wall (A) or midway between two walls 80 cm apart (B). The ants’ paths in the two cases vary with feeding state, such that unfed ants match a memory appropriate to their foodward journey in which the wall is in their left visual field, and fed ants match a homeward memory with the wall in their right visual field. (C and D) Ants traveled a foraging route through a Y maze (C), which displayed two different patterns. The ants’ first trip through the maze was to reach food, when they had to approach the white–black edge, and their second trip was to reach home, when the correct pattern to approach was the black–white edge. (D) Ants learnt to approach the appropriate patterns on their foodward and homeward journeys. Choices from each ant were binned across 10 trials. Plot shows mean choices with 95% CI (adapted from Harris et al., 2005).

162

THOMAS S. COLLETT ET AL.

(Harrison and Breed, 1987; Menzel et al., 1996, 1998; Ribbands, 1964; Schatz et al., 1999; Wahl, 1932). Time of day can also control which visual or olfactory stimuli honeybees approach at their destination. For instance, bees exposed to a feeder on blue‐green paper in the morning and on violet paper in the afternoon prefer blue‐green over violet paper in the morning and the reverse in the afternoon (Koltermann, 1971). The bee’s sense of time shown in such experiments is tied strongly to its internal circadian clock. A single visit to a scented feeder on day 1 is enough to bring the bee back to the same smelling feeder at the same time on the next day. Attempts to train bees to time intervals other than 24 hr were not successful (Gallistel, 1990). It seems that bees become disposed to find a particular resource at a particular time of day in a particular place. Bees will even use cues from time of day to select the appropriate branch of a route at a crossing point some way from the hive (Fig. 22). Jander (2005) trained bees to feed from vanilla scented sucrose solution at a central site 100 m from their hive. Once bees visited the central site regularly, the sucrose solution was removed and replaced by vanilla scented water. The scented sucrose was placed instead at an auxiliary site 10 m away. When bees flew straight to the auxiliary site, without visiting the central site, the sucrose was removed from the auxiliary site and returned to the center. Repetition of this training cycle ensured that on each trip bees visited first the central site and then the auxiliary site. In the morning the auxiliary site was to the west of the central site; around noon it was moved to the south, and in late afternoon to the east. After 2‐week training, the vanishing

Fig. 22. Time‐linked directional memories in honeybees. (A) Plan view of experimental arrangement. Bees fly 100 m south to a central table and from there to an auxiliary feeder that is placed in a different position at different times of day; from 0800 to 1000 hr in the morning site, from 1130 to 1330 hr in the noon site, and from 1500 to 1700 hr in the afternoon site. (B) Histograms show the directions in which bees flew from the central table in tests given at the start of each time period with the auxiliary feeder removed. Bees have learnt to link flight direction with time of day (adapted from Jander, 2005).

NAVIGATIONAL MEMORIES IN ANTS AND BEES

163

bearings of bees from the central site were measured with the auxiliary site absent. Measurements were taken to coincide with the start of training to the auxiliary site in the morning, noon, or afternoon sessions. Bees flew west in the morning, south at noon, and east in the late afternoon, showing that they had associated three directions from the central site with three times of day. Further evidence that vector memories are linked to time of day comes from observations by Lindauer (1960) on bees performing ‘‘marathon’’ waggle dances. Bees sometimes dance for long periods at night advertising sites that they have visited the previous day. If such marathon dancers have been trained to one site in the morning and another in the late afternoon, their dances switch during the course of the night from advertising the afternoon site to advertising the morning site. C. OLFACTORY CUES AND ROUTE MEMORIES Another way that bees are turned on to currently profitable flowers is by encountering their scent in the hive. von Frisch (1967) made the remarkable discovery that bees foraging together on the same flowers rest close together in the hive. When a forager of that group returns, bearing the relevant floral scent on its body, its resting companions, inferring that the site is profitable once more, are encouraged to resume foraging, even when the returning bee does not dance. Several studies have shown that scent blown into a hive will trigger bees to return to a familiar feeder that has the same scent (Free, 1969; Jakobsen et al., 1995; Johnson, 1967; Ribbands, 1954; von Frisch, 1967). It is supposed that the scent provokes the performance of a stored vector that takes the bee from the hive to the feeding site. Reinhard et al. (2004) have confirmed that navigational memories are linked to odors through the more powerful method of making bees choose between two sets of routes, each of which is evoked by a different scent. Bees were trained individually over 2–3 days to visit two differently scented feeders, each 50 m from the hive and 30 m apart. In occasional tests, the scented feeders were replaced by unscented ones, and one or other of the two familiar odors was blown into the hive. Trained bees then visited predominantly the now odorless feeder that normally carried the scent, which the bees had just experienced in the hive. Bees thus choose between two routes according to a priori evidence (the scent), indicating that a particular route is likely to be profitable. Still missing is an experiment to test whether different odors trigger different vectors independently of familiar route landmarks. Reinhard et al. (2004) also trained bees to associate two colors with different scents. Individual bees learnt to visit two feeders each with a

164

THOMAS S. COLLETT ET AL.

different color and associated scent. Positional cues were eliminated by moving the two feeders from trial to trial. In tests, one or other of the scents was blown into the hive. Bees then searched preferentially at the colored but now unscented feeder that they had previously associated with the odor blown into the hive. D. PRIMING

BY

REWARD?

The honeybee waggle dance encodes the profitability as well as the location of a food site (Seeley, 1995). Bees signal profitability by the number of waggle runs that they produce, showing that they attribute a measure of value to a PI‐defined location. Bees can also learn the reward value of multiple feeder locations, if those locations are defined by landmarks. Greggers and Mauelshagen (1997) trained bees to visit four feeders, which were placed within an array of landmarks and which delivered sucrose at different flow rates. The faster a feeder was replenished, the more frequently it was visited. But, if bees were trained with the same arrangement of feeders and flow rates, with no landmarks present, they never matched the frequency of their visits to the reward level of the site. With landmarks present, the reward level can be attached to the landscape memory viewed from the feeder so that the bee is attracted preferentially to sites offering high rewards. As already mentioned, there is some evidence for associative links between the sequence of visual memories that underpin a route (Zhang et al., 1999). It remains an open question whether the priming of a memory by reward experienced at a goal will propagate through a group of memories so that the whole route acquires the value experienced at the goal. Equally intriguing is whether route selection is sensitive to short‐term changes in value. Does a sudden drop in reward quality at the end of a route lower the insect’s likelihood of performing that route in favor of another? Does an unexpected increase in reward lead to stronger priming of the memories belonging to that route? E. VERSATILE DECISION‐MAKING The flexibility of an insect’s choice of route resides in the values of a variety of internal and external parameters that set which memories an insect retrieves and which route it takes. Gallistel (1990) makes the important distinction between memories that are strictly state dependent, for instance ones that are only retrievable at particular phases of an insect’s circadian rhythm and memories that are primed at a particular phase, but are still accessible at other times. Wahl (1932), reviewed in Gallistel (1990),

NAVIGATIONAL MEMORIES IN ANTS AND BEES

165

shows that the honeybees’ learnt time‐linked behavior is of the latter less restrictive form. Foragers were familiarized over several days to two feeding sites (Fig. 23). Site A offered sucrose from 0900 to 1030 hr and site B from 1600 to 1730 hr. After a few days, bees restricted their foraging visits to the appropriate site from about 1 to 2 hr before each site was open to its closing time. When both sites were unexpectedly closed throughout the day, bees would visit B as well as A in the morning, and they visited A and B in the late afternoon. But they refrained from visiting either site in the middle of the day, when there was normally no sucrose to be had. This behavior suggests that memories for both food routes are primed more during the morning and afternoon slots than at other times, but that A is primed more than B in the morning and vice versa in the afternoon. It seems that a single priming factor (e.g., time of day) can act simultaneously at different points within a control system, influencing both what to do (forage or not) and where to do it (take route A or B). And, further, route memories can be primed simultaneously by separate factors (e.g., feeding state and time of day). A related experiment by Menzel et al. (2001) gives analogous results. Bees were trained in the morning to one site, where they approached a blue target for food, and to a different site in the afternoon, where they approached a yellow target. When bees visited the morning site in the afternoon or the afternoon site in the morning, they chose the color appropriate to the location rather than the time of day. The flexible decision‐making seen in these experiments comes from the permissive and multilevel way in which priming cues seem to operate. Motivational states may be more restrictive in the memories that they allow to be retrieved. Thus, the ant, M. bagoti, did not recognize landmarks on its food‐bound route when it was motivated to go home (Wehner et al., 2006). This behavior is less rigid than it might appear. Whereas an insect cannot change the time of day, it can and does change its motivational state. Dyer (1991) finds that bees displaced unexpectedly to a familiar location on leaving the hive often switch from a foodward to a homeward route and Menzel et al. (2005) give examples of a possible switch in the reverse direction. Flexible route choice can arise through a variety of mechanisms and undoubtedly there are more to find.

VI. SUMMARY Ants and bees have an impressively rich store of navigational memories that are linked together in complex ways. We have stressed that to accompany their large store of memories, insects need adequate mechanisms that

166

THOMAS S. COLLETT ET AL.

Fig. 23. Visits of trained honeybees to two feeding sites during a test day when no sucrose was available. In training, site A gave sucrose from 0900 to 1030 hr and site B from 1600 to 1730 hr. On this test day, bees visited both sites, but landed more at the temporally appropriate site. During training almost all the bees’ visits were to site A in the morning and to B in the afternoon (data from Wahl, 1932).

NAVIGATIONAL MEMORIES IN ANTS AND BEES

167

allow the appropriate memories to be retrieved reliably. One way that insects, like other animals, seem to reduce possible retrieval errors is by linking together the different parts of a memory and by associating a memory with its own spatial and temporal context. That way memories are primed by the correct combination of cues; a flexible many‐pronged key to unlock the appropriate memory that only works if most of the prongs find their proper slot. These learnt cues are encountered as an insect travels along a route through familiar terrain. Route following gives a significant sequential component to memory retrieval such that a cue experienced at one moment can contribute to priming a memory a short while later. A second part to memory retrieval is choosing memories appropriate to a particular route. This process is somewhat like picking out the right collection of keys. It is uncertain whether memories of a route are handily bunched together on their own ring and can be addressed as a group, or whether memories are fingered individually. That insects prime the memories that belong to a particular route, while ignoring those that are irrelevant, is one more reason for supposing that visual, navigational memories serve fixed routes (Collett et al., 2003; Dyer, 1991; Wehner and Menzel, 1990; Wehner et al., 1996, 2006) rather than being organized in a graph‐like structure that could mediate the planning of novel routes (Gould, 1986; Menzel et al., 2005).

REFERENCES ˚ kesson, S., and Wehner, R. (1997). Visual snapshot memory of desert ants, Cataglyphis fortis. A In ‘‘Proceedings of the 25th Go¨ ttingen Neurobiology Conference’’ (N. Elsner and H. Wa˜ ssle, Eds.), p. 482. Georg Thieme Verlag, Stuttgart. ˚ kesson, S., and Wehner, R. (2002). Visual navigation in desert ants Cataglyphis fortis: Are A snapshots coupled to a celestial system of reference? J. Exp. Biol. 205, 1971–1978. Beugnon, G., Chagne´, P., and Dejean, A. (2001). Colony structure and foraging behaviour in the tropical formicine ant Gigantiops destructor. Insectes Sociaux 48, 347–351. Beugnon, G., Lachaud, J.‐P., and Chagne´, P. (2005). Use of long‐term stored vector information in the neotropical ant Gigantiops destructor. J. Insect Behav. 18, 415–432. Bisch‐Knaden, S., and Wehner, R. (2003). Local vectors in desert ants: Context‐dependent landmark learning during outbound and homebound runs. J. Comp. Physiol. A 189, 181–187. Cartwright, B. A., and Collett, T. S. (1983). Landmark learning in bees: Experiments and models. J. Comp. Physiol. A 151, 521–543. Cheng, K. (2005). Context cues eliminate retroactive interference effects in honeybees Apis mellifera. J. Exp. Biol. 208, 1019–1024. Cheng, K., Collett, T. S., Pickhard, A., and Wehner, R. (1987). The use of visual landmarks by honeybees: Bees weight landmarks according to their distance from the goal. J. Comp. Physiol. A 161, 469–475.

168

THOMAS S. COLLETT ET AL.

Chittka, L., Kunze, J., and Geiger, K. (1995a). The influences of landmarks on distance estimation of honeybees. Anim. Behav. 50, 23–31. Chittka, L., Kunze, J., Shipman, C., and Buchman, S. L. (1995b). The significance of landmarks for path integration of homing honey bee foragers. Naturwissenschaften 82, 341–343. Collett, M., and Collett, T. S. (2000). How do insects use path integration for their navigation? Biol. Cybern. 83, 245–259. Collett, M., Collett, T. S., Bisch, S., and Wehner, R. (1998). Local and global vectors in desert ant navigation. Nature 394, 269–272. Collett, M., Collett, T. S., and Wehner, R. (1999). Calibration of vector navigation in desert ants. Curr. Biol. 9, 1031–1034. Collett, M., Harland, D., and Collett, T. S. (2002). The use of landmarks and panoramic context in the performance of local vectors by navigating honeybees. J. Exp. Biol. 205, 807–814. Collett, M., Collett, T. S., Chameron, S., and Wehner, R. (2003). Do familiar landmarks reset the global path integration system of desert ants? J. Exp. Biol. 206, 877–882. Collett, T. S. (1995). Making learning easy: The acquisition of visual information during the orientation flights of social wasps. J. Comp. Physiol. A 177, 737–747. Collett, T. S., and Baron, J. (1994). Biological compasses and the coordinate frame of landmark memories in honeybees. Nature 368, 137–140. Collett, T. S., and Baron, J. (1995). Learnt sensorimotor mappings in honeybees: Interpolation and Its possible relevance to navigation. J. Comp. Physiol. A 177, 287–298. Collett, T. S., and Kelber, A. (1988). The retrieval of visuo‐spatial memories by Honeybees. J. Comp. Physiol. A 163, 145–150. Collett, T. S., and Land, M. F. (1975). Visual spatial memory in a hoverfly. J. Comp. Physiol. A 100, 59–84. Collett, T. S., and Lehrer, M. (1993). Looking and learning: A spatial pattern in the orientation flight of the wasp Vespula vulgaris. Proc. R. Soc. Lond. B. Biol. Sci. 252, 129–134. Collett, T. S., Dillmann, E., Giger, A., and Wehner, R. (1992). Visual landmarks and route following in desert ants. J. Comp. Physiol. A 170, 435–442. Collett, T. S., Fry, S. N., and Wehner, R. (1993). Sequence learning by honeybees. J. Comp. Physiol. A 172, 145–150. Collett, T. S., Baron, J., and Sellen, K. (1996). On the encoding of movement vectors by honeybees. Are distance and direction represented independently? J. Comp. Physiol. A 179, 395–406. Collett, T. S., Fauria, K., Dale, K., and Baron, J. (1997). Places and patterns: A study of context learning in honeybees. J. Comp. Physiol. A 181, 243–353. Collett, T. S., Collett, M., and Wehner, R. (2001). The guidance of desert ants by extended landmarks. J. Exp. Biol. 204, 1635–1639. Davis, R. L. (2005). Olfactory memory formation in Drosophila: From molecular to systems neuroscience. Annu. Rev. Neurosci. 28, 275–302. Deisig, N., Lachnit, H., Giurfa, M., and Hellstern, F. (2001). Configural olfactory learning in honeybees: Negative and positive patterning discrimination. Learn. Mem. 8, 70–78. Deisig, N., Lachnit, H., Sandoz, J. C., Lober, K., and Giurfa, M. (2003). A modified version of the unique cue theory accounts for olfactory compound processing in honeybees. Learn. Mem. 10, 199–208. Deneubourg, J. L., Goss, S., Pasteels, J. M., Fresneau, D., and Lachaud, J.‐P. (1987). Self‐organization mechanisms in ant societies (II): Learning in foraging and division of labour. In ‘‘From Individual to Collective Behaviour in Social Insects, Experientia Supplementum’’ (J. M. Pasteels and J. L. Deneubourg, Eds.), Vol. 54, pp. 177–196. Birkhauser‐Verlag, Basel.

NAVIGATIONAL MEMORIES IN ANTS AND BEES

169

Dickinson, J. A. (1994). Bees link local landmarks with celestial compass cues. Naturwissenschaften 81, 465–467. Durier, V., Graham, P., and Collett, T. S. (2003). Snapshot memories and landmark guidance in wood ants. Curr. Biol. 13, 1614–1618. Dyer, F. C. (1991). Bees acquire route‐based memories but not cognitive maps in a familiar landscape. Anim. Behav. 41, 239–246. Dyer, F. C., Gill, M., and Sharbowski, J. (2002). Motivation and vector navigation in honey bees. Naturwissenschaften 89, 262–264. Ernst, R., and Heisenberg, M. (1999). The memory template in Drosophila pattern vision at the flight simulator. Vision Res. 39, 3920–3933. Fauria, K., Colborn, M., and Collett, T. S. (2000). The binding of visual patterns in bumblebees. Curr. Biol. 10, 935–938. Free, J. B. (1969). Influence of the odour of a honeybee colony’s food stores on the behaviour of its foragers. Nature 222, 778. Frier, H. J., Edwards, E., Smith, C., Neale, S., and Collett, T. S. (1996). Magnetic compass cues and visual pattern learning in honeybees. J. Exp. Biol. 199, 1353–1361. Gallistel, C. R. (1990). ‘‘The Organization of Learning.’’ MIT Press, Cambridge, MA. Gerber, B., Tanimoto, H., and Heisenberg, M. (2004). An engram found? Evaluating the evidence from fruit flies. Curr. Opin. Neurobiol. 14, 737–744. Giurfa, M., Zhang, S. W., Jenett, A., Menzel, R., and Srinivasan, M. V. (2001). The concepts of ‘sameness’ and ‘difference’ in an insect. Nature 410, 930–933. Giurfa, M., Schubert, M., Reisenman, C., Gerber, B., and Lachnit, H. (2003). The effect of cumulative experience on the use of elemental and configural visual discrimination strategies in honeybees. Behav. Brain Res. 145, 161–169. Gould, J. L. (1986). The locale map of honey bees: Do insects have cognitive maps? Science 232, 861–863. Graham, P., and Collett, T. S. (2002). View‐based navigation in insects: How wood ants (Formica rufa L.) look at and are guided by extended landmarks. J. Exp. Biol. 205, 2499–2509. Graham, P., Fauria, K., and Collett, T. S. (2003). The influence of beacon‐aiming on the routes of wood ants. J. Exp. Biol. 206, 535–541. Graham, P., Durier, V., and Collett, T. S. (2004). The binding and recall of snapshot memories in wood ants (Formica rufa L.). J. Exp. Biol. 207, 393–398. Greggers, U., and Mauelshagen, J. (1997). Matching behaviour of honeybees in a multiple choice situation: The differential effect of environmental stimuli on the choice process. Anim. Learn. Behav. 25, 458–477. Hammer, M. (1997). The neural basis of associative reward learning in honeybees. Trends Neurosci. 20, 245–252. Harris, R. A., Hempel de Ibarra, N., Graham, P. R., and Collett, T. S. (2005). Priming of visual route memories. Nature 438, 302. Harrison, J. M., and Breed, M. D. (1987). Temporal learning in the giant tropical ant, Paraponera clavata. Physiol. Entomol. 12, 317–320. Hempel de Ibarra, N., Giurfa, M., and Vorobyev, M. (2002). Discrimination of coloured patterns by honeybees through chromatic and achromatic cues. J. Comp. Physiol. A 188, 503–512. Ho¨lldobler, B., and Wilson, E. O. (1990). ‘‘The Ants.’’ Belknap Press of Harvard University Press, Cambridge, MA. Horridge, G. A. (1998). Spatial coincidence of cues in visual learning by the honeybee (Apis mellifera). J. Insect Physiol. 44, 343–350.

170

THOMAS S. COLLETT ET AL.

Jakobsen, H. B., Kristjanson, K., Rohde, B., Terkildsen, M., and Olsen, C. (1995). Can social bees be influenced to choose a specific feeding station by adding the scent of the station to the hive air? J. Chem. Ecol. 21, 1635–1648. Jander, R. (2005). Mapping of scattered foraging sites in the honeybee’s navigation memory. In ‘‘Royal Institute of Navigation 5th Quadrennal Conference,’’ University of Reading. Johnson, D. L. (1967). Communication among honey bees with field experience. Anim. Behav. 15, 487–492. Judd, P. D., and Collett, T. S. (1998). Multiple stored views and landmark guidance in ants. Nature 39, 710–714. Junger, W. (1991). Waterstriders (Gerris paludum F.) compensate for drift with a discontinuously working visual position servo. J. Comp. Physiol. A 169, 633–639. Knaden, M., and Wehner, R. (2005). Nest mark orientation in desert ants Cataglyphis: What does it do to the path integrator? Anim. Behav. 70, 1349–1354. Kohler, M., and Wehner, R. (2005). Idiosyncratic route‐based memories in desert ants, Melophorus bagoti: How do they interact with path‐integration vectors? Neurobiol. Learn. Mem. 83, 1–12. Koltermann, R. (1971). 24‐Std‐Periodik in der Langzeiterinnerung an Duft‐ und Farbsignale bei der Honigbiene. Z. vergl. Physiol. 75, 49–68. Kwon, H. W., Lent, D. D., and Strausfeld, N. J. (2004). Spatial learning in the restrained American cockroach Periplaneta americana. J. Exp. Biol. 207, 377–383. Lambrinos, D., Moller, R., Labhart, T., Pfeifer, R., and Wehner, R. (2000). A mobile robot employing insect strategies for navigation. Rob. Auton. Syst. 30, 39–64. Lehrer, M. (1999). Dorsoventral asymmetry of colour discrimination in bees. J. Comp. Physiol. A 184, 195–206. Lindauer, M. (1960). Time‐compensated sun orientation in bees. Cold Spring Harb. Symp. Quant. Biol. 25, 371–377. Liu, G., Seiler, H., Wen, A., Zars, T., Ito, K., Wolf, R., Heisenberg, M., and Liu, L. (2006). Distinct memory traces for two visual features in the Drosophila brain. Nature 439, 551–556. Menzel, R. (1969). Das geda¨chtnis der honigbiene fu¨r spektralfarben. II. Umlernen und mehrfachlernen [The honey bee’s memory of spectral colours. II. Reversal learning and learning of several colours]. Z. vergl. Physiol. 63, 290–309. Menzel, R., and Lieke, E. E. (1983). Antagonistic colour effects in spatial vision of honeybees. J. Comp. Physiol. A 151, 441–448. Menzel, R., Geiger, K., Chittka, L., Joerges, J., Kunze, J., and Mu¨ller, U. (1996). The knowledge base of bee navigation. J. Exp. Biol. 199, 141–146. Menzel, R., Geiger, K., Joerges, J., Mu¨ller, U., and Chittka, L. (1998). Bees travel novel homeward routes by integrating separately acquired vector memories. Anim. Behav. 55, 139–152. Menzel, R., Giurfa, M., Gerber, B., and Hellstern, F. (2001). Cognition in insects: The honeybee as a study case. In ‘‘Brain Evolution and Cognition’’ (G. Roth and M. Wullimann, Eds.), pp. 333–366. John Wiley & Sons, New York. Menzel, R., Greggers, U., Smith, A., Berger, S., Brandt, R., Brunke, S., Bundrock, G., Hulse, S., Plumpe, T., Schaupp, F., Schuttler, E., Stach, S., et al. (2005). Honey bees navigate according to a map‐like spatial memory. Proc. Natl. Acad. Sci. USA 102, 3040–3045. Nicholson, D. J., Judd, S. P. D., Cartwright, B. A., and Collett, T. S. (1999). Learning walks and landmark guidance in wood ants (Formica rufa). J. Exp. Biol. 202, 1831–1838. Pratt, S., Brooks, S. E., and Franks, N. F. (2001). The use of edges in visual navigation by the ant Leptothorax albipennis. Ethology 107, 1125–1136. Reinhard, J., Srinivasan, M. V., Guez, D., and Zhang, S. W. (2004). Floral scents induce recall of navigational and visual memories in honeybees. J. Exp. Biol. 207, 4371–4381.

NAVIGATIONAL MEMORIES IN ANTS AND BEES

171

Ribbands, C. R. (1954). Communication between honeybees, I: The response of crop‐attached bees to the scent of their crop. Proc. R. Entomol. Soc. Lond. 29, 141–144. Ribbands, C. R. (1964). ‘‘The Behaviour and Social Life of Honeybees.’’ Dover Publications, New York. Riley, J. R., Greggers, U., Smith, A. D., Stach, S., Reynolds, D. R., Stollhoff, N., Brandt, R., Schaupp, F., and Menzel, R. (2003). The automatic pilot of honeybees. Proc. R. Soc. Lond. B. Biol. Sci. 270, 2421–2424. Riley, J. R., Greggers, U., Smith, A. D., Reynolds, D. R., and Menzel, R. (2005). The flight paths of honeybees recruited by the waggle dance. Nature 435, 205–207. Santschi, F. (1913). Comment s’orientent les fourmis. Rev. Suisse. Zool. 21, 347–425. Sassi, S., and Wehner, R. (1997). Dead reckoning in desert ants, Cataglyphis fortis: Can homeward‐bound vectors be reactivated by familiar landmark configuration? In ‘‘Proceedings of the 25th Go¨ttingen Neurobiology Conference’’ (N. Elsner and H. Wa˜ssle, Eds.), p. 484. Georg Thieme Verlag, Stuttgart. Schatz, B., Lachaud, J.‐P., and Beugnon, G. (1999). Spatio‐temporal learning by the ant Ectomma ruidum. J. Exp. Biol. 202, 1897–1907. Schmid‐Hempel, P. (1984). Individually different foraging methods in the desert ant Cataglyphis bicolor (Hymenoptera, Formicidae). Behav. Ecol. Sociobiol. 14, 263–271. Schubert, M., Lachnit, H., Francucci, S., and Giurfa, M. (2002). Nonelemental visual learning in honeybees. Anim. Behav. 64, 175–184. Seeley, T. D. (1995). ‘‘The Wisdom of the Hive.’’ Harvard University Press, London. Sommer, S., and Wehner, R. (2004). The ant’s estimation of distance travelled: Experiments with desert ants, Cataglyphis fortis. J. Comp. Physiol. A 190, 1–6. Srinivasan, M. V., Zhang, S. W., and Witney, K. (1994). Visual discrimination of pattern orientation by honeybees: Performance and implications for cortical processing. Philos. Trans. R. Soc. Lond. B Biol. Sci. 343, 199–210. Srinivasan, M. V., Zhang, S. W., and Bidwell, N. J. (1997). Visually mediated odometry in honeybees en route to the goal: Visual flight control and odometry. J. Exp. Biol. 200, 2513–2522. Srinivasan, M. V., Zhang, S. W., and Zhu, H. (1998). Honeybees link sight to smells. Nature 396, 637–638. Stach, S., Benard, J., and Giurfa, M. (2004). Local‐feature assembling in visual pattern recognition and generalization in honeybees. Nature 429, 758–761. Tang, S., Wolf, R., Xu, S., and Heisenberg, M. (2004). Visual pattern recognition in Drosophila is invariant for retinal position. Science 305, 1020–1022. Vardy, A., and Mo¨ller, R. (2005). Biologically plausible visual homing methods based on optical flow techniques. Connection Science 17, 47–89. von Frisch, K. (1967). ‘‘The Dance Language and Orientation of Bees.’’ Oxford University Press, London. Wahl, O. (1932). Neue untersuchungen u¨ber das zietgeda¨chtnis der bienen [New experiments concerning the time‐sense of bees]. Z. vergl. Physiol. 16, 529–589. Wehner, R. (1987). Spatial organization of foraging behaviour in individually searching desert ants, Cataglyphis (Sahara desert) and Ocymyrmex (Namib desert). In ‘‘From Individual to Collective Behaviour in Social Insects, Experientia Supplementum’’ (J. M. Pasteels and J. L. Deneubourg, Eds.), Vol. 54, pp. 15–41. Birkhauser‐Verlag, Basel. Wehner, R. (1992). Arthropods. In ‘‘Animal Homing’’ (F. Papi, Ed.), pp. 44–144. Chapman and Hall, London. Wehner, R. (2003). Desert ant navigation: How miniature brains solve complex tasks. J. Comp. Physiol. A 189, 579–588.

172

THOMAS S. COLLETT ET AL.

Wehner, R., and Menzel, R. (1990). Do insects have cognitive maps? Annu. Rev. Neurosci. 13, 403–414. Wehner, R., and Ra¨ber, F. (1979). Visual spatial memory in desert ants, Cataglyphis bicolor. Experientia 35, 1569–1571. Wehner, R., and Srinivasan, M. V. (2003). Path integration in insects. In ‘‘The Neurobiology of Spatial Behaviour’’ (K. J. Jeffery, Ed.). Oxford University Press, Oxford. Wehner, R., Harkness, R. D., and Schmid‐Hempel, P. (1983). Foraging strategies in individually searching ants, Cataglyphis bicolor (Hymenoptera, Formicidae). Akad. Wiss. Lit. Mainz. Math. Naturwiss Kl. Fischer, Stuttgart. Wehner, R., Michel, B., and Antonsen, P. (1996). Visual navigation in insects: Coupling of egocentric and geocentric information. J. Exp. Biol. 199, 129–140. Wehner, R., Meier, C., and Zollikofer, C. (2004). The ontogeny of foraging behaviour in desert ants, Cataglyphis bicolor. Ecol. Entomol. 29, 240–250. Wehner, R., Boyer, M., Loertscher, F., Sommer, S., and Menzi, U. (2006). Ant navigation: One‐way routes rather than maps. Curr. Biol. 16, 75–79. Wolf, H., and Wehner, R. (2000). Pinpointing food sources: Olfactory and anemotactic orientation in desert ants, Cataglyphis fortis. J. Exp. Biol. 203, 857–868. Yates, F. A. (1966). ‘‘The Art of Memory.’’ University of Chicago, Chicago. Zeil, J. (1993a). Orientation flights of solitary wasps (Cerceris, Sphecidae, Hymenoptera). 1. Description of flight. J. Comp. Physiol. A 172, 189–205. Zeil, J. (1993b). Orientation flights of solitary wasps (Cerceris, Sphecidae, Hymenoptera). 2. Similarities between orientation and return flights and the use of motion parallax. J. Comp. Physiol. A 172, 207–222. Zeil, J., Hofmann, M. I., and Chahl, J. S. (2003). The catchment areas of panoramic snapshots in outdoor scenes. J. Opt. Soc. Am. A 20, 450–469. Zhang, S. W., Lehrer, M., and Srinivasan, M. V. (1999). Honeybee memory: Navigation by associative grouping and recall of visual stimuli. Neurobiol. Learn. Mem. 72, 180–201. Zhang, S. W., Bock, F., Si, A., Tautz, J., and Srinivasan, M. V. (2005). Visual working memory in decision making by honey bees. Proc. Natl. Acad. Sci. USA 102, 5250–5255.

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 36

Functional Genomics Requires Ecology Lara S. Carroll* and Wayne K. Potts{ *howard hughes medical institute, university of utah utah 84112, usa { department of biology, university of utah, utah 84112, usa

The problems faced by pre‐ and post‐genomic genetics are therefore much the same—they all involve bridging the chasm between genotype and phenotype. Sydney Brenner (Nobel Laureate, 2002). The End of the Beginning, Science 287, 2173

I. THE PROBLEM: MANY GENES SEEM

TO

BE UNNECESSARY

Since Mendel’s time, most genes have been identified by the effects mutations (including knockouts) have on the morphology, physiology, or behavior of individuals. Thus, almost by definition, there could be no ‘‘mutant genes without phenotypes.’’ In the molecular era, however, it has become possible to identify genes from DNA sequences. This alternative path to gene discovery has already led to the completion of several genome projects yielding ‘‘complete parts lists’’ for representatives of several major branches in the tree of life. This achievement is drawing widespread attention to a paradox that has troubled biologists for more than a decade: many genes lack obvious phenotypes. For example, fewer than half of the estimated 14,000 genes revealed by the recently completed genomic sequence of Drosophila melanogaster had been previously identified by ‘‘forward’’ (phenotype based) genetics, despite the fact that all genes had been hit multiple times during mutant screens (Rubin and Lewis, 2000). More importantly, a substantial proportion of engineered knockout and knockdown mutations of well‐conserved genes in yeast (Saccharomyces cerevisiae) (Giaever et al., 2002; Thatcher et al., 1998), Caenorhabditis elegans (Kamath et al., 2003; Maeda et al., 2001), and mice (Mus musculus) (Shastry, 1995) produce no discernable phenotypic effects. This problem is even more troubling in light of the discovery that 0065-3454/06 $35.00 DOI: 10.1016/S0065-3454(06)36004-4

173

Copyright 2006, Elsevier Inc. All rights reserved.

174

LARA S. CARROLL AND WAYNE K. POTTS

vertebrate genomes contain many fewer genes than expected (Brookfield, 1997; Lander et al., 2001). These nonessential genes must have functions or they would wither away like all pseudogenes under the continual rain of deleterious mutations. Yet the phenotypic ‘‘invisibility’’ of these genes frustrates efforts to learn what they do, because function cannot be studied effectively (or at all) without phenotypes. Whether a gene encodes an indispensable structural protein or a protein cofactor embedded in a complex biosynthetic pathway, each gene directly or indirectly influences the ultimate coordinated assembly of cells, tissues, organs, and their precise functioning. Any protein in any pathway, no matter how minor its role, is optimized to facilitate some molecular event, whether it is precise timing, expression level, or tissue specificity of other genes in the complex, the binding specificity of the complex, the function of the complex as a whole, or the ultimate regulation of downstream genes and events. The absence of a knockout phenotype, rather than providing evidence for a developmental safety net, is more likely indicating that one or more metabolic pathways are operating at reduced efficiency. The mutant, suspiciously intact in its petri dish or laboratory cage, will actually be compromised in some quantitative way, however small. Function is fitness is function! The only function of each and every gene that ultimately matters during its evolutionary history is how it contributes to fitness (lifetime reproductive success; Darwin, 1859). Consequently, characterization of gene function will always be incomplete without fitness measurements in settings that simulate the rigorous test environment of nature. Fortunately, fitness‐based assays can be exploited to reveal organismal function of mutants. These assays attempt to measure critical components of fitness in the context of important ecological conditions. Almost every biological character is potentially a component of fitness, but critical components of fitness are typically major integrative characters such as mating success, reproductive success, weaning success, survival, and social dominance. The measured components of fitness most dramatically affected will guide attempts to identify more specific effects. For example, if the effect occurs only when comparing across multiple generations then one might look for ‘‘transgenerational characters’’ such as parenting and other factors influencing the developmental environment. The important ecological conditions will differ dramatically among species and will also depend on the functions of the mutant that require testing. For many species, fitness assays will require studies conducted in nature (Endler, 1986). But for other species, critical ecological conditions can be simulated under lab settings, for example, Dictyostelium (Queller et al., 2003), Drosophila (Shabalina et al., 1997), plants (Hayes et al., 2005), and house mice (Meagher et al., 2000). Any lab simulation studies must be cautious

FITNESS AND GENE FUNCTION

175

in their conclusions, because such studies will always miss many factors present in nature. Such assays establish the relative importance of the mutant gene for fitness, but more importantly, for the many genes with missing (cryptic) functions, fitness assays will be the most sensitive test for detecting phenotypic change, allowing molecular characterization to proceed. A. HOW BIG IS

THE

PROBLEM?

1. How Common Are No‐Phenotype Gene Knockouts? The development of molecular technologies, such as transgenes, gene targeting, gene jumping transposons, RNA‐mediated interference (RNAi), and other antisense nucleic acid approaches to name a few, have granted biologists the ability to disrupt their favorite gene in a variety of prokaryotic and eukaryotic organisms. Subsequent comparative analysis of mutant and wild‐type phenotypes often reveals the disrupted gene’s function. Efforts are underway to provide mutants for any desired gene in C. elegans (http://www.celeganskoconsortium.omrf.org/), Drosophila melanogaster (www.openbiosystems.com; http://flyrnai.org), and every known gene in yeast (http://www.sequence.stanford.edu/group/yeast_deletion_ project/deletions3.html) and mice (Austin et al., 2004) (http://www.jax.org/ imr/index.html). Knockout studies in each of these model organisms have consistently yielded the surprising result that many genes are nonessential. About 30% of random yeast knockouts show no phenotypic change from wild type (Thatcher et al., 1998), and about 10% of mouse knockouts including a similar proportion of conserved developmental Hox gene knockouts show no phenotype (Duboule, 2000). This mouse figure will be an underestimate because the most important genes have been knocked out first and studies failing to detect a phenotype go unpublished or are slower to be published. 2. How Common Are No‐Phenotype Gene Mutants? Not surprisingly, the frequency of no‐phenotype mutants generated by random mutagenesis is much higher than for knockouts of specific genes. Chemical mutagens, such as ethylmethane sulfonate (EMS) in worms and flies, or N‐ethyl‐N‐nitrosourea (ENU) in mice can be used to induce random point mutations. These techniques yield functional knockouts, as well as hypomorphic, hypermorphic, or neomorphic alleles that differ in overall expression levels or in the spatial and temporal domains of expression. One problem with chemical mutagenesis in diploid organisms is that unless mutant alleles are dominant or semidominant, the screening process requires two generations of reproduction to generate homozygotes capable

176

LARA S. CARROLL AND WAYNE K. POTTS

of expressing phenotypes. Hence the absence of a phenotype among second generation offspring might merely be due to the chance lack of homozygous mutants. This will artificially boost the number of mutations scored as lacking a phenotype. As might be expected, only 4% of EMS‐induced amino acid substitutions in C. elegans lowered lifetime fecundity in the lab, yet we know that most of these substitutions are under purifying selection based on the low ratios of nonsynonymous to synonymous mutations (Davies et al., 1999). In both C. elegans and Drosophila, a low rate of phenotype discovery is further confounded by the potential rescue of defective mutants by functional RNA inherited from their mothers. RNAi is a targeted knockdown method that works well in invertebrates (and for some vertebrate applications). The development of high‐ throughput RNAi is allowing functional analysis of C. elegans genes on a genome‐wide basis (Kamath et al., 2003; Sugimoto, 2004). Ectopic introduction of RNA complementary to the target gene causes degradation of the endogenous transcript. Although there are great advantages to this gene‐specific, high‐throughput approach, RNAi knockdowns suffer the same limitations as knockouts, no conclusions can be drawn in the absence of a phenotype. B. CONVENTIONAL EXPLANATION: FUNCTIONAL REDUNDANCY One possible explanation to the problem of mutants without phenotypes is that other genes are serving a ‘‘backup’’ role—the genome has built‐in functional redundancy. This has become the conventional default assumption, particularly for gene knockouts that are phenotypically indistinguishable from wild type. The prevalence of gene duplication events throughout evolutionary history has provided the basic framework and fuel for the functional redundancy explanation. However, we make the case in Section II.A that functional redundancy, far from being a general phenomenon, is probably limited to a small proportion of genetic mutants for which phenotypes are lacking. We expect that the majority of genes described as nonessential will reveal phenotypes when subjected to the ecological conditions under which they arose. C. ECOLOGICAL EXPLANATION Phenotypes too subtle to be detected in lab assays may nonetheless have large effects on the performance, health, and overall fitness of individuals living under natural conditions. Organisms living under benign laboratory conditions are free from many challenges and stresses that would confront them in the real world. For example, a 5% reduction of metabolic efficiency would have no detectable effect on the health, longevity, or reproduction of

FITNESS AND GENE FUNCTION

177

mice in laboratory mouse cages, but it might be effectively lethal under the food‐limited and socially competitive conditions in nature (Carr and Dudash, 1995; Jime´nez et al., 1994; Miller, 1994). What matters for a gene’s evolutionary survival is that the selection coefficient is greater than the reciprocal of the effective population size. If selection exceeds this threshold, a genetic mutant with no detectable phenotype in the laboratory would be eliminated from the wild with almost as much certainty as an embryonic lethal mutation would be. In the sections below, we review theoretical and empirical studies that address our central thesis—that fitness assays are a powerful tool for elucidating gene function, especially for phenotypes that are cryptic using traditional methods. To a large extent this chapter focuses on model organisms: yeast, Drosophila, C. elegans, and mice because the tools to generate genetic mutations and gene disruptions to study specific gene function are already available. However, the principles are general and can be applied to similar enterprises in any species. In Sections II.B and II.C, we provide numerous examples where ecological approaches reveal phenotypes that were undetectable in the laboratory. We also demonstrate that even for mutations with known phenotypes, fitness testing is a useful screen for additional unknown phenotypes and to quantify the relative importance of the mutation in the evolutionary currency of fitness.

II. GENES LACKING PHENOTYPES: EXPLANATIONS AND EXPERIMENTAL APPROACHES FOR THEIR ELUCIDATION Various lines of reasoning have been invoked to explain the surplus of gene knockouts that yield no phenotype in animals homozygous for the mutation. These are: (1) complete functional redundancy, where redundant genes are equally efficient at carrying out a particular function and can provide complete rescue if either counterpart is rendered dysfunctional, (2) partial functional redundancy among genes that are unequal in their efficiency or have additional unique functions, (3) genetic backgrounds that mask defects, (4) genes with small effects, or (5) genes with ecological functions that are not detectable in typical laboratory tests (Cooke et al., 1997). Functional redundancy has become the default assumption when a phenotype is not detected, but laboratory methods are largely ineffective for testing the alternative hypotheses listed above. Thus, the inherent limitations of laboratory analyses to replicate nature have erroneously established functional redundancy as a general explanation for minimal or absent phenotypes. In the following subsections we discuss and evaluate these alternative hypotheses.

178

LARA S. CARROLL AND WAYNE K. POTTS

A. FUNCTIONAL REDUNDANCY: THEORY

AND

EVIDENCE

1. Theory It is generally thought that most cases of functional redundancy will occur among related genes. Although it is theoretically possible for some degree of functional overlap to arise in unrelated genes in integrated pathways, the majority of known genes with suspected redundant partners are themselves the direct descendents of gene duplication events (Conant and Wagner, 2004; Gu et al., 2003). Gene duplication is a pervasive process that has been responsible for many evolutionary innovations and the expansion of phenotypic complexity in general. Once duplicated, newly redundant counterparts are theoretically freed from selection and become more susceptible to mutations that either destroy one paralogue entirely, or modify one or both duplicates within coding or cis‐regulatory regions ultimately changing the extent, timing, or spatial pattern of gene expression. This leads either to subfunctionalization of the ancestral role (Force et al., 1999; Gibert, 2002), with each paralogue evolving toward greater functional specificity, or more rarely, to the origin of evolutionary novelty (Carroll et al., 2004b; Cheng and Chen, 1999). With the availability of genomic databases, the strength of selection following duplication events can be measured indirectly via sequence comparison. Such studies have indeed found that gene duplication events are followed by a brief period of relaxed selection. In contrast, orthologous genes (sharing an ancestral origin in related species) with divergence times comparable to within‐ species duplicates do not experience relaxed selection (Kondrashov et al., 2002; Lynch and Conery, 2000). So the problem is not how functional redundancy arises, it is how functional redundancy can be maintained despite the inevitable disruptive effects of mutation. In some cases, gene conversion may extend the lifetime of redundant paralogues, homogenizing sequences of duplicated genes that are in the process of drifting apart (Gao and Innan, 2004). Even without the assistance of gene conversion or natural selection, recently duplicated genes can maintain functional redundancy for millions of years assuming both genes experience similar mutation rates (Cooke et al., 1997). Although recent gene duplications may explain a minority of putative functional overlap among some paralogues (Lynch and Conery, 2000), it does not explain how functional redundancy can be maintained among very old paralogues from ancient duplication events that characterize much of the genome in yeast (Wolfe and Shields, 1997), fish (Taylor et al., 2001), and other vertebrates (Spring, 1997). The continued maintenance of complete functional redundancy, like we routinely find on human‐engineered systems, such as airliners, is effectively impossible under evolutionary engineering. Whereas airline

FITNESS AND GENE FUNCTION

179

mechanics have mechanisms to test redundant systems independently of each other, natural selection cannot detect a broken backup system because the crippling of one gene paralogue will have no effect on the function of its redundant counterpart. At an average mutation rate (10
8–10
9 mutations per base pair per generation), the functional gene could remain fully operational for thousands of generations without the organism ever missing the ancestral backup system. The best explanations proposed for the maintenance of functional backup genes rely on mechanisms that yield only partial redundancy, with the genes being maintained by selection for reasons other than exclusively backup functions. The case of recently duplicated genes described earlier is the only clear situation for which mutually redundant genes might exist (Cooke et al., 1997). However, Cooke et al. (1997) developed models to show how partial redundancy might be maintained provided genes are asymmetrically efficient at performing their function (and experience unequal mutation rates), or as long as paralogues perform unique roles in addition to the redundant function (Cooke et al., 1997). Finally, genes might appear to be functionally redundant if they are prone to functional failure in the face of frequent developmental and/or environmental perturbations which favors the maintenance of backup genes. This example is rather more closely associated with ecological contingencies, where different genes function under dissimilar ecological conditions to achieve homeostasis. Consequently, both genes are being maintained by selection due to their specialization to effectively solve different (but related) ecological problems. 2. Evidence As we first began discussing these issues with colleagues and students, one confused student innocently asked if a no‐phenotype knockout meant the animal became invisible. Fair enough. We define the term ‘‘no‐phenotype knockout’’ (or mutant) to be a mutant that shows no phenotypic change from wild type. Correspondingly, a ‘‘minimal phenotype mutant’’ would have a phenotype differing mildly from wild type. It is becoming apparent that many cases of no‐phenotype and minimal phenotype knockouts are due to limitations of the testing environment. Many genes are predicted to be essential for development based on their embryonic expression patterns and involvement in important developmental pathways. Yet these same genes lack clear knockout phenotypes or have unexpectedly mild effects. The homeobox family of transcription factors known as Hox genes provides many examples of this. Hox genes are responsible for patterning the metazoan body plan from worms to humans. These genes, clustered in four separate chromosomal linkage groups in tetrapods, are expressed during

180

LARA S. CARROLL AND WAYNE K. POTTS

embryogenesis in spatially and temporally restricted domains. Genes at 30 ends of the Hox clusters turn on first, with anterior limits of expression in the hindbrain of the developing embryo. Additional 50 Hox genes along the clusters are sequentially turned on, with each newly active Hox gene being expressed in progressively more caudal embryonic domains. This particular expression pattern of Hox genes, termed spatial and temporal colinearity, exhibits significant trans‐species conservation. Figure 1 compares Hox genes in Drosophila and Mus and illustrates the conservation of this spatial colinearity between these two distantly related species. Also broadly conserved across taxa are the sequences of peptide residues making up the ‘‘homeodomain.’’ This region of a Hox protein, about 60 amino acids in length, recognizes and binds to DNA, permitting Hox proteins to operate as transcriptional repressors or activators of their downstream targets. These features initially distinguished the Hox genes as strong candidates for having critical functions in development. Consequently, Hox genes were among the first in mice to be systematically knocked out (Capecchi, 1997). However, the majority of the 39 mouse Hox genes are not required for embryonic or postnatal survival. Single homozygous knockouts that affect cervical, thoracic, lumbar, and sacral axial structures tend to have subtle defects, with minor transformations of vertebrae toward the identity of slightly more anterior or posterior structures. Functional redundancy among genes within paralogous groups has been and continues to be the general explanation for such results. It was found in mice that all members of a Hox paralogous group might require disruption before a complete phenotype emerges. For example, all six alleles of the three Hox10 paralogues (HoxA10, HoxC10, and HoxD10) had to be simultaneously disrupted before the role of these genes in suppressing rib development could be identified. The ribs of tetrapods are normally attached to thoracic vertebrae only. Knocking out all six alleles of the Hox10 group produces a mouse with ribs attached to all thoracic, lumbar, and sacral vertebrae. However, even a single functional Hox10 allele from any of the three Hox10 paralogues precludes this phenotype (Wellik and Capecchi, 2003). In another study, Greer et al. (2000) found no phenotype after swapping the HoxA3 and HoxD3 coding sequences. It was speculated that perhaps protein identity is irrelevant and what matters is the overall amount of protein supplied by paralogues within overlapping domains. Far from being a rare exception, mild or absent phenotypes have been repeatedly documented among the thousands of mutant mouse lines produced from gene targeting and mutant screens. This general phenomenon prompted the Journal of Molecular and Cellular Biology in 1999 to begin dedicating a section in each issue to mammalian genetic models with minimal or complex phenotypes.

FITNESS AND GENE FUNCTION

181

Fig. 1. A comparison of Hox (homeobox) genes in Drosophila and Mus showing the conservation of spatial colinearity and regions of the embryos controlled by each orthologous Hox gene.

182

LARA S. CARROLL AND WAYNE K. POTTS

How relevant are such studies as supportive evidence for functional redundancy? When the rescue of a genetic mutation is credited to a paralogous partner, an immediate question arises. Even if the paralogous partner is functionally capable of assuming the role of its crippled counterpart, what is the adaptive significance of this capacity, if any? Perhaps the redundant roles performed by paralogues are never actually required in the context of an intact genome, but remain as evolutionary baggage, performing in suboptimal capacity as useless in nature as the eyes of blind cave fish. To answer these questions, it is first necessary to know how commonly paralogues are coregulated within individual cells. Expression timing and location is a necessary requirement for predicting when genes might function redundantly and/or cooperatively to orchestrate subsequent downstream developmental processes. A recent study supplies a plausible mechanism for apparent redundancy in yeast genes (Kafri et al., 2005). Distantly related paralogues (resulting from ancient duplications) with the highest backup efficiency following mutation are dissimilarly expressed in most growth conditions and are therefore likely to perform distinct roles in the wild‐type cell. Although these efficient backup paralogues show low levels of coexpression in wild‐ type cells, they are capable of conditionally switching their expression profiles in a coordinated manner and show an intermediate degree of similarity among their cis‐regulatory motifs (Kafri et al., 2005). The authors speculated that this partial sharing among regulatory elements enabled transcriptional reprogramming when one paralogue was disrupted. This reprogramming then led to novel expression of the intact gene, ultimately providing backup for its crippled partner (Kafri et al., 2005). Dissimilar expression profiles clearly indicate that yeast paralogues are not functionally equivalent. So if yeast backup circuits are indeed evolutionarily conserved, it is overwhelmingly among genes that are not truly redundant, but serve unique functions in addition to their backup capability. In summary, although there may be cases of selectively maintained functional redundancy, they are likely to be rare. Authors are beginning to acknowledge that the functional redundancy they observe in lab is likely to reflect the limitations of test conditions and does not indicate that paralogues are equally dispensable in long‐term evolution (Gu, 2003). In cases where backup circuits are shown to buffer gene disruptions, the gene partners involved are increasingly accepted as performing unique roles in addition to buffering. It is these unique roles mediated by new regulatory motifs that we expect to be under strong selection, with the conserved functions being maintained possibly via genetic hitchiking for longer than would be expected under selective neutrality. Finally, only a few possible examples of functionally redundant mutations have been subjected to fitness assays to evaluate the alternative hypotheses. However, in cases where they were

FITNESS AND GENE FUNCTION

183

tested, the general result is that fitness assays reveal phenotypes in formerly no‐phenotype mutants/variants (Sections II.B.5 and II.C.1). B. ECOLOGICAL FUNCTIONS The most important alternative hypothesis to functional redundancy is that many genes function during specific ecological conditions, most of which are not included in standard laboratory assays. The list of such ecological conditions is long and certainly incomplete for any species, but usually includes predators, infectious agents, wounding, nutritional deficiencies, sporadic food and water availability, toxins, thermal extremes, and social ecology. Social ecology includes competition among conspecifics, which is often of special importance because it involves aggressive, physical contests that demand maximum performance from most physiological systems. Consequently, these contests represent a particularly sensitive and integrative screen for many molecular and physiological defects. Accordingly, competition among conspecifics is an ecological condition we treat in its own section (Section II.C) due to its special importance. In Section II.C and later, we review nine examples where natural or seminatural population conditions allowed the discovery of phenotypes that were previously missed or likely to be missed in laboratory studies. Most of these nine examples involve social ecology, in part because other ecological conditions have seldom been manipulated in relevant studies. One reason that we have not manipulated many of these ecological variables in our studies on Mus is because the extreme competition among individuals that occurs in mouse society appears to be a sensitive screen for health, vigor, and fitness differences. We have framed the general problem around no‐phenotype knockouts/ mutants. However, only two of the nine examples represent cases of specific knockouts/mutants (Sections II.B.5 and II.C.1) tested under ecologically relevant conditions. We believe that the lack of good examples in the literature underscores the general problem that no‐phenotype knockouts are not typically being tested under ecological or competitive conditions. The other seven examples represent cases of natural variants (Sections II.B.1, II.B.2, and II.B.4; see also Section II.C.4) or groups of uncharacterized mutants (such as inbreeding and mutation accumulation; Sections II.B.3, II.C.2, and II.C.3) where fitness assays revealed previously unknown phenotypes. 1. MHC‐Mediated Sexual Selection Alleles of the major histocompatibility complex (MHC) have several unique features. MHC genes are characterized by tremendous polymorphisms (up to 400 alleles at some loci; Klein, 1986). They show allele sharing

184

LARA S. CARROLL AND WAYNE K. POTTS

among anciently diverged species, and allelic frequencies are uniform across populations. These features all point directly to some form of strong balancing selection maintaining allelic diversity at these immunologically critical loci. It has long been assumed that pathogens and parasites are the dominant form of selection acting on these loci, either by favoring MHC heterozygotes that are capable of resisting a broader array of immunological challenges than homozygotes, and/or by favoring individuals that carry rare MHC alleles. A variety of empirical and theoretical models have predicted both heterozygote and rare allele advantages to be capable of maintaining the degree of polymorphisms found in nature (Bodmer, 1972; Doherty and Zinkernagel, 1975; Hill et al., 1991; Hughes and Nei, 1988). However, studies have only recently begun to yield the predicted results that in the presence of pathogens different MHC genes yield phenotypes that differ in their susceptibility to disease. In the search for pathogen‐ mediated selection, incorporating multiple pathogens or strains of pathogens as is encountered in nature appears to be necessary for detecting MHC effects (Carrington et al., 1999; McClelland et al., 2003; Penn et al., 2002; Thursz et al., 1997; Wegner et al., 2004). In general, it is MHC heterozygotes that are better protected against infection than homozygotes. Potts et al. (1991) set up seminatural populations of house mice designed to determine whether MHC heterozygotes were more fit than homozygous conspecifics. The authors reasoned that the best way to test for differences among genotypes was to allow natural selection to reveal fitness using free‐ living mice maintained for a year in nonsterile seminatural enclosures. The authors reasoned that mice carrying less favored MHC haplotypes would have lower survival and hence lower fitness. In a surprise twist, the authors found that mice were relatively unencumbered by the mild pathogenic conditions of the enclosures. Instead, what prevailed was massive sexual selection. Compared to random mating expectations, MHC genotypes of pups born to the original founders revealed a reduction of homozygotes in all nine populations (Fig. 2) with a mean 27% deficiency overall. In evaluating the potential causes for this deficiency, only MHC‐based disassortative mating preferences could explain the pup genotype patterns. Behavioral observations of the populations suggested that females were the protagonists of the study, leaving their territories to engage in extra‐ pair matings when the MHC genotype of their neighboring territorial male ‘‘smelled better’’ (had greater genetic complementarity) (Potts et al., 1992). MHC‐based mating preferences had been previously detected in the laboratory using inbred strains of mice (Yamazaki et al., 1976; for reviews see Jordan and Bruford, 1998; Penn and Potts, 1999), although the discovery was serendipitous and could have been easily missed. Also, these studies have been difficult to interpret for many reasons, not the least of

FITNESS AND GENE FUNCTION

185

Fig. 2. The percentage deficiency of MHC homozygotes relative to Hardy–Weinberg expectations in each of nine seminatural house mouse populations. The overall mean deficiency was 27% (Potts et al., 1991).

which is that reproductive behaviors of inbred animals have been affected by hundreds of generations of artificial selection (Manning et al., 1992a). In the case of MHC‐based mating preferences, social ecology is clearly critical to release the phenotype to selection, because the phenotype is social behavior. Testing mice in the context of Mus populations demonstrated that the selection coefficient arising from nonrandom mating was strong enough to maintain the allelic diversity found in surveys of wild populations, suggesting that mating preferences could indeed be the elusive source of selection maintaining MHC polymorphisms (Hedrick, 1992; Potts et al., 1991). 2. MHC‐Mediated Kin Recognition for Communal Nesting Partners House mice sometimes form communal nests where two or more females share nursing duties, apparently directed without bias toward all pups (Fig. 3) (Manning et al., 1992b). Evidence suggests that communal nesting functions at least in part to reduce infanticide in house mice (Manning et al., 1995), presumably because one female can guard the nest while other females are foraging. Communal nursing is a rare trait for mammals and it makes females vulnerable to cheating by communal nesting partners, who would do less than their fair share of the work or bias their nursing toward their own offspring. One way to reduce this conflict is for females to

186

LARA S. CARROLL AND WAYNE K. POTTS

Fig. 3. Female house mouse nursing members of three litters in a communal nest.

communally nest with relatives, thereby lowering costs by directing behavior toward kin. This prediction emerges from kin selection theory (Hamilton, 1964) and was tested by evaluating communal nesting patterns in seminatural populations of house mice. When familiar sisters were present in populations, they almost always chose each other as communal nesting partners. More importantly, communally nesting females with no sisters in their population showed a significant preference for settling with MHC‐similar females relative to random expectations. Due to the breeding design, MHC similarity was not correlated with relatedness, excluding the possibility that non‐MHC genetic cues were being used as indicators of relatedness. This preference for MHC‐similar communal nesting partners was the first example of a genetic‐based kin recognition system in vertebrates. In the context of this chapter, it is an example of the discovery of an ecologically specific function for naturally occurring genetic variants. This function would have been difficult to discover in a lab setting, but it was easily revealed under conditions allowing seminatural social ecology. 3. Sexual Conflict in Drosophila The sexes are predicted to have conflict in many aspects of their biological interactions. For example, in species with sperm competition, male adaptations might reduce the fitness of females, because the chemical warfare among ejaculates might have toxic effects in females. Drosophila

FITNESS AND GENE FUNCTION

187

provides us with striking examples that illustrate how mating is not necessarily a cooperative venture for the purpose of producing offspring. Indirect evidence suggests that male‐specific adaptations reduce female fitness in Drosophila because seminal fluid from a male is toxic to females, it reduces the propensity of females to remate, and decreases the competitive ability of sperm from other males, among other effects (Ravi Ram et al., 2005). Such interactions could lead to cycles of antagonistic coevolution between males and females. To test this hypothesis Rice (1996) designed an elegant set of experiments that allowed males to adapt to sperm competition occurring in females, but prevented females from making counteradaptations. Rice predicted that such unilateral male evolution would result in reduced fitness in females when interacting with these adapted males. After 30 generations of unilateral evolution, males showed 24% increased fitness relative to control males in population assays. The unilateral evolution was allowed to continue for a total of 41 generations after which adapted males caused significantly higher female mortality and the mortality rate was correlated with the mating rate. In this case, sperm competition within the female reproductive tract was ecologically critical for promoting the evolution of increasingly competitive sperm as well as for detecting the fitness consequences for both males and females. The genes involved in this remarkable case of experimental evolution would, when disrupted, likely look like no‐phenotype knockouts without the use of the specific ecology surrounding sperm competition. These approaches have continued to reveal remarkable insights into related aspects of Drosophila reproduction (Gibson et al., 2002; Holland and Rice, 1999; Rice and Holland, 2005; Rice et al., 2005). 4. Timing of Flowering in Arabidopsis The initiation of many behaviors in plants and animals are controlled by environmental cues such as day length. Day length can be manipulated in the laboratory to successfully initiate many seasonal behaviors, such as timing of flowering, which has been extensively studied in Arabidopsis under laboratory conditions. To determine if the genetic basis of timing of flowering in lab studies duplicates that in nature, quantitative trait loci (QTL) studies in these two environments were performed (Weinig et al., 2002). The surprising findings were that QTL important under lab conditions were often undetectable under field conditions and vice versa. These data suggest that many ecological cues important in nature are missing under laboratory conditions and that lab conditions initiate pathways that are silent under some field conditions. In a companion study, Weinig et al. (2003) went on to show that different field conditions and different genetic backgrounds favored different alleles at these QTL. Taken together, these

188

LARA S. CARROLL AND WAYNE K. POTTS

data underscore the importance of ecology and epistatic interactions (Section II.D) for understanding gene function. 5. Dictyostelium Under starving conditions, social Dictyostelium amoebas strike a coordinated venture in pursuit of a common interest—survival. Dictyostelium, with its simple physiology and behavior, plus the availability of molecular tools for analysis, has presented sociobiologists with an ideal model organism to study the molecular underpinnings of social evolution and cooperation. As bacterial food sources become meager, free‐living Dictyostelium secrete and track cAMP signals to form aggregations which may often be composed of genetically distinct clones (Fortunato et al., 2003). From these aggregations, a motile multicellular slug emerges and migrates to the soil surface to form a fruiting body. During this social phase of the life cycle, a ball of fertile reproductive spores differentiates from a large percentage of amoebas, while about 20% of cells assemble themselves altruistically into a slender stalk and die while elevating the spores for optimal dispersal (Bonner and Slifkin, 1949). As survival depends on directing one’s genes into a spore, cheaters that manage to escape the dead‐end fate of the stalk would be highly favored by natural selection. The search for genes that mediate social conflict in Dictyostelium led researchers to csA, a gene which codes for the homophilic cell adhesion protein, gp80. Aggregation behavior in Dictyostelium is regulated in part by csA which is expressed during the preaggregation and stalk formation stages. Although no other proteins are capable of compensating for csA’s EDTA‐resistant cell adhesion role during aggregation (Ponte et al., 1998), knockouts initially had no obvious developmental phenotype. Cells lacking the csA gene had similar aggregation timing and development to wild‐type amoebas in laboratory assays (Harloff et al., 1989). Even more curious, when knockouts and wild‐type amoebas were allowed to assemble into chimeric aggregates, the knockouts had a distinct advantage in becoming spores. Since the gp80 protein‐binding site recognizes and binds to copies of itself on cell membranes, the advantage of knockouts is partly due to a weakened intercellular binding which loosens them to the back of the slug where they are more likely to become spores. This puzzling result suggests a strong selective advantage to mutants that lose the csA allele, yet its presence in wild Dictyostelium and its specific expression pattern suggests some critical function during aggregation. Dictyostelium normally inhabits a complex three‐dimensional environment composed of soil, decaying leaves, and other forest detritus. Yet laboratory assays are typically performed on smooth two‐dimensional test surfaces of agar, nitrocellulose filters, or glass coverslips. With this in mind,

FITNESS AND GENE FUNCTION

189

Ponte et al. (1998) retested knockouts in petri dishes containing either agar or moistened soil. In comparison with wild‐type cells, aggregation behavior of csA knockouts was delayed 8–10 hr on soil, but not on agar. Moreover, fruiting body formation of knockouts was only 15% that of wild types—a reduction of approximately 99% overall when compared to performance on agar plates. Actual spore production by csA knockout cells was reduced overall by a similar amount, and when wild type and csA knockout cells were mixed on soil plates to test differential fitness, only 18% of the resulting colonies came from the knockout spores. This reduction in fitness is likely to result from the same cell adhesion deficiency that gives csA
a spore‐forming advantage on smooth surfaces. On soil or other complex substrates, the lack of csA protein product limits a cell’s ability to get into aggregates in the first place, resulting in a strong reproductive disadvantage (Queller et al., 2003). Other genes that affect social aggregation in Dictyostelium have proven simpler to phenotypically characterize in standard laboratory assays. For example, the product of dimA responds to the signaling molecule DIF‐1 which triggers differentiation into prestalk cells. By disrupting dimA, a cell could theoretically increase its chances of becoming a spore by avoiding the stalk entirely. However, although they outnumber wild‐type cells in the prespore phase, dimA
cells have a high failure rate during spore differentiation (Foster et al., 2004), preventing their reproductive domination over wild‐type cells. That is not to say that stronger or additional phenotypes are not waiting to be discovered by testing dimA and other ‘‘cheater’’ genes that have been recovered in mutant screens (Dao et al., 2000) under more natural conditions. C. COMPETITION AMPLIFIES SMALL PERFORMANCE DIFFERENCES INTO LARGER FITNESS DIFFERENCES The power of competition to amplify small differences among competitors has been a major theme in the ecological literature for decades (Koella, 1988; Latter and Sved, 1994; Smith and Holt, 1996; West Eberhard, 1983). However, this same literature largely fails to appreciate the power of competition to amplify phenotypic differences among genotypes when the examined genes are not specifically ‘‘social/sexual competition’’ genes (Carroll and Potts, in press). It is perhaps then no surprise that ecological approaches as a general screen for gene function have been largely overlooked by the functional genomics community. There are hundreds of papers and dozens of meetings per year on functional genomics; few consider the role of ecological approaches for revealing gene function (Feder and Mitchell‐Olds, 2003). As we illustrate in the examples that

190

LARA S. CARROLL AND WAYNE K. POTTS

follow, the strength of social competition to discriminate among genotypes extends far beyond genes that code for bright feathers, elaborate displays, and other sexually and socially selected traits. Figure 4 provides a hypothetical illustration of how competition in the real world alters the fitness distribution of mutants making many more detectable under ecological versus lab conditions. Panel (A) illustrates the fitness distribution of yeast knockouts under lab conditions. When these same genes are assayed under competitive, ecological conditions the fitness distribution is shifted down, resulting in many more genes with detectable phenotypes (Panel B).

Fig. 4. Competition amplifies small performance differences into larger, detectable fitness differences (see text).

FITNESS AND GENE FUNCTION

191

Intraspecific competition can take two major forms. Direct (interference) competition results in direct encounters between competitors, for example fights for territory ownership. During indirect (exploitation) competition, a resource is used by one individual thereby removing it from the resource pool for other competitors who might never be encountered directly. Traits that influence the efficiency of resource exploitation are favored for indirect competiton, whereas traits for fighting ability (or other forms of direct competition) are paramount for direct competition. Direct and indirect forms of competition often have dramatically different dynamics in their quantitative influence on fitness. During indirect competition, the difference in competitive ability is often proportional to fitness outcomes. For example, individuals that feed 10% more efficiently have 10% more offspring. In contrast, during direct competition, difference in competitive ability is often amplified into much larger fitness effects. As in the inbreeding example later, outbred males may only be 10% better duelists, but since they win most fights over territories (and nonterritorial males do not breed), the fitness consequences are dramatically amplified. Consequently, it may be easier to detect fitness consequences of a similar genetic defect (mutation) in species with direct competition compared to species with indirect competition. Later we provide examples in yeast, Drosophila, and mice where competition amplified fitness differences dramatically, turning no‐phenotype mutants into major phenotype mutants. 1. Gene Knockouts in Saccharomyces Many gene knockouts in yeast (S. cerevisiae) reveal no phenotypic change from wild type when grown under normal laboratory conditions. To determine if competition might reveal phenotypes, Thatcher et al. (1998) measured the fitnesses of a random collection of these disruption mutants in direct competition with their wild‐type progenitor. Figure 5 shows the fitness distribution of 34 no‐phenotype yeast knockout mutants (under no competition) when subjected to direct competition with wild type. Approximately 1/3 maintained their no‐phenotype status, but 2/3 expressed significant fitness declines ranging from 0.3 to 22%; two knockouts showed a significant fitness increase compared to wild type. Competition became a microscope that made the formerly invisible phenotypes visible and subsequent studies have now incorporated this approach (Giaever et al., 2002). 2. Mutation Accumulation in Drosophila Similar results have been demonstrated in Drosophila by Kondrashov and coworkers (Shabalina et al., 1997). They allowed mutations to accumulate

192

LARA S. CARROLL AND WAYNE K. POTTS

Fig. 5. Fitness distribution of 34 no‐phenotype yeast knockout mutants during direct competition with wild type (Thatcher et al., 1998). These 34 mutants showed no phenotypic change from wild type when grown separately under standard laboratory conditions (ns = not significant).

in populations of Drosophila for 30 generations. These mutation accumulation lines were allowed to compete with wild type to test for fitness declines either under benign or harsh competitive conditions. In benign conditions, food was not limited, eliminating most competition among adults and among their larvae. Consequently larval survival was high. In harsh conditions food was limiting, promoting competition among adults and among larvae, which resulted in larval survival of approximately 10%. The fitness declines of mutants under harsh population conditions were approximately 70% (2% per generation). However, the final fitness decline of mutants under benign population conditions was only 5%, an order of magnitude lower than under harsh conditions (Fig. 6). This represents a case where ecological stressors other than social stressors were also manipulated. However, since social competition and harsh ecological conditions were not manipulated independently, it is unclear what proportion of the large fitness declines were due to each variable or their interactions. Similar competition‐ amplified fitness effects have been demonstrated for inbreeding in Drosophila (Charlesworth and Charlesworth, 1987). 3. Inbreeding in Mus The primary cause of inbreeding depression is the expression of deleterious recessive alleles that are expressed at a higher rate in inbred individuals (Latter, 1998). These negative consequences have been well established for centuries. Two major studies have been conducted on mice and the reproductive consequences of one generation of full‐sib matings

FITNESS AND GENE FUNCTION

193

Fig. 6. Competitive performance (fitness) under harsh versus benign conditions of Drosophila lines allowed to accumulate mutations for 30 generations. Competition is against wild type. Means and regression lines are shown (adapted from Shabalina et al., 1997, with permission: # 1997, National Academy of Sciences, USA).

were estimated at about a 10% decline (Connor and Belucci, 1979; Lynch, 1977); almost all of the effect was due to reduced litter size. No attempt was made to measure fitness in any type of competitive social conditions. Meagher et al. (2000) repeated these experiments with the goal of adding fitness measures in competitive social conditions. Wild‐caught mice were bred so that the F2 generation came from either outbred or full‐sib matings. These progeny became the founders for six experimental populations. It was found that outbred males had five times more offspring than inbred males (Fig. 7A). This represented a tenfold amplification over the reproductive declines observed for males in breeding cages. Significant fitness declines were found for inbred females, but they were an order of magnitude smaller than the observed male declines (Fig. 7B). There was no significant difference between laboratory and enclosures results for females. These gender differences were attributed to the fact that males compete aggressively over territories and nonterritorial (subordinate) males have little reproductive success. In contrast, females had no limiting resources. It remains an open question whether the fitness consequences of inbreeding in females would approach males if they had to compete over critical resources such as food or nest sites. The dramatic fitness declines in inbred males were due both to a 41% reduced ability to gain territories and decreased survival. This was particularly true for territorial inbred males where 90% had died by the end of the experiment as compared to only 24% of outbred territorial males (Fig. 8). These results suggest that inbred males had difficulty maintaining territories, as well as gaining them.

194

LARA S. CARROLL AND WAYNE K. POTTS

Fig. 7. Relative reproductive success of inbred (solid) and outbred (open) males (A) and females (B) (Meagher et al., 2000). Male reproductive success is measured using a genetic marker on the Y‐chromosome, which explains why only sons are counted in (A).

Figure 7 shows the relative reproductive success of inbred and outbred males and females over time. This analysis demonstrates that the relative differences were increasing at the end of the experiment, suggesting that all the inbreeding depression estimates were conservative. If the populations had been allowed to continue to obtain lifetime reproductive success measures, the fitness differences between inbred and outbred animals would have been much larger. The same 10% reduction in litter size was observed under colony housing as was found in the two major previous studies on Mus inbreeding (Connor

FITNESS AND GENE FUNCTION

195

Fig. 8. Survivorship analysis of inbred (solid) and outbred (open) territorial males (Meagher et al., 2000).

and Belucci, 1979; Lynch, 1977), suggesting the inbreeding load in all three wild‐caught populations were similar. However, the analysis of adult male fitness added an additional 500% effect; outbred males had five times more offspring than inbred males. A recent survey of inbreeding studies demonstrates that in most cases stress amplifies the deleterious effects of inbreeding (Armbruster and Reed, 2005). Since inbreeding depression is primarily the expression of defective mutant genes (Latter, 1998), these results are particularly instructive as to the power and sensitivity of fitness assays for other gene function studies involving mutants or knockouts. Competition and other forms of stress increase the deleterious effects of mutants, making such tools useful for revealing phenotypes of mutants. 4. Resolving the Paradox of the Selfish t Complex The mouse t complex on chromosome 17 is a classic example of a selfish gene which increases its own genetic representation at the expense of its bearer. Across the globe, all subspecies of house mice (M. musculus and M. domesticus) carry versions of this segregation distorter complex, held genetically intact by four nonoverlapping inversions that effectively prevent crossing over and recombination within its 400 megabase span. Although females transmit the t complex in Mendelian frequencies, a heterozygote male will transmit the t complex to up to 100% of his offspring. This nearly perfect meiotic drive in males is accompanied by a well‐characterized cost—homozygosity at the t complex causes lethality

196

LARA S. CARROLL AND WAYNE K. POTTS

or sterility in males, depending on which combination of t haplotypes is inherited. However, this costly phenotype is only sufficient to keep the t complex from achieving complete fixation in populations. It is not sufficient to prevent the t complex from spreading to high frequencies. Early studies estimated that the dual effects of segregation distortion and homozygote lethality should yield population frequencies around 70% (Bruck, 1957). Yet the t complex staggers along at puzzlingly low levels around 6–25% (Ardlie and Silver, 1998; Dunn and Levene, 1961; Figueroa et al., 1988; Lenington et al., 1988; Myers, 1973), less than half of its expected frequency. In the 50 years since its discovery, the t complex has been studied empirically to determine the effects of fertility, fecundity, juvenile survival, and female choice (Dunn and Suckling, 1955; Dunn et al., 1958; Johnston and Brown, 1969; Lenington et al., 1994; Levine et al., 1980). Models have been constructed and computer simulations have been run to sort out the effects of drift, migration, and selection (Baker, 1981; Berry et al., 1991; Durand et al., 1997; Levin et al., 1969; Lewontin, 1968; Petras and Topping, 1983). Since then, many phenotypes of the t complex have been discovered and much theory has been published regarding the population dynamics of genetic elements possessing the peculiar characteristics of the t complex. Yet, perhaps not surprisingly, many of these results are in conflict with one another and no single study accounts for a significantly large proportion of the discrepancy between observed and expected frequencies. What these studies do show is that there are clearly many different relevant factors which limit the spread of the selfish t complex, making it nearly impossible to integrate all available data into a cohesive model for predicting the fitness of t haplotypes in nature. In an attempt to measure t haplotype fitness directly, Carroll et al. (2004a) analyzed pup genotypes to estimate lifetime reproductive success in 10 seminatural populations of wild house mice over the approximate span of a generation. This study of competing t‐bearing and non‐t‐bearing mice revealed a strong heterozygote disadvantage in both males and females. Heterozygote disadvantage had been predicted by previous models, but had not been convincingly demonstrated by laboratory assays. The novel phenotype emerging from long‐term competitive populations was a significant impairment of heterozygous t‐bearing males in their ability to gain territories—only 32% of heterozygous males gained territories, whereas 67% of non‐t‐bearing males gained territories (Carroll et al., 2004a). Female mice overwhelmingly prefer to breed with dominant males, which helps explain why in a single generation the t complex was at frequencies nearly 50% lower than expected (when both segregation distortion and male homozygote sterility were considered). An additional novel phenotype was the increased mortality of both t‐bearing male and

FITNESS AND GENE FUNCTION

197

female population founders under competition. These data collectively suggest that selection against t‐bearing heterozygotes in natural populations can easily resolve the paradox of why t frequencies in nature are so low. Although the populations were not run long enough to determine an equilibrium between heterozygote disadvantage and meiotic drive, the dramatic loss of t haplotypes from the enclosures in a single generation suggests this trend would lead to the ultimate exclusion of t‐bearing animals from the reproductive pool. Yet the t complex has survived over millions of years, and it is tempting to speculate that heterozygote disadvantage of t‐bearing mice is a phenotypically plastic phenomenon affected by social and ecological context. Without competition, t‐bearing animals are quite prolific. t‐Bearing males that successfully emigrate to found new populations could easily produce rapid increases in t frequencies by virtue of meiotic drive, serving as primary reservoirs of t haplotypes. In larger populations, individuals carrying t haplotypes will face competition and suffer lowered fitness, driving down t complex frequencies. Ardlie and Silver (1998) obtained t frequency data from a variety of natural populations. Their results suggest that small‐ and medium‐sized populations (<60 individuals) experience the largest fluctuations in t frequency and carry more t haplotypes than large populations. Large populations (>60 individuals) tend to carry low numbers of t haplotypes (average 3%) or none at all. This prediction of density‐dependent selection not only explains why t complex phenotypes have been so difficult to pin down in the laboratory, it also adds another dimension to the detection of subtle phenotypes, underscoring the argument that the appropriate context for studying a gene is the ecological circumstance in which its function evolved. D. GENETIC BACKGROUND PROBLEM It has become clear that many phenotypic effects of mutants depend on epistatic interactions with background genes (Leiter, 2002; Nadeau, 2001, 2003). Ten years ago, Threadgill et al. (1995) radically raised the awareness on this issue by illustrating the tremendous influence genomes can have on specific gene disruptions. Knocking out the gene for epidermal growth factor receptor caused early embryonic lethality in the CF‐1 mouse stain. However, CD‐1 mice carrying the mutation survived past birth for up to 3 weeks (Strunk et al., 2004; Threadgill et al., 1995). Such epistatic interactions could explain some cases of no‐phenotype mutants. The common solution to this problem is to breed a mutant onto many different inbred strains, but this is a slow and expensive process (Bucan and Abel, 2002). No approach is perfect, however, one feasible alternative is to test phenotypes in the context of wild‐derived, segregating backgrounds. This approach has the advantage

198

LARA S. CARROLL AND WAYNE K. POTTS

of reducing some of the inescapable effects of drift and artificial selection that afflict inbred strains. However, the reluctance of wild rodent females to breed in the laboratory potentially introduces extreme selection for animals predisposed to breed under artificial conditions. Inbred strains come with a tremendous load of accumulated genetic baggage from the unavoidable side effect of spontaneous deleterious mutations becoming genetically fixed through inbreeding and low effective population size. Data documenting these effects in inbred strains primarily come from the mouse literature. The mouse, with human homologues to 99% of its genes, has held distinction as the principal animal model for human disease, making it vitally important to characterize phenotypic variation among the established strains. However, even sublines of strains separated in different breeding colonies have been shown to carry fixed mutational differences (Simpson et al., 1997; Weiss et al., 1989). Phenotypic divergence of sublines has been documented for such phenotypes as aggressive behavior (Sluyter et al., 1999), response to cocaine (Henricks et al., 1997), susceptibility to Theiler’s virus‐induced demyelinating disease (Nicholson et al., 1994), and susceptibility to experimental Salmonella infections (McClelland et al., 2004). The Jackson Laboratory currently manages a comprehensive database supplying information on phenotypic strain differences in mice (www.jax.org/phenome). Mutation accumulation lines in C. elegans have shown similar effects, including degradation in behavior (Ajie et al., 2005) and other specific components of fitness occurring over a short period of time (Estes et al., 2005). A related problem that arises with inbred mice during characterization of knockout phenotypes is the potential misinterpretation of phenotypes that arise from linked lethal mutations to the gene of interest. In mice, gene targeting is typically performed in embryonic stem (ES) cells from the 129 inbred stains. Subtypes of this strain carry a number of known defects that can greatly confound interpretation of the targeted gene when they occur within its flanking regions. This problem garnered enough concern to a prompt a Banbury Conference on Genetic Background in Mice, which generated numerous recommendations for its remedy (Silva, 1997; Wolfer et al., 2002). However, although such recommendations may be relevant when a phenotype is detected, they are not expected to improve the detection of null phenotypes. As a whole, inbreeding in laboratory animals creates strong selection to adapt to the peculiar conditions of laboratory housing and breeding (Miller, 1994) so that inbred animals often display aberrant behaviors and physiological traits (Manning et al., 1992a) compared to their wild counterparts. For detecting behavioral and physiological phenotypes that are expressed in only a subset of genetic backgrounds, breeding mutations

FITNESS AND GENE FUNCTION

199

onto a wild, segregating background might be a straightforward compromise. Of course, the disadvantage of breeding onto outbred genomes is that this approach produces higher variance in data sets due to uncontrolled segregating genes. This was the major impetus for producing inbred laboratory strains in the first place. However, when testing performance or fitness differences, a wild outbred background will greatly facilitate expression of the full range of physiology and behaviors that a mutant animal would normally experience in nature. Despite the inherent selection favoring wild mice willing to breed in the laboratory, we argue that using outbred genomes may often be a more effective approach than the current approach of relying on many inbred strains that can have aberrant physiology, behavior, and accompanying epistatic effects. E. FITNESS DIFFERENCES

TOO

SMALL

TO

MEASURE

Genes with small effects may have functions that are ultimately too subtle for even the most exhaustive analyses to detect. Although these phenotypes might defy our keenest efforts to identify them, they are hardly invisible to natural selection, because what matters for a gene’s survival in nature is that the selection coefficient is roughly greater than the reciprocal of the effective population size (Kimura, 1985; Tautz, 2000). As effective population size increases, even a vanishingly small selective advantage would be enough to maintain a seemingly functionless gene against the effects of mutation and drift. Just as population size exposes genes to the discriminating sweep of natural selection, sample size might be a crucial factor for obtaining the statistical power to detect small genetic effects. For this reason, studies in bacteria, yeast, and other organisms that can be tested within the context of large populations might be amenable for testing the generality of small effects as an alternative explanation for genetic redundancy. For example, Thatcher et al. (1998) used competitive yeast cultures to monitor fitness declines in mutant versus wild‐type Saccharomyces cervisiae (Section II.C.1). This strategy permitted the detection of fitness declines as small as 0.3%, reducing the number of yeast genes with no known phenotypes from 100% to only 20%. Similarly, Smith et al. (1995) used a novel assay to measure the ‘‘genetic footprints’’ of random gene mutations competed in batch culture. By this method, fitness declines were detected in over 60% of 255 randomly derived mutant strains (Smith et al., 1995; Thatcher et al., 1998). In mice, competitive population studies are capable of detecting fitness differences on the order of 10–15%. This means that many mutations with strong selective effects (s > 0.01) will still be undectable in fitness assays. For such genes, sequence analysis will remain the leading method for inferring function by detecting evidence of selection.

200

LARA S. CARROLL AND WAYNE K. POTTS

Genetic sequence comparisons among related species with divergent population sizes could help determine whether a gene is maintained due to a small fitness effect or whether its maintenance is not a direct effect of population size but is likely due to an unidentified yet significant function. F. WHY IS BEHAVIOR SO CRITICAL WHEN MEASURING FITNESS? In the postgenomics era, we may hope to find few if any genes chiefly dedicated to specific behaviors. Rather, genes that affect behavior are pleiotropic so that a behavioral phenotype will result from mutations in genes that affect many physiological processes whether these are fundamentally metabolic or neurobiological. Stated otherwise, behavior is the whole organismal response to various combinations of specific cellular, molecular, and physiological processes. Therefore, the collective outcome of these processes can be studied by measuring behavioral performance. In most metazoans, fitness is achieved primarily through successful behavior such as predator avoidance and intra‐ and interspecific competition for resources. The remaining organismal biology largely becomes infrastructure for these activities because behavior puts physiology to its greatest tests. Thus, defects in this behavioral infrastructure below the detectable threshold (e.g., cryptic‐phenotype mutants) might still manifest noticeably during the performance of behaviors that demand energy, endurance, neuromuscular coordination, and so on. This is particularly true in light of the numerous examples where relatively small differences in physiological performance are amplified into large fitness differences by intraspecific competition (Section II.C). There are few physiological systems in house mice (and other behavior‐rich metazoans) whose deficiencies will not result in fitness‐ reducing behavioral impairment. Under this view, almost all genes become behavioral genes and consequently, when phenotypes are cryptic, behaviorists may be the best biologists at detecting the resulting phenotypes, as well as the components of fitness most affected. Studying behavior under natural conditions sufficient to measure fitness is one major way to reveal phenotypes of mutants. Unfortunately, there is almost no mention of this approach from either the phenomic or functional genomic communities. This failure to appreciate the power of behavior‐ related fitness measures is a major rationale for writing this chapter. G. WHY SEMINATURAL MAY OFTEN BE MORE EFFICIENT THAN NATURAL: SHOULD YOU TEST YOUR MOUSE AGAINST A CAT OR ANOTHER MOUSE? Whether your favorite organism is predator or prey, the ultimate measure of fitness is lifetime reproductive success. When resources are limiting,

FITNESS AND GENE FUNCTION

201

there are generally fewer breeding opportunities than there are fertile individuals, and since the most physically robust, pathogen‐free, predator savvy individuals are those that win reproductive opportunities, this means that reproduction falls to those who win the competition for food, basking spots, predator‐free hiding sites, and other limited resources. For this reason, competition in experimental populations might serve as a useful proxy for natural selection, even when experimental populations lack many of the important components of natural selection. In nature, the losers of intrasexual competition are killed by starvation, predators, disease, and other difficult to measure effects. By eliminating these natural selective factors while simultaneously creating competition for the resources that would serve to restrict them, potential breeders are excluded from territories not by predators and starvation, but by competitors. Reproductive winners are those that successfully gain access to mates and to sites appropriate for the rearing of offspring. Staged seminatural conditions are impossible for many species. For these species nature becomes the only place to obtain realistic fitness measures. Many long‐term field studies have shown that an amazing level of detail can be revealed by studying animal populations in nature. Just a few examples include lions (Packer et al., 2005), Darwin’s finches (Grant, 1986), Florida scrub jays (Wolfenden and Fitzpatrick, 1996), and acorn woodpeckers (Koenig and Mumme, 1987). For species that are amenable to a seminatural approach, measuring selection in competitive experimental populations offers a practical compromise between nature and the laboratory. For vertebrates in particular, selection is difficult to measure in the laboratory. Forcing reproduction in caged breedings can only give a narrow range of results regarding the mechanisms underlying reproductive differences among genotypes. However, studies performed in the wild have problems of their own. Stochastic environmental conditions (weather, food, shelter, and so on) add noise to already statistically complex data sets, and lifetime measures of fitness which could be easily measured in artificial populations, are confounded in nature by the loss of subjects to dispersal and various sources of mortality. That is to say, testing your mouse against another mouse might be a less stochastic, more tractable solution for determining exactly which one is more adept at evading the cat. H. GENE FUNCTION STUDIES WILL SELDOM BE COMPLETE WITHOUT FITNESS ASSAYS Even if a phenotype is detected in the laboratory for a gene knockout or mutant, there remain at least two important aspects of gene function

202

LARA S. CARROLL AND WAYNE K. POTTS

that require fitness studies in order to comprehensively understand the function(s) of that gene. First, we need to find the true fitness consequence of lab phenotypes because their relative importance in the real world may be difficult to predict from lab‐assayed phenotypes. Second, there may be additional, important phenotypes that were missed in the laboratory screens. 1. Relative Importance of a Particular Gene Must Ultimately Be Measured in the Currency of Evolution: Fitness Fitness measurements are important for determining how essential or nonessential a gene is—the strength of selection acting against its knockout. Such measurements provide a quantitative measure of the relative importance (essentialness) of a gene. It will often be difficult to estimate the actual fitness declines of a given lab‐assayed phenotype that is not lethal or near lethal. This is because estimations require extrapolation from minor phenotypes in the lab to their fitness consequences in the context of complex epistatic and ecological interactions as well as the harsh competitive conditions of nature. This is demonstrated by all four of our examples in Sections II.C.1 , II.C.2, II.C.3, and II.C.4 where phenotypes were initially invisible or minor, but had major fitness consequences under harsh competitive conditions. The relative fitness decline is the accurate measure of how important that mutation would be to its bearer in nature. Are phenotypes trivial if detectable only in fitness assays? The answer is obviously no if you consider the inbreeding results in Section II.C.3. Being an inbred male is equivalent to having a lethal gene with 80% penetrance. The reduced health and vigor of inbred males prevent them from effectively competing against conspecifics. This should be of foremost interest to conservation biologists concerned with the genetic health of species communities and of no less interest to the biomedical community concerning human welfare. It is not that inbreeding‐associated declines in health and vigor are trivial, but rather, that our previous phenotyping methods were insensitive. For example, quantitative defects in most metabolic pathways and organ function would go undetected until they became debilitating. Many neurological disorders in animals, such as migraine headaches, would go undetected under most lab assays. However, these conditions in humans would be considered disease and they would be detectable during competition in mouse and other vertebrate populations. The danger of misinterpreting laboratory artifacts or detecting nonsense phenotypes is yet another important reason for characterizing gene function using an ecological approach. Genes have evolved to function in the context of the natural environment, so artificial environments can cause the expression of inappropriate phenotypes. For example, the genetic basis of flowering time in Arabidopsis is one of this model organism’s most studied

FITNESS AND GENE FUNCTION

203

traits and many QTL have been identified in laboratory studies (Section II. B.4). It was a great surprise to find out that when similar QTL studies were conducted in natural field experiments, many new loci were found that had not and could not have been detected in laboratory experiments (Weinig et al., 2002). Furthermore, many QTL important in the lab had no detectable effects in nature. 2. Discovery of Additional Phenotypes A single gene can influence many phenotypic traits (pleiotropy) and this is probably the general rule rather than the exception (Fraser and Marcotte, 2004). Consequently, if a phenotype is already known for a mutant or for a natural genetic variant, additional unknown phenotypes may await discovery. Most of the examples previously described in Sections II.B and II.C are cases where fitness assays revealed major new roles for genes that already had well‐characterized phenotypes. For example, our early MHC experiments used seminatural populations in house mice to test for pathogen‐associated selection (Sections II.B.1 and II.B.2). Consistent with the idea that homozygotes would be more susceptible to pathogens, we found a deficiency of MHC homozygous offspring. However, analysis of the components of fitness revealed not one but two novel phenotypes for MHC genes: first, the observed deficiency of homozygotes was not because they were dying from pathogens, but rather because females were preferring to mate with MHC dissimilar males (Potts et al., 1991) (Section II.B.1). Later we were able to show that these same MHC genes also allowed the recognition of unfamiliar kin during the choice of communal nesting/nursing partners (Manning et al., 1992b) (Section II.B.2). Most genetic mutants will probably have multiple phenotypes, many of which may be invisible in laboratory tests, but may be revealed during ecological competition.

III. GENE FUNCTION STUDIES DEMAND INTEGRATIVE APPROACHES The era of functional genomics affords a great opportunity for organismal biologists to collaborate with molecular biologists to truly evaluate how genes function through all levels of biological organization (Feder and Mitchell‐Olds, 2003). One might say that the ultimate reductionist act has been committed—sequencing of genomes. Genome projects will largely be failures until the functions of these genes are clarified, a task that will often require organismal and ecological approaches. This endeavor promises to be a major application of integrative biology that could begin to heal the divisive wounds that tore apart our great biology departments in the last decades of the twentieth century.

204 A. INTEGRATING

LARA S. CARROLL AND WAYNE K. POTTS

THE

FITNESS COMPONENT

OF

PHENOMICS

Our central thesis is that testing fitness will often be integral to understanding gene function. Competitive population studies are capable of providing the most direct fitness measures while simultaneously providing a comprehensive comparison of genotypes with respect to important variables such as male and female activity patterns, dominance, reproduction, longevity, and offspring‐rearing capacity. However, setting up population studies are by no means trivial, especially for larger metazoans and nonsocial species. Researchers working on vertebrate species might be wise to start with simpler approaches to learn as much as possible about the gene or trait of interest using tools that are readily available in a laboratory setting. Despite the surfeit of mutants with no obvious phenotypes, there are nevertheless many cases where a little or a lot of concentrated effort in the laboratory will be rewarded. The basic problem is how to best proceed with phenotype analysis. As behavior represents the combined organismal response to all molecular, cellular, and physiological processes, it is certainly the most complex, but also perhaps the most fruitful area to begin the search. Most researchers find it prudent to begin with a battery of behavioral tests. A variety of guidelines and recommended protocols exist for this purpose, which are intended to help improve across‐ laboratory standardization and rigor (Bolivar et al., 2000; Crawley, 2000; Crawley and Paylor, 1997; Hatcher et al., 2001). Accordingly, the relatively new field of behavioral phenomics is an especially ripe area for the elucidation of gene function. Organisms with complex behavioral repertoires present the greatest challenge for efficient phenotyping. At the forefront of testing technology, sophisticated equipment is becoming available for automated behavioral monitoring and testing of mice and rats (Gerlai, 2002; Tecott and Nestler, 2004). The vast datasets these instruments are capable of producing are once again raising the bar for bioinformatics to facilitate the handling, processing, organization, and retrieving of tremendous information flow. The hope is that improved across‐laboratory consistency, reliability, and comparative analysis will not only help reveal hidden phenotypes, but will simultaneously avoid the opposite pitfall—detecting a phenotype when none exists or misinterpreting a phenotype. Phenotypes represent not only the effect of a disrupted gene, but depend also on genetic background (Strunk et al., 2004; Threadgill et al., 1997), age (Crabbe et al., 1999; Heiman‐Patterson et al., 2005; Hultcrantz and Li, 1993; McIlwain et al., 2001), experience (McIlwain et al., 2001), and environment (Crabbe et al., 1999). Therefore, although the entire behavioral phenome is likely to occupy an enormous space, a large segment of the phenome will

205

FITNESS AND GENE FUNCTION

undoubtedly reside within ecological space, involving the extended interplay of genes and environment. Phenomics technologies are still largely based on measuring the physiology and behavior of individual animals, and therefore have a long way to go before replicating the complex social milieu of experimental population studies. Nevertheless, automated technologies have many uses, from tracking motion, to measuring the duration of such complex behaviors as eating and grooming. Some of the more clever technologies are even beginning to integrate a more naturalistic social environment into the testing design. One such example is IntelliCage, manufactured by NewBehavior Inc. (Zurich, Switzerland; http://www. newbehavior.com). This instrument enables the simultaneous tracking and testing of multiple interacting animals. Although laboratory‐based phenomics testing does not yet offer a substitute for long‐term fitness studies, these technologies have proven to be extremely informative and continue to make rapid technological advances as researchers demand more from their assays.

B. HOW DO FITNESS MEASURES CONTRIBUTE MOLECULAR BASIS OF PHENOTYPES?

TO

UNDERSTANDING

THE

One criticism of the ecological approach espoused here is that ‘‘fitness differences in population cages will not easily lead to understanding the function of these genes in a more mechanistic sense.’’ However, we are presenting the ecological approach for understanding gene function not as a substitute for mechanistic studies, but as a vital first step in the process, because determining the function of a gene and the mechanistic basis of its associated phenotype is greatly aided by a full characterization of the phenotype. Most diseases are first discovered as an organismal defect, usually with symptoms that do not reveal the molecular and physiological basis of the malady. Once the disease phenotype is characterized, we then go on to characterize its molecular, cellular, and physiological bases. This has often taken decades. Diseases characterized in seminatural conditions are no different than diseases characterized any other way. The struggle to elucidate biochemical and biological details will proceed in identical ways as diseases identified by any other means. The advantage of an ecological approach is that forward and reverse genetic studies are both possible once fitness defects of knockouts or known mutations are revealed. We are therefore much farther ahead at characterizing the mechanistic basis of a mutant than when we are fooled into thinking there is no defect, which is the case any time functional redundancy is falsely invoked.

206

LARA S. CARROLL AND WAYNE K. POTTS

Our proposed approach simply identifies disease states that are difficult to detect in other ways. It gives voice to mice who can now tell us, ‘‘Bearing a t allele causes me discomfort; I am only half the mouse I used to be.’’ As a consequence, we can combine an advantage of human medicine (where the patient tells you it hurts) with the advantages of experimental animal studies. Our ecological approach revealed defects in t‐bearing mice having massive evolutionary consequences, equivalent to a lethal gene with 29% penetrance. We can now proceed to identify and characterize the molecular basis of these defects which were invisible under four decades of traditional approaches. In this age of evo‐devo, developmental and evolutionary biologists are increasingly eager to share ideas and insights across fields, using the principles of natural selection and evolution along with biological and molecular tools to attack problems of mutual interest. Despite these melding of interests, there is a general lack of appreciation for the idea that genes may be developmentally critical if they are regulated during embryogenesis but only manifest phenotypes at later stages of development or adulthood, and furthermore, that genes which are only expressed during later stages of development and adulthood are nevertheless essential if they mediate successful reproduction. This includes, but is not limited to genes which enable procurement of resources critical to obtaining mates. For this reason, phenotypic changes that show up under competitive circumstances are utterly relevant to the study of development. The ultimate and only meaningful test of all development is how it influences adult performance (fitness). Developmental genes that fail this test will be discarded by natural selection. Successful embryogenesis is the intermediate process on the way to high‐performance adults. Thus, testing adult performance is requisite for evaluating successful embryogenesis. If we are going to take seriously the challenge of determining the function of genes in the postgenomic era, we must have sensitive methods for detecting less obvious phenotypes. The ultimate function of many genes will be to increase competitiveness by enhancing what might be called ‘‘general health and vigor.’’ Enhanced vigor can be achieved in innumerable ways such as increasing metabolic efficiency, neuromuscular coordination, and so on. Each of these mutants will have a molecular and physiological basis and when we discover it we will not call it general vigor anymore, we will call it by its specific name, such as a metabolic defect. But without sensitive methods to identify organismal defects, these molecular defects will largely remain undetected. The ecological approaches proposed here do not replace current functional genomic tools; they add a sensitive screen allowing detection of important but cryptic functions.

FITNESS AND GENE FUNCTION

C. NONMODEL ORGANISMS

AND

207

FUNCTIONAL GENOMICS

Female zebra finches, with an acoustic call structure far simpler than that of their musical mates, were long assumed to lack the vocal skills capable of allowing males to distinguish them individually. That is, until Christopher Sturdy (2004) discovered that males can and do respond to their mate’s call—it just takes the right social environment. Male zebra finches respond to their mate’s call twice as often as to that of an unfamiliar female when he finds himself in the presence of a mated pair of zebra finches. But a male’s brain simply does not activate the same way when he is alone (Vignal et al., 2004). Clearly, his ability to judge the importance of social context is more sophisticated than our own naı¨ve attempts. The field of Sociogenomics (Robinson, 1999) takes such experiments a step further, by asking not just ‘‘why,’’ but ‘‘how.’’ The goal of Sociogenomics is to dissect the molecular underpinnings of social life, and as such, focuses well beyond the familiar model organisms examined in this chapter, to all creatures displaying potentially complex social behaviors, from Dictyostelium to hymenoptera to birds and other beasts. To understand social behavior and how it evolves, sociogenomic researchers track down genes and regulatory pathways that underlie development, physiology, and behavior using the same genomics tools as do conventional molecular and developmental geneticists. What distinguishes this field from that of connected molecular and genetic research is its special focus on species that live in societies and its emphasis on naturalistic conditions as a prerequisite for study (Robinson et al., 2005). The related fledgling field of evolutionary and ecological functional genomics (Feder and Mitchell‐Olds, 2003) similarly seeks to understand which genes effect ecological success and influence fitness in nature and how they do it. Integration of these two approaches with conventional genomics offers the opportunity to broaden genetic studies to include phenotypes that are not found in model organisms and moreover, to allow inferences into the evolution of traits through comparative studies with outgroups of species carrying genes of interest.

IV. SUMMARY The enterprise of determining the function of genes is by far the most difficult portion of genome projects. This reflects the sheer complexity of the genome, with genes interacting to influence function (epistasis), genes influencing more than one function (pleiotropy), the involvement of many genes to effect one function (polygenic traits), and countless gene‐associated

208

LARA S. CARROLL AND WAYNE K. POTTS

phenotypes yet to be discovered. A particular problem emerging from targeted gene‐disruption technologies is that many of these gene knockouts seem to have no phenotypic effect on the organism. The conventional explanation of such observations is to invoke functional redundancy in genomes. Although this may explain some cases, our review of the literature here suggests that many, if not the majority of such observations represent situations where if the mutant gene was tested under the ecological stresses and contingencies in which they evolved, functional defects could be measured as substantial declines in specific components of fitness. Here we review and develop this ecological approach for evaluating the functional effects of gene mutants, knockouts, or variants. Such ecological approaches are already in use in nonmodel organisms, largely for evaluating functional consequences of genetic variants. Thus the research program does not represent anything particularly new other than pointing out what should be obvious—to succeed over long‐term evolution, alleles must outperform the fitness contribution of genetic variants (and mutants) within the ecological conditions where they function. Yet, when one looks at what is published in functional genomic journals or topics at functional genomics meetings, one seldom observes attempts to test gene function under the ecologies in which they evolved. In the same journals functional redundancy emerges as the default explanation in cases where genes are knocked out but with little to no phenotypic effect. When functional redundancy is accepted as the explanation for no phenotypic change, research on that mutant largely comes to a halt. Here we review many cases where fitness‐based assays under seminatural ecological conditions revealed phenotypes (often major phenotypes) that were missed in laboratory studies. Developing such a research program provides a great opportunity for the development of a truly integrative biology, where we begin to understand how genetic change influences molecular, cellular, and physiological changes that ultimately control the fitness‐influencing performance of whole organisms. We conclude that functional genomics will often require an understanding of ecology and behavior to gain a useful understanding of gene function.

Acknowledgments We thank Jon Seger and Leda Ramoz who made important contributions to our thinking during the development of these ideas. Jon Seger created Figure 4. We also thank Adam Nelson, Jane Brockmann, David Queller, and an anonymous referee for important comments on the chapter. This chapter was written while WKP was supported by grants from NSF (IBN‐0344907) and NIH (RO1‐GM039578).

FITNESS AND GENE FUNCTION

209

REFERENCES Ajie, B. C., Estes, S., Lynch, M., and Phillips, P. C. (2005). Behavioral degradation under mutation accumulation in Caenorhabditis elegans. Genetics 170, 655–660. Ardlie, K. G., and Silver, L. M. (1998). Low frequency of t haplotypes in natural populations of house mice (Mus musculus domesticus). Evolution 52, 1185–1196. Armbruster, P., and Reed, D. H. (2005). Inbreeding depression in benign and stressful environments. Heredity 95, 235–242. Austin, C. P., Battey, J. F., Bradley, A., Bucan, M., Capecchi, M., Collins, F. S., Dove, W. F., Duyk, G., Dymecki, S., Eppig, J. T., Grieder, F. B., Heintz, N., et al. (2004). The knockout mouse project. Nat. Genet. 36, 921–924. Baker, A. E. M. (1981). Gene flow in house mice: Introduction of a new allele into free‐living populations. Evolution 35, 243–258. Berry, R. J., Triggs, G. S., King, P., Nash, H. R., and Noble, L. R. (1991). Hybridization and gene flow in house mice introduced into an existing population on an island. J. Zool. 225, 615–632. Bodmer, W. F. (1972). Evolutionary significance of the HL‐A system. Nature 237, 139–145. Bolivar, V., Cook, M., and Flaherty, L. (2000). List of transgenic and knockout mice: Behavioral profiles. Mamm. Genome 11, 260–274. Bonner, J. T., and Slifkin, M. K. (1949). A study of the control of differentiation: The proportions of stalk and spore cells in the slime mold Dictyostelium discoideum. Am. J. Botany 36, 727–734. Brenner, S. (2000) Genomics. The end of the beginning. Science 287, 2173–2174. Brookfield, J. F. (1997). Genetic redundancy: Screening for selection in yeast. Curr. Biol. 7, R366–R368. Bruck, B. (1957). Male segregation ratio advantage as a factor in maintaining lethal alleles in wild populations of house mice. Genetics 43, 152–158. Bucan, M., and Abel, T. (2002). The mouse: Genetics meets behaviour. Nat. Rev. Genetics 3, 114–123. Capecchi, M. R. (1997). Hox genes and mammalian development. Cold Spring Harb. Symp. Quant. Biol. 62, 273–281. Carr, D. E., and Dudash, M. R. (1995). Inbreeding depression under a competitive regime in Mimulus guttatus: Consequences for potential male and female function. Heredity 75, 437–445. Carrington, M., Nelson, G. W., Martin, M. P., Kissner, T., Vlahov, D., Goedert, J. J., Kaslow, R., Buchbinder, S., Hoots, K., and O’Brien, S. J. (1999). HLA and HIV‐1: Heterozygote advantage and B*35‐Cw*04 disadvantage. Science 283, 1748–1752. Carroll, L. S., and Potts, W. K. (in press). Sexual selection: Using social ecology to determine fitness differences. In ‘‘Rodent Societies’’ (J. O. Wolff and P. W. Sherman, Eds.), University of Chicago Press, Chicago. Carroll, L. S., Meagher, S., Morrison, L., Penn, D. J., and Potts, W. K. (2004a). Fitness effects of a selfish gene (the Mus t complex) are revealed in an ecological context. Evolution 58, 1318–1328. Carroll, S., Grenier, J., and Weatherbee, S. (2004b). ‘‘From DNA to Diversity.’’ Blackwell Publishing, Oxford. Charlesworth, D., and Charlesworth, B. (1987). Inbreeding depression and its evolutionary consequences. Annu. Rev. Ecol. Syst. 18, 237–268. Cheng, C. H., and Chen, L. (1999). Evolution of an antifreeze glycoprotein. Nature 401, 443–444. Conant, G. C., and Wagner, A. (2004). Duplicate genes and robustness to transient gene knock‐downs in Caenorhabditis elegans. Proc. Roy. Soc. Bio. Sci. 271, 89–96.

210

LARA S. CARROLL AND WAYNE K. POTTS

Connor, J. L., and Belucci, M. J. (1979). Natural selection resisting inbreeding depression in captive wild house mice (Mus musculus). Evolution 33, 929–940. Cooke, J., Nowak, M. A., Boerlijst, M., and Maynard‐Smith, J. (1997). Evolutionary origins and maintenance of redundant gene expression during metazoan development. Trends Genet. 13, 360–364. Crabbe, J. C., Wahlsten, D., and Dudek, B. C. (1999). Genetics of mouse behavior: Interactions with laboratory environment. Science 284, 1670–1672. Crawley, J. N. (2000). ‘‘What’s Wrong With My Mouse?’’ John Wiley & Sons, New York. Crawley, J. N., and Paylor, R. (1997). A proposed test battery and constellations of specific behavioral paradigms to investigate the behavioral phenotypes of transgenic and knockout mice. Horm. Behav. 31, 197–211. Dao, D. N., Kessin, R. H., and Ennis, H. L. (2000). Developmental cheating and the evolutionary biology of Dictyostelium and Myxococcus. Microbiology 146(Pt. 7), 1505–1512. Darwin, C. R. (1859). ‘‘On the Origin of Species by Means of Natural Selection, or the Preservation of Favoured Races in the Struggle for Life.’’ John Murray, London. Davies, E. K., Peters, A. D., and Keightley, P. D. (1999). High frequency of cryptic deleterious mutations in Caenorhabditis elegans. Science 285, 1748–1751. Doherty, P. C., and Zinkernagel, R. M. (1975). A biological role for the major histocompatibility antigens. Lancet 1, 1406–1409. Duboule, D. (2000). Developmental genetics. A Hox by any other name. Nature 403, 609–610. Dunn, L. C., and Levene, H. (1961). Population dynamics of a variant t allele in a confined population of wild house mice. Evolution 15, 385–393. Dunn, L. C., and Suckling, J. A. (1955). A preliminary comparison of the fertilities of wild house mice with and without a mutant at locus T. Am. Nat. 89, 231–233. Dunn, L. C., Beasley, A. B., and Tinker, H. (1958). Relative fitness of wild house mice heterozygous for a lethal allele. Am. Nat. 92, 215–220. Durand, D., Ardlie, K., Buttel, L., Levin, S. A., and Silver, L. M. (1997). Impact of migration and fitness on the stability of lethal t‐haplotype polymorphism in Mus musculus: A computer study. Genetics 145, 1093–1108. Endler, J. A. (1986). ‘‘Natural Selection in the Wild.’’ Princeton University Press, Princeton. Estes, S., Ajie, B. C., Lynch, M., and Phillips, P. C. (2005). Spontaneous mutational correlations for life‐history, morphological and behavioral characters in Caenorhabditis elegans. Genetics 170, 645–653. Feder, M. E., and Mitchell‐Olds, T. (2003). Evolutionary and ecological functional genomics. Nat. Rev. Genet. 4, 651–657. Figueroa, F., Neufeld, E., Ritte, U., and Klein, J. (1988). t‐Specific DNA polymorphisms among wild mice from Israel and Spain. Genetics 119, 157–160. Force, A., Lynch, M., Pickett, F. B., Amores, A., Yan, Y. L., and Postlethwait, J. (1999). Preservation of duplicate genes by complementary, degenerative mutations. Genetics 151, 1531–1545. Fortunato, A., Strassmann, J. E., Santorelli, L., and Queller, D. C. (2003). Co‐occurrence in nature of different clones of the social amoeba, Dictyostelium discoideum. Mol. Ecol. 12, 1031–1038. Foster, K. R., Shaulsky, G., Strassmann, J. E., Queller, D. C., and Thompson, C. R. (2004). Pleiotropy as a mechanism to stabilize cooperation. Nature 431, 693–696. Fraser, A. G., and Marcotte, E. M. (2004). A probabilistic view of gene function. Nat. Genet. 36, 559–564. Gao, L. Z., and Innan, H. (2004). Very low gene duplication rate in the yeast genome. Science 306, 1367–1370.

FITNESS AND GENE FUNCTION

211

Gerlai, R. (2002). Phenomics: Fiction or the future? Trends Neurosci. 25, 506–509. Giaever, G., Chu, A. M., Ni, L., Connelly, C., Riles, L., Veronneau, S., Dow, S., Lucau‐Danila, A., Anderson, K., Andre, B., Arkin, A. P., Astromoff, A., et al. (2002). Functional profiling of the Saccharomyces cerevisiae genome. Nature 418, 387–391. Gibert, J. M. (2002). The evolution of engrailed genes after duplication and speciation events. Dev. Genes Evol. 212, 307–318. Gibson, J. R., Chippindale, A. K., and Rice, W. R. (2002). The X chromosome is a hot spot for sexually antagonistic fitness variation. Proc. Biol. Sci. 269, 499–505. Grant, P. R. (1986). ‘‘Ecology and Evolution of Darwin’s Finches.’’ Princeton University Press, Princeton, New Jersey. Greer, J. M., Puetz, J., Thomas, K. R., and Capecchi, M. R. (2000). Maintenance of functional equivalence during paralogous Hox gene evolution. Nature 403, 661–665. Gu, X. (2003). Evolution of duplicate genes versus genetic robustness against null mutations. Trends Genet. 19, 354–356. Gu, Z., Steinmetz, L. M., Gu, X., Scharfe, C., Davis, R. W., and Li, W. H. (2003). Role of duplicate genes in genetic robustness against null mutations. Nature 421, 63–66. Hamilton, W. D. (1964). The genetical evolution of social behaviour. J. Theor. Biol. 7, 1–52. Harloff, C., Gerisch, G., and Noegel, A. A. (1989). Selective elimination of the contact site A protein of Dictyostelium discoideum by gene disruption. Genes Dev. 3, 2011–2019. Hatcher, J. P., Jones, D. N., Rogers, D. C., Hatcher, P. D., Reavill, C., Hagan, J. J., and Hunter, A. J. (2001). Development of SHIRPA to characterise the phenotype of gene‐ targeted mice. Behav. Brain Res. 125, 43–47. Hayes, C. N., Winsor, J. A., and Stephenson, A. G. (2005). Multigenerational effects of inbreeding in Cucurbita pepo ssp. texana (Cucurbitaceae). Evolution 59, 276–286. Hedrick, P. W. (1992). Female choice and variation in the major histocompatibility complex. Genetics 132, 575–581. Heiman‐Patterson, T. D., Deitch, J. S., Blankenhorn, E. P., Erwin, K. L., Perreault, M. J., Alexander, B. K., Byers, N., Toman, I., and Alexander, G. M. (2005). Background and gender effects on survival in the TgN(SOD1‐G93A)1Gur mouse model of ALS. J. Neurol. Sci. 236, 1–7. Henricks, K. K., Miner, L. L., and Marley, R. J. (1997). Differential cocaine sensitivity between two closely related substrains of C57BL mice. Psychopharmacology (Berlin) 132, 161–168. Hill, A. V., Allsopp, C. E., Kwiatkowski, D., Anstey, N. M., Twumasi, P., Rowe, P. A., Bennett, S., Brewster, D., McMichael, A. J., and Greenwood, B. M. (1991). Common west African HLA antigens are associated with protection from severe malaria. Nature 352, 595–600. Holland, B., and Rice, W. R. (1999). Experimental removal of sexual selection reverses intersexual antagonistic coevolution and removes a reproductive load. Proc. Natl. Acad. Sci. USA 96, 5083–5088. Hughes, A. L., and Nei, M. (1988). Pattern of nucleotide substitution at major histocompatibility complex class I loci reveals overdominant selection. Nature 335, 167–170. Hultcrantz, M., and Li, H. S. (1993). Inner ear morphology in CBA/Ca and C57BL/6J mice in relationship to noise, age and phenotype. Eur. Arch. Otorhinolaryngol. 250, 257–264. Jime´nez, J. A., Hughes, K. A., Alaks, G., Graham, L., and Lacy, R. C. (1994). An experimental study of inbreeding depression in a natural habitat. Science 266, 271–273. Johnston, P. G., and Brown, G. H. (1969). A comparison of the relative fitness of genotypes segregating for the tw2 allele in laboratory stock and its possible effect on gene frequency in mouse populations. Am. Nat. 103, 5–21.

212

LARA S. CARROLL AND WAYNE K. POTTS

Jordan, W. C., and Bruford, M. W. (1998). New perspectives on mate choice and the MHC. Heredity 81, 127–133. Kafri, R., Bar‐Even, A., and Pilpel, Y. (2005). Transcription control reprogramming in genetic backup circuits. Nat. Genet. 37, 295–299. Kamath, R. S., Fraser, A. G., Dong, Y., Poulin, G., Durbin, R., Gotta, M., Kanapin, A., Le Bot, N., Moreno, S., Sohrmann, M., Welchman, D. P., Zipperlen, P., et al. (2003). Systematic functional analysis of the Caenorhabditis elegans genome using RNAi. Nature 421, 231–237. Kimura, M. (1985). ‘‘The Neutral Theory of Molecular Evolution.’’ Cambridge University Press, Cambridge, UK. Klein, J. (1986). ‘‘Natural History of the Histocompatibility Complex.’’ Wiley, New York. Koella, J. C. (1988). The tangled bank: The maintenance of sexual reproduction through competitive interactions. J. Evol. Biol. 1, 95–116. Koenig, W. D., and Mumme, R. L. (1987). ‘‘Cooperatively Breeding Acorn Woodpecker.’’ Princeton University Press, Princeton. Kondrashov, F. A., Rogozin, I. B., Wolf, Y. I., and Koonin, E. V. (2002). Selection in the evolution of gene duplications. Genome Biol. 3, RESEARCH0008. Lander, E. S., Linton, L. M., Birren, B., Nusbaum, C., Zody, M. C., Baldwin, J., Devon, K., Dewar, K., Doyle, M., FitzHugh, W., Funke, R., Gage, D., et al. (2001). Initial sequencing and analysis of the human genome. Nature 409, 860–921. Latter, B. D. (1998). Mutant alleles of small effect are primarily responsible for the loss of fitness with slow inbreeding in Drosophila melanogaster. Genetics 148, 1143–1158. Latter, B. D., and Sved, J. A. (1994). A reevaluation of data from competitive tests shows high levels of heterosis in Drosophila melanogaster. Genetics 137, 509–511. Leiter, E. H. (2002). Mice with targeted gene disruptions or gene insertions for diabetes research: Problems, pitfalls, and potential solutions. Diabetologia 45, 296–308. Lenington, S., Franks, P., and Williams, J. (1988). Distribution of t‐haplotypes in natural populations of wild house mice. J. Mammal. 69, 489–499. Lenington, S., Coopersmith, C. B., and Erhart, M. (1994). Female preference and variability among t‐haplotypes in wild house mice. Am. Nat. 143, 766–784. Levin, B. R., Petras, M. L., and Rasmussen, D. I. (1969). The effect of migration on the maintenance of a lethal polymorphism in the house mouse. Am. Nat. 103, 647–661. Levine, L., Rockwell, R. F., and Grossfield, J. (1980). Sexual selection in mice. V. Reproductive competition between þ/þ and þ/tw5 males. Am. Nat. 116, 150–156. Lewontin, R. C. (1968). The effect of differential viability on the population dynamics of t alleles in the house mouse. Evolution 22, 262–273. Lynch, C. B. (1977). Inbreeding effects upon animals derived from a wild population of Mus musculus. Evolution 31, 526–537. Lynch, M., and Conery, J. S. (2000). The evolutionary fate and consequences of duplicate genes. Science 290, 1151–1155. Maeda, I., Kohara, Y., Yamamoto, M., and Sugimoto, A. (2001). Large‐scale analysis of gene function in Caenorhabditis elegans by high‐throughput RNAi. Curr. Biol. 11, 171–176. Manning, C. J., Potts, W. K., Wakeland, E. K., and Dewsbury, D. A. (1992a). What’s wrong with MHC mate choice experiments? In ‘‘Chemical Signals in Vertebrates’’ (R. L. Doty and D. Mu¨ller‐Schwarze, Eds.), pp. 229–235. Plenum Press, New York, NY. Manning, C. J., Wakeland, E. K., and Potts, W. K. (1992b). Communal nesting patterns in mice implicate MHC genes in kin recognition. Nature 360, 581–583.

FITNESS AND GENE FUNCTION

213

Manning, C. J., Dewsbury, D. A., Wakeland, E. K., and Potts, W. K. (1995). Communal nesting and communal nursing in house mice, Mus musculus domesticus. Anim. Behav. 50, 741–751. McClelland, E. E., Penn, D. J., and Potts, W. K. (2003). Major histocompatibility complex heterozygote superiority during coinfection. Infect. Immun. 71, 2079–2086. McClelland, E. E., Damjanovich, K., Gardner, K., Groesbeck, Z. J., Ma, M. S., Nibley, M., Richardson, K. S., Wilkinson, M., Morrison, L. C., Bernhardt, P., and Potts, W. K. (2004). Infection‐dependent phenotypes in MHC‐congenic mice are not due to MHC: Can we trust congenic animals? BMC Immunol. 5, 14. McIlwain, K. L., Merriweather, M. Y., Yuva‐Paylor, L. A., and Paylor, R. (2001). The use of behavioral test batteries: Effects of training history. Physiol. Behav. 73, 705–717. Meagher, S., Penn, D. J., and Potts, W. K. (2000). Male‐male competition magnifies inbreeding depression in wild house mice. Proc. Natl. Acad. Sci. USA 97, 3324–3329. Miller, P. S. (1994). Is inbreeding depression more severe in a stressful environment? Zoo Biol. 13, 195–208. Myers, J. H. (1973). The absence of t alleles in feral populations of house mice. Evolution 27, 702–704. Nadeau, J. H. (2001). Modifier genes in mice and humans. Nat. Rev. Genet. 2, 165–174. Nadeau, J. H. (2003). Modifier genes and protective alleles in humans and mice. Curr. Opin. Genet. Dev. 13, 290–295. Nicholson, S. M., Peterson, J. D., Miller, S. D., Wang, K., Dal Canto, M. C., and Melvold, R. W. (1994). BALB/c substrain differences in susceptibility to Theiler’s murine encephalomyelitis virus‐induced demyelinating disease. J. Neuroimmunol. 52, 19–24. Packer, C., Hilborn, R., Mosser, A., Kissui, B., Borner, M., Hopcraft, G., Wilmshurst, J., Mduma, S., and Sinclair, A. R. (2005). Ecological change, group territoriality, and population dynamics in Serengeti lions. Science 307, 390–393. Penn, D., and Potts, W. (1999). The evolution of mating preferences and major histocompatibility genes. Am. Nat. 153, 145–164. Penn, D. J., Damjanovich, K., and Potts, W. K. (2002). MHC heterozygosity confers a selective advantage against multiple‐strain infections. Proc. Natl. Acad. Sci. USA 99, 11260–11264. Petras, M., and Topping, J. C. (1983). The maintenance of polymorphism at two loci in house mouse (Mus musculus) populations. Can. J. Genet. Cytol. 25, 190–201. Ponte, E., Bracco, E., Faix, J., and Bozzaro, S. (1998). Detection of subtle phenotypes: The case of the cell adhesion molecule csA in Dictyostelium. Proc. Natl. Acad. Sci. USA 95, 9360–9365. Potts, W. K., Manning, C. J., and Wakeland, E. K. (1991). Mating patterns in seminatural populations of mice influenced by MHC genotype. Nature 352, 619–621. Potts, W. K., Manning, C. J., and Wakeland, E. K. (1992). MHC‐based mating preferences in Mus operate through both settlement patterns and female controlled extra‐territorial matings. In ‘‘Chemical Signals in Vertebrates VI’’ (R. L. Doty, Ed.). Plenum Press, New York. Queller, D. C., Ponte, E., Bozzaro, S., and Strassmann, J. E. (2003). Single‐gene greenbeard effects in the social amoeba Dictyostelium discoideum. Science 299, 105–106. Ravi Ram, K., Ji, S., and Wolfner, M. F. (2005). Fates and targets of male accessory gland proteins in mated female Drosophila melanogaster. Insect Biochem. Mol. Biol. 35, 1059–1071. Rice, W. R. (1996). Sexually antagonistic male adaptation triggered by experimental arrest of female evolution. Nature 381, 232–234.

214

LARA S. CARROLL AND WAYNE K. POTTS

Rice, W. R., and Holland, B. (2005). Experimentally enforced monogamy: Inadvertent selection, inbreeding, or evidence for sexually antagonistic coevolution? Evolution 59, 682–685. Rice, W. R., Linder, J. E., Friberg, U., Lew, T. A., Morrow, E. H., and Stewart, A. D. (2005). Inter‐locus antagonistic coevolution as an engine of speciation: Assessment with hemiclonal analysis. Proc. Natl. Acad. Sci. USA 102(Suppl. 1), 6527–6534. Robinson, G. E. (1999). Integrative animal behaviour and sociogenomics. Trends Ecol. Evol. 14, 202–205. Robinson, G. E., Grozinger, C. M., and Whitfield, C. W. (2005). Sociogenomics: Social life in molecular terms. Nat. Rev. Genet. 6, 257–270. Rubin, G. M., and Lewis, E. B. (2000). A brief history of Drosophila’s contributions to genome research. Science 287, 2216–2218. Shabalina, S. A., Yampolsky, L., and Kondrashov, A. S. (1997). Rapid decline of fitness in panmictic populations of Drosophila melanogaster maintained under relaxed natural selection. Proc. Natl. Acad. Sci. USA 94, 13034–13039. Shastry, B. S. (1995). Genetic knockouts in mice: An update. Experientia 51, 1028–1039. Silva, A. J. (1997). Mutant mice and neuroscience: Recommendations concerning genetic background. Banbury Conference on genetic background in mice. Neuron 19, 755–759. Simpson, E. M., Linder, C. C., Sargent, E. E., Davisson, M. T., Mobraaten, L. E., and Sharp, J. J. (1997). Genetic variation among 129 substrains and its importance for targeted mutagenesis in mice. Nat. Genet. 16, 19–27. Sluyter, F., Marican, C. C., and Crusio, W. E. (1999). Further phenotypical characterisation of two substrains of C57BL/6J inbred mice differing by a spontaneous single‐gene mutation. Behav. Brain Res. 98, 39–43. Smith, V., Botstein, D., and Brown, P. O. (1995). Genetic footprinting: A genomic strategy for determining a gene’s function given its sequence. Proc. Natl. Acad. Sci. USA 92, 6479–6483. Smith, V. H., and Holt, R. D. (1996). Resource competition and within‐host disease dynamics. Trends Ecol. Evol. 11, 386–388. Spring, J. (1997). Vertebrate evolution by interspecific hybridisation—are we polyploid? FEBS Lett. 400, 2–8. Strunk, K. E., Amann, V., and Threadgill, D. W. (2004). Phenotypic variation resulting from a deficiency of epidermal growth factor receptor in mice is caused by extensive genetic heterogeneity that can be genetically and molecularly partitioned. Genetics 167, 1821–1832. Sturdy, C. B. (2004). Animal behaviour: A social call. Nature 430, 414. Sugimoto, A. (2004). High‐throughput RNAi in Caenorhabditis elegans: Genome‐wide screens and functional genomics. Differentiation 72, 81–91. Tautz, D. (2000). A genetic uncertainty problem. Trends Genet. 16, 475–477. Taylor, J. S., Van de Peer, Y., Braasch, I., and Meyer, A. (2001). Comparative genomics provides evidence for an ancient genome duplication event in fish. Phil. Trans. Roy. Soc. Lond., B, Biol. Sci. 356, 1661–1679. Tecott, L. H., and Nestler, E. J. (2004). Neurobehavioral assessment in the information age. J. Neuroimmunol. 7, 462–466. Thatcher, J. W., Shaw, J. M., and Dickinson, W. J. (1998). Marginal fitness contributions of nonessential genes in yeast. Proc. Natl. Acad. Sci. USA 95, 253–257. Threadgill, D. W., Dlugosz, A. A., Hansen, L. A., Tennenbaum, T., Lichti, U., Yee, D., LaMantia, C., Mourton, T., Herrup, K., Harris, R. C., Barnard, J. A., Yuspa, S. H., et al. (1995). Targeted disruption of mouse EGF receptor: Effect of genetic background on mutant phenotype. Science 269, 230–234.

FITNESS AND GENE FUNCTION

215

Threadgill, D. W., Yee, D., Matin, A., Nadeau, J. H., and Magnuson, T. (1997). Genealogy of the 129 inbred strains: 129/SvJ is a contaminated inbred strain. Mamm. Genome 8, 390–393. Thursz, M. R., Thomas, H. C., Greenwood, B. M., and Hill, A. V. (1997). Heterozygote advantage for HLA class‐II type in hepatitis B virus infection. Nat. Genet. 17, 11–12. Vignal, C., Mathevon, N., and Mottin, S. (2004). Audience drives male songbird response to partner’s voice. Nature 430, 448–451. Wegner, K. M., Kalbe, M., Schaschl, H., and Reusch, T. B. (2004). Parasites and individual major histocompatibility complex diversity—an optimal choice? Microbes Infect. 6, 1110–1116. Weinig, C., Ungerer, M. C., Dorn, L. A., Kane, N. C., Toyonaga, Y., Halldorsdottir, S. S., Mackay, T. F., Purugganan, M. D., and Schmitt, J. (2002). Novel loci control variation in reproductive timing in Arabidopsis thaliana in natural environments. Genetics 162, 1875–1884. Weinig, C., Dorn, L. A., Kane, N. C., German, Z. M., Halldorsdottir, S. S., Ungerer, M. C., Toyonaga, Y., Mackay, T. F., Purugganan, M. D., and Schmitt, J. (2003). Heterogeneous selection at specific loci in natural environments in Arabidopsis thaliana. Genetics 165, 321–329. Weiss, S., Uematsu, Y., D’Hoostelaere, L., and Alanen, A. (1989). Gross genetic differences among substrains of NZB mice. J. Immunogenet. 16, 217–221. Wellik, D. M., and Capecchi, M. R. (2003). Hox10 and Hox11 genes are required to globally pattern the mammalian skeleton. Science 301, 363–367. West Eberhard, M. J. (1983). Sexual selection, social competition and speciation. Q. Rev. Biol. 58, 155–183. Wolfe, K. H., and Shields, D. C. (1997). Molecular evidence for an ancient duplication of the entire yeast genome. Nature 387, 708–713. Wolfenden, G. E., and Fitzpatrick, J. W. (1996). ‘‘The Florida Scrub Jay: Demography of a Cooperative Breeding Bird.’’ Princeton University Press, Princeton. Wolfer, D. P., Crusio, W. E., and Lipp, H. P. (2002). Knockout mice: Simple solutions to the problems of genetic background and flanking genes. Trends Neurosci. 25, 336–340. Yamazaki, K., Boyse, E. A., Mike, V., Thaler, H. T., Mathieson, B. J., Abbott, J., Boyse, J., and Zayas, Z. A. (1976). Control of mating preferences in mice by genes in the major histocompatibility complex. J. Exp. Med. 144, 1324–1335.

This page intentionally left blank

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 36

Signal Detection and Animal Communication R. Haven Wiley department of biology, university of north carolina chapel hill, north carolina 27599, usa

I. INTRODUCTION Although communication consists of associations between signals from one individual and responses by another, in reality these associations are often weak. In recent decades there has been a tendency to explain these weak associations as the result of attempts by signalers to manipulate or exploit receivers and of receivers to resist these attempts. This chapter takes a different approach, although the underlying question remains the same—how can signalers and receivers optimize their behavior? The present approach develops an earlier suggestion that it is the inevitability of errors by receivers that limits optimal behavior by both parties in communication (Wiley, 1994). Signal detection theory provides the basic theory for this approach. The previous applications of this theory, however, have been in psychophysics. To justify its application to the evolution of communication is the purpose of the present chapter. The problems of signal detection arise especially for signals in their natural contexts. The properties of signals perceived by a receiver inevitably differ from those emitted by the signaler. For instance, acoustic signals like bird songs are altered by attenuation and degradation during propagation through the environment (Naguib, 2003; Naguib and Wiley, 2001; Wiley, 1991; Wiley and Richards, 1982). Although in any one situation, on average, some features of attenuation and degradation are predictable, much remains unpredictable in detail. Furthermore, a receiver perceives this attenuated and degraded signal against a background of irrelevant energy that shares some features with the signal. An acoustic signal, for instance, is often perceived against a background of sounds with more or less similar frequencies, intervals, or other patterns. These sounds come from nearby individuals of the same or different species and from physical features of the environment such as 0065-3454/06 $35.00 DOI: 10.1016/S0065-3454(06)36005-6

217

Copyright 2006, Elsevier Inc. All rights reserved.

218

R. HAVEN WILEY

wind and water. Finally, any receiver’s sensory, associative, and motor neurons always include some unpredictability. As a result of all of these processes, it is not surprising that signals usually have weak associations with responses. Sometimes when a stimulus occurs, the intended receiver fails to respond; sometimes the receiver responds when there is no stimulus. In the first case, the receiver seems to mistake a stimulus for the background; in the latter it seems to mistake the background for a stimulus. Because of the pervasiveness of these mistakes, receivers fail to achieve maximal performance and signals fail to reach maximal efficiency. At first sight, these weak associations of signals and responses seem to be just noise in the system without fundamental implications for communication. This chapter, however, develops the view that these mistakes are a result of inescapable constraints on the performance of receivers and that these constraints in turn influence the evolution of both producing and responding to signals (Wiley, 1994). Many current issues in the study of communication, such as honesty and exploitation and the multiplicity and exaggeration of signals, become clearer once we understand the constraints on the performance of receivers. These constraints on receivers are addressed by signal detection theory (Green and Swets, 1966; Macmillan and Creelman, 1991; McNicol, 1972). Developed originally by electronics engineers, in recent decades this theory has provided a rationale for the psychophysical study of sensory thresholds and perception. Despite its success in these studies, its application to the evolution of communication is still rudimentary. The initial sections of this chapter provide an introduction to signal detection theory and its applications in psychophysics. The objective of these sections is to identify general principles for the study of adaptations in animal communication. These principles can clarify the properties of signals that affect a receiver’s performance. They also suggest ways to extend the theory to the classification as well as detection of signals. These steps lead to hypotheses about the evolution of both signaling and receiving. In particular, signal detection theory leads to natural explanations for the evolution of deception and exaggeration in communication. The final sections take up the design and interpretation of experiments for studying communication in natural situations. The objective of these sections is to suggest practical ways to study the performance of receivers under conditions like those in which communication evolved.

II. ESSENTIAL FEATURES OF SIGNAL DETECTION To apprehend the essential features of signal detection theory, it helps to consider a simple situation. Suppose an individual listens for a conspecific vocalization characterized by some feature such as a particular frequency.

SIGNAL DETECTION AND COMMUNICATION

219

In this case, the signal has a single feature, a particular frequency, which varies along a single dimension, its intensity. Even in this simple case, a receiver in natural situations faces a formidable problem. By the time the signal reaches the receiver, its intensity varies irregularly, as a result of variable attenuation and degradation of the signal during propagation. With some ingenuity and proper instruments, we can measure the intensity of the arriving signal in appropriate time intervals (for instance, the temporal resolution of the receiver’s hearing). From this information, we can determine the probabilities of different intensities of the characteristic frequency as the signal reaches the receiver. These probabilities constitute the probability density function (PDF) for the intensity of that frequency during a signal. At the same time, the receiver usually experiences background stimulation that can also include this characteristic frequency. For instance, this frequency might occur in other species’ or individuals’ vocalizations or in other environmental sources of sound, all irrelevant to the listener. Again, with some care we can determine the PDF for the intensity of this frequency in the background stimulation reaching the receiver. If the distributions of intensities during the signal and background stimulation overlap, then the receiver (a listener) cannot avoid mistakes. Errors are inevitable whenever a receiver cannot completely separate signal and background. Only an observer with independent access to the source of the signal and the background can measure their properties separately. A receiver has no independent access to the signal. It must instead decide whether or not a particular intensity of the characteristic frequency merits response or not. Past experience with different intensities might lead to different expectations for the frequency of the signal and thus different levels of confidence in its decision to respond or not, but an isolated perception itself provides no basis for certainty. A simple graph can introduce the issues that arise in this situation. We can plot the overlapping PDFs for intensity during the signal and background stimulation along the same axis (Fig. 1). The subject’s criterion for a decision is then represented by a threshold for response. In Fig. 1, the PDFs are represented by normal distributions, with equal variances but different means. This simplified situation applies when background stimulation has a normal distribution of intensities and the signal has a fixed intensity, which is added to the background. Complications are addressed later, but they do not change the basic issues. Once a threshold for response is chosen, then the total probability of a correct response (responding when a signal has occurred) is the integral of the PDF for signals from the threshold to infinity. The probability of a missed detection (failing to respond when a signal has occurred) is the integral of the same PDF from the threshold to negative infinity.

220

R. HAVEN WILEY

Fig. 1. The basic situation described by signal detection theory. (A) The levels of background stimulation with and without a signal are represented by the outputs from a perceptual channel. The probability of an output as a function of the level of the output is a probability density function, PDF, for the output. A decision to respond involves selecting a criterion (in this case, a threshold in the output of the channel above which a response occurs). (B) Any such threshold results in a probability of correct detections, PCD, the area under the PDF for background plus signal to the right of the threshold. (C) Any threshold also results in a probability of false alarms, PFA, the corresponding area under the PDF for background alone.

Thus when a signal occurs, the probability of correct detection by the receiver equals one minus the probability of missed detection, PCD ¼ 1
PMD. Similarly, we can find the probabilities of false alarm (responding to background stimulation), PFA, and correct rejection (no response to background stimulation), PCR, from integrals of the PDF for background stimulation. When only background stimulation occurs, PFA ¼ 1
PCR.

221

SIGNAL DETECTION AND COMMUNICATION

The first essential feature of signal detection is now apparent. As a receiver changes its threshold for response, the PFA varies with the probability of a correct detection (PCD). By shifting the threshold for response to the right (toward higher intensities of the characteristic frequency), for example, a receiver can reduce its false alarms (responding when there is no signal present). Simultaneously, however, it increases its missed detections (not responding when a signal occurs). Clearly a receiver in this situation cannot simultaneously both minimize PFA and maximize PCD. This trade‐off between correct detections and false alarms has fundamental implications for the evolution of communication (Wiley, 1994). Another essential feature of signal detection is a distinction between the receiver’s criterion for a response and the detectability of the signal. In this simple case, the receiver’s criterion is represented by a threshold for response; the detectability of the signal is represented by the separation of the PDFs for signal alone and signal plus background (the difference between the means in relation to the standard deviation). A receiver’s performance is determined by both of these variables. When we present signals to animals, such as recordings of calls or songs, we often want to determine the subjects’ attitude or responsiveness toward the signal. It is thus the subjects’ criterion that interests us. In other cases, such as determination of sensory thresholds, it is the detectability of the signals that interests us. Signal detection theory allows us to separate the criterion for response from the detectability of signals. To see how, we can turn to a well‐established application of this theory.

III. APPLICATION OF SIGNAL DETECTION THEORY EXPERIMENTAL PSYCHOPHYSICS

IN

The earliest application of signal detection theory to a behavioral problem was the determination of human sensory thresholds. Signal detection theory solved the problem of measuring the detectability of a signal despite differences in subjects’ thresholds for responses. Procedures for this purpose are now well established (Green and Swets, 1966; Macmillan and Creelman, 1991; McNicol, 1972). Before the application of signal detection theory, psychophysicists determined the absolute threshold for hearing sounds of a particular frequency by asking subjects to respond to faint sounds, barely separable from the background. The activity of auditory neurons in response to these sounds would barely differ from their spontaneous activity. These experiments confronted an insurmountable problem, because there was no satisfactory

222

R. HAVEN WILEY

way to standardize the criteria different subjects used for responding, in other words, their thresholds. Signal detection theory provides a solution to this problem by a simple modification of the experimental procedure. Subjects listen for a tone during brief intervals indicated by a cuing stimulus, for instance illumination of a light. During half of these intervals, selected at random, there occurs a tone of a particular frequency and intensity; during the remaining intervals there is no tone. The intervals with a tone allow an estimate of PCD; those with no tone allow an estimate of PFA. If the tone is loud enough, subjects detect the tone with high efficiency (high PCD and low PFA). If the tone is faint, this efficiency drops. The subject’s performance in this situation depends on both the detectability of the stimulus and the subject’s criterion for response (in this case, a threshold). The literature in psychophysics often refers to a subject’s criterion as a bias. For any constant level of detectability (the distance between the means of the two PDFs relative to the standard deviation), as a subject’s threshold increases, PCD increases as a function of PFA. This function, called the receiver operating characteristic (ROC), increases monotonically from (0,0) to (1,1) in the unit square (Fig. 2). As an exercise, try generating Fig. 2 from Fig. 1, by varying the threshold for response. To obtain an ROC, we must measure PCD and PFA at different thresholds for response. Psychophysicists use two basic methods. One involves direct manipulation of the subjects’ thresholds, by rewards or instructions that place different weights on correct detections and false alarms. Another method involves asking subjects to rate their certainty for each response (for instance, 0 ¼ absolutely certain no signal occurred, 10 ¼ absolutely certain a signal occurred) (Egan et al., 1959; Macmillan and Creelman, 1991; McNicol, 1972). In the latter case, the experimenter uses different levels of certainty for different thresholds of response. For instance, for a high threshold, take all responses with certainty greater than 9 as positive responses for determining both PCD and PFA. For a lower threshold, take all responses with certainty greater than 8, and so forth. Accuracy in estimating PCD and PFA at each threshold requires repeated tests for each subject. The ROC then allows us to determine the detectability of a signal in a way that is independent of subjects’ thresholds for response. As the detectability of a signal increases (the PDFs for signal alone and signal plus background move apart), the ROC moves away from the positive diagonal toward the upper left corner of the unit square, the point where performance is ideal (PCD ¼ 1, PFA ¼ 0). The closer the ROC approaches the upper left corner, the greater the detectability of the signal. As the subject’s threshold changes, on the other hand, its performance moves one way or the other along the ROC. As its threshold increases,

SIGNAL DETECTION AND COMMUNICATION

223

Fig. 2. A receiver operating characteristic (ROC) results from plotting PCD as a function of PFA, as the threshold for response varies. The ROC is symmetrical about the negative diagonal of the unit square provided the two PDFs have normal distributions and equal variances. The separation of the means of the PDFs determines how far the ROC lies from the positive diagonal and thus how nearly it approaches the point of ideal performance, the upper left‐ hand corner. This illustration shows the ROC when the means are separated by one standard deviation (d0 ¼ 1).

a subject’s performance approaches the origin (PFA ¼ 0, PCD ¼ 0). As its threshold decreases, its performance approaches the upper right corner (PCD ¼ 1, PFA ¼ 1). Thus changes in detectability of a signal shift the ROC away from or toward the diagonal, while changes in the subject’s threshold shift its performance upward or downward along the ROC. Some study of Figs. 1 and 2 can clarify these relationships between the detectability of a signal, the threshold for response, and a subject’s performance (its PCD and PFA). Measurement of detectability is straightforward when the PDFs for background alone and for signal plus background are normally distributed with equal variance. The ROC in this case is symmetrical about the negative diagonal. If we plot the normal deviates or z‐transforms of PCD and PFA, then the ROC is a straight line with unit slope (Fig. 3), and the difference in z‐scores, z(PFA)
z(PCD), is the same for all points on this line. This difference, usually represented by d0 , represents the detectability of the signal. It equals the separation of the PDFs for background alone

224

R. HAVEN WILEY

Fig. 3. An ROC plotted on probability (z‐transformed) axes is a straight line with slope ¼ 1 in the case of normally distributed PDFs with equal variance. This illustration shows the same ROC as Fig. 2.

and signal plus background divided by their standard deviation. For alternative measures of detectability, all highly correlated with d0 , see discussions by Green and Swets (1966), McNicol (1972), or Macmillan and Creelman (1991). Detectability is a measure of a receiver’s ability to separate a signal from background stimulation; the analogous measure of ability to separate two signals is discriminability. The methods just described for measurement of the detectability of a signal also permit measurement of the discriminability of two signals. Instead of comparing responses to a signal and background stimulation, we compare responses to two signals in the presence of constant background stimulation.

IV. GENERAL ASSUMPTIONS OF SIGNAL DETECTION THEORY The theory of signal detection derives from assumptions about the nature of signals and their processing by receivers. This section considers these assumptions in order to establish the wide application of this theory.

SIGNAL DETECTION AND COMMUNICATION

225

General assumptions are separated from some specific ones so that we do not discard the general theory entirely on the basis of questions about specifics. This section addresses general assumptions; the next considers specifics. The theory accommodates a broad definition of a signal. Elsewhere, I have proposed that a signal is any pattern of energy produced by one individual (the signaler) and evoking a response from another individual (the receiver) without providing all of the power necessary to effect the response (Wiley, 1994). Some power is necessary to produce an alteration in the receiver’s sensors, but the receiver itself provides essential power for the response. It is the necessary role of the receiver in producing a response that creates the essential elements of signal detection and, ultimately, all communication. Although a signal is similar to any stimulus that evokes a response, the term ‘‘signal’’ serves to emphasize the crucial importance of the limited contribution of power for the response. The restriction of the sources and receivers of signals to living individuals (or their components) serves to include just those cases in which signalers and receivers might coevolve. This restriction is not essential, however, as signal detection theory addresses the optimization of a receiver’s performance regardless of the source of the signals. Nevertheless, when both source and receiver are living organisms or their components, the possibility of coevolution raises particularly interesting issues, a topic we discuss later. Signal detection theory also accommodates a broad scope for receivers. The two essential components of a receiver are a sensor and a mechanism for decisions. Each sensor is a perceptual channel tuned to a particular feature or dimension of stimulation (such as a particular band of frequencies of sound, a particular direction of a visual object, or a particular spectrotemporal pattern of sound). A decision to respond then depends on the output from one or more of these perceptual channels (Fig. 4). Any channel is specified by its characteristic feature (for instance, the frequency of sound for maximal response from an auditory neuron) and its selectivity (often presented as its tuning curve or pass band). Each channel produces an output that depends on the energy in its pass band within the broader range of energy impinging on the organism. This stimulation can include background energy of no interest to the organism (including irrelevant signals produced by other species or individuals and energy from the physical environment). The physiological mechanisms of channels often also produce spontaneous output. Consequently, a decision to respond based on the output of a channel often includes the possibility of false alarms and missed detections. This model has broad generality (Green and Swets, 1966, Chapter 1). It is perhaps the most general model for an organism’s responses to

226

R. HAVEN WILEY

Fig. 4. The general model for signal detection involves perceptual channels that analyze features or patterns in stimulation impinging on the receiver. The output of one or more channels forms the basis for a decision to respond (in the form of a multidimensional criterion for response). Channels and decisions might represent distinct neurons or populations of neurons, or a single neuron might combine these two properties.

stimulation: a decision to respond or not depends on the output of a channel that receives combined signal and background. Green and Swets (1966) showed that the best rule for a decision to respond is a likelihood ratio that takes into account the expected frequencies of occurrence of signals. These basic ideas have a long history in psychology (Broadbent, 1958) and are familiar to ethologists and neuroethologists studying releasing mechanisms, stimulus filtering, and feature detectors. The literature of psychophysics often contrasts ‘‘signal detection theory’’ with ‘‘threshold theory’’ (Green and Swets, 1966; Luce, 1963; Luce and Green, 1974; Macmillan and Creelman, 1991; Yonelinas, 2002). The distinction, however, is not fundamental. Threshold theory assumes some threshold above which a signal is always detected without error. Below this threshold, signals are detected with some fixed PFA and PCD (or some fixed ratio of these values). This theory thus requires at least two channels for the analysis of any one feature of a signal, one error‐free for signals above the threshold, the other error‐prone for signals below the threshold. These two channels, however, are equivalent to a single channel without normally distributed PDFs for background and for signal plus background (in this case the PDFs are rectangular; for full discussion, see Green and Swets, 1966; McNicol, 1972; Macmillan and Creelman, 1991). Only if we restrict the term ‘‘signal detection theory’’ to normally distributed PDFs with equal variance, are we forced to draw a distinction between this theory and ‘‘threshold theory.’’ If we relax these restrictions, threshold theory

SIGNAL DETECTION AND COMMUNICATION

227

becomes a special case of a general signal detection theory, based on a model of signal detection without restrictions on the distributions of outputs from perceptual channels. Debate about these alternatives complicates much of the psychological literature on signal detection. In many cases, signal detection theory with additional assumptions of normality and equal variance can explain the properties of experimentally determined ROCs. The assumptions of normality and equal variance are best approached by examining the procedures for measuring detectability.

V. SPECIFIC ASSUMPTIONS OF SIGNAL DETECTION THEORY: MEASURING DETECTABILITY Signal detection theory, as applied routinely in psychophysical determinations of sensory thresholds, involves calculation of d0 from measurements of PCD as a function of PFA. As shown in Section III earlier, this calculation is made simple by assuming normal PDFs with equal variances. In this special case, a single pair of measurements of PCD and PFA determines the ROC and thus d0 , as calculated from the standardized deviates, or z‐scores, of PCD and PFA. Calculation of d0 from a single pair of measurements and determinations of absolute sensory thresholds requires some specific conditions: (1) normally distributed PDFs with equal variance; (2) fixed criteria for responses; and (3) cuing of responses. This section considers each of these requirements. Although each is critical in special cases, none is necessary for measurements of detectability in general. A. NORMAL DISTRIBUTIONS

WITH

EQUAL VARIANCE

For sensory discriminations under laboratory conditions, the relevant PDFs are often nearly normal with nearly equal variances. The clearest evidence is an ROC symmetrical around the negative diagonal in the unit square and linear in probability space (with z‐transformed axes for PCD and PFA) with slope equal to 1 (Green and Swets, 1966; McNicol, 1972). In this case, d0 ¼ z(PFA) – z(PCD) provides an unambiguous measure of detectability. If the PDFs are not normally distributed or have unequal variances, then the picture changes. If variances are not equal, the ROC lacks symmetry around the negative diagonal. When plotted in z‐transform space, the ROC has a slope equal to the ratio of variances. If the PDFs are not normally distributed, the ROC changes shape and is no longer linear in z‐transform space.

228

R. HAVEN WILEY

Consequently, when either normality or equal variance is violated, d0 ¼ z(PFA)
z(PCD) makes little sense as a measure of detectability. When normality or equal variance does not apply, we must use an alternative measure of detectability. A simple one is the area between the ROC and the positive diagonal of the unit square. This area measures the displacement of the ROC away from the positive diagonal and toward the point of maximal performance at the upper left‐hand corner; d0 provides a measure of this displacement only for a symmetrical ROC. B. OPTIMAL CRITERIA Accurate measurement of absolute sensory thresholds requires that subjects use an optimal criterion or rating scale for any set of experimental conditions. Variation among subjects, or variation among trials for any one subject, results in an underestimate of d0 for maximal performance and also an underestimate of any difference in variances between signal and background. In carefully conducted psychophysical experiments, these possible errors turn out to be slight (Macmillan and Kaplan, 1985; McNicol, 1972, pp. 202–204). This assumption that subjects use an optimal criterion is less critical for an investigation of communication, when an organism’s actual performance has greater interest than its maximally possible performance. In this case, we can combine observations from different subjects by averaging z‐scores to obtain a composite value of d0 (Macmillan and Kaplan, 1985; McNicol, 1972, p. 112). If subjects’ criteria or ratings vary, these composite measurements of detectability reflect expected average performance. Alternatively, we could study each individual’s ability to detect or to discriminate signals. C. CUING

OF

RESPONSES

Any measurement of the detectability of a stimulus requires null (background only) presentations, which permit measurement of PFA, the probability of response without the signal present. In laboratory experiments, a cuing stimulus identifies intervals in which the subject must make a decision. This procedure assures equal decisions with and without the signal present. In field experiments this device is not possible. However, we can still include null presentations with no stimulus; even better, white noise or prerecorded background sounds might serve as a null stimulus. Alternatively, one could abandon attempts to measure the absolute detectability of any one stimulus and consider only the discriminability of two signals. In this case, a balanced experimental design could include equal numbers of presentations of the two signals.

SIGNAL DETECTION AND COMMUNICATION

229

The absence of null presentations confounds interpretation of a large body of research on human vigilance (Davies and Parasuraman, 1982; Mackie, 1977). Studies of vigilance and field studies of responses to playback have some similarities. In both cases, subjects experience long intervals between infrequent occurrences of a stimulus. The long periods without signals inevitably make PFA very small during any brief interval when the signal is absent. Consistently small PFA makes a meaningful ROC difficult to construct. Despite some suggestions for ways to circumvent this problem (Egan et al., 1961a; Watson and Nichols, 1976), there seems to be no convincingly satisfactory solution. When we cannot measure false alarms, by means of cuing, null presentations, or comparisons of two signals, determination of an ROC is problematic. Measurement of PFA is essential for a full understanding of a receiver’s performance. A later section discusses some practical possibilities for solving this problem in field studies of animal communication by means of playbacks. The two general results of signal detection theory—the interdependence of PCD and PFA and the distinction between the receiver’s criterion and the detectability of the signal—do not depend on the specific assumptions of normality and equal variance and are not affected by the practical difficulties of measuring detectability or discriminability. These two general features of signal detection are alone sufficient to clarify the determinants of a receiver’s performance.

VI. PROPERTIES OF SIGNALS THAT AFFECT

A

RECEIVER’S PERFORMANCE

Signal detection theory makes it clear that any receiver’s performance in detecting or discriminating signals has limits. Furthermore, these limits are in part determined by properties of the signals. Predictions about these determinants of a receiver’s performance have in many cases been repeatedly confirmed by psychophysical studies of humans, but the results of these studies have broad application to signal detection in general and thus to all forms of communication. Consider three properties of a signal that influence a receiver’s performance: (1) contrast, (2) redundancy, and (3) uncertainty. We shall see that the inevitable effects of these three properties of signals explain a lot of ‘‘receiver psychology.’’ A. CONTRAST Contrast and detectability are so closely related that it requires care to distinguish them carefully. As explained earlier, detectability is the difference between the means, in relation to the standard deviations, of background

230

R. HAVEN WILEY

alone and background plus signal in the output of some perceptual channel (for instance, in the responses of an experimental subject). Contrast is an analogous difference in the stimulation at the input to a channel (in the stimulation impinging on the subject). Unlike detectability, contrast depends only on the properties of the external stimulation reaching an organism and not on the properties of the organism’s perceptual channels. Contrast usually increases detectability. The influence of contrast on a subject’s performance is so clear that it has received little explicit study by psychophysicists. One such study, included in one of the first applications of signal detection theory to perception, showed that log d0 increased linearly with log intensity for a signal in the presence of constant background stimulation (Tanner and Swets, 1954). Because we define contrast by the properties of a signal in relation to the background stimulation impinging on an organism, detectability of the signal depends on both its contrast and the selectivity of the perceptual channel. This dual determination of detectability is the basis for a procedure in psychophysics for determining bandwidths of sensory channels. In the case of hearing, the intensity of broad‐spectrum background sound (white noise) that can mask a signal of a particular frequency depends on the bandwidth of the auditory channel. In fact, the signal‐to‐noise ratio (a measure of contrast) for complete masking of a single frequency with broad‐spectrum noise equals the effective bandwidth of the auditory channel for that frequency. The dual determination of detectability implies that the intensities of signals and background stimulation impinging on an organism do not alone allow us to predict an organism’s performance. For instance, the intensity of a particular frequency of sound, or hue of light in a signal, and in the background are not enough to allow us to predict the detectability of that sound or light for a particular organism. To determine the influence of contrast on detectability, we must study the organism’s responses, at either the neural or behavioral levels. Study of contrast and detectability in natural situations is still rudimentary (Klump, 1996). For instance, despite many studies of sound propagation in natural environments and its influence on the evolution of bird songs (reviewed by Naguib and Wiley, 2001; Wiley, 1991), we know little about the properties of background sound in relation to acoustic signals in natural situations. Such studies of acoustic contrast would require recordings of signalers with omnidirectional microphones at typical positions for conspecific listeners. To extend these studies to detectability would require adjustments for the directionality and selectivity of the listeners’ hearing. Only one study has shown how background noise affects the detectability of acoustic signals in natural situations. Measurements of auditory

SIGNAL DETECTION AND COMMUNICATION

231

thresholds in great tits Parus major, in the absence of noise, reveal greatest sensitivity to frequencies between 2 and 4 kHz, lower than most of this species’ vocalizations. However, critical bandwidths remain nearly constant over a wide range of frequencies up to 8 kHz. Consequently, in the presence of wind in a forest, which produces noise decreasing exponentially in intensity with increasing frequency, the frequency for greatest detectability shifts to 8 kHz (Langemann et al., 1998). It is also clear that birds and mammals can increase the intensity of vocal signals in the presence of background sound, presumably to improve the contrast of their signals with the background (Brumm, 2004; Brumm and Todt, 2002; Brumm et al., 2004; Cynx et al., 1998; Leonard and Horn, 2005). Shifts in frequency to increase contrast with background noise are not so well documented. The clearest case is again the great tit, which uses higher dominant frequencies in its songs in urban environments with predominantly low‐frequency noise (Slabbekoorn and Peet, 2003). Contrast and detectability of visual signals is more complex. Unlike acoustic signals, for which the signaler generates the power to produce the signal, visual signals usually rely on reflectance or scattering of light from other sources. As Endler (1990, 1993) explains, the spectrum of light arriving at a receiver’s eyes from an object depends on the product of the irradiance spectrum, the reflectance spectrum of the object, and the transmission spectrum (the spectra of the incident, reflected, and transmitted light, Q, R, and T). The contrast between a visual signal and its background thus depends on the contrast between QRT for the signal and the background. Q, which depends on the photic properties of the environment, can vary substantially with microhabitat (Endler, 1993; Gomez and The´ry, 2004). These principles apply to male manakins, small birds that use bright colors in their plumage to produce visual displays at leks in the understory of tropical forests. Both the reflectance spectra of patches in their plumage and the placement of their leks in the forest serve to increase the contrast of their displays with the visual background (Endler and The´ry, 1996; Heindl and Winkler, 2003). Furthermore, Uy and Endler (2004) have shown that, in one species, males increase the contrast of their plumage with the background by clearing fallen leaves from their display sites. Contrast between different parts of a signal is also affected by choice of location (Endler, 1993; Heindl and Winkler, 2003), but this within‐signal contrast is a form of structural redundancy, discussed in the next Section VI.B. One consequence of the dependence of visual signals on environmental irradiance is that changes in habitats can drastically alter contrast of signals with background. A case in point are the numerous endemic species of cichlids in Lake Victoria. Many of these recently evolved species differ mainly in male coloration and mate choice by females. Increased turbidity

232

R. HAVEN WILEY

of some parts of the lake in recent decades, as a result of sedimentation from human activities, is associated with a loss of many species (Seehausen et al., 1997). Apparently, the species‐specific colorations of the males no longer contrast enough to allow females to differentiate them. Contrast applies to complex signals as well as to signals with a single characteristic feature. As with simpler signals, there has been little investigation of complex signals in the presence of background stimulation. One exception is human speech. Early experiments showed that human subjects have trouble understanding one person speaking in the presence of others, the so called ‘‘cocktail‐party problem’’ (Cherry, 1953; Cherry and Taylor, 1954). Similar tasks requiring discrimination of one conspecific’s vocalizations from those of other conspecifics in the background recur in many natural situations, for instance in choruses of frogs or insects, colonies of seabirds, and dawn choruses of birds or primates. Detection and discrimination in these situations have received little attention. One such study in a colony of king penguins (Aptenodytes patagonicus) confirmed that the presence of large numbers of conspecifics increased attenuation and degradation of the adults’ calls that allow chicks to recognize their parents (Aubin and Jouventin, 1998). The situation is particularly difficult because the noise has nearly the same spectral distribution as the signals of interest to a chick. Nevertheless, these chicks can recognize their parents’ calls even when the overall signal‐to‐noise ratio is less than 1. In such ‘‘cocktail‐party’’ situations, birds as well as humans use cues for spatial localization to increase the effective signal‐to‐noise ratio of signals in more evenly distributed noise (Cherry, 1953; Cherry and Taylor, 1954; Dooling, 1982). In this case, contrast between signals consists mostly of differences in location.

B. REDUNDANCY Redundancy results from predictable relationships between the parts of a stimulus, either in time or space. It takes two forms, both of which improve detectability of a signal. Sequential redundancy consists of fixed temporal relationships between components of a signal. Repetition of a signal, the simplest form of sequential redundancy, increases its detectability (Swets and Birdsall, 1978; Swets et al., 1959). In fact, the detectability of tones increases with the square root of the number of presentations. This result is consistent with an assumption that each instance of a stimulus is assessed independently (Swets et al., 1959). All psychophysical experiments on detectability use an alerting signal to tell the subject when to respond. An alerting signal, one with high contrast and low uncertainty, accompanying a more informative signal is a special

SIGNAL DETECTION AND COMMUNICATION

233

case of redundancy. Although many natural signals might include alerting components (Richards, 1981a), this possibility has received little attention. Simultaneous redundancy consists of fixed relationships between concurrent dimensions of a signal. Simultaneous redundancy can take the form of multiple components with fixed spatial relationships in a visual signal, multiple molecular components in an olfactory signal, or multiple components with fixed spectral relationships in an acoustic signal. Such a multidimensional stimulus is more detectable than one with a single feature. The increase in detectability with the number of features characterizing a stimulus again suggests that human observers assess each feature independently (Macmillan and Creelman, 1991; Mulligan and Shaw, 1980; Shaw, 1982). An interesting twist on redundancy involves predictable relationships within the background noise rather than within the signal of interest. If different frequencies in noise are subject to synchronized amplitude modulation (called comodulation), then it is possible to use the properties of noise in one band of frequencies to improve detection of a signal in another band. This ‘‘comodulation masking release’’ has been demonstrated in both humans and birds (Klump and Langemann, 1995; Langemann and Klump, 2001; Nieder and Klump, 2001). C. UNCERTAINTY

AND

UNFAMILIARITY

Uncertainty about signals takes two forms, each of which decreases detectability. Intrinsic uncertainty occurs when a subject lacks prior information about a signal’s features, including the interval of time and location in which it might occur. Extrinsic uncertainty occurs when a subject must respond to many different signals. A subject can have prior information about the features of each signal but still face uncertainty about which signal will occur. Multiplicity of signals reduces the detectability of each of them. Uncertainty about the features of signals reduces their detectability (Green, 1961; Pelli, 1985). Detectability also decreases when observers are uncertain about the locations or intervals of time in which signals might occur (Egan et al., 1961b; Watson and Nichols, 1976; Starr et al., 1975; Swensson and Judy, 1981). These latter situations are in fact special cases of the detection of signals with uncertain features. Uncertainty about which of several signals might occur also reduces their detectability. For instance, if human observers are asked to report any of several possible signals, the overall detectability of the signals decreases as the number of alternatives increases (Cary and Reder, 2003; Nolte and Jaarsma, 1967). Human performance in detecting multiple signals again implicates independent perceptual channels. It is as if a separate channel assesses each

234

R. HAVEN WILEY

signal’s characteristic feature, and the subject decides that a signal has occurred when the criterion in any channel is met (Cohn, 1978; Green and Birdsall, 1978). These conclusions rest on a comparison of the ROCs of subjects detecting different numbers of signals. This analysis also confirms that the reduction in detectability of signals in this situation results from the uncertainty of the task, not from any change in the observers’ criterion for response. Thus detection of signals from a repertoire of possibilities is inherently more difficult than detection of a single signal specified in advance. Unfamiliarity also makes signals more difficult to detect. For instance, the frequencies of words in common usage influence their thresholds for visual recognition (Pierce, 1963). Other studies have confirmed that high‐ frequency words are more detectable than low‐frequency ones (although memory of high‐frequency words presented previously is less accurate) (Broadbent, 1967; Glanzer and Adams, 1985; Glanzer et al., 1993; Pollack et al., 1959). Thus greater familiarity with a stimulus increases its detectability, just as greater uncertainty reduces it. Human performance during vigilance fits the same pattern. The greater the uncertainty about the features, timing, or location of possible signals, the lower the efficiency of the observer (Davies and Parasuraman, 1982; Davies and Tune, 1970; Loeb and Alluisi, 1977; Warm, 1977). Studies of vigilance have not provided definitive evidence that detectability changes, as opposed to the subject’s criterion, because such studies, as explained earlier, do not allow analysis of the ROC. Nevertheless, these results resemble those of studies with a full analysis of detectability and thus reinforce the conclusion that uncertainty about a stimulus, in any form, reduces its detectability.

VII. CLASSIFICATION

OF

SIGNALS

IN

ADDITION TO DETECTION

Although in many situations it is reasonable to assume that an animal’s task involves no more than detection of an appropriate signal, in others some classification of a stimulus is essential. Detection, for example, is involved when an individual responds to a suitable mate or to its offspring or chooses a diet based on profitability of prey. Classification, on the other hand, is required when it recognizes several social partners or chooses a diet with an optimal mixture of nutrients. The discussion so far has focused on detection of a signal in noise. This section considers the use of signal detection theory to understand a receiver’s performance when classification is as important as detection. An experiment to show detection of a signal is designed so that the subject must make a binary decision about the occurrence of the signal,

SIGNAL DETECTION AND COMMUNICATION

235

‘‘yes’’ or ‘‘no,’’ go or no‐go. An experiment to show discrimination likewise requires only a single binary decision, either ‘‘signal 1’’ or ‘‘signal 2.’’ Other situations, however, require both detection and subsequent classification of signals. Detection plus classification requires one of at least three responses (‘‘no,’’ ‘‘1,’’ or ‘‘2’’) as a result of at least two binary decisions (‘‘yes’’ or ‘‘no;’’ if ‘‘yes’’ then ‘‘1’’ or ‘‘2’’). Detection plus classification is the basis for recognition or identification, as these terms are often used. In some discussions, however, recognition means detection of multidimensional signals or detection of signals with uncertain features, situations discussed in the previous section. These cases require single binary responses to a multiplicity of possible signals. The distinguishing feature of a classification of signals, in contrast, is the multiplicity of possible responses. A few experiments confirm that classification in addition to detection is a more difficult task for receivers than detection alone. For instance, the task of identifying a stimulus as familiar or not requires less attention during previous exposures to the stimulus than does recollecting specific associations of a stimulus (Dobbins et al., 2004). Female frogs (Hyla ebraccata) detect a conspecific male’s calls in background noise from a natural chorus at signal‐to‐noise ratios above 3 dB. Yet they express a preference for those calls with lower fundamental frequencies only at signal/ratios greater than 9 dB (Wollerman and Wiley, 2002). At intermediate signal‐to‐noise ratios, females did not discriminate between otherwise preferred and nonpreferred males’ calls, even though she could detect these calls. Classification in addition to detection has surprisingly complex influences on a receiver’s performance. To analyze these complexities and to assess their influence on receivers, we first consider a basic experiment. This approach leads to more complex ones and ultimately to a theoretical justification for a general principle: a receiver’s performance in a task requiring classification is inevitably lower than in a comparable task requiring only detection. To investigate detection plus classification, an experiment might present background alone and background in combination with each of two signals. With human subjects, we can simply ask for two responses, first ‘‘yes’’ or ‘‘no’’ for the presence of a stimulus, then ‘‘1’’ or ‘‘2’’ for the class of stimulus, provided one has been detected. Because classification presupposes correct detection of signals, the probability of correct classification can never exceed the PCD. Some evidence for ‘‘subliminal’’ classification does not alter the situation significantly (Macmillan and Creelman, 1991, p. 255). One approach in a study of this sort is to calculate both an ROC and an analogous identification operating characteristic (IOC). To construct the ROC for this case, one measures PCD as the probability of a correct ‘‘yes’’

236

R. HAVEN WILEY

response when either signal occurs and PFA as the probability of a ‘‘yes’’ response when no signal occurs. For the analogous IOC, one measures PCD as the probability of correct identification of a signal when it occurs; PFA is still the probability of a ‘‘yes’’ response when no signal occurs (Benzschawel and Cohn, 1985; Green and Birdsall, 1978; Green et al., 1977; Macmillan and Creelman, 1991). The IOC, thus defined, can be derived from the ROC for detection of uncertain signals discussed earlier. Despite this theoretical advantage, the IOC fails to consider errors of classification once a signal is detected and thus provides an unrealistic measure of a receiver’s performance. A better approach in a study of detection plus classification is to consider a bivariate plot of PDFs, with one axis for a measure of the characteristic feature of each stimulus (Fig. 5). If the characteristic features of the two signals are orthogonal (in other words, if they vary independently), the PDFs for background only and for each signal in combination with background lie along two perpendicular axes. A receiver’s performance then depends on three thresholds: two that separate background from each signal in combination with background (T1 and T2) and a third that separates the two signals (T3, Fig. 5). This third threshold differentiates the two signals based on the ratio of measures of their respective characteristic features. The slope of threshold T3 changes, as the receiver alters its criterion for classifying the signals. This experiment thus allows measurement of three d0 values (Macmillan, 2002; Macmillan and Creelman, 1991; Tanner, 1956): between background (B) and background plus one of the signals (B þ S1), between B and B þ S2, and between B þ S1 and B þ S2. Suppose the receiver processes the characteristic features of the two signals independently, as predicted for orthogonal features, and the variances of the three PDFs are equal, as predicted for constant signals added to background, with equal variance in each signal’s characteristic feature. Then these three d0 values have a qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 0 02 Pythagorean relationship, d3 ¼ ðd1 þ d022 Þ as seen by geometry in Fig. 5, in which each d0 is proportional to the distance between the means of the respective PDFs. An even more robust experiment would include a fourth stimulus, background in combination with both signals at once, B þ S1 þ S2. The six d0 values in this case specify the nature of any interaction in processing the features of the two signals (masking of one signal by the other, inhibitory interaction between channels, correlation of the background in the two channels) (Klein, 1985; Thomas, 1985). To understand the consequences of detection plus classification for a receiver’s overall performance, we can compare PCD and PFA for detection plus classification with those for simple detection. The probability of

SIGNAL DETECTION AND COMMUNICATION

237

Fig. 5. (A) A bivariate plot of probability densities for combined detection plus classification shows the PDFs (now represented topographically by circles of equal probability density) for background stimulation alone, B, and in combination with each of two signals, S1 and S2. Decisions in this case require three thresholds: T1 for detection of B þ S1 from B; T2 for detection of B þ S2 from B; and T3 for classification of a signal once detected. (B) Threshold T2 results in a PFA (shaded) for responses appropriate for S2 when background alone occurs. (C) Threshold T3 results in a PFA (shaded) for responses appropriate for S2 when S1 occurs. (D) A combination of thresholds T1 and T3 results in a PCD (shaded) for correct detection and classification of S2.

correct response to a particular signal (PCD for detection plus classification) is always less than or equal to that for simple detection. As the threshold for classification, T3, decreases in slope, PCD for detection plus classification increases from near 0 to a value approaching PCD for simple detection (Fig. 5). The situation for PFA is more complex, because it involves two kinds of false alarm responding when only background occurs or when the alternative signal occurs. Because classification must follow detection, the two kinds of false alarm are not independent. Consequently, to combine the PFA for simple detection and the PFA for detection plus classification

238

R. HAVEN WILEY

requires information about the relative frequencies of these two situations. A full analysis of this situation is not yet available. Analysis of this situation is simplified by considering only false alarms for detection. False alarms in this narrow sense include only responses to background stimulation and thus include only false alarms for detection and exclude those for classification. For any level of false alarm in this narrow sense, classification in addition to detection reduces correct responses to signals in comparison to detection alone. Conversely, for any level of correct detections of signals, classification in addition to detection increases false alarms in this narrow sense (Macmillan, 2002; Starr et al., 1975). Classification plus detection, in comparison to simple detection, thus inevitably results in more false alarms by a receiver, even in the narrow sense. Classification thus inevitably reduces a receiver’s performance in comparison to detection alone.

VIII. COMPLEX PATTERNS: EXTENSION

OF THE

CONCEPT

OF

CHANNELS

Signal detection theory, as we have seen, describes decisions based on the outputs of perceptual channels. Detection and discrimination, the focus of discussion so far, suggest that the perceptual channels under consideration are sensory receptors and their immediate neural connections. Peripheral mechanisms of perception have been the main concern of many applications of signal detection theory, especially in studies of hearing. Nevertheless, the theory applies equally well to more cognitive aspects of nervous systems. A channel can in fact represent any step in the hierarchy of perceptual analysis of a signal. It could represent ‘‘detection’’ of a species‐specific song, for instance, when the issue is not whether or not a listening bird can hear each of the component frequencies but whether or not the entire pattern fits some criterion for a decision to respond. Such pattern detection has all the same general properties as feature detection. A channel for pattern detection produces an output that reflects the presence of components with particular sequential or simultaneous relationships. Irrelevant background stimulation can include similar relationships, differing in unpredictable ways from those in the signal, and the mechanism of the channel can itself include some unpredictability. A criterion for a decision to respond based on the output from such a pattern‐detecting channel inevitably results in false alarms and missed detections, just as from a feature‐detecting channel. Thus all of the preceding discussion of signal detection theory applies equally well to complex, as well as simple, perception. It applies to recognition of conspecific songs, to recognition of the vocalizations of mates,

SIGNAL DETECTION AND COMMUNICATION

239

offspring, or neighboring individuals, to mate choice based on complex repertoires, and to interpretation of subtle innuendos in the close‐range vocalizations of group‐living animals—signal detection theory applies to all communication.

IX. EVOLUTION

OF

SIGNALING AND RECEPTION

Signal detection theory suggests ways that receivers and signalers could coevolve (Wiley, 1994). We can understand many features of this coevolution by applying principles of signal detection first to optimizing receivers’ performance and then to optimizing signalers’ behavior. Because receivers provide the essential power for responses, their adaptation is primary. Nevertheless, signal detection theory shows that receivers do not necessarily get what they want. Because of the inevitable limitations on their performance, receivers can attain optimal, but not ideal, performance. Signalers can then evolve in response to the conditions set by their intended receivers. If changes in signalers’ behavior alter the features or frequency of signals, receivers might evolve new optima for their own performance. Then signalers might evolve new features of signals. It seems probable that this form of coevolution could either reach an equilibrium or propagate perpetual lags between the adaptations of signalers and receivers. Receivers can optimize the net utility of their decisions to respond or not by adjusting their criteria for response. The net utility for a receiver’s decision depends on the probabilities and payoffs (net gains, positive or negative) of correct detections, missed detections, false alarms, and correct rejections (for details, see Wiley, 1994). The payoffs from these four possible outcomes must be measured in units relevant to natural selection. The probabilities of these outcomes, we have seen, depend on the discriminability of signals and the receiver’s criterion. Depending on these payoffs and probabilities, the optimal criterion for a receiver can lie anywhere between adaptive gullibility (a low criterion for response when missed detections are especially costly) and adaptive fastidiousness (a high criterion for response when false alarms are especially costly). Gullability of receivers should result in the evolution of dishonest signals, fastidiousness in the evolution of exaggerated signals (Wiley, 1994). Signaling should evolve to increase the predictability of responses from intended receivers. As a result, signals should often evolve to improve detectability (Wiley, 1983, 1994), so receivers can in turn evolve criteria that permit high PCD and low PMD. Greater contrast and redundancy and less uncertainty about a signal’s features, including its timing and location,

240

R. HAVEN WILEY

all increase detectability and thus ultimately increase the probability of responses by receivers. Signal detection theory can explain why these properties of a stimulus affect detection and consequently learning and memory (the ‘‘receiver psychology’’ of Guilford and Dawkins, 1991, 1993). The widely reported phenomenon of peak shift in discrimination learning (Enquist and Arak, 1993; Guilford and Dawkins, 1991, 1993; ten Cate and Bateson, 1988; Weary et al., 1993) follows from maximizing the net utility of a receiver’s criterion for response (Lynn et al., 2005). When false alarms are more costly than missed detections, it pays for a receiver to adapt a strict criterion for response. Because the adaptive solution is to respond to extremes of signal properties in one direction rather than the other, in order to minimize false alarms, peak shift is the result. On the other hand, unintended receivers (eaves‐dropping predators and parasites or conspecific rivals, for instance) can reduce the advantages of highly detectable signals. Properties that improve detectability, such as redundancy and predictability, also limit possibilities for encoding of complex information, which requires variation rather than constancy in signals (Wiley, 1994). Signals might thus evolve a compromise between advantages of detectability and advantages of privacy or complex coding.

X. INTERPRETATION OF PLAYBACK EXPERIMENTS DETECTION THEORY

IN

TERMS

OF

SIGNAL

Experimental studies of communication depend on presentations of signals to subjects in order to record their responses. Signal detection theory suggests new approaches for designing and interpreting such experiments. First of all, it calls into question the use of clear signals. Because the ability of animals to detect or to discriminate any signals depends on background stimulation, experiments with intense signals and weak background stimulation often have little relevance to communication in natural situations. Signal detection theory, however, does not simply suggest cautious interpretation of playback experiments. It also identifies two distinct reasons why results should depend on background stimulation: both the features of effective signals and a receiver’s criterion for response should change with the level of background stimulation. Many investigations of the features of signals that make them effective in eliciting responses have employed clear signals and minimal background stimulation. This approach is unlikely to provide a full understanding of communication because, as the preceding review has indicated, the features of effective signals, those that optimize receivers’ performance, differ in the presence of high and low background stimulation. Signals effective

SIGNAL DETECTION AND COMMUNICATION

241

when background stimulation is low could prove much less so when background stimulation is high. In the latter case, we should expect greater emphasis on features that contribute to detectability of signals (contrast, redundancy, low uncertainty, familiarity). Experiments with playbacks have so far never considered the possible effects of background stimulation on detectability of signals. The interpretation of responses is also complicated by the possibility of confounding detectability of signals with criteria for responses. In studies of animal communication, experiments are usually interpreted in terms of the subjects’ attitude toward the experimental signals. For instance, do subjects have a lower threshold for a particular response to one type of signal in comparison to another? Yet the probability of response depends both on the listener’s attitude (its threshold or criterion) and on the level of the signal in relation to background stimulation as perceived by the listener (the detectability of the signal). A few studies of responses to bird songs in the field have considered both of these possibilities (Brenowitz, 1982; Richards, 1981b), but all have so far relied on indirect evidence to separate them. Even differences in responses to loud, repeated, clean signals might reflect differences in detectability of signals rather than differences in receivers’ criteria for response. When it is important to be sure that the receivers’ attitude (criterion) differs, only an ROC analysis can separate these possibilities. Signal detection theory also shows how to characterize the general properties of perceptual channels by comparing responses to at least three types of signals. Each pair of signals elicits responses that depend on outputs from a perceptual channel or combination of channels. Although only neurophysiology can determine the neural components and mechanisms of these channels, we can nevertheless learn something about their overall properties even without knowing the details of their mechanisms. For instance, are the pattern‐detecting channels for each of the three possible pairs of signals independent (A‐B, B‐C, A‐C)? Measuring the discriminabilities for the three possible pairs of signals can provide an answer. As explained earlier, discriminabilities that summed would indicate completely shared channels; discriminabilities with Pythagorean relationships would indicate completely independent channels; intermediate results would suggest partially correlated channels.

XI. PRACTICALITIES OF EXPERIMENTS

IN

NATURAL SITUATIONS

To take advantage of these possibilities, we must measure detectabilities and discriminabilities in the field. To accomplish this task, we have to broaden the way we think about experiments with playbacks. Presentation of

242

R. HAVEN WILEY

loud, repeated, clear signals close to subjects provides little information for comparisons of detectability or discriminability of signals. Instead, for this purpose, it would be better for each trial to present a brief (perhaps a single) stimulus in combination with background stimulation. Furthermore, the nature of the background must become part of the experimental design. To determine the detectability of a single stimulus, we can use background stimulation as a null stimulus (background only) for comparison with the signal (background plus signal). To determine the discriminability of two signals, the problem of a null stimulus does not arise. Nevertheless, including a null stimulus in the experimental design adds the possibility of a full analysis of detection plus classification, as described earlier. An ROC can then allow evaluation of normality and variance in the outputs of the channels involved and thus choice of an appropriate measure of detectability or discriminability. To construct an ROC from field studies of animals, a rating scale is likely to be the method of choice. To do so, we must first determine the distribution of some measure of response (perhaps the first principal component of all behavioral measures) across all trials. Depending on sample sizes, we can assign scores, for instance, to quartiles or deciles of this distribution. These scores provide nonverbal ratings of the subjects’ levels of confidence in discriminating between the two signals. The distributions of scores for each signal then generate pairs of PCD and PFA for the construction of an ROC. A practical problem in measuring ROCs in the field is the limited number of trials. Experiments with animals in the field can rarely expect, as psychophysical experiments do, to present signals hundreds of times to each subject and then to examine each subject’s ROC separately. Field studies will probably have to combine data from different subjects and thus determine only characteristics of populations, ideally ones as homogeneous as possible. Nevertheless, practical numbers of trials could yield useful measures of detectability in experiments with rating scales (McNicol, 1972, Chapter 5). Once an ROC is constructed, we can apply standard procedures for calculating detectability (or discriminability) of the signals. Furthermore, each pair of scores used to construct the ROC reveals the subjects’ average criterion under particular conditions. Procedures for calculating detectability or discriminability from a rating scale, summarized earlier, are thoroughly reviewed by McNicol (1972). The location of the criterion for response under particular conditions is best specified by its absolute location with respect to the underlying PDFs. Macmillan and Creelman (1990) recommend simple measures, such as (PCD þ PFA)/2 or
[z(PCD) þ z (PFA)]/2. With these procedures, the application of signal detection theory

SIGNAL DETECTION AND COMMUNICATION

243

to field studies of animal communication seems unlikely to encounter insurmountable problems.

XII. SUMMARY Signal detection theory involves a level of abstraction unfamiliar in field studies of animal communication. Mastering its implications, however, leads to some strong predictions about the evolution of signals and responses and to some new procedures for investigating animal communication. A consequence of this approach to communication is the fundamental conclusion that a receiver cannot independently adjust its PCD and PFA. The only exception is the limiting case in which the output of a channel in the presence of a signal is perfectly distinct from the output in its absence, so PFA ¼ 0. Otherwise, no matter how the criterion for response changes, any change in PCD is accompanied by a corresponding change in PFA. This compromise leads ultimately to a prediction that receivers evolve to optimize the net utility of their responses. The optimum might lie anywhere between extremes of gullibility or fastidiousness. In turn, signalers should evolve to balance the often incompatible advantages of increased detectability of signals, increased complexity of encoding, and restriction of signals to intended receivers. A second consequence of signal detection theory is the fundamental distinction between the detectability of a signal and the receiver’s criterion for a response. Detectability depends on the contrast of the signal impinging on the subject and on the selectivity of the subject’s perceptual channels. A receiver’s criterion for response depends on its attitude toward the output of its perceptual channels, as a result of a decision to accept particular PFA and PCD. Because any receiver’s responses to stimulation depend on both detectability of the stimulus and criterion for response, a definitive interpretation of responses requires attention to both. For a full interpretation of a receiver’s performance, it is necessary to include null presentations in experiments in order to measure false alarms as well as correct detections. Signal detection theory thus suggests new ways to design and to interpret experiments that compare responses to stimulation. Although some practical difficulties face any application of signal detection theory to field studies, none seems insurmountable. With this approach, we stand to learn more about (1) the adaptations for communication in situations with high background stimulation, such as in

244

R. HAVEN WILEY

choruses or complex social groups or at long range, (2) the effects of contrast, redundancy, reduced uncertainty, and familiarity on receivers’ abilities to detect and discriminate signals, and (3) the evolution of exaggeration or dishonesty in signals as a consequence of the evolution of receivers’ performance. In all of these ways, signal detection theory can advance our understanding of both the physiology and the evolution of communication.

Acknowledgments I thank many current and former colleagues for discussions on the ideas presented here, but especially Lori Wollerman, Marc Naguib, Jean Boal, and Stephen Nowicki.

REFERENCES Aubin, T., and Jouventin, P. (1998). Cocktail‐party effect in king penguin colonies. Proc. R. Soc. Lond. B 265, 1665–1673. Benzschawel, T., and Cohn, T. E. (1985). Detection and recognition of visual targets. J. Opt. Soc. Am. A 2, 1543–1550. Brenowitz, E. A. (1982). The active space of red‐winged blackbird song. J. Comp. Physiol. A 147, 511–522. Broadbent, D. E. (1958). ‘‘Perception and Communication.’’ Pergamon Press, London. Broadbent, D. E. (1967). Word‐frequency effect and response bias. Psychol. Rev. 74, 1–15. Brumm, H. (2004). The impact of environmental noise on song amplitude in a territoral bird. J. Anim. Ecol. 73, 434–440. Brumm, H., and Todt, D. (2002). Noise‐dependent song amplitude regulation in a territorial songbird. Anim. Behav. 63, 891–897. Brumm, H., Voss, K., Ko¨llmer, I., and Todt, D. (2004). Acoustic communication in noise: Regulation of call characteristics in a New World monkey. J. Exp. Biol. 207, 443–448. Cary, M., and Reder, L. M. (2003). A dual‐process account of the list‐length and strength‐ based mirror effects in recognition. J. Mem. Lang. 49, 231–248. Cherry, E. C. (1953). On the recognition of speech with one, and with two ears. J. Acoust. Soc. Am. 25, 975. Cherry, E. C., and Taylor, W. K. (1954). Some further experiments upon the recognition of speech with one, and with two ears. J. Acoust. Soc. Am. 26, 554–559. Cohn, T. (1978). Detection of 1‐of‐M orthogonal signals: Asymptotic equivalence of likelihood ratio and multiband models. Opt. Lett. 3, 22–23. Cynx, J., Lewis, R., Tavel, B., and Tse, H. (1998). Amplitude regulation of vocalizations in noise by a songbird, Taeniopygia guttata. Anim. Behav. 56, 107–113. Davies, D. R., and Parasuraman, R. (1982). ‘‘The Psychology of Vigilance.’’ Academic Press, London. Davies, D. R., and Tune, G. S. (1970). ‘‘Human Vigilance Performance.’’ Staples, London. Dobbins, I. G., Kroll, N. E. A., and Yonelinas, A. P. (2004). Dissociating familiarity from recollection using rote rehearsal. Mem. Cognition 32, 932–944.

SIGNAL DETECTION AND COMMUNICATION

245

Dooling, R. J. (1982). Auditory perception in birds. In ‘‘Acoustic Communication in Birds’’ (D. E. Kroodsma and E. H. Miller, Eds.), Vol. 1, pp. 95–130. Academic Press, New York. Egan, J. P., Schulman, A. I., and Greenberg, G. Z. (1959). Operating characteristics determined by binary decisions and by ratings. J. Acoust. Soc. Am. 31, 768–773. Egan, J. P., Greenberg, G. Z., and Schulman, A. I. (1961a). Operating characteristics, signal detectability and the method of free response. J. Acoust. Soc. Am. 33, 993–1007. Egan, J. P., Greenberg, G. Z., and Schulman, A. I. (1961b). Interval of time uncertainty in auditory detection. J. Acoust. Soc. Amer. 33, 771–778. Endler, J. A. (1990). On the measurement and classification of color in studies of animal colour patterns. Biol. J. Linn. Soc. London 41, 315–352. Endler, J. A. (1993). The color of light in forests and its implications. Ecol. Monogr. 63, 1–27. Endler, J. A., and The´ry, M. (1996). Interacting effects of lek placement, display behavior, ambient light, and color patterns in three Neotropical forest‐dwelling birds. Am. Nat. 148, 421–452. Enquist, M., and Arak, A. (1993). Selection of exaggerated male traits by female aesthetic senses. Nature 361, 446–448. Glanzer, M., and Adams, J. K. (1985). The mirror effect in recognition memory. Mem. Cognit. 13, 8–20. Glanzer, M., Adams, J. K., Iverson, G. J., and Kim, K. (1993). The regularities of recognition memory. Psych. Rev. 99, 546–567. Gomez, D., and The´ry, M. (2004). Influence of ambient light on the evolution of colour signals: Comparative analysis of a Neotropical rainforest bird community. Ecol. Lett. 7, 279. Green, D. M. (1961). Detection of auditory sinusoids of uncertain frequency. J. Acoust. Soc. Am. 33, 897–903. Green, D. M., and Birdsall, T. G. (1978). Detection and recognition. Psychol. Rev. 85, 192–206. Green, D. M., and Swets, J. A. (1966). ‘‘Signal Detection Theory and Psychophysics.’’ Wiley & Sons, New York. Green, D. M., Weber, D. L., and Duncan, J. E. (1977). Detection and recognition of pure tones in noise. J. Acoust. Soc. Am. 62, 948–954. Guilford, T., and Dawkins, M. S. (1991). Receiver psychology and the evolution of animal signals. Anim. Behav. 42, 1–14. Guilford, T., and Dawkins, M. S. (1993). Receiver psychology and the design of animal signals. Trends Neurosci. 16, 430–436. Heindl, M., and Winkler, H. (2003). Vertical lek placement of forest‐dwelling manakin species (Aves, Pipridae) is associated with vertical gradients of ambient light. Biol. J. Linn. Soc. 80, 647–658. Klein, S. A. (1985). Double‐judgment psychophysics, problems and solutions. J. Opt. Soc. Am. A 2, 1560–1585. Klump, G. M. (1996). Bird communication in the noisy world. In ‘‘Ecology and Evolution of Acoustic Communication in Birds’’ (D. E. Kroodsma and E. H. Miller, Eds.), pp. 321–338. Cornell University Press, Ithaca. Klump, G. M., and Langemann, U. (1995). Comodulation masking release in a songbird. Hearing Res. 87, 157–164. Langemann, U., and Klump, G. M. (2001). Signal detection in amplitude‐modulated maskers: I. Behavioral auditory threshold in a songbird. Eur. J. Neurosci. 13, 1025–1032. Langemann, U., Gauger, B., and Klump, G. M. (1998). Auditory sensitivity in the great tit perception of signals in the presence and absence of noise. Anim. Behav. 56, 763–769. Leonard, M. L., and Horn, A. G. (2005). Ambient noise and the design of begging signals. Proc. Biol. Sci. 272, 651–656.

246

R. HAVEN WILEY

Loeb, M., and Alluisi, E. A. (1977). An update of findings regarding vigilance and a reconsideration of underlying mechanisms. In ‘‘Vigilance: Theory, Operational Performance, and Physiological Correlates’’ (R. R. Mackie, Ed.), pp. 719–749. Plenum Press, New York. Luce, R. D. (1963). A threshold theory for simple detection experiments. Psych. Rev. 70, 61–79. Luce, R. D., and Green, D. M. (1974). Detection, discrimination and recognition. In ‘‘Handbook of Perception’’ (E. C. Carterette and M. P. Friedman, Eds.), Vol. 2. Academic Press, New York. Lynn, S. K., Cnaani, J., and Papaj, D. R. (2005). Peak shift discrimination learning as a mechanism of signal evolution. Evolution 59, 1300–1305. Mackie, R. R. (Ed.) (1977). ‘‘Vigilance: Theory, Operational Performance, and Physiological Correlates.’’ Plenum Press, New York. Macmillan, N. A. (2002). Signal detection theory. In ‘‘Stevens’ Handbook of Experimental Psychology, Third Edition. Methodology in Experimental Psychology’’ (H. Pashler and J. Wixted, Eds.), Vol. 4, pp. 43–90. John Wiley & Sons, New York. Macmillan, N. A., and Creelman, C. D. (1990). Response bias, characteristics of detection theory, threshold theory, and ‘‘nonparameteric’’ indexes. Psychol. Bull. 107, 401–413. Macmillan, N. A., and Creelman, C. D. (1991). ‘‘Detection Theory: A User’s Guide.’’ Cambridge University Press, Cambridge. Macmillan, N. A., and Kaplan, H. L. (1985). Detection theory analysis of group data, estimating sensitivity from average hit and false‐alarm rates. Psychol. Bull. 98, 185–199. Mulligan, R., and Shaw, M. L. (1980). Multimodal signal detection, independent decisions vs. integration. Percept. Psychophys. 28, 471–478. McNicol, D. (1972). ‘‘A Primer of Signal Detection Theory.’’ Allen and Unwin, London. Naguib, M. (2003). Reverberation of rapid and slow trills: Implications for signal adaptations to long range communication. J. Acoust. Soc. Am. 133, 1749–1756. Naguib, M., and Wiley, R. H. (2001). Estimating the distance to a source of sound: Mechanisms and adaptations for long‐range communication. Anim. Behav. 62, 825–837. Nieder, A., and Klump, G. M. (2001). Signal detection in amplitude‐modulated maskers: II. Processing in the songbird’s auditory forebrain. Eur. J. Neurosci. 13, 1033–1044. Nolte, L. W., and Jaarsma, D. (1967). More on the detection of one of M orthogonal signals. J. Acoust. Soc. Am. 41, 497–505. Pelli, D. G. (1985). Uncertainty explains many aspects of visual contrast detection and discrimination. J. Opt. Soc. Am. A 2, 1508–1532. Pierce, J. (1963). Some sources of artifact in studies of tachistoscopic perception of words. J. Exp. Psychol. 66, 363–370. Pollack, I., Rubenstein, H., and Decker, L. (1959). Intelligibility of known and unknown message sets. J. Acoust. Soc. Am. 31, 273–279. Richards, D. G. (1981a). Alerting and message components in songs of rufous‐sided towhees. Behaviour 76, 223–249. Richards, D. G. (1981b). Estimation of distance of singing conspecifics by the Carolina wren. Auk 98, 127–133. Seehausen, O., van Alphen, J. J. M., and Witte, F. (1997). Cichlid fish diversity threatened by eutrophication that curbs sexual selection. Science 277, 1808–1811. Shaw, M. L. (1982). Attending to multiple sources of information, I. The integration of information in decision making. Cognit. Psychol. 14, 353–409. Slabbekoorn, H., and Peet, M. (2003). Birds sing at a higher pitch in urban noise. Nature (Lond.) 424, 267. Starr, S. J., Metz, C. E., Lusted, L. B., and Goodenough, D. J. (1975). Visual detection and localization of radiographic images. Radiology 116, 533–538.

SIGNAL DETECTION AND COMMUNICATION

247

Swensson, R. G., and Judy, P. F. (1981). Detection of noisy visual targets: Models for the effects of spatial uncertainty and signal‐to‐noise ratio. Percept. Psychophys. 29, 521–534. Swets, J. A., and Birdsall, T. G. (1978). Repeated observation of an uncertain signal. Percep. Psychophys. 23, 269–274. Swets, J. A., Shipley, E. F., JMcKee, J. M., and Green, D. M. (1959). Multiple observations of signals in noise. J. Acoust. Soc. Am. 31, 514–521. Tanner, W., and Swets, J. (1954). The human use of information—I: Signal detection for the case of the signal known exactly. IRET. Inform. Theor. 4, 213–221. Tanner, W. P., Jr. (1956). Theory of recognition. J. Acoust. Soc. Amer. 28, 882–888. ten Cate, C., and Bateson, P. P. G. (1988). Sexual selection: The evolution of conspicuous characteristics in birds by means of imprinting. Evolution 42, 1355–1358. Thomas, J. P. (1985). Detection and identification, how are they related? J. Opt. Soc. Am. A 2, 1457–1467. Uy, J. A. C., and Endler, J. A. (2004). Modification of the visual background increases the conspicuousness of golden‐collared manakin displays. Behav. Ecol. 15, 1003–1010. Warm, J. S. (1977). Psychological processes in sustained attention. In ‘‘Vigilance: Theory, Operational Performance, and Physiological Correlates’’ (R. R. Mackie, Ed.), pp. 623–644. Plenum Press, New York. Watson, C. S., and Nichols, T. L. (1976). Detectability of auditory signals presented without defined observation intervals. J. Acoust. Soc. Amer. 59, 655–668. Weary, D., Guilford, T. C., and Weisman, R. G. (1993). A product of discriminative learning may lead to female preferences for elaborate males. Evolution 47, 333–336. Wiley, R. H. (1983). The evolution of communication, information and manipulation. In ‘‘Animal Behaviour, Volume 2, Communication’’ (T. R. Halliday and P. J. B. Slater, Eds.), pp. 131–181. Blackwell Scientific Publications, Oxford. Wiley, R. H. (1991). Associations of song properties with habitats for territorial oscine birds of eastern North America. Am. Nat. 138, 973–993. Wiley, R. H. (1994). Errors, exaggeration, and deception in animal communication. In ‘‘Behavioral Mechanisms in Evolutionary Ecology’’ (L. R. Real, Ed.), pp. 157–189. University of Chicago Press, Chicago. Wiley, R. H., and Richards, D. G. (1982). Adaptations for acoustic communication in birds, sound propagation and signal detection. In ‘‘Acoustic Communication in Birds’’ (D. E. Kroodsma and E. H. Miller, Eds.), Vol. 1, pp. 131–181. Academic Press, New York. Wollerman, L., and Wiley, R. H. (2002). Background noise from a natural chorus alters female discrimination of male calls in a Neotropical frog. Anim. Behav. 63, 15–22. Yonelinas, A. P. (2002). The nature of recollection and familiarity: A review of 30 years of research. J. Mem. Lang. 46, 441–517.

This page intentionally left blank

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 36

Preexisting Male Traits Are Important in the Evolution of Elaborated Male Sexual Display Gerald Borgia department of biology, university of maryland college park, maryland 20742, usa

I. INTRODUCTION The evolutionary cause of elaborate male sexual display traits remains controversial despite extensive recent research. R. A. Fisher is credited with developing the most widely discussed models: the good genes hypothesis (Fisher, 1915; see also Hamilton and Zuk, 1982; Maynard Smith, 1976; Zahavi, 1975, 1977) and runaway selection (Fisher, 1930, 1954; see also Arnold, 1983; Heisler, 1985; Lande, 1981, 1987). The good genes hypothesis has gained support from models showing how male traits and good genes preferences could coevolve (Houle and Kondrashov, 2002; Iwasa and Pomiankowski, 1999; Pomiankowski, 1987, 1988) and, most importantly, by strong empirical support (Go¨ransson et al., 1990; Hasselquist et al., 1996; Hill, 1991; Hoikkala et al., 1998; Kempenaers et al., 1992; Moore, 1994; Norris, 1993; Partridge, 1980; Petrie, 1994; Reynolds and Gross, 1992; von Schantz et al.,‘ 1989; Welsh et al., 1998; Wilkinson et al., 1998). Similar strong empirical support is lacking for runaway selection (Ryan, 1997). Developing on a largely separate track has been preexisting preference (Burley, 1985) and related models (Basolo and Endler, 1995; Ryan and Rand, 1990). In these models, the females commonly have preferences for male traits that are not currently expressed in males. Males that appear with novel traits suited to that preference are selected by these females. These preexisting preference models differ from all other sexual selection models because the female preference evolves as a pleiotropic side effect rather than from the benefits of mate choice (Burley, 1985) and may involve maladaptive female preferences (Ryan and Rand, 1990). Preexisting preferences are not coevolutionary and do not require genetic correlations between traits and preferences. This is seen by some (Kirkpatrick and Ryan, 0065-3454/06 $35.00 DOI: 10.1016/S0065-3454(06)36006-8

249

Copyright 2006, Elsevier Inc. All rights reserved.

250

GERALD BORGIA

1991; Ryan, 1998) as an important advantage over competing sexual selection models. Models of female preference and sexual display trait evolution (Iwasa et al., 1991; Lande, 1981) usually assume that there are no other female preferences already present. A more realistic approach is to consider how novel preferences might fare in competition with likely alternatives (Houle and Kondrashov, 2002) that are already established in the existing repertoire of female choice behaviors. Here I take an adaptive approach which suggests that multiple female preferences function as a coadapted set weighted in importance so that the total effect of mate choice brings the highest net payoff for females. Clearly then, new preferences must not only be functional, and must outcompete alternatives, but they will be selected in relation to their importance relative to other female preferences. This adaptive view is supported by the occurrence of situation‐specific female choice behavior that is dependent on the choosing female’s age (Coleman et al., 2004; Morris et al., 2003), past experience (Hebets, 2003), threat from predators (Breden and Stoner, 1987), and social circumstance (Doutrelant and McGregor, 2000; Otter et al., 1999) that appear to enhance the fitness of choosing females. Here I propose a broadened version of a model Fisher (1930) described as ‘‘war propaganda.’’ That model suggests that females use preexisting male aggressive traits in mate choice. We have suggested previously that these preexisting traits can indicate male genetic quality as sires and can result in females evolving preferences for using (co‐opting) these male cues in mate assessment (Borgia, 1979; Borgia and Coleman, 2000; Borgia et al., 1985; see also Berglund et al., 1996). I suggest that a wider array of traits can be co‐ opted for use as indicators of male quality. Co‐option of preexisting traits for mate choice should be viewed as an important model for the evolution of elaborate display because (1) it provides an explanation for how good genes preferences evolve with fewer of the limitations than other sexual selection models, and because (2) there is widespread evidence of co‐option of preexisting traits for use in sexual display traits and mate choice by females.

II. ALTERNATIVE MODELS

OF

DISPLAY TRAIT EVOLUTION

A. WAR PROPAGANDA MODEL Fisher considered trait borrowing (or co‐option) as a third mechanism for the evolution of elaborated male sexual display traits. He argued (Fisher, 1954, p. 151) that traits exaggerated by runaway selection might sometimes require ‘‘. . . an initial advantage not due to sexual preference.’’

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

251

Thus, an initial nonrunaway trait was suggested to be co‐opted for use in mate choice by a female preference that evolved under runaway selection. He also raised the possibility that male display might have dual functions in aggression and courtship saying, ‘‘. . . a sprightly bearing with fine feathers and triumphant song are quite as well adapted for war propaganda as for courtship’’ (p. 115). But he then plays down the idea saying, ‘‘Male ornaments acquired in this way might be striking but could scarcely ever become extravagant’’ (p. 116). Beebe (1929) and Wynne‐Edwards (1962) noted the similarity between aggressive and courtship displays and suggested dual use of these traits. The co‐option of traits for male display has been extensively discussed in the ethological literature (Schenkel, 1956; see Bradbury and Vehrencamp, 1993, Chapter 13). These discussions suggest the widespread occurrence of co‐option of display traits, but they are focused on the evolution of these traits as ritualized signals, not as indicators of mate quality in mate choice. Borgia (1979, see also Borgia, 1995; Borgia and Coleman, 2000; Borgia and Presgraves, 1998; Borgia et al., 1985; Loffredo and Borgia, 1986a) and Berglund (Berglund and Rosenqvist, 2001; Berglund et al., 1996) argued in support of the war propaganda hypothesis proposing that already elaborated male aggressive signals may be co‐opted for use in sexual display serving as effective indicators of good genes. The same vigorous, aggressive displays that are useful for intimidating competitors and that honestly indicate males’ ability to win fights may also indicate to females high male genetic quality. Females that evolve preferences for these display traits can gain a good genes benefit. Males producing aggressive display are policed by other males, so use of these displays by inferior males is often checked (Hurd, 2004; Parker and Ligon, 2002). On leks, males are often arrayed by their relative quality with more preferred males in more central positions (Kokko et al., 1998; Wiley, 1991), thereby enhancing female ability to find high‐quality males. Additionally, the displays themselves may be intense, for example, vocal displays often involving broadband calls, so that only especially fit males may be able to produce them effectively (Loffredo and Borgia, 1986a). Females also appear to incite males to fight and then use this information in mate choice (Bisazza et al., 1989; Borgia, 1981; Cox and LeBoeuf, 1977; Farr and Travis, 1986; Thornhill, 1988). Berglund et al. (1996) reviewed more than 200 cases of traits with aggressive and nonaggressive functions across a wide variety of taxa as evidence for the co‐option of aggressive traits for use in courtship display. They renamed the ‘‘war propaganda’’ hypothesis the ‘‘armament–ornament’’ hypothesis and classified it as a preexisting trait (as compared to a preexisting preference) model. Several studies have supported general predictions of the war propaganda/armament–ornament hypothesis

252

GERALD BORGIA

(Hein et al., 2003; Mateos and Carranza, 1999; Parker and Ligon, 2002; Thusius et al., 2001) and have used phylogenetic comparisons to show the direction of trait co‐option (Borgia and Coleman, 2000). Currently, runaway, good genes and preexisting female preference models dominate the sexual selection literature. Neither the war propaganda (armaments–ornaments) model nor the preexisting traits model is considered in any of the recent major reviews of sexual selection (Andersson, 1994; Andersson and Iwasa, 1996; Arnold, 1983, 1987; Candolin, 2003; Cuervo and Møller, 1999; Endler and Basolo, 1998; Fuller et al., 2005; Jennions and Brooks, 2001; Jennions and Petrie, 1997; Jennions et al., 2001; Kokko et al., 2002, 2003; Mead and Arnold, 2004; Møller, 1994; Ryan, 1997; Sargent et al., 1998; Zeh and Zeh, 2003). This omission is important because each of these widely discussed models has controversial aspects that may limit its application. By contrast, there are no similar limitations to the application of the war propaganda and related models. Additionally, because preexisting trait models predict adaptive female preferences that can lead to good genes without genetic correlations and can explain the evolution of costly displays, they may be most suited to explaining highly elaborated male displays. These models do not require genetic correlations between male traits and female preferences because male traits already exist, and the female preferences evolve to choose male traits that indicate male genetic quality or other benefits. These models lead only to adaptive mating preferences, in contrast to preexisting trait models, because the female preferences that are expressed are those that evolve as they increase female fitness in competition with already existing preferences. Because the male trait is already present, the problem of how initially rare female preferences are able to find initially rare male traits is avoided. New preference variants that appear can be readily tested by selection and if they increase female fitness then the preference can evolve to replace already existing preferences. This opportunity for new female preferences to be readily expressed indicates the potential for a high level of adaptive tuning of mate choice based on the frequent emergence of new preferences and competition among these preferences. B. PREEXISTING TRAIT MODEL Advocates of the war propaganda model suggest that aggressive display traits might be unique in providing information to females about the quality of males (Borgia, 1979), but what was not recognized is that there are multiple ways of indicating good genes in addition to aggressive display traits. Elaborate traits, such as the finely crafted bowers of bowerbirds, and the elaborate nests of cichlid fish, appear to have evolved initially for

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

253

functions not related to a good genes indicator function. Male variation in the construction and/or development of these traits may indicate genetic differences that are important to females in mate choice. While these male traits initially did not have an indicator function, differences in male performance and morphology may allow females who attend to these traits to choose males of high genetic quality. Those traits that indicate heritable differences in brain and motor development (Nowicki and Searcy, 2004), developmental stability (Thornhill and Moller, 1998), disease resistance (Hamilton and Zuk, 1982), or other characters that positively affect the performance and success of offspring may be used by females, or co‐opted, to indicate differences in male quality as sires. The relatively simple preexisting trait model offers an important alternative to existing models because (1) it needs to explain only the evolution of a female preference for an already existing male trait; (2) it explains how costly sexual displays can evolve; (3) there is no requirement for genetic correlations between male traits and female preferences or coevolution of these traits; and (4) it allows new female preferences to evolve readily and compete with alternatives leading to females with a repertoire of highly adaptive female preferences. Given that traits that are not aggressive can be co‐opted, the war propaganda or armament–ornament labels are no longer appropriate. Instead, a more suitable name for this expanded hypothesis is the ‘‘preexisting trait hypothesis.’’ While this chapter is focused on the evolution of female preferences for male genetic quality indicators, it is noteworthy that other benefits can be indicated by preexisting traits, for example, high‐quality male parental care (Soler et al., 1998a,b) and lowered risk for disease transmission (Borgia and Collis, 1990).

III. PROBLEMS

WITH

CURRENT MODELS OF ELABORATE DISPLAY TRAIT EVOLUTION

A. ZAHAVIAN HANDICAP MODELS While there is strong empirical support for the hypothesis that females choose males for good genes, the widely held view that these preferences coevolve with male traits dependent on genetic correlations leading to costly Zahavian handicaps (1977) has not been well supported. Zahavi’s requirement for costly male traits has two important problems that receive little attention: costs lower the male viability (Maynard Smith, 1976) and, if the traits are not completely sex limited, there will be costs to females. Also, the requirement for costs raises the issue of how good genes are

254

GERALD BORGIA

honestly indicated when these traits are in their incipient stages. Many versions of Zahavi’s handicap also depend on male condition to explain variation in the expression of elaborated traits. However, condition dependence in a variable environment may obscure rather than amplify differences among males and can reduce rather than improve the likelihood that females will choose males of high genetic quality. Here I review these potential complications of the handicap hypothesis as part of a comparison with the other explanation for good genes: the preexisting traits hypothesis. B. ARE GOOD GENES INDICATOR TRAITS HANDICAPS? Zahavi (1975, 1977) was among the first to use the good genes hypothesis to explain the origins of highly elaborated male display. He has focused on the cost of display proposing that costly handicaps are necessary to allow females to reliably identify high‐quality males. Zahavi (1977) proposed that males with handicaps produce displays of varying size and cost, and only the genetically best males are able to bear the high costs associated with producing the largest displays. He argued that by choosing males with these large costly displays, females are guaranteed to receive a good genes benefit. But Zahavi’s view (Zahavi, 1991, 1993; Zahavi and Zahavi, 1997) on the role of costs is extreme, arguing for ‘‘inefficiency’’ and ‘‘waste’’ as critical to insuring honesty (John, 1997). He says, ‘‘. . . the evolution of signals differs fundamentally from the logic by which all other characters are selected. All other characters are selected for efficiency . . .’’ (Zahavi, 1991). But the high cost of handicaps might outweigh the expected good genes benefits to offspring (Borgia, 1979; Davis and O’Donald, 1976; Maynard Smith, 1976). Despite this criticism, the handicap hypothesis has become the basis for many models that emphasize the role of costly traits in some form for producing honest advertisement of male genetic quality (Folstad and Karter, 1992; Getty, 1998; Grafen, 1990, 1991; Johnstone, 1995; Kokko et al., 2002; Kotiaho, 2001a; Nur and Hasson, 1984; Zahavi, 1975, 1977, 1991). Among these models, there has been surprisingly little effort directed at separating these models from Zahavi’s extreme views on the role of costs. Zahavi’s hypothesis has become so pervasive that some texts (Krebs and Davies, 1993) refer to all good genes indicator traits as handicaps. Overreliance on the handicap has caused some authors to assume that the presence of costly male display traits justifies a conclusion of good genes function (Alatalo et al., 1998; Kotiaho et al., 1998; Møller and Pomiankowski, 1993; van Doorn and Weissing, 2004; Verhulst et al., 1999) without considering alternative explanations. Møller and Pomiankowski (1993; see

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

255

also Candolin, 2003) have claimed that males with multiple display traits could only afford a single good genes trait because of the necessarily high cost of these displays. And with little other support, they have claimed that the remaining display elements must be inexpensive results of runaway or functionless vestigial traits. But there are several problems with this argument. First, male displays could be costly for a variety of reasons that are not due to selection for a wasteful Zahavian handicap. For example, male display traits could be selected for high signal value as in a passive attraction display (Parker, 1983) or as an advertisement call. Traits that are products of runaway are also predicted to be costly when there are intense female preferences (Arnold, 1983; Lande, 1981; Mead and Arnold, 2004), or expensive male displays that evolved in another context could be secondarily co‐opted for use as cues for male quality (Borgia and Coleman, 2000). Second, evidence supporting the role of waste and high cost as necessary components in male sexual display is still not established. Kotiaho (2001a) reviewed evidence for costly display and found that ‘‘. . . the data do not provide direct general support for the assumption that sexual traits are costly in line with the indicator mechanism models of sexual selection.’’ He concluded that there may be a problem with how costs are measured but did not consider the possibility that cost may not always be critical for honest display. Third, theoretical studies that consider the evolution of handicaps disagree about the necessity for costs to insure honest display. For example, contrary to Zahavi’s arguments that male displays must be generally expensive, Getty (1998) and Johnstone and Grafen (1993) suggest that only poor‐quality males must pay a cost for there to be honest displays. The alternative hypothesis that males can reliably indicate good genes without costly displays has been given little attention. Several authors (Borgia, 1979, 1981, 1993; Maynard Smith, 1991; Maynard Smith and Harper, 1995; von Schantz et al., 1989) have proposed that athletic displays indicate intrinsic differences in male genetic quality that cannot be easily cheated, for example, by the input of extra investment (see also Viljugrein, 1997; Wedekind, 1994 for other models of cheap honest sexual signaling). Lachmann et al. (2001) developed a model in which cost‐free signals evolve, but this model relies on the unrealistic assumption that male signals greater than their true quality are lethal. However, male signals can be constrained to signal their true quality in a more realistic way. For example, if they are limited by individual physiological, neurological, or athletic abilities, then low‐cost honest advertising of individual quality could occur. The existence of human and animal (e.g., horse and dog) championship performers who consistently win races and other athletic competitions with few obvious costs that lower survivorship or future reproduction suggests that inexpensive cues that honestly signal quality are

256

GERALD BORGIA

common and are not difficult to choose. In satin bowerbirds, males show extreme skews in mating success and individual male display quality and success are correlated across multiple years, which suggests that these displays are not extremely costly (Borgia, 1993). Additionally, there is no evidence of higher male mortality during the mating season when bowers and male display sites are maintained than at other times. In other species, high‐intensity athletic displays, for example, display rate, display intensity, strut rate or singing rate, are commonly used to indicate differences in genetic quality among displaying males but often these displays do not carry significant costs (Aparicio et al., 2003; Borgia, 1993; Jennions et al., 2001; Kotiaho, 2001a,b). Female preferences for costly displays in males could also raise mate searching costs for females. In satin bowerbirds, females tend to remate with successful males over successive years and appear to benefit from this remating because it lowers the cost of their mate searching. Females who lose a mate they have mated with over multiple years put the greatest effort into mate searching after that male dies. Thus, if males can indicate their quality with displays that differ only in cost, males with low cost displays will live longer and females who choose them will have lower mate searching costs.

C. COSTS

OF

HONESTY

IN INCIPIENT

TRAITS

If large costly traits are necessary to reliably indicate male quality, a critical problem for the handicap hypothesis is to explain how incipient male display traits can function as honest good genes indicators when they are still small and have relatively low cost. Such traits are unlikely to stress even poor‐quality males and thus reduce their ability to use these displays. Thus, in the early stages of their evolution, these traits would provide little honest information to females about good genes and they would be unlikely to be selected for their good genes indicator function. Alternatively, if already enlarged traits are co‐opted for an indicator function, then the problem of how incipient traits function as honest indicators of good genes is resolved because these traits evolved initially because of another function. Zahavi’s handicap hypothesis suggests that elaborated traits are designed to be costly. Alternatively, costs of display traits may be associated with their construction. Maynard Smith and Harper (1995) argue that if females were interested in what Zahavi (1991) refers to as waste, they would prefer males with asymmetric tails that would handicap their flight. I suggested (Borgia, 1979) that in most avian species, bright and enlarged crests and other plumage elements used exclusively for sexual displays

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

257

(and not as weapons) appear to have a high signal value relative to their cost, for example, light‐weight feathers that are hidden or folded away when not used in courtship (Gadagkar, 2003; Gilliard, 1969). This suggestion is supported by Barbosa and Møller (1999) and Aparicio et al. (2003) (see also Møller in Guilford, 1995) who found that elaborated feathers are often reduced in thickness indicating a design to lower production and aerodynamic costs. So, while some cost is necessary to produce any highly visible structure or display, there is no evidence that these traits are designed to enhance cost as expected from handicap models and there is evidence for cost reduction in many displays. Some traits like elongated male peafowl coverts appear to be costly, raising the question that if costly traits reduce benefits and are not a necessary requirement for honest display, why do they evolve? One likely answer is that already costly traits that evolved for another function have been co‐opted for a secondary function as indicator traits. For example, weapons that initially evolved for combat, like large antlers, provide an important immediate benefit to their owner that requires a high‐cost investment. These traits may be co‐opted as good genes indicators at little or no additional cost (Borgia and Coleman, 2000). The high cost of growing antlers combined with the use of these weapons to limit the opportunity of inferior males to cheat may provide a reliable signal of male fighting ability that also functions as a reliable cue indicating good genes (Berglund and Rosenqvist, 2001; Berglund et al., 1996; Borgia, 1979, 1981). The co‐option of traits as male quality indicators may provide the best opportunity for cost to function in enforcing honest signaling because it avoids many of the difficulties associated with the coevolution of costly male traits and female preferences. Since the good genes indicator function evolves only after the trait is already elaborated, there is no requirement that this trait produce honest signals when they are small and not very costly. Co‐option can also explain why costly indicators might evolve if there are cheaper low‐cost alternatives. If the original function of the co‐opted indicator trait remains important, then the costs of building that trait are tied to its original function, for example, as a weapon. Because these costs were there before the co‐option occurred, the addition of the secondary indicator function may occur with little or no additional cost, yet the initial costliness of the trait can help enforce honest advertising in its indicator function. Thus, the de novo evolution of costly genetic quality indicators may be limited because their costs must be subtracted from their benefits, but the evolution of costly good genes indicators may be more likely to arise where the indicator function has secondarily evolved in a preexisting, already expensive trait.

258

GERALD BORGIA

D. ARE CONDITION‐DEPENDENT TRAITS GENES INDICATORS?

THE

BEST GOOD

Another problem for the handicap hypothesis is that male condition‐ dependent traits may not reliably indicate male quality. The handicap hypothesis including most recent models (Andersson, 1994; Andersson and Iwasa, 1996; Getty, 1998; Iwasa et al., 1991; Johnstone, 1995; Rowe and Houle, 1996) claims that males in better condition are able to invest more in display and that the differences in ability to invest honestly indicate differences in male genetic quality. This condition dependence of traits introduces a strong environmental component into a process that is designed to assess genetic quality (David et al., 2000). Experiments designed to measure the genetic contribution of traits typically control for and reduce the effect of environmental variation (Falconer and Mackay, 1996). While females are unable to control the histories of males they are choosing among, they can choose traits less subject to environmental effects. Because male condition can be strongly influenced by the environment in ways that may not be representative of male genetic quality, for example, the quality of parental care received (Clutton‐Brock et al., 1982), local differences in the availability of resources, competition, past reproductive effort (Kokko, 1997), stress during development (Leitner et al., 2001; Nowicki et al., 2000, 2002; Polak et al., 2004; Spencer et al., 2003) or cheating on future reproductive investment (Candolin, 1999), and so on, females should assess male genetic quality with relatively condition‐independent traits if their choices are to be reliable indicators of quality. For example, females could choose males based on the display length (to test their endurance) that may tire them after each courtship or peak call frequency (Howard and Young, 1998) which may be less costly and is a more repeatable and reliable signal because it is influenced less by the demands of previous courtships. Condition dependence may allow cheating by genetically inferior males in several ways. In a cost‐dependent handicap system in which males mate over multiple years, inferior males might cheat by saving investment across years, allowing them to build up their condition, then spend these accumulated resources to perform well in one year (Kokko, 1997; Kotiaho, 2001a); or they may invest heavily in one year at the expense of future reproduction. These life history adjustments could allow cheaters to match or even surpass the investment of high‐quality males for at least one year (Kokko, 1997), improving their chance of reproducing and exposing females to unreliable signals of male quality. Female mammals adjust their reproduction based on past investment often skipping reproduction or investing in less expensive female offspring in the year after producing more expensive

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

259

male offspring (Clutton‐Brock et al., 1982). Candolin (1999) found evidence for cheating behavior in male sticklebacks which mate repeatedly through the year. Poor‐condition males cheat by developing the red color of good condition males, and she suggests that cheating males develop these displays at the expense of future reproduction. Badyaev and Duckworth (2003) found that male house finches that did not breed the previous year invested more in chest patch coloration. In these cases, there is evidence for advantageously adjusting investment between reproductive bouts. It is unclear if such adjustment occurs in species where males do not invest parentally and display across multiple years. Delayed maturation and plumage expression and lower attendance at display sites by young males are common among these species (e.g., bowerbirds, Marshall, 1954; birds of paradise, Gilliard, 1969; black grouse, Ho¨glund and Alatalo, 1995). We found that male satin bowerbirds may take on permanent bower sites at different ages, raising the possibility that males who delay bower holding may be saving resources for this task. But males with high‐quality displays maintained them across successive years, and there was no evidence that low‐quality males could enhance the quality of their displays in a single year (Borgia, 1993). This may be related to the important role of experience accumulated across multiple years in allowing males to construct successful displays such that cheating is suppressed by the lack of experience rather than by the costs of display. Good genes models differ in the extent to which they rely on costly handicaps to insure honesty. The limited evidence for cost associated with male display and evidence showing design for reduced costs imply that there has been no selection for waste. Reliable low‐cost male displays should have an advantage in competition with costly condition‐dependent displays because they provide the offspring of choosing females higher net benefits and allow the male display trait to be a reliable signal across a variety of environmental conditions, and make him available for mating across multiple years. Thus, cost may have a more limited role than Zahavi’s handicap model suggests. Studies differ in the role of male condition in affecting female choice; some show that females choose on the basis of male condition (Holzer et al., 2003; Rantala et al., 2003), whereas others do not (Gray and Eckhardt, 2001) and some show mixed results (Badyaev and Duckworth, 2003; Hunt et al., 2004). Experimental studies commonly show that males reared on depleted resources are less attractive to females than males who are not (Leitner et al., 2001; Nowicki et al., 2000, 2002; Spencer et al., 2003). The positive results of these experiments show that strong environmental effects can be sufficient to override the effects of genetic quality. They offer no clear evidence to support the hypothesis that females gain genetic

260

GERALD BORGIA

benefits from choosing males in good condition. More suitable experiments would involve designs that show a connection between genotype, condition and female choice.

IV. EVALUATING GENETIC CORRELATION MODELS A. GENETIC CORRELATIONS

AND

MATE CHOICE

The two most discussed sexual selection models, runaway and coevolutionary good genes models, both require a genetic correlation between male traits and female preferences (Andersson, 1994; Hall et al., 2004; Iwasa and Pomiankowski, 1995, 1999; Iwasa et al., 1991; Kokko et al., 2002; Lande, 1981; Pomiankowski and Iwasa, 1998). These genetic correlations result from gametic phase disequilibrium (linkage disequilibrium not due to physical linkage; Andersson, 1994). In these models, males with attractive traits obtain a mating advantage because of female preferences for that trait. This causes an increase in frequency of both the male trait and the female preference among offspring in the next generation. The statistical association of the male trait and the female preference in offspring produces the gametic phase disequilibrium. The occurrence of these correlations has been viewed as critical in assessing the plausibility of both good genes and runaway models (Andersson, 1994; Arnold, 1983; Bakker and Pomiankowski, 1995; Kirkpatrick and Ryan, 1991; Ryan, 1998) but remains controversial. Genetic correlations may be difficult to maintain under variable selection pressures (Barton and Turelli, 1991; Breden et al., 1994; Nichols and Butlin, 1989, 1992). Bakker and Pomiankowski (1995) indicate that when selection is suspended for one generation, the genetic correlation will be reduced by 50%. Sexual selection studies suggest a complicated mate choice dynamic that could limit the occurrence of genetic correlations in natural populations. Female preferences needed to maintain genetic correlations may be suppressed or altered by reductions in efficiency or increases in costs of mate searching resulting from predation threat (Breden and Stoner, 1987), parasitization (Simmons et al., 1999), male–male competition (Houde, 1994), or loss of top males requiring additional searching by females (Uy et al., 2000). The development of genetic correlations may also be limited by age‐dependent (Coleman et al., 2004) or learned (Hebets, 2003) mating preferences, mate choice copying (Gibson et al., 1991; Grant and Green, 1996; Ho¨glund et al., 1995), frequency‐dependent preference for male morphs (Hughes et al., 1999; Qvarnstro¨m et al., 2004), or other factors that reduce the association between a particular female preference genotype and the corresponding male trait genotype. Female preferences for multiple traits

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

261

(Borgia, 1985a; Mays and Hill, 2004; Møller and Pomiankowski, 1993) could also complicate choice and limit the evolution of genetic correlations if different mixes of male trait values suit females, thus lowering the intensity of selection on any single male trait. Additionally, with inbreeding avoidance females may be programed to discriminate against male relatives as mates who would, if genetic correlations were present, carry their most preferred traits. Thus, many mechanisms known to affect mate choice can reduce genetic correlations, so it is not clear that correlations with sufficient strength to drive and maintain correlation‐based sexual selection are present in natural populations. Artificial selection experiments have been used to show a correlated response to selection (Bakker and Pomiankowski, 1995; Houde, 1994; Wilkinson and Reillo, 1994) that has been interpreted as evidence for genetic correlations. But Gray and Cade (1999) argue that this correlated response test overestimates the genetic correlation. Genetic correlations also have been reported in some unselected populations (Bakker, 1993; Isyengar et al., 2002) but not others (Jang, 1997; Mu¨hlha¨user and Blanckenhorn, 2004). Evaluating these results is complicated because genetic correlations could occur because of pleiotropy (Kokko et al., 2002), physical linkage (Gilburn et al., 1993), intrapopulational mate choice polymorphisms, or gametic phase disequilibrium, with only the latter being consistent with genetic correlation‐based sexual selection models. Where genetic correlations have been found without artificial selection, the male displays are not the extreme types of highly elaborated traits that Fisher (1930) and others have sought to explain with runaway and good genes models. Bakker (1993) found a genetic correlation between the red coloration of male sticklebacks and a female preference, but the evolution of this trait could also be explained by its role in male–male territorial signaling with a secondary use as a quality indicator, perhaps for parental care (Candolin, 1999; Ku¨nzler and Bakker, 2001). Blows (1999) followed the evolution of genetic correlations on Drosophila across multiple generations and found correlations between traits and preferences. Initially, the correlations increased but they eventually collapsed as predicted by Nichols and Butlin (1989). Similarly, Houde (1994) found that divergence in female preferences in high and low selected lines in the first two generations decreased or reversed in the third generation. She attributed this loss of divergence in all four of her selection experiments to a breakdown in the genetic correlation. These results do not support models requiring the ongoing maintenance of genetic correlations (Hall et al., 2000; Kirkpatrick, 1982; Lande, 1981) and suggest that the importance of genetic correlations in shaping sexual display is still unresolved.

262

GERALD BORGIA

B. COMPARING GENETIC CORRELATION MODELS Kokko et al. (2002) proposed that good genes and runaway models should be merged into a larger model because in both models female preferences enhance the reproduction of males with attractive traits (see also Andersson, 1986; Eshel et al., 2000; Mead and Arnold, 2004). Thus, while differences among these models are usually cast as being between the male mating (sexy son) advantage of runaway models versus a viability enhancement to both sons and daughters of good genes models, a more appropriate comparison is the sexy son benefit alone from runaway models versus a sexy son and good genes benefits from good genes models. Females initially choosing males for good genes give males they choose, if as in many cases females choose the same males, a mating advantage as a side effect of their choices with the result that the sons of females receive both kinds of benefits. On the other hand, females choosing because of a runaway trait would not necessarily choose males that provide high viability to offspring (Lande, 1981). Although there are similarities between coevolutionary good genes and runaway models, this and other important differences suggest that these models should not be merged. There are at least three different good genes models that differ in their dependence on genetic correlations between the male trait and the female preference and on the coevolution of male traits and female preferences. The genetic correlation models (Iwasa et al., 1991) discussed earlier are the most widely discussed versions of the good genes models, but it remains unresolved if genetic correlations are critical to explaining elaborated male display. The second kind of good genes model is the simpler coevolution model that does not require a genetic correlation between male traits and female preferences. Male indicator traits increase because of the enhanced survivorship of offspring of males indicating high quality of males with these traits and because males have a mating advantage with females showing a preference for the indicator trait. Females mating with males having viability indicator traits gain an advantage because their offspring have higher fitness than females who do not attend to this male trait. This causes the female preferences for the male trait to increase. Because this kind of coevolutionary model does not depend on genetic correlations between male traits and female preferences, the inability to maintain genetic correlations is not critical for the successful coevolution of traits and preferences (although genetic correlations may occur). The simplest good genes model is the preexisting traits model. There is no genetic correlation or coevolution required because the male trait is already present and the female preference evolves because of gains in offspring quality. One potential problem for this

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

263

hypothesis taken alone is that sexual selection via female preferences is not involved in trait elaboration. It may be that female choice‐based sexual selection has little role in trait elaboration. Another possibility is that co‐option of preexisting male traits acts as a starting point for coevolutionary models that can lead to further trait elaboration. With the male trait already present in the population, it is not difficult for females to find males that may have variable expressions of the trait correlated with their genetic quality (the problem of males with incipient traits that do not correlate with fitness is bypassed if these traits are large), and by choosing males with more developed versions of the trait, females can enhance the fitness of their offspring which in turn selects for females choosing more extreme versions of the trait. This process may lead to elaboration of the male trait beyond the size at which co‐option occurred, particularly if this enlargement is not costly for top males. The disparity in the evidence for good genes versus runaway may be explained because of two advantages for the good genes models when they are in competition with runaway. First, at least two good genes models do not rely on genetic correlations and thus can evolve with less demanding requirements. Co‐option of preexisting traits does not require genetic correlations or the evolution of a novel male trait. The coevolution good genes model is more complex because it requires the evolution of the male trait; but because it does not require genetic correlations, it may allow good genes preferences to evolve under conditions when runaway cannot operate, for example, when genetic correlations cannot be maintained. Second, because good genes models provide both sexy son and good genes benefit, they should evolve more readily when in competition with pure runaway models that provide only a sexy son benefit. Consistent with the more difficult requirements for evolving runaway traits, there is scant evidence clearly supporting Fisher’s runaway hypothesis. For example, it is suggested that highly variable male display among sister groups at the tips of phylogenies provides evidence of runaway (Candolin, 2003; Omland, 1996a,b; Prum, 1997). But, there are many reasons for lability in male display among sister taxa including adaptation to different local sensory environments (Boughman, 2001; Endler et al., 2005; McKinnon and Rundle, 2002; Seehausen, 2000; Uy and Borgia, 2000); different levels of sexual competition regulated by the mating site (Panhuis and Wilkinson, 1999); or sexual isolation (Danley and Kocher, 2001) that does not depend on runaway selection. Alternatively, the existence of genetic correlations between traits and preferences is cited as evidence for runaway (Arnold, 1983), but this could occur for different reasons as discussed earlier.

264

GERALD BORGIA

V. EVALUATING

THE

PREEXISTING PREFERENCE MODEL

A. GENERAL ISSUES Preexisting preference models, sometimes called sensory exploitation, are attractive because, like preexisting traits models, they are relatively simple and do not require genetic correlations to explain the evolution of female preferences for male display traits. Sherman and Reeve (1999) discuss limitations in the operation of preexisting preference models arguing that because these preferences start as unselected side effects, it is unlikely that they provide genetic or other benefits and may, as Ryan and Rand (1990) suggest, produce maladaptive consequences for choosing females (see also Fuller et al., 2005). Such traits should be less likely to evolve and resist invasion if they were to become established than alternative positive benefit‐providing (e.g., good genes) traits (Houle and Kondrashov, 2002). Unfortunately, preexisting preferences are commonly considered as part of a larger model also involving sensory bias and sensory drive (Endler and Basolo, 1998), but this includes a range of different models that differ in their likely importance in shaping sexual selection. Sensory bias used in its original sense (Endler, 1992) to indicate that the environment affects the transmission characteristics of light and sound and therefore affects the form of signals is well supported (Boughman, 2001; Endler et al., 2005; McKinnon and Rundle, 2002; Seehausen, 2000; Uy and Borgia, 2000), but this is different from the question of whether there are preexisting preferences that are important in sexual selection (Fuller et al., 2005). Here I review three cases often cited as providing the best support for preexisting preferences and point out significant problems with each of these examples.

B. RECONSIDERING PREEXISTING PREFERENCES Preexisting preferences are suggested to be simple by‐products of the sensory system (Autumn et al., 2002; Basolo and Endler, 1995; Ryan, 1998). But in order for these preferences to operate, they may require much more complex and sophisticated mechanisms than are typically suggested. For example, in the Tu´ngara frog, Ryan et al. (1990) propose that male ancestors produced whine calls in mate sexual advertisement and that in a descendent species they evolved an additional and acoustically distinct chuck element in response to a preexisting female preference for a chuck call. Two different auditory structures are used to detect these call components. The whine component is perceived by the amphibian papilla, and the basilar papilla is used to detect the chuck elements. They argue that the

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

265

basilar papilla of the female is tuned to respond to the frequency range of the chuck and is more responsive to slightly lower than average frequency chucks in the population that are associated with larger males. Females thus choose larger males and gain a reproductive advantage (Autumn et al., 2002; Ryan et al., 1990). One significant problem for this hypothesis is how the proposed preexisting preference for chucks involving an auditory structure (the basilar papilla) can be fully functional in mate choice if it has not been previously used in that capacity. In Ryan and Rand’s experiments, they play chuck calls to females from species in which males do not give chucks and females show evidence of a preference for these calls. Their hypothesis requires a complex of interaction among functional traits that seems unlikely to be present in a basilar papilla that had not been previously used to detect chucks. For such a system to operate (1) there must be already existing neural circuits that detect the chuck as distinct from environmental noise; (2) these particular chuck‐sensitive neuronal elements must be linked to brain centers affecting mate choice; (3) but not other centers where stimulation would cause inappropriate or harmful effects; (4) the centers stimulated by the chuck call must cause females to be more inclined to mate with chuck‐producing males; (5) in contrast to being indifferent to or less inclined to mate; and (6) females are tuned to respond to a lower than average frequency of chucks that allow them to choose larger than average males. While there is little doubt that natural selection can shape an auditory system to achieve these tasks, it is unlikely that such a complex set of integrated capabilities tuned to function in an adaptive way (tuned so that females would choose large males) could arise, as these authors propose, without selection. Shaw (1995) suggested an alternative hypothesis consistent with the possibility that selection has directly shaped the functioning of the basilar papillae for mate choice. She hypothesized that ancestral calls in this lineage contained both the chuck and whine elements but that chuck elements were lost in some species while females retained their ancestral (now atavistic) preference for these call elements. Because these atavistic preferences for lost male traits had been shaped by selection, this hypothesis provides a more plausible explanation for how a female from a species in which males do not produce chucks can immediately and apparently adaptively respond to experimentally provided chucks in a way that indicates a preference for these calls. Ryan’s discovery (1985) that predatory bats use chucks to locate male frogs as prey is also consistent with this second hypothesis. Ryan and associates (Ryan, 1990, 1997; Ryan and Rand, 1993, 1995; Ryan et al., 1990) justified their preexisting preference hypothesis with a parsimony argument based on the mapping of chuck calls onto the phylogeny of this frog genus. Shaw’s hypothesis (1995) leads to an equally parsimonious mapping of these vocal displays as

266

GERALD BORGIA

compared with the preexisting preference hypothesis calling into question the validity of that hypothesis as applied to the evolution of chuck calls and their perception by females. This analysis suggests an even stronger argument against the hypothesis that preexisting preferences favored the evolution of chuck calls. It is extremely unlikely that the Tungara frog’s auditory system would have, without selection for hearing and responding to chuck calls, basilar papillae in females capable of detecting these calls and then causing them to respond by identifying and mating with high‐quality males, all without the benefit of selection for these functions. Another frequently cited example is the suggested preexisting preference for swordtails in the platyfish genus Xiphophorus and its close relatives. Basolo (1990) found that females from Xiphophorus species with unsworded males associate more with males from other species with swords and with conspecific males with artificially appended swords (Basolo, 1990, 1995a,b). This and a mapping of traits onto the Xiphophorus phylogeny led her to suggest a preexisting preference for swords in females of these unsworded species (but see Borowsky et al., 1995; Meyer et al., 1994). As with the Tungara frogs, an alternative hypothesis is that the preference for the elaborated male trait (swords) evolved in an ancestor and that they were lost in the lineages that do not have them. Suggesting why swords might be lost, Rosenthal et al. (2001) found that sworded males were more subject to predation than unsworded males. Additionally, Rosenthal and Evans (1998) found that female Xiphophorus prefer video images of males with large male body size and that this replaced the female preference for swords. They suggested an alternative interpretation for the behavior of females in nonsworded species (see also Sherman and Reeve, 1999) that females prefer to associate and perhaps mate with males with a large body size and that swords make males appear larger than similar sized counterparts. Basolo (2002) found an association preference in three of four unsworded species for experimentally sworded members of the opposite sex. She suggested that males share the latent preference shown by females, but this result also supports the hypothesis that this is a general preference for association with individuals with large body size (Gabor, 1999) not necessarily associated with mating preferences. Preexisting female mating preferences are also suggested to be important in mate choice in unionicolid water mites (Autumn et al., 2002; Proctor, 1991, 1992; Ryan, 1998). Proctor (1992) argues that this may be one of the clearest cases of preexisting preferences because of strong supporting behavioral and cladistic evidence, but close examination of this evidence suggests a less convincing case. Proctor (1991, 1992) hypothesized that male water mites produce water surface vibrations during courtship that mimic copepod swimming motions that attract the predatory females

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

267

of their species. Thus, males are able to exploit the female ‘‘preference’’ for copepod vibrations to gain matings. As support for her hypothesis, Proctor (1991) argues that (1) the frequency of male trembling matches that of copepod water surface vibrations; (2) females grasp males using the same motions they use to grasp prey suggesting that females are deceived to initially identify males as prey; and (3) hungry females appear more sexually receptive than well‐fed females, supporting her contention that females are mistaking males for prey. The most significant problem for Proctor’s hypothesis (1991) is that the behavior she reports is not consistent with the claim (see also Autumn et al., 2002) that male vibrations are used to mimic copepods and attract females from a distance. She says (Proctor, 1992), ‘‘Males search for mates by walking or swimming until they contact a female, whereupon they vibrate their first and second legs near the female (‘male courtship trembling’)’’ (p. 745). Thus, males touch females first and then vibrate; they are not attracting females from a distance. This is critical because by touching her first, the male water mite alerts her to his presence, making it unlikely that she would be fooled (sensorially exploited) by male mimicry of copepod vibrational signals. If vibrational signals are not used in mimicry, why are they present? Such signals are common in water mites in species for which copepod mimicry has not been suggested and are used in positioning the female, directing her to spermatophores and in postcopulatory mate assessment (Proctor and Smith, 1994). Also, the frequencies of copepod and male water mite vibrations are not so similar to provide unambiguous evidence of convergence necessary to support a claim for mimicry: trembling male water mites produce vibrations at 10–23 cycles/sec and copepods produce vibrations at 8–45 cycles/sec (Proctor, 1991). Overlap could occur because the mechanics of leg movements may constrain the possible frequency range of these vibrations. Proctor (1991) argues that hungry females show a greater response to courtship vibrations than well‐fed females as an explanation for why females grasp males. But, as suggested earlier, because males tremble after they touch females, hungry females should not be deceived into reacting to male vibrations as if they were indicators of prey. An alternative hypothesis is that hungry females may gain a nutritional contribution from male spermatophores and are thus more likely to seek matings. Proctor also suggests that similarity in how females mount males and attack prey indicates that females are being deceived by male mimicry of copepods. Female water mites commonly mount males for courtship even in species in which mimicry is not suggested to occur (Proctor and Smith, 1994), and it is common for individuals in predatory species to use predatory movements to gain access to potential mates. For example, in the yellow dung fly Scatophaga, males capture females for copulation in the same way

268

GERALD BORGIA

they capture prey (Borgia, 1982; Parker, 1970), and there is no suggestion that male dung flies are deceived into reacting to females as prey as has been suggested for female water mite’s reaction to males. Proctor’s cladistic analysis (1992) also does not convincingly prove the case for mimicry and sensory exploitation. She argues that net stance (the position males and females use in waiting to capture prey) occurs before or simultaneous with male courtship trembling that she associates with male mimicry of copepods. Her cladogram of 13 species shows branching into two major clades with no net stance and no trembling in one, and in the other 7 of 8 species show both of these behaviors and one does not show trembling. This results in two equally parsimonious maps requiring three transitions: net stance and male trembling evolving simultaneously and then one loss of male trembling which supports her hypothesis, or net stance evolving first and then male trembling evolving later in two separate events which fails to support her hypothesis. She concludes from this analysis that ‘‘. . . when taken together with previous behavioral evidence, this cladistic study strongly supports sensory exploitation as an explanation for male trembling in Unioncoidal mites’’ (1992, p. 745). But the cladogram at best suggests that these two alternatives are equally likely and that sensory exploitation may be less likely if losses are considered more likely than gains. Thus, the behavioral or cladistic data offered to support Proctor’s predation hypothesis are not clear‐cut, and there are alternatives that are at least equally plausible that do not support sensory exploitation. C. THEORETICAL ISSUES The argument for preexisting preferences becomes less convincing when set in the context of a new mate preference evolving in competition with other already existing adaptive mating preferences. While most models consider the evolution of new preferences in species where there are no other mating preferences, the widespread occurrence of mate choice among animal species suggests that this may be rare. In many species, male displays involve multiple traits that females use in mate choice (Borgia, 1985a; Candolin, 2003; Møller and Pomiankowski, 1993; Schluter and Price, 1993). The incorporation of a new unselected preference into the existing repertoire of female preferences should lessen the importance of other preferences. If these already existing preferences are advantageous to females, then replacement or lessening of importance of these preferences by a new unselected preexisting preference should reduce female fitness and lead to selection against these less advantageous preferences. Thus, the expression of a novel preexisting preference may be selected against both because it is unlikely to be beneficial and its use reduces the benefits provided by other previously established preferences.

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

269

There may also be selection against latent preexisting preferences as they await the evolution and expression of a suitable male trait. First, latent (not used in mate choice) preexisting preferences have costs necessary to make them operational so that females can choose appropriate male traits as they appear. These costs include the costs of structures and energy needed to allow females to identify appropriate novel male traits preferred by the latent preference and the cost of attending to nonexistent male signals at times during courtship when this attention may be more profitably directed toward assessing existing male display elements or threats such as predation. Although these costs maybe small, in the absence of a suitable male trait, there is no compensating benefit, so there should be selection for the elimination of these traits while they are still latent and before compatible male traits appear. This problem may be lessened if the latent preferences are adjuncts to existing preferences. For example, Burley and Symanski (1995) found in estrildine finches that both sexes have a preference for natural plumage colors in the opposite sex and when these are applied to artificial head crests. They interpret this as a preexisting preference, but it may also be viewed as an extension of an existing preference to other areas of the bird. Such ‘‘latent’’ preferences that use the same underlying mechanisms for mate choice as operational preferences seem more plausible given that there may be fewer added costs to expressing these preferences. Arguments supporting preexisting traits have relied heavily on mapping of male display traits and female preferences onto phylogenies. However, the mapping of male display traits onto well‐established phylogenies in a variety of other species shows a high level of rapid evolution at the tips including frequent reversals and convergences (Baker and Wilkinson, 2001; Ellsworth et al., 1995; Johnson, 1999; Kusmierski et al., 1997; Omland, 1997; Omland and Lanyon, 2000; Prum, 1997; Sturmbauer et al., 1996; Wiens, 2001). This suggests that parsimony use in trait mapping may neither be reliable for interpreting the evolutionary history of these traits (Cunningham et al., 1998; Losos, 1999; Reeve and Sherman, 1993; Shaw, 1995) nor useful for evaluating competing hypothesis, particularly when a small difference in the number of character state changes affects which hypothesis is supported. Wiens (2001) has suggested that the high level of turnover of male display traits and female preferences would rapidly deplete the store of latent female preferences. Thus, it is unlikely that preexisting preferences give rise to the rapid evolution of diverse traits seen in many species with highly elaborate display. Also, because these preferences are not selected for their mate choice function, they are unlikely to explain the evolution of complex adaptive mate choice behaviors such as conditional preferences that change as females age (Coleman et al., 2004) or with female social circumstances (Doutrelant and McGregor, 2000; Otter et al., 1999) or that involve complex

270

GERALD BORGIA

courtship communication (Patricelli et al., 2002). Preexisting biases that affect mate choice may be important in some contexts. For example, in satin bowerbirds, females are threatened by high‐intensity male courtship display (Coleman et al., 2004; Patricelli et al., 2002). Females are often chased at feeding sites by the larger and more dominant males so it is not surprising that they would be sensitive to threat when courted with high‐intensity displays that have aggressive elements by these same males (Patricelli et al., 2002, 2003). Female signaling of their level of comfort in courtship and males modulating in reaction to signals of discomfort provide a means by which males and females can overcome the threat associated with attractive high‐intensity displays. In this case, not all females are threatened to the same degree with older females showing less discomfort from high‐intensity displays than younger females (Patricelli et al., 2004; Coleman and Borgia, submitted for publication). Also, this preexisting trait is adaptive in providing protection to females in what they perceive to be dangerous circumstances and this causes females to avoid rather than prefer particular males. It may be relatively more common for biases that are generally protective to influence mate choice by limiting danger to females rather than compelling them to mate based on traits not related to adaptively evolved preferences. (For a comparison of sexual selection models see Table I.)

TABLE I COMPARISON OF MODELS OF SEXUAL SELECTION

Model Runaway Coevolutionary good genes Preexisting preference Preexisting trait good genes a

Genetic correlation required

Coevolution required

Early costly trait problem

Evidence

Adaptive preferences

Yes Yes

Yes Yes

No Yes

No Maybea

No Yes

No

No

No

Maybeb

No

No

No

No

Maybea

Yes

Good genes have been related to particular phenotypic traits females choose in males in a variety of species, but it has not been resolved if these traits evolve by coevolution or from preexisting preferences. There is independent evidence of many male display traits having a preexisting function. b While there are several studies that claim to show a preexisting preference evolution, most subject to alternative interpretation. One likely case is discussed by Burley and Symanski (1995), but it may have limited application, see text.

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

VI. EVIDENCE FOR

THE

CO‐OPTION

OF

271

PREEXISTING TRAITS

A. INTRODUCTION The preceding review suggests an important role for the preexisting traits version of the good genes model. This model is particularly attractive because of its relative simplicity and because it makes clear testable predictions. Since the model assumes that male indicator traits are co‐opted from traits that have previous function, a useful test is to determine if the existing indicator display shows evidence of a previous function. For this, detailed phylogenies which allow the order of trait evolution to be resolved can be valuable, particularly if there is evidence of multiple co‐options. But even then, assessing the order in which trait functions evolve can be difficult if there has been rapid evolution of display functions (Kusmierski et al., 1997), or if, as is often true, it has not been determined whether a trait has an indicator function across a group of species. Without phylogenies we can identify traits that have dual functions as being likely candidates in which one was co‐opted but we cannot resolve the order in which they evolved. Sometimes traits are widespread and have obviously long histories, for example, bird nests, so that a secondary use in a small set of species can be inferred even without detailed phylogenetic information. Despite these limitations, there is growing evidence that co‐option has been important in the evolution of indicator displays, and with more reliable phylogenies and information on trait functions as indicators, we can better assess the importance of the preexisting trait model.

B. ICONIC NONMORPHOLOGICAL DISPLAY Many cases of iconic (exemplar), highly elaborated sexual display traitsshow evidence of co‐option and suggest that co‐option may be generally important in the evolution of elaborated displays. Display trait co‐option is likely affected by the preexisting conditions associated with courtship. The kinds of traits that may be most readily co‐opted are especially large or difficult to build structures or other traits expressed near courtship sites that are effective in showing differences among males (e.g., neurological function, parasite resistance, developmental stability, or resistance to interference by other males) that correlate with and can indicate good genes. For some traits, there may be post‐co‐option evolution such that those traits shown only briefly during display may be selected to be exposed longer or presented where they can be more easily viewed by females. The nests of birds, sticklebacks, and mouth‐breeding cichlids that have long histories with a clearly defined initial function as a repository for eggs

272

GERALD BORGIA

now show evidence of a secondary function as indicators of male quality. Barber et al. (2001) suggest that stickleback nests secondarily function as male quality indicators. Males differ in their ability to produce Spiggin, a glycoprotein glue used to hold nest‐building materials together, and this results in nests that differ in quality. They suggest that variation in ability to produce costly Spiggin and its use in building high‐quality nests has secondarily become a condition‐dependent indicator that females assess as part of mate choice. Bird nests have been discussed as having a secondary function in advertising male quality (Collias and Victoria, 1978). Soler et al. (1998a,b) suggest that females discriminate among male nests to assess the quality of male parental care. Several studies have shown that males of some species build multiple nests and those males with more nests are more likely to attract a mate (Verner and Englesen, 1970). The ability of males to build multiple nests may be attractive to females because they indicate male quality, or because freshly built nests may be better nests that attract females because eggs are better protected. The first case would be consistent with a co‐opted indicator function, while the second would suggest a proximate benefit for females. Quader (2005) found that in baya weavers (Ploceus philippinus), females chose nests based on location (e.g., over water and height) and architecture (neatness of weave). It is possible that neatness may indicate differences in male quality. Mouth‐breeding African cichlids build large volcano‐shaped sand display structures that originated as nests (McKaye et al., 1993; Tweddle et al., 1998) and currently function both as a site where eggs are briefly deposited before being picked up by the mouth‐breeding females and as a trait used in mate selection. Outgroup comparisons suggest that ancestral nests were small, and it remains unclear when and how these sand structures became enlarged. McKaye et al. (2001) suggest that these sand structures function to protect eggs from sneaker males who might eat them during transfer, and enlargement may have occurred for this function. Their hypothesis suggests that there were two co‐option events in the evolution of mouth‐breeding cichlid sand structures, the first involving the co‐option and modification (enlargement) of the sand structure for use as a site for protected egg transfer and a secondary co‐option of these structures as display elements that females use in assessing males (Taylor et al., 1998). Bowerbird bowers were initially thought to be modified nests (Sodderberg, 1929), but the absence of evidence for egg laying in bowers and differences between nests and bowers in shape, location (tree vs ground) and builder (male vs female) indicate a separate origin (Borgia et al., 1985). Across different bower types, bowers show a design most consistent with

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

273

protecting females from forced copulation by courting males. Two species that have lost bower‐building show alternative adaptations that allow female choice (Borgia, 1995), and generally females not protected by the bower during courtship are susceptible to forced copulation. Bowers may serve to calm females and increase female visitation which, for high‐quality males, likely outweighs the reduced opportunity for forced copulations (Borgia, 1995; Borgia et al., 1985). Bowers are also used in mate choice in at least one of the two clades of bower‐building bowerbirds (Borgia, 1985a; Borgia and Mueller, 1992), and I proposed that they have been co‐opted for this mate choice function secondarily after they evolved initially as barriers that, for males, increase female visitation and, for females, protect them from forced copulation (Borgia, 1995). Under this hypothesis, bowers were present on the display court and available for inspection by females while serving their initial protective function and then females evolved to use already existing differences in the quality of bower construction to assess males. In satin bowerbirds, various characteristics of the bower, including its symmetry, neatness in construction, and the fineness and density of sticks, are strongly correlated with male mating success (Borgia, 1985a), and these traits may indicate to females heritable differences in male motor skills and resistance to destruction by competing males (Borgia, 1985b). Among structure‐building species, there are also cases where the preexisting traits hypothesis is not supported. Several species of fiddler crabs build sand hoods and pillars near their burrows. Christy et al. (2003) propose that these structures are built from sand leftover from burrow construction and that they now function as markers allowing males to quickly relocate their burrows when threatened by a predator. They suggest that females also use these structures in finding burrow entrances when threatened by predators. Males benefit from this behavior because females tend to mate with males once in burrows, although females do not show a preference for males with pillars. It may be that sand pillars do not reflect owner’s quality with sufficient accuracy for females to use them in mate choice or that the relatively simple visual system of these crabs may not allow such discrimination. Weakly electric fish (order Gymnotiformes) generate multifunctional electric organ discharges (EODs) for electrolocation (e.g., finding prey) and social communication. Hagedorn and Zelick (1989) suggest that the strength of the EODs provide information about the internal state of the animal, including their state of health, which may be useful to females in assessing males as mates. Phylogenetic analysis suggests that EODs have been selected for greater signal complexity resulting in lower detectability by key predators. For extant species in the families Gymnotidae, Hypopomidae, Rhamphichthyidae, and Apteronotidae, an additional wave phase

274

GERALD BORGIA

was added to the ancestral monophasic signal that shifts its spectrum above the most sensitive frequencies of electroreceptive predators (Stoddard, 1999). Signals in the family Hypopomidae are sexually dimorphic, and males have extended the duration and amplified electric signal pulses of these secondary EODs for use in sexual display. Curtis and Stoddard (2003) found that female Brachyhypopomus pinnicaudatus preferred males with longer duration and higher amplitude EODs in mate choice. These traits correlate with male body size and success in encounters with other males. They suggest that these secondary EOD elements that evolved to enhance crypsis to predators have been co‐opted by females for assessment of male quality. Many moths have independently evolved ears on various parts of their body to respond to ultrasonic pulses of predatory bats. In many groups, this is associated with the evolution of ultrasonic clicks and other calls by both sexes. These calls may have initially functioned to jam bat signals (Fullard et al., 1994) or warn bats about the distastefulness of the signaling moth (Hristov and Conner, 2005). Quite remarkably, a wide variety of sound production mechanisms have independently evolved in different groups of moths (Connor, 1999), and in many cases these have resulted in a series of co‐options resulting in the use of male ultrasonic signals in mate choice. Wax moth, Achroia grisella (Pyralidae), males call continuously near the wax combs of honeybees (Greenfield and Colfelt, 1983; Snedden et al., 1994). Although males simultaneously release a sex pheromone (Dahm et al., 1971), the ultrasonic acoustic signal alone appears critical for the female approach to the male (Jia and Greenfield, 1997; Spangler, 1984; Spangler et al., 1984). Jang and Greenfield (1996) found that females more often approach synthetic calls with high pulse amplitude, pulse duration, pulse repetition rate, and pulse asynchrony, providing strong evidence that these ultrasonic calls are important in mate choice. In the rice moth Corcoran cephalonica, these ultrasonic calls attract virgin females. Jang (1997) showed that females prefer males who produce calls with high levels of acoustic energy and/or asynchrony. Noctuid moth Hecatesia exultans males perch in vegetation producing high rates of chirp calls. Male–male agonistic interactions involve ultrasonic calls and calling males increase chirp duration in response to conspecifics (Alcock and Bailey, 1995). These calls function in mate attraction as females approach calling males on their lek and solicit copulations. Alcock and Bailey (1995) suggest that females may be choosing among the lekking males who have proven their quality through aggressive interactions with other males. It remains unclear to what extent females use differences in male chirps in mate choice (Alcock et al., 1989; Surlykke and Fullard, 1989). The noctuid Amyna natalis also displays on well‐exposed vegetation

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

275

and makes a buzzing call by twisting its vertically held wings at a high frequency which activates a thinly scaled forewing ‘‘tymbal.’’ Sound production in this species may be associated with the release of a pheromone from putative scent‐disseminating structures on the underside of the forewing tymbal (Heller and Achmann, 1993). These studies suggest that ultrasonic calls that have evolved to confuse, or signal distaste fitness to bats have gained a secondary function in mate choice. C. CO‐OPTION

OF

MORPHOLOGICAL TRAITS

Many morphological traits appear to be co‐opted for mate choice. Often these traits are co‐opted from aggressive display (Beebe, 1929; Berglund et al., 1996). Deer antlers and similar weapons of antelope clearly evolved for aggressive functions (Clutton‐Brock et al., 1982; Darwin, 1871), and several authors suggest an additional role in mate choice (Barrette and Vandal, 1986, 1990; Geist, 1971; Lincoln, 1994; Markusson and Folstad, 1997). Ditchkoff et al. (2001) have related enhanced antler size to the expression of a particular major histocompatability (MHC) genotype and resistance against parasites, indicating that females may use antler size as a good genes indicator. Sivinski (1997) in his review of dipteran ornamentation says, ‘‘Ornaments that appear to be used in aggressive interactions with members of the same sex seem to be concentrated on the head. Since the head is often used in the pushing style of confrontation and combat typical of Diptera, such elaborations are probably embellishments of weapons or advertisements of size and the ability to use weapons. They may then take on a presumably secondary function by advertising sexual competitiveness to potential mates (e.g., stalk‐eyes).’’ Male Diopsid flies have eyes on the end of long stalks sometimes with eyespans wider than their body length. These structures are sexually dimorphic and function in male–male aggression (Panhuis and Wilkinson, 1999) and allow males to assess size directly and the fighting ability of rivals. In highly dimorphic species, females use male eyespan in mate selection (Panhuis and Wilkinson, 1999). In root aggregating Malaysian flies, those that defend large groups of females have larger eyestalks (Wilkinson and Dodson, 1997). This suggests that male–male aggression associated with female control has had a key role in driving eyestalk elaboration and that female use of male eyespan in mate choice may be secondarily evolved in these species. Male fiddler crabs have an asymmetrically enlarged claw that may account for half of their mass. Co‐option has clearly reshaped the use of these claws from an initial feeding function (still retained by the males’ minor and females’ claws) to an enlarged and robust design for male–male

276

GERALD BORGIA

combat associated with burrow acquisition and defense (Hyatt and Salmon, 1978; Jennions and Backwell, 1996). Studies show that male major claw waving rates increase sharply when females are present (Pope, 2000), suggesting that waving functions primarily in mate attraction. Females do not discriminate in favor of males with more robust nonregenerated claws but these males win fights against males with narrower regenerated claws (Backwell et al., 2000), suggesting that the robustness of claws is primarily associated with their fighting function. It remains to be resolved whether the initial cause of claw enlargement was for fighting ability or for mate attracting through waving, but what is clear is that there have been successive co‐options of the major claw for its ultimate use in sexual combat and sexual display. Combs in male junglefowl are used as signals of dominance status and are also used by females in mate choice (Ligon and Zwartjes, 1995; Zuk et al., 1990). Parker and Ligon (2002) showed that the comb size in male junglefowl is a dominance indicator that is suppressed when subordinate males are in the presence of more dominant males. Female fowl generally prefer dominant males (Leonard and Zanette, 1998). A likely hypothesis consistent with the occurrence of small combs in females is that comb size originally functioned as a signal of dominance that was co‐opted for use in mate choice. D. CO‐OPTION

OF

VISUAL

AND

CHEMICAL TRAITS

Similar kinds of bioluminescent signals are used in fireflies and ostracode crustaceans (Herring, 2000) and appear to have been co‐opted for use in sexual display. Branham and Wenzel (2000) argue that bioluminescence in the beetle family Lampyridae and close relatives originally functioned as an aposematic warning in larvae and was later co‐opted for this same function in adults (see also Sivinski, 1981). Only larvae are bioluminescent in the basal‐most luminous taxon, and they have only laterally located light organs that are used for signaling their unpalatability to predators. Adults in more derived species have lateral and ventral organs and use the latter for sexual signals and have also developed the capability of pulsing these signals (Ghiradella, 1998). The relatively late use of this historically old trait in sexual display suggests that it was co‐opted for that function. Flash patterns differ among firefly species, and sexually ready females respond to conspecifics suggesting that flashes function in species recognition (Lloyd, 1971). Also, female Photinus fireflies discriminate among conspecific males (Branham and Greenfield, 1996) based on flash intensity (Vencl and Carlson, 1998) and duration (Cratsley and Lewis, 2003; Lewis et al.,

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

277

2004a). Flash characteristics are good predictors of spermatophore mass (Lewis et al., 2004b), suggesting that females may benefit from using these signals in assessing males. Although different aspects of ventral male flashes are associated with species recognition and courtship, the dual use of the ventral light organ suggests that one function was co‐opted from the other. Fireflies present a particularly interesting but not unique case in which there appears to have been multiple successive co‐options, first the co‐option of bioluminescence from larval predator defense displays, then use in predator defense in later life‐history stages, and then use of this trait for producing ventral light organs for sexual display, and then possibly co‐option of this species recognition signal for use in mate assessment. Like the fireflies, the Caribbean ostracode Vargulae uses bioluminescence both in defense and sexual signaling (Morin, 1986). When attacked, the ostracode squirts a pulse of luminescent fluid from its upper lip into the water as antipredatory behavior. The same pulsing system is used by males signaling to sexually receptive females. A phylogenetic analysis of the ostracodes (Cohen and Morin, 2003) suggests that, like fireflies, the evolution of bioluminescence as a defensive signal preceded its use in sexual display function. This is consistent with the hypothesis that bioluminescence which was used first as an antipredator adaptation has been co‐opted for sexual display. It is unresolved whether these sexual signals are used by females to assess male quality. Co‐option is extremely common at the molecular level and is increasingly important in understanding the evolution of new genes and gene families (Holland et al., 1994). Like sexual signaling, the evolution of the unique firefly bioluminescence enzyme luciferase appears to be the product of co‐option. Day et al. (2004) report that this is a bifunctional enzyme catalyzing light emission and functioning as a fatty acid CoA ligase. They suggest that the light‐emitting function was initially a side effect that was co‐opted for display. Other unrelated nonbioluminescent beetles have the ability to support luciferin‐dependent bioluminescence, indicating that this capability is not unique to the Lampyridae and preceded the evolution of bioluminescent organs, which appear to be co‐opted from fat storage organs. Volatile olfactory signals are used in many aspects of animal communication such as scent marking and mate assessment (Blaustein et al., 1993; Zala et al., 2004). Sweat, urine, and dung commonly contain many biochemical by‐products that can potentially reveal the condition and other aspects of the physiological state (Gosling and Roberts, 2001; Zala et al., 2004) and genetic characteristics of individuals. These may be viewed as preexisting traits that have been co‐opted as indicators of male condition or health

278

GERALD BORGIA

by male opponents and are used by females for assessing male quality (Isvaran, 2004). Where these odorants have taken on important functions in social interactions and their production has positive benefits for at least some individuals producing them, there may be selection for increased production to better advertise these traits. Specialized structures, such as scent glands, probably evolved to enhance and control the production of particular components of sweat apart from the normal physiological functions associated with sweating. It is possible that scent glands may have first functioned as part of a territory marking system used to communicate with other males and females, and were then co‐opted for assessment by females, although the reverse pattern of co‐option could have also occurred. Chemical cues associated with the MHC loci in sticklebacks are used in mate choice (Reusch et al., 2000). MHC odorants may have been released as by‐products across permeable membranes into the water, and females began to use these odorants as effective cues in mate choice (Haberli and Aeschlimann, 2004), preferring more genetically diverse males. Singer et al. (1997) claim that co‐option may be common in the evolution of chemical signals used in mate choice, ‘‘Organisms as diverse as marine invertebrates and mice and humans may have seized these serendipitously available volatile signals of individual identity [MHC] to identify appropriate mates, thereby avoiding inbreeding, or to recognize siblings, parents, or offspring.’’ Extending Singer et al.’s argument, MHC by‐products may have started as a mechanism for inbreeding avoidance and then generalized to allow selection of mates that contribute toward more disease‐resistant MHC genotypes (see Penn and Potts, 1999). In less viscous populations, selection for discrimination among male MHC by‐products may have been more direct without first involving inbreeding avoidance. E. CO‐OPTION

OF

VOCAL DISPLAYS

AND

DISPLAY MECHANISMS

Searcy and Andersson (1986) point out that the songs of birds, frogs, and insects, although acoustically very different, have important functional similarities. The origins of these songs represent independent evolutionary events in each of these groups with multiple independent origins occurring among the insects, for example, Drosophila, Caribbean fruit flies, Orthoptera, cicadas, moths, and beetles. In many cases, these signals have dual uses in female choice and male contests (see also Brenowitz and Beecher, 2005; Nowicki and Searcy, 2004) and use the same anatomical and neurological mechanisms. The commonness of these dual use vocal displays suggests widespread co‐option, but it is unclear which function occurred first. In birds, Beecher and Brenowitz (2005) claim that small repertoires

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

279

are associated with male territorial contests and large song repertoires are more often associated with female choice. The more widespread occurrence of small repertoires in birds suggests a possible origin for song learning associated with territoriality and then a later co‐option of the song mechanism for use in sexual display, but this is far from conclusive. What may be overlooked in these discussions is that females may use male calls for locating males and for assessing male quality. Female use of male calls to locate males seems a less demanding task than assessing male quality. So the initial function of these calls may have been advertisement in which females already recognized males and then, secondarily, females used differences in this preexisting male trait for mate assessment. The use of learned song by passerine birds provides a mechanism particularly susceptible to co‐option. The vocal systems of passerine birds are built for young birds to learn songs from tutors on the same or nearby territory, which they will use for their lifetime (Marler and Peters, 1981, 1982, 1987, 1988; Nelson, 1992; O’Loghlen and Beecher, 1997). Some species have a more open‐ended ability to learn songs (Nottebohm and Nottebohm, 1978) which may be particularly suited for the co‐option of calls from other species through mimicry. Males of many species mimic song elements from other species (Baylis, 1982; Dobkin, 1979; Harcus, 1977; Hindmarsh, 1986; Robinson and Curtis, 1996) or sounds used in other contexts (Bostwick, 2000). Several experimental studies provide direct evidence that avian vocal mimicry is learned (Payne et al., 1998; Pepperberg et al., 1998) and mimicry is commonly used in mate attraction. There is strong evidence for co‐option for the ‘‘skraa’’ calls of bowerbirds (Borgia and Coleman, 2000), which are used in aggressive displays across the bowerbirds and are also used in the courtship displays of some species. The high level of similarity of skraa calls used in courtship and aggressive display suggests that one was co‐opted from the other. Mapping of these calls onto a molecular phylogeny of the bowerbirds shows the more limited distribution of skraa calls used in courtship which first occurred in the lineage leading to the Chlamydera bowerbirds in which all five species are the only bowerbirds to use these calls in courtship. This more restricted distribution as a courtship display element suggests that skraa calls were first used in aggressive display and that there was a later co‐option event before the diversification of the Chlamydera bowerbirds. This scenario is consistent with predictions of the war propaganda version of the preexisting traits model. Sexual selection models differ in their suitability in explaining learned mimicry. Because runaway (Lande, 1981; Mead and Arnold, 2004) and some versions of good genes (Eshel et al., 2000; Iwasa et al., 1991; Kokko

280

GERALD BORGIA

et al., 2002) depend on genetic correlations between male traits and female preferences, they are not well suited to explain learned displays where the particular form of the display is important in mate choice. Preexisting female preferences (Burley, 1985) and related models (e.g., sensory exploitation, Ryan and Rand, 1990; or sensory drive, Endler and Basolo, 1998) assume genetically determined female preferences that are side effects of selection for other traits. Searcy (1992) suggested that the evolution of starling calls was driven by preexisting female preferences for complex calls, but this was not supported in a phylogenetic analysis (Gray and Hagelin, 1996). Several studies show that mimicry enhances repertoire size that is used in mate selection by females (Catchpole, 1987; Hasselquist et al., 1996; Howard, 1974; Yasukawa, 1981; but see Forstmeier and Leisler, 2004; Hamao and Eda‐Fujiwara, 2004). Females might have preexisting preferences for variable and/or prolonged male song output that would cause them to favor males who enlarge their repertoire by mimicking songs of other species. Alternatively, females might have preexisting preferences that coincidentally favor songs from another species and the males of their species mimic these calls. In either case, females may be able to select for imprecise mimetic songs seen in some species (Dobkin, 1979; Searcy, 1992). Because there is no selection on preexisting preferences to allow preferences to precisely match male calls, they are unlikely to discriminate high‐frequency allospecific songs of multiple species mimicked by male bowerbirds (Loffredo and Borgia, 1986b; Coleman et al., submitted for publication), lyrebirds (Robinson and Curtis, 1996), and manakins (Trainer et al., 2002). Female preferences for precise mimetic songs could evolve for preexisting traits. In satin bowerbirds, Patricelli et al. (2002, 2004, 2005) found that the intense and threatening broadband ‘‘mechanical’’ portion of the male courtship song startles females causing rapid movements upward out of a crouching position that may lead to the courted female leaving the bower without copulation. Male satin bowerbirds may have interspersed calming melodic mimetic songs between intense mechanical elements to lower the threat to females during male courtship. The inclusion of this threat reducing mimicry should result in more complete courtships and in more copulations for the displaying male. Mate choice based on mimetic quality may have evolved later as females observing male courtship displays that incorporated mimicry began discriminating in favor of higher quality mimicry because it indicated male quality. Thus, females may have started to use these threat‐reducing mimetic display elements for a secondary function as indicators of male quality. Male mimetic abilities, although learned, may indicate heritable differences in neural circuitry that can affect individual survival and serve as an important good genes indicator (Leitner et al., 2001; Nowicki et al., 2000, 2002; Spencer et al., 2003).

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

281

F. PREEXISTING TRAITS EXPLAIN GENERAL INDICATOR MECHANISMS Several modes of display trait evolution that have attracted wide attention in the literature may have evolved as preexisting traits. Developmental stability, measured by fluctuating asymmetry, is suggested to be used by females to assay male genetic quality (Møller, 1988, 1989; Thornhill and Moller, 1998) and has become a controversial topic in sexual selection (Bjorksten et al., 2000; Lens et al., 2002; Markow, 1995; Simmons et al., 1999). Traits measured for symmetry, such as tail feathers, wings, and so on, typically have clear designs for other functions. Swaddle (1999) points out that in initial stages of trait evolution small symmetry differences may be undetectable, lending support to the hypothesis that differences in symmetry are best detected in already large preexisting traits that have been secondarily used (co‐opted) by females for assessment of male symmetry. Male barn swallow tail streamers that have been at the center of many discussions of fluctuating asymmetry are known to have an aerodynamic function (Norberg, 1994), and both length and symmetry are used in mate choice (Møller, 1988). Natural tail streamers increase aerodynamic function in barn swallows, and the addition of artificial streamers in the streamerless sand martins also increases their maneuverability. These and later experiments by Evans et al. (2004) on barn swallows led them to state ‘‘. . . variation in pre‐existing naturally selected traits may provide a starting point for the evolution of ornamental traits.’’ This is probably true for most other traits that are used for symmetry assessment (Møller, 1990) in that it is unlikely that any traits have evolved specifically to exhibit symmetry differences. Among the many traits mentioned including wings, tails, facial structure, breasts, and so on, all have already established functions before they were used for symmetry assessment. As such, symmetry indicating traits used in mate choice are preexisting traits co‐opted for a secondary function. Hamilton and Zuk (1982) suggested that bright plumage or integument color functions as an indicator of parasite resistance with only high‐fitness individuals able to make the brightest colors. This hypothesis has received mixed support (Borgia et al., 2004; Hamilton and Poulin, 1997). Plumage and integument color are unlikely to evolve de novo as an indicator because incipient colorful male displays are not likely to be sufficiently bright to allow females to effectively identify males with low levels of infection. This may be particularly true if large expensive displays are necessary for indicating differences in male quality (Folstad and Karter, 1992). Alternatively, co‐option of these preexisting colorful displays for a secondary indicator function may occur if differences in already existing colorful male plumage or integument displays are coincidentally inversely

282

GERALD BORGIA

correlated with parasite infection resistance, as might occur if sickly individuals are unable to make these displays. Bright plumage and integument color are used across a wide variety of species for many different functions, for example, status signaling (Rohwer and Rohwer, 1978), territorial display (Wolfenbarger, 1999), species identification signals (Alatalo et al., 1994), and so on. Females may then evolve to use these already existing traits as indicators of heritable male parasite resistance.

G. PREEXISTING TRAITS

AND

MULTIPLE DISPLAY ELEMENTS

There is growing evidence that multiple display elements are used in mate choice. Most focus on multiple display elements has focused on distinctly different traits used in display such as plumage and behavioral displays. However, studies suggest that multiple components of a single male signal are used in mate choice. For example, Scheuber et al. (2004) found that male chirp rate and carrier frequency are important in mate choice by females (Holzer et al., 2003). It seems unlikely that these preferences would evolve simultaneously. A more likely possibility is that females selected for one of these traits, and, while being exposed to males who varied in the other attribute, females who chose on that trait could increase their fitness further. Thus, the complexity of female choice can increase as females utilize (co‐opt) additional information from the signals they are already using in mate choice.

H. TIMING

OF

DISPLAY AND CO‐OPTION

It is not always clear when in the history of trait elaboration co‐option has occurred. A critical issue is to what extent co‐opted traits are elaborated before or after co‐option. In cases where nests are co‐opted for a secondary function, comparisons with related species that do not show evidence of co‐option may indicate if the co‐opted nest is or is not more elaborate than others. However, as has been suggested for African mouth‐breeding cichlids or sticklebacks, there may be enlargement before co‐option. If elaboration occurs after co‐option (Box I, Model 1), then other mechanisms may be needed to explain elaboration, and it suggests that even less than fully elaborated traits may be effective indicator traits and that co‐option may function as a starting point of a coevolutionary process leading to enhancement for an even better indicator function. But if commonly co‐ opted displays are already in a fully elaborated state (Box I, Model 2), then full elaboration is available for the indicator function and the need for other

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

283

BOX I THE PREEXISTING TRAIT MODEL OF SEXUAL SELECTION: DIFFERENT MODES OF OPERATION The co‐option of male display traits can occur in many different patterns with respect to when traits are elaborated. Evolution is opportunistic so there are many different ways that co‐option can occur. Co‐option could occur before or after elaboration and involve nonsexually selected traits or male–male aggressive traits that become preferred by females. It can also allow for selection for good genes or other male attributes. MODEL 1: FISHER’S INITIATOR HYPOTHESIS Naturally selected and relatively unelaborated traits are co‐opted by sexual selection and then elaborated. This process was first suggested by Fisher as a possible initiator for the start of the runaway sexual selection model. Because trait elaboration occurs after co‐option, it may be mostly coevolutionary. It remains unclear how common this kind of co‐option is. Evidence for this kind of evolution might include traits shared among a set of related species that show evidence of co‐option but which show different degrees of elaboration resulting from different strengths of female preferences. A possible example is sex combs of Drosophila studied by Polak et al. (2004) in which they show that in a Queensland, Australia population there is a female preference for enlargement of the second comb; since these combs are widely distributed in Drosophila, it appears that this role in mate choice is a secondary function and has driven the enlargement of these combs. Post‐co‐option enlargement may not be great. For example, Rowe et al. (2001) estimate that 9–20% of barn swallow tail streamer length occurs as a result of sexual selection which likely occurred after streamers had evolved for aerodynamic function. MODEL 2: PREEXISTING TRAIT Naturally selected and already elaborated traits are co‐opted by sexual selection. Traits are already present and indicate differences in male quality. Co‐option occurs when females evolve a preference for males showing versions of the trait which indicate that they are a high‐quality mate. This should work most easily for traits present where females are being courted, such as at nests or bowers, and where parts of the male phenotype, such as colored plumage, are brought with him to the courtship site. Model 2a: Aggressive Preexisting Traits Fully elaborated traits used in male–male competition (or more generally for aggression) are co‐opted for female choice. This is the best‐described model for the evolution of display, and evidence to support it comes from the very common occurrence of traits of dual function. For example, males on leks may fight for position and females evolve to use aggressive displays in mate choice that occur in conjunction with these fights (Alexander, 1975). What remains unclear is the proportion of cases in which aggressive display was co‐opted for courtship display and vice versa. The good genes traits from courtship displays could also be co‐opted for signaling in aggressive displays; however, there are several reasons to suspect that the direction of most of these co‐options has involved aggressive displays co‐opted for courtship. First, the costs of large and otherwise expensive display would be less likely to evolve under female choice because

284

GERALD BORGIA

the costs to sons have to be subtracted from the benefits to them of good genes, whereas the benefits from aggression are immediate and can justify a higher cost for aggressive displays. Second, co‐opted traits functioning as courtship traits often have the appearance of aggressive displays, for example, broadband vocalizations (Loffredo and Borgia, 1986a), which are not typical of courtship traits that show no evidence of co‐ option. Third, in one study in which the direction of trait evolution has been tested for the aggressive trait function appears to have occurred first (Borgia and Coleman, 2000). So while the direction of evolution remains to be resolved in most cases, there is suggestive evidence that aggressive traits will more often be co‐opted for use in courtship than vice versa. Model 2b: Nonaggressive Preexisting Traits Some co‐opted traits that indicate male quality did not evolve in the context of aggression. These include morphological, electrical, chemical, or behavioral traits that evolve for a variety of functions such as nests, aerodynamic tail streamers, protective bowers, antipredator signals, or chemical by‐products that are co‐opted by females to assess male genetic quality. Indicators of male symmetry may commonly evolve from preexisting morphological traits.

sexual selection models to explain elaboration is reduced. While it is clear that co‐option contributes importantly toward the use of already enlarged traits as indicators, it remains to be determined what proportion of all elaborated displays involve co‐option and what proportion of these are already fully elaborated or have required further elaboration to reach current levels of exaggeration. VII. IMPLICATIONS

AND

CONCLUSIONS

The occurrence of co‐option of already elaborated traits for use in sexual display offers to radically change our view of sexual display trait evolution. It raises the possibility that relatively simple co‐option models may be sufficient to explain elaborated display traits in many cases rather than more controversial coevolutionary models. A. CO‐OPTION IS

A

COMMON SOURCE

OF

SEXUAL DISPLAY TRAITS

This survey suggests that sexual display traits from a large number of taxa are the products of co‐option. Many traits have dual functions commonly with one but not the other involved in mate choice. In some cases, it is clear that the sexual display function is derived, as suggested by the preexisting traits hypothesis, but in others the order of origin of traits is unclear. Phylogenetic mapping can help resolve some of these cases. Also, where co‐opted traits are used for sexual display, it is sometimes unclear if these sexual displays function as good genes indicators or for some other function.

285

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

This will require more detailed studies of how these traits are used in sexual selection. Even with this limited information, the numerous cases of co‐ option of traits for use as indicators suggest that this is an important mechanism for the evolution of elaborate male traits in sexual display. Development of the preexisting traits model should lead to more detailed studies directed at understanding the role of co‐option in shaping good genes indicator displays, indicators of other aspects of male quality, and sexual displays generally. B. CO‐OPTION OF INDICATOR TRAITS NEED NOT BE RESTRICTED AGGRESSIVE DISPLAYS

TO

The war propaganda/armament–ornament model has been the focus of previous discussions of preexisting traits being co‐opted for use as indicator traits. This survey of male display traits indicates that co‐option is common and is not restricted to, but certainly includes, aggressive displays. Any trait that shows differences in male performance correlated with male quality that is accessible to females choosing mates can be co‐opted to function as a male quality indicator. Thus, differences in preexisting male traits like nests, male acoustic calls, electrical organ discharges, light flashes, plumage colors, symmetry differences can be used by females as indicators of male quality. C. AT WHAT STAGE IN MALE TRAIT ELABORATION DOES CO‐OPTION OCCUR? Co‐option can occur early or late in the process of elaboration of traits that evolve indicator functions. An important role for pre‐co‐option elaboration suggests a dramatic change in how we view sexual selection and provides an interesting solution to some difficult aspects of sexual display trait evolution. If most elaboration occurs pre‐co‐option, then the processes that build traits before co‐option are critical for explaining elaboration. A prominent role for co‐option of elaborated naturally selected traits for use in sexual display breaks down the separation of natural and sexual selection as causes of elaborated male displays. With co‐option, natural selection is no longer just the brake on display trait evolution as suggested by Lande (1981) but may have a central role in trait evolution. Traits evolved by natural selection (and by male–male sexual competition) may be co‐opted because of the evolution of a female preference. Sexual selection can then transform these naturally selected male traits to function as an indicator of male quality. Initial elaboration of male traits could also occur as a result of male–male competition which may also be co‐opted to function as indicator

286

GERALD BORGIA

traits (Borgia, 1979), suggesting that it will be useful to determine the relative importance of sexual and natural selection in this pre‐co‐option phase of trait evolution. Alternatively, if elaboration occurs post‐co‐option, then other sexual selection processes are needed to explain this additional elaboration. Nonetheless, co‐option can account for initiation of these display traits and to the degree that elaboration occurs may help bypass the difficult initial stages of trait evolution associated with other models. Collecting information to resolve the question of when co‐option occurs as traits are elaborated is critical to understanding the role of co‐option in evolution of elaborated displays. D. CO‐OPTION OF PREEXISTING TRAITS MAY SOLVE HOW COSTLY DISPLAY TRAITS EVOLVE

THE

PROBLEM

OF

Costly displays are suggested to evolve because they are more resistant to cheating, but in incipient stages in the evolution of these traits, costs are expected to be small and thus ineffective in preventing cheating. Thus, if honesty is dependent on near full elaboration of traits, then costliness of display traits cannot explain why they initially evolve. I argued earlier that cost may not be necessary to insure honest advertising, although under some conditions it may be important as one mechanism for insuring honest advertising. The co‐option of already elaborated expensive traits for use as indicators of male quality in sexual display may explain the existence of costly indicator traits. Because their costs are associated with the initial trait function, indicator functions that evolve secondarily may bring no added cost and are thus not limited by this constraint but may benefit from the existing trait cost to limit the ability of other inferior males to cheat. Thus, when traits are already enlarged when co‐opted, that trait can take on an indicator function without taking on additional costs and the honesty of the indicator insured by the costs needed to originally develop the trait.

E. HOW IMPORTANT ARE CURRENTLY POPULAR MATE CHOICE MODELS DISPLAY TRAIT EVOLUTION?

IN

Current reviews assume that genetic correlation‐based models are the only means of explaining good genes preferences. As a result, the growing evidence that females choose males for good genes has been taken as support for these genetic correlation‐based models. Preexisting traits offer an alternative explanation for the occurrence of good genes selection.

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

287

Evidence in this review indicating the widespread occurrence of co‐option in the evolution of sexual display traits suggests that mating preferences based on genetic correlations may be of limited significance. Preexisting female preferences have been offered as an alternative to genetic correlation‐based models. These preferences are suggested to evolve as side effects and are not specifically shaped by selection to enhance the selection of quality mates by females and should only infrequently produce a successful preference. The chances of a preexisting preference becoming successful are greatest for those similar to already functioning preferences. Thus, preexisting preferences at best may cause the evolution of successful mate selection patterns that are not much different from current patterns. They are unlikely to contribute to rapid divergence characteristic of sexual selection because of the limited supply of hidden female preferences established in populations and the likelihood that most present will fail when placed in competition with already selected alternatives. Burley and Symanski’s claim (1995) that preexisting preferences may give rise to coevolutionary good genes traits seems improbable because there is no reason to expect that preexisting preferences would be biased toward selecting good genes in males. Preexisting preferences are most consistent with very simple patterns of mate selection and are not suited for explaining complex and highly integrated sets of mating preferences that are now being found.

VIII. SUMMARY The evolution of highly elaborated male sexual display traits remains an important and controversial issue in evolutionary and behavioral biology. Nearly all discussion of the evolution of these traits has focused on runaway, preexisting preference and coevolutionary good genes models. Here I evaluate each of these models, considering growing empirical support for good genes traits, and analyze the difficulties of currently popular versions of these models that limit their suitability as explanations for the evolution of elaborated male sexual displays. Co‐option of preexisting traits provides an important alternative that can explain the evolution of good genes indicator traits with fewer limiting requirements, such as genetic correlations, between male traits and female preferences. The current preexisting trait model focuses on the co‐option of aggressive traits for use as good genes indicators. I present a broadened version of this model which considers that females may evolve to use a wide array of preexisting male traits as indicators of differences in male genetic quality. This approach fits with the current trend in evolutionary biology to view co‐option as critical in the evolution of many complex traits.

288

GERALD BORGIA

Co‐option of preexisting traits emphasizes contributions from both natural and sexual selection in shaping traits used in elaborated male sexual display. This review of elaborated male display traits suggests that co‐option of preexisting male traits for mate assessment is very common and has been important in mate choice and the evolution of elaborated male sexual display. The preexisting traits models must be included in any comprehensive discussion of the evolution of elaborated male display traits.

Acknowledgments Thanks to Leo Borgia, Jane Brockmann, Seth Coleman, Brian Coyle, Jason Keagy, Jan Lauridsen, Carrie Long, Marc Naguib, Todd Oakley, Sheila Reynolds, Peter Slater, Kerry Shaw, and Claus Wedekind for helpful discussions and comments on this chapter. The NSF, Animal Behavior and Systematics Programs (USA), The National Geographic Society, and Universities of Wollongong, Melbourne, and Maryland, and James Cook University have supported this work. The federal and state governments of Australia, Papua New Guinea, and Indonesia have granted me permits and together with local landowners have allowed me access to their lands for which I am grateful. Numerous graduate students, volunteers, collaborators, and other kind individuals have made substantial contributions to this work. My desire to understand why bowerbirds build and decorate bowers inspired this work.

REFERENCES Alatalo, R. V., Gustafsson, L., and Lundberg, A. (1994). Male coloration and species recognition in sympatric flycatchers. Proc. R. Soc. Lond. B 256, 113–118. Alatalo, R. V., Kotiaho, J., Mappes, J., and Parri, S. (1998). Mate choice for offspring performance: Major benefits or minor costs? Proc. R. Soc. Lond. B 265, 2297–2301. Alcock, J., and Bailey, W. J. (1995). Acoustical communication and the mating system of the Australian whistling moth Hecatesia exultans (Noctuidae: Agaristinae). J. Zool. 237, 337–352. Alcock, J., Gwynne, D. T., and Dadour, I. R. (1989). Acoustic signalling, territoriality and mating in whistling moths, Hecatesiathyridion (Agaristidae). J. Insect Behav. 2, 27–37. Alexander, R. D. (1975). Natural selection and specialized chorusing behavior in acoustical insects. In ‘‘Insects, Science, and Society’’ (D. Pimental, Ed.), pp. 35–77. Academic Press, New York. Andersson, M. (1986). Evolution of condition‐dependent sex ornaments and mating preferences: Sexual selection based on viability differences. Evolution 40, 804–816. Andersson, M. (1994). ‘‘Sexual Selection.’’ Princeton University Press, Princeton, NJ. Andersson, M., and Iwasa, Y. (1996). Sexual selection. Trends Ecol. Evol. 11, 53–58. Aparicio, J. M., Bonal, R., and Cordero, P. J. (2003). Evolution of the structure of tail feathers: Implications for the theory of sexual selection. Evolution 57, 397–405. Arnold, S. J. (1983). Sexual selection: The interface of theory and empiricism. In ‘‘Mate Choice’’ (P. P. G. Bateson, Ed.), pp. 67–107. Cambridge University Press, Cambridge.

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

289

Arnold, S. J. (1987). Quantitative genetic models of sexual selection: A review. Experientia Suppl. 55, 283–315. Autumn, K., Ryan, M. J., and Wake, D. B. (2002). Integrating historical and mechanistic biology enhances the study of adaptation. Q. Rev. Biol. 77, 383–408. Backwell, P. R. Y., Christy, J. H., Telford, S. R., Jennions, M. D., and Passmore, N. I. (2000). Dishonest signaling in a Fiddler crab. Proc. R. Soc. Lond. B 267, 719–724. Badyaev, A. V., and Duckworth, R. A. (2003). Context‐dependent sexual advertisement: Plasticity in development of sexual ornamentation throughout the lifetime of a passerine bird. J. Evol. Biol. 16, 1065–1076. Baker, R. H., and Wilkinson, G. S. (2001). Phylogenetic analysis of eye stalk allometry and sexual dimorphism in stalk‐eyed flies (Diopsidae). Evolution 55, 1373–1385. Bakker, T. C. M. (1993). Positive genetic correlation between female preference and preferred male ornament in sticklebacks. Nature 363, 255–257. Bakker, T. C. M., and Pomiankowski, A. (1995). The genetic basis of female mate preferences. J. Evol. Biol. 8, 129–171. Barber, I., Narin, D., and Hungerford, F. (2001). Nests as ornaments: Revealing construction by male sticklebacks. Behav. Ecol. 12, 390–396. Barbosa, A., and Møller, A. P. (1999). Aerodynamic costs of long tails in male barn swallows Hirundo rustica and the evolution of sexual size dimorphism. Behav. Ecol. 10, 128–135. Barrette, C., and Vandal, D. (1986). Social rank, dominance, antler size, and access to food in snow‐bound wild woodland caribou. Behaviour 97, 118–146. Barrette, C., and Vandal, D. (1990). Sparring, relative antler size, and assessment in male caribou. Behav. Ecol. Sociobiol. 26, 383–387. Barton, N., and Turelli, M. (1991). Natural and sexual selection on many loci. Genetics 127, 229–255. Basolo, A. L. (1990). Female preference predates the evolution of the sword in swordtail fish. Science 250, 808–810. Basolo, A. L. (1995a). Phylogenetic evidence for the role of a pre‐existing bias in sexual selection. Proc. R. Soc. Lond. B 259, 307–311. Basolo, A. L. (1995b). A further examination of a pre‐existing bias favouring a sword in the genus. Xiphophorus. Anim. Behav. 50, 365–375. Basolo, A. L. (2002). Congruence between the sexes in preexisting receiver responses. Behav. Ecol. 13, 832–837. Basolo, A. L., and Endler, J. A. (1995). Sensory biases and the evolution of sensory systems. Trends Ecol. Evol. 10, 489. Baylis, J. R. (1982). Avian vocal mimicry: Its function and evolution. In ‘‘Acoustic Communication in Birds. Volume 2, Song Learning and its Consequences’’ (D. E. Kroodsma and E. H. Miller, Eds.), pp. 51–83. Academic Press, New York. Beebe, W. (1929). ‘‘Pheasants, Their Lives and Homes.’’ Doubleday, Page and Co, New York. Beecher, M. D., and Brenowitz, E. A. (2005). Functional aspects of song learning in songbirds. Trends Ecol. Evol. 20, 143–148. Berglund, A., and Rosenqvist, G. (2001). Male pipefish prefer dominant over attractive females. Behav. Ecol. 12, 402–406. Berglund, A., Bisazza, A., and Pilastro, A. (1996). Armaments and ornaments: An evolutionary explanation of traits of dual utility. Biol. J. Linn. Soc. Lond. 58, 385–399. Bisazza, A., Marconato, A., and Marin, G. (1989). Male competition and female choice in Padogobius martensi (Pisces, Gobiidae). Anim. Behav. 38, 406–413.

290

GERALD BORGIA

Bjorksten, T., David, P., Pomiankowski, A., and Fowler, K. (2000). Fluctuating asymmetry of sexual and nonsexual traits in stalk‐eyed flies: A poor indicator of developmental stress and genetic quality. J. Evol. Biol. 13, 89–97. Blaustein, A. R., Yoshikawa, T., Asoh, K., and Walls, S. C. (1993). Ontogenetic shifts in tadpole kin recognition: Loss of signal and perception. Anim. Behav. 46, 525–538. Blows, M. W. (1999). Evolution of the genetic covariance between male and female components of mate recognition: An experimental test. Proc. R. Soc. Lond. B 266, 2169–2174. Borgia, G. (1979). Sexual selection and the evolution of mating systems. In ‘‘Sexual Selection and Reproductive Competition’’ (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 male controlled system. Anim. Behav. 29, 71–80. Borgia, G. (1982). Experimental changes in resource structure and male density: Size‐related differences in mating success among male Scatophaga stercoraria. Evolution 36, 307–315. Borgia, G. (1985a). Bowers as markers of male quality. Test of a hypothesis. Anim. Behav. 33, 266–271. Borgia, G. (1985b). Bower destruction and sexual competition in the satin bowerbird (PtiloPtilonorhynchus violaceus). Behav. Ecol. Sociobiol. 18, 91–100. Borgia, G. (1993). The cost of display in the non‐resource based mating system of the satin bowerbird. Am. Nat. 141, 729–743. Borgia, G. (1995). Why do bowerbirds build bowers? Am. Scientist 83, 542–547. Borgia, G., and Coleman, S. W. (2000). Co‐option of male courtship signals from aggressive display in bowerbirds. Proc. R. Soc. Lond. B 267, 1735–1740. Borgia, G., and Collis, K. (1990). Parasites and bright male plumage in the satin bowerbird (Ptilonorhynchus violaceus). Am. Zool. 30, 279–285. Borgia, G., and Mueller, U. (1992). Bower destruction, decoration stealing, and female choice in the spotted bowerbird (Chlamydera maculata). Emu 92, 11–18. Borgia, G., and Presgraves, D. C. (1998). Coevolution of elaborated male display traits in the spotted bowerbird: An experimental test of the threat reduction hypothesis. Anim. Behav. 56, 1121–1128. Borgia, G., Pruett‐Jones, S., and Pruett‐Jones, M. (1985). Bowers as markers of male quality. Zeitschr. Tierpsychol. 67, 225–236. Borgia, G., Egeth, J., Uy, A. C., and Patricelli, G. L. (2004). Juvenile infection and male display: Testing the bright male hypothesis across individual life histories. Behav. Ecol. 15, 722–728. Bostwick, K. S. (2000). Display behaviors, mechanical sounds, and evolutionary relationships of the club‐winged manakin (Machaeropterus deliciosus). Auk 117, 465–478. Boughman, J. W. (2001). Divergent sexual selection enhances reproductive isolation in sticklebacks. Nature 411, 944–948. Borowsky, R. L., Mcclelland, M., Cheng, R., and Welsh, J. (1995). Arbitrarily primed DNA fingerprinting for phylogenetic reconstruction in vertebrates: The Xiphophorus model. Mol. Biol. Evol. 12, 1022–1032. Bradbury, J. W., and Veherencamp, S. (1993). ‘‘Principles of Animal Communication.’’ Sinauer Associates, Sunderland, MA. Branham, M. A., and Greenfield, M. D. (1996). Flashing males win mate success. Nature 381, 745–746. Branham, M. A., and Wenzel, J. W. (2000). The evolution of bioluminescence in cantharoids (Coleoptera: elateroidea). Florida Entomol. 84, 566–587.

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

291

Breden, F., and Stoner, G. (1987). Male predation risk determines female preference in the Trinidad guppy. Nature 329, 831–833. Breden, F., Geherardt, C., and Butlin, R. K. (1994). Female choice and genetic correlations. Trends Ecol. Evol. 9, 343. Brenowitz, E. A., and Beecher, M. D. (2005). Song learning in birds: Diversity and plasticity, opportunities and challenges. Trends Ecol. Evol. 28, 127–132. Burley, N. (1985). The organization of behavior and the evolution of sexually selected traits. In ‘‘Avian Monogamy’’ (P. A. Gowaty and D. W. Mock, Eds.), pp. 22–44. American Ornithologists’ Union Publication 37. McLean, VA, USA. Burley, N. T., and Symanski, R. (1995). ‘‘A taste for the beautiful’’: Latent aesthetic mate preferences for white crests in two species of Australian grassfinches. Am. Nat. 152, 792–802. Candolin, U. (1999). The relationship between signal quality and physical condition: Is sexual signalling honest in the three‐spined stickleback? Anim. Behav. 58, 1261–1267. Candolin, U. (2003). The use of multiple cues in mate choice. Biol. Rev. 78, 575–595. Catchpole, C. K. (1987). Bird song, sexual selection and female choice. Trends Ecol. Evol. 2, 94–97. Christy, J. H., Baum, K., and Backwell, P. (2003). Attractiveness of sand hoods built courting male fiddler crabs, Uca musica: Test of a sensory trap hypothesis. Anim. Behav. 66, 89–94. Clutton‐Brock, T. H., Guinness, F. E., and Albon, S. D. (1982). ‘‘Red Deer: Behavior and Ecology of Two Sexes.’’ University of Chicago Press, Chicago. Cohen, A. C., and Morin, J. G. (2003). Sexual morphology, reproduction and the evolution of bioluminescence in Ostracoda. In ‘‘Bridging the Gap: Trends in the Ostracod Biological and Geological Sciences’’ (L. E. Park and A. J. Smith, Eds.), Vol. 94, pp. 37–70. The Paleontological Society Papers. Lancaster, PA, USA. Coleman, S. W., Patricelli, G. L., and Borgia, G. (2004). Variable female preferences drive complex male displays. Nature 428, 742–745. Collias, N. E., and Victoria, J. K. (1978). Nest and mate selection in the village weaverbird (Ploceus cucullatus). Anim. Behav. 26, 470–479. Connor, W. E. (1999). ‘Un Chant D’appel Amoureux’: Acoustic communication in moth. J. Exp. Biol. 202, 1711–1723. Cox, C. R., and Le Boeuf, B. J. (1977). Female incitation of male competition: A mechanism in sexual selection. Am. Nat. 111, 317–335. Cratsley, C. K., and Lewis, S. M. (2003). Female preference for male courtship flashes in Photinus ignitus fireflies. Behav. Ecol. 14, 135–140. Cuervo, J. J., and Møller, A. P. (1999). Phenotypic variation and fluctuating asymmetry in sexually dimorphic feather ornaments in relation to sex and mating system. Biol. J. Linn. Soc. 68, 505–529. Cunningham, C. W., Omland, K. E., and Oakley, T. H. (1998). Reconstructing ancestral character states: A critical reappraisal. Trends Ecol. Evol. 13, 361–366. Curtis, C. C., and Stoddard, P. K. (2003). Active mate choice in an electric fish, Brachyhypopomus pinnicaudatus. Anim. Behav. 66, 329–336. Dahm, K. H., Meyer, D., Finn, W. E., Reinhold, V., and Ro¨ller, H. (1971). The olfactory and auditory mediated sex attraction in Achroia grisella (Fabr.). Naturwissenschaften 58, 265–266. Danley, P. D., and Kocher, T. D. (2001). Speciation in rapidly diverging systems: Lessons from Lake Malawi. Mol. Ecol. 10, 1075–1086.

292

GERALD BORGIA

Darwin, C. (1871). ‘‘The Descent of Man, and Selection in Relation to Sex.’’ John Murray, London. David, P., Bjorksten, T., Fowler, K., and Pomiankowski, A. (2000). Condition‐dependent signaling of genetic variation in stalk‐eyed flies. Nature 406, 186–187. Davis, G. W. F., and O’Donald, P. (1976). Sexual selection for a handicap. A critical analysis of Zahavi’s model. J. Theor. Biol. 57, 345–354. Day, J. C., Tisi, L. C., and Bailey, M. J. (2004). Evolution of beetle bioluminescence: The origin of beetle luciferin. Luminescence 19, 8–20. Ditchkoff, S. S., Lochmiller, R. R., Masters, R. E., Hoofer, S. E., and Van Den Bussche, R. A. (2001). Major‐histocompatibility‐complex‐associated variation in secondary sexual traits of white‐tailed deer (Odocoileus virginianus): Evidence for good‐genes advertisement. Evolution 55, 616–625. Dobkin, D. S. (1979). Functional and evolutionary relationships of vocal copying phenomenon in birds. Z. Tierpsychol. 50, 348–363. Doutrelant, C., and McGregor, P. K. (2000). Eavesdropping and mate choice in female fighting fish behaviour. Behaviour 137, 1655–1669. Ellsworth, D. I., Honeycutt, R. I., and Silvy, N. J. (1995). Phylogenetic relationships among North American Grouse inferred from restriction endonuclease analysis of mitochondrial DNA. Condor 97, 492–502. Endler, J. A. (1992). Signals, signal conditions, and the direction of evolution. Am. Nat. 139 (Suppl.), S125–S153. Endler, J. A., and Basolo, A. (1998). Sensory ecology, receiver biases and sexual selection. Trends Ecol. Evol. 13, 415–420. Endler, J. A., Westcott, D. A., Madden, J. R., and Robson, T. (2005). Processing illuminates signal evolution animal visual systems and the evolution of color patterns: Sensory processing illuminates signal evolution. Evolution 59, 1795–1818. Eshel, I., Volovik, I., and Sansone, E. (2000). On Fisher–Zahavi’s handicapped sexy son. Evol. Ecol. Res. 2, 509–523. Evans, M. R., Bro‐Jorgenson, J., and Johnstone, R. A. (2004). Uninformative exaggeration of male sexual ornaments in barn swallows. Abstract. International Society for Behavioral Ecology, Jyva¨skyla¨, Finland. Falconer, D., and Mackay, T. (1996). ‘‘Introduction to Quantitative Genetics,’’ 4th edn. Prentice Hall, New York. Farr, J. A., and Travis, J. (1986). Fertility advertisement by female sailfin mollies, Poecilia latipinna (Pisces: Poeciliidae). Copeia 1986, 467–472. Fisher, R. A. (1915). The evolution of sexual preference. Eug. Rev. 7, 184–192. Fisher, R. A. (1930). ‘‘The Genetical Theory of Natural Selection.’’ Clarendon Press, Oxford. Fisher, R. A. (1954). ‘‘The Genetical Theory of Natural Selection,’’ 2nd edn. Clarendon Press, Oxford. Folstad, I., and Karter, A. J. (1992). Parasites, bright males and the immunocompetence handicap. Am. Nat. 139, 603–622. Forstmeier, W., and Leisler, B. (2004). Repertoire size, sexual selection, and offspring viability in the great reed warbler: Changing patterns in space and time. Behav. Ecol. 15, 555–563. Fullard, J. H., Simmons, J. A., and Saillant, P. A. (1994). Jamming bat echolocation: The dogbane tiger moth Cycnia tenera times its clicks to the terminal attack calls of the big brown bat Eptesicus fuscus. J. Exp. Biol. 194, 285–298. Fuller, R. C., Houle, D., and Travis, J. (2005). Sensory bias as an explanation for the evolution of mate preferences. Am. Nat. 166, 437–446.

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

293

Gabor, C. R. (1999). Association patterns of sailfin mollies (Poecilia latipinna): Alternative hypotheses. Behav. Ecol. Sociobiol. 46, 333–340. Gadagkar, R. (2003). Is the peacock merely beautiful or also honest? Curr. Sci. 85, 1012–1020. Geist, V. (1971). ‘‘Mountain Sheep: A Study of Behavior and Evolution.’’ University of Chicago Press, Chicago, IL. Getty, T. (1998). Handicap signaling: When viability and fecundity do not add up. Anim. Behav. 56, 127–130. Ghiradella, H. (1998). Anatomy of light production: The firefly lantern. In ‘‘Microscopic Anatomy of Invertebrates’’ (M. Locke and F. W. Harrison, Eds.), Vol. 11A, pp. 363–381. Wiley and Sons, New York. Gibson, R. M., Bradbury, J. W., and Vehrencamp, S. L. (1991). Mate choice in lekking sage grouse revisited: The roles of vocal display, female site fidelity, and copying. Behav. Ecol. 2, 165–180. Gilburn, A. S., Foster, S., and Day, T. H. (1993). Genetic correlation between a female mating preference and the preferred male character in seaweed flies Coelopa frigida. Evolution 47, 1788–1795. Gilliard, E. T. (1969). ‘‘Birds of Paradise and Bower Birds.’’ Weidenfeld & Nicolson, London. Go¨ransson, G., von Schantz, T., Fro¨berg, I., Helge´e, A., and Wittzell, H. (1990). Male characteristics, viability and harem size in the pheasant, Phasianus colchicus. Anim. Behav. 40, 89–104. Gosling, L. M., and Roberts, S. C. (2001). Scent‐marking by male mammals: Cheat‐proof signals to competitors and mates. Adv. Study Behav. 30, 169–217. Grafen, A. (1990). Sexual selection unhandicapped by the Fisher process. J. Theor. Biol. 144, 473–516. Grafen, A. (1991). Modelling in behavioural ecology. In ‘‘Behavioural Ecology’’ (J. R. Krebs and N. B. Davies, Eds.), 3rd edn., pp. 5–31. Blackwell Scientific Publications, Oxford. Grant, J. W. A., and Green, L. D. (1996). Mate copying versus preference for actively courting males by female Japanese medaka (Oryzias latipes). Behav. Ecol. 7, 165–167. Gray, D. A., and Cade, W. H. (1999). Correlated‐response‐to‐selection experiments designed to test for a genetic correlation between female preferences and male traits yield biased results. Anim. Behav. 58, 1325–1327. Gray, D. A., and Eckhardt, G. (2001). Is cricket courtship song condition dependent? Anim. Behav. 62, 871–877. Gray, D. A., and Hagelin, J. H. (1996). Song repertoires and sensory exploitation: Reconsidering the case of the common grackle. Anim. Behav. 52, 795–800. Greenfield, M. D., and Colfelt, J. A. (1983). Reproductive behaviour of the lesser wax moth, Achroia grisella (Pyralidae: Galleriinae): Signalling, pair formation, male interactions and mate guarding. Behaviour 84, 287–315. Guilford, T. (1995). Animal signals: All honesty and light? Trends Ecol. Evol. 10, 100–101. Haberli, M. A., and Aeschlimann, P. B. (2004). Male traits influence odor‐based mate choice in the three‐spined stickleback. J. Fish Biol. 64(3), 702–710. Hagedorn, M., and Zelick, R. (1989). Relative dominance among males is expressed in the electric organ discharge characterics of a weakly electric fish. Anim. Behav. 38, 1409–1413. Hall, D. W., Kirkpatrick, M., and West, B. (2000). Runaway sexual selection when female preferences are directly selected. Evolution 54, 1862–1869. Hall, M., Lindholm, A. K., and Brooks, R. (2004). Direct selection on male attractiveness and female preference fails to produce a response. BMC Evol. Biol. 4, 1.

294

GERALD BORGIA

Hamao, S., and Eda‐Fujiwara, H. (2004). Vocal mimicry by the Black‐browed Reed Warbler, Acrocephalus bistrigiceps: Objective identification of mimetic sounds. Ibis 146, 61–68. Hamilton, W. D., and Zuk, M. (1982). Heritable true fitness and bright birds: A role for parasites? Science 218, 384–387. Hamilton, W. J., and Poulin, R. (1997). The Hamilton and Zuk hypothesis revisited: A meta‐ analytical approach. Behaviour 134, 299–320. Harcus, J. L. (1977). The functions of mimicry in the vocal behaviour of the Chorister Robin. Z. Tierpsychol. 44, 178–193. 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 381, 229–232. Hebets, E. A. (2003). Subadult experience influences adult mate choice in an arthropod: Exposed female wolf spiders prefer males of a familiar phenotype. Proc. Natl. Acad. Sci. USA 100, 13390–13395. Hein, W. K., Westneat, D. F., and Poston, J. P. (2003). Sex of opponent influences response to a potential status signal in house sparrows. Anim. Behav. 65, 1211–1221. Heisler, I. L. (1985). Quantitative genetic models of female choice based on ‘‘arbitrary’’ male characters. Heredity 55, 187–198. Heller, K. G., and Achmann, R. (1993). The ultrasonic song of the moth, Amyna natalis (Lepidoptera: Noctuidae: Acontiinae). Bioacoustics 5, 89–97. Herring, P. J. (2000). Bioluminescent signals and the role of reflectors. J. Opt. A: Pure Appl. Opt. 2, R29–R38. Hill, G. E. (1991). Plumage coloration is a sexually selected indicator of male quality. Nature 350, 337–339. Hindmarsh, A. M. (1986). The functional significance of vocal mimicry in song. Behaviour 99, 87–100. Ho¨glund, J., and Alatalo, R. V. (1995). ‘‘Leks.’’ Princeton University Press, Princeton. Ho¨glund, J., Alatalo, R. V., Gibson, R. M., and Lundberg, A. (1995). Mate‐choice copying in black grouse. Anim. Behav. 49, 1627–1633. Hoikkala, A., Aspi, J., and Suvanto, L. (1998). Male courtship song frequency as an indicator of male genetic quality in an insect species, Drosophila montana. Proc. R. Soc. Lond. B 265, 503–508. Holland, P. W., Garcia‐Fernandez, J., Williams, N. A., and Sidow, A. (1994). Gene duplications and the origins of vertebrate development. Dev. Suppl. 1994, 125–133. Holzer, B., Jacot, A., and Brinkhof, M. (2003). Condition‐dependent signaling affects male sexual attractiveness in field crickets. Gryllus campestris. Behav. Ecol. 14, 353–359. Houde, A. E. (1994). Effect of artificial selection on male colour patterns on mating preference in female guppies. Proc. R. Soc. Lond. B 256, 125–130. Houle, D., and Kondrashov, A. S. (2002). Coevolution of costly mate choice and condition‐ dependent display of good genes. Proc. R. Soc. Lond. B 269, 97–104. Howard, R. D. (1974). The influence of sexual selection and interspecific competition on Mockingbird song (Mimus polyglottos). Evolution 28, 428–438. Howard, R. D., and Young, J. R. (1998). Individual variation in male vocal traits and female preferences in Bufo americanus. Anim. Behav. 55, 1165–1179. Hristov, N. I., and Conner, W. E. (2005). Sound strategy: Acoustic aposematism in the bat–tiger moth arms race. Naturwissenschaften 92, 164–169. Hughes, K. A., Du, L., Rodd, F. H., and Reznick, D. N. (1999). Familiarity leads to female mate preference for novel males in the guppy, Poecilia reticulata. Anim. Behav. 58, 907–916.

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

295

Hunt, J., Brooks, R., Jennions, M. D., Smith, M. J., Bentsen, C. L., and Bussiere, L. F. (2004). High‐quality male field crickets invest heavily in sexual display but die young. Nature 432, 1024–1027. Hurd, P. L. (2004). Conventional displays: Evidence for socially mediated costs of threat displays in a lizard. Aggress. Behav. 30, 326–431. Hyatt, G. W., and Salmon, M. (1978). Combat in the fiddler crabs, Uca pugilator and U. pugnax: A quantitative analysis. Behaviour 65, 182–211. Isyengar, V. K., Reeve, H. K., and Eisner, T. (2002). Paternal inheritance of a female moth’s mating preference: Consequences for sexual selection. Nature 419, 830–832. Isvaran, K. (2004). Scent marking in a lekking antelope. Its role in female choice and territory maintenance. ISBE 2004 Abstracts, p. 102. Iwasa, Y., and Pomiankowski, A. (1995). A continual change in mate preferences. Nature 377, 420–422. Iwasa, Y., and Pomiankowski, A. (1999). A good parent and good genes models of handicap evolution. J. Theor. Biol. 200, 97–109. Iwasa, Y., Pomiankowski, A., and Nee, S. (1991). The evolution of costly male preferences II. The ‘Handicap’ principle. Evolution 45, 1431–1442. Jang, Y. (1997). Evolution and genetics of mate preference in an ultrasonic pyralid moth. PhD Dissertation, University of Kansas, Lawrence. Jang, Y., and Greenfield, M. D. (1996). Ultrasonic communication and sexual selection in wax moths: Female choice based on energy and asynchrony of male signals. Anim. Behav. 51, 1095–1106. Jennions, M. D., and Backwell, P. R. Y. (1996). Residency and size affect fight duration and outcome in the fiddler crab, Uca annulipes. Biol. J. Linn. Soc. 57, 293–306. Jennions, M. D., and Brooks, R. (2001). A sensory of history. Trends Ecol. Evol. 16, 113–115. Jennions, M. D., and Petrie, M. (1997). Variation in mate choice and mating preferences: A review of causes and consequences. Biol. Rev. 72, 283–327. Jennions, M. D., Moller, A. P., and Petrie, M. (2001). Sexually selected traits and adult survival: A meta‐analysis. Q. Rev. Biol. 76, 3–36. Jia, F., and Greenfield, M. D. (1997). When are good genes good? Variable outcomes of female choice in wax moths. Proc. R. Soc. Lond. B 264, 1057–1063. John, J. L. (1997). Handicap signal selection is not selection for inefficiency. Anim. Behav. 54, 225–227. Johnson, K. P. (1999). The evolution of bill coloration and plumage dimorphism supports the transference hypothesis in dabbling ducks. Behav. Ecol. 10, 63–67. Johnstone, R. A. (1995). Sexual selection, honest advertisement and the handicap principle: Reviewing the evidence. Biol. Rev. 70, 1–65. Johnstone, R. A., and Grafen, A. (1993). Dishonesty and the handicap principle. Anim. Behav. 46, 759–764. Kempenaers, B., Verheyen, G. R., Vandenbroeck, M., Burke, T., Vanbroeck, M., Burke, T., Vanbroeckhoven, C., and Dhondt, A. A. (1992). Extra‐pair paternity results from female preference for high‐quality males in the blue tit. Nature 357, 494–496. Kirkpatrick, M. (1982). Sexual selection and the evolution of female choice. Evolution 36, 1–12. Kirkpatrick, M., and Ryan, M. J. (1991). The evolution of mating preferences and the paradox of the lek. Nature 350, 33–38. Kokko, H. (1997). Evolutionarily stable strategies of age‐dependent sexual advertisement. Behav. Ecol. Sociobiol. 41, 99–107.

296

GERALD BORGIA

Kokko, H., Lindstrom, J., Alatalo, R. V., and Rintamaki, P. (1998). Queuing for territory positions in the lekking black grouse (Tetrao tetrix). Behav. Ecol. 9, 376–383. Kokko, H., Brooks, R., McNamara, J. M., and Houston, A. I. (2002). The sexual selection continuum. Proc. R. Soc. Lond. B 269, 1331–1340. Kokko, H., Brooks, R., Jennions, M. D., and Morely, J. (2003). The evolution of mate choice and mating biases. Proc. R. Soc. Lond. B 270, 653–664. Kotiaho, J. S. (2001a). Costs of sexual traits: A mismatch between theoretical considerations and empirical evidence. Biol. Rev. 76, 365–376. Kotiaho, J. S. (2001b). Sexual selection and condition dependence of courtship display in three species of horned dung beetles. Behav. Ecol. 13, 791–799. Kotiaho, J. S., Alatalo, R. V., Mappes, J., Nielsen, M. G., Parri, S., and Rivero, A. (1998). Energetic costs of size and sexual signaling in a wolf spider. Proc. R. Soc. Lond. B 265, 2203–2207. Krebs, J. R., and Davies, N. B. (1993). ‘‘An Introduction to Behavioural Ecology,’’ 3rd edn. Blackwell Science, Oxford. Ku¨nzler, R., and Bakker, T. C. M. (2001). Female preferences for single and combined traits in computer animated stickleback males. Behav. Ecol. 12, 681–685. Kusmierski, R., Borgia, B., Uy, A., and Crozier, R. (1997). Molecular information on bowerbird phylogeny and the evolution of exaggerated male characters. Proc. R. Soc. Lond. 264, 307–313. Lachmann, M., Sza´mado´, S., and Bergstrom, C. T. (2001). Cost and conflict in animal signals and human language. Proc. Natl. Acad. Sci. USA 98, 13189–13194. Lande, R. (1981). Models of speciation sexual selection on polygenic traits. Proc. Natl. Acad. Sci. USA 78, 3721–3725. Lande, R. (1987). Genetic correlations between the sexes and the evolution of sexual dimorphism and mating preferences. In ‘‘Sexual Selection: Testing the Alternatives’’ (J. W. Bradbury and M. B. Andersson, Eds.), pp. 83–94. John Wiley, Chichester. Leitner, S, Voigt, C., and Gahr, M. (2001). Seasonal changes in the song pattern of the non‐ domesticated island canary (Serinus canaria), a field study. Behaviour 138, 885–904. Lens, L., Van Dongen, S., Kark, S., and Matthysen, E. (2002). Fluctuating asymmetry as an indicator of fitness: Can we bridge the gap between studies? Biol. Rev. 77, 27–38. Leonard, M., and Zanette, L. (1998). Female mate choice and male behaviour in domestic fowl. Anim. Behav. 56, 1099–1105. Lewis, S. M., Cratsley, C. K., and Demary, K. (2004a). Mate recognition and choice in Photinus fireflies. Ann. Zool. Fennici 41, 809–921. Lewis, S. M., Cratsley, C. K., and Rooney, J. A. (2004b). Nuptial gifts and sexual selection in Photinus fireflies. Integr. Comp. Biol. 44, 234–237. Ligon, J. D., and Zwartjes, P. W. (1995). Ornate plumage of male red junglefowl does not influence mate choice by females. Anim. Behav. 49, 117–125. Lincoln, G. A. (1994). Teeth, horns and antlers: The weapons of sex. In ‘‘The Differences Between Sexes’’ (R. V. Short and E. Balaban, Eds.), pp. 131–158. Cambridge University Press, Cambridge. Lloyd, J. E. (1971). Bioluminescent communication in insects. Annu. Rev. Entomol. 16, 97–122. Loffredo, C., and Borgia, G. (1986a). Sexual selection, mating systems, and the evolution of avian acoustical displays. Am. Nat. 128, 773–794. Loffredo, C., and Borgia, G. (1986b). Male courtship vocalizations as cues for mate choice in the satin bowerbird (Ptilonorhynchus violaceus). Auk 103, 189–195.

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

297

Losos, J. (1999). Uncertainty in the reconstruction of ancestral character states and limitations on the use of phylogenetic comparative methods. Anim. Behav. 58, 1319–1324. Markusson, E., and Folstad, I. (1997). Reindeer antlers: Visual indicators of individual quality? Oecologia 110, 501–507. Markow, T. A. (1995). Evolutionary ecology of developmental instability. Ann. Rev. Entomol. 40, 105–120. Marler, P., and Peters, S. (1981). Sparrows learn adult song and more from memory. Science 213, 780–782. Marler, P., and Peters, S. (1982). Developmental overproduction and selective attrition: New processes in the epigenesis of birdsong. Dev. Psychobiol. 15, 369–378. Marler, P., and Peters, S. (1987). A sensitive period for song acquisition in the song sparrow, Melospiza melodia, a case of age‐limited learning. Ethology 77, 125–149. 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. Marshall, A. J. (1954). ‘‘Bower‐Birds: Their Displays and Breeding Cycles, a Preliminary Statement by Alexander James Marshall.’’ Clarendon Press, Oxford. Mateos, C., and Carranza, J. (1999). Effects of male dominance and courtship display on female choice in the ring‐necked pheasant. Behav. Ecol. Sociobiol. 45, 235–244. Maynard Smith, J. (1976). Sexual selection and the handicap principle. J. Theor. Biol. 57, 239–242. Maynard Smith, J. (1991). Must reliable signals always be costly? Anim. Behav. 47, 1115–1120. Maynard Smith, J., and Harper, D. G. C. (1995). Animal signals: Models and terminology. J. Theor. Biol. 177, 305–311. Mays, H. L. Jr., and Hill, G. E. (2004). Choosing mates: Genes that are good versus genes that are a good fit. Trends Ecol. Evol. 19, 554–559. McKaye, K. R., Howard, J., Stauffer, J., Morgan, R., and Shonhiwa, F. (1993). Sexual selection and genetic‐relationships of a sibling species complex of bower building cichlids in lake Malawi, Africa. Jpn. J. Ichthyol. 40, 15–21. McKaye, K. R., Stauffer, J. R., Turner, G. F., Konings, A., and Sato, T. (2001). Fishes, as well as birds, build bowers. Aquariol Aquat Sci 9, 121–133. Mead, L. S., and Arnold, S. J. (2004). Quantitative genetic models of sexual selection. Trends Ecol. Evol. 19, 264–271. Meyer, A., Morrissey, J., and Schartl, M. (1994). Recurrent origin of a sexually selected trait in Xiphophorus inferred from a molecular phylogeny. Nature 368, 539–542. McKinnon, J. S., and Rundle, H. D. (2002). Speciation in nature: The threespine stickleback model systems. Trends Ecol. Evol. 17, 490–498. Møller, A. P. (1988). Female choice selects for male sexual tail ornaments in the monogamous swallow. Nature 332, 640–642. Møller, A. P. (1989). Viability costs of male tail ornaments in a swallow. Nature 339, 132–134. Møller, A. P. (1990). Fluctuating asymmetry in male sexual ornaments may reliably reveal male quality. Anim. Behav. 40, 1185–1187. Møller, A. P. (1994). ‘‘Sexual Selection and the Barn Swallow.’’ Oxford University Press, Oxford. Møller, A. P., and Pomiankowski, A. (1993). Why have birds got multiple sexual ornaments? Behav. Ecol. Sociobiol. 32, 167–176. Moore, A. J. (1994). Genetic evidence for the ‘‘good genes’’ process of sexual selection. Behav. Ecol. Sociobiol. 35, 235–241. Morin, J. G. (1986). ‘Firefleas’ of the sea: Luminescent signaling in marine ostracode crustaceans. Florida Entomol. 69, 105–121.

298

GERALD BORGIA

Morris, M. R., Nicoletto, P. F., and Hesselman, E. (2003). A polymorphism in female preference for a polymorphic male trait in the swordtail fish Xiphophorus cortezi. Anim. Behav. 65, 45–52. Mu¨hlha¨user, C., and Blanckenhorn, W. U. (2004). The quantitative genetics of sexual selection in the dung fly Sepsis cynipsea. Behaviour 141, 327–342. Nelson, D. A. (1992). Song overproduction and selective attrition lead to song sharing in the field sparrow (Spizella pusilla). Behav. Ecol. Sociobiol. 30, 415–424. Nichols, R. A., and Butlin, R. K. (1989). Does runaway sexual selection work in finite populations? J. Evol. Biol. 2, 299–313. Nichols, R. A., and Butlin, R. K. (1992). Sexual preference and genetic correlations. Trends Ecol. Evol. 7, 29–30. Norberg, R. A. (1994). Swallow tail streamer is a mechanical device for self‐deflection of tail leading edge, enhancing aerodynamic efficiency, and flight maneuverability. Proc. R. Soc. Lond. B 257, 227–233. Norris, K. (1993). Heritable variation in a plumage indicator of viability in male great tits Parus major. Nature 362, 537–539. Nottebohm, F., and Nottebohm, M. E. (1978). Relationship between song repertoire and age in the canary, Serinus canarius. Z. Tierpsychol. 46, 298–305. Nowicki, S., and Searcy, W. A. (2004). Song function and the evolution of female preferences: Why birds sing, why brains matter. Ann. NY Acad. Sci. 1016, 704–723. Nowicki, S., Hasselquist, D., Bensch, S., and Peters, S. (2000). Nestling growth and song repertoire sire in great reed warblers: Evidence for song learning as an indicator mechanism in mate choice. Proc. R. Soc. Lond. B 267, 2419–2424. Nowicki, S., Searcy, W. A., and Peters, S. (2002). Brain development, song learning and mate choice in birds: A review and experimental test of the ‘‘nutritional stress hypothesis.’’ J. Comp. Physiol. A 188, 1003–1014. Nur, N., and Hasson, O. (1984). Phenotypic plasticity and the handicap principle. J. Theor. Biol. 110, 275–297. O’Loghlen, A. L., and Beecher, M. D. (1997). Sexual preferences for mate song types in female song sparrows. Anim. Behav. 53, 835–841. Omland, K. E. (1996a). Female mallard preferences for multiple male ornaments: I. Natural variation. Behav. Ecol. Sociobiol. 39, 353–360. Omland, K. E. (1996b). Female mallard preferences for multiple male ornaments: II. Experimental variation. Behav. Ecol. Sociobiol. 39, 361–366. Omland, K. E. (1997). Examining two standard assumptions of ancestral state reconstruction: Repeated loss of dichromatism in dabbling ducks (Anatini). Evolution 51, 1636–1646. Omland, K. E., and Lanyon, S. M. (2000). Reconstructing plumage evolution in orioles (Icterus): Repeated convergence and reversal in patterns. Evolution 54, 2119–2133. Otter, K. A., McGregor, P. K., Terry, A. M. R., Burford, F. R. L., Peake, T. M., and Dabelsteen, T. (1999). Do female great tits (Parus major) assess extra‐pair males by eavesdropping? A field study using interactive song playback. Proc. R. Soc. Lond. B 266, 1305–1310. Panhuis, T. M., and Wilkinson, G. S. (1999). Exaggerated male eye span influences contest outcome in stalk‐eyed flies (Diopsidae). Behav. Ecol. Sociobiol. 46, 221–227. Parker, G. A. (1970). The reproductive behaviour and the nature of sexual selection in Scathophaga stercoraria L.: The fertilization rate and the spatial and temporal relationships of each sex around the site of mating and oviposition. J. Anim. Ecol. 28, 205–228. Parker, G. A. (1983). Mate quality and mating decisions. In ‘‘Mate Choice’’ (P. Bateson, Ed.), pp. 141–166. Cambridge University Press, Cambridge.

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

299

Parker, T., and Ligon, J. D. (2002). Dominant male red junglefowl (Gallus gallus) test the dominance status of other males. Behav. Ecol. Sociobiol. 53, 20–24. Partridge, L. (1980). Mate choice increases a component of offspring fitness in fruit flies. Nature 283, 290–291. Patricelli, G. L., Uy, J. A. C., Walsh, G., and Borgia, G. (2002). Male displays adjusted to female’s response—Macho courtship the satin bowerbird is tempered to avoid frightening the female. Nature 415, 279–280. Patricelli, G. L., Uy, J. A. C., and Borgia, G. (2003). Multiple male traits interact: Attractive bower decorations facilitate attractive behavioural displays in satin bowerbirds. Proc. R. Soc. Lond. B 270, 2389–2395. Patricelli, G. L., Uy, J. A. C., and Borgia, G. (2004). Female signals enhance the efficiency of mate assessment in satin bower‐birds (Ptilonorhynchus violaceus). Behav. Ecol. 15, 297–304. Patricelli, G. L., Coleman, S., and Borgia, G. (2005). Male satin bowerbirds, Ptilonorhynchus violaceus, adjust their display intensity in response to female startling: An experiment with robotic females. Anim. Behav. 71, 49–59. Payne, R. B., Payne, L. L., and Woods, J. L. (1998). Song learning in brood parasitic indigobirds Vidua chalybeata: Song mimicry of the host species. Anim. Behav. 55, 1537–1553. Penn, D. J., and Potts, W. K. (1999). The evolution of mating preferences and major histocompatibility complex genes. Am. Nat. 153, 145–164. Pepperberg, I. M., Naughton, J. R., and Banta, P. A. (1998). Allospecific vocal learning by grey parrots (Psittacus erithacus): A failure of videotaped instruction under certain conditions. Behav. Processes 42, 139–158. Petrie, M. (1994). Improved growth and survival of offspring of peacocks with more elaborate trains. Nature 371, 598–599. Polak, M., Starmer, W. T., and Wolf, L. L. (2004). Sexual selection for size and symmetry in a diversifying secondary sexual character in Drosophila bipectinata duda (Diptera: Drosophilidae). Evolution 58, 597–607. Pomiankowski, A. (1987). Sexual selection: The handicap principle does work sometimes. Proc. R. Soc. Lond. B 231, 123–145. Pomiankowski, A. (1988). The evolution of mate preferences for male genetic quality. Oxf. Surv. Evol. Biol. 5, 136–184. Pomiankowski, A., and Iwasa, Y. (1998). Runaway ornament diversity caused by Fisherian sexual selection. Proc. Natl. Acad. Sci. USA 95, 5106–5111. Pope, D. S. (2000). Video playback experiments testing the function of claw waving in the sand fiddler crab. Behaviour 137, 1349–1360. Proctor, H. C. (1991). Courtship in the water mite Neumania papillator: Males capitalize on female adaptations for predation. Anim. Behav. 42, 589–598. Proctor, H. C. (1992). Sensory exploitation and the evolution of male mating behaviour: A cladistic test using water mites (Acari: Parasitengona). Anim. Behav. 44, 745–752. Proctor, H. C., and Smith, B. P. (1994). Mating behaviour of the water mite Arrenurus manubriator (Acari: Arrenuridae). J. Zool. Lond. 232, 473–483. Prum, R. O. (1997). Phylogenetic tests of alternative intersexual selection mechanisms: Macroevolution of male traits in a polygynous clade (Aves: Pipridae). Am. Nat. 149, 668–692. Quader, S. (2005). Elaborate nests in a weaverbird: A role for female choice? Ethology 111, 1073–1088.

300

GERALD BORGIA

Qvarnstro¨m, A., Blomgren, V., Wiley, C., and Svedin, N. (2004). Female collared flycatchers learn to prefer males with an artificial novel ornament. Behav. Ecol. 15, 543–548. Rantala, M. J., Kortet, R., Kotiaho, J. S., Vainikka, A., and Suhonen, J. (2003). Condition dependence of pheromones and immune function in the grain beetle. Tenebrio molitor. Funct. Ecol. 17, 534–540. Reeve, H. K., and Sherman, P. W. (1993). Adaptation and the goals of evolutionary research. Q. Rev. Biol. 68, 1–32. Reusch, T., Ha¨berli, M., Aeschlimann, P. B., and Milinski, M. (2000). Female sticklebacks count alleles in a strategy of sexual selection explaining MHC polymorphism. Nature 414, 300–302. Reynolds, J. D., and Gross, M. R. (1992). Female mate preferences enhance offspring growth and reproduction in a fish, Poecilia reticulata. Proc. R. Soc. Lond. B 250, 57–62. Robinson, F. N., and Curtis, H. S. (1996). The vocal displays of the lyrebirds (Menuridae). Emu 96, 258–275. Rohwer, S., and Rohwer, F. C. (1978). Status signaling in Harris’ Sparrows: Experimental deceptions achieved. Anim. Behav. 26, 1012–1022. Rosenthal, G. G., and Evans, C. S. (1998). Female preference for swords in Xiphophorus helleri reflects a bias for large apparent size. Proc. Natl. Acad. Sci. USA 95, 4431–4436. Rosenthal, G. G., Flores Martinez, T. Y., Garcia de Leon, F. J., and Ryan, M. J. (2001). Shared preferences by predators and females for male ornaments in swordtails. Am. Nat. 158, 146–154. Rowe, L., and Houle, D. (1996). The lek paradox, condition dependence and genetic variance in sexually selected traits. Proc. R. Soc. Lond. B 263, 1415–1421. Rowe, L. V., Evans, M. R., and Buchanan, K. L. (2001). The function and evolution of the tail streamer in hirundines. Behav. Ecol. 12, 157–163. Ryan, M. J. (1985). ‘‘The Tu´ngara Frog: A Study in Sexual Selection and Communication.’’ University of Chicago Press, Chicago. Ryan, M. J. (1990). Sensory systems, sexual selection, and sensory exploitation. Oxf. Surv. Evol. Biol. 7, 157–195. Ryan, M. J. (1997). Sexual selection and mate choice. In ‘‘Behavioral Ecology’’ (J. R. Krebs and N. B. Davies, Eds.), pp. 173–202. Blackwell, Maulden, MA. Ryan, M. J. (1998). Sexual selection, receiver biases, and the evolution of sex differences. Science 281, 1999–2003. Ryan, M. J., and Rand, A. S. (1990). The sensory basis of sexual selection for complex calls in the Tu´ngara frog, Physalaemus pustulosus (sexual selection for sensory exploitation). Evolution 44, 305–314. Ryan, M. J., and Rand, A. S. (1993). Sexual selection and signal evolution: The ghost of biases past. Philos. Trans. R. Soc. Lond. B 340, 187–195. Ryan, M. J., and Rand, A. S. (1995). Female responses to ancestral advertisement calls in Tu´ngara frogs. Science 269, 390–392. Ryan, M. J., Fox, C., Wilczynski, W., and Rand, A. S. (1990). Sexual selection for sensory exploitation in the frog. Physalaemus pustulosus. Nature 343, 66–67. Sargent, R. C., Rush, V. N., Wisenden, B. D., and Yan, H. Y. (1998). Courtship and mate choice in fishes: Integrating behavioral and sensory ecology. Am. Zool. 38, 82–97. Scheuber, H., Jacot, A., and Brinkhof, M. W. G. (2004). Female preference for multiple condition‐dependent components of a sexually selected signal. Proc. R. Soc. Lond. B 271, 2453–2457.

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

301

Schenkel, R. (1956). Zur Deutung der Balzleistungen einiger Phasianiden und Tetraoniden. Der Ornithologische Beobachter 53, 182–201. Schluter, D., and Price, T. (1993). Honesty, perception and population divergence in sexually selected traits. Proc. R. Soc. Lond. B 253, 117–122. Searcy, W. A. (1992). Song repertoire and mate choice in birds. Am. Zool. 32, 71–80. Searcy, W. A., and Andersson, M. (1986). Sexual selection and the evolution of song. Ann. Rev. Ecol. Syst. 17, 507–533. Seehausen, O. (2000). Explosive speciation rates and unusual species richness in haplochromine cichlids—effects of sexual selection. Adv. Ecol. Res. 31, 237–274. Shaw, K. (1995). Phylogenetic tests of the sensory exploitation model of sexual selection. Trends Ecol. Evol. 10, 117–120. Sherman, P., and Reeve, H. K. (1999). Sexual selection and sensory exploitation. Science 283, 1083a. Simmons, L., Tomkins, J., Kotiaho, J., and Hunt, J. (1999). Fluctuating paradigm. Proc. R. Soc. Lond. B 266, 593–595. Singer, A. G., Beauchamp, G. K., and Yamazaki, K. (1997). Volatile signals of the major histocompatibility complex in male mouse urine. Proc. Natl. Acad. Sci. USA 94, 2210–2214. Sivinski, J. (1981). The nature and possible functions of luminescence in Coleoptera larvae. Coleopt. Bull. 35, 167–179. Sivinski, J. (1997). Ornaments in the Diptera. Florida Entomol. 80, 142–162. Snedden, W. A., Tosh, C. R., and Ritchie, M. G. (1994). The ultrasonic mating signal of the male lesser wax moth. Physiol. Entomol. 19, 367–372. Sodderberg, R. (1929). Genesis of decorative and building instincts of bower birds (Ptilonorhynchidae). Verh. VI. Int. Cong. Ornith. Kopenhagen 297–337. Soler, J. J., Cuervo, J. J., Møller, A. P., and de Lope, F. (1998a). Nest building is a sexually selected behaviour in barn swallows. Anim. Behav. 56, 1435–1442. Soler, J. J., Møller, A. P., and Soler, M. (1998b). Nest building, sexual selection and parental investment. Evol. Ecol. 12, 427–441. Spangler, H. G. (1984). Attraction of female lesser wax moths (Lepidoptera: Pyralidae) to male‐produced and artificial sounds. J. Econ. Entomol. 77, 346–349. Spangler, H. G., Greenfield, M. G., and Takessian, A. (1984). Ultrasonic mate calling in the lesser wax moth. Physiol. Entomol. 9, 87–95. Spencer, K. A., Buchanan, K. L., Goldsmith, A. R., and Catchpole, C. K. (2003). Song complexity signals past developmental stress in the zebra finch. Horm. Behav. 44, 132–139. Stoddard, P. K. (1999). Predation enhances complexity in the evolution of electric fish signals. Nature 400, 254–256. Sturmbauer, C., Levinton, J. S., and Christy, J. (1996). Molecular phylogeny analysis of fiddler crabs: Test of the hypothesis of increasing behavioral complexity in evolution. Proc. Natl. Acad. Sci. USA 93, 10855–10857. Surlykke, A., and Fullard, J. H. (1989). Hearing of the Australian whistling moth, Hecatesia thyridion. Naturwissenschaften 76, 132–134. Swaddle, J. (1999). Limits to length asymmetry detection in starlings: Implications for biological signalling. Proc. R. Soc. Lond. B 266, 1299–1303. Taylor, M. I., Turner, G. F., Robinson, R. L., and Stauffer, J. R. (1998). Sexual selection, parasites and bower height skew in a bower‐building cichlid fish. Anim. Behav. 56, 379–384.

302

GERALD BORGIA

Thornhill, R. (1988). The jungle fowl hen’s cackle incites male competition. Verh Deutschen. Zool. Ges. 81, 145–154. Thornhill, R., and Moller, A. P. (1998). The relative importance of size and asymmetry in sexual selection. Behav. Ecol. 9, 546–551. Thusius, K., Peterson, K., Dunn, P. O., and Whittingham, L. A. (2001). Male mask size is correlated with mating success in common yellowthroats. Anim. Behav. 62, 435–446. Trainer, J. M., McDonald, D. B., and Learn, W. A. (2002). The development of coordinated singing in cooperatively displaying long‐tailed manakins. Behav. Ecol. 13, 65–69. Tweddle, D., Eccles, D. H., Frith, C. B., Fryer, G., Jackson, P. B. N., Lewis, D. S. C., and Lowe‐McConnell, R. H. (1998). Cichlid spawning structures—bowers or nests? Environ. Biol. Fishes 51, 107–109. Uy, A., Patricelli, G., and Borgia, G. (2000). Dynamic mate searching tactic allows female satin bowerbirds to reduce searching. Proc. R. Soc. Lond. B 267, 251–256. Uy, J. A. C., and Borgia, G. (2000). Sexual selection drives rapid divergence in bowerbird display traits. Evolution. 54, 273–278. van Doorn, G. S., and Weissing, F. J. (2004). The evolution of female preferences for multiple indicators of quality. Am. Nat. 164, 173–186. Verner, J., and Englesen, G. H. (1970). Territories, multiple nest building, and polygyny in the Long‐billed Marsh Wren. Auk 87, 557–567. Vencl, F. V., and Carlson, A. D. (1998). Proximate mechanisms of sexual selection in the firefly Photinus pyralis (Coleoptera: Lampyridae). Insect Behav. 11, 191–207. Verhulst, S., Dieleman, S. J., and Parmentier, H. K. (1999). A tradeoff between immunocompetence and sexual ornamentation in domestic fowl. Proc. Natl. Acad. Sci. USA 96, 4478–4481. Viljugrein, H. (1997). The cost of dishonesty. Proc. R. Soc. Lond. B 264, 815–821. von Schantz, T., Go¨ransson, G., Andersson, G., Fro¨berg, I., Grahn, M., Helge´e, A., and Wittzell, H. (1989). Female choice selects for a viability—male trait in pheasants. Nature 337, 166–169. Wedekind, C. (1994). Handicaps not obligatory in sexual selection for resistance genes. J. Theor. Biol. 170, 57–62. Welsh, A. M., Semlitsch, R. D., and Gerhardt, H. C. (1998). Call duration as an indicator of genetic quality in male gray tree frogs. Science 280, 1928–1930. Wiens, J. (2001). Widespread loss of sexually selected traits: How the peacock lost its spots. Trends Ecol. Evol. 16, 517–521. Wiley, R. H. (1991). Lekking in birds and mammals: Behavioral and evolutionary issues. Adv. Study Behav. 20, 201–291. Wilkinson, G. S., and Dodson, G. N. (1997). Function and evolution of antlers and eye stalks in flies. In ‘‘The Evolution of Mating Systems in Insects and Arachnids’’ (J. Choe and B. Crespi, Eds.), pp. 310–328. Cambridge University Press, Cambridge. Wilkinson, G. S., and Reillo, P. R. (1994). Female choice response to artificial selection on an exaggerated male trait in a stalk‐eyed fly. Proc. R. Soc. Lond. B 255, 1–6. Wilkinson, G. S., Presgraves, D. C., and Crymes, L. (1998). Male eye span in stalk‐eyed flies indicates genetic quality by meiotic drive suppression. Nature 391, 276–279. Wolfenbarger, L. L. (1999). Red coloration of male Northern Cardinals correlates with mate quality and territory quality. Behav. Ecol. 10, 80–90. Wynne‐Edwards, V. C. (1962). ‘‘Animal Dispersion in Relation to Social Behaviour.’’ Oliver and Boyd, Edinburgh. Yasukawa, K. (1981). Male quality and female choice of mate in the red‐winged blackbird (Agelaius phoeniceus). Ecology 62, 922–929.

CO-OPTION OF MALE TRAITS FOR SEXUAL DISPLAY

303

Zahavi, A. (1975). Mate selection—a selection for a handicap. J. Theor. Biol. 53, 205–214. Zahavi, A. (1977). The cost of honesty (further remarks on the handicap principle). J. Theor. Biol. 67, 603–605. Zahavi, A. (1991). On the definition of sexual selection, Fisher’s model and the evolution of waste and of signals in general. Anim. Behav. 42, 501–503. Zahavi, A. (1993). The fallacy of conventional signalling. Phil. Trans. R. Soc. B. 340, 227–231. Zahavi, A., and Zahavi, A. (1997). ‘‘The Handicap Principle.’’ Oxford University Press, Oxford. Zala, S. M., Potts, W. K., and Penn, D. J. (2004). Scent‐marking displays provide honest signals of health and infection. Behaviour 15, 338–344. Zeh, J., and Zeh, D. W. (2003). Toward a new sexual selection paradigm: Polyandry, conflict and incompatibility. Ethology 109, 929–950. Zuk, M., Johnson, K., Thornhill, R., and Ligon, D. J. (1990). Mechanisms of female choice in red jungle fowl. Evolution 44, 477–485.

This page intentionally left blank

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 36

Adaptation, Genetic Drift, Pleiotropy, and History in the Evolution of Bee Foraging Behavior Nigel E. Raine,* Thomas C. Ings,* Anna Dornhaus,{ Nehal Saleh,* and Lars Chittka*,1 *school of biological and chemical sciences queen mary university of london london e1 4ns, united kingdom { department of ecology and evolutionary biology university of arizona, tucson arizona 85721, usa

I. INTRODUCTION The formal study of foraging behavior began in the mid 1960s, using an approach that later became known as optimal foraging theory (Emlen, 1966; MacArthur and Pianka, 1966). Practitioners would use modeling to identify an optimal strategy for an animal facing a given number of foraging options, and then compare this to the strategy actually chosen by the animal (Maynard Smith, 1978; Orzack and Sober, 2001; Stephens and Krebs, 1991). This approach was instrumental in predicting quantitatively which types of food an animal should choose to consume (Pyke et al., 1977; Stephens and Krebs, 1991; Waddington and Holden, 1979), when to abandon a patch of food (Cuthill et al., 1990; Kacelnik and Krebs, 1985), how variance in food supply might affect forager choice (Fu¨lo¨p and Menzel, 2000; Real, 1981; Shafir et al., 1999), and what currencies animals use in making decisions about food quality (McNamara et al., 1993; Schmid‐ Hempel et al., 1985). The field thrived and expanded rapidly throughout the 1970s and 1980s, receiving further impetus from studies on the neurobiological mechanisms that underlie and constrain foraging during the 1990s (Chittka et al., 1999; Clayton, 1995; Clayton and Krebs, 1994; Greggers and Menzel, 1993), and from studies into the genetic basis of foraging behavior (Ben‐Shahar et al., 2002; Rueppell et al., 2004a). 1

Corresponding author.

0065-3454/06 $35.00 DOI: 10.1016/S0065-3454(06)36007-X

305

Copyright 2006, Elsevier Inc. All rights reserved.

306

NIGEL E. RAINE ET AL.

Despite these successes, a number of fundamental questions with respect to the adaptiveness of foraging strategies remain relatively unexplored. In a study on bee foraging in a natural environment by Schmid‐Hempel and Heeb (1991), a large percentage of foragers were removed at regular intervals during the colony cycle. The authors found no significant effects of this apparent decimation of the forager workforce on colony growth, life history, or ultimate colony reproductive success. So how can the precise subtleties of minute‐to‐minute foraging strategies of individuals matter, if not even the individuals’ existence matters for colony reproductive success? Perhaps foraging strategies are crucial only under adverse conditions (Schmid‐Hempel and Schmid‐Hempel, 1998), but the point here is much more general: we do not yet understand at all well how foraging strategies contribute to the fitness of animals in the wild. How well does a given strategy perform relative to other strategies, used by another individual or species? The shape of the adaptive landscape with respect to foraging remains relatively unexplored. If foraging strategies are sometimes placed on fitness plateaus, rather than steep adaptive peaks, genetic drift may make traits meander in random directions, before an animal falls down the cliff of severe fitness loss. In small populations, the effects of evolutionary chance should be especially pronounced (Adkison, 1995; Crow and Kimura, 1970; Ford, 1955), which is why we have devoted special attention to island bumblebee populations. In other cases perhaps, we might be better able to explain an extant animal’s foraging behavior by its evolutionary history, rather than the conditions under which it presently forages. While the power of studying adaptive hypotheses in foraging behavior through comparisons between species, or individuals, with different behavioral strategies was recognized early on (Clutton‐Brock and Harvey, 1977; Maynard Smith, 1978; Stephens and Krebs, 1991), these methods have received relatively little attention. Instead optimality modeling remained the favored tool of the trade. Here, we advocate using the toolbox of modern evolutionary biology, which has already been successfully applied to study adaptive patterns in many branches of animal behavior (Alcock, 1996), to the study of foraging in bees. We employ a comparative approach (Harvey and Purvis, 1991) to correlate differences in foraging styles, at both the species and population level, with features in the bees’ respective environments. We use reciprocal transplant experiments (Kawecki and Ebert, 2004; Riechert and Hall, 2000), comparing the foraging performance of native bees with those stemming from populations operating in different (foreign) environments, to test hypotheses about local foraging adaptation. We manipulate the foraging environment to remove the possibility that bees can use particular foraging strategies (Schmid‐Hempel and Schmid‐Hempel, 1998), such as forming traplines, to tease apart the effects of each of these strategies individually. We use experimental manipulations to create artificial

EVOLUTION OF BEE FORAGING BEHAVIOR

307

foraging phenotypes (Curio, 1973), selectively eliminating the foraging‐ related abilities of wild‐type bees, to determine the adaptive significance of the manipulated traits. Where such manipulations are not possible, we use agent‐based simulations to assess the success of phenotypes that are not naturally available (Dornhaus et al., 1998). We focus especially on the following traits: flower constancy, floral color preference, learning behavior, traplining, and communication about food sources. We also correlate some of these with foraging performance. In some cases, we show that forager behavior has been tuned to function adaptively in a given niche. In other cases, however, the observed differences in behavior patterns can be better explained by chance processes, or by the historical conditions under which bees operated in their evolutionary past.

II. COMPARISON BETWEEN SPECIES: FLOWER CONSTANCY Aristotle observed that ‘‘during each flight the bee does not settle on flowers of different kinds, but flies, as it were, from violet to violet, and touches no other till it returns to the hive’’ (quoted in Christy, 1884). This phenomenon, now termed flower constancy, is defined as follows: an individual insect is flower constant if it visits only a restricted number of flower species, even if other species are available and equally rewarding, and if the insect has no innate or imprinted predisposition to visit only flowers of a restricted plant taxon, which must be confirmed by the observation that other individuals of the same insect species visit other plant species within the same array (Chittka et al., 1999; Waser, 1986). Is flower constancy an optimal foraging behavior? It is hard to see how such behavior could be adaptive per se, since there is rarely only a single best food source, and specializing on one flower type, while skipping other valuable resources encountered en route, is not necessarily the best strategy to maximize energy intake rate (Chittka, 2002; Chittka et al., 1999; Waser, 1986). Thus, flower constancy can only be considered adaptive in the face of behavioral limitations that might make switching between species costly. Short‐term memory limitations are one likely explanation (Chittka, 1998; Chittka et al., 1997, 1999; Raine and Chittka, 2005a). While generalist bees are able to store the sensory cues and motor patterns for several flower species in long‐term memory, there appear to be delays in retrieving the sensory cues of flowers that have not been visited in the bee’s immediate history (Bar‐Shai et al., 2004; Chittka and Thomson, 1997; Greggers and Menzel, 1993). In addition, several workers have found that switching between plant species with different morphologies increases flower handling time. While such costs are often negligible for easily accessible flowers

308

NIGEL E. RAINE ET AL.

(Chittka et al., 1997; Laverty, 1994), they can be substantial when bees have to retrieve multiple but drastically different motor patterns from memory (Chittka and Thomson, 1997; Woodward and Laverty, 1992). Under these conditions, when flowers of the same and novel species are available at equal distances, foraging insects should remain flower constant to minimize switching costs. Conversely, as travel time between flowers increases, or if all flowers are poorly rewarding, the costs of bypassing alternative species may exceed the costs of switching, which should favor inconstancy (Chittka et al., 1999). In reality, it is difficult to rigorously test these specific predictions in the economy of nature because controlling the range of floral species, morphologies, and patterns of reward provision available to free foraging bees is virtually impossible. An alternative, and perhaps more direct, test of the adaptive benefits of flower constancy could be to examine bumblebee species that differ consistently in the extent to which they are flower constant and to compare their relative foraging performance. Do we actually find that more flower constant species forage more effectively? In a study where foraging bumblebees were monitored in a meadow, containing five plant species near Berlin, Germany (Chittka et al., 1997), Bombus terrestris (L.) switched in 15% of 107 observed flights (transitions) between plants, Bombus lapidarius (L.) switched in 18% of 867 transitions, and Bombus pascuorum (Scopoli) switched in 26% of 2368 transitions. In this study, B. pascuorum switched significantly more often than B. lapidarius (w2 ¼ 19.52, p < 0.00005), but B. lapidarius and B. terrestris did not differ (w2 ¼ 0.78, p > 0.1: Chittka et al., 1997). We consistently found the same rank order of flower constancy, among the same three bumblebee species near Wu¨rzburg, Germany, in controlled field trials (‘‘bee interviews,’’ sensu; Thomson, 1981), where bee choices between specific pairs of plant species were observed (Chittka et al., 2001; Raine and Chittka, 2005a; Fig. 1). Likewise, in a study near Southampton, England, B. terrestris foragers were observed to be more constant than B. pascuorum (Stout et al., 1998). The results of all of these studies suggest that B. terrestris is consistently more flower constant than B. lapidarius and B. pascuorum. To what extent, then, is this consistent difference in foraging strategy mirrored in the foraging performance of these bee species? In a first approach, we placed colonies reared from wild‐caught queens of B. lapidarius (two colonies in 1999, three colonies in 2001) and B. terrestris (two colonies in 1999, five colonies in 2001) at a field site near Wu¨rzburg, Germany (Raine and Chittka, 2005a). We were unsuccessful at rearing B. pascuorum from wild‐caught queens, but in 1999, we found a small colony in the wild, placed it into a nest box, and raised the colony to a relatively large size in the laboratory before placing it in the field, alongside

EVOLUTION OF BEE FORAGING BEHAVIOR

309

Fig. 1. Consistent differences in flower constancy across three bumblebee species. Bees were tested using the bee interview technique (Thomson, 1981) using three pairs of plant species. The plant species used were red clover (Trifolium pretense L.), white clover (Trifolium repens L.), bird’s foot trefoil (Lotus corniculatus L.), and cow vetch (Vicia cracca L.). Common plant names are given on the x‐axis labels. Higher values of the flower constancy index indicate that bees are more likely to move between individual plants of the same species when foraging. Constancy indices were calculated according to Chittka et al. (2001) and can vary from 1 (complete constancy), through 0 (random flights between species), to –1 (complete inconstancy). Data from Raine and Chittka (2005a) with permission.

two colonies each of B. terrestris and B. lapidarius. The field site was typical central European bumblebee habitat, including dry grassland, deciduous forest, and farmland within the bees’ foraging range (Darvill et al., 2004; Dramstad, 1996; Osborne et al., 1999; Walther‐Hellwig and Frankl, 2000). Individually marked foragers were weighed at the start and the end of each foraging trip, allowing us to determine the foraging rate of individual workers by dividing the difference in body mass (i.e., return minus outgoing weight) by the trip duration (Chittka et al., 2004; Ings et al., 2005b; Raine and Chittka, 2005a; Spaethe and Weidenmu¨ller, 2002). At first inspection, the more flower constant B. terrestris foragers performed consistently better in both 1999 and 2001 than the less constant B. lapidarius (Fig. 2; Frauenstein, 2002; Raine and Chittka, 2005a). From this, one might conclude that a higher degree of floral constancy is beneficial in this habitat. However, there are several complications with this

310

NIGEL E. RAINE ET AL.

Fig. 2. Interspecific comparison of foraging performance in three bumblebee species. The foraging rate of individual workers from each colony was determined by dividing the difference in body mass (i.e., incoming minus outgoing weight) by the duration of their foraging trip. Colony foraging performance was evaluated by averaging each bee’s performance across all foraging bouts, then averaging across all bees tested. Column heights are colony mean (
1 SE) foraging rates/flight durations in each year tested. The number of foragers evaluated per colony is indicated at the foot of each column. For two species (B. terrestris and B. lapidarius), the experiment was performed in two different years (1999 and 2001), while for B. pascuorum it was only performed in 1999. Data from Raine and Chittka (2005a) with permission.

interpretation. B. pascuorum, the least flower constant species, performed even better than B. terrestris: hence flower constancy appears to be a poor predictor of foraging performance at the species level. This suggests that factors besides flower constancy may be decisive in determining foraging performance. Body mass might be one such factor as larger bees appear to bring home more nectar per unit time (Chittka et al., 2004; Goulson et al., 2002; Ings et al., 2005b; Spaethe and Weidenmu¨ller, 2002). While body size puts the larger B. terrestris (mean body mass
1 SD ¼ 166
43 mg) at an advantage over the smaller B. lapidarius (mean body mass ¼ 114
35 mg), once again it cannot explain the superior performance of B. pascuorum (mean body mass ¼ 138
18 mg), which is much smaller than B. terrestris. Tongue length and foraging range could be other important factors. B. pascuorum has a longer proboscis than B. terrestris or B. lapidarius (Goulson and Darvill, 2004; Hagen, 1990; Prys‐Jones and Corbet, 1991), which allows B. pascuorum workers to collect nectar from flowers with

EVOLUTION OF BEE FORAGING BEHAVIOR

311

longer corolla tubes that would not be accessible to the other two species (Barrow and Pickard, 1984). B. pascuorum also flies shorter distances to foraging patches than other species (Darvill et al., 2004; Free and Butler, 1959; Goulson, 2003; Hedkte, 1996), which might give it an additional edge. Hence, even if flower constancy is an important factor in determining foraging performance, each bee species might effectively choose microhabitats with a plant species composition best suited to its particular foraging strategies (Chittka et al., 1999; Thomson and Chittka, 2001). We conclude that using species comparisons to determine the adaptive significance of foraging strategies in the field is difficult because species will typically differ with respect to multiple foraging‐related traits. This is an important general lesson about the evolution of foraging behavior: typically animals proceed along multiple alternative evolutionary pathways to optimize foraging behavior, and constraints imposed by one foraging‐related trait might be easily compensated for by alterations of another trait.

III. COMPARISON BETWEEN SPECIES: FLORAL COLOR PREFERENCE Comparisons between species can be more rewarding when we compare many closely related species of known phylogeny. Attempts to identify evolutionary adaptations in foraging by focusing only on a single species, or sets of unrelated species, were common in earlier studies (Dukas and Real, 1991; Greggers and Menzel, 1993; Pyke, 1978). However, this is problematic since correlation and optimality cannot be equated with adaptation (Chittka, 1996a; Chittka and Dornhaus, 1999; Maynard Smith, 1978): in order to show that a trait is adapted for the task we think it is, we need to demonstrate that the ancestors of the animal in question which did not share the same environment also do not share the trait under scrutiny (Brooks and McLennan, 1991; Chittka and Briscoe, 2001; Losos and Miles, 1994). The comparative phylogenetic method, which seeks to reconstruct the traits of ancestral species through comparing closely related extant species, is a powerful tool to study patterns of adaptation (Armbruster, 1992; Chittka and Dornhaus, 1999; Phelps and Ryan, 2000; Ryan and Rand, 1999). This has been used to some extent to study adaptation in the foraging strategies of beetles (Betz, 1998), birds (Barbosa and Moreno, 1999), and primates (Clutton‐Brock and Harvey, 1977) but not, to our knowledge, bees. We start by applying this method to a foraging‐related trait, the floral color preferences of bees. Many newly emerged insects that have never seen flowers prefer certain colors over others (Briscoe and Chittka, 2001; Chittka and Wells, 2004; Lunau et al., 1996). Such innate color preferences

312

NIGEL E. RAINE ET AL.

help naive insects to find food, and, possibly, to select profitable flowers among those available. Floral preferences can be overwritten by learning to some degree, but there is evidence that in some situations (for example when rewards are similar across flower species), bees will revert to their initial preferences (Banschbach, 1994; Gumbert, 2000; Heinrich et al., 1977). Our hypothesis is that these innate preferences reflect the traits of local flowers that are most profitable for bees. In one study, Giurfa et al. (1995) found a good correlation between the color preferences of naive honeybees (Apis mellifera L.) and the nectar offerings of different flowers in a nature reserve near Berlin. These honeybees preferred violet (bee UV–blue, i.e., stimulating most strongly the bees’ UV and blue receptors) and blue (bee blue, i.e., stimulating predominantly the bees’ blue receptors), which were also the colors most associated with high nectar rewards. However, correlation does not imply causality. Hence, to show that color preferences actually evolved to match floral offerings, we could compare a set of closely related bee species that live in habitats in which the association of floral colors with reward is different. We tested the color preferences of eight bumblebee species from three subgenera: four species from central Europe (B. terrestris, B. lucorum (L.), B. pratorum (L.), and B. lapidarius); three from temperate East Asia (Bombus diversus (Smith), Bombus ignitus (Smith), and Bombus hypocrita (Pe´rez)); and one from North America (Bombus occidentalis (Greene)). Note that all data were collected by naive observers, who were given no background information on the bees’ foraging biology (Chittka et al., 2001). We rotated observers between the experimental setups containing different species to minimize any effect of observer bias on observed interspecific patterns. All colonies were raised under identical temperature and humidity conditions in a dark laboratory. Feeding and other necessary colony manipulations (e.g., marking workers) were conducted under dim red light, otherwise colonies were kept in unlit conditions. Bees had never been exposed to flower colors prior to experiments. This rearing procedure minimizes the risk that any observed between‐species differences were caused by nongenetic factors. One cannot entirely exclude the possibility that different species respond differentially to identical rearing conditions, but we think that any effect of this on color preferences is most unlikely. Colony nest boxes were connected to a flight arena (120  100  35 cm3), where workers were allowed to forage for sucrose solution (50% w/w) from colorless, UV‐transmittent Plexiglas square chips (25  25 mm2) placed on transparent glass cylinders (diameter ¼ 10 mm; height ¼ 40 mm). Workers that foraged on these transparent chips were individually marked with Opalith numbered tags. To test bee color preference, these rewarding, colorless

EVOLUTION OF BEE FORAGING BEHAVIOR

313

Plexiglas chips were replaced by 18 unrewarding ‘‘flowers’’ of 6 different colors [i.e., 3 flowers of each color: violet (bee UV–blue), blue (bee blue), white (bee blue–green, i.e., producing a strong signal in the blue and green receptors of bees), yellow, orange, and red (all bee green, stimulating most strongly the bees’ green receptors)]. These ‘‘flowers’’ were painted Plexiglas squares on glass cylinders (dimensions as above) located at random in the arena. Only one forager was allowed into the arena for testing at a time, and each bee was tested for a single foraging bout during which the number of times it chose flowers of each color was recorded. Flowers were changed between each foraging bout to ensure that the next bee received no odor cues from the previously tested forager. We superimposed the behavioral data from these species onto their phylogeny, as established by Williams (1994). All species tested preferred the violet–blue range, which therefore presumably represents a phylogenetically ancient preference (Fig. 3). This preference is likely to be advantageous, since flowers of these colors have been found to contain high nectar rewards in a variety of habitats (Chittka et al., 2004; Giurfa et al., 1995; Menzel and Shmida, 1993). Since all tested species share this trait, it is impossible to conclude that it has been adapted specifically by bumblebees in the context of flower visitation. However, we did also find interspecific differences in color preference. B. occidentalis had a much stronger preference for red than any other bumblebee species tested. This is particularly intriguing because B. occidentalis is frequently observed foraging, or robbing nectar, from red flowers whose morphology seems well adapted for pollination by hummingbirds (Chittka and Waser, 1997; Irwin and Brody, 1999). Our comparative phylogenetic analysis strongly suggests that this preference is derived and is therefore likely to represent an adaptation to this unique foraging strategy of B. occidentalis (Chittka and Wells, 2004; Raine and Chittka, 2005b). We conclude that the approach of superimposing foraging‐related traits onto the known phylogeny (Harvey and Purvis, 1991) is a powerful tool to study evolutionary adaptation of foraging behavior, so we recommend that this approach be used more frequently in similar such studies to determine the adaptiveness of foraging traits.

IV. COMPARISON BETWEEN POPULATIONS: FLORAL COLOR PREFERENCES Comparisons between populations of the same species are attractive because they reveal patterns of adaptation among very closely related individuals operating under divergent ecological conditions. We became especially interested in island populations, which are ‘‘natural laboratories’’ because of their relatively small population sizes, risk of genetic

314

NIGEL E. RAINE ET AL.

Fig. 3. Color preferences of eight bumblebee species superimposed on their phylogeny (following Williams, 1994). Each bee was experimentally naive at the start of the experiment, and only the first foraging bout was evaluated. At least three colonies were tested per species and at least 15 workers per colony. Bees were individually tested in a flight arena in which they were offered the colors V, violet (bee UV–blue); B, blue (bee blue); W, white (bee blue–green); Y, yellow; O, orange; R, red (the latter three are all bee green). Column height denotes the mean (
1 SE) of choice percentages. The sequence of species in the histogram (top panel) left to right maps onto those from the phylogeny, top to bottom; hence the leftmost column is B. diversus. Data from Chittka et al. (2001, 2004) and Chittka and Wells (2004).

EVOLUTION OF BEE FORAGING BEHAVIOR

315

bottlenecks, and occasionally more pronounced local adaptation because of disruption to gene flow with other populations adapted to different conditions (Adkison, 1995; Barton, 1998; Chittka et al., 2004; Ford, 1955; Stanton and Galen, 1997). As well as being one of the commonest bumblebee species in Europe, B. terrestris has managed to colonize all Mediterranean islands. These island populations of B. terrestris are particularly interesting because they are genetically differentiated from one another and from the mainland population (Estoup et al., 1996). In contrast, the entire mainland population, stretching across central, southern, and eastern Europe, appears to be much more genetically homogeneous (Widmer et al., 1998). We tested the unlearned color preferences of laboratory‐raised colonies obtained from eight B. terrestris populations: B. t. terrestris (L.) from Holland and Germany; B. t. dalmatinus (Dalla Torre) from Israel, Turkey, and Rhodes; B. t. sassaricus (Tournier) from Sardinia; B. t. xanthopus (Kriechbaumer) from Corsica; and B. t. canariensis (Pe´rez) from the Canary Islands. Color preference tests and rearing conditions were identical to those in Section III. All populations preferred colors in the violet to blue range of the spectrum, but there were some differences in the relative preference for violet and blue (Fig. 4). This largely matches the picture seen in most species tested in Section III, and this preference for violet and blue flowers makes biological sense since these flowers have been identified as most rewarding in a variety of habitats (Chittka et al., 2004; Giurfa et al., 1995). One might ask why flowers have not exploited these preferences, so that flowers with colors that are innately preferred might ultimately produce less nectar, while maintaining the same pollination success. It is necessary to bear in mind that innate preferences typically govern only the first few flower visits of a naive bee so that overall visitation rates of plants will largely be governed by informed choices of experienced bees. However, some island populations displayed a different pattern of color preference. B. t. sassaricus and B. t. canariensis exhibited an additional red preference (Chittka et al., 2001). Thus, there clearly is evolutionary plasticity in flower color preference within B. terrestris, and tests with laboratory‐ bred offspring colonies show that such between‐population differences are heritable (Chittka and Wells, 2004). The adaptive significance of such a red preference is not easy to understand. Some red, UV‐absorbing, pollen‐rich flowers exist in the Mediterranean basin, particularly toward the eastern end, with the highest concentration in Israel (Dafni et al., 1990). However, in Israel, bumblebees do not show a red preference, and the red flowers which grow there appear to be predominantly visited by beetles (Dafni et al., 1990). In Sardinia, red, UV‐absorbing flowers are neither more common than on the European

316

NIGEL E. RAINE ET AL.

Fig. 4. Biogeography of floral color preference in B. terrestris. Bees were individually offered the colors: V, violet (bee UV–blue); B, blue (bee blue); W, white (bee blue–green); Y, yellow; O, orange; R, red (the latter three are bee green). Column height denotes the mean (
1 SE) of colony choices. At least five colonies were tested per population. The shaded area shows the distribution of B. terrestris (this range was provided with kind permission of P. Rasmont). Data from Chittka et al. (2001, 2004).

mainland nor more rewarding than flowers of other colors in Sardinia (Chittka et al., 2004). The Canary Islands do harbor several orange‐red flower species (Vogel et al., 1984), which are most probably relics of a Tertiary flora, and some species seem strongly adapted to bird pollination. In fact, bird visitation has been observed in at least some of these species (Olesen, 1985; Valido et al., 2002), but it is not known whether bumblebees use them at all. Thus, we are left with an interesting observation: flower color preferences are clearly variable within B. terrestris, and these differences are heritable (Chittka and Wells, 2004). But we cannot easily correlate the color preferences in different habitats with differences in local floral colors. The possibility that genetic drift has produced the color preferences in some island populations certainly deserves consideration. However, it is also possible that the red preference of these bumblebee populations is a ‘‘behavioral fossil,’’ which dates back to an age when red, bird‐pollinated flowers were common in Europe. The discovery of fossil hummingbirds in the Old World (Germany) provides putative pollinators for such bird‐pollinated flowers (Mayr, 2004). Mayr conjectured that some flower species, seemingly adapted to bird pollination, might be relics from

EVOLUTION OF BEE FORAGING BEHAVIOR

317

times when these birds were common in Europe. If this is true, and if bumblebees exploited some of these flowers (as some species do in North America: Chittka and Waser, 1997), then the red preference of some of our B. terrestris populations might be a result of history rather than either recent adaptation or chance.

V. VARIATION WITHIN POPULATIONS: COLOR PREFERENCE AND FORAGING PERFORMANCE Many scientists studying insects have long ignored interindividual variation in behavior: some have even regarded it as noise that needed to be eliminated by averaging (reviewed in Chittka and Dornhaus, 1999). However, heritable differences between individuals represent the raw material for evolution. If no such variation exists (as in the number of legs in insects), selection has nothing to act on. In the social bumblebees, matters are somewhat more complicated because reproduction is restricted to a subset of individuals: here then, the unit of selection is not the individual, but the entire colony, which works together to maximize the contribution of sexually active individuals to the next generation. Hence, for bumblebees, intercolony, rather than interindividual, variation allows us to test the adaptive benefits of foraging behavior within a given ecological framework. To test if floral color preference, or any foraging‐related trait, is adaptive, one would ultimately want to show that the trait confers greater fitness to its bearers, compared to animals lacking the trait, or that have it in a modified form (Chittka and Briscoe, 2001). One indirect measure of biological fitness is foraging performance (Alcock, 1996, p. 159), as the amount of food available to a bumblebee colony is positively correlated with the production of males and new queens (Ings et al., 2005a, 2006; Pelletier and McNeil, 2003; Schmid‐Hempel and Schmid‐Hempel, 1998). Here we explore within‐population variation of floral color preference, a heritable foraging‐related trait, to measure the extent to which such preferences can be regarded as adaptive, that is, improving the foraging performance of individual bees, and hence indirectly colony fitness. In the vicinity of Wu¨rzburg, Germany, we made two interesting observations. First, that plant species with violet (bee UV–blue) flowers contain the highest nectar rewards (Chittka et al., 2004). Second, that there is appreciable variation among colonies in the extent to which bees prefer either blue or violet flowers (Raine and Chittka, 2005b; Fig. 5). To establish any potential correlation between a preference for violet (highly rewarding) flowers and good foraging performance, we needed to test both for each colony. To enable us to achieve this within the lifespan of a single

318

NIGEL E. RAINE ET AL.

Fig. 5. Correlation of unlearned floral color preference and foraging performance in the wild measured in the bumblebee B. terrestris near Wu¨rzburg (rs ¼ 0.82; N ¼ 5; p ¼ 0.089). Each data point represents mean (
1 SE) performance for each of these traits for one test colony. Data from Raine and Chittka (2005b).

colony, we simplified the laboratory color preference tests from those in Section III. We tested the color preference of each forager individually in a flight arena, which contained eight violet and eight blue artificial flowers (Frauenstein, 2002; Raine and Chittka, 2005b). Each bee was tested for a single foraging bout, after which the flowers in the arena were changed to ensure that the subsequent test bee received no odor cues. We tested 12 foragers from each of 5 colonies (i.e., 60 bees in total). All five tested colonies were subsequently taken into the field and their foraging performance tested over a 3‐week period in July 2001 (see Section II; Chittka et al., 2004; Raine and Chittka, 2005b for site description and methods). In the five colonies tested, the average percentage preference for violet over blue ranged from 41 to 56% (Raine and Chittka, 2005b), although other colonies tested in a separate study exhibited a violet preference of up to 62% (Frauenstein, 2002). In our study, colonies with a higher average unlearned preference for violet in the laboratory harvested more nectar per unit time in the field (Raine and Chittka, 2005b; Fig. 5). This is as one might expect, given that the violet flowers around Wu¨rzburg appear to contain more nectar than blue flowers (Chittka et al., 2004), but the correlation narrowly misses statistical significance (rs ¼ 0.82, N ¼ 5, p ¼ 0.089; Raine and Chittka, 2005b) possibly because of the small sample size. We left these test colonies in the field for a further 5 weeks after the foraging

EVOLUTION OF BEE FORAGING BEHAVIOR

319

tests to allow us to quantify the production of new queens (gynes) from each one—a more direct measure of biological fitness than foraging performance. We reduced the nest entrance diameter to 7 mm to prevent the escape of newly emerged queens from their natal nest, while allowing the smaller foraging workers to pass freely (worker thorax width 3–7 mm: Goulson, 2003; Goulson et al., 2002). Queen productivity per colony ranged from 4 to 39, with the highest number of queens being produced by the colony with the strongest violet preference (Raine and Chittka, 2005b). However, while the overall correlation between violet preference and queen production was positive, it was far from significant (rs ¼ 0.46, N ¼ 5, p ¼ 0.43; Raine and Chittka, 2005b). In conclusion, while there is an overall trend for colonies with a stronger violet preference to perform better in an environment with highly rewarding violet flowers, we need more data to ascertain whether this trend is actually biologically meaningful. This study clearly illustrates a number of the challenges faced when trying to quantify the fitness impacts of foraging‐related traits in bees. First, the traits of interest (e.g., color preference) and foraging performance must both be measured for a large number of colonies, which requires a large and motivated workforce. Second, even if the traits under examination are somehow correlated with foraging performance, they may have no measurable impact on biological fitness within one generation. However, even if any fitness effect is difficult to measure within a single generation, the effects of that trait may still be important over evolutionary relevant time scales. Finally, other traits, notably parasite resistance (Baer and Schmid‐ Hempel, 1999), may be so important that they obscure the potential impact of the trait(s) under examination. This is further complicated by the fact that the parasite load may itself also affect foraging behavior (Ko¨nig and Schmid‐Hempel, 1995; Otterstatter et al., 2005; Schmid‐Hempel and Stauffer, 1998) and learning performance (Mallon et al., 2003). Therefore, this is not just a lesson in the difficulties involved in measuring adaptive significance—it is also a lesson related to the evolution of foraging behavior itself. If the effects of foraging‐related traits on biological fitness are relatively hard to measure, or are often obscured by other unrelated traits, then selection on foraging strategies may itself be relatively weak. Thus, foraging‐related traits may well be sitting on relatively broad adaptive peaks, where deviations from the optimum may not be strongly penalized in terms of fitness costs because of the shape of the adaptive landscape (Gilchrist and Kingsolver, 2001; Whitlock, 1997). If variation in foraging strategies is sometimes selectively neutral, evolutionary chance processes may play a greater role in between‐species or between‐population differences than is generally thought.

320

NIGEL E. RAINE ET AL.

VI. VARIATION WITHIN POPULATIONS: LEARNING BEHAVIOR The ‘‘pollination market’’ in which bees forage will typically contain dozens of flower species, which differ greatly in the nectar and pollen rewards on offer, their handling costs, and spatial distribution. The average rewards in a flower species may change rapidly over the course of the day, depending on patterns of reward production and the activities of other flower visitors (Harder, 1990; Heinrich, 1979; Inouye, 1978; Stone et al., 2003; Willmer and Stone, 2004). Since floral rewards differ strongly among plant species and fluctuate rapidly over time, generalist foragers, such as bumblebees and honeybees, need to assess such differences in reward and respond accordingly (Chittka, 1998; Menzel, 2001). For this reason, learning floral traits, such as color, pattern, and scent, as predictors of floral reward is vital to efficient foraging (Chittka et al., 1999). But is the speed at which bees form associations, such as those between floral color and reward, adaptive? To examine this question, we set out to assess the variability in colony learning performance within the British population of B. terrestris (B. t. audax (Harris)). We tested bumblebee workers (240 workers from 16 colonies) in a simple foraging situation in which they had to distinguish yellow, rewarding artificial flowers from blue, unrewarding ones (Raine et al., 2006). Test colonies, produced from wild caught queens, were raised entirely in the laboratory and were therefore unbiased by previous experience at the start of the experiments. During testing, each bee foraged alone in a flight arena containing 10 blue and 10 yellow artificial flowers. The yellow flowers contained a sucrose reward (15 ml of 50% sucrose solution w/w), while blue flowers were empty (unrewarded). The behavior of each test bee was observed until it approached, or landed on, at least 100 flowers after it first fed from (probed) a yellow flower. The learning performance of each bee was quantified as the number of errors made, that is, choices of unrewarding (blue) flowers, as a function of the total number of flowers chosen. We found striking variation in learning performance among the 16 bumblebee colonies tested. First, we found significant variation in the average number of flower choices made by a bee before probing a yellow flower, the point at which associative learning between yellow flowers and reward could begin (Raine et al., 2006). While the vast majority (88%) of bees probed their first yellow flower after fewer than 100 flower choices, bees from some colonies did not feed from a yellow flower until after several hundred choices, and the highest recorded number of choices was 373. There was also significant intercolony variation in the speed at which bees subsequently learned to associate yellow flowers with reward (Raine

EVOLUTION OF BEE FORAGING BEHAVIOR

321

et al., 2006). A comparison of the learning curves for the fastest (A99), a medium (A62), and the slowest (A228) learning colonies shows that they differed most in the number of errors they make during the earlier stages of the learning process, predominantly during the first 60 choices (1–60) after probing their first yellow, rewarding flower (Fig. 6). In all colonies, the largest improvement in task performance happened during the first 10 flower choices after, but including, the first time the bee probed a yellow, rewarding flower. However, the magnitude of this improvement in task performance varies greatly among colonies. The task performance of the fastest learning colony (A99) improved by 70% during the first 10 flower choices after probing a yellow flower, while the slowest learning colony improved by 49%. After this very large improvement in task performance, learning continues, but the rate at which task performance improves declines until the bee’s task performance eventually saturates.

Fig. 6. Learning performance of bees from a fast (A99), medium (A62), and slow (A228) learning colony. The behavior of 15 bees in each colony was observed for 100 flower choices after they first fed from (probed) a yellow, rewarding flower. An ‘‘error’’ was categorized as a bee approaching or visiting (landing on) a blue, unrewarding flower. The first column (py) represents the mean (
1 SE) percentage of errors made by bees from each colony during the first 10 flower choices they made in the test arena, that is, before they probed a rewarding, yellow flower. The remaining columns (N ¼ 10) represent the colony mean percentage error for each consecutive sequence of 10 flower choices made after probing the first yellow flower (choices 1–10, 11–20, and so on). Data from Raine et al. (2006).

322

NIGEL E. RAINE ET AL.

We can therefore conclude that there is significant variability in the ability of bumblebee colonies to learn color as a predictor of floral reward. This raises the question whether there might be an optimal learning speed for foraging under natural conditions. We often tacitly assume that behavioral traits (including learning ability) are sitting on narrow adaptive peaks (Price et al., 2003) so that deviations from the most common wild type will be strongly penalized in terms of losses in fitness. Indeed, tests with honeybees (Benatar et al., 1995; Brandes, 1988; Scheiner et al., 2001) and fruit flies (Lofdahl et al., 1992; Tully, 1996) have shown that measurably faster or slower learners can be bred in very few generations. If artificial selection can easily produce faster‐than‐wild‐type learners, why has natural selection not done the same? The fact that bees do not learn as fast as they could do, indicates that natural selection stabilizes learning ability at an intermediate level, and that both faster and slower learners might have lower fitness and are therefore selected against. But why would faster learning be selected against? In nature’s dynamic pollination market in which the most profitable flower type is constantly changing, it would seem advantageous for foragers to be able to learn new associations quickly to keep pace with changing floral rewards. However, if the speed with which bees form associations compromise the fitness returns of a second trait (Mery and Kawecki, 2004), then this could produce a trade‐off between learning speed and this other trait. One such potential trade‐off could be between learning speed and efficient memory retrieval (Chittka, 1998). Foraging bees are continually amassing experience, learning many new associations, such as those between floral morphology, scent or color and reward, and new sensorimotor skills to obtain rewards from flowers effectively. While long‐term memory has sufficient capacity to store much of this information (Chittka, 1998; Greggers and Menzel, 1993; Menzel, 1990), problems might arise regarding the organization and retrieval of this stored information. Since information is very hard to eliminate once stored in long‐term memory (Chittka, 1998) and information retrieval becomes both slower (Chittka and Thomson, 1997) and less efficient (Chittka et al., 1995, 1997) as more information is stored, it makes adaptive sense to limit both the amount and the rate of information input to long‐term memory. One potential way to regulate this problem is by limiting the input to long‐term memory to information which has shown its salience in large numbers of trials. The high levels of intercolony variation we have demonstrated in learning performance also raise some important methodological considerations of sample size. Clearly, care must be taken when making comparisons between species or populations based on small number of colonies, or when examining correlations between learning performance and other

EVOLUTION OF BEE FORAGING BEHAVIOR

323

parameters. While there was no overall correlation between bee age and learning performance, we did find significant correlations between bee age and learning speed in 3 out of 16 colonies (2 positive and 1 negative correlation: Raine et al., 2006). Thus, randomly selecting a single colony from this population would produce a significant correlation between age and learning speed in almost 1 in 5 (20%) cases. Thus, when designing experiments it is important to consider the potential significance of variation among, as well as within, colonies when deciding how to allocate finite sampling effort. Animal species differ widely in their cognitive capacities, and it is commonly assumed that such differences reflect adaptations to the natural conditions under which these animals operate (Dukas, 1998; Gallistel, 1990; Shettleworth, 1998). The evidence for this view comes from interspecific comparisons and correlative studies (Dukas and Real, 1991; Sherry, 1998). For example, vole species with larger home range size have, on average, better spatial memory, and the hippocampi (brain areas which store spatial memories) in such animals are typically larger (Sherry and Healy, 1998). An alternative way to address the question of the adaptive value of variation in cognitive capacities could be to examine the link between intraspecific variation in learning ability and fitness under ecologically relevant conditions. As the colony represents the unit of selection in social insects, the intercolony variation we have demonstrated represents the raw material on which selection for learning ability might act. This forms a solid basis from which to explore the potential adaptive value and constraints imposed on such variation in the economy of nature.

VII. RECIPROCAL POPULATION TRANSPLANT EXPERIMENTS: A TEST OF LOCAL ADAPTATION A rarely used but potentially powerful method of testing the adaptiveness of a (foraging) behavior is by testing an animal’s (foraging) performance under natural conditions in its native habitat and then transplanting this animal into a second animal’s native environment and retesting its performance. Crucially, the second animal’s foraging performance must also be measured in both its native habitat and that of the first animal—hence a reciprocal transplant experiment (Chittka et al., 2004; Ings et al., 2005b; Riechert and Hall, 2000). A necessary implication of the notion that animals are best adapted to foraging in their own habitat is that native animals should outcompete animals from other populations in terms of foraging performance in their native environment. Therefore, we set up reciprocal transplant experiments in which we compared the foraging

324

NIGEL E. RAINE ET AL.

performance of B. t. terrestris from central Europe with B. t. sassaricus from Sardinia and B. t. canariensis from the Canary Islands. All 27 tested colonies had been raised under identical conditions, including ad libitum provision of nectar and pollen. We therefore conjecture that any between‐ population differences at the start of the foraging career of individuals would be genetically determined. Test colonies were at a comparable developmental stage at the start of each experiment, that is, colonies were young and vigorous, and had 30–50 workers. We measured the nectar collection rate (weight of nectar collected per unit foraging time) of bee colonies from each of these populations at three sites: Costa Rei (southern Sardinia, autumn 2000), Monte Padru (northern Sardinia, spring 2001), and Wu¨rzburg (Germany, summer 2002). Ideally we would have liked to test our bee populations at a field site in the Canary Islands, but this was impossible as local authorities prohibit the import of nonnative bees. At the three sites chosen, we tested the foraging performance of nine bee colonies, that is, three from each population. All foragers were individually marked, and their flight departure and arrival times and weights were recorded for each foraging bout. There was no selection of foragers to be tested: we simply monitored all bees motivated to forage (Chittka et al., 2004). We expected that Sardinian B. terrestris would perform better in their native Sardinian habitat than either bees from Germany or the Canary Islands. Likewise, we expected that mainland B. terrestris would be the superior foragers in their native Germany. We also predicted that B. t. canariensis, as a nonnative of either site, would perform worse than either native population in their native habitats. Surprisingly, however, B. t. canariensis performed best at all three sites. B. t. sassaricus was consistently second: it performed better than German B. t. terrestris not only in its native Sardinia but also most surprisingly in Germany (Fig. 7; Ings et al., 2005b). Thus, our hypothesis that each population is best adapted to its native habitat in terms of foraging behavior cannot be upheld. One possible explanation for between‐population differences in foraging performance could be that members of different populations differ in body size, since body size is a good predictor of foraging rate within populations (Goulson et al., 2002; Spaethe and Weidenmu¨ller, 2002). We measured body mass of all foragers tested as body mass is highly correlated with size (Goulson et al., 2002). It turns out that body sizes of the three populations tested fall into the following order: B. t. canariensis >B. t. sassaricus >B. t. terrestris, that is, exactly the same rank order as that established for foraging rates (Chittka et al., 2004; Ings et al., 2005b). These differences in body size are not a consequence of variation in foraging performance as all colonies were fed pollen and nectar ad libitum prior to the start of field

EVOLUTION OF BEE FORAGING BEHAVIOR

325

Fig. 7. Nectar foraging performance of three populations of B. terrestris in different test locations. One ‘‘native’’ bee population (B. t. sassaricus in Sardinia and B. t. terrestris in Germany) is compared against two nonnative bee populations at each test location. Columns represent pooled mean (
1 SE) nectar foraging rates of bees from three colonies per population at each location. Numbers in bars are sample sizes, that is, the number of bees that performed three or more foraging trips. Data from Ings et al. (2005b).

trials using freely foraging bees. We monitored each nest for less time than it takes for a worker to develop (ca. 22 days from newly laid eggs to eclosion: Duchateau and Velthuis, 1988; Shykoff and Mu¨ller, 1995), so worker size could not be a result of colony foraging performance during the experiments. These results strongly suggest that worker size is an important factor in determining the foraging intake of a bumblebee colony, in fact, perhaps so important that between‐population differences in forager size may obscure the effects of other traits such as those of color preference (Ings et al., 2005b). There are a variety of reasons why larger foragers might be better foragers, but why are island foragers larger in the first place? In general, small‐bodied animals tend to be larger on islands than on the mainland: Foster’s (1964) ‘‘Island Rule.’’ Palmer (2002) showed that beetle body size increases with island size until reaching its maximum and then subsequently decreases with further increases in island size. One explanation for the island rule is that ecological release from predators and competition leads to an initial increase in body size, while resource limitation leads to

326

NIGEL E. RAINE ET AL.

size reductions at larger island size (Brown and Lomolino, 1998). Clearly we need comparative data on resource availability and predation levels on islands and the mainland to resolve this issue. But one important lesson here is this: since worker size might be under selective pressures wholly unrelated to foraging, for example, thermoregulation (Bishop and Armbruster, 1999; Corbet et al., 1993; Willmer and Stone, 2004) or predator pressure (Dukas and Morse, 2003), apparent foraging adaptations may in fact be exaptations: that is, the result of traits historically evolved for other purposes (Gould and Lewontin, 1979).

VIII. MANIPULATION

OF THE

FORAGING ENVIRONMENT: SCENT MARKING TRAPLINING

AND

One possible approach to studying the adaptive significance of a foraging strategy is to manipulate the environment in such a way that the foraging strategy cannot be used. For example, bees use the scent marks they deposit when visiting a flower as an olfactory cue to minimize the risk of revisiting recently emptied flowers (Giurfa and Nu´n˜ez, 1992, 1993; Goulson et al., 2000; Saleh et al., 2006). In order to test the adaptive benefits of bees’ ability to respond to these cues, Giurfa and Nu´n˜ez (1992) eliminated these floral scent marks by means of an air extractor in a flight arena and found that this resulted in significant decrease in the number of recently visited flowers rejected when the fan was turned on (mean
1 SE ¼ 11.43
0.79 rejections per flower visit with fan off vs 0.13
0.05 with extractor on: t ¼ 14.24, p <0.001; Giurfa and Nu´n˜ez, 1992), suggesting that the ability to correctly interpret scent marks is a highly important and adaptive component of bee foraging. Here we test the adaptive significance of another foraging strategy, traplining. In analogy with a trapper checking his traps in a fixed stable sequence, bees often visit flowers, or patches of flowers, in repetitive orders (Collett, 1993; Heinrich, 1976; Manning, 1956; Thomson, 1996; Thomson et al., 1982, 1987, 1997). In a field study, Williams and Thomson (1998) found that traplining bees harvested more nectar per unit time than casual foragers (bees foraging opportunistically within the same flower patch). But how can the advantages of traplining be explained? Williams and Thomson (1998) found that the greater efficiency of traplining bees in collecting rewards primarily resulted from greater selectivity. Traplining bees could select, on average, more rewarding flowers within a patch than those selected by casual foragers. This ability to select the most profitable flowers appeared to be the result of the fact that traplining bees were better able to reject recently visited, resource‐depleted flowers, that is, those

EVOLUTION OF BEE FORAGING BEHAVIOR

327

bearing scent marks (Williams and Thomson, 1998). But why did trapliners respond more strongly to such scent cues? One possibility is that bees with extensive local experience might be better able to respond to repellent scent cues within a floral patch. Traplining bees would build up such local experience while making repeated circuits of visits to the same flowers, plants, and flower patches. In addition, bees might also be able to distinguish their own scent marks from those deposited by other bees (Giurfa and Nu´n˜ez, 1993). If so, traplining foragers might use scent marks as a backup strategy to minimize the risk of visiting recently depleted flowers (Thomson and Chittka, 2001). In order to tease apart the relative benefits of using scent marks and traplining, we used an experimental design that removed the possibility for bees to visit flowers in a stable sequence, that is, they could no longer trapline. Bumblebee workers (Bombus impatiens Cresson) were trained to empty six large artificial flowers (colored plastic chips, diameter ¼ 3 cm), each containing a sucrose solution reward, placed in a flight arena (Thomson and Chittka, 2001). We ensured workers needed to visit all six flowers by adjusting the total volume of sucrose solution available in the flowers to the size of their honeycrop. Since bumblebees foragers vary in size (Goulson et al., 2002), and therefore in honeycrop capacity (Spaethe and Weidenmu¨ller, 2002), we needed to determine maximum honeycrop load size for each individual worker to be tested. This was done by presenting each bee with 15 large artificial flowers, each containing a 10 ml sucrose solution reward, and counting the number of flowers it visited per foraging bout (Thomson and Chittka, 2001). For subsequent tests, each large flower was filled with a reward equal to one‐sixth of the test bee’s honeycrop volume. Two groups of bumblebees were tested for 40 foraging bouts per individual. We evaluated the performance of each forager in the final 20 bouts to ensure that bees had reached saturation level in terms of familiarizing themselves with the task (Thomson and Chittka, 2001). Bees in the first group found the flowers in fixed positions in subsequent bouts, while flower positions varied randomly between foraging bouts for bees in the second group. Thus, bees foraging from the random arrangements of flowers had no opportunity to form traplines: that is, they had to seek out the positions of the six flowers de novo in each successive foraging bout. Large flowers (diameter ¼ 3 cm) were used in all of these tests, irrespective of whether the spatial arrangement of flowers was held constant, or randomized, between subsequent foraging bouts (Thomson and Chittka, 2001). We measured the search time taken by each bee to find all six rewarding flowers: that is, the flight time from entering the flight arena to when the bee first visited the sixth rewarding flower minus the time spent feeding from the other five flowers.

328

NIGEL E. RAINE ET AL.

Surprisingly, bees foraging from large flowers appeared to be entirely unaffected by being unable to form traplines. The total flight time taken to find all six flowers was statistically indistinguishable between bees allocated to the random or the constant spatial arrangement of flowers (Mann–Whitney U ¼ 13, p ¼ 0.86; Thomson and Chittka, 2001; Fig. 8A). Also, while the number of revisits made to already emptied flowers was higher in the group of bees foraging from the random (mean ¼ 3.4) as opposed to the constant flower arrangement (mean ¼ 2.0), this difference was not significant (U ¼ 8.5, p ¼ 0.29: Thomson and Chittka, 2001). Bees foraging from the constant arrangement of flowers clearly visited flowers in a highly repeatable sequence (Thomson and Chittka, 2001), but this gave them no measurable advantage over bees that had to actively search for all six flowers in each new foraging bout. So does this mean traplining represents a behavioral pattern without adaptive benefits? It is possible that using a stable sequence of flight vectors (traplining) is particularly advantageous when flowers are hard to find, that is, when they

Fig. 8. The relative benefits of traplining (visiting flowers in a stable sequence) depend on whether flowers are (A) large or (B) small. Bumblebees (B. impatiens) were trained to empty six artificial flowers placed in a spatial arrangement which either remained stable (open columns) across, or was randomized between (shaded columns), subsequent foraging bouts. Bees foraging from flowers in a constant, stable arrangement could form traplines, those foraging from randomly arranged flowers could not. In the first experiment (A) all bees foraged from large flowers (diameter ¼ 3 cm), while in the second (B) all flowers were small (diameter ¼ 1 cm). Column heights indicate the mean flight time (
1 SE) for bees to find all six flowers in each test group (minus the time spent on flowers and imbibing nectar). Numbers in each column are the number of bees tested in each treatment (N ¼ 40 foraging bouts per bee tested). Significant differences between stable and random arrangements of flowers for each experiment are indicated with an asterisk. Data from Thomson and Chittka (2001) and Saleh and Chittka (unpublished).

EVOLUTION OF BEE FORAGING BEHAVIOR

329

are either far apart or sufficiently small that they are difficult to detect. Flowers with a diameter of 3 cm, like the large ones used in our first test above, would be detectable from a distance of 34 cm: given that a target (here a flower) needs to subtend an angle of approximately 5 to be detected by an average‐sized worker bumblebee (Spaethe and Chittka, 2003; Spaethe et al., 2001). Thus, a forager in our test flight arena (floor dimensions: 105  75 cm2), containing six randomly arranged large flowers, will almost always be able to detect the nearest flower(s) from wherever it is currently foraging. Thus, it might simply not be very challenging for bees to locate such large flowers at a relatively high density. In a scenario in which flowers are smaller, and thus more difficult to detect, a bee would probably need to search considerably harder to find each flower (unless the bee already knows their location). For a bee foraging in such an environment, a strategy allowing it to memorize flower locations, and learn to link them with a set of flight vector instructions (e.g., ‘‘first, fly 50 cm in a northeast direction, then 20 cm west,’’and so on), might have a clear advantage over a strategy in which flowers must to be located afresh in each foraging bout. To test this idea, we repeated our first experiment with flowers of smaller size: diameter ¼ 1 cm (Thomson and Chittka, 2001; Saleh and Chittka, unpublished). In this situation, randomizing the spatial arrangement of flowers from one bout to the next had a strong effect (U ¼ 31, p ¼ 0.023: Thomson and Chittka, 2001; Saleh and Chittka, unpublished): the time taken to locate all six flowers increased by more than 60% (Fig. 8B). Likewise, the mean number of revisits to previously emptied flowers increased from 2.4 (stable) to 4.9 (random), and this difference was also highly significant (U ¼ 29, p ¼ 0.009: Thomson and Chittka, 2001; Saleh and Chittka, unpublished). As bees in all treatments had equal access to the scent marks (those the forager itself left) on flowers, any differences in the frequency of revisits to empty flowers could only have been produced by differences in the spatial arrangement of flowers. Our findings strongly suggest that bees use a combination of traplining and scent‐marking flowers to avoid revisiting resource‐depleted flowers. However, it seems that the adaptive benefits of traplining are context dependent: in situations where flowers are hard to detect (because they are either small and/or widely spaced), traplining gives bees a clear advantage over others which do not implement a stable flight route connecting memorized flower locations. When floral detection imposes no constraints on foraging performance, that is, when flowers are large (highly apparent) and/or closely packed together, more ‘‘random’’ spatial movements do not appear to be detrimental to foraging performance. In accordance with these findings, wild bumblebees (Bombus ternarius) foraging from natural flowers displayed a clear tendency to trapline when foraging from widely spaced sarsaparilla (Aralia hispida

330

NIGEL E. RAINE ET AL.

Vent.) plants (Thomson et al., 1982), whereas they show no such tendency when foraging from dense stands of goldenrod (Solidago spp.) plants (Thomson and Chittka, 2001).

IX. MANIPULATING FORAGING PHENOTYPES: THE HONEYBEE DANCE The honeybee dance language is regarded by many as one of the most intriguing communication systems in nonhuman animals (Chittka, 2004; Frisch, 1955). A successful scout bee returns from the field and advertises the location of a newly discovered food source to nestmates. To do this, the forager performs a repetitive figure‐eight‐shaped sequence of movements, the so‐called ‘‘waggle dance.’’ In the darkness of the hive, the successful forager waggles her abdomen from side to side, while moving forward in a straight line: the ‘‘waggle (wagtail or wagging) run.’’ Then she runs in a half circle to the left, back to her starting point, before performing another straight waggle run, after which she circles to the right to reach her starting point once again, thereby completing a waggle dance circuit. This pattern is repeated multiple times and is eagerly attended by bees in the hive. Shortly after such dances commence, scores of newly recruited foragers will arrive at the food source being advertised (Frisch, 1967; Seeley, 1995). But what were the ecological conditions under which such a dance language evolved, and what are its benefits to colony foraging performance? An ideal approach to studying this question would be to study a knockout animal, or mutant, in which dance communication is disrupted, but which otherwise functions completely normally. Unfortunately, such study systems are not currently available in honeybees. Therefore, we examined this question by creating experimental phenotypes in which the location information of the dances was eliminated. In order to try to understand the adaptive significance of the dance language, we decided to measure the performance of bee colonies under natural conditions and compare it to conditions under which the information flow between dancers and recruits was disrupted (Dornhaus, 2002; Frisch, 1967; Kirchner and Grasser, 1998; Sherman and Visscher, 2002). To these ends, we used a simple trick to disrupt the normal process of information transfer from dancer to recruit. Under normal conditions, the angle of the forager’s waggle run relative to the direction of gravity on the vertical comb indicates the direction of the food source relative to the azimuth of the sun (Frisch, 1955, 1967). However, by tilting the combs into a horizontal position we eliminated the possibility for bees to use gravity as a reference (Dornhaus and Chittka, 2004). Therefore bees performed dances in chance directions, so that dances lose their directional

EVOLUTION OF BEE FORAGING BEHAVIOR

331

information component. Having ‘‘interpreted’’ these nondirectional dances, recruits leave the hive in random directions (Dornhaus, 2002; Frisch, 1967; Kirchner and Grasser, 1998). However, if bees are offered a direct view of the sun or polarized light, then a returning forager can perform a correctly oriented waggle dance (with respect to the sun rather than to gravity) on a horizontal surface (Frisch, 1967). We used specially constructed hives in which combs were arranged horizontally. The top was fitted with a window, so the first comb would be exposed to the sun if the window was uncovered. Hence covering this window allowed us to eliminate the directional component of a returning forager’s waggle dance (Dornhaus, 2002; Dornhaus and Chittka, 2004). Initially, we compared the success of colonies that were able, or unable, to communicate the direction of profitable food sources in two temperate locations representative of the present distribution of European honeybees, A. mellifera, in spring. The experimental sites were a typical Mediterranean habitat in the Sierra Espada´n Nature Reserve, Spain, and a site near Wu¨rzburg, Germany, where agricultural land is mixed with natural meadows. We placed a pair of hives with 10 horizontal combs and ca. 5000 workers in each location. This is the sort of colony size one might expect to find in the wild, and it ensured that colony foraging would not be limited by the space available for honey storage. Each colony was switched from oriented to disoriented dancing every 2 days by uncovering or covering the window on top of the hive, respectively. Colony success was assessed using the daily weight gain of hives, which predominantly reflects nectar intake (Seeley, 1995). Surprisingly, we found no difference in weight gain, at either European site, between days in which colonies were able to follow oriented or disoriented waggle dances from returning foragers (Dornhaus and Chittka, 2004). To confirm that this was not simply a consequence of the time of year, we repeated the same experiment with two three‐comb hives monitored from May to September in Wu¨rzburg, Germany. However, even over this extended timescale, we again found no effect of obscuring the directional dance information (Dornhaus and Chittka, 2004). In both experiments, hive net weight changes were quite often negative, that is, the hive lost weight over a 24‐hr period, except on those days when bees apparently discovered a rich nectar flow. This is similar to the patterns Seeley (1995) has found in his foraging experiments in North America. So why bother communicating the direction to profitable food sources? Are the elaborate dances of European honeybees a useless behavioral feat? It seems highly counterintuitive, especially when one considers the enormous efficiency of the dance language to recruit bees to single points in space (Dyer, 2002; Frisch, 1967; Gould, 1975; Towne and Gould, 1988).

332

NIGEL E. RAINE ET AL.

However, to understand why animals behave the way they do, we must consider their ecological history as well as the conditions under which they currently operate. A. mellifera, the European honeybee in which the dance language was first described, occurred historically in temperate habitats west of the Iranian desert (Ruttner, 1987). The honeybee spread unassisted into sub‐Saharan Africa from Europe via Arabia, whereas its colonization of the new world tropics and Australia is the result of human intervention (Ruttner, 1987). However, A. mellifera shares the dance with all other species of honeybees (genus Apis), most of which are limited in their distribution to tropical Asia (Ruttner, 1988). The evolutionary origins of these dances are therefore thought to have occurred in an open‐nesting tropical ancestor of extant honeybees (Dyer and Seeley, 1989). These ancestral honeybees foraged under conditions wholly different from those in which modern European A. mellifera colonies find themselves. In tropical forests, floral food sources are predominantly arboreal and patchily distributed in space. Individual trees frequently offer many thousands of flowers at a very precise spatial location within the forest, and there are often large distances between trees flowering at the same time (Bawa, 1983, 1990; Roubik, 1992). This is in marked contrast to most temperate habitats in which widely distributed herbs and shrubs form a significant component of a bee’s diet (Heinrich, 1979). To test if the dance language is more essential to efficient foraging in tropical than in temperate habitats, we repeated our experiment with A. mellifera in the tropical dry deciduous forest of Bandipur Biosphere Reserve, India. We found no difference in the foraging capability of hives with vertical combs (the natural comb orientation) compared to hives with horizontal combs in which bees could perform oriented dances (i.e., the window atop the hive was uncovered: Dornhaus and Chittka, 2004). But scrambling the information content of the dance, by covering this window, reduced the number of successful foraging days by 85% (Fig. 9). The median weight gain on days with oriented dances was 5 g compared to 65 g on days when location communication was disrupted (Kolmogorov‐ Smirnov Test, p ¼ 0.02, N ¼ 45: Dornhaus and Chittka, 2004). In a similar study, Sherman and Visscher (2002) showed that season may also be a factor influencing whether the waggle dance actually increases foraging success. One explanation for differential effects of preventing bees from communicating in different habitats or seasons is different spatial distribution of resources. Since mapping the actual flower distribution in the bees’ foraging range (100 km2; Seeley, 1995) is effectively impossible, we used the information that the bees themselves provide in their waggle dances to map the locations where they forage (Visscher and Seeley, 1982). Using this

EVOLUTION OF BEE FORAGING BEHAVIOR

333

Fig. 9. Foraging performance of A. mellifera colonies with (oriented dances) and without (disoriented dances) the ability to communicate directional information about the location of food sources to nestmates through their dance language. Columns indicate the percentage of days on which each colony increased in weight. A significant effect of disrupting information between dancers and recruits was found in the tropical (indicated by the asterisk), but not in the two temperate habitats, where bees foraged equally well with and without directional communication about location of food sources. Data from Dornhaus and Chittka (2004).

approach, we created foraging maps for the Indian site by extracting information on the distance and direction of foraging sites from the hive from videotaped dances of returning foragers (Dornhaus and Chittka, 2004). This method has previously been used to create forage maps of honeybees in several habitats: temperate forest (Visscher and Seeley, 1982), African tropical forest (Schneider, 1989), a disturbed suburban habitat (Waddington et al., 1994), and a disturbed habitat mixed with more natural open moors (Beekman and Ratnieks, 2000). To see if the degree of clustering varied between different habitats, we calculated the patchiness of foraging sites (following Clark and Evans, 1954) for our maps and those previously published. We found that bees at our Indian site foraged up to 10 km from the hive, but that most dances indicated foraging sites much closer (ca. 500 m) to the colony. Honeybee foraging sites were very patchily distributed within the Indian dry deciduous forest (Dornhaus and Chittka, 2004). Indeed it appears that floral resources are significantly more patchily distributed in tropical forests (Dornhaus and Chittka, 2004; Schneider, 1989) than temperate habitats (Beekman and Ratnieks, 2000; Visscher

334

NIGEL E. RAINE ET AL.

and Seeley, 1982; Waddington et al., 1994). There was also appreciable variation in the patchiness of honeybee foraging sites among these temperate habitats, with temperate forests showing the most aggregation of floral resources. Therefore, the degree of forest cover could be an important factor determining the patchiness of honeybee food sources. Our findings suggest that the honeybee dance language is an adaptation to the tropical conditions under which the genus Apis diversified and may no longer be essential for efficient foraging in the temperate habitats studied. Here, it may have been maintained simply because it confers no selective disadvantage. In support of the argument that the dance language is more crucial under tropical conditions, Towne and Gould (1988) found that the precision of direction communication is higher in tropical than in temperate species. When food is less aggregated in space than in tropical forest, foraging by individual initiative, or communication through floral scent, may be as efficient as dance communication (Dornhaus and Chittka, 1999). Alternatively, stabilizing selection might have occurred through nonforaging applications of the dance such as indicating the location of nesting sites (Weidenmu¨ller and Seeley, 1999).

X. GENETIC BASIS

OF

FORAGING BEHAVIOR

If we understood the genetic basis of foraging behavior, that is, the identity and number of genes involved, this would clearly give us a better understanding of the evolvability of traits that influence foraging and the extent to which foraging behavior is adapted to a given niche (Ben‐Shahar et al., 2002; Whitfield et al., 2003). It is likely that most behavioral traits are polygenic, and linked through pleiotropies, that is, correlated characters (Amdam et al., 2004; Chittka et al., 2001), and therefore selection on any of them might have complex effects (Rueppell et al., 2004a,b). This notion is confirmed by a series of studies by R. E. Page and colleagues, who have explored the genetic architecture, as well as the physiological and molecular basis of a variety of foraging‐related traits in the behavior of the honeybee (A. mellifera: Page and Robinson, 1991; Page et al., 1995; Pankiw et al., 2002; Robinson et al., 1989; Rueppell et al., 2004a,b). They started by selecting two strains of honeybee colonies for a single characteristic: the amount of pollen collected and stored (Page and Fondrk, 1995; Page et al., 1995). Within a few generations, they had selectively bred two lines of bees that strongly differed in the relative effort they devoted to nectar and pollen foraging. The resulting bee strains differed in multiple aspects of foraging behavior that could either be linked directly or through pleiotropies to pollen foraging. The ‘‘high strain’’ colonies (those which hoarded

EVOLUTION OF BEE FORAGING BEHAVIOR

335

more pollen) not only had more pollen foragers, and collected larger pollen loads (Pankiw and Page, 2001), but they also initiated foraging at a younger age and collected smaller and less concentrated nectar loads (Pankiw and Page, 2001). However, because foraging loads were not measured as a function of foraging flight duration, these data are not indicators of differential foraging performance. On the sensory level, proboscis extension reflex experiments showed that ‘‘high strain’’ bees were more sensitive to low concentrations of sucrose (Page et al., 1998), and the authors conjecture that this might explain their higher relative acceptance level for poor nectar quality (Pankiw and Page, 2000). Indeed sensitivity to other chemosensory stimuli, such as pheromones produced by the brood, might also be elevated (Pankiw and Page, 2001). This suggests that there might be an overall improvement of sensory function in these bees, which could in turn explain their superior performance in both olfactory and tactile learning paradigms (Scheiner et al., 2001). However, there may also be changes at the level of the central nervous system: Humphries et al. (2003) found higher levels of protein kinases A and C in the brain of bees selected for high pollen hoarding—both of these kinases play roles in memory consolidation and avoidance conditioning (Shobe, 2002). Also, Amdam et al. (2004) proposed that all of these differences might be pleiotropically linked to reproductive behavior. There are multiple implications of these findings for the study of the adaptiveness of foraging behavior. The good news is that researchers are homing in on the genetic architecture underlying foraging behavior, that several foraging‐related traits are heritable, and that therefore the raw material for selection, both natural and experimental, exists. This opens up the possibility to study the adaptive benefits of these traits in the wild, especially since nonlethal DNA sampling techniques have been refined for bees (Chaˆline et al., 2004; Holehouse et al., 2003). However, the interpretation of the potential differences in fitness will be difficult. This is because selection on any one trait is likely to drag along a host of other traits, which may all operate under a variety of environmental constraints, and might therefore affect fitness in different ways.

XI. MODELING In behavioral ecology, two types of models have traditionally been used to study adaptation (Judson, 1994; Ydenberg and Schmid‐Hempel, 1994). Mathematical descriptions of a behavior and its fitness consequences are often very abstract, and therefore simplified, but generally applicable (Maurer and Se´guinot, 1995). They can usually be solved analytically, making predictions about the optimal trait value that maximizes fitness

336

NIGEL E. RAINE ET AL.

and how fitness will change away from that optimum. Computational models, on the other hand, cannot be solved without using numerical values for the parameters involved (Grimm, 1999). Such models can, for example, be rule‐based descriptions of behavior and its fitness consequences, as is often the case in individual‐based simulation models, or they can be equation‐based models that are too complex to be solved analytically. Both of these model types are powerful tools to test whether animals are adapted to particular environments; however, each also has its own potential pitfalls, which may tempt the observer to infer optimality of behaviors for the wrong reasons. For example, mathematical models of optimal foraging and load size predict that bees should return from a food source without collecting a full load if the bee is trying to maximize energetic efficiency rather than reward collection rate (Schmid‐Hempel, 1987; Schmid‐Hempel et al., 1985). However, several other models also predict such submaximal loads (Cuthill and Kacelnik, 1990): for example, those assuming diminishing returns at the food source (Ydenberg and Hurd, 1998), or those cases where there is some chance of sharing (Varju and Nu´n˜ez, 1991, 1993) or receiving (Dornhaus et al., in press) information on high‐quality food sources when the bee returns. Each of the models by itself can be used to argue that not collecting a full load at a food source is the optimal strategy. Unless some of these models use assumptions that do not reflect the situation of foraging bees, the bee’s load size is the result of the combined effects of all these factors. In this case, none of the models alone would be sufficient to explain the full deviation from the maximal load size in foraging bees. It is therefore important not to exclude alternative hypotheses because an observed effect is consistent with one model. Like all scientific hypotheses, models that are falsified can ultimately be more interesting than those that are consistent with data, because we can deduce how the biological system does not work; whereas models that are consistent with data may, or may not, reflect the true mechanisms underlying real biological processes. There is an additional difficulty associated with the ‘‘exact’’ solutions achieved in analytically solved models. In such models, analysis often focuses on the mathematically ‘‘interesting’’ areas. However, it might well be that what is mathematically interesting is not biologically relevant. While a model might show a trait to have several optima, only one of these may be at biologically feasible values of that trait. It is therefore crucial to apply the model to experimental data and to check that the concluded effects apply in a biologically relevant region of parameter space (Grimm, 1994; Kacelnik et al., 1986; May, 2004). Similarly, it is important to derive quantitative predictions from a model (Orzack and Sober, 1994), for example, about foraging behavior. Not only should the model make quantitative predictions about the optimal value of the studied foraging trait, it should

EVOLUTION OF BEE FORAGING BEHAVIOR

337

also predict the magnitude of the benefits of optimizing this trait. In other words, it should estimate how big an advantage is gained by optimizing this particular trait, something that is seldom addressed in studies of optimal foraging. Very small effects can be hard to detect in biological data, and may indeed be too small to cause significant selection pressure in the predicted direction. Also, if effects predicted by a model were much smaller than those observed, this would indicate that additional factors influence the measured trait, and that the model does not provide a full explanation. By quantifying the trait values predicted by a mathematical model, one loses some of its generality and exactness; but at the same time, one makes a more accurate assessment of the biological relevance of the model’s predicted effects. Computational models avoid such difficulties because they require the experimenter to think about relevant parameter values from the start. However, estimation of biological parameters is inherently inexact. This means that we cannot base conclusions on the assumption that any particular estimate is correct; we have to conduct a sensitivity analysis to test for the effects of all parameters involved, within their biologically plausible limits (Chittka et al., 1992). If this is not done systematically, it is all too easy to tune parameter values so that a particular result is achieved (Ginzburg and Jensen, 2004). However, with proper sensitivity analysis, computational models can be powerful tools in understanding which environmental and other factors are likely to have contributed to the evolution of observed traits (Chittka, 1996b; Judson, 1994). Full (or even limited) sensitivity analyses are very rarely published with computational modeling studies (exceptions are Bautista et al., 2001; Chittka et al., 1992; Schmid‐Hempel et al., 1985). However, a full sensitivity analysis was performed in a study of benefits of recruitment to food sources in bees (Dornhaus et al., 2006). Recruitment systems vary considerably between species of social bees (Chittka and Dornhaus, 1999; Dyer and Seeley, 1989; Lindauer and Kerr, 1958), and to develop hypotheses about the evolution of such systems, it is necessary to identify which social or ecological factors favor the evolution of recruitment. In the study by Dornhaus et al. (2006), an individual‐based model of honeybee foraging was developed to quantify the benefits of recruitment. These were measured under different spatial resource distributions and colony sizes. Benefits of recruitment in the simulations were found to be strongly dependent on resource patch quality, density, and variability. Communication was especially beneficial if patches were poorly rewarding, few in number, and variable (Fig. 10; Dornhaus et al., 2006). This result would not have been achieved had the interaction effect of environmental parameters on bee foraging success in the model not been studied. A sensitivity analysis was carried out in which each parameter value put

338

NIGEL E. RAINE ET AL.

Fig. 10. Foraging success in the individual‐based model was dependent on environmental parameters such as resource density and quality. Unsurprisingly, most energy was collected by the bee colony when there were many high‐quality resources. The model also predicts that recruitment has different effects under different conditions. The highest relative increase in energy collected is achieved by recruitment under conditions of few and poor resources. Each data point represents the average of 10 simulation runs. The shading on each model landscape indicates the amount of energy collected (same as y‐axis; black is a net energy loss) in the period simulated (50 hr). Data from Dornhaus et al. (2006).

into the model was varied to study its effect on foraging success of the modeled bees. Such a sensitivity analysis can be very time consuming, particularly if many parameters are involved, which is often the case particularly in individual‐based models. In the cited study, 4600 simulation runs were carried out and analyzed (Dornhaus et al., 2006). Some parameters that were varied within their biologically plausible limits had no effect at all; while others strongly influenced colony foraging success. The sensitivity analysis showed, for example, that under conditions of high resource density, recruitment could even become detrimental if foraging bout duration was short, the tendency to scout was high, or the recruits

EVOLUTION OF BEE FORAGING BEHAVIOR

339

needed a long time to find communicated locations. Colony size, the other main factor studied, has often been suspected to influence recruitment evolution but had no significant effect in the model (Dornhaus et al., 2006). These results may explain the experimental findings that in honeybees, benefits of waggle dance recruitment seem to vary seasonally and with habitat (Dornhaus and Chittka, 2004; Sherman and Visscher, 2002). Finally, when predictions of any models are compared with experimental results, it is important to distinguish between adaptive (‘‘optimal’’) behaviors and the mechanisms that enable animals to achieve them (Kacelnik, 1984). These mechanisms may not be identical with the way the optimal decision is computed in a model. Animal behavior arises from natural selection in a given environment, and certain rules of thumb may lead to the optimal behaviors in this environment but not necessarily in very artificial laboratory test situations (Herre, 1995). Such a situation would show animals behaving nonadaptively, but that does not prove that the trait under consideration is not under selection. Any modeling studies that produce quantitative predictions about traits and their fitness values, that provide full sensitivity analyses, and that test predictions by comparing them with the behavior of animals in their natural environment will advance our understanding of the evolution of these traits.

XII. DISCUSSIONS We have illustrated the value of a number of approaches taken from the toolbox of the modern evolutionary biologist, which can be used to study the adaptive nature of foraging behavior. When trying to establish the role and importance of the extant behaviors, we must consider the evolutionary processes by which these traits have been forged: adaptation, chance, and history are all likely to have played their part. So to determine the adaptiveness of a particular behavioral trait we must conceive our experiments such that we can distinguish adaptation from the effects of chance and history on the behavior in question (Adkison, 1995; Clutton‐Brock and Harvey, 1977). Putting this into practice in the economy of nature is never as straightforward as it may sound, due to the interrelated nature of many behavioral traits. Animals will typically be able to proceed along multiple evolutionary pathways to optimize foraging behavior, and constraints imposed by one foraging‐related trait might be easily compensated for by alterations in another trait (cf. Endler et al., 2001). For this reason, it is often necessary to use several different approaches to tease apart the effects of different traits and to establish whether, or under what conditions, any (or all) of them are adaptive.

340

NIGEL E. RAINE ET AL.

The essential first step is to quantify the raw material for selection of any behavioral trait of interest, that is, variation among individuals and/or colonies in the case of social insects. Where such variation exists, we can then attempt to correlate the trait with foraging performance in the wild, and ideally, with biological fitness (Maynard Smith, 1978). Where such variation is lacking, selection might have eliminated it in the past (Chittka et al., 2001), which makes a direct study of the adaptiveness of these traits more challenging. In such cases, modeling (McNamara et al., 1993), manipulations of the environment (Schmid‐Hempel and Schmid‐Hempel, 1998), or alterations of the behavioral phenotype (Curio, 1973) help us to understand the adaptive benefits conferred on their bearer by a given trait. Reciprocal transplant experiments are a useful tool to examine hypotheses of local adaptation (Kawecki and Ebert, 2004), and the comparative phylogenetic method (Harvey and Purvis, 1991) allows us to identify patterns of adaptation by comparing closely related species. We have applied this package of methods to a variety of foraging‐related behavior patterns, that is, flower constancy, flower color preference, flower color learning, traplining behavior, and bee communication about floral resources. We also aim to highlight some of the promising areas of future research: further foraging‐related traits which deserve attention using existing approaches to study their potential adaptive value (e.g., risk sensitivity, memory dynamics, and pollen foraging) and new techniques which could potentially be used to great effect in the study of adaptation (e.g., correlating foraging performance with actual biological fitness, and molecular genetic methods). Foraging bees face a complex challenge to assess accurately the floral rewards being offered in the dynamically changing pollination market. When trying to assess which flower type is currently the most profitable, a foraging bee must not only contend with differences in quality and quantity of reward among flower species but also the variation among plants within a species and even across flowers on an individual plant. There are many experimental laboratory studies on bee ‘‘risk’’ sensitivity to variance in reward (Chittka, 2002; Chittka and Wells, 2004; Fu¨lo¨p and Menzel, 2000; Shafir et al., 1999; Waddington, 2001). There are also many models that examine the potential adaptive benefits of responsiveness to reward variance and the mechanisms underlying it (Bateson and Kacelnik, 1998). Given this complex foraging problem, it would seem intrinsically interesting to investigate the potential effect of a bee’s risk sensitivity on its foraging performance, exploiting between‐species variation in this trait, or by examining performance in natural environments that differ in reward variance.

EVOLUTION OF BEE FORAGING BEHAVIOR

341

Memory dynamics and recall seem to be important in many aspects of bee foraging behavior, such as associative learning of floral cues and reward, and spatial learning of flower positions in traplining. Menzel (2001) has suggested that honeybee working memory dynamics are tuned to the flight interval between flower visits (ca. 3–5 sec). However, while such adaptive speculations for these cognitive capacities might seem intuitively appealing, the ideas require rigorous testing. If memory dynamics are tuned to the foraging process, then related insects which do not forage from flowers (e.g., some cleptoparasitic bees; Roubik, 1989) would be expected to have memory phases with different temporal dynamics. Likewise, studying the foraging performance of learning mutants, should these become available in bees, may be a rewarding avenue of future research; in Drosophila melanogaster, scientists can make use of a wide variety of memory mutants in which only specific phases of memory are rendered nonfunctional (Reif et al., 2002; Tully, 1991). The vast majority of foraging studies on bees concentrate on the collection of nectar. This is not altogether surprising as nectar foraging provides a convenient and much more easily manipulated model system. However, pollen collection is also crucial to the success of any bee colony, and the intrinsic differences between pollen and nectar mean that bees collect them in different ways. Given the need for bees to develop such divergent strategies to harvest these distinct floral resources, we cannot reasonably extend conclusions drawn from studies investigating nectar foraging bees to questions concerning pollen collection. When collecting nectar bees automatically receive instant feedback on its quality via taste receptors (Kuwabara, 1957) and quantity via stretch receptors as the honeycrop is filled (Neese, 1988). In contrast, bees gain only indirect information on the pollen quantity from the mass they collect in their corbiculae (Ford et al., 1981; Harder, 1990; Robertson et al., 1999; Schikora and Chittka, 1999), and any information about pollen quality (such as the relative composition and richness of essential amino acids) is harder to collect (Erhardt and Baker, 1990), except perhaps by odor (Dobson et al., 1996; Robertson et al., 1999), taste, or indirect feedback through colony development. Pollen quality may be particularly important because many bees which will opportunistically collect nectar from a variety of different flower species are much more particular about finding specific flower species from which to collect pollen (Waser et al., 1996; Westrich, 1989). This all begs the obvious question: what are the strategies that bees use in harvesting pollen and are these strategies adaptive? To show the biological relevance of a foraging‐related trait, we should ideally be able to quantify its impact on fitness. However, in foraging

342

NIGEL E. RAINE ET AL.

studies fitness is seldom directly measured but is frequently inferred through changes in proxy measures or correlates of fitness. In social bees, foraging performance is well correlated with colony production of sexuals (males and new queens), which will leave the nest, mate and set up the next generation (Ings et al., 2005b; Pelletier and McNeil, 2003; Schmid‐Hempel and Schmid‐Hempel, 1998). As such, foraging performance represents a good, but indirect, measure of fitness, while the number (or biomass) of sexuals produced by a colony gives us a more direct measure of fitness. To really understand the adaptiveness of (foraging) behavior, we need to link variation in a behavioral trait to changes in fitness. Measuring the fitness consequences of traits is challenging but obviously a desirable thing to do. While this has been done in some studies of parasitism in bees (Baer and Schmid‐Hempel, 1999; Mu¨ller and Schmid‐Hempel, 1992), it still needs to be achieved in the field of foraging behavior. Another desirable avenue of future research would be to gain an insight into the adaptiveness of behavioral traits at the genetic level. Researchers are closing in on isolating the genes that encode particular behavioral traits (Ben‐Shahar et al., 2002, 2003; Rueppell et al., 2004a,b; Whitfield et al., 2003). In the future, it might be possible to modify behavioral phenotypes by knocking out their expression using double‐stranded RNA interference (dsRNAi: Fire et al., 1998) or perhaps by creating more traditional knockout mutants (Lipp, 2002; Wolfer and Lipp, 2000). While it is occasionally possible to create behavioral phenotypes for traits without genetic techniques, such as removing the ability to encode distance information in the honeybee waggle dance (Section IX), the use of dsRNAi could extend the potential of this powerful approach (i.e., modification of natural behavioral phenotypes) for many other traits of interest. dsRNAi is being used to study functional mechanisms by knocking out gene function (Booth, 2004; Marie et al., 2000). Farooqui et al. (2003) have modified the behavioral phenotype using dsRNAi techniques to block the octopaminergic pathway in the antennal lobe of honeybees. As a result, these bees were unable to learn an odor paired with a sucrose reward because octopamine mediates the unconditioned stimulus (the reward) in this associative learning task. Continued advances in the search for other behaviorally important genes and refinements in the dsRNAi techniques could herald the beginning of a very powerful future tool for the study of adaptation in behavioral ecology.

XIII. SUMMARY Our goal in this chapter is to determine whether particular behavioral traits represent actual adaptations in the context of foraging. Social bees are our chosen study system because they provide a convenient and tractable

EVOLUTION OF BEE FORAGING BEHAVIOR

343

biological system with which to study the potential adaptiveness of a wide range of foraging traits such as flower constancy, floral color preference, learning to associate floral color as a predictor of reward, traplining, and communication about food sources. This variety of behavioral traits allows us to demonstrate the strengths and weaknesses of applying five approaches (four experimental and one theoretical) to the study of foraging at the species, population, and colony level. (1) The comparative approach allows us to contrast behavioral traits of extant species with those of their common ancestor. We correlated differences in floral color preference between closely related species (and populations), with a known phylogeny, with features in each bee’s respective environment. (2) Reciprocal transplant experiments allowed us to test for local adaptation. We compared the relative foraging performance of distinct bee populations in both of their respective native environments. (3) Manipulating the foraging environment to eliminate specific behavioral traits permitted a direct comparison of animals’ foraging performance in their normal and experimentally manipulated environment, allowing us to quantify the effect of the trait in question (traplining) on foraging performance. (4) Manipulating the foraging phenotype to eliminate specific behavioral traits is another valuable approach. Unless suitable behavioral mutants, knockouts, or molecular techniques to selectively block gene expression are available, creating such artificial foraging phenotypes is only possible for a very small number of specific traits, for example, the honeybee dance language. (5) Integrating biologically realistic modeling with experimental studies allows us to test predictions about the adaptive significance of foraging‐related traits not amenable to experimental manipulation and to identify the ranges over which these traits might affect fitness. Do these approaches provide evidence that foraging behaviors are adaptive? In some cases, we show that forager behavior has indeed been tuned to function adaptively in a given niche, although the adaptive benefits of such behavioral traits are often strongly context dependent. However, in other cases, the observed patterns of behavior were more parsimoniously explained by chance evolutionary processes, or by the historical conditions under which bees operated in their evolutionary past. Acknowledgments We would like to thank Chris Armstrong, Petra Frauenstein, Adrienne Gerber‐Kurz, Natalia Lopez, Oscar Ramos Rodrı´guez, Juliette Schikora, Annette Schmidt, Rohini Simbodyal, Kristina Stu¨ber, and Tulay Yilmaz for help with the experiments, and Jane Brockmann, Marc Naguib, Alice Sharp Pierson, Peter Slater, and an anonymous referee for comments on an earlier version of this chapter. This work was supported by grants from the NERC (NER/A/S/2003/00469) to L.C. and N.E.R., the Central Research Fund (University of London) to N.S. and T.C.I., and the DFG (Emmy Noether program) to A.D.

344

NIGEL E. RAINE ET AL.

REFERENCES Adkison, M. D. (1995). Population differentiation in Pacific salmon: Local adaptation, genetic drift, or the environment. Can. J. Fish. Aquat. Sci. 52, 2762–2777. Alcock, J. (1996). ‘‘Das Verhalten der Tiere.’’ Gustav Fischer, Stuttgart. Amdam, G. V., Norberg, K., Fondrk, M. K., and Page, R. E. (2004). Reproductive ground plan may mediate colony‐level selection effects on individual foraging behavior in honey bees. Proc. Natl. Acad. Sci. USA 101, 11350 –11355. Armbruster, W. S. (1992). Phylogeny and the evolution of plant‐animal interactions. Bioscience 42, 12–20. Baer, B., and Schmid‐Hempel, P. (1999). Experimental variation in polyandry affects parasite loads and fitness in a bumble‐bee. Nature 197, 151–154. Banschbach, V. S. (1994). Colour association influences honey bee choice between sucrose concentrations. J. Comp. Physiol. A 175, 107–114. Bar‐Shai, N., Samuels, R., Keasar, T., Motro, U., and Shmida, A. (2004). Flight durations in bumblebees under manipulation of feeding choices. J. Insect Behav. 17, 155–168. Barbosa, A., and Moreno, E. (1999). Evolution of foraging strategies in shorebirds: An ecomorphological approach. Auk 116, 712–725. Barrow, D. A., and Pickard, R. S. (1984). Size‐related selection of food plants by bumblebees. Ecol. Entomol. 9, 369–373. Barton, N. H. (1998). Natural selection and random genetic drift as causes of evolution on islands. In ‘‘Evolution on Islands’’ (P. R. Grant, Ed.), pp. 102–123. Oxford University Press, Oxford. Bateson, M., and Kacelnik, A. (1998). Risk‐sensitive foraging: Decision making in variable environments. In ‘‘Cognitive Ecology: The Evolutionary Ecology of Information Processing and Decision Making’’ (R. Dukas, Ed.), pp. 297–341. University of Chicago Press, Chicago. Bautista, L. M., Tinbergen, J., and Kacelnik, A. (2001). To walk or to fly? How birds choose among foraging modes. Proc. Natl. Acad. Sci. USA 98, 1089–1094. Bawa, K. S. (1983). Patterns of flowering in tropical plants. In ‘‘Handbook of Experimental Pollination Ecology’’ (C. E. Jones and R. J. Little, Eds.), pp. 394–410. Van Nostrand Reinhold, New York. Bawa, K. S. (1990). Plant‐pollinator interactions in tropical forests. Annu. Rev. Ecol. Syst. 21, 399–422. Beekman, M., and Ratnieks, F. L. W. (2000). Long range foraging by the honey bee, Apis mellifera. Funct. Ecol. 14, 490–496. Benatar, S. T., Cobey, S., and Smith, B. H. (1995). Selection on a haploid genotype for discrimination learning performance: Correlation between drone honey bees (Apis mellifera) and their worker progeny (Hymenoptera: Apidae). J. Insect Behav. 8, 637–652. Ben‐Shahar, Y., Robichon, A., Sokolowski, M. B., and Robinson, G. E. (2002). Influence of gene action across different time scales on behavior. Science 296, 741–744. Ben‐Shahar, Y., Leung, H. T., Pak, W. L., Sokolowski, M. B., and Robinson, G. E. (2003). cGMP‐dependent changes in phototaxis: A possible role for the foraging gene in honey bee division of labor. J. Exp. Biol. 206, 2507–2515. Betz, O. (1998). Comparative studies on the predatory behaviour of Stenus spp. (Coleoptera: Staphylinidae): The significance of its specialized labial apparatus. J. Zool. 244, 527–544. Bishop, J. A., and Armbruster, W. S. (1999). Thermoregulatory abilities of Alaskan bees: Effects of size, phylogeny and ecology. Funct. Ecol. 13, 711–724.

EVOLUTION OF BEE FORAGING BEHAVIOR

345

Booth, D. (2004). Sensory ecology of nocturnal insects: Two evolutionary case studies. D. Phil. Thesis, University of Sussex. Brandes, C. (1988). Estimation of heritability of learning behaviour in honeybees (Apis mellifera capensis). Behav. Genet. 18, 119–132. Briscoe, A. D., and Chittka, L. (2001). The evolution of color vision in insects. Annu. Rev. Entomol. 46, 471–510. Brooks, D. R., and McLennan, D. H. (1991). ‘‘Phylogeny, Ecology, and Behavior.’’ University of Chicago Press, Chicago. Brown, J. H., and Lomolino, M. V. (1998). ‘‘Biogeography.’’ Sinauer Associates, Sunderland. Chaˆline, N. G., Ratnieks, F. L. W., Raine, N. E., Badcock, N. S., and Burke, T. (2004). Non‐ lethal sampling of honey bee, Apis mellifera, DNA using wing tips. Apidologie 35, 311–318. Chittka, L. (1996a). Does bee colour vision predate the evolution of flower colour? Naturwissenschaften 83, 136–138. Chittka, L. (1996b). Optimal sets of colour receptors and opponent processes for coding of natural objects in insect vision. J. Theor. Biol. 181, 179–196. Chittka, L. (1998). Sensorimotor learning in bumblebees: Long‐term retention and reversal training. J. Exp. Biol. 201, 515–524. Chittka, L. (2002). Influence of intermittent rewards in learning to handle flowers in bumblebees (Hymenoptera: Apidae: Bombus impatiens). Entomol. Gen. 26, 85–91. Chittka, L. (2004). Dances as windows into insect perception. PLoS Biol. 2, 898–900. Chittka, L., and Briscoe, A. (2001). Why sensory ecology needs to become more evolutionary— insect color vision as a case in point. In ‘‘Ecology of Sensing’’ (F. G. Barth and A. Schmid, Eds.), pp. 19–37. Springer‐Verlag, Berlin. Chittka, L., and Dornhaus, A. (1999). Comparisons in physiology and evolution, and why bees can do the things they do. Ciencia al Dia 2, 1–17. Chittka, L., and Thomson, J. D. (1997). Sensori‐motor learning and its relevance for task specialization in bumble bees. Behav. Ecol. Sociobiol. 41, 385–398. Chittka, L., and Waser, N. M. (1997). Why red flowers are not invisible for bees. Isr. J. Plant Sci. 451, 69–183. Chittka, L., and Wells, H. (2004). Color vision in bees: Mechanisms, ecology and evolution. In ‘‘Complex Worlds from Simpler Nervous Systems’’ (F. Prete, Ed.), pp. 165–191. MIT Press, Boston. Chittka, L., Beier, W., Hertel, H., Steinmann, E., and Menzel, R. (1992). Opponent colour coding is a universal strategy to evaluate the photoreceptor inputs in Hymentoptera. J. Comp. Physiol. A 170, 545–563. Chittka, L., Geiger, K., and Kunze, J. (1995). The influences of landmarks on distance estimation of honey bees. Anim. Behav. 50, 23–31. Chittka, L., Gumbert, A., and Kunze, J. (1997). Foraging dynamics of bumble bees: Correlates of movements within and between plant species. Behav. Ecol. 8, 239–249. Chittka, L., Thomson, J. D., and Waser, N. M. (1999). Flower constancy, insect psychology, and plant evolution. Naturwissenschaften 86, 361–377. Chittka, L., Spaethe, J., Schmidt, A., and Hickelsberger, A. (2001). Adaptation, constraint, and chance in the evolution of flower color and pollinator color vision. In ‘‘Cognitive Ecology of Pollination’’ (L. Chittka and J. D. Thomson, Eds.), pp. 106–126. Cambridge University Press, Cambridge. Chittka, L., Ings, T. C., and Raine, N. E. (2004). Chance and adaptation in the evolution of island bumblebee behaviour. Popul. Ecol. 46, 243–251. Christy, R. M. (1884). On the methodic habits of insects when visiting flowers. J. Linn. Soc. 17, 186–194.

346

NIGEL E. RAINE ET AL.

Clark, P. J., and Evans, F. C. (1954). Distance to nearest neighbor as a measure of spatial relationships in populations. Ecology 35, 445–453. Clayton, N. S. (1995). Development of memory and the hippocampus: Comparison of food‐ storing and nonstoring birds on a one‐trial associative memory task. J. Neurosci. 15, 2796–2807. Clayton, N. S., and Krebs, J. R. (1994). Memory for spatial and object‐specific cues in food‐ storing and non‐storing birds. J. Comp. Physiol. A 174, 371–379. Clutton‐Brock, T. H., and Harvey, P. H. (1977). Species differences in feeding and ranging behaviour in primates. In ‘‘Primate Ecology’’ (T. H. Clutton‐Brock, Ed.), pp. 557–584. Academic Press, London. Collett, T. S. (1993). Localized learning in insects. Nature 365, 693. Corbet, S. A., Fussell, M., Ake, R., Fraser, A., Gunson, C., Savage, A., and Smith, K. (1993). Temperature and the pollinating activity of social bees. Ecol. Entomol. 18, 17–30. Crow, J. F., and Kimura, M. (1970). ‘‘An Introduction to Population Genetics Theory.’’ Harper and Row, New York. Curio, E. (1973). Towards a methodology of teleonomy. Experientia 29, 1045–1048. Cuthill, I. C., and Kacelnik, A. (1990). Central place foraging: A reappraisal of the ‘‘loading effect’’ Anim. Behav. 40, 1087–1101. Cuthill, I. C., Kacelnik, A., Krebs, J. R., Haccou, P., and Iwasa, Y. (1990). Starlings exploiting patches: The effect of recent experience on foraging decisions. Anim. Behav. 40, 625–640. Dafni, A., Bernhardt, P., Shmida, A., Greenbaum, S., O’Toole, C., and Losito, L. (1990). Red bowl‐shaped flowers: Convergence for beetle pollination in the Mediterranean region. Isr. J. Bot. 39, 81–92. Darvill, B., Knight, M. E., and Goulson, D. (2004). Use of genetic markers to quantify bumblebee foraging range and nest density. Oikos 107, 471–478. Dobson, H. E. M., Groth, I., and Bergstrom, G. (1996). Pollen advertisement: Chemical contrasts between whole‐flower and pollen odors. Am. J. Bot. 83, 877–885. Dornhaus, A. (2002). When do dances make a difference? Significance of honey bee recruitment depends on foraging distance. Entomol. Gen. 26, 93–100. Dornhaus, A., and Chittka, L. (1999). Evolutionary origins of bee dances. Nature 401, 38. Dornhaus, A., and Chittka, L. (2004). Why do honey bees dance? Behav. Ecol. Sociobiol. 55, 395–401. Dornhaus, A., Klu¨gl, F., Puppe, F., and Tautz, J. (1998). Task selection in honeybees— experiments using multi‐agent simulation. In ‘‘Third German Workshop on Artificial Life’’ (C. Wilke, S. Altmeyer, and T. Martinetz, Eds.), pp. 171–183. Verlag Harri Deutsch, Frankfurt, Germany. Dornhaus, A., Klu¨gl, F., Oechslein, C., Puppe, F., and Chittka, L. (2006). Benefits of recruitment in honey bees: Effects of ecology and colony size in an individual‐based model. Behav. Ecol. 17, 336–344. Dornhaus, A., Collins, E. J., Dechaume‐Moncharmont, F.‐X., Houston, A., Franks, N. R., and McNamara, J. (in press). Paying for information: Partial loads in central place foragers. Behav. Ecol. Sociobiol. Dramstad, W. E. (1996). Do bumblebees (Hymenoptera: Apidae) really forage close to their nests? J. Insect Behav. 2, 163–182. Duchateau, M. J., and Velthuis, H. (1988). Development and reproductive strategies in the bumble bee, Bombus terrestris. Behaviour 107, 186–207. Dukas, R. (1998). Evolutionary ecology of learning. In ‘‘Cognitive Ecology: The Evolutionary Ecology of Information Processing and Decision Making’’ (R. Dukas, Ed.), pp. 129–174. University of Chicago Press, Chicago.

EVOLUTION OF BEE FORAGING BEHAVIOR

347

Dukas, R., and Morse, D. H. (2003). Crab spiders affect flower visitation by bees. Oikos 101, 157–163. Dukas, R., and Real, L. A. (1991). Learning foraging tasks by bees: A comparison between social and solitary species. Anim. Behav. 42, 269–276. Dyer, F. C. (2002). The biology of the dance language. Annu. Rev. Entomol. 47, 917–949. Dyer, F. C., and Seeley, T. D. (1989). On the evolution of the dance language. Am. Nat. 133, 580–590. Emlen, J. M. (1966). The role of time and energy in food preference. Am. Nat. 100, 611–617. Endler, J. A., Basolo, A., Glowacki, S., and Zerr, J. (2001). Variation in response to artificial selection for light sensitivity in guppies (Poecilia reticulata). Am. Nat. 158, 36–48. Erhardt, A., and Baker, I. (1990). Pollen amino acids—an additional diet for a nectar feeding butterfly? Plant Syst. Evol. 169, 111–121. Estoup, A., Solignac, M., Cornuet, J.‐M., Goudet, J., and Scholl, A. (1996). Genetic differentiation of continental and island populations of Bombus terrestris (Hymenoptera: Apidae) in Europe. Mol. Ecol. 5, 19–31. Farooqui, T., Robinson, K., Vaessin, H., and Smith, B. H. (2003). Modulation of early olfactory processing by an octopaminergic reinforcement pathway in the honeybee. J. Neurosci. 23, 5370–5380. Fire, A., Xu, S., Montgomery, M. K., Kostas, S. K., Driver, S. E., and Mello, C. C. (1998). Potent and specific genetic interference by double‐stranded RNA in Caenorhabditis elegans. Nature 391, 806–811. Ford, D. M., Hepburn, H. R., Moseley, F. B., and Rigby, R. J. (1981). Displacement sensors in the honeybee pollen basket. J. Insect Physiol. 227, 339–346. Ford, E. B. (1955). Rapid evolution and the conditions which make it possible. Cold Spr. Harb. Symp. Quant. Biol. 20, 230–238. Foster, J. B. (1964). The evolution of mammals on islands. Nature 202, 234–235. Frauenstein, P. (2002). Sammelstrategien und Fitness bei einheimischen Hummelarten. Masters Thesis, University of Wu¨rzburg. Free, J. B., and Butler, C. G. (1959). ‘‘Bumblebees.’’ Collins, London. Frisch, K. v. (1955). ‘‘The Dancing Bees.’’ Methuen, London. Frisch, K. v. (1967). ‘‘The Dance Language and Orientation of Bees.’’ Harvard University Press, Cambridge. Fu¨lo¨p, A., and Menzel, R. (2000). Risk‐indifferent foraging behaviour in honeybees. Anim. Behav. 60, 657–666. Gallistel, C. R. (1990). ‘‘The Organization of Learning.’’ MIT Press, Cambridge. Gilchrist, G. W., and Kingsolver, J. G. (2001). Is optimality over the hill? The fitness landscapes of idealized organisms. In ‘‘Adaptationism and Optimality’’ (H. Orzack and E. Sober, Eds.), pp. 219–241. Cambridge University Press, Cambridge. Ginzburg, L. R., and Jensen, C. X. J. (2004). Rules of thumb for judging ecological theories. Trends Ecol. Evol. 19, 121–140. Giurfa, M., and Nu´n˜ez, J. A. (1992). Honeybees mark with scent and reject recently visited flowers. Oecologia 89, 113–117. Giurfa, M., and Nu´n˜ez, J. A. (1993). The repellent scent‐mark of the honeybee Apis mellifera ligustica and its role as communication cue during foraging. Insect. Soc. 40, 59–67. Giurfa, M., Nu´n˜ez, J., Chittka, L., and Menzel, R. (1995). Colour preferences of flower‐naive honeybees. J. Comp. Physiol. A 177, 247–259. Gould, J. L. (1975). Honey bee recruitment: The dance‐language controversy. Science 189, 685–693. Gould, S. J., and Lewontin, R. C. (1979). The spandrels of San Marco and the Panglossian paradigm: A critique of the adaptationist programme. Proc. R. Soc. Lond. B 205, 581–598.

348

NIGEL E. RAINE ET AL.

Goulson, D. (2003). ‘‘Bumblebees: Behaviour and Ecology.’’ Oxford University Press, Oxford. Goulson, D., and Darvill, B. (2004). Niche overlap and diet breadth in bumblebees; are rare species more specialized in their choice of flowers? Apidologie 35, 55–63. Goulson, D., Stout, J. C., Langley, J., and Hughes, W. O. C. (2000). Identity and function of scent marks deposited by foraging bumblebees. J. Chem. Ecol. 26, 2897–2911. Goulson, D., Peat, J., Stout, J. C., Tucker, J., Darvill, B., Derwent, L. C., and Hughes, W. O. C. (2002). Can alloethism in workers of the bumblebee, Bombus terrestris, be explained in terms of foraging efficiency? Anim. Behav. 64, 123–130. Greggers, U., and Menzel, R. (1993). Memory dynamics and foraging strategies of honeybees. Behav. Ecol. Sociobiol. 32, 17–29. Grimm, V. (1994). Mathematical‐models and understanding in ecology. Ecol. Model. 75, 641–651. Grimm, V. (1999). Ten years of individual‐based modelling in ecology: What have we learned and what could we learn in the future? Ecol. Model. 115, 129–148. Gumbert, A. (2000). Color choices by bumble bees (Bombus terrestris): Innate preferences and generalization after learning. Behav. Ecol. Sociobiol. 48, 36–43. Hagen, E. v. (1990). ‘‘Hummeln: Bestimmen, Ansiedeln, Vermehren, Schu¨tzen.’’ Natur Verlag, Augsburg. Harder, L. D. (1990). Behavioral responses by bumble bees to variation in pollen availability. Oecologia 85, 41–47. Harvey, P. H., and Purvis, A. (1991). Comparative methods for explaining adaptations. Nature 351, 619–624. Hedkte, C. (1996). Heimfinden von Apis mellifera and 4 Bombus‐Arten im Vergleich. Apidologie 27, 320–323. Heinrich, B. (1976). The foraging specializations of individual bumblebees. Ecol. Monogr. 2, 105–128. Heinrich, B. (1979). ‘‘Bumblebee Economics.’’ Harvard University Press, Cambridge. Heinrich, B., Mudge, P. R., and Deringis, P. G. (1977). Laboratory analysis of flower constancy in foraging bumblebees: Bombus ternarius and B. terricola. Behav. Ecol. Sociobiol. 2, 247–265. Herre, E. A. (1995). Tests of optimality models. Trends Ecol. Evol. 10, 121. Holehouse, K. A., Hammond, R. L., and Bourke, A. F. G. (2003). Non‐lethal sampling of DNA from bumble bees for conservation genetics. Insect. Soc. 50, 277–285. Humphries, M. A., Muller, U., Fondrk, M. K., and Page, R. E. (2003). PKA and PKC content in the honey bee central brain differs in genotypic strains with distinct foraging behavior. J. Comp. Physiol. A 189, 555–562. Ings, T. C., Raine, N. E., and Chittka, L. (2005a). Mating preference of commercially imported bumblebees (Bombus terrestris) in Britain (Hymenoptera: Apidae). Entomol. Gen. 28, 233–238. Ings, T. C., Schikora, J., and Chittka, L. (2005b). Bumblebees, humble pollinators or assiduous invaders? A population comparison of foraging performance in Bombus terrestris. Oecologia 144, 508–516. Ings, T. C., Ward, N. L., and Chittka, L. (2006). Can commercially imported bumble bees out‐compete their native conspecifics? J. Appl. Ecol. 43, 940–948. Inouye, D. W. (1978). Resource partitioning in bumblebees: Experimental studies in foraging behavior. Ecology 59, 672–678. Irwin, R. E., and Brody, A. K. (1999). Nectar‐robbing bumble bees reduce the fitness of Ipomopsis aggregata (Polemoniaceae). Ecology 80, 1703–1712. Judson, O. P. (1994). The rise of the individual‐based model in ecology. Trends Ecol. Evol. 9, 9–14.

EVOLUTION OF BEE FORAGING BEHAVIOR

349

Kacelnik, A. (1984). Central place foraging in starlings (Sturnus vulgaris) 1: Patch residence time. J. Anim. Ecol. 53, 283–299. Kacelnik, A., and Krebs, J. R. (1985). Learning to exploit patchily distributed food. In ‘‘Behavioural Ecology’’ (R. M. Silby and R. H. Smith, Eds.), pp. 189–205. Oxford University Press, Oxford. Kacelnik, A., Houston, A. I., and Schmid‐Hempel, P. (1986). Central‐place foraging in honey bees: The effect of travel time and nectar flow on crop filling. Behav. Ecol. Sociobiol. 19, 19–24. Kawecki, T. J., and Ebert, D. (2004). Conceptual issues in local adaptation. Ecol. Lett. 7, 1225–1241. Kirchner, W., and Grasser, A. (1998). The significance of odor cues and dance language information for the food search behavior of honey bees (Hymenoptera: Apidae). J. Insect Behav. 11, 169–178. Ko¨nig, C., and Schmid‐Hempel, P. (1995). Foraging activity and immunocompetence in workers of the bumble bee, Bombus terrestris L. Proc. R. Soc. Lond. B 260, 225–227. Kuwabara, M. (1957). Bildung des bedingten Reflexes von Pavlovs Typus bei der Honigbiene, Apis mellifica. J. Fac. Sci. Hokkaido Univ. Ser. VI Zool. 13, 458–464. Laverty, T. M. (1994). Costs to foraging bumble bees of switching plant species. Can. J. Zool. 72, 43–47. Lindauer, M., and Kerr, W. (1958). Die gegenseitige Versta¨ndigung bei den stachellosen Bienen. Z. Vergl. Physiol. 41, 405–434. Lipp, H. P. (2002). The tortuous path from genotype to phenotype: Genes and cognition in mutant mice. Eur. J. Hum. Genet. 10, 55. Lofdahl, K. L., Holliday, M., and Hirsch, J. (1992). Selection for conditionability in Drosophila melanogaster. J. Comp. Psychol. 106, 172–183. Losos, J. B., and Miles, D. B. (1994). Adaptation, constraint, and the comparative phylogenetic method: Phylogenetic issues and methods. In ‘‘Ecological Morphology: Integrative Organismal Biology’’ (P. C. Wainwright and S. M. Reilly, Eds.), pp. 60–98. University of Chicago Press, Chicago. Lunau, K., Wacht, S., and Chittka, L. (1996). Colour choices of naive bumble bees and their implications for colour perception. J. Comp. Physiol. A 178, 477–489. MacArthur, R. H., and Pianka, E. R. (1966). On optimal use of patchy environment. Am. Nat. 100, 603–609. Mallon, E. B., Brockmann, A., and Schmid‐Hempel, P. (2003). Immune response inhibits associative learning in insects. Proc. R. Soc. Lond. B 270, 2471–2473. Manning, A. (1956). Some aspects of the foraging behaviour of bumble‐bees. Behaviour 9, 164–201. Marie, B., Bacon, J. P., and Blagburn, J. M. (2000). Double‐stranded RNA interference shows that Engrailed controls the synaptic specificity of identified sensory neurons. Curr. Biol. 10, 289–292. Maurer, R., and Se´guinot, V. (1995). What is modelling for? A critical review of the models of path integration. J. Theor. Biol. 175, 457–475. May, R. (2004). Uses and abuses of mathematics in biology. Science 303, 790–793. Maynard Smith, J. (1978). Optimization theory in evolution. Annu. Rev. Ecol. Syst. 9, 31–56. Mayr, G. (2004). Old world fossil record of modern‐type hummingbirds. Science 304, 861–864. McNamara, J. M., Houston, A. I., and Weisser, W. W. (1993). Combining prey choice and patch use—what does rate‐maximizing predict? J. Theor. Biol. 164, 219–238.

350

NIGEL E. RAINE ET AL.

Menzel, R. (1990). Learning, memory, and ‘‘cognition’’ in honey bees. In ‘‘Neurobiology of Comparative Cognition’’ (R. P. Kesner and D. S. Olten, Eds.), pp. 237–292. Erlbaum Inc., Hillsdale. Menzel, R. (2001). Behavioral and neural mechanisms of learning and memory as determinants of flower constancy. In ‘‘Cognitive Ecology of Pollination’’ (L. Chittka and J. D. Thomson, Eds.), pp. 21–40. Cambridge University Press, Cambridge. Menzel, R., and Shmida, A. (1993). The ecology of flower colours and the natural colour vision of insect pollinators: The Israeli flora as a case study. Biol. Rev. 6, 262–293. Mery, F., and Kawecki, T. J. (2004). An operating cost of learning in Drosophila melanogaster. Anim. Behav. 68, 589–598. Mu¨ller, C. B., and Schmid‐Hempel, P. (1992). Correlates of reproductive success among field colonies of Bombus lucorum: The importance of growth and parasites. Ecol. Entomol. 17, 343–353. Neese, V. (1988). Die Entfernungsmessung der Sammelbiene: Ein energetisches und zugleich sensorisches. BIONA‐report 6, 1–15. Olesen, J. M. (1985). The Macaronesian bird‐flower elements and its relation to bird and bee opportunists. Bot. J. Linn. Soc. 91, 395–414. Orzack, S. H., and Sober, E. (1994). How (not) to test an optimality model. Trends Ecol. Evol. 9, 265–267. Orzack, S. H., and Sober, E. (2001). ‘‘Adaptationism and Optimality.’’ Cambridge University Press, Cambridge. Osborne, J. L., Clark, S. J., Morris, R. J., Williams, I. H., Riley, J. R., Smith, A. D., Reynolds, D. R., and Edwards, A. S. (1999). A landscape‐scale study of bumble bee foraging range and constancy, using harmonic radar. J. Appl. Ecol. 36, 519–533. Otterstatter, M. C., Gegear, R. J., Colla, S. R., and Thomson, J. D. (2005). Effects of parasitic mites and protozoa on the flower constancy and foraging rate of bumble bees. Behav. Ecol. Sociobiol. 58, 383–389. Page, R. E., and Fondrk, M. K. (1995). The effects of colony‐level selection on the social organization of honey bee (Apis mellifera L.) colonies: Colony‐level components of pollen hoarding. Behav. Ecol. Sociobiol. 36, 135–144. Page, R. E., and Robinson, G. E. (1991). The genetics of division of labor in honey bee colonies. In ‘‘Advances in Insect Physiology’’ (P. D. Evans, Ed.), Vol. 23, pp. 117–169. Academic Press, London. 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. Page, R. E., Erber, J., and Fondrk, M. K. (1998). The effect of genotype on response thresholds to sucrose and foraging behavior of honey bees (Apis mellifera L.). J. Comp. Phys. A 182, 489–500. Palmer, M. (2002). Testing the ‘Island Rule’ for a tenebrionid beetle (Coleoptera Tenebrionidae). Acta Oecol. 223, 103–107. Pankiw, T., and Page, R. E. (2000). Response thresholds to sucrose predict foraging division of labor in honeybees. Behav. Ecol. Sociobiol. 47, 265–267. Pankiw, T., and Page, R. E. (2001). Genotype and colony environment affect honeybee (Apis mellifera L.) development and foraging behavior. Behav. Ecol. Sociobiol. 51, 87–94. Pankiw, T., Tarpy, D. R., and Page, R. E. (2002). Genotype and rearing environment affect honeybee perception and foraging behaviour. Anim. Behav. 64, 663–672. Pelletier, L., and McNeil, J. N. (2003). The effect of food supplementation on reproductive success in bumblebee field colonies. Oikos 103, 688–694. Phelps, S. M., and Ryan, M. J. (2000). History influences signal recognition: Neural network models of tu´ngara frogs. Proc. R. Soc. Lond. B 267, 1633–1639.

EVOLUTION OF BEE FORAGING BEHAVIOR

351

Price, T. D., Qvarnstrom, A., and Irwin, D. E. (2003). The role of phenotypic plasticity in driving genetic evolution. Proc. Roy. Soc. Lond. B 270, 1433–1440. Prys‐Jones, O. E., and Corbet, S. A. (1991). ‘‘Bumblebees.’’ The Richmond Publishing Company, London. Pyke, G. H. (1978). Optimal foraging: Movement patterns of bumblebees between inflorescences. Theor. Popul. Biol. 13, 72–98. Pyke, G. H., Pulliam, H. R., and Charnov, E. L. (1977). Optimal foraging: A selective review of theory and tests. Q. Rev. Biol. 52, 137–154. Raine, N. E., and Chittka, L. (2005a). Comparison of flower constancy and foraging performance in three bumblebee species (Hymenoptera: Apidae: Bombus). Entomol. Gen. 28, 81–89. Raine, N. E., and Chittka, L. (2005b). Colour preferences in relation to the foraging performance and fitness of the bumblebee Bombus terrestris. Uludag Bee J. 5, 145–151. Raine, N. E., Ings, T. C., Ramos‐Rodrı´guez, O., and Chittka, L. (2006). Intercolony variation in learning performance of a wild British bumblebee population (Hymenoptera: Apidae: Bombus terrestris audax). Entomol. Gen. 28, 241–256. Real, L. A. (1981). Uncertainty and pollinator‐plant interactions: The foraging behavior of bees and wasps on artificial flowers. Ecology 62, 20–26. Reif, M., Linsenmair, E., and Heisenberg, M. (2002). Evolutionary significance of courtship conditioning in Drosophila melanogaster. Anim. Behav. 63, 143–155. Riechert, S. E., and Hall, R. F. (2000). Local population success in heterogeneous habitats: Reciprocal transplant experiments completed on a desert spider. J. Evol. Biol. 13, 541–550. Robertson, A. W., Mountjoy, C., Faulkner, B. E., Roberts, M. V., and Macnair, M. R. (1999). Bumble bee selection of Mimulus guttatus flowers: The effects of pollen quality and reward depletion. Ecology 80, 2594–2606. 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, 109–112. Roubik, D. W. (1989). ‘‘Ecology and Natural History of Tropical Bees.’’ Cambridge University Press, New York. Roubik, D. W. (1992). Loose niches in tropical communities: Why are there so few bees and so many trees? In ‘‘Effects of Resource Distribution on Animal‐Plant Interactions’’ (M. D. Hunter, T. Ohgushi, and P. W. Price, Eds.), pp. 327–354. Academic Press, New York. Rueppell, O., Pankiw, T., Nielsen, D. I., Fondrk, M. K., Beye, M., and Page, R. E. (2004a). The genetic architecture of the behavioral ontogeny of foraging in honeybee workers. Genetics 167, 1767–1779. Rueppell, O., Pankiw, T., and Page, R. E. (2004b). Pleiotropy, epistasis and new QTL: The genetic architecture of honey bee foraging behavior. J. Hered. 95, 481–491. Ruttner, F. (1987). The evolution of honeybees. In ‘‘Neurobiology and Behavior of Honeybees’’ (R. Menzel and A. Mercer, Eds.), pp. 8–20. Springer‐Verlag, Berlin. Ruttner, F. (1988). ‘‘Biogeography and Taxonomy of Honeybees.’’ Springer‐Verlag, Berlin. Ryan, M. J., and Rand, A. S. (1999). Phylogenetic influence on mating call preferences in female tu´ngara frogs, Physalaemus pustulosus. Anim. Behav. 57, 945–956. Saleh, N., Ohashi, K., Thomson, J. D., and Chittka, L. (2006). Facultative use of repellent scent mark in foraging bumblebees: Complex versus simple flowers. Anim. Behav. 71, 847–854. Scheiner, R., Page, R. E., and Erber, J. (2001). Responsiveness to sucrose affects tactile and olfactory learning in preforaging honey bees of two genetic strains. Behav. Brain Res. 120, 67–73. Schikora, J., and Chittka, L. (1999). Learning a complex motor skill: Bumblebees and pollen foraging. In ‘‘Go¨ttingen Neurobiology Report 1999. Proceedings of the 1st Go¨ttingen

352

NIGEL E. RAINE ET AL.

Conference of the German Neuroscience Society 1999’’ (N. Elsner and U. Eysel, Eds.), Vol. II, ppp. 256. Georg Thieme Verlag, Stuttgart. Schmid‐Hempel, P. (1987). Efficient nectar‐collecting by honey bees: I. Economic models. J. Anim. Ecol. 56, 209–218. Schmid‐Hempel, P., and Heeb, D. (1991). Worker mortality and colony development in bumblebees, Bombus lucorum (L.) (Hymenoptera, Apidae). Mitt. Schweiz. Entomol. Ges. 64, 93–108. Schmid‐Hempel, P., and Stauffer, H.‐P. (1998). Parasites and flower choice of bumblebees. Anim. Behav. 55, 819–825. Schmid‐Hempel, P., Kacelnik, A., and Houston, A. (1985). Honey bees maximise efficiency by not filling their crop. Behav. Ecol. Sociobiol. 17, 61–66. Schmid‐Hempel, R., and Schmid‐Hempel, P. (1998). Colony performance and immunocompetence of a social insect, Bombus terrestris, in poor and variable environments. Funct. Ecol. 12, 22–30. Schneider, S. S. (1989). Spatial foraging patterns of the African honey bee Apis mellifera scutellata. J. Insect Behav. 2, 505–521. Seeley, T. D. (1995). ‘‘The Wisdom of the Hive.’’ Harvard University Press, Cambridge. Shafir, S., Wiegmann, D. D., Smith, B. H., and Real, L. A. (1999). Risk‐sensitive foraging: Choice behaviour of honeybees in response to variability in volume of reward. Anim. Behav. 57, 1055–1061. Sherman, G., and Visscher, P. K. (2002). Honeybee colonies achieve fitness through dancing. Nature 419, 920–922. Sherry, D. F. (1998). The ecology and neurobiology of spatial memory. In ‘‘Cognitive Ecology: The Evolutionary Ecology of Information Processing and Decision Making’’ (R. Dukas, Ed.), pp. 261–296. University of Chicago Press, Cambridge. Sherry, D. F., and Healy, S. (1998). Neural mechanisms of spatial representation. In ‘‘Spatial Representations in Animals’’ (S. Healy, Ed.), pp. 133–158. Oxford University Press, Oxford. Shettleworth, S. J. (1998). ‘‘Cognition, Evolution, and Behavior.’’ Oxford University Press, New York. Shobe, J. (2002). The role of PKA, CaMKII, and PKC in avoidance conditioning: Permissive or instructive? Neurobiol. Learn. Mem. 77, 291–312. Shykoff, J. A., and Mu¨ller, C. B. (1995). Reproductive decisions in bumble‐bee colonies—the influence of worker mortality in Bombus terrestris (Hymenoptera, Apidae). Funct. Ecol. 9, 106–112. Spaethe, J., and Chittka, L. (2003). Interindividual variation of eye optics and single object resolution in bumblebees. J. Exp. Biol. 206, 3447–3453. Spaethe, J., and Weidenmu¨ller, A. (2002). Size variation and foraging rate in bumblebees (Bombus terrestris). Insect. Soc. 49, 142–146. Spaethe, J., Tautz, J., and Chittka, L. (2001). Visual constraints in foraging bumblebees: Flower size and color affect search time and flight behavior. Proc. Natl. Acad. Sci. USA 98, 3898–3903. Stanton, M. L., and Galen, C. (1997). Life on the edge: Adaptation versus environmentally mediated gene flow in the snow buttercup, Ranunculus adoneus. Am. Nat. 150, 143–178. Stephens, D. W., and Krebs, J. R. (1991). ‘‘Foraging Theory.’’ Princeton University Press, Princeton. Stone, G. N., Raine, N. E., Prescott, M., and Willmer, P. G. (2003). The pollination ecology of Acacias (Fabaceae, Mimosoideae). Aust. Syst. Bot. 16, 103–118.

EVOLUTION OF BEE FORAGING BEHAVIOR

353

Stout, J. C., Allen, J. A., and Goulson, D. (1998). The influence of relative plant density and floral morphological complexity on the behaviour of bumblebees. Oecologia 117, 543–550. Thomson, J. D. (1981). Field measures of flower constancy in bumblebees. Am. Midl. Nat. 105, 377–380. Thomson, J. D. (1996). Trapline foraging by bumblebees: I. Persistence of flight‐path geometry. Behav. Ecol. 7, 158–164. Thomson, J. D., and Chittka, L. (2001). Pollinator individuality: When does it matter? In ‘‘Cognitive Ecology of Pollination’’ (L. Chittka and J. D. Thomson, Eds.), pp. 191–213. Cambridge University Press, Cambridge. Thomson, J. D., Maddison, W. P., and Plowright, R. C. (1982). Behavior of bumble bee pollinators on Aralia hispida Vent. (Araliaceae). Oecologia 54, 326–336. Thomson, J. D., Peterson, S., and Harder, L. D. (1987). Response of traplining bumble bees to competition experiments: Shifts in feeding location and efficiency. Oecologia 71, 295–300. Thomson, J. D., Slatkin, M., and Thomson, B. A. (1997). Trapline foraging by bumble bees: II. Definition and detection from sequence data. Behav. Ecol. 8, 199–210. Towne, W. F., and Gould, J. L. (1988). The spatial precision of the honey bees’ dance communication. J. Insect Behav. 1, 129–155. Tully, T. (1991). Genetic dissection of learning and memory in Drosophila melanogaster. In ‘‘Neurobiology of Learning, Emotion and Affect’’ (J. Madden, Ed.), pp. 29–66. Raven Press Ltd., New York. Tully, T. (1996). Discovery of genes involved with learning and memory: An experimental synthesis of Hirschian and Benzerian perspectives. Proc. Natl. Acad. Sci. USA 93, 13460–13467. Valido, A., Dupont, Y. L., and Hansen, D. M. (2002). Native birds and insects, and introduced honey bees visiting Echium wildpretii (Boraginaceae) in the Canary Islands. Acta Oecol. 23, 413–419. Varju, D., and Nu´n˜ez, J. A. (1991). What do foraging honeybees optimize? J. Comp. Physiol. A 169, 729–736. Varju, D., and Nu´n˜ez, J. (1993). Energy balance versus information exchange in foraging honeybees. J. Comp. Physiol. A 172, 257–261. Visscher, P. K., and Seeley, T. D. (1982). Foraging strategy of honey bee colonies in a temperate deciduous forest. Ecology 63, 1790–1801. Vogel, S., Westerkamp, C., Thiel, B., and Gessner, K. (1984). Ornithophilie auf den Canarischen Inseln. Plant Syst. Evol. 146, 225–248. Waddington, K. D. (2001). Subjective evaluation and choice behavior by nectar and pollen‐ collecting bees. In ‘‘Cognitive Ecology of Pollination’’ (L. Chittka and J. D. Thomson, Eds.), pp. 41–60. Cambridge University Press, Cambridge. Waddington, K. D., and Holden, L. R. (1979). Optimal foraging: On flower selection by bees. Am. Nat. 114, 179–196. Waddington, K. D., Visscher, P. K., Herbert, T. J., and Richter, M. R. (1994). Comparisons of forager distributions from matched honey bee colonies in suburban environments. Behav. Ecol. Sociobiol. 35, 423–429. Walther‐Hellwig, K., and Frankl, R. (2000). Foraging distances of Bombus muscorum, Bombus lapidarius, and Bombus terrestris (Hymenoptera, Apidae). J. Insect Behav. 13, 239–246. Waser, N. M. (1986). Flower constancy: Definition, cause, and measurement. Am. Nat. 127, 593–603. Waser, N. M., Chittka, L., Price, M. V., Williams, N., and Ollerton, J. (1996). Generalization in pollination systems, and why it matters. Ecology 77, 1043–1060.

354

NIGEL E. RAINE ET AL.

Weidenmu¨ller, A., and Seeley, T. D. (1999). Imprecision in waggle dances of the honey bee (Apis mellifera) for nearby food sources: Error or adaptation? Behav. Ecol. Sociobiol. 46, 190–199. Westrich, P. (1989).‘‘Die Wildbienen Baden‐Wu¨rttembergs.’’ Vols. I and II, Ulmer Verlag, Stuttgart. Whitfield, C. W., Cziko, A., and Robinson, G. E. (2003). Gene expression profiles in the brain predict behavior in honey bees. Science 302, 296–299. Whitlock, M. C. (1997). Founder effects and peak shifts without genetic drift: Adaptive peak shifts occur easily when environments fluctuate slightly. Evolution 51, 1044–1048. Widmer, A., Schmid‐Hempel, P., Estoup, A., and Scholl, A. (1998). Population genetic structure and colonization history of Bombus terrestris s.I. (Hymenoptera: Apidae) from the Canary Islands and Madeira. Heredity 81, 563–572. Williams, N. M., and Thomson, J. D. (1998). Trapline foraging by bumble bees: III. Temporal patterns of visitation and foraging success at single plants. Behav. Ecol. 9, 612–621. Williams, P. H. (1994). Phylogenetic relationships among bumble bees (Bombus Latr.): A reappraisal of morphological evidence. Syst. Entomol. 19, 327–344. Willmer, P. G., and Stone, G. N. (2004). Behavioral, ecological, and physiological determinants of the activity patterns of bees. Adv. Stud. Behav. 34, 347–466. Wolfer, D. P., and Lipp, H. P. (2000). Dissecting the behaviour of transgenic mice: Is it the mutation, the genetic background, or the environment? Exp. Physiol. 85, 627–634. Woodward, G. L., and Laverty, T. M. (1992). Recall of flower handling skills by bumble bees: A test of Darwin’s interference hypothesis. Anim. Behav. 44, 1045–1051. Ydenberg, R., and Hurd, P. (1998). Simple models of feeding with time and energy constraints. Behav. Ecol. 9, 49–53. Ydenberg, R., and Schmid‐Hempel, P. (1994). Modelling social insect foraging. Trends Ecol. Evol. 9, 491–493.

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 36

Kin Selection, Constraints, and the Evolution of Cooperative Breeding in Long‐Tailed Tits Ben J. Hatchwell and Stuart P. Sharp department of animal and plant sciences university of sheffield, sheffield s10 2tn united kingdom

I. INTRODUCTION Charles Darwin (1859) struggled with the evolutionary puzzle of apparent altruism among social insects, famously declaring that they posed ‘‘by far the most serious special difficulty, which my theory has encountered.’’ The evolutionary difficulty of cooperation among supposedly selfish individuals remained unresolved, and to a large extent unrecognized (Cronin, 1991), for over a century. The great advance in our understanding of the evolution of social organisms which has occurred in the past four decades originated with Hamilton’s introduction (1963, 1964a,b) of the concept of inclusive fitness and the idea of kin selection (Maynard Smith, 1964). Social insects and the cooperative breeding systems of vertebrates have since proved to be fertile testing grounds for evolutionary theories stemming from Hamilton’s insight. Among social vertebrates the observation, common to many long‐term studies, that cooperative behavior routinely occurs among relatives (Stacey and Koenig, 1990) led to the obvious conclusion that kin selection played a key role in the evolution of such systems. This conclusion dovetailed neatly with explanations for the ecological conditions in which such cooperative behavior was expected to evolve. Selander (1964) first proposed that habitat saturation might cause grown offspring to delay dispersal and remain on their natal territory as helpers. The development of subsequent ideas on the factors responsible for the evolution of cooperative breeding (Brown, 1974; Koenig and Pitelka, 1981) has crystallized into three alternative, although not exclusive, hypotheses: the ecological constraints hypothesis (Emlen, 1982), the benefits of philopatry hypothesis (Stacey and Ligon, 0065-3454/06 $35.00 DOI: 10.1016/S0065-3454(06)36008-1

355

Copyright 2006, Elsevier Inc. All rights reserved.

356

BEN J. HATCHWELL AND STUART P. SHARP

1987, 1991), and the life‐history hypothesis (Arnold and Owens, 1998). The predictions, similarities, and differences of these alternative visions have been discussed at length in the literature (Cockburn, 1996, 1998; Dickinson and Hatchwell, 2004; Emlen, 1991, 1994, 1997; Hatchwell and Komdeur 2000; Koenig et al., 1992) and, so far, it is fair to conclude that no general consensus has emerged (Koenig and Dickinson, 2004). In particular, there is still no clear distinction that can be drawn between the ecology and life history of cooperative species compared to noncooperative species. Nevertheless, the various hypotheses do share the basic similarity that they result in kin‐structured populations, hence the easy assumption that kin selection has played a crucial role in the evolution of helping among individuals within such social groups. Despite the substantial number of correlational studies implicating kin selection as a major force in the evolution of cooperative breeding, recent reviews have concluded that the evidence for kin selection providing a general explanation for cooperative societies among vertebrates is less compelling now than it was 25 years ago (Clutton‐Brock, 2002; Cockburn, 1998). The criticisms of evidence focus on five main points. First, there is no evidence that the genetic structure and, hence, degree of relatedness of cooperatively breeding species is any higher than in noncooperative species, so kin‐directed helping may simply be a consequence of philopatry rather than the selective basis for cooperation (Clutton‐Brock, 2002). Second, intraspecific variation in the effort invested in a brood by helpers is often unrelated to helper kinship (Clutton‐Brock et al., 2001; Dunn et al., 1995; but see Curry, 1988; Emlen and Wrege, 1988; Griffin and West, 2003; Komdeur, 1994); moreover, several studies have shown that helpers are often unrelated to helped broods (Clutton‐Brock et al., 2000; Magrath and Whittingham, 1997). Third, the kin‐selected fitness benefits may have been routinely overestimated by incorrect calculation of indirect fitness (Creel, 1990) and by failure to recognize the costs of competing with relatives for resources (West et al., 2002). Fourth, a positive correlation between productivity and group size or the number of helpers may be confounded by the effects of territory or individual quality (Brown et al., 1982; Cockburn, 1998). Finally, the direct fitness benefits of helping may have been underestimated. For example, the application of molecular genetic techniques has revealed that in some species ‘‘helpers’’ are in fact cobreeders (Richardson et al., 2001, 2002); helpers may also benefit directly through group augmentation (Kokko et al., 2001). In this chapter, we describe some aspects of our research on the cooperative breeding system of the long‐tailed tit Aegithalos caudatus (Fig. 1). The principal aim of the chapter is to bring together evidence from various sources concerning the role of kin selection and ecological constraints in

KIN SELECTION AND COOPERATION IN LONG‐TAILED TITS

357

Fig. 1. Adult long‐tailed tit at its nest.

the evolution of cooperative breeding in this species. The study was started because long‐tailed tits have many attributes that make them ideally suited to studying the process of kin selection. In particular, they have an atypical cooperative system in which groups do not defend exclusive territories and individuals switch back and forth between breeding and helping throughout their lives. Moreover, those lives are generally short, with few individuals living beyond their third birthday, so we have been able to accumulate data on lifetime reproductive success relatively quickly when compared to studies of most other cooperative species that are characterized by slow life histories (Arnold and Owens, 1998). Finally, long‐tailed tits have the endearing trait (for evolutionary biologists) of being extremely accommodating when challenged with experimental manipulations. Here, we draw on observational and experimental results amassed over 12 years

358

BEN J. HATCHWELL AND STUART P. SHARP

with the aim of answering four questions that were posed at the start of the study: (1) Do helpers discriminate kin from non‐kin? (2) What is the mechanism of kin recognition? (3) What are the fitness consequences of cooperation? and (4) What is the ecological basis for cooperation in this species? Our chapter is focused on long‐tailed tits but, as indicated by our review of the debates over the roles of ecological constraints and kin selection in the evolution of animal societies, we suggest that many of the findings are of broader relevance, and we have sought to draw more general conclusions where appropriate.

II. STUDY SPECIES, STUDY SITES, A. STUDY SPECIES

AND

AND

GENERAL METHODS

STUDY SITES

The long‐tailed tit is one of seven species comprising the Aegithalidae. The five congeners of the long‐tailed tit are found only in the Himalayas and China and are poorly known; in contrast, the long‐tailed tit has an extensive range, occurring right across the Palearctic from Iberia to Japan. The seventh member of the family, the bushtit Psaltriparus minimus, is found in western North America from Guatemala north to British Columbia (Harrap and Quinn, 1996). The pygmy tit Psaltria exilis, from Java, is also included in the Aegithalidae by some authors, but its affinities remain unclear (Perrins, 2004). The bushtit is known to be a cooperative breeder (Sloane, 1996), differing in some respects from the long‐tailed tit, but the social organization and breeding system of the remaining members of the family are unknown. We have studied two populations of long‐tailed tits in South Yorkshire, United Kingdom. One population in the Rivelin Valley, Sheffield (53
230 N, 1
340 W) comprised 18–69 pairs and was studied continuously from 1994 to 2005; the second population in Melton Wood, Doncaster (53
310 N, 1
130 W) comprised 32–90 pairs and was studied from 1996 to 1998 and 2001 to 2003. Both sites contain woodland (predominantly deciduous), scrub, and small areas of garden and farmland, and they are situated 27 km apart. Long‐tailed tits in our study populations spend the nonbreeding season (June–February) in fluid winter flocks containing an average of around 16 individuals, including overlapping generations of kin from one or more families and also unrelated male and female immigrants. Each flock occupies a large nonexclusive range (mean area of ranges in Rivelin Valley ¼ 59 ha). The typical composition of a flock toward the end of winter prior to their breakup is 30–40% previous breeders, 25–40% philopatric recruits, and 30–40% immigrants (Hatchwell et al., 2001a; Russell, 2001).

KIN SELECTION AND COOPERATION IN LONG‐TAILED TITS

359

Monogamous pairs form in early spring with each pair occupying a nonexclusive breeding range within the range occupied by their winter flock. All birds start the season by attempting to breed independently, and there are no helpers associated with nests at this stage. Long‐tailed tits are single brooded, raising a maximum of one brood per year, but they often have several breeding attempts because of the failure of early nests. The elaborate domed nest is built by both sexes and comprises an outer structure of moss, plant fibers, spider silk, and lichen flakes, lined with up to 2600 feathers (Gaston, 1973; Hansell, 1993, 2000; Lack and Lack, 1958; McGowan et al., 2004). The nest may take several weeks to complete, although building time decreases through the breeding season: mean time to build first nests ¼ 38 days, later nests ¼ 11 days (McGowan et al., 2004). Clutch size ranges from 7 to 12 eggs (typically 9–10) with incubation starting on the day the final egg is laid. Females incubate alone for about 15 days, during which time they forage for themselves, although they are also fed by their partner while on the nest. Nestlings hatch synchronously and are fed in the nest for 16–17 days prior to fledging and then for several weeks before reaching independence. Nest failure, caused mainly by predators, is frequent at all stages of the breeding cycle (Hatchwell et al., 1999), and failed breeders may either renest or abandon breeding; some of those failed breeders that do not attempt to breed again independently may become helpers at the nest of another pair, assisting that pair in provisioning nestlings and subsequently fledglings (Glen and Perrins, 1988; Hatchwell et al., 2004). B. GENERAL METHODS Systematic field observations have been conducted in a standard way in each year at both study sites. Adults were captured using mist nets and banded with unique color‐ring combinations (mean proportion of adults color‐ringed: Rivelin Valley 94%, n ¼ 12 years; Melton Wood 90%, n ¼ 6 years). Most adults were caught during the building phase of their first breeding attempt, but a few were caught at other stages of the breeding season. Nestlings were also uniquely color‐ringed, weighed (0.1 g), and their right tarsus was measured (0.1 mm) on day 9–13 (mostly day 11) of the nestling period (hatch day ¼ day 0). The breeding attempts of all members of our study populations were closely monitored throughout the breeding season in each year of the study. Nests were found by following nest‐building pairs and were then checked at regular intervals until fledging or the breeding attempt failed. The date of egg laying and hatching was determined by feeling inside nests for the presence of eggs or hatchlings; the clutch size in accessible nests was determined by removing eggs from nests with a plastic teaspoon, counting them, and then replacing them.

360

BEN J. HATCHWELL AND STUART P. SHARP

A small proportion of nests was inaccessible, in which case the onset of incubation and date of hatching were determined by observation of breeder behavior; we assumed that egg laying started 10 days before the start of incubation because the modal clutch size is 10 eggs in both populations. When pairs failed in a breeding attempt, the study area was searched thoroughly for repeat attempts. Once broods hatched (day 0), most nests were observed either every 2 days (Rivelin Valley) or every 4 days (Melton Wood) from day 2 up to fledging or nest failure in order to record provisioning rates and the identities of all carers. Observation periods typically lasted 1 hr (for further details of provisioning observations, see MacColl and Hatchwell, 2003a).

III. KIN DISCRIMINATION

BY

HELPERS

Many studies of vertebrate cooperative breeding systems have shown that helpers assist relatives in raising their offspring (Brown, 1987; Dickinson and Hatchwell, 2004; Stacey and Koenig, 1990). The typical pattern in such species is that grown offspring remain on their natal territory or in their natal group and assist one or both of their own parents in raising subsequent broods, which are therefore either full or half‐siblings of the helper. At some stage helpers will then either die, disperse to breed, float or help, or assume breeding status in their natal group. The chronological sequence of delayed dispersal followed by helping means that helpers are almost inevitably related to the helped brood. A major reason for initiating a study on long‐ tailed tits was that their helping behavior is atypical and does not conform to this sequence of events because all helpers are failed breeders that ‘‘redirect’’ their care to become helpers (Emlen, 1982; Hatchwell et al., 2004). This feature of their cooperative system makes them ideal for testing the hypothesis that helpers exhibit a kin preference in their cooperative behavior. There are two additional features of long‐tailed tit ecology that predispose them to such an investigation. (1) They do not live in extended family groups on stable exclusive territories; instead, they occupy nonexclusive ranges that may be shared with both kin and non‐kin. (2) An individual’s status as a potential helper or as a recipient of help is determined principally by whether or not its own independent breeding attempt is destroyed by a predator. This means that status is the product of a largely stochastic process; moreover, helping and breeding status are not determined by age because all individuals that survive to their first breeding season breed independently in the first instance, becoming potential helpers only when they fail in their own breeding attempt, and one bird may switch back and forth between helping and breeding throughout its life.

KIN SELECTION AND COOPERATION IN LONG‐TAILED TITS

A. OBSERVATIONAL STUDY

OF

361

KIN DISCRIMINATION

Using pedigrees of banded long‐tailed tits, Glen and Perrins (1988) found that most helpers of known sex were male and those of known relatedness always helped to feed a brood that belonged to their brother. In a larger sample, using pedigrees and genotyping data from eight polymorphic microsatellite loci (Hatchwell et al., 2002), a similar pattern was found in our study populations. The majority of helpers were male (85%, n ¼ 90), and out of 52 helpers whose relatedness could be determined with some certainty, 41 (79%) were closely related to either the male breeder (25, 48%), or the female breeder (11, 21%), or both (5, 10%). The most frequent relationship was that the helper was a brother of the breeder, but they may also be a parent, offspring, nephew, or uncle (Russell and Hatchwell, 2001). We are using a larger number of microsatellite loci to genotype a much bigger sample of helpers, which will give greater resolution of relatedness and allow us to give a more definitive description of helper–breeder kinship. Nevertheless, the data available indicate that helpers usually assist close kin (Glen and Perrins, 1988; Russell and Hatchwell, 2001). However, just as in many previous studies of other cooperative bird species, long‐tailed tit populations are spatially structured with respect to kinship. Limited dispersal of the main helping sex (males) means that if a failed breeder chose to help at the nearest available nest containing chicks, there is a reasonable probability that the nest would belong to a relative. Two bits of evidence suggest that failed breeders are more discriminating than this when deciding whether and where to help. First, 50% of male failed breeders did not help even when active nests were available in the vicinity of their last failed breeding attempt. Second, in a typical year it was estimated that approximately one‐third of nests within 300 m, and just 15% of nests within 600 m of a failed bird’s last nest, belonged to close kin. The mean distance traveled between a failed nest and a helped nest is 290 m (Hatchwell et al., 2004), so a higher proportion of helpers care for the broods of close kin than would be expected if they were randomly selecting among nearby nests (Russell and Hatchwell, 2001). Therefore, these data suggest that helpers exhibit a kin preference when making helping decisions. However, to conduct a more rigorous test of the hypothesis that long‐tailed tit helpers discriminate between kin and non‐kin, a controlled experiment was required.

B. EXPERIMENTAL TEST

OF

KIN DISCRIMINATION

Emlen and Wrege (1988) provided the first compelling evidence for kin discrimination by helpers using the redirected caring behavior of white‐ fronted bee‐eaters Merops bullockoides. In that species, potential helpers

362

BEN J. HATCHWELL AND STUART P. SHARP

often have broods of varying relatedness available to care for. Emlen and Wrege (1988) showed in a ‘‘natural experiment’’ that these potential helpers were more likely to help when closely related broods were available. In long‐ tailed tits, a technique pioneered by Glen (1985) allowed the manipulation of breeding success so that the identity of potential helpers and potential recipients of help could be engineered to test experimentally the hypothesis that helpers preferred to assist broods of relatives. By enveloping nests and their supporting branches with 6‐cm diameter wire mesh, the avian predators (corvids) that are responsible for 85% of nest depredation (Hatchwell et al., 1999) can be effectively excluded, although protected nests are still vulnerable to predation by mustelids. The wire mesh causes no disturbance to the adults at a nest (Glen, 1985; Russell, 1999). Using this technique, some birds were chosen as potential recipients of care by protecting their nests, while the nests of other birds were selected as potential helpers (hereafter termed ‘‘focal birds’’) and their nests were left unprotected. Unprotected nests were very likely to be depredated, but in a few instances where predators did not oblige, broods were removed from a pair and fostered with other pairs (under English Nature license), thereby inducing ‘‘failure’’ of focal birds. Focal birds were given two kinds of choice. Nine potential helpers had the option of helping at nests belonging to nonrelatives only; these potential helpers had no close relative within the same social group (i.e., winter flock) with an active nest. A further 17 potential helpers had the opportunity to help at the active nest of either a close relative or a nonrelative within the same social group. The results of the experiment were clear‐cut: 17/17 (100%) birds with a nest belonging to close kin in the same group became helpers, while 0/9 (0%) birds without close kin in their group helped (binomial test: p < 0.005); no birds left their social group to become helpers at nests of related or unrelated pairs in other groups. The second choice concerned those 17 birds with close kin with active nests in the same group. These potential helpers had a choice between helping at the nest of close kin or a nest of non‐kin within the same social group (the two nests being equidistant from the potential helper’s failed nest). In this case 16/17 (94%) helpers chose to help their close kin rather than their distant or non‐kin (binomial test: p < 0.001). Thus, this experiment generated two key results: first, failed breeders became helpers only when the brood of a relative was available to be helped in the same social group. Second, helpers were capable of active discrimination between kin and non‐kin when deciding at which nest to help (Russell and Hatchwell, 2001). This was the first controlled experiment to demonstrate a kin preference in helping behavior, and a subsequent meta‐analysis of observational studies suggested that such discriminatory helping may be more widespread than is sometimes recognized (Griffin and West, 2003).

KIN SELECTION AND COOPERATION IN LONG‐TAILED TITS

363

However, even in species in which helpers show kin discrimination in deciding whether and where to help, as in white‐fronted bee‐eaters and long‐ tailed tits, it is not necessarily the case that helpers also show fine discrimination in the effort that they invest in broods of varying degrees of relatedness. The absence of such fine discrimination has sometimes been proposed as evidence against kin selection (e.g., Clutton‐Brock, 2002), but this overlooks the possibility that the cues used in discrimination of kin from non‐kin may not always permit individuals to discriminate among potential recipients at a fine level. This is likely to be the case for long‐tailed tits, as shown by our investigations of the mechanism of kin recognition.

IV. KIN RECOGNITION MECHANISM The ability to discriminate between kin and non‐kin plays a major role in the evolution of social behavior. Although the recognition of relatives per se is not necessarily a prerequisite for kin selection, we would expect that for individuals to maximize their inclusive fitness, selection will favor a mechanism for discrimination among conspecifics according to kinship. Consequently, kin recognition has been the focus of a great deal of empirical and theoretical research, and there is evidence for its occurrence in a wide range of animal taxa (Fletcher and Michener, 1987; Hepper, 1991). Avian recognition systems have been particularly well studied, and vocalizations are the most commonly used cues for discrimination in birds (Halpin, 1991). However, previous research has focused on recognition between parents and offspring, neighbors and strangers, or flock members; surprisingly few studies have investigated kin recognition in complex social systems. Vocal discrimination has been demonstrated in a few cooperatively breeding birds (Payne et al., 1988; Price, 1998), but the precise mechanism for kin recognition in these species is poorly understood (Komdeur and Hatchwell, 1999). In long‐tailed tits, kin‐biased helping was shown to occur in the absence of reliable spatial cues to kinship (Russell and Hatchwell, 2001), suggesting that recognition based on other cues must operate in this species. Given the prevalence of acoustic cues in avian systems, it seemed likely that studies of vocal communication in long‐tailed tits would offer important insights into how kin discrimination is achieved. A. INDIVIDUALITY

IN

VOCALIZATIONS

Long‐tailed tits have a limited vocal repertoire consisting of five major calls and a very rarely used song (Cramp and Perrins, 1993). Two call types in particular, the ‘‘churr’’ and the ‘‘triple,’’ act as contact calls and therefore seemed likely to play some role in recognition. The churr is given

364

BEN J. HATCHWELL AND STUART P. SHARP

frequently by both sexes and is important for short‐range communication, for example, in the coordination of breeding activities or during aggressive interactions; the triple is also given by both sexes but is predominantly used to establish or maintain long‐range contact (Cramp and Perrins, 1993; Sharp, 2003). In order to investigate the potential role of these calls as recognition cues, it was important to first examine the variability in their acoustic structure and in particular assess the relative variation in calls of the same individual and those of different individuals. We obtained multiple recordings of the churr calls and triple calls of color‐ ringed birds in Melton Wood during the breeding seasons of 2000–2003 and produced sonograms of these calls in order to compare their acoustic structure. Spectrographic cross‐correlation (SPCC) revealed that for both call types, the variation between individuals was significantly greater than that within individuals (Sharp and Hatchwell, 2005). In other words, these vocalizations are individually distinctive (Fig. 2) and thus have the potential to act as reliable cues to identity. However, kin recognition may be achieved using family‐specific rather than individual‐specific signals, and some degree of individuality in potential cues may be expected in either case (Halpin, 1991; Waldman, 1987). It was therefore important to compare the calls of individuals of known relatedness. B. FAMILY SPECIFICITY

IN

VOCALIZATIONS

The precise nature of the recognition mechanism has important implications for the cooperative breeding system. The use of individual‐specific cues may permit fine adjustments in behavior according to the perceived degree of relatedness; family‐specific cues may lead to simple decisions regarding behavior according to whether or not conspecifics are considered as relatives (Komdeur and Hatchwell, 1999). By restricting our library of recordings to those involving individuals of known relatedness, we compared the churr and triple calls of siblings and non‐siblings; insufficient recordings were available of birds related in other ways. However, as most helpers provide care at the nest of a sibling (Russell and Hatchwell, 2001), these comparisons were of primary biological interest. Using SPCC, we found that for both call types, the calls of siblings were significantly more similar than those of non‐siblings (Sharp and Hatchwell, in press). Thus both of these vocalizations have the potential to act as individual‐specific or family‐specific kin recognition cues. Clearly, however, these results alone were not enough to demonstrate that vocally mediated discrimination operates in this system; playback experiments were required in order to determine if individuals use these calls as cues for kin recognition.

KIN SELECTION AND COOPERATION IN LONG‐TAILED TITS

365

Fig. 2. Sonograms of two calls from each of two individual long‐tailed tits, (i) and (ii), showing (A) the churr and (B) the triple vocalizations (adapted from Sharp and Hatchwell, 2005; reproduced with permission from Brill Academic Publishers).

C. EXPERIMENTAL TEST

OF

VOCAL KIN RECOGNITION

Kin recognition cues, whether individual specific or family specific, must exhibit greater interindividual than intraindividual variation, and the individuals themselves must be able to detect and respond to this variation (Falls, 1982; Halpin, 1991). Sufficient recordings of the churr call were available to measure and compare a suite of call characteristics in color‐ringed individuals; multivariate analysis revealed that two parameters in particular, the maximum and minimum frequency of calls, were highly individual specific (Sharp and Hatchwell, 2005). Sharp et al. (2005) carried out a playback experiment to determine whether individuals could discriminate between kin and non‐kin, based on variation in these parameters alone. Playback trials with four treatments were conducted at the nests of focal birds using the following stimuli: (1) the churr calls of a close relative

366

BEN J. HATCHWELL AND STUART P. SHARP

(coefficient of relatedness, r ¼ 0.5); (2) the churr calls from treatment (1) but with maximum and minimum frequency manipulated; (3) the churr calls of a nonrelative (r < 0.125); and (4) the churr calls from treatment (3) but with maximum and minimum frequency manipulated. Using the software package Avisoft (Raimund Specht, Berlin, Germany), calls were manipulated by shifting them along the frequency axis, thus altering the maximum and minimum frequency while leaving other call characteristics unchanged; manipulated calls remained within the observed range of natural variation. The difference in each of four behavioral responses during periods of playback and periods of quiet was calculated to give four ‘‘net’’ responses for each treatment. For three of these, the net response during playback of the unmanipulated churr calls of a close relative was significantly different from that during the other three treatments, between which there were no significant differences (Fig. 3). Long‐tailed tits therefore responded differently to the churr calls of kin unless maximum and minimum frequency were manipulated, in which case birds responded in the same way as to the calls of non‐kin, manipulated or otherwise. Thus individuals were able to discriminate between the vocalizations of kin and non‐kin based at least in part on variation in individual‐specific call characteristics (Sharp et al., 2005). Although these results suggest that the churr indeed acts as a cue for kin recognition, it does not offer any insights into whether discrimination operates on an individual‐specific or family‐specific recognition basis. These two processes are not necessarily mutually exclusive and are difficult to study separately (Halpin, 1991; Waldman, 1987). However, the use of individual‐specific or family‐specific cues will depend in part on the manner in which they are acquired (Halpin, 1991; Sherman et al., 1997). Furthermore, the nature of cue development has important implications for the reliability of the recognition system. D. EXPERIMENTAL INVESTIGATION

OF

CALL DEVELOPMENT

Discrimination based on genetically determined cues may lead to recognition errors due to the effects of recombination, whereas cues derived from the environment are only reliable if acquired at a time when there is good evidence of kinship (Halpin, 1991; Sherman et al., 1997; Waldman, 1987). By recording the vocalizations of nestling long‐tailed tits from hatching through to fledging, we were able to show that both the churr and the triple develop during the course of the nestling period (Sharp and Hatchwell, in press), but this reveals little about the relative contribution of genetic and environmental factors in their development. Avian calls were traditionally thought to be genetically determined, but more recently

KIN SELECTION AND COOPERATION IN LONG‐TAILED TITS

367

Fig. 3. Responses to playback trials (n ¼ 8) using the churr calls of kin and non‐kin with maximum and minimum frequency unmanipulated (Unmanip.) and manipulated (Manip.). Net responses (error bars: mean  S.E.) were calculated as the difference in response during playback and quiet periods. (A) Closest approach to the speakers (Friedman test, s ¼ 0.63, p ¼ 0.889). (B) Time spent within 10 m of the speakers (s ¼ 14.85, p ¼ 0.002). (C) Churr rate (s ¼ 12.15, p ¼ 0.007). (D) Triple rate (s ¼ 15.93, p ¼ 0.001). Tests remained significant after sequential Bonferroni correction. Asterisks indicate significant differences after treatment comparison tests (sensu Siegel and Castellan, 1988) (adapted from Sharp et al., 2005).

learning has been shown to play an important role in call development (Baptista, 1996; Hughes et al., 1998; Price, 1998). In order to investigate the pattern of development in the churr, Sharp et al. (2005) carried out a cross‐fostering experiment. Nestlings from partial broods were marked and swapped between synchronous nests of unrelated birds (r < 0.125). The churr calls of recruits from cross‐fostered broods were then recorded in the following year and compared using SPCC. The calls of foster siblings were found to be just as similar as those of true

368

BEN J. HATCHWELL AND STUART P. SHARP

siblings reared together, whereas those of true siblings reared apart were significantly less similar (Fig. 4). Furthermore, the correlation coefficients for foster siblings and true siblings reared apart were comparable with those previously obtained for siblings and non‐siblings respectively. Recordings were also available of the churr calls of the foster and biological parents of cross‐fostered birds; further analysis revealed that the calls of individuals were significantly more similar to those of their foster parents than to those of their biological parents, whether comparisons were made with female (Fig. 5A) or male (Fig. 5B) parents. Together, these results demonstrate a significant learned component in the development of the churr. Reliance on cues that develop through learning may result in recognition errors (i.e., the acceptance of non‐kin as kin) if interactions with non‐kin occur during the development period, or if social relationships are not reliable predictors of kinship (Sherman et al., 1997). However, extra‐pair paternity and brood parasitism are rare in long‐tailed tits (Hatchwell et al., 2002; see Section V.A.1) and offspring associate with their relatives over a relatively long period, as in most cooperatively breeding birds (Langen, 2000); the risk of making recognition errors is therefore reduced. Thus, the churr seems to make an effective kin recognition cue given the social system of these birds; we would therefore expect its use as a family‐specific signal to be reflected in the pattern of helping behavior shown by this species.

Fig. 4. Call similarity between different groups of siblings (sibs). Correlation coefficients (dashed horizontal bars indicate means) for pairwise comparisons of churr calls in each group were obtained using SPCC (Kruskal–Wallis test, w2 ¼ 9.752, p ¼ 0.008). Asterisk indicates significant differences after treatment comparison tests (adapted from Sharp et al., 2005).

KIN SELECTION AND COOPERATION IN LONG‐TAILED TITS

369

Fig. 5. Call similarity between recruits from cross‐fostered broods and their true and foster parents. Correlation coefficients (dashed horizontal bars indicate means) for pairwise comparisons of churr calls were obtained using SPCC. (A) Comparisons between the calls of recruits and their true and foster mothers (Wilcoxon signed rank test, z ¼ 2.201, p ¼ 0.028). (B) Comparisons between the calls of recruits and their true and foster fathers (Wilcoxon signed rank test, z ¼ 2.201, p ¼ 0.028) (adapted from Sharp et al., 2005).

E. THE PATTERN

OF

HELPING BEHAVIOR

The use of learned recognition cues does not necessarily mean that the act of discrimination itself is the result of individuals learning the cues of conspecifics; the process may instead entail the matching of perceived cues to

370

BEN J. HATCHWELL AND STUART P. SHARP

some genetically encoded template (Halpin, 1991; Sherman et al., 1997; Waldman, 1987). However, if long‐tailed tits learn vocal cues in the nest from the individuals that raise them, then we might expect recognition to be achieved through association, with individuals learning the cues of conspecifics at a time when they interact almost exclusively with close relatives. Sharp et al. (2005) provided evidence for this by investigating whether those helpers whose entire life history was documented (n ¼ 64) became helpers at the nests of individuals with whom they had been directly associated during the nestling phase, either as siblings or as a recipient or donor of care; in 89% of cases, this was indeed the case. In 5% of cases helpers assisted at nests belonging to a sibling of either a parent or a helper who fed it as a nestling. These instances suggest that recognition might occasionally be achieved indirectly through shared call characteristics, but the possibility that some direct association went undetected cannot be excluded. Further evidence that kin recognition is achieved by learning through association came from another cross‐fostering experiment. Using the same procedures described earlier, Hatchwell et al. (2001b) showed that among a small sample of survivors from experimentally cross‐fostered broods, failed breeders did not discriminate between true and foster siblings when making helping decisions; all failed breeders who were not the sole survivors from their nest became helpers, and brood‐mates were treated as siblings regardless of their true relatedness. Therefore, the results of these observations and experimental manipulations show that long‐tailed tits use a simple rule to identify kin and the pattern of helping is consistent with kin recognition, using cues learned during a period of close association. V. FITNESS CONSEQUENCES

OF

COOPERATION

Having shown that helpers exhibit a kin preference in their cooperative behavior and having identified their kin recognition mechanism, the aim of this section is to assess the contributions of the indirect component of fitness toward inclusive fitness in long‐tailed tits. Individuals can obviously gain direct fitness via successful personal reproduction, but what are the fitness benefits of cooperation by helpers? We first consider the various potential sources of direct fitness from helping for long‐tailed tits and then consider the indirect or kin‐selected benefits of cooperation. A. DIRECT FITNESS BENEFITS The direct fitness benefits of helping are those that enhance the personal reproductive success of helpers. There are many routes through which this might be achieved, which are well reviewed in the literature (Brown, 1987;

KIN SELECTION AND COOPERATION IN LONG‐TAILED TITS

371

Dickinson and Hatchwell, 2004; Emlen, 1991). In brief, hypothesized direct fitness benefits of helping include: reciprocal help from recipient broods or breeders (Wiley and Rabenold, 1984); experience in brood care (Skutch, 1961); payment of rent for the benefits of living on the breeders’ territory or in their group (Gaston, 1978); access to direct reproduction in the current brood (Richardson et al., 2001, 2002); inheritance of either a territory or breeding partner (Reyer, 1990); and group augmentation and the enhancement of the benefits of group living (Kokko et al., 2001). Not all of these potential benefits apply to the cooperative breeding system of long‐tailed tits. For example, the absence of territoriality eliminates the possibility of territory inheritance, and the fact that most helping occurs within a generation reduces the potential for cross‐generation reciprocity. Therefore, here we first consider three potential direct fitness benefits that are likely to accrue to long‐tailed tit helpers: shared reproduction in helped broods, acquisition of experience, and survival benefits. We then consider the opportunity for enhanced future reproduction. 1. Shared Reproduction by Helpers Helper reproduction is likely to be constrained by inbreeding avoidance in nuclear families (Pusey and Wolf, 1996; Ralls et al., 1986; but see Keane et al., 1996; Reeve et al., 1990; Richardson et al., 2001), but some cooperatively breeding species have more complex social or reproductive strategies in which helpers have a reproductive stake in the brood that they assist in rearing, technically making them cobreeders rather than helpers. Their help therefore has a direct fitness return. The ecological or evolutionary basis for this rich diversity of reproductive strategies among cooperative breeding systems is poorly understood, despite the extensive body of theory that has been developed to explain reproductive skew in social animals (Clutton‐ Brock, 1998; Keller and Reeve, 1994; Magrath et al., 2004; Reeve et al., 1998; Vehrencamp, 1983, 2000). However, it has become clear that helper reproduction is more widespread than was once thought (Cockburn, 1998, 2004). In the case of long‐tailed tits, most males are helpers and they typically help at the nest of a male relative who is paired to an unrelated female, so incest avoidance is unlikely to constrain helpers from seeking paternity in a brood for which they subsequently provide food. In this section, we describe the outcome of molecular genetic investigations into parentage in long‐tailed tits with the aim of testing the hypothesis that shared parentage of broods plays a role in the caring decisions of helpers. Eight polymorphic microsatellite loci were used to assign paternity and maternity of 296 offspring from 39 complete families; a further 73 offspring (12 broods) were compared with their putative mothers only and 84 offspring (10 broods) with putative fathers only. Virtually all offspring (365/369, 98.9%) were assigned to putative mothers with a high degree of confidence;

372

BEN J. HATCHWELL AND STUART P. SHARP

3 of the 4 exceptions (all of which were in one brood) were assigned to a female that laid a full clutch of eggs in a nest before she disappeared (presumably dead) and a second female, the putative mother, laid a second clutch and reared a brood that contained offspring of both females (Hatchwell et al., 2002). We have recorded just two other cases of joint nesting by females (one with synchronous laying by two females and another in which a second female laid eggs in a nest containing nestlings) in over 1200 breeding attempts over 12 years. Thus, intraspecific brood parasitism occurs at negligible frequency in long‐tailed tits and in no case was a female assigned parentage in a brood at which she was a helper (Hatchwell et al., 2002). Assignment of paternity is difficult in a kin‐structured population with closely related males available as potential sires (Thompson and Meagher, 1987). Nevertheless, it is clear that extra‐pair paternity (EPP) was infrequent: putative fathers were excluded for just 2.4% of nestlings in 16% of broods in likelihood‐based estimates of paternity using CERVUS (Marshall et al., 1998) and the upper estimate of extra‐pair paternity from inclusion analyses (which is likely to be confounded by the presence of closely related males in the population and hence overestimate extra‐pair paternity) was just 6.9% of nestlings in 29% of broods (Hatchwell et al., 2002). Unusually among cooperative breeders, divorce is frequent between years, and some mate switching also occurs within years, with 7–9% of birds breeding with at least two partners within a single breeding season (Hatchwell et al., 2000). However, field observations showed that mate switching was not responsible for any of the observed cases of extra‐pair paternity in the sample of genotyped nestlings. It is more likely that observed EPP results from extra‐pair copulations, and the low rate of EPP is consistent with the facts that copulations occur at a low rate (Hatchwell et al., 2002) and that long‐tailed tits have relatively small testes (Birkhead, T. R., personal communication). Some male helpers (6/28, 21%) were assigned paternity of one or more nestlings in the broods they helped to rear, but these cases accounted for just 11/288 (3.8%) nestlings whose paternity was assigned. Furthermore, paternity assignment in these cases was generally equivocal because of the close kinship of the putative father and helper; indeed, in an exclusion analysis the putative father was excluded for just 1 of these 11 nestlings, giving a lower estimate of just 1/28 (3.6%) helpers with paternity in the helped brood. A substantial fraction of any extra‐pair young would be expected to be fathered by males that became helpers simply because around a third of close neighbors are also close kin, and it is these relatives that are also likely to become helpers (Russell and Hatchwell, 2001). It should also be noted that any shared parentage in a brood is not a direct benefit of helping per se because males do not need to help to achieve extra‐pair paternity, although of course by feeding a brood containing their extra‐pair offspring

KIN SELECTION AND COOPERATION IN LONG‐TAILED TITS

373

they enhance the survival prospects of those offspring (Section V.B.1). Therefore, we concluded that any tendency for the few extra‐pair offspring to be fathered by helpers is a consequence of male philopatry rather than an important route to fitness for helpers (Hatchwell et al., 2002). We are genotyping a much larger sample of broods with microsatellites giving substantially better resolution of parentage, permitting more accurate assignment of paternity among potential fathers. 2. Experience This idea, originally proposed by Skutch (1961) as the ‘‘skills’’ hypothesis, argues that young delay dispersal and assist their parents while they acquire experience that enables them to breed independently or enhances their own future reproductive success. Support for this hypothesis comes from Seychelles warblers Acrocephalus sechellensis, the benefit of helping experience for future productivity being as large as the benefit of breeding experience (Komdeur, 1996). However, the hypothesis is difficult to test because the acquisition of experience is often confounded with age or dispersal decisions (Dickinson and Hatchwell, 2004). Long‐tailed tit nests suffer a high rate of predation, but the risk of predation is significantly lower for nests sited low in vegetation (<2 m) than those placed in trees (Gaston, 1973; Hatchwell et al., 1999; Lack and Lack, 1958; Riehm, 1970). In an analysis of long‐tailed tit nest site selection, Hatchwell et al. (1999) found that males that became helpers at successful nests placed their own subsequent nests lower than nonhelpers and lower than the nests they had built prior to becoming helpers, suggesting that helpers might learn the characteristics of a good nest site during their period of helping. It was argued that experience gained as a helper was more valuable than that gained as a breeder because the failure rate of nests is so high that even a breeder’s low nests are three times more likely to fail than to succeed; in contrast, helpers are associated with nests (which will tend to be low because high ones have been depredated before the nestling period) for a relatively short period prior to fledging. The flaw in this argument is that it is difficult to explain why a failed breeder needs to become a helper in order to identify the characteristics of nest sites that are likely to avoid predation. A subsequent analysis on a larger sample (Section V.A.4) suggests that helping does not enhance an individual’s chance of future successful reproduction. Further indirect evidence against the experience hypothesis derives from analyses of factors influencing reproductive measures (Hatchwell et al., 2004) that found no significant effect of male or female age on nestling provisioning rates (n ¼ 156 broods), nestling mass on day 11 (n ¼ 123 broods), brood survival (n ¼ 113 broods), or recruitment (n ¼ 93 broods).

374

BEN J. HATCHWELL AND STUART P. SHARP

3. Survival Benefits The potential benefits of group living (Krause and Ruxton, 2002) may be gained by simply living in close proximity to conspecifics, and do not necessarily require apparently cooperative behavior among group members (Hamilton, 1964a,b). However, even if helpers gain nothing through helping per se, their help may be exchanged for the direct fitness benefits they gain from being allowed to remain on a territory or to join a social group. This hypothesis has been termed the ‘‘payment of rent’’ (Gaston, 1978) or ‘‘pay to stay’’ (Kokko et al., 2002; Mulder and Langmore, 1993) hypothesis. Glen and Perrins (1988) suggested that gaining access to the thermoregulatory benefits of communal roosting may be a principal reason for helping by long‐tailed tits that failed in their own breeding attempt and were therefore without a group with which to spend the winter. Likewise, group members may enhance their own direct fitness benefits of group‐ living by increasing the size of their social group through helping, termed the group augmentation hypothesis (Kokko et al., 2001), an idea that does not require that help be directed toward kin. Using 7 years of data, survival parameters were estimated using MARK (White and Burnham, 1999) for long‐tailed tits from the Rivelin Valley with the purpose of comparing the survival rates of adults of differing reproductive status (n ¼ 136 males and 114 females). The objective was to test the hypothesis that failed breeders that became helpers had higher survival than failed breeders that did not become helpers. In the best‐fitting model, the survival of adult long‐tailed tits varied with breeding status; failed breeders that became helpers had the same survival probability (56%) as successful breeders, while those failed breeders that did not become helpers had a lower survival probability of 46% (McGowan et al., 2003). Thus, helpers appear to gain survival benefits through their cooperative behavior. However, neither the payment of rent hypothesis nor the group augmentation hypothesis is likely to apply to long‐tailed tits. The payment of rent hypothesis requires that access to a territory or group should be conditional on the provision of help during the preceding breeding season, but several lines of evidence suggests that provisioning young is not a prerequisite for group membership. First, nonbreeding flocks in October–January (when thermoregulatory demands are expected to be at their highest) typically include a variable proportion of apparently unrelated immigrants who were not associated with other members of the flock during the preceding breeding season and so could not have ‘‘paid rent’’ (Hatchwell et al., 2001b; McGowan, 2002; Russell, 1999). Second, group augmentation is unlikely to work because nonbreeding flocks are not composed solely of nuclear families and their helpers. Instead, winter flocks usually include birds from two or more nuclear

KIN SELECTION AND COOPERATION IN LONG‐TAILED TITS

375

families of surviving breeders, helpers, and their offspring (often following amalgamation of flocks as they decrease in size through emigration and mortality), in addition to the immigrants mentioned earlier (Hatchwell et al., 2001b; Russell, 1999). Furthermore, among birds of different breeding status (i.e., helper vs nonhelper) there was no significant difference in the size of their nonbreeding flock (McGowan, 2002). Therefore, we have no evidence that the difference in survival between failed breeders that did and did not help is related to helping per se. Neither is there any evidence that there are intrinsic differences in quality between these two categories of birds (Hatchwell et al., 2004; McGowan et al., 2003). Instead, we suggest that this differential survival might result from the nepotistic benefits of living with kin during the nonbreeding season. McGowan (2002) showed that when compared to failed breeders that did not help, failed breeders that helped (and successful breeders) spent the winter in flocks containing a significantly higher proportion of close relatives, suggesting that there may be benefits to associating with kin. In other words, the apparent association between helping and higher survival may simply be a consequence of both being correlated with the presence of relatives. In Siberian jays Perisoreus infaustus, parental concession of food, reduced levels of parental aggression, and nepotistic alarm calling confer benefits on philopatric young compared to dispersers (Ekman et al., 2000, 2004; Griesser and Ekman, 2004), and similar benefits could operate in long‐tailed tits. We are investigating, using captive flocks, the effects of relatedness and other factors on social dominance and the outcome of competition for positions within communal roosts (McGowan et al., in press; Hatchwell, B. J., and Sharp, S. P., unpublished data). In this purely correlational investigation of the effects of sociality, helping, and other factors on breeder and helper survivorship, it is extremely difficult to demonstrate causation, but, at present, we have no unequivocal evidence that helping per se has a positive effect on an individual’s chance of survival. Nevertheless, we do not discount the possibility that such effects may become apparent in future analyses of survivorship. 4. Enhanced Probability of Successful Future Reproduction The acid test for demonstrating a direct fitness benefit of helping is to ask whether failed breeders that became helpers enhanced their probability of successful personal reproduction in the future, relative to nonhelpers. In the same MARK analysis as described earlier, McGowan et al. (2003) used the same sample of adults to determine whether the probability of a bird surviving to year n þ 1 and breeding successfully in that year (measured as offspring fledged or no offspring fledged) depended on their helping status (failed breeder that helped or failed breeder that did not

376

BEN J. HATCHWELL AND STUART P. SHARP

help) in year n. In fact, a helper’s chance of surviving and reproducing successfully (15%) was remarkably close to that of a nonhelper (17%), and it was concluded that the direct fitness payoffs for future personal reproduction from adopting either strategy are roughly equal. This analysis is simplistic in that it fails to capture the full intricacies of the long‐tailed tit cooperative breeding system. However, the outcome of this analysis is in striking contrast to our findings regarding the indirect fitness benefits of helping in this species. B. INDIRECT FITNESS BENEFITS Helpers may gain indirect fitness benefits either by helping relatives to increase productivity of their current breeding attempt, or by reducing the reproductive costs of related breeders, thereby enhancing their survival. We consider these possibilities in turn. 1. Increased Productivity of Relatives We have described observational and experimental evidence that helpers preferentially direct their care toward relatives. Here we ask whether that care provided by helpers increases the productivity of assisted breeders over a range of timescales. In the short term, the females of some species are able to anticipate the additional help that they are likely to receive in raising a brood and lay a correspondingly large clutch of eggs (Davies and Hatchwell, 1992). There is no evidence that female long‐tailed tits do this (Hatchwell et al., 2004), which is unsurprising because females cannot know with any certainty whether they will have helpers at the time that they are laying eggs. Helpers are associated with nests only during the nestling period and their presence is dependent on a nearby relative failing in its own breeding attempt. Moreover, most helpers are relatives of the male breeder and females may be unaware whether a male has any surviving relatives, that is, potential helpers. Many studies have reported a positive relationship between group size (or number of helpers) and brood size at fledging (Emlen, 1991, 1997; Stacey and Koenig, 1990), although as pointed out earlier there have been few experimental studies and correlative investigations are likely to be confounded by variation in individual and/or habitat quality (Cockburn, 1998; Legge, 2000). In long‐tailed tits, a correlative approach was taken to ask whether helpers increase brood size at fledging by reducing either brood depredation or nestling starvation (Hatchwell et al., 2004). First, 34% of broods (n ¼ 113) in unprotected nests failed between hatching and fledging, mainly through predation by corvids or mustelids, and helpers

KIN SELECTION AND COOPERATION IN LONG‐TAILED TITS

377

had neither a positive nor a negative effect on the probability of failure. Long‐tailed tits are ill equipped to repel either avian or mammalian predators (their nests may even be taken over during building or egg‐laying by bumblebee Bombus queens seeking nest sites), and there was no evidence that the additional activity at helped nests attracted more predators (Hatchwell et al., 2004). Second, in those nests that were not depredated, nestling survival from hatching (day 0) to ringing on day 11 of the 16‐day nestling period was extremely high: 98% of the mean brood of 9.1 nestlings (n ¼ 133 broods) survived, on average. Unsurprisingly, given the very low nestling mortality rate, there was no significant effect of helpers on nestling survival and hence brood size; nor was there any effect of maternal or paternal identities, breeder ages, or nest location (Hatchwell et al., 2004). Therefore, no short‐term effect of helpers on productivity has been detected. However, helpers do increase the rate at which broods are provisioned, the total amount of food delivered to nests by all carers showing a highly significant increase as the number of helpers assisting breeders increases (Fig. 6A). There is a corresponding increase in nestling mass within broods as the number of helpers increases (Fig. 6B) and, as earlier, there was no effect of either breeder identity, age, or nest location in either analysis (Hatchwell et al., 2004). Recruitment is positively related to nestling mass in many bird species (Garnett, 1981; Magrath, 1991), and so it is reasonable to propose that these effects of helpers on food provision to nestlings and hence their mass should also influence subsequent recruitment. The proportion of young from fledged broods that recruited locally within the study areas averaged 19.5% (0–80%, n ¼ 93 broods). The number of helpers that fed the brood had a significant effect on recruitment (Fig. 6C); those fledglings that had been fed by a pair plus three or more helpers had a 41% chance of recruiting compared to just 12% for pair‐fed fledglings. Recruitment was not significantly influenced by individual breeder identity, age, or location (Hatchwell et al., 2004). This dramatic impact of helpers on fledgling survival to their first year does not appear to persist beyond then because McGowan et al. (2003) found no long‐term effect of helpers on the survival of recipients beyond their first year. Using the results in Fig. 6C, the productivity of broods (in terms of genetic equivalents) can be calculated from the perspective of helpers and of breeders (Fig. 6D). This shows that from the breeder’s perspective there is no reduction in the marginal benefit of having helpers as the number of helpers increases, at least over the range of helper numbers used in this analysis. In other words, potential helpers need not take into account the current number of helpers at a nest when deciding where or whether to help. Likewise, breeders should welcome as many helpers as are

378

BEN J. HATCHWELL AND STUART P. SHARP

A

B 35 Nestling mass (mean mass per brood)

8.0

Mean provisioning rate (feeds/hr)

30 25 20 15 10

7.5

7.0

6.5

5 0

6.0 0

2 1 Number of helpers

3+

1 2 Number of helpers

3+

0

1 2 Number of helpers

3+

D

C

Productivity per fledged brood (genetic equivalent)

50 Mean probability of recruitment (%)

0

40 30 20 10 0 0

1 2 Number of helpers

3+

2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

Fig. 6. The effect of variation in the number of long‐tailed tit helpers on: (A) Total provisioning rates (residual maximum likelihood, REML, model estimates  S.E., n ¼ 963 observation periods on 156 broods; number of helpers: p < 0.0001). (B) Mean nestling mass per brood (REML model estimates  S.E., n ¼ 68, 23, 20, 12 broods for 0, 1, 2, 3þ helpers, respectively; number of helpers: p < 0.001). (C) Probability of recruitment per fledged brood (iterated residual maximum likelihood, IRREML, model estimates  S.E., n ¼ 41, 24, 14, 14 broods for 0, 1, 2, 3þ helpers, respectively; number of helpers: p ¼ 0.003). (D) Productivity per fledged brood, measured in genetic equivalents for breeders (hatched bars) and helpers (open bars), estimated from recruitment rates in Fig. 6C, fledging brood size ¼ 8.9, coefficient of relatedness of breeder to fledgling ¼ 0.48, and of helper to fledgling ¼ 0.22 (adapted from Hatchwell et al., 2004; reproduced with permission from the International Society for Behavioral Ecology).

willing to feed their brood. These situations can only persist up to a certain group size because there must inevitably be some diminution in the marginal effect of helpers as fledgling recruitment approaches 100%. Nevertheless, over the normal range of helper numbers the argument holds. The other conclusion to be drawn from Fig. 6D is that when considering individual broods, marginal productivity as a helper (mean ¼ 0.14 genetic equivalents per helper) is substantially lower than productivity as a breeder without any helpers (0.51 genetic equivalents). This comparison is consistent

KIN SELECTION AND COOPERATION IN LONG‐TAILED TITS

379

with the view that, on average, it is better for an individual to breed independently than it is to help (Emlen, 1982). In Section VI, we consider the ecological circumstances that cause long‐tailed tits to switch from attempting to breed independently to becoming helpers even though it appears to be a less rewarding option. 2. Increased Survival of Related Breeders The other potential source of indirect benefits for long‐tailed tit helpers is to reduce the reproductive costs of breeders by allowing them to work less hard when provisioning their brood. Observations of natural variation in provisioning rates at 135 nests showed that breeders reduced their provisioning rate by around 20% when they had one or more helpers. This reduction in effort by breeders when helped was associated with the arrival of their first helper, but neither breeders nor helpers reduced their work rate any further with the arrival of additional helpers (MacColl and Hatchwell, 2003a). In an experimental test of the hypothesis that helpers lighten breeders’ reproductive load, Hatchwell and Russell (1996) conducted experiments in which helpers were temporarily removed; the work rate of all carers was determined during observations on the day before the removal, during the removal, and on the day after the removal when helpers had been returned and had resumed provisioning of the brood. This experiment confirmed that breeders significantly increased their provisioning effort when helpers were absent and then significantly reduced effort when they were returned. Across species, such a reduction in the provisioning effort of breeders when assisted by helpers is associated with a low nestling starvation rate. In contrast, when nestling starvation is relatively high, breeders maintain their normal provisioning effort so that the care of helpers is additive, increasing the total rate of food delivery to the brood (Hatchwell, 1999). Long‐tailed tit breeders have an intermediate response in which breeders initially reduce their effort, but the care of further helpers is purely additive. The logic of the breeders’ strategic reduction in effort is that there should be a corresponding increase in breeder survival (Crick, 1992; Heinsohn, 2004), and this opportunity to reduce costs is taken when the marginal benefit of additional food for offspring survival is low (Hatchwell, 1999). Therefore, given the reduced effort of breeders when helped, it would be predicted that they should have reduced mortality rate compared to breeders without helpers. However, in the MARK analysis described earlier, McGowan et al. (2003) compared the survival to the following breeding season of breeders with and without helpers and found no significant difference between them (both 56%). Likewise, McGowan et al. (2003) found no evidence that the presence of helpers in year n increased the probability

380

BEN J. HATCHWELL AND STUART P. SHARP

of survival and successful reproduction in year n þ 1 (both 34%). The reason that load lightening has no significant impact on breeder survival is unknown, and in the absence of such an effect, breeders’ tactics of reducing their effort when helped appears paradoxical, especially given the apparently strong effect of increased provisioning on fitness (Hatchwell et al., 2004; MacColl and Hatchwell, 2004). It may be that the stochastic nature of many of the other influences on adult survival limits our ability to detect any effect without a very large sample of adults. Whatever the reason, at present we can conclude only that helpers do not appear to acquire additional indirect fitness by increasing the survival of related breeders.

C. INCLUSIVE FITNESS We have concluded that there is strong evidence for indirect or kin‐ selected benefits from helping via the increased productivity of kin, but, as yet, we have no unequivocal evidence that helping has any substantial direct fitness benefit for helpers. Thus, we think that the cooperative breeding system of long‐tailed tits is the product of kin selection, but how important is the kin‐selected component of inclusive fitness? More specifically, how does the indirect fitness gain of helping compare with the direct fitness gain of independent reproduction? To answer this question, MacColl and Hatchwell (2004) quantified lifetime reproductive success (LRS, the sum of all local recruits produced in an individual’s lifetime) of 228 long‐tailed tits for which complete reproductive histories were available, and used these data to calculate individual fitness (lambda, which approximates inclusive fitness) using the method of McGraw and Caswell (1996) and Oli (2003). The two measures of fitness were highly correlated and both were significant predictors of the number of grand‐offspring an individual had. As in many small passerines (Newton, 1989), fitness measures were strongly skewed: just 55/228 (24%) of birds achieved nonzero LRS (0–13 local recruits) and just 70/228 (31%) achieved nonzero individual fitness. Of the 70 birds with nonzero individual fitness, 15 (21%) accrued fitness only indirectly, that is, they had zero LRS and gained inclusive fitness only through helping. For these 70 birds, the average contribution of indirect fitness to individual fitness was 22%. However, it was unusual for individuals to accrue fitness through both routes because just two birds that achieved some LRS also gained fitness by helping (MacColl and Hatchwell, 2004). This means that the indirect component of inclusive fitness is substantial and, most interestingly, helping provides an alternative route to achieving some fitness for a proportion of individuals in the population that would not otherwise have achieved any personal LRS.

KIN SELECTION AND COOPERATION IN LONG‐TAILED TITS

381

Our earlier conclusion regarding the significance of parental investment for offspring fitness was also borne out in this analysis. Local recruitment of offspring was positively related to the provisioning effort of their mothers (although not their fathers), and as a result the individual fitness of females was positively related to their provisioning effort (MacColl and Hatchwell, 2004). The provisioning effort of parents is repeatable within individuals, is not confounded by environmental effects, and most notably, has been shown using parent–offspring regression and an animal model approach to have a significant heritable component (h2 ¼ 0.43; MacColl and Hatchwell, 2003b); at present it is not clear what trade‐off maintains this variation in a fitness‐determining trait.

VI. ECOLOGICAL BASIS

FOR

COOPERATIVE BREEDING

Finally, we address the question of what ecological factors have led to the evolution of cooperative breeding in long‐tailed tits. Globally, cooperative breeding is an unusual phenomenon, and within the Palearctic avifauna it is particularly rare (Arnold and Owens, 1998; Cockburn, 1998). Long‐tailed tits are clearly atypical cooperative breeders because they do not follow the usual sequence of events of delayed dispersal of offspring and deferred personal reproduction, helping of subsequent broods, followed by breeding either after dispersal to another territory, or inheritance of the natal territory. In contrast, there appears to be no constraint on dispersal, or shortage of mates, or habitat for breeding, because all birds attempt to breed each year and helping only occurs following breeding failure. Therefore, at first sight, the ecological constraints hypothesis of Emlen (1982) appears not to apply to this species. However, this model states that cooperative behavior arises because of constraints on independent reproduction, and Emlen (1982) identified four specific kinds of constraint that might result in cooperative breeding: high risks of early dispersal, shortage of territories, shortage of mates, and prohibitive costs of independent reproduction. The latter constraint is most likely to apply here. Several other cooperative breeders switch from breeding to helping within a season, behavior termed ‘‘redirected helping,’’ and this may be explained by poor success later in the breeding season (Dickinson et al., 1996; Emlen, 1982; Lessells, 1990). A. MODEL

OF

FITNESS PAYOFFS

FROM

BREEDING AND HELPING

No study had demonstrated a quantitative link between the relative fitness payoffs from breeding and helping and the timing of the switch from breeding to helping in species with redirected helping behavior. Therefore,

382

BEN J. HATCHWELL AND STUART P. SHARP

to determine whether temporal constraints on reproduction drives the switch from breeding to helping in long‐tailed tits, seasonal variation in reproductive success was calculated from long‐term data (1994–2000) and used to model temporal variation in the expected fitness payoffs from the alternative reproductive tactics of breeding and helping (MacColl and Hatchwell, 2002). A seasonal decline was observed in brood size (n ¼ 85 fledged broods), nestling weight on day 11 (n ¼ 931 nestling from 93 broods), and in the probability of fledglings recruiting into the breeding population in the following spring (n ¼ 490 fledglings from 64 fledged broods). Similar declines in reproductive success, mediated via clutch or brood size and offspring survival, are widely reported among other temperate passerine species (Both et al., 1999; Magrath, 1991; van Noordwijk et al., 1995). Seasonal changes in these reproductive parameters were then used to generate models of within‐season temporal variation in the expected fitness payoffs from independent breeding and from helping. The models described the expected fitness payoff for a breeder that fails in a breeding attempt at time t and then adopts one of two behavioral tactics: attempt to breed again or help at an existing nest. The expected reproductive payoff from breeding again (Eb) is: Eb ¼ ntþx  stþx  rb where n is brood size, s is fledgling survival, rb is the coefficient of relatedness of parents to their own brood, and x is the delay between failing in one attempt and fledging a new brood of chicks. The expected reproductive payoff in terms of increased inclusive fitness of adopting the alternative tactic of becoming a helper (Eh) is modeled as: Eh ¼ ntþy  ðsh  su Þtþy  rh where su is the survival of fledglings from broods without helpers, sh is the survival of fledglings from a helped brood, rh is the coefficient of relatedness of a helper to the brood, and y is the delay between failing in one attempt and fledging the brood that is being helped. These models are a simple representation of the likely fitness payoffs but they capture all of the key factors thought to affect the payoffs from the alternative reproductive tactics of breeding and helping. Specific assumptions are that helpers derive no direct fitness benefit from helping and that the costs of breeding and helping are the same. Both of these assumptions are reasonable because elsewhere it has been shown that although helpers feed nestlings less frequently than breeders, the probability of survival to the following year of helpers and breeders is the same (McGowan et al., 2003), and, as discussed in the previous section, evidence suggests that helpers

KIN SELECTION AND COOPERATION IN LONG‐TAILED TITS

383

do not derive any substantial direct fitness benefits through their helping behavior. Other notable assumptions of the models are that breeders are the parents of chicks in their own nest (rb ¼ 0.5) and that helpers are first order relatives of one breeder so that rh ¼ 0.25. Again these are fair assumptions because intraspecific brood parasitism and extra‐pair paternity are rare in long‐tailed tits (Section V.A.1.; Hatchwell et al., 2002), and most helpers (69%) are first order kin of one breeder, although in some instances they are either related to both breeders, are more distant relatives, or are apparently unrelated to both breeders (Russell and Hatchwell, 2001). In the models presented earlier, the risk of nest failure of repeat nests is ignored so the equation describes the maximum expected payoff from breeding. This will not always be achieved because long‐tailed tit nests suffer high rates of nest failure (Hatchwell et al., 1999), so a more realistic scenario might incorporate the risk of nest failure (measured as a daily rate of failure over time x or time y for breeders and helpers, respectively), generating the minimum expected fitness payoff. Of course, it is not known whether long‐tailed tits take into account the probability of nest predation when making their decision of whether to renest or help, so these minimum and maximum payoffs were both used to calculate expected payoffs of the alternative tactics and to generate predictions about the timing of the switch from independent breeding to cooperation. For further details of the model and its assumptions, see MacColl and Hatchwell (2002). The patterns of seasonal variation in brood size and fledgling survival determined from statistical models were then used to parameterize the expected fitness payoff models (Fig. 7A). The fitness payoff from breeding was initially high, but declined rapidly through the season, while the fitness payoff from helping was initially lower than that from breeding, but declined at a relatively slow rate. The same is true when the probability of nest predation was taken into account, although the starting point for each curve was lower. As a result, the fitness payoff from helping exceeded that of breeding after day 57 (April 27) when nest predation was taken into account generating minimum payoff curves and after day 73 (May 13) when nest failure was not included generating maximum payoff curves. The model therefore predicts that long‐tailed tits that fail in a breeding attempt should switch from independent breeding to helping during a 16‐day period of the breeding season. The principal reason why the model generates this switch in the relative magnitudes of the payoffs from breeding and helping is the difference between x and y, the time delays between failing in one attempt and fledging a brood as a breeder or helper, respectively. In the model, the average time before a brood fledges for helpers is just 6 days (because helpers usually join nests midway through the nestling period), while the time taken to fledge a brood for birds that nest again is 40–43 days, depending on clutch size.

384

BEN J. HATCHWELL AND STUART P. SHARP

Fig. 7. Seasonal change in payoffs of independent breeding and helping tactics, and change in observed adoption of different tactics. (A) Model predictions of seasonal change in fitness payoffs of breeding (solid lines) and helping (dashed lines) for birds that take account of the probability of nest failure (gray), and for those that do not (black). The stippled area indicates the window of time during which a switch from breeding to helping behavior is expected. (B) Observed seasonal change in the probability of birds renesting following failure (¨, n ¼ 209 pairs). The bold line is the fitted relationship from a logistic regression. The faint line (?) is the cumulative percent of helpers arriving at nests (n ¼ 94 helpers) (adapted from MacColl and Hatchwell, 2002; reproduced with permission from University of Chicago Press, # 2002 by The University of Chicago Press).

KIN SELECTION AND COOPERATION IN LONG‐TAILED TITS

B. MATCHING MODEL PREDICTIONS

WITH

385

OBSERVED BEHAVIOR

How did the predicted time of switching match the observed switch in reproductive tactics? First, using data from 1994–2000, the change in the proportion of failed breeders that renested as the season progressed showed that early in the season all birds that failed in one breeding attempt tried to nest again, but later in the season all failed pairs abandoned breeding for that year. Second, the number of helpers at nests increased through the season as the number of birds abandoning breeding increased. The switch from independent breeding to helping coincided very closely with the predicted switch (Fig. 7B). This analysis suggests that the seasonal decline in reproductive success acts as a temporal constraint on independent reproduction, so that failed breeders switch from breeding to helping when the inclusive fitness gain anticipated from helping kin exceeds that expected from breeding. It is a notable feature of this model that it includes only indirect fitness benefits from helping; the close fit of the observed time of switching to the time predicted by the model suggests that this indirect‐fitness‐only model provides a reasonable assessment of the fitness consequences of helping. The environmental factor, or specific ecological constraint, that drives the seasonal decline in breeding success, and hence the switch to helping by failed breeders, is currently unidentified. However, there is evidence that the cues used to initiate and terminate breeding may not be the same. The start of breeding is highly variable between years: the median lay date of first clutches in the Rivelin Valley population varies by up to 21 days (March 30–April 20) over 12 years, and is closely correlated with mean daily temperature during the nest‐building period, February–March (MacColl and Hatchwell, 2002; Hatchwell, B. J., unpublished data). This effect of temperature presumably influences the timing of egg laying indirectly through its effects on female nutrition or energetics during the nest‐ building and egg‐formation periods, as found in many other temperate passerines (Crick and Sparks, 1999; Perrins, 1965; Stevenson and Bryant, 2000). In contrast, the timing of the switch between reproductive tactics from breeding to helping does not vary significantly between years, occurring predictably in the first 10 days of May in each year (MacColl and Hatchwell, 2002), and this is also true using a longer run of 11 years of data (Hatchwell, B. J., unpublished data). Therefore, we suspect that a temporally fixed cue, for example, photoperiod, rather than a variable biotic cue, such as food supply, causes failed breeders to abandon breeding. One other factor that clearly has some influence on the decision of failed breeders to abandon breeding is the stage of the reproductive cycle at the time of nest failure. The timing of the switch for failed breeders that were

386

BEN J. HATCHWELL AND STUART P. SHARP

either at the nest‐building, egg‐laying, or incubating stage of the cycle did not differ significantly, but breeders whose nest failed after their eggs had hatched very rarely renested, even if the failure occurred earlier than the usual switch point (Hatchwell et al., 2004). Physiological changes that breeders experience when they start provisioning nestlings, for example, a reduction in testosterone levels and regression of reproductive organs may be responsible for this effect (Ketterson and Nolan, 1994). Another factor that might be expected to influence the decision of whether a failed breeder should renest is the presence or absence of close relatives with active nests in the population. The fitness payoff that failed breeders derive from helping is accrued only if they assist relatives because, as shown earlier, we have not been able to measure a substantial direct fitness benefit from helping. Therefore, the time at which failed breeders decide to abandon breeding should be a function of kin availability: failed breeders without a related brood to help should be more likely to renest than those failed breeders with the option of gaining indirect fitness through helping. However, the detailed genetic information on the presence or absence of close kin in the breeding population that is required to test this prediction rigorously is not yet available. VII. CONCLUSIONS A. KIN SELECTION The main conclusions regarding the role of kin selection in the evolution of cooperative breeding in long‐tailed tits are that: (1) helpers exhibit a kin preference in helping behavior using a learned vocal kin recognition mechanism; (2) helpers increase the productivity of their relatives by increasing recruitment of fledglings of the helped brood; and (3) the kin‐selected fitness benefit of helping is the sole source of inclusive fitness for a substantial proportion of individuals. The first of these conclusions derives from observations and field experiments, so we think that the case for active kin discrimination by helpers, using learned vocal recognition cues, is strong. Few experimental investigations of kin discrimination in cooperative societies exist, so this represents an important finding. Even within the apparently kin‐selected cooperative breeding system of long‐tailed tits, we have no evidence that helpers make fine adjustments of effort in relation to their precise kinship to a brood (Hatchwell, B. J., unpublished data), and we think this is a consequence of using a simple rule of thumb for kin recognition. A learned kin recognition cue would not function effectively as a kinship label for individuals that were not associated during the appropriate period

KIN SELECTION AND COOPERATION IN LONG‐TAILED TITS

387

of development because of the rapid diluting effect of learning from parents in an outbred population. Thus, this learning mechanism limits the pool of potential beneficiaries of kin‐directed cooperation to the subset of kin within the population with whom the helper has had direct association. In this sense the recognition mechanism is conservative because it should not result in helping of non‐kin. This makes evolutionary sense because our analyses of the fitness consequences of cooperation suggest that helping is beneficial only when directed toward kin. The other important general point to make is that the absence of fine discrimination among kin of varying degrees of relatedness is not necessarily evidence against kin selection, and an understanding of recognition mechanisms may be crucial for understanding helping behavior, the role of kin selection, and the limits of adaptive behavior. Our second conclusion is derived from observational data and so is open to the charge that the positive effect of helpers on recruitment is actually a function of parental or habitat quality (Section I). We do not believe this to be the case for three reasons. First, although parental effort is repeatable and heritable (MacColl and Hatchwell, 2003b) we have included individual identity and age in statistical models when testing for helper effects on recruitment and they have always been nonsignificant. Second, long‐tailed tits are not territorial and have extremely large home ranges and again there is no evidence that habitat affects either provisioning rates, nestling condition, or recruitment (Hatchwell et al., 2004; MacColl and Hatchwell, 2003b). Finally, the presence of helpers at a specific nest is a largely stochastic process depending on the timing of breeding attempts and the lottery of predation. We cannot be so confident of the process through which helpers increase recruitment. Their positive effect on nestling mass is compelling, but the long postfledging association of helpers and parents with juveniles also provides opportunities for nepotism that we are currently investigating. Our third conclusion is consistent with the widespread finding that helping is a best‐of‐a‐bad‐job strategy for individuals that are not able to breed independently or successfully. However, an interesting outcome of our analysis of LRS data is that the individuals that acquire inclusive fitness indirectly via helping rarely acquire any fitness directly through personal reproduction. Therefore, in practice, helping is an alternative route to acquiring fitness for a significant portion of the population that would not otherwise achieve any fitness. Therefore, while researchers have traditionally asked why cooperative behavior has evolved in a particular species, given the fact that LRS is usually strongly skewed within bird populations (Newton, 1989), we suggest that it is equally valid to ask why this alternative route to acquiring fitness has not been adopted more widely. If one takes the view that cooperative breeding is the ancestral condition among

388

BEN J. HATCHWELL AND STUART P. SHARP

Passeriformes (Cockburn, 1996; Ligon and Burt, 2004), the equivalent question might be why helping has been lost in so many lineages. B. CONSTRAINTS A seasonal switch in the relative magnitude of the inclusive fitness payoffs from breeding and helping may explain the evolution of cooperative behavior in a number of other species in which some individuals redirect their effort from breeding to helping within a season; such species include western bluebirds Sialia mexicana (Dickinson et al., 1996), European bee‐ eaters Merops apiaster (Lessells, 1990), white‐fronted bee‐eater (Emlen, 1982), bushtits (Sloane, 1996), Galapagos mockingbird Nesomimus parvulus Curry and Grant, 1990), and rifleman Acanthisitta chloris (Sherley, 1990). An equivalent switch from breeding to helping, as their relative fitness payoffs change through a lifetime, has been proposed as an explanation for the phenomenon of menopause in human females (Lahdenpera¨ et al., 2004). The fascinating question that arises is why this reproductive tactic (with a potentially substantial inclusive fitness benefit, as in long‐tailed tits) is not more widespread among other species that exhibit similar seasonal declines in productivity (Linden et al., 1992; Magrath, 1991; Verboven and Visser, 1998; Verhulst and Tinbergen, 1991). A key point is likely to be the availability of relatives. In long‐tailed tits, helpers accrue inclusive fitness through cooperation only if they assist kin, and they show an appropriate degree of discrimination, with the result that many failed breeders do not adopt the tactic of helping when they fail in their own independent breeding attempts. Redirected helping is also known to be kin directed in several of the species listed earlier. Therefore, one explanation for the general absence of redirected helping despite widespread seasonal declines in productivity in noncooperative species is that populations of most species lack the genetic structure necessary to make kin routinely available for helping. Limited dispersal in viscous populations may favor the evolution of cooperative behavior (Hamilton, 1964a,b; Queller, 1992), but appropriate conditions for kin‐directed cooperation may also be a consequence of certain life history or demographic traits. For example, cooperative breeding tends to be associated with ‘‘slow’’ life histories, for example, high longevity, that might be expected to produce genetically‐structured populations (Arnold and Owens, 1998; Brown, 1987). We are exploring theoretically and empirically the roles of dispersal, life history, and demographic traits in generating fine‐scale genetic structure in bird populations with a view to answering the question of whether there are fundamental differences between cooperative and noncooperative species in this respect.

KIN SELECTION AND COOPERATION IN LONG‐TAILED TITS

389

VIII. SUMMARY Despite extensive research on cooperative breeding systems over 30 years, the role of kin selection in the evolution of animal societies is still the subject of considerable debate. Likewise, the ecological factors that promote cooperation and differentiate cooperative from noncooperative species are still not well understood. The aim of this study is to use a case study of the long‐tailed tit Aegithalos caudatus to address these broad questions in social evolution. The long‐tailed tit has an atypical cooperative breeding system that is unusually well suited to investigating the fitness consequences of helping and especially the role of kin selection in the evolution of cooperative behavior. A combination of long‐term systematic field observations and experimental manipulations in the field have been used to demonstrate that: (1) helpers exhibit kin discrimination in their helping behavior, preferentially directing their care toward close kin; (2) long‐tailed tits use vocalizations to recognize their relatives; and (3) these recognition cues are learned while still in the nest, when association provides reliable information on relatedness. Selection for an effective means of recognition would be expected in this system because the evidence for indirect, kin‐selected fitness benefits from cooperation is strong, while the evidence for direct fitness benefits that might be acquired via less discriminating helping is weak. Furthermore, inclusive fitness calculations using LRS data indicate that helping behavior by failed breeders provides a substantial source of indirect fitness for individuals that would otherwise achieve zero fitness in their lifetime. Finally, we show that cooperative behavior in long‐tailed tits is consistent with the ecological constraints hypothesis. The reproductive success of breeders declines sharply through the breeding season and failed breeders switch from a tactic of independent breeding to one of helping relatives to reproduce when the expected payoff from independent breeding drops below that expected from helping.

Acknowledgments Many people have collaborated and given invaluable help with the long‐tailed tit project over the past 12 years. Andrew MacColl in particular took the project in new, unexpected directions that proved particularly rewarding; however, all who have assisted in the field, in the lab, and in providing analytical solutions have made important contributions. In roughly chronological order, our thanks go to Andy Russell, Martin Fowlie, Douglas Ross, Andy McGowan, Nic Chaline, Carl Anderson, Rick Woodburn, Nicky Green, Dan Richardson, Andy Bamford, Sarah Collins, Rob Briers, Matt Wood, Michelle Simeoni, Beth Woodward, Ed Sykes, Kate Howe, and Richard Gill. We thank Dave Hazard, Geoff Mawson, and members of the Sorby‐ Breck Ringing Group for their help and support, and Doncaster City Council, Sheffield City Council, Hallam Golf Club, and Yorkshire Water for allowing us to work on their land. Genetic analyses were all performed in the Sheffield Molecular Genetics Facility under the guidance of

390

BEN J. HATCHWELL AND STUART P. SHARP

Terry Burke and Deborah Dawson. We also thank Jane Brockman and an anonymous reviewer for their helpful comments on the chapter. We are extremely grateful for the financial support provided by the Natural Environment Research Council, Nuffield Foundation, Association for the Study of Animal Behaviour, and University of Sheffield.

REFERENCES Arnold, K. E., and Owens, I. P. F. (1998). Cooperative breeding in birds: A comparative test of the life history hypothesis. Proc. R. Soc. London Ser. B 265, 739–745. Baptista, L. F. (1996). Nature and its nurturing in avian vocal development. In ‘‘Ecology and Evolution of Acoustic Communication in Birds’’ (D. E. Kroodsma and E. H. Miller, Eds.), pp. 39–60. Cornell University Press, Ithaca. Both, C., Visser, M. E., and Verboven, N. (1999). Density‐dependent recruitment rates in great tits: The importance of being heavier. Proc. R. Soc. London Ser. B 266, 465–469. Brown, J. L. (1974). Alternate routes to sociality in jays with a theory for the evolution of altruism and communal breeding. Am. Zool. 14, 63–80. Brown, J. L. (1987). ‘‘Helping and Communal Breeding in Birds.’’ Princeton University Press, Princeton. Brown, J. L., Brown, E. R., Brown, S. D., and Dow, D. D. (1982). Helpers: Effects of experimental removal on reproductive success. Science 215, 421–422. Clutton‐Brock, T. H. (1998). Reproductive skew, concessions and limited control. Trends Ecol. Evol. 13, 288–292. Clutton‐Brock, T. H. (2002). Breeding together: Kin selection and mutualism in cooperative vertebrates. Science 296, 69–72. Clutton‐Brock, T. H., Brotherton, P. N. M., O’Riain, M. J., Griffin, A. S., Gaynor, D., Sharpe, L., Kansky, R., Manser, M. B., and McIlrath, G. M. (2000). Individual contributions to babysitting in a cooperative mongoose, Suricata suricatta. Proc. R. Soc. London Ser. B 267, 301–305. Clutton‐Brock, T. H., Brotherton, P. N. M., O’Riain, M. J., Griffin, A. S., Gaynor, D., Kansky, R., Sharpe, L., and McIlrath, G. M. (2001). Contributions to cooperative rearing in meerkates. Anim. Behav. 61, 705–710. Cockburn, A. (1996). Why do so many Australian birds cooperate: Social evolution in the Corvida? In ‘‘Frontiers in Population Ecology’’ (R. B. Floyd, A. W. Sheppard, and P. J. De Barro, Eds.), pp. 451–472. CSIRO Publishing, Melbourne. Cockburn, A. (1998). Evolution of helping behaviour in cooperatively breeding birds. Annu. Rev. Ecol. Syst. 29, 141–177. Cockburn, A. (2004). Mating systems and sexual conflict. In ‘‘Ecology and Evolution of Cooperative Breeding in Birds’’ (W. D. Koenig and J. L. Dickinson, Eds.), pp. 81–101. Cambridge University Press, Cambridge. Cramp, S., and Perrins, C. M. (1993). ‘‘The Birds of the Western Palearctic. Volume VII.’’ Oxford University Press, Oxford. Creel, S. (1990). How to measure inclusive fitness. Proc. R. Soc. London Ser. B 241, 229–231. Crick, H. Q. P. (1992). Load‐lightening in cooperatively breeding birds and the costs of reproduction. Ibis 134, 56–61. Crick, H. Q. P., and Sparks, T. H. (1999). Climate change related to egg‐laying trends. Nature 399, 423–424. Cronin, H. (1991). ‘‘The Ant and the Peacock.’’ Cambridge University Press, Cambridge. Curry, R. L. (1988). Influence of kinship on helping behaviour in Galapagos mockingbirds. Behav. Ecol. Sociobiol. 22, 141–152.

KIN SELECTION AND COOPERATION IN LONG‐TAILED TITS

391

Curry, R. L., and Grant, P. R. (1990). Galapagos mockingbirds: Territorial cooperative breeding in a climatically variable environment. In ‘‘Cooperative Breeding in Birds’’ (P. B. Stacey and W. D. Koenig, Eds.), pp. 289–331. Cambridge University Press, Cambridge. Darwin, C. R. (1859). ‘‘On the Origin of Species by means of Natural Selection or the Preservation of Favoured Races in the Struggle for Life.’’ John Murray, London. Davies, N. B., and Hatchwell, B. J. (1992). The value of male parental care and its influence on reproductive allocation by male and female dunnocks. J. Anim. Ecol. 61, 259–272. Dickinson, J. L., and Hatchwell, B. J. (2004). The fitness consequences of helping. In ‘‘Cooperative Breeding in Birds’’ (W. D. Koenig and J. L. Dickinson, Eds.), pp. 48–66. Cambridge University Press, Cambridge. Dickinson, J. L., Koenig, W. D., and Pitelka, F. A. (1996). Fitness consequences of helping behavior in the western bluebird. Behav. Ecol. 7, 168–177. Dunn, P. O., Cockburn, A., and Mulder, R. A. (1995). Fairy‐wren helpers often care for young to which they are unrelated. Proc. R. Soc. London Ser. B 259, 339–343. Ekman, J., Bylin, A., and Tegelstrom, H. (2000). Parental nepotism enhances survival of retained offspring in the Siberian jay. Behav. Ecol. 11, 416–420. Ekman, J., Dickinson, J. L., Hatchwell, B. J., and Griesser, M. (2004). Delayed dispersal. In ‘‘Ecology and Evolution of Cooperative Breeding in Birds’’ (W. D. Koenig and J. L. Dickinson, Eds.), pp. 35–47. Cambridge University Press, Cambridge. Emlen, S. T. (1982). The evolution of helping. I. An ecological constraints model. Am. Nat. 119, 29–39. Emlen, S. T. (1991). Evolution of cooperative breeding in birds and mammals. In ‘‘Behavioural Ecology: An Evolutionary Approach’’ (J. R. Krebs and N. B. Davies, Eds.), pp. 301–337. Blackwell Scientific Publications, Oxford. Emlen, S. T. (1994). Benefits, constraints and the evolution of the family. Trends Ecol. Evol. 9, 282–285. Emlen, S. T. (1997). Predicting family dynamics in social vertebrates. In ‘‘Behavioural Ecology: An Evolutionary Approach’’ (J. R. Krebs and N. B. Davies, Eds.), pp. 228–253. Blackwell Scientific Publications, Oxford. Emlen, S. T., and Wrege, P. H. (1988). The role of kinship in helping decisions among white‐ fronted bee‐eaters. Behav. Ecol. Sociobiol. 23, 305–315. Falls, J. B. (1982). Individual recognition by sound in birds. In ‘‘Acoustic Communication in Birds. Volume 2: Song Learning and its Consequences’’ (D. E. Kroodsma and E. H. Miller, Eds.), pp. 237–278. Academic Press, New York. Fletcher, D. J. C., and Michener, C. D. (1987). ‘‘Kin Recognition in Animals.’’ John Wiley and Sons, Chichester. Garnett, M. C. (1981). Body size: Its heritability and influence on juvenile survival among great tits, Parus major. Ibis 123, 31–41. Gaston, A. J. (1973). The ecology and behaviour of the long‐tailed tit. Ibis 115, 330–351. Gaston, A. J. (1978). The evolution of group territorial behavior and cooperative breeding. Am. Nat. 112, 1091–1100. Glen, N. W. (1985). The cooperative breeding behaviour of the long‐tailed tit (Aegithalos caudatus) Unpublished PhD thesis, University of Oxford, Oxford. Glen, N. W., and Perrins, C. M. (1988). Cooperative breeding by long‐tailed tits. Br. Birds 81, 630–641. Griesser, M., and Ekman, J. (2004). Nepotistic alarm calling in the Siberian jay Perisoreus infaustus. Anim. Behav. 67, 933–939. Griffin, A. S., and West, S. A. (2003). Kin discrimination and the benfit of helping in cooperatively breeding vertebrates. Science 302, 634–636.

392

BEN J. HATCHWELL AND STUART P. SHARP

Halpin, Z. T. (1991). Kin recognition cues of vertebrates. In ‘‘Kin Recognition’’ (P. G. Hepper, Ed.), pp. 220–258. Cambridge University Press, Cambridge. Hamilton, W. D. (1963). The evolution of altruistic behavior. Am. Nat. 97, 354–356. Hamilton, W. D. (1964a). The genetical evolution of social behaviour. I. J. Theor. Biol. 7, 1–16. Hamilton, W. D. (1964b). The genetical evolution of social behaviour. II. J. Theor. Biol. 7, 17–52. Hansell, M. H. (1993). The long‐tailed tit’s nest of many parts. BTO News 186, 20. Hansell, M. H. (2000). ‘‘Bird Nests and Construction Behaviour.’’ Cambridge University Press, Cambridge. Harrap, S., and Quinn, D. (1996). ‘‘Tits, Nuthatches and Treecreepers.’’ Christopher Helm, London. Hatchwell, B. J. (1999). Investment strategies of breeders in avian cooperative breeding systems. Am. Nat. 154, 205–219. Hatchwell, B. J., and Komdeur, J. (2000). Ecological constraints, life history traits and the evolution of cooperative breeding. Anim. Behav. 59, 1079–1086. Hatchwell, B. J., and Russell, A. F. (1996). Provisioning rules in cooperatively breeding long‐ tailed tits Aegithalos caudatus: An experimental study. Proc. R. Soc. London Ser. B 263, 83–88. Hatchwell, B. J., Russell, A. F., Fowlie, M. K., and Ross, D. J. (1999). Reproductive success and nest‐site selection in a cooperative breeder: Effect of experience and a direct benefit of helping. Auk 116, 355–363. Hatchwell, B. J., Russell, A. F., Ross, D. J., and Fowlie, M. K. (2000). Divorce in cooperatively breeding long‐tailed tits: A consequence of inbreeding avoidance? Proc. R. Soc. London Ser. B 267, 813–819. Hatchwell, B. J., Anderson, C., Ross, D. J., Fowlie, M. K., and Blackwell, P. G. (2001a). Social organisation in cooperatively breeding long‐tailed tits: Kinship and spatial dynamics. J. Anim. Ecol. 70, 820–830. Hatchwell, B. J., Ross, D. J., Fowlie, M. K., and McGowan, A. (2001b). Kin discrimination in cooperatively breeding long‐tailed tits. Proc. R. Soc. London Ser. B 268, 885–890. Hatchwell, B. J., Ross, D. J., Chaline, N., Fowlie, M. K., and Burke, T. (2002). Parentage in the cooperative breeding system of long‐tailed tits, Aegithalos caudatus. Anim. Behav. 64, 55–63. Hatchwell, B. J., Russell, A. F., MacColl, A. D. C., Ross, D. J., Fowlie, M. K., and McGowan, A. (2004). Helpers increase long‐term but not short‐term productivity in cooperatively breeding long‐tailed tits. Behav. Ecol. 15, 1–10. Heinsohn, R. G. (2004). Parental care, load‐lightening and costs. In ‘‘Ecology and Evolution of Cooperative Breeding in Birds’’ (W. D. Koenig and J. L. Dickinson, Eds.), pp. 67–80. Cambridge University Press, Cambridge. Hepper, P. G. (1991). ‘‘Kin Recognition.’’ Cambridge University Press, Cambridge. Hughes, M., Nowicki, S., and Lohr, B. (1998). Call learning in black‐capped chickadees (Parus atricapillus): The role of experience in the development of ‘chick‐a‐dee’ calls. Ethology 104, 232–249. Keane, B., Creel, S. R., and Waser, P. M. (1996). No evidence of inbreeding avoidance or inbreeding depression in a social carnivore. Behav. Ecol. 7, 480–489. Keller, L., and Reeve, H. K. (1994). Partitioning of reproduction in animal societies. Trends Ecol. Evol. 9, 98–102. Ketterson, E. D., and Nolan, V. (1994). Male parental behaviour in birds. Annu. Rev. Ecol. Syst. 25, 601–628. Koenig, W. D., and Dickinson, J. L. (2004). ‘‘Ecology and Evolution of Cooperative Breeding in Birds.’’ Cambridge University Press, Cambridge. Koenig, W. D., and Pitelka, F. A. (1981). Ecological factors and kin selection in the evolution of cooperative breeding in birds. In ‘‘Natural Selection and Social Behaviour’’ (R. D. Alexander and D. W. Tinkle, Eds.), pp. 261–280. Chiron Press, New York.

KIN SELECTION AND COOPERATION IN LONG‐TAILED TITS

393

Koenig, W. D., Pitelka, F. A., Carmen, W. J., Mumme, R. L., and Stanback, M. T. (1992). The evolution of delayed dispersal in cooperative breeders. Quart. Rev. Biol. 67, 111–150. Kokko, H., Johnstone, R. A., and Clutton‐Brock, T. H. (2001). The evolution of cooperative breeding through group augmentation. Proc. R. Soc. London Ser. B 268, 187–196. Kokko, H., Johnstone, R. A., and Wright, J. (2002). The evolution of parental and alloparental effort in cooperatively breeding groups: When should helpers pay to stay? Behav. Ecol. 13, 291–300. Komdeur, J. (1994). The effect of kinship on helping in the cooperative breeding Seychelles warbler (Acrocephalus sechellensis). Proc. R. Soc. London Ser. B 256, 47–52. Komdeur, J. (1996). Influence of helping and breeding experience on reproductive performance in the Seychelles warbler: A translocation experiment. Behav. Ecol. 7, 326–333. Komdeur, J., and Hatchwell, B. J. (1999). Kin recognition: Function and mechanism in avian societies. Trends Ecol. Evol. 14, 237–241. Krause, J., and Ruxton, G. D. (2002). ‘‘Living in Groups.’’ Oxford University Press, Oxford. Lack, D., and Lack, E. (1958). The nesting of the long‐tailed tit. Bird Study 5, 1–19. Lahdenpera¨, M., Lummaa, V., Helle, S., Tremblay, M., and Russell, A. F. (2004). Fitness benefits of prolonged post‐reproductive lifespan in women. Nature 428, 178–181. Langen, T. A. (2000). Prolonged offspring dependence and cooperative breeding in birds. Behav. Ecol. 11, 367–377. Legge, S. (2000). Helper contributions in the cooperatively breeding laughing kookaburra: Feeding young is no laughing matter. Anim. Behav. 59, 1009–1018. Lessells, C. M. (1990). Helping at the nest in European bee‐eaters: Who helps and why? In ‘‘Population Biology of Passerine Birds, an Integrated Approach’’ (J. Blondel, A. Gosler, J. D. Lebreton, and R. McCleery, Eds.), pp. 357–368. Springer‐Verlag, Berlin. Ligon, J. D., and Burt, D. B. (2004). Evolutionary origins. In ‘‘Cooperative Breeding in Birds’’ (W. D. Koenig and J. L. Dickinson, Eds.), pp. 5–34. Cambridge University Press, Cambridge. Linden, M., Gustafsson, L., and Part, T. (1992). Selection on fledging mass in the collared flycatcher and the great tit. Ecology 73, 336–343. MacColl, A. D. C., and Hatchwell, B. J. (2002). Temporal variation in fitness pay‐offs promotes cooperative breeding in long‐tailed tits Aegithalos caudatus. Am. Nat. 160, 186–194. MacColl, A. D. C., and Hatchwell, B. J. (2003a). Sharing of caring: Nestling provisioning behaviour of long‐tailed tit (Aegithalos caudatus) parents and helpers. Anim. Behav. 66, 955–964. MacColl, A. D. C., and Hatchwell, B. J. (2003b). Heritability of parental care in a passerine bird. Evolution 57, 2191–2195. MacColl, A. D. C., and Hatchwell, B. J. (2004). Determinants of lifetime fitness in a cooperative breeder, the long‐tailed tit Aegithalos caudatus. J. Anim. Ecol. 73, 1137–1148. Magrath, R. D. (1991). Nestling weight and juvenile survival in the blackbird Turdus merula. J. Anim. Ecol. 60, 335–351. Magrath, R. D., and Whittingham, L. A. (1997). Subordinate males are more likely to help if unrelated to the breeding female in cooperatively breeding white‐browed scrubwrens. Behav. Ecol. Sociobiol. 41, 185–192. Magrath, R. D., Johnstone, R. A., and Heinsohn, R. G. (2004). Reproductive skew. In ‘‘Ecology and Evolution of Cooperative Breeding in Birds’’ (W. D. Koenig and J. L. Dickinson, Eds.), pp. 157–176. Cambridge University Press, Cambridge. Marshall, T. C., Slate, J., Kruuk, L. E. B., and Pemberton, J. M. (1998). Statistical confidence for likelihood‐based paternity inference in natural populations. Mol. Ecol. 7, 639–655. Maynard Smith, J. (1964). Group selection and kin selection. Nature 201, 1145–1147. McGowan, A. (2002). Fitness benefits of cooperative breeding in long‐tailed tits (Aegithalos caudatus) Unpublished PhD thesis, University of Sheffield, Sheffield.

394

BEN J. HATCHWELL AND STUART P. SHARP

McGowan, A., Woodburn, R. J. W., and Hatchwell, B. J. (2003). The effect of helping behaviour on the survival of juvenile and adult long‐tailed tits (Aegithalos caudatus). J. Anim. Ecol. 72, 491–499. McGowan, A., Sharp, S. P., and Hatchwell, B. J. (2004). The structure and function of nests of long‐tailed tits Aegithalos caudatus. Funct. Ecol. 18, 578–583. McGowan, A., Sharp, S. P., Simeoni, M., and Hatchwell, B. J. (2006). Competition for position in the communal roosts of long‐tailed tits. Anim. Behav. (in press). McGraw, J. B., and Caswell, H. (1996). Estimation of individual fitness from life‐history data. Am. Nat. 147, 47–64. Mulder, R. A., and Langmore, N. E. (1993). Dominant males punish helpers for temporary defection in superb fairy‐wrens. Anim. Behav. 45, 830–833. Newton, I. (1989). ‘‘Lifetime Reproduction in Birds.’’ Academic Press, London. Oli, M. K. (2003). Hamilton goes empirical: Estimation of inclusive fitness from life‐history data. Proc. R. Soc. London Ser. B 270, 307–311. Payne, R. B., Payne, L. L., and Rowley, I. (1988). Kin and social relationships in splendid fairy‐ wrens—recognition by song in a cooperative bird. Anim. Behav. 36, 1341–1351. Perrins, C. M. (1965). Population fluctuations and clutch‐size in the great tit, Parus major L. J. Anim. Ecol. 34, 601–647. Perrins, C. M. (2004). ‘‘The New Encyclopaedia of Birds.’’ Oxford University Press, Oxford. Price, J. J. (1998). Family‐ and sex‐specific vocal traditions in a cooperatively breeding songbird. Proc. R. Soc. London Ser. B 265, 497–502. Pusey, A. E., and Wolf, M. (1996). Inbreeding avoidance in animals. Trends Ecol. Evol. 11, 201–206. Queller, D. C. (1992). Does population viscosity promote kin selection? Trends Ecol. Evol. 7, 322–324. Ralls, K., Harvey, P. H., and Lyles, A. M. (1986). Inbreeding in natural populations of birds and mammals. In ‘‘Conservation Biology: The Science of Scarcity and Diversity’’ (M. E. Soule´, Ed.), pp. 35–56. Sinauer Associates, Sunderland, MA. Reeve, H. K., Westneat, D. F., Noon, W. A., Sherman, P. W., and Aquadro, C. F. (1990). DNA ‘‘fingerprinting’’ reveals high levels of inbreeding in colonies of the eusocial naked mole‐ rat. Proc. Natl. Acad. Sci. USA 87, 2496–2500. Reeve, H. K., Emlen, S. T., and Keller, L. (1998). Reproductive sharing in animal societies: Reproductive incentives or incomplete control by dominant breeders? Behav. Ecol. 9, 267–278. Reyer, H.‐U. (1990). Pied kingfishers: Ecological causes and reproductive consequences of cooperative breeding. In ‘‘Cooperative Breeding in Birds’’ (P. B. Stacey and W. D. Koenig, Eds.), pp. 529–557. Cambridge University Press, Cambridge. Richardson, D. S., Jury, F. L., Blaakmeer, K., Komdeur, J., and Burke, T. (2001). Parentage assignment and extra‐group paternity in a cooperative breeder: The Seychelles warbler (Acrocephalus sechellensis). Mol. Ecol. 10, 2263–2273. Richardson, D. S., Burke, T., and Komdeur, J. (2002). Direct benefits and the evolution of female‐biased cooperative breeding in Seychelles warblers. Evolution 56, 2313–2321. Riehm, V. H. (1970). Okologie und verhalten der schwanzmeise (Aegithalos caudatus L.). Zool. Jb. Syst. 97, 338–400. Russell, A. F. (1999). Ecological constraints and the cooperative breeding system of the long‐tailed tit (Aegithalos caudatus) University of Sheffield, SheffieldUnpublished PhD thesis,. Russell, A. F. (2001). Dispersal costs set the scene for helping in an atypical avian cooperative breeder. Proc. R. Soc. London Ser. B 268, 95–99. Russell, A. F., and Hatchwell, B. J. (2001). Experimental evidence for kin‐biased helping in a cooperatively breeding vertebrate. Proc. R. Soc. London Ser. B 268, 2169–2174.

KIN SELECTION AND COOPERATION IN LONG‐TAILED TITS

395

Selander, R. K. (1964). Speciation in wrens of the genus Campylorhynchus. Univ. Calif. Publ. Zool. 74, 1–224. Sharp, S. P. (2003). Kin recognition in the cooperative breeding system of the long‐tailed tit, Aegithalos caudatus. Unpublished PhD thesis, University of Sheffield, Sheffield. Sharp, S. P., and Hatchwell, B. J. (2005). Individuality in the contact calls of cooperatively breeding long‐tailed tits. Behaviour 142, 1559–1575. Sharp, S. P., and Hatchwell, B. J. (in press). Development of family‐specific contact calls in the long‐tailed tit Aegithalos caudatus. Ibis. Sharp, S. P., McGowan, A., Wood, M. J., and Hatchwell, B. J. (2005). Learned kin recognition cues in a social bird. Nature 434, 1127–1130. Sherley, G. H. (1990). Cooperative breeding in riflemen (Acanthisitta chloris): Benefits to parents, offspring and helpers. Behaviour 112, 1–22. Sherman, P. W., Reeve, H. K., and Pfennig, D. W. (1997). Recognition systems. In ‘‘Behavioural Ecology: An Evolutionary Approach’’ (J. R. Krebs and N. B. Davies, Eds.), pp. 69–96. Blackwell Science, Oxford. Siegel, S., and Castellan, N. J. (1988). ‘‘Nonparametric Statistics for the Behavioural Sciences.’’ McGraw‐Hill, Boston. Skutch, A. F. (1961). Helpers among birds. Condor 63, 198–226. Sloane, S. A. (1996). Incidence and origins of supernumeraries at bushtit (Psaltriparus minimus) nests. Auk 113, 757–770. Stacey, P. B., and Koenig, W. D. (1990). ‘‘Cooperative Breeding in Birds.’’ Cambridge University Press, Cambridge. Stacey, P. B., and Ligon, J. D. (1987). Territory quality and dispersal options in the acorn woodpecker, and a challenge to the habitat saturation model of cooperative breeding. Am. Nat. 130, 654–676. Stacey, P. B., and Ligon, J. D. (1991). The benefits of philopatry hypothesis for the evolution of cooperative breeding: Variance in territory quality and group size effects. Am. Nat. 137, 831–846. Stevenson, I. R., and Bryant, D. M. (2000). Avian phenology: Climate change and constraints on breeding. Nature 406, 366–367. Thompson, E. A., and Meagher, T. R. (1987). Parental and sib likelihoods in genealogy reconstruction. Biometrics 43, 585–600. van Noordwijk, A. J., McCleery, R. H., and Perrins, C. M. (1995). Selection for the timing of great tit breeding in relation to caterpillar growth and temperature. J. Anim. Ecol. 64, 451–458. Vehrencamp, S. E. (2000). Evolutionary routes to joint‐female nesting in birds. Behav. Ecol. 11, 334–344. Vehrencamp, S. L. (1983). A model for the evolution of despotic versus egalitarian societies. Anim. Behav. 31, 667–682. Verboven, N., and Visser, M. E. (1998). Seasonal variation in local recruitment of great tits: The importance of being early. Oikos 81, 511–524. Verhulst, S., and Tinbergen, J. M. (1991). Experimental evidence for a causal relationship between timing and success of reproduction in the great tit Parus major. J. Anim. Ecol. 60, 269–282. Waldman, B. (1987). Mechanisms of kin recognition. J. Theor. Biol. 128, 159–185. West, S. A., Pen, I., and Griffin, A. S. (2002). Cooperation and competition between relatives. Science 296, 72–75. White, G. C., and Burnham, K. P. (1999). Program MARK: Survival estimation from populations of marked animals. Bird Study 46(Suppl.), 120–139. Wiley, R. H., and Rabenold, K. N. (1984). The evolution of cooperative breeding by delayed reciprocity and queuing for favorable social positions. Evolution 38, 609–621.

This page intentionally left blank

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 36

How Do Little Blue Penguins ‘‘Validate’’ Information Contained in Their Agonistic Displays? Joseph R. Waas department of biological sciences university of waikato, hamilton new zealand

I. INTRODUCTION Animals improve their ability to defend or secure resources by using agonistic displays to manage the activities of those around them during disputes. However, individuals that perceive displays should only react to signals which contain properties that minimize or eliminate their chances of being duped. If bluff could not be ruled out on average, signalers would be expected to abuse existing display conventions to their own advantage and, in response, perceivers would eventually be expected to ignore the displays; such a signaling system could not be evolutionarily stable (Maynard Smith, 1982; Maynard Smith and Parker, 1976). Considerable research has been directed at identifying how animals corroborate or ‘‘validate’’ the information contained in their displays, so perceivers remain attentive and respond accordingly (reviewed in Bradbury and Vehrencamp, 1998; see Fitch and Hauser, 2002 for a review focusing on acoustic signals). Four types of validations relevant to the present review follow. 1. In many species, the properties of agonistic signals are inextricably linked to physical characteristics that influence fighting ability or resource holding power (RHP; Parker, 1974); later, the term resource holding potential was favored (Maynard Smith and Parker, 1976). For example, between‐individual variation in the amplitude of vocalizations produced by male elephant seals (Mirounga sp.) reflects differences in body size (note, however, that the effect is stronger for interspecific comparisons than intraspecific comparisons; Sanvito and Galimberti, 2003)—larger animals can produce louder calls (a similar relationship occurs in humans; Titze, 1994). Bluff is precluded because small individuals are physically incapable 0065-3454/06 $35.00 DOI: 10.1016/S0065-3454(06)36009-3

397

Copyright 2006, Elsevier Inc. All rights reserved.

398

JOSEPH R. WAAS

of producing a signal that misrepresents their size. While this type of signal can provide a good immediate account of an animal’s physical attributes, important underlying qualities like stamina, battle weariness, or nutritional state, may remain hidden (Payne and Pagel, 1996). 2. Contests of endurance or wars of attrition (Bishop and Cannings, 1978; Hammerstein and Parker, 1982; Maynard Smith, 1982), which often involve the repetition of a given display (Payne and Pagel, 1997), can provide a better indication of an animal’s underlying physical state. For example, side‐blotched lizards (Uta stansburiana) engage in repeated ‘‘push‐up’’ displays during agonistic encounters with rivals. The number of repetitions or amount of time animals engage in the display correlates well with endurance (Brandt, 2003). Further, when a treadmill is used to reduce an animal’s endurance experimentally, there is a corresponding decline in duration of the push‐up display. Bluff is precluded because the weak or weary are not physically capable of falsifying the signal. When this information is combined with that obtained by immediate assessments of physical attributes like size or weaponry (discussed earlier), an accurate ‘‘extended account’’ of RHP can be obtained. 3. However, ‘‘size is not everything’’ when it comes to winning resources, nor is being in top physical condition or possessing great stamina. An animal’s motivational state can be crucial. A well‐fed animal with high RHP may well give up a morsel of food to a smaller individual that is motivated by starvation to secure the resource at any cost. But how can such an ethereal quality as motivation be conveyed honestly? Animals may signal motivation by using displays that advertise risks they are willing to take while contesting resources (Zahavi, 1977). For example, the threat displays of American goldfinches (Carduelis tristis) differ in terms of how vulnerable they make the signaler to opponent retaliation (Popp, 1987a). A signaler using high‐risk displays proves that it is highly motivated to secure the resource—bluff is precluded because the animal must take the risk to express the display. In accordance, Popp found that high‐risk displays were more effective at causing rivals to retreat than lower risk versions. Thus, risk‐based displays provide opponents with motivational information that can be considered in concert with information on RHP. 4. In social systems where opponents interact regularly, store memories of encounters, and are capable of individual recognition, signals of intent can also be evolutionarily stable (van Rhijn, 1980; van Rhijn and Vodegel, 1980). Rivals may learn to associate different displays in a signaler’s repertoire with different consequences, as long as the signaler consistently pairs each display with a given consequence during previous encounters.

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

399

For example, collared lizards (Crotaphytus collaris) are capable of individual recognition (Husak and Fox, 2003) and use a graded display to signal how likely they are to engage in aggressive acts at territory boundaries; a neighbor performing the low‐cost lateral display at a high rate is more likely to attack than one performing at a low rate (Husak, 2004). Although bluff is never totally precluded, each instance of bluff weakens the association between the display and its predicted consequence for that signaler. As a result, bluff must be contained for the signaler to gain the benefits of using the display. Signals of intent are interesting, in that they provide information not only on intent but also on motivation and RHP, as derived from previous encounters. The purpose of this chapter is to review research designed to determine how little blue penguins (Eudyptula minor) validate the information contained in their agonistic displays. The research program involved investigating all four validation methods identified earlier: (1) signals advertising resource holding potential, (2) wars of attrition, (3) displays that advertise risks, and, finally, (4) signals of intent. The work suggests that little blue penguins use all four of these methods to corroborate the information contained in their displays, so perceivers remain attentive. I then outline how animals invest energy to ensure signal efficacy, and discuss the possibility that each display in an animal’s repertoire may contain ‘‘overlapping validations,’’ that is, perceivers may be presented with a variety of validations simultaneously when a given display is performed. The amount of information available to the perceiver will be dependent on the ‘‘depth’’ of its social relationship with the signaler. Before discussing the different signaling strategies, I provide an account of little blue penguin natural history to provide the setting in which the displays are actually used.

II. NATURAL HISTORY Penguins are superb subjects for the study of agonistic communication. They are highly territorial during the breeding period and use a wide range of obvious offensive and defensive displays during disputes with neighbors and intruders. Both their graded (i.e., signals that can be performed at a range of intensities; see Fig. 1 for an example of their graded vocalizations) and discrete signals typically involve a simple combination of visual and acoustic components (Waas, 1988, 1990a,b). The birds are also site faithful, forming historically stable communities of regularly interacting individuals; as a result, the same groups of individuals can be observed

400

JOSEPH R. WAAS

Fig. 1. Graded vocalizations of little blue penguins, from low‐intensity growls to low, medium, and finally full braying (adapted from Waas, 1990b). The growl–bray vocalizations accompany both sexual and agonistic displays.

interacting on a regular basis. Furthermore, penguins are tolerant of stationary observers positioned a few meters from them, allowing one to easily obtain information on signaling interactions from undisturbed subjects. Here I provide a brief account of little blue penguin natural history and major events within their breeding cycle (see Waas, 1990b for further detail, and references for the specific ecological details). Little blue penguins are the smallest of all living penguin species. The species is the sole member of the genus Eudyptula and populations around New Zealand and Australia differ widely in terms of behavior, ecology, morphology, and genetics (Banks et al., 2002; Waas 1990a,b). Most of the research reviewed here has been conducted on the white‐flippered morph inhabiting New Zealand’s Banks Peninsula and Motunau Island (Miyazaki and Waas, 2002; Waas, 1988, 1990a,b, 1991a,b), the northern morph occupying the country’s upper North Island (Banks et al., 2002; Miyazaki and Waas, 2003a,b;

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

401

Nakagawa et al., 2001), or the southern South Island morph (Miyazaki and Waas, 2003c). The subjects are the only penguin species to restrict their movements and most social interactions on land to the hours of darkness, with their loud calls making them an obvious and familiar component of Australasian coastlines. Little blue penguins are monogamous with long‐term pair bonds and can live well into their twenties (Challies, C. N., personal communication). Pairs that successfully rear young tend to remain together for several years, but ‘‘divorces’’ can occur (mainly when the pair fails to breed). Early in the breeding season, birds will sometimes ‘‘keep company’’ and copulate with members of the opposite sex prior to the arrival of their partners in colonies. The birds breed under cover (e.g., in caves, burrows, under dense vegetation) in loosely colonial aggregations, with pairs returning to the same area to breed each year. Little blue penguins first attempt to breed as 2‐ to 3‐year olds. There is little sexual dimorphism, although males tend to be slightly larger with stouter bills than females. Unmated males start advertising for females from late April (Austral fall; but this is highly population dependent as are dates for other events in the cycle listed later) by performing ‘‘solo calls’’ (Waas, 1988; the ‘‘full bray’’ vocal component of this display is shown in Fig. 1) from positions beneath rocks, tree roots, or thick brush along the shorelines below established breeding colonies. In cave colonies, the males may call from central nonbreeding areas away from established nest sites (small clubs of three to six calling males occurred regularly, and aggressive interactions between these males were common; Waas, 1990a). Attracted females keep company and perform mutual sexual displays with selected males before the bonding pair begins looking for suitable nesting sites (e.g., unused burrows or nesting hollows along cave walls). Once pairs settle on a nest site, both birds initiate nest building; the male typically forages for the nest material, while the female positions the material within the nest. Mutual sexual displays, copulations, and territorial interactions increase steadily as birds move toward the egg‐laying period. Two‐egg clutches are initiated from August or September (Austral spring) and are incubated for 5 weeks with pair members swapping incubation duties every few days (relieved partners return to sea to replenish fat reserves). The number of aggressive interactions declines during this phase in comparison to what is observed during courtship. Parents guard hatched chicks for 10–25 days, with a nightly change in guard duties; during the postguard stage, adults only make brief feeding visits to the chicks during the night, with the visits becoming less frequent toward the fledging period (50–55 days). After chicks fledge (December–February), the breeders no longer return to colonies, spending most of their time at sea replenishing

402

JOSEPH R. WAAS

depleted fat reserves. Finally, the birds return for a short molting period of 2–3 weeks before spending further time readying themselves at sea for the next breeding season.

III. VALIDATIONS FOR INFORMATION CONTAINED

IN

AGONISTIC DISPLAYS

Here I review four types of signals that little blue penguins use to contest or defend resources. For each signal type, a brief introduction to the underlying theory is provided before I outline research completed by our group to test predictions developed from each theory; I end each of the four subsections with conclusions derived from the research findings.

A. SIGNALS THAT ADVERTISE RHP RHP‐based signals simply advertise attributes (e.g., weaponry, musculature) that impact on an animal’s ability to secure or defend resources. Bluff is precluded by an inextricable link between the highlighted characteristic(s) and fighting ability. For example, body size in many animals is strongly and positively correlated with attributes like experience and strength, both of which impact an animal’s ability to contest resources (reviewed by Archer, 1988). Body size variation in birds may be revealed reliably by vocalizations because the size of sound‐producing structures and the volume of the cavities required to hold air (lungs and air sac) correlate with body size (King and Molony, 1971; Ryan and Brenowitz, 1985; Wallschlager, 1980); however, more information on relationships between the properties of vocalizations and bird size (especially at an intraspecific level) is required. Most little blue penguin vocalizations are based on low‐pitched growled exhalations and inhalation–exhalation brays (Waas, 1988; Fig. 1). The brayed solo calls are loud, high‐intensity vocalizations used by males to advertise for females and announce ownership of the nesting site (Waas, 1988). Growls are quiet, low‐ intensity threats used not only during low‐risk offensive displays but also when birds lock bills and engage in overt aggression (Fig. 1); they are the most common vocalization used by the birds. Masamine Miyazaki and I assessed whether little blue penguins might convey information on body size using these calls (Miyazaki and Waas, 2003a,c). For both types of calls, we predicted that larger penguins would produce: (1) low‐pitched calls as the result of having a more expansive vocal apparatus and (2) longer inhalation and/or exhalation phrases as the result of being capable of drawing more air in and out of the body cavity.

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

403

1. Correlations Between Body Size and the Acoustic Features of Solo Calls a. Method We addressed this issue by recording the solo calls of 12 males in a single population of southern little blue penguins (Oamaru, South Island, New Zealand; see Miyazaki and Waas, 2003c for full details). Calls were recorded at night, by quietly crawling to a position ca. 2 m from each bird and directing a shotgun microphone toward the advertising bird’s bill. Once a clear bout of braying was recorded, the bird was captured and we measured bill length, bill depth, foot size, flipper size, and head length with digital calipers (see Miyazaki and Waas, 2003a,c for details). We also weighed each male. As the mass of little blue penguins can be quite variable (Dann et al., 1995; Hocken, 2000), morphometric parameters and mass were analyzed separately. Principal component 1 (PC1) scores from a principal components analysis (PCA) of the five morphometric values were used to represent body size. Using the inhalation–exhalation recording with the highest signal‐to‐ noise ratio for each male, we measured the following acoustic parameters: (1) the dominant frequency (DF, in kHz), (2) the highest frequency (HF), (3) the lowest frequency (LF), and (4) the duration of both the inhalation and exhalation phrases (PD, in sec; Fig. 2). Linear regression techniques were then used to examine the relationship between each of the acoustic parameters and the physical attributes of the males. b. Results Table I presents the relationships between call parameters and both PC1 and mass values (Fig. 2; see Miyazaki and Waas, 2003c for details).

Fig. 2. Sonogram of the advertising call of a male little blue penguin showing seven analyzed call properties (DF, dominant frequency; HF, highest frequency; LF, lowest frequency; and PD, phrase duration). Reprinted from J. Avian Biol., Miyazaki and Waas (2003c), with permission from Blackwell Publishing.

404

JOSEPH R. WAAS

TABLE I PEARSON CORRELATION COEFFICIENTS BETWEEN SEVEN CALL PROPERTIES a AND OVERALL BODY SIZE (PC1) AND BODY MASS (N ¼ 12) Call parameter Exhalation DF HF PD Inhalation DF HF LF PD

PC1

Body mass


0.28 0.05
0.04


0.72**
0.91*** 0.66*


0.62* 0.20 0.21
0.001


0.05
0.11
0.38 0.68*

a Reprinted from J. Avian Biol. in Miyazaki and Waas (2003c), with permission from Blackwell Publishing. * p < 0.05. ** p < 0.01. *** p < 0.001.

The lowest frequency of the exhaled phrase was excluded from the analysis because it fell below the recording device’s frequency response or was obscured by low‐frequency background noise in each case. Variation in body size accounted for over 32% of the variance in this analysis. Inhalation DF was the only acoustic parameter to vary significantly with body size as represented by the PC1 values, while four of the seven acoustic parameters varied significantly with body mass (Table I). As we predicted, pitch parameters were negatively correlated with body mass, while the duration of both inhaled and exhaled phrases was positively correlated (Table I). The relationship between PD and the other acoustic parameters was also examined: phrase length only varied significantly with HF (a negative relationship, at p < 0.05). The acoustic characteristics of solo calls revealed information on the signaler’s physical attributes: low‐pitched calls with long exhalation and inhalation phrases indicated a larger, heavier bird; these features may have important impacts on RHP. 2. Correlations Between Body Size and the Acoustic Features of ‘‘Growled Threats’’ a. Method To address this issue, Masamine Miyazaki and I (see Miyazaki and Waas, 2003a for details) recorded growls from 26 males in a single

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

405

population of northern little blue penguins on Tiritiri Matangi Island. Recordings were obtained by playing the call of a stranger 1 m from each male’s burrow or moving a penguin model to within 30 cm of the occupant. Samples with high signal‐to‐noise ratios were obtained on calm nights. The males were captured and measured as described earlier (see also Miyazaki and Waas, 2003a,c). However, male mass was not measured as it varied dramatically before and after the feeding trips of this diminutive subspecies, and recordings were obtained across stages of the breeding cycle (mainly from courtship and incubation). Once again, the PC1 scores from a PCA of the five morphological measures were used to represent body size; head length was also used as a simple estimator of body size, as it explained most variation in PC1. The following acoustic parameters were measured for the growl of each male: (1) the DF in kHz, (2) the HF, (3) call duration (CD, in seconds), and (4) the number of syllables per second. Once again, linear regression techniques were used to examine the relationship between each of the acoustic parameters and the physical attributes of the males. b. Results A significant positive relationship occurred between male size and the DF of their growls, regardless of whether PC1 or head length was used to represent size (Fig. 3; see Miyazaki and Waas, 2003a for details). Male size did neither correlate significantly with the highest frequency or number of syllables per second in the growl nor influence growl duration.

Fig. 3 The relationship between peak frequency of growl and head length. Reprinted from Ibis, Miyazaki and Waas (2003a), with permission from Blackwell Publishing.

406

JOSEPH R. WAAS

The acoustic characteristics of the growl threat call revealed information on the signaler’s physical attributes. However, in contrast to the results for solo calls, larger birds produced higher pitched calls. 3. Conclusions Our analysis (Miyazaki and Waas, 2003a,c) of the relationship between the physical attributes of displaying penguins and the acoustic parameters of their calls indicate that their displays reveal information on body size and mass. Large, heavy opponents will have a distinct advantage during escalated contests that involve ‘‘pushing matches’’ (e.g., breast butts) and bill‐to‐bill contact, where opponents attempt to flip one another with judo‐ like swings of the head (e.g., bill‐lock fights, bill twists; Waas, 1990a). The acoustic features of solo calls varied with body measures in the predicted way: large birds had deep‐pitched calls with long phrases. Studies across a range of birds (Badyaev and Leaf, 1997; Tubaro and Mahler, 1998; Wallschlager, 1980) and other animals (Davies and Halliday, 1978; Clutton‐Brock and Albon, 1979 for classic examples) provide similar results. However, our finding that the pitch of growl calls was positively correlated with body size is perplexing in this context—larger penguins should have been capable of producing sounds at lower frequencies than smaller ones, suggesting that this particular call may not function as a signal of RHP. Growl calls grade, with increasing intensity and amplitude, to low brays which in turn grade to medium and finally full brays (i.e., as used in the solo call display; Fig. 1). The increase in the intensity of the exhaled component corresponds to an increase in pitch. Thus, it is possible the higher pitched calls of large birds reflect a stronger, more ‘‘intense’’ response to the stimulus. For example, the pitch gradient may reveal variation in the animal’s motivational state or willingness to escalate. We will return to this issue later in this chapter. Our studies of RHP‐based signals in little blue penguins are preliminary. It is clear that information on body size and mass is available in little blue penguin calls, but not whether opponents actually recognize or use the information. The next step will be to use playback techniques to broadcast calls that vary in pitch and/or duration to determine whether opponents process the information and vary their behavior accordingly. Miyazaki and Waas (2003a) have shown that male size is positively correlated with a range of ‘‘quality’’ indicators with respect to reproductive success (e.g., early breeding, rapid chick growth), while this was not the case for female size. Also, when females were presented with male solo calls that varied in pitch, they were more likely to respond vocally to low or medium, than high‐pitched, male calls (Miyazaki and Waas, 2003c, but see Miyazaki and

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

407

Waas, 2005). Thus, it seems likely that rivals would also be attentive to the information on male size and quality present in calls. B. WARS

OF

ATTRITION

My research has focused primarily on agonistic signals that little blue penguins use to acquire and defend high‐quality nesting sites—a critical terrestrial resource for these marine birds (Miyazaki and Waas, 2003b). However, for single males, the most important intrasexual competition occurs early in the breeding season when males use solo calls to compete for single females. The males appear to engage in wars of attrition (Bishop and Cannings, 1978; Hammerstein and Parker, 1982), calling loudly and repeatedly in the presence of many other calling males, from areas just below burrow colonies or in central display sites within caves. Males that can sustain a high calling rate over many nights would maximize their chances of capturing the attention of fertile females moving from the sea to the colonies; a high calling rate might also reveal information on male attributes like stamina and body condition, important quality indicators that females and potential rivals should attend to. This logic has been used to explain signal repetition in a range of species and circumstances (Briffa et al., 1998; Payne and Pagel, 1996, 1997). Another important feature of the calling contests that occur between male penguins advertising for females is call overlap. Little blue penguin males regularly ‘‘call over’’ the advertising calls of other males, often causing the initial caller to interrupt its display, as Jouventin (1982) has reported for the emperor penguins (Aptenodytes forsteri). One explanation for this phenomenon is that males overlap rivals’ displays to ‘‘jam’’ information contained in the calls (Naguib and Todt, 1997) and to potentially draw more attention to their own call. Overlapping may be particularly effective when crucial information is contained in the middle or terminal part of the calls (Hultsch and Todt, 1982), as maybe the case for little blue penguins; advertising males start with low amplitude, irregular brays and work up to full amplitude, rhythmic brays as they proceed to a crescendo. Call overlap may also act as a challenge to other callers (Dabelsteen et al., 1997), that the surrounding ‘‘audience’’ (e.g., females or rival males) may be attentive to. Overlap can be understood in the simple context of the war of attrition. Males in good condition with great stamina may not only be capable of repeating their displays at a high rate to secure the attention of available females, but also afford to expend additional energy to jam the signals of rivals. I used playback experiments early in the breeding season to determine if little blue penguin males adjusted their calling rate in response to acoustic

408

JOSEPH R. WAAS

competition created by broadcasting additional advertising calls from loudspeakers (Waas, 1988). Miyazaki and Waas (2002) then examined how call overlap during calling contests influenced an ‘‘audience’’ of males, females, and prebreeders during courtship. If males use solo calls in a war of attrition with other advertising males, I predicted that (1) the playback of additional solo calls would elevate the solo calling rate within colonies relative to playback of other types of vocalizations (e.g., calls used by mated pairs, control calls of other species). Also, if call overlap acts as a mechanism to jam an opponent’s signal and reveal information on stamina (and possibly dominance), we predicted that (2) females would find it easier to locate, or be more attracted to, calls of overlapping males and that males and prebreeders should avoid such callers. 1. Male Penguins ‘‘Escalate’’ Advertising Call Rates in Response to Acoustic Competition from Other Males a. Method I addressed this issue by conducting playback experiments in three little blue penguin colonies on the Banks Peninsula and one on Rangatira Island in the Chathams group during the pair formation period (see Waas, 1988 for details). Two widely separated experimental sites were selected in each colony and a central point in each site was used to broadcast calls and make observations. I ran experiments on calm evenings. Recordings for playback were obtained from areas outside but near (<0.5 km) playback sites. Each 2‐hr experiment (N ¼ 47) was made up of four half‐hour tests, each composed of a 10‐min control period followed by 5 min of playback and ending with a 15‐min postplayback period (see Waas, 1988 for details). The playback consisted of single 15–20 sec calls separated by 30 sec silent periods. Playback treatments included: (1) male solo calls, (2) territorial mutual displays (used by mated pairs in response to intruders), (3) sexual mutual displays (a display that precedes copulation), and (4) the calls of crested penguins (Eudyptes pachyrhynchus and E. robustus) as a control. The playback sequence was selected randomly before each test and amplitude was matched by ear to a level equivalent to 1 m from a displaying bird. All three little blue penguin displays involved the same inhalation– exhalation braying pattern, but the territorial mutual displays tend to have a faster delivery and higher pitch than the sexual mutual displays or the solo calls. All acoustic behavior occurring during each 2‐hr test was noted. The postcontrol period (i.e., the playback and postplayback phases) for each treatment was divided into four 5‐min blocks and mean display rates (per minute) were calculated for each. The mean display rate (per minute) for the 10‐min preplayback control periods was then subtracted from the

409

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

means for each 5‐min block to obtain dependent variables (playback, PP1, PP2, and PP3) for treatments in each sampling area. Therefore, the dependent variable represents variation from calling rates before playback began. A one‐way analysis of variance (ANOVA) was used to detect differences across the four treatments in the eight sampling sites. Orthogonal comparisons (Sokal and Rohlf, 1981) were used to distinguish variation in response between (1) the crested penguin treatment and little blue penguin treatments and (2) the behavior used as the treatment and the other two little blue penguin displays. b. Results Colony occupants produced solo calls at a higher rate when they heard little blue penguin playback (of any type) than crested penguin playback during playback and the first 5 min of the postplayback period (Table II; Fig. 4). More importantly, the playback of solo calls caused birds to solo call at a higher rate than did playback of the other two little blue penguin calls (an effect that was maintained throughout the playback period and the first 10 min of the postplayback period) (Table II; Fig. 4). There was no significant difference in the effect that the two mutual display treatments had on the facilitation of solo calls. Playback of solo calls caused the birds to maintain their solo call rate significantly above that given in response to crested penguin calls throughout the playback and postplayback periods (Fig. 4). Playback of solo calls appeared to be largely ignored by mated pairs of little blue penguins. The solo call treatment only caused a mutual display response significantly different from that observed during the crested TABLE II RESULTS OF ANOVA ACROSS SOLO CALL RESPONSES TO PLAYBACK TREATMENTS FOR EACH 5‐MIN TIME SEGMENT AND ORTHOGONAL COMPARISONS

Response Solo call

Orthogonal comparisonsa

ANOVA

Time (5‐min intervals, 1¼ playback)

F (df ¼ 3,28)

p

A

B

1 2 3 4

23.373 7.164 3.408 1.499

0.000 0.001 0.031 0.235

0.001 0.01 n. s. n. s.

0.001 0.01 0.01 n. s.

a Orthogonal comparison A: response to crested penguin treatments versus response from little blue penguin treatments. Orthogonal comparison B: response to little blue penguin solo call versus response from other little blue penguin treatments. n. s., not significant. Reprinted from Anim. Behav. 36, in Waas (1988), with permission from Elsevier. Playback treatments include solo call, territorial mutual display, sexual mutual display, and crested penguin call.

410

JOSEPH R. WAAS

Fig. 4 Mean response rates per minute for playback treatments. The dashed line represents pre‐experiment performance rates. Each histogram shows the change in response rate from the initiation of playback (0 min) to the end of the postplayback period (20 min). Response change within histograms was not tested statistically. Vertical comparisons made with respect to time (i.e., follow equivalently hatched bars downward) indicate variation in response rate between playbacks and were tested statistically. Bars with any overlapping letters are not significantly different from each other. Bars not sharing any letter are significantly different (p < 0.05; Bonferroni confidence limits). SC, solo call; MD, mutual display; SMD, sexual mutual display; TMD, territorial mutual display; COP, copulation; CON, composite of crested penguin calls. Reprinted from Animal Behaviour 36, in Waas (1988), with permission from Elsevier.

penguin playback during the first 5 min of the postplayback period (Fig. 4). Solo call playback only increased copulation rates significantly above those observed during the crested penguin treatment during the second 5 min of the postplayback period (Fig. 4). Thus, little blue penguins elevate solo call rates in response to an apparent increase in the use of solo calls by birds around them, as created by using playback to increase solo call levels above background levels. Solo calling birds in the colony doubled the rate at which they used the display prior to playback (or in response to crested penguin playback; Fig. 4). Little

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

411

blue penguins appear to have solo calling contests that suggest single males engage in a war of attrition to advance their chances of attracting single females that approach colonies from the sea. 2. Do Costs Associated with Overlapping a Competitor’s Solo Calls Enhance a Male’s Ability to Attract Females and Repel Males? a. Method Masamine Miyazaki and I conducted an experiment to address this issue on Motunau Island near the Banks Peninsula on New Zealand’s South Island (see Miyazaki and Waas, 2002 for details). Recordings of solo calls for the experiment were collected during the courtship period of 1984 from the Wainui Cave colony on the Banks Peninsula. Ten high‐quality solo calls were obtained from each of 10 different males. The calls from each bird were digitized, adjusted to the same recording level, and then copied to playback tapes. Only full exhalation/inhalation brays performed in the middle of an advertising bout were used to compose playback tapes to avoid the incomplete versions given at the beginning of a bout. We obtained three exhalation/inhalation phrases from each call, giving us a total of 30 phrases from the 10 calls obtained from each of the 10 recorded males. Each call in a given matched‐pitch pair played the role of the overlapping call on one tape and the role of the overlapped call on a second tape. Calls were edited so the first male’s call, broadcast from one speaker, was consistently overlapped by the call of the second male, broadcast from a second speaker. All playbacks consisted of three exhalation/ inhalation sets from each of the males, with call overlap initiated just after the first exhalation of the first male’s call (Fig. 5). The 10‐sec vocal segments were separated by 60‐sec silent periods, and were continuously broadcast over 2 hr. The two 2‐hr experiments were conducted after nightfall. For each test, three circular arenas 8 m in diameter were established side‐by‐side and parallel to the shoreline to create: (1) an ‘‘overlapping arena,’’ (2) an ‘‘overlapped arena,’’ and (3) a control arena (see Miyazaki and Waas, 2002 for full details). Amplified speakers were placed in the middle of the first two arenas while an unconnected speaker was placed in the third (all pointing out to sea). The observer was positioned 10 m from the site behind rocks or shrubs. The two speakers that broadcast calls were connected to the playback system to broadcast the overlapped and overlapping call in the appropriate arenas at 85 dB at 1 m (matching the amplitude of a natural solo call). All arenas were checked before playback began to confirm no birds were present. Every 30 min after playback began, we entered the three arenas to count the number of birds found among the rocks and boulders strewn along the shoreline. Any bird found within 4 m of a speaker was deemed as selecting that treatment (i.e., overlapper, overlapped, or silent control). These birds were captured to

412

JOSEPH R. WAAS

Fig. 5 Example of call overlap on the playback tapes. The song of Bird B overlaps that of Bird A after the first exhalation. The short pulses between exhalations are inhalations (adapted from Miyazaki and Waas, 2002).

measure their bills for sexing and to record their band numbers (playback continued while this was done). Captured birds were then released well away from the experimental site. A total of 18 tests were conducted at 18 different and widely separated sites during our 9 nights on the island. A one‐way factorial ANOVA and Fisher’s multiple comparison tests were used to compare the numbers of males, females, and young entering each arena (see Miyazaki and Waas, 2002 for details). b. Results Call overlap influenced female distribution, but not that of males or young little blue penguins (Table III). Significantly more females were found near the speaker that continuously overlapped the call being broadcast in the other playback arena or the control area (Table III; see Miyazaki and Waas, 2002 for details); note, however, that females were only detected during 7 of the 18 tests. Males were recorded during a larger number of tests (12 of the 18) but there was no significant difference in the number of males found in the three types of arenas (Table III). Similarly, young birds (about 1 year and 8 months of age) were found to occupy the arenas more regularly than females (11 of 18 tests) but showed no clear preference for occupying any particular treatment arena (Table III). Two birds responded vocally to playback during the tests, producing short contact calls just after a bout of playback. Both were females (from bill measures): one gave a single contact call while entering the overlapping playback arena, while the other produced a series of 16 contact calls in the overlapped playback arena.

413

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

TABLE III THE MEAN SD NUMBER OF BIRDS FOUND WITHIN EACH TREATMENT ARENA Treatment

Overlapping

Overlapped

Control

Males (N ¼ 12) Females (N ¼ 7) Young (N ¼ 11)

1.08  1.08 2.14  1.95a 0.24  0.31

1.24  0.87 0.43  0.54b 0.06  0.08

0.42  0.90 0.43  0.79b 0.09  0.25

Data were excluded from tests where no bird landed on any arena (N shows the number of tests that each type of individual was represented). Values with the same letter in superscript were not significantly different (Fisher’s multiple comparison tests; p < 0.05) (adapted from Miyazaki and Waas, 2002).

3. Conclusions Male little blue penguins appear to use solo calls to compete acoustically for available females. By using playback to artificially exaggerate the solo call rate in colonies, I showed that males attempted to ‘‘meet the challenge’’ by increasing their own calling rate. Such a war of attrition will be won by males who call often and over the longest period of time because they will be more conspicuous to available females than males that call infrequently and for shorter periods; frequently calling males may also reveal honest information on stamina and physical condition to females. Similarly, Farr (1976) argued that the social facilitation of male display behavior in guppies, Poecilia reticulate, was the result of intrasexual competition for females, and evidence for such intrasexual wars of attrition now exist for many other animals (e.g., frogs, Bosch and Marquez, 2000; butterflies, Kemp, 2002). Continuous playback of male royal penguin (Eudyptes schlegeli) advertising calls (the ‘‘vertical head swinging display’’) during the pre‐pairing phase caused males to quadruple their calling rates (Waas et al., 2000); further, male royal penguins were more responsive to the advertising calls of their neighbors than to those of strangers (see Waas, 1995 and Waas et al., 2000 for more detail). Similar calling contests also appear to occur in zebra finch (Taeniopygia guttata) colonies (Waas et al., 2005). We are now designing playback experiments to determine whether females are more attracted to speakers broadcasting advertising calls at a high rate than to those broadcasting at a lower rate. Greenfield and Roizen (1993) suggest that synchronous chorusing in katydids, Neoconocephalus spiza, and other animals (Greenfield et al., 1997) occurs as the result of males attempting to jam one another’s calls during contests for females. This hypothesis fits well within the war of attrition model as males who can afford to jam the signals of others, in addition to maintaining high conspicuousness by calling often themselves,

414

JOSEPH R. WAAS

demonstrate superiority in terms of stamina and physical condition. In katydids, the crucial information is concentrated near the beginning of the male advertising call, and females have been shown to orient most toward the ‘‘leading’’ or overlapped call (Greenfield and Roizen, 1993). In contrast, the most valuable information in penguin calls is likely to occur near the end of the call (Jouventin, 1982). Correspondingly, we found that female little blue penguins showed a preference for the ‘‘following’’ or overlapping calls during the two‐speaker playback experiment (Miyazaki and Waas, 2002). Whether this result is the product of (1) important information near the end of the calls becoming unavailable for overlapped males (e.g., Naguib and Todt, 1997), (2) a simple logistic effect associated with it being easier to locate the last call heard in a train of calls (e.g., Miyazaki and Waas, 2002), or (3) a female preference for dominant males (e.g., Dabelsteen et al., 1997) or those who can afford to jam the calls of competitors in addition to maintaining their own advertisements, will provide fertile ground for further research. C. RISK‐BASED SIGNALS Risk‐based signals are validated by the potential and real costs of opponent retaliation. For example, a signaler that uses a display that placed its head within easy striking distance of the opponent’s bill reveals a greater motivation to secure the resource than one that uses a posture that protects the head. The opponent should be attentive to these differences because bluff is totally precluded—the signaler must take the risk to perform the display. The conspicuous elements of signals, like vocalizations, may simply advertise the risks that the signaler is taking, ensuring the rival is aware of them. If animals used this signaling convention, one would predict that (1) the displays in the agonistic repertoire would differ in risk, perhaps on a gradient that allowed individuals to express a range of motivational levels. Further, one would expect that, on average, (2) high‐risk displays would be more effective at repelling opponents than lower risk versions. 1. Little Blue Penguin Displays Differ in Risk a. Method I addressed this issue (see Waas, 1990a, 1991a for details) by collecting 200 hr of focal animal observations from two cave and two burrow colonies of penguins over three breeding seasons (May–February; austral spring–summer). Continuous samples were obtained from focal males and females defending nest sites (N ¼ 29 in caves and N ¼ 6 in burrows; 398 sampling sessions in total). Individuals were marked with numbered metal flipper bands or were recognized by distinctive patterns on their flippers. Night‐vision equipment was used to avoid disturbing the

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

415

subjects during observations. In burrow habitats, it was necessary to use observation boxes with Perspex ceilings to observe nest occupants (see Waas, 1990a for details); in caves, I set up the equipment behind rocks or driftwood. A total of 3360 agonistic interactions were observed (Waas, 1990a) from which I was able to identify the complete agonistic repertoire and measure physical properties of the displays that might contribute to differences in performance risk. i. Subjective estimates of display risk I used four criteria to provide a ‘‘common sense’’ ranking of agonistic displays in order of increasing cost (see Waas, 1990a,b for details) measured by the risk of sustaining an injury during the performance of the display: (1) distance (displays initiated near the opponent were considered higher in risk than those used at greater distances); (2) orientation (displays positioning the head toward the opponent were considered higher in risk than those shielding the head and eyes); (3) movement (displays that close the distance between the subject and its opponent were considered higher in risk than those performed while stationary or moving away); and (4) conspicuousness [displays with components that made the subject and its position obvious to the opponent (e.g., vocalizations, postures revealing the white underbelly) were considered higher in risk than those that made the actor less obvious]. ii. Quantitative estimates of display risk To provide a more objective measure of display risk, I determined how often penguins were attacked, bitten, or pecked during the performance of each display in their agonistic repertoire (i.e., including both offensive and defensive acts; Waas, 1991a). Displays during which actors were likely to be attacked by the opponent were considered higher in risk and more costly to perform than displays that rarely led to an overt form of opponent retaliation. Only data from cave‐dwelling territorial males (N ¼ 29) and females (N ¼ 29) were used for this analysis due to sample size constraints in the other contexts. b. Results Table IV summarizes the agonistic repertoire of little blue penguins, and ranks acts in terms of performance risk using the four outlined criteria (for full details see Waas, 1990a,b, 1991a). The birds used an extensive repertoire of agonistic acts that varied in form and size between cave and burrow habitats (see Waas, 1990a for a description of each display). Two major types of agonistic activities were distinguishable: defensive and offensive behavior. Defensive activities appeared to reduce opportunities for opponent‐inflicted injuries, while offensive activities included properties that increased the risk of sustaining an injury during the interaction. For example, defensive behaviors always involved moving or orienting away from the opponent, and tended to be performed silently (Table IV), making the actor inconspicuous. Offensive activities, on the

TABLE IV AGONISTIC BEHAVIORS (WITH ABBREVIATIONS USED IN SUBSEQUENT TABLES) USED BY CAVE‐ AND BURROW‐DWELLING PENGUINS, CATEGORIZED ON THE BASIS OF PERFORMANCE RISK Criteria used to categorize and rank agonistic behaviors

Categorya

Risk level

Behaviors

416

Defense Distance increasing (D1) Stationary (defensive) (D2) Offense Stationary (offensive) (O1)

1

Low walk Submissive hunch

2

Face away Indirect look

3

Distance reducing (O2)

4

Contact (O3)

5

Direct look Directed flipper spread Point Zigzag approach Directed flipper spread approach Bill to back Breast butt Bill to bill Bill slap Bill lock/twist

Performance distance (m)b Cave‐dwelling birds n. s. <1; 1–2

Orientationc

Movementd

Percentage of cases with vocalization (types of vocalizatione)

Away Away

Away Away

– –

<1; 1–2 <1; 1–2

Away Away

– –

– 7.6 (Growl) f

>3; 2–3 2–3; 1–2

Toward Toward

– –

2.2 (Growl) 35.4 (Bray/Growl)

1–2; 2–3 1–2; 2–3 1–2; <1

Toward Toward Toward

– Toward Toward

13.9 (Growl) 4.9 (Growl) 12.0 (Growl/Bray)

In contact In contact In contact In contact In contact

Toward Toward Toward Toward Toward

Toward Toward – – –

– – – – 100.0 (Growl/ Aggressive yell)

1–2; <1 In contact In contact

417

Overt aggression (OV)

6

Attack Bite Fight

Defense Distance increasing (D1) Stationary (defensive) (D2) Offense Stationary (offensive) (O1)

1

Low walk

2

Face away

<1; 1–2

3

Distance reducing (O2) Contact (O3) Overt aggression (OV)

4

Stretch‐neck look Directed flipper spread Bill vibe Lunge/Hiss Lung/Peck Attack Bite Fight

a

5 6

Toward Toward Toward

Toward – –

2.4 (Aggressive bark) 42.9 (Growl) 100.0 (Growl/ Aggressive yell and aggressive bark)

Away

–

Away

–

–

2–3; >3 >3; 1–2

Toward Toward

– –

2.1 (Growl) 100.0 (Bray)

1–2; 2–3 1–2; <1

Toward Toward

– Toward

In contact n. s. In contact In contact

Toward Toward Toward Toward

Toward Toward – –

81.8 (Growl) 100.0 (Hiss/ Aggressive bark) – – 20.0 (Growl) 100.0 (Growl/ Aggressive yell and aggressive bark)

Burrow‐dwelling birds n. s. Away

Behaviors are categorized and ranked with respect to performance risk using four criteria (see text). Performance distance was measured with respect to the opponent’s position (i.e., <1 m, 1–2 m, 2–3 m, or >3 m away). When a one‐way ANOVA indicated that a given behavior was used more frequently at some distances than others (p < 0.05), I list the two distances at which that behavior was most commonly used (i.e., most frequent first). n. s., not significant. c Orientation of the actor’s head and body with respect to the opponent’s position. d Direction the actor is moving with respect to the opponent’s position. e Vocalizations are listed with respect to how frequently they accompanied the behavior (i.e., most common first). f Only territorial females used a vocalization (i.e., growl) during indirect look. Adapted from Waas (1990a,b), with permission from Elsevier. b

418

JOSEPH R. WAAS

other hand, always involved moving or orienting toward the opponent, and tended to include a vocal or tactile component, making the actor’s position conspicuous even under the low light conditions (Table IV). Note that I have used the terms offensive and defensive in preference to aggressive and submissive. Defensive displays and related acts are not necessarily submissive in nature—the displays may simply allow the animal to position itself or create a context, where it may be eventually able to gain the ‘‘upper hand’’ later in the agonistic interaction. The four criteria suggested two subcategories of defensive acts (from lowest to highest risk; see Waas, 1990a,b, 1991a for details): (1) distance‐ increasing and (2) stationary defensive activities. The two types shared many properties (Table IV), but movement away from the opponent, suggested that distance‐increasing activities reduced retaliation risks. Four subcategories of activities composed the offensive category (from lowest to highest risk): (3) stationary, (4) distance reducing, (5) contact displays, and finally, (6) overt aggression. All involved orienting toward the opponent and had components that made actors conspicuous (by acoustic or tactile means; Table IV). Distance‐reducing displays appeared higher in risk than stationary displays because they were performed in closer proximity to the rival and involved further reducing the opponent’s striking distance (Table IV). Contact displays were considered higher in risk than distance‐reducing displays because the inter‐rival striking distance was reduced to zero; the displays also involved postures that drew rivals’ bills toward one another’s head and eyes (Table IV). Overt aggression represents the pinnacle of risk—the rivals attacked, bit, and fought one another, and injuries were likely to occur. Most fights involved one animal biting hold of the other’s nape while delivering a flipper bashing (bite nape fights). However, a more dangerous form of fighting sometimes occurred: bill‐lock fights (11.9% and 50.0% of fights in caves and burrows) involved interlocking bills so the hooked bill tips could scrape across one another’s eyes while delivering flipper beatings to the head and body. The quantitative estimates of risk provided support for the subjective categorization but not complete support. Table V shows the percentage of cases in which a signaling animal was attacked, bitten, or pecked while using the two categories of defensive displays and the three categories of offensive displays (the z‐scores reveal the probability of an opponent resorting to overt aggression following the actor’s use of each display type). When using an offensive display, both territorial males and females were most likely to be attacked, bitten, or pecked during the performance of the high‐risk contact displays than during stationary or distance‐reducing displays. Major differences between stationary and distance‐reducing displays were not apparent. Furthermore, actors were more likely to be attacked

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

419

TABLE V PERCENTAGE OF CASES (Z‐SCORE) THAT TERRITORIAL MALES OR TERRITORIAL FEMALES WERE ATTACKED, BITTEN, OR PECKED DURING THE PERFORMANCE OF EACH TYPE OF AGONISTIC DISPLAYa Defensive displays

Offensive displays

Actor

D1

D2

O1

O2

O3

TM TF

3.6 (þ1.73) 4.2 (þ0.80)

5.8 (þ3.33) 5.1 (þ0.46)

1.2 (–2.34) 0.0 (–0.98)

0.0 (–2.43) 0.3 (– 0.05)

4.8 (þ1.44) 3.8 (þ1.79)

a See Table IV for abbreviations; from Waas (1991a) with kind permission of Springer Science and Business Media. z‐Scores approximate a standardized normal distribution, so values  þ1.96 indicate a significant positive association with overt aggression, while values  –1.96 indicate a negative association at the 1 ¼ 0.05 level. TM, territorial males; TF, territorial females.

during defensive activities than during offensive acts (Table V), so a defensive display was not an effective way of avoiding attack. However, the risk of injury for each attack was likely to be much lower than that associated with the offensive activities. During defensive activities, vulnerable body surfaces (e.g., eyes, head) were positioned away from the attacker and the attacker simply bit or pecked the opponent on the nape or flank. The ‘‘head‐on’’ collisions suffered during interactions between animals using offensive acts would involve much greater risks—the attacker usually bit or pecked the opponent on the head or throat. These results suggest that ‘‘frequency of attack’’ should not be the only criterion used to distinguish differences in performance risk. Accurate estimates of performance risks may require measures of both attack frequency and the cost of each attack in terms of the injury potential. Subjective and quantitative analyses provided support for the premise that the agonistic displays of penguins differ in performance risk. For example, not only were there differences in risk between the main types of offensive displays (i.e., stationary, distance reducing, and contact displays), but fine scale differences in risk were apparent across displays that composed each subcategory. For example, the three stationary offensive behaviors in cave habitats (direct look, directed flipper spread, and point; Table IV) tended to be used at different distances from the rival (i.e., ca. 3, 2, and 1 m, respectively; Waas, 1990a). This difference would influence the opponent’s striking distance, and therefore, the signaler’s chances of being injured while performing the display. It is noteworthy that cave dwellers had a much larger agonistic repertoire than burrow dwellers (22 vs 13 acts). Cave dwellers nest colonially and interact with one another three times as often as the semicolonial burrow dwellers.

420

JOSEPH R. WAAS

However, burrow dwellers were more likely to attack opponents (caves: 0.5% of interactions; burrows: 4.2%), engage in more dangerous fights (caves: 11.9% were bill‐lock fights; burrows: 50%) and had longer fights (caves: 14.1 sec; burrows: 34.5 sec) than birds in cave colonies (Waas, 1990a). Further, when burrow dwellers bit an opponent, they held on longer (cave: 3.1 sec; burrow 46.0 sec). A diverse repertoire of agonistic acts, varying along a fine risk gradient, may reduce the number of interactions that escalate to overt aggression, reflecting the cave dweller’s need to restrict the risks of injury and energy expended during any one interaction. 2. Are High‐Risk Displays More Effective at Repelling Opponents? a. Method To address this issue, I used lag sequential analysis (Sackett, 1979) to identify nonrandom sequences of behavior or ‘‘events’’ occurring during the 150 hr of focal animal observations collected from cave dwelling penguins (see Waas, 1991a for full details). I restricted the analysis to interactions between territorial males (N ¼ 16; ca. 15 interactions from each) and nonterritorial birds who invaded the males’ nesting space. The analysis organizes interactions as trains of events, with each event occupying a single ‘‘unit.’’ An event of interest (e.g., an offensive display) is then set as the ‘‘criterion’’ from which a count is made of how often all the other events occupy previous (e.g., Lag – 1, Lag – 2) or following (e.g., Lag þ 1, Lag þ 2) units. Observed probabilities for a given event, at a specified lag from the criterion, are assessed by dividing the number of times the event occurred at that lag by the total number of times the criterion occurred in the data. Expected probabilities for the event (calculated by dividing the total number of times the event occurred, without respect to a specific lag, by the total number of occupied units in the sample) were then subtracted from observed probabilities. The product is divided by an error term, producing a z‐score, which approximates a standardized normal distribution. Thus, positive scores higher than þ1.96 indicate that the event is more likely to occur than expected at the specified lag from the criteria, and a value below
1.96 means the event is less likely to occur, at the 1 ¼ 0.05 level (see Bradbury and Fincham, 1991 for step‐by‐step instructions). For each little blue penguin display or category of interest (criterion), I identified activities that intruders were most likely to be performing immediately before (Lag – 1) and after (Lag þ 1) the territorial bird’s signal, allowing me to estimate the effect that the display had on the opponent. I also examined the territorial bird’s behavior at Lag þ 2, with respect to the intruder’s behavior at Lag þ 1, to estimate what if anything a given display (criterion) predicted about the actor’s next move (Waas, 1991a). Finally, the activities at Lag – 2 were examined to assess what

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

421

influenced the Lag – 1 distribution and what activities predicted the display set as the criterion. b. Results Table VI displays the transitional probabilities (%) for the three major categories of offensive displays [stationary displays (O1), distance‐reducing displays (O2), and contact displays (O3); data from the three to five displays comprising each category were pooled for this initial analysis; see Waas, 1991a for further details]. The sequences represent interactions that are 5‐steps long, and alternate between the actor’s and the opponent’s actions; the third step in the sequence represents the activity of interest (i.e., the actor’s use of each type of aggressive display O1, O2, and O3). The table shows both the actor’s (1. Lag – 2) and opponent’s (2. Lag – 1) behavior leading up to the performance of each type of aggressive display used by the actor (3. Lag 0; the criterion), and then the opponent’s (4. Lag þ 1) and actor’s (5. Lag þ 2) actions following the display of interest (i.e., the criterion). In the following paragraphs, I describe the (1) effect that each type of offensive display had on the intruder and (2) what the display category predicted about the actor’s next act. The low‐risk offensive stationary displays appeared to have little effect on opponents (Table VI; Waas, 1991a). Following the actor’s performance of these displays, rivals were less likely to retreat and more likely to use an offensive behavior (i.e., O2, O3, and OV) than they were before the actor used the display. The displays predicted that the actor would escalate to distance‐reducing displays, when the opponent failed to retreat (see steps 3–5 in the stationary category and 1–3 in the distance‐reducing category; Table VI). The intermediate‐risk distance‐reducing displays were more effective at deterring rivals than offensive stationary displays (Table VI). After the actor used one of these displays, the intruder was much more likely to retreat than they were before the display was performed. These displays predicted that the actor would escalate to a contact display if the opponent failed to retreat (see steps 3–5 in the distance‐ reducing category and 1–3 in the contact display category; Table VI). If, however, the intruder retreated or used a defensive behavior, the actors were most likely to drop back to a stationary display (see steps 3–5 in the distance‐reducing category and 1–2 in the stationary category; Table VI). High‐risk contact displays had a major impact on intruder behavior. Following the actor’s use of a contact display, the intruder was far more likely to retreat than it was before the actor used the display (Table VI). Intruders typically responded to contact displays by retreating, thus, the train of events appeared to cease at this point (Table VI).

422

JOSEPH R. WAAS

TABLE VI TRANSITIONAL PROBABILITIES (%) FOR BEHAVIORS LEADING TO CATEGORY OF AGGRESSIVE DISPLAY

AND

FOLLOWING EACH

Behavioral sequence

Criterion

Behavior

1 Lag – 2 (actor)

2 Lag – 1 (opponent)

3 Lag 0 (actor)

4 Lag þ 1 (opponent)

5 Lag þ 2 (actor)

Offensive stationary

D1 D2 O1 O2 O3 OV

1.0a 13.5a 45.1 28.4b 6.7 5.3b

34.0b 26.7b 32.3a 6.9a 0.5a 0.0a

– –  – – –

16.3 29.1b 31.8a 17.8 3.7 1.2

19.0a 22.2 39.1 15.2b 4.0 0.5

Distance reducing

D1 D2 O1 O2 O3 OV

0.8 13.9a 58.1b 14.9 9.3 3.2

9.4 44.7b 42.0 3.7a 0.2a 0.0

– – –  – –

47.2b 34.0b 14.5a 3.0a 1.3 0.0

1.3a 14.8 53.5b 8.1 15.1b 7.2b

Contact

D1 D2 O1 O2 O3 OV

1.9 5.5a 19.7 45.7b 7.9 0.5

24.6 13.1 23.9 3.1 16.5b 0.0

– – – –  –

45.4b 11.5 2.8a 0.3a 16.5b 4.8

0.3 3.5 29.4 31.8 4.0 0.3

Less often than expected by chance (p  0.05). More often than expected by chance (p  0.05). –Criterion set at Lag 0. See Table IV for abbreviations; from Waas (1991a) with kind permission of Springer Science and Business Media. The sequential Bonferroni technique was used to avoid type I errors in each block of comparisons (Rice, 1989). a

b

To further investigate the relationship between the performance cost and the ‘‘deterrence potential’’ of offensive displays, I isolated conspicuousness as a key risk factor (see Waas, 1991a for details). Little blue penguins enhance the visual components of many displays with acoustic elements, and many of the offensive displays used by them were performed either silently or with calls (Table IV; Waas, 1990a,b). By using a call, these nocturnal animals make their position and presence conspicuous to opponents, conferring a risk to the performer. I predicted that the displays with

423

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

vocal components would be more effective in deterring opponents than silent versions of the same displays. Table VII displays the transitional probabilities (%) for silent and vocal offensive stationary displays (see Waas, 1991a for full details). In the following paragraphs, I describe (1) the effect that each type of offensive display had on the intruder and (2) what the display category predicted about the actor’s next act. Silent offensive stationary displays were not effective at deterring intruders. Following the use of the silent display, intruders were much less likely to retreat and appeared more likely to escalate to high‐risk offensive acts than they were prior to the display (Table VII). Actors typically followed silent offensive behavior with escalation if the opponent did not retreat. Vocal offensive stationary displays were more effective than the silent versions at preventing intruders from escalating to higher risk behavior and maintaining their movement away from the actor (Table VII). As most intruders moved away, or continued to perform defensive behavior in

TABLE VII TRANSITIONAL PROBABILITIES (%) FOR BEHAVIORS LEADING TO AND FOLLOWING SILENT VOCAL OFFENSIVE STATIONARY DISPLAYS

AND

Behavioral sequence

Criterion

Behavior

1 Lag – 2 (actor)

2 Lag – 1 (opponent)

3 Lag 0 (actor)

4 Lag þ 1 (opponent)

5 Lag þ 2 (actor)

Silent O1

D1 D2 Silent O1 Vocal O1 O2 O3 OV

1.4a 16.5a 34.7 6.1 31.3b 5.4 4.6

34.1b 24.5 30.3 3.2 7.2a 0.6a 0.0a

– –  – – – –

11.8a 27.9b 27.3 6.8 20.3 4.5 1.5

22.5 24.9 28.4 3.7 15.4b 4.6 0.5

Vocal O1

D1 D2 Silent O1 Vocal O1 O2 O3 OV

0.7 13.5 26.5 15.3 23.1 7.8 6.9b

25.3b 37.9 24.5 0.7 5.3 0.0 0.0

– – –  – – –

36.9b 32.7 23.6 0.8 6.0 0.0 0.0

9.7 23.6 33.8 15.6 16.5 0.7 0.0

See footnotes of Table VI. See Table IV for abbreviations; from Waas (1991a) with kind permission of Springer Science and Business Media.

424

JOSEPH R. WAAS

response to the vocal displays, actors tended to follow the display with another stationary behavior (Table VII). There was some evidence to suggest that actors would escalate to distance‐reducing behavior if the intruder did not move off. A very similar set of patterns was observed for silent and vocal distance‐ reducing displays (detailed in Waas, 1991a), with intruders being driven off the territory more effectively with vocal versions of the displays, and actors escalating to higher risk behavior when this did not occur. Furthermore, an analysis of differences in displays within each general category of behavior (i.e., across the three stationary, two distance‐reducing, and four contact displays composing each of the three categories) provided further support for a positive relationship between performance risk and effectiveness in deterring intruders (Waas, 1991a). The entire offensive display repertoire appeared to form a fine scale gradient from the lowest to highest risk activities, and this gradient correlated with how effective displays were at deterring intruders. 3. Conclusions My analysis of displays comprising the agonistic repertoire of little blue penguins produced both qualitative and quantitative support for the idea that the activities differ in how vulnerable they make actors to opponent retaliation. Furthermore, there was a positive correlation between performance risk and the displays’ effectiveness in deterring intruders. Animals using high‐risk displays physically demonstrate a high motivation to secure the resource; the stylized components of displays may simply act to advertise the risks being taken. Bluff is precluded by the real and potential costs of social remediation, so conveying information in this way can be evolutionarily stable. Actors followed a step‐by‐step escalation in risk when opponents failed to ‘‘back down,’’ such a process would ensure that the opponent was deterred at the lowest possible cost (Waas, 1991b). Biological modelers of animal contests provide theoretical support for this risk‐benefit approach (e.g., Adams and Mesterton‐Gibbons, 1995; Deag and Scott, 1999; Grafen, 1990) and studies across a wide range of taxa provide empirical support (e.g., Andersson, 1976; Breithaupt and Atema, 2000; Breithaupt and Eger, 2002; Enquist et al., 1985; Hansen, 1986; Jennings et al., 2002; Nelson, 1984; Popp, 1987a,b; Waas, 1991a,b; see Bradbury and Vehrencamp, 1998 for a review). For example, a strikingly similar relationship occurs between the way that little blue penguins use vocal components, and crayfish (Astacus leptodactylus) use chemical components, to deter opponents by making their presence and position obvious to the rival. Both species are nocturnal and perform offensive displays with or without the vocal or chemical component. A lag sequential

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

425

analysis of crayfish agonistic interactions revealed that chemical‐added displays were more effective in deterring opponents than the same displays performed without the chemical component (Breithaupt and Eger, 2002), just as I demonstrated above for the vocal offensive displays of penguins. However, I would be remiss not to point out that empirical support for the risk‐benefit model is not universal (e.g., Lange and Leimar, 2003; Wilson, 1994); I will return to this point later in the chapter and explain why we should not expect universal support. D. SIGNALS OF INTENT Animals that interact regularly, with easily distinguishable opponents, may be able to use signals of intent to settle disputes over resources, providing they can retain a memory of previous interactions (Bossema and Burgler, 1980; van Rhijn, 1980; van Rhijn and Vodegel, 1980). For example, a given actor might always follow its rendition of Display A by an immediate, outright attack of an approaching rival, while Display B is typically followed by a lunge and bite, and Display C by a lunge alone. Providing that the actor pairs each signal (conditional stimuli) with each consequence (unconditional stimuli) consistently, a rival who was familiar with the actor and has interacted regularly with it would come to associate the three displays with different postdisplay consequences through Pavlovian conditioning. Displays used in this way need not vary in terms of immediate production costs or risks (the different signals could be arbitrary acts as long as they were obvious and easily distinguishable to the rival). Instead, the signaling protocol is validated by past investment made by the actor to condition rivals it interacted with regularly. Bluff, on the actor’s part, is never totally precluded; however, bluff dilutes the actor’s ability to signal effectively because it weakens the association between the display and its associated outcome for that rival. Thus, acts of bluff must be contained if the signaler is to benefit from the use of intention signals. The graded displays used by many animals (Andersson, 1980), including little blue penguins (Fig. 1), are likely candidates for signals of intent. Immediate cost differentials across graded displays, like the low‐ and high‐pitched growls of little blue penguins or their low and medium brays, are typically trivial. However, by using a display that varies along an easily recognized sensory continuum (e.g., associated with shifts in pitch and amplitude), signalers may be able to use the conditioning process identified earlier to convey detailed information on what they are likely to do following their performance of each display. The graded displays of little blue penguins were studied to test three predictions associated with

426

JOSEPH R. WAAS

signals of intent. (1) The vocal displays of little blue penguins should be individually distinctive so that a given actor can condition rivals to associate the consequences of its postdisplay actions with its individually distinctive displays. (2) The individual distinctiveness of displays would only be useful if other individuals can actually detect the ‘‘signatures,’’ so little blue penguins should be capable of individual recognition on the basis of the sound. (3) As an actor would need to maintain a consistent relationship between its displays and its postdisplay actions, the behavior the actor uses after each display should be predictable. To measure the predictive value of agonistic displays effectively, the rival’s actions following the display need to be standardized (Waas, 1990b, 1991b), so playback and model experiments were used to investigate this issue. 1. Acoustic ‘‘Signatures’’ Exist in Little Blue Penguin Calls a. Method To assess whether little blue penguins possess vocal signatures [an attribute that may be important in a variety of signaling categories, e.g., intentional and ‘‘convention’’ signals (Deag and Scott, 1999)], the solo calls of the 12 male southern little blue penguins and the growl calls of the 26 male northern little blue penguins described in the previous section were analyzed (Section III.A; see Miyazaki and Waas, 2003a,c for details of this work). For the solo calls, we simply calculated coefficients of variation (CV) for the dominant frequency, the highest frequency, and the duration of both the inhalation and exhalation phrases, and calculated call individuality by dividing the intermale by the intramale CVs (see Jouventin, 1982). For growls, a one‐way ANOVA (across males) was used to obtain F‐ratios for each acoustic measure (i.e., the dominant frequency, the highest frequency, call duration, and the number of syllables per second). In addition to identifying parameters that differ significantly across individuals, F‐ratios indicate which parameters best discriminate interindividual differences because a large F‐ratio represents greater interindividual than intraindividual variability (e.g., Campbell, 1989). If signals of intent are an important feature of the penguins’ communication system, we might also expect call individuality and individual recognition to be initiated early in life. This way, individuals in stable social groups could maximize the amount of information they had on the behavior of rivals and improve their own ability to modify rival behavior through the conditioning process discussed earlier. Thus, Shinichi Nakagawa, Masamine Miyazaki, and I also determined whether the inhalation–exhalation begging calls (an early version of the solo call; Fig. 6) of 11 northern little blue penguin chicks that were about 1‐month old contained acoustic signatures (see Nakagawa et al., 2001 for details). Here we used one‐way ANOVAs

427

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

Chick 2A

Frequency (kHz)

5

Chick 11F

d

n b

4

p c

g, (h)

3 o

2 1

i, (j)

a m

e, (f)

k l

0 0.5 Time (sec) Fig. 6 Measurements obtained from sonograms of chick begging calls (left: chick 2A, right: chick 11F). See Table VIII for descriptions of acoustic variables (a–p). Reprinted from Behav. Ecol. Sociobiol., in Nakagawa et al. (2001), with kind permission of Springer Science and Business Media.

(across chicks) on 14 different acoustic variables to obtain F‐ratios for each parameter (Fig. 6, Table VIII). A discriminant‐function analysis was also carried out to determine if the 110 calls sampled from the 11 chicks (10 per chick) could be correctly allocated to the original callers (only 10 acoustic parameters were used for this analysis as four variables were either highly correlated with or a linear combination of other variables; see Nakagawa et al., 2001 for details). b. Results The interindividual CVs for all three solo call parameters we measured were higher than the intraindividual CVs, indicating that each bird had an individually distinctive call (see Miyazaki and Waas, 2003c for details). The maximum difference occurred for the highest exhalation frequency (CV ratio ¼ 4.4), while exhalation PD produced the lowest CV ratio of 1.3 (Table IX). ANOVAs conducted across males for each acoustic parameter measured from growls showed that the only significant difference between individuals was in the dominant frequency and the number of syllables per second; across individual differences in the highest frequency fell just short of being statistically significant (see Miyazaki and Waas, 2003c for details). Differences across males in call duration were not apparent.

428

JOSEPH R. WAAS

ACOUSTIC VARIABLES MEASURED

TABLE VIII EACH LITTLE BLUE PENGUIN BEGGING CALL

FOR

Acoustic variable

Description (unit of measurement)

a. Inhalation duration b. Exhalation duration c. Gap duration

Length of inhalation (sec) Length of exhalation (sec) Length of gap between inhalation and exhalation (sec) Total length of a call (a þ b þ c in seconds) Frequency (kHz) of inhalation at darkest (loudest) section of sonogram Amplitude (dB) at main inhalation frequency Frequency (kHz) of exhalation at darkest (loudest) section of sonogram Amplitude (db) at main exhalation frequency Main exhalation frequency minus main inhalation frequency (g–e in kHz) Maximum exhalation amplitude minus maximum inhalation amplitude (h–f in dB) Highest frequency of inhalation (kHz) Lowest frequency of inhalation (kHz) Maximum inhalation frequency minus minimum inhalation frequency (k–l in kHz) Highest frequency of exhalation (kHz) Lowest frequency of exhalation (kHz) Maximum exhalation frequency minus minimum exhalation frequency (n–o in kHz)

d. Total duration e. Main inhalation frequency f. Maximum inhalation amplitudea g. Main exhalation frequency h. Maximum exhalation amplitudea i. Main frequency change j. Amplitude change k. Maximum inhalation frequency l. Minimum inhalation frequency m. Inhalation frequency change n. Maximum exhalation frequency o. Minimum exhalation frequency p. Exhalation frequency change a

These variables were not used for statistical analyses. From Nakagawa et al. (2001) with kind permission of Springer Science and Business Media.

COMPARISON

TABLE IX COEFFICIENTS OF VARIATION ALONG CORRESPONDING CV RATIOSa

OF INTER‐ AND INTRAINDIVIDUAL

Exhalation

Interindividual (%) Intraindividual (%) CV ratio

WITH THE

Inhalation

DF

HF

PD

DF

HF

LF

PD

15.6 5.1 3.1

11.4 2.6 4.4

23.3 17.3 1.3

5.6 2.1 2.7

4.6 2.7 1.7

6.2 2.0 3.1

19.7 6.5 3.0

a See Fig. 2 for definitions of abbreviated terms. CV, coefficients of variation. From Miyazaki (2002).

429

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

RESULTS

OF A

SERIES

TABLE X ONE‐WAY ANOVAS ON 14 ACOUSTIC MEASURES BEGGING CALLS OF 11 CHICKS

OF

OF

Acoustic variable Frequency Maximum inhalation frequency Minimum exhalation frequency Main inhalation frequency Inhalation frequency changea Minimum inhalation frequency Exhalation frequency changea Maximum exhalation frequency Main frequency changea Main exhalation frequency Time Total duration Inhalation duration Exhalation duration Gap duration Amplitude Amplitude changeb

F10,99* 99.54 49.93 48.15 44.35 43.97 37.21 35.11 24.49 19.92 16.68 16.15 12.21 7.58 7.74

a

Frequency modulation (FM) parameters. Amplitude modulation (AM) parameter. * All F ratios were significant at the p < 0.001 level. From Nakagawa et al. (2001) with kind permission of Springer Science and Business Media. b

The series of one‐way ANOVAs on call parameters measured from chicks revealed significant variation in all 14 acoustic parameters (Table X; Nakagawa et al. 2001). The greatest interindividual versus intraindividual difference occurred for the maximum inhalation frequency. Frequency parameters had higher R‐ratios than time and amplitude parameters (Table X). The discriminant function analysis, using either the original groupings or cross‐validated groupings of calls (see Nakagawa et al., 2001 for details), showed that more than 90% of the 110 calls were successfully allocated to the original callers (Fig. 7). The percentages were much greater than those predicted by chance (ca. 9%). The results for adult males clearly indicate that growl and solo calls have individually distinctive properties, which could act as vocal signatures in a recognition system. Our analysis of chick calls suggests that signatures develop early in life, by 1 month of age. In general, frequency parameters showed greater interindividual variation than temporal parameters.

430

JOSEPH R. WAAS

Fig. 7 Scatterplot of discriminant scores for begging calls and each chick’s centroid on the first two discriminant functions. The symbols used to represent each chick (1–11), and family associations (A–F), are presented at the base of the plot. Reprinted from Behav. Ecol. Sociobiol., in Nakagawa et al. (2001), with kind permission of Springer Science and Business Media.

2. Little Blue Penguins Are Capable of Individual Recognition a. Method To determine if northern little blue penguins were capable of individual recognition [an attribute that may be important in a variety of signaling categories, e.g., intentional and convention signals (Deag and Scott, 1999)], Sarrah Winter and I used an artificial egg (based on the design of Nimon et al., 1996) to measure the heart rates of incubating birds presented with a control sound (i.e., the call of a Fiordland crested penguin, Eudyptes pachyrhyncus) or the brays of their partner, a neighbor, or a stranger (Winter, 2000; Winter and Waas, unpublished data). An infrared sensor in the egg measured changes in heart rate by detecting pulses of blood through the skin of the brood patch. Clear recordings for each treatment were obtained by positioning a microphone 1 m from the bill. On calm nights, lone incubating adults were removed from their burrow to swap their egg(s) with a warmed artificial egg. The bird’s egg(s) were placed in a portable incubator. The incubating bird was then released and quickly accepted the artificial egg. After a 30‐min settling period, a speaker positioned 1.5 m from the burrow mouth broadcast the first of the four playback treatments. Each playback consisted of three, single 15‐ to 20‐sec call segments separated by 20 sec of silence; a 10‐min recovery period

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

431

separated each of the four vocal treatments presented to each bird (the sequence of treatments was randomly assigned). A balanced ANOVA was used to determine if birds’ heart rates during playback and the 10‐min postplayback phases differed in response to the four acoustic treatments. Given that penguin chicks possess individually distinctive calls at 1 month of age (see previous section), we also attempted to determine if chicks were capable of individual recognition (Nakagawa et al., 2001). Early recognition would allow chicks to maximize the intake of information on rivals and allow chicks to start conditioning associations between its own displays and postdisplay actions. Once again, heart rate was used to assess recognition, using the northern subspecies (see Nakagawa et al., 2001 for details). A portable electric cardiograph (ECG) was used to measure changes in heart rate; contacts were placed on the chick’s back and chest. Each chick (N ¼ 11) was then placed in an artificial burrow and allowed to settle for 15 min before being exposed, in random order, to the five auditory treatments: (1) a sibling’s begging calls (N ¼ 10), (2) a neighboring chick’s begging calls (N ¼ 10), (3) a stranger chick’s begging calls, (4) the begging calls of a soft‐plumaged Petrel, Pterodroma mollis (Control 1), and (5) a piece of classical music by the Penguin Cafe´ Orchestra (Control 2). Each of the five treatments lasted 140 sec and consisted of six 7‐sec auditory segments preceded and separated by 16 sec of blank tape noise. Eight minutes of silence separated treatments. Dependent variables were obtained by subtracting the frequency of heartbeats recorded during each 140‐sec treatment period from the 140 sec preceding each treatment. A one‐way repeated measures ANOVA was then used to compare responses. b. Results Incubating penguins distinguished between the calls of mates, neighbors, strangers, and control sounds generally reducing their heart rates in comparison to the predisturbance levels (Fig. 8); the sex of the tested bird had an important influence on the response. The lowest heart rates were associated with the most familiar stimulus, the mate’s calls, with sequentially higher heart rates expressed with a reduction in familiarity, from neighbors, to strangers, to the calls of a species they would have not encountered previously. Reductions in heart rates, associated with exposure to familiar calls outside the burrow, were most pronounced in females and, while males showed a similar response pattern to the calls of conspecifics, their heart rates varied much less from predisturbance or control conditions (Fig. 8). However, when we examined data from males and females separately, significant treatment effects remained for both sexes. A Tukey’s test indicated that females discriminated between the two types of calls that they would have heard before, mates and neighbors, but their response to neighbors and strangers was not significantly different;

432

JOSEPH R. WAAS

Fig. 8 The mean (SE) cardiac response of male (solid) and female (hatched) incubating little blue penguins in response to predisturbance conditions and playback treatments (adapted from Winter, 2000; Winter and Waas, unpublished data).

the Tukey’s test revealed no across‐treatment differences in male heart rates. Behavioral responses, such as vocalizations, movements toward the speaker, or offensive activities were never observed. Our study of heart rate changes of chicks in response to the playback of different calls revealed that little blue penguins may also have sophisticated recognition skills, from as early as 1 month of age. Chick heart rates, normally between 205 and 238 beats per minute (bpm), increased during sibling and neighbor treatments, but decreased during stranger, petrel call and music treatments (Fig. 9). A least significant difference (LSD) test showed that chicks had significantly higher heart rates when hearing sibling calls than when hearing either the calls of an unfamiliar little blue penguin chick or the calls of a petrel (Fig. 9). However, differences in the chick’s response to the two types of calls that would have been familiar to it, sibling and neighbor calls, were not detected (Fig. 9; see Nakagawa et al., 2001 for details). Little blue penguins appear to be capable of individual recognition, which requires animals to detect differences between familiar conspecifics

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

433

Fig. 9 Heart rate changes (mean  SE) between 140‐sec periods before and during the five vocal treatments chicks experienced. Bars not sharing any letters are significantly different (LSD, p < 0.05). Reprinted from Behav. Ecol. Sociobiol. in Nakagawa et al. (2001), with kind permission of Springer Science and Business Media.

of a given social category (see Waas and Colgan, 1994). Adult females were able to discriminate between the calls of two mature territorial males (i.e., their mate and the neighboring male) who were familiar to them; the drop in heart rate associated with hearing familiar calls probably reflected a ‘‘relief’’ response [i.e., sounds associated with the researcher’s presence near the burrow before playback may have created a fearful situation (e.g., suggesting a potential predator) and raising heart rate—the subsequent perception of a familiar, unthreatening stimulus might then have dropped the heart rate]. The ability to recognize individuals appears to be shared by males and chicks as young as 30 days of age, as they also discriminated between treatments, although we did not detect significant differences in their reaction to the two familiar stimuli. However, the ability of chicks to distinguish siblings from unfamiliar chicks suggests that little blue penguins have sophisticated recognition processes that develop early in life. 3. Do Penguin Displays Reveal Information About ‘‘Intentions’’? a. Method I used call playback and stuffed penguin models to determine if little blue penguins use signals to provide reliable information about their postdisplay actions (see Waas, 1991b for details). The study was conducted over two field seasons in a colony of banded, known age and sex white‐ flippered penguins on the Banks Peninsula, New Zealand. To simulate an intrusion, I played the solo call of a male 2 m from the burrow entrance at

434

JOSEPH R. WAAS

an amplitude equivalent to that from a displaying bird. I then remained still for 30 sec and noted any threat calls or other activities displayed by the occupant. At the end of the 30 sec, I moved the stuffed model to within 30 cm of the burrow entrance to simulate an approach by the stranger. After waiting a further 30 sec, I noted the occupants’ subsequent reaction to the model. The identity of the occupant(s) was confirmed after the test was over. A total of 783 experiments were conducted (during night) on lone males (N ¼ 40), lone females (N ¼ 37), or pairs (N ¼ 59) defending burrows. Lone birds were members of pairs, but their partners were at sea when the tests were run. To avoid habituation, six different solo call recordings were used to simulate the initial presence of an intruder; these were selected at random before each test, but no subjects were tested with the same solo call more than twice in a given breeding season. For each category of occupant, I determined the percentage of cases that each response to the model (i.e., attack, lunge/hiss, vocalize or remain silent) was preceded by each of the responses displayed to playback at the initiation of each test (i.e., lunge/ hiss, bray, growl, or remain silent). I then used 4  4 tests of independence to determine whether a penguins’ choice of threat behavior (given in response to the initial solo call playback) was independent of its choice of post‐threat behavior (given in response to the penguin model; see Waas, 1991b for details). b. Results The signals penguins used on hearing the call of an intruder revealed information on what they were likely to do when an intruder model was moved toward their burrow (see Waas, 1991b for details), that is, their choice of threat behavior was not independent of their choice of post‐threat behavior for males, females, or pairs of penguins (Fig. 10). The values shown in Fig. 10 (calculated by pooling data within each group) were similar to those obtained using the mean values of each individual (the average deviation was <2.5% for each group). Thus, the relationship between an animal’s choice of threat behavior and its subsequent actions were consistent across individuals. Solitary males that hissed were significantly more likely to attack or peck the intruder than those that growled or brayed (Fig. 10); further, males that growled were more likely to attack or peck than those that remained silent. Similarly, pairs of penguins that hissed were significantly more likely to attack or peck the model than pairs that growled, brayed, or remained silent (Fig. 10). In total contrast, solitary females rarely attacked models, and those that at least lunged at the model were most likely to do so if their initial response was to remain silent (Fig. 10). Thus, males or pairs of penguins that hissed revealed that they are highly likely to attack an

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

435

Fig. 10 Percentage of cases in which growl, bray, lunge/hiss (L/H), and silence led to overt aggression, lunge/hiss, vocalizations (growl and bray), or silence when the model was moved toward the burrow. Vertical comparisons (i.e., follow bars downward) indicate whether the initial threats (i.e., responses to playback) differed in how often they lead to overt aggression (or other behavior directed toward the model). Bars not sharing any letter are significantly different (p < 0.05, tests of independence). Reprinted from Anim. Behav. 41, in Waas (1991b), with permission from Elsevier.

intruder that moved toward their burrow. Growls (and perhaps brays as well; Fig. 10) revealed an intermediate attack propensity for males only. In no case did the two graded vocalizations used by little blue penguins (growl and bray) differ in what they allowed intruders to predict about the subsequent behavior of the burrow occupant. 4. Conclusions The ‘‘social conditioning hypothesis’’ adopts the principles of associative learning to explain how learning processes and individual recognition can modify the behavior of animals engaged in agonistic interactions (see for

436

JOSEPH R. WAAS

example Miklosi et al., 1997). Much ethological research on this topic has been conducted on fish (e.g., Bakker et al., 1989; Francis, 1983; McDonald et al., 1968; Miklosi et al., 1997), where there is clear evidence for individual recognition (Myrberg and Riggio, 1985; Peek and Veno, 1973; Waas and Colgan, 1994); the work suggests that pain inflicted by a recognizable opponent can become associated with the opponent’s individually distinctive features, allowing animals to develop or foster ‘‘reputations’’ (see also Caldwell, 1986 for similar work on stomatopods). A more advanced form of this process is also hypothesized to occur—animals may recognize sets of individually distinctive signals provided by a familiar opponent and associate specific consequences with the different signals in the set, as long as the opponent maintained consistent signal‐consequence associations (e.g., Bossema and Burgler, 1980; van Rhijn, 1980; van Rhijn and Vodegel, 1980). Such a process is critical to understanding how signals of intent can be evolutionarily stable. Our research shows that little blue penguins have all the tools required to use social conditioning to gain benefits from an intentional signaling system: (1) The birds live in stable burrow or cave communities and interact regularly with other penguins in the group (Waas, 1990a,b). (2) They also possess individually distinctive acoustic signatures that would allow opponents to clearly distinguish one familiar opponent from another (Miyazaki and Waas, 2003a,c). These acoustic signatures also develop very early in life, so individuals have a large window of opportunity to learn about the individuals around them (Nakagawa et al., 2001). (3) Little blue penguins are capable of using individually distinctive cues to distinguish between familiar individuals (Winter, 2000; Winter and Waas, unpublished data), and this process, like the development of acoustic signatures, is initiated early in life (Nakagawa et al., 2001). (4) Finally, little blue penguins appear to use their individually distinctive displays in a way that suggests that they signal their intentions during aggressive interactions (Waas, 1991b). By hissing, males and pairs revealed that they were more likely to attack an intruder who continued to approach than those that brayed or growled; further, growling males were more likely to attack than those that remained silent. Thus, little blue penguins appear to use signals to reveal information on their intentions. An important problem, however, exists for the interpretation provided in the preceding paragraph. While the vocalizations used to simulate intrusions were from the same population of little blue penguins on the Banks Peninsula, none were actually from the colony where the birds were tested. As a result, it seems unlikely that burrow occupants were familiar with the simulated intruders. Signals of intent may not be evolutionarily stable in situations where opponents are usually unfamiliar because such a signaling

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

437

system would become susceptible to bluff (Maynard Smith, 1982; Maynard Smith and Parker, 1976). However, in the stable colonies formed by penguins, strangers quickly and regularly become familiar members of the colony. Under these socioecological conditions, it may benefit animals to signal their intentions reliably regardless of whether the intruder is a stranger or a familiar rival. Basically, each bird must ‘‘start somewhere’’ to establish its reputation and signal‐consequence associations with a new acquaintance, potentially explaining why reliable information on intentions was provided to strangers. Such a process may remain evolutionarily stable if the recognition process allowed animals to quickly detect and discriminate against ‘‘cheats,’’ who provided bogus information during agonistic interactions. In some respects, this process may be similar to that envisaged in conventional signaling systems, where low‐cost displays are validated by receiver retaliation in future contests rather than through past or ‘‘real‐ time’’ investments in signal production (Molles and Vehrencamp, 2001). Signals of intent may help to explain how conventional signals may have evolved in the first place. An alternative explanation for my discovery that penguins signal accurate information on their intentions to strangers (despite the potentially unstable nature of such a signaling system from a Game Theory perspective) exists. Only the lunge/hiss display predicted attack significantly more often than the other threat displays used by male or pairs of little blue penguins. The low and high intensities of the graded signal (growl and bray; Fig. 1) did not differ in what they would allow an opponent to predict about the occupant’s subsequent behavior. As discussed earlier, lunge/hiss is actually a ‘‘distance‐reducing behavior,’’ and would be expected to entail greater risks than the stationary behavior associated with the graded displays because the occupant flings itself forward, exposing its head at the mouth of the burrow entrance. The greater risk associated with hissing may help explain why it appears to be a better predictor of attack. As outlined earlier in this chapter, these penguins may settle disputes by using an escalation process during which the animals progress step‐by‐step from low‐ to high‐risk displays (Waas, 1990b). Escalation continues until a fight ensues or until one of the two animals is no longer willing to take further risks and backs down. The process of progressive escalation could explain why hiss lead to attacks more frequently than growl or bray. If a risk‐based escalation process exists, attacks should be preceded by hisses and hisses should be preceded by the growl and bray calls associated with the stationary displays (Table IV). As expected, males and pairs that hissed were more likely to attack the simulated intruder when it was placed near the burrow entrance than those that growled or brayed. However, males and pairs that growled or brayed did not progress to the lunge/hiss displays as would be predicted

438

JOSEPH R. WAAS

by the idea of progressive escalation; furthermore, females were more likely to escalate to lunge/hiss if they remained silent initially than if they produced a graded call. Thus, progressive escalation may not provide a satisfactory explanation for the results. At least for males, threat displays appear to provide intruders with an attack probability: a silent occupant attacked approaching intruders in ca. 20% of cases, while growls and brays led to attack in 40% of cases, and hisses in 80% of cases. Little blue penguins provide information on attack probabilities to complete strangers.

IV. INVESTMENT STRATEGIES VALIDATING SIGNALS

AND

SIGNAL SYNERGY

I have argued that little blue penguins and other animals must corroborate or validate the information in their displays by shouldering costs or risks to ensure that perceivers remain attentive and respond accordingly (Zahavi, 1977), without such costs, bluff would destabilize the signaling system (Maynard Smith, 1982). The way in which signal costs are bourn by the displaying animal may vary across displays in the animal’s repertoire. In this chapter, I have considered four types of signals that little blue penguins use during contests with rivals: RHP, war of attrition, risk‐based, and intentional signals. The ‘‘landscape’’ of energy investments validating each signal, from the perceiver’s perspective, differs for the four types of displays (Table XI). In the following paragraphs, I distinguish between the four types of displays on the basis of how animals pay the costs associated with validating their signals, before going on to consider how the different types of information available in the displays may be considered in concert by perceivers. RHP displays simply advertise or highlight attributes that affect fighting ability, which the animal has been able to develop overtime as the result of past effort (Table XI). For example, the bills of adult male blackbirds (Turdus merula) range from yellow to deep orange, depending on how successful the bird has been in obtaining carotenoid‐rich food from its environs (Bright et al., 2004; Faivre et al., 2003). During male–male contests, rivals find orange‐billed opponents more threatening (Bright and Waas, 2002) probably because the orange color provides an honest account of the signaler’s ability to successfully maintain a large territory and forage effectively in the past (birds cannot manufacture carotenoids—they must gain them from the environment). While the primary factor making displays of RHP effective from the perceiver’s perspective is past investments made by the signaler, signalers would also incur maintenance costs to ensure that they retain condition for future signaling episodes (Table XI).

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

439

TABLE XI PRIMARY ENERGY INVESTMENTS FOR FOUR CATEGORIES OF SIGNALS USED BY LITTLE BLUE PENGUINS AND ASSOCIATED MAINTENANCE COSTS Investment regime Signal type

Past

Present

Future

RHP Intent War of attrition Risk based

PRIMARY PRIMARY Maintenance –

Maintenance Maintenance PRIMARY PRIMARY

Maintenance Maintenance Maintenance –

War of attrition signals, unlike signals of RHP, are primarily corroborated by costs associated with energy or time expenditure experienced during the contest (Table XI). The winner must engage in an energetically expensive activity to demonstrate its superiority or greater motivation to secure the resource. For example, during fights for shells, hermit crabs (Pagurus bernhardus) will rap their shells vigorously and repeatedly against the opponent’s shell—clearly, an energetically costly activity. Contests are settled on the basis of which animal can rap more vigorously and persistently (Briffa et al., 1998); the winner demonstrates in ‘‘real time’’ during the contest that it can exhaust the opponent. Like signals of RHP, signals involving a war of attrition have maintenance costs. While the primary factor contributing to the signal’s success is the costly display of stamina during the contest, stamina that must be built up and retained; thus, maintenance costs would be expected through a degree of past investments and must continue to be paid in the future if the signals are to remain effective. Risk‐based signals, like signals associated with wars of attrition, are validated by costs experienced during the actual conflict (Table XI). Unlike war of attrition signals, the costs are associated with the real or potential risks of social remediation during the contest, not necessarily energetic costs. For example, Mediterranean field crickets (Gryllus bimaculatus) engage in agonistic encounters involving a stereotyped escalation sequence from ritualized displays in the beginning of the encounter to physical combat toward the end (Hofmann and Schildberger, 2001). The risk level to which an animal is willing to escalate would provide accurate information on the animal’s motivation to secure the resource because the cost of social remediation prevents animals from bluffing. Unlike all the signals discussed earlier, risk‐based signals have no maintenance costs (Table XI). An animal of any size or ability can provide a realistic account of its

440

JOSEPH R. WAAS

motivation to secure a resource by advertising its willingness to expose itself to risks during the contest—no past or future maintenance costs are necessarily incurred. Signals of intent, like RHP signals, are validated or corroborated by past investments; however, instead of having invested in physical attributes that improve their fighting ability, animals invest in their reputation as fighters or honest signalers (Table XI). For example, mantis shrimps (Gonodactylus bredini) live in small, stable communities of burrows on coral reefs. Each stomatopod has an individually distinctive odor that rivals learn to associate with the outcome of past interactions; individuals will actively avoid burrows containing the odor of animals that have regularly defeated them, but will invade those that they have regularly defeated in past encounters (reviewed by Caldwell, 1986). In some species, rivals may also learn to associate different displays in the signaler’s repertoire with different consequences, providing the signaler consistently linked each signal with a specific consequence (van Rhijn, 1980; van Rhijn and Vodegel, 1980). While the primary factor making signals of reputation or intent effective from the perceiver’s perspective is past investments made by the signaler (through social conditioning), signalers would have to maintain the display‐consequence associations by continuing to invest in honest signaling during current and future contests (Table XI). Over the course of this chapter, I have considered the four types of signals used by little blue penguins as separate entities. However, a given signal may convey several types of information synergistically. For example, I have argued that the calls used by penguins during agonistic displays act to highlight or advertise the risks that the actor is willing to take during contests that escalate in risk. One call commonly used in these contests is the bray call. As demonstrated earlier, the pitch of the bray call also reveals information on the size of the displayer. Thus, during the encounter the rival obtains information not only on the risks the signaler is willing to take but also on the size of the signaler. The rival may consider the information in concert—for example, a small opponent taking high risks may be perceived as less of a threat than a large opponent taking high risks. Similarly, male penguins calling for females in an acoustic war of attrition may take note of the pitch of their competitors’ bray calls. A small male may avoid calling with low‐pitched competitors (e.g., by positioning itself elsewhere along the shoreline), as low pitch reveals large size, and large birds are likely to have greater stamina and be preferred by any available females. The depth of the social relationship between rivals is also likely to influence information synergy in displays. As outlined earlier, little blue penguins are capable of individual recognition and live in stable communities. Therefore, individuals have opportunities to establish reputations or

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

441

to condition associations between their displays and postdisplay consequences; this may influence their perception of other information available in the displays. For example, a small opponent as revealed by its high‐ pitched calls may be perceived as a greater threat than one judged by call pitch to be much larger, if the small opponent has repeatedly demonstrated greater willingness to attack following the display than the larger individual. Thus, to assess the threat posed by an opponent, information on size derived from assessing pitch must be used in concert with information on the recognized individual’s temperament during previous conflicts. Recognizing that signals may convey several types of information simultaneously, and that the depth of social relationships influences how signals are used, may help to explain regular observations of a ‘‘lack of fit’’ between theoretical models and the results of empirical studies. It should not be surprising, for example, that support for the risk‐benefit model is not universal (e.g., Lange and Leimar, 2003; Wilson, 1994)—risk advertisement may only represent one factor explaining a display’s effectiveness. Other factors contributing to the display’s effect, like information on RHP or the recognition of reputations, may need to be considered in concert to truly understand how signals are used during the contests. It is now time to develop theoretical models that can cope more realistically with the complexities of communication processes occurring during agonistic interactions (Jennings et al., 2005). Recognizing the possibility of information overlap and synergies associated with such overlap may not only help establish more realistic models of animal conflicts, but also help us come to a better understanding of existing empirical information.

V. SUMMARY Everyone, soon or late, sits down to a banquet of consequences. Robert Louis Stevenson

Little blue penguins improve their ability to secure or defend resources by using agonistic displays to manage the activities of rivals during disputes over resources. (1) RHP‐based signals reveal reliable information on body mass, an attribute that is likely to correlate positively with fighting ability. Larger birds had lower pitched braying calls and longer phrases than smaller birds. Perceivers should be attentive to this relationship because bluff is totally precluded: small animals would be physically incapable of producing a signal that misrepresents their size. (2) Energetic wars of attrition are used by male penguins engaged in intrasexual competitions for females. Males that call most, over long periods of time, enhance their

442

JOSEPH R. WAAS

chances of being detected by females approaching breeding colonies from the sea. By using playback to artificially exaggerate calling rates within colonies, I showed that males adjust their calling rates to ‘‘meet the challenge’’ as would be expected in a war of attrition. Males in top condition may also be able to afford to use their vocalizations to jam the signals of potential competitors. Playback experiments investigating call overlap showed that females were attracted to males that regularly overlapped the signal of a nearby rival, perhaps because males that can afford to jam the signals of other males in addition to maintaining their own calling effort are viewed as being superior males. (3) Risk‐based signals have been best studied in little blue penguins—the aggressive display repertoire varies along a gradient from low‐ to high‐risk displays that precede overt aggression. Penguins escalate step‐by‐step up the risk gradient until the opponent is seen off or until the actor is no longer willing to take further risks. The conspicuous components of displays (elaborate postures and calls) may simply function to make the opponent aware of the risks the actor is taking. Perceivers should respect the actor’s signal because the actor exposes itself to the possibility of social remediation by performing risky displays. Accordingly, I found that a display’s effectiveness in deterring opponents was positively correlated with risk. (4) Signals revealing intent are also used by little blue penguins. I found that the penguins paired the different individually distinctive displays in their repertoire with different postdisplay consequences: birds that hissed at a rival followed the threat with attack in 80% of cases, while those that growled attacked the rival in 40% of cases and those that remained silent attacked in 20% of cases. Little blue penguins may condition opponents to associate different displays (conditional stimuli) with specific consequences (unconditional stimuli) through Pavlovian conditioning. Perceivers should be attentive to such signals because the actor has demonstrated a consistent relationship between its signals and its postdisplay actions—essentially creating a reputation for reliability. Actors invest energy to validate or corroborate information in their displays. Signals of RHP and intent are validated primarily through past investments, in the form of time and energy investments associated with gaining physical attributes that improve fighting ability and social conditioning, respectively. War of attrition and risk‐based signals are corroborated primarily by costs or risks that animals incur in real time during the course of a contest. While the four types of signals used by little blue penguins appear to operate in different ways, it is clear that a given signal may present several types of information simultaneously. Recognizing the possibility of information overlap and synergies associated with such overlap may not only help establish more realistic models of animal conflicts, but also help us come to a better understanding of existing empirical information.

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

443

Acknowledgments I would like to acknowledge my colleagues and graduate students, Masamine Miyazaki, Shinichi Nakagawa, Sarah Winter, Jonathon Banks, and Alex Eagles, for their important individual contributions to the research on little blue penguins described in this chapter. Support for the work was provided by the Department of Biological Sciences at the University of Waikato, the Department of Biology at Queen’s University, and the Department of Zoology at the University of Canterbury. Chris Challies is gratefully acknowledged for sharing his knowledge of little blue penguins and providing valuable suggestions during the field component of my early work. Marc Naguib, Jane Brockmann, Tim Roper, Peter Slater, and an anonymous referee provided valuable comments on earlier versions of this chapter. The New Zealand Department of Conservation provided the permits to conduct the research and the University of Waikato Animal Ethics Committee approved the described protocols.

REFERENCES Adams, E. L., and Mesterton‐Gibbons, M. (1995). The cost of threat displays and the stability of deceptive communication. J. Theor. Biol. 175, 405–421. Andersson, M. (1976). Social behaviour and communication in the great skua. Behaviour 49, 40–77. Andersson, M. (1980). Why are there so many threat displays? J. Theor. Biol. 86, 773–781. Archer, J. (1988). ‘‘The Behavioural Biology of Aggression.’’ Cambridge University Press, Cambridge. Badyaev, A. V., and Leaf, E. S. (1997). Habitat associations of song characteristics in Phylloscopus and Hippolais warblers. Auk 114, 40–46. Bakker, Th. C. M., De Bruijn, E. F., and Sevenster, P. (1989). Asymmetrical effects of prior winning and losing on dominance in sticklebacks (Gasterosteus aculeatus). Ethology 82, 224–229. Banks, J. C., Mitchell, A. D., Waas, J. R., and Paterson, A. M. (2002). An unexpected pattern of molecular divergence within the blue penguin (Eudyptula minor) complex. Notornis 49, 29–38. Bishop, A. T., and Cannings, C. (1978). A generalized war of attrition. J. Theor. Biol. 70, 85–125. Bosch, J., and Marquez, R. (2000). Acoustical interference in the advertisement calls of the midwife toads (Alytes obstetricans and Alytes cisternasii). Behaviour 137, 249–263. Bossema, I., and Burgler, R. R. (1980). Communication during monocular and binocular looking in European jays (Garrulus g. glandarius). Behaviour 74, 274–283. Bradbury, J. W., and Vehrencamp, S. L. (1998). ‘‘Principles of Animal Communication.’’ Sinauer Associates, Sunderland, MA. Bradbury, T. N., and Fincham, F. D. (1991). The analysis of sequence in social interaction. In ‘‘Personality, Social Skills, and Psychopathology: An Individual Differences Approach’’ (D. G. Gilbert and J. J. Conley, Eds.), pp. 257–290. Plenum Press, New York. Brandt, Y. (2003). Lizard threat display handicaps endurance. Proc. Roy. Soc. Lond. B 270, 1061–1068. Breithaupt, T., and Atema, J. (2000). The timing of chemical signalling with urine in dominance fights of male lobsters (Homarus americanus). Behav. Ecol. Sociobiol. 49, 67–78. Breithaupt, T., and Eger, P. (2002). Urine makes the difference: Chemical communication in fighting crayfish made visible. J. Exp. Biol. 205, 1221–1231.

444

JOSEPH R. WAAS

Briffa, M., Elwood, R. W., and Dick, J. T. A. (1998). Analysis of repeated signals during shell fights in the hermit crab Pagurus bernhardus. Proc. R. Soc. Lond. B 265, 1467–1474. Bright, A., and Waas, J. R. (2002). Effects of bill pigmentation and UV reflectance during territory establishment in blackbirds. Anim. Behav. 64, 207–213. Bright, A., Waas, J. R., King, C. M., and Cuming, P. D. (2004). Bill colour and correlates of male quality in blackbirds: An analysis using canonical ordination. Behav. Processes 65, 123–132. Caldwell, R. L. (1986). The deceptive use of reputation by stomatopods. In ‘‘Deception: Perspective on Human and Nonhuman Deceit’’ (R. W. Mitchel and N. S. Thompson, Eds.), pp. 129–145. State University of New York Press, Albany. Campbell, R. C. (1989). ‘‘Statistics for Biologists.’’ Cambridge University Press, Cambridge. Clutton‐Brock, T. H., and Albon, S. D. (1979). The roaring of red deer and the evolution of honest advertisement. Behaviour 69, 145–170. Dabelsteen, T., McGregor, P. K., Holland, J., Tobias, J. A., and Pedersen, S. B. (1997). The signal function of overlapping singing in male robins. Anim. Behav. 53, 249–256. Dann, P., Cullen, J. M., and Jessop, R. (1995). Cost of reproduction in little penguins. In ‘‘The Penguins: Ecology and Management’’ (P. Dann, F. I. Norman, and P. Reilly, Eds.), pp. 39–55. Surrey Beatty and Sons, Sydney. Davies, N. B., and Halliday, T. R. (1978). Deep croaks and fighting assessment in toads Bufo bufo. Nature 274, 683–685. Deag, J. M., and Scott, G. W. (1999). ‘‘Conventional’’ signals in avian agonistic displays: Integrating theory, data and different levels of analysis. J. Theor. Biol. 196, 155–162. Enquist, M., Plane, E., and Roed, J. (1985). Aggressive communication in fulmars (Fulmarus glacialis) competing for food. Anim. Behav. 33, 1007–1020. Faivre, B., Gregoire, A., Preault, M., Cezilly, F., and Sorci, G. (2003). Immune activation rapidly mirrored in a secondary sexual trait. Science 300, 103. Farr, J. A. (1976). Social facilitation of male sexual behaviour, intrasexual competition, and sexual selection in the guppy, Poecilia reticulate (Pisces: Poeciliidae). Evolution 30, 707–717. Fitch, W. T., and Hauser, M. D. (2002). Unpacking ‘‘honesty’’: Vertebrate vocal production and the evolution of acoustic signals. In ‘‘Acoustic Communication’’ (A. Simmons, A. N. Popper, and R. R. Fay, Eds.), pp. 65–137. Springer, New York. Francis, R. C. (1983). Experiential effects on agonistic behaviour in the paradise fish, Macropodus opercularis. Behaviour 85, 292–313. Grafen, A. (1990). Biological signals as handicaps. J. Theor. Biol. 144, 517–546. Greenfield, M. D., and Roizen, I. (1993). Katydid synchronous chorusing is an evolutionarily stable outcome of female choice. Nature 364, 618–620. Greenfield, M. D., Tourtellot, M. K., and Snedden, W. A. (1997). Precedence effects and the evolution of chorusing. Proc. R. Soc. Lond. B 264, 1355–1361. Hammerstein, P., and Parker, G. A. (1982). The asymmetric war of attrition. J. Theor. Biol. 96, 647–682. Hansen, A. J. (1986). Fighting behaviour in bald eagles: A test of game theory. Ecology 67, 787–797. Hocken, A. G. (2000). Internal organ weights of the blue penguin Eudyptula minor. New Zeal. J. Zool. 27, 299–304. Hofmann, H. A., and Schildberger, K. (2001). Assessment of strength and willingness to fight during aggressive encounters in crickets. Anim. Behav. 62, 337–348. Hultsch, H., and Todt, D. (1982). Temporal performance roles during vocal interactions in nightingales (Luscinia megarhynchos). Behav. Ecol. Sociobiol. 11, 253–260.

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

445

Husak, J. F. (2004). Signal use by collared lizards, Crotaphytus collaris: The effects of familiarity and threat. Behav. Ecol. Sociobiol. 55, 602–607. Husak, J. F., and Fox, S. F. (2003). Adult male collared lizards, Crotaphytus collaris, increase aggression towards displaced neighbours. Anim. Behav. 65, 391–396. Jennings, D. J., Gammell, M. P., Carlin, C. M., and Hayden, T. J. (2002). Does lateral presentation of the palmate antlers during fights by fallow deer (Dama dama L.) signify dominance or submission? Ethology 108, 389–401. Jennings, D. J., Gammell, M. P., Payne, R. J. H., and Hayden, T. J. (2005). An investigation of assessment games during fallow deer fights. Ethology 111, 511–525. Jouventin, P. (1982). ‘‘Visual and Vocal Signals in Penguins, their Evolution and Adaptive Characters.’’ Paul Parey, Berlin. Kemp, D. J. (2002). Butterfly contests and flight physiology: Why do older males fight harder? Behav. Ecol. 13, 456–461. King, A. S., and Molony, V. (1971). The anatomy of respiration. In ‘‘Physiology and Biochemistry of Domestic Fowl’’ (D. J. Bell and B. M. Freeman, Eds.), Vol. 1, pp. 93–169. Academic Press, London. Lange, H., and Leimar, O. (2003). The function of threat display in wintering great tits. Anim. Behav. 65, 573–584. Maynard Smith, J. (1982). Do animals convey information about their intentions? J. Theor. Biol. 97, 1–5. Maynard Smith, J., and Parker, G. A. (1976). The logic of asymmetric contests. Behaviour 24, 159–175. McDonald, A. L., Heimstra, N. W., and Damkot, D. K. (1968). Social modification of agonistic behaviour in fish. Anim. Behav. 16, 437–441. Miklosi, A., Haller, J., and Csanyi, V. (1997). Learning about the opponent during aggressive encounters in paradise fish (Macropodus opercularis L.): When it takes place? Behav. Processes 40, 97–105. Miyazaki, M. (2002). Relationships among male size, reproductive success, female choice and male advertising calls in little penguins (Eudyptula minor). PhD Thesis, University of Waikato, Hamilton, New Zealand. Miyazaki, M., and Waas, J. R. (2002). ‘‘Last word effects’’ of male advertising calls on female preference in little blue penguins. Behaviour 139, 1413–1423. Miyazaki, M., and Waas, J. R. (2003a). Correlations between body size, defensive behaviour and reproductive success in male little blue penguins Eudyptula minor: Implications for female choice. Ibis 142, 645–656. Miyazaki, M., and Waas, J. R. (2003b). Influence of parental body size on sea to nest distances and food provisioning in little penguins Eudyptula minor. Emu 103, 239–243. Miyazaki, M., and Waas, J. R. (2003c). Acoustic properties of male advertisement and their impact on female responsiveness in little penguins Eudyptula minor. J. Avian Biol. 34, 229–232. Miyazaki, M., and Waas, J. R. (2005). Effects of male call pitch on female behaviour and mate fidelity in little penguins. J. Ethol. 23, 167–171. Molles, L. E., and Vehrencamp, S. L. (2001). Songbird cheaters pay a retaliation cost: Evidence for auditory conventional signals. Proc. R. Soc. Lond. B 268, 2013–2019. Myrberg, A. A., and Riggio, R. J. (1985). Acoustically mediated individual recognition by a coral reef fish (Pomacentus partitus). Anim. Behav. 33, 411–416. Naguib, M., and Todt, D. (1997). Effects of dyadic vocal interactions on other conspecific receivers in nightingales. Anim. Behav. 54, 1535–1543.

446

JOSEPH R. WAAS

Nakagawa, S., Waas, J. R., and Miyazaki, M. (2001). Heart rate changes reveal that little blue penguin chicks (Eudyptula minor) can use vocal signatures to discriminate familiar from unfamiliar chicks. Behav. Ecol. Sociobiol. 50, 180–188. Nelson, D. A. (1984). Communication of intentions in agonistic contexts by the pigeon guillemot, Cepphus columba. Behaviour 88, 145–187. Nimon, A. J., Schroter, R. C., and Oxenham, R. K. (1996). Artificial eggs: Measuring heart rate and effects of disturbance in nesting penguins. Physiol. Behav. 60, 1019–1022. Parker, G. A. (1974). Assessment strategy and the evolution of fighting behavior. J. Theor. Biol. 47, 223–243. Payne, R. J. H., and Pagel, M. (1996). Escalation and time costs in displays of endurance. J. Theor. Biol. 183, 185–193. Payne, R. J. H., and Pagel, M. (1997). Why do animals repeat displays? Anim. Behav. 54, 109–119. Peek, H. V. S., and Veno, A. (1973). Stimulus specificity of habituated aggression in three‐ spined sticklebacks (Gasterosteus aculeatus L.). Behav. Biol. 8, 427–432. Popp, J. W. (1987a). Risk and effectiveness in the use of agonistic displays by American goldfinches. Behaviour 103, 141–156. Popp, J. W. (1987b). Agonistic communication among wintering purple finches. Wilson Bull. 99, 97–100. Rice, W. R. (1989). Analyzing tables of statistical tests. Evolution 43, 223–225. Ryan, M. J., and Brenowitz, E. A. (1985). The role of body size, phylogeny, and ambient noise in the evolution of bird song. Am. Nat. 126, 87–100. Sackett, G. P. (1979). The lag sequential analysis of contingency and cyclicity in behavioural interaction research. In ‘‘Handbook of Infant Development’’ (J. D. Osofsky, Ed.), pp. 623–652. Wiley, New York. Sanvito, S., and Galimberti, F. (2003). Source level of male vocalisations in the genus Mirounga: Repeatability and correlates. Bioacoustics 14, 47–59. Sokal, R. R., and Rohlf, F. J. (1981). ‘‘Biometry. The Principles and Practice of Statistics in Biological Research.’’ Freeman, New York. Titze, I. R. (1994). ‘‘Principles of Voice Production.’’ Prentice Hall, Hemel Hampstead. Tubaro, P. L., and Mahler, B. (1998). Acoustic frequencies and body mass in New World doves. Condor 100, 54–61. van Rhijn, J. G. (1980). Communication by agonistic displays: A discussion. Behaviour 74, 284–293. van Rhijn, J. G., and Vodegel, R. (1980). Being honest about one’s intentions: An evolutionarily stable strategy for animal conflicts. J. Theor. Biol. 85, 623–641. Waas, J. R. (1988). Acoustic displays facilitate courtship in little blue penguins, Eudyptula minor. Anim. Behav. 36, 366–371. Waas, J. R. (1990a). Intraspecific variation in social repertoires: Evidence from cave‐ and burrow‐dwelling little blue penguins. Behaviour 115, 63–98. Waas, J. R. (1990b). An analysis of communication during the aggressive interactions of little blue penguins (Eudyptula minor). In ‘‘Penguin Biology’’ (L. S. Davis and J. T. Darby, Eds.), pp. 345–376. Academic Press, San Diego. Waas, J. R. (1991a). The risks and benefits of signalling aggressive motivation: A study of cave‐ dwelling little blue penguins. Behav. Ecol. Sociobiol. 29, 139–146. Waas, J. R. (1991b). Do little blue penguins signal their intentions during aggressive interactions with strangers? Anim. Behav. 41, 375–382. Waas, J. R. (1995). Social stimulation and reproductive schedules: Does the acoustic environment influence the egg‐laying schedule in penguin colonies? In ‘‘The Penguins: Ecology

‘‘VALIDATIONS’’ FOR PENGUIN AGONISTIC DISPLAYS

447

and Management’’ (P. Dann, F. I. Norman, and P. Reilly, Eds.), pp. 111–137. Surrey Beatty and Sons, Sydney. Waas, J. R., and Colgan, P. W. (1994). Male sticklebacks can distinguish between familiar rivals on the basis of visual cues alone. Anim. Behav. 47, 7–13. Waas, J. R., Caulfield, M., Colgan, P. W., and Boag, P. T. (2000). Colony sound facilitates sexual and agonistic activities in royal penguins. Anim. Behav. 60, 77–84. Waas, J. R., Colgan, P. W., and Boag, P. T. (2005). Playback of colony sound alters the breeding schedule and clutch size in zebra finch (Taeniopygia guttata) colonies. Proc. R. Soc. Lond. B 272, 383–388. Wallschlager, D. (1980). Correlation of song frequency and body weight in passerine birds. Experientia 36, 412. Wilson, J. M. (1994). Variation in initiator strategy in fighting by silvereyes. Anim. Behav. 47, 153–162. Winter, S. J. A. (2000). Heart rate as a measure of individual recognition in little blue penguins (Eudyptula minor). MSc Thesis, University of Waikato, Hamilton, New Zealand. Zahavi, A. (1977). Reliability in communication systems and the evolution of altruism. In ‘‘Evolutionary Ecology’’ (B. Stonehouse and M. C. Perrins, Eds.), pp. 253–259. Macmillan, London.

This page intentionally left blank

Index

A A. mellifera, 331–332 Achroia grisella, 274 Acorn woodpecker, 201 Acrocephalus sechellensis, 373 Adaptive fastidiousness, 239 Adaptive gullibility, 239 Aegithalos caudatus, 356 Alerting signal, 232–233 Alloparental care, 29 Alzheimer’s disease, 43 Amniotic fluid, 10, 13 -coated breasts, 12 Amniotic sac, 1 Amphetamine injections, 35 Amphibian papilla, 264 Amyna natalis, 274 Animal foraging behavior choice of route and destination feeding state and route memories, 157–160 olfactory cues and route mechanisms, 163–164 priming by reward, 164 time of day and route memories, 160–163 versatile decision making, 164–165 foraging routes, 124–127 navigational memories, 127–129 retrieval of memories, along route binding with snapshots, 149–151 interference effects without cues, 154–155 in panoramic contexts, 151–153 recognition, retrieval and binding, 149

sequential priming, 153–154 snapshot matching image matching, 141–143 orientation of snapshots, 139–141 vector memories global vectors, 144–147 local vectors, 147–148 visual landmark memories extent of snapshots, 137–139 features encoded in snapshots, 135–137 retinotopic and non-retinotopic encoding of snapshots, 132–135 Animal mate selection in Apteronotidae, 273–274 bioluminescent signals, role of, 276 comparisons of models of, 270 and copepod vibrations, 267 female song preferences, 73 in Gymnotidae, 273–274 head ornamentation, role of, 275 and high-intensity displays, 270 in Hypopomidae, 273–274 IEG induction in, 80–87 male courtship trembling, 267 mouth-breeding females as a trait in, 272 in Rhamphichthyidae, 273–274 Anteriori forebrain pathway, 64 Ants, 125–126 of Australian desert, 126–127, 129, 144, 157 diurnal rhythm of, 160–163 homeward routes of, 128 trained, 150 unfed, 160 449

450

INDEX

Ants (continued ) wood, 130, 141 visual memories of, 161 Apis mellifera, see Honeybees Appetitive learning pivots, 23 Aptenodytes patagonicus, 232 Aralia hispida, 328–329 Armament–ornament hypothesis, 251 Aromas, of mother’s diet, 11 Artificial head crests, 269 Astacus leptodactylus, 424 Attachment behavior, 1 Auditory feedbacks, 66 Autumnal singing, 69 Avian species, 256 vocal mimicry, 279 Avian telencephalon, 66 Avisoft computer program, 366

B Banbury Conference on Genetic Background in Mice, 198 Barn swallow tail streamers, 281 Basilar papilla, 264–265 Bats, 274 Bee foraging behavior discussion of, 339–342 floral color preference comparison between populations, 313–317 comparison between species, 311–313 vs foraging performance, 317–319 flower constancy, 307–311 genetic basis of, 334–335 learning behavior, 320–323 manipulating foraging phenotypes, 330–334 manipulation of foraging environment, 326–330 models of, 335–339 reciprocal population transplant experiments, 323–326 Behavioral fossil, 316

Bird’s own song, 87 Birth, initial hours of colostrum, specificity of, 41–43 sensitive period, of suckling act, 41 vitality, 40–41 Birth fluids, 1 Black-capped chickadees, 78 Bluff signals, 397–398 Bombus diversus, 312 Bombus hypocrita, 312 Bombus ignitus, 312 Bombus impatiens, 327 Bombus lapidarius, 308–311 Bombus occidentalis, 312–313 Bombus pascuorum, 308–311 Bombus terrestris, 308–311, 315 canariensis, 315, 324 dalmatinus, 315 sassaricus, 315, 324 terrestris, 315–316 xanthopus, 315 BOS, see Bird’s own song Bottle-fed infants, 28, 34, 36 Bottom-up mechanism, 89 Bovine milk, 17, 22, 32, see also Milk colostrum of, 41–43 Bowerbirds, 252, 256, 280 bowers of, 272–273 Brachyhypopomus pinnicaudatus, 274 Brain–behavior relationships, 76–77 Brain endorphin reward system, 37 Breast-fed babies, 28 Bumblebee species, 308 color preferences of, 314 foraging performance in, 310

C Caenorhabditis elegans, 173, 175–177, 198 Calcium ions, 22, 74 Calming effect, of pacifiers, 36–37, 40 -Calmodulin protein kinase II, 74 cAMP signals, 188 Cannulae, 31

INDEX

Carbohydrates, 22–23 Carduelis tristis, 398 Caribbean ostracodes, 277 Caseins, 19, 22 -Casomorphin, 43 Cataglyphis bicolor, 125, 157 Cataglyphis fortis, 142 Caudal nidopallium, 66 CB1 cannabinoid receptor inhibits, 86 CCK, see Cholecystokinin CCK-1 receptor antagonists, 38 Central nervous system, 73 Cerceris rybyensis, 141 CF-1 mouse stain, 197 c-fos, 75 CGB, see Corticosteroid-binding globulin Chemosensory signals, 13 Chirp calls, 274 Chlamydera bowerbirds, 279 Chloride ions, 22 Cholecystokinin, 35, 37–40 Chuck calls, 264–266 Cichlids, endemic species of, 231, 282 cis-regulatory motifs, 182 Citrate, 22 Clutch initiation, 63 CNM, see Caudal nidopallium CoA ligase acid, 277 Cocaine, 198 Cockroaches, 133, 135, 143 Cocktail-party problem, 232 Color preferences comparison between population, 313–317 comparison between species, 311–313 vs foraging performance, 317–319 Colostrum, 16, 35, see also Milk bovine milk, 41–43 and geniohyoid activity, 42 ingestion by lambs, 35–36 lactose concentration, 18 and masseter activity, 42 protein concentration, 17–18 specificity of, 41–43 thermogenic effect of, 40

451

Comforting state, 36 Comodulation, 233 Conditioned responses, in species, 270 human baby, 28 rabbit pup, 27 rat pup, 26–27 Co-operative breeding, in long-tailed tits constraints in, 381–389 fitness consequences, of co-operation direct fitness benefits enhanced probability of successful future reproduction, 375 experiences, 373 shared reproduction by helpers, 371–372 survival benefits, 373–375 ecological basis for co-operative breeding matching model predictions with observed behavior, 385 model of fitness payoffs from breeding and helping, 381–383 inclusive fitness, 380 indirect fitness benefits increased productivity of relatives, 376–380 increased survival of related breeders, 379 general methods, 358–359 kin discrimination, by helpers experimental test of kin discrimination, 361–362 observational study of kin discrimination, 361–362 kin recognition mechanism experimental investigation of call development, 367–368 experimental test of vocal kin recognition, 365–366 family specificity in vocalizations, 364 individuality in vocalizations, 363

452

INDEX

Co-operative breeding, in long-tailed tits (continued ) pattern of helping behavior, 368–370 kin selection, 386–387 study species and study sites, 357–358 Cooper’s ligaments, 7 Co-option of traits as common source of display, 284–285 of indicator traits, 285 in morphological form, 275–276 occurrence of, 285–286 of preexisting traits, 286 and timing of display, 282–284 of visual and chemical form, 276–278 of vocal display and display mechanisms, 278–280 Corcoran cephalonica, 274 Correct rejection, 220 CORT, see Corticosterone Corticosteroid-binding globulin, 98 Corticosterone, 98 Cost-dependent handicap system, 258 CREB, see Cyclic AMP response element-binding protein Crotaphytus collaris, 399 CsA gene, 188 CsA knockout cells, 189 Cyclic AMP response element-binding protein, 74

D Damselflies, 82 Darwin’s finches, 201 Deer antlers, 275 Dehydroepiandrosterone, 69 Delayed matching, 153 Delayed maturation, 259 Detection plus classification, of signals, 234–238 Devazepide, 37

Developmental stress hypothesis, 99–102 Dictyostelium amoebas, 174, 188–189, 207 Di-methyl disulfide, 10 Diopsid flies, 275 Dipteran ornamentation, 275 Discriminability of signals, 224 Discrimination learning, 240 Dorsolateral thalamic nucleus, 65–66 Drosophila, 135, 261 mutation accumulation in, 191–192 sexual conflict in, 186–187 Drosophila melanogaster, 173, 175–176, 341 Dung flies, 267–268

E Early olfactory learning, 2 Ecological functions, in functional genomics in Dictyostelium amoebas, 186–189 MHC-mediated kin recognition, for communal nesting partners, 185–186 MHC-mediated sexual selection, 183–185 sexual conflict, in Drosphilia, 186–187 timing of flowering, in Arabidopsis, 187–188 Egr-1, 68 Electric organ discharges, 273 Embryogenesis, 206 Embryonic stem cells, 198 EMS, see Ethylmethane sulfonate Enhanced vigor, 206 ENU, see N-ethyl-N-nitrosourea EOD, see Electric organ discharges Erithacus rubecula, 68 ES cells, see Embryonic stem cells Estradiol, 11

453

INDEX

Estrildine finches, 269 Ethylmethane sulfonate, 175 Eudyptes pachyrhyncus, 430 Eudyptes schlegeli, 413 Eudyptula, 400 minor, 399 Eukaryotic gene transcription, 75 European starlings Belgian population of, 91 Canadian population of, 91 distribution of, 61 polygynous males of, 103 song behavior and perception in, see European starling, song behavior and perception in European starlings, song behavior and perception in electrophysiological responses to song, 88–90 functional basis of song preferences direct benefits, 90–91 indirect benefits developmental stress hypothesis, 99–102 future directions, 103–104 immunocompetence handicap hypothesis, 92–99 sexy son hypothesis, 103 functions of, 63–64 fundamentals of, 61–63 song behavior, control of hormonal control, 68–71 neural, 64–68 perception of songs behavioral experiments of song recognition, 71–73 female song preferences, in mate choice context, 73 physiological responses to songs IEG expression, 74–80 IEG induction in mate choice, 80–87 Eutherian milks, 21 Extracellular signal-regulated kinase, 86 Extrinsic uncertainty, 233

F False alarm, 220, 240 Female preferences, for costly displays, 256 F2 generation, 193 Fiddler crabs, 273, 275 Fireflies, 276–277 Fisherian sexy son hypothesis, see Sexy son hypothesis Fisher’s initiator hypothesis, 283 Fisher’s runaway hypothesis, 263 Fitness assays, 201–203 Fitness measurements, 200 Floral rewards, 320, 322 Florida scrub jays, 201 Flower constancy, 307–311 consistent differences in, across bumblebee species, 309 Forced reproduction, in caged breedings, 201 Forebrain auditory regions, 73 FOS expression, 84–87 Foster’s Island Rule, 325 Freud concepts, on mother child relations, 26, 44 Fruit flies, 278 Functional genomics behavioral performance and measuring fitness, 200 competition and phenotypic differences gene knockouts in Saccharomyces, 191 inbreeding in Mus, 192–195 mutation accumulation, in Drosophila, 191–192 resolving the paradox, in selfish gene, 195–197 ecological explanation, 176–177 fitness differences, 199–200 functional redundancy, 176 genes lacking phenotypes ecological functions in Dictyostelium amoebas, 186–189

454

INDEX

Functional genomics (continued ) MHC-mediated kin recognition, for communal nesting pattern, 185–186 MHC-mediated sexual selection, 183–185 sexual conflict, in Drosophila, 186–187 timing of flowering, in Arabidopsis, 187–188 functional redundancy evidence, 179–183 theory, 178–179 genetic background problem, 197–199 integrative approaches to integration of fitness components in phenomics, 204–205 nonmodal organisms and functional genomics, 207 role of fitness measures in understanding phenotypes, 205–206 phenotype gene knockouts, 175 phenotype gene mutants, 175–176 relevance of fitness assays additional phenotypes, role of, 203 relative importance of genes, 202–203 and staged seminatural conditions, 200–201 Functional redundancy, 176 theory and evidence, 178–183 Fur, role in feeding, 10

G GABA, see Gamma-aminobutyric acid Gametic phase disequilibrium, 260–261 Gamma-aminobutyric acid, 86 Genetic correlations, 261 models of, 262–263 Genetic footprints, of gene mutations, 199 Gigantiops destructor, 158–159

Glycoprotein glue, see Spiggin Gonadal steroids, 70 Gonodactylus bredini, 440 Good genes indicators, 256–260 vs runaway genes, 262–263 Gp80 protein-binding site, 188 Growled threats, 404–406 Gryllus bimaculatus, 439

H Hamilton and Zuk’s original hypothesis, 95 Harsh competitive conditions, 192 Hecatesia exultans, 274 High-quality males, 254 High strain colonies, 334 Honeybees, 135, 138–140, 151, 153–154 dance of, 330–334 diurnal rhythm of, 160–163 foodward and homeward vectors in, 145 marathon waggle dance, 163 time-linked directional memories in, 162 trajectories of, 156 vector switching in, 158 Hox genes, 175 in Drosphilia, 180 in Mus, 180 Hox paralogous groups, 180 Hox protein, 180 Hummingbirds, 77 HVC lesions, 88 Hyla ebraccata, 235 Hypoglossal nucleus, 64

I ICHH, see Immunocompetence handicap hypothesis Iconic nonmorphological display, 271–275

455

INDEX

Identification operating characteristic, 235 IEG expression IEG induction in song birds, 76–80 neural induction, of IEGs, 74–76 Immunocompetence handicap hypothesis, 92–99, 101 Immunoglobulins, 8 Inbred females, 193 Inbred males, 193, 202 Incipient male display traits, 256 Information validation methods, of little blue penguins investment strategies, validating signals and signal synergy, 438–441 natural history, 399–402 risk breed signals risk display and repelling behavior, 420–424 risk pattern in little blue penguins 414–420 signals of intent acoustic signatures, 426–429 capability for individual recognition, 430–433 information, revealing of, 433–438 validations for information contained in agonistic displays signals that advertise RPH correlations between body size and acoustic cues of growled threats, 404–406 correlations between body size and acoustic cues of solo calls, 403–404 wars of attrition male penguin advertising call rates, 408–411 overlapping competitor calls, 411–412 Intranasal tube feeding, 31 Intraoral cannula, 26 Intrinsic uncertainty, 233

IOC, see Identification operating characteristic Irradiance spectrum, 231

J Jackson laboratories, 198 Japanese quail, 76 Journal of Molecular and Cellular Biology, 180 Jungle fowl, 276

K Keyhole limpet hemocyanin, 94, 97 Kin discrimination mechanism, see Co-operative breeding, in long-tailed tits Kin recognition mechanism, see Co-operative breeding, in long-tailed tits KLH, see Keyhole limpet hemocyanin Knockout phenotypes, 174 Hox gene knockouts, 175 Krox-24, 68

L Lab-assayed phenotypes, 202 -Lactalbumin, 20 Lactase, 21 Lactiferous duct, 7 Lactiferous sinus, 7 -Lactoglobulin, 20 Lactose, 21, 32, 38 Lagomorphs, 4 Landmarks, in foraging behavior, 151 Larval predator defense displays, 276–277 8L:16D photoperiods, 83 LED, see Light-emitting diode Lekking males, 274

456

INDEX

Licking behavior, see Maternal behavior Light-emitting diode, 133, 143, 150 Lions, 201 Listener’s attitude, 241 LMAN projects, 66 Long-bout songs, 73, 83 Long-chain polyunsaturated fatty acids, 21 Lordosis, in rats, 69

M Magnesium ions, 22 Major histocompatibility genotypes, 275 Major histocompatibility complex (MHC) by-products, 278 homozygous offspring, 203 mediated kin recognition for communal nesting partners, 185–186 mediated sexual selection, 183–185 odorants, 278 Malaysian flies, 275 Male–male agonistic interactions, 274 Male–male territorial signaling, 261 Male manakins, 231 Male morphs, 82, 260 Male sexual display traits, see also Animal mate selection; Signal perception, in animals acoustic calls, 285 alternative models pre-existing trait model, 252–253 war propaganda model, 250–252 bright plumage and integument color, use of, 282 co-option as common source of display, 284–285 co-option of indicator traits, 285 co-option of pre-existing traits, 286 and different models of operation

Fisher initiator hypothesis, 283 pre-existing trait, 283–284 electrical organ discharges, 285 genetic correlation models genetic correlations and mate choice, 260–261 good genes vs runaway models, 262–263 larval predator defense displays, in fireflies, 277 light flashes, 285 mate choice models in display trait evolution, 286–287 and mouth-breeding cichlids, 271 and nest of birds, 271 nests, 285 occurrence of co-option, 285–286 plumage colors, 285 pre-existing preference models general issues, 264 pre-existing preferences, 264–268 theoretical issues, 268–270 pre-existing trait version of good genes model co-option of morphological traits, 275–276 co-option of visual and chemical traits, 276–278 co-option of vocal displays and display mechanisms, 278–280 and general indicator mechanisms, 281–282 iconic nonmorphological display, 271–275 and multivariate display elements, 282 timing of display and co-option, 282–284 problems with elaborate display traits good genes hypothesis, 254–256 honest good genes indicators, 256–257 male contradiction dependant traits, 258–260 Zahavian handicap models, 253–254

457

INDEX

and stickle backs, 271 symmetry differences, 285 Male spermatophores, 267 Mammary gland, 2 in eutherian mammals, 5 in metatherian mammals, 5 Mammary pheromone, 27 Mammary region anatomy of mammary glands ewe, 7 human female, 7–8 rabbit doe, 7 rat, 6–7 maternal cues, understanding of human baby, 12 lamb, 11–12 rabbit pup, 10–11 rat pup, 9–10 patterns of nursing behavior human, 16–17 rabbit, 14–15 rat, 14 sheep, 15–16 source of milk preparation to source, 4–5 Maternal behavior, see also Neonates among predator species, 23 animal licking/stroking in aquatic species, 2 in mammals, 1–2 purpose of, 1 in primates, 2 and exposure to milk, 2 in humans, 2 nursing of neonates, 2–3 suckling-nursing relationship, between young and mother, 2 Maternal immunoglobulins, 17 Medial nidopallium, 66, 80, 86 Melophorus bugati, 126, 157, 165 Melopsittacus undulatus, 79 Melospiza georgiana, 101 Melospiza melodia, 68 Mendelian frequencies, of t complex, 195

Merino, 7 Merops bullockoides, 361 Metazoans, 200, 204 2-Methylbut-2-enal, 11 Migraine headaches, 202 Milk biochemical composition of carbohydrate, 21–22 milk fat, 17–19, 21 milk protein, 19–20 minerals, 22 other components, 22 water, 19 bovine, see Bovine milk colostrum, 17 early milk production, around parturition, 2 ovine, see Ovine milk secretion of, after nursing bouts in rabbits, 14–15 in rats, 14 species differences in milk composition and suckling interaction, 22–23 Mimetic songs, 280 Mirounga spp., 397 Molothrus ater, 82 Monoparental species, 24–25 Montgomery glands, 8 Morphine injections, 35 Motor theory, of song perception, 88 Mus domesticus, 195 inbreeding in, 192–195 musculus, 173

N Navigational memories, see Animal foraging behavior NCM, see Medial nidopallium Negative geotropism, 8 Neoconocephalus spiza, 413

458

INDEX

Neonates, see also Maternal behavior motor activities, 2 nursing behavior, of mothers, 2–3 among predator species, 23 in humans, 16–17 in rabbits, 14–15 rats, 14 in sheep, 15–16 rewards for, 23–26 suckling-nursing relationship, with mother, 2 Nestling-feeding period, 70 N-ethyl-N-nitrosourea, 175 Neuroactive estrogen, 68 Neuroethology, 60 Neurological disorders, in animals, 202 Neurons, 67, 85, 88–89, 218 Neurotransmitters, 86 NewBehavior Inc., 205 NGFI-A, 68 Nipples, 4–6 of human females, 7–8 nonlactating, of rats, 26 of rats, 6–7 Nipple-search pheromone, 10 NMDA receptor-dependent LTP formation, 75 Noctuid moth, 274 Nonnutritive sucking activity, 34, 36 Non-Oscine budgerigar, 79 Normal filial attachment process, 36 Nucleus HVC, 64 Nucleus mesencephalicus lateralis dorsalis, 66 Nucleus robustus arcopallialis, 64 Nutrients, 234 Nutritional stress hypothesis, see Developmental stress hypothesis

O Odor cues, of skin, 7 Olfactory bulbectomy, 9 Olfactory cues, 9–10, 12, 28, 163–164, 326

Omnidirectional microphones, 230 Opioids, 35, 37–40 Oral milk infusions, act of, 27 Oro-gastrointestinal sphere, 31 Ostracode crustaceans, 276 Outbred males, 193 Ovine milk, 17, 32, see also Milk colostrum of, 41–43

P Pagurus bernhardus, 439 Palearctic avifauna, 381 Panoramic views, 151–153 Parent–offspring contact, 1 Parrots, 77 Parus major, 231 Passerine birds, 279 Path integration, 125, 144–146, 159, 164 PCA, see Principal components analysis PDF, see Probability density function Peripheral mechanisms, of perception, 238 Perisoreus infaustus, 375 PHA, see Phytohemagglutinin injection Phenomics technologies, 205 Phenotype gene knockouts, 175 Phenotype gene mutants, 175–176 Photinus fireflies, 276 Phylogenetic mapping, 284 Phylogenies, 269, 271, 311 Phytohemagglutinin injection, 93 PI, see Path integration Pleiotropy, 203, 207 Ploceus philippinus, 272 Plumage expression, 259 Poecilia reticulate, 413 Pollination market, 320 Polytocous species, 8 POM, see Preoptic medial nucleus Postpartum period, in polytocous species, 18 Potassium ions, 22 Preexisting preference model general issues, 264

459

INDEX

preexisting preferences, 264–268 theoretical issues, 268–270 Preexisting trait hypothesis, 253 Preoptic medial nucleus, 77 Principal components analysis, 403 Probability density function, 219–220, 222, 236 Proctor’s cladistic analysis, 268 Proctor’s hypothesis, 267 alternative to, 267 Progesterone, 11 Prolactin, 11 Proximate causation, 49 Psaltria exilis, 358 Psaltriparus minimus, 358 Psychophysicists, 221–222, 230 Pterodroma mollis, 431 Pythagorean relationships, 236, 241

Q QTL, see Quantitative trait loci Quantitative trait loci, 187, 203

R Receiver operating characteristic, 222, 227, 229 Receiver psychology, 229 Red-winged blackbirds, 69 Rennin, 19 Repertoire size, of song, 90–91 RNAi, see RNA-mediated interference RNA-mediated interference, 175 ROC, see Receiver operating characteristic Rosenblatt’s proposal, 29 Route mechanisms and feeding state, 157–160 foraging routes, 124–127 navigational memories, 127–129 and olfactory cues, 163–164 role of rewards, 164 snapshot matching, 139–143

time aspects and route memories, 160–163 vector memories, 144–148 visual landmark memories, 132–139 Ryan and Rand’s experiments, 265

S Saccharomyces cerevisiae, 173, 199 gene knockouts in, 191 Saline, 32, 34 Scatophaga, 268 Selfish gene, 195–197 Self-licking behavior, 4 Sensory stimuli, different forms of gastrointestinal stimuli, 31–34 oral stimuli, 34–35 Sensory thresholds, 221–222 optimal criteria of, 228 Sequential priming, 153–154 Sequential redundancy, 232–233 Serinus canaria, 64 Serum albumin, 20 Sexually selected traits, 80 Sexy son hypothesis, 103 Short-bout songs, 73, 83 Short-term memory limitations, 307 Signal detection theory application of, in experimental psychophysics, 221–224 essential features of, 218–221 general assumptions of, 224–227 specific assumption of cuing of responses, 228–229 normal distribution with equal variance, 227–228 optimal criteria, 228 Signal perception, in animals, see also Male sexual display traits application of signal detection theory, 221–224 classification of signals, 235–238 complex pattern of signals, 238–239

460

INDEX

Signal perception, in animals (continued ) essential patterns in, 218–221 evolution of signals reception, 239–240 general assumptions of signal detection theory, 224–227 interpretation of playback experiments, 240–241 practicality of experiments in natural situations, 241–243 properties of signals, affecting receiver performance contrast, 229–232 redundancy, 232–233 uncertainty and unfamiliarity, 233–234 specific assumptions of signal detection theory cuing of responses, 228–229 normal distribution with equal variance, 227–228 optimal criteria, 228 Signals of intent, 398 Signal-to-noise ratio, 230, 232, 235 Silent offensive stationary displays, 423 Single homozygous knockouts, 180 Skraa calls, 279 Snapshots extent of, 137–138 features encoded in, 135–137 image matching, 141–143 orientation of, 139–141 retinotopic and nonretinotopic encoding of, 132–135 Social ecology, 183 Social isolation, 1 Sociogenomics, 207 Sodium ions, 22 Solidago spp., 330 Solo calls, 401, 403–404, 407–413, 433 Somatosensory cues, 24 Song classification strategy, 72–73 Song preferences, functional benefits in direct benefits, 90–91

indirect benefits developmental stress hypothesis, 99–102 immunocompetence handicap hypothesis, 92–99 sexy son hypothesis, 103 Song sparrows, 69 SPCC, see Spectrographic cross-correlation Spectrographic cross-correlation, 363 Spectrotemporal receptive fields, 67 Sperm competition, 187 Spiggin, 272 SRF, see Spectrotemporal receptive fields Steroid-induced immunosuppression, 104 Stochastic environmental conditions, 201 STRFs, see Spectrotemporal receptive fields Sturdy, Christopher, 207 Sturnus vulgaris, 60 Subtractive hybridization techniques, 75 Suckling and early learning biological mechanisms immediate effects, 35–37 opioids and cholecystokinin, 37–40 development of conditioned responses human baby, 28 rabbit pup, 27 rat pup, 26–27 key roles of first suckling episodes, 28–31 neonatal rewards, 23–26 sensory stimuli embedded in gastro intestinal stimuli, 31–34 oral stimuli, 34–35 Suckling–nursing relationship, between young and mother, 3 Sulfate, 22 Swamp sparrows, 70 Sweat glands, 5

461

INDEX

T Tactile stimulation, of face, 11 Taeniopygia guttata, 64, 413 T complex, 196–197 T concentrations, 70–71, 95, 105 t haplotypes, 196–197 Theiler’s virus-induced demyelinating disease, 198 Thermal cues, 11, 13 Thermotactile characteristics, of mother’s ventrum, 9–10 Threshold theory, 226 Top-down mechanism, 89 Transgenerational characters, 174 Traplining bees, 326–327 relative benefits of, 328 Triglycerides, 21 Tungara frog’s auditory system, 266 Turdus merula, 438

U Ultimate causation, 49 Unionicolid water mites, 266 Uta stansburiana, 398 UV-transmittent Plexiglas square chips, 312–313

V Vagal mechanoreceptive endings, 39 Vanilla scented sucrose solution, 162 Vargulae, 277 Vector memories global, 144–147 local, 147–148 Ventral tegmental area, 77 Vernix caseosa, 2 Vespula vulgaris, 140 Visual landmark memories, see Snapshots Vitality, 40–41

Vocally familiar males, 72 Vocal offensive stationary displays, 423 Volatile olfactory signals, 277 VTA, see Ventral tegmental area

W Waggle dance, 144, 330 Warble motifs, 61 Warm maternal womb warm mammary zone, 13 War propaganda model, in mate selection, 250–252 Wars of attrition, 398, 407–413, 439 Wasps, 139–140 Water striders, 129, 132 Wax moth, 274 Weakly electric fish, 273 Weaning ages among pinnipeds, 13 in primates, 13 Whey proteins, 19 White-crowned sparrows, 70, 79, 82 Wild-caught mice, 193 Wolf spiders, 82

X Xiphophorus species, 266

Y Yeast backup circuits, 182 Y maze, 135, 153, 160

Z Zahavian handicap model, 253–254 Zebra finches, 69, 71, 77, 84, 88, 101, 105, 207 Zenk, 68, 75, 77

462 ZENK expression, 86–87, 89 ZENK responses, 77–80, 83–84 Zif-268, 68 Zif268 IEG transcription factor, 75

INDEX

Zonotrichia leucophrys, 69 Zonotrichia leucophrys oriantha, 79 Z-transforms of PCD and PFA, 223

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 Behavioral Aspects of Sperm Competition in Birds T. R. BIRKHEAD Neural Mechanisms of Perception and Motor Control in a Weakly Electric Fish WALTER HEILIGENBERG Behavioral Adaptations of Aquatic Life in Insects: An Example ANN CLOAREC The Circadian Organization of Behavior: Timekeeping in the Tsetse Fly, A Model System JOHN BRADY

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 Mode Selection and Mode Switching in Foraging Animals GENE S. HELFMAN Cricket Neuroethology: Neuronal Basis of Intraspecific Acoustic Communication FRANZ HUBER Some Cognitive Capacities of an African Grey Parrot (Psittacus erithacus) IRENE MAXINE PEPPERBERG

Volume 19 Volume 20 Polyterritorial Polygyny in the Pied Flycatcher P. V. ALATALO AND A. LUNDBERG Kin Recognition: Problems, Prospects, and the Evolution of Discrimination Systems C. J. BARNARD Maternal Responsiveness in Humans: Emotional, Cognitive, and Biological Factors CARL M. CORTER AND ALISON S. FLEMING

Social Behavior and Organization in the Macropodoidea PETER J. JARMAN The t Complex: A Story of Genes, Behavior, and Population SARAH LENINGTON The Ergonomics of Worker Behavior in Social Hymenoptera PAUL SCHMID-HEMPEL 463

464

CONTENTS OF PREVIOUS VOLUMES

‘‘Microsmatic Humans’’ Revisited: The Generation and Perception of Chemical Signals BENOIST SCHAAL AND RICHARD H. PORTER

Parasites and the Evolution of Host Social Behavior ANDERS PAPE MOLLER, REIJA DUFVA, AND KLAS ALLANDER

Lekking in Birds and Mammals: Behavioral and Evolutionary Issues R. HAVEN WILEY

The Evolution of Behavioral Phenotypes: Lessons Learned from Divergent Spider Populations SUSAN E. RIECHERT

Volume 21

Proximate and Developmental Aspects of Antipredator Behavior E. CURIO

Primate Social Relationships: Their Determinants and Consequences ERIC B. KEVERNE The Role of Parasites in Sexual Selection: Current Evidence and Future Directions MARLENE ZUK Conceptual Issues in Cognitive Ethology COLIN BEER Response to Warning Coloration in Avian Predators W. SCHULER AND T. J. ROPER Analysis and Interpretation of Orb Spider Exploration and Web-Building Behavior FRITZ VOLLRATH Motor Aspects of Masculine Sexual Behavior in Rats and Rabbits GABRIELA MORALI AND CARLOS BEYER On the Nature and Evolution of Imitation in the Animal Kingdom: Reappraisal of a Century of Research A. WHITEN AND R. HAM

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

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 Genetic Correlations and the Control of Behavior, Exemplified by Aggressiveness in Sticklebacks THEO C. M. BAKKER Territorial Behavior: Testing the Assumptions JUDY STAMPS Communication Behavior and Sensory Mechanisms in Weakly Electric Fishes BERND KRAMER

CONTENTS OF PREVIOUS VOLUMES

Volume 24 Is the Information Center Hypothesis a Flop? HEINZ 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

465

An Overview of Parental Care among the Reptilia CARL GANS 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 GONZA¨LEZ-MARISCAL AND JAY S. ROSENBLATT Parental Behavior in Voles ZUOXIN WANG AND THOMAS R. INSEL Physiological, Sensory, and Experiential Factors of Parental Care in Sheep F. LE¨VY, K. M. KENDRICK, E. B. KEVERNE, R. H. PORTER, AND A. ROMEYER

Cognition in Cephalopods JENNIFER A. MATHER

Socialization, Hormones, and the Regulation of Maternal Behavior in Nonhuman Simian Primates CHRISTOPHER R. PRYCE

Volume 25

Field Studies of Parental Care in Birds: New Data Focus Questions on Variation among Females PATRICIA ADAIR GOWATY

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

Parental Investment in Pinnipeds FRITZ TRILLMICH Individual Differences in Maternal Style: Causes and Consequences of Mothers and Offspring LYNN A. FAIRBANKS

466

CONTENTS OF PREVIOUS VOLUMES

Mother–Infant Communication in Primates DARIO MAESTRIPIERI AND JOSEP CALL Infant Care in Cooperatively Breeding Species CHARLES T. SNOWDON Volume 26 Sexual Selection in Seaweed Flies THOMAS H. DAY AND ANDRE¨ S. GILBURN Vocal Learning in Mammals VINCENT M. JANIK AND PETER J. B. SLATER Behavioral Ecology and Conservation Biology of Primates and Other Animals KAREN B. STRIER How to Avoid Seven Deadly Sins in the Study of Behavior MANFRED MILINSKI Sexually Dimorphic Dispersal in Mammals: Patterns, Causes, and Consequences LAURA SMALE, SCOTT NUNES, AND KAY E. HOLEKAMP Infantile Amnesia: Using Animal Models to Understand Forgetting MOORE H. ARNOLD AND NORMAN E. SPEAR Regulation of Age Polyethism in Bees and Wasps by Juvenile Hormone SUSAN E. FAHRBACH Acoustic Signals and Speciation: The Roles of Natural and Sexual Selection in the Evolution of Cryptic Species GARETH JONES

Volume 27 The Concept of Stress and Its Relevance for Animal Behavior DIETRICH VON HOLST Stress and Immune Response VICTOR APANIUS Behavioral Variability and Limits to Evolutionary Adaptation P. A. PARSONS Developmental Instability as a General Measure of Stress ANDERS PAPE MOLLER Stress and Decision-Making under the Risk of Predation: Recent Developments from Behavioral, Reproductive, and Ecological Perspectives STEVEN L. LIMA Parasitic Stress and Self-Medication in Wild Animals G. A. LOZANO Stress and Human Behavior: Attractiveness, Women’s Sexual Development, Postpartum Depression, and Baby’s Cry RANDY THORNHILL AND F. BRYANT FURLOW Welfare, Stress, and the Evolution of Feelings DONALD M. BROOM Biological Conservation and Stress HERIBERT HOFER AND MARION L. EAST

Volume 28

Understanding the Complex Song of the European Starling: An Integrated Ethological Approach MARCEL EENS

Sexual Imprinting and Evolutionary Processes in Birds: A Reassessment CAREL TEN CATE AND DAVE R. VOS

Representation of Quantities by Apes SARAH T. BOYSEN

Techniques for Analyzing Vertebrate Social Structure Using Identified

CONTENTS OF PREVIOUS VOLUMES

Individuals: Review and Recommendations HAL WHITEHEAD AND SUSAN DUFAULT Socially Induced Infertility, Incest Avoidance, and the Monopoly of Reproduction in Cooperatively Breeding African Mole-Rats, Family Bathyergidae NIGEL C. BENNETT, CHRIS G. FAULKES, AND JENNIFER U. M. JARVIS Memory in Avian Food Caching and Song Learning: A General Mechanism or Different Processes? NICOLA S. CLAYTON AND JILL A. SOHA Long-Term Memory in Human Infants: Lessons in Psychobiology CAROLYN ROVEE-COLLIER AND KRISTIN HARTSHORN Olfaction in Birds TIMOTHY J. ROPER Intraspecific Variation in Ungulate Mating Strategies: The Case of the Flexible Fallow Deer SIMON THIRGOOD, JOCHEN LANGBEIN, AND RORY J. PUTMAN

Volume 29 The Hungry Locust STEPHEN J. SIMPSON AND DAVID RAUBENHEIMER Sexual Selection and the Evolution of Song and Brain Structure in Acrocephalus Warblers CLIVE K. CATCHPOLE Primate Socialization Revisited: Theoretical and Practical Issues in Social Ontogeny BERTRAND L. DEPUTTE

467

Ultraviolet Vision in Birds INNES C. CUTHILL, JULIAN C. PARTRIDGE, ANDREW T. D. BENNETT, STUART C. CHURCH, NATHAN S. HART, AND SARAH HUNT What Is the Significance of Imitation in Animals? CECILIA M. HEYES AND ELIZABETH D. RAY Vocal Interactions in Birds: The Use of Song as a Model in Communication DIETMAR TODT AND MARC NAGUIB

Volume 30 The Evolution of Alternative Strategies and Tactics H. JANE BROCKMANN Information Gathering and Communication during Agonistic Encounters: A Case Study of Hermit Crabs ROBERT W. ELWOOD AND MARK BRIFFA Acoustic Communication in Two Groups of Closely Related Treefrogs H. CARL GERHARDT Scent-Marking by Male Mammals: Cheat-Proof Signals to Competitors and Mates L. M. GOSLING AND S. C. ROBERTS Male Facial Attractiveness: Perceived Personality and Shifting Female Preferences for Male Traits across the Menstrual Cycle IAN S. PENTON-VOAK AND DAVID I. PERRETT The Control and Function of Agonism in Avian Broodmates HUGH DRUMMOND

468

CONTENTS OF PREVIOUS VOLUMES

Volume 31 Conflict and Cooperation in a Female-Dominated Society: A Reassessment of the ‘‘Hyperaggressive’’ Image of Spotted Hyenas MARION L. EAST AND HERIBERT HOFER Birdsong and Male–Male Competition: Causes and Consequences of Vocal Variability in the Collared Dove (Streptopelia decaocto) CAREL TEN CATE, HANS SLABBEKOORN, AND MECHTELD R. BALLINTIJN Imitation of Novel Complex Actions: What Does the Evidence from Animals Mean? RICHARD W. BYRNE Lateralization in Vertebrates: Its Early Evolution, General Pattern, and Development LESLEY J. ROGERS Auditory Scene Analysis in Animal Communication STEWART H. HULSE Electric Signals: Predation, Sex, and Environmental Constraints PHILIP K. STODDARD How to Vocally Identify Kin in a Crowd: The Penguin Model THIERRY AUBIN AND PIERRE JOUVENTIN

Volume 32 Self-Organization and Collective Behavior in Vertebrates IAIN D. COUZIN AND JENS KRAUSE Odor-Genes Covariance and Genetic Relatedness Assessments: Rethinking

Odor-Based Recognition Mechanisms in Rodents JOSEPHINE TODRANK AND GIORA HETH Sex Role Reversal in Pipefish ANDERS BERGLUND AND GUNILLA ROSENQVIST Fluctuating Asymmetry, Animal Behavior, and Evolution JOHN P. SWADDLE From Dwarf Hamster to Daddy: The Intersection of Ecology, Evolution, and Physiology That Produces Paternal Behavior KATHERINE E. WYNNE-EDWARDS Paternal Behavior and Aggression: Endocrine Mechanisms and Nongenomic Transmission of Behavior CATHERINE A. MARLER, JANET K. BESTER-MEREDITH, AND BRIAN C. TRAINOR Cognitive Ecology: Foraging in Hummingbirds as a Model System SUSAN D. HEALY AND T. ANDREW HURLY

Volume 33 Teamwork in Animals, Robots, and Humans CARL ANDERSON AND NIGEL R. FRANKS The ‘‘Mute’’ Sex Revisited: Vocal Production and Perception Learning in Female Songbirds KATHARINA RIEBEL Selection in Relation to Sex in Primates JOANNA M. SETCHELL AND PETER M. KAPPELER

CONTENTS OF PREVIOUS VOLUMES

Genetic Basis and Evolutionary Aspects of Bird Migration PETER BERTHOLD Vocal Communication and Reproduction in Deer DAVID REBY AND KAREN MCCOMB Referential Signaling in Non-Human Primates: Cognitive Precursors and Limitations for the Evolution of Language KLAUS ZUBERBU«HLER Vocal Self-stimulation: From the Ring Dove Story to Emotion-Based Vocal Communication MEI-FANG CHENG

469

Odor Processing in Honeybees: Is the Whole Equal to, More Than, or Different from the Sum of Its Parts? HARALD LACHNIT, MARTIN GIURFA, AND RANDOLF MENZEL Begging, Stealing, and Offering: Food Transfer in Nonhuman Primates GILLIAN R. BROWN, ROSAMUNDE E. A. ALMOND, AND YFKE VAN BERGEN Song Syntax in Bengalese Finches: Proximate and Ultimate Analyses KAZUO OKANOYA Behavioral, Ecological, and Physiological Determinants of the Activity Patterns of Bees P. G. WILLMER AND G. N. STONE

Volume 34

Volume 35

Reproductive Conflict in Insect Societies ˆ RGEN HEINZE JU

Mechanisms and Evolution of Communal Sexual Displays in Arthropods and Anurans MICHAEL D. GREENFIELD

Game Structures in Mutualistic Interactions: What Can the Evidence Tell Us About the Kind of Models We Need? REDOUAN BSHARY AND JUDITH L. BRONSTEIN Neurobehavioral Development of Infant Learning and Memory: Implications for Infant Attachment TANIA L. ROTH, DONALD A. WILSON, AND REGINA M. SULLIVAN Evolutionary Significance of Sexual Cannibalism MARK A. ELGAR AND JUTTA M. SCHNEIDER Social Modulation of Androgens in Vertebrates: Mechanisms and Function RUI F. OLIVEIRA

A Functional Analysis of Feeding GEORGE COLLIER The Sexual Behavior and Breeding System of Tufted Capuchin Monkeys (Cebus apella) MONICA CAROSI, GARY S. LINN, AND ELISABETTA VISALBERGHI Acoustic Communication in Noise HENRIK BRUMM AND HANS SLABBEKOORN Ethics and Behavioral Biology PATRICK BATESON Prenatal Sensory Ecology and Experience: Implications for Perceptual and Behavioral Development in Precocial Birds ROBERT LICKLITER

470

CONTENTS OF PREVIOUS VOLUMES

Conflict and Cooperation in Wild Chimpanzees MARTIN N. MULLER AND JOHN C. MITANI

Trade-Offs in the Adaptive Use of Social and Asocial Learning RACHEL L. KENDAL, ISABELLE COOLEN, YFKE VAN BERGEN, AND KEVIN N. LALAND