No title

Advances in THE STUDY OF BEHAVIOR VOLUME 31 Advances in THE STUDY OF BEHAVIOR Edited by PETER J. B. SLATER JAY S. R...

Author: Peter J.B. Slater | Jay S. Rosenblatt | Charles T. Snowdon | Timothy J. Roper

23 downloads 502 Views 4MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Advances in

THE STUDY OF BEHAVIOR VOLUME 31

Advances in THE STUDY OF BEHAVIOR Edited by

PETER J. B. SLATER JAY S. ROSENBLATT CHARLES T. SNOWDON TIMOTHY J. ROPER

Advances in THE STUDY OF BEHAVIOR Edited by PETER J. B. SLATER School of Biology University of St. Andrews Fife, United Kingdom

JAY S. ROSENBLATT Institute of Animal Behavior Rutgers University Newark, New Jersey

CHARLES T. SNOWDON Department of Psychology University of Wisconsin Madison, Wisconsin

TIMOTHY J. ROPER School of Biological Sciences University of Sussex Sussex, United Kingdom

VOLUME 31

San Diego San Francisco New York Boston London Sydney Tokyo

∞ This book is printed on acid-free paper.

Copyright C 2002 by ACADEMIC PRESS All Rights Reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the Publisher. The appearance of the code at the bottom of the ﬁrst page of a chapter in this book indicates the Publisher’s consent that copies of the chapter may be made for personal or internal use of speciﬁc clients. This consent is given on the condition, however, that the copier pay the stated per copy fee through the Copyright Clearance Center, Inc. (222 Rosewood Drive, Danvers, Massachusetts 01923), 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-2002 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. 0065-3454/2002 $35.00 Explicit permission from Academic Press is not required to reproduce a maximum of two ﬁgures or tables from an Academic Press chapter in another scientiﬁc or research publication provided that the material has not been credited to another source and that full credit to the Academic Press chapter is given.

Academic Press A division of Harcourt, Inc. 525 B Street, Suite 1900, San Diego, California 92101-4495, USA http://www.academicpress.com

Academic Press Harcourt Place, 32 Jamestown Road, London NW1 7BY, UK http://www.academicpress.com International Standard Book Number: 0-12-004531-1 PRINTED IN THE UNITED STATES OF AMERICA 02 03 04 05 06 07 SB 9 8 7 6 5 4 3 2

1

Contents

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

ix xi

Conﬂict and Cooperation in a Female-Dominated Society: A Reassessment of the “Hyperaggressive” Image of Spotted Hyenas MARION L. EAST AND HERIBERT HOFER I. II. III. IV. V. VI. VII. VIII. IX. X. XI. XII. XIII. XIV.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Serengeti Hyena Social Organization . . . . . . . . . . . . . Conﬂict over Access to Food . . . . . . . . . . . . . . . . . . . Sibling Rivalry: A By-Product of Androgenization or an Adaptation to Fluctuating Levels of Prey? . . . . . . Siblicide and Biases in the Sex Composition of Twin Litters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Female Dominance and Virilization . . . . . . . . . . . . . . Virilization and Androgens . . . . . . . . . . . . . . . . . . . . The Fitness Consequences of Virilization . . . . . . . . . . The Male-Dominance Hierarchy: Queuing for Social Status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Male–Female Behavior . . . . . . . . . . . . . . . . . . . . . . . Natal Postdispersal Males . . . . . . . . . . . . . . . . . . . . . . Conﬂict Resolution . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

..... ..... .....

1 4 5

.....

8

. . . .

. . . .

. . . .

. . . .

. . . .

10 13 14 19

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

20 21 22 22 24 24 25

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 I. II. III. IV. V.

Introduction . . . . . . . . . . . . . . . . . . . . . . . Song: Mate Attraction and Male Contest . The Collared Dove as a Model System . . . Perch-Coo Variability among Individuals . Modulated Elements . . . . . . . . . . . . . . . . v

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

31 32 40 44 46

vi

CONTENTS

VI. Variation in Element Number . . . . . . . . . . . . . . . VII. Vocal Variability and Male Contest: Conclusions . VIII. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

57 62 64 65

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

77 79 81 85 87 97 99 100 101

. . . . . . . .

. . . . . . . .

. . . . . . . .

107 111 123 133 139 144 151 153

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Auditory Scene Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . Demonstrations of Auditory Scene Analysis by Songbirds Further Possibilities for Research on Auditory Scene Analysis in Nonhuman Animals . . . . . . . . . . . . . . . . . . . . . V. Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . .

163 164 169

. .

182 191

Imitation of Novel Complex Actions: What Does the Evidence from Animals Mean? RICHARD W. BYRNE I. II. III. IV. V. VI. VII. VIII.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Social Learning without Imitation . . . . . . . . . . . . . . Learning When and Learning How . . . . . . . . . . . . . Imitation: Social Mimicry and Picking Up Methods . Production Learning by Imitation . . . . . . . . . . . . . . What Limits Production Learning by Imitation? . . . Wider Implications . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

Lateralization in Vertebrates: Its Early Evolution, General Pattern, and Development LESLEY J. ROGERS I. II. III. IV. V. VI. VII.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Feeding and Prey Capture . . . . . . . . . . . . . . . . . . . . . Predator-Escape and Fear Responses . . . . . . . . . . . . Spatial Maps and Attention . . . . . . . . . . . . . . . . . . . . Aggressive Responses . . . . . . . . . . . . . . . . . . . . . . . . Advantages and Disadvantages of Being Lateralized . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . .

. . . . . . . .

Auditory Scene Analysis in Animal Communication STEWART H. HULSE I. II. III. IV.

vii

CONTENTS

VI. Final Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

195 196 196

Electric Signals: Predation, Sex, and Environmental Constraints PHILIP K. STODDARD I. II. III. IV. V. VI. VII. VIII.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . A Gymnotiform Primer . . . . . . . . . . . . . . . . . . . The Perils of Predators . . . . . . . . . . . . . . . . . . . . ’Twas Ever Thus: Sex versus Predators . . . . . . . . Electrosensory Influences on Niche Partitioning The Story Thus Far . . . . . . . . . . . . . . . . . . . . . . . Future Investigations . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

201 204 215 224 228 233 234 235 236

How to Vocally Identify Kin in a Crowd: The Penguin Model THIERRY AUBIN AND PIERRE JOUVENTIN I. II. III. IV. V. VI. VII.

Introduction . . . . . . . . . Locations and Methods The Constraints . . . . . . The Solutions Found . . Perspectives . . . . . . . . . Conclusion . . . . . . . . . . Summary . . . . . . . . . . . References . . . . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

243 245 248 254 269 271 272 273

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

279

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

287

This Page Intentionally Left Blank

Contributors

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

THIERRY AUBIN (243), NAM-CNRS URA 1491, Universit´e Paris-XI, F-91400 Orsay, France MECHTELD R. BALLINTIJN (31), Behavioural Biology, Institute of Evolutionary and Ecological Sciences, Leiden University, 2300 RA Leiden, The Netherlands RICHARD W. BYRNE (77), Scottish Primate Research Group, School of Psychology, University of St. Andrews, Fife KY16 9JU, Scotland, United Kingdom MARION L. EAST (1), Institute of Zoo and Wildlife Research, D-10315 ¨ Verhaltensphysiologie, D-82305 Berlin, Germany; and Max-Planck-Institut fur Seewiesen, Germany HERIBERT HOFER (1), Institute of Zoo and Wildlife Research, D-10315 ¨ Verhaltensphysiologie, D-82305 Berlin, Germany; and Max-Planck-Institut fur Seewiesen, Germany STEWART H. HULSE (163), Department of Psychology, Johns Hopkins University, Baltimore, Maryland 21218 PIERRE JOUVENTIN (243), CEFE-CNRS, 34293 Montpellier Cedex 5, France LESLEY J. ROGERS (107), Division of Zoology, School of Biological Sciences, University of New England, Armidale, New South Wales 2351, Australia HANS SLABBEKOORN (31), Center for Tropical Research and Department of Biology, San Francisco State University, San Francisco, California 94132 PHILIP K. STODDARD (201), Department of Biological Sciences, Florida International University, Miami, Florida 33199 CAREL TEN CATE (31), Behavioural Biology, Institute of Evolutionary and Ecological Sciences, Leiden University, 2300 RA Leiden, The Netherlands

ix

This Page Intentionally Left Blank

Preface

The aim of Advances remains as it has been since the series began: to serve the increasing number of scientists who are engaged in the study of animal behavior by presenting their theoretical ideas and research to their colleagues and to those in neighboring ﬁelds. We hope that the series will continue its “contribution to the development of cooperation and communication among scientists in our ﬁeld,” as its intended role was phrased in the preface to the ﬁrst volume in 1965. Since that time, traditional areas of animal behavior research have achieved new vigor through the links they have formed with related ﬁelds and through the closer relationship that now exists between those studying animal subjects and those studying human subjects. Scientists studying behavior today range more widely than ever before: from ecologists and evolutionary biologists to geneticists, endocrinologists, pharmacologists, neurobiologists, and developmental psychobiologists, not forgetting the ethologists and comparative psychologists, whose prime domain is the study of behavior. It is our intention not to focus narrowly on one or a few of these ﬁelds, but to publish articles covering the best behavioral work from a broad spectrum. The skills and concepts of scientists in such diverse ﬁelds necessarily differ, making the task of developing cooperation and communication among them a difﬁcult one. However, it is one that is of great importance, and one to which the editors and publisher of Advances in the Study of Behavior are committed. We will continue to provide the means to this end by publishing critical reviews, inviting extended presentations of signiﬁcant research programs, encouraging the writing of theoretical syntheses and reformulations of persistent problems, and highlighting especially penetrating research that introduces important new concepts. The aspirations of the series are well matched by the current volume. The nature of and evidence for imitation are certainly persistent problems, discussed by Byrne, as is the issue of how animals detect signals in the midst of noise, a remarkable capacity nicely illustrated and illuminated by two chapters (Aubin and Jouventin; Hulse). Behavior borders on ecology in Stoddard’s discussion of the selective forces shaping the signals of electric ﬁsh, on neurobiology in Rogers’ review of lateralization, and on physiology in East and Hofer’s discussion of the female-dominated social system of spotted hyenas. Finally, ten Cate, Slabbekoorn, and Ballintijn nicely show how combining ﬁeld and laboratory studies can shed light on something as apparently simple as the cooing of doves; there is certainly far more to it than meets the uninitiated ear. All of these chapters tackle important topics of wide signiﬁcance to those interested in the study of behavior. xi

This Page Intentionally Left Blank

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 31

Conﬂict and Cooperation in a Female-Dominated Society: A Reassessment of the “Hyperaggressive” Image of Spotted Hyenas MARION L. EAST and HERIBERT HOFER INSTITUTE OF ZOO AND WILDLIFE RESEARCH D-10315 BERLIN, GERMANY; AND

¨ VERHALTENSPHYSIOLOGIE MAX-PLANCK-INSTITUT FUR D-82305 SEEWIESEN, GERMANY

I. INTRODUCTION The spotted hyena (Crocuta crocuta) displays a suite of unusual traits that include female dominance, the masculinization of female genitalia (virilization), and sibling rivalry that can lead to the death of one sibling (siblicide). There has been much speculation about the evolution of these traits, probably because the assumed links between androgens, aggressive behavior, and masculinized traits in spotted hyena females point to fundamental issues concerning the evolution of sex-speciﬁc (aggressive) behavior (Gowaty, 1997; Hrdy, 1999). The prevailing assumption has been that elevated levels of androgens in female spotted hyenas produce large, hyperaggressive females that are able to outcompete males in feeding situations and thus acquire absolute dominance over males (Gould, 1981; Glickman et al., 1993; Frank et al., 1995; Frank, 1997; Hrdy, 1999). Some authors have even suggested that the assumed high levels of androgens in adult female spotted hyenas may induce pathological problems during reproduction such as difﬁculties in completing pregnancies or a delay in the onset of reproduction (Packer et al., 1995). Female spotted hyenas have an enlarged clitoris that is erectile and resembles the penis of the male (Matthews, 1939). The prevailing opinion on the evolution of this “pseudo-penis” has been that it was acquired as a by-product of selection for androgenized, aggressive females (Gould, 1981; Hamilton et al., 1986; Frank et al., 1991; Frank, 1996, 1997). Similarly, a phenomenon described as “habitual neonatal siblicide” in spotted hyenas has also been viewed as a by-product of selection for androgenized, aggressive females (Frank et al.,1991; Frank, 1996, 1997). 1

C 2002 by Academic Press Copyright All rights of reproduction in any form reserved. 0065-3454/02 $35.00

TABLE I EVIDENCE THAT SPOTTED HYENAS EVOLVED BEHAVIORAL AND LIFE HISTORY TRAITS IN THE CONTEXT OF CONFLICT RESOLUTION THAT ARE FLEXIBLE, RESOURCE-DEPENDENT, INDEPENDENT OF BODY SIZE, OR RELY ON CONFLICT AVOIDANCE Behavioral or life history trait or context

Manifestation

2

Behavioral response to territorial intruders

Highly context-speciﬁc, ranging from tolerance of intruders in transit to quickly escalated aggressive attacks on intruders at carcasses or kills

Territorial defense through encounters with neighboring clans

Territorial encounters are won by the side that recruits the largest number of clan members after the encounter begins

Acquisition of social status (females)

Linear dominance hierarchy based on aggressive and submissive interactions in female dyads

Acquisition of social status (immigrant males)

Immigrant males queue for social status: social status is based on tenure, agonistic encounters usually consist of submissive avoidance responses rather than aggressive dominance actions, and social status is independent of body size Natal males acquire top rank in the immigrant male social hierarchy when they start competing with immigrant males for access to females

Acquisition of social status (nondispersing natal males)

Remarks One of the few vertebrates where such ﬂexibility has been demonstrated; consistent with game theoretical arguments Group size, not the physical qualities of the participants, decides the outcome of the encounter Independent of body mass, muscular measures, or skeletal body size, but dependent on coalition quality (i.e., the social status of the mother) Queues contain a much higher number of members and last for a substantially longer period of time than other mating queues in vertebrates, yet queuing discipline is strict in comparison to queues described for other species “Nondispersal” natal males do not queue for high social status probably due to the status they acquired from their mothers when cubs. All nondispersal natal males submit to all females, irrespective of their mother’s rank

References Hofer and East, 1993b

Hofer and East, 1993b Frank, 1986b; Hofer and East, 1995a; Holekamp et al., 1996 East and Hofer, 2001

East and Hofer, 2001

3

Maintenance of social status (males)

Position in queue mostly maintained by coalitions among males and between high ranking males and females

Independent of measures of body size

East and Hofer, 2001

Rate of aggression between siblings

Most intense sibling rivalry occurs during suckling bouts; depends on food availability and maternal social status, which determines access to food resources in the territory

Consistent with avian models of facultative sibling rivalry

Golla et al., 1999

Age at ﬁrst reproduction (females)

Increases as social status declines

Reconciliation

Afﬁliative behavior after an agonistic encounter is more likely after escalated conﬂicts; winners of conﬂicts regularly initiate afﬁliative behavior with losers of conﬂicts Clan members regularly forage outside the territory in areas containing large concentrations of migratory herds when such herds are absent from the clan territory

Dependent on coalition quality (maternal social status) when daughters acquired their rank One of the few nonprimate species where reconciliation has been documented after conﬂicts

Hofer and East, 1996; Holekamp et al., 1996 Hofer and East, 2000

Avoidance of conﬂict with high-ranking clan members inside the territory

Hofer and East, 1993a,b,c

Commuting system

Acoustic defense of resources

Rate of loud calling (whoops) by females is related to social status, prey abundance, level of intrusion and food competition, and expected beneﬁts in terms of increased food intake and improved cub survival

The rate of acoustic defense of the territory and the resources it contains vary with both intrusion pressure and expected ﬁtness beneﬁts through the resources that are being defended

East and Hofer, 1991a,b

Greeting ceremonies

Ritualized greeting in which the erect “penis” is used as a signal of submission and clan members compete for “valuable” greeting partners

Individuals value these afﬁliative interactions and chose greeting partners likely to provide maximum social beneﬁts in accordance with primate models of optimal partner choice

East et al., 1993

4

MARION L. EAST AND HERIBERT HOFER

Given the volume of literature that has promoted the idea that female dominance in spotted hyenas evolved due to selection favoring androgenized, large, hyperaggressive females, it is perhaps surprising that this idea has received little critical assessment. In this review we will use data from our long-term research on a population of several hundred individually known spotted hyenas in several social groups in the Serengeti National Park, Tanzania to examine this prevailing view. Using current knowledge of the behavior, ecology, and endocrinology of free-ranging spotted hyenas, we will present an alternative evolutionary scenario for the occurrence of habitual male submission resulting in female dominance in this species. We will outline a scenario that links elevated levels of androgens in neonates of both sexes with selection for early sibling rivalry leading to resourcedependent, facultative siblicide (East et al., 1993; Hofer and East, 1995a, 1997; Golla et al., 1999) and producing as a side effect virilized genitalia in female offspring (East et al., 1993; Hofer and East, 1995a). During our discussion of these scenarios we will consider some of the conﬂicts that exist within spotted hyena society and the mechanisms used to resolve them. Throughout our review we aim to highlight how both conﬂict and cooperation shape hyena society and to emphasize that spotted hyenas most likely use social relationships with other group members as an important resource (Table I). This view of social relationships coincides with a major paradigm shift among many animal behavior researchers (Aureli and de Waal, 2000) who view aggression not as the driving force that determines the fabric of social life, but rather as one of several forms of conﬂict resolution that are in an evolutionary balance with mechanisms of conﬂict avoidance and with afﬁliative behaviors that encourage the maintenance of social relationships. This argument is particularly pertinent for long-lived, slowly reproducing, and highly social mammals, such as spotted hyenas, where the social environment may exert as strong a selective force as the natural environment. We thus hope that our review of the evidence relating to the assumed links between aggressive behavior, androgens, and masculinization in spotted hyenas leads to a more balanced appraisal of this topic.

II. SERENGETI HYENA SOCIAL ORGANIZATION Similar to spotted hyenas elsewhere (Kruuk, 1972; Frank, 1986a; Henschel and Skinner, 1987; Mills, 1990; Hofer, 1998), Serengeti hyenas live in stable social groups called clans that defend territories against neighboring clans (East and Hofer, 1991a,b; Hofer and East, 1993a). Females remain in their natal clan throughout their life. As in other spotted hyena populations (Hofer, 1998), there is a strong social hierarchy among female Serengeti

CONFLICT AND COOPERATION IN A FEMALE-DOMINATED SOCIETY

5

clan members (East and Hofer, 1991b; East et al., 1993), with high-ranking females maintaining priority of access to food resources (Hofer and East, 1996). Coalitions among Serengeti females are important for the acquisition and maintenance of social status (Table I). As a consequence of such coalitions, the social status of adult females is highly positively correlated with that of their mother (most daughters acquire a social status just below that of their mother). Also, adult females may dramatically decline in social status after the death of a strong coalition partner, such as a mother or adult daughter (Hofer and East, 1996; Hofer and East, unpubl. data). All adult females breed and their cubs are typically reared in a communal den inside the clan territory (Hofer and East, 1995b). Serengeti males normally disperse from their natal clan after reaching sexual maturity at about 2 years of age (Hofer and East, 1995b). When males disperse into a new clan, they join the immigrant male dominance hierarchy at the bottom. Immigrant males do not ﬁght for social status, but instead low-ranking males with short tenure increase in rank as higher ranking, longer tenured males die or leave the clan (Table I). Thus, immigrant males can be considered to queue for social status (East and Hofer, 1991b; East et al., 1993; East and Hofer, 2001). All immigrant males are subordinate to all female clan members; males do not participate in parental care.

III. CONFLICT OVER ACCESS TO FOOD A. THE COMMUTING SYSTEM In the Serengeti National Park, the herbivore biomass is dominated by migratory species, chieﬂy wildebeest (Connochaetes taurinus), Thomson’s gazelles (Gazella thomsoni), and zebras (Equus burchelli), and these species are the primary prey of Serengeti spotted hyenas (Hofer and East, 1993a). Due to the movements of these migratory herbivores, the abundance of prey within clan territories ﬂuctuates markedly. Large numbers of herbivores are present within any given territory for only 22–30% of any year and, during such periods, the abundance of herbivores is high (mean of 238 animals/km2). Irrespective of social status, all clan members forage within the clan territory when large migratory herds are present (Hofer and East, 1993b). When such herds are absent from a territory, the abundance of prey is dramatically reduced to a mean of 31 animals/km2 during periods of medium prey abundance (small herds of migratory and resident herbivores are present) and to a mean of 3.3 animals/km2 during periods of low prey abundance when only resident herbivores are present (Hofer and East, 1993a,b).

6

MARION L. EAST AND HERIBERT HOFER

Conﬂict between clan members over access to food resources in the clan territory increases as the abundance of herbivores declines. One method of decreasing conﬂict over access to declining food resources is to leave the clan territory on short-term (6–8 days for nonlactating animals), longdistance (40–80 km) foraging trips to the nearest concentration of migratory herbivores (Table I). The proportion of clan members regularly commuting to forage outside the territory increases as the abundance of herbivores declines (Hofer and East, 1995b). Individuals of low social status regularly commute when there are moderate densities of herbivores in the group territory. In contrast, females of high social status only commute when conditions of low prey abundance prevail. Although commuting decreases conﬂict with socially dominant group members, commuting dramatically increases foraging effort. For example, nonlactating, low-ranking females undertake approximately 15–18 commuting trips annually, and such trips entail traveling a minimum average distance of 1000–1300 km per year (Hofer and East, 1993c). B. TERRITORIAL DEFENSE AND INTRUSION PRESSURE The evolution of the commuting system in Serengeti spotted hyenas does not indicate near-pathological hyperaggression among females. Rather, it demonstrates considerable behavioral ﬂexibility that allows the year-round exploitation of abundant migratory prey, copes with high levels of intrusion by commuters, and permits a larger group size than would be feasible given the low abundance of resident prey in group territories (Hofer and East, 1993a,b). Clan territory owners have a highly context-speciﬁc response to intruding hyenas. Intruders in transit, moving swiftly along recognized thoroughfares through a clan territory, are tolerated or ignored by residents, whereas foraging intruders are usually subjected to low-level aggressive challenges by residents when the latter encounter the former. Territory owners escalate aggressive attacks at food resources if intruders do not immediately relinquish the prey they have killed or carcass they have scavenged (Hofer and East, 1993b). Such ﬂexible, context-speciﬁc adjustment of the territorial response to intruders is expected from game-theoretical arguments (Hofer and East 1993b) but, to our knowledge, our study of Serengeti hyena territoriality is the only example where such a ﬂexible territorial response has been demonstrated in a carnivore society (Table I). Territory owners rarely need to resort to physical contests as they invest in olfactory (Mills and Gorman, 1987; Hofer et al., 2001) and acoustic signals to advertise ownership of resources in their territory (East and Hofer, 1991a,b). Clan females “loud call” or whoop most often when their territory contains a high abundance of herbivores. Thus females adjust their vocal defense of resources in their territory in relation to (1) levels of competition from

CONFLICT AND COOPERATION IN A FEMALE-DOMINATED SOCIETY

7

intruding commuters and other clan members and (2) the expected beneﬁt that an increased abundance of carcasses provides in terms of food and high cub growth rates (Hofer and East, 1993b,c). High-ranking resident females who stand to beneﬁt the most from ensuring that they have priority of access to carcasses also whoop at the highest rates (East and Hofer, 1991b). C. MATERNAL CARE Parental effort (the effort required to obtain resources) and parental input (the quality and quantity of resources delivered to offspring that are converted into growth; Evans, 1990) comprise parental care. Beneﬁts of parental care are usually measured in terms of the survival and reproductive success of the offspring. Although the cost of parental care is the reduced ﬁtness obtained from future offspring caused by the level of care invested in current offspring (Trivers, 1972), parental energy expenditure has often been considered an appropriate approximation of parental costs (Clutton-Brock, 1991). Due to ﬂuctuating levels of conﬂict over access to food resources in Serengeti clan territories, the options available to females with respect to parental effort and parental care differ according to social status. During periods of high prey abundance, maternal effort is low, maternal input is high as offspring are nursed each day, and thus, cub growth rate is high irrespective of the social status of mothers (Hofer and East, 1993c, 1996; Golla et al., 1999). In contrast, all Serengeti females, irrespective of social status, regularly leave the clan territory to forage on distant migratory prey during periods of low prey abundance. Maternal effort is high because mothers have to travel long distances to foraging areas and return to the clan territory to nurse their offspring at the communal den. Maternal input is low because commuting mothers are absent from their offspring for several (2–9) days. Mothers do not compensate for infrequent nursing of their young by increasing milk quality or volume delivered during nursing (Hofer and East, 1993c, 1996). Consequently, the comparatively low maternal input during commuting periods results in minimum or even negative growth rates of offspring over periods of several weeks (Hofer and East, 1993c, 1996; Golla et al., 1999). During periods of medium prey abundance, females of high social status monopolize low-cost foraging opportunities inside the clan territory, whereas females of medium and low status continue to exercise the high-cost foraging option of traveling long distances to feed on migratory prey. By remaining “at home,” females of high status ensure that their parental effort is kept as low as possible, and they continue to nurse their offspring frequently. By comparison, commuting medium- and low-ranking females have a substantially higher maternal effort and lower maternal input. Medium- and low-ranking lactating females undertake between 42–51 commuting trips

8

MARION L. EAST AND HERIBERT HOFER

annually, traveling a minimum average distance of 2900–3700 km, whereas high-ranking mothers may travel only half this distance (Hofer and East, 1993c, and unpubl. data). These commuting distances exceed the total distance covered by the migratory herds throughout their annual migratory cycle by a factor of 3 or more (Hofer and East, 1995b). The two foraging options available to Serengeti mothers substantially differ in their consequences for maternal effort (costs to the mother) and maternal input (beneﬁts to the cubs). For instance, the average one-way distance of a commuting trip of 40 km exceeds by an order of magnitude the distance of 4 km that is equivalent to the approximate radius of territories defended by clans in our study (Hofer and East 1993b, 1995b). Similarly, commuting mothers nurse their cubs on average at intervals of 3.4 days, which is substantially less often than mothers feeding inside the clan territory who nurse their cubs typically twice a day (Hofer and East, 1993c). Thus, commuting is clearly the worse option, yet low- and mid-ranking females spend most of the year commuting to avoid conﬂict over food with high-ranking clan members. Offspring of high-ranking females generally grow faster and survive better than those of low- and mid-ranking mothers because their mothers only commute during periods of low prey abundance (Hofer and East, 1996; Golla et al., 1999). These effects have far-reaching consequences for the reproductive careers of daughters. Parental input by Serengeti mothers, measured in terms of cub growth rate to independence, is a signiﬁcant predictor of adult mass, adult skeletal size, and age at ﬁrst parturition of daughters (Hofer and East, 1996). Age at ﬁrst parturition is considered a key life history trait (Roff, 1992; Stearns, 1992) because it determines, among other things, the expected duration of the lifetime reproductive period. In species such as spotted hyenas, with small litter size and long interbirth intervals due to their long period of offspring dependence, total duration of the reproductive period is a signiﬁcant predictor of lifetime reproductive success (Clutton-Brock, 1988). Since daughters of high-ranking Serengeti mothers grow faster, obtain a higher social status, and reproduce earlier than daughters of low-ranking mothers, offspring quality in Serengeti hyenas is strongly and positively correlated between generations (Hofer and East, 1996).

IV. SIBLING RIVALRY: A BY-PRODUCT OF ANDROGENIZATION OR AN ADAPTATION TO FLUCTUATING LEVELS OF PREY? Intense sibling rivalry, leading to the death of a sibling, has been observed in numerous avian species (Mock and Parker, 1997). Such siblicide has been described as obligate if rivalry invariably results in the death of a sib and

CONFLICT AND COOPERATION IN A FEMALE-DOMINATED SOCIETY

9

facultative if brood reduction only occurs when parental food provisioning falls below that required to sustain the entire brood (Mock and Parker, 1997). Spotted hyenas are one of the few mammalian species that exhibit sibling rivalry leading to siblicide. Frank et al. (1991) and Frank (1996, 1997) considered siblicide in spotted hyenas to be independent of resources, “habitual” (i.e., presumably obligate) in twin litters composed of cubs of the same sex, and a by-product of selection for large, aggressive, androgenized females that could dominate males. In contrast, we have described in detail cases of resource-dependent, facultative siblicide in Serengeti spotted hyenas (Hofer and East, 1997). Variation in the intensity of sibling rivalry within Serengeti litters conforms to predictions derived from avian models of facultative siblicide (Golla et al., 1999), and facultative siblicide in Serengeti litters normally occurs within the ﬁrst 2 months of life (Hofer and East, 1997; Golla et al., 1999). “Habitual” siblicide in same-sex twin litters has not been observed during our long-term study (East and Hofer, 1997; Hofer and East, 1997). Smale et al. (1999) also argued that siblicide in spotted hyenas was unlikely to be “habitual,” and more likely to be facultative, due to the low incidence of siblicide in their study clan in the Masai Mara of Kenya. In the Serengeti, litter size is one or two, rarely three (Hofer and East, 1997), and cubs are born throughout the year (Hofer and East, 1995b). Mothers normally nurse only their own offspring, and cubs are stationed at the communal den for approximately 12 months (Hofer and East, 1993c). Due to a long period of lactation (12–18 months), Serengeti mothers experience large ﬂuctuations in prey abundance in their territory irrespective of the timing of parturition (Hofer and East, 1995b). The milk of spotted hyenas is highly nutritious (protein content 14.9%, fat content 14.1%, gross energy density 9.70 kJ/g), with a protein content higher than that recorded for any other land carnivore, a fat content only exceeded by Palearctic bears and sea otters (Enhydra lutris), and a gross energy density higher than that of most other land carnivores (Hofer and East, 1995b). Given the considerable maternal investment of spotted hyenas in terms of high-quality milk and long lactation period, the existence of siblicide in this species might be expected, as a siblicidal survivor would gain exclusive access to highly nutritious maternal milk during the prolonged period of lactation. Although a siblicidal cub might gain an additional share of the maternal milk supply, a cub should only sacriﬁce a littermate when the increase in ﬁtness acquired by such an act is greater than the ﬁtness cost incurred (Parker et al., 1989; McNamara et al., 1994). Such a situation will occur when maternal food supply is insufﬁcient to sustain the entire litter (Table I), and under such circumstances, the dominant offspring should seek to improve its chance of survival by monopolizing maternal resources (Mock et al., 1990).

10

MARION L. EAST AND HERIBERT HOFER

Severe sibling rivalry between spotted hyena siblings results in the exclusion by the dominant cub of the subordinate cub from access to maternal teats. This results in large asymmetries in body mass between littermates and the eventual death due to starvation of the subordinate offspring (Hofer and East, 1997). If, as theory predicts, insufﬁcient parental provisioning causes despotic behavior by the dominant offspring (Mock et al., 1990; McNamara et al., 1994; Mock and Parker, 1997), then escalated sibling rivalry within spotted hyena twin litters should be expected when mothers fail to provide sufﬁcient milk to sustain a litter. Consistent with this prediction, rates of aggression among Serengeti littermates increased as levels of maternal input, measured in terms of cub growth, decreased (Golla et al., 1999). Rates of aggression were calculated as the frequency of aggressive behaviors per minute during suckling bouts derived by a bout criterion analysis (Sibly et al., 1990). Also in accordance with this prediction, littermates belonging to highranking mothers and receiving high levels of maternal input (see earlier) had lower rates of aggression than those belonging to lower ranking mothers that provided less adequate levels of input (Golla et al., 1999). Thus all the evidence currently available suggests that siblicide in spotted hyenas is facultative and resource-driven, just as in any other species with facultative siblicide, and not a pathological by-product of selection for another trait.

V. SIBLICIDE AND BIASES IN THE SEX COMPOSITION OF TWIN LITTERS Facultative siblicide in spotted hyenas is an interesting phenomenon, as enforced starvation does not appear to occur randomly with respect to the sex composition of twin litters. Twin litters may be composed of cubs of the same sex (i.e., both male or both female) or of mixed sex (male and female cubs). Considering the sex composition of 114 twin Serengeti litters at the age of 3 months, when cubs can ﬁrst be sexed in the ﬁeld by the shape of their phallic glans (Frank et al., 1990), mixed-sex litters occurred signiﬁcantly more often than expected, whereas there were too few all-female litters (Hofer and East, 1997). This suggests that facultative siblicide is more likely to occur in all-female than all-male or mixed-sex twin litters. If facultative siblicide most often occurs in all-female twin litters, then many all-female twin litters will be represented by a singleton female after siblicide has occurred (Hofer and East, 1997). The overall sex ratio expressed as the proportion of males in singleton Serengeti litters was 0.429 (N = 91 cubs), a signiﬁcant female bias consistent with the idea that females are more likely to commit siblicide (Hofer and East, 1997).

CONFLICT AND COOPERATION IN A FEMALE-DOMINATED SOCIETY

11

If females are more likely to commit siblicide, then the twin litters that survive intact to the age of 3 months should show a male bias. In the Serengeti, the proportion of males in twin litters at this age was 0.557, a bias toward male cubs. James and Hofer (1999) showed that siblicide could only account for the subbinomial variance of the sex composition of Serengeti twin litters at the age of 3 months if the probability of siblicide depended on the sex of the victim, or if it was caused by preparturition processes, as suggested by Smale et al. (1999). East and Hofer (1997) suggested that if facultative siblicide is most common in all-female litters, then the sex composition of twin litters in spotted hyena populations should alter with the amount of food available to lactating mothers. If so, the low incidence (8%) of all-female litters in a Masai Mara clan, when group size was considered at carrying capacity, and the higher incidence of such litters (41%) after ﬁssion reduced group size to below carrying capacity (Holekamp and Smale, 1995), may have been caused by a decline in facultative siblicide after ﬁssion (East and Hofer, 1997). A comparison of the sex composition of twin litters in ecosystems with large differences in prey abundance shows that the proportion of all-female twin litters increased as the availability of resident prey increased (Table II, Spearman’s ρ = 0.886, N = 6, P < 0.05). The highest incidence of all-female (twin/triplet) litters was observed, as predicted, in a well-fed captive colony of spotted hyenas. Sex-speciﬁc biases in sibling competition may result from unequal investment in offspring of different sex (Clutton-Brock, 1991), causing more extreme competition for parental resources by the sex that receives more parental investment (Mock and Parker, 1997). In Serengeti spotted hyenas, singleton female cubs grow at a higher rate than singleton males, suggesting that investment in females is higher than in males (H. Hofer and M. L. East, unpubl. data). Also, as previously mentioned, females remain in their natal groups and normally acquire a social rank immediately below that of their mother, with social status directly inﬂuencing access to food resources. Thus, due to philopatry among females, higher levels of competition may be expected within all-female litters than within all-male litters (Holekamp and Smale, 1995; Golla et al., 1999). Aggression rates among Serengeti cubs within all-male and all-female litters were higher than in mixed-sex litters, but did not differ between all-male and all-female litters (Golla et al., 1999). These results may be due to escalated contests between competitors with identical interests (Maynard Smith, 1982). More likely, they reﬂect the combined aggressive behavior of both siblings and not the tactics used by dominant and subordinate contestants during competition for access to maternal milk supplies (Golla et al., 1999).

TABLE II RESIDENT PREY DENSITIES (IN ANIMALS/KM2), AGGRESSION RATES WITHIN TWIN LITTERS (AGGRESSIVE MOVES/MIN), OCCURRENCE OF SIBLICIDE, AND SEX-RATIO BIASES IN DIFFERENT ENVIRONMENTSa

Ecosystem Kalahari Serengeti

Prey density 2.6b 3.3

Sex composition of litters

Aggression rate

Facultative siblicide

? 0.49 ± 0.31

? Yes

89∗∗ 61∗∗

11 25

% Mixed-sex

% All-male

% All-female

N

References

0§ 14∗∗

9 114

Mills, 1990; Hofer and East, 1993a Hofer and East, 1993a,1997; Golla et al., 1999 Frank et al., 1991 Holekamp et al.,1993; Holekamp and Smale, 1995; Smale et al., 1995, 1999 Holekamp et al., 1997; Smale et al., 1995, 1999 B. Wachter and O. P. Honer, ¨ unpubl. data Drea et al., 1996; Frank, 1996

Masai Marac Masai Marad

? 153e

? 0.035e,i

Yes Possible

85∗∗∗ 81∗∗

12§§§ 13

Masai Marab

163e

0.035e,i

No g

48

27

25

48

Ngorongoro

184.2

0.023 ± 0.06

No

49

31

20

35

0.05e

Possibleh

56

17

28

18

Captivity

Ad libitum

3∗∗∗ 6§§

33 16

Note: §p = 0.075; §§p = 0.063; §§§p = 0.058; ∗∗ p < 0.005; ∗∗∗ p < 0.001. Sex composition of twin litters from binomial expectations is 50% mixed-sex, 25% all-male, and 25% all-female litters. b Minimum estimate. c Study period 1979–1987. d Study period 1988–1990 before clan ﬁssion, high competition between females. e Derived from graphs or data published in the original citations. f Study period 1990–1998 after clan ﬁssion, low competition between females. g No evidence for siblicide in twin litters but one case in a triplet litter. h The subordinate in a twin litter and the runts in all four triplet litters sustained serious injuries and observers intervened. i Aggression rates may have been recorded before and after clan ﬁssion. a

CONFLICT AND COOPERATION IN A FEMALE-DOMINATED SOCIETY

13

VI. FEMALE DOMINANCE AND VIRILIZATION The female genitalia of spotted hyenas are virilized and resemble the genitals of the males. The clitoris is enlarged and forms an erectile pseudopenis, the labia are fused to form a pseudo-scrotum, and the urogenital tract passes through the pseudo-penis (Matthews, 1939; Neaves et al., 1980; Frank et al., 1990). Female spotted hyenas thus urinate, copulate, and give birth through their clitoris. Several hypotheses have been advanced to explain the virilization of female genitalia in spotted hyenas. The nonadaptive theories view female virilization as an incidental by-product of selection for elevated prenatal androgen levels. Such elevation occurred because (1) it lead to large size, increased aggression, and social dominance of adult females over males (Gould, 1981; Frank et al., 1991; Frank, 1997); it was (2) a consequence of a punctuated genetic translocation (van Jaarsveld and Skinner, 1987); it was (3) favored in the evolution of sibling rivalry (East et al., 1993, Hofer and East, 1995a). Adaptive theories considered the enlarged clitoris to have the direct selective advantage of a conspicuous structure used to facilitate the reestablishment of social bonds (Wickler, 1965; Kruuk, 1972). Another theory, which is an adaptive extension of the nonadaptive theory of Gould (1981), suggests that the virilization of female genitalia evolved as a by-product of selection for androgen-facilitated dominance and that the virilized clitoris was incorporated into greetings, favoring selection for its enlargement and erectile nature (Hamilton et al., 1986). What evidence is there to support the idea that virilization of the female external genitalia was driven by selection for androgen-facilitated dominance? We will look at the evidence on androgen concentrations in the next section and concentrate here on the behavior. Adult female spotted hyenas are commonly assumed to be highly aggressive (Hamilton et al., 1986; Frank et al., 1991, 1995; Glickman et al., 1993; Monaghan and Glickman, 1992; Packer et al., 1995), even though levels of aggression among female spotted hyenas have not been compared with those of females in other mammalian species living in a similar social context of multimatrilineal groups that live in a ﬁssion–fusion society (Goymann, 1999). Rates of aggression among competing immigrant males are low, and males are rarely aggressive to females (East and Hofer, 2001). Thus, females may appear to be aggressive because the aggressiveness of immigrant males is so low (East et al., 1993). In mammalian species other than the spotted hyena (e.g., European mole Talpa europaea: Matthews, 1935; platyrrhine monkeys: Ploog and MacLean, 1963) virilization of female external genitalia is rare but where it occurs it is not associated with female dominance (East et al., 1993; Hofer and East, 1996; Teltscher, 1996). In mammalian species, female dominance is uncommon, but does occur in a variety of taxa (Ralls, 1976; Hrdy, 1981) and is typical for

14

MARION L. EAST AND HERIBERT HOFER

Lemuriformes (Jolly, 1984; Kappeler, 1990; Young et al., 1990). However, extreme masculinization of female external genitalia has not been recorded in any female-dominated species except for the spotted hyena (Teltscher, 1996). Female dominance in other mammals is not dependent on females being larger than males, and dominance is not a function of aggressiveness (Rowell, 1974). Even in spotted hyenas, females are not necessarily larger than males (Hamilton et al., 1986; East and Hofer, 1997, unpubl. data), and body size in males and females is not related to social status in our Serengeti population (East and Hofer 1997, 2001), nor in a clan in the Masai Mara of Kenya (Holekamp et al., 1996). In several primate species (Saimiri, Cebus, Cercopithecus talapoin, Macaca mulatta, Erythrocebus patas), coalitions of females can successfully challenge larger males (Harvey et al., 1987; Smuts, 1987), and in spotted hyenas coalitions are important in the maintenance of female rank (East et al., 1993; Hofer and East, 1996).

VII. VIRILIZATION AND ANDROGENS It has been argued that spotted hyena females have androgen proﬁles that are unusual for female mammals. On an interspeciﬁc level, they are thought to have higher androgen levels than other female mammals (e.g., Glickman et al., 1993; Packer et al., 1995; Frank, 1996). On an intraspeciﬁc level, the quoted “aggressiveness” of females is assumed to be caused by elevated levels of androgens. What is the evidence? Comparative data demonstrate that testosterone and androstenedione levels in nonpregnant free-ranging spotted hyena females are below or similar to those in a variety of other nonpregnant female mammals, including those in female brown and striped hyenas (Tables III and IV). In intraspeciﬁc comparisons, some studies reported similar testosterone levels in male and female spotted hyenas, whereas others reported significantly higher testosterone levels in males than in females (Table V). Androstenedione levels were similar in females and males in some studies and signiﬁcantly higher in females than in males in other studies (Table VI). Confusion about androgen levels in spotted hyenas probably stems from three sources (Goymann et al., 2001). First, some studies have not considered the impact of reproductive status on androgen levels in males. This is a problem because androgen levels in young, nonbreeding natal males (“predispersal” males) are signiﬁcantly lower (Holekamp and Smale, 1998) than levels in older, mostly immigrant, reproductively active males (“postdispersal” males) (Table V and VI). Second, since the challenge hypothesis (Wingﬁeld et al., 1990) predicts that androgen levels should increase when individuals experience elevated levels of social challenge, androgen levels in

CONFLICT AND COOPERATION IN A FEMALE-DOMINATED SOCIETY

15

TABLE III EXAMPLES OF PUBLISHED DATA ON PLASMA TESTOSTERONE (ng/ml) IN NONPREGNANT FEMALE CARNIVORES AND OTHER MAMMALS

Species Spotted hyena (Crocuta crocuta) Brown hyena (Hyaena brunnea) Striped hyena (Hyaena hyaena) Domestic dog (Canis familaris) Human (Homo sapiens) Chimpanzee (Pan troglodytes) Baboon (Papio hamadryas) Talapoin monkey (Miopithecus talapoin) Syrian hamster (Mesocricetus auratus) Musk shrew (Suncus murinus) European mole (Talpa europaea)

Plasma concentration (ng/ml) 0.32 0.52 0.64 0.05–0.29 0.30–0.60 0.35–0.73 0.15–0.23 2.2–4.0 0.06–0.09 0.20 0.79–1.44

References Goymann et al., 2001 Racey and Skinner, 1979 Racey and Skinner, 1979 Concannon and Castracane, 1985 Feder, 1985 Nadler et al., 1985 Feder, 1985 Batty et al., 1986 Feder, 1985 Rissman et al., 1990 Whitworth et al., 1999

captive (e.g., Glickman et al., 1987, 1992) and free-ranging spotted hyenas may not be comparable. Unlike free-ranging spotted hyenas that live in ﬁssion–fusion societies, captive individuals are housed together, have constant social contact, and have limited opportunities to avoid interactions. These socially challenging conditions are likely to result in elevated levels of androgens. Following the same line of argument, Serengeti spotted hyenas that are caught in wire snares set by game meat hunters, but escape by biting through the tethering wire (Hofer et al., 1993), experience an elevated level of social challenge. This is because the snare is often embedded in the TABLE IV EXAMPLES OF PUBLISHED DATA ON PLASMA ANDROSTENEDIONE (ng/ml) IN NONPREGNANT FEMALE CARNIVORES AND OTHER MAMMALS

Species Spotted hyena (Crocuta crocuta) Brown hyena (Hyaena brunnea) Domestic dog (Canis familaris) Human (Homo sapiens) Baboon (Papio hamadryas) Talapoin monkey (Miopithecus talapoin) Syrian hamster (Mesocricetus auratus) European mole (Talpa europaea)

Plasma concentration (ng/ml) 2.44 4.70 0.41–0.79 2.25 1.2–2.0 3.6–6.0 1.0–1.9 1.22

References Goymann et al., 2001 Racey and Skinner, 1979 Concannon and Castracane, 1985 Cashdan, 1995 Feder, 1985 Batty et al., 1986 Feder, 1985 Whitworth et al., 1999

TABLE V MEAN TESTOSTERONE CONCENTRATIONS (ng/ml PLASMA) IN MALE AND FEMALE SPOTTED HYENAS OF DIFFERENT REPRODUCTIVE STATE

16

Female, neither pregnant nor lactating

Pregnant female

0.76 (5) — 0.5a (1) 0.35 (12) 0.46 (16) 0.3a,b (11) 0.75 (12)

1.20 (10) 1.33 (4) 0.5a (1) — 1.24 (8) — —

1.06 (5) — — — — — —

1.2a (5)

—

0.6a (6)

0.4a,b (8) 0.3a,b (5) — — 0.32 ± 0.12 (7)

— 0.3–7.0a,b (5) 1.7a,b (3) — 0.29 ± 0.02 (3)

Lactating female

— — — — 0.16 ± 0.02 (24)

Young, natal male before dispersal — — — — — — — 1.5a,c (6) — — — 0.3d (12) 0.51 ± 0.20 (4)

Sexually mature, reproductively active male (usually immigrants) — — — — — 1.5a,b (16) —

Male of unknown reproductive state 1.48 (18) — — 3.15 (9) 0.84 (4) — 4.52 (6)

10.5a (3)

—

3.6 a,b (6) — — 1.30 (30) 4.26 ± 0.77 (19)

— — — — —

References Racey and Skinner, 1979 Lindeque and Skinner, 1982 Gombe, 1985 Frank et al., 1985 Lindeque et al., 1986 Glickman et al., 1987 van Jaarsveld and Skinner, 1991a van Jaarsveld and Skinner, 1991b Glickman et al., 1992 Licht et al., 1992 Drea et al., 1998 Holekamp and Smale, 1998 Goymann et al., 2001

Note: Numbers in parentheses refer to “sample sizes” (i.e., some references may contain repeat samples from the same individual, thus, sample sizes are not necessarily identical with the number of individuals sampled). a The original paper did not supply numeric means; thus, mean values have been approximated from published graphs. b Data from captive individuals. c Described as “resident natal males,” included subadults older than one year. d Labeled as “natal males.”

TABLE VI MEAN ANDROSTENEDIONE CONCENTRATIONS (ng/ml PLASMA) IN MALE AND FEMALE SPOTTED HYENAS OF DIFFERENT REPRODUCTIVE STATE

Female, neither pregnant nor lactating

17

— — 1.92 (16) 2.7a,b (11) 1.67 (12) 2.1a (5) 4.5a,b (8) 3.42b (5) 2.44 ± 1.02 (7)

Sexually mature, reproductively active male (usually immigrants)

Male of unknown reproductive state

— —

0.66 (18) — 0.97 (4) — 1.14 (6)

Lactating female

Young, natal male before dispersal

— — — — —

— — — — —

—

1.6a (6)

1.0a,c (6)

3.4a (3)

—

— 6.06b (5) 2.32 ± 1.03 (3)

— — 1.14 ± 0.21 (24)

— — 0.18 ± 0.06 (4)

2.5a,b (6) — 1.57 ± 0.28 (19)

— — —

Pregnant female 1.18 (10) 11.1 (3) 6.79 (8) — —

1.5a,b (16) —

References Racey and Skinner, 1979 Lindeque and Skinner, 1982 Lindeque et al., 1986 Glickman et al., 1987 van Jaarsveld and Skinner, 1991a van Jaarsveld and Skinner, 1991b Glickman et al., 1992 Licht et al., 1992 Goymann et al., 2001

Note: Numbers in parentheses refer to “sample sizes” (i.e., some references may contain repeat samples from the same individual, thus, sample sizes are not necessarily identical with the number of individuals sampled). a The original paper did not supply numeric means, thus, mean values have been approximated from published graphs. b Data from captive individuals. c Described as “resident natal males,” included subadults older than one year.

18

MARION L. EAST AND HERIBERT HOFER

ﬂesh and the wound created by the snare, and the wire protruding from the ﬂesh draw the attention of other clan members, including individuals with higher social status than the snared animal, and contact with the wound probably causes pain. Elevated levels of social challenge may explain why nonpregnant females with snares have elevated levels of androgens in comparison to nonpregnant females without snares (Goymann, 1999). Third, studies have applied different analytical techniques for the measurement of androgen levels (Goymann et al., 2001). Often studies purporting to measure testosterone do not distinguish between testosterone and 5-αdihydrotestosterone, as common antisera against testosterone cross-react highly with 5-α-dihydrotestosterone, and vice versa, and it requires special techniques to prevent these kinds of cross-reactions from biasing the results. For the ﬁrst time in hyena studies, Goymann et al. (2001) measured testosterone, 5-α-dihydrotestosterone, and androstenedione concentrations in plasma samples from Serengeti spotted hyenas after partial puriﬁcation through chromatography by using a modiﬁcation (Fusani et al., 2000) of the method described by Wingﬁeld and Farner (1975) and thus eliminating the usual cross-reactions. Thus Goymann et al. (2001) provide a highly speciﬁc measure of these three androgens. The results show that free-ranging spotted hyena females follow the standard mammalian pattern, with testosterone concentrations in reproductively active males being an order of magnitude higher than in females (Table V), and androstenedione levels being similar in both sexes (Table VI). In summary, there is little evidence to support the idea that androgen levels in adult female spotted hyenas are abnormally high, or likely to cause hyperaggression or pathological problems during reproduction. We have proposed an alternative idea that links prenatal androgen levels to facultative siblicide in spotted hyenas (East et al., 1993; Hofer and East, 1995a; East and Hofer, 1997). We consider it probable that the elevated levels of androgens (androstenedione in females, testosterone in males) recorded during fetal development, at birth, and during the initial months of life (Glickman et al., 1987; Frank et al., 1991; Licht et al., 1992) have been selected to produce neonates primed for high levels of sibling rivalry (East et al., 1993). High concentrations of circulating androgens (especially testosterone and 5-α-dihydrotestosterone) in adult females during the ﬁnal period of pregnancy are thought to be one source of testosterone circulating in the fetus (Matthews, 1939; Licht et al., 1992). If this is the case, then we suggest that the elevation of androgens during the ﬁnal period of pregnancy may have been selected to contribute to the production of neonates primed for sibling rivalry (East et al., 1993). It has been thought that high androgen levels in female fetuses and neonates were involved in the formation of the hypertrophied clitoris (Frank

CONFLICT AND COOPERATION IN A FEMALE-DOMINATED SOCIETY

19

et al., 1991). However, as antiandrogen treatment during fetal development has not produced females without a pseudo-penis, it is unlikely that androgen-dependent mechanisms are solely responsible, and androgenindependent mechanisms are likely to play an important role in the development of the enlarged clitoris (Drea et al., 1998).

VIII. THE FITNESS CONSEQUENCES OF VIRILIZATION When female spotted hyenas give birth, the hypertrophied clitoris ruptures on the caudal side, causing a large, bleeding wound that may be a source of infection. Many primiparous captive females produce stillbirths, whereas subsequent deliveries are easier (Glickman et al., 1992; Frank et al., 1995). In the Serengeti, the loss of a ﬁrst litter at parturition implies a reduction in terms of lifetime reproductive success of at least 5% (East et al., 1993). Thus, the maintenance of the hypertrophied clitoris entails a ﬁtness cost for females. If so, direct selection pressure must have operated to maintain this large (1% of body weight, Matthews, 1939) erectile organ (East et al., 1993). A possible selective advantage for females that have the opening to their reproductive tract (urogenital meatus) displaced to a position forward of their hind legs at the tip of their hypertrophied clitoris is that males have considerable difﬁculty locating and penetrating the clitoris opening (Kruuk, 1972). Although the hypertrophied clitoris is not erect during mating, its presence and position prevent males from achieving intromission without the complete cooperation of the female (East et al., 1993). We have proposed that this provides females with control over the mechanics of copulation and that male mating success might depend on close male–female relationships (East et al., 1993). With strong female control over copulation, the importance of sexual selection for male ﬁghting prowess is likely to be reduced. If males stand to gain little by ﬁghting with each other, because females do not select mating partners on the basis of their ﬁghting ability, then game theory predicts that ﬁghts between males should be rare and/or ﬁghting intensity low (Maynard Smith, 1982). This idea is consistent with the observed low levels of aggression among competing males (Frank, 1986b; East and Hofer, 1991b, 2001). The challenge hypothesis (Wingﬁeld et al., 1990) has successfully explained how different mating strategies may affect patterns of androgen secretion in several avian species, but the hypothesis has rarely been tested in mammalian species. The hypothesis predicts that if direct competition for females is low, as is the case among male spotted hyenas, then there should be no association between male social status and plasma androgen concentrations. Goymann (1999) found no association between plasma androgen levels and male social

20

MARION L. EAST AND HERIBERT HOFER

status among Serengeti postdispersal males. Similarly, in some stable primate groups, where challenges are rare among group males, a lack of correlation between androgens and male social status has also been reported (Sapolsky, 1993). The challenge hypothesis (Wingﬁeld et al., 1990) predicts a rise in male androgen levels during periods of mate guarding and courtship, and consistent with this prediction Goymann (1999) found that Serengeti males that excluded other males from approaching females had signiﬁcantly higher androgen levels than males that were not defending females. If a male spotted hyena is more likely to be selected as a mating partner if he has developed a relationship with a female, then males should actively foster relationships with females. The high investment by males in the courtship of speciﬁc females and the submission of males to females at food resources are consistent with the idea that males invest in fostering relationships with females (East and Hofer, 2001; see later). Thus, in contrast to the prevailing idea that females had to be virilized, large, and aggressive to dominate males, we suggest that the maintenance of the pseudo-penis imposed on males a need to cultivate relationships with females, and an important factor in this process was male submission (East and Hofer, 2001). The activation of these sex-reversed traits (decreased male aggression, female dominance) in adult spotted hyenas may be mediated by mechanisms other than elevated or decreased levels of peripheral androgens (Goymann, 1999).

IX. THE MALE-DOMINANCE HIERARCHY: QUEUING FOR SOCIAL STATUS We propose that the female pseudo-penis has not only decreased selection for male aggression, but has also inﬂuenced the manner in which immigrant males acquire social status (Table I). Social status among immigrant male hyenas is not linked to physical characteristics such as measures of skeletal size or body mass, and males do not ﬁght to acquire status (East and Hofer, 2001). Instead, the social status of immigrant males is correlated with duration of tenure, and thus males acquire social status by queuing (East and Hofer, 1991b, 2001). The queuing convention observed by immigrant males is relatively strict (cf. Kokko et al., 1998) as successively arriving males wait their turn to rise in social status and rarely use aggressive contests to increase in status (East and Hofer, 2001). Why do immigrant males observe queuing conventions and why do newcomers not attempt to jump the queue? We have shown that several factors are likely to contribute to the stability of the male social hierarchy (Table I). When males reach high social status they form coalitions with other queuing males more often than when they are mid- or low-ranking. High-ranking

CONFLICT AND COOPERATION IN A FEMALE-DOMINATED SOCIETY

21

males also form coalitions with females against lower ranking males, whereas mid- and low-ranking males do not form coalitions with females. Coalitions among high-ranking males and between high ranking males and the socially dominant females may contribute to the stability of the hierarchy among the males and deter queue jumping (East and Hofer, 2001). Furthermore, greeting ceremonies (see later) among immigrant males chieﬂy involve individuals of similar social status and may help conﬁrm their relative status to both participants (East et al., 1993).

X. MALE–FEMALE BEHAVIOR Immigrant males of high social status invest considerable time in following (shadowing) particular females irrespective of whether or not the females are in estrus (East and Hofer, 2001). Shadowing behavior appears to be a tactic employed by immigrant males to foster relationships with particular clan females. Immigrant males also attempt to exclude possible competitors from the vicinity of females. High-ranking males shadow and defend females more often than midand low-ranking males. High-ranking males also shadow a larger number of clan females and concentrate their shadowing effort mostly on high-ranking females (East and Hofer, 2001). High-ranking males probably prefer to associate with high-ranking females because they are more valuable mating partners, since cub growth rate, cub survival, and reproductive success of Serengeti females increase with social status (Hofer and East, 1993c, 1996; Golla et al., 1999). Relationships between immigrant males and clan females vary with male social status and tenure. Females are generally more tolerant of males with high social status and long tenure than males with shorter tenure and low social status (East and Hofer, 2001). Males with high social status and longer tenure display higher levels of friendly behavior toward females. Such females in turn are more tolerant of long-tenured males than more recent immigrant males (Table I). Thus there are status-related differences in the quality of relationships between males and females (East and Hofer, 2001). These ﬁndings are consistent with our idea that relationships between immigrant males and females are strengthened with increased male tenure (East and Hofer, 1991b; East et al., 1993). The reproductive success of a spotted hyena male may not be directly linked to his investment in building relationships with and defending females, but rather depend on female mate choice. The strength of male–female bonds may inﬂuence female mate choice, but other factors such as genetic quality may also be involved. In the Serengeti, where females commute and breed

22

MARION L. EAST AND HERIBERT HOFER

throughout the year, female mate choice is likely to be restricted to those males that are present in the clan territory during her period of estrus (East and Hofer, 2001).

XI. NATAL POSTDISPERSAL MALES It was thought that male spotted hyenas always disperse from their natal groups (Kruuk, 1972; Frank, 1986a,b; Mills, 1990). However, in the Serengeti sexually mature males do not always disperse. In contrast to immigrant males, natal males that do not disperse acquire top rank in the male social hierarchy soon after they begin to compete with immigrant males for access to females. Genetic analysis of paternity has demonstrated that these natal postdispersal males father cubs (M. L. East, H. Hofer, T. Burke, and K. Wilhelm, unpubl. data), so a lack of dispersal does not prevent successful reproduction. Holekamp and Smale (1998) have suggested that low levels of testosterone in natal males before they disperse may be due to persistent exposure to female kin, and that following dispersal and exposure to novel sexually receptive females in a new clan, male testosterone levels increase. However, the existence of reproductively active, postdispersal natal males in Serengeti clans suggests that testosterone levels in males may be more closely linked to male–male conﬂict over access to mates (Goymann, 1999). It is probable that natal postdispersal males gain immediate high rank in the immigrant male hierarchy due to the social dominance they acquired from their mother over all immigrant males during their development. Natal males are dominant to female clan members ranked below their mother, but when natal males start to compete with immigrant males for access to females they habitually submit to and are socially subordinate to the females they previously dominated (East and Hofer, 2001). This reversal of behavior is not due to increased aggressiveness from adult females, but rather because postdispersal males habitually submit when they begin to approach clan females as potential mates.

XII. CONFLICT RESOLUTION Much has been written about conﬂict and aggression in hyena society but afﬁliative behavior, cooperation, and conﬂict resolution among clan members has received much less attention. There is certainly conﬂict among members of spotted hyena clans over access to resources, but there is also much cooperation. Coalitions are important for the maintenance of social status in both male and female dominance hierarchies (East et al., 1993; Hofer

CONFLICT AND COOPERATION IN A FEMALE-DOMINATED SOCIETY

23

and East, 1996; East and Hofer, 2001). Clan members also cooperate in the acquisition and defense of carcasses against intra- and interspeciﬁc competition, defense of territory boundaries against neighboring groups, defense of young at the communal den against predators, group hunting and allosuckling (Kruuk, 1972; Mills, 1985; East and Hofer, 1991b; Knight et al., 1992; East et al., 1993; Hofer and East, 1993c). Thus clan members need to strike a balance between conﬂicts that arise in competitive situations and the cooperative pursuit of shared interests against clan members, competitors from other clans, and interspeciﬁc predators and competitors (Hofer and East, 2000). When members of a hyena clan meet, they often participate in a greeting ceremony in which two animals stand side by side, head to tail, with their hind legs lifted while they investigate each other’s anal-genital area (Wickler, 1965). We have shown that greetings are ritualized ceremonies in which socially subordinate animals actively signal submission to more dominant group members (East et al., 1993). The distribution of chosen greeting partners is consistent with the distribution predicted by optimality models of primate afﬁliative behavior that assume that individuals value social relationships as a resource (Table I, Seyfarth, 1977). Thus, hyena greeting ceremonies help to build relationships in a manner reminiscent of grooming in primates, and thus contribute to the stability of the social hierarchy (East et al., 1993). If female spotted hyenas value relationships with other females in their clan, in a similar manner to that seen in primates (Seyfarth, 1977), then a conﬂict that is behaviorally expressed as an aggressive encounter might damage that relationship (de Waal and van Roosmalen, 1979). If so, both opponents might be expected to actively repair the damage to the relationship by initiating neutral or friendly reunions after an aggressive encounter. We have shown that the behavior of females after an aggressive encounter differs from that before the encounter. Losers are more likely to initiate a friendly reunion than winners, and the probability of losers initiating a friendly reunion increases if after the initial conﬂict there is an escalated conﬂict with further aggression between opponents (Hofer and East, 2000). In a ﬁssion–fusion society, individuals have the option to reduce conﬂict by simply avoiding possible sources of conﬂict. For example, in the Serengeti, subordinate females undertake long-distance foraging trips more often than dominants (see earlier) and by doing so reduce possible conﬂict with dominant females over food resources within the territory. We predict that simple avoidance of conﬂict may be a more common tactic in ﬁssion–fusion societies than in more close-knit societies and that active reconciliation may be a more prominent aspect of close-knit societies (Hofer and East, 2000).

24

MARION L. EAST AND HERIBERT HOFER

XIII. CONCLUSION Previously it has been assumed that the evolution of female dominance, the masculinization of female genitalia, and siblicide in spotted hyenas was driven by selection for elevated levels of androgens that increased female aggressiveness and body size. We argue that current knowledge of the ecology, behavior, and endocrinology of spotted hyenas does not support this view. Adult female spotted hyenas do not have abnormally elevated levels of androgens, and female aggression should not be seen as the only force that shapes hyena social life. Cooperative and afﬁliative behaviors that foster the maintenance of relationships are equally important in the functioning of hyena society. To explain habitual submission by males to females we link female control over copulation with their enlarged clitoris or pseudo-penis. As males are unable to mate with uncooperative females they need to cultivate relationships with females. We suggest that habitual submission is an important part of this process. We propose that female control over mating has also reduced selection for male ﬁghting prowess, as relationships with females are a nontransferable resource. This, we suggest, has led to the observed low levels of physical male–male conﬂict, to a lack of a correlation between social status and plasma androgen levels in immigrant males, and to the evolution of a male social queue. There is little evidence to support the notion that elevated levels of androgens in both male and female neonates is a by-product of selection for androgenized, large, aggressive adult females. Data from free-ranging hyena populations also cast doubt on the idea that androgenized neonates commit “habitual” (obligate) siblicide in same-sexed twin litters. Sibling rivalry in spotted hyenas is linked to levels of maternal provisioning of milk, and facultative siblicide occurs when the milk supplied by mothers is insufﬁcient to support twin litters.

XIV. SUMMARY It has been proposed that the evolution of female dominance in spotted hyenas (Crocuta crocuta) was due to selection for increased levels of androgens leading to increased aggressiveness and body size as well as to the development of an enlarged penile clitoris. Siblicide in spotted hyena twin litters has also been viewed as a by-product of selection for androgenized, aggressive adult females. These ideas, however, do not ﬁt current knowledge of the behavior, ecology, and endocrinology of this species. We outline an alternative route to female dominance that links female control

CONFLICT AND COOPERATION IN A FEMALE-DOMINATED SOCIETY

25

over mating (due to their enlarged clitoris) to the need for males to establish relationships with females before successful copulation is possible. This idea provides a likely explanation for habitual male submission to females, for low levels of aggression among competing males, for a lack of association between male social status and plasma androgen concentration, and for the investment by males in establishing relationships with females. Relationships with females improve with male tenure and are a nontransferable resource, and these are factors that are likely to have contributed to the evolution of a queue for social status among immigrant males. Free-ranging adult female spotted hyenas have perfectly ordinary, and thus not abnormally elevated, levels of androgens, and dominance in spotted hyena society is not related to body size. Hence, the assumed link between androgens, aggression, and body size does not exist. Aggression is not the only means of resolving conﬂicts among females, and there is substantial evidence that social relationships are a resource valued by females. As females that receive antiandrogen treatment during fetal development still develop an enlarged clitoris, androgen-dependent mechanisms are unlikely to be solely responsible for the development of this structure. Elevated levels of androgens in neonates of both sexes do not result in the “habitual” (obligate) death of a cub in same-sexed twin litters, and they are unlikely to result in contrasting behavior in adults (increased aggression in females, decreased aggression in males). Elevated levels of androgens in neonates are more likely linked to sibling rivalry and facultative siblicide when the milk provided by mothers falls below the level required to rear two cubs.

Acknowledgments We are grateful to the many people who have assisted our research, particularly A. Francis, W. Golla, O. Honer, ¨ R. Klein, B. Knauer, G. Orio, T. Shabani, D. Thierer, C. and L. Trout, A. Turk, ¨ B. Wachter, W. Wickler, and K. Wilhelm. We are grateful to the Tanzanian Commission of Science and Technology for permission to conduct the study; to the Tanzanian Wildlife Research Institute and Tanzania National Parks for cooperation and support; and also to the Fritz-Thyssen-Stiftung, the Stifterverband der deutschen Wissenschaft, the Max-PlanckGesellschaft, and the Institute for Zoo and Wildlife Research Berlin for ﬁnancial assistance.

References Aureli, F., and de Waal, F. B. M. (2000). “Natural Conﬂict Resolution.” Univ. of California Press, Berkeley. Batty, K. A., Herbert, J., Keverne, E. B., and Velucci, S. V. (1986). Difference in blood levels of androgens in female Talapoin monkeys related to their social status. Neuroendocrinology 44, 347–354.

26

MARION L. EAST AND HERIBERT HOFER

Cashdan, E. (1995). Hormones, sex and status in women. Horm. Behav. 29, 354–366. Clutton-Brock, T. H. (1988). “Reproductive Success.” Univ. of Chicago Press, Chicago. Clutton-Brock, T. H. (1991). “The Evolution of Parental Care.” Princeton Univ. Press, Princeton, NJ. Concannon, P. W., and Castracane, V. D. (1985). Serum androstenedione and testosterone concentrations during pregnancy and nonpregnancy cycles in dogs. Biol. Reprod. 33, 1078–1083. de Waal, F. B. M., and van Roosmalen, A. (1979). Reconciliation and consolation among chimpanzees. Behav. Ecol. Sociobiol. 5, 55–66. Drea, C. M., Hawk, J. E., and Glickman, S. E. (1996). Aggression decreases as play emerges in infant spotted hyaenas: Preparation for joining the clan. Anim. Behav. 51, 1323–1336. Drea, C. M., Weldele, M. L., Forger, N. G., Coscia, E. M., Frank, L. G., Licht, P., and Glickman, S. E. (1998). Androgens and masculinization of genitalia in the spotted hyena (Crocuta crocuta). 2. Effects of prenatal anti-androgens. J. Reprod. Fert. 113, 117–127. East, M. L., and Hofer, H. (1991a). Loud calling in a female-dominated mammalian society: I. Structure and composition of whooping bouts, Crocuta crocuta. Anim. Behav. 42, 637– 649. East, M. L., and Hofer, H. (1991b). Loud calling in a female-dominated mammalian society: II. Behavioral contexts and functions of whooping of spotted hyaenas, Crocuta crocuta. Anim. Behav. 42, 651–669. East, M. L., and Hofer, H. (1997). The peniform clitoris of female spotted hyaenas. Trends Ecol. Evol. 12, 401–402. East, M. L., and Hofer, H. (2001). Male spotted hyaenas Crocuta crocuta queue for status in social groups dominated by females. Behav. Ecol. In press. East, M. L., Hofer, H., and Wickler, W. (1993). The erect “penis” as a ﬂag of submission in a female-dominated society: Greetings in Serengeti spotted hyenas. Behav. Ecol. Sociobiol. 33, 355–370. Evans, R. M. (1990). The relationship between parental input and investment. Anim. Behav. 39, 797–798. Feder, H. H. (1985). Peripheral plasma levels of gonadal steriods in adult male and in adult, nonpregnant female mammals. In “Handbook of Behavioral Neurobiology” Vol. 7 (N. Adler, D. Pfaff, and R. W. Goy, Eds.), pp. 299–370. Plenum Press, New York. Frank, L. G. (1986a). Social organization of the spotted hyaena (Crocuta crocuta). I. Demography. Anim. Behav. 35, 1500–1509. Frank, L. G. (1986b). Social organization of the spotted hyaena (Crocuta crocuta). II. Dominance and reproduction. Anim. Behav. 35, 1510–1527. Frank, L. G. (1996). Female masculinization in the spotted hyena: Endocrinology, behavioral ecology and evolution. In “Carnivore Behavior, Ecology, and Evolution” Vol. 2 (J. L. Gittleman, Ed.), pp. 78–131. Cornell Univ. Press, Ithaca, NY. Frank, L. G. (1997). Evolution of genital masculinization: Why do female hyaenas have such a large “penis”. Trends Ecol. Evol. 12, 58–62. Frank, L. G., Davidson, J. M., and Smith, E. R. (1985). Androgen levels in the spotted hyaena. J. Zool. London 206, 525–531. Frank, L. G., Glickman, S. E., and Powch, I. (1990). Sexual dimorphism in the spotted hyaenas (Crocuta crocuta). J. Zool. London 221, 308–313. Frank, L. G., Glickman, S. E., and Licht, P. (1991). Fatal sibling aggression, precocial development, and androgens in neonatal spotted hyaenas. Science 252, 702–704. Frank, L. G., Weldele, M. L., and Glickman, S. E. (1995). Masculinization costs in hyaena. Nature 377, 584–585.

CONFLICT AND COOPERATION IN A FEMALE-DOMINATED SOCIETY

27

Fusani, L., Van’t Hof, T., Hutchison, J. B., and Gahr, M. (2000). Seasonal expression of androgen receptors, estrogen receptors, and aromatase in the canary brain in relation to circulating androgens and estrogens. J. Neurobiol. 43, 254–268. Glickman, S. E., Frank, L. G., Davidson, J. M., Smith, E. R., and Siiteri, P. K. (1987). Androstenedione may organize or activate sex-reversed traits in female spotted hyenas. Proc. Natl. Acad. Sci. USA 84, 3444–3447. Glickman, S. E., Frank, L. G., Pavgi, S., and Licht, P. (1992). Hormonal correlates of “masculinization” in female spotted hyaenas (Crocuta crocuta). 1. Infancy to sexual maturity. J. Reprod. Fert. 95, 451–462 . Glickman, S. E., Frank, L. G., Holekamp, K. E., and Licht, P. (1993). Costs and beneﬁts of “androgenization” in the female spotted hyena: The natural selection of physiological mechanisms. In “Perspectives in Ethology” Vol. 10 (P. P. G. Bateson, Ed.), pp. 87–117. Plenum Press, New York. Golla, W., Hofer, H., and East, M. L. (1999). Within-litter sibling aggression in spotted hyaenas: Effect of maternal nursing, sex and age. Anim. Behav. 58, 715–726. Gombe, S. (1985). Short-term ﬂuctuations in progesterone, oestradiol and testosterone in pregnant and non-pregnant hyaenas (Crocuta crocuta). Afr. J. Ecol. 23, 269–271. Gould, S. J. (1981). Hyena myths and realities. Nat. Hist. 90, 16–24. Gowaty, P. A. (1997). “Feminism and evolutionary biology: Boundaries, intersections and frontiers.” Chapman & Hall, New York. Goymann, W. (1999). Hormone physiology and life history of spotted hyaenas, Crocuta crocuta. PhD dissertation, Ludwig-Maximilians-University Munich. Goymann, W., East, M. L., and Hofer, H. (2001). Androgens and the role of female “hyperaggressiveness” in spotted hyenas (Crocuta crocuta). Horm. Behav. 39, 83–92. Hamilton, W. J., Tilson, R. L., and Frank, L. G. (1986). Sexual monomorphism in spotted hyaenas, Crocuta crocuta. Ethology 71, 63–73. Harvey, P. H., Martin, R. D., and Clutton-Brock, T. H. (1987). Life histories in comparative perspectives. In “Primate Societies” (B. B. Smuts, D. L. Cheney, R. M. Seyfarth, R. W. Wrangham, and T. T. Struhsaker, Eds.), pp. 181–196. Univ. of Chicago Press, Chicago. Henschel, J. R., and Skinner, J. D. (1987). Social relationships and dispersal patterns in a clan of spotted hyaenas, Crocuta crocuta, in the Kruger National Park. S. Afr. J. Zool. 22, 18–24. Hofer, H. (1998). Spotted hyaena Crocuta crocuta (Erxleben, 1777). In “Hyaenas. Status Survey and Conservation Action Plan” (M. G. L. Mills and H. Hofer, Eds.), pp. 29–38. IUCN, Gland. Hofer, H., and East, M. L. (1993a). The commuting system of Serengeti spotted hyaenas: How a predator copes with migratory prey. I. Social organisation. Anim. Behav. 46, 547–557. Hofer, H., and East, M. L. (1993b). The commuting system of Serengeti spotted hyaenas: How a predator copes with migratory prey. II. Intrusion pressure and commuters’ space use. Anim. Behav. 46, 559–574. Hofer, H., and East, M. L. (1993c). The commuting system of Serengeti spotted hyaenas: How a predator copes with migratory prey. III. Attendance and maternal care. Anim. Behav. 46, 575–589. Hofer, H., and East, M. L. (1995a). Virilised sexual genitalia as adaptations of female spotted hyaenas. Rev. Suisse. Zool. 102, 895–906. Hofer, H., and East, M. L. (1995b). Population dynamics, population size, and the commuting system of Serengeti spotted hyaenas. In “Serengeti II” (A. R. E. Sinclair and P. Arcese, Eds.), pp. 332–363. Univ. of Chicago Press, Chicago. Hofer, H., and East, M. L. (1996). The components of parental care and their ﬁtness consequences: A life history perspective. Verh. Dtsch. Zool. Ges. 89.2, 149–164.

28

MARION L. EAST AND HERIBERT HOFER

Hofer, H., and East, M. L. (1997). Skewed offspring sex ratios and sex composition of twin litters in Serengeti spotted hyaenas (Crocuta crocuta) are a consequence of siblicide. Appl. Anim. Behav. Sci. 51, 307–316. Hofer, H., and East, M. L. (2000). Conﬂict management in female dominated spotted hyaenas. In “Natural Conﬂict Resolution” (F. Aureli and F. B. M. de Waal, Eds.), pp. 232–234. Univ. of California Press, Berkeley. Hofer, H., Campbell, K. L. I., and East, M. L. (1993). Snares, commuting hyaenas and migratory herbivores: Humans as predators in the Serengeti. Symp. Zool. Soc. London 65, 347–366. Hofer, H., East, M. L., Sammang, ¨ I., and Dehnhard, M. (2001). Analysis of volatile compounds in scent-marks of spotted hyenas (Crocuta crocuta) and their possible function in olfactory communication. In “Chemical Signals in Vertebrates” Vol. 9 (A. Marchlewska-Koj, J. J. Lepri, and D. Muller-Schwarze, ¨ Eds.), Plenum Press, New York. Holekamp, K. E., and Smale, L. (1995). Rapid change in offspring sex ratios after clan ﬁssion in the spotted hyena. Am. Nat. 145, 261–278. Holekamp, K., and Smale, L. (1998). Dispersal status inﬂuences hormones and behavior in the male spotted hyena. Horm. Behav. 33, 205–216. Holekamp, K. E., Ogutu, J. O., Dublin, H. A. T., Frank, L. G., and Smale, L. (1993). Fission of a spotted hyaena clan: Consequences of prolonged female absenteeism and causes of female emigration. Ethology 93, 285–299. Holekamp, K., Smale, L., and Szykman, M. (1996). Rank and reproduction in female spotted hyena. J. Reprod. Fert. 108, 229–237. Holekamp, K. E., Smale, L., Berg, R., and Cooper, S. M. (1997). Hunting rates and hunting success in the spotted hyena (Crocuta crocuta). J. Zool. London 242, 1–15. Hrdy, S. B. (1981). “The Woman That Never Evolved.” Harvard Univ. Press, Cambridge, MA. Hrdy, S. B. (1999). “Mother nature. Natural selection and the female of the species.” Random House, New York. James, W. H., and Hofer, H. (1999). A note on sex ratios and sex combinations in Serengeti spotted hyaenas: Siblicide and sub-binomial variance. Appl. Anim. Behav. Sci. 65, 153–158. Jolly, A. (1984). The puzzle of female feeding priority. In “Female Primates: Studies by Woman Primatologists” (M. Small, Ed.), pp. 197–215. Alan R. Liss, New York. Kappeler, P. M. (1990). Female dominance in Lemur catta: More than just female feeding priority? Folia. Primatol. 55, 92–103. Knight, M. H., van Jaarsveld, A., and Mills, M. G. L. (1992). Allo-suckling in spotted hyaenas (Crocuta crocuta): An example of behavioural ﬂexibility in carnivores. Afr. J. Ecol. 30, 245–251. Kokko, H., Lindstrom, ¨ J., Alatalo, R. V., and Rintamaki, ¨ P. T. (1998). Queuing for territory position in the lekking black grouse (Tetrao tetrix). Behav. Ecol. 9, 376–383. Kruuk, H. (1972). “The Spotted Hyena.” Univ. of Chicago Press, Chicago. Licht, P., Frank, L. G., Yalcinkaya, T. M., Siiteri, P. K., and Glickman, S. E. (1992). Hormonal correlates of “masculinization” in female spotted hyaenas (Crocuta crocuta). 2. Maternal and fetal steroids. J. Reprod. Fert. 95, 463–474. Lindeque M., and Skinner, J. D. (1982). Aseasonal breeding in spotted hyaenas in Southern Africa. Afr. J. Ecol. 20, 217–278. Lindeque, M., Skinner, J. D., and Millar, R. P. (1986). Adrenal and gonadal contributions to circulating androgens in spotted hyaenas as revealed by LHRH, hCG and ACTH stimulation. J. Reprod. Fert. 78, 211–217. Matthews, L. H. (1935). The oestrous cycle and intersexuality in the female mole Talpa europaea (Linn). Proc. Zool. Soc. London 106, 347–383. Matthews, L. H. (1939). Reproduction in the spotted hyaena, Crocuta crocuta (Erxleben). Philos. Trans. R. Soc. London Ser. B 230, 1–78.

CONFLICT AND COOPERATION IN A FEMALE-DOMINATED SOCIETY

29

Maynard Smith, J. (1982). “Evolution and the Theory of Games.” Cambridge Univ. Press, Cambridge, MA. McNamara, J. M., Houston, A. I., and Webb, J. N. (1994). Dynamic kin selection. Proc. R. Soc. London Ser. B 258, 23–28. Mills, M. G. L. (1985). Related spotted hyaenas forage together but do not cooperate in rearing young. Nature 316, 61–62. Mills, M. G. L. (1990). “Kalahari Hyaenas. Comparative Behavioural Ecology of Two Species.” Unwin Hyman, London. Mills, M. G. L., and Gorman, M. L. (1987). The scent-marking behaviour of the spotted hyaena Crocuta crocuta in the southern Kalahari. J. Zool. London 212, 482–497. Mock, D. W., and Parker, G. A. (1997). “The Evolution of Sibling Rivalry.” Oxford Univ. Press, London. Mock, D. W., Drummond, H., and Stinson, Ch. H. (1990). Avian siblicide. Am. Sci. 78, 438–449. Monaghan, E. P., and Glickman, S. E. (1992). Hormones and aggressive behavior. In “Behavioral Endocrinology” (B. J. Becker, S. M. Breedlove, and D. Crews, Eds.), pp. 261–285. MIT Press, Cambridge, MA. Nadler, R. D., Graham, C. E., Gosselin, R. E., and Collins, D. C. (1985). Serum levels of gonadotropins and gonadal steriods including testosterone during the menstrual cycle of the chimpanzee (Pan troglodytes). Am. J. Primatol. 9, 273–284. Neaves, W. B., Grifﬁn, J. E., and Wilson, J. D. (1980). Sexual dimorphism of the phallus in spotted hyaenas (Crocuta crocuta). J. Reprod. Fert. 59, 509–513. Packer, C., Collins, D. A., Sindimwo, A., and Goodall, J. (1995). Reproductive constraints on aggressive competition in female baboons. Nature 373, 60–63. Parker, G. A., Mock, D. W., and Lamey, T. C. (1989). How selﬁsh should stronger sibs be? Am. Nat. 133, 846–868. Ploog, D. W., and MacLean, P. D. (1963). Display of penile erection in squirrel monkeys (Saimiri sciureus). Anim. Behav. 11, 32–39. Racey, P. A., and Skinner, J. D. (1979). Endocrine aspects of sexual mimicry in spotted hyaenas, Crocuta crocuta. J. Zool. London 187, 315–326. Ralls, K. (1976). Mammals in which females are larger than males. Quart. Rev. Biol. 51, 245– 276. Rissman, E. F., Clendenon, A. L., and Krohmer, R. W. (1990). Role of androgens in the regulation of sexual behavior in the female musk shrew. Neuroendocrinology 51, 468–473. Roff, D. A. (1992). “The Evolution of Life Histories.” Chapman & Hall, London. Rowell, T. H. (1974). The concept of social dominance. Behav. Biol. 11, 131–154. Sapolsky, R. M. (1993). Endocrinology alfresco: Psychoendocrine studies of wild baboons. Rec. Prog. Horm. Res. 48, 437–468. Seyfarth, R. M. (1977). A model of social grooming among adult female monkeys. J. Theor. Biol. 65, 671–698. Sibly, R. M., Nott, H. M. R., and Fletcher, D. J. (1990). Splitting behaviour into bouts. Anim. Behav. 39, 63–69. Smale, L., Holekamp, K. E., Weldele, M., Frank, L. G., and Glickman, S. E. (1995). Competition and cooperation between litter-mates in the spotted hyaena, Crocuta crocuta. Anim. Behav. 50, 671–682. Smale, L., Holekamp, K. E., and White, P. A. (1999). Siblicide revisited in the spotted hyaenas: Does it conform to obligate or facultative models? Anim. Behav. 58, 545–551. Smuts, B. B. (1987). Gender, aggression, and inﬂuence. In “Primate Societies” (B. B. Smuts, D. L. Cheney, R. M. Seyfarth, R. W. Wrangham, and T. T. Struhsaker, Eds.), pp. 400–412. Univ. of Chicago Press, Chicago. Stearns, S. C. (1992). “The Evolution of Life Histories.” Oxford Univ. Press, London.

30

MARION L. EAST AND HERIBERT HOFER

Teltscher, C. (1996). “Die Evolution virilisierter Genitalien bei weiblichen Primaten. Eine vergleichende Analyse.” Diploma thesis, Ludwig-Maximilians-University Munich. Trivers, R. L. (1972). Parental investment and sexual selection. In “Sexual Selection and the Descent of Man” (B. Campbell, Ed.), pp. 136–179. Aldine, Chicago. van Jaarsveld, A. S., and Skinner, J. D. (1987). Spotted hyaena monomorphism: An adaptive “phallusy.” S. Afr. J. Sci. 83, 612–615. van Jaarsveld, A. S., and Skinner, J. D. (1991a). Plasma androgen concentrations in initial samples from spotted hyaenas immobilized with zoletile reﬂect hormone status estimated by GnRH challenge and immobilization stress response. S. Afr. J. Zool. 26, 1–5. van Jaarsveld, A. S., and Skinner, J. D. (1991b). Plasma androgens in spotted hyaenas (Crocuta crocuta): Inﬂuence of social and reproductive development. J. Reprod. Fert. 93, 195–201. Whitworth, D. J., Licht, P., Racey, P. A., and Glickman, S. E. (1999). Testis-like steroidogenesis in the ovotestis of the European mole, Talpa europaea. Biol. Reprod. 60, 413–418. Wickler, W. (1965). Die au ¨ ßeren Genitalien als soziale Signale bei einigen Primaten. Naturwiss. 52, 269–270. Wingﬁeld, J. C., and Farner, D. S. (1975). The determination of ﬁve steroids in avian plasma by radioimmunoassay and competitive protein-binding. Steroids 26, 311–327. Wingﬁeld, J. C., Hegner, R. E., Dufty, A. M., and Ball, G. F. (1990). The “challenge hypothesis”: Theoretical implications for patterns of testosterone secretion, mating systems and breeding strategies. Am. Nat. 136, 829–846. Young, A. L., Richard, A. F., and Aiello, L. C. (1990). Female dominance and maternal investment in strepsirrhine primates. Am. Nat. 135, 473–488.

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 31

Birdsong and Male–Male Competition: Causes and Consequences of Vocal Variability in the Collared Dove (Streptopelia decaocto) CAREL TEN CATE,1 HANS SLABBEKOORN,2 and MECHTELD R. BALLINTIJN1 1 BEHAVIOURAL BIOLOGY INSTITUTE OF EVOLUTIONARY AND ECOLOGICAL SCIENCES LEIDEN UNIVERSITY, 2300 RA LEIDEN, THE NETHERLANDS 2 CENTER FOR TROPICAL RESEARCH AND DEPARTMENT OF BIOLOGY SAN FRANCISCO STATE UNIVERSITY SAN FRANCISCO, CALIFORNIA 94132

I. INTRODUCTION During the breeding season, many birds produce species-speciﬁc, relatively complex and loud vocalizations, usually referred to as “songs.” These vocalizations are predominantly produced by males and serve to attract mates, to compete with conspeciﬁcs, and to defend territories by deterring rivals. The structure of these songs may vary widely among species, ranging from the “hoots” of owls and the “coos” of doves and pigeons to the elaborate songs of many songbirds. Signals produced in these contexts are subject to sexual selection and birdsong is no exception. In fact, many aspects of sexual selection have been tested using birdsong. The relative ease with which sound recordings can be analyzed allows for a detailed quantitative description of signal structure and its variation. Recordings can be edited and played back to both males and females to examine their responses. In our studies we also exploit the possibilities that the birdsong model system provides to examine aspects of sexual selection and, in particular, intrasexual selection. In this chapter we concentrate on the role of birdsong in one aspect of male–male competition: territorial defense. We will ﬁrst brieﬂy review this subject, arriving at the conclusion that the role of variation in vocal signals in territorial defense and male contest has received limited attention. We outline the questions to be addressed and next introduce our model species, the collared dove (Streptopelia decaocto) and its vocalizations. We describe 31

C 2002 by Academic Press Copyright All rights of reproduction in any form reserved. 0065-3454/02 $35.00

32

CAREL TEN CATE et al.

the vocal variation in the perch-coo, the prime signal in the territorial behavior of this species. Current knowledge about causes and consequences of perch-coo variation is presented. We end with a discussion of the importance of a better understanding of the linkage between vocal variability and sender quality.

II. SONG: MATE ATTRACTION AND MALE CONTEST A. MATE ATTRACTION As a conspicuous secondary sex trait, song is assumed to have evolved because it serves a role in competition over mates (Kroodsma and Byers, 1991; Andersson, 1994; Catchpole and Slater, 1995). This competition for mates occurs if females choose mates on the basis of their songs or if songs are involved in male–male interactions, giving some males a reproductive advantage over others. Song characteristics may thus evolve through both inter- and intrasexual selection, that is, song can be a trait of dual utility (Berglund et al., 1996). With respect to mate attraction, playback experiments have shown that song attracts females and that some songs or singing styles are more attractive than others (Catchpole, 1980; Catchpole et al., 1984; Eriksson and Wallin, 1986; Eens et al., 1991; Mountjoy and Lemon, 1991; Vallet and Kreutzer, 1995; for a review, see Searcy and Yasukawa, 1996). It is still a matter of debate what beneﬁts females may gain from making such choices and how female choice evolved. Several, not mutually exclusive, hypotheses have been put forward to explain the evolution of elaborate mate attraction signals in general (for reviews see Cronin, 1991; Andersson, 1994; Bradbury and Vehrencamp, 1998; Stearns and Hoekstra, 2000). In the “Fisherian” or “runaway” model (Fisher, 1930; Lande, 1981; Kirkpatrick, 1982) and in the “sensory exploitation” model (Ryan et al., 1990) the signal may, in theory, be arbitrary in the sense that it need not be related to speciﬁc qualities of the male. In contrast, some alternative theories hypothesize that signals bear a close relation to male quality. One such hypothesis is the “genetic indicator” hypothesis (Andersson, 1994). This states that conspicuous traits like bright colors or elaborate songs evolve because they reﬂect heritable viability differences, such as resistance against parasites (Hamilton and Zuk, 1982; cf. Read and Weary, 1990, 1992). Another hypothesis is that elaborate traits may reﬂect a male’s current phenotypic condition or other qualities (e.g., the willingness to invest in parental care that might be directly beneﬁcial to a female; Grafen, 1990a,b). Both the genetic indicator model and the current quality model are also known as “handicap” models (after Zahavi,

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

33

1975, 1977; Zahavi and Zahavi, 1997) because various theoretical studies have shown that such signals only evolve if production of the signal is costly (Grafen, 1990a,b). Birdsong has been used to examine and test the various previously mentioned hypotheses (for a review: see Searcy and Yasukawa, 1996). So far there is not much evidence that runaway selection has played a role in the evolution of varied songs or elaborate repertoires (Catchpole and Slater, 1995; Searcy and Yasukawa, 1996). Also the role of sensory exploitation in song evolution awaits further study (Searcy, 1992; Gray and Hagelin, 1996; Collins, 1999; Neubauer, 1999). Most studies on the function of song concentrate on the possibility that females may get direct or indirect (genetic) beneﬁts from mating with males with certain types of song or singing styles. In several species, for instance, song repertoires increase with age (e.g., Nottebohm and Nottebohm, 1978; Cosens and Sealy, 1986; Derrickson, 1987; Lampe and Espmark, 1994; Hasselquist et al., 1996; Mountjoy and Lemon, 1996; Eens, 1997). Therefore a female preference for larger repertoires may lead to getting more experienced, or perhaps healthier, mates. Females may also prefer vocal characteristics that might be more demanding to produce, indicating a male of good physical quality (e.g., Gibson and Bradbury, 1985; Vehrencamp et al., 1989; Vallet and Kreutzer, 1995; Vallet et al., 1998). Several studies have obtained correlational evidence for a relationship between male song features and reproductive success (Horn et al., 1993; Kempenaers et al., 1997; Galeotti, 1998; Hasselquist, 1998; Martin-Vivaldi et al., 1999; Gil and Slater, 2000), but in many of these cases it has not been possible to disentangle the contribution of direct and indirect effects. However, the evidence keeps accumulating that selection through female preferences plays a role in the evolution of birdsong (Searcy and Yasukawa, 1996). B. MALE CONTEST In the context of male contest, some processes through which female choice may drive signal evolution, such as selection for “good genes,” are not relevant, or less so. However, the handicap theory in particular provides a good framework to explain signal evolution in the context of male contest. Birdsong may be used in territorial defense (reviews by Kroodsma and Byers, 1991; Catchpole and Slater, 1995; Searcy and Nowicki, 2000). The ﬁrst experimental evidence showing this was provided by Peek (1972) who showed that muted red-winged blackbirds (Agelaius phoeniceus) suffered from more territorial intrusions than nonmuted controls. Later studies in this and other species conﬁrmed this ﬁnding (Smith, 1979; McDonald, 1989). Krebs (1977) took a different experimental approach. He captured all

34

CAREL TEN CATE et al.

territorial great tits (Parus major) from a woodland area and replaced them by speakers playing either song, noise (a whistle), or no sound. The area from which no song or noise was played was the ﬁrst one to be occupied by new great tits. These experiments and similar ones on other species (Falls, 1988; Nowicki et al., 1998b) indicate that song can have a role in territorial defense, and therefore, that it may be subject to intrasexual selection. They do, however, not show whether some song variants are more effective than others. But, rather than testing whether interindividual variation in song features correlated with aspects of male–male competition, many subsequent studies have focused on interspecific song variation and the function of song in species isolation (e.g., Nuechterlein, 1981; Becker, 1982; Martens, 1996; Miller, 1996; de Kort and ten Cate, 2001). Playback experiments tested whether altering species-speciﬁc features, such as temporal structure, frequency range, and syntax, affected the responses of territorial birds (e.g., Becker, 1982; Dabelsteen and Pedersen, 1985; Nelson, 1988). Weaker responses to a modiﬁed song were interpreted as being due to a loss of species speciﬁcity and the results were interpreted within a theoretical framework of species recognition, testing theories such as the “invariant features” (Nelson, 1989) or the “room for variation” (e.g., Dabelsteen and Pedersen, 1992) hypotheses. Responding to a species-speciﬁc territorial song and not to the songs of other species will usually be of advantage to males. Hence, it is likely that selection will favor the evolution of species speciﬁcity in songs, as it prevents the waste of time and energy or the risk of injuries involved in aggressive interactions with birds that are not competitors. However, it is realized more and more that the border between signal evolution under the inﬂuence of species recognition and through inter- and intrasexual selection is a gradual one. Various authors state that species isolation has been overemphasized as an explanation for the evolution of secondary sex traits (Ryan and Rand, 1993a; Andersson, 1994, p. 207). Rather, sexual selection may lead to rapid evolution in secondary sex signals in isolated populations, which next may result in speciation (West-Eberhard, 1983; Andersson, 1994; Pfennig, 1998). Hence, we are back to the question of the extent to which the evolution of male territorial signals can be driven by male contest. For a long time, this possibility received noticeably less attention than selection by female choice. Andersson (1994), for instance, in his review of empirical studies of sexually selected traits lists 65 putative cases of sexual selection in birds, of which 55 concern selection through female choice and only 10 through male contest. When limiting the overview to male song, there remain 31 cases, involving only 5 instances where a trait may have been selected through male competition. Berglund et al. (1996) reviewed the studies in which the function of signals for both intra- and intersexual communication has been

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

35

examined. They also state that studies on sexual selection are heavily biased to intersexual selection. However, they suggest that many of the signals that serve a dual function ﬁrst evolved in the context of male–male competition as an indicator of male competitive ability and only later became subject to female choice for high-quality males (see also Borgia and Coleman, 2000). Whether this suggestion is correct or not, it underscores the importance of studying the role of male–male competition on signal evolution. The survey of Berglund et al. (1996) involves, among others, studies of the vocal signals of six bird species. These six cases, and also the ﬁve mentioned by Andersson (1994), include the studies by Peek (1972) and Krebs (1977) on the effects of song absence on territorial defense. However, these shed little light on the question of whether the characteristics of vocal signals once they are there have been or still are molded by intrasexual selection. Bradbury and Vehrencamp (1998) also discuss territorial signaling among males. They also emphasize the role of these signals in informing other individuals about the sender’s species, individual identity, motivation, location, or distance; that is, on messages that are predominantly about species recognition and identifying territory owners. Such messages are important in territorial signaling. But does this imply there is no scope for communication about competitive abilities through territorial signals? Various factors may affect the outcome of a conﬂict or ﬁght between two opponents. One of them is motivation. In a ﬁght between a territory owner and an intruder of similar strength, the owner will usually win (Krebs, 1982; Enquist and Leimar, 1987; Englund and Otto, 1991). Ownership seems to increase the animal’s willingness to escalate a ﬁght, in line with gametheoretical predictions (Bradbury and Vehrencamp, 1998). However, strongly motivated intruders are also predicted to escalate (Grafen, 1987), and this may lead to long ﬁghts. The outcome of such ﬁghts will depend on physical strength, and superior intruders may beat owners (Smith and Arcese, 1989; Rosenberg and Enquist, 1991; Olsson, 1992). The factors that determine physical strength thus become important. Such factors are body size or weight, condition, or energy reserves (Austad, 1983; Englund and Olsson, 1990; Haley, 1994). Therefore, if a territory owner can produce a signal that reﬂects any of these qualities, and that is recognized as such by receivers, then he may suffer less from intrusions, and conﬂicts can be settled without the need to ﬁght. Apart from reducing the chance of injuries, this also means that more time and energy can be used for activities with a direct impact on reproduction. Just as in the context of female choice, such a signal is only likely to evolve if it cannot be cheated; that is, if the production of songs, certain types of vocal characteristics, or singing styles involves size- or energy-related constraints and have higher costs for low-quality males than for better quality ones. If this is the case, then such features may

36

CAREL TEN CATE et al.

evolve through male–male competition. The handicap model thus also applies to male–male contest. C. MALE SONG VARIATION AND MALE COMPETITIVE ABILITY: A BRIEF REVIEW As shown earlier, there are theoretical considerations making it likely that variation in territorial signals reﬂects competitive ability. But what is known about the relation between male competitive ability and the characteristics of vocalizations used as territorial signals? In addressing this question, we focus on the vocalizations themselves, and their variability, rather than on how they are used during vocal interactions. Todt and Naguib (2000) have reviewed this latter issue. They demonstrate that the relative timing of vocal signals given by two participants in a vocal exchange and the use of speciﬁc song patterns in reply to the opponent may have important communicative signiﬁcance. However, the relation between the use of various singing styles or types of vocal interactions and either male quality or male ﬁtness consequences is still largely unclear and remains to be explored (Todt and Naguib, 2000). For this reason we leave this issue aside and concentrate on the structure of the signals used in such interactions and on whether and how their acoustic features may be shaped by selection through male–male competition. To assess the hypothesis that selection through male competitive ability may shape the evolution of song features that are related to quality, the following conditions have to be met: (1) males should show interindividual variation in song features; (2) this variation has to correlate with some aspect of competitive ability (e.g., strength); (3) the signal has to be one for which some variants can only be produced by males with the highest competitive ability; and, ﬁnally, (4) receivers of the signal, in this case other males, should note the signal variation and differentiate their responses according to the perceived quality of the sender. So, what evidence is available that birdsong meets these requirements and which song features may contain meaningful variation? 1. Repertoires and Male Quality One song variable examined in the context of male–male interaction is song- and syllable-repertoire size. In both great tits (Krebs et al., 1978) and red-winged blackbirds (Yasukawa, 1981) playbacks of large repertoires were more effective in preventing intrusions into territories than those of small repertoires. Krebs (1977) suggested that the effect was due to the sender creating the impression of being several birds, rather than one. Whether this is really the cause of the observed effect is still unclear (Catchpole

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

37

and Slater, 1995), and later observations on red-winged blackbirds are not unequivocally consistent with the hypothesis (Yasukawa and Searcy, 1985). However, other studies have also indicated that larger repertoires may deter rival males. In starlings (Sturnus vulgaris), for instance, Mountjoy and Lemon (1991) showed that playback of “complex song” was more effective in preventing males from entering a nestbox than playback of “simple song,” whereas Eens (1997) obtained a positive correlation between repertoire size and proportion of encounters won in a group of captive starlings. So, as with mate attraction, larger repertoires may be more effective in male– male competition. However, whereas females may use a large repertoire to get an older male, with a better territory or of a higher genetic quality, it is not immediately clear whether such characteristics make these males better competitors in a male conﬂict. Between species, larger repertoire size correlates with size of the HVC (Higher Vocal Center—DeVoogd et al., 1993). There is also some evidence for a correlation between repertoire size and HVC within a species (e.g., Airey et al., 2000), and it has been suggested that only ﬁtter males can afford the costs associated with the increase in brain size (Catchpole, 1996). Such males may be generally stronger as well. But, although repertoire may thus reﬂect some aspect of quality (e.g., Brenowitz et al., 1995), this parameter need not necessarily bear a relation to the current competitiveness or ﬁghting ability of the singer. A more direct relationship with ﬁghting ability would be expected in vocal parameters that are constrained by body size, energy supplies, and physiological or respiratory limitations. 2. Song Frequency and Body Size Body size may act as a constraint on the production of low-frequency vocalizations. A larger body size will generally be correlated with a larger sound-producing organ and this will allow the production of lower frequency sounds in the same way that larger musical instruments produce lower frequency sounds than smaller versions of the same kind. As body size can be an important component in the outcome of ﬁghts, selection may have favored the evolution of low-frequency sounds, indicative of a larger body size, in male contest signals (Morton, 1977). In several groups of animals as diverse as insects, ﬁsh, anurans, and mammals, larger species or individuals do indeed produce lower frequency sounds (Davies and Halliday, 1978; Clutton-Brock and Albon, 1979; Arak, 1983; Ryan, 1985, 1988; Robertson, 1986a,b; Wagner, 1989; Hauser, 1993; Myrberg et al., 1993; Simmons, 1995; Brown et al., 1996). In line with this, it has been found that larger bird (sub)species generally produce lower pitched vocalizations and that within a species the larger sex is usually the one that produces lower frequencies (e.g., Ryan and Brenowitz, 1985; James and Robertson, 1989; Rosenﬁeld and Bielefeldt, 1991; Farquhar,

38

CAREL TEN CATE et al.

1993; Bretagnolle, 1996; Ballintijn and ten Cate, 1997a; Bried and Jouventin, 1997; Tubaro and Mahler, 1998; Whitehead, 1999). Also, it has been suggested that the tracheal elongation shown by several groups of birds has evolved because sexual selection for large body size gave an advantage to those birds that, by enlarging their vocal apparatus, could fake a larger body size than they actually possess ( Fitch, 1999). This phenomenon is very interesting as it suggests that birds may break through the constraints of body size on the production of low-frequency vocalizations. As a consequence, the reliability of low-frequency vocalizations as a predictor of size and competitive abilities is weakened. Thus, most critical to our discussion, is whether within a species larger or heavier individuals really produce lower frequency vocalizations than smaller or lighter same-sexed individuals. This has been found in some studies (Wurdinger, ¨ 1970; Handford and Lougheed, 1991; Beani and Dessi-Fulgheri, 1995; Appleby and Redpath, 1997; Galeotti et al., 1997). However, others have not found such a correlation (e.g., Bowman, 1979; Shy, 1983; Genevois and Bretagnolle, 1994). In roosters (Gallus gallus domesticus) the relation between competitive abilities and frequency even seems reversed as the most dominant males produce crows with higher frequencies (Leonard and Horn, 1995). Hence, although the expected positive correlation between frequency use and body size has been found in some species, it is not a universal and reliable correlation. This demonstrates that this correlation is not caused by simple allometry. The reason for this may be that birds generate their sounds in more complex ways than insects or anurans. A consequence of these ﬁndings is that even if a negative correlation between body size and vocal frequency is obtained it becomes necessary to assess how the relationship is brought about before it can be assessed whether frequency acts as an honest signal of body size. In addition, controlled experiments comparing the effects on other males of playbacks of high- and low-frequency vocalizations that are identical in other parameters are necessary to establish whether the frequency difference has a role in communication. Such experiments are still few and far between. 3. Condition and Song Rate It has been suggested that energetic or physiological constraints are reﬂected in parameters such as total song output and loudness, repetition rate, and duration of vocalizations (Bradbury and Vehrencamp, 1998). Song rate is a parameter that may be affected by condition or energetic constraints. In line with this, it has been found that condition (weight) correlates with vocalization rate in some species (Galeotti, 1998; Butchart et al., 1999). Correlations have also been obtained between measures of condition and other vocal parameters. Male blue petrels (Halobaena caerulea), for instance, show a positive correlation between body weight and how fast the call is delivered

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

39

(Genevois and Bretagnolle, 1994), whereas in hoopoes (Upupa epops) body condition shows a positive correlation with number of elements in their song (Martin-Vivaldi et al., 1998). However, in these examples the call variation has been examined in relation to female mate choice rather than to male– male competition. Several studies have reported that food supplementation may increase song output (Reid, 1987; Alatalo et al., 1990; Cuthill and Macdonald, 1990), demonstrating that body condition may affect this parameter. However, it is still not quite clear how expensive vocal production is in birds (Eberhardt, 1994; Chappell et al., 1995; Horn et al., 1995; Eberhardt, 1996; Gaunt et al., 1996). Also, song output may be affected by aspects of condition other than energy, such as parasite load (Møller, 1991). Hence, it is not clear how direct the linkage is between ﬁghting ability, energy, and song rate or other song characteristics. The correlation between weight and song features, as mentioned earlier, may, for instance, not be a consequence of an energetic constraint, but be due to greater strength of heavier individuals. In addition, song rate requires a receiver to invest time to sample this parameter. Therefore, although condition and energy reserves may be a factor affecting ﬁghting potential, their role in the evolution of assessment signals for competitive ability remains to be tested. 4. Neuromotor Mechanisms and Vocal Characteristics Fighting, like vocalizing, requires muscular and respiratory activity. Therefore, vocal parameters that reﬂect superior muscular or respiratory qualities provide good candidates to serve in assessment. Studies on the mechanisms underlying vocal production have identiﬁed potential constraints in neuromotor mechanisms, such as the coordination between the left and right sides of the syrinx or between respiratory and syringeal muscles, and constraints on vocal tract movements and neuromotor coordination (Allan and Suthers, 1994; Suthers et al., 1994; Podos, 1996, 1997; Suthers, 1997; Suthers and Goller, 1997). However, we still lack experimental evidence (e.g., from playbacks) that these constraints affect meaningful parameters of male territorial signals. Also, for some cases in which particular vocal features correlate with male quality (Appleby and Redpath, 1997; Galeotti et al., 1997; Rehsteiner et al., 1998), the mechanism of vocal production involved is not clear, making it difﬁcult to assess their link to some neuromotor constraint. Nevertheless, studies on vocal production suggest that quality might potentially reﬂect itself in parameters such as high repetition or trill rates, the occurrence of a large proportion of sound within a duty cycle, long-lasting sounds, and rapid vocal delivery. For instance, Lambrechts and Dhondt (1986) showed that winter dominance in great tits was correlated with a higher number of phrases per song.

40

CAREL TEN CATE et al.

Lambrechts has suggested that high repetition rate might lead to muscular exhaustion and hence may reﬂect superior quality of a sender (Lambrechts and Dhondt, 1987, 1988; Lambrechts, 1996, 1997; see also Hailman et al., 1987). There is some evidence of this phenomenon in frogs (Wells and Taigen, 1986). Lambrechts suggested that muscular exhaustion is also reﬂected in certain features of great tit song, such as an increase in intersong interval during the course of a bout or a decrease in time spent singing a song with increasing bout length (Lambrechts and Dhondt, 1988; Weary et al., 1991). This phenomenon of “drift” in prolonged singing is also known from other species (McGregor, 1991). However, it is controversial whether the phenomenon really reﬂects exhaustion (Weary et al., 1988; 1991). Furthermore, it has not been tested whether variation in drift affects receiver responses. 5. Conclusion To summarize, studies from various perspectives have provided plausible arguments and some correlational support for a linkage between competitive ability and speciﬁc vocal characteristics. Nevertheless, the direct evidence for this is very limited, as are experiments demonstrating that the identiﬁed vocal characteristics affect the responses of other males. The aim of our studies on coo-vocalizations of the collared dove is to examine the link between interindividual vocal variability, variability in sender quality, and receiver responses in a case study on one speciﬁc model species.

III. THE COLLARED DOVE AS A MODEL SYSTEM We speciﬁcally chose a nonsongbird to avoid the complications arising from the sometimes extensive intraspeciﬁc song variability in songbirds and from the learning processes underlying their song structure. The vocalizations of Streptopelia doves (turtledoves) provide an excellent model system to examine the relationship between vocal variation and sexual selection. Studying these vocalizations has a long history. Darwin (1871, p. 60) suggested that the vocalizations of the European turtledove (S. turtur) had a role in mate attraction. Craig (1908) identiﬁed various functions of cooing, in particular for the ringdove (S. risoria), and also noticed the existence of individual differences in vocalizations. Lehrman (e.g., 1965) and others (e.g., Cheng, 1979; 1986; ten Cate and Hilbers, 1991; ten Cate et al., 1993; Cheng et al., 1998) examined in detail the role of cooing in the reproductive behavior of the ringdove. Other studies showed that the species-speciﬁc coo-pattern of ringdoves emerged in deafened birds (Nottebohm and Nottebohm, 1971) and that cross-fostering doves of various Streptopelia species to related

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

41

species did not affect their conspeciﬁc coo-patterns whereas hybridization did (Lade and Thorpe, 1964; Davies, 1970; Baptista, 1996). These latter ﬁndings suggest that genetic differences underly the vocal differences between, and maybe also within, various species. It is this background of knowledge that made us select a Streptopelia species that we could study in the laboratory and in the ﬁeld: the collared dove (Streptopelia decaocto). A. THE COLLARED DOVE AND ITS TERRITORIAL BEHAVIOR The collared dove (Streptopelia decaocto) is one of the 17 species in the genus (Johnson et al. 2001). It is a sister species to the African collared dove (S. roseogrisea), the ancestor of the domesticated ringdove (S. risoria). The species originates from the Indian subcontinent. In historical times, however, it had invaded large parts of Asia and Europe (Nowak, 1965), and currently it is rapidly expanding in the United States (Romagosa and McEneaney, 1999). In the Netherlands it was ﬁrst observed in 1947 and it now is a very common inhabitant of suburban areas (SOVON, 1987). Our study population is located in the village of Oegstgeest (52◦ 11 N, 4◦ 28 E). Blocks of terrace houses with moderately sized front and back gardens characterize the habitat. These gardens, with their mixture of deciduous and coniferous trees, provide shelter, nesting sites, and food. In our study area breeding may start as early as February and continue until October (Slabbekoorn, 1998). During this period, a pair can raise several broods successively. Male collared doves defend a territory for most of the year, and the same territory may be occupied for several years (own observations). In winter, when attendance at the territory may be absent or very limited, most birds stay in the area, gathering at communal feeding sites and roosts. These feeding sites, often locations where humans provide food, and roosts are used all year round by nonbreeding birds. Even territorial males may regularly leave their territories to visit these sites during the breeding season. Some observations suggest that pair formation may occur at these locations (Hofstetter, 1954), although males may also attract females at their territories. Pair-bonds may last for several years (Hofstetter, 1954), but males and females do not stay in close vicinity to each other all the time. Even during the breeding season females may temporarily leave the territory. The territory is predominantly a breeding territory. As part of the feeding and roosting may occur elsewhere, territory size can be limited. In our study area the number of territories is about 100 per 30 ha, ranging in size from 100–3000 m2 (Slabbekoorn, 1998). Especially when densities are high, territorial conﬂicts can be observed regularly. If other birds ﬂy over or approach the territory, display ﬂights are given. These are upward ﬂights, accompanied with wing beats, followed by a

42

CAREL TEN CATE et al.

downward glide with wings and tail fully spread. These glides may be circular or may lead to escorting an intruder off the territory. Display ﬂights are usually accompanied by calls (see later). Intruders may also be approached directly, usually resulting in a chase. If intruders do not ﬂy off, this may lead to a ﬁght. Obviously, territorial birds may beneﬁt from preventing ﬁghts and using displays to advertise territoriality and ﬁghting ability. Display ﬂights fall in this category, but the most prominent role is played by coo-vocalizations. B. VOCALIZATIONS Four types of vocalizations are linked to territorial behavior and/or mate attraction in the collared dove: three coo-types and one call. The call is mainly produced in ﬂight. It is frequently given during display ﬂights, but also during chases and other ﬂights inside the territory. It has been mentioned as the “excitement call” in the literature (Goodwin, 1970; Cramp, 1985), but we prefer the more neutral term “ﬂight call” (other calls are produced in different contexts, but when we use the term “call” in this chapter, we refer to this call). It sounds like a nasal “weh” and may be given in short series of up to about ﬁve calls. Although some authors have described different variants of this call for different contexts (Bodenstein, 1949; Hofstetter, 1954; Nowak, 1965), we have not been able to detect clear context-speciﬁc variation in it (C. ten Cate and S. Rantala, unpublished data). As with many other doves and pigeons (Goodwin, 1970), the collared dove has three coo-types: the perch-coo, the bow-coo, and the nest-coo. These coo-types can be distinguished on the basis of accompanying behavior, by the contexts in which they are shown, and also on acoustic parameters. In some species of the genus Streptopelia the three coo-types differ in number of elements and in acoustic structure (e.g., tonal vs trill sound), in others, such as the collared dove, they share the same overall structure, differing in a quantitative way only. In the collared dove, all three coo-types have a three-syllabic, more or less tonal structure with a similar temporal pattern [Fig. 1 provides two examples of perch-coos; see ten Cate (1992) and Ballintijn and ten Cate (1999a) for additional sonograms and sonagrams of bow- and nest-coo]. The perch-coo is by far the most common vocalization (Fig. 1). It is a typical, long-distance advertising display that is predominantly shown by males. They show it from singing posts at conspicuous positions inside the territory, such as trees, chimneys, rooftops, and streetlights. During perchcooing, the head is kept in front of the body, with the throat inﬂated and the bill pointing downward. Perch-coos are usually produced in series (bouts) of repeated coos (Fig. 1). Although perch-coo activity may ﬂuctuate during the breeding cycle, it can be heard every day for most of the year. Cooing usually peaks in the morning, and there may be an additional smaller peak in the afternoon (Slabbekoorn, 1998).

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

43

FIG. 1. Sonagrams of perch-coo bouts from two individuals; 1, 2, and 3 indicate the three elements of a coo. The ﬁrst sonagram consists of perch-coos with nonmodulated elements. This bout ends with a coo lacking the third element. The second sonagram is of a coo with modulated elements—the elements show a sudden upward shift in frequency at the start of the element and a decline toward the end.

The bow-coo is almost exclusively shown by males. It is directed toward another individual at close quarters and accompanied by bowing movements of the head and hopping and trampling movements of the feet. The behavior may be directed at the partner or at an intruder that has landed in the territory. Studies on the ringdove have shown that it is an aggressively motivated behavior, although it plays a role in the initial phase of pair formation (Davies, 1974; Lovari and Hutchison, 1975). Bow-coos are also given in bouts of repeated coos, similar to perch-coos. The nest-coo of the collared dove is linked to activities and locations related to nesting, such as inspecting potential nest sites, nest building, and nest relief during the breeding phase. It is shown by both members of a pair, but predominantly by males. During nest-cooing the body is tilted forward and the wings may perform a scissoring movement. In the ringdove (S. risoria), male nest-cooing stimulates female nest-cooing (Cheng, 1979), and the latter has a self-stimulating effect on ovulation (Cheng, 1986; Cheng et al., 1998). It thus peaks just before egg laying (ten Cate and Hilbers, 1991), and ﬁeld observations indicate a similar pattern in the collared dove. Nest-coos are given in bouts that may last for over 50 coos (Ballintijn and ten Cate, 1999a). Within such a bout the interval between coos is variable but almost always longer than the duration of the coos themselves. Apart from the way a bout is structured, a detailed analysis of the three coo-types of wild caught and ﬁrst generation offspring of collared doves kept in our laboratory revealed several other acoustic differences between them. These were in the duration and amplitude of the ﬁrst element (shortest and softest in the nest-coo, longest in the bow-coo, and loudest in the perch-coo) and in the duration of the second element (longest in the nest-coo, shortest

44

CAREL TEN CATE et al.

in the perch-coo) (Ballintijn and ten Cate, 1999a). A ﬁrst study indicated differences in frequency between the coo-types (ten Cate, 1992), but this could not be conﬁrmed in a later study (Ballintijn and ten Cate, 1999a). Another difference is that elements of the bow- and perch-coo may show so-called “modulations” (Fig. 1); sudden upward shifts in frequency at the start of elements, followed by a decrease toward the end. The proportion of such modulated elements is very small in the nest-coo and signiﬁcantly larger in bow- and perch-coos (ten Cate, 1992; Ballintijn and ten Cate, 1999a). Also, this proportion decreases signiﬁcantly from the ﬁrst to the third element in the bow-coo, but not in the perch-coo (ten Cate, 1992). Finally, perch- and bow-coo series may contain coos lacking the third element (Ballintijn and ten Cate, 1999b) (see also the top sonagram of Fig. 1). This has not been observed for nest-coos (Ballintijn and ten Cate, 1999a). C. THE PERCH-COO AND SEXUAL SELECTION The perch-coo has all the characteristics of a long-distance territorial signal. Although pair formation may have occurred outside the territory (Hofstetter, 1954), one function of the perch-coo may be mate attraction, for instance to get extrapair copulations, a phenomenon that has been observed in collared doves (Hofstetter, 1954; H. Slabbekoorn and C. ten Cate, personal observations). A role for intersexual selection on the evolution of this signal thus cannot be excluded. However, an important, if not the main, function of perch-coos is communication between males. For instance, perch-cooing continues after mating, whereas vocalizations serving primarily for mate attraction usually decrease substantially during this period (e.g. Catchpole, 1973; Møller, 1991). Males also respond to each others cooing by cooing themselves, and playback of coos signiﬁcantly reduces the interval between successive bouts and increases the amount of perch-cooing of territorial males (Gurtler, ¨ 1977; Slabbekoorn and ten Cate, 1996). Cooing also occurs in association with agonistic interactions or in response to such interactions in neighboring territories (authors’ personal observations). This signal is thus a very suitable one with which to investigate interindividual variation and its meaning in the context of male–male competition.

IV. PERCH-COO VARIABILITY AMONG INDIVIDUALS The starting point of our research was the search for interindividual acoustic variation that might indicate some aspect of male quality and hence serve as an assessment signal. One parameter that differs between individuals is bout length. A bout is deﬁned as a series of coos in which the pause between

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

45

coos is shorter than the duration of the coo. In our study population, the average bout length per individual varied from 4 to 11 coos, with a population mean of 7 coos (N = 36, H. Slabbekoorn and C. Goed, unpublished data). Similar values were obtained in the laboratory (ten Cate, 1992; Ballintijn and ten Cate, 1999a; ten Cate and Ballintijn, 1996), where we obtained a mean of 8 coos per bout, with signiﬁcantly different bout lengths between individuals (ten Cate, 1992; Ballintijn and ten Cate, 1999a). Another difference between males is the fundamental frequency of both nonmodulated and modulated elements; this also differs signiﬁcantly between individuals (ten Cate, 1992). However, although these differences may be meaningful, our attention was drawn to two other conspicuous parameters for which the coos of individuals can differ. Both are related to the structure of the coo: the proportion of modulated elements and the proportion of coos lacking the third element. A. FREQUENCY MODULATIONS The ﬁrst researcher who described the two types of elements in the perchcoos of the collared dove was Gurtler ¨ (1973, 1977). However, he did not examine inter- and intraindividual variation in any detail. In our study population, the proportion of all elements that were modulated varied between 0 and 100% per individual (N = 36, H. Slabbekoorn and C. Goed, unpublished data). Wild-caught males kept in the laboratory differed signiﬁcantly in the proportion of modulated elements (ten Cate, 1992) and the same is true for laboratory-raised males (Ballintijn and ten Cate, 1997b). Again the number of modulated elements varied between 0 and 100% per individual. Although the proportion of modulated elements produced by a male in the middle of this distribution range may ﬂuctuate somewhat between different bouts (possibly depending on the context), the average intraindividual variation is usually smaller than the interindividual variation, even when intraindividual variation between successive years is considered (ten Cate, 1992). B. NUMBER OF ELEMENTS Various authors have described from ﬁeld observations the variation in the number of elements, ranging from 2–4, in perch-coos (Bodenstein, 1949; Dyrcz, 1956; Bettman, 1959; Frieling, 1960; Nowak, 1965). In our study population we also observed such variation. Coos with four elements were rare. A study of one bird showing this variant in our laboratory suggests that in this case three-element coos had an additional ﬁrst element (Ballintijn and ten Cate, 1999b). A possible interpretation of this variant is that it is some

46

CAREL TEN CATE et al.

sort of “stutter” in the motor program underlying the coo, a phenomenon also observed when a cooing dove is interrupted by a distracting light ﬂash (ten Cate and Ballintijn, 1996). In our study population most birds showed two-element coos on an occasional or regular basis. However, there was a clear and consistent interindividual variation in the proportion of coos consisting of two elements only. Of 26 birds recorded in our laboratory, only 4 birds never showed two-element coos, whereas in other birds the proportion reached 50% (Ballintijn and ten Cate, 1999b—see also Figs. 1 and 7). For both the proportion of modulated elements and the proportion of two-element coos, the variation is noticeable to the human ear. Although this may not mean that it is also noticeable to the doves, it suggests that it is. In our studies we addressed the causes and consequences of this variation in the structure of the vocalizations. We will ﬁrst address the variation in the proportion of modulated elements (Section V) and then the variation in the number of elements (Section VI).

V. MODULATED ELEMENTS A. SENDER CHARACTERISTICS The interindividual variation in the proportion of modulated elements raises the question of their message: does this variation relate to any particular feature of the sender? A ﬁrst indication of this was obtained when we examined the development of perch-cooing and sex differences in perchcoo parameters in laboratory-reared, individually housed collared doves (Ballintijn and ten Cate, 1997a,b). Young males and females start to coo at around the age of 12–16 weeks. At that time coo-bouts were still short, and coo structure was variable and showed several harmonics. Apart from the frequency level, which was higher in females, there was no clear sex difference in vocal characteristics. Neither males nor females produced any modulated elements at this stage. For females vocal structure changed only slightly when they became sexually mature. Males, on the other hand, changed in vocal structure at around 44 weeks of age. At that point several changes occurred more or less simultaneously, leading to signiﬁcant differences between male and female coo characteristics. Vocal activity, bout length, stereotypy, and proportion of sound per coo all increased, whereas the number of harmonics decreased (Ballintijn and ten Cate, 1997b). These changes occurred to a greater or lesser degree in all males. At the same time some males started to produce modulated elements. However, other males continued to produce no modulations at all, and this difference remained present when adult

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

47

FIG. 2. (A) Mean ( ± SE) proportion of modulated elements for males (ﬁlled circles, solid line; N = 9) is higher than for females (open circles, dashed line; N = 5) during the ﬁrst 1.5 years of age. (B) Individual differences in the proportion of modulated elements for all nine males. Each line is one individual. After Ballintijn and ten Cate (1997b).

(Fig. 2). At the same age at which the sexual differentiation in vocal structure became apparent, the ﬁrst instances of nest- and bow-cooing in males started to occur. The time of occurrence of the ﬁrst bow-coos correlated strongly with the time of occurrence of the ﬁrst modulated elements in perch-coos (r = 0.96, N = 8, p = .001; Ballintijn and ten Cate, 1997b). The absence of modulated elements in young doves and in adult females, and their emergence at the same time as other sexually differentiated acoustic traits means that modulated elements can be considered a signal of sexual maturity in males. But not all males showed modulations, even if all other vocal changes suggested that they had reached sexual maturity. So, do modulations carry another, more subtle message, for instance, one about male quality? Weight in birds can be an indicator of male quality, with heavier males generally being bigger and stronger. For this reason we next examined whether weight (of both males and females) correlated with any of the vocal parameters measured. We obtained a signiﬁcant positive correlation between male weight and the frequency (pitch) of unmodulated elements (r = 0.679, N = 9, p = 0.025) and a near signiﬁcant one (r = 0.563, N = 9, p = 0.067) between male weight and the proportion of modulated elements (Ballintijn and ten Cate, 1997b). Both this study and an earlier one (ten Cate, 1992) also showed a positive correlation between these two acoustic parameters. When we extended our sample size, the correlation between modulated elements

48

CAREL TEN CATE et al.

FIG. 3. Male body weight correlates positively with the proportion of modulated elements in perch-coos (r = 0.442, p = 0.024, N = 26).

and weight became signiﬁcant (Fig. 3). So, the proportion of modulated elements is an indicator of body weight in males. As weight of animals in general correlates with strength or ﬁghting ability (see Section II.B), these data suggest a link between the presence of modulations and ﬁghting abilities. As we demonstrate later, circumstantial evidence makes this a strong working hypothesis. The observed relation between weight and the proportion of modulated elements raises two questions: (1) How come heavier males produce more modulations? (2) Is the variation in the proportion of modulated elements detected and responded to by receivers? The ﬁrst question is important as it may provide insight into whether modulations can only be produced by certain physically superior individuals (i.e., they are handicaps) or whether all males might produce them but some do not for strategic reasons. Such a further examination may also reveal more about which precise qualities are reﬂected in this vocal parameter. We shall return to this issue in Section V.D, but will ﬁrst address the receiver side. B. RECEIVER RESPONSES TO MODULATED ELEMENTS Various techniques have been used to assess whether a vocal signal functions in territorial competition. A frequently used paradigm is to play back

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

49

song in a territory, mimicking an intruder, and to observe the behavior of the owner. The assumption is that owners will show stronger responses (closer approach, more aggressive behavior, more vocal display) to more threatening signals. Gurtler ¨ (1973) used this method to examine the responses of territorial collared doves to various coos and artiﬁcial coo-variants. Although his pioneering experiments suggest differential responses to different sounds, his data allowed no clear conclusions about the meaning of modulations. Hence, we decided to do an extensive series of more-controlled playbacks. We used a standardized design in which we examined the responses of territorial birds to various sets of two stimuli that differed in a systematic way. The two stimuli were played 2 min after each other, thus creating a paired design. Each stimulus usually consisted of a series of four bouts made up of eight identical coos. A loudspeaker was placed in the territory and the behavior of the owner was recorded from 1 min before the ﬁrst playback to 2 min after the second playback. Playback sets were controlled for sequence effects and avoided pseudo-replication (see Slabbekoorn and ten Cate, 1996, 1997, for details). As response measures we used the time in ﬂight, the number of calls, and the number of perch-coos. In our ﬁrst experiment (Slabbekoorn and ten Cate, 1996), we examined whether and how doves responded to nonmanipulated coos and coovariants. When the response to dove coos was compared with that to playback of songs of the winter wren (Troglodytes troglodytes—a common bird in our study area), the dove sound led to more time in ﬂight and a higher number of calls during playback and to more coos after playback. These results validated our method. We subsequently used this design to test the responses to modulated elements. If modulations reﬂect the competitive abilities of senders we would expect territorial males to respond more strongly to modulated coos than to nonmodulated ones. This was indeed the outcome when the responses to playback of modulated coos were compared with those to nonmodulated coos of the same individuals (Slabbekoorn and ten Cate, 1996). However, modulated coos may not only differ from nonmodulated ones by the presence of the modulation, but also in temporal parameters or in amplitude. So, further experiments were needed to control for such factors. To this end, we compared the responses to coos without modulation with responses to the same coos in which an artiﬁcial modulation was made in either the ﬁrst or in all three elements of the coo (Slabbekoorn and ten Cate, 1997). All other parameters of the nonmodulated coo (temporal structure, amplitude) remained intact in the modulated version. Playback of both manipulated coo-variants (i.e., those with one and those with three modulated elements per coo) led to stronger responses than the nonmodulated coos

50

CAREL TEN CATE et al.

FIG. 4. Responses to paired, sequential playbacks of different combinations of stimuli with and without modulated elements. (Left) Schematic representation of the various stimuli— NM = nonmodulated; M3 = all three elements modulated; M1 = ﬁrst element modulated; INF = nonmodulated coo with frequency increased. (Right) Percentage of time spent in ﬂight during playback of the stimuli. ∗ Indicates signiﬁcantly stronger responses to playback of modulated coos ( p < 0.05). See Slabbekoorn and ten Cate (1997) for full ﬁndings.

(Fig. 4). Playback of nonmodulated coos, however, led to a signiﬁcantly increased response compared to the behavior before playback; that is, nonmodulated coos are also a meaningful signal. Whether all three elements or only the ﬁrst one showed modulations did not matter in terms of response strength (Fig. 4). In later experiments (Slabbekoorn and ten Cate, 1998b) we showed that coos with modulations in the second element, or with one modulation per coo in an arbitrarily chosen element differing for each subsequent coo, also gave rise to stronger responses than did coos with nonmodulated elements. Although this suggests that the presence of modulations induced the increased responses, there is a caveat. Presence of a modulation also increases the average frequency level of that element. As mentioned in the previous section, heavier birds use higher frequencies in their (nonmodulated) coos. So, was it the higher frequency that produced the stronger response, or was it the modulation as such? This was tested by comparing the response to nonmodulated coos raised in frequency to match that of modulated coos (Slabbekoorn and ten Cate, 1997). An increase in frequency did not lead to a signiﬁcantly stronger response compared to nonmodulated coos, whereas they induced a signiﬁcantly lower response than modulated coos did (Fig. 4). Thus, it is the presence of modulations as such and not the use of a high frequency that induced stronger responses to coos with modulated elements. So, this set of experiments showed that presence of modulated elements has a clear and positive effect on the response strength of territorial collared

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

51

doves. The effect is in line with the hypothesis that modulations may signal some aspect of better quality or ﬁghting ability. C. SUPERMODULATIONS? Some experiments on intersexual selection have shown that signal versions beyond, or at the extreme of, the natural signal range are more effective than the natural signal in attracting mates (e.g., the tail of widowbirds, Euplectes jacksoni, Andersson, 1982; the calls of Physalaemus frogs, Ryan and Rand, 1993a,b). This phenomenon may drive signal evolution to more extreme versions. Although an established phenomenon in the context of intersexual selection (Andersson, 1994), it is not clear whether the same may occur in the competitive context. Therefore we did some experiments to examine whether it was possible to create modulated elements that lead to a stronger response than naturally occurring modulated elements. To create “supermodulated” coos, we cannot increase the number of modulated elements beyond three without altering coo structure itself. However, what can be examined is whether the magnitude or frequency level of the modulations affects the response strength. In the experiments described earlier, an artiﬁcial modulation was made by a 3% decrease in frequency of the ﬁrst part of a nonmodulated element (the so-called “base frequency”) followed by an 11% increase in frequency for the remainder of the element (the “modulated frequency”). This leads to a modulation with a magnitude of 14%; that is, the rise in frequency at the start of the modulation is 14% of the frequency of the nonmodulated element. So, when the nonmodulated element had a frequency level of 550 Hz, the magnitude of the modulation was 0.14 × 550 = 77 Hz. The 14% was used because it was well within the natural range of modulation magnitudes. For our experiments (Slabbekoorn and ten Cate, 1998a), we varied two parameters of the modulations independently of each other. The ﬁrst parameter was the magnitude of the modulation, the second was the frequency level of the base frequency. In a ﬁrst series of experiments, we examined the responses to coos with modulations of 3, 28, and 42% by testing each against nonmodulated coos. Of these values, 3 and 42% are outside the natural range, for which the average is around 20% (Slabbekoorn and ten Cate, 1998a). In a second series, modulations of 3, 21, 28, and 42% were tested against the previously used modulation of 14%. Finally, we tested whether it mattered if the base frequency was within or outside a range of 3 SD (standard deviation) around the population average. It turned out (Slabbekoorn and ten Cate, 1998a) that receivers were most responsive to modulated coos with a modulation of 14 or 21% (i.e.,

52

CAREL TEN CATE et al.

similar to that of the average in the population) and a base frequency within 2 SD of the population average. Responses to other variants were generally weaker, although still of the same magnitude as responses to nonmodulated elements. Only the variants with a base frequency of more than 3 SD above the population average (>650 Hz) produced no meaningful responses. So, altering the magnitude or the base of modulated elements does not create a supernormal signal. This means that intrasexual selection, as mediated by receiver responses, will have a stabilizing effect on modulation characteristics. This raises a new question about the sender, which is whether birds showing modulations within the average species range are the strongest competitors, and consequently, whether those below or above this average are the weaker ones. At this moment we cannot answer this question. However, given the persistently stronger responses to coos with modulated elements and the ﬁnding that the proportion of modulations is correlated with weight, we return to the question of why all males do not produce coos with modulated elements.

D. THE SENDER REVISITED: PRODUCTION OF MODULATED ELEMENTS Modulated elements seem to carry a message about the sender. However, what precisely is this message? Why is weight reﬂected in the proportion of modulated elements? Does their production require more effort than the production of nonmodulated elements? To address these questions we turned to the mechanisms of vocal production. Here, however, we face the problem that, in spite of all the research into birdsong, there still is no complete understanding about how birds sing and how different sound characteristics are produced (for reviews, see Gaunt and Nowicki, 1998; Suthers, 1999). The reason for this lack of understanding is the difﬁculty of measuring the processes occurring within the syrinx (the sound production organ) in an actively singing bird. Thus, most of the traditional hypotheses concerning sound production are based on indirect measurements, analyses of acoustic structure and syrinx morphology, and biophysical modeling (Greenewalt, 1968; Stein, 1968; Gaunt et al., 1982; Fletcher, 1988; Brackenbury, 1989). It is only recently that new technologies and methods have become available to measure several features of sound production more directly (Goller and Larsen, 1997; Suthers, 1997). Although this is rapidly changing the existing views on vocal production in birds (Gaunt and Nowicki, 1998; Suthers, 1999), inferences on how vocal variation is linked to sender characteristics are still limited for any bird species. The starting point for our investigation of the production of modulated elements was that circumstantial evidence suggests that modulated elements

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

53

are more difﬁcult to produce than nonmodulated ones. For instance, those birds that produce modulations only in part of their coos have fewer modulations in the very ﬁrst and last coos of a bout (ten Cate, 1992). During these phases, the birds do not seem to use their full capacity to vocalize. Furthermore, if a cooing bird is interrupted by a light ﬂash, this may cause a modulated element to change into a nonmodulated one, but never the other way around (ten Cate and Ballintijn, 1996). We also noticed that some birds that are transferred to a new cage may temporarily produce fewer modulated elements. So, what can be said about the way modulated elements are produced, and is it really likely that their production requires more effort than the production of nonmodulated elements? To address this question we turn to the dove syrinx and various models that have been proposed for vocal production in birds. E. SOUND PRODUCTION IN DOVES The syrinx of the collared dove, as with that of other Columbideae (Warner, 1972; King, 1989), consists of a number of membranes between several modiﬁed rings of the trachea and the primary bronchi (Fig. 5, Ballintijn et al., 1995; Ballintijn and ten Cate, 1997a). Several authors have suggested different prime sources for the sound. Abs (1980), studying domestic pigeons (Columba livia), suggested that vibration of the median tympaniformic membranes, caused by airﬂow along them, was the prime source of sound production. To test this, he constructed an artiﬁcial syrinx and showed that a membrane did indeed vibrate and produce sound when air was blown along it. This model also showed another feature. A gradual rise in airﬂow led to an increase in frequency of the sound that was to some extent proportional to the increase in ﬂow. However, at a certain level of ﬂow, the frequency level abruptly changed to higher frequencies, a phenomenon also known from various types of whistles (Gaunt and Gaunt, 1985). Using the syrinx of a dead pigeon, Abs (1980) demonstrated that here too a gradual rise in air speed led to an increase in frequency of the sound produced, although the sudden change to a higher frequency was lacking. Nevertheless, the ﬁnding of higher airﬂow resulting in higher frequencies being produced, combined with the occurrence of sudden frequency increases, is reminiscent of the ﬁndings in the collared dove, where a higher frequency of nonmodulated elements is linked to a higher proportion of modulations (ten Cate, 1992). Fee et al. (1998) showed that nonlinear dynamics may underlie sudden changes from one stable state of the sound production mechanisms to another, a phenomenon also present in the vocalizations of mammals (Wilden et al., 1998). A gradual rise in airﬂow through an in vitro preparation of a zebra ﬁnch syrinx led to sudden changes in the acoustic structure of elements at

54

CAREL TEN CATE et al.

certain critical ﬂow levels. These observations suggest that a possible hypothesis for the occurrence of modulations may be that the ﬂow through the syrinx reaches a critical level, at which a nonlinear transition in the sound pattern occurs. This proposed hypothesis is based on the assumption that vibrating membranes are the prime source of sound production in doves. However, this mode of sound production has been seriously questioned for doves. Gaunt et al. (1982) and Gaunt and Gaunt (1985) suggested dove coos to be whistles, with the medial tympaniformic membranes (Fig. 5) acting as a static slot through which air is blown, leading to sound production through formation of stable vortices. They arrived at their conclusion after measuring EMGs of syringeal and abdominal muscles and air sac pressure patterns in

FIG. 5. Slightly schematic ventral view (left) and horizontal section (right) of the syrinx of an adult male collared dove. In the ventral view the dark structures are ossiﬁed tissue, the lighter gray ones are muscles and the light parts of the tracheal and bronchial rings are cartilaginous. Not shown is the dorsal (tracheosyringeal) membrane. See Ballintijn et al. (1995) and Ballintijn and ten Cate (1997a) for more details.

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

55

the abdominal air sac of the ringdove (S. risoria). As the ringdove is the closest living relative of the collared dove (Johnson et al., 2001), the latter model of sound production may also be applicable to the collared dove. However, our detailed study of the syrinx morphology of the collared dove (Ballintijn et al., 1995), indicated that the system probably cannot function in the way suggested by Gaunt et al. (1982). For instance, the structure of the medial tympaniformic membranes is highly elastic, ideal for vibrating rather than for staying in a stationary position to create a slot. For this reason, we decided to put the whistle model to a test in the collared dove. If sound is produced in a whistlelike manner, the fundamental frequency is dependent on resonances which stabilize the acoustic disturbance in the airﬂow (Chanaud, 1970). Consequently, altering the resonance frequency of the whistle will result in a correlated shift in its fundamental frequency. Such a shift is not expected if vibrating membranes are the prime source for the sound. A way to alter the resonance properties of the vocal tract is to alter the medium from normal air to a gas with a different density, altering sound velocity. In our experiment (Ballintijn and ten Cate, 1998) we used a closed cage in which we gradually replaced air by heliox (a mixture of helium and oxygen). This mixture allows the birds to breathe normally while increasing sound velocity. The prediction is that if coos are whistled sounds, the frequency should increase in proportion to the heliox concentration. If, on the other hand, a vibrating membrane produces sounds, the frequency should not be affected, or only marginally so. A toy ﬂute could be operated from outside the cage and was used to assess the increase in frequency that a coo would show if it was a whistled sound. We recorded six male collared doves in heliox as well as in normal air. As expected, the frequency for the ﬂute increased in heliox, on average 46%. Nonmodulated coos also showed a signiﬁcant increase in frequency, but with just 3.6% it was marginal compared to the ﬂute rise in heliox. This demonstrated that, like the pure tonal sounds in several songbirds and the calls of budgerigars (Nowicki, 1987; Nowicki and Marler, 1988; Brittan-Powell et al., 1997), the nonmodulated elements of dove coos are not produced by a whistle-type mechanism. However, we did obtain a substantial and signiﬁcant decrease in amplitude of the fundamental frequency of the nonmodulated elements (Fig. 6). This ﬁnding suggests that, although there was no direct coupling between acoustic resonance and the frequency at the sound source, resonances of some sort may be involved in its ampliﬁcation (cf. Hoese et al., 2000). The result of the heliox experiment was noticeably different for modulated elements (Ballintijn and ten Cate, 1998). The ﬁrst phenomenon we observed was a signiﬁcant reduction in the number of modulated elements. However, the most remarkable effect was a systematic change in the frequency proﬁle of the modulated elements

56

CAREL TEN CATE et al.

FIG. 6. Comparison of sonagrams and amplitude envelopes of coos in normal air (left) and in heliox (right) of two collared doves. Dove 111 produced nonmodulated elements only. The heliox atmosphere did not affect frequency, but the amplitude was strongly reduced. Dove 015 produced coos with a modulation in the ﬁrst element in normal air. In heliox, the modulated ﬁrst elements (arrow), but not the nonmodulated ones, show a higher frequency level than in normal air. Again, the amplitude is strongly reduced in heliox (note the different amplitude scales in normal air and heliox for this bird). From Ballintijn and ten Cate (1998).

(Fig. 6). With increasing heliox concentration, the frequency of the modulated part of the elements increased to about 1.5 times the frequency of the base of the elements, after which the increase stopped (Ballintijn and ten Cate, 1998). This change occurred in each of the four birds that showed modulated elements. Although the results show details that raise questions about what precisely is going on, the observed changes are different from the predictions of the whistle model. So, the whistle model of sound production, suggested by studies of the ringdove, can be discarded for the production of modulated as well as nonmodulated elements. This is also suggested by experiments on feral pigeons, Columba livia, using endoscopic techniques (Goller and Larsen, 1997; Larsen and Goller, 1999). Direct observations of vocalizing in anesthetized pigeons, induced by brain stimulation, showed that opening and closing of the lateral tympaniformic membranes (Fig. 5) was correlated with sound production (Larsen and Goller, 1999). This suggests the vibrating movement of these membranes produces the sound, possibly in a way similar to the vocal folds in mammals.

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

57

So, to summarize, the coos in doves are not produced by a whistle-type mechanism. At the same time, the different response to heliox by modulated and nonmodulated elements suggests that different, albeit it as yet unspeciﬁed, modes of production of resonance may underlie their production. It conﬁrms that the complexity of the processes underlying vocal production is greater than earlier models suggested (Gaunt and Nowicki, 1998). To return to the hypothesis that an increased ﬂow of air through the syrinx may induce the occurrence of modulations, this is still a feasible one, with nonlinear processes being a candidate principle for their origin. Assuming for the moment that the hypothesis is correct, is the production of modulations so demanding that some males are unable to produce them, that is, are they true handicaps? One constraint on signal production may be the energetic costs (see Section II.C.3). But one may wonder whether energetic costs make much of a difference between the production of modulated and nonmodulated elements and also whether such a difference is really worth broadcasting. If a signal is needed to broadcast superior energetic reserves, then a high output rate (i.e., a high amount of singing) may be a much better signal than a speciﬁc sound variant. However, if a high airﬂow is required to produce modulations, then the prime message conveyed by modulations might be about respiratory muscle strength: a bird able to produce a higher driving force may also be a stronger ﬁghter in case of a conﬂict. It will be obvious that the mechanism of sound production has to be understood before the cause of individual differences in vocal structure can be fully assessed. Nevertheless, we tentatively conclude that modulations may act as an honest signal of ﬁghting ability.

VI. VARIATION IN ELEMENT NUMBER A. RELATIONSHIP TO SENDER CHARACTERISTICS We now turn to the second parameter for which noticeable individual variation is present: the number of elements within a coo. In our laboratory colony, coos consisting of two elements are quite common. Out of 26 individuals, only 4 never produced a two-element coo. In 15 individuals, two-element coos, if produced, were always the last ones of a bout (Fig. 1). Finally, seven individuals showed two-element coos in other positions during a bout, usually toward the end (Ballintijn and ten Cate 1999b—Fig. 7). Two-element coos not only occurred more toward or at the end of coo-bouts, but the chances of having a bout ending with one or more two-element coos increased with increasing bout length (Ballintijn and ten Cate, 1999b). Over

58

CAREL TEN CATE et al.

FIG. 7. Two coo-bouts of one individual, each consisting of six coos. In A, all coos have three elements. In B, coo three and further lack the third element. Note that this has no effect on the duration of the duty cycle of the coo (i.e., on the interval between the start of one coo and start of the next). After Ballintijn and ten Cate (1999b).

all birds there was a signiﬁcant negative correlation between body weight and proportion of two-element coos per bout (Fig. 8). A detailed comparison of the temporal structure of coos and coo bouts with and without missing third elements revealed that the duration from the beginning of one coo to the beginning of the next one did not differ between the two (Ballintijn and ten Cate, 1999b; see also Fig. 7). This suggests that while no third element is heard, the underlying rhythm of coo production is maintained. For one bird we measured the pattern of tracheal airﬂow and pressure in the abdominal air sacs simultaneously with sound production. This showed that the pressure pattern, but not the ﬂow pattern, of coos with two and three elements had the same duration (G. Beckers, R. Suthers, and C. ten Cate, unpublished). This supports the idea of constancy in the underlying rhythm, at least partly. But why is no sound produced if at least some aspects of the underlying pattern of neuromuscular actions seem maintained? Three, not necessarily mutually exclusive, mechanisms may account for this phenomenon. The ﬁrst one may be declining motivation, resulting in a reduced intensity of the signal during prolonged reiteration of the coo. Although this hypothesis cannot be excluded it does not readily explain why decreasing motivation results in incomplete coos rather than in short bouts. Nor does it explain why elements disappear only toward the end of long bouts, and not short ones. Finally, the observed correlation between body weight and the occurrence of two-element coos does not ﬁt a motivational explanation. A second hypothesis to account for the ﬁndings is the “neuromuscular exhaustion” hypothesis proposed by Lambrechts (Lambrechts and Dhondt,

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

59

FIG. 8. Male body weight correlates negatively with the proportion of coos lacking the third element. Stars: individuals that never produced two-element coos; open circles: individuals that produced two-element coos only at the end of bouts; closed circles: individuals that produced two-element coos at various positions during bouts (r = −0.41, p = 0.036, N = 26). From Ballintijn and ten Cate (1999b).

1988; Lambrechts, 1996). As discussed in Section II.C.4, this hypothesis postulates that high repetition rates of vocal elements may lead to exhaustion in some of the muscles involved in their production. The repetitive coo pattern might be a candidate to suffer from this phenomenon. This hypothesis would explain the pattern of more missing elements toward the end of long bouts. Also, heavier or stronger males might suffer less from exhaustion than lighter ones. The exhaustion hypothesis thus provides a potential explanation for the ﬁndings. However, direct measurements of muscle activity and metabolism are necessary to conﬁrm it. Such measurements are also required to test a third and related hypothesis. This hypothesis is that with increasing bout length the bird may run into problems with keeping up the respiratory activity required to vocalize. Sound production is strongly coupled with respiration (e.g., Brackenbury, 1989; Wild et al., 1998). Rather than a constraint at the level of syringeal muscle activity, as postulated by Lambrechts’ hypothesis, the constraint may be an inability of the respiratory system, causing the bird to run out of breath (Ballintijn and ten Cate, 1999b). In summary, although not conclusive yet, the data suggest that differences in the number of elements may reﬂect honest differences in sender quality.

60

CAREL TEN CATE et al.

B. RECEIVER RESPONSES To test whether coos lacking elements were perceived as less threatening by receivers, we used the same methodology as we did to examine the effect of modulated elements. Playbacks were done, comparing the effect of two coo-variants, differing in one parameter (Slabbekoorn and ten Cate, 1999). In this case the manipulated parameter was presence or absence of a particular element. A total of ﬁve different tests were done each against a matched control series of nonmanipulated coos. The experimental manipulations involved three series, which were: (1) the ﬁrst, second, or third elements of coos were removed and replaced by a silent interval, (2) removal of the third element led to a corresponding shortening of the interval between coos, and (3) the pause between elements one and two was exchanged with the pause between elements two and three (Fig. 9). In contrast to our expectations, replacing the third element by silence did not affect receiver responses. Territorial males responded equally strongly to coos with and without the third element. When the second element was replaced by silence we obtained the same response. It was only when the ﬁrst element was replaced by silence that territorial males showed signiﬁcantly weaker responses. Weaker responses were also obtained in response to shortened coos and to coos in which the pauses between the elements were exchanged (Slabbekoorn and ten Cate, 1999—Fig. 9). The ﬁnding that replacing the ﬁrst element by silence has a stronger effect is in line with results obtained in other species. Several studies, both

FIG. 9. Responses to playbacks of stimuli with manipulated temporal structure against a nonmanipulated control. (Left) Schematic representation of the various stimuli. NM = nonmodulated coos with normal temporal structure (control); NM-3, NM-2, NM-1 are nonmodulated coos for which, respectively, the third, second, and ﬁrst element are replaced by silence. In NM-3∗ the third element is also removed, but not replaced by silence. NM∗ is a nonmodulated coo in which the pause between elements one and two is exchanged with that between two and three. (Right) Percentage of time spent in ﬂight during playback of the stimuli [see Slabbekoorn and ten Cate (1999) for details]. ∗ Indicates signiﬁcant differences at p < 0.05.

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

61

in the ﬁeld and in the laboratory, have shown that modiﬁcations in the ﬁrst part of songs lead to a stronger decline in response than modiﬁcations to later parts (Schubert, 1971; Bremond, ´ 1972; Helb, 1973; Wolffgramm and Todt, 1982; Johnsrude et al., 1994). Also, laboratory studies of zebra ﬁnches (Taeniopygia guttata) have shown that replacing part of a song by silence may go seemingly unnoticed (Cynx, 1993). This latter phenomenon is reminiscent to ﬁndings in humans. Obliteration of words in a sentence, replacing them by noise or silence may go unnoticed by listeners (Warren et al., 1972; Bashford et al., 1988). Finally, studies on other species also indicate that affecting element or interval duration of songs, or the rhythm of sounds, may reduce responsiveness (Falls, 1963; Reinert, 1965; Emlen, 1972; Heidemann and Oring, 1976; Romanowski, 1979; Brenowitz, 1983; Hulse et al., 1984). So, in several respects the results of our experiments on the collared dove are comparable to those obtained in other species. At the same time, however, the outcome raises questions when the communicative context is taken into account. From a functional perspective it makes sense for males to respond only marginally to shortened coos or altered pause durations. These coo variants are outside the species typical range as we observed it in the ﬁeld and the laboratory. At the same time, differences in temporal parameters are the most distinguishing features among the perch-coos of the various species of the genus Streptopelia (Slabbekoorn et al., 1999). So, unless there are instances of interspeciﬁc competition with related species, there is no beneﬁt and there may even be a disadvantage in responding to coos with considerably modiﬁed temporal characteristics. Various Streptopelia species do indeed show reduced responding to the perch-coos of other members of the genus (de Kort and ten Cate, 2001). Also, laboratory studies indicate that temporal features are one parameter that Streptopelia species use to distinguish coo-like sounds (Beckers and ten Cate, 2001). However, as outlined in the previous section, it may beneﬁt a receiver to differentiate between coos with a third element and coos without a third element, but with the same overall duration. Intruders that produce three-element coos only, should be more of a threat than intruders producing two-element coos. Why this is not reﬂected in the responses to playbacks is not clear. The ﬁndings of other species might be taken to suggest that replacing an element by silence goes perceptually unnoticed. If so, lack of a difference in response to stimuli carrying different messages is due to a perceptual constraint. An alternate, and to us more plausible, explanation is that the difference may be noticed by the birds, but is not reﬂected in their response or in our measurements. In our experiments we used nonmodulated coos. As shown in Section V, such coos give rise to weaker responses than modulated coos. Maybe the response to nonmodulated coos with a third element represents the weakest

62

CAREL TEN CATE et al.

response given to any vocalizing intruder. Weaker responses were only obtained in experiments with stimuli outside the species range and indicate the virtual absence of any reactions. If a “ﬂoor effect” is the reason for ﬁnding no difference in response, then we predict that replacing the third element by silence in a modulated coo will reduce the response to playbacks. So, whether the lack of differentiation between two- and three-element coos is a general phenomenon remains an open question.

VII. VOCAL VARIABILITY AND MALE CONTEST: CONCLUSIONS The aim of our study was to examine whether variability in territorial signals provides information about sender quality and whether receivers take this information into account. We have shown that two features of the perch-coo of the collared dove— the occurrence of modulations and the number of elements—are related to weight of the sender. Weight is a potential indicator of competitive strength. As either more muscle strength or a higher respiratory capacity may be needed to produce modulated elements and coos with three elements, the interindividual vocal variation may be an honest reﬂection of sender quality. Although the currently available evidence thus ﬁts the hypothesis that vocal signals used to deter rivals carry information about competitive ability, several aspects of the system await further investigation. For instance, our playback experiments show that receiving males are not only very sensitive to the presence of modulations, but also that male responses to modulated elements are tuned to modulations with a magnitude similar to the population average. We do not know whether the magnitude of modulations is in any way related to other features of males. Further study is also required to analyze whether modulations and element number really affect the success of intruders or defenders in acquiring or defending a territory. Nevertheless, our study shows, at least, the potential for intrasexual selection in shaping territorial signals. It indicates that, in general, more attention should be given to the possibility that the structure of such signals is not arbitrary (cf. Bradbury and Vehrencamp, 1998), but shows linkage to their content, and that interindividual variation is related to differences in quality. Another issue requiring further study concerns the precise linkage between vocal variation, physiology, and strength or competitive abilities of the sender, not just in the context of male–male contest, but also in that of mate choice. The production of vocalizations requires an appropriate morphology of the vocal tract, precise neural coordination, and extensive muscular activity. Vocal variation may thus potentially reveal individual differences in a whole range of morphological, neural, and physiological parameters that

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

63

are also relevant for the ﬁtness of the sender. Just as some aspects of withinindividual motivational variation may give rise to a physiological state that is inextricably linked to the ability to produce certain types of vocalizations (Morton, 1982), physiological and morphological differences between individuals may have the same effect. Examining such relationships provides a challenge for future research, as our current understanding of such links is still limited (see also our review in Section II.C). It will help to understand whether partially correlated variation in separate sound parameters reﬂects different underlying mechanisms or whether these are a consequence of a common causal factor. It will also help to get a better understanding of the precise messages in interindividual vocal variation. Such variation may originate from very different sources. For instance, certain aspects of a vocalization may depend on the morphology of the sound-producing system (e.g., its size or its membrane structure), whereas others depend on the way the system is operated. While the ﬁrst may reﬂect something about the conditions of the birds during rearing (Nowicki et al., 1998a), the latter may reﬂect current condition. Knowing the relationship between vocal parameters and their main determinant may thus allow a much more detailed interpretation of the information present in the signal. This will improve an assessment of their potential or actual relevance in various contexts. It will also reveal whether birds that learn their vocalizations, such as songbirds, differ in this respect from birds that do not, such as doves. This may shed more light on the advantages or disadvantages of song learning. Further studies on the relationship between vocal parameters, vocal production, physiology, and morphology are also relevant to assess whether vocal signals contain redundant information. Redundancy occurs when several parameters of a signal carry the same type of information (Bradbury and Vehrencamp, 1998). Redundancy can be seen as an adaptive trait, evolved to guarantee that receivers detect a message. It is undoubtedly true that signal efﬁcacy might be improved by adding features that exploit “receiver psychology” (Guilford and Dawkins, 1991). In examining potential cases of such redundancy, vocal signals have, for instance, been considered as one component and movement or morphology as another (e.g., Fusani et al., 1997; Møller et al., 1998). One general problem with assessing redundancy is that its assessment is based upon disproving the alternative that the separate components carry different messages (Møller and Pomiankowski, 1993; Johnstone, 1997). One can imagine that the more detailed the level of analysis of the message of a signal, the less likely one is to ﬁnd redundancy. Another issue with particular relevance to vocal signals is whether they are single signals. We suggest that vocalizations, through their complexity and, hence, multitude of sound parameters, might better be viewed as multicomponent signals, potentially carrying multiple messages. When examined in

64

CAREL TEN CATE et al.

some detail, certain vocal parameters may turn out to provide one particular message, others another. In such a case, these different messages may be addressing different audiences, for example, potential competitors and potential mates. So, although at one level “song” can perhaps be seen as one signal with “dual utility” (Berglund et al., 1996), at a different level it can serve as carrier for multiple messages for different audiences. When such messages may give rise to conﬂicting selection pressures, the result may be the evolution of different singing styles or songs for different contexts—a phenomenon present in several bird species (e.g., Morse, 1970; Catchpole, 1983; Groschupf, 1985; Kroodsma et al., 1989). There are certain systematic differences between songs serving for mate attraction and songs involved in territorial behavior (Catchpole and Slater, 1995). It is, however, still hardly explored whether these differences also reﬂect different qualities of the senders. More generally, the conclusion can be that understanding signal content and whether and how it links to signal design still represents a challenge for future research. In spite of all the progress that has been made with respect to theories about why and how animals communicate, there is still a lot to be discovered when it comes down to explaining why birds sing the way they do.

VIII. SUMMARY Bird songs are assumed to have a dual function: attracting mates and deterring rivals. Hence, they are likely to be subject to sexual selection. Most studies on signal evolution through sexual selection concentrate on the ﬁrst function: mate attraction. In this chapter we concentrated on vocal signals involved in male–male competition, and, in particular, territorial defense. Broadcasting species or individual identity and territory ownership are usually considered to be the main functions of territorial signals. Far less is known about whether such signals may also act as assessment signals, providing honest information about the sender’s competitive abilities. On theoretical grounds, such a function is to be expected. We review the evidence that interindividual variation in vocal characteristics may provide a message about male competitive ability. This evidence is fragmentary. Some vocal parameters may be suitable indicators of differences in such abilities, but few studies demonstrate that they are actually used in this way. As a case study on this issue, we examined the role of vocal variation in the territorial “perch-coo” vocalization of the collared dove (Streptopelia decaocto). The perch-coo consists of three elements that are produced in series. Heavier males show a higher proportion of perch-coo

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

65

elements with a “modulation,” a sudden increase in frequency at the start of an element. They also show fewer coos in which the third element is lacking. The available evidence about how doves vocalize suggests that production of both modulated elements and three-element coos require greater strength or respiratory capacity. Playback experiments show that coos with modulated elements give rise to stronger responses from territorial males, that is, they are perceived as originating from a stronger competitor. The strongest responses are given to modulations with a magnitude close to the population average. Presence or absence of the third element only leads to different responses when combined with reduced length of the coo. These results show that at least part of the interindividual variation in acoustic parameters of the perch-coo conveys meaningful information about sender quality to receivers. This demonstrates the potential for intrasexual selection as a factor shaping territorial signals. The ﬁndings highlight the point that the study of signal variation in a competitive context is a neglected but fruitful area for research. The challenge for future studies will be to link vocal variation to speciﬁc modes of vocal production and to relate it to more speciﬁc aspects of “sender quality.” This will help us to understand the relationship between signal content and signal design and may shed light upon the role of birdsong as a multicomponent signal.

Acknowledgments For discussions and comments concerning this paper we thank Gabriel ¨ Beckers, Selvino de Kort, Katharina Riebel, and Jean Secondi. ´ Gerdi Dijkstra provided indispensable secretarial assistance, and Martin Brittijn prepared the ﬁgures.

References Abs, M. (1980). Zur Bioakustik des Stimmbruchs bei Vogeln. ¨ Zool. Jb. Physiol. 84, 289–382. Airey, D. C., Buchanan, K. L., Szekely, T., Catchpole, C. K., and DeVoogd, T. J. (2000). Song, sexual selection, and a song control nucleus (HVc) in the brains of European sedge warblers. J. Neurobiol. 44, 1–6. Alatalo, R. V., Glynn, C., and Lundberg, A. (1990). Singing rate and female attraction in the pied ﬂycatcher: An experiment. Anim. Behav. 39, 601–603. Allan, S. E., and Suthers, R. A. (1994). Lateralization and motor stereotypy of song production in the brown-headed cowbird. J. Neurobiol. 25, 1154–1166. Andersson, M. (1982). Female choice selects for extreme tail length in a widow bird. Nature 299, 818–820. Andersson, M. (1994). “Sexual Selection.” Princeton Univ. Press, Princeton, NJ. Appleby, B. M., and Redpath, S. M. (1997). Indicators of male quality in the hoots of Tawny owls (Strix aluco). J. Raptor. Res. 31, 65–70.

66

CAREL TEN CATE et al.

Arak, A. (1983). Sexual selection by male–male competition in the natterjack toad choruses. Nature 306, 261–262. Austadt, S. (1983). A game theoretical interpretation of male combat in the bowl and doily spider Frontinella pyramitela. Anim. Behav. 31, 59–73. Ballintijn, M. R., and ten Cate, C. (1997a). Sex differences in the vocalizations and syrinx of the collared dove (Streptopelia decaocto). Auk 114, 22–39. Ballintijn, M. R., and ten Cate, C. (1997b). Vocal development and its differentiation in a non-songbird: The Collared dove (Streptopelia decaocto). Behaviour 134, 595–621. Ballintijn, M. R., and ten Cate, C. (1998). Sound production in the collared dove: A test of the “whistle” hypothesis. J. Exp. Biol. 201, 1637–1649. Ballintijn, M. R., and ten Cate, C. (1999a). Acoustic differentiation in the coo vocalizations of the collared dove. Bioacoustics 110, 1–17. Ballintijn, M. R., and ten Cate, C. (1999b). Variation in number of elements in the perch-coo vocalization of the collared dove (Streptopelia decaocto) and what it may tell about the sender. Behaviour 136, 847–864. Ballintijn, M. R., ten Cate, C., Nuijens, F. W., and Berkhoudt, H. (1995). The syrinx of the Collared dove (Streptopelia decaocto): Structure, interindividual variation and development. Neth. J. Zool. 45, 455–479. 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 Univ. Press, Ithaca, NY. Bashford, J. A., Meyers, M. D., Brubaker, B. S., and Warren, R. M. (1988). Illusory continuity of interrupted speech: Speech rate determines durational limits. J. Acoust. Soc. Am. 84, 1635–1638. Beani, L., and Dessi-Fulgheri, F. (1995). Mate choice in the grey partridge, Perdix perdix: A role of physical and behavioural male traits. Anim. Behav. 49, 347–356. Becker, P. H. (1982). The coding of species-speciﬁc characteristics in bird sounds. In “Acoustic Communication in Birds” (D. E. Kroodsma and E. M. Miller, Eds.), pp. 213–252. Academic Press, New York. Beckers, G. J. L., and ten Cate, C. (2001). Perceptual relevance of species-speciﬁc differences in acoustic signal structure in Streptopelia doves. Anim. Behav. In press. Berglund, A., Bisazza, A., and Pilastro, A. (1996). Armaments and ornaments: An evolutionary explanation of traits of dual utility. Biol. J. Linn. Soc. 58, 385–399. Bettman, H. (1959). Der Rhythmus des Taubenrufes. Orn. Mitt. 11, 161–163. Bodenstein, G. (1949). Zur Ethologie and Biologie der Turkentaube ¨ (Streptopelia d. decaocto Friv.). Ornithol. Beobachter 46, 107–116. Borgia, G., and Coleman, S. W. (2000). Co-option of male courtship signals from aggressive display in bowerbirds. Proc. R. Soc. London 267, 1735–1740. Bowman, R. I. (1979). Adaptive morphology of song dialects in Darwin’s ﬁnches. J. Ornithol. 120, 353–389. Brackenbury, J. H. (1989). Functions of the syrinx and control of sound production. In “Form and Function in Birds” (A. S. King and J. McLelland, Eds.), pp. 193–220. Academic Press, New York. Bradbury, J. W., and Vehrencamp, S. L. (1998). “Principles of Animal Communication.” Sinauer, Sunderland, MA. Bremond, ´ J.-C. (1972). Comparison entre l’apprentissage du chant chez le jeune pinson (Fringilla coelebs) et les el ´ ements ´ reactog ´ enes ` du chant territorial de l’adulte. Rev. Comp. Anim. 6, 191–195. Brenowitz, E. A. (1983). The contribution of temporal song cues to species recognition in the red-winged blackbird. Anim. Behav. 31, 1116–1127.

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

67

Brenowitz, E. A., Lent, K., and Kroodsma, D. E. (1995). Brain space for learned song in birds develop independently of song learning. J. Neurosci. 15, 6281–6286. Bretagnolle, V. (1996). Acoustic communication in a group of non-passerine birds, the petrels. In “Ecology and Evolution of Acoustic Communication in Birds” (D. E. Kroodsma and E. H. Miller, Eds.), pp. 160–177. Cornell Univ. Press, Ithaca, NY. Bried, J., and Jouventin, P. (1997). Morphological and vocal variation among subspecies of the black-faced sheathbill. Condor 99, 818–825. Brittan-Powell, E. F., Dooling, R. F., Larsen, O. N., and Heaton, J. T. (1997). Mechanisms of vocal production in budgerigars (Melopsittacus undulatus.) J. Acoust. Soc. Am. 101, 578–589. Brown, W. D., Wideman, J., Andrade, M. C. B., Mason, A. C., and Gwynne, D. T. (1996). Female choice for an indicator of male size in the song of the black-horned tree cricket, Oecanthus nigricornis (Orthoptera: gryllidae: oecanthinae). Evolution 50, 2400–2411. Butchart, S. H. M., Seddon, N., and Ekstrom, J. M. M. (1999). Yelling for sex: Harem males compete for female access in bronze-winged jacanas. Anim. Behav. 57, 637–646. Catchpole, C. K. (1973). The functions of advertising song in the sedge warbler (Acrocephalus schoenobanus) and reedwarbler (A. scirpaceus). Behaviour 46, 300–320. Catchpole, C. K. (1980). Sexual selection and the evolution of complex songs among warblers of the genus Acrocephalus. Behaviour 74, 149–166. Catchpole, C. K. (1983). Variation in the song of the great reed warbler Acrocephalus arundinaceus in relation to mate attraction and territorial defense. Anim. Behav. 31, 1217– 1225. Catchpole, C. K. (1996). Song and female choice: Good genes and big brains? TREE 11, 358–360. Catchpole, C. K., and Slater, P. J. B. (1995). “Bird Song: Biological Themes and Variations.” Cambridge Univ. Press, Cambridge, UK. Catchpole, C. K., Dittami, J., and Leisler, B. (1984). Differential responses to male song repertoires in female songbirds implanted with oestradiol. Nature 3122, 563–564. Chanaud, R. C. (1970). Aerodynamic whistles. Sci. Am. 221, 40–46. Chappell, M. A., Zuk, M., Kwan, T. H., and Johnsen, T. S. (1995). Energy cost of an avian vocal display: Crowing in red junglefowl. Anim. Behav. 49, 255–257. Cheng, M.-F. (1979). Progress and prospects in ring dove research: A personal view. Adv. Study Behav. 9, 97–129. Academic Press, New York. Cheng, M.-F. (1986). Female cooing promotes ovarian development in ring doves. Physiol. Behav. 37, 371–374. Cheng, M.-F., Peng, J. P., and Johnson, P. (1998). Hypothalamic neurons preferentially respond to female nest coo stimulation: Demonstration of direct acoustic stimulation of luteinizing hormone release. J. Neurosci. 18, 5477–5489. Clutton-Brock, T. H., and Albon, S. D. (1979). The roaring of red deer and the evolution of honest advertisement. Behaviour 69, 145–170. Collins, S. A. (1999). Is female bias for male repertoires due to sensory bias? Proc. R. Soc. London B. 266, 2309–2314. Cosens, S. E., and Sealy, S. G. (1986). Age-related variation in song repertoire size and repertoire sharing of yellow warblers (Dendroica petechia). Can. J. Zool. 64, 1926–1929. Craig, W. (1908). The voices of pigeons regarded as a means of social control. Am. J. Sociol. 14, 86–100. Cramp, S. (1985). “The Birds of the Western Palearctic,” Vol. IV, Oxford Univ. Press, Oxford. Cronin, H. (1991). “The Ant and the Peacock.” Cambridge Univ. Press, Cambridge, UK. Cuthill, I. C., and Macdonald, W. A. (1990). Experimental manipulation of dawn and dusk chorus in the blackbird Turdus merula. Behav. Ecol. Sociobiol. 26, 209–216. Cynx, J. (1993). Conspeciﬁc song perception in zebra ﬁnches (Taeniopygia guttata). J. Comp. Psychol. 107, 395–402.

68

CAREL TEN CATE et al.

Dabelsteen, T., and Pedersen, S. B. (1985). Correspondence between messages in the full song of the blackbird Turdus merula and meanings to territorial males, as inferred from responses to computerized modiﬁcations of natural song. Z. Tierpsychol. 69, 149–165. Dabelsteen, T., and Pedersen, S. B. (1992). Song features essential for species discrimination and behaviour assessment by male blackbirds (Turdus merula). Behaviour 121, 259–287. Darwin, C. (1871). “The Descent of Man, and Selection in Relation to Sex.” J. Murray, London, UK. Davies, N. B., and Halliday, T. R. (1978). Deep croaks and ﬁghting assessment in toads Bufo bufo. Nature 274, 683–685. Davies, S. J. J. F. (1970). Patterns of inheritance in the bowing display and associated behaviour of some hybrid Streptopelia doves. Behaviour 36, 187–214. Davies, S. J. J. F. (1974). Studies of the three coo-calls of the male barbary dove. Emu 74, 18–26. de Kort, S., and ten Cate, C. (2001). Response to interspeciﬁc vocalisations is affected by degree of phylogenetic relatedness in Streptopelia doves. Anim. Behav. 61, 239–247. Derrickson, K. C. (1987). Yearly and situational changes in the estimate of repertoire size in northern mockingbirds (Mimus polyglottos). Auk 104, 198–207. DeVoogd, T. J., Krebs, R., Healy, S., and Purvis, A. (1993). Relations between sing repertoire and the volume of brain nuclei related to song: Comparative evolution analyses amongst oscine birds. Proc. R. Soc. London B 254, 75–82. Dyrcz, R. J. (1956). On the biology and distribution of the Collared Turtle dove, Streptopelia Decaocto, Friv. in Poland. Zool. Poloniae 7, 433–454. Eberhardt, L. S. (1994). Oxygen consumption during singing by male Carolina Wrens (Thryothorus ludovicianus). Auk 111, 124–130. Eberhardt, L. S. (1996). Energy expenditure during singing: A reply to Gaunt et al. Auk 113, 721–723. Eens, M. (1997). Understanding the complex song of European Starling: An integrated ethological approach. Adv. Study Behav. 26, 355–434. Eens, M., Pinxten, R., and Verheyen, R. F. (1991). Male song as a cue for mate choice in the European starling. Behaviour 116, 210–238. Emlen, S. (1972). An experimental analysis of the parameters of bird song eliciting species recognition. Behaviour 41, 130–171. Englund, G., and Olsson, T. I. (1990). Fighting and assessment in the net-spinning caddis larva Arctopsyche ladogensis: A test of the sequential assessment game. Anim. Behav. 39, 55–62. Englund, G., and Otto, C. (1991). Effects of ownership status, weight asymmetry and case ﬁt on the outcome of case contests in two populations of Agrypnia pagetana (Trichoptera: Phryganeidae) larvae. Behav. Ecol. Sociobiol. 29, 113–120. Enquist, M., and Leimar, O. (1987). Evolution of ﬁghting behaviour: The effect of variation in resource value. J. Theor. Biol. 127, 187–205. Eriksson, D., and Wallin, L. (1986). Male bird song attracts females—A ﬁeld experiment. Behav. Ecol. Sociobiol. 19, 297–299. Falls, J. B. (1963). Properties of bird song eliciting responses from territorial males. Proc. XIII Int. Orn. Congr. 259–271. Falls, J. B. (1988). Does song deter intrusion in white-throated sparrows (Zonotrichia albicollis)? Can. J. Zool. 66, 206–211. Farquhar, C. C. (1993). Individual and intersexual variation in the alarm calls of the White-tailed Hawk. Condor 95, 234–239. Fee, M. S., Shraiman, B., Pesaran, B., and Mitra, P. P. (1998). The role of nonlinear dynamics of the syrinx in the vocalizations of a songbird. Nature 395, 67–71. Fisher, R. A. (1930). “The Genetical Theory of Natural Selection.” Oxford Univ. Press, Oxford, UK.

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

69

Fitch, W. T. (1999). Acoustic exaggeration of size in birds by tracheal elongation: Comparative and theoretical analyses. J. Zool. (London) 248, 31–49. Fletcher, N. H. (1988). Bird song—A quantitative acoustic model. J. Theor. Biol. 135, 455–481. Frieling, H. (1960). Dialekt bei Turkentauben. ¨ Orn. Mitt. 12, 120. Fusani, L., Hutchison, R. E., and Hutchison, J. B. (1997). Vocal-postural co-ordination of a sexually dimorphic display in a monomorphic species: The Barbary dove. Behaviour 134, 321–335. Galeotti, P. (1998). Correlates of hoot rate and structure in male Tawny Owls Strix aluco: Implications for male rivalry and female mate choice. J. Avian Biol. 29, 25–32. Galeotti, P., Saino, N., Sacchi, R., and Møller, A. P. (1997). Song correlates with social context, testosterone and body condition in male Barn swallows. Anim. Behav. 53, 687–700. Gaunt, A. S., and Gaunt, S. L. L. (1985). Syringeal structures and avian phonation. In “Current Ornithology” (R. F. Johnston, Ed.), pp. 213–245. Plenum Press, New York. Gaunt, A. S., and Nowicki, S. (1998). Sound production in birds: Acoustics and physiology revisited. In “Animal Acoustic Communication, Sound Analysis, and Research Methods” (S. L. Hopp, M. J. Owren, and C. S. Evans, Eds.), pp. 291–321. Springer, Heidelberg/Berlin. Gaunt, A. S., Gaunt, S. L. L., and R. M. Casey. (1982). Syringeal mechanics reassessed: Evidence from Streptopelia. Auk 99, 474–494. Gaunt, A. S., Bucher, T. L., Gaunt, S. L. L., and Baptista, L. F. (1996). Is singing costly? Auk 113, 718–721. Genevois, F., and Bretagnolle, V. (1994). Male blue petrels reveal their body mass when calling. Ethol. Ecol. Evol. 6, 377–384. Gibson, R. M., and Bradbury, J. W. (1985). Sexual selection in lekking sagegrouse: Phenotypic correlates of male mating success. Behav. Ecol. Sociobiol. 18, 112–123. 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. Goller, F., and Larsen, O. N. (1997). A new mechanism of sound generation in songbirds. Proc. Nat. Acad. Sci. U.S.A. 94, 14787–14791. Goodwin, D. (1970). “Pigeons and Doves of the World.” British Museum, London, UK. Grafen, A. (1987). The logic of divisively asymmetric contests. Respect for ownership and the desperado effect. Anim. Behav. 35, 462–467. Grafen, A. (1990a). Biological signals as handicaps. J. Theor. Biol. 144, 517–546. Grafen, A. (1990b). Sexual selection unhandicapped by the Fisher process. J. Theor. Biol. 144, 473–516. Gray, D. A., and Hagelin, J. C. (1996). Song repertoires and sensory exploitation: Reconsidering the case of the common grackle. Anim. Behav. 52, 795–800. Greenewalt, C. H. (1968). “Bird Song: Acoustics and Physiology.” Smithsonian Institution Press, Washington DC. Groschupf, K. (1985). Changes in ﬁve-striped sparrow song in intra- and intersexual contexts. Wilson Bull. 97, 102–105. Guilford, T., and Dawkins, M. S. (1991). Receiver psychology and the evolution of animal signals. Anim. Behav. 42, 1–15. Gurtler, ¨ W. (1973). Artisolierende Parameter des Revierrufes der Turkentaube ¨ (Streptopelia decaocto). J. Orn. 114, 305–316. Gurtler, ¨ W. (1977). Analyse der artkennzeichnenden Parameter im Revierruf der Turkentaube ¨ (Streptopelia decaocto) mit weiteren Untersuchungen ihres Rufrepertoires. Ph.D. Thesis, Friedrich-Alexander-Universitat, ¨ Erlangen-Nurnberg. ¨ Hailman, J. P., Ficken, M. S., and Ficken, R. W. (1987). Constraints on the structure of combinatorial “chick-a-dee” calls. Ethology 75, 62–80.

70

CAREL TEN CATE et al.

Haley, M. P. (1994). Resource-holding power asymmetries, the prior residence effect, and reproductive payoffs in male northern elephant seal ﬁghts. Behav. Ecol. Sociobiol. 34, 427–434. Hamilton, W. D., and Zuk, M. (1982). Heritable true ﬁtness and bright birds: A role for parasites? Science 218, 384–387. Handford, P., and Lougheed, S. C. (1991). Variation in duration and frequency characters in the song of the rufous-collared sparrow, Zonotrichia capensis, with respect to habitat, trill dialects and body size. Condor 93, 644–658. Hasselquist, D. (1998). Polygyny in great reed warblers: A long-term study of factors contributing to male ﬁtness. Ecology 79, 2376–2390. Hasselquist, D., Bensch, S., and von Schantz, T. (1996). Correlation between male song repertoire, extra-pair paternity and off-spring survival in the great reed warbler. Nature 381, 229–232. Hauser, M. D. (1993). The evolution of nonhuman primate vocalizations: Effects of phylogeny, body weight and motivational state. Am. Nat. 142, 528–542. Heidemann, M. K., and Oring, L. W. (1976). Functional analysis of spotted sandpiper (Actitis macularia) song. Behaviour 56, 181–193. Helb, H. -W. (1973). Analyse der artisolierende Parameter im Gesang des Fitis (Phylloscopus t. trochilus) mit Untersuchungen zur objectivierung der analytischen Methode. J. Orn. 114, 145–206. Hoese, W. J., Podos, J., Boetticher, N. C., and Nowicki, S. (2000). Vocal tract function in birdsong production: Experimental manipulation of beak movements. J. Exp. Biol. 203, 1845–1855. Hofstetter, F. B. (1954). Untersuchungen an einer Population der Turkentaube. ¨ J. Orn. 95, 348–411. Horn, A., Leonard, M. L., and Weary, D. M. (1995). Oxygen consumption during crowing by roosters: Talk is cheap. Anim. Behav. 50, 1171–1175. Horn, A. G., Dickinson, T. E., and Falls, J. B. (1993). Male quality and song repertoires in western meadowlarks (Sturnella neglecta). Can. J. Zool. 71, 1059–1061. Hulse, S. H., Humpal, J., and Cynx, J. (1984). Discrimination and generalization of rhythmic and arrhythmic patterns by European starlings (Sturnus vulgaris). Music Percep. 1, 442–464. James, P. C., and Robertson, H. A. (1989). Sexual dimorphism of chatter-calls and vocal sex recognition in Leach’s Storm-Petrel (Oceanodroma leucorboa). Auk 106, 498–501. Johnson, K., de Kort, S., Dinwoodey, K., Mateman, A. C., ten Cate, C., Lessells, C. M., and Clayton, D. H. (2001). A molecular phylogeny of the dove genus Streptopelia. Auk. In press. Johnsrude, I. S., Weary, D. M., Ratcliffe, L. M., and Weisman, R. G. (1994). Effect of motivational context on conspeciﬁc song discrimination by brown-headed cowbirds (Molothrus ater). J. Comp. Psychol. 108, 172–178. Johnstone, R. A. (1997). The evolution of animal signals. In “Behavioural Ecology: An Evolutionary Approach.” (J. R. Krebs and N. B. Davies, Eds.), pp. 155–178. Blackwell Science, Oxford, UK. Kempenaers, B., Verheyen, G. R., Dhondt, A. A. (1997). Extrapair paternity in the blue tit (Parus caeruleus): Female choice, male characteristics, and offspring quality. Behav. Ecol. 8, 481–492. King, A. S. (1989). Functional anatomy of the syrinx. In “Form and Function in Birds,” Vol. 4 (A. S. King and J. McLelland, Eds.), pp. 105–191. Academic Press, New York. Kirkpatrick, M. (1982). Sexual selection and the evolution of female choice. Evolution 36, 1–12. Krebs, J. R. (1977). Song and territory in the great tit Parus major. In “Evolutionary Ecology” (B. Stonehouse and C. Perrins, Eds.), pp. 47–62. Macmillan, London. Krebs, J. R. (1982). Territorial defense in the great tit (Parus major): Do residents always win? Behav. Ecol. Sociobiol. 11, 185–194.

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

71

Krebs, J. R., Ashcroft, R., and Webber, M. (1978). Song repertoires and territory defense in the great tit. Nature 271, 539–542. Kroodsma, D. E., and Byers, B. E. (1991). The function(s) of bird song. Am. Zool. 31, 318–328. Kroodsma, D. E., Bereson, R. C., Byers, B. E., and Minear, E. (1989). Use of song types by the chestnut-sided warbler: Evidence for both intra- and inter-sexual functions. Can. J. Zool. 67, 447–456. Lade, B. I., and Thorpe, W. H. (1964). Dove songs as innately coded patterns of speciﬁc behaviour. Nature 202, 366–368. Lambrechts, M. M. (1996). Organization of birdsong and constraints on performance. In “Ecology and Evolution of Acoustic Communication in Birds” (D. E. Kroodsma and E. H. Miller, Eds.), pp. 305–320. Cornell Univ. Press, Ithaca, NY. Lambrechts, M. M. (1997). Song frequency plasticity and composition of phrase versions in Great tits Parus major. Ardea 85, 99–109. Lambrechts, M., and Dhondt, A. A. (1986). Male quality, reproduction, and survival in the great tit (Parus major). Behav. Ecol. Sociobiol. 19, 57–64. Lambrechts, M., and Dhondt, A. A. (1987). Differences in singing performance between male great tits. Ardea 75, 43–52. Lambrechts, M. M., and Dhondt, A. A. (1988). The anti-exhaustion hypothesis: A new hypothesis to explain song performance and song switching in the great tit. Anim. Behav. 36, 327–334. Lampe, H., and Espmark, Y. O. (1994). Song structure reﬂects male quality in pied ﬂycatchers. Anim. Behav. 47, 869–876. Lande, R. (1981). Models of speciation by sexual selection on polygenic traits. Proc. Natl. Acad. Sci. U.S.A. 78, 3721–3725. Larsen, O. N., and Goller, F. (1999). Role of syringeal vibrations in bird vocalizations. Proc. R. Soc. London B 266, 1609–1615. 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. Wiley, New York. Leonard, M. L., and Horn, A. G. (1995). Crowing in relation to status in roosters. Anim. Behav. 49, 1281–1290. Lovari, S., and Hutchison, J. B. (1975). Behavioural transitions in the reproductive cycle of barbary doves (Streptopelia risoria L.). Behaviour 53, 126–150. Martens, J. (1996). Vocalisations and speciation of Palearctic birds. In “Ecology and Evolution of Acoustic Communication in Birds” (D. E. Kroodsma and E. H. Miller, Ed.), pp. 221–241. Cornell Univ. Press, Ithaca, NY. Martin-Vivaldi, M., Palomino, J. J., and Soler, M. (1998). Song structure in the Hoopoe (Upapa epops). Strophe length reﬂects male condition. J. Ornithol. 139, 287–296. Martin-Vivaldi, M., Palomino, J. J., Soler, M., and Martinez, J. G. (1999). Song strophe-length and reproductive success in a non-passerine bird, the Hoopoe Upupa epops. Ibis 141, 670–679. McDonald, M. V. (1989). Function of song in Scott’s seaside sparrow, Ammodramus maritimus peninsulae. Anim. Behav. 38, 468–485. McGregor, P. K. (1991). The singer and the song: On the receiving end of bird song. Biol. Rev. 66, 57–81. Miller, E. H. (1996). Acoustic differentiation and speciation in shorebirds. In “Ecology and Evolution of Acoustic Communication in Birds” (D. E. Kroodsma and E. H. Miller, Ed.), pp. 241–257. Cornell Univ. Press, Ithaca, NY. Møller, A. P. (1991). Parasite load reduces song output in a passerine bird. Anim. Behav. 41, 723–730.

72

CAREL TEN CATE et al.

Møller, A. P., and Pomiankowski, A. (1993). Why have birds got multiple sexual ornaments? Behav. Ecol. Sociobiol. 3, 167–176. Møller, A. P., Saino, N., Taramino, G., Galeotti, P., and Ferrario, S. (1998). Paternity and multiple signalling: Effects of a secondary sexual character and song on paternity in the Barn Swallow. Am. Nat. 151, 236–242. Morse, D. H. (1970). Territorial and courtship songs of birds. Nature 226, 659–661. Morton, E. S. (1977). On the occurrence and signiﬁcance of motivation-structural rules in some bird and mammal sounds. Am. Nat. 111, 855–869. Morton, E. S. (1982). Grading, discreteness, redundancy and motivational-structural rules. In “Evolution and Ecology of Acoustic Communication in Birds” (D. E. Kroodsma and E. H. Miller, Eds.), pp. 183–212. Academic Press, New York. Mountjoy, D. J., and Lemon, R. E. (1991). Song as an attractant for male and female European starlings, and the inﬂuence of song complexity on their response. Behav. Ecol. Sociobiol. 28, 97–100. 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. Myrberg, A. A., Ha, S. J., and Shamblott, M. J. (1993). The sounds of bicolor damselﬁsh (Pomacentrus partitus)—Predictors of body size and a spectral basis for individual recognition and assessment. J. Acoust. Soc. Am. 94, 3067–3070. Nelson, D. A. (1988). Feature weighting in species song recognition by the ﬁeld sparrow (Spizella pussilla). Behaviour 106, 158–182. Nelson, D. A. (1989). The importance of invariant and distinctive features in species recognition of bird song. Condor 91, 120–130. Neubauer, R. L. (1999). Super-normal length song preferences of female zebra ﬁnches (Taeniopygia guttata) and a theory of the evolution of bird song. Evol. Ecol. 13, 365–380. Nottebohm, F., and Nottebohm, M. E. (1971). Vocalizations and breeding behaviour of surgically deafened ring doves, Streptopelia risoria. Anim. Behav. 19, 313–327. Nottebohm, F., and Nottebohm, M. E. (1978). Relationship between song repertoire and age in the canary Serinus canaria. Z. Tierpsychol. 46, 298–305. Nowak, E. (1965). “Die Turkentaube.” ¨ Kosmos-verlag, Stuttgart, Germany. Nowicki, S. (1987). Vocal tract resonances in oscine bird sound production. Evidence from birdsong in a helium atmosphere. Nature (London) 325, 53–55. Nowicki, S., and Marler, P. (1988). How do birds sing? Music Percep. 5, 391–426. Nowicki, S., Peters, S., and Podos, J. (1998a). Song learning, early nutrition and sexual selection in songbirds. Am. Zool. 38, 179–190. Nowicki, S., Searcy, W. A., and Hughes, M. (1998b). The territory defence function of song in song sparrows: A test with the speaker occupation design. Behaviour 135, 615–628. Nuechterlein, G. L. (1981). Variations and multiple functions of the advertising display of western grebes. Behaviour 76, 289–317. Olsson, M. (1992). Contest success in relation to size and residency in the sand lizards, Lacerta agilis. Anim. Behav. 44, 386–388. Peek, F. W. (1972). An experimental study of the territorial function of vocal and visual display in the male red-winged blackbird. Anim. Behav. 20, 112–118. Pfennig, K. S. (1998). The evolution of mate choice and the potential for conﬂict between species and mate-quality recognition. Proc. R. Soc. London B 265, 1743–1748. Podos, J. (1996). Motor constraints on vocal development in a song bird. Anim. Behav. 51, 1060–1070. Podos, J. (1997). A performance constraint on the evolution of trilled vocalizations in a songbird family (Passeriformes: Emberizidae). Evolution 51, 537–551. Read, A. F., and Weary, D. M. (1990). Sexual selection and the evolution of bird song: A test of the Hamilton-Zuk hypothesis. Behav. Ecol. Sociobiol. 26, 47–56.

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

73

Read, A. F., and Weary, D. M. (1992). The evolution of bird song: Comparative analyses. Philos. Trans. R. Soc. London B 338, 165–187. Rehsteiner, U., Geisser, H., and Reyer, H.-U. (1998). Singing and mating success in water pipits: One speciﬁc song element makes all the difference. Anim. Behav. 55, 1471–1481. Reid, M. L. (1987). Costliness and reliability in the singing vigour of Ipswich sparrows. Anim. Behav. 35, 1735–1743. Reinert, J. (1965). Takt- und Rhythmusunterscheidung bei Dohlen. Z. Tierpsychol. 22, 623–671. Robertson, (1986a). Male territoriality, ﬁghting and assessment of ﬁghting ability in the Australian frog Uperoleria rugosa. Anim. Behav. 34, 763–772. Robertson, (1986b). Female choice, male strategies and the role of vocalizations in the Australian frog Uperoleria rugosa. Anim. Behav. 34, 773–784. Romagosa, C. M., and McEneaney, T. (1999). Eurasian collared-dove in North America and the Carribbean. North Am. Birds 53, 348–353. Romanowski, E. (1979). Der Gesang von Sumpf- und Weidenmeise (Parus palustrus und Parus montanus)—Reaktionsauslosende ¨ Parameter. Vogelwarte 30, 48–65. Rosenberg, R. H., and Enquist, M. (1991). Contest behaviour in Weidemeyer’s admiral butterﬂy Limenitis weidemeyerii (Nymphalidae): The effect of size and residency. Anim. Behav. 42, 805–811. Rosenﬁeld, R., and Bielefeldt, J. (1991). Vocalizations of Cooper’s Hawk during the incubation stage. Condor 93, 659–665. Ryan, M. J. (1985). “The Tungara Frog.” Univ. of Chicago Press, Chicago. Ryan, M. J. (1988). Energy, calling and selection. Am. Nat. 28, 885–898. Ryan, M. J., and Brenowitz, E. A. (1985). The role of the body size, phylogeny, and ambient noise in the evolution of bird song. Am. Nat. 126, 87–100. Ryan, M. J., and Rand, A. S. (1993a). Species recognition and sexual selection as a unitary problem in animal communication. Evolution 47, 647–657. Ryan, M. R., and Rand, A. S. (1993b). Sexual selection and signal evolution: The ghost of biases past. Philos. Trans. R. Soc. London B 340, 187–195. Ryan, M. J., Fox, W., Wilczyski, W., and Rand, A. S. (1990). Sexual selection for sensory exploitation in the frog Physalaemus pustulosus. Nature 343, 66–67. Schubert, M. (1971). Untersuchungen uber ¨ die reaktionsauslosenden ¨ Signalstrukturen des Fitisgesanges, Phylloscopus t. trochilus (L.), und das Verhalten gegenuber ¨ arteigenen Rufen. Behaviour 38, 250–288. Searcy, W. A. (1992). Song repertoire and mate choice in birds. Am. Zool. 32, 71–80. Searcy, W. A., and Nowicki, S. (2000). Male–male competition and female choice in the evolution of vocal signaling. In “Animal Signals: Signalling and Signal Design in Animal Communication” (Y. Espmark, T. Amundsen, and G. Rosenqvist, Eds.), pp. 301–315. Tapir Academic Press, Trondheim. Searcy, W. A., and Yasukawa, K. (1996). Song and female choice. In “Ecology Evolution of Acoustic Communication in Birds” (D. E. Kroodsma and E. H. Miller, Ed.), pp. 454–473. Cornell Univ. Press, Ithaca, NY. Shy, E. (1983). The relation of geographical variation in song to habitat characteristics and body size in North American tanagers (Thraupinae: Piranga). Behav. Ecol. Sociobiol. 12, 71– 76. Simmons, L. W. (1995). Correlates of male quality in the ﬁeld cricket, Gryllus campestris L.: Age, size, and symmetry determine pairing success in ﬁeld populations. Behav. Ecol. 6, 376–381. Slabbekoorn, H. (1998). Perception of coo variation by the collared dove. A parameter analysis using playbacks in the ﬁeld. Thesis, Leiden University, The Netherlands. Slabbekoorn, H., and ten Cate, C. (1996). Responses of collared doves to playback of coos. Behav. Processes 38, 169–174.

74

CAREL TEN CATE et al.

Slabbekoorn, H., and ten Cate, C. (1997). Stronger territorial responses to frequency modulated coos in collared doves. Anim. Behav. 54, 955–965. Slabbekoorn, H., and ten Cate, C. (1998a). Perceptual tuning to frequency characteristics of territorial signals in collared doves. Anim. Behav. 56, 847–857. Slabbekoorn, H., and ten Cate, C. (1998b). Multiple parameters in the territorial coo of the collared dove: Interactions and meaning. Behaviour 135, 879–895. Slabbekoorn, H., and ten Cate, C. (1999). Collared dove responses to playback: Slaves to the rhythm. Ethology 105, 377–391. Slabbekoorn, H., de Kort, S., and ten Cate, C. (1999). Comparative analysis of perch-coo vocalizations in Streptopelia doves. Auk 116, 737–748. Smith, D. G. (1979). Male singing ability and territory integrity in red-winged blackbirds (Agelaius phoeniceus). Behaviour 68, 193–206. Smith, J. N. M., and Arcese, P. (1989). How ﬁt are ﬂoaters? Consequences of alternative territorial behaviors in a non-migratory sparrow. Am. Nat. 133, 830–845. SOVON (1987). “Atlas van de Nederlandse Vogels.” Arnhem, The Netherlands. Stearns, S., and Hoekstra, R. F. (2000). “Evolution: An Introduction.” Oxford Univ. Press, Oxford. Stein, R. C. (1968). Modulation in bird sound. Auk 94, 229–243. Suthers, R. A. (1997). Peripheral control and lateralization of birdsong. J. Neurobiol. 33, 632–652. Suthers, R. A. (1999). The motor basis of vocal performance in songbirds. In “The Design of Animal Communication” (M. D. Hauser and M. Konishi, Eds.), pp. 37–62. MIT Press, Cambridge, MA. Suthers, R., and Goller, F. (1997). Motor correlates of vocal diversity in songbirds. In “Current Ornithology,” Vol. 14 (V. Nolan, Jr., E. Ketterson, and C. F. Thompson, Eds.), pp. 235–288. Plenum Press, New York. Suthers, R., Goller, F., and Hartley, R. S. (1994). Motor dynamics of song production by mimic thrushes. J. Neurobiol. 25, 917–936. ten Cate, C. (1992). Coo types in the collared dove Streptopelia decaocto: One theme, distinctive variations. Bioacoustics 4, 161–183. ten Cate, C., and Ballintijn, M. R. (1996). Dove coos and ﬂashed lights: Interruptibility of “song” in a nonsongbird. J. Comp. Psychol. 110, 267–275. ten Cate, C., and Hilbers, J. (1991). Effect of brood size on inter-clutch intervals, offspring development and male–female interactions in the ring dove Streptopelia risoria. Anim. Behav. 41, 27–36. ten Cate, C., Lea, R. W., Ballintijn, M. R., and Sharp, P. J. (1993). Brood size affects behavior, interclutch interval, LH levels and weight in ring dove (Streptopelia risoria) breeding pairs. Horm. Behav. 27, 539–550. Todt, D., and Naguib, M. (2000). Vocal interactions in birds: The use of song as a model in communication. Adv. Study Behav. 29, 247–296. Tubaro, P. L., and Mahler, B. (1998). Acoustic frequencies and body mass in new world doves. Condor 100, 54–61. Vallet, E., and Kreutzer, M. (1995). Female canaries are sexually responsive to special song phrases. Anim. Behav. 49, 1603–1610. Vallet, E., Beme, I., and Kreutzer, M. (1998). Two-note syllables in canary songs elicit high levels of sexual display. Anim. Behav. 55, 291–297. Vehrencamp, S. L., Bradbury, J. W., and Gibson, R. M. (1989). The energetic cost of display in male sage grouse. Anim. Behav. 38, 885–986. Wagner, Jr., W. E. (1989). Fighting, assessment and frequency of alteration in Blanchard’s cricket frog. Behav. Ecol. Sociobiol. 25, 429–436.

SONG AND MALE–MALE COMPETITION IN COLLARED DOVES

75

Warner, R. W. (1972). The syrinx in the family Columbidae. J. Zool. London 166, 385–390. Warren, R. M., Obusek, C. J., and Ackroff, J. M. (1972). Auditory induction: Perceptual synthesis of absent sounds. Science 176, 1149–1151. Weary, D. M., Krebs, J. R., Eddyshaw, R., McGregor, P. K., and Horn, A. (1988). Decline in song output by great tits: Exhaustion or motivation? Anim. Behav. 36, 1242–1244. Weary, D. M., Lambrechts, M. M., and Krebs, J. R. (1991). Does singing exhaust male great tits? Anim. Behav. 41, 540–542. Wells, K. D., and Taigen, T. L. (1986). The effect of social interactions on calling energetics in the Gray Treefrog (Hyla versicolor). Behav. Ecol. Sociobiol. 19, 8–18. West-Eberhard, M. J. (1983). Sexual selection, social competition, and speciation. Q. Rev. Biol. 58, 155–183. Whitehead, P. J. (1999). Boofheads with deep voices: Sexual dimorphism in the Magpie Goose Anseranas semipalmata. Wildfowl 49, 72–91. Wild, J. M., Goller, F., and Suthers, R. A. (1998). Inspiratory muscle activity during bird song. J. Neurobiol. 36, 441–453. Wilden, I., Herzel, H., Peters, G., and Tembrock, G. (1998). Subharmonics, biphonation, and deterministic chaos in mammal vocalisation. Bioacoustics 9, 171–196. Wolffgramm, J., and Todt, D. (1982). Pattern and time speciﬁcity in vocal responses of blackbirds Turdus merula L. Behaviour 81, 264–286. Wurdinger, ¨ I. (1970). Erzeugung, Ontogenie und Funktion der Lautausserungen ¨ bei vier Gansearten ¨ (Anser indicus, A. caerulescens, A. albifrons, und Branta canadensis). Z. Tierpsychol. 27, 257–302. Yasukawa, K. (1981). Song repertoires in the red-winged blackbird (Agelaius phoeniceus): A test of the Beau Geste hypothesis. Anim. Behav. 29, 114–125. Yasukawa, K., and Searcy, W. A. (1985). Song repertoires and density assessment in red-winged blackbirds: Further tests of the Beau Geste hypothesis. Behav. Ecol. Sociobiol. 16, 171–175. 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., and Zahavi, A. (1997). “The Handicap Principle.” Oxford Univ. Press, Oxford, UK.

This Page Intentionally Left Blank

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 31

Imitation of Novel Complex Actions: What Does the Evidence from Animals Mean? RICHARD W. BYRNE SCOTTISH PRIMATE RESEARCH GROUP SCHOOL OF PSYCHOLOGY UNIVERSITY OF ST. ANDREWS FIFE KY16 9JU, SCOTLAND, UNITED KINGDOM

I. INTRODUCTION Despite a ﬂurry of interest in imitation of bodily actions by nonhumans, little clear agreement has resulted as to its cognitive mechanism, adaptive function, and species distribution (Galef, 1988; Heyes, 1993; Heyes and Ray, 2000; Meltzoff, 1996; Mitchell, 1987; Tomasello et al., 1993; Whiten and Ham, 1992; Zentall, 2001). These three deﬁciencies may be interlinked: What is meant by imitation varies signiﬁcantly in different theoretical formulations, affecting both the possible adaptive value of having the capacity to imitate and the range of mechanisms that are sufﬁcient to account for it. Ambiguities about the species distribution of imitation, in turn, impede understanding of the phylogeny of imitation. In textbook accounts of animal behavior and animal mind, too often the topic of imitation is still consigned to a small and isolated subheading. By contrast, in developmental psychology imitation has long been considered a hallmark of intelligent function in children (Piaget, 1945/1962; Vygotsky, 1962); yet, even there, clear models for its operation are often lacking. The purpose of this chapter is to put forward a mechanistic theory that can explain the most challenging data on imitation, that is, how novel motor behavior with complex organization may be learned by observation. This is the kind of imitation that enables us to learn by apprenticeship, patiently watching a skilled practitioner. The question of whether, and to what extent, nonhumans share this ability can best be judged when the phenomenon is better understood. A secondary aim, therefore, is to articulate a process model clearly enough to aid proper interpretation of animal behavior, in the hope of avoiding the danger that “imitation” becomes a wastebasket 77

C 2002 by Academic Press Copyright All rights of reproduction in any form reserved. 0065-3454/02 $35.00

78

RICHARD W. BYRNE

category of those behaviors that cannot be understood in any other way. It is not intended that imitation of novel, complex motor behavior should be seen as the only kind of imitation, and it will, in fact, be argued that several types of imitation serve different adaptive functions and may have different evolutionary origins. If imitation has evolved in more than one functional context, it is expected that its properties and underlying mechanism will be different in each case, and conﬂating them in a single category will be unhelpful. (These words are partly offered as an apology for introducing yet more novel terminology.) It may reasonably be asked why this should be necessary: Why is there not already a range of adequate process models for imitation of novel, complex behavior that require only to be tested against evidence? At least three reasons may be offered. First, much research has been phenomenon-driven, focusing attention on isolated behaviors that are labeled “imitation” and need to be explained (or explained away). This has distracted from the overall context and adaptive functions of social learning in general and has led to reliance on contrived laboratory tasks or unusual natural circumstances as gold-standard indicators of “true imitation.” If these phenomena were indeed pure cases of imitation, this might be a good research strategy, but that is most unlikely. This is the second point: It is unlikely that it will ever prove possible to devise a demonstration of imitation uncontaminated by other social inﬂuences and ways of learning. In the untidy world that real animals inhabit, such tidy partitionings are rare: Behavior that is entirely imitated cannot be expected any more than can behavior that results entirely from nature or entirely from nurture. Instead, it is more likely that several different mechanisms contribute to the development of any single behavior. Even if one of these mechanisms is imitation, the acquisition process will also be inﬂuenced by genetic predispositions, other learning mechanisms, and other social inﬂuences. Third, clarity has not been increased by the ambiguity and vagueness of the terms employed in this arena, in particular, the word imitation itself. Even now, there is no consensus on whether, to count as imitation, behavior needs to be a close physical match to the behavior that was demonstrated, and, if so, how exact this match should be; whether it need involve new learning or merely evocation of already practiced actions; and whether or not imitation implies some form of empathy or identiﬁcation of mental state with the imitated individual (see peer commentaries to Byrne and Russon, 1998). The net result is that, although an abundance of different categories of social learning have now been deﬁned, based often on somewhat hairsplitting differences (see Whiten and Ham, 1992, for an attempted taxonomy of this plethora), less progress has been made on how imitation works, mechanistically, and why, functionally, it has evolved in some species. Numerous hypothetical mechanisms have, in fact, been proposed to avoid attributing

IMITATION OF NOVEL COMPLEX ACTIONS

79

imitation to nonhumans and to account, in a simpler way, for animal behavior suggestive of imitation. Often the existence of these mechanisms lacks independent validation, and the precise meaning of “simpler” is not often explored; there is thus heavy reliance on Lloyd Morgan’s canon, rather than on discriminatory data, when deciding between explanations (Roitblat, 1998). Since the everyday meaning of imitation is loose, this is to some extent understandable; but, if making advances in scientiﬁc understanding of social learning is regarded as a process rather like peeling off unwanted layers of husk, then the current concern must be whether the kernel will ﬁnally be palatable. As Heyes and Ray (2000) have shown, the majority of theoretical models of imitation do not ﬁt long-established data, and many models are not well-enough speciﬁed even to qualify as explanations at all. In this chapter, I ﬁrst set out what is not in serious contention: the explanation, by means of quite simple and now well-accepted learning mechanisms, of much behavior that is still called imitation in lay circumstances. Then, I use a distinction found useful in the vocal learning literature, contextual versus production learning, to resolve a controversy as to what is or is not good operational evidence of imitation. This will clear the way for a discussion of two forms of imitation, which are apparently widely divergent in adaptive function: social mimicry and method learning. A theoretical treatment is offered of how and why they might differ in mechanism. Finally, narrowing the focus to the production learning of complex, novel behavior, I discuss several different process models, emphasizing their strengths and limitations, with the aim of enabling proper categorization of animal imitation in the future. Throughout, attempts will be made to integrate imitation with wider cognitive domains and wider evolutionary functions: “imitation for action,” not for its own sake.

II. SOCIAL LEARNING WITHOUT IMITATION Before sharpening the focus toward observations that have been treated as motor imitation in the research literature, I think it is useful to recognize a baseline of agreement on a range of other phenomena to which the term is sometimes applied. It has long been understood by animal learning theorists that a general tendency to respond more vigorously (stimulus enhancement: Spence, 1937) or attend more closely (local enhancement: Thorpe, 1956, 1963) toward those parts of the environment with which a conspeciﬁc is seen to interact will usually have the beneﬁcial consequence of speeding learning in social circumstances. More rapid learning in social species makes it more likely that local traditions of dealing with the environment can develop and persist, giving rise to differences in characteristic behavior patterns between family

80

RICHARD W. BYRNE

lines or populations (“cultures”; see Roper, 1983). Popular accounts of these differences normally refer to imitation (and also teaching), but the mere existence of such differences does not on its own constitute evidence for imitation in any of the senses in which the term is used by ethologists or psychologists. Similarly, imitation is often popularly attributed when individuals are seen to do the same thing together, a coincidence of action which may be due to many things, including environmental prompting (e.g., when several birds on a bird feeder table all peck at seeds because that is their purpose in ﬂying there, not because they imitate the others). Zajonc (1965) proposed that animals and people were often seen to do things together because of a general tendency for the presence of conspeciﬁcs to increase the activity level of all of them, an effect he called social facilitation. The existence of this phenomenon has been doubted (e.g., Byrne, 1994, 1995), but it has been shown under controlled conditions in capuchin monkeys (Visalberghi and Addessi, 2000). The researchers suggest it may play a role in overcoming neophobia. However, neophobia itself presumably evolved as a protection mechanism, so the trade-off may not be simple. In different species, synchrony of behavior might be adaptive for a number of reasons (reduction of predation risk, enhanced prey capture rate, locomotor efﬁciency, etc.). According to current theoretical treatments, social facilitation, as with stimulus enhancement, may result in behavior that mimics the results of true imitation, but may be explained away be simpler means (Galef, 1988). Moreover, it is recognized that interactions with humans may inﬂuence animals in unintended ways. Where wild or feral populations are provisioned, the distribution of food items by people can potentially condition behavior patterns, giving rise to behavioral traditions that appear to be the product of learning by imitation. This has been advanced as a possible explanation of the dialects found in Japanese macaque monkeys (Green, 1975). Three populations show identical call repertoires except for their foodassociated calls, which are quite different in form in each population. The rewarding consequence of food, applied differentially toward the more noisy and conspicuous individuals, may have resulted in a learned association between vocalization and food; coincidental differences in the particular calls that attracted human attention resulted in each population’s “food calls” developing from different founding vocalizations (coos in one, screams in another, and so on). Further, as Green noted, this might also apply to the various feeding skills, which are unique to the provisioned monkeys of Koshima Islet and have often been attributed to imitation (but see Huffman, 1984, for a behavioral tradition that appears to have spread without human inﬂuence). Further doubt has been cast upon the idea that these traditions resulted from motor imitation, based on the slowness of their original spread (Galef, 1990, 1992).

IMITATION OF NOVEL COMPLEX ACTIONS

81

Rapid social learning, frequently observed co-action, and populationspeciﬁc traditions of behavior will no doubt continue to be attributed to motor imitation in popular accounts. Although demonstrating the negative is always problematic, researchers in animal behavior now routinely require stronger evidence (Heyes, 1993). In particular, the demand is for control data, to show that the opportunity for observation of a skilled model is essential, and evidence that the effect of having a model to observe is mediated by more than individual learning bolstered by general tendencies such as stimulus enhancement and social facilitation. If genetic differences and other potential confounds can be ruled out, then population-speciﬁc traditions provide control for the availability of skilled models, but innovations in artiﬁcially provisioned populations must remain suspect. Even under conditions unaffected by human intervention, such as the insect-gathering and nut-cracking traditions of tool use in wild chimpanzees, maintenance by imitation may be disputed, since even long-standing traditions might potentially be maintained by socially enhanced individual learning (Tomasello, 1990). Further, population isolation introduces concern about genetic differences in some cases. For example, all records of nut-cracking in chimpanzees are in the Western subspecies Pan troglodytes verus, so it is possible to argue that P. t. troglodytes and P. t. schweinfurthii lack some innate capacity that makes learning to use hammers peculiarly straightforward for verus. In the light of these problems, many researchers have turned to laboratory experimentation for incontrovertible evidence of imitation.

III. LEARNING WHEN AND LEARNING HOW Use of controlled experiments has shown that, even with these concerns removed, there remains a wide range of circumstances in which an animal, if allowed to see another individual performing a particular motor behavior, is likely itself to show behavior of the same form; whereas if it is denied that sight it does not do so. These cases have typically been hailed as “true imitation” in literature (e.g., Voelkl and Huber, 2000), but rejoicing may be premature. Several different evolved mechanisms may produce behavior that ﬁts the basic pattern of “A does X after seeing B do X, but not otherwise.” Lumping all cases together leads to expectations that a single explanation will serve for all, which may be quite wrong. Moreover, some mechanisms that can explain the experimental results are inadequate to account for everyday imitative learning. Instead, I propose that drawing two distinctions may aid our understanding of evolutionary function and cognitive mechanism in the different cases. The ﬁrst of these is a distinction already found useful in the literature on vocal imitation, it is that between contextual and production learning

82

RICHARD W. BYRNE

(Janik and Slater, 2000). The second distinction, separating production learning into two types according to whether the imitation is action-level or program-level, derives instead from studies of primate skill acquisition (Byrne, 1994, 1999; Byrne and Russon, 1998). Taken together, they create three categories of imitation: contextual imitation, production learning by action-level imitation, and production learning by program-level imitation. In contextual learning, there is a change in the circumstances in which an action is performed; the action itself is not new but was present in the existing repertoire of the animal. Applying this to imitation, contextual imitation would amount to learning to employ an action already in the repertoire in different circumstances, but not learning its form. Consider the progressive narrowing down during development of the range of referents to which vervet monkeys, Cercopithecus aethiops, give eagle alarm calls (Seyfarth and Cheney, 1986). Young infants give these alarms to any looming airborne object, even a falling leaf; juveniles restrict their use to a range of broad-winged soaring birds; ﬁnally, adults reserve them speciﬁcally for the martial eagle, Polemaetus bellicosus, the major predator of the monkeys. If this developmental change is to be ascribed to imitation, it is imitation of context not product; the acoustic form of the call appears to be independent of any social learning. Contextual imitation may aid problem solving as well as reliable communication: A behavior in the species-typical repertoire might be applied to a novel problem, or be used in novel circumstances, after observation of a conspeciﬁc doing the same. A range of experiments has shown that a bird, seeing a conspeciﬁc using one particular action to obtain food from an unfamiliar container, will use the same action when it is given the chance to open a similar container, whereas individuals shown another efﬁcient action acquire that technique instead (Dawson and Foss, 1965; Galef et al., 1986; Lefebvre et al., 1997; Palameta and Lefebvre, 1985; Zentall, 1996). In several species of primate, also, observing a demonstration has been shown to cause individuals to apply a particular technique to an unfamiliar problem; these include chimpanzees (Whiten et al., 1996), capuchin monkeys (Cebus apella; Custance et al., 1999) and marmosets (Callithrix jacchus; Bugnyar and Huber, 1997; Voelkl and Huber, 2000). For instance, seeing a human use a twisting motion, to remove a bolt that locks a box containing food, makes the nonhuman primate observer more likely to twist rather than pull the bolt out, and vice versa. Crucially, for both bird and primate examples, the two methods are equally effective. This two-action methodology (the label is retained even if more than two methods are tested) has become the paradigm case for experimentally demonstrating “true imitation.” Indeed, these data have been described as the best evidence for motor imitation in animals (Heyes, 1993, who explicitly deﬁnes imitation as including both this “R–S learning” (response–stimulus) and “learning of novel R”). However, if the differences between experimental groups are to be ascribed to imitation,

IMITATION OF NOVEL COMPLEX ACTIONS

83

it must be to contextual imitation—contextual learning by observation. In none of these cases is there evidence that a new action has been added to the individual’s repertoire; rather, in every case it may simply be that a familiar action is applied to an altered range of environmental stimuli. The result under many circumstances will be improved ﬁtness, since effective actions are prompted into use, but no learning of new actions necessarily occurs; contextual learning is therefore a feasible and arguably simpler explanation. Moreover, contextual learning of motor behavior can be explained at a cognitive level in a remarkably simple way (Byrne, 1994, 1995, 1998), one closely allied to stimulus enhancement. Stimulus enhancement may be viewed as priming of brain records of environmental stimuli, such that subsequent attention to these stimuli is increased and exploratory behavior is directed more toward them. In the same way, the data of the twoaction methodology, described as “true imitation,” may instead be treated as response facilitation. In this case, observation results in the priming of brain records of motor behavior, with the result that these motor actions will more likely be elicited if similar circumstances recur. A proviso is that there must be a straightforward way of recognizing that the action is “the same”: a clear overlap in perceptual features between the act as done by another and by the self, or some other way of making the link. This is not an all-or-nothing distinction. Although one’s own vocalizations, perceived through the skull as well as the air, do not sound the same as those of others, heard through air alone, the two are similar and the differences are highly predictable ones. Correspondingly, hand actions done by a conspeciﬁc and by the self do differ in visual perspective, but they retain many invariances. The difference between self- and other-view is greater for movements of the head and beak in birds, although the sounds made by pecking or tearing at environmental objects will be much the same whether self or other performs the action. For the human facial gestures imitated by newborn babies the feature overlap is presumably nil (Meltzoff and Moore, 1977). Heyes and Ray (2000) introduce the useful term perceptual opacity for this dimension of variation. In response facilitation, the correspondence between an act in the observer’s repertoire and one performed by another individual automatically triggers an increase in the likelihood of the observer performing the act (Byrne, 1994). (Thus it is not strictly a learning process at all—although, if the observer thereby gains reward, as with the experimental two-action methodology, then contextual learning would occur.) This interpretation is consistent with the discovery of mirror neurons in monkey cortex, a population of cells that respond equally to a purposive gesture whether performed by the monkey itself or by another (Gallese et al., 1996; Rizzolatti et al., 1996). These researchers believe that all goal-directed motor actions in a monkey’s current repertoire will be matched by mirror

84

RICHARD W. BYRNE

neurons, so that any action the monkey can perform will trigger an appropriate mirror cell even when it is seen done by another. Given the nontrivial differences in perspective between self and other, it may appear that the existence of mirror neurons simply shows that the matching occurs, not how it is achieved. However, Laland and Bateson (2001) have used a neural net simulation to test the idea that such matching can indeed result from past experiences, in which self and other simultaneously performed the same action, with encouraging results (note that they label the effect “imitation,” rather than response facilitation, as herein): If their conclusions are accepted, the origin of mirror neurons is not problematic. Rizzolatti (2001) suggests that mirror neurons enable a monkey to understand the dispositions of others (intentions in his terms, but there is no implication of mental state attribution) and thus to anticipate their next actions. This provides a very different explanation of the adaptive function for copying familiar acts, shown in experiments designed to reveal “true imitation,” than the usual assumption of instrumental learning. If mirror neurons underlie response facilitation, as argued by both Byrne (2001) and Rizzolatti (2001), then the tendency to copy observed motor acts is a by-product of a mechanism whose primary adaptive function is to reveal behavior’s purpose (in the sense of its immediate consequences). This would account for one apparent paradox. It would seem to be an inevitable prediction, in the case of an individual showing a general tendency to copy the actions of conspeciﬁcs, that its trial-and-error exploration should be reduced in social circumstances, thus reducing its chances of discovering novel and effective solutions. No such conservatism has been noted to be a consequence of group living. However, if the effect of observation was to reveal the purpose of an act, in many cases copying would be inappropriate. In the wild, only where the immediate consequence was a previously unattainable reward would copying the actions pay. In most tasks used by psychologists to demonstrate this sort of imitation experimentally, the end result is indeed normally a reward (food), so copying brings rewards and does make sense; thus contextual learning is seen. The limitation on this contextual learning is that the motor act must already exist in the individual’s repertoire, and it must be recognizable as similar when seen performed by another (Byrne, 1994). (In neuronal terms, this means mirror neurons for the particular action do exist which automatically trigger when self, or other, performs an action which largely resembles it.) In contrast, Heyes and Ray (2000) suggest a model in which pairings between observed and copied actions are arbitrary because they are acquired by association learning. Whereas normally the result is a link between an act as seen and the same act as done, in principle quite different links could be learned. For instance, repeated sight of one’s face with a particular expression, viewed in a mirror, may become linked to the kinesthetic experience of making that facial expression, despite minimal lack of perceptual feature

IMITATION OF NOVEL COMPLEX ACTIONS

85

overlap (see also Mitchell, 1987). Thus, with appropriate experimental manipulation, Heyes and Ray predict that it would be possible to produce an animal whose “imitations” were distinctively unusual: for instance, it might do action X when it sees action Y, and vice versa. This interesting idea has yet to be tested. Neither response facilitation nor Heyes and Ray’s related theory can account for neonatal imitation, where highly speciﬁc facial gestures of an adult are imitated by very young babies, since in this case not only is there minimal perceptual overlap in visual appearance, but the phenomenon is found long before the infant has access to mirrors which could allow the stimulus– response pairing to be learned (Meltzoff and Moore, 1977, 1983). One possibility is that matching behavior is the result of a few innate linkages between trigger-features and ﬁxed action patterns (which happen to look the same to external observers), an account favored by those who dispute the generality of the infants’ copying (Anisfeld, 1991). Alternatively, the child might be equipped at birth with a very general mechanism for making a 1 : 1 correspondence between body movements of others and the muscular contractions needed to match them in the self, an account favored by those who consider that the range of gestures that can be copied is limited only by the restricted motor repertoire of the child (Meltzoff, 1988a). This latter possibility, of course, envisages a sufﬁciently powerful mechanism to underwrite production learning even of novel actions. (Neonatal imitation is sometimes of “novel” actions, in the trivial sense that the infant has never performed them, but there is no doubt that the facial movements are in the child’s latent repertoire and would be shown at some point without any need to see a demonstration.) With contextual learning by imitation, no new motor behavior is acquired. By contrast, in production learning, the result of observation is to learn a new action. The evidence for production learning of motor behavior in animals is much sparser and more controversial, but I suggest that one reason for this has been the conﬂation of several different processes, each underwritten by a different mechanism and each probably evolved in response to different environmental demands. New distinctions need to be introduced to capture these differences. To begin, we consider the function of copying behavior from another.

IV. IMITATION: SOCIAL MIMICRY AND PICKING UP METHODS As is clear from everyday experience, imitation can serve a social function, by increasing identiﬁcation between the imitator and the imitated. Signaling may be directed at third parties, where the identiﬁcation is used to comment upon some aspect of the imitated, for instance, in the theater or in

86

RICHARD W. BYRNE

the use of imitation for playground mockery. A purely dyadic interaction is also possible, and here increasing the meshing of behavior between two individuals may serve to build up or cement their relationship (see papers in Giles, 1984). Such a function has been argued to be central to both the unconscious tendency of adults to match the body posture of their interlocutor (Bull, 1987) and the tendency of neonates to copy facial gestures of adults (Meltzoff, 1990; Meltzoff and Moore, 1994). In this social mimicry, the copying is not a matter of improving individual performance, so acquiring new actions from what is copied is not important. Instead, even if this is not a conscious process, the copy must be seen as a copy and treated as a signal. The fidelity of the match is what is crucial for such an interpretation. Social mimicry seems peculiarly important to humans, from birth onward (Meltzoff, 1996). Even at birth, human infants copy some of the mother’s expressions and facial movements, and during childhood the prominence of imitation is striking (Meltzoff, 1988b; Meltzoff and Gopnik, 1993) and clearly serves at times to allow experimentation and practice with the social roles and actions of adult life. Copying the actions of other signiﬁcant individuals, apparently for the pleasure of copying, is almost as prominent in the behavior of human-reared great apes. Neonatal imitation of facial gestures of human caretakers has been carefully documented in infant chimpanzees (Myowa, 1996; Myowa-Yamakoshi and Matsuzawa, 2000). Juvenile and adult gorillas, orangutans, and chimpanzees, after being exposed to everyday human living, copy a bewildering range of human actions, including tying shoelaces, brushing teeth with toothbrush and paste, sawing wood, lighting a ﬁre by fanning embers dipped in kerosene, washing clothes with soap, and many other examples (Hoyt, 1941; Russon, 1996; Russon and Galdikas, 1993; SavageRumbaugh, 1986; Savage-Rumbaugh and Lewin, 1994). In some cases, it may be difﬁcult to sharply distinguish copying for its own sake from copying for performance reasons (compare Russon, 1999; Tomasello, 1996), and this is also true of imitation by human children. Indeed, Meltzoff argues that there is a continuum between neonatal imitation of facial gestures and the ability of older children to learn novel manual skills and to copy intended actions rather than performance errors (Meltzoff, 1995; see also Want and Harris, 2001). This interpretation is by no means established: The extent to which children do learn novel skills by imitation, rather than by socially enhanced individual learning, has not been explored in children and is no easier to decide than in the case of animal learning. It is just as plausible that the “adult version” of neonatal imitation is ideomotor behavior (the phenomenon in which, for instance, seeing a car about to crash causes depression of the braking foot in pedestrian observers); the function throughout development may remain a purely social one, increasing identiﬁcation and empathy, at all stages of development. On such an interpretation, social mimicry has nothing to do

IMITATION OF NOVEL COMPLEX ACTIONS

87

with method learning, in either its ontogeny or its functional origin; imitation of methods is a separate phenomenon. Certainly, in most cases of social mimicry, skill repertoire is not affected, and the function appears instead to be that of inﬂuencing the social relationship or immediate social interaction between mimicker and mimicked. This creates a clear parallel with vocal production learning in birds and mammals. Bottlenose dolphins, Tursiops truncatus, imitate the individually distinctive signature whistles of others, apparently as a means of engaging their attention (Janik, 2000). Mated pairs of bou-bou shrikes, Laniarius, develop elaborate duets, but the fact that social mimicry is important in this process is evident from the fact that each member of the pair can sing the complete duet when alone (Thorpe, 1967). Indeed, one major function of expert mimicry in birds, for instance parrots and mynah birds, has been suggested to be in developing and retaining a social bond (Bertram, 1970). In both gestural and vocal cases of social mimicry, since matching to the visual or acoustic form of the observed behavior is crucial, copying the surface form rather than the underlying organization is sufﬁcient for adequate function and may be necessary to achieve high ﬁdelity. Precisely which cognitive models can account for social mimicry depends upon the novelty, perceptual opacity, and complexity of what is copied. If the mimicked action is one already in the individual’s repertoire, and there is a reasonably transparent perceptual relationship between the action as done by model and by mimic, then response facilitation by behavior priming is a sufﬁcient explanation. (As noted earlier, this may apply to most of the experimental demonstrations of “true imitation” of motor actions in primates and birds.) Neonatal imitation of facial gestures is perceptually opaque, and theatrical mimicry of adult humans and vocal imitation by dolphins and many birds are of signals that are highly complex: In these cases, a cognitively more complex process is required, and what this might be is discussed in the next section.

V. PRODUCTION LEARNING BY IMITATION To researchers in the human sciences, acutely aware of the technological gulf between humans and other animals, the possibility that humans possess an evolved suite of superior learning processes is an attractive one. Along with pedagogical teaching and apprenticeship, imitation has often been invoked as part of this repertoire. In this case, it is speciﬁcally production learning of methods that is implied: the case where an individual augments its repertoire of useful ways of dealing with the world by observation of a skilled practitioner. Instead of a painstaking, incremental process of building

88

RICHARD W. BYRNE

up a novel skill by individual experimentation, crucial aspects of the skill of another individual can be directly acquired by the imitator plagiarizing the success of another. Of course, most complex activities also require individual practice to perfect the skill, and pedagogical teaching if available may be the most efﬁcient aid; however, imitation may play a signiﬁcant part. Here, the means to acquire novel behavior is essential, and in most cases the acquired behavior is liable to be complex in organization, since it is in this circumstance that individual learning fails and imitation can give advantage. For behavior that is not complex in organization, but needs to be performed with precise timing and force, optimization by individual practice is likely to be efﬁcient, since the resulting optimum will be adjusted for the precise mechanical capabilities of the learner. (When methods are copied by imitation, especially of an adult by a child, there is always the risk that the actions will be unsuited to the size and power of the learner, raising actual impediments to progress.) To account adequately for the production learning by observation of novel and complex behavior is a considerable challenge, whether the function of the imitation is to acquire new methods of dealing with the world or for the purpose of social mimicry. Although in principle one powerful mechanism could subserve both functions, the possibility needs to be kept in mind that two or more separate evolutionary trajectories have converged on one ﬁnal outcome, the ability to copy complex, novel behavior. If so, very different processes may be involved, with contrasting strengths and weaknesses. A. EMULATION Among the ways in which an animal might acquire novel, complex behavior by observational learning, most attention has been devoted to a process that has, in fact, been set up in opposition to imitation; that is, emulation (Tomasello, 1990; Tomasello et al., 1993). Although deﬁned by the Oxford English Dictionary as “to imitate zealously,” emulation in this technical sense means to copy a result, but not necessarily the behavioral method of achieving it.1 How the result is copied is less clear. In the simplest sense (called goal emulation by Whiten and Ham, 1992), observation might simply bring a desirable result to the animal’s attention as 1

Tomasello refers the term emulation to the developmental psychologist Wood (1989) whose meaning was, however, somewhat different. Wood distinguished it as one form of imitation, in which the aim is to achieve similar effects on the world as the model, in contrast to impersonation, in which the aim is to behave “as like someone else as possible” (p. 72), and he notes that some children do this so well that they may one day become professional impersonators. Others have used emulation in yet other ways: Lloyd Morgan describes it as “a desire not only to imitate but to improve upon the copy” (Morgan, 1896, p. 173), which seems to suggest intelligent modiﬁcation of a copy.

IMITATION OF NOVEL COMPLEX ACTIONS

89

a possible goal, after which the animal would itself work toward this new goal in its own way. This may be modeled as a matter of priming brain records, in much the same way as can stimulus enhancement and response facilitation (Byrne, 1994, 1995). In the example originally used to illustrate the distinction, seeing another individual using a rake to pull in peanuts caused an observer chimpanzee to rake peanuts itself—but not to copy the precise detail of the movements (Tomasello et al., 1987). However, it is unclear in this case whether what was learned was a result, juxtaposition of rake tip and peanut, as argued by Tomasello, or a behavior, the raking action, as suggested by Byrne and Russon (1998; see also Whiten and Custance, 1996). A better example, showing the utility of the distinction, comes from vocal learning. Mynah birds copy human speech so accurately that the formant transitions of speech (pitch changes in the bands of major energy) are clearly visible on a spectrogram (Bertram, 1970). However, formants in speech are produced by resonance in the different areas of the human supralaryngeal tract which can be independently varied in size and shape, thus ﬁltering the original sound signal (Fry, 1979). Formant transitions correspond to changes in the shape of the supralaryngeal tract. Mynah birds, however, must use the avian syrinx, a very different anatomical structure, to copy the human voice. Nevertheless, they are able to produce the same sound pattern to a high degree of ﬁdelity; a similar case could be made for song thrushes copying the electronic ringing tone of some telephones (Slater, 1986). These remarkable cases suggest that, more generally, imitation in birds may be mediated by copying the ﬁnal result, not the behavioral means of achieving it. In birds which learn species-typical songs party by imitation, they will of course tend to produce these sounds by using similar muscular movements, but this may be a by-product of attempting to duplicate the end result. Indeed, in some species there is a steady improvement in match to the original model with practice, even without further opportunity to hear the original, and auditory feedback is necessary for full song to develop (Marler, 1976b): both these ﬁndings suggest that what the bird does is to modify its output toward ever closer match to a remembered original. This perfectly meets the original definition of emulation and suggests one cognitive mechanism that can generate emulation: trial-and-error modiﬁcation of behavior until it meets a criterion of match to a remembered template, itself learned from another individual. However, since the term emulation was ﬁrst applied to animal learning, its meaning has changed somewhat. In addition to goal-setting, it is now used to include observational learning about the “affordances” of objects (Tomasello, 1998). Affordance, a term borrowed from the perceptual psychology of J. J. Gibson, refers to what can be done with an object. Seeing a chimpanzee smash a nut between two stones, for instance, reveals that some nuts are breakable and that they contain edible kernels. This revelation,

90

RICHARD W. BYRNE

rather than any ability to copy the movement of hammering, may cause the observer to take up nut-cracking, whereas other chimpanzees may never discover the rich food source within hard nuts. What is learned may be cognitively complex, including in this instance the notion of containment and information about the structural form of an unfamiliar object; or, it may be no more than an association between nuts and eating a new and delicious food. In either case, the important distinction from other kinds of imitation is that motor behavior per se is not copied. Thus, in principle, emulation could be as effective if the behavior were not seen—if the nut fell by gravity, for instance, and broke as a result. This points to an obvious experimental control for emulation, but note that the spontaneous movements of objects (especially contrived gadgets in psychologists’ laboratories) may be intrinsically of less interest to animals than the actions of conspeciﬁcs, so care is needed in interpretation of negative results. (Moreover, for complex tasks, elaborate videographic editing would be needed to present the unfolding of subtle, sequential transformations of an object with no hand visible.) B. ACTION LEVEL IMITATION Where it is clear that copying rather than learning about the world has taken place and that novel behavior rather than its physical consequences has been copied a ﬁnal distinction needs to be introduced. This is the question of whether it is the surface form that is imitated, action-level imitation, or the underlying organization of behavior, program-level imitation (Byrne and Russon, 1998). In neither case are there many, well-articulated cognitive models as candidates. A close match to the surface form of observed behavior is crucial to social mimicry, but even for production learning by imitation this may be all that is required for success. How might this action-level imitation be achieved? The following two processes have been suggested, differing in the nature of what is supposed to be copied. 1. Copying a Fluid Movement Complex behavior as performed by a skilled practitioner appears as a ﬂuid whole, and any highly practiced act is literally a single continuous movement in its surface form. It has often been proposed that imitation might be an empathic process, in which the observer can directly “identify” with the movement that is seen and so copy it. The cognitive mechanism would be some form of kinesthetic-motor mapping (Mitchell, 1987). In humans, this elaborate 1 : 1 visual-kinesthetic mapping of movements has been proposed to be innately prespeciﬁed (Meltzoff, 1996), so no further explanation is needed to account for human imitative ability, that is, provided the body of

IMITATION OF NOVEL COMPLEX ACTIONS

91

the observer and model approximately match. Although this may be sufﬁcient to account for social mimicry and some forms of production imitation, such as gaining sporting prowess by bodily imitation, the lack of any analytic power limits the usefulness of such a mechanism for skill learning. 2. Copying a String of Beads If instead the ﬂuid movement is seen as a series of single, discrete actions, and each one of these actions were something that the observer could already do, then the sequence could be copied—and thus novel behaviors could be imitated. Although the underlying structure would be lost, complex behaviors can nevertheless be imitated in this way, by dissection into a string of component, simpler sequential parts that are already in the observer’s repertoire (Byrne and Russon, 1998). An associative version of this model has been developed by Heyes and Ray (2000), in which the correspondence between the seen-version and the done-version of each action is learned by association. They do not consider how the ﬂuid movement is segmented into the “string of beads,” each action of which can be copied (see Fig. 1a). Response facilitation by mirror neurons provides a straightforward explanation of how, and precisely under what circumstances, this segmentation could take place. Speciﬁcally, the segments (“beads”) will be just those sections of the model’s ﬂuid behavior which map onto actions already in the observer’s repertoire, and thus already have corresponding representation

FIG. 1. Schematic illustration of indications of underlying planning structure, visible in strings of separately recognized actions. Each action is coded by a letter: A, B, C, . . . , I. (a) “String of beads,” no organization discernible; (b) modular structure can be shown by interruptions; (c) shorter and longer pauses; (d) omissions; and (e) iteration can be signaled by repeated strings.

92

RICHARD W. BYRNE

as mirror neurons. Thus the coarseness or ﬁneness of segmentation will be automatically prescribed by the observer’s existing motor knowledge: experienced observers, whose motor repertoire already includes quite elaborate actions, chunk observed behavior into fewer segments and are confronted with a simpler sequence learning task as a result. The fact that some animals, as well as humans, have a tendency to copy sequences of familiar actions when engaged in trying to solve a novel puzzle was shown by Whiten (1998) for chimpanzees. It appears difﬁcult to decide between these two theoretical accounts, particularly in cases of social mimicry. However, note that the decision to copy one particular segment of observed behavior presupposes that some sort of segmentation has taken place, before kinesthetic-visual matching could take place. Thus, even where it appears that ﬂuid movements are copied, this may always equate to string-of-beads copying: a large number of very small 1 : 1 correspondences, between observed and performed movements, produces a smooth action in the ﬁnal output. Kinesthetic-visual matching may never be more than a convenient label for a process of action-level imitation which is essentially a matter of matching segmented units of behavior. C. PROGRAM LEVEL IMITATION For production learning of methods in general, it is ideal to acquire not only the surface form but also the underlying organization appropriate to the task. The hierarchical organization that mountain gorillas show in processing plant foods gives an illustration of this sort of complexity (see Fig. 2, and Byrne and Byrne, 1993). If behavioral organization could be discerned in the actions of a skilled practitioner, then the observer would be able to copy a whole range of alternatives: (1) one particular subprocess, ﬁlleted out; or, (2) the whole operation at a broad-brush level of description, with details ﬁlled in by individual trial and error; or, (3) the whole operation exactly as seen, including both the underlying organization and the surface form. But is it possible to “see” underlying organization in behavior? As noted by Heyes and Ray (2000), although a number of models have been put forward to explain such ability, the mental transformations involved have never been clearly speciﬁed; indeed, most of these models have a somewhat magical quality, in which the crucial elements are never speciﬁed. In contrast, Byrne (1999) proposes an entirely mechanistic model of program-level imitation that is held to be sufﬁcient to account for imitation of the underlying hierarchical structure as well as the surface form of novel, complex behavior. This model is called string parsing. The model begins from the assumption that behavior observed in another individual can be seen as made up of a sequence of simpler elements. The

IMITATION OF NOVEL COMPLEX ACTIONS

93

FIG. 2. Mountain gorilla nettle leaf processing. The technique used by gorillas is to rapidly amass a bundle of leaf blades, without the stems of petioles that host the strongest stings, and fold the bundle, so that only a single leaf underside is exposed when the parcel is popped into the mouth (Byrne and Byrne, 1993). This minimizes the number of stings that contact the palm, the ﬁngers, and, especially, the lips. The cognitive organization of the task must include at least the level of organization shown here, on the basis of current observational data. “Get leaf blades” and “fold blades” are both known to be separate modules because they are separated by the optional action, “clean.” In addition, “get leaf blades” may or may not be iterated, again implying that it is a distinct module. “Get leaf blades” itself is a routine composed of at least two modules: we know that “strip stem” and “tear off petioles” are separate modules because “strip stem” may or may not be iterated.

most straightforward way in which this might be achieved is by means of an array of units, each of which responds to an action that matches the corresponding action in the observer’s own repertoire: response facilitation. As with action-level imitation, the system of mirror neurons in premotor and associated temporal cortex of macaque monkeys (Gallese et al., 1996; Rizzolatti et al., 1996; see also Rizzolatti et al., 2000) provides a feasible neural basis for the process of segmenting ﬂuid action into a string of discrete elements, each one an action in the observer’s repertoire—effectively, “reading the letters” of action by means of response facilitation, action by action. (Alternatively, every possible visible movement in a conspeciﬁc’s musculature may be prewired to alert the motor program for a corresponding movement in the observer, as in Meltzoff’s account of neonatal imitation; or arbitrary pairings may be learned, between seen movements and performed actions, which may or may not visibly correspond, as in Heyes and Ray’s account of action-level imitation.) At some level of detail, any action must be composed of units which a conspeciﬁc has in its repertoire, so that response facilitation is sufﬁcient to allow segmentation of all actions that might be seen. As noted in the case of action-level imitation, the level of segmentation

94

RICHARD W. BYRNE

would be determined by the observer’s experience, and hence its preexisting repertorie: If existing repertoire units are already large and complex, observed behavior would be segmented into only a few chunks, and vice versa. String parsing operates on the resulting linear sequences of elements. Sequences are compared in a search for data reduction by the discovery of patterns; thus the model assumes that, in some sense, sequence orders of elements can be retained, but this may not involve explicit remembrance. Although the underlying structure of any one string is entirely ambiguous, as more and more strings are cross-correlated the persisting regularities of pattern can betray the organization. Over time and the observation of a number of these one-dimensional, linear strings of elements, an organized structure is generated, in which the links are correlational—a “statistical sketch” of the action. The regularities exist because the underlying structure, present in the brain of the individual that generated the behavior, imposes a distinctive signature on the linear sequence of elements that serves to show how the ﬂuid stream of action may appropriately be “carved at the joints” (Byrne, 1999; and see Fig. 1). In a hierarchical organization, modules occur in which the elements seem to be more tightly bound together, whereas at a junction between modules the link is weaker (Dawkins, 1976; Lashley, 1951). Interruptions occurring at these junction points allow smooth resumption once the distraction is past; in contrast, interruptions within a module will force the animal to “begin at the beginning again,” at either the beginning of the module or the entire program. Interruptability is a correlate of structure: Interruptions will thus tend to occur between, not within, modules (Fig. 1b). Even when behavior is not interrupted, relatively tighter or weaker linkage will betray itself in the pause structure of output behavior (Fig. 1c). Moreover, hierarchical organization of behavior may allow ﬂexibility, so that unnecessary stages or modules can be omitted on the basis of local circumstances (Dawkins, 1976). So, in repeated strings that are broadly similar, certain sections will occur in some strings but not in others, and the omissions signal the underlying modular structure (Fig. 1d). If we represent a modular structure as A = B—C = D—E = F, where elements joined by “ = ” are more tightly bound together than those joined by “—,” then interruptions will tend to occur more at B/C and D/E than at other points, and omissions will tend to give rise to shorter strings such as A,B,E,F as well as the full sequence A,B,C,D,E,F. Thus, analysis of sequence orders and of pause structure, including interruptions, can betray an underlying modularity of organization. Hierarchical organization allows modules to be used as subroutines in a larger organization. For instance, they may be applied iteratively until some criterion is reached (cf. the test-operate-test-exit unit of Miller et al., 1960). When control “loops” around a subroutine in this way, a distinctive pattern

IMITATION OF NOVEL COMPLEX ACTIONS

95

of sequential elements is generated: a series of repeated sequences, embedded within the main sequence (Fig. 1e). Using the same conventions as before, the sequence A,B,C,B,C,B,C,D,E signals that B = C is an iterated subroutine within the main program. Hierarchical organization allows efﬁciency, in that modules may be omitted if they are not required, on the basis of local circumstances at execution time; this is reﬂected in sequential orderings, with omissions signaling underlying structure. Finally, hierarchical organization enables a subroutine to be used in more than one program or one program to be used as a subroutine in another. Again, this can aid detection of structure in observed behavior. Once some strings have been identiﬁed as forming discrete modules or more complex structures, these familiar patterns can be picked out in as-yet-unparsed strings of elements and serve to reveal their structure. It is a truism of motor psychology, that every execution of a behavior is slightly different. The variation is necessarily minor and trivial, or else the behavior would fail in its purpose. This leads to another way of detecting underlying structure from string parsing. Because of this structure, deviations at some points will be far less innocuous than at others, and the range of variation over repeated strings will indicate which points are most constrained. The invariant features, occurring in regular positions in every string of elements leading to the same outcome, can therefore be identiﬁed as necessary ones, whereas those that are not are revealed as relatively unimportant ones. This means that, by comparing a series of strings that lead to the same outcome, the ordered sequence of necessary elements that leads to it can be identiﬁed. (Where the behavior serves to modify some environmental object, as in food processing, the recurring states will be visible as physical objects as well as behavior actions. It may be easier for observers, both human and nonhuman, to attend to objects than to body movements; see Byrne, 1999.) To an adult human observing the process, these necessary intermediate steps are probably seen as “subgoals,” achieved along the route of a successful completion of the main goal. To an organism that has no conception of goal and intentions, invariant features of sequence can nevertheless reveal their existence and sequential order, mechanistically and automatically. Thus, without any prior understanding of goals and intentions, an observer with string-parsing capability is able to extract the underlying structure from repeated observations of organized behavior (but from random actions, nothing can be extracted). The process is one of pattern-matching, and any such process is aided by extrinsic information about starts and ends of sequences. Cues to these features are often present in goal-directed behavior. Often, the goal is a consummatory activity of some sort (e.g., eating a food item), coming after a sequence of elements, and so a proper end to that sequence is indicated. In some cases, the proper start to a sequence might also

96

RICHARD W. BYRNE

be visible in behavior, if no other activity occurred immediately beforehand; thus some “complete” strings of elements corresponding to goal-directed behavior can be identiﬁed. More generally, in this way indication of the appropriate phase in which sequences need to be compared to detect any recurring patterns can be obtained. The adequacy of string parsing to detect structure has been partially tested by Furse (2001), who constructed an algorithm which operated according to some of these principles and was effective in simulating some aspects of children’s learning of arithmetical procedures, such as long division, by observation of worked examples. Moreover, Saffran et al. (1996) have shown that 8-month-old human infants are sensitive to recurring patterns in spoken strings of nonsense syllables: they learn repeated patterns quickly, with only 2 min of exposure necessary to pick up the statistical regularities. In a rather literal way, therefore, the organization of complex behavior can in principle be “perceived.” Such organization includes the starting conditions, the outcome or goal, the modular structure, any use of modules as iterated subroutines, and, in general, the statistical regularities of elements that connect start to goal. In this manner, the behavior may also be copied— at program level (Byrne, 1998). A string-parsing algorithm which is sensitive to these regularities will generate a statistical sketch which tends to mirror the actual organization of any planned, complex behavior. If each element were an action in the observer’s repartoire, as would necessarily be the case if segmentation took place by mirror neurons, the use of the statistical sketch to build a novel behavioral routine is particularly straightforward. If the output of segmentation by mirror neurons were the input to the string-parsing process, then the statistical sketch would mirror behavioral organization directly. The major limitation on string-parsing is that numerous instances are liable to be needed to provide sufﬁcient data to extract all the crucial regularities: “One shot” imitation of complex behavior is not possible, if imitation depends on string parsing. It has already been noted that there may be circumstances in which it is easier for observers to see changes to objects than the movement of a limb or other effector. String parsing could also operate in this case. If each perceived element were a state of the physical world—for example, how an object looked after some manipulation had been applied to it—then the statistical sketch would specify the organization and sequence of transformations necessary to reach the goal. What actions are needed to make each transformation would then still need to be acquired, based on prior knowledge of how to attain these subgoals, or by a process of trial-and-error learning. It would probably be unhelpful to distinguish this possibility as “program-level emulation,” however, because even in this case the organization, by which the actions are assembled into a behavioral routine, would

IMITATION OF NOVEL COMPLEX ACTIONS

97

be derived by string parsing: that is, repeated observation of a sequence which was itself generated by behavior. With videographic technology, it is possible to produce sequential object transformation in the absence of any view of the transforming bodily agent, and sequential transformation behavior on an invisible object, but the extreme artiﬁciality of these things is some indication of the irrelevance of this distinction in everyday behavior.

VI. WHAT LIMITS PRODUCTION LEARNING BY IMITATION? The emphasis of this chapter has been on mechanistic devices by which observational learning of complex and novel behavior can be achieved, in a relatively automatic and wholly nonmysterious way—in particular, copying of hierarchical organization by means of string parsing to extract statistical regularities. The associative mechanism underlying classical conditioning has been interpreted as a system evolved for extracting correlational structure from sequences of observed behavior (see Dickinson, 1980). If repeated sequences reveal underlying complex organization, when subject to correlational analysis, does this mean that almost any animal should be capable of program-level imitation? Not necessarily. It might be that the underlying correlation–extraction mechanisms do overlap considerably between classical conditioning and string parsing in many species which nevertheless differ widely in other capacities, and it is these other capacities that limit the use that they can make of observational learning. Some species, those lacking the ﬂexibility to augment an innately prescribed repertoire of actions, may be capable of contextual learning by observation; their “imitation” being restricted to changes in the circumstances and frequency of deployment of their behavioral repertoire (as indexed by the two-action methodology). Other species may be capable of learning new actions by sequential combination of existing elements in their repertoire, but yet be unable to learn novel hierarchical structure because they cannot mentally manipulate hierarchical structures. Such a limitation would be a matter of working memory capacity and so could be indexed by memory experiments, in addition to direct tests of program-level imitation. These species should show production learning by imitation, but the results would have the “string-of-beads” organization of action-level imitation, without hierarchical structure. This would be entirely adequate for the conformity needed in social mimicry but of more limited use in acquiring novel skills. Only species with the capacity to manipulate subgoals mentally into novel structures—in other words, to plan—would be capable of gaining full beneﬁt from string parsing and thus of production learning of methods by programlevel imitation. In many cases, this would be hard to distinguish from the

98

RICHARD W. BYRNE

results of action-level imitation, without careful analysis. Pause structure might betray modularity, but its absence would be inconclusive. Experimentally, interruption could be used to interrogate the underlying organization: If interruption at any point had the same effect, whether this were catastrophic or minimal, then string-of-beads organization would be indicated. Perhaps most tellingly, an individual that can imitate at program level has the facility to copy selectively from a complicated sequence of observed behavior, extracting the module or subroutine that particularly serves its current need, and integrate this into an existing structure of behavior. Thus, when we watch an expert mechanic deal with our blocked carburetor, we do not copy every mannerism of undoing bolts or maneuvering within the engine—only the method of freeing the blockage. Currently, at a time when many of the distinctions used here are rather new, it is unlikely that evidence exists to state categorically which species have these various capacities; moreover, experimental tests inevitably lag behind observational evidence. From observational data, it has been argued that all great apes, not just humans, show program-level imitation (Byrne and Russon, 1998). No experimental evidence so far supports this, but chimpanzees do apparently have quite humanlike working memory capacity (Kawai and Matsuzawa, 2000). On the hypotheses of this chapter then, they should be able to manipulate mental representations of plans and thus beneﬁt from string-parsing dissections of the structure of behavior of skilled practitioners. Many of the clearest examples of production learning by imitation come from birdsong learning. Evidence from avian imitation of human speech suggests that when birds show very exact imitation of sound strings, their copying should technically be classed as “emulation.” However, many oscine passerines acquire their song patterns by imitation: often, sounds matching a broadly speciﬁed innate template are memorized by ﬂedglings and performed and improved with practice in the ﬁrst breeding season (Marler, 1976a; Slater, 1983). Since new elements can sometimes later be inserted into these early learned structures, some hierarchical organization is implied, and indeed experimental interruption of birdsong suggests this is the case (Cynx, 1990; Riebel and Todt, 1997). Whether this implies that birdsong learning depends sometimes on program-level imitation (as has been argued by Todt, 1998) is not yet clear; if so, their behavior might be expected to show ﬂexibility in other ways, whereas the bird taxa with the greatest known ﬂexibility (parrots and crows) are not those for which evidence of hierarchical song structure is available. Several mammalian species show contextual learning by motor imitation, but so far no sign of production learning by imitation. When it is clear just which species do show program-level imitation, the crucial test of these ideas will be the corresponding measures of working memory capacity.

IMITATION OF NOVEL COMPLEX ACTIONS

99

VII. WIDER IMPLICATIONS From title onward, I have tried in this chapter to treat the various mechanisms hypothesized to underlie imitation, not as special-purpose explanations of singular phenomena in laboratory experiments or ﬁeld observations, but as parts of a functional cognitive system in which imitation is only one small part of the repertoire. Imitation is not an isolated and peculiar capacity, but one manifestation of a broader suite of capacities—and not even necessarily the evolutionary driving force that enabled them to evolve. Taking this broader perspective, where does production learning by program-level imitation “ﬁt in”? Program-level imitation is built upon a more primitive ability to identify discrete units in ﬂuid streams of observed behavior. If this primitive ability is accepted as a matter of response facilitation by mirror neurons, then its evolutionary function must be something within the capacity of rhesus monkeys—which does not on present knowledge include program-level imitation (Visalberghi and Fragaszy, 1990). Rizzolatti (2001) has instead suggested that mirror neurons act to pick out “intentions” in the goal-directed action of conspeciﬁcs, by which evidently he means dispositions toward certain future actions rather than the embedded attribution of second-order and higher intentionality (Dennett, 1983). Monkeys show extensive ability to predict and react appropriately to the actions of their social companions (Cheney and Seyfarth, 1990; Dunbar, 1988), so this hypothesis makes good sense. That monkeys and birds are slightly more likely to perform an action which is observed demonstrated several times, as shown in the two-action methodology of testing for imitation (Bugnyar and Huber, 1997; Dawson and Foss, 1965; Whiten et al., 1996; Zentall, 1996), would thus be an epiphenomenon of a mechanism whose primary adaptive function is for social categorization. String parsing, acting on the products of this evolutionarily more ancient mechanism, enables organizational structure to be detected, and hence copied, by any species with the ability to construct hierarchically embedded programs of motor action (Byrne, 2001). Those abilities are part of a much wider constellation of general planning and problem-solving mechanisms, which have yet to be studied in any detail in animals. (In humans, they have largely been studied in the context of verbal reports or quasi-mathematical tasks: Byrne, 1983; Newell and Simon, 1972.) But string parsing itself may not simply be part of an imitation mechanism, since what it delivers is of far more general utility. Just as mirror neurons enable “perception” of disposition and immediately likely future actions, so string parsing enables perception of hierarchical programs lying behind output behavior. This is a crucial part of discerning the plans of other individuals, if those plans can be imagined.

100

RICHARD W. BYRNE

Discerning the plans of others and copying their skills are both plausible functional origins of string parsing; currently, it would be premature to judge their relative plausibility. On this interpretation, string parsing paves the way for identiﬁcation of the intentions of other individuals, including the level of intentionality that is sometimes called “theory of mind.” It may therefore be that all species that show program-level imitation can also represent the intentions of others and thereby compute new actions based on these unobservable constructs; this possibility remains a target for future research. VIII. SUMMARY Underlying the various behaviors that are classiﬁed as imitation, there may be several distinct mechanisms, differing in adaptive function, cognitive basis, and computational power. Experiments reporting “true motor imitation” in animals do not as yet give evidence of production learning by imitation; instead, contextual imitation can explain their data, and this can be explained by a simple mechanism (response facilitation) which matches known neural ﬁndings. When imitation serves a function in social mimicry, which applies to a wide range of phenomena from neonatal imitation in humans and great apes to pair-bonding in some bird species, the ﬁdelity of the behavioral match is crucial. Learning of novel behavior can potentially be achieved by matching the outcome of a model’s action, and it is argued that vocal imitation by birds is a clear example of this method (which is sometimes called emulation). Alternatively, the behavior itself may be perceived in terms of actions that the observer can perform, and thus it may be copied. If the imitation is linear and stringlike (action level), following the surface form rather than the underlying plan, then its utility for learning new instrumental methods is limited. However, the underlying plan of hierarchically organized behavior is visible in output behavior, in subtle but detectable ways, and imitation could instead be based on this organization (program level), extracted automatically by string parsing. Currently, the most likely candidates for such capacities are all great apes. It is argued that this ability to perceive the underlying plan of action, in addition to allowing highly ﬂexible imitation of novel instrumental methods, may have resulted in the competence to understand the intentions (theory of mind) of others.

Acknowledgments The development of the ideas within this chapter has been a cumulative process over many years and has beneﬁted from stimulating discussions with too many people to list accurately now,

IMITATION OF NOVEL COMPLEX ACTIONS

101

but I particularly mention Anne Russon, whose ability to keep on at vague and weak concepts until clarity emerges has so often helped. For taking the time to read and comment upon the long ﬁrst draft in these busy times, I especially thank Jeff Galef, Vincent Janik, Giacomo Rizzolatti, Tim Roper, Peter Slater, and Charles T. Snowdon. Their encouragement and criticism have been a great stimulus, and I can only apologize if I have not always managed to rise to their challenges.

References Anisfeld, M. (1991). Neonatal imitation. Dev. Rev. 11, 60–97. Bertram, B. (1970). The vocal repertoire of the Indian hill mynah, Gracula religiosa. Anim. Behav. Monogr. 3, 79–192. Bugnyar, T., and Huber, L. (1997). Push or pull: An experimental study of imitation in marmosets. Anim. Behav. 54, 817–831. Bull, P. E. (1987). “Posture and Gesture.” Pergamon Press, Oxford. Byrne, R. W. (1983). Protocol analysis in problem-solving. In “Thinking and Reasoning: Psychological Approaches” (J. S. B. T. Evans, Ed.), pp. 227–249. Routledge and Kegan Paul, London. Byrne, R. W. (1994). The evolution of intelligence. In “Behaviour and Evolution” (P. J. B. Slater and T. R. Halliday, Eds.), pp. 223–265. Cambridge Univ. Press, Cambridge, UK. Byrne, R. W. (1995). “The Thinking Ape: Evolutionary Origins of Intelligence.” Oxford Univ. Press, Oxford. Byrne, R. W. (1998). Imitation: The contributions of priming and program-level copying. In “Intersubjective Communication and Emotion in Early Ontogeny” (S. Braten, Ed.), pp. 228–244. Cambridge Univ. Press, Cambridge, UK. Byrne, R. W. (1999). Imitation without intentionality. Using string parsing to copy the organization of behaviour. Anim. Cogn. 2, 63–72. Byrne, R. W. (2001). Seeing actions as hierarchically organized structures. Great ape manual skills. In “The Imitative Mind: Development, Evolution, and Brain Bases” (A. Meltzoff and W. Prinz, Eds.). Cambridge Univ. Press, Cambridge, UK. Byrne, R. W., and Byrne, J. M. E. (1993). Complex leaf-gathering skills of mountain gorillas (Gorilla g. beringei): Variability and standardization. Am. J. Primatol. 31, 241–261. Byrne, R. W., and Russon, A. E. (1998). Learning by imitation: A hierarchical approach. Behav. Brain Sci. 21, 667–721. Cheney, D. L., and Seyfarth, R. M. (1990). “How Monkeys See the World: Inside the Mind of Another Species.” Univ. Chicago Press, Chicago. Custance, D., Whiten, A., and Fredman, T. (1999). Social learning of an artiﬁcial fruit task in capuchin monkeys (Cebus apella). J. Comp. Psychol. 113(1), 13–23. Cynx, J. (1990). Experimental determination of a unit of song production in the zebra ﬁnch (Taeniopygia guttata). J. Comp. Psychol. 104, 3–10. Dawkins, R. (1976). Hierarchical organisation: A candidate principle for ethology. In “Growing Points in Ethology” (P. P. G. Bateson and R. A. Hinde, Eds.), pp. 7–54. Cambridge Univ. Press, Cambridge, UK. Dawson, B. V., and Foss, B. M. (1965). Observational learning in budgerigars. Anim. Behav. 13, 470–474. Dennett, D. C. (1983). Intentional systems in cognitive ethology: The “Panglossian paradigm” defended. Behav. Brain Sci. 6, 343–390. Dickinson, A. (1980). “Contemporary Animal Learning Theory.” Cambridge Univ. Press, Cambridge, UK.

102

RICHARD W. BYRNE

Dunbar, R. I. M. (1988). “Primate Social Systems.” Croom Helm, London. Fry, D. B. (1979). “The Physics of Speech.” Cambridge Univ. Press, Cambridge, UK. Furse, E. (2001). A model of imitation learning of algorithms from worked examples. Cybernet. Syst. 32, 121–154. Galef, B. G. (1988). Imitation in animals: History, deﬁnitions, and interpretation of data from the psychological laboratory. In “Social Learning: Psychological and Biological Perspectives” (T. Zentall and B. G. Galef, Jr., Eds.), pp. 3–28. Erlbaum, Hillsdale, NJ. Galef, B. G. (1990). Tradition in animals: Field observations and laboratory analyses. In “Interpretations and Explanations in the Study of Behaviour: Comparative Perspectives” (M. Bekoff and D. Jamieson, Eds.), pp. 74–95. Westview Press, Boulder, CO. Galef, B. G. (1992). The question of animal culture. Hum. Nat. 3, 157–178. Galef, B. G., Manzig, L. A., and Field, R. M. (1986). Imitation learning in budgerigars: Dawson and Foss (1965) revisited. Behav. Proc. 13, 191–202. Gallese, V., Fadiga, L., Fogassi, L., and Rizzolatti, G. (1996). Action recognition in the premotor cortex. Brain 119, 593–609. Giles, H. (Ed.) (1984). “The Dynamics of Speech Accommodation.” Mouton, Berlin. Green, S. (1975). Dialects in Japanese monkeys: Vocal learning and cultural transmission of locale-speciﬁc vocal behaviour? Z. Tierpsychol. 38, 304–314. Heyes, C. M. (1993). Imitation, culture, and cognition. Anim. Behav. 46, 999–1010. Heyes, C. M., and Ray, E. D. (2000). What is the signiﬁcance of imitation in animals? Adv. Study Behav. 29, 215–245. Hoyt, A. M. (1941). “Toto and I: A Gorilla in the Family.” Lippincott, Philadelphia. Huffman, M. A. (1984). Stone play of Macaca fasciata in Arashiyama B troop: Transmission of a non-adaptive behavior. J. Hum. Evol. 13, 725–735. Janik, V. M. (2000). Whistle matching in wild bottlenose dolphins (Tursiops truncatus). Science 289, 1355–1357. Janik, V. M., and Slater, P. J. B. (2000). The different roles of social learning in vocal communication. Anim. Behav. 60, 1–11. Kawai, N., and Matsuzawa, T. (2000, 6 January 2000). Numerical memory span in a chimpanzee. Nature 403, 39–40. Laland, K. N., and Bateson, P. (2001). The mechanisms of imitation. Cybernet. Syst. 32, 195–224. Lashley, K. S. (1951). The problem of serial order in behaviour. In “Cerebral Mechanisms in Behaviour: The Hixon Symposium” (L. A. Jeffress, Ed.), pp. 112–136. Wiley, New York. Lefebvre, L., Templeton, J., Brown, K., and Koelle, M. (1997). Carib grackles imitate conspeciﬁc and zenaida dove tutors. Behaviour 134, 1003–1017. Marler, P. (1976a). An ethological theory of the origin of vocal learning. Ann. N. Y. Acad. Sci. 280, 386–395. Marler, P. (1976b). Sensory templates in species-speciﬁc behaviour. In “Simpler Networks and Behaviour” (J. C. Fentress, Ed.), pp. 314–329. Sinauer, Sunderland, MA. Meltzoff, A. (1996). The human infant as imitative generalist: A 20-year progress report on infant imitation with implications for comparative psychology. In “Social Learning in Animals: The Roots of Culture” (B. G. Galef and C. M. Heyes, Eds.), pp. 347–370. Academic Press, New York. Meltzoff, A. N. (1988a). The human infant as Homo imitans. In “Social Learning: Psychological and Biological Perspectives” (T. Zentall and B. Galef, Eds.), pp. 319–341. Erlbaum, Hillsdale, NY. Meltzoff, A. N. (1988b). Infant imitation and memory: nine-month-olds in immediate and deferred tests. Child Dev. 59, 217–225. Meltzoff, A. N. (1990). Towards a developmental cognitive science: The implications of crossmodal matching and imitation for the development of representation and memory in infancy. Ann. N. Y. Acad. Sci. 608, 1–37.

IMITATION OF NOVEL COMPLEX ACTIONS

103

Meltzoff, A. N. (1995). Understanding the intentions of others: Re-enactment of intended acts by 18-month-old children. Dev. Psychol. 31(5), 838–850. Meltzoff, A. N., and Gopnik, A. (1993). The role of imitation in understanding persons and developing a theory of mind. In “Understanding other Minds: Perspectives from Autism” (S. Baron-Cohen, H.Tager-Flusberg and D. J. Cohen, Eds.), pp. 335–366. Oxford Univ. Press, Oxford. Meltzoff, A. N., and Moore, M. K. (1977). Imitation of facial and manual gestures by human neonates. Science 198, 75–78. Meltzoff, A. N., and Moore, M. K. (1983). Newborn infants imitate adult facial gestures. Child Dev. 54, 702–709. Meltzoff, A. N., and Moore, M. K. (1994). Imitation, memory, and the representation of persons. Infant Behav. Dev. 17, 83–99. Miller, G. A., Galanter, E., and Pribram, K. (1960). “Plans and the Structure of Behavior.” Holt, Rinehart, and Winston, New York. Mitchell, R. W. (1987). A comparative developmental approach to understanding imitation. Perspect. Ethol. 7, 183–215. Morgan, C. L. (1896). “Habit and Instinct.” Edward Arnold, London. Myowa, M. (1996). Imitation of facial gestures by an infant chimpanzee. Primates 37,207–213. Myowa-Yamakoshi, M., and Matsuzawa, T. (2000). Imitation of Intentional Manipulatory Actions in Chimpanzees (Pan troglodytes). J. Comp. Psychol. 114, 381–391. Newell, A., and Simon, H. A. (1972). “Human Problem Solving.” Prentice-Hall, New York. Palameta, B., and Lefebvre, L. (1985). The social transmission of a food technique in pigeons: What is learned. Anim. Behav. 33, 892–896. Piaget, J. (1945/1962). “Play, Dreams, and Imitation in Childhood.” Norton, New York. Riebel, K., and Todt, D. (1997). Light ﬂash stimulation alters the nightingale’s singing style: Implications for song control mechanisms. Behaviour 134, 789–808. Rizzolatti, G. (2001). Cortical mechanisms for recognizing actions made by others. In “The Imitative Mind: Development, Evolution, and Brain Bases” (A. Meltzoff and W. Prinz, Eds.), Cambridge Univ. Press, Cambridge, UK. Rizzolatti, G., Fadiga, L., Fogassi, L., and Gallese, V. (1996). Premotor cortex and the recognition of motor actions. Brain Res. 3, 131–141. Rizzolatti, G., Fogassi, L., and Gallese, V. (2000). Cortical mechanisms subserving object grasping and action recognition: A new view on the cortical motor functions. In “The New Cognitive Neurosciences,” 2nd ed. (M. S. Gazanniga, Ed.), pp. 539–552. MIT Press, Cambridge, MA. Roitblat, H. L. (1998). Mechanisms of imitation: The relabeled story. Behav. Brain Sci. 21, 701–702. Roper, T. J. (1983). Learning as a biological phenomenon. In “Animal Behaviour, Volume 3: Genes, Development and Learning” (T. R. Halliday and P. J. B. Slater, Eds.), pp. 178–212. Blackwell Scientiﬁc, Oxford. Russon, A. E. (1996). Imitation in everyday use: Matching and rehearsal in the spontaneous imitation of rehabilitant orangutans (Pongo pygmaeus). In “Reaching into Thought: The Minds of the Great Apes” (A. E. Russon, K. A. Bard, and S. T. Parker, Eds.), pp. 152–176. Cambridge Univ. Press, Cambridge, UK. Russon, A. E. (1999). Orangutans’ imitation of tool use: A cognitive interpretation. In “The Mentalities of Gorillas and Orangutans” (S. T. Parker, H. L. Miles, and R. M. Mitchell, Eds.), pp. 117–146. Cambridge Univ. Press, Cambridge, UK. Russon, A. E., and Galdikas, B. M. F. (1993). Imitation in free-ranging rehabilitant orangutans. J. Comp. Psychol. 107, 147–161. Saffran, J. R., Aslin, R. N., and Newport, E. L. (1996). Statistical learning by 8-month-old infants. Science 274, 1926–1928.

104

RICHARD W. BYRNE

Savage-Rumbaugh, E. S. (1986). “Ape Language: From Conditional Response to Symbol.” Columbia Univ. Press, New York. Savage-Rumbaugh, S., and Lewin, R. (1994). “Kanzi. The Ape at the Brink of the Human Mind.” Wiley, New York. Seyfarth, R. M., and Cheney, D. L. (1986). Vocal development in vervet monkeys. Anim. Behav. 34, 1640–1658. Slater, P. J. B. (1983). The development of individual behaviour. In “Animal Behaviour. Vol 3: Genes, Development and Learning” (T. R. Halliday and P. J. B. Slater, Eds.), pp. 52–81. Blackwell Scientiﬁc, Oxford. Slater, P. J. B. (1986). The cultural transmission of bird song. Trends Ecol. Evol. 1, 94–97. Spence, K. W. (1937). Experimental studies of learning and higher mental processes in infrahuman primates. Psychol. Bull. 34, 806–850. Thorpe, W. H. (1956). “Learning and Instinct in Animals.” Methuen, London. Thorpe, W. H. (1963). “Learning and Instinct in Animals,” 2nd ed. Methuen, London. Thorpe, W. H. (1967). Vocal imitation and antiphonal song and its implications. In “Proceedings of the 14th International Ornithological Congress,” pp. 245–263. Todt, D. (1998). Hierarchical learning of song in birds: A case of vocal imitation? Behav. Brain Sci. 21, 702–703. Tomasello, M. (1990). Cultural transmission in the tool use and communicatory signaling of chimpanzees? In “ ‘Language’ and Intelligence in Monkeys and Apes” (S. T. Parker and K. R. Gibson, Eds.), pp. 274–311. Cambridge Univ. Press, Cambridge, UK. Tomasello, M. (1996). Do apes ape? In “Social Learning in Animals: The Roots of Culture” (C. M. Heyes and B. G. Galef, Eds.), pp. 319–346. Academic Press, San Diego. Tomasello, M. (1998). Emulation learning and cultural learning. Behav. Brain Sci. 21, 703–704. Tomasello, M., Davis-Dasilva, M., Camak, L., and Bard, K. (1987). Observational learning of tool-use by young chimpanzees. Hum. Evol. 2, 175–183. Tomasello, M., Kruger, A. C., and Ratner, H. H. (1993). Cultural learning. Behav. Brain Sci. 16, 495–552. Visalberghi, E., and Addessi, E. (2000). Seeing group members eating a familiar food enhances the acceptance of novel foods in capuchin monkeys. Anim. Behav. 60, 69–76. Visalberghi, E., and Fragaszy, D. M. (1990). Do monkeys ape? In “‘Language’ and Intelligence in Monkeys and Apes” (S. T. Parker and K. R. Gibson, Eds.), pp. 247–273. Cambridge Univ. Press, Cambridge, UK. Voelkl, B., and Huber, L. (2000). True imitation in marmosets. Anim. Behav. 60, 195–202. Vygotsky, L. (1962). “Thought and Language.” MIT Press, Cambridge, MA. Want, S. C., and Harris, P. L. (2001). Learning from other people’s mistakes: Casual understanding in learning to use a tool. Child Dev. 72, 431–443. Whiten, A. (1998). Imitation of the sequential structure of actions by chimpanzees (Pan troglodytes). J. Comp. Psychol. 112, 270–281. Whiten, A., and Custance, D. (1996). Studies of imitation in chimpanzees and children. In “Social Learning in Animals: The Roots of Culture” (C. M. Heyes and B. G. Galef, Eds.), pp. 291–318. Academic Press, San Diego. Whiten, A., and Ham, R. (1992). On the nature and evolution of imitation in the animal kingdom: Reappraisal of a century of research. In “Advances in the Study of Behavior” (P. J. B. Slater, J. S. Rosenblatt, C. Beer, and M. Milinski, Eds.), pp. 239–283. Academic Press, San Diego. Whiten, A., Custance, D. M., Gomez, J.-C., Teixidor, P., and Bard, K. A. (1996). Imitative learning of artiﬁcial fruit processing in children (Homo sapiens) and chimpanzees (Pan troglodytes). J. Comp. Psychol. 110, 3–14.

IMITATION OF NOVEL COMPLEX ACTIONS

105

Wood, D. (1989). Social interaction as tutoring. In “Interaction in Human Development” (M. H. Bornstein and J. S. Bruner, Eds.), pp. 59–80. Erlbaum, Hillsdale, NJ. Zajonc, R. B. (1965). Social facilitation. Science 149, 269–274. Zentall, T. R. (1996). An analysis of imitative learning in animals. In “Social Learning in Animals: The Roots of Culture” (C. M. Heyes and B. G. Galef, Eds.), pp. 221–243. Academic Press, San Diego. Zentall, T. R. (2001). Imitation in animals: Evidence, function, and mechanisms. Cybernet. Syst. 32, 53–96.

This Page Intentionally Left Blank

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 31

Lateralization in Vertebrates: Its Early Evolution, General Pattern, and Development LESLEY J. ROGERS DIVISION OF ZOOLOGY SCHOOL OF BIOLOGICAL SCIENCES UNIVERSITY OF NEW ENGLAND ARMIDALE, NEW SOUTH WALES 2351, AUSTRALIA

I. INTRODUCTION Over the past two or three decades, a large number of examples of left– right differences in brain functioning and responding have been reported for a range of vertebrate species. This has provided convincing evidence against the earlier held notion of lateralization as a unique characteristic of the human species that is associated with tool use, language, and consciousness. It is now possible to see that there is a basic pattern of lateralization common to all vertebrates, including humans. This chapter is concerned with asymmetry of the behavioral responses, in some cases dependent on structural asymmetry either of the central nervous system or of physical features but, mostly, not associated with any known structural asymmetry. It may be manifested as left- or right-side differences in responding or in differential responsiveness to various forms of sensory stimulation according to whether those stimuli are perceived by receptors on the left or right side of the animal’s midline. The evidence reported in this chapter shows that there is a common basic pattern of lateralization among vertebrates, indicating that the lateralization of birds and mammals is homologous, that is, it is inherited from their common ancestor. The possible advantages and disadvantages of being lateralized are considered.

A. INDIVIDUAL AND POPULATION LATERALIZATION There are two distinct kinds of lateralization. The ﬁrst is present in the majority of individuals in a population (or species), but there is no lateral bias 107

C 2002 by Academic Press Copyright All rights of reproduction in any form reserved. 0065-3454/02 $35.00

108

LESLEY J. ROGERS

FIG. 1. Frequency plots to show distributions characteristic of individual lateralization, not present at the population level (A and B), and population lateralization in which the majority of individuals are lateralized in the same direction (C and D). The lateralization index is plotted on the x-axis (L − R/L + R).

in the population as a whole. Paw preference in rodents (some strains but not all; Waters and Denenberg, 1994) is the best-known example of this form of lateralization: Half of the individuals prefer to use the left paw to reach into a tube to retrieve food and the other half prefer to use the right paw (Bulman-Flemming et al., 1997; Collins, 1985). Thus, a frequency histogram of right minus left values (or of the laterality index, right minus left divided by right plus left) is bimodal with peaks on either side of the no-bias value (Fig. 1A and 1B). The majority of individuals have a paw preference but the population has no handedness (“pawedness”). The second kind of lateralization is present in the population (or species) as well as in individuals. The population bias results from the fact that most individuals are lateralized in the same direction, as in the case of righthandedness in humans. A frequency histogram of the lateralization index in this case has an obvious skew to the right side of the no-bias point (Fig. 1C). A similar rightward skewed bias at the population level is seen in the data for

LATERALIZATION IN VERTEBRATES

109

righting responses of the frog Litoria latopalmata (discussed later; Fig. 4). Other such examples of population lateralization are skewed leftward of the no-bias point (Fig. 1D), as is known to be the case for foot preference in holding food in several species of parrots (Harris, 1989; Rogers, 1980) and for hand preferences in orangutans (Pongo pygmaeus) when they manipulate parts of their face using ﬁnger ﬂexion and ﬁne movements of the ﬁngers (Rogers and Kaplan, 1996). Such left-handedness has also been found in prosimians when they reach out to grasp food (Ward et al., 1990). The ﬁrst type of lateralization (present in individuals but not in the population) is referred to as “individual lateralization” to distinguish it from “population lateralization.” The term “antisymmetry” is also used to refer to asymmetry at the individual level occurring with no consistent side bias (i.e., with no lateralization at the population level) (Møller and Swaddle, 1997). It should be noted that antisymmetry (Fig. 1B) differs from ﬂuctuating asymmetry, of interest to evolutionary biologists (Swaddle, 1999), in that all or most of the individuals have some degree of asymmetry. Fluctuating asymmetry refers to populations in which lateralization occurs in only a small percentage of the population and with random bias to the left or right side. It is, therefore, characterized by a bell-shaped curve (see Palmer, 1996, for further explanation). In populations with ﬂuctuating asymmetry, symmetry in the population is maintained despite the vicissitudes of development (e.g., of lateralized disturbances caused by chance injury or parasite attack) that lead to some individuals, but not the majority, being lateralized in one direction and others in the other direction. “Individual lateralization” does not refer to this kind of population but instead to antisymmetry (e.g., asymmetry of claw size in ﬁddler crabs; Neville, 1976). Nevertheless, the bimodal distribution of hand preferences in mice has been considered in terms of chance effects occurring during the course of development, leading to lateralization at the individual but not at the population level (Collins et al., 1993). Provided such inﬂuences were common enough to affect the majority of the individuals in the population, this could be a possible explanation for the population distribution of paw preferences, but, usually, the chance effects of lateralized disturbance of development are less severe and affect only a minority of the population. Therefore, although ﬂuctuating asymmetry resulting from chance inﬂuences on development may apply to some forms of functional lateralization, it seems rather unlikely to apply to paw preference in mice and rats because one might expect to ﬁnd many individuals that have escaped the chance effects of lateralized disturbance of development and so show no preference. This is not the case; there are very few rats and mice that are ambipreferent (Fig. 1A). Paw preference in rodents is, therefore, an extreme example of antisymmetry with a substantial number of individuals showing zero bias.

110

LESLEY J. ROGERS

Research conducted over the years has provided us with a large number of examples of behavioral lateralization at the population level (i.e., directional asymmetry; Palmer, 1996). Contrary to the initial impression made, primarily, by earlier work showing bimodal distributions of paw preferences and turning biases in rats, it now seems that directional bias of populations is the more common type of lateralization in brain function and behavior. For example, it had been thought that primates acquired individual hand preferences through practice (Walker, 1980), until closer examination of the evidence for hand preferences in primates revealed that population biases were present in a number of species (McNeilage et al., 1987). It remains possible, of course, that individual preferences could be enhanced by practice effects in the case of some lateralized characteristics, on some tasks and in some species. Even in species with bimodally distributed motor biases (as for paw preferences in rodents), there may be clear population lateralization for attention, sensory processing, and emotional responses. Rats provide a clear example of this (Bianki, 1988; Denenberg, 1981). Despite the absence of a population bias for paw preference to retrieve food, rats are lateralized at a population level for emotional responding, spatial ability, and a number of other cortical functions (discussed later). Many of these population biases ﬁt a general model applicable across vertebrate species. Examples of population lateralization in different species will be discussed later, but, before this, it should be helpful to give a general framework into which most, if not all, of the examples may be ﬁtted (also outlined in Andrew and Rogers, 2001). B. GENERAL MODEL OF HEMISPHERIC SPECIALIZATION It is hypothesized that the left hemisphere (Lhemi) is preferentially involved in the control of responses that require weighing out alternative options before responding. To do this it must be able to inhibit responding until the appropriate decision has been made. This includes inhibiting responses to inappropriate stimuli (e.g., objects that are inedible) and directing responses to appropriate stimuli (e.g., food objects). The Lhemi appears to achieve this outcome, in part by its ability to inhibit the right hemisphere (Rhemi). Also, as the data discussed later indicate, use of the Lhemi causes the animal to focus its attention on local features of the stimulus and to analyze information sequentially. The hypothesized general model assigns an entirely different set of functions to the Rhemi. Use of the Rhemi appears to be characterized by diffuse (or global) attention and control of responses that are spontaneous, given without inhibition and without weighing out alternative possibilities.

LATERALIZATION IN VERTEBRATES

111

In other words, the Rhemi controls intense responses to stimuli that release species-typical behavior. Thus, it seems, the Rhemi is used in the expression of intense emotions. It also attends to spatial cues that can be used to construct a topographical map of the animal’s surrounding environment. In fact, the constellation of Rhemi functions is, apparently, designed for escape or attack responses. Intense expression of fear, under control of the Rhemi, would be expected to lead to immediate escape responses, or attack responses if escape is not an alternative, and use of the Rhemi enables processing of topographical information to guide locomotion along known escape routes. This would occur in the absence of inhibition by the Lhemi. These complementary hemispheric specializations are characteristic of humans (Hellige, 1993) as well as other vertebrates. As the following examples show, the hemispheric specializations are expressed in some animal species as side biases, or as eye and ear preferences. The various ways of revealing lateralization include unilateral insult of the hemispheres, monocular or monaural testing, head turning to achieve optimal sensory inputs for the task at hand, and measurement of motor biases. The following sections attempt to discuss lateralization from an ecological perspective, ﬁrst by considering responses essential for survival and, where possible, by considering the evidence for lateralization in species-typical behavior performed in the natural environment. Second, the probable advantages and disadvantages of being lateralized are discussed. As avian species, and especially the domestic chick, have been the most studied in terms of lateralization, birds are considered ﬁrst under each heading. This approach allows discussion of the detailed pattern of lateralization before extending the discussion to other vertebrate species. Although, in the following sections, lateralization may often be discussed as if it were an absolute phenomenon, this is merely to facilitate expression, and the reader is reminded that all of the examples are probabilistic and refer to a signiﬁcant population bias away from symmetry.

II. FEEDING AND PREY CAPTURE Feeding requires considered responses to be made after recognition processes have been completed and alternative responses have been assessed; for example, grain has to be discriminated from similar inedible objects and living prey has to be identiﬁed and discriminated from toxic species with characteristics similar to the prey. Food items must also be discriminated from background. Several examples show that feeding responses are controlled by the Lhemi, and, in some species and conditions, they are directed preferentially to the animal’s right side.

112

LESLEY J. ROGERS

A. BIRDS Specialization of the Lhemi for control of feeding responses was ﬁrst demonstrated in the young, domestic chick. The original method was to administer cycloheximide to the left or right hemisphere and examine the long-term effect of this treatment on the chick’s ability to peck at grain and avoid pecking at the pebbles forming a background among which the grains had been scattered (Rogers and Anson, 1979). Treatment of the left hemisphere impaired the chick’s ability to perform this task, whereas treatment of the right hemisphere had no effect. The same effect was caused by unihemispheric treatment with a low dose of glutamate (Howard et al., 1980), and localized injections of very small volumes and concentrations of glutamate in various regions of the forebrain have revealed that the effect of the treatment is located in the visual Wulst region of the left hemisphere (Deng and Rogers, 1997). This is the region of the brain receiving the thalamofugal visual projections, as opposed to the other visual region, the ectostriatal region of the hemispheres, which receives input from the optic tectum (Fig. 2). Apparently, both cycloheximide and glutamate treatment renders the left Wulst dysfunctional, not merely by lesioning it but by causing proliferation of inappropriate neural connections (for details on the mechanism involved see Khyentse and Rogers, 1997; Rogers, 1993, 1996). It appears that the right Wulst is forced to assume control, and so the chick is unable to inhibit pecking at pebbles. A large number of such experiments have demonstrated a clear role of the Lhemi in controlling both pecking at grain and avoiding pebbles. To perform the task, the chick has to inhibit any spontaneous pecks that might be directed at pebbles, and so make a considered decision each time before a peck is delivered. The difﬁculty of the task is important, therefore, because a key feature of the task is that the grains and pebbles are hard to distinguish, differing only in texture and not shape, size, or color. Although a chick pecks rapidly in the task, it would appear that each peck requires decision making. In fact, the chicks are often seen to inhibit a peck directed at a pebble just before their beak hits its target. This type of response control is a feature of the Lhemi and not the Rhemi. Were no such decisions necessary on each peck, the chick might well use its Rhemi (see later). The same lateralization of feeding responses in the same task has also been revealed by monocular testing, and this is possible because the optic nerve ﬁbers decussate completely in birds and, therefore, input from one eye goes, at ﬁrst, entirely to the opposite (contralateral) side of the brain (thalamus or optic tectum). Figure 2 summarizes the visual inputs to the chick forebrain hemispheres. In the tectofugal pathway, the rapidly conducting (myelinated) projections send input exclusively to the hemisphere contralateral to the eye

LATERALIZATION IN VERTEBRATES

113

FIG. 2. The two visual pathways of the chick are presented diagrammatically to give an impression of the projection of information to each hemisphere of the forebrain. The black symbols indicate a strong projection: note that each eye sends this information to its contralateral hemisphere. The gray symbols indicate a lesser projection (less detailed information), going from each eye to its ipsilateral hemisphere. The dotted cross in A (left hemisphere) indicates an even less well-developed projection from the left eye (LE) to its contralateral hemisphere (i.e., showing the asymmetry in the thalamofugal visual projections). LE and RE refer to the left and right eyes. The thickness of the lines with arrows gives some indication of the number of projections.

114

LESLEY J. ROGERS

providing those inputs. Then, after a short delay, inputs from that eye are sent to both hemispheres via slowly conducting (unmyelinated) projections. This secondary bilateral input may be the mechanism allowing the animal to shift its attention (and turn its head) to the other hemisphere should the task demand it (Deng and Rogers, 1998a). As Kinsbourne (1975) hypothesized for humans, the nature of the task may determine which hemisphere processes the information and controls responses. It is noted from Fig. 2 that, although a “sketchy” visual input is transmitted from each eye to its ipsilateral hemisphere via a lesser number of ﬁbers that recross the midline at thalamic level (in both visual pathways), the main input from each eye goes to its contralateral hemisphere. In addition, the input from each eye to its contralateral hemisphere is conducted rapidly through myelinated ﬁbers, whereas the recrossing ﬁbers, in the tectofugal system at least, are slower conducting, unmyelinated ﬁbers (Deng and Rogers, 1998a,b). This means that the primary visual input from each eye is processed by its contralateral hemisphere, and this is the case for both sets of visual projections to the forebrain. Note also that, as shown in Fig. 2A, there is a structural asymmetry of the thalamofugal visual projections of the chick. There are more projections from the left side of the thalamus to the right Wulst than from the right side of the thalamus to the left Wulst (Adret and Rogers, 1989; Rogers and Deng, 1999; Rogers and Sink, 1988). This structural asymmetry underlies some of the lateralized functions of the chick (Rogers, 1990). Asymmetry of the organization of the visual projections to the forebrain has also been found in the adult pigeon but the location of the structural asymmetry in adult pigeons differs from that in the chick: It is present in the tectofugal system only, in the projections from the optic tecta to the rotundal nuclei (Skiba et al., 2000). It is not the focus of this chapter to discuss the neural correlates of visual lateralization in detail, but it is important to mention that, at least to some extent, the structural asymmetries determine lateralizations of visual responses and that they underlie lateralization revealed when birds are tested monocularly. In line with the results obtained by injecting cycloheximide or glutamate in the hemispheres, monocular testing has demonstrated that chicks can avoid pecking at pebbles, and so peck preferentially at grain, when they use the right eye (RE) but not when they use the left eye (LE) (Mench and Andrew, 1986; Rogers, 1997). Given ethical considerations for research using animals, the technique of monocular testing lends itself to revealing lateralization in other species with similar decussation of the optic nerves. Such species include ﬁsh, amphibians, and mammals that have their eyes placed laterally so that they have a small degree of binocular overlap and large lateral monocular ﬁelds of vision.

LATERALIZATION IN VERTEBRATES

115

In fact, by careful choice of testing procedures, it is possible to reveal biases in responding by the LE and RE in animals with laterally placed eyes. A study by Andrew et al. (2000) showed that chicks searching binocularly for grains scattered over a ﬂoor prefer to peck at grains located by the RE, even when no eye patches are applied and the chicks can use both eyes. Andrew et al. (2000) also showed that, when chicks have to manipulate an object to obtain a food reward, they display a bias in their direction of approach that enables preferential use of the RE. The object that had to be manipulated was a lid on the food dish, and, in a series of tests, it had to be removed in different ways, such as by grasping a corner and displacing it or by lifting it by a short string at the top. The direction of approach to the dish and the eye used for viewing was determined by videotaping the chick from overhead and then using frame-by-frame analysis of the tape. In all cases, the RE (and Lhemi) was used prior to performing the manipulation act. When the food was presented in a dish without a lid, the chicks approached it with the opposite bias, using the LE and Rhemi. The researchers concluded that preferential eye use was determined by the need for visual guidance of the bill to grasp and manipulate an object. This explanation would be consistent with the general model in that use of the Lhemi would allow inhibition of spontaneous, and misdirected, pecks to allow manipulation of the lid. When the food can be seen during approach, there is no need for inhibition of immediate pecking and the Rhemi is used. Similar specialization of the RE and Lhemi for controlling feeding responses has been shown in the pigeon, Columba livia (Gunt ¨ urk ¨ un, ¨ 1985), and also in the zebra ﬁnch, Taeniopygia guttata (Alonso, 1998). Gunt ¨ urk ¨ un ¨ and colleagues tested pigeons monocularly on a task used to reveal lateralization in chicks (Gunt ¨ urk ¨ un, ¨ 1985; Gunt ¨ urk ¨ un ¨ and Kesch, 1987). Adult pigeons were trained to discriminate safﬂower seeds from pebbles. The seeds were mixed in with the pebbles, and so the bird had to inhibit pecks that might have been misdirected toward pebbles. The pigeons were given several trials, and the pecking rate as well as the number of seeds ingested was scored. The pigeons using their RE consumed signiﬁcantly more seeds and pecked more accurately and faster than pigeons using the LE. Further testing showed that this difference in performance between pigeons using the LE or RE was not due to differences in visual acuity between the eyes. On a visual acuity task using high-contrast, square-wave gratings, no signiﬁcant difference was found between birds using the LE and those using the RE (Gunt ¨ urk ¨ un ¨ and Hahmann, 1994). This lateralization for control by the Lhemi in the seed–pebble test appears to be located at higher levels of neural processing, as shown earlier in the chick by administering cycloheximide or glutamate to the Wulst regions of the forebrain (see earlier discussion). It is also associated with the ability both to make considered decisions and to inhibit responses to the negative stimulus.

116

LESLEY J. ROGERS

Entirely consistent with this interpretation is the ﬁnding that pigeons perform better in a reversal learning paradigm when using the RE than they do when using the LE (Diekamp et al., 1999). The birds were trained on a color discrimination operant task and then reverse tested by rewarding the previously negative key and not rewarding the previously positive key. Another study has shown that such reversal learning is a function of the Wulst region of the forebrain (Shimizu and Hodos, 1989), which is consistent with the role of the Wulst in the Lhemi of the chick in performance of the pebble–grain task (see earlier discussion). Pigeons also have better recall of a complex visual discrimination task when they use the RE than when they use the LE (Fersen and Gunt ¨ urk ¨ un, ¨ 1990), consistent with an earlier ﬁnding of the same result in chickens, tested on a much simpler discrimination task (Gaston and Gaston, 1984). It is likely that this particular specialization of the Lhemi for recall reﬂects its role in controlling feeding responses that involve inhibition (considered decisions) rather than specialized involvement of the Lhemi in memory storage. Tests on a food storing bird, the marsh tit (Parus palustris), demonstrate a similar RE (Lhemi) superiority in recall some 50 h after the tits have been allowed to cache food when binocular (Clayton, 1993). The marsh tits tested using the LE only might, possibly, have access to memory of the sites in which they had previously cached seeds but be unable to inhibit responding to potential cache sites that they had not used. This interpretation differs from that given by Clayton (1993), who preferred the alternative hypothesis that birds using the RE have no access to long-term memory of the cache sites (i.e., no memory store in the Rhemi). Although there is, so far, no available evidence in support of either one of these alternative explanations, the data for marsh tits would be consistent with that of chicks and pigeons if the ﬁrst explanation is correct. In fact, there is an even more striking demonstration of the role of the RE of birds in feeding responses that require discrimination and manipulation. Hunt (1996) discovered that New Caledonian crows (Corvus moneduloides) manufacture and use two different kinds of tools, hooked tools made from twigs and probing tools made from pandanus leaves, and that the RE guides the birds’ responses during tool manufacture. The crows use these tools for probing holes in trees or the debris around trees to ﬁnd insects to eat. Unfortunately, eye use during probing for food has not been studied but eye use in making the tool from pandanus leaves has been deduced by examining the shape of the cutouts on the leaves. The bird uses its beak to sculpture a piece of pandanus leaf from the right edge of the main leaf. To do this, the bird must angle the beak to cut in a series of different directions. Hunt (2000) has shown that the crows make their tools to a standard shape by following a system of rules and that they have a form of handedness. The

LATERALIZATION IN VERTEBRATES

117

crows cut more tools from the left edges of the pandanus leaves than they do from the right edges (left and right being determined when looking down at the upper surface of the leaf and outward from the trunk). Since more tools are cut from the left edges of the pandanus leaves and the birds start at the narrow end of the tool working away from the trunk, it can be deduced that they use their right eye to guide their tool manufacturing behavior and that they primarily use the right side of their beak for cutting (for details and an illustration see Hunt, 2000). This interesting ﬁnding suggests that the Lhemi is used to control the cutting behavior and to ensure that the tool made matches a template. This specialization of the Lhemi may be a more complex evolution of the basic pattern described in chicks by Andrew et al. (2000); viz., use of the Lhemi when manipulation using the beak is required to obtain food from a dish. We also note here Hunt’s conclusion that crows have “handedness” for tool manufacture, as was known previously to be the case only in humans. Andrew and Rogers (2001) extended this view to draw attention to the involvement of the Lhemi in skilled manipulation in birds (right “eyedness”) and humans (right-handedness). Here it is worth mentioning a preliminary result (A. Hoffman and L. Rogers, unpublished) showing that juvenile Australian magpies (Gymnorhina tibicen) may display a signiﬁcant preference to approach on the parent bird’s right side when begging for food (out of six birds observed, all were biased to approach on the parent’s right side). Although a larger sample size is required to conﬁrm this observation, it suggests that lateral asymmetries might inﬂuence social behavior. In this case, the juvenile might position itself on its parent’s right side to ensure that the parent uses skilled manipulation of its beak when feeding the juvenile. All of these results, obtained from ﬁve different species of bird (and perhaps six, if the magpie is included), ﬁt with the general pattern of Lhemi specialization for control of the beak when manipulation is required or for feeding responses that require selection of food targets and inhibition of pecks at inedible targets. The same seems to apply to other vertebrate species. B. AMPHIBIANS Toads (Bufo bufo and Bufo viridus) strike preferentially at prey to their right side, suggesting control by the Lhemi (Vallortigara et al., 1998). Each toad was tested by placing it inside a glass cylinder and moving a live prey around it, outside the cylinder, in either a clockwise or counterclockwise direction. Tongue strikes were aimed at prey within the binocular ﬁeld. The angle between the toad’s lengthwise axis and the line between the point of contact of the tongue on the glass cylinder and midway between the toad’s eyes was determined (Fig. 3). When the prey was rotated clockwise

118

LESLEY J. ROGERS

FIG. 3. Striking at prey by two species of toads. The position of the strike has been determined by noting the point of contact of the tongue on the glass cylinder, outside which the prey is rotated, and by measuring the angle of this from the toad’s midline. The shaded areas represent the binocular ﬁeld of vision, and the white areas represent the monocular ﬁelds. Note that, when the prey is rotated clockwise (ﬁgures on the left side), striking is in the right hemiﬁeld region of the binocular ﬁeld. When the prey is rotated counterclockwise, the strikes are more widely distributed within the binocular ﬁeld. Modiﬁed from Vallortigara et al. (1998).

(as viewed from above the toad), almost all strikes were to the right of the toad’s midline (i.e., in the right visual hemiﬁeld). This was true for both species. When the direction was counterclockwise, no obvious side bias was seen, although in Bufo viridis there was still a tendency for more strikes to be directed to prey to the right of the midline than to left of the midline (Fig.3).

LATERALIZATION IN VERTEBRATES

119

Although all strikes are made in the toad’s binocular ﬁeld, the processes that lead to striking may be initiated when the toad ﬁrst catches sight of the prey in its extreme lateral, monocular ﬁeld. We might consider this as initial priming for the response, which is not given until the prey advances into the binocular ﬁeld. If so, it is interesting to note that priming in the left monocular ﬁeld (Rhemi) leads to a delayed response, not occurring until the prey passes the toad’s midline and enters the right hemiﬁeld (i.e., for clockwise rotation of the prey). Priming in the right monocular ﬁeld is more likely to trigger a response before the prey passes the midline, particularly in Bufo viridis. Considered together, these results suggest that the Lhemi is specialized to control prey catching responses in toads, as it is in birds. The prey striking responses of Bufo marinus have also been tested in the same way with prey rotated around the toad (Robins and Rogers, manuscript in preparation). In this species, clockwise rotation of the prey led to frequent striking at the prey, although not until it had entered the right visual hemiﬁeld. This indicates that in Bufo bufo, as in the other two species of toads, striking at prey is controlled by the Lhemi. When Bufo marinus was tested with the prey rotating counterclockwise, very few strikes were made. In this case, it seems that initial detection of the stimulus in the right, monocular visual ﬁeld leads to an inhibition of feeding responses. Although this result is consistent with the Lhemi’s ability to inhibit responding, it differs from the results obtained with the other two species, perhaps because the prey used in testing Bufo marinus (a cricket tethered to a vertical rod) was less preferred by this species. This bias of toads to strike at prey rotated clockwise only after it has passed the midline deserves further discussion. Some strikes are on the midline but, for clockwise rotation of the prey, the midline represents a clear dividing point with strikes to the right side and not the left side (Fig. 3). In contrast to the avian species discussed earlier, the toads tested have a large binocular ﬁeld of vision (approximately 70◦ in Bufo viridis, 90◦ in Bufo bufo, and 110◦ in Bufo marinus) resulting from location of their eyes somewhat frontally and toward the top of the head (Lippolis et al., 2001). Therefore, the side biases in toads make an interesting contrast to those found earlier in birds. In toads, the lateralization is seen clearly in the binocular ﬁeld (i.e., not in the lateral, monocular ﬁelds of vision, as is the case in birds discussed earlier, but in relation to the animal’s midline). It is as if the toad suffers from hemineglect syndrome, described for humans (Myslobodsky, 1983), for prey on its left side. These results suggest that toads may have mechanisms operating with respect to the midline and extracorporeal space, as reported previously only in humans. The midline of extracorporeal space is obvious in humans, despite the wide binocular ﬁeld and, more importantly, the fact that each eye sends input to both sides of the brain (Heilman et al., 1985; Mesulam, 1985). The midline division of responding may depend on different cognitive processes

120

LESLEY J. ROGERS

or mechanisms in toads and humans, but, as it is an aspect of lateralization not considered previously in nonhuman species, it deserves further investigation in species with wide binocular ﬁelds. C. FISH The rightward bias for feeding responses has been traced to an appearance during evolution as early as teleost ﬁsh. Fish direct bites at small “prey” objects to their right side. Zebraﬁsh (Brachydanio rerio) use the RE to ﬁxate on a target, a small bead, that the ﬁsh intends to bite, presumably for feeding (Mikløsi and Andrew, 1999). There is a strong, positive association between biting and RE use. No such eye bias occurs if the ﬁsh does not bite at the bead. The researchers found that the RE preference disappeared as the ﬁsh habituated to the bead (ceased to bite it) and reappeared when the bead was changed in appearance in a way that elicited biting again. Thus, use of the RE and Lhemi is associated with the decision to bite or not to bite the small bead. These bites were likely to be feeding bites, and not aggressive bites, since the ﬁsh had been fed previously at the site where the testing bead was located. Other experiments have shown that zebraﬁsh use the LE to view scenes and larger objects at a distance when no considered decision about responding has to be made (Mikløsi et al., 1998). By contrast, a novel scene evokes RE viewing when seen for the ﬁrst time, when the ﬁsh has to decide whether or not to respond to it. Presentation of the same scene a second time requires no such decision making and evokes viewing by the LE. This ﬁnding, along with a large number of other discoveries of lateralization in ﬁsh of various species (summarized in Bisazza et al., 1998a; also see later), attests to an early evolution of lateralization. In fact, Andrew et al. (2000) have hypothesized that lateralization evolved in association with the ﬁrst visually controlled predation. They have argued that the brain of the earliest chordates may have been as asymmetrical as their viscera. Taking amphioxus as an example, they note that the position of the mouth is on the left side in the larval stages of development and it sends its neural inputs to the Lhemi. The mouth of an amphioxus does not migrate to the midline until the adult stage is reached. Andrew et al. (2000) reason that this asymmetry in the larval form put the left side of the brain in control of feeding responses, which are largely tactile in amphioxus, and that this lateralization was retained when visually guided predation evolved. Thus, the earliest visually guided motor responses for feeding were, Andrew and associates suggest, controlled by the RE and its inputs to the left hemisphere. It is noted that this hypothesis for the evolution of lateralization departs radically from the alternative, and currently accepted hypothesis, that

LATERALIZATION IN VERTEBRATES

121

vertebrates evolved to become increasingly more lateralized (Corballis, 1991). Given the numerous and now well-documented examples of lateralization in lower vertebrates (Vallortigara et al., 1999), including teleost ﬁsh, there seems to be no reason to adhere to the notion of increasing lateralization as vertebrates evolved. As Andrew (2001) has pointed out, the nervous systems of the earliest vertebrates may have been strongly lateralized. D. NONPRIMATE MAMMALS No research has speciﬁcally investigated lateralization of feeding responses in nonprimate mammals, although many of the tasks with which Bianki (1988) tested for lateralization in rats relied on conditioning for a food reward. His approach was to cause cortical spreading depression to the left or right hemisphere. The results showed that rats with spreading depression in the Rhemi (and therefore using their Lhemi) were superior in conditioned reﬂexes and that they adapted to the reward procedure better than those forced to use the Rhemi. Right-side responding was found to be better in rats trained to intercept a beam of light with their snouts. The rat had to hold its snout in a central hole until a light was illuminated on its left or right side, at which time it had to move its head to the side of the light and this intercepted a photocell beam that delivered a food reward (Mittleman et al., 1988). This result is suggestive of Lhemi specialization for responses akin to manipulation for a food reward, as described for the chick. Certainly, it would be predicted from the general model for lateralization of feeding responses. Response to sickness induced by injecting lipopolysaccharide is, perhaps, more relevant to discussion of specialization of the Lhemi for feeding responses. Nevue et al. (1998) found that the sickness response, characterized by depressed social behavior, increased immobility, and reduced intake of food for 6 h following the injection, is more pronounced in right-pawed mice than in left-pawed mice. This suggests that it is controlled by the Lhemi. This conclusion is based on the reasoning that preferred use of one paw activates its contralateral hemisphere or that it may reﬂect a precondition for higher activity in that hemisphere, as suggested for handedness and hemispheric activity in primates (Hellige et al., 1994; King, 1995; Ward et al., 1990; discussed in detail in Section III.E). In fact, research has shown that this is, indeed, the case at least in the motor cortex and during movement of the hand: In humans, movement of the preferred (dominant) hand is associated with a greater volume of activation of the contralateral motor cortex than is movement of the nondominant hand (Dassonville et al., 1997). It is not unreasonable to assume that the activation extends to other regions of the cortex, although further research is required to conﬁrm this assumption.

122

LESLEY J. ROGERS

It is, of course, recognized that all of these examples of lateralization of feeding-related responses in rodents may have other explanations not strongly related to specialization of the Lhemi speciﬁcally for feeding responses. Nevertheless, they are not inconsistent with the specialization of the Lhemi for control of responses requiring manipulation of objects and inhibition of responding while the animal decides on the appropriate response. E. PRIMATES Primates represent a special case because most of their feeding and prey catching responses use the hands, at least in the initial grasping of the prey, whereas all of the examples preceding refer to species in which a single effector is used to obtain food (e.g., the mouth in ﬁsh, the beak in birds, and the tongue in toads). Bias to the right side and control by the left hemisphere might be expected to put the right hand of primates in control for grasping food, but in lower primates, for example, the left hand is used more often than the right. Prosimians seize moving prey with a bias for the left hand, using a ballistic movement of the hand and arm (Ward, 1999). Squirrel monkeys also use the left hand to capture swimming ﬁsh (King and Landau, 1993). One can only assume that the ability to use two effectors (hands) severed the link between Lhemi control and right-side bias for control of the effector organ. It would be logical to predict that primates when prey catching would make use of the Rhemi’s advantage in processing topographical cues and so use the left hand (Andrew and Rogers, 2001). In fact, a similar preference to use the left hand for grasping or tracking moving objects has been reported in other species. Cats also show an advantage when using the left paw to reach for a moving target (Lorincz and Fabre-Thorpe, 1996), and humans have superior ability with the left hand for throwing (Watson and Kimura, 1989) and producing ballistic movements (Guiard et al., 1983). Planned manipulation requiring visual guidance may, by contrast, remain under the control of the Lhemi and so be carried out by the right hand (Andrew and Rogers, 2001). This means that it may reﬂect the ancient evolutionary pattern of specialization of the Lhemi and not, in essence, be determined by postural demands, as proposed by McNeilage et al. (1987). Lhemi control of visual manipulation may, thus, explain right-handedness in humans for writing and other visuomotor tasks and maybe also right-handedness for ﬁne manipulation in higher primates (Hopkins and Morris, 1993). Captive chimpanzees (Pan troglodytes) exhibit a right-hand preference for manipulating a joystick (Morris et al., 1993), but no such handedness has been seen in wild chimpanzees (Marchant and McGrew, 1996). Hand preferences in the apes remain controversial. The left-handedness of orangutans when manipulating parts of the face (Rogers and Kaplan, 1996)

LATERALIZATION IN VERTEBRATES

123

seems to be opposite to the prediction of Lhemi control of manipulation but this response is not visually guided. Instead it relies on tactile feedback, shown in other primates and contexts to be a function of the left hand and Rhemi (Lacreuse and Fragaszy, 1996, 1999), as mentioned previously.

III. PREDATOR-ESCAPE AND FEAR RESPONSES As the evidence presented later shows, predator-escape and a broad range of fear responses are primarily controlled by the Rhemi, and, in the case of species with lateral placement of the eyes, they are elicited more rapidly and strongly by stimuli on the animal’s left side. It is possible that the function of the Rhemi in fear and predator-escape responses is one aspect of this hemisphere’s specialization for expressing intense emotions and is closely linked to more effective control of sympathetic outﬂow by the Rhemi, compared to control of parasympathetic outﬂow by the Lhemi (Wittling, 1997). The Rhemi’s control of sympathetic mechanisms of the heart has been shown in humans (Hugdahl, 1995), dogs, and cats (Lane and Jennings, 1995). A. BIRDS Phillips and Youngren (1986) investigated fear responses of chicks following placement of a lesion in the left or right archistriatal region of the forebrain. The lesioned chicks were placed in a novel environment and fearinduced peeping (distress calling) was scored. Peeping was reduced to a greater extent by lesions of the archistriatum on the right side than by lesions on the left side. Hence, it would seem that the Rhemi is specialized for the expression of distress and fear responses. Entirely consistent with this result, chicks peep more when they detect a novel stimulus in the left, lateral visual ﬁeld than when they detect it in the right, lateral visual ﬁeld (Andrew, 1983), although age is a factor in this response (Andrew and Brennan, 1983). As there have been no published reports of experiments testing whether predator-escape responses of birds are lateralized, it is appropriate to mention here a study of young domestic chicks completed in my laboratory (Lippolis and Rogers, in preparation). The test involved positioning the chick in the center of a circular arena with its head through a hole in a small piece of card standing vertically. A simulated predator resembling the head of a snake was introduced to the arena at various points from behind a wall constructed of cloth and advanced into the chick’s left or right monocular ﬁeld. The model was moved mechanically by a motor programmed to advance it a ﬁxed distance at two speeds. The latency from the ﬁrst appearance of the model in the arena to the time when the chick withdrew its head from the

124

LESLEY J. ROGERS

hole was scored from video recordings and found to be signiﬁcantly shorter for presentations on the chick’s left side compared to the right side. This result indicates that the Rhemi is specialized for control of predatorescape responses as well as for expressing fear responses. A similar result was found in another experiment in which a novel stimulus was advanced toward the chick from behind while the bird was feeding (Rogers and Anson, 1979). On each trial the stimulus entered the monocular ﬁeld of either the LE or the RE. The distance of the stimulus from the chick at the time when the chick stopped feeding and showed the startle response was measured and found to be greater for the LE than for the RE. Thus, the Rhemi is also more responsive to novel stimuli. As it is not possible to separate this specialization from that of the response to a model predator, it might be preferable to see the Rhemi as being specialized to control immediate or spontaneous responses that are elicited by novel, and potentially dangerous, stimuli. These responses do not require any form of manipulation to be carried out by the animal, as discussed previously for feeding, but are rapid responses made without weighing up possible alternative responses. Lateralization is, thus, a matter of the amount, as well as the kind, of attention given to a stimulus on its ﬁrst detection in the left or right visual ﬁelds. Consistent with these specializations of the Rhemi, adult chickens have been found to scan overhead using the LE when they hear their speciesspeciﬁc alarm call signaling the presence of an aerial predator. Evans et al. (1993) played recorded aerial alarm calls to hens and noted that they responded by tilting their heads to look overhead using the LE. This engages the Rhemi in detection as well as the rapid motor responses that may be required to escape the predator. In other words, the alarm call puts the Rhemi in charge and so makes use of its specialization for a number of linked functions; viz., novelty detection, rapid responding, fear expression, and predator-escape responses. A recent study (A. Hoffman and L. Rogers, in preparation) has found a similar lateralization in the fear/escape responses of wild Australian magpies. The experimenter walked toward the bird (sample size of 34 birds) from behind when it was feeding. The eye used by the bird to view the experimenter could be determined easily since the bird turns its head and ﬁxates using one or the other monocular visual ﬁeld. After looking thus, the bird ran away a few paces; the human then approached again allowing a new score of eye use to be obtained. Between 10 and 30 scores were obtained per bird. Although some of the magpies displayed a signiﬁcant eye preference, there was no bias in the population and many of the birds showed no signiﬁcant preference. Alarm calling was also recorded. The percent RE preference was determined for each bird and these scores were correlated with alarm calling (scores adjusted for the number of recorded events of eye use).

LATERALIZATION IN VERTEBRATES

125

The percent RE preference correlated negatively with alarm calling (r = −0.43, p = 0.01), indicating that fear and escape responses are elicited more strongly when the LE (and Rhemi) is used than vice versa. This result forms a basis for testing experimentally the hypothesized role of the Rhemi in control of fear responses in other avian species in their natural environment. The Rhemi is also more likely to control immediate responses to noxious olfactory stimuli (Rogers et al., 1998). Chicks were presented with a novel blue bead mounted on a hollow rod from which the irritant odor, clove oil, was released and could be detected by the chick as it pecked the bead. The chicks were tested with either the right or the left nostril occluded. Those using the right nostril only reacted to the odor by shaking their head in a characteristic disgust response, whereas this response was rare in those using their left nostril. Since olfactory input from each nostril goes only to the ipsilateral hemisphere, it can be concluded that the innate disgust response is elicted more readily when the Rhemi is processing the information. The Lhemi is able to suppress this response and, presumably, direct the chick’s attention to the colored bead. Although head shaking in response to an odor is not a fear response, it may be associated with the same fundamental specialization of the Rhemi for expression of intense emotions. B. AMPHIBIANS Toads also are more reactive to predators advancing from their left side than to those advancing from their right side. This left-side bias may demonstrate that the Rhemi responds sooner to a potential predator and expresses fear more strongly. This result was obtained by testing toads (three species of Bufo) in a circular arena by advancing a mechanical model of a snake (the same one used for the chicks) toward the toad in its midline (binocular) or in the left or right lateral monocular ﬁelds (Lippolis et al., 2001). The toads were more likely to react, most often by jumping away, when the simulated predator was in their binocular ﬁeld or left monocular ﬁeld than when it was in their right monocular ﬁeld. It can be suggested, therefore, that toads might be somewhat more likely to succumb to predators approaching them from their right side than from their left side. In fact, predators might exploit this side bias, present at a population level, by preferentially approaching their prey on its right side and turning to strike leftward. There has been no research addressing this possibility. Also, it is not yet known whether the lateral bias for reactivity to the simulated predator is present only in the monocular, lateral ﬁelds of vision or whether, as for the lateralization of striking at prey, the toad’s midline is the dividing point. Given the large binocular overlap in most species of toads (including the three species tested so far), it would be important to determine whether the lesser reactivity on

126

LESLEY J. ROGERS

the right side is conﬁned to the relatively small monocular ﬁeld or whether it extends over the entire right hemiﬁeld of vision. If it is conﬁned to the monocular ﬁeld of the right eye, it would be tempting to suggest that toads minimize the side bias in reactivity to a predator by maximizing the size of their binocular ﬁeld in the horizontal plane. The preceding discussion has considered lateralization of reactivity (whether or not an escape response occurs) to an advancing predator but not lateralization of the motor response itself. The motor responses elicited by an advancing predator may also be lateralized in some anurans. Dill (1977) scored the direction of jumping in Paciﬁc tree frogs, Hyla regilla, evoked by swinging a ball toward the frog along its midline. Leftward jumps were more common, a result that Dill attributed to the frogs having a longer right than left hindlimb, although there was no direct concordance of this character and the lateralized jumping scores within individuals. Later experiments have found no such motor bias for jumping in Bufo marinus faced with an advancing model predator in the frontal ﬁeld (Lippolis et al., 2001) even though Bufo marinus does exhibit a right hindlimb preference for initiating a successful righting response from a supine position (Robins et al., 1998). Motor biases in righting responses seem to be common in anurans, as shown by the limb or side preference in righting in three species of toads (Bisazza et al., 1996, 1997b) and two species of Australian tree frogs (Litoria caerulea and L. latopalmata). As shown in Fig. 4, Litoria sp. show a population bias to rotate their body around its right side when righting on a dry surface. It is relevant to consider that an anuran might be required to right itself from a supine position as part of its repertoire of predator-escape responses. Therefore, it is worth noting that rotation around the right side of the body allows the frog to continue viewing the predator using its LE. This might be coincidental, but it is suggested on the basis of the Rhemi’s specialization for predator monitoring and control of escape responses in other species, especially relevant in the case of the teleost ancestors of anurans (discussed later). Tadpoles also show lateralized escape responses, after they have surfaced to breathe air. Wassersug et al. (1999) tested two species of frog tadpoles in water with reduced oxygen content, which forces the tadpoles to rise to the surface to gulp air. They found a signiﬁcant population bias for bullfrog tadpoles (Rana catesbeiana) to turn leftward during the C-turn escape response that follows the air-breathing but no such bias in clawed frog tadpoles (Xenopus laevis). The difference between the two species could, as the researchers suggested, be attributed to placement of the spiracle through which water that enters the mouth is expelled. Bullfrog tadpoles have a single spiracle on the left side of the body, whereas clawed frog tadpoles have a spiracle

LATERALIZATION IN VERTEBRATES

127

FIG. 4. Motor lateralizations in toads and tadpoles of the Australian tree frog Litoria sp. is presented. The two species Litoria latopalmata and L. lesueuri were tested. (A) Righting responses in toads from a supine position: the number of turns around the right side of the body out of a total of 10 trials is plotted on the x-axis. Note the population bias for a rightside preference. (B) Direction of turning in tadpoles scored for escape turns given to a visual stimulus and for turns after breathing air at the surface of the water. As individuals could not be identiﬁed, the number of tadpoles recorded as turning left or right is plotted. Note the signiﬁcant rightward turning bias to escape and the signiﬁcant leftward bias of turns after breathing air. These data were collected from toads in their natural environment, Glenreagh, New South Wales.

128

LESLEY J. ROGERS

on each side of the body. However, this explanation would be unlikely if the direction of turning in a given species varies with the context in which it occurs. This is the case in Litoria sp.: As in the bullfrog tadpole, Litoria tadpoles have a population bias to turn leftward after breathing air and a population bias to turn rightward in escape turns elicited by a visual stimulus (Fig. 4; L. Rogers, in preparation). A tadpole breathes air by ﬁrst drifting up to slightly below the surface of the water and then it swims rapidly to the surface, gulps air, and turns rapidly in a C-turn, and swims down to deeper levels in the water. It is probable that the rapid-turning phase is triggered by the activity of the Mauthner cell on one side of the body. This response may not be driven by visual inputs. A visually driven escape response was elicited by waving a hat to elicit the escape response from tadpoles in a pond. Litoria tadpoles often remain stationary at the surface of still water in a pond, possibly because it is warmer, and a sudden visual stimulus overhead elicits rapid turning followed by swimming downward or ﬁrst along but just under the surface of the water away from the stimulus and then downward. As shown in Fig. 4, more tadpoles turned right than left, the position of the stimulus being randomized with respect to the position of the tadpole. This rightward turning, when vision is used, ﬁts with the toad’s greater reactivity to predators seen in the lateral ﬁeld of the LE and with use of the LE during escape turning in the teleost ancestors of anurans (see later). It suggests the presence of a population bias toward use of the Rhemi during escape in both tadpoles and toads of various anuran species. C. FISH The turning preferences of poeciliid ﬁsh, Girardinus falcatus, in escaping from a model predator have been tested by presenting the model to each ﬁsh from in front as it was swimming toward the stimulus (Cantalupo et al., 1995). After detecting the simulated predator in the frontal ﬁeld, the majority of the ﬁsh escaped with a rightward turning bias, signiﬁcant at the population level. As Vallortigara and Bisazza (2001) suggest, this bias is likely to be generated at the level of the optic tectum, where visual information is processed before being relayed to the Mauthner cell that drives the C-start ballistic escape turning movement (Zottoli et al., 1987). There is a Mauthner cell on each side of the brain, and the axon of each one crosses the animal’s midline to innervate the musculature on the contralateral side of the body (Canﬁeld and Rose, 1993). This means that rightward turning is caused by greater activity in the left Mauthner cell compared to the right, and so it may be generated by excitatory inputs to the left Mauthner cell from the left optic tectum and,

LATERALIZATION IN VERTEBRATES

129

thus, originally from the RE. This implies that the Lhemi is controlling the response, although, once the ﬁsh has turned rightward, it would engage the LE and right side of the brain in continued monitoring of the predator. After repeated presentations of the simulated predator, Girardinus falcatus shifted to leftward turning, again with a bias at the population level (Cantalupo et al., 1995). By this stage, it was likely that the ﬁsh had habituated to the stimulus and were no longer treating it as a potential predator. The results obtained with Girardinus falcatus are consistent and explicable in terms of the general model for hemispheric lateralization (see earlier discussion) only in so far as during turning the LE and Rhemi are used by the ﬁsh. Escape responses in ﬁsh might, however, be rather variable from species to species, depending on whether or not the species schools and on the habitat it occupies. The teleost ﬁsh, Jenynsia lineata, exhibits individual, but not population, lateralization of turning responses when faced with a simulated predator (Bisazza et al., 1997a). Half of the group tested turned leftward, and so continued to ﬁxate on the predator with their right eye, and the other half turned rightward. These preferences were retained when the same ﬁsh were tested 1 month later. In fact, Bisazza et al. (2000) have conducted a study of 16 species of ﬁsh presented with a predator model behind a barrier with vertical bars. Ten of the species showed a signiﬁcant population bias for turning but only six of these turned rightward. The eye and the hemisphere driving turning may, of course, depend on whether the model predator was seen as a potential predator or not by each particular species, since habituation to the model leads to a change in the direction of turning (see earlier discussion). The direction of turning to go around the barrier must depend on when the ﬁnal decision to approach is made. It may be made at the point of turning and, depending on the species, with or without further monitoring as the ﬁsh swims around the barrier. The most notable ﬁnding of this study by Bisazza et al. (2000) was that a population bias for turning (left or right) was present in all of the species that school but in less than half of the species that do not school (discussed later). It is interesting to note that more scars have been reported to occur on the left side of the whiteﬁsh, Coregonus nasus and C. clupeaformis (Reist et al., 1987). This asymmetrical distribution presumably indicates unsuccessful attempts at predation by lampreys, birds, or bears, as claimed by the researchers. It is impossible to tell whether the lateralization results from less successful leftward escape responses by the ﬁsh (leftward escaping ﬁsh might, actually, succumb to the predator), more successful rightward escapes by the ﬁsh (if the rightward ones escape unscathed), or rightward prey-catching responses of the predators. However, all of these alternatives assume that there is no population bias for turning. If there is a population

130

LESLEY J. ROGERS

bias to turn rightward, as for Girardinus falcatus, whiteﬁsh may be more likely to be damaged on their left side since that side would be next to the predator. In some circumstances approaching a potential predator, instead of escaping, occurs, and here ﬁsh show a preference to use the RE to view the predator. Some ﬁsh leave their school in twos or threes to approach a predator and examine it more closely. Bisazza et al. (1998b) have shown that poeciliid ﬁsh turn leftward when they reach a barrier separating them from a simulated model predator. This direction of turning allows the ﬁsh to view the predator using the RE (Facchin et al., 1999). As an approach would demand inhibition of the escape response, it is entirely consistent with the general model that the Lhemi is engaged to perform the task. Added to this, a mosquito ﬁsh (Gambusia holbrooki) will approach the predator more closely when it can see a conspeciﬁc with its LE at the same time that it approaches the predator. Bisazza et al. (1999) discovered this by placing a mirror on the left or right side of a tank in which a mosquito ﬁsh could approach a predator ﬁsh in an adjoining tank. The test ﬁsh approached the predator more closely when the mirror was on its left side than when the mirror was on its right side. This preference for a ﬁsh to position itself so that a conspeciﬁc (actually, the ﬁsh’s own image in a mirror) is on its left side appears to be a general characteristic of teleosts: It has been found in females of ﬁve other species of teleosts in the absence of predator inspection (Sovrano et al., 1999) and conﬁrmed to occur in the mosquito ﬁsh in the absence of a predator (De Santi et al., 2001). The preferred eye for viewing a conspeciﬁc during inspection of a predator, however, appears to vary with the ﬁsh’s prior experience with the predator. Guppies bred in captivity and given prior experience with a hungry predator in their home tank showed more tendency to approach the predator when the image of a conspeciﬁc was on their right side (De Santi et al., 2000). They had a preference to use their LE to view the familiar predator and so could respond to it spontaneously and rapidly. However, this does not explain why the ﬁsh should approach the familiar predator unless the escape response had been inhibited by the absence of chemical or other cues of the predator ﬁsh because it was in a separate tank. There is clearly a need to investigate the responses of ﬁsh to predators in more detail: They are an excellent group for systematic investigations. D. NONPRIMATE MAMMALS There have been no speciﬁc tests to see whether nonprimate mammals have lateralized predator-escape responses, but some tests have shown that rats display heightened fear and withdrawal responses after lesioning of the

LATERALIZATION IN VERTEBRATES

131

Lhemi, thus forcing use of the Rhemi only. The ﬁrst experiment showing this relied on a learned withdrawal response—that of taste aversion to a sweet solution (Denenberg et al., 1980). Lithium chloride treatment was used to induce the taste aversion and the left or right hemisphere of the rat was ablated before retention was tested. Aversion and withdrawal behavior was greater in the rats with an intact Rhemi than in those with an intact Lhemi. Of course, there may be alternative explanations for this result. The role of the Rhemi in fear responses at a physiological level is shown by that fact that lesions of the Rhemi and not the Lhemi decrease circulating noradrenalin levels (Robinson, 1985). Added to this, it is apparent that the Rhemi has greater control of the hypothalamic-pituitary-adrenal axis than the Lhemi, as shown by the fact that, in response to stress induced by physical restraint, plasma corticosterone increases to a higher level in left-pawed mice compared to ambidextrous and right-pawed mice (Nevue and Moya, 1997). These data for rats do not provide strong evidence in support of the hypothesized general model associating the Rhemi with control of fear responses, but they do not contradict it. E. PRIMATES Primates express fear more strongly on the left side of the face, as found in rhesus monkeys (Macaca mulatta) (Hauser, 1993), common marmosets (Callithrix jacchus) (Hook-Costigan and Rogers, 1998) and also in humans (Sackheim et al., 1978). Also, as shown in rhesus monkeys, the left side of the mouth opens before the right during facial expressions of fear made alone or together with a vocalization (Hauser and Akre, 2001; see also Weiss et al., 2001). Since the musculature of the left side of the face is controlled by the Rhemi, this asymmetry reﬂects specialization of the Rhemi for the expression of intense emotions. Although the facial expression is one of fear, this cannot be used as direct evidence that the Rhemi controls fear responses in a more general sense, but it does, at least, suggest that this may be so. The validity of this suggestion is, of course, strengthened by the evidence for Rhemi control of fear responses in a range of species. Also, fear responses have been shown to be expressed more strongly in left-handed primates. Hand preferences in primates are usually measured as the preferential use of one hand over the other for simple tasks, particularly tasks involving picking up and holding food (Ward and Hopkins, 1993). Hand preference is used to indicate dominance of the contralateral hemisphere, on the grounds that consistent use of one hand in a number of tasks either activates the contralateral hemisphere, which controls that hand, or reﬂects a precondition of higher activity in that hemisphere (Hellige et al., 1994; King, 1995; Ward et al., 1990). To say this in another way, preferential use of a hand

132

LESLEY J. ROGERS

is associated with an attentional asymmetry that extends to higher cortical levels in the contralateral hemisphere (Peters, 1995). Thus left-handed individuals might show diffuse attention and intense emotional responses, including fear responses and withdrawal, whereas right-handed individuals might be better able than left-handed ones to inhibit fear and withdrawal to allow exploration of novel objects and situations (Davidson, 1992, 1995). Results obtained by testing chimpanzees (Hopkins and Bennett, 1994) and marmosets (Cameron and Rogers, 1999) suggest that this is, indeed, the case. Hopkins and Bennett (1994) found that the right-handed chimpanzees approached novel toys with a shorter latency than non-right-handed (i.e., lefthanded and no preference) chimpanzees. The right-handers also touched more toys than the non-right-handers. Thus, approach behavior and exploration appeared to be associated with greater activation of the left hemisphere, and avoidance behavior appeared to be associated with greater activation of the right hemisphere. A similar result has been found in common marmosets (Callithrix jacchus). Individual marmosets display hand preferences consistent over time and in similar tasks, but there is no population bias in this species. Half of the population is left-hand preferent and half is right-hand preferent (Box, 1977; Hook and Rogers, 2000; Hook-Costigan and Rogers, 1995, 1996; Rothe, 1973). Cameron and Rogers (1999) found that marmosets with a right-hand preference for holding food displayed a shorter latency to enter, on their own, a novel room containing novel structures and objects, touched more objects, and performed more touches and more parallax movements than subjects with a left-hand preference. The same marmosets were tested 2 years later and, again, the left-handed subjects took signiﬁcantly longer than the right-handed ones to emerge from their nest box into an unfamiliar room (J. Hall and L. Rogers, paper in preparation). In fact, left-handed males had the longest latency to emerge. The results obtained with chimpanzees and marmosets support the hypothesized specialization of the Rhemi for expressing fear and other intense emotional states and thus its association with withdrawal from novel situations and novel objects. The Lhemi inhibits fear and so its activity permits approach of novel objects and entry into unfamiliar environments. In humans, Rhemi control has been variously associated with intense emotions or negative emotions and withdrawal (Davidson, 1992, 1995). Dimond et al. (1976) have found that the use of the left visual ﬁeld to view a ﬁlm resulted in more negative assessment than did use of the right visual ﬁeld to view the same ﬁlm. Frontal and anterior regions of the Rhemi are selectively activated during withdrawn emotional states involving fear and disgust and, consistent with this, PET scans have revealed elevated activation of the Rhemi during resting in panic-prone subjects (Davidson, 1995). Also,

LATERALIZATION IN VERTEBRATES

133

schizotypy with social and emotional withdrawal is associated with Rhemi dominance, seen in terms of scoring better memory of faces than words and poverty of speech (Gruzelier and Doig, 1996). Damage to the frontal region of the Lhemi, presumably forcing the equivalent region of the Rhemi to take control, leads to decreased interaction with other people and difﬁculty in initiating voluntary actions (Davidson, 1995). Moreover, patients with injury to the left hemisphere resulting from stroke are signiﬁcantly more depressed than those with equivalent injury to the right hemisphere (Robinson and Price, 1982; Robinson et al., 1984). Rhemi involvement in human emotion, therefore, seems to be associated with negative and intense states and with social withdrawal. It is not known whether control of predator-escape responses is a specialization of the Rhemi in primates but it is a possibility worth investigating. The evidence discussed earlier makes Rhemi involvement highly likely.

IV. SPATIAL MAPS AND ATTENTION Perhaps in association with its specialization for attending to novel stimuli and in controlling escape responses, the Rhemi is specialized to process spatial information using a topographical map. These abilities might be linked because there is a functional need to attend to the spatial location of a novel stimulus and to use spatial information to escape effectively from attack by a predator. A. BIRDS The ﬁrst evidence for use of the Rhemi to assess topographical cues and to utilize spatial memory was provided by Rashid and Andrew (1989), who tested chicks in a search task requiring them to ﬁnd food buried under sawdust. The chicks were able to locate the buried food using two types of topographical cues, both distant cues from outside the testing arena (i.e., structures within the room and mainly overhead) and local cues inside the arena (two cylinders placed next to the buried food during training). The chicks were trained binocularly and tested monocularly. During testing either the walls of the arena were rotated relative to the distal cues in the testing room or the local cues were shifted. To prevent new learning during testing, no food was buried under the sawdust. The site(s) where the chicks scratched in their attempts to uncover food were recorded. The results showed that the chicks used the distal cues to locate the expected food site when they were using their LE (Rhemi) and they used the local cues when they were using their RE (Lhemi).

134

LESLEY J. ROGERS

Evidence that chicks using the LE attend to spatial cues is also shown by the fact that the response to a colored bead dishabituates when the bead is introduced into their cage from a different angle, whereas chicks using the RE ignore this change (Andrew, 1991). The spatial information is taken into account only when the Rhemi is used. Although it would be simpler to think of the Rhemi as processing spatial information and the Lhemi as being unable to do so, such an absolute division of function is not correct. In fact, both eyes and hemispheres are able to use spatial information of very different kinds. The LE and Rhemi use a topographical map taking into account the distal cues, whereas the RE and Lhemi use location relative to nearby features. Both types of information are available to chicks tested in the binocular condition, but the topographical cues take precedence over the local cues, as Tommasi and Vallortigara (2000) were able to show. They trained chicks to ﬁnd buried food in the center of an arena and then tested them in arenas of different sizes, the chicks being able to use both eyes in training and testing. When the chicks were tested in a smaller arena, they searched in the center, showing that they were using topographical cues. When they were tested in a larger arena, they focused their searching in two locations, in the center and, to a lesser extent, at distances from the walls equivalent to the distance at which the center had been located in the smaller, training arena. Most of their searching relied on a concept of the center, but relative distance from the walls was also used. Further tests by Tommasi and Vallortigara (2000) showed that the chicks were using angular estimates of the height of the walls to specify the center rather than a detailed topographical map, but, in their tests, a screen placed over the arena prevented the chicks from seeing any features of the room. Angular estimation of height would, in fact, be a basis on which to construct a topographical map if more features were available for the chick to view. Such a topographical map would be constructed using rules, and it would encode spatial information in a more global and adaptable form than the other kind of spatial information that relies on measuring relative distance from conspicuous, local features. Subsequent experiments by Tommasi, Vallortigara, and colleagues have shown that the Rhemi can operate on rules that use spatial information to construct a topographical map, whereas the Lhemi cannot. If chicks are trained to search for buried food in the center of an arena, they choose to search in the center of distorted versions of the arena when they use their LE but not when they use their RE (Tommasi et al., 1997; summarized in Vallortigara, 2000). When using their RE they simply search at a distance from the walls equal to where the center was in the training arena (i.e., they search in a strip around the walls at this distance). The chicks using their RE rely on local, object-speciﬁc cues, instead of on a spatial map plus a rule specifying the topographical location of food.

LATERALIZATION IN VERTEBRATES

135

In another task, chicks were trained binocularly to ﬁnd buried food in the center of a square arena and then tested monocularly in a larger arena, also square (Tommasi and Vallortigara, 2001). Chicks using the LE during testing searched in the center of the enlarged arena and chicks using the RE searched at a distance from the walls equivalent to the distance to the center in the training arena (Tommasi and Vallortigara, 2001). When the testing arena was made smaller than the training arena, LE chicks again searched at the center, whereas RE chicks searched erratically and in a dispersed manner, presumably because they could no longer specify the distance from the walls to the center. When the location of the food was marked by a vertical rod inserted into the sawdust during training and then placed in a corner during testing, LE chicks were able to ignore it and search at the center but RE chicks searched next to the shifted landmark. All of these tests show that the Rhemi uses relational spatial information to build a map, whereas the Lhemi uses absolute measurement of distance speciﬁed by local cues (Tommasi and Vallortigara, 2001). The hen’s preferential use of the LE to monitor overhead for an aerial predator (Evans et al., 1993; discussed earlier) may make use of the spatial abilities of the Rhemi since the position in the visual ﬁeld of a potential predator is important information for the bird to take into account. Indeed, the LE is expected to dominate in tasks requiring the bird to monitor topographical information. This specialization may also explain the observation that kookaburras (Dacelo gigas) tilt their head to use the LE when monitoring the ground beneath them for moving prey. Almost two-thirds of wild kookaburras (total of 90 birds) spotted sitting on telegraph wires while searching for prey on the ground were recorded as using the LE to look downward, the remainder mostly used both eyes (Rogers, 2001). Such a preference for the LE and Rhemi enables use of spatial information and it also ensures that the bird responds rapidly. A limited number of observations suggest that the kookaburra looks brieﬂy with its RE just before it swoops to capture the prey. If the latter is conﬁrmed by more observations, this shift to the RE would reﬂect use of Lhemi in object-related tasks and therefore feeding (discussed earlier). Spatial learning and memory is highly developed in marsh tits (Parus palustris) and lateralized in similar ways to that shown in the chick (Clayton, 1993; Clayton and Krebs, 1993). Lateralization is shown during recovery of cached food items, when the birds are tested with one or the other eye covered (Clayton and Krebs, 1994). When the RE is in use, for retrieval just after food has been cached, the marsh tits make use of the local cues associated with the hole used for storing the food. When the LE is in use, spatial position relative to the test room is used instead to locate the cache. Hence the Rhemi is specialized to use and to store spatial information in this species also.

136

LESLEY J. ROGERS

Successful foraging for food may require use of spatial cues (Rhemi) in some circumstances and local cues (Lhemi) in other circumstances. Tommasi et al. (2000) have shown that parallel processing of information in the two hemispheres allows competition leading to choice of the hemisphere more suited to carry out the task at hand. They trained and tested chicks binocularly in a task requiring the chick to ﬁnd a dish of food in one arm of a cross-shaped arena, and there were two different conditions of searching. One condition required the chick to use spatial cues of conspicuous objects in the testing room because all arms of the arena had dishes that looked the same but the dish in one arm was baited and always in the same position. In this case, the chicks began to search by making a series of turns to the left, indicating that the Rhemi was in control. In the other condition, the dish with food was distinguished by colored stripes and it was moved between the arms of the maze on each trial. Hence the chick had to use the local cues (stripes) to locate the food dish and, in this case, searching began with a series of right turns, indicating control by the Lhemi. Therefore, task demands had set the bias for which of the hemispheres would assume control. B. AMPHIBIANS No research has been conducted to see whether spatial processing is lateralized in amphibia. It is known that toads update the position of their escape routes each time they move and that this guides their escape leaps (Ingle and Hoff, 1990). Furthermore, toads are more likely to respond to a predator on their left side (see earlier discussion). On the basis of this result, one would predict that the Rhemi has an advantage processing topographical information, but this remains to be investigated. C. FISH Little can be said about the processing of spatial information in ﬁsh. Preferential use of the LE and Rhemi to view familiar scenes and objects when no speciﬁc response has to be given may depend on diffuse attention (Mikløsi et al., 1998), and this function of the Rhemi may be an aspect of processing spatial information. At least, it could be argued that diffuse attention would allow detection of stimulus change anywhere in the visual ﬁeld and effective response would require processing of spatial cues. As mentioned previously, schooling ﬁsh approach a predator more closely when they swim with their mirror image on their left side (Bisazza et al., 1999), and, even in the absence of a predator, they prefer to swim with their mirror image on their left side (De Santi et al., 2001). This preference for having a “conspeciﬁc” on the left side might affect the structure of a school. Since ﬁsh space themselves at rather precise distances from each other when

LATERALIZATION IN VERTEBRATES

137

they are in a school, the preference for swimming with a conspeciﬁc on the left could reﬂect a need to monitor the position of that ﬁsh precisely. The need to monitor spatial information to maintain the species-typical spacing between ﬁsh in a school may explain the preference for having a conspeciﬁc on the left side, since it would allow use of the LE and Rhemi. Escape responses inevitably involve processing of topographical cues to monitor the position of the predator and to enable effective movement away from the predator to a safer location. In fact, in teleost ﬁsh the direction of escape C-turns, driven by one of the Mauthner cells, is inﬂuenced by obstacles around the ﬁsh (Eaton and Emberly, 1991). Given the rapidity of the C-turn, it is likely that the spatial information used by the escaping ﬁsh has been acquired during regular monitoring of the topography before turning is elicited. There is no evidence associating this monitoring of topography to either one of the hemispheres, but it may be relevant to note that the preferential right-side turning of Girardinus falcatus as it approaches a predator (Cantalupo et al., 1995) permits monitoring of the spatial location of the predator relative to obstacles with the LE and Rhemi. D. NONPRIMATE MAMMALS Studies on rats have indicated that the Rhemi is involved in processing spatial information. Bianki (1981) tested rats on a discrimination task requiring them to locate the position of a spot of light using depth cues and found that this performance was impaired by causing spreading depression in the Rhemi but not in the Lhemi. In other words, the rat can determine depth provided that the Rhemi is functioning. Perhaps the best demonstration of the Rhemi specialization for spatial learning has been obtained by testing rats monocularly in the Morris water maze. This task requires a swimming rat to locate an escape platform submerged beneath the water’s surface using spatial cues only. These spatial cues are features in the testing room (i.e., distal to the water maze). Cowell et al. (1997) found that rats with the RE occluded, and so using their Rhemi, were able to ﬁnd the hidden escape platform but those with the LE occluded had great difﬁculty doing so. However, this lateralization was present only in rats that had been handled in early life. Handling involved taking each pup away from its mother and placing it in isolation for 3 min/day from birth to day 20 of life. Rats that had not received such handling in early life showed no difference between performance with the LE or the RE. When tested using their LE, nonhandled rats were unable to perform the task as well as handled rats, and, when tested using their RE, nonhandled rats performed better than handled rats. Hence, handling in early life unmasks the superior ability of the Rhemi for spatial navigation and, at the same time, decreases the ability of the Lhemi to use this information. Handling in early life also

138

LESLEY J. ROGERS

unmasks other forms of lateralization in rats (summarized in Bradshaw and Rogers, 1993), including lateralization in direction of moving off in openﬁeld tests (handled rats have a population bias to move leftward from their starting position; Sherman et al., 1980) and emotional behavior (Maier and Crowne, 1993). E. PRIMATES In humans the Rhemi, it seems, subserves a broader area of hemispace than does the Lhemi (Posner et al., 1987). Lesions of either hemisphere cause inattention to the contralateral side of the body, but inattention to the ipsilateral side of the body sometimes occurs and it is more frequent following a lesion of the Rhemi (Heilman, 1995). Thus, inattention to one side of the body or even to one-half of any object (i.e., hemineglect syndrome) is more likely to occur after damage to the Rhemi than to the Lhemi (Heilman et al., 1985). The Rhemi is said to direct attention to visual extrapersonal space, whereas the Lhemi directs attention to within the person’s body space (Heilman, 1995). These observations may reﬂect the role of the Rhemi in processing spatial information using distal cues. Some evidence indicates that primates may, likewise, have specialization of the Rhemi for spatial performance. For example, capuchins have a stronger left-hand preference when tested on haptic-visual tasks than they do on simple reaching tasks (Lacreuse and Fragaszy, 1999). This may be because the former task requires the integration of spatial and sensorimotor components of the task. Gorillas (Gorilla gorilla) and baboons (Papio papio) also have been found to use the left hand preferentially on spatial tasks requiring them to align transparent doors to obtain a food reward (Fagot and Vauclair, 1988a,b). Specialization of the Rhemi for spatial processing might be the reason why Rigamonti et al. (1998) found that left-handed pigtail macaques (Macaca nemestrina) have a higher speed of performance than right-handed ones on a task requiring them to search for small food rewards from a vertical array. Other indirect evidence points to specialization of the Rhemi for spatial processing. “Global processing” is an ability that may reﬂect spatial processing and it is a function of the Rhemi: Deruelle and Fagot (1997) tested baboons in an operant task in which they had to respond to a letter of the alphabet composed of smaller letters, which could be the same as, or different from, the larger letter. Attention to the larger letter was interpreted as “global precedence,” and this was associated with use of the Rhemi. Attention to the smaller letters indicated local attention and was associated with use of the Lhemi. Chimpanzees have a Rhemi advantage for locating a short line contained within a geometric ﬁgure (Hopkins and Morris, 1989), which suggests

LATERALIZATION IN VERTEBRATES

139

specialization of the Rhemi for spatial processing. The important feature of the task was the location of the line, not merely attention to the line itself. Extrapolating from the more extensive data on the processing of spatial information in the chick, the Rhemi is used for processing information about distant cues and the Lhemi for local cues. It is difﬁcult to say whether the preceding experiments required the primates to use distant or local cues, but, where an operant task was used that required the subjects to judge the distance between a line and a dot displayed on the same key (short vs. long distance discrimination), they can only have relied on local cues. As predicted, baboons and humans performed such a task better with the Lhemi (Depy ´ et al., 1998).

V. AGGRESSIVE RESPONSES Intraspeciﬁc agonistic displays and aggressive interactions are lateralized, as shown in a number of species, and direction of bias is to the left side. The Rhemi is, therefore, specialized for control of aggressive interactions and this may be associated with, or a manifestation of, its specialization for expressing intense emotions. It is a clear example of the specialization of the Rhemi for controlling species-typical patterns of behavior given rapidly in response to speciﬁc releasing stimuli. A. BIRDS The ﬁrst evidence for lateralized control of aggressive responses came from studies of chicks with glutamate injected into the left or right hemisphere on day 2 posthatching. Subsequent to this treatment, the chicks treated in the left hemisphere had elevated levels of attack, as well as copulation, responses when tested with standard hand-thrust tests (Howard et al., 1980). In fact, these chicks responded just as if they had been treated with testosterone (Andrew, 1966), whereas the chicks treated with glutamate in the right hemisphere showed no such elevation of attack or copulation. Although it was likely that the elevation of aggressive behavior reﬂected control by the Rhemi, resulting from impaired function of the Lhemi following the glutamate treatment, this was not certain until chicks were tested monocularly. First the chicks were administered testosterone to elevate their attack and copulation behavior, and then they were tested monocularly. Provided that they were tested using the LE, the chicks showed elevated attack and copulation, but, when using the RE, they had levels of attack and copulation no different from those of untreated controls (Rogers et al., 1985).

140

LESLEY J. ROGERS

The same lateralization of agonistic interactions has been found in adult hens (Rogers, 1991). The hens were tested with partial occlusion of the left or right eye (the frontal ﬁeld and at least half of the lateral ﬁeld was occluded). Social interactions in groups were scored and it was found that the agonistic interactions of hens with an unoccluded LE were elevated above control levels (hens with no occlusion of the eyes), whereas hens with an unoccluded RE had fewer agonistic interactions than controls. It is apparent, therefore, that elevated aggressive and sexual responses occur when the Rhemi is processing visual inputs and controlling responses, whereas control by the Lhemi suppresses the aggressive behavior. An experiment by Vallortigara et al. (2001) has found that agonistic interactions in chicks are directed leftward even in the binocular condition. Social pecking between pairs of chicks unfamiliar with each other was recorded on videotape and the angle of ﬁxation used by each chick just before it delivered a peck to the other was determined using frame-by-frame analysis of the videotape. It is typical for the chick to turn its head to ﬁxate its opponent brieﬂy, using its lateral monocular ﬁeld of vision, before pecking, mostly at a region of the head. Comparison of ﬁxations in which the chick used its left versus its right monocular ﬁeld revealed a bias to use the LE in the majority of chicks (Fig. 5). Therefore, consistent with control of aggression by the Rhemi, aggressive pecks are delivered with a bias for LE use even in chicks not treated with testosterone and tested monocularly. Nonaggressive pecks, given to features in the cage, showed no such lateral bias (Vallortigara et al., 2001), a result which demonstrates that the bias is not simply a lateralized motor response and that it relates speciﬁcally to aggressive pecks. Preferential use of the LE to direct aggressive pecks at a conspeciﬁc occurs in chicks exposed to light before hatching and in chicks hatched from eggs incubated in the dark but the latter group also has lateralized use of the frontal binocular ﬁeld (Vallortigara et al., 2001). Chicks hatched from eggs incubated in the dark show a signiﬁcant bias to use the right half of the frontal ﬁeld in preference to the left half of the frontal ﬁeld. In other words, they have complementary specialization of the left lateral ﬁeld and right frontal ﬁeld for controlling aggressive pecking. This complementary specialization is not seen in chicks exposed to light during incubation. This ﬁnding leads us to reexamine earlier work showing that light exposure of the embryo is essential for establishing lateralization for attack, copulation, and feeding responses. For reasons that are outlined later, exposure of the chick embryo to light for as little as 2 h during the last few days before hatching establishes the lateralization seen in monocular tests of pecking at grains against a background of pebbles (Rogers, 1997). Such light exposure also establishes the lateralization of attack and copulation responses revealed by treatment of the left or right hemisphere with glutamate (Rogers, 1982, 1990). However, we can now see that this result

LATERALIZATION IN VERTEBRATES

141

FIG. 5. The preferred eye (or side) for chicks and toads attacking a conspeciﬁc is illustrated. LE and RE indicate the left and right eyes (or sides), respectively. The data for chicks (Gallus gallus domesticus) were collected in tests with pairs of birds that had not seen each other previously. A bird was said to attack with its LE if it delivered more pecks to a conspeciﬁc after ﬁxating it with its left lateral ﬁeld of vision, and vice versa for being right eyed. Note that the majority of chicks were left eyed. (Data from Vallortigara et al., 2001.) The same method was used to collect the data shown for toads (Bufo marinus). (Data from Robins et al., 1998.)

may be due to the complementary specialization of the left lateral and right frontal visual ﬁelds in chicks hatched from eggs incubated in the dark. Hence light exposure of the embryo inﬂuences the development of lateralization, but lateralization still develops even when the eggs are not exposed to light. It is difﬁcult to say what the natural condition might be in Gallus gallus, but, since as little as 2 h of light exposure is sufﬁcient to alter lateralization (Rogers, 1982), it is likely that the eggs receive an equivalent exposure to light in the natural environment. At this stage of incubation, the hen frequently leaves the nest to feed and stands to move the eggs (discussed in Rogers, 1995). Whether or not this is the case, and it may vary between clutches, the left lateral visual ﬁeld would be used preferentially for aggressive pecking since this bias is independent of the amount of light exposure received by the embryos. B. AMPHIBIANS Toads direct more aggressive responses to conspeciﬁcs on their left side than on their right side, as do chicks. Robins et al. (1998) scored strikes at

142

LESLEY J. ROGERS

conspeciﬁcs given by Bufo marinus toads when feeding in a group. A tongue strike at another toad, often at its eye, was recorded as being left or right according to whether the point at which the tongue contacted the other toad’s body was in the attacker’s left or right visual hemiﬁeld. Individual toads were followed to allow comparison between them, and the bias in the group was also noted (Vallortigara et al., 1998). There was a signiﬁcant bias to strike in the left hemiﬁeld (of 22 toads tested, 17 had a left bias, 4 had a right bias, and 1 had no bias; Fig. 5). This is clear evidence for the control of attack strikes by the Rhemi, and it is in contrast to control of feeding strikes by the Lhemi of the toad (see earlier discussion). In addition to being more common, the leftward attack strikes by Bufo marinus have a more focused target than the rightward attack strikes. The leftward strikes hit one of the opponent’s eyes two to three times more often than do the rightward strikes (Robins et al., 1998). Rightward strikes hit a number of sites on the opponent’s body and appear to be directed away from the eyes. This result indicates that the leftward strikes may be more aggressive than the rightward strikes. C. REPTILES For the ﬁrst time in this review, it is possible to mention lateralization in a reptilian species. Deckel (1995, 1998) has shown that lizards (Anolis spp.) tend to direct their agonistic encounters leftward. Nonaggressive encounters show no such bias. Aggressive responses are, therefore, controlled by the Rhemi in lizards. The aggressive behavior of Anolis includes bobbing of the head, extension of the colored dewlap under the throat of males, changes in body posture, biting, and a lightening of skin color. Deckel and Jevitts (1997) reported the lightening of skin color as “greening,” compared to a brown skin color when aggression is low. Since the lizards have their eyes on the sides of their heads and because their snouts occlude the frontal visual ﬁeld, they view each other by using the left or right monocular ﬁeld. This means that they orient with the head, and often the body also, positioned sideways to the lizard at which they are directing their agonistic displays. Hence, from video recordings, Deckel could score which eye was used in agonistic displays. Not only was a LE preference found for such interactions but also the leftwarddirected aggressive responses of Anolis were accompanied by a greening of the lizard’s skin color. Nonaggressive, and nonlateralized, assertion displays were accompanied by a darker skin color. Since serotonin is known to darken the skin color of lizards (Deckel, 1995), the Rhemi of the lizard apparently has control of the neurochemical system involved in changing skin color, as well as behavioral responses associated with aggression.

LATERALIZATION IN VERTEBRATES

143

D. FISH Siamese ﬁghting ﬁsh (Betta splendens) are the obvious choice for investigating lateralized aggressive displays in ﬁsh. Cantalupo et al. (1996) scored the number and duration of lateral threat displays given by male Siamese ﬁghting ﬁsh to their image in a mirror. They found individual biases for both of these scores but no population bias. These same ﬁsh were tested 2 months later, but this time their lateral displays toward a female conspeciﬁc, not a mirror, were scored. Individual ﬁsh retained their same side bias for displaying and there was still no bias in the population. It seems, therefore, that the side bias for an individual is consistent over time and context but that there is no indication of Rhemi control of these aspects of agonistic (or sexual) displays. Further investigation measuring the viewing eye preferences of ﬁsh during agonistic displays might be more revealing. E. NONPRIMATE MAMMALS Jarman (1972) reported that more scars are found on the right side of the pelts of impalas than on the left side. These scars are most common in the neck and shoulder region and are most likely a reﬂection of lateralized intraspecies interactions than of interspecies attempts at predation. In fact, it is reasonable to assume that the injuries are inﬂicted during agonistic encounters in which the aggressor launches an attack to its left side. F. PRIMATES Casperd and Dunbar (1996) found that gelada baboons (Theropithecus gelada) direct more agonistic responses toward conspeciﬁcs on their left side than on their right side. In other words, they were more likely to respond rapidly to direct aggression at a conspeciﬁc sighted in the left peripheral ﬁeld of vision, indicating control by the Rhemi. Here it is interesting to note the observation of Drews (1996) that pelts of wild savanna baboon, Papio cynocephalus, have more scars on the right side of the head region than on the left side. This is similar to Jarman’s earlier report of more scars on the right side of the pelts of impalas. In baboons also, the lateralized distribution of scars may result from intraspecies agonistic encounters and the bias is likely to reﬂect leftward-directed attacks by the aggressor. The inhibition of emotions by the Lhemi is sometimes suggested in the case of humans (e.g., Nestor and Safer, 1990) and, therefore, impairment of the Lhemi’s inhibition results in the expression of more intense emotions (discussed earlier) and, sometimes, in aggression. Subjects with epileptic seizures focused in the left temporal lobe (and so causing impaired Lhemi

144

LESLEY J. ROGERS

function) have been reported to express higher than average levels of hostile feelings (Devinsky et al., 1994). Also, reduced activity in posterior regions of the Lhemi has been associated with suicidal and aggressive behavior (Graae et al., 1996). These suggested associations between the activity of the Rhemi and aggression in humans are interesting in light of the data for other species showing the same association but they are made cautiously here. Considerably more research is needed to conﬁrm these initial observations.

VI. ADVANTAGES AND DISADVANTAGES OF BEING LATERALIZED The examples given earlier demonstrate that vertebrates share a common pattern of lateralization that is present in the central nervous system and frequently manifested as side biases in responding. These side biases in responding could be disadvantageous as, for example, in the case of an animal that tends to respond to prey on its right side but to ignore them on its left side. An animal that is more likely to attack an opponent on its left side but ignore one on its right side may also be at a disadvantage. In fact, the disadvantage of being lateralized at a population level seems most evident in the left-side bias of responding to predators. Animals that may be symmetrical in physical features have, for some reason, retained consistent lateralization of behavior, and this is the case for vertebrates from ﬁsh to humans, although some species may diverge from the rule at least in some forms of lateralization. The persistence of lateralization across vertebrate species and, presumably, over time leads one to conclude that brain lateralization has advantages (Rogers, 2000) even in cases in which the animal has to adopt particular behavior patterns to counteract centrally controlled side biases. The kookaburra, for example, turns its head to position its LE for scanning directly below its perch. Nevertheless, toads, chicks, ﬁsh, and primates tested in seminatural conditions display side biases that might be to their detriment unless additional beneﬁts are conferred by having a lateralized brain. A number of potential advantages of being lateralized are discussed later, beginning with the hypothesis that differential specialization of the hemispheres enhances cognitive capacity. A. ENHANCED COGNITIVE CAPACITY AND EFFICIENCY One way to increase a brain’s capacity to carry out simultaneous processing is to channel or ﬁlter the different types of input and/or output so that processing takes place in parallel in different regions of the brain (Deng and Rogers, 1998a). Neurophysiologists have concentrated on how this takes place in a vertical sense, from one level of brain organization to another,

LATERALIZATION IN VERTEBRATES

145

but mostly they have ignored ﬁltering that occurs laterally at the same levels of neural organization. Generally, it has been assumed that differential specialization of the hemispheres enhances neural and cognitive capacity (Dunaif-Hattis, 1984; Gazzaniga and Le Doux, 1978) but little evidence in support of this hypothesis has, so far, been provided. There is an obvious way to conduct behavioral tests to assess the potential advantage of being lateralized at an individual level. This is to compare the performance of strongly lateralized individuals with that of weakly lateralized individuals. The chick presents itself as a model for such experimentation since it is possible to produce chicks that are weakly or strongly lateralized for attack, copulation, and feeding responses (see later), although they are still strongly lateralized for other responses in the visual as well as the other sensory modalities (Rogers, 1982, 1997; Rogers et al., 1998). As mentioned earlier, chicks hatched from eggs exposed to light just prior to hatching are strongly lateralized for attack and feeding responses, but chicks hatched from eggs incubated in the dark are not lateralized for control of feeding responses and the specialization of their left lateral visual ﬁeld for aggressive pecking is balanced by specialization of their right frontal ﬁeld for aggressive pecking. Light exposure affects the development of lateralization because the embryo is turned in the egg so that it occludes its left eye, whereas light entering through the shell can stimulate the right eye (Rogers, 1990). The light stimulation enhances the development of visual pathways receiving input from the right eye and projecting to the forebrain (Rogers, 1996) and, as a consequence, the thalamofugal visual projections that cross the midline of the brain become asymmetrical (Rogers and Deng, 1999). There are more projections from the left side of the thalamus to the forebrain than from the right side to the forebrain (Fig. 2). Following incubation in the dark, no asymmetry develops in these visual projections (Rogers and Bolden, 1991), and, after hatching, the chicks show no lateralization of attack or feeding responses as revealed by treatment of the left or right hemisphere with glutamate (Rogers, 1990, 1997). However, dark-incubated chicks tested without treatment still direct more pecks at an unfamiliar conspeciﬁc in the left visual ﬁeld compared to the right (Vallortigara et al., 2001). The data for feeding responses show clearly that, after incubation in the dark, there is no lateralization at either the group or the individual level (Rogers, 1997). Hence, dark-incubated chicks are less lateralized than light-incubated ones but the difference is a matter of degree, rather than being absolute, and it is limited to certain visual functions. Despite this qualiﬁcation, it has still been fruitful to test the hypothesis relating lateralization to enhanced cognitive capacity by comparing dark-incubated and light-exposed chicks. Dark-incubated chicks perform less successfully than light-incubated chicks in a task requiring dual attention to entirely different categories of stimuli (Rogers, 2000). The task tested the chick’s vigilance for a simulated

146

LESLEY J. ROGERS

aerial predator while the chick was pecking at grain and mealworms on the ﬂoor. This placed demands on two known and opposite lateralizations of the chick. The ﬁrst was the use of the RE (Lhemi) for pecking to feed, and the second was use of the LE (Rhemi) to monitor overhead for an aerial predator (Evans et al., 1993; discussed earlier). Thus, a chick feeding in a situation in which it might expect a predator to appear overhead is expected to use the Rhemi to monitor overhead and the Lhemi to control feeding pecks. Dark-incubated chicks might ﬁnd this channeling of tasks to the separate hemispheres more difﬁcult, or impossible, and be less efﬁcient at detecting the overhead predator. The results indicated that this was, in fact, the case: When the predator approached from the chick’s left side, lightexposed chicks detected it sooner than did dark-incubated chicks (Rogers, 2000). The chicks were tested in the second week posthatching and after several hours of food deprivation (Rogers, 2000). Once a chick had started to peck at grains and mealworms, a model predator, shaped to resemble a raptor, was passed overhead. The latency to detect the predator stimulus was scored, as indicated by the chick ceasing to peck and, usually, giving a startle trill and a twitch of the head. The light-exposed chicks detected the simulated predator after a very short latency when it advanced on their left side (LE) and with a signiﬁcantly longer latency when it advanced on their right side (RE). This lateralization was expected. The dark-incubated chicks showed no such lateralization and their latency scores were between those of the LE and RE scores for the lateralized group of chicks. Their latency scores were both more variable and longer than those of the light-incubated chicks using their LE. Therefore, in the dark-incubated chicks, the Rhemi fails to show an advantage in detecting the stimulus. Controls showed that these results were due neither to differences in the motivation to feed in darkincubated and light-exposed chicks nor to differences in fear responses to the simulated predator (Rogers, 2000). The results demonstrate enhanced vigilance in light-incubated chicks, and this could confer an advantage in the natural environment, even though it is limited to the LE only. A somewhat similar result has been found for cats tested on a task requiring them to track a moving spot of light using one or the other forepaw. Fabre-Thorpe et al. (1993) compared the reaction times of cats with and without a preference to use one paw over the other when performing the task and found that those with lateralization of paw preference had a shorter reaction time to commence tracking than those without such lateralization. The enhanced ability of the lateralized cats appears to depend on channeling responses to one or the other forelimb and paw (and thus to control by one hemisphere). In this case, it is lateralization of motor output that enhances performance, rather than perceptual input as is the case with chicks.

LATERALIZATION IN VERTEBRATES

147

In the cats, enhanced ability appears to be attained by specializing each paw to perform a different function, one for tracking the moving stimulus and the other for supporting and balancing the body. A similar result has been found in capuchin monkeys (Cebus appella); individuals with a greater degree of hand preference were found to solve a problem with shorter latency than individuals with a lesser degree of hand preference (Fragaszy and Mitchell, 1990). Another similar result has been found in wild chimpanzees foraging for termites. Both hands are used by chimpanzees in termite ﬁshing, one to hold the twig used as a probe and the other to act as a stabilizer across which the twig covered in termites is rubbed when the chimpanzee eats them. Some individuals always use the same hand to probe and the other to stabilize the twig, whereas others vary which hand is used for either purpose. McGrew and Marchant (1999) studied the efﬁciency of termite ﬁshing by the chimpanzees (Pan troglodytes schweinfurthii) at Gombe and found that completely lateralized chimpanzees gathered more prey for a given amount of effort than did incompletely lateralized chimpanzees. A correlation between efﬁciency of response and degree of lateralization has been shown also in the pigeon. Gunt ¨ urk ¨ un ¨ (2000) tested pigeons on a task requiring them to discriminate grain from grit, a task which they perform better with the RE than the LE when tested monocularly (Gunt ¨ urk ¨ un ¨ and Kesch, 1987), as do chicks in similar situations. The researchers ﬁrst determined the LE–RE difference in performance for each bird and used that as an index of their degree of lateralization. They then tested the birds binocularly to see whether there was an association between their efﬁciency on the task and the degree of lateralization. There was a signiﬁcant correlation: The stronger the lateralization, the better the visually guided foraging. This result substantiates the preceding results for chicks. It is also worth noting that in pigeons, as in chicks, visual lateralization depends on exposure of the developing embryo to light (Gunt ¨ urk ¨ un, ¨ 1993), although the lightdependent asymmetry in pigeons is located in the tectofugal visual system, whereas in chicks it is in the thalamofugal visual system (Gunt ¨ urk ¨ un ¨ et al., 1998). The positive association between degree of lateralization in the avian visual system and efﬁciency of predator detection and foraging deserves consideration in an ecological context, especially if it extends to other avian species. In the natural environment, eggs may receive varying amounts of exposure to light and, depending on the embryo’s stage of development at the time of receiving exposure to light, individual differences in both the degree of lateralization and efﬁciency of performance may develop. It is known for the chick that as little as 2 h of light-exposure is sufﬁcient to establish lateralization but it must occur during a sensitive period during the last stages before hatching (Rogers, 1982, 1990; Zappia and Rogers, 1983).

148

LESLEY J. ROGERS

Added to this, the level of steroid hormones in the embryo inﬂuences its sensitivity to the effects of light: High levels of the sex steroids decrease the effect of the light exposure (Rajendra and Rogers, 1993; Schwarz and Rogers, 1992) and so too do high levels of corticosterone (Deng and Rogers, 2001). The level of these hormones circulating in the embryo depends not only on their secretion by the embryo itself but also on the amount that the hen deposits in the egg before it is laid (Schwabl, 1997). In fact, the order of laying affects the amount of testosterone deposited in the egg, either increasing it or decreasing it depending on the species (Schwabl, 1993, 1996, 1998; Gil et al., 1999). In addition, stress of the hen may affect the amount of corticosterone so deposited. Therefore, there are individual differences in light-sensitivity during development of the embryo and subsequent differences in the degree of visual lateralization. If so, individual differences in predator detection and foraging efﬁciency may depend, at least to some extent, on early developmental processes. Of course, there are direct effects of the steroid hormones on behavior after hatching (e.g., competitive or aggressive behavior; Schwabl, 1993, 1996) but, in addition to the classical effects of steroid hormones on behavior, patterns of behavior associated with lateralization of the visual pathways, at least, may be affected. The examples discussed in this section suggest that lateralization may be beneﬁcial for an individual, although this may be context dependent. It is conceivable that not so strongly lateralized individuals may perform better than more lateralized ones in tasks with different cognitive demands. The advantages of lateralization discussed so far, however, operate at an individual level and are just as effective regardless of the direction in which the lateralization is present. They do not, therefore, provide an explanation for population lateralization. B. POPULATION LATERALIZATION: ADVANTAGE IN SOCIAL SITUATIONS? Population lateralization is likely to exist when there is a need to maintain coordination among individuals, that is, in a social context. In populations in which the majority, if not all, of the individuals are lateralized in the same direction, each individual could rely on the predictable lateralization of other members of the group and this could be used to a social advantage. For example, an individual might lower its chances of being attacked by approaching a higher ranking conspeciﬁc on that individual’s right side. Such behavior might be particularly important for avoiding aggression and maintaining the social hierarchy. Rogers and Workman (1989) reported that, in the ﬁrst 2 weeks of life, dark-incubated chicks form less stable hierarchies than do chicks exposed to light just before hatching. Six groups of eight chicks hatched from eggs

LATERALIZATION IN VERTEBRATES

149

exposed to light during the ﬁnal days before hatching (i.e., lateralized at a population level for the visual functions mentioned already) were compared to six groups of chicks incubated in the dark (i.e., not lateralized at a population level for those same visual functions). A rigid (or stable) hierarchy was maintained in the groups that had received exposure to light, as indicated by the fact that the lowest ranking chicks in these groups rarely gained access to the food dish. The lowest ranking chicks in the groups hatched from eggs incubated in the dark gained access to the food dish more often than did their light-exposed counterparts, and their entry scores varied more from day to day. Another possible advantage of population lateralization might be maintaining cohesion of a ﬂock, a shoal, or a herd. If all members of the group have the same bias to turn, say, to the right when they see a predator, the integrity of the group will be maintained during escape avoidance. There is evidence in support of this hypothesis. Bisazza et al. (2000) have shown that schooling in ﬁsh is associated with a population bias to turn in one direction (either left or right) when faced by a barrier of vertical bars through which the ﬁsh could see a model that simulated a predator. Each ﬁsh was tested singly and it had to swim down a corridor toward the simulated predator placed behind the barrier. On reaching the barrier, the ﬁsh turned right or left and this was recorded. Each ﬁsh was given 10 such trials and from this score an index of lateralization was calculated. Next the social behavior of the species was determined in terms of tendency to school. Groups of ﬁsh were placed in a tank together and an index of their proximity to each other was determined. Six species were found to school and all six were lateralized for turning bias at the population level. Ten species were found to be nonschooling and six of these were not lateralized at the population level. The species not lateralized at a population level were, however, composed of individuals that were lateralized. It seems possible, therefore, that a group bias for the majority of individuals to turn in the same direction may be maintained to advantage in those species that school. A population lateralization for turning would maintain the integrity of the school on contact with predators or other features of the environment that evoke a change in the direction of swimming. This would operate effectively provided that the school is moving in one direction or, say, facing upstream so that all of the individuals are aligned in the same direction. Once the school remains stationary with the individuals facing in different directions (e.g., when the school has been cornered), this mechanism would no longer be effective and a predator might exploit this to dissipate the school, a tactic used by yellowtail (Seriola lalandei) ﬁsh when they prey on schooling jack mackerel (Trachurus symmetricus) (Dugatkin, 1997).

150

LESLEY J. ROGERS

C. LATERALIZATION AND IMMUNE RESPONSES Immune responses differ according to whether they are under the control of the left or the right hemisphere. Mice with a lesion in the right parietooccipital lobe of the neocortex show depressed mitogen-induced lymphocyte proliferation and enhanced antibody production, whereas mice with an equivalent lesion placed in the left parieto-occipital lobe show no modiﬁcation of these immune responses (Barneoud ´ et al., 1987). Another study found that lesioning of the left neocortex depressed T-cell functions, whereas right side lesioning either enhanced T-cell functions or had no effect (Nevue, 1988). Also, individual differences in turning bias have been associated with different immune responses: Rats which circle to the left have higher indices of lymphocyte stimulation than those that circle to the right (Nevue, 1988). Paw preference in mice is also associated with immune responses, almost certainly because it reﬂects preferential activation of one or the other hemisphere. Nevue et al. (1991) reported that left-pawed mice (determined for reaching into a tube to obtain food) have higher mitogen-induced T-lymphocyte proliferation than right-pawed mice. Ablation of the left cortex abolished the difference in T-cell function between left- and right-pawed mice whereas ablation of the right hemisphere had no effect on this difference. In humans also, an association between hand preference and immune reactivity has been suggested, and there is some evidence, albeit not wellsubstantiated, indicating that left-handers have a higher incidence of immune disorders (Geschwind and Behan, 1982). Although these data for humans need conﬁrmation in well-controlled conditions, which are difﬁcult to achieve in studies of humans (Peters, 1995), considered together with the research on paw-preference and immune responses, the evidence makes a reasonably strong case for an association between hand preference and immune responses. In my opinion, it is worth speculating that in some populations and some environments, there has been direct selection of a particular type of immune response, and that limb preference and other forms of lateralization are a secondary aspect of this. Some populations might have been exposed to conditions favoring the immune responses controlled by the left hemisphere and others to conditions favoring those of the right hemisphere. Such differing immune types would be associated with differences in hemispheric activity at a population level. Hence, it is possible that some types of lateralization, including handedness, may occur because they are associated with particular types of immune response. The association between paw/hand preference and immune responsiveness in mammals has laid a strong basis for investigating similar associations

LATERALIZATION IN VERTEBRATES

151

between preferential limb use and the immune system in birds, amphibians, and even ﬁsh. Right-pawedness has been reported in laboratory studies of three species of toads (Bisazza et al., 1996, 1997b), and Litoria spp. exhibit side biases (Fig. 4). There may be a number of reasons for the population bias but selection for a particular type of immune response presents itself as a possibility worth investigating in these and other species. VII. SUMMARY A pattern of hemispheric specialization common to all vertebrates has been presented (summarized in Fig. 6). It has been considered, as far as possible, in terms of behavior patterns that are important in the natural environment. Responses in foraging and prey capture are typical specializations of the left hemisphere and they are often apparent as rightward-directed responses. Responses to novel stimuli and potential predators are specializations of the right hemisphere and responses to these stimuli are directed preferentially to the animal’s left side. The right hemisphere is also specialized for the expression of fear and aggression, and to process and memorize topographical information, essential in association with predator-escape responses. The baseline differences between the hemispheres are (1) rapid and

FIG. 6. Summary of the general pattern of lateralization of the hemispheres.

152

LESLEY J. ROGERS

species-typical responses being under the control of the right hemisphere and (2) responses requiring considered decision making being under control of the left hemisphere. Side biases in responding to prey to the right side and agonistic responses delivered to conspeciﬁcs on the left side occur not only in species with eyes placed on the sides of their head and with only a small binocular ﬁeld of vision, but also in species with smaller lateral ﬁelds and with a large binocular overlap. At least in one case, that of prey striking in toads, there is clear evidence that the side bias occurs within the binocular ﬁeld and manifests itself as a hemiﬁeld neglect of prey to the left side of the toad’s midline. Chicks also, provided they have been hatched from eggs incubated in the dark, have a bias within the binocular ﬁeld: They show a preference to ﬁxate with the right half of the binocular ﬁeld before delivering an aggressive peck. It should be noted, however, that the majority of aggressive pecks made by chicks are delivered after use of the left, monocular ﬁeld of vision in dark-incubated chicks as well as those exposed to light before hatching. Lateralized use of the visual hemiﬁelds deserves further investigation in a range of species. Side biases in responding are not easily associated with any evolutionary advantage but they may be balanced against other advantages of lateralization. Lateralization may enhance cognitive abilities and more efﬁcient responding in some species and contexts. It may also be associated with specialized functioning of the immune system selected on the basis of past experience of the species in terms of physiological demand on the immune system, although this hypothesis has yet to be tested. Finally, population bias for the majority of individuals to be lateralized in the same direction is considered as a potential beneﬁt to group cohesion, as in schooling ﬁsh, and a mechanism by which individuals can interact in predictable ways within a social group. For example, the bias for directing attack responses leftward may be used by an approaching animal to avoid an aggressive encounter; approach on the potential aggressor’s right side would minimize the chances of attack. Until now, most examples of lateralization have come from experimentation in the laboratory or using captive animals. There are relatively few examples demonstrating lateralization in wild populations, despite the obvious implications of lateralization for survival of species and individuals. Those lateralizations found in species in their natural habitat have been mentioned. They highlight the promise of gaining a better understanding of behavior of animals in the wild if lateralization is taken into account as a characteristic of all vertebrates. Acknowledgment Financial support was provided by a Special Investigator Award to the author from the Australian Research Council.

LATERALIZATION IN VERTEBRATES

153

References Adret, P., and Rogers, L. J. (1989). Sex difference in the visual projections of young chicks: A quantitative study of the thalamofugal pathway. Brain Res. 478, 59–73. Alonso, Y. (1998). Lateralization of visual guided behaviour during feeding in zebra ﬁnches (Taeniopygia guttata). Behav. Processes 43, 257–263. Andrew, R. J. (1966). Precocious adult behaviour in the young chick. Anim. Behav. 14, 485–580. Andrew, R. J. (1983). Lateralization of emotional and cognitive function in higher vertebrates, with special reference to the domestic chick. In “Advances in Vertebrate Neuroethology” (J.-P. Ewert, R. R. Capranica, and D. Ingle, Eds.), pp. 477–509. Plenum Press, New York. Andrew, R. J. (1991). The nature of behavioural lateralization in the chick. In “Neural and Behavioural Plasticity: The Use of the Domestic Chick as a Model” (R. J. Andrew, Ed.), pp. 536–554. Oxford Univ. Press, Oxford. Andrew, R. J. (2001). The earliest origins and subsequent evolution of lateralization. In “Comparative Vertebrate Lateralization” (L. J. Rogers and R. J. Andrew, Eds.). Cambridge Univ. Press, New York. In press. Andrew, R. J., and Brennan, A. (1983). Lateralization of fear behaviour in the male domestic chick: A developmental study. Anim. Behav. 31, 1166–1176. Andrew, R. J., and Rogers, L. J. (2001). The nature of lateralization in tetrapods. In “Comparative Vertebrate Lateralization” (L. J. Rogers and R. J. Andrew, Eds.). Cambridge Univ. Press, New York. In press. Andrew, R. J., Tommasi, L., and Ford, N. (2000). Motor control by vision and the evolution of cerebral lateralization. Brain and Language 73, 220–235. Barneoud, ´ P., Nevue, P. J., Vitiello, S., and LeMoal, M. (1987). Functional heterogeneity of the right and left cerebral neocortex in the modulation of the immune system. Physiol. Behav. 41, 525–530. Bianki, V. L. (1981). Lateralization of functions in the animal brain. Int. J. Neurosci. 15, 37–47. Bianki, V. L. (1988). The Right and Left Hemispheres of the Animal Brain: Cerebral Lateralisation of Function. In “Monographs in Neuroscience,” Vol. 3. Gordon & Breach, New York. Bisazza, A., Cantalupo, C., Robins, A., Rogers, L. J., and Vallortigara, G. (1996). Right pawedness in toads. Nature 379, 408. Bisazza, A., Cantalupo, C., and Vallortigara, G. (1997a). Lateral asymmetries during escape behavior in species of teleost ﬁsh (Jenynsia lineata). Physiol. Behav. 61, 31–35. Bisazza, A., Cantalupo, C., Robins, A., Rogers, L. J., and Vallortigara, G. (1997b). Pawedness and motor asymmetries in toads. Laterality 2(1), 49–64. Bisazza, A., Rogers, L. J., and Vallortigara, G. (1998a). The origins of cerebral asymmetry: A review of evidence of behavioral and brain lateralization in ﬁshes, reptiles and amphibians. Neurosci. Biobehav. Rev. 22, 411–426. Bisazza, A., Facchin, L., Pignatti, R., and Vallortigara, G. (1998b). Lateralization of detour behaviour in Poeciliid ﬁshes: The effect of species, gender and sexual motivation. Behav. Brain Res. 91, 157–164. Bisazza, A., De Santi, A., and Vallortigara, G. (1999). Laterality and cooperation: Mosquitoﬁsh move closer to a predator when the companion is on their left side. Anim. Behav. 57, 1145–1149. Bisazza, A., Cantalupo, C., Capocchiano, M., and Vallortigara, G. (2000). Population lateralization and social behaviour: A study with sixteen species of ﬁsh. Laterality 5, 269–284. Bisazza, A., Lippolis, G., and Vallortigara, G. (2001). Lateralization of ventral ﬁn use during object exploration in the blue gourami (Trichogaster trichopterus). Physiol. Behav. 72, 575–578.

154

LESLEY J. ROGERS

Box, H. O. (1977). Observations on spontaneous hand use in the common marmoset (Callithrix jacchus). Primates 18, 395–400. Bradshaw, J. L., and Rogers, L. J. (1993). “The Evolution of Lateral Asymmetries, Language, Tool Use, and Intellect.” Academic Press, San Diego. Bulman-Fleming, M. B., Bryden, M. P., and Rogers, T. T. (1997). Mouse paw preference: Effects of variations in testing protocol. Behav. Brain Res. 86, 79–87. Cameron, R., and Rogers, L. J. (1999). Hand preference of the common marmoset (Callithrix jacchus): Problem solving and responses in a novel setting. J. Comp. Psychol. 113, 149–157. Canﬁeld, J. G., and Rose, G. J. (1993). Activation of Mauthner neurons during prey capture. J. Comp. Physiol. A 172, 611–618. Cantalupo, C., Bisazza, A., and Vallortigara, G. (1995). Lateralization of predator-evasion response in a teleost ﬁsh (Girardinus falcatus: Poeciliidae). Neuropsychologia 33, 1637–1646. Cantalupo, C., Bisazza, A., and Vallortigara, G. (1996). Lateralization of displaying aggressive and courtship behavior in the Siamese ﬁghting ﬁsh (Betta splendens). Physiol. Behav. 60, 249–252. Casperd, J. M., and Dunbar, R. I. M (1996). Asymmetries in the visual processing of emotional cues during agonistic interactions in gelada baboons. Behav. Processes 37, 57–65. Clayton, N. S. (1993). Lateralization of unilateral transfer of spatial memory in marsh tits. J. Comp. Physiol. A 171, 799–806. Clayton, N. S., and Krebs, J. R. (1993). Lateralization in Paridae: Comparison of a storing and nonstoring species on a one-trial associative memory task. J. Comp. Physiol. A 171, 807–815. Clayton, N. S., and Krebs, J. R. (1994). Memory for spatial and object-speciﬁc cues in foodstoring and non-storing birds. J. Comp. Physiol. A 174, 371–379. Collins, R. L. (1985). On the inheritance of direction and degree of asymmetry. In “Cerebral Lateralization in Nonhuman Species” (S. D. Glick, Ed.), pp. 41–71. Academic Press, New York. Collins, R. L., Sargent, E. E., and Neumann, P. E. (1993). Genetic and behavioral tests of the McManus hypothesis relating response to selection for lateralization of handedness in mice to degree of heterozygosity. Behav. Genetics 23, 413–421. Corballis, M. C. (1991). “The Lopsided Ape.” Oxford Univ. Press, New York. Cowell, P. E., Waters, N. S., and Denenberg, V. H. (1997). The effects of early environment on the development of functional laterality in Morris maze performance. Laterality 2, 221–232. Dassonville, P., Zhu, X. H., Ugurbil, K., Kim, S. G., and Ashe, J. (1997). Functional activation in motor cortex reﬂects the direction and the degree of handedness. Proc. Natl. Acad. Sci. U.S.A. 94, 14015–14018. Davidson, R. J. (1992). Emotion and effective style. Psychol. Sci. 3, 39–43. Davidson, R. J. (1995). Cerebral asymmetry, emotion, and affective style. In “Brain Asymmetry” (R. J. Davidson and K. Hugdahl Eds.), pp. 361–387. MIT Press, Cambridge, MA. Deckel, A. W. (1995). Laterality of aggressive responses in Anolis. J. Exp. Zool. 272, 194–200. Deckel, A. W. (1998). Hemispheric control of territorial aggression in Anolis carolinensis: Effects of mild stress. Brain Behav. Evol. 51, 33–39. Deckel, A. W., and Jevitts, E. (1997). Left- vs. right-hemisphere regulation of aggressive behaviors in Anolis carolinensis: Effects of eye-patching and ﬂuoxetine administration. J. Exp. Zool. 278, 9–21. Denenberg, V. H. (1981). Hemispheric lateralization in animals and the effects of early experience. Behav. Brain Sci. 4, 1–49. Denenberg, V. H., Hofmann, M., Garbonati, J. A., Sherman, G. F., Rosen, G. D., and Yutzey, D. A. (1980). Handling in infancy, taste aversion, and brain laterality in rats. Brain Res. 200, 123–133.

LATERALIZATION IN VERTEBRATES

155

Deng, C., and Rogers, L. J. (1997). Differential contributions of the two visual pathways to functional lateralization in chicks. Behav. Brain Res. 87, 173–182. Deng, C., and Rogers, L. J. (1998a). Bilaterally projecting neurones in the two visual pathways of the chick. Brain Res. 794, 281–290. Deng, C., and Rogers, L. J. (1998b). Organisation of the tecto-rotundal and SP/IPS-rotundal projections in the chick. J. Comp. Neurol. 394, 171–185. Deng, C., and Rogers, L. J. (2001). Factors affecting the development of lateralization in chicks. In “Comparative Vertebrate Lateralization” (L. J. Rogers and R. J. Andrew, Eds.), Cambridge Univ. Press, New York. In press.. Depy, ´ D., Fagot, J., and Vauclair, J. (1998). Comparative assessment of distance processing and hemispheric specialisation in humans and baboons (Papio papio). Brain Cogn. 38, 165– 182. Deruelle, C., and Fagot, J. (1997). Hemispheric lateralisation and global precedence effects in the processing of visual stimuli by humans and baboons (Papio papio). Laterality 2, 233–246. De Santi, A., Bisazza, A., Cappelletti, M., and Vallortigara, G. (2000). Prior exposure to a predator inﬂuences lateralization of cooperative predator inspection in the guppy, Poecilia reticulata. Ital. J. Zool. 67, 175–178. De Santi, A., Sovrano, V. A., Bisazza, A., and Vallortigara, G. (2001). Mosquitoﬁsh display differential left- and right-eye use during mirror image scrutiny and the predator inspection responses. Anim. Behav. 61, 305–310. Devinsky, O., Ronsaville, D., Cox, C., Witt, E., Fedio, P., and Theodore, W. H. (1994). Interictal aggression in epilepsy: The Buss-Durkee hostility inventory. Epilepsia 35, 585–590. Diekamp, B., Prior, H., and Gunt ¨ urk ¨ un, ¨ O. (1999). Lateralization of serial color reversal learning in pigeons (Columba livia). Anim. Cogn. 2, 187–196. Dill, L. M. (1977). “Handedness” in the Paciﬁc tree frog (Hyla regilla). Can. J. Zool. 55, 1926– 1929. Dimond, S. J., Farrington, L., and Johnson, P. (1976). Differing emotional response from right and left hemispheres. Nature 261, 690–692. Drews, C. (1996). Context and patterns of injuries in free-ranging male baboons (Papio cynocephalus). Behavior 133, 443–474. Dugatkin, L. A. (1997). “Cooperation among Animals: An Evolutionary Perspective.” Oxford Univ. Press, Oxford. Dunaif-Hattis, J. (1984). “Doubling the Brain.” Peter Lang, New York. Eaton, R. C., and Emberly, D. S. (1991). How stimulus direction determines the trajectory of the Mauthner-initiated escape response in a teleost ﬁsh. J. Exp. Biol. 161, 469–487. Evans, C. S., Evans, L., and Marler, P. (1993). On the meaning of alarm calls: Functional references in an avian vocal system. Anim. Behav. 46, 23–28. Fabre-Thorpe, M., Fagot, J., Lorincz, E., Levesque, F., and Vauclair, J. (1993). Laterality in cats: Paw preference and performance in a visuomotor activity. Cortex 29, 15–24. Facchin, L., Bisazza, A., and Vallortigara, G. (1999). What causes lateralization of detour behavior in ﬁsh? Evidence for asymmetries in eye use. Behav. Brain Res. 103, 229–234. Fagot, J., and Vauclair, J. (1988a). Handedness and manual specialisation in the baboon. Neuropsychologia 26, 795–804. Fagot, J., and Vauclair, J. (1988b). Handedness and bimanual coordination in the lowland gorilla. Brain Behav. Evol. 32, 89–95. Fersen, von L., and Gunt ¨ urk ¨ un, ¨ O. (1990). Visual memory lateralization in pigeons. Neuropsychologia 28, 1–7. Fragaszy, D. M., and Mitchell, S. R. (1990). Hand preference and performance on unimanual and bimanual tasks in capuchin monkeys (Cebus apella). J. Comp. Psychol. 104, 275–282.

156

LESLEY J. ROGERS

Gaston, K. E., and Gaston, M. G. (1984). Unilateral memory after binocular discrimination training: Left hemisphere dominance in the chick? Brain Res. 303, 190–193. Gazzaniga, M. S., and Le Doux, J. E. (1978). “The Integrated Mind.” Plenum Press, New York. Geschwind, N., and Behan, P. (1982). Left-handedness: Association with immune disease, migraine, and development of learning disorders. Proc. Natl. Acad. Sci. U.S.A. 79, 5097– 5100. Gil, D., Graves, J., Hazon, N., and Wells, A. (1999). Male attractiveness and differential testosterone investment in zebra ﬁnch eggs. Science 286, 126–128. Graae, F., Tenke, C., Bruder, G., Rotheram, M.-J., Piacentini, D. C.-B., Leite, P., and Towey, J. (1996). Abnormality of EEG alpha asymmetry in female adolescent suicide attempters. Biol. Psychiatr. 40, 706–713. Gruzelier, J. H., and Doig, A. (1996). The factorial structure of schizotypy: Part II. Cognitive asymmetry, arousal, handedness, and sex. Schizophrenia Bull. 22, 621–634. Guiard, Y., Diaz, G., and Beaubaton, D. (1983). Left hand advantage in right handers for spatial constant error: Preliminary evidence in a manual ballistic aimed movement. Neuropsychologia 21, 111–115. Gunt ¨ urk ¨ un, ¨ O. (1985). Lateralization of visually controlled behavior in pigeons. Physiol. Behav. 34, 575–577. Gunt ¨ urk ¨ un, ¨ O. (1993). The ontogeny of visual lateralization in pigeons. German J. Psychol. 17, 276–287. Gunt ¨ urk ¨ un, ¨ O. (2000). Asymmetry pays: Visual lateralization improves discrimination success in pigeons. Curr. Biol. 10, 1079–1081. Gunt ¨ urk ¨ un, ¨ O., and Hahmann, U. (1994). Visual acuity and hemispheric asymmetries in pigeons. Behav. Brain Res. 60, 171–175. Gunt ¨ urk ¨ un, ¨ O., and Kesch, S. (1987). Visual lateralization during feeding in pigeons. Behav. Neurosci. 101, 433–435. Gunt ¨ urk ¨ un, ¨ O., Hellmann, B., Melsbach, G., and Prior, H. (1998). Asymmetries of representation in the visual system of pigeons. Neuroreport 9, 4127–4130. Harris, L. J. (1989). Footedness in parrots: Three centuries of research, theory, and mere surmise. Can. J. Psychol. 43, 369–396. Hauser, M. D. (1993). Right hemisphere dominance for the production of facial expression in monkeys. Science 261, 475–477. Hauser, M. D., and Akre, K. (2001). Asymmetries in the timing of facial and vocal expressions by rhesus monkeys: Implications for hemispheric specialization. Anim. Behav. 61, 391–400. Heilman, K. M. (1995). Attentional asymmetries. In “Brain Asymmetry” (R. J. Davidson and K. Hugdahl, Eds.), pp. 217–234. MIT Press, Cambridge, MA. Heilman, K. M., Watson, R. T., and Valenstein, E. (1985). Neglect and related disorders. In “Clinical Neuropsychology” (K. M. Heilman and E. Valenstein, Eds.), pp. 377–401. Oxford Univ. Press, New York. Hellige, J. B. (1993). “Hemispheric Asymmetry: What’s Right and What’s Left.” Harvard Univ. Press, Cambridge, MA. Hellige, J. B., Bloch, M. I., Cowin, E. L., Eng, T. L., Eviatar, Z., and Sergent, V. (1994). Individual variation in hemisphere asymmetry: Multitask study of effects related to handedness and sex. J. Exp. Psychol. Gen. 123, 257–263. Hook, M. A., and Rogers, L. J. (2000). Development of hand preferences in marmosets (Callithrix jacchus) and effects of ageing. J. Comp. Psychol. 114, 263–271. Hook-Costigan, M. A., and Rogers, L. J. (1995). Hand, mouth and eye preferences in the common marmoset (Callithrix jacchus). Folia Primatol. 64, 180–191. Hook-Costigan, M. A., and Rogers, L. J. (1996). Hand preferences in New World primates. Int. J. Comp. Psychol. 9, 173–207.

LATERALIZATION IN VERTEBRATES

157

Hook-Costigan, M. A., and Rogers, L. J. (1998). Lateralized use of the mouth in production of vocalizations by marmosets. Neuropsychologia 36, 1265–1273. Hopkins, W. D., and Bennett, A. (1994). Handedness and approach-avoidance behavior in chimpanzees. J. Exp. Psychol. 20, 413–418. Hopkins, W. D., and Morris, R. (1989). Laterality for visual-spatial processing in two languagetrained chimpanzees (Pan troglodytes). Neuropsychologia 35, 343–348. Hopkins, W. D., and Morris, R. (1993). Handedness in great apes: A review of ﬁndings. Int. J. Primatol. 14, 1–25. Howard, K. J., Rogers, L. J., and Boura, A. L. A. (1980). Functional lateralisation of the chicken fore-brain revealed by use of intracranial glutamate. Brain Res. 188, 369–382. Hugdahl, K. (1995). Classical conditioning and implicit learning: The right hemisphere hypothesis. In “Brain Asymmetry” (R. J. Davidson and K. Hugdahl, Eds.), pp. 235–267. MIT Press, Cambridge, MA. Hunt, G. R. (1996). Manufacture and use of hook-tools by New Caledonian crows. Nature 379, 249–251. Hunt, G. R. (2000). Human-like, population-level specialisation in the manufacture of pandanus leaves by New Caledonian crows Corvus moneduloides. Proc. R. Soc. London B 267, 403– 413. Ingle, D. J., and Hoff, K. vS. (1990). Visually evoked evasive behaviour in frogs. BioScience 40, 284–291. Jarman, P. J. (1972). The development of the dermal shield in impala. J. Zool. (London) 166, 349–356. Khyentse, M. D., and Rogers, L. J. (1997). Glutamate affects the development of the thalamofugal visual projections of the chick. Neurosci. Lett. 230, 65–68. King, J. E. (1995). Laterality in hand preferences and reaching accuracy of cotton-top tamarins (Saguinus oedipus). J. Comp. Psychol. 109, 34–41. King, J. E., and Landau, V. I. (1993). Manual preferences in varieties of reaching in squirrel monkeys. In “Primate Laterality: Current Evidence of Primate Asymmetries” (J. P. Ward and W. D. Hopkins, Eds.), pp. 107–124. Springer-Verlag, New York. Kinsbourne, M. (1975). The mechanism of hemispheric control of the lateral gradient of attention. In “Attention and Performance V” (P. M. A. Rabbitt and S. Dornic, Eds.), pp. 81–97. Academic Press, London. Lacreuse, A., and Fragaszy, D. M. (1996). Hand preferences for a haptic searching task by tufted capuchins. Int. J. Primatol. 17, 613–632. Lacreuse, A., and Fragaszy, D. M. (1999). Left hand preferences in capuchins (Cebus apella): Role of spatial demands in manual activity. Laterality 4, 65–78. Lane, R. D., and Jennings, J. R. (1995). Hemispheric asymmetry, autonomic asymmetry and the problem of sudden cardiac death. In “Brain Asymmetry” (R. J. Davidson and K. Hugdahl, Eds.), pp. 271–304. MIT Press, Cambridge, MA. Lippolis, G., Bisazza, A., Rogers, L. J., and Vallortigara, G. (2001). Lateralization of predator avoidance responses in three species of toads. Laterality, in press. Lorincz, E., and Fabre-Thorpe, M. (1996). Shift of laterality and compared analysis of paw performances in cats during practice of a visuomotor task. J. Comp. Psychol. 110, 307–315. Maier, S. E., and Crowne, D. P. (1993). Early experience modiﬁes the lateralization of emotionality in parietal lesioned rats. Behav. Brain Res. 56, 31–42. Marchant, L. F., and McGrew, W. C. (1996). Laterality of limb function in wild chimpanzees of Gombe National Park: Comprehensive study of spontaneous activities. J. Hum. Evol. 30, 427–443. McGrew, W. C., and Marchant, L. F. (1999). Laterality of hand use pays off in foraging success for wild chimpanzees. Primates 40, 509–513.

158

LESLEY J. ROGERS

McNeilage, P. F., Studdert-Kennedy, M. G., and Lindblom, B. (1987). Primate handedness reconsidered. Behav. Brain Sci. 10, 247–263. Mench, J., and Andrew, R. J. (1986). Lateralization of a food search task in the domestic chick. Behav. Neural Biol. 46, 107–114. Mesulam, M.-M. (1985). Attention, confusional states and neglect. In “Principles of Behavioral Neurology” (M.-M. Mesulam, Ed.), pp. 125–168. Davis, Philadelphia. Mikløsi, A., and Andrew, R. J. (1999). Right eye use associated with decision to bite in zebraﬁsh. Behav. Brain Res. 105, 199–205. Mikløsi, A., Andrew, R. J., and Savage, H. (1998). Behavioural lateralisation of the tetrapod type in the zebraﬁsh (Brachydanio rerio). Physiol. Behav. 65, 127–135. Mittleman, G., Whishaw, I. Q., and Robbins, T. W. (1988). Cortical lateralization of function in rats in a visual reaction time task. Behav. Brain Res. 31, 29–36. Møller, A. P., and Swaddle, J. P. (1997). “Asymmetry, Developmental Stability, and Evolution.” Oxford Univ. Press, Oxford. Morris, R. D., Hopkins, W. D., and Bolser-Gilmore, L. (1993). Assessment of hand preference in two language-trained chimpanzees (Pan troglodytes): A multimodal method analysis. J. Clin. Exp. Neuropsychol. 15, 487–502. Myslobodsky, M. S. (1983). “Hemisyndromes: Psychobiology, Neurology, Psychiatry.” Academic Press, New York. Nestor, P. G., and Safer, M. A. (1990). A multi-method investigation of individual differences in hemisphericity. Cortex 26, 409–421. Neville, A. C. (1976). Animal asymmetry. “The Institute of Biology’s Studies on Biology,” No. 67. Arnold, London. Nevue, P. J. (1988). Cerebral neocortex modulation of immune functions. Life Sci. 42, 1917– 1923. Nevue, P. J., and Moya, S. (1997). In the mouse, the corticoid stress response depends on lateralization. Brain Res. 749, 344–346. Nevue, P. J., Betancur, C., Barneoud, ´ S., Vitiello, S., and LeMoal, M. (1991). Functional brain asymmetry and lymphocyte proliferation in female mice: Effects of right- and left-cortical ablations. Brain Res. 550, 125–128. Nevue, P. J., Bluthe, ´ R.-M., Liege, ´ S., Moya, S., Michaud, B., and Dantzer, R. (1998). Interleukin1-induced sickness behavior depends on behavioral lateralization in mice. Physiol. Behav. 63, 587–590. Palmer, A. R. (1996). Waltzing with asymmetry. Bioscience 46, 518–532. Peters, M. (1995). Handedness and its relation to other indices of cerebral lateralization. In “Brain Asymmetry” (R. J. Davidson and K. Hugdahl, Eds.), pp. 183–214. MIT Press, Cambridge, MA. Phillips, R. E., and Youngren, O. M. (1986). Unilateral kainic acid lesions reveal dominance of right archistriatum in avian fear behavior. Brain Res. 377, 216–220. Posner, M. I., Walker, J. A., Friedrich, F. A., and Rafal, R. D. (1987). How do the parietal lobes direct covert attention? Neuropsychologia 25, 135–146. Rajendra, S., and Rogers, L. J. (1993). Asymmetry is present in the thalamofugal projections of female chicks. Exp. Brain Res. 92, 542–544. Rashid, N., and Andrew, R. J. (1989). Right hemisphere advantage for topographical orientation in the domestic chick. Neuropsychologia 27, 937–948. Reist, J. D., Bodaly, R. A., Fudge, K. J., Cash, K. J., and Stevens, T. V. (1987). External scarring of whiteﬁsh, Coregonus nasus and C. clupeaformis complex, from the western Northwest Territories, Canada. Can. J. Zool. 65, 1230–1239. Rigamonti, M. M., Previde, E. P., Poli, M. D., Marchant, L. F., and McGrew, W. C. (1998). Methodology for motor skill and laterality: Test of hand preference in Macaca nemistrina. Cortex 34, 693–705.

LATERALIZATION IN VERTEBRATES

159

Robins, A., Lippolis, G., Bisazza, A., Vallortigara, G., and Rogers, L. J. (1998). Lateralization of agonistic responses and hind-limb use in toads. Anim. Behav. 56, 875–881. Robinson, R. G. (1985). Lateralised behavioral and neurochemical consequences of unilateral brain injury in rats. In “Cerebral Lateralisation in Nonhuman Species” (S. D. Glick, Ed.), pp. 135–156. Academic Press, New York. Robinson, R. G., and Price, T. R. (1982). Post-stroke depressive disorders: A follow-up study of 103 patients. Stroke. 13, 635–641. Robinson, R. G., Kubos, K. L., Starr, L. B., Rao, K., and Price, T. R. (1984). Mood disorders in stroke patients: Importance of location of lesion. Brain 107, 81–93. Rogers, L. J. (1980). Lateralisation in the avian brain. Bird Behav. 2, 1–12. Rogers, L. J. (1982). Light experience and asymmetry of brain function in chickens. Nature 297, 223–225. Rogers, L. J. (1990). Light input and the reversal of functional lateralization in the chicken brain. Behav. Brain Res. 38, 211–221. Rogers, L. J. (1991). Development of lateralization. In “Neural and Behavioral Plasticity: The Use of the Domestic Chick as a Model” (R. J. Andrew, Ed.), pp. 507–535. Oxford Univ. Press, Oxford. Rogers, L. J. (1993). The molecular neurobiology of early learning, development and sensitive periods, with emphasis on the avian brain. Mol. Neurobiol. 7, 161–187. Rogers, L. J. (1995). “The Development of Brain and Behaviour in the Chicken.” CAB International, Wallinford. Rogers, L. J. (1996). Behavioural, structural and neurochemical asymmetries in the avian brain: A model system for studying visual development and processing. Neurosci. Biobehav. Rev. 20, 487–503. Rogers, L. J. (1997). Early experiential effects on laterality: Research on chicks has relevance to other species. Laterality 2, 199–219. Rogers, L. J. (2000). Evolution of hemispheric specialisation: Advantages and disadvantages. Brain and Language 73, 236–253. Rogers, L. J. (2001). Advantages and disadvantages of lateralization. In “Comparative Vertebrate Lateralization” (L. J. Rogers and R. J. Andrew, Eds.), Cambridge Univ. Press, New York. In press. Rogers, L. J., and Anson, J. M. (1979). Lateralisation of function in the chicken fore-brain. Pharm. Biochem. Behav. 10, 679–686. Rogers, L. J., and Bolden, S. W. (1991). Light-dependent development and asymmetry of visual projections. Neurosci. Lett. 121, 63–67. Rogers, L. J., and Deng, C. (1999). Light experience and lateralization of the two visual pathways in the chick. Behav. Brain Res. 98, 277–287. Rogers, L. J., and Kaplan, G. (1996). Hand preferences and other lateral biases in rehabilitated orang-utans (Pongo pygmaeus pygmaeus). Anim. Behav. 51, 13–25. Rogers, L. J., and Sink, H. S. (1988). Transient asymmetry in the projections of the visual hyperstriatum of the chicken, and reversal of its direction by light exposure. Exp. Brian Res. 70, 378–384. Rogers, L. J., and Workman, L. (1989). Light exposure during incubation affects competitive behaviour in domestic chicks. Appl. Anim. Behav. Sci. 23, 187–198. Rogers, L. J., Zappia, J. V., and Bullock, S. P. (1985). Testosterone and eye-brain asymmetry for copulation in chickens. Experientia 41, 1447–1449. Rogers, L. J., Andrew, R. J., and Burne, T. H. J. (1998). Light exposure of the embryo and development of behavioural lateralisation in chicks: I. Olfactory responses. Behav. Brain Res. 97, 195–200. Rothe, H. (1973). Handedness in the common marmosets (Callithrix jacchus). Am. J. Physical Anthropol. 64, 417–433.

160

LESLEY J. ROGERS

Sackheim, H. A., Gur, R. C., and Saucy, M. C. (1978). Emotions are expressed more intensively on the left side of the face. Science 202, 434–436. Schwabl, H. (1993). Yolk is a source of maternal testosterone for developing birds. Proc. Natl. Acad. Sci. U.S.A. 90, 11444–11450. Schwabl, H. (1996). Maternal testosterone in the avian egg enhances postnatal growth. Comp. Biochem. Physiol. A 114, 271–276. Schwabl, H. (1997). Maternal steroid hormones in the egg. In “Perspectives in Avian Endocrinology” (R. Etches and S. Harvey, Eds.), pp. 3–13. J. Endocrinology Ltd., Bristol. Schwabl, H. (1998). Developmental changes and among-sibling variation of corticosterone levels in an altricial avian species. Gen. Comp. Endocrinol. 116, 403–408. Schwarz, I. M., and Rogers, L. J. (1992). Testosterone: A role in the development of brain asymmetry in the chick. Neurosci. Lett. 146, 167–170. Sherman, G. F., Garbanati, J. A., Rosen, G. D., Yutzey, D. A., and Denenberg, V. H. (1980). Brain and behavioral asymmetries for spatial preference in rats. Brain Res. 192, 61–67. Shimizu, T., and Hodos, W. (1989). Reversal learning in pigeons: Effects of selective lesions of the Wulst. Behav. Neurosci. 103, 262–272. Skiba, M., Diekamp, B., Prior, H., and Gunt ¨ urk ¨ un, ¨ O. (2000). Lateralized interhemispheric transfer of color cues: Evidence for dynamic coding principles of visual lateralization in pigeons. Brain and Language 73, 254–273. Sovrano, V. A., Rainoldi, C., Bisazza, A., and Vallortigara, G. (1999). Roots of brain specializations: Preferential left-eye use during mirror-image inspection in six species of teleost ﬁsh. Behav. Brain Res. 106, 175–180. Swaddle, J. P. (1999). Limits to length asymmetry detection in starlings: Implications for biological signalling. Proc. R. Soc. London B 266, 1299–1303. Tommasi, L., and Vallortigara, G. (2000). Searching for the centre: Spatial cognition in the domestic chick (Gallus gallus). J. Exp. Psychol: Anim. Behav. Proc. 26, 477–486. Tommasi, L., and Vallortigara, G. (2001). Encoding geometric and landmark information in the left and right hemisphere of the avian brain. Behav. Neurosci. 115, 602–613. Tommasi, L., Vallortigara, G., and Zanforlin, M. (1997). Young chicks learn to localize the centre of a spatial environment. J. Comp. Physiol. A 180, 567–572. Tommasi, L., Andrew, R. J., and Vallortigara, G. (2000). Eye use in search is determined by the nature of the task in the domestic chick (Gallus gallus). Behav. Brain Res. 112, 119–126. Vallortigara, G. (2000). Comparative neuropsychology of the dual brain: A stroll through animals’ left and right perceptual worlds. Brain and Language 73, 189–219. Vallortigara, G., and Bisazza, A. (2001). How ancient is brain lateralization? In “Comparative Vertebrate Lateralization” (L. J. Rogers and R. J. Andrew, Eds.), chap. 1. Cambridge Univ. Press, New York. In press. Vallortigara, G., Rogers, L. J., Bisazza, A., Lippolis, G., and Robins, A. (1998). Complementary right and left hemiﬁeld use for predatory and agonistic behavior. Neuroreport 9, 3341–3344. Vallortigara, G., Rogers, L. J., and Bisazza, A. (1999). Possible evolutionary origins of cognitive brain lateralization. Brain Res. Rev. 30, 164–175. Vallortigara, G., Cozzutti, C., Tommasi, L., and Rogers, L. J. (2001). How birds use their eyes: Opposite left–right specialisation for the lateral and frontal visual hemiﬁeld in the domestic chick. Curr. Biol. 11, 29–33. Walker, S. F. (1980). Lateralization of functions in the vertebrate brain: A review. Br. J. Psychol. 71, 329–367. Ward, J. P. (1999). Left-hand advantage for prey capture in the galago (Galago Moholi). Int. J. Comp. Psychol. 11, 173–184. Ward, J. P., and Hopkins, W. D. (1993). “Primate Laterality: Current Behavioral Evidence of Primate Asymmetries.” Springer-Verlag, New York.

LATERALIZATION IN VERTEBRATES

161

Ward, J. P., Milliken, G. W., Dodson, D. L., Stafford, D. K., and Wallace, M. (1990). Handedness as a function of sex and age in a large population of lemur. J. Comp. Psychol. 104, 167–173. Wassersug, R. J., Naitoh, T., and Yamashita, M. (1999). Turning bias in tadpoles. J. Herpetol. 33, 543–548. Waters, N. S., and Denenberg, V. H. (1994). Analysis of two measures of paw preference in a large population of inbred mice. Behav. Brain Res. 63, 195–204. Watson, N. V., and Kimura, D. (1989). Right-hand superiority for throwing but not for intercepting. Neuropsychologia 27, 1399–1414. Weiss, D. J., Ghazanfar, A. A., Miller, C. T., and Hauser, M. D. (2001). Specialized processing of primate facial vocal expressions: Evidence for cerebral asymmetries. In “Comparative Vertebrate Lateralization” (L. J. Rogers and R. J. Andrew, Eds.). Cambridge Univ. Press, New York. In press. Wittling, W. (1997). The right hemisphere and the human stress response. Acta Physiol. Scand. 161(suppl. 640), 55–59. Zappia, J. V., and Rogers, L. J. (1983). Light experience during development affects asymmetry of fore-brain function in chickens. Dev. Brain Res. 11, 93–106. Zottoli, S. J., Hordes, A. R., and Faber, D. S. (1987). Localization of optic tectal input to the ventral dendrite of the goldﬁsh Mauthner cell. Brain Res. 401, 113–121.

This Page Intentionally Left Blank

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 31

Auditory Scene Analysis in Animal Communication STEWART H. HULSE DEPARTMENT OF PSYCHOLOGY JOHNS HOPKINS UNIVERSITY BALTIMORE, MARYLAND 21218

I. INTRODUCTION It is an early spring morning, and I have decided to take a walk in the woods. Many sounds surround me. I listen to my feet in the leaves on the path and, perhaps, to the sound of a wood thrush in a tree on one side and a blue jay in some scrub on the other. A farmer’s tractor in the distance breaks the silence. So does a car on a nearby road and a jet plane ﬂying overhead. Even an everyday walk is a journey through a multitude of natural and man-made sounds that come from all directions. Yet somehow I am able to focus on these sounds and hear each of them as a unitary, coherent whole, as if they represented functional objects—auditory objects. And this happens effortlessly even though the sounds may frequently overlap each other in time. Although my perception of all the auditory events on my walk seems completely natural and simplicity itself, it is truly a remarkable perceptual process that parses the surrounding world of sounds, the auditory scene, into functionally signiﬁcant auditory objects. All the sounds identifying an object strike my sensory system through a single common pathway—my ears. Still, I somehow perceive the objects as distinct qualitatively and spatially. Furthermore, a spectrographic analysis of the mixture of sounds I hear would be remarkably complex, especially if two, three, or more sounds overlapped each other. Yet my auditory system accomplishes this auditory scene analysis with ease. How can this be so? Albert Bregman (1990) published a major introduction to the problem of auditory scene analysis in human auditory perception in a seminal book published some years ago, and the ﬁeld has become increasingly active since then (for more summaries, see Darwin and Carlyon, 1995, and Yost, 1997). Previously, however, there has been little or no work done with nonhuman species, despite the fact that any animal that depends on sound to get along 163

C 2002 by Academic Press Copyright All rights of reproduction in any form reserved. 0065-3454/02 $35.00

164

STEWART H. HULSE

in the world must certainly face the problem of auditory scene analysis. A short life awaits the animal that fails to detect the sound of a predator among other benign sounds that comprise the animal’s acoustic environment. That is not to say that the problem which scene analysis tries to solve has not been recognized by students of animal behavior. Early on, Busnel (1968) noted speciﬁcally that animal communication involves the detection of a signal in biologically signiﬁcant background noise produced by a crowd of social animals and identiﬁed this with the “cocktail party” phenomenon (Cherry, 1953). The cocktail party phenomenon refers to the ease with which a person’s voice can be identiﬁed in the midst of a noisy gathering. The remainder of this chapter examines auditory scene analysis in the context of nonhuman auditory perception. The next section explains very brieﬂy some of the general perceptual processes that govern the formation of auditory objects and the parsing of the auditory scene. Following that, I discuss some work with songbirds from my own laboratory to introduce methods for studying auditory scene analysis in animals and to provide some basic evidence for the validity of the process in nonhuman species. Remaining sections review the animal literature for data, few though they may be, that suggest auditory scene analysis at work. Finally, I discuss brieﬂy the directions in which future research with animals might go.

II. AUDITORY SCENE ANALYSIS I have observed that the prevailing auditory scene is composed of one or more auditory objects, but what does that mean in more detail? A. AUDITORY OBJECTS AND THE AUDITORY SCENE According to Bregman (1990), auditory objects are coherent, organized auditory stimuli that come from a single source and signal a functional acoustic event. Function is a key concept here; auditory objects are sounds imbued with ecological signiﬁcance. Although one could analyze acoustically the complex of sounds that strike my ear on a morning walk, albeit with some technical difﬁculty perhaps, the starting point of a scene analysis is not an acoustic analysis, but a functional analysis based on the source of the sound, its likely identity, whether it is approaching or receding, and so on. For example, the stream of acoustic events that comprise the auditory object of a “birdsong” are the sounds uttered by that particular wood thrush, in that particular tree, in that particular location. More generally, auditory objects as the listener experiences them are not merely complex acoustic spectra (which they may well be), but the separate sounds of footsteps in the forest, birds, tractors, autos, airplanes, and the unique quality of my voice as I speak.

AUDITORY SCENE ANALYSIS IN ANIMAL COMMUNICATION

165

That is not to say that auditory objects are not amenable to acoustic analysis. Indeed, an important challenge is to identify those acoustic features and processes that assure the creation of an auditory object and its perceptual segregation from the complete auditory scene. But such an acoustic analysis will never sufﬁce completely. The ecological signiﬁcance of auditory objects necessarily derives from processes other than a formal analysis of an acoustic waveform by the auditory system. B. AUDITORY OBJECTS, GROUPING PRINCIPLES, AND STREAM SEGREGATION An important characteristic of all stimuli is that they change in time. A bout of birdsong as an auditory object is not a momentary event, but a ﬂow of acoustic information through time—an auditory stream. What is it about this ﬂow that makes it cohere into a streaming auditory object that is unique and differentiable from other objects in the auditory scene? As Bregman (1981, 1990) suggests, the answer lies in the principles of Gestalt psychology. These include the principles of grouping (proximity, similarity, etc.), closure, and belongingness. They also include the principle of exclusive allocation, which refers to the fact that the perceptual system assigns a sensory event to the perception of just one object at a time. Visual analogies help to demonstrate some of these principles at work. For example, the familiar reversible ﬁgure shown in Fig. 1 can be seen as either a vase or two human proﬁles facing each other, but the ﬁgure cannot be seen as both concurrently, a demonstration of the principle of exclusive allocation. An example in the auditory domain would be that a given blob of sound can be part of a birdsong or a tractor noise, but it cannot be part of both simultaneously. The perceptual system assigns the auditory event to one auditory object or the other, but not to both at the same time. The principles of similarity and proximity are illustrated in Fig. 2. Visually, the ﬁlled and open ﬁgures at the top are each grouped together on the basis of their similarity; like objects tend to group with like objects. At the bottom, the same ﬁgures have also been separated by their proximity; the ﬁlled objects are closer to each other in space than they are to the open objects, and so ﬁlled and open objects each hang together even more cohesively. Note, too, that the open and ﬁlled objects are interleaved, that is, open and ﬁlled objects alternate in time. However, interleaving fails to disturb the perceptual grouping afforded by similarity and proximity. To bring the schemas of Fig. 2 into the auditory world, observe that the open and closed ﬁgures are plotted on coordinates of frequency (pitch) and time. At the bottom of the ﬁgure, it follows from grouping principles that the objects (tones) high in frequency should cluster together into two high groups of ascending “melodies,” whereas those low in frequency should cluster into two low groups of descending “melodies.” By the principle of

166

STEWART H. HULSE

FIG. 1. A familiar “reversible ﬁgure.” The ﬁgure can be seen either as a black vase or as two white faces in proﬁle, but not both at the same time.

FIG. 2. Examples of similarity and proximity at work. At the top of the ﬁgure black ﬁgures are grouped together visually by similarity into two descending sequences while open ﬁgures group into two ascending sequences. At the bottom of the ﬁgure, grouping principles based on proximity have been added to those of similarity. The segregation of black and white ﬁgures into ascending and descending sequences has been enhanced by the combination of the two grouping principles.

AUDITORY SCENE ANALYSIS IN ANIMAL COMMUNICATION

167

exclusive allocation, one ought to be able to attend to the high, ascending melodies or to the low, descending melodies, but not both at the same time. If the frequency separation between the high and low tones and the rate at which the successive tones occur are set appropriately, that is exactly what happens. One might think that interleaved melodies might be heard in the sense that the objects comprising the melodies alternate in time—high–low–high– low. . .. In fact, at low repetition rates (and at appropriate frequency separations between high and low tones), a listener does hear the tones bouncing back and forth between high and low. But as repetition rates and frequency differences increase, the alternating tones segregate and “split” to be heard as two separate auditory objects streaming through time (Miller and Heise, 1950). In fact, functions can be obtained that relate frequency separation, repetition rate, and the likelihood that tone sequences will segregate perceptually into separate streams (Rose and Moore, 2000; Van Noorden, 1975). In discussing this phenomenon, Bregman notes that Gestalt principles of visual apparent motion can also be applied to such auditory motion, in particular, Korte’s ¨ third law (Bregman and Campbell, 1971). In addition to pitch differences, grouping and stream segregation can occur on the basis of other acoustic dimensions. Timbre, or the spectral composition of a sound, is an especially important one (Culling and Darwin, 1993; Hartmann and Johnson, 1991; Singh and Bregman, 1997). Differences in spatial location is another, albeit one that—surprisingly—may serve most importantly as a vehicle for other auditory grouping cues (Arbogast and Kidd, 2000; Bregman, 1990). The distinction between primitive segregation and schema-based segregation (Bregman, 1990) provides another important aspect of auditory scene analysis. The former refers to innate, unlearned capacities to segregate auditory objects in the natural world and, presumably, depends primarily on bottom-up auditory processes that have evolved in many animals over time— an idea that Gestalt psychologists stressed at the outset. Schema-based principles, on the other hand, are top-down and learned through individual experience. For example, the way we hear music or our spoken language may depend in part on unlearned organizational principles, but much of our ability to perceive music (e.g., to hear distinctly the sounds of one musical instrument from another) depends on learning the harmonic structure that deﬁnes the sound of the instrument in question. As we shall see, the themes of unlearned, bottom-up principles and learned, top-down principles will recur frequently in this review. It is impossible in this chapter to explore the ﬁeld of human auditory scene analysis in any great detail; for an elegant exposition, Bregman’s (1990) book is without comparison. In it you will ﬁnd many further illustrations of the

168

STEWART H. HULSE

parallels between perceptual organization in the visual and auditory worlds and a far more detailed discussion of the factors that enter into auditory scene analysis. McAdams and Bigand (1993) provide another useful general source which discusses principles of auditory scene analysis (as well as memory and some other functions) in auditory cognition. Before we move on, we should reemphasize that Bregman develops the relevance of scene analysis for the ﬂow of auditory information characteristic of the perception of both language and music. In fact, viewing language as a stream of coherent auditory information accounts for many important phenomena in language perception. And music composers discovered the principles of scene analysis and stream segregation many centuries ago. For example, such principles underlie the construction of melodic counterpoint, and as you listen you can partition the larger sound of a symphony orchestra into the sounds of its component instruments—the strings, the brass, the woodwinds, and so on. Given that birdsong and other forms of acoustic communication by animals are, similarly, ﬂows of acoustic information in time, the application of the principles of auditory scene analysis to acoustic communication has obvious utility. C. THE ROLE OF TIME IN SCENE ANALYSIS It is worth emphasizing that the ﬂow of acoustic events in time is a hallmark of auditory scene analysis. I have introduced auditory scene analysis in part by drawing upon some of the basic Gestalt organizing principles. Although it is possible to do this with visual examples by using static ﬁgure–ground relations (as in Fig. 1), it is never possible to illustrate the organizing principles in auditory scene analysis unless sounds are moving in time. To be sure, many visual examples of Gestalt principles rely on time and motion for their production, but with audition this is necessarily the case. A brief glimpse at a stable visual ﬁgure can generate grouping (e.g., as in the grouping of open and ﬁlled objects in Fig. 2), but a brief burst of sound will never allow the organizing principles of auditory scene analysis to take hold. Instead, sounds must move through time for their organization to occur. Indeed, although open and ﬁlled objects in Fig. 2 group together visually, the auditory stream segregation they are designed to illustrate will appear only if the objects (now conceived as tones) move through time. D. THE ROLE OF ATTENTION IN SCENE ANALYSIS There is yet another aspect of scene analysis that warrants discussion, and that is the role of attention in our perception of auditory objects, especially those that overlap in time.

AUDITORY SCENE ANALYSIS IN ANIMAL COMMUNICATION

169

The familiar “cocktail party” phenomenon provides an example. Cherry (1953) wondered how we can distinguish one speaker among a multitude of others in a crowded auditory scene such as a cocktail party. He proposed that the auditory system might do so by an appropriately chosen set of auditory ﬁlters, but today we also recognize the cocktail party phenomenon as a functional auditory scene analysis. One of the critical features of the phenomenon is that it involves the process of selective attention. I can easily attend to the voice of the person with whom I am speaking directly at the party, but if I try, I can also switch my attention to the voice of the person who is chatting with someone else behind me—even though in each case both speakers are talking simultaneously and the sounds of their voices strike my ears almost simultaneously. In either instance, the voice to which I am not attending recedes into the background—much as the two opposing faces in the reversible ﬁgure (see Fig. 1) recede into the background if one attends to the vase. Also, recall my spring walk in the woods. The process of listening to the jet plane or the thrush or the auto is an active, goal-driven process of focusing on one auditory object to the exclusion of others—which recede perceptually into the background of the overall auditory scene. Once again it is far beyond the scope of this chapter to discuss the complex factors that govern the process of selective attention. Bregman (1990), Yost (1997), and Jones and Yee (1993) have discussed the problem in the context of auditory scene analysis, but most of the work on attention has been done in the visual domain: see Pashler (1998) and Yantis (2000) for summaries. Yantis (2000) makes the general claim that selective attention is both a top-down (learned and goal directed) and a bottom-up (stimulus driven) process, which ﬁts the emphasis of an auditory scene analysis very well. In other words, as I single out objects from the auditory scene, my attention is actively driven (goal directed) toward some functionally signiﬁcant sound in particular, but my attention is also drawn by stimulus features unique to the object, such as its loudness, spectral composition, and so on. These ideas parallel Bregman’s notions of primitive and schema-driven organizational principles. It cannot be overemphasized that selective attention among a panoply of potential auditory objects is a hallmark of auditory scene analysis.

III. DEMONSTRATIONS OF AUDITORY SCENE ANALYSIS BY SONGBIRDS With the foregoing brief introduction to the general process of auditory scene analysis in mind, let us turn now to some examples of experiments in which auditory scene analysis has been the focus of attention in nonhuman animals. Almost all the work has been done with songbirds, especially the European starling (Sturnus vulgaris)—although as we shall see Benney and

170

STEWART H. HULSE

Braaten (2000) have extended the work to zebra ﬁnches (Taenopygia guttata) and Bengalese ﬁnches (Lonchura striata var. domestica) Starlings are a semicolonial species of birds that gather and roost in large ﬂocks during the winter months, then disperse to breed in the early spring. Both males and females are adept singers with complicated, multisyllable (multimotif) songs that may last for 40 s or more (Eens, 1997). Starlings also show an ability to mimic both natural and synthetic sounds which indicates that they perceive and respond to their auditory scene with some facility. The birds are also excellent for the operant studies of auditory perception and scene analysis to which I turn now. Thus far, experiments aimed directly at auditory scene analysis in starlings have been addressed to three issues. First, can a starling identify one species’ song when that song is mixed with a song from another species? Is starling song privileged in this regard, that is, are starlings especially sensitive to conspeciﬁc as contrasted with heterospeciﬁc song as an auditory object? Second, can a starling discriminate between the songs of two other starlings, thus showing individual identiﬁcation, and can that discrimination be maintained when the songs of even more starlings are added to the background auditory scene? Third, is auditory scene analysis a general perceptual process for the starling? That is, can auditory object formation and stream segregation be achieved with synthetic, nonnatural stimuli? A. SEGREGATING TWO SIMULTANEOUS BIRDSONGS INTO SEPARATE AUDITORY OBJECTS The ﬁrst demonstration of auditory scene analysis was obtained in an experiment in which my colleagues and I tested starlings for their ability to single out the birdsong of one species from that of another (Hulse et al., 1997). Table I shoes the scheme of the study. TABLE I SONG PAIRINGS AND THE CORRECT RESPONSE (A OR B) FOR THE SUPERIMPOSED STIMULUS MIXTURES DURING BASELINE TRAINING

N

Song pairs with starling song Category A/Peck A

Song pairs without starling song Category B/Peck B

2 2 2

Starling + brown thrasher Starling + nightingale Starling + mockingbird

Nightingale + mockingbird Mockingbird + brown thrasher Nightingale + brown thrasher

Note: Five exemplars from each stimulus combination were used in the experiment. From Table 1 in Hulse et al. (1997). Copyright C 1997 by the American Psychological Association. Adapted with permission.

AUDITORY SCENE ANALYSIS IN ANIMAL COMMUNICATION

171

Six starlings were trained in an operant conditioning task with food as a reward to discriminate superimposed 10-s mixtures of birdsongs which combined the song of one species with that of another. One category of songs (Category A) contained many exemplars of a simultaneous mixture of a starling song with a song from one of three other species, brown thrasher (Toxostoma rufum), nightingale (Luscinia megarynchos), or mockingbird (Mimus polyglottos). The other category of songs (Category B) contained many exemplars of a simultaneous mixture of a brown thrasher song, say, with a nightingale or a mockingbird song, but no starling song. The birds’ task was to peck one key in an operant conditioning chamber if they heard an exemplar from Category A played through the single loud speaker in the chamber, and a second key if they heard an exemplar from Category B. Correct responses were rewarded with food; incorrect responses turned the light out in the chamber for a few seconds and led to a correction trial. Spectrograms of a representative sample of the stimuli appear in Fig. 3. A 10-s segment of starling song appears at the top of the ﬁgure, and a 10-s segment of nightingale song appears just below the starling sample. The third line contains a spectrogram of the starling and nightingale songs when they are combined. Finally, at the bottom of the ﬁgure, the starling and nightingale songs are mixed with a sample of the dawn chorus. (I will discuss this last condition shortly). All starlings learned the discrimination between exemplars from Category A and exemplars from Category B to a criterion of at least 85% correct in 20–35 daily sessions each lasting about 2 h. On what basis might the discrimination between the two categories have been learned? 1. Stream Segregation The most interesting possibility is that the starlings learned the discrimination by segregating the song of one species from that of the other in the pair, perhaps attending to the song of one species on some trials and the song of the other species on other trials. This would be an example of auditory scene analysis at work. One might also imagine that exemplars containing conspeciﬁc starling song would be especially salient to the starling subjects and therefore more easily segregated, but that was not the case—at least in the sense that song mixtures that contained starling song were not discriminated more accurately than those that did not. 2. Memorization Another possibility is that the birds were merely memorizing acoustic features that were unique to each of the stimulus exemplars. The use of many exemplars from each of the categories was designed to minimize this possibility, but it exists nevertheless. To check for memorization, the birds were

172

STEWART H. HULSE

FIG. 3. Representative samples of songs from the European starling, nightingale, European starling mixed with nightingale, and European starling mixed with both nightingale and the dawn chorus. The high-amplitude tracing in the dawn chorus sonograph, which is especially prominent after about 3 s, is the song of a wood peewee (Contopus virens) that was near the microphone when the dawn chorus was recorded. (Figure 1 from Hulse et al., 1997.) Copyright C 1997 by the American Psychological Association. Reprinted with permission.

transferred to a new discrimination in which some of the stimuli were novel exemplars from the same two categories in the initial baseline discrimination. If the test starlings were simply memorizing acoustic features, then the novel stimuli should lead to an immediate drop in discrimination performance because the features would be new and, therefore, unlearned. When the transfer was undertaken, the starlings immediately transferred their accurate discrimination performance to the novel song exemplars. Posttransfer performance with the novel exemplars was virtually identical (>85%

AUDITORY SCENE ANALYSIS IN ANIMAL COMMUNICATION

173

correct) to pretransfer performance with the original baseline exemplars. One might still argue that features of the novel exemplars were similar to those of the baseline exemplars, but the feasibility of that argument becomes increasingly unlikely. Remember that the features in question would arise from combinations of the two superimposed songs, and novel exemplars reduce the likelihood that such combinations would recur. 3. Tests with Songs from Each Species Heard Alone As a further, perhaps most critical, assay for song segregation, the starlings were tested next with novel stimuli that consisted of single-species song exemplars for the four species songs that were used in the previous experiment. The novel exemplars were introduced as occasional probe trials, that is, single-species song exemplars from each of the four species were presented amid regular baseline trials incorporating the usual song-pair mixtures. Single-species probes occurred on 10% of the trials in a session. This test is especially telling because the starling subjects had never heard these stimuli in the experiment—neither individually nor as part of song mixtures. If the starlings were in fact selectively attending to and segregating the individual songs from the mixtures, then they should maintain accurate discrimination performance with the single-species songs. However, if their baseline performance was based on stimulus features or other acoustic factors unique to the song mixtures, their accuracy should fall. This should be especially so for nightingale song exemplars because this species is native to Europe and our starling subjects, trapped in the United States, had certainly never heard nightingale song before. Fig. 4 tells the story. Clearly, the starlings maintained above-chance performance on the singlesong probe stimuli. Although there was some (statistically reliable) overall decrease in percent correct relative to performance on the baseline mixtures during the same session, mean performance was still better than 70% correct for all species. Interestingly, performance on the novel starling songs was reliably better than that for any of the other three species, showing in this case that conspeciﬁc starling song was indeed a privileged stimulus relative to songs of the other species. 4. Song Segregation from the Dawn Chorus In a ﬁnal transfer in this experiment, the starlings were returned to the original baseline discrimination of song-pair mixtures and asked to maintain the discrimination when recordings of the dawn chorus were added to all the two-species song mixtures (see Fig. 3 for a spectrogram of a representative sample of this stimulus condition). This transfer was an attempt to add additional natural distracters to the auditory scene to press the starlings’ abilities

174

STEWART H. HULSE

FIG. 4. Auditory segregation of birdsongs by species. The bars at the left show performance on baseline song pairs that either included starling song or did not. The bars on the right show performance when probe tests were inserted in which the song of each species was presented by itself. Error bars are standard errors. (Figure 3 from Hulse et al., 1997.) Copyright C 1997 by the American Psychological Association. Reprinted with permission.

to maintain the baseline discrimination between song pairs. Although the results showed a decline in performance relative to baseline, discrimination accuracy was still roughly 80% correct in the presence of the dawn chorus— substantially and reliably above chance. Auditory object formation and song segregation are apparently quite robust, as indeed they ought to be in nature. 5. Summary This initial work ﬁrmly establishes auditory scene analysis and song segregation as principles that are at work in the natural world of songbirds. Starlings were able to single out the songs of single species from a variety of auditory mixtures including the song of a second species and an auditory melange ´ as complex as the dawn chorus. B. INDIVIDUAL RECOGNITION AND AUDITORY SCENE ANALYSIS A second series of experiments examined the ability of starlings to correctly identify and classify the songs of two other conspeciﬁcs and to maintain

AUDITORY SCENE ANALYSIS IN ANIMAL COMMUNICATION

175

that identiﬁcation and classiﬁcation in the face of an increasing number of conspeciﬁc distracters (Wisniewsky and Hulse, 1997). The general procedures were virtually identical to those of the work described above for mixed-song pairs. Test starlings were ﬁrst trained to discriminate ten 10-s song segments from Starling A from ten 10-s song segments from Starling B using food as a reward for correct responses on the corresponding keys in the operant chambers. The song segments were sampled randomly from longer complete songs. Training began with discrimination between one song segment from Starling A and one from Starling B, and when that pair was mastered to a criterion of 85% correct, a second pair was added, then a third, and so on. Of course, unlike the mixed-pair stimuli in the ﬁrst series of experiments, just one song-segment exemplar was presented by itself on any given trial. Training was very rapid. When the 10th pair of song segments was added (and the starlings were now classifying 20 song segments), that ﬁnal pair was mastered in less than 20 trials. In fact, the rapid classiﬁcation of the A and B song segments into two classes suggests that the starlings learned to treat the song segments as separate higher order stimulus categories. That is, the test starlings had abstracted some general characteristics of the song segments from Starling A as compared with Starling B and learned to distinguish the songs of the two individuals on that general basis. Subsequent tests established that fact more ﬁrmly (see the original article by Wisniewsky and Hulse, 1997, for details). 1. Individual Recognition with an Added Distracter Given the initial discrimination between the songs of Starling A and those of Starling B, a transfer test began in which a set of 10-s segments of song from Starling C, identical in length and loudness to those of Starlings A and B, were mixed (overlapped) by computer with both the original Starling A and Starling B song stimuli. Note that Starling C songs were not valid discriminative stimuli because they were common to both A and B stimuli. However, the C songs did serve as distracters because the test starlings had to selectively attend and segregate the relevant A and B songs from the irrelevant C songs to maintain the A versus B discrimination. Figure 5 displays the results of this experiment. The addition of distracter C songs in the transfer test reduced discrimination accuracy initially by about 20%, but overall accuracy was still substantially (and reliably) above chance. The starlings also showed improvement as transfer sessions progressed. 2. Individual Recognition with Four Added Distracters To press the starlings to some kind of limit, a ﬁnal transfer required them to maintain the discrimination between the songs of Starling A and those of Starling B when distracter songs from four starlings—Starlings C, D, E, and F—were combined with both the A and B songs. To human ears, the

176

STEWART H. HULSE

FIG. 5. Auditory segregation when additional songs are combined with A and B starling songs learned in a baseline discrimination. At the left is pretransfer performance on discrimination between Starling A and Starling B songs. During transfer sessions that followed, songs from a third starling, C, was added to the baseline pairs, or songs from four additional starlings were added to the baseline pairs. Error bars are standard errors. (From Figure 6 in Wisniewsky and Hulse, 1997.) Copyright C 1997 by the American Psychological Association. Adapted with permission.

blend of four distracters with the songs of Starling A or Starling B creates a cacophony from which it is utterly impossible (for me and colleagues in the lab, at any rate) to identify either the target A or B songs—even after some practice listening to the target songs. Not so for starlings. Although four distracters reduced discrimination accuracy for the Song A and Song B stimuli to 60–65% correct, performance remained reliably above chance (Fig. 5). This is strong testament to the species’ capacity for auditory scene analysis and object formation. Lest too much credit be given to starlings’ ability to identify individuals in the midst of conspeciﬁc song, think of the analogous situation of the crowded cocktail party. In all likelihood, people would have little trouble identifying the voices of two familiar friends despite a substantial number of other distracting voices in the background. As far as I know, the limits on human capacity in this regard have not been determined. Neither has the capacity for starlings to identify two human voices in the midst of many human distracters.

AUDITORY SCENE ANALYSIS IN ANIMAL COMMUNICATION

177

But there can be little doubt that familiarity with target and distracter stimuli must have a substantial effect on the capacity to segregate auditory objects from noisy environments. 3. Summary In this section I have broadened our demonstration of starlings’ accomplishments as auditory scene analyzers to include, ﬁrst, the discrimination of the songs of two target starlings, and second, the maintenance of that discrimination when sets of conspeciﬁc songs from as many as four other starlings are added to the stimulus complex as distracters. As Wisniewsky and Hulse (1997) point out the ﬁrst fact suggests that starlings can form abstract categories for the songs of individual conspeciﬁcs, an interesting observation in its own right (for which further discussion would take us beyond the scope of this chapter). Given that starlings are a semicolonial species that roost in large groups in the winter, it is perhaps not surprising that they are also accomplished at identifying the song of other individuals in the complex auditory scene created by the ﬂock. C. SCENE ANALYSIS IN ZEBRA FINCHES AND BENGALESE FINCHES Some of the observations just described for starlings have been replicated with two Estrildid ﬁnches, the zebra ﬁnch and the Bengalese ﬁnch. Benney and Braaten (2000), in an elegant experiment, used operant techniques similar to those just described for starlings to see if one ﬁnch species, say zebra ﬁnches, could hear examples of their own song when those examples were mixed with songs from two other species, a Bengalese ﬁnch and one of four unrelated species, brown thrasher (Toxostoma rufum), northern cardinal (Cardinalis cardinalis), American robin (Turdus migratorius), or song sparrow (Melospiza melodia). The operant task was a Go–NoGo task in which responses to one stimulus type were rewarded, while those to an alternative stimulus type were not. The discrimination emerged over trials as rapid responding appeared for the rewarded stimuli and slow (or no) responding appeared for the unrewarded stimulus type. The results showed, ﬁrst of all, that both zebra ﬁnches and Bengalese ﬁnches were quite adept at attending to and discriminating the target song when it was combined with two other songs. That is, the birds successfully singled out the critical song stimulus from the larger auditory scene and responded accurately to it. The results also showed that, if both ﬁnch species were trained to detect conspeciﬁc song as a rewarded target in the three-song mixture, learning proceeded faster than if the ﬁnches were trained to detect heterospeciﬁc song as a rewarded target. That is, there was a signiﬁcant bias among the birds for rapid learning based on conspeciﬁc song targets in the

178

STEWART H. HULSE

mixtures. Further analysis of the zebra ﬁnches and Bengalese ﬁnches taken separately showed that the zebra ﬁnches were especially sensitive in this regard. Zebra ﬁnches trained to detect zebra ﬁnch song as compared with Bengalese ﬁnch song as the target were much faster to learn the discrimination. However, for Bengalese ﬁnches, learning proceeded at about the same pace regardless of whether or not the target song was a Bengalese ﬁnch or a zebra ﬁnch. These results not only generalize the results regarding auditory scene analysis with starlings to two other songbird species, but also add to data which show that conspeciﬁc song is a privileged stimulus within the auditory scene. D. AUDITORY SCENE ANALYSIS AND STREAM SEGREGATION AS GENERAL PERCEPTUAL PROCESSES Now that I have shown that several songbird species can perform auditory scene analyses for a naturally relevant stimulus like birdsong, I can now go a step further and ask if scene analysis is a process that can be used for sounds in general. In other words, would songbirds demonstrate auditory object formation and stream segregation for synthetic sounds that are not part of their natural environment? If so, that would suggest the songbird brain has some general mechanism for scene analysis that is independent of any avian-speciﬁc auditory stimulus. To this end, my students and I (MacDougall-Shackleton et al., 1998) studied the effect of pitch separation on auditory stream segregation (Bregman and Campbell, 1971; Rose and Moore, 2000; Van Noorden, 1975). We trained starlings initially to discriminate sequences of sine (pure) tones that varied in their temporal structure. The scheme of the experiment appears in Fig. 6. Each object in the ﬁgure represents a time–amplitude plot of a sine tone of one frequency (e.g., 1000 Hz) that is 100 ms in duration. The objects are organized into one of three stimuli used in baseline training: a “galloping” sequence (so-called because the temporal spacing makes the sequence sound like the hoof beats of a horse, especially at quicker tempos), and two isochronous sequences in which the 1000-Hz objects are spaced at equal temporal intervals, but the interval length differs between the two sequences. Starlings were taught in baseline to peck one key when they heard the galloping sequence and another key when they heard either of the two isochronous sequences. The birds were ﬁrst trained with sine-tone stimuli of one frequency, say 1000 Hz, then trained in later sessions with frequencies of 1050, 1710, and 4538 Hz. Within any training session, however, the stimuli were all of just one frequency. After the starlings were responding to baseline stimuli to a criterion of 85% correct within daily training sessions, probe stimuli were introduced on

AUDITORY SCENE ANALYSIS IN ANIMAL COMMUNICATION

179

FIG. 6. Scheme of the test for generalized stream segregation in starlings. Birds were trained to discriminate the “galloping” pattern from either of the two “isochronous” patterns. Then they were tested with an interleaved pattern (bottom) in which a galloping structure was used, but tones were separated by two different pitches. If stream segregation occurs the probe stimuli should be heard as one or the other of the isochronous patterns and should be so classiﬁed by the birds. (Based on Figure 1 from MacDougall-Shackleton et al., 1998.)

10–20% of the trials in a session. The probe stimuli, schematized at the bottom of Fig. 6, had the temporal structure of the galloping baseline stimulus. However, the frequency of sine tones changed within the stimulus sequence (denoted by the upward displacement of some of the tone-objects in the ﬁgure). The more frequently occurring frequency was 1000 Hz, but on a random one-third of the probe trials, the displaced frequency was either 1050, 1710, or 4538 Hz (the same frequencies with which the starlings had been accustomed in different segments of baseline training). This created frequency separations of 50 (near threshold for starlings), 710, and 3538 Hz. If the starlings were able to segregate the probe sequences on the basis of pitch differences, they would peck the key correlated with isochronous sequences because the low-pitched sequence (correlated with 1000-Hz tones) had a temporal structure identical to the Isochronous 1 sequence in baseline training (see Fig. 6), and the high-pitched sequence (correlated with higher frequency tones) had a temporal structure equivalent to the Isochronous 2 sequence in baseline training. If, however, stream segregation failed to

180

STEWART H. HULSE

FIG. 7. Results of the test for generalized stream segregation. The left two pairs of bars show discrimination performance between galloping and isochronous patterns during initial training and during the test sessions when probes were introduced. The bars at the right show the probability that the birds classiﬁed the probes as isochronous. The tendency to do so increases with the frequency separation among tones in the galloping probe patterns. Error bars are standard errors. (Figure 3 from MacDougall-Shackleton et al., 1998.)

occur on the basis of pitch differences, then the birds should peck the key correlated with the galloping sequence because the probe sequences had the galloping temporal structure of baseline stimuli. Figure 7 shows the results. The bars on the left of the ﬁgure show a high probability of pecking the key correlated with isochronous sequences when isochronous sequences occurred during initial baseline training and on trials with baseline stimuli during sessions that included probe trials, and they show a corresponding low probability of pecking the isochronous key when galloping sequences occurred. The bars on the right of the ﬁgure show a low probability of pecking the key correlated with isochronous stimuli when the frequency difference

AUDITORY SCENE ANALYSIS IN ANIMAL COMMUNICATION

181

on probe trials was just 50 Hz. However, the probability of an isochronouskey response increases sharply as the frequency difference increases to 710 and 3538 Hz. Clearly, as humans do in comparable situations (Van Noorden, 1975), the starlings segregated the galloping probe sequences into two isochronous auditory streams on the basis of pitch differences among the tones in the sequences. 1. Summary The foregoing results establish that stream segregation with synthetic stimuli occurs in at least one species of songbird. There is every reason to believe that such segregation would be found in other species as well. Apparently stream segregation (based on pitch differences, at any rate) is a general process that does not depend in some fashion on avian-speciﬁc auditory stimuli. E. OVERVIEW To this point, I have shown that starlings (a) can identify a target song when it is mixed with the songs of conspeciﬁcs and other species, (b) can do so even when the target song is from an unfamiliar species (e.g., nightingales; though there is a distinct bias favoring a conspeciﬁc target song), and (c) can maintain starling target song identiﬁcation when distracting songs are added from several additional conspeciﬁcs. We also know that scene analysis is not limited to the perception of birdsong—starlings were able to segregate streams of synthetic sounds (sine tones) when the streams differed in pitch. Moreover, data (Benney and Braaten, 2000) show that many of these observations are not limited to starlings, but generalize to Estrildid ﬁnches. It is worth nothing that there is at least a functional parallel between segments of overlapping birdsong and sequences of synthetic sine tones. Both are acoustic events moving in time, and both offer an opportunity for the formation of segregated auditory streams. Overlapping songs, say, from two or more individuals, differ from moment to moment in pitch, spectral structure (timbre), and loudness. Any one or a combination of these acoustic cues could be used to afford the perceptual segregation of the aggregated sound into separate auditory streams—the voices of individual conspeciﬁcs in this case. Although we have focused on pitch as a cue to produce segregation of synthetic auditory streams, work with human observers has shown that people can also segregate on the basis of synthesized spectral structure (Iverson, 1995; Vliegen et al., 1999) and loudness changes (Bregman and Dannenbring, 1977), among other possible cues. Without doubt, further use of human perception as a heuristic tool for work with nonhuman animals would be worthwhile.

182

STEWART H. HULSE

IV. FURTHER POSSIBILITIES FOR RESEARCH ON AUDITORY SCENE ANALYSIS IN NONHUMAN ANIMALS Because so little work has been done directly on auditory scene analysis in nonhuman animals, very little literature bears directly on the problem. There are many studies that have been done on acoustic interactions (especially among conspeciﬁcs) that would lend themselves nicely to auditory scene analysis, however, and in this section I want to draw attention to a few of them. It is not my purpose to review this literature in detail. Instead, I wish merely to suggest possibilities for future research if a scene analytic approach were to be adopted. Scene analysis arises only when there are melanges ´ of sound from which a functionally signiﬁcant acoustic signal must be perceptually isolated. At ﬁrst glance, this suggests the signal-in-noise problem of classical auditory psychophysics, and there are, of course, a plethora of studies on both humans and nonhuman animals examining an observer’s capacity for identifying an acoustic signal in noise—especially white noise. However, although examinations of signals in white noise are useful for establishing basic auditory functions such as critical ratios, for example, auditory signals in the psychophysical laboratory rarely have functional signiﬁcance, and white noise rarely occurs in nature. A. WHAT IS NATURAL NOISE? So, then, what characterizes “natural” noise? For the purposes of scene analysis it is useful to identify three classes of noise that occur in nature. The classiﬁcation is quite arbitrary in the sense, for example, that a given acoustic event, such as a birdsong, can be a signal on one occasion and part of a noisy environment (the sound of a ﬂock) on another. Remember that the function of a sound is an essential part of auditory scene analysis, and this holds true for identifying classes of noise in the auditory scene. 1. Chorusing among Conspecifics The acoustic chatter, or chorus, generated by individuals of the same species interacting simultaneously in groups provides one class of natural noise. The cacophony produced by large ﬂocks of birds, such as European starlings in a winter roost, is a case in point in the avian world. The chorus of amphibians such as frogs in the neighborhood of a mating resource such as a breeding pond is another. In each instance, from the vantage point of a single perceiver in the group, the complex auditory scene must be analyzed to identify the output from a single signiﬁcant sender, such as a potential breeding mate or territorial competitor. But in all cases, functionally signiﬁcant interactions arise through acoustic interactions among members of the same

AUDITORY SCENE ANALYSIS IN ANIMAL COMMUNICATION

183

species, and the task of the perceiver is to identify important signals in the midst of a simultaneous mixture of signals with which the perceiver is highly familiar and highly engaged. 2. Heterospecific Noise Potentially, sounds from other species within the same taxonomic group provide a functionally signiﬁcant background from which important signals might have to be perceptually isolated. In the case of songbirds, the dawn chorus is a case in point. Certainly, the dawn chorus constitutes a natural noise from which conspeciﬁcs must isolate sounds that are functionally relevant for purposes of mating and territorial defense. To my knowledge, however, we know relatively little about the extent to which a given individual attends to the sounds of heterospeciﬁc species in that chorus. Of course the sounds of any potential predators (birds or otherwise) that might occur along with the dawn chorus is an interesting special case in its own right. 3. Other Environmental Sounds Finally, there are sounds, possibly distinct from the foregoing, that occur in nature but which have no clear functional signiﬁcance for the listener. The farmer’s tractor in his ﬁeld, the airplane ﬂying overhead, or the wind in the trees are probably of little import for the songbird singing in the hedgerow nearby, and to that end, and at any given moment, they would appear to have little signiﬁcance for a functional auditory scene analysis. Instead, they merely add to the general environmental noise in which all animals live. That is not to say that such noise has no consequences; clearly it has masking effects in the psychoacoustic sense on all the sounds an animal hears. And, of course, something may happen within the general acoustic background that suddenly creates a functionally signiﬁcant auditory object—the cry of a hawk overhead, for example. 4. Selective Attention—Again At this point a reminder about the importance of selective attention in auditory scene analysis seems in order. The process of identifying important auditory objects in noise—no matter what the nature of the noise may be— involves the process of selective attention. Whether that process is bottomup, depending on low-level identiﬁcation of stimulus features, or top-down, depending on some goal-driven process, auditory scene analysis implies a search of the auditory scene for functionally signiﬁcant signals. B. SCENE ANALYSIS UNDER THE SEA Because sound travels so well under water, life under the surface of the sea occurs in an environment that is even noisier than the environment

184

STEWART H. HULSE

on land. Granted that the underwater world is different acoustically (e.g., the speed of sound in water is about ﬁve times greater than the speed in air), all the same necessities for auditory scene analysis exist nevertheless. Underwater creatures must presumably sort their auditory world into the functionally signiﬁcant sounds that mark their own species, that identify potential predators, and so on. 1. Fish Fay (1998, 2000), using classical conditioning techniques, has undertaken a psychophysical analysis of the goldﬁsh’s (Carassius auratus) capacity for stream segregation of two pulse trains with different repetition rates (19 and 85 pps) and spectral envelopes (centered at 238 and 625 Hz, respectively). In one experiment (Fay, 1998), the ﬁsh were trained initially with simultaneous mixtures of the two sound types and then tested for stimulus generalization for pulse rate at each of the two spectral envelopes separately. The data showed that generalization was greatest as the pulse rates approached those associated with the particular spectrum used in initial training. This indicates that the ﬁsh had learned initially to perceptually segregate the two pulse streams on the basis of their spectral composition. Fay also found that if one stream began 500 ms before the other (onset asynchrony), it facilitated stream segregation—just as it does in humans. In a later experiment, Fay (2000) extended these results to show that if sounds of different spectra were interleaved, then stream segregation was a function of the center frequency differences between the interleaved sounds. Again, this replicates similar results with humans (e.g., Miller and Heise, 1950; Van Noorden, 1975). 2. Underwater Mammals It is far beyond the scope of this chapter to review the vast amount of auditory research that has been undertaken with marine mammals—from whales to dolphins (especially the humpbacked whale, Megaptera novaeangliae, and the bottlenose dolphin, Tursiops truncatus). Tyack (1998) has provided an elegant review of this fascinating world. He describes, for example, the sound interactions between humpbacked whales as they sing extremely long songs when by themselves—which may be advertisement displays to attract females—and shorter social sounds, which occur when males ﬁght in groups. Songs are also used for individual recognition among whales, and dolphins, too, have their individual acoustic signatures. I am unaware, however, of any research on underwater communication from a scene analytic point of view. That is true despite the fact that the sounds of underwater creatures are extremely varied and often very loud. Furthermore—given the transmission characteristics of sound in

AUDITORY SCENE ANALYSIS IN ANIMAL COMMUNICATION

185

water—sounds disperse rapidly over very long distances. This must make the underwater auditory scene immensely more complicated—at least potentially—than that above the sea. I believe underwater communication systems may be especially fruitful scenarios for studies of stream segregation and auditory grouping. Perhaps, for example, the long, loud, individually distinctive songs of humpback whales have evolved in part because they are especially easy signals for a listener to segregate from the surrounding acoustic clutter. Interesting questions abound. C. SCENE ANALYSIS IN ANURANS In this section I want to touch very brieﬂy on a few of the data obtained from anurans that speak to the process of auditory scene analysis. A very large body of research exists with anurans, especially frogs, and my review will seek merely to draw attention to the data from this new perspective. Among all research with nonhuman animals, the work on frogs probably constitutes the best example of an attempt to understand conspeciﬁc interactions in terms of scene analysis—without actually using that term. Essentially, the research has been aimed at understanding how frogs detect important signals such as advertisement calls and other mating signals in the midst of a complicated chorus of conspeciﬁcs and other environmental sounds. Frogs have evolved to use just about all possible spectral, loudness, and temporal features of an acoustic signal to communicate with one another. As a general strategy they seek “spaces” in the acoustic characteristics of a chorus in which to place their own species-typical acoustic signals. These include nonoverlapping spectral frequencies, temporal adjustments in call rates (e.g., which can be exquisitely ﬁne-tuned to avoid overlap with other individuals), and, of course, adjustments in call amplitude. Narins and his co-workers (e.g., Brush and Narins, 1989; Narins and Zelick, 1988) have recognized that the problem is one of detecting a signal in functional noise. A great many other researchers have identiﬁed in great detail the acoustic features to which anurans are sensitive in detecting, discriminating, and interacting with conspeciﬁcs, although their focus is only infrequently based on the environmental noise problem (e.g., a brief list includes Gerhardt, 1988, 1989, 1991; Gerhardt and Klump, 1988; Klump and Gerhardt, 1992; Littlejohn, 1969; Ryan, 1988; Simmons et al., 1992; Wells, 1988; Wilczynski et al., 1995; Xu et al., 1994). The interesting characteristic of all this work is that it has essentially mapped out the correlates of auditory streaming for the frog communication system. For example, by making advertisement calls with highly speciﬁc spectral characteristics (relative to the chorus in which the call occurs), the male

186

STEWART H. HULSE

frog has evolved to assure that his call will segregate as a streaming auditory object from the larger environmental surround. Add temporal structure and interactive timing to the spectral structure, and streaming is greatly enhanced. Frogs even have a system in the inferior colliculus that represents the direction of a sound at the single-neuron level (Xu et al., 1994), and the direction from which a sound comes is a relevant feature for stream segregation. It remains to be seen if the speciﬁc application of a scene analysis approach to this work will add new information to what we know about the anuran communication system in general, but the approach certainly suggests that much more attention should be paid to the characteristics of auditory streams in segregating signiﬁcant auditory objects from a frog chorus. I have stressed the importance of selective attention in scene analysis, and this raises the interesting question of whether or not anurans are capable of selectively attending to auditory objects in their environment. Given the probable capacity limitations of the anuran central nervous system, it is doubtful that selective attention occurs as a top-down process in which the listener actively searches in a goal-driven way for important environmental auditory objects. The information suggests, instead, that selective attention—if it occurs at all—must be a bottom-up activity that has evolved to ﬁlter highly speciﬁc information at low levels in the nervous system. Early work by Capranica (e.g., Capranica, 1966; Rose and Capranica, 1983, 1985) describes activity in the frog peripheral nervous system that would have to serve for that analysis. For the future, anuran auditory perception could serve as an ideal model for studying bottom-up processes in scene analysis and vertebrate hearing. D. SCENE ANALYSIS IN BIRDS Oddly enough, apart from the work discussed earlier, studies that would ﬁt the scheme of auditory scene analysis are much less common among avian species than among anurans. There has been emphasis on dyadic interactions through playback studies and so on that have examined territorial and mating strategies based on one-on-one interactions between individual birds or between a bird and song playback from an audio speaker. And Marler (1959) identiﬁed the potential signiﬁcance of “background noise” in animal communication. There are interesting data that can be interpreted from a scene analytic viewpoint, however, and we turn to a brief survey of some of them now. 1. The Dawn Chorus Just as frogs chorus around a mating pool, songbirds chorus in their natural habitat, chieﬂy at dawn. The dawn chorus fosters mating behavior and territorial defense, signalers advertise their sexual attractiveness and defend

AUDITORY SCENE ANALYSIS IN ANIMAL COMMUNICATION

187

their territories, and perceivers react accordingly. The properties of the dawn chorus and its function have attracted a relatively small measure of research attention. The dawn chorus is, of course, a mixture of conspeciﬁc and heterospeciﬁc noise, and offers an extraordinarily complex environment for scene analysis. Henwood and Fabrick (1979), for example, did a quantitative analysis of the acoustic properties of the dawn chorus, especially from the point of view of sound transmission. They concluded that the chorus may have evolved to occur at dawn because atmospheric conditions are most conducive to sound transmissions over long distances at that time of day. Staicer, Spector, and Horn (1996) provide a detailed discussion of why singing may occur at dawn from an ecological and functional perspective. To my knowledge, however, only Hulse et al. (1997) have studied the ability of a songbird, the European starling, to detect speciﬁc birdsong signals in the midst of the dawn chorus, and that work was done in a laboratory setting not in the ﬁeld. Clearly, the ability of listeners to segregate signiﬁcant auditory streams in the dawn chorus and the conditions which foster that segregation deserve more attention than they have received. The extent to which evolution has encouraged the use of easily segregated acoustic channels for communication in the background of the dawn chorus must be as important in songbirds as it is in the chorusing of frogs. 2. Flocking Birds and Colonial Species If the dawn chorus provides an example of multiple species creating an auditory scene, a related example occurs when conspeciﬁcs interact vocally in ﬂocks or groups. Anyone who has heard a winter roost of European starlings, for example, knows that information is presumably being exchanged among individuals in the midst of a true cacophony of conspeciﬁc song. Again, some, but not much, work has been done to study how individuals distinguish themselves in such a situation or what information can, in fact, be transmitted and perceived. Wiley (1976) studied communication among a colony of grackles (Quiscalus quiscula). He found that wide-spectrum sounds were characteristic of the species and noted that they would be especially useful for individual identiﬁcation at close quarters. He even mentioned the cocktail party effect and the solutions it affords for communication among individuals at close range. As with anurans, this work ﬁts the scene analysis model because frequency spectrum is a very potent variable for stream segregation (Bregman and Pinker, 1978; Darwin and Carlyon, 1995). Another example comes from Beecher’s work with cliff swallows (Beecher et al., 1986, 1989). Here, parents easily identify the signature call of their own chicks, and this takes place among large colonies of breeding birds. The evolutionary beneﬁts of segregating the calls of one’s own chicks from the auditory scene in which many chicks are begging are clearly obvious.

188

STEWART H. HULSE

Auditory scene analysis suggests cues that ought to facilitate such segregation, for example, spectral cues and amplitude modulation. Temporal effects such as call repetition and the spacing of calls with respect to other chicks might be especially promising cues to examine for further information, given the importance of a ﬂow of acoustic information in scene analysis. Finally, there is evidence for a cocktail party effect in colonial king penguins (Aubin and Jouventin, 1998). Penguin chicks can identify the calls of their parents at sound levels which are less than the general sound level of the background colony noise. This ability has obvious adaptive signiﬁcance. 3. Environmental Noise In the larger auditory scene provided by multitudes of mixed natural sounds, some research has focused on the abilities of perceivers to detect important signals in that environment. Part of the work has dealt primarily with the acoustic ﬁlters provided by the environment that modulate the transmission of functional signals (e.g., Dabelsteen et al., 1993; Holland et al., 1998; Wiley and Richards, 1982). This research has often emphasized the degradation of signals as a ranging cue and the establishment of the distances in various environments over which functional signals can be transmitted and identiﬁed. Other research has adopted a more psychophysical model to determine thresholds and critical ratios for the detection of auditory objects (e.g., birdsong) in background noises, including white noise (for a review see, e.g., Klump, 1996; see Langemann et al., 1998, for a speciﬁc application to the great tit, Parus major). Klump makes the point that estimates of masked thresholds, rather than absolute thresholds, are important for estimating the properties of functional signals in a larger, noisy environment. I would simply add that we need to know more about the aspects of masking that facilitate (or hinder) the identiﬁcation of auditory objects in the larger, noisy environment. It is possible, for example, that stream segregation and the cues that facilitate it will work to overcome the general effects of masking that are reﬂected in critical ratios based on signal detection in noise. Klump has spoken to this issue, and we shall return to it in due course. 4. Song Timing and Communication Networks There is another aspect of birdsong structure that may be especially interesting from a scene analytic point of view. I refer to duetting, song overlapping, and other scenarios where two or more birds are engaged in acoustic interaction. There is evidence, for example, that the timing of an overlap (onset asynchrony) in the song of two interacting conspeciﬁcs can indicate the intended receiver and escalate behavioral arousal (e.g., Dabelsteen et al., 1997; Hultsch and Todt, 1982; Naguib, 1999; Naguib and Todt, 1997; for an especially good review, see Todt and Naguib, 2000). From a scene analytic

AUDITORY SCENE ANALYSIS IN ANIMAL COMMUNICATION

189

point of view, this may be because an auditory stream that leads in time compared to another auditory stream will capture attention ﬁrst (Bregman, 1990). Especially interesting is the fact that nightingales who are merely listening to a dyadic interaction between two other nightingales also respond to such timing effects (Naguib and Todt, 1997). Similar effects in duetting may occur where the timing of the interaction between the duetting individuals may facilitate segregation of the duet from the larger auditory scene (Todt and Naguib, 2000). There is another possible form of interaction among conspeciﬁcs that, if true, may involve scene analysis in an especially interesting way. I refer to the form of acoustic interaction that occurs potentially between conspeciﬁcs interacting in networks that are larger than simple dyads (for example, Dabelsteen, 1992; McGregor, 1993; McGregor and Dabelsteen, 1996; Todt and Naguib, 2000). The idea that conspeciﬁcs may eavesdrop on an ongoing acoustic interaction and gain useful information that way is an interesting possibility that comes from this work. Birds might assess the ﬁtness of a conspeciﬁc by listening to the bird’s interaction with some third bird, for example. Also, networks of signalers might be involved in the spread of alarm calls. In each case, any auditory cue that can specify individuals and their location in the network is especially important, and scene analysis suggests where to look for relevant processes. In particular, the timing of interactions in networks and the function of timing in segregating the auditory objects represented by the sound of each interacting participant ought to be especially important. That would also be true for any eavesdropper listening in on the auditory scene. There is evidence that onset asynchrony facilitates the salience of an auditory object in perception. In particular, sounds that lead others in time tend to dominate in perception (e.g., Darwin and Ciocca, 1992; Grafe, 1996). E. SCENE ANALYSIS IN PRIMATES Virtually all primates use auditory means for communication, especially for social communication in groups. By the same token, primates live in the same noisy environments that other animals do, environments that for some species of primates become noisier all the time. (As I write this, there is a large construction machine operating not 6 ft outside my window—an unwelcome object in my auditory scene.) Apart from the current interest in human auditory scene analysis, however, there is little, if any, comparable work that I am aware of that has been done with nonhuman primates. That is not to say, of course, that systems of auditory communication among nonhuman primates have not received their due share of attention. There have been many studies, in numbers too far beyond the scope of

190

STEWART H. HULSE

this chapter to mention, dealing with vocal interactions for many species of nonhuman primates. These studies range from analyses of the acoustic features of vocalizations that are used for communication to studies of the interactions that in fact take place within and between social groups. Hauser (1996) has provided an elegant summary and discussion of this work, and the reader is encouraged to turn to that resource for further information. But despite this plethora of information, the fact remains that there is virtually no work on auditory scene analysis in nonhuman primates designed with that particular process in mind. Snowdon (personal communication, 2001) has observed that only a few species of primates produce highly elaborate calls similar to song in birds (titi monkeys, gibbons, possibly long calls of marmosets and tamarins): such species are rarely found sympatrically and, when they are found together, may be likely to call at different times of day or in different strata. In this case, it may be that evolution has solved the scene analysis problem by assuring that the relevant communication signals rarely interfere with one another. Still, one must be on the lookout for scene analysis at work—if frequency differences between simultaneous calls are used by primates in the same habitat, that would be a clear example of stream segregation and auditory scene analysis. Sometimes scene analysis is so obvious it is overlooked. In any case, further work with primates should be useful. Research with humans provides a general heuristic tool to guide research, and my discussion of anurans and birds deﬁnes situations, especially social situations, that ought to be especially apt for work with nonhuman primates. Given that primates have central auditory systems that are far more developed than those of, say, frogs, nonhuman primates also ought to be especially worthy candidates for studies of the top-down, goal-driven, selective-attention aspects of auditory scene analysis. One wonders, for example, if dominant members of a group have learned to use onset asynchrony in communication to establish primary auditory streams and, thus, dominance hierarchies. Do dominant males establish a salient auditory stream by speaking ﬁrst? The possibilities are intriguing. Fishman, Reser, Arezzo and Steinschneider (2001) have reported measurements of neurological evidence for auditory scene analysis in rhesus monkeys (Macaca fascularis). They recorded from area A1 in primary auditory cortex with arrays of 14 microelectrodes. Recordings were made while awake monkeys listened to sequences of A and B tones that alternated between two frequencies at 5-, 10-, 20-, or 40-Hz pulse rates. At fast pulse rates (i.e., 20 or 40 Hz) they found suppression of neural responding to the B tone such that responding to the A tone was half the stimulus pulse rate. Furthermore, the degree of suppression grew as the frequency difference between A and B increased. They attribute the change in suppression to forward

AUDITORY SCENE ANALYSIS IN ANIMAL COMMUNICATION

191

masking. To my knowledge, this is the ﬁrst report of a neurological process that clearly mimics the psychophysical data that appear with stream segregation based on changes in frequency and pulse rate (Bregman and Campbell, 1971; Hulse et al., 1997; MacDougall-Shackleton et al., 1998). This is also an important result because it suggests, among other things, that the neural representation of stream segregation exists in nonhuman primates, at least in part, at a fairly high level in the auditory system.

V. THEORY I turn ﬁnally to a brief presentation of theoretical schemes that have been suggested to draw together the empirical processes that deﬁne auditory scene analysis. There are three tacks to be mentioned: top-down, goal-driven approaches; bottom-up peripheral explanations; and, ﬁnally, theories based on mathematical modeling. All of these involve an appreciation of the physics and mathematics of acoustical analysis and are, therefore, formally quite complex. In the interests of brevity and simplicity, I attempt to convey the essence of the theories rather than their details. At the same time, I suggest resources to which the interested reader may turn for further information. A. TOP-DOWN, GOAL-DRIVEN THEORY Perhaps the best example of a top-down approach to auditory scene analysis comes from the seminal book by Bregman (1990). The initial portions of this chapter describe the essence of his theory. The theory is not mathematical, nor is it based for the most part on a complex analysis of the waveforms and formal structure of auditory signals. It does, however, draw heavily upon the principles of Gestalt psychology, the principles of grouping, proximity, similarity, apparent motion, and so on, that describe functionally the way auditory scene analysis seems to work. We have discussed the most salient organizing principles at the outset of this chapter, stressing the idea of functional auditory objects. The theory also draws upon the ideas of primitive and schema-driven processes, as we have also discussed earlier. Though the distinction is sometimes hard to make, primitive principles of organization are unlearned and independent of experience, while schema-driven principles are learned and goal-(object-)directed, leading to the formation of functional auditory objects. Bregman is by no means alone in adopting this Gestalt approach. McAdams and his colleagues (e.g., McAdams and Bigand, 1993) have also worked within the top-down approach, identifying the acoustic conditions that produce auditory objects (McAdams uses the term auditory

192

STEWART H. HULSE

images)—even in young children (McAdams et al., 1990). Similarly, Warren (1999) has drawn heavily upon top-down principles in his treatment of auditory perception. Jones and her colleagues (e.g., Boltz, 1994; Jones and Yee, 1993) have studied the role of the temporal organization of sound patterns in establishing auditory objects. And when it comes to selective attention—which is always of potential importance in scene analysis—the distinction by Yantis (2000) between topdown, goal-directed and bottom-up attention should always be borne in mind. B. BOTTOM-UP THEORIES Bottom-up theories of scene analysis stress the idea that there is sufﬁcient information in the acoustic properties of the auditory signal itself to produce stream segregation and scene analysis. In Bregman’s terms, this approach refers to primitive principles in scene analysis, that is, principles that are unlearned. 1. Peripheral Channeling Hartmann and Johnson (1991) have reported a series of experiments and a theoretical analysis that is virtually a pure bottom-up description of scene analysis. They stress the proposition that stream segregation can be accounted for by the acoustic information contained in peripheral channels. By a peripheral channel they mean basic acoustic modes such as those based on frequency (tonotopic cues), spectrum, and ear presentation (lateralization cues). They studied the situation where two interleaved melodies were varied in pitch, location, temporal structure, and spectrum. Their experiments show that, indeed, if such peripheral channels are manipulated either to foster stream segregation or to work against it, then a peripheral channeling model describes most (but not all) of the data quite well. Acoustic cues which did not involve peripheral channeling, such as intensity and stimulus duration, could lead to stream segregation, but always to a lesser extent than if peripheral channels were involved. As Hartmann and Johnson themselves put it (1991, p. 175), “The results of the experiments lead to the practical conclusion that if a composer, or orchestrator, or gambler needs to make a prediction as to whether two melodies will be segregated or not, then the best strategy is to base the prediction on whether the two melodies excite different peripheral channels or not.” It is important to recognize, however, that stream segregation was obtained if cues that did not involve peripheral channels were used, so this bottom-up approach cannot provide the entire story—as Hartmann and Johnson freely acknowledge.

AUDITORY SCENE ANALYSIS IN ANIMAL COMMUNICATION

193

2. Comodulation Masking Release Klump and his associates (Klump, 1996; Klump and Langemann, 1995; Langemann et al., 1998) have studied another low-level system based on principles of auditory masking that could well be involved in auditory scene analysis. They note that animals in the ﬁeld are able to detect signals in noise at a lower signal-to-noise ratio than if masking levels are obtained in the quiet conditions of the laboratory. This leads to the possibility that natural ﬂuctuations in the levels of the background noise and the signal may facilitate perception of the signal. This could happen in at least three ways. First, assume that signal and noise are broken into different frequency channels (remember Hartmann and Johnson’s approach earlier). Assume further that amplitude ﬂuctuations in the frequency channels of the noise are correlated. Then if an introduced signal reduces that correlation, the change in correlation could be detected by the auditory system. This approach emphasizes cross talk between frequency channels. Second, assume that a signal is detected because the auditory system can detect a change in the modulation pattern of the envelope of the combined masker and signal when the signal is introduced. This emphasizes events in a single frequency channel. Third, assume that the frequency channels of a noise are amplitude modulated in coherent fashion. When this happens, there will be large ﬂuctuations in the amplitude envelope of the masker; the masker will be more effective at large amplitudes, but less effective at low amplitudes. During periods of low amplitude, the likelihood that an added signal will be detected will be increased—in contrast to the condition in which the masker is not amplitude modulated. This approach emphasizes temporal factors in detection. Klump and Langemann (1995; Langemann et al., 1998) studied comodulation masking release in starlings and great tits and found substantial release from masking for variations in the bandwidth of the masker and the rate of envelope ﬂuctuations in the masker. As Klump (1996) points out, the data from the laboratory on birds appears to compare quite well with those obtained from humans. There is no question that the natural habitat of animals is full of noise, and sometimes the interaction of signal and noise can work in fact to facilitate detection of the signal. Wiley and Richards (1982), for example, found that there was a 15 dB improvement in signal detection during exposure to (ﬂuctuating) sounds from the birds’ natural habitat. Fluctuation of the noise may serve to facilitate the segregation of song. There are some data from humans that seem especially relevant here. Oh and Lutﬁ (1999) found, contrary to what one would expect from our discussion, that people were no more skillful in identifying a 1 kHz signal in maskers of everyday sounds (such as a dog barking, a car horn, a baby crying)

194

STEWART H. HULSE

than they were in white noise. However, there was less masking by everyday sounds than by noise on occasions when the observers reported that they easily recognized the sounds. One would like to know how the converse of this experiment would work, that is, the extent to which the detection of familiar sounds, as distinct from sine tones, in noise would be facilitated. Arbogast and Kidd (2000) have shown that informational masking is a factor when people are asked to identify where in space a target signal was emanating—accuracy and response times were improved when the target came from an expected location. Regardless of the particular bottom-up approach that one favors, the important point is that there are large amounts of auditory information in mixtures of signals and noise that co-occur and interact with each other. This information is based on things that happen in the auditory periphery at the level of frequency analysis, spatial location, and the interactions of information in peripheral acoustic channels. There is little reason to doubt, as Hartmann and Johnson (1991) show so clearly, that this information can be used in auditory scene analysis. The question remains whether or not peripheral mechanisms are entirely adequate for a full account of auditory scene analysis. The idea that auditory perception is guided by learning and by a functional analysis of the objects in the auditory scene certainly suggests that bottom-up mechanisms are, in and of themselves, not sufﬁcient for all species. C. FORMAL MODELS OF AUDITORY SCENE ANALYSIS We should not be surprised to ﬁnd that the phenomena associated with auditory scene analysis have been studied closely by those who build formal, mathematically oriented models of perceptual phenomena. This is a large literature growing at a rapid rate, and it would be far beyond our perspective to do more than mention a few articles that will outline the ideas involved and lead the interested reader to places where further information can be found. Beauvois and Meddis (1991, 1996) have developed a model which describes auditory stream segregation phrased in terms of known properties of the peripheral auditory nervous system. The model is based on the idea that mechanisms may exist in the nervous system that will render one of two overlapping sounds as accentuated over the other; both sounds are heard, but one has salience, and hence captures attention. The claim is that this process arises because of the stochastic nature of nervous system activity in the periphery. The model is appropriately complex but, in brief, comprises bandpass ﬁlters that provide an initial frequency analysis, simulations of inner-hair-cell/auditory-nerve synapses (Meddis, 1990), a series of three excitation-level pathways, a comparator that selects the dominant channel in the excitation-level pathways, and an attenuator that reduces the excitation

AUDITORY SCENE ANALYSIS IN ANIMAL COMMUNICATION

195

in the nondominant pathways. Stream segregation is determined by the relative amplitudes in the dominant and attenuated nondominant paths. The model works surprisingly well in accounting for some of the basic phenomena of scene analysis, for example, the frequency and rate boundaries which determine whether or not stream segregation will occur. The theory has received conﬁrmation by Rose and Moore (2000) who also ﬁnd that their data support Hartmann and Johnson’s peripheral channeling model described earlier (Hartmann and Johnson, 1991). McCabe and Denham (1997) have also developed a model that accounts for stream segregation and selective attention. For further reading about the application of formal modeling to auditory scene analysis, Rosenthal and Okuno (1998) have provided an excellent collection of papers by a number of researchers, and that volume will lead the interested reader to many other resources.

VI. FINAL COMMENTS After reading the material in this chapter, I hope the reader will have a basic appreciation of the fundamental principles of auditory scene analysis. Even more, I hope the reader will see how those principles can be applied to further our understanding of auditory communication in animals. There has not been very much research with nonhuman animals aimed directly at auditory scene analysis, but I have tried to show that there are many facets of such communication that are amenable to study with the scene analytic approach. Over the years some researchers have noted, usually in passing, that animals live in a noisy environment, and animal communication must somehow deal with this noise if it is to succeed. The principles offered by scene analysis can be a tool to guide investigations of this problem. Work with animals can be especially important because research can use evolutionary principles as a tool to analyze the capacity for auditory scene analysis. Where in evolution did scene analysis arise, and what were the anatomical and physiological structures that enabled it to take place? We know from Fay’s work (Fay, 1998) that scene analysis exists in ﬁsh, and my descriptions of our work with starlings establishes the process in songbirds. The work of Fishman et al. (2001) established the process in nonhuman primates at the neurological level. But what are the true roots in phylogeny for auditory scene analysis? My suspicion is that the roots are very old because, as I have noted before, there is an enormous adaptive advantage for any creature that parses the auditory world into objects which have signiﬁcance for survival: the sounds of environmental dangers, predators, rivals, and potential mates.

196

STEWART H. HULSE

In all likelihood, early scene analysis was based on bottom-up processes which depended extensively on organizing principles contained in the structure of the stimulus itself. If that is so, when did top-down, learned, goaldirected processes emerge? These almost certainly involved selective attention among auditory objects. When did that capacity ﬁrst appear? These are fundamental questions in the study of perception. It seems to me that auditory scene analysis provides an arena in which these questions can be studied with proﬁt. VII. SUMMARY In this chapter, I ﬁrst introduced the concept of auditory scene analysis, the principle that the auditory perceptual system somehow parses a noisy world into individual auditory components—auditory objects—that often occur simultaneously but are attended to selectively. I then discussed some initial studies which demonstrated for the ﬁrst time that auditory scene analysis occurs in nonhuman animals, European starlings in particular. I turned next to studies of other taxa (e.g., other songbirds, ﬁsh, nonhuman primates, and especially anurans) which suggest that auditory scene analysis may be an important principle helping to organize the perception of auditory events. Finally, I discussed two classes of theory—top-down and bottom-up—that have been developed to account for auditory scene analysis. Throughout, I have tried to identify situations in which it may be especially proﬁtable to pursue further research on auditory scene analysis in nonhuman species. References Arbogast, T., and Kidd, Jr. G. (2000). Evidence for spatial tuning in informational masking using the probe-signal method. J. Acoust. Soc. Am. 108, 1803–1810. Aubin, T., and Jouventin, P. (1998). Cocktail-party effect in king penguin colonies. Proc. R. Soc. London Ser. B 265, 1665–1673. Beauvois, M. W., and Meddis, R. (1991). A computer model of auditory stream segregation. Q. J. Exp. Psychol. 43(3), 517–541. Beauvois, M. W., and Meddis, R. (1996). Computer simulation of auditory stream segregation in alternating-tone sequences. J. Acoust. Soc. Am. 99(4), 2270–2280. Beecher, M. D., Medvin, M. B., Stoddard, P. K., and Leask, P. (1986). Acoustic adaptations for parent–offspring recognition in swallows. Exp. Biol. 45, 179–193. Beecher, M. D., Leask, P., Stoddard, P. K., and Medvin, M. B. (1989). Individual recognition by voice in swallows: Signal or perceptual adaptation? In “The comparative psychology of audition: Perceiving complex sounds” ( R. J. Dooling and S. H. Hulse, Eds.), pp. 277–294. Erlbaum, Hillsdale, NJ. Benney, K. S., and Braaten, R. F. (2000). Auditory scene analysis in estrildid ﬁnches (Taenopygia guttata and Lonchura striata domestica): A species advantage for detection of conspeciﬁc song. J. Comp. Psychol. 114(2), 174–182.

AUDITORY SCENE ANALYSIS IN ANIMAL COMMUNICATION

197

Boltz, M. G. (1994). Changes in internal tempo and effects on the learning and remembering of event durations. J. Exp. Psychol. Hum. Percep. Perform. 20(5), 1154–1171. Bregman, A. S. (1981). Asking the “what for” question in auditory perception. In “Perceptual Organization” (M. Kubovy and J. Pomerantz, Eds.), pp. 99–118. Erlbaum, Hillsdale, NJ. Bregman, A. S. (1990). “Auditory scene analysis.” MIT Press, Cambridge, MA. Bregman, A. S., and Campbell, J. (1971). Primary auditory stream segregation and perception of order in rapid sequences of tones. J. Exp. Psychol. 89(2), 244–249. Bregman, A. S., and Dannenbring, G. L. (1977). Auditory continuity and amplitude edges. Can. J. Psychol. 31, 151–159. Bregman, A. S., and Pinker, S. (1978). Auditory streaming and the building of timbre. Can. J. Psychol. 32, 19–31. Brush, J. S., and Narins, P. M. (1989). Chorus dynamics of a neotropical assemblage: Comparison of computer simulation and natural behaviour. Anim. Behav. 37(1), 33–44. Busnel, R.-G. (1968). Acoustic communication. In “Animal Communication: Techniques of Study and Results of Research” (T. A. Sebeok, Ed.), pp. 233–251. Indiana Univ. Press, Bloomington. Capranica, R. R. (1966). Vocal response of the bullfrog to natural and synthetic mating calls. J. Acoust. Soc. Am. 40(5), 1131–1139. Cherry, E. C. (1953). Some experiments on the recognition of speech, with one and with two ears. J. Acoust. Soc. Am. 25(5), 975–979. Culling, J. F., and Darwin, C. J. (1993). The role of timbre in the segregation of stimulants voices with intersecting Fo contours. Percep. Psychophys. 54(3), 303–309. Dabelsteen, T. (1992). Interactive playback: A ﬁnely tuned response. In “Playback and Studies of Animal Communication” (P. K. McGregor, Ed.), pp. 97–109. Plenum Press, New York. Dabelsteen, T., Larsen, O., and Pedersen, S. B. (1993). Habitat-induced degradation of sound signals: Quantifying the effects of communication sounds and bird location on blur ratio. J. Acoust. Soc. Am. 93, 2206–2220. Dabelsteen, T., McGregor, P. K., Holland, J., Tobias, J., and Pedersen, S. B. (1997). The Signal function of overlapping singing in male robins. Anim. Behav. 53, 249–256. Darwin, C. J., and Carlyon, R. P. (1995). Auditory grouping. In “Hearing,” 2nd ed. (B. C. J. Moore, Ed.), pp. 387–424. Academic Press, San Diego. Darwin, C. J., and Ciocca, V. (1992). Grouping in pitch perception: Effect of onset asynchrony and ear of presentation of a mistuned component. J. Acoust. Soc. Am. 91, 3381–3390. Eens, M. (1997). Understanding the complex song of the European starling: An integrated ethological approach. In “Advances in the Study of Behavior” (P. J. B. Slater, J. S. Rosenblatt, C. T. Snowdon, and M. Milinski, Eds.), Vol. 26, pp. 355–434. Academic Press, New York. Fay, R. R. (1998). Auditory stream segregation in goldﬁsh (Carassius auratus). Hear. Res. 129, 69–76. Fay, R. R. (2000). Spectral constrasts underlying auditory stream segregation in goldﬁsh (Carassius auratus). J. Assoc. Res. Otolaryng. 1, 120–128. Fishman, Y., Reser, D. H., Arezzo, J. C., and Steinschneider, M. (2001). Neural correlates of auditory stream segregation in primary auditory cortex of the awake monkey. Hear. Res. 151, 167–187. Gerhardt, H. C. (1988). Acoustic properties used in call recognition by frogs and toads. In “The Evolution of the Amphibian Auditory System” (B. Fritzsch, M. J. Ryan, W. Wilczynski, T. E. Hetherington, and W. Walkowiak, Eds.), pp. 455–483. Wiley, New York. Gerhardt, H. C. (1989). Acoustic pattern recognition in anuran amphibians. In “The Comparative Psychology of Audition: Perceiving Complex Sounds” (R. J. Dooling and S. H. Hulse, Eds.), pp. 175–197. Erlbaum, Hillsdale, NJ.

198

STEWART H. HULSE

Gerhardt, H. C. (1991). Female mate choice in treefrogs: Static and dynamic acoustic criteria. Anim. Behav. 42, 615–635. Gerhardt, H. C., and Klump, G. M. (1988). Masking of acoustic signals by the chorus background noise in the green treefrog: A limitation on mate choice. Anim. Behav. 36, 1247–1249. Grafe, T. U. (1996). The function of call alternation in the African reed frog (Hyperolius marmoratus): Precise call timing prevents auditory masking. Behav. Ecol. Sociobiol. 38(3), 149–158. Hartmann, W. M., and Johnson, D. (1991). Stream segregation and peripheral channeling. Mus. Percep. 2, 155–184. Hauser, M. D. (1996). “The Evolution of Communication.” MIT Press, Cambridge, MA. Henwood, K., and Fabrick, A. (1979). A quantitative analysis of the dawn chorus: Temporal selection for communicatory optimization. Am. Nat. 114, 260–274. Holland, J., Debelsteen, T., and Pedersen, S. B. (1998). Degradation of wren Troglodytes troglodytes song: Implications for information transfer and ranging. J. Acoust. Soc. Am. 103(4), 2154–2166. Hulse, S. H., MacDougall-Shackleton, S. A., and Wisniewski, A. B. (1997). Auditory scene analysis by songbirds: Stream segregation of birdsong by European starlings (Sturnus vulgaris). J. Comp. Psychol. 111, 3–13. Hultsch, H., and Todt, D. (1982). Temporal performance roles during vocal interactions in nightingales (Luscinia megarynchos). Behav. Ecol. Sociobiol. 11, 253–260. Iverson, P. (1995). Auditory stream segregation by musical timbre: Effects of static and dynamic acoustic attributes. J. Exp. Psychol.: Hum. Percep. Perform. 21, 751–763. Jones, M. R., and Yee, W. (1993). Attending to auditory events: The role of temporal organization. In “Thinking in Sound” (S. McAdams and E. Bigand, Eds.), pp. 69–112. Clarendon Press, Oxford. 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 Univ. Press, Ithaca, NY. Klump, G. M., and Gerhardt, H. C. (1992). Mechanisms and functions of call-timing in male– male interactions in frogs. In “Playback and Studies of Animal Communication” (P. K. McGregor, Ed.), pp. 153–174. Plenum Press, New York. Klump, G. M., and Langemann, U. (1995). Comodulation masking release in a songbird. Hear. Res. 87, 157–164. 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. Littlejohn, M. J. (1969). Acoustic interaction between two species of leptodactylid frogs. Anim. Behav. 17(4), 785–791. Loesche, P., Beecher, M. D., and Stoddard, P. K. (1992). Perception of cliff swallow calls by birds (Hirundo pyrrhonota and Sturnus vulgaris) and Humans (Homo sapiens). J. Comp. Psychol. 106(3), 239–247. MacDougall-Shackleton, S. A., Hulse, S. H., Gentner, T. Q., and White, W. (1998). Auditory scene analysis by European starlings (Sturnus vulgaris): Perceptual segregation of tone sequences. J. Acoust. Soc. Am. 103, 3581–3587. Marler, P. (1959). Developments in the study of animal communication. In P. R. “Darwin’s Biological Work: Some Aspects Reconsidered” (P. R. Bell, Ed.), pp. 150–206. Cambridge Univ. Press, Cambridge, UK. McAdams, S., and Bigand, E. (Eds.) (1993). “Thinking in Sound.” Clarendon Press, Oxford. McAdams, S., Betoncini, J., and Bobrow, J. (1990). Organization and discrimination of repeating sound sequences by newborn in fants. Paper presented at the 120th meeting of the Acoustical Society of America, San Diego, CA.

AUDITORY SCENE ANALYSIS IN ANIMAL COMMUNICATION

199

McCabe, S. L., and Denham, M. J. (1997). A model of auditory streaming. J. Acoust. Soc. Am. 101, 1611–1621. McGregor, P. K. (1993). Signaling in territorial systems: A context for individual recognition, ranging and eavesdropping. Trans. R. Soc. London Ser. B. 340, 237–244. McGregor, P. K., and Dabelsteen, T. (1996). Communication networks. In “Ecology and Evolution of Acoustic Communication in Birds” (D. E. Kroodsma and E. H. Miller, Eds.), pp. 409–425. Cornell Univ. Press, Ithaca, NY. Meddis, R. (1990). Implementation details of a computational model of the inner haircell/auditory-nerve synapse. J. Acoust. Soc. Am. 87, 1813–1816. Miller, G. A., and Heise, G. A. (1950). The trill threshold. J. Acoust. Soc. Am. 22, 637–638. Naguib, M. (1999). Effects of song overlapping and alternating on nocturnally singing nightingales. Anim. Behav. 58, 1061–1067. Naguib, M., and Todt, D. (1997). Effects of dyadic vocal interactions on other conspeciﬁc receivers in nightingales. Anim. Behav. 54, 1535–1543. Narins, P. M., and Zelick, R. (1988). The effect of noise on auditory processing and behavior in amphibians. In “The Evolution of the Amphibian Auditory System” (M. J. Ryan, W. Wilczynski, T. E. Hetherington, and W. Walkowiak, Eds.), pp. 511–536. Oh, E. L., and Lutﬁ, R. A. (1999). Information masking by everyday sounds. J. Acoust. Soc. Am. 106(6), 3521–3528. Pashler, H. (Ed.). (1998). “Attention.” Psychology Press. Hove, East Sussex, England. Rose, G. J., and Capranica, R. R. (1983). Temporal selectivity in the central auditory system of the leopard frog. Science 219, 1087–1089. Rose, G. J., and Capranica, R. R. (1985). Sensitivity to amplitude modulated sounds in the anuran nervous system. J. Neurophysiol. 53(2), 446–465. Rose, M. M., and Moore, B. C. J. (2000). Effects of frequency and level on auditory stream segregation. J. Acoust. Soc. Am. 108(3), 1209–1214. Rosenthal, D., and Okuno, H. G. (Eds.) (1998). “Computational Auditory Scene Analysis.” Erlbaum, Hillsdale, NJ. Ryan, M. J. (1988). Constraints and patterns in the evolution of anuran acoustic communication. In “The Evolution of the Amphibian Auditory System” (B. Fritzsch, M. J. Ryan, W. Wilczynski, T. E. Hetherington, and W. Walkowiak, Eds.), pp. 637–677. Wiley, New York. Simmons, A. M., Schwartz, J. J., and Ferragamo, M. (1992). Auditory nerve representation of a complex communication sound in background noise. J. Acoust. Soc. Am. 91(5), 2831– 2844. Singh, P. G., and Bregman, A. S. (1997). The inﬂuence of different timbre attributes on the perceptual segregation of complex-tone sequences. J. Acoust. Soc. Am. 102, 1943–1952. Staicer, C. A., Spector, D. A., and Horn, A. G. (1996). The dawn chorus and other diel patterns in acoustics signaling. In “Ecology and Evolution of Acoustic Communication in Birds” (D. E. Kroodsma and E. H. Miller, Eds.), pp. 426–453. Cornell Univ. Press, Ithaca, NY. Todt, D., and Naguib, M. (2000). Vocal interaction in birds: The use of song as a model in communication. In “Advances in the Study of Behavior” (P. J. B. Slater, J. S. Rosenblatt, C. T. Snowdon, and T. J. Roper, Eds.), Vol. 29, pp. 247–296. Academic Press, New York. Tyack, P. (1998). Acoustic communication under the sea. In “Animal Acoustic Communication” (S. L. Hopp, M. J. Owren, and C. S. Evans, Eds.), pp. 163–220. Springer-Verlag, Berlin. Van Noorden, L. P. A. S. (1975). “Temporal Coherence in the Perception of Tone Sequences.” Eindhoven Univ. Technol. The Netherlands. Vliegen, J., Moore, B. C. J., and Oxenham, A. J. (1999). The role of spectral structure and periodicity cues in auditory stream segregation, measured using a temporal discrimination task. J. Acoust. Soc. Am. 106, 938–945.

200

STEWART H. HULSE

Warren, R. M. (1999). “Auditory Perception: A New Analysis and Synthesis.” Cambridge Univ. Press, New York. Wells, K. D. (1988). The effect of social interactions on anuran vocal behavior. In “The Evolution of the Amphibian Auditory System” (B. Fritzsch, M. J. Ryan, W. Wilczynski, T. E. Hetherington, and W. Walkowiak, Eds.), pp. 433–454. Wiley, New York. Wilczynski, W., Rand, A. S., and Ryan, M. J. (1995). The processing of spectral cues by the call analysis system of the tungara frog, Physalaemus pustilosis. Anim. Behav. 49, 911–929. Wiley, R. H. (1976). Communication and spatial relationships in a colony of common grackles. Anim. Behav. 24, 570–584. Wiley, R. H., and Richards, D. G. (1982). Adaptations for acoustic communication in birds: Sound transmission and signal detection. In “Acoustic Communication in Birds: Production, Perception, and Design Features of Sounds” (D. E. Kroodsma and E. H. Miller, Eds.), Vol. 1, pp. 132–181. Academic Press, New York. Wisniewsky, A. B., and Hulse, S. H. (1997). Auditory scene analysis in European starlings (Sturnus vulgaris): Discrimination of starling song segments, their segregation from conspeciﬁc songs, and evidence for conspeciﬁc song categorization. J. Comp. Psychol. 111, 337–350. Xu, J., Gooler, D. M., and Feng, A. S. (1994). Single neurons in the frog inferior colliculus exhibit direction-dependent frequency selectivity to isointensity tone bursts. J. Acoust. Soc. Am. 95, 2160–2170. Yantis, S. (2000). Goal-directed and stimulus-driven determinants of attentional control. In “Attention and Performance” (S. Monsell and J. Driver, Eds.), Vol. XVIII, pp. 73–103. MIT Press, Cambridge, MA. Yost, W. A. (1997). The cocktail party phenomenon: Forty years later. In “Binaural and Spatial Hearing in Real and Virtual Environments” (R. H. Gilkey and T. R. Anderson, Eds.), (pp. 329–347). Erlbaum, Mahwah, NJ.

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 31

Electric Signals: Predation, Sex, and Environmental Constraints PHILIP K. STODDARD DEPARTMENT OF BIOLOGICAL SCIENCES FLORIDA INTERNATIONAL UNIVERSITY MIAMI, FLORIDA 33199

I. INTRODUCTION Animal signals have adapted to the environment in ways both obvious and subtle. Signals of many animals show spectral and temporal adaptations to improve transmission, to circumvent environmental noise, and to avoid predators (reviews by Dusenbery, 1992; Bradbury and Vehrencamp, 1998). Not surprisingly, as animals move to new habitats with different conditions, their signals must evolve as well. We are familiar with the idea that habitat use can guide or constrain signal evolution, but what about the converse? Is the relationship between signal and habitat strong enough that preexisting signal traits with limited plasticity can constrain an animal’s movements between habitats or into new habitats? This question is probably not best addressed in the oscine songbirds because of their extreme evolutionary plasticity. Territorial song of Emberizine sparrows illustrates this point: white-throated sparrows (Zonotrichia albicollis) sing a simple song of two or three pure-toned whistles whereas white-crowned sparrows (Z. leucophrys) sing complex songs consisting of buzzes, FM transients, and pure-toned whistles. In electric ﬁsh, however, a picture is emerging that signal adaptations to particular habitats may constrain further habitat choices. This story is at the model stage, synthesized from ecological studies in the ﬁeld and physiology studies in the laboratory. Signiﬁcant portions of these models need further testing, particularly in the ﬁeld. I hope this chapter focuses future research efforts along these lines. The electric organ discharge is known universally by the acronym EOD. The same EODs used for active electrolocation can double as communication signals. Thus, it is easy to see how adaptations driven by one function (i.e., electrolocation) would have signiﬁcant consequences for the other function 201

C 2002 by Academic Press Copyright All rights of reproduction in any form reserved. 0065-3454/02 $35.00

202

PHILIP K. STODDARD

(i.e., communication). In this chapter I explore how the external forces of electrostatic law, predation, and aquatic environment interact with internal constraints to shape the evolution of electric communication signals. The key internal constraints I address are, of course, the need to electrolocate and communicate, but also the neural substrates these ﬁsh had before their divergent radiation, as revealed by structures common to extant members of their clades. A. A FORK IN THE ROAD: PULSE VERSUS WAVE EODS Weakly electric ﬁsh have evolved at least four times: in the Neotropical knifeﬁsh, order Gymnotiformes; in the West African elephant noses and allies, superfamily Mormyroidae; and in isolated species within two families of African catﬁsh, order Siluriformes (Fig. 1) (Fink and Fink, 1981; Finger et al., 1986; Baron, 1994; Baron et al., 1994). In this chapter, I focus on the

FIG. 1. Phylogeny of the teleost bony ﬁshes (after Lauder and Liem, 1983), showing the families with electroreception (∗) and electrogenesis (◦ ). A few individual genera and species have evolved electrogenesis independently (indicated in parentheses).

ELECTRIC SIGNALS

203

FIG. 2. Phylogeny of the neotropical electric ﬁsh order Gymnotiformes (after Albert and Campos-da-Paz, 1998) showing representative EOD waveforms. Arrows represent likely independent origins of the multiphasic EODs. The independent origin of biphasy in the Apteronotidae is inferred from its novel electric organ. The site of independent origin within the pulse clade is more speculative. An alternative scenario is that the family Gymnotidae evolved a multiphasic EOD independently from its sister clades, although it shares with them the mechanism for generation of the large negative phase. Most of the EODs shown here have equal amounts of energy above and below 0 V (dashed line), which nulls the dc in the EOD spectrum and reduces energy that may be too detectable by predators with low-frequency electroreception such as catﬁsh. These EODs came from the following species: Electrophoridae (Electrophorus electricus) electric eel, Gymnotidae (Gymnotus cf. carapo), Rhamphichthyidae (Steatogenys sp.), Hypopomidae (Brachyhypopomus pinnicaudatus) feathertail knifeﬁsh, Sternopygidae (Eigenmannia virescens) glass knifeﬁsh, and Apteronotidae (Apteronotus leptorhynchus) brown ghost.

gymnotiform electric ﬁsh of the New World, but I draw some parallels to their mormyroid counterparts. The most recent phylogeny of the Gymnotiformes (Albert and Campos-da-Paz, 1998) divides these electric ﬁsh into two clades—those with continuous EODs (“wave ﬁsh”) and those with discontinuous EODs (“pulse ﬁsh”) (Fig. 2). Because electromotor and electrosensory systems are so tightly connected through the two functions of electric sense, the evolutionary split of Gymnotiformes into two signal strategies has had profound consequences for every aspect of electroreception and electric communication. The divergence of pulse and wave ﬁsh lets us explore how gymnotiform electric ﬁsh were forced to evolve two sets of solutions to the same problems. In the rest of this chapter, I look at how these problems have been addressed differently by pulse and wave signalers:

204

PHILIP K. STODDARD

1. Avoiding predation—Electric signals should attract electroreceptive predators. Most of these predators are endowed with ampullary electroreceptors sensitive to the lower frequencies present in the ancestral EOD spectrum. Pulse and wave ﬁsh have evolved different tricks to shift their electric spectra upward. 2. Risky sex—The power spectra of ancestral gymnotiform signals included more low-frequency energy (0–100 Hz) than signals of most modern gymnotiforms. When predators drove gymnotiforms to evolve signals with a higher frequency spectrum, female gymnotiforms may not have lost their preference for low-frequency courtship signals. Males in modern pulse and wave groups have different mechanisms for inserting low-frequency energy back into their courtship signals and for keeping this energy transitory. 3. Limits imposed by signal strategy—The bifurcation of gymnotiforms into wave and pulse clades has strong implications for the versatility of these animals. Models indicate that the higher discharge rates of wave ﬁsh might impose severe constraints on other aspects of behavior and life history. I suggest that the wave signal strategy may constrain options for communication, interfere with energy conservation strategies, and lock these ﬁsh out of productive habitat with variable dissolved oxygen or daily temperature cycles.

II. A GYMNOTIFORM PRIMER For those not familiar with the gymnotiform electric ﬁsh, I include the background information needed to follow the main inquiries. More complete coverage of these issues can be found in excellent reviews (Bullock and Heiligenberg, 1986; Hopkins, 1988; Heiligenberg, 1991; Kramer, 1995; Moller, 1995; Hopkins, 1999b). Gymnotiforms are generally referred to by scientiﬁc names, most often by the genus name, though a few have common names from the aquarium trade. Gymnotiforms currently constitute six families with about 150 species, though both numbers could increase with further taxonomic work. The interesting comparisons made in this chapter hinge on distinctions between the clades. I list these in Table I along with the type of EOD (pulse or wave) and the approximate range of discharge rates. The remainder of the background section is divided between the electromotor system that produces electric signals and the electrosensory system that detects electric signals. Electric signals and electric sense serve two key functions: electrolocation and communication. Electroreception can be passive or active. In passive electroreception, ﬁsh use ampullary electroreceptors in their skin to sense

205

ELECTRIC SIGNALS

TABLE I NAMES OF THE BETTER KNOWN GYMNOTIFORM FISHES Familya

Well-known genus

Common nameb

Electrophoridae Gymnotidae Hypopomidae Rhamphichthyidae Sternopygidae

Electrophorus Gymnotus Brachyhypopomus Steatogenys Sternopygus

Apteronotidae

Eigenmannia Apteronotus Sternarchorhynchus

Electric eel Banded knife Pintail knife Leaf knife Black knife, Golden-line knife Glass knife Ghost Elephant nose knife

EODc Pulse Pulse Pulse Pulse Wave Wave Fast wave

Rates 0–8 15–85 4–115 30–120 60–230 200–630 500–1800

a

Family taxonomy after Albert and Campos-da-Paz (1998). Common names derive from the aquarium trade. c EOD discharge rates given are typical. b

weak electric ﬁelds of prey or the electric signals of other ﬁsh. With active electroreception, ﬁsh emit a weak electrostatic ﬁeld then sense the distortion of that ﬁeld caused by nearby objects that differ in complex impedance from the surrounding water (Fig. 3). Active electric sense enables these ﬁsh to navigate dense tangles of vegetation under nocturnal conditions where vision is useless. Gymnotiforms can detect objects through active electrolocation at distances of about half a body length. The range of passive electrolocation is harder to characterize because it depends not only on the spectrum and electric ﬁeld intensity of the external signal but also on how that signal sums with the ﬁsh’s own EOD. Gymnotiforms inhabit freshwater ponds, streams, and rivers of the Neotropical lowlands where some member of this order has evolved to eat just about any sort of animal material available. Small invertebrates constitute the majority of diets, but species have been found that subsist on ﬁsh, plankton, freshwater sponges, and even tails of other gymnotiforms (Lundberg et al., 1987; Marrero and Winemiller, 1993; Crampton, 1996; Lundberg et al., 1996). A. ELECTROMOTOR SYSTEM The EOD makes an electrostatic ﬁeld around the ﬁsh, much in the way a dry-cell battery would if dipped brieﬂy into water. Current ﬂows out of the ﬁsh, through the water, and back into the ﬁsh. The EOD produces no signiﬁcant electromagnetic emission and does not propagate. At a distance, the electric ﬁeld resembles that emanating from an electric dipole, attenuating at the reciprocal of the cube of the distance—this in contrast to sound

206

PHILIP K. STODDARD

FIG. 3. Weakly electric ﬁsh use active electrolocation to sense nearby objects. The electrostatic ﬁeld is distorted by objects of different impedance than the surrounding water. Conductive objects (c) concentrate the electric ﬁeld, creating a spot on the skin of greater electric ﬁeld intensity. Resistive objects (r) divert the electric ﬁeld, creating on the skin a shadow of lesser electric ﬁeld intensity surrounded by zones of greater intensity.

intensity which attenuates with the square of the distance (Knudsen, 1975; Brenowitz, 1986). Electric ﬁeld strengths within a body length are signiﬁcantly stronger than predicted by the inverse cube law. Close to the body, electric ﬁelds of most gymnotiforms are far from an ideal dipole, displaying remarkable spatial and temporal heterogeneity (Rasnow and Bower, 1996; Assad et al., 1998; Stoddard et al., 1999). EODs are the composite action potentials of electrocytes, excitable cells comprising bilateral electric organs (Fig. 4). Individual EOD pulses last 0.5–5.0 ms depending on the species. Electrocytes are arrayed in series within an insulated tube, much like batteries in a ﬂashlight. The serial arrangement of electrocytes increases the EOD source voltage. The serial arrays of electrocytes are often arranged in parallel to magnify the source current. Tropical soils are noted for their low ionic concentration and, as a consequence, tropical freshwaters have low conductivity. Low water conductivity enables a small electric ﬁsh to generate a wide enough electric ﬁeld to serve some useful purpose. The small electric organs of gymnotiforms would short-circuit in salt water.

ELECTRIC SIGNALS

207

FIG. 4. Electric ﬁsh generate electricity with a bilateral pair of electric organs. In ﬁsh of the genus Brachyhypopomus shown here, electrogenic cells (electrocytes) are arrayed in series within insulating tubes, an arrangement that increases voltage. Tubes of electrocytes are arranged in parallel, a mechanism to increase current. Spinal motoneurons trigger action potentials in the posterior face of the electrocyte, creating a net-forward ﬂux of positively charged sodium. Summation of these currents throughout the electric organ make the initial, headpositive phase of the EOD. The action potential in the posterior face triggers a second action potential in the anterior face, resulting in a tailward ﬂux of sodium and the second, head-negative phase of the EOD.

208

PHILIP K. STODDARD

FIG. 5. Electromotor and electrosensory systems of a gymnotiform electric ﬁsh (after Heiligenberg, 1991). Shaded structures are part of the electromotor system, unshaded structures are part of the electrosensory system, nucleus electrosensorius (nE) is both. Abbreviations: ALLG, anterior lateral line ganglion; EG, eminentia granularis; ELL, electrosensory lateral-line lobe; nP, nucleus praeminentialis; OT, optic tectum; Pn, pacemaker nucleus; PPn, prepacemaker nucleus; TE, telencephalon; TS, torus semicircularis.

The command signal for an EOD originates in the pacemaker nucleus of the medulla (Fig. 5). Pacemaker neurons discharge spontaneously, setting a regular rhythm for the EOD. The baseline rate of pacemaker neurons is set by the inward leak of sodium and calcium ions (Smith and Zakon, 2000). Action potentials in the pacemaker cells trigger action potentials in the adjoining relay cells. Relay cells in turn project long axons down the spinal cord where they make electrotonic synapses onto a subpopulation of spinal motoneurons called electromotoneurons. Electromotoneurons form nicotinic (cholinergic) synapses onto electrocytes, giant excitable cells derived embryonically from muscle cells. In development, future electrocytes lose the contractile apparatus of muscle and take on a nervelike function, generating enormous action potentials through the rapid inﬂux of sodium ions (Ferrari and Zakon, 1993; Unguez and Zakon, 1998; Zakon and Unguez, 1999). Wave ﬁsh generate EODs with onset intervals lasting one to two times the duration of the EOD waveform (Fig. 2). The even duty cycle results in a near sinusoidal discharge. Played through an audiospeaker, a wave discharge sounds tonal. Fish with longer intervals between EODs are called “pulse ﬁsh.” Played through a speaker, EOD trains of pulse ﬁsh sound like a nasal buzz. In general, discharge rates of wave ﬁsh are faster than those of pulse ﬁsh, although wave ﬁsh such as Sternopygus may discharge slower than some pulse ﬁsh such as Microsternarchus. Gymnotiforms of the family Apteronotidae produce high-frequency wave EODs at discharge rates ranging from 600 to 1800 Hz (Hopkins and Heiligenberg, 1978; Crampton, 1998b). In Apteronotids, the electromotoneurons themselves generate the EOD (Bennett, 1961). Myogenic electrocytes appear during

ELECTRIC SIGNALS

209

development, function brieﬂy, then regress (Kirschbaum, 1995). Apteronotid electromotoneurons oscillate freely if severed from the pacemaker (Schaefer and Zakon, 1996). Both the oscillatory properties and the loss of a chemical synapse enable the apteronotids to achieve higher discharge rates than are possible with myogenic electric organs. The neurogenic organ does impose other constraints on signal performance that will be discussed later. Wave EODs have a narrow power spectrum centered about the discharge frequency, with energy peaks appearing at the various harmonic intervals (Bullock et al., 1975; Hopkins and Heiligenberg, 1978; Crampton, 1998b). These harmonics are not caused by resonance in the electric organ, but by the imperfect ﬁt of the EOD to a pure sine wave. EODs of pulse ﬁsh have much broader power spectra than those of wave ﬁsh, and their discharge rate has much less relation to the peak power frequency (Heiligenberg, 1977; Hopkins and Heiligenberg, 1978). Unlike in wave EODs, where peak power frequency is a critical metric, peak power frequency of pulse EODs tells relatively little about the shape of their EOD or energy distribution in their power spectrum. Information for communication can be encoded either within the EOD waveform or in modulations of the discharge rate. When a gymnotiform discharges at a constant rate, the EOD waveform still conveys information about the signaler. EOD waveforms can identify the signaler’s family, genus, species, and in some cases age, sex, social status, or individual identity (Hopkins, 1988). Electric ﬁsh have become specialists in the control and modulation of the action potential waveforms that make up their EODs (Zakon, 1996; Zakon et al., 1999). Individual electric ﬁsh can vary EOD waveforms over three distinct time courses: (1) organizational changes over the lifetime of the individual, (2) activational changes over the course of the day, and (3) instantaneous changes from one EOD to the next (∼10 ms). I illustrate these with examples from the hypopomid genus, Brachyhypopomus, a widely distributed pulse ﬁsh with a biphasic EOD. During development, the EOD begins as a monophasic pulse, then becomes biphasic as more myocytes convert to electrocytes (Franchina, 1997). Once the development of the electric organ is complete, the second phase of the EOD continues to change (Fig. 6). In juveniles, the second phase is shorter than the ﬁrst phase and greater in magnitude. In adult females the second phase is about the same shape as the ﬁrst phase. In adult males the second phase is longer than the ﬁrst phase and lower in magnitude. Juveniles and females can be induced to produce masculine EODs by injecting or implanting them with androgens (Hagedorn and Carr, 1985; Silva et al., 1999, unpublished data). Two sexually dimorphic characters, EOD duration and amplitude, are surprisingly plastic. Males change their EOD between day and night, enhancing

210

PHILIP K. STODDARD

FIG. 6. Ontogeny of EODs of Brachyhypopomus pinnicaudatus. Amplitudes are scaled to equate the ﬁrst phase. The second phase of the EOD develops over the ﬁrst 3 months, cutting into the ﬁrst phase and shortening it. The second phase continues to change, showing strong sexual dimorphism at maturity.

the sexual dimorphism during the early evening when spawning occurs and decreasing sexual dimorphism at other times (Hagedorn, 1995; Franchina and Stoddard, 1998). Social isolation decreases the degree of sexual dimorphism over the course of several days and social stimulation restores males’ EOD traits, the time course of recovery depending on the stimulation and the EOD trait (Franchina et al., 2001). Capture stress can decrease sexual dimorphism in minutes and release from conﬁnement restores it equally fast (Franchina and Stoddard, 1998). The neural and hormonal bases of these rapid changes are currently under investigation. Brachyhypopomus produces instantaneous alterations in the EOD waveform in the context of reproduction. During courtship, these ﬁsh produce a series of stereotypic rate modulations, the more dramatic of which alter

ELECTRIC SIGNALS

211

FIG. 7. Acceleration and chirp, two of the complex courtship signals of male Brachyhypopomus pinnicaudatus. The acceleration signal is given early in courtship, most often while the male runs his nose into the female’s midsection. The female’s discharge is also visible in the upper trace as the smaller EODs amid the larger ones of the male. Seen in expanded view, the male’s EODs are unchanged from normal. The slight diminution in amplitude is due to the male’s movement with respect to the recording electrodes. The chirp signal is given later in courtship, immediately before and during spawning. EOD amplitudes are reduced during the chirps and the EOD waveform loses much of its second phase.

brieﬂy the EOD itself by causing EODs to fail partially, reducing brieﬂy the EOD amplitude (Fig. 7). Pulse ﬁsh appear to engage in more subtle, dyadic interactions, involving interactions between the EODs of the two participants. Members of courting pairs of Brachyhypopomus track each other’s discharge rates in what appear to be rate duets (Fig. 8). During such interchanges, discharge rates can be different but highly correlated. If either member of the pair changes its rate, then the other member changes in the same direction. This rate correlation could be a form of courtship communication, or it could be an artifact of jamming avoidance in which the leader changes its discharge rate for some reason and the follower shifts its rate to avoid electric interference. Detailed analyses of rate and phase to evaluate these possibilities are in progress.

212

PHILIP K. STODDARD

FIG. 8. EOD rate tracking by a courting pair of Brachyhypopomus pinnicaudatus, interspersed with male chirps (indicated by ∗). Either member of the pair may lead or follow. The signiﬁcance of this behavior is not yet known; it might be a courtship duet or it might be involved in jamming avoidance.

In the ﬁrst study of gymnotiform ecology, Lissmann (1958) proposed that wave ﬁsh were adapted to faster waters than pulse ﬁsh, their higher discharge rates serving the faster pace of life in a moving medium. Presumably, the higher the EOD rate, the shorter the potential latency to detect novel features in the aquatic environment. The correlation between water ﬂow and signal strategy is good but not absolute; both wave ﬁsh and pulse ﬁsh live in slow streams, and a few wave ﬁsh are found in still waters (Hopkins and Heiligenberg, 1978; Crampton, 1998a,b). Key environmental differences between fast and slow waters include not only ﬂow rate itself but also oxygen availability and predator distribution (Marrero and Taphorn, 1991; Crampton, 1998a), all of which appear to affect gymnotiform distribution and signal strategies. B. ELECTRIC SENSE Animals that sense electricity have specialized electroreceptive cells in the skin called electroreceptors, similar to mechanosensory hair cells in derivation. Electroreceptors are divided into two broad classes, ampullary and tuberous (Fig. 9). Ampullary electroreceptors are tuned to low-frequency electric ﬁelds (∼0–100 Hz) and can sense electric ﬁeld intensities on the order of a microvolt per centimeter, such as the weak electric ﬁelds generated by the muscle action of nearby animals (e.g., wiggling worms, locomoting daphnia, ventilating ﬁsh). Ampullary electroreceptors are useful for detecting prey under conditions of poor visibility and, potentially, for listening to or eavesdropping on EODs of other individuals.

ELECTRIC SIGNALS

213

FIG. 9. Ampullary and tuberous electroreceptors from gymnotiforms (drawn from Szamier and Wachtel, 1970; Szabo, 1974) and their respective tuning curves. Gymnotiform ampullary electroreceptors are very sensitive to low-frequency energy. Tuberous electroreceptors are less sensitive but are tuned to the spectrum of the EOD. Both sets of tuning curves shown here are from pulse ﬁsh. The ampullary electroreceptor threshold was measured in Gymnotus cf. carapo (adapted from Dunning, 1973). Three representative threshold curves of tuberous receptors were recorded in the pulse ﬁsh Brachyhypopomus occidentalis (adapted from Shumway and Zelick, 1988). Tuning curves take on three shapes: narrow bandpass, broad bandpass, and lowpass. Tuberous receptors of wave ﬁsh resemble the more narrow of these.

Ampullary electroreceptors are found in the Gymnotiformes (New World electric knifeﬁsh) and in their sister order the Siluriformes (catﬁsh) (Fink and Fink, 1981). Catﬁsh use olfactory whiskers to locate prey, then they guide the ﬁnal strike with passive electrolocation (Atema, 1969; Roth, 1972). An acute electric sense gives large piscivorous catﬁsh the potential ability to detect the EODs of tasty little electric ﬁsh at signiﬁcant distances. Ampullary electroreceptors are found in all primitive ﬁsh, including those that gave

214

PHILIP K. STODDARD

rise to the tetrapods (Northcutt, 1986). Ampullary electroreceptors were lost in the early teleost ﬁsh then reevolved in two teleost lineages (Fig. 1) (Finger et al., 1986). Until someone sequences the genes for electroreceptors, we can only guess whether their appearance in electroreceptive teleosts represents complete evolutionary novelty or some reexpression of an ancestral developmental program. Tuberous electroreceptors are derived specialties of the teleost electric ﬁsh. In gymnotiforms, tuberous electroreceptors code either the amplitude or the time of the EOD. Most of an individual’s tuberous electroreceptors are tuned to its own EOD rather than to that of ﬁsh it may need to listen to (Hopkins, 1976) such as members the opposite sex which may have a different EOD spectrum (Hopkins, 1974a,b; Meyer, 1983). The amplitude coding receptors have a narrow dynamic range (∼10–12 dB), approximately centered around the individual’s own EOD. If tuberous electroreceptors of the gymnotiforms were adapted for electric communication, we would expect broader tuning and more extreme sensitivity to EODs than they display. African mormyrid electric ﬁsh have evolved a unique type of electroreceptor, the Knollenorgan, with precisely those characteristics. Knollenorgans are broadly tuned to the entire EOD spectrum, they are extremely sensitive, and their afferent signals to the brain are gated off during the listener’s own EOD (Bell and Grant, 1989). Based on the evidence presented earlier, it is probably safe to assume that tuberous electroreceptors in gymnotiforms evolved for active electrolocation rather than communication. I am less certain that the gymnotiform EOD evolved for active electrolocation, as has been assumed (Lissmann, 1958). The source of my hesitation is an unrelated group of electric ﬁsh, the saltwater skates. Skates possess ampullary electroreceptors (the ampullae of Lorenzini) and an electric organ in their tail, a location more conducive to communication than electrolocation (Bass, 1986). In the breeding season, androgens shift the spectral sensitivity of the electroreceptors in male stingrays to better match the ventilation signal, a shift which presumably helps them locate females at the expense of prey detection (Tricas et al., 1995; Sisneros and Tricas, 2000). Based on the skate and stingray, one can envision an alternate scenario in which early gymnotiforms evolved low-frequency electric signals to stimulate the ampullary electroreceptors of prospective mates. These early gymnotiforms might have shifted their EOD spectrum to higher frequencies to facilitate active electroreception with their newly evolving tuberous electroreceptors. Sensory information from the electroreceptors is gathered by afferent ganglion neurons of the anterior lateral line nerve (ALLn). The ALLn projects to an enormously hypertrophied region of the brainstem called the electrosensory lateral-line lobe (ELL) (Maler et al., 1981; Carr and Maler, 1986). Information from the different electroreceptor classes (ampullary, tuberous

ELECTRIC SIGNALS

215

time coders, tuberous amplitude coders) remains segregated throughout the hindbrain and much of the midbrain, undergoing extensive parallel processing (Carr et al., 1981; Shumway, 1989a,b) before converging in the upper layers of the torus semicircularis (Metzner and Heiligenberg, 1991). Unlike the electroreceptive periphery, the electrosensory regions of the gymnotiform brain shows signiﬁcant specialization for communication. The key feature of the gymnotiform brain, however, is its specialization for detecting distortion of the ﬁsh’s own EOD. In particular, amplitude distortion is detected in the ELL, and phase distortion is detected in the torus. Neural networks of the ELL and associated structures ﬁlter the input stream to delete predictable information, thus revealing and enhancing novelty in the ﬁsh’s electric world (Bell et al., 1997). The ELL is a cortical structure bearing four somatotopic maps of the electroreceptive periphery (Carr and Maler, 1986). The medial map segment receives afferent input from the ampullary receptors, and the outer-three map segments process information from the amplitude-coding tuberous receptors in parallel (Heiligenberg and Dye, 1982). Each tuberous map segment performs a unique optimization of the trade-off in spatiotemporal acuity. Pyramidal cells of the innermost map segment have the smallest receptive ﬁelds and display the best spatial resolution, whereas pyramidal cells of the outermost segment have the largest receptive ﬁelds and the fastest temporal resolution (Shumway, 1989a,b). Different map segments appear specialized for different electrosensory tasks as well. Fish with lesions of the innermost map segments cannot perform the jamming avoidance response necessary for electrolocation whereas those with lesions of the outermost map segments can shift their discharge rate to avoid jamming but cease to emit aggressive chirps in response to conspeciﬁc signals (Metzner and Juranek, 1997; Juranek and Metzner, 1998). Temporal information converges slightly in spherical neurons of ELL and then is relayed directly to the torus semicircularis of the midbrain (Maler et al., 1981). Layer 6 of the torus transforms spike time over the body surface into phase differences between the different receptive ﬁelds (Bastian and Heiligenberg, 1980; Carr et al., 1986; Rose and Heiligenberg, 1986), encoding EOD phase distortion with submicrosecond acuity (Kawasaki et al., 1988).

III. THE PERILS OF PREDATORS A catﬁsh’s face is bespeckled with sensitive ampullary electroreceptors, ideal for locating a small prey species that gives off a robust electric signal (Roth, 1972). Large piscivorous catﬁsh course the river channels of tropical South America and Africa, consuming electric ﬁsh when they are encountered (Marrero and Taphorn, 1991; Winemiller and Kelso-Winemiller, 1996). Many catﬁsh of the New World family Pimelodidae hunt gymnotiforms

216

PHILIP K. STODDARD

opportunistically whereas at least one species, Pseudoplatystoma tigrinum, specializes on gymnotiforms (Reid, 1983). In Africa, where mormyrid electric ﬁsh occupy the electric ﬁsh niche, clariid catﬁsh hunt them in packs (Merron, 1993), attracted to low-frequency spectral components in the EOD (Hanika and Kramer, 1999, 2000). Having evolved electroreceptors before their split with the Gymnotiformes, catﬁsh have probably been a strong selective agent ever since gymnotiforms emitted their ﬁrst EOD. Not surprisingly, most gymnotiforms exhibit electric adaptations that render their signals less attractive to piscivorous catﬁsh. The electric eel has an extra, high-voltage electric organ that can discharge hundreds of volts into any predator that fails to sever its spinal cord on the ﬁrst bite (Faraday, 1839; Hagiwara et al., 1965). Even baby electric eels can deliver a painful shock (personal observation). The electric eel is itself a serious predator of gymnotiforms (Westby, 1988, M. Hagedorn, personal communication). The rest of the gymnotiforms have no high-voltage electric organ, but instead appear to rely on a suite of characters that shift their EOD spectrum upward and out of the spectral sensitivity band of the catﬁsh’s electroreceptors (∼0–30 Hz—Peters and Buwalda, 1972). Two aspects of the EOD work to keep the spectrum high: (1) discharge rate set by the pacemaker nucleus and (2) dc balance of the remote EOD waveform. A good treatment of spectral structure is given by Bradbury and Vehrencamp (1998). If the discharge rate falls within the spectral range of the ampullary electroreceptors, some amount of EOD energy will fall there also. Pulse gymnotiforms generally discharge at low rates (8–20 Hz) during the day but increase their discharge rate by a factor of 2–5 when the risks are higher (i.e., at night or when they are disturbed during the day). Wave ﬁsh maintain high and relatively constant discharge rates all the time. The fundamental frequencies of their EOD spectra lie far above the sensory range of the ampullary electroreceptors. Apteronotid wave ﬁsh, in particular, dominate the ﬁsh fauna of river channels where water ﬂows quickly and large catﬁsh are plentiful (Lundberg et al., 1987). If the net current under the EOD curve (between the curve and 0 V) is not zero, then the EOD has a net dc component with a spectral frequency of 0 Hz. Because the ampullary electroreceptors are sensitive to very low frequencies, residual dc components in the EOD spectrum may be detected by electroreceptive predators. Gymnotiforms have evolved two mechanisms to reduce or eliminate the dc spectrum of the EOD. The pulse ﬁsh clade and the wave ﬁsh family Apteronotidae independently evolved multiphasic waveforms that matched positive and negative energy to produce a net dc energy close to zero (Fig. 2). The wave ﬁsh

ELECTRIC SIGNALS

217

clade Sternopygidae produces head-positive monophasic EODs but offsets them with a head-negative dc (Bennett and Grundfest, 1961) that cancels the net dc in the EOD. Bennett (1971) speculated that the head-negative dc of the sternopygids evolved to maintain ionic balance in the electrocytes. I favor predation avoidance as the basis for reduction of dc energy in the EOD because it explains dc balance for all taxa whereas the ionic homeostasis hypothesis applies only to the Sternopygidae. For instance, in those Brachyhypopomus species with biphasic EODs, currents close to the body are extremely asymmetric with respect to 0 V dc and also have a large low-frequency spectral component (Fig. 10). Only at distances greater than a body length from the ﬁsh does the waveform assume dc symmetry and the dc component approach zero (Fig. 11) (Stoddard, 1994). Given that the asymmetry in the local EOD is reﬂected in the ionic conductances of individual electrocytes (Bennett, 1961), ionic ﬂux would not be maintained by the EOD itself. In contrast, laboratory trials have shown that dc-balanced waveforms are signiﬁcantly less detectable by electric eels and clariid catﬁsh than dc-imbalanced waveforms (Hanika and Kramer, 1999, 2000; Stoddard, 1999). At least three gymnotiforms not only produce monophasic EODs (Fig. 12), but apparently have evolved from ancestors with multiphasic EODs (AlvesGomes et al., 1995; Sullivan, 1997; Albert and Campos-da-Paz, 1998). The reversion of the EOD to a form with a large dc component would seem to challenge any adaptive hypothesis for dc symmetry. So far, each species I have examined seems to have some scheme for reducing risk predation by electroreceptive predators. The electric eel (Electrophorus electricus) produces a high-voltage discharge. Gymnotus cylindricus lives in streams lacking large electroreceptive predators. Brachyhypopomus sp. (“monophasus”) appears to be an electric Batesian mimic of young electric eels with which it is sympatric, though we still need behavioral conﬁrmation that this species deceives predators as readily as it fools ﬁeld biologists (Stoddard, 1999). Both ﬁeld and lab data suggest that the modern gymnotiform weakly electric ﬁsh has not yet won the evolutionary arms race with electroreceptive predators. Catﬁsh appear to present an ongoing source of predation. Pimelodid catﬁsh consume gymnotiforms in the large river channels, sometimes in large numbers (Reid, 1983; D. Taphorn, personal communication). In fact, live Gymnotus are the preferred catﬁsh bait of commercial ﬁshermen in the Orinoco. Whether they use electroreception or olfaction is not known. Among electroreceptive predators, the electric eel (Electrophorus electricus) may present a special problem for gymnotiforms. Electrophorus has been observed hunting the pulse ﬁsh Gymnotus carapo in the ﬁeld and is readily attracted to them in the lab (Westby, 1988). When Electrophorus get trapped in oxbow lakes with smaller gymnotiforms during the annual

218

PHILIP K. STODDARD

FIG. 10. EODs local to the body of female Brachyhypopomus pinnicaudatus express significant energy in the range of the ampullary electroreceptors due to dc imbalance in the local EODs. At more remote locations, the local electric ﬁelds merge to form a symmetric, biphasic EOD with much less energy in the low-frequency spectrum. In the example shown here, the remote EOD has 27 dB less energy at 10 Hz than at 1000 Hz. EOD spectra at the head and tail are nearly ﬂat through the spectral range of both electroreceptor types. Cancellation of low-frequency energy makes the EODs less detectable by electroreceptive predators and has been proposed as the initial selective force for the evolution of multiphasic EODs (Stoddard, 1999).

ELECTRIC SIGNALS

219

FIG. 11. Electric ﬁeld measurements in the midplane of a female Brachyhypopomus beebei show how the dc symmetry varies with distance from the body (Stoddard, 1994; Stoddard et al., 1999). For this ﬁgure, I calculated the RMS of the three-dimensional dc component of the signal across a 20 ms interval which approximates the nighttime discharge rate. At distances beyond 5 cm, the dc component of the EOD drops below the detection threshold of ampullary electroreceptors, presumably reducing risk of predation.

dry-down, the smaller gymnotiforms disappear completely (M. Hagedorn, personal communication). The electric eel is a remarkable predator and the most unusual gymnotiform by any measure. Where other gymnotiforms have a single bilateral pair of electric organs, Electrophorus has three morphologically distinct pairs of electric organs (Hunter, Sachs, main). The Sachs organ emits the low-amplitude EODs used for electrolocation and communication, and the Hunter and the main electric organs are used to generate highvoltage discharges for stunning prey and for defense (Coates, 1950; Keynes and Martins-Ferreira, 1953; Hagiwara et al., 1965). Other gymnotiforms emit

220

PHILIP K. STODDARD

FIG. 12. A few gymnotiforms produce monophasic EODs with no dc offset to conceal lowfrequency energy. These are the electric eel (Electrophorus electricus), and two species with low-voltage EODs, Gymnotus cylindricus of Middle America, and Brachyhypopomus nov. sp. monophasus (Sullivan, 1997). Brachyhypopomus nov. sp. monophasus may be a Batesian mimic of the sympatric electric eel. G. cylindricus live in streams devoid of large electroreceptive predators. Phylogenetic analysis shows secondary loss of multiple EOD phases in the latter two species (Sullivan, 1997, unpublished data).

EODs continually, driven by an oscillating pacemaker nucleus, but Electrophorus lies silently for minutes on end, discharging only when it is active. Hopkins (1973) proposed that irregular discharges of mormyrid electric ﬁsh blend into the background of electrical noise from tropical electric storms. Perhaps the irregular discharge of the electric eel facilitates a stealthy approach upon electroreceptive prey. Most pulse gymnotiforms can gulp air when oxygen levels are low, a trait that allows them to inhabit oxygen poor waters (Crampton, 1998a), a possible refuge from large, active predatory catﬁsh. But electric eels are obligate air breathers (Farber and Rahn, 1970; Garey and Rahn, 1970; Johansen, 1970) and they can penetrate oxygen poor waters. All these characters—multiple discharge types, facultative signaling, air breathing—equip Electrophorus as a formidable predator of smaller gymnotiforms. In keeping with this role, Electrophorus begins hunting in the late afternoon while other gymnotiforms are still inactive (Westby, 1988). Based on his ﬁeld observations, Westby (1988) speculated that Electrophorus uses its tuberous electroreceptors to detect smaller gymnotiforms. There is an unresolved problem with this proposal: to detect the dc-balanced EOD of a gymnotiform at any useful distance (e.g., 1 m), the eel would need tuberous electroreceptors signiﬁcantly more sensitive than those of any known

ELECTRIC SIGNALS

221

FIG. 13. EODs of Brachyhypopomus pinnicaudatus used as playback stimuli with an electroreceptive predator, the electric eel (Electrophorus electricus). The half EOD (solid line) has about 30 dB more energy in the spectrum of the ampullary electroreceptors than does the full EOD (dashed line). The eel was signiﬁcantly better at detecting the half EOD than the full EOD (Stoddard, 1999). The eel thus appears to use its ampullary electroreceptors to detect signals of prey. Signals that conceal the low-frequency spectrum will be less readily detected, even if their overall amplitude is greater.

gymnotiform. Such electroreceptors would be equivalent in sensitivity to the Knollenorgans of mormyrids or the type I gymnarchomasts of Gymnarchus niloticus (review by Zakon, 1986). So far, no evidence suggests Electrophorus is endowed with high-sensitivity tuberous receptors. Reported thresholds for the eel’s tuberous electroreceptors range from 2 to 30 mV/cm (Hagiwara et al., 1965). EOD intensities of 2 mV/cm appear between 1 and 5 cm from the tail of adult Brachyhypopomus (Stoddard et al., 1999). So tuberous electroreceptors, by themselves, do not appear sufﬁciently sensitive to enable electric tracking of small gymnotiform prey over signiﬁcant distances. Electric playbacks to a resting electric eel suggest the eel might have a clever solution to this problem. When we played monophasic EODs with signiﬁcant energy in the spectrum of its ampullary electroreceptors (Fig. 13), it swam silently until it reached the electrode. Not surprisingly, when we played biphasic, dc-symmetric EODs expressing little energy within the spectral sensitivity of the ampullary receptors the eel generally did not react (Stoddard, 1999). But a few times the eel began to discharge rapidly as it homed in on the playback electrode. Although the eel’s tuberous electroreceptors may be too insensitive to detect the remote EOD in isolation, its gymnotiform brain is exquisitely sensitive to distortion of its own EOD. By discharging rapidly and continually, the eel can create a carrier signal that biases the weak, remote EOD up into the active range of its tuberous electroreceptors. The eel thus could use its active electrolocation system to track the remote EOD to the source, a variant of the “scan sampling hypothesis”

222

PHILIP K. STODDARD

FIG. 14. (A) Complex social signals of three gymnotiform species. The two recordings from Brachyhypopomus pinnicaudatus include both male and female EODs. The female EODs remain at constant amplitude and rate throughout, but their coincidences with the male EOD constitute the unusually high and low traces. Because of their dc symmetry, the female EODs do not contribute to the low-frequency components in the recording. The chirp of Eigenmannia virescens was resynthesized from an EOD and a dc-coupled chirp recording from Metzner and Heiligenberg (1991). (B) Complex electric signals used in courtship and aggression (A) contain signiﬁcant energy in the spectral range of the ampullary electroreceptors. The

ELECTRIC SIGNALS

223

FIG. 14. (Continued) low-frequency components shown here are revealed by passing the digitized waveform through a digital 2-pole Butterworth low-pass ﬁlter with a corner of 8 Hz. Of the complex signals shown here, Eigenmannia chirps generate the greatest proportion of low-frequency energy, but even Apteronotus leptorhynchus produces enough lowfrequency energy that it should stimulate the ampullary electroreceptors of conspeciﬁcs at close range.

224

PHILIP K. STODDARD

(Hopkins and Westby, 1986). The eel’s electrolocation EOD is at least two orders of magnitude stronger than that of a small gymnotiform, so electrical silence could allow the eel to get close to its prey without revealing itself. Our data here are suggestive, but anecdotal. This hypothetical mechanism requires systematic investigation. When they are not contending with predators that would swallow them whole, gymnotiforms have to contend with electroreceptive “partial-predators” vying to eat them one bite at a time. Deep channels of the Amazon harbor two apteronotids, Magosternarchus raptor and M. duccis, that, based on their stomach contents, seem to exist solely on the tails of other gymnotiforms (Lundberg et al., 1996). The blind Amazonian catﬁsh Pseudocetopsis sp. has a tuberous electroreceptor with morphological similarity to the mormyrid Knollenorgan (Andres et al., 1988). The physiology of this electroreceptor is unknown, but Pseudocetopsis lacks an electric organ of its own and may use its novel receptor to track gymnotiforms. Andres et al. suggested that this small catﬁsh, like Magosternarchus, may survive by cropping the tails of gymnotiforms. Behavioral data on the foraging habits of these two groups would be most interesting. IV. ’TWAS EVER THUS: SEX VERSUS PREDATORS Modern gymnotiforms have evolved mechanisms that avoid the production of low-frequency energy that might otherwise attract predators endowed with ampullary electroreceptors. In the sort of turnabout we have come to expect in natural systems, male gymnotiforms have evolved secondary mechanisms to put the low-frequency energy back into their courtship signals. As with other aspects of their signal behavior, wave and pulse species have different mechanisms for resynthesizing low-frequency energy. Courtship signals of male Eigenmannia consist of gaps in the continuous EOD wave train (Fig. 14). These gap signals are called “chirps” for their sound on an audio monitor (Hopkins, 1974b). As mentioned earlier, Eigenmannia produce monophasic EODs against a head-negative baseline. Periods without EODs are not electrically silent in Eigenmannia, but instead reveal continuous, head-negative dc electric ﬁelds. Thus when Eigenmannia chirps the net dc ﬁeld goes from zero during EOD pulses, to negative during the period of silence, and back to zero when the EOD resumes. Eigenmannia readily discriminates between synthetic signals with and without the low-frequency spectral component (Naruse and Kawasaki, 1998). Neurophysiological and behavioral evidence is consistent with the assumption that low-frequency energy should be attractive to female Eigenmannia. The pyramidal cells of the ELL are sensitive to amplitude modulation. Tuberous map segments respond to the high-frequency AM modulation and

ELECTRIC SIGNALS

225

the ampullary map segment responds to modulation of the dc component (Metzner and Heiligenberg, 1991; Naruse and Kawasaki, 1998). The different electroreceptor streams, segregated in ELL, converge in the torus semicircularis (Metzner and Heiligenberg, 1991). Some neurons found in the torus of Eigenmannia are excited only by amplitude-modulated signals that simultaneously excite both the tuberous and the ampullary electroreceptor pathways (Heiligenberg et al., 1991). Neurons in this region of the torus send projections to the preoptic area (POA), a conserved hypothalamic region that controls sexual behavior throughout the vertebrates. In keeping with the neurophysiological data, captive female Eigenmannia spawned when played chirps with the dc component intact but did not spawn when the playback was ac coupled (Hagedorn and Heiligenberg, 1985). The idea that ampullary information is a critical component of Eigenmannia’s courtship signals has led me to look for low-frequency content in the sexual signals of other gymnotiforms. Males of the pintail knifeﬁsh Brachyhypopomus use their sexually dimorphic EOD to attract females to defended ovoposition sites then court them with a pair of complex rate and amplitude modulated signals. As with the wave ﬁsh Eigenmannia, the pulse ﬁsh Brachyhypopomus synthesizes low-frequency energy in its complex signals. But Brachyhypopomus also adds low-frequency energy to the simple EOD waveform as well. Male Brachyhypopomus enhance the sexually dimorphic characters in their EODs on a daily cycle, reaching maximum values about 1 h after sunset when nightly spawning activity begins, and reducing these characters 3 h later when spawning typically ends (Franchina and Stoddard, 1998). The EOD spectrum changes signiﬁcantly between day and night, most noticeably in the spectral range of the ampullary electroreceptors (Fig. 15).

FIG. 15. Spectra of the EODs of a reproductive male Brachyhypopomus pinnicaudatus during the day and at night. Extension of the second phase of the EOD waveform at night boosts energy in the spectral range of the ampullary electroreceptors by 15 dB (560%), presumably to attract females.

226

PHILIP K. STODDARD

EOD energy in the spectral range of the tuberous electroreceptors increases by about 3 dB at night, a change of 40%. More striking, energy in the range of the ampullary electroreceptors increases by 15 dB, an increase of 560%. Early in courtship male Brachyhypopomus produce stereotypic rapid acceleration signals. Because a reproductive male’s EOD is net dc negative at night, increasing the EOD discharge rate increases the proportion of time the total wave train is dc negative (Fig. 14). An acceleration signal thus synthesizes a head-negative energy blip in the spectral range of the ampullary electroreceptors. As spawning begins, the male switches from acceleration signals to “chirps.” Brachyhypopomus chirps are a high-frequency series of small EODs in which the second EOD phase is missing or much reduced relative to the ﬁrst phase. The result is a positive blip in the net dc component of the signal. Chirps given in rapid succession create a continuous low-frequency wave in the spectral range of the ampullary electroreceptor system (Fig. 14). EODs of the wave ﬁsh Apteronotus leptorhynchus are dc balanced, but their chirps are not. The dc component of the A. leptorhynchus chirp shown in Fig. 14 is 0.7% of the peak-to-peak EOD amplitude. The behavioral threshold for low-frequency signals in A. albifrons is 0.6–0.9 µV/cm (Knudsen, 1974). Based on electric ﬁeld strength measured for a 21 cm A. leptorhynchus (Rasnow and Bower, 1996), the dc component in a chirp would be 1 µV/cm at a distance of 8 cm from the signaler. This distance is well within the range of courtship in this species (Hagedorn and Heiligenberg, 1985). EODs of some other apteronotids with higher frequency EODs are not dc symmetric but rather are negatively biased (Fig. 2). I have not measured the dc components of their chirps but they should be greater than that seen in Apteronotus. Dual-band stimulation of gymnotiform chirp signals are reminiscent of the “whine-chuck” courtship vocalizations of the Tungara ´ frog Physalaemus pustulosus (Ryan, 1985). Whines focus energy in the spectral band of the amphibian papilla and “chucks” focus energy in the lower spectral band of the basilar papilla (Ryan et al., 1990). Chucks also attract the frog-eating bat Trachops cirosus (Ryan et al., 1982, 1983). Male Tungara ´ frogs calling in isolation produce only whine calls, but switch to whine-chucks when competing for mates with nearby males. Like Tungara ´ frogs, the low-frequency spectrum of Eigenmannia is entirely transitory, appearing only during chirp signals. In contrast, male Brachyhypopomus produce signiﬁcant low-frequency energy all the time, more energy during the spawning hours, and the most energy during the transitory courtship signals. Indirect evidence from two ﬁeld studies in Venezuela suggests something in their behavior, perhaps the low frequencies in the EOD, increases their risk of predation. X rays of B. pinnicaudatus found that half of the sexually mature male had regenerated tails in contrast to nearly none of the females (Hopkins et al., 1990). First, during the breeding season, mature male B. diazi disappear from streams while

ELECTRIC SIGNALS

227

mature females survive in good numbers (P. Stoddard and E. Setteducati, unpublished data). Second, perhaps as a response to predation risk, perhaps for energy savings, male Brachyhypopomus reduce the energy in their EODs signals when socially isolated, particularly the low-frequency energy. They restore the low-frequency spectrum in the company of conspeciﬁcs, fastest in the company of males (Franchina et al., 2001). Nobody has investigated the ampullary electroreceptive system of Brachyhypopomus, so I can only assume for now that it resembles that of Gymnotus and Eigenmannia, other gymnotiforms in which it has been studied. The dual-frequency band of gymnotiform courtship signals might have arisen for any of several reasons: sensory exploitation of a preexisting female sensory bias, phylogenetic inertia, or through female choice for a signal of male quality. I consider these in order. A. SENSORY BIAS Several aspects of the male Brachyhypopomus signal ﬁt a simple sensory bias model. During spawning hours, reproductive male Brachyhypopomus maximize EOD amplitude and duration (Franchina and Stoddard, 1998). Males also discharge at higher rates than females (Fig. 8). By directing energy into ampullary and tuberous spectrum bands, males increase the signal bandwidth and thus its potential conspicuousness. Higher signal amplitudes increase the active space of the signal and thus the female’s chance of detecting that signal (Brenowitz, 1986). Both greater EOD duration and higher discharge rates increase the probability that the female’s EOD will overlap with the male EOD in time. Because the gymnotiform electrosensory system is tuned to detect EOD distortion, temporal EOD overlap increases the sensitivity to other ﬁsh’s EODs by two orders of magnitude (Shumway and Zelick, 1988). Reproductive male Eigenmannia have lower frequencies than females (Hopkins, 1974b; Hagedorn and Heiligenberg, 1985) but wave ﬁsh always overlap each other’s signals, so increasing discharge rate would not enhance detection. B. PHYLOGENETIC INERTIA If early gymnotiforms courted with electric signals before the origin of high-frequency EODs and tuberous electroreceptors, they would have relied on their ampullary electroreceptors to process those signals, as Atlantic stingrays do today (Tricas et al., 1995; Sisneros et al., 1998). Although predation pressure acted against the low-frequency EOD spectrum, the preexisting female attraction to the low-frequency spectrum might have acted to retain low-frequency energy in the EOD even as the basic species EOD shifted upward in frequency.

228

PHILIP K. STODDARD

C. MALE QUALITY Females in general, and female ﬁsh in particular, are often attracted to conspicuous, expensive, and dangerous male traits, honest indicators of survival ability (Zahavi, 1977a,b; Andersson, 1994; Sargent et al., 1998). Attraction to low-frequency signals might have been selected based on their value as indicators of male quality. Low-frequency signals might indicate a “willingness” to take risks. By increasing EOD duration, male Brachyhypopomus increases the total energy expended on EODs. We do not yet know the costliness of EOD production so it is hard to say whether increased energy expenditure would be a reliable indicator of male condition. I can make a case, however, that total EOD amplitude could be used as a reliable indicator of male quality. EOD amplitude in B. pinnicaudatus depends on body size (Hopkins et al., 1990; Franchina and Stoddard, 1998). Sexually receptive females, given a choice, prefer to mate with larger males (Curtis, 1999). Under controlled lab conditions, male body size at sexual maturity depends on the amount of food we provide. Presumably body size in wild males indicates the ability to acquire food or allocate it toward growth. Behavioral traits are often consistent with multiple mechanisms of sexual selection. Controlled mate choice studies are needed to determine whether such traits as EOD amplitude, duration, and higher discharge rate signal attractiveness per se, or whether they merely increase probability of detection by the female. V. ELECTROSENSORY INFLUENCES ON NICHE PARTITIONING The split into pulse and wave discharge types has further facilitated habitat partitioning by the gymnotiforms. The physiology of the electrosensory system is affected by external environment and thus has provided some novel axes for specialization and compromise. Dissolved oxygen, water ﬂow, temperature, and water conductivity appear to be key environmental variables affecting both the physiology and spatial distribution of different gymnotiform taxa. A. OXYGEN AND ENERGY During the day, pulse ﬁsh lie motionless, slowing their electric discharge as much as ﬁve-fold from nighttime rates. In contrast, the elongate anal ﬁns of the wave ﬁsh often remain in motion, and their high EOD rates remain unchanged from nighttime levels. The high activity levels of wave ﬁsh might be ideal for life in moving water, but logic dictates that such constant output must command a price. We have no data on energetic needs of gymnotiforms,

ELECTRIC SIGNALS

229

but oxygen consumption can serve as a useful proxy. Gymnotiform wave ﬁsh require high concentrations of dissolved oxygen to sustain their high output. I saw a most striking demonstration of this point when a colleague adjusted the EOD discharge frequency of an Apteronotus on a physiology rig merely by regulating the water ﬂow over the ﬁsh’s gills. Crampton (1998a) studied the oxygen needs of captive gymnotiforms, recording EOD rates as he lowered oxygen content in the water. Pulse ﬁsh require less oxygen to maintain their EOD rate than wave ﬁsh. Every pulse ﬁsh Crampton studied could survive anoxic conditions for hours; most gulped air from the surface, one shut off its EOD while remaining motionless at the bottom. The only wave ﬁsh capable of surviving anoxia was Eigenmannia virescens which survived hypoxia by respiring at the oxygen-rich air–water interface. Other Eigenmannia species required high oxygen. As a rule, Crampton found that wave ﬁsh avoided Amazonian habitats with protracted or intermittent periods of low oxygen availability. Oxygen proved a better predictor of gymnotiform pulse versus wave ﬁsh distribution than water speed per se (Crampton 1996, 1998a). It seems likely that the high oxygen demand of wave ﬁsh is a direct consequence of higher metabolic rate, possibly related to their high discharge rate. The alternative is that wave ﬁsh require no more energy, but are less efﬁcient at extracting dissolved oxygen from the water. We need to know how much it costs an electric ﬁsh to produce its EOD and whether that portion of the energetic budget is signiﬁcant relative to the cost of other activities such as locomotion. Wave ﬁsh produce lower peak-to-peak EOD amplitudes than do pulse ﬁsh (Rasnow and Bower, 1996; Assad et al., 1998; Stoddard et al., 1999), perhaps as a way to conserve energy. Because their sensory tuning is narrow, a wave ﬁsh can lower the total amplitude of its EOD while concentrating the remaining energy in the narrow sensory band of its tuberous electroreceptors. Constriction of the EOD spectrum thus allows a wave ﬁsh to save energy and make its signals less conspicuous to hostile eavesdroppers without losing sensory acuity (Hopkins, 1976). Because the spectrum of a wave EOD is narrow and low in amplitude, a variety of wave ﬁsh with different frequencies should be able to coexist in the same active space without jamming their active electric sense. Noise immunity and predator avoidance thus could have been signiﬁcant evolutionary advantages to the wave signal strategy. The wave strategy may not be universally advantageous. Maintaining a wave EOD may impose some unique costs on the signaler. Narrow spectral tuning obliges the wave ﬁsh to maintain a constant discharge rate. Were a wave ﬁsh to lower its discharge rate during the day as a pulse ﬁsh does, it would also lower the fundamental frequency of the EOD spectrum out of the sensory range of its tuberous electroreceptors. The ﬁsh would be unable to sense its own EOD, and thus would become electrically blind to

230

PHILIP K. STODDARD

much of its environment. Electric ﬁsh show no facility for rapid alteration of electroreceptor tuning. Electroreceptor tuning changes gradually, under control of sex steroids, not feedback from the EOD itself. Steroid implants retune the tuberous electroreceptors even after the EOD has been silenced surgically (Keller et al., 1986; Ferrari and Zakon, 1989). Not surprisingly, androgens masculinize the electrosensory system (electroreceptors, pacemaker, and electric organ) and estrogens feminize it (reviews by Zakon, 1993, 1996, 1998; Zakon et al., 1999). B. THERMAL TRAPS Narrow receptor tuning and pure EOD frequencies may have costs beyond energetics. The tight spectral match of narrowly tuned signal and sense may trap the wave ﬁsh in a narrow thermal zone. Many neural processes are temperature dependent. Electroreceptor tuning, discharge rate, and EOD pulse width all vary with the ambient temperature (Feng, 1976; Hopkins, 1976; Dunlap et al., 2000). If these all change the same amount over the temperature range of interest, then receptor spectrum tracks the EOD spectrum and the ﬁsh can continue to electrolocate (Hopkins, 1976). If tuning of these different structures changes differently with temperature, then the ﬁsh would experience a tuning mismatch which, if extreme enough, would interfere with active electrolocation. Subtropical Uruguay represents the southern geographic limit of the gymnotiforms. Brachyhypopomus pinnicaudatus lives in the shallowest water (40–80 cm) which undergoes extreme daily temperature swings. Typical daily water temperatures during the breeding season range from 18◦ C at sunrise to 33◦ C in the late afternoon (Silva et al., 1999). Eigenmannia virescens inhabits deeper water in the same ponds, deeper water being notably more stable in temperature (A. Silva, personal communication). Based on published measurements of temperature effects on tuning of electric organ and electroreceptors of Eigenmannia, we can estimate what would happen to its electrosensory system if this wave ﬁsh were to dwell in the thermally unstable shallows. EOD tuning in Eigenmannia varies with temperature following a Q10 of 1.5 below 26◦ C and a Q10 of 1.0 above 26◦ C (Feng, 1976). The Q10 of tuberous electroreceptors is higher—Hopkins (1976) estimated the Q10 for time-coding tuberous electroreceptors (T-units) in Eigenmannia and Sternopygus to be about 2.9 over the temperature range 20–22◦ C. In Fig. 16, I calculate best frequencies for T-units and the fundamental frequency of the EOD over the temperature range experienced by B. pinnicaudatus in shallow water. I use the published Q10 values and assume the spectral match between the signal and receptor occurs at 26◦ C. I make several

ELECTRIC SIGNALS

231

FIG. 16. Tuning of electroreceptors and EOD discharge rates vary with ambient temperature. EOD discharge rate and electroreceptor best frequency are calculated here for the wave ﬁsh Eigenmannia using published Q10 values (Feng, 1976; Hopkins, 1976). The temperature range shown here is experienced daily by another ﬁsh, Brachyhypopomus pinnicaudatus in marshes of Uruguay. Based on these calculations, were Eigenmannia to experience the same temperature swing without retuning its electrosensory or electromotor system, the ﬁsh would lose sense of its own electric signal.

critical assumptions here, chief among them that Q10 estimates and spectral width are constant for electroreceptors across the temperature range considered. Hopkins (1976) shows that the frequency response of Eigenmannia’s T-units is sharp, particularly the upper shoulder. Thus at both temperature extremes, the spectral mismatch between tuberous receptors and the EOD appears sufﬁcient to drop the response of the receptors down to the spontaneous ﬁring rate where the ﬁsh would become deaf to its own signal. According to this model, without a mechanism for rapid retuning of electroreceptors a wave ﬁsh should not be able electrolocate throughout the broad temperature swings of shallow water in the subtropics. As no such mechanism has been found, the wave ﬁsh may be in a thermal trap, restricted to habitat with constant temperature. EODs of pulse ﬁsh have broad spectra and electroreceptors of pulse ﬁsh are broadly tuned (Bastian, 1976, 1977; Watson and Bastian, 1979). Pulse ﬁsh thus would not have the same thermal range restrictions as wave ﬁsh, a feature which may allow them to exploit habitats with more variable temperatures. A useful set of experiments would be to explore the effects of temperature ﬂuctuation on acuity of active electrolocation in wave and pulse gymnotiforms.

232

PHILIP K. STODDARD

C. IMPEDANCE-MATCHING TO THE EXTERNAL ENVIRONMENT Power transfer between two media is most efﬁcient if the impedances of those two media are identical. The ossicles of the terrestrial vertebrate’s middle ear are an example of a structural adaptation to match the low impedance of air in the transmission environment to the high impedance of liquid in the inner ear. To transfer electric energy out of the electric organ and into the water with greatest efﬁciency, an electric ﬁsh needs to match the impedance of the electric organ and surrounding body tissues to the impedance of the water (Kramer and Kuhn, 1993; Caputi and Budelli, 1995). The higher the conductivity of the water, the greater is the current output of the electric organ needed to match the external impedance. Pulse ﬁsh of the genus Brachyhypopomus show characteristic species and racial differences in the number of parallel columns of electrocytes in their electric organ in association with the conductivity of the water in which they are naturally found. Brachyhypopomus species residing in high-conductivity water (∼250–350 µS/cm) have ﬁve to six columns, those in medium-conductivity water (∼70–100 µS/cm) have four columns, and those found in low-conductivity water (<10 µS/cm) have three columns (Sullivan et al., 1996; Hopkins, 1999a). The number of electrocyte columns is not inﬂuenced by the water conductivity in which the individual hatches and develops (Franchina, 1997). Impedancematching by the electric organ structure may act as a mechanism for niche partitioning within a geographic region (Crampton, 1998b; Hopkins, 1999a).

D. THE PAUCITY OF MORMYROID WAVE FISH As with the Gymnotiformes, the mormyroid electric ﬁsh of Africa split into two clades with different EOD characteristics, Mormyridae, the pulse ﬁsh, and Gymnarchidae, the wave ﬁsh (Sullivan et al., 2000). Unlike the order Gymnotiformes, which has tremendous species diversity among wave EOD taxa, the African wave family Gymnarchidae is represented by a single species, Gymnarchus niloticus, a large marsh-dweller that gulps air from the surface (Budgett, 1901; Svensson, 1933). Missing from Africa is the diverse radiation of small wave ﬁsh. What could be the basis for continental differences in diversity of wave species? Evolutionary chance is always a default explanation, but given that Gymnarchus splits off the base of the African mormyroids (i.e., it has been around a long time) and given the enormous radiation of the mormyrid pulse ﬁshes, it does not seem plausible that evolution has not had a chance to produce a radiation of gymnarchid wave ﬁsh. Nor should thermal ﬂuctuation

ELECTRIC SIGNALS

233

constrain wave ﬁsh in the tropical African rivers. That would leave as possible restrictions (1) anoxia, (2) predation, and (3) niche competition: 1. Are African rivers prone to periodic oxygen depletion? Although Gymnarchus breathes air, air-breathing is most common among ﬁsh of shallow waters for whom frequent trips to the surface do not interrupt the normal course of life. Gymnarchus is the largest mormyroid electric ﬁsh, reaching lengths in excess of 1 m. Its large body size may allow air-breathing to sustain an oxygen-costly EOD in two ways. (1) large ﬁsh may be safer while traveling to the surface for the oxygen needed to sustain a continuous wave EOD. (2) Large body size may increase the relative amount of oxygen that can be banked in the blood and spent on activities other than basal metabolism. 2. Are the rivers inhabited by predators that are even more effective than those in South America? Unlike the sternopygid wave ﬁsh of South America, Gymnarchus has little or no dc offset to its monophasic EOD (Bennett, 1971). 3. Are mormyrid pulse ﬁsh such effective competitors that they excluded all the wave ﬁsh except Gymnarchus niloticus? The scarcity of mormyroid wave ﬁsh is an interesting puzzle still to be solved. VI. THE STORY THUS FAR I have suggested here that the signal strategies of electric ﬁsh do in fact impose constraints on their life histories. Not surprisingly, the division of the weakly electric ﬁsh into pulse and wave strategists has forced independent solutions to the problems of communication in habitat with hostile eavesdroppers. But the division into two duty-cycle strategies has proven more signiﬁcant than this. The suite of specializations that make up the wave sensory strategy appear ideally adapted for noise reduction, but it also may have imposed signiﬁcant restrictions on use of habitat by ﬁsh with this signal type. Electrosensory models predict that the tight spectral match between signal and receptors imposes two critical constraints on wave ﬁsh: they may experience problems electrolocating in habitat subject to rapid temperature ﬂuctuations, and they cannot lower their discharge rates to save energy without compromising their electrolocation sense. Further restricting the wave ﬁsh, their overall metabolic needs appear greater than those of pulse ﬁsh, as determined by their greater need for oxygen and inability to penetrate anoxic or hypoxic habitats. In spite of these restrictions, the wave strategy, however specialized, has proven extremely successful in the freshwater river systems of South America.

234

PHILIP K. STODDARD

VII. FUTURE INVESTIGATIONS The pictures I have pieced together here have been synthesized from a mix of ﬁeld and lab studies. They have beneﬁted from discussions with many colleagues. Many of these ideas are explanations proposed to address unsolved questions posed by members of my lab in our weekly meetings and undoubtedly from similar discourses while I was a member of Carl Hopkins a decade earlier. I view this discourse as setting a framework for future study more than as a closed-door explanation of how the electric world came to be. Our understanding of these issues has glaring holes which should be seen as invitations for study. These gaps present exciting opportunities for young scientists looking for a ﬁeld where they can make signiﬁcant contributions to our understanding of animal communication. I outline below some areas in need of work. A. ELECTROSENSORY PREDATION r Our knowledge of electrosensory predation has been assembled from disjunct fragments. We need a good series of behavioral studies showing the foraging habits and hunting techniques of electrosensory predators and their effects on electric ﬁsh behavior and populations. r It would be interesting to see if the electric eel engages in active tracking by using its electrolocation EOD to bias weak signals of gymnotiform prey into the dynamic range of its tuberous electroreceptors. B. NEUROETHOLOGY OF ELECTRIC MATE CHOICE r Sensory networks seem very similar among gymnotiform families (Kennedy and Heiligenberg, 1994; Heiligenberg et al., 1996; Wong, 1997a,b), but it would be worth ﬁnding out whether other gymnotiforms besides Eigenmannia have ampullary-tuberous combination neurons that project from the torus semicircularis to the preoptic area as reported by Heiligenberg et al. (1991), and whether courtship signals with dc spectral components excite these circuits. r Much work has been done on the effects of hormones on the electrosensory system (Keller et al., 1986; Dulka et al., 1995; Zakon, 1996) but very little is known of electrosensory inﬂuence over the endocrine system (Wong, 1997a). r Controlled mate choice studies are needed to determine whether such traits as EOD amplitude, duration, and higher discharge rate signal attractiveness per se, or whether they merely increase probability of detection by the female.

ELECTRIC SIGNALS

235

C. ENVIRONMENTAL PHYSIOLOGY OF ELECTROLOCATION r Is water ﬂow a reasonable basis for evolutionary differentiation of pulse and wave strategies as originally proposed by Lissmann (1958)? We need a systematic study to determine whether discharge rate actually inﬂuences the speed of sensory integration. r Crampton (1998a) has shown wave ﬁsh need more oxygen to maintain their EOD output than do pulse ﬁsh. Is their energetic expenditure higher, or are they simply less efﬁcient at extracting dissolved oxygen from water? How much of their energy goes to electrogenesis? My lab is currently working on this problem. r What are the effects of temperature on electrolocation? Are wave ﬁsh really more compromised by rapid temperature swings as the model suggests? How rapidly can the electrosensory system compensate for temperature changes and by what mechanisms? Over the past two decades, students of animal behavior have been drawn in one of two directions, toward the evolutionary track of behavioral ecology or the mechanistic track of neuroethology. The advent of molecular biology is so alluring that many young neuroethologists are failing to anchor their mechanistic studies in a ﬁrm understanding of the behavioral endpoint. On the other side of the divide, many behavior ecologists have adopted molecular techniques to answer questions of kinship and phylogeny, but only a few have risked the learning curve involved with applying neurobiological techniques. While neuroethology has the capacity to bridge between molecular biology and behavioral ecology, I watch a further widening of the divide. In spite of this schism, or perhaps because of it, exciting potential remains for integration of the two approaches. The Holy Grail of this integrative approach is to identify every link in the chain between natural selection on behaviors and the molecular mechanisms that produce these behaviors. My heroes in this ﬁeld are the scientists who have integrated studies of neural mechanisms with studies of natural behavior. Electric ﬁsh are ideal subjects for such a synthesis and the ﬁeld remains wide open.

VIII. SUMMARY Animals that depend on electric signals for navigation and communication have to balance (1) physical constraints on electrostatic conduction and detection, (2) physiological constraints on sensation and signal production, and (3) selective pressures of predation and communication. Electric ﬁsh have evolved two broad signal strategies, each with its inherent advantages and constraints with respect to these conﬂicting environmental demands.

236

PHILIP K. STODDARD

Discontinuous “pulse” signals are more ﬂexible in that their waveforms and discharge rates can be varied to suit the energetic and communication needs of the signaler. But pulse signals are also more prone to electric interference from environmental noise and other electric ﬁsh and they may be more detectable by electrosensory predators eavesdropping on the lowfrequency spectrum. Continuous “wave” signals are less prone to electric interference and are probably much less conspicuous to electroreceptive predators. But the tight match between the narrow spectrum of the wave signal and the narrow-band electroreceptors also constrains wave signalers to a narrow range of environmental conditions (high oxygen and constant temperature) that can sustain their continuous and unvarying electric output. Pulse and wave signalers have evolved various techniques for suppressing low-frequency energy in their electrolocation signals and for adding it back into their courtship signals, probably in response to mating preferences of females, but also at the peril of increased predation. Both pulse and wave strategies evolved independently in gymnotiform electric ﬁsh of the neotropics and the mormyroid electric ﬁsh of Africa. Although wave ﬁsh are abundant in the New World, they are represented by only a single species in Africa; the reason for the disparity is as yet undetermined. Acknowledgments I thank Carl Hopkins for introducing me to many of the ideas discussed in this chapter. These ideas have been reﬁned in discussions with James Albert, Lauren Chapman, Will Crampton, Cheryl Franchina, Mary Hagedorn, Walter Heiligenberg, Mark Kilburn, Bernd Kramer, Brian Rasnow, Vicky Salazar, and John Sullivan. My work has been supported by NIH grant GM08205. Special thanks to my wife Gray Read for “holding the fort” long enough for me to commit these ideas to paper.

References Albert, J. S., and Campos-da-Paz, R. (1998). Phylogenetic systematics of the Gymnotiformes with diagnoses of 58 clades: A review of available data. In “Phylogeny and Classiﬁcation of Neotropical Fishes” (L. R. Malabarba, R. E. Reis, R. P. Vari, Z. M. Lucena, and C. A. S. Lucena, Eds.). Edipucrs, Porto Alegre, Brazil. Alves-Gomes, J. A., Orti, G., Haygood, M., Heiligenberg, W., and Meyer, A. (1995). Phylogenetic analysis of the South American electric ﬁshes (order Gymnotiformes) and the evolution of their electrogenic system: A synthesis based on morphology, electrophysiology, and mitochondrial sequence data. Mol. Biol. Evol. 12, 298–318. Andersson, M. (1994). “Sexual Selection.” Princeton Univ. Press, Princeton, NJ. Andres, K. H., von During, M., and Petrasch, E. (1988). The ﬁne structure of ampullary and tuberous electroreceptors in the South American blind catﬁsh Pseudocetopsis spec. Anat. Embryol. 177, 523–535. Assad, C., Rasnow, B., Stoddard, P. K., and Bower, J. M. (1998). The electric organ discharges of the gymnotiform ﬁshes: II. Eigenmannia. J. Comp. Physiol. A 183, 419–432.

ELECTRIC SIGNALS

237

Atema, J. (1969). The chemical senses in feeding and social behavior of the catﬁsh, Ictalurus natalis. Ph.D. thesis. University of Michigan, Ann Arbor, Michigan. Baron, V. D. (1994). African Clarias catﬁsh emits long-lasting weak electric pulses. Experientia 50, 644–647. Baron, V. D., Morshnev, K. S., Olshansky, V. M., and Orlov, A. A. (1994). Electric organ discharges of two species of African catﬁsh (Synodontis) during social behaviour. Anim. Behav. 48, 1472–1475. Bass, A. H. (1986). Electric organs revisited. In “Electroreception.” (T. H. Bullock and W. Heiligenberg, Eds.), pp. 13–70. Wiley, New York. Bastian, J. (1976). Frequency response characteristics of electroreceptors in weakly electric ﬁsh (Gymnotoidei) with a pulse discharge. J. Comp. Physiol. 112, 165–190. Bastian, J. (1977). Variations in the frequency response of electroreceptors dependent on receptor location in weakly electric ﬁsh (Gymnotoidei) with a pulse discharge. J. Comp. Physiol. 121, 53–64. Bastian, J., and Heiligenberg, W. (1980). Phase-sensitive midbrain neurons in Eigenmannia: Neural correlates of the jamming avoidance response. Science 209, 828–831. Bell, C., Bodznick, D., Montgomery, J., and Bastian, J. (1997). The generation and subtraction of sensory expectations within cerebellum-like structures. Brain Behav. Evol. 50, 17–31. Bell, C. C., and Grant, K. (1989). Corollary discharge inhibition and preservation of temporal information in a sensory nucleus of mormyrid electric ﬁsh. J. Neurosci. 9, 1029–1044. Bennett, M. L. V. (1961). Modes of operation of electric organs. Ann. N. Y. Acad. Sci. 94, 458–509. Bennett, M. L. V. (1971). Electric organs. In “Fish Physiology” (W. S. Hoar and D. J. Randall Eds.), pp. 347–491. Academic Press, London. Bennett, M. L. V., and Grundfest, H. (1961). Studies on the morphology and electrophysiology of electric organs. III. Electrophysiology of electric organs in mormyrids. In “Electrogenesis: A Comparative Survey of Its Mechanisms with Particular Emphasis on Electric Fishes” (C. Chagas and A. P. de Carvalho, Eds.), pp. 113–135. Elsevier, New York. Bradbury, J. W., and Vehrencamp, S. L. (1998). “Principles of Animal Communication.” Sinauer, Sunderland, MA. Brenowitz, E. A. (1986). Environmental inﬂuences on acoustic and electric animal communication. Brain Behav. Evol. 28, 32–42. Budgett, J. S. (1901). On the breeding habits of some West African ﬁshes with an account of the external features in development of Protopterus annectens, and a description of the larva of Polypterus lapradei. Trans. Zool. Soc. London 16, 115–137. Bullock, T. H., and Heiligenberg, W. (1986). “Electroreception.” Wiley, New York. Bullock, T. H., Behrend, K., and Heiligenberg, W. (1975). Comparison of the jamming avoidance responses in gymnotoid and gymnarchid electric ﬁsh: A case of convergent evolution of behavior and its sensory basis. J. Comp. Physiol. 103, 97–121. Caputi, A., and Budelli, R. (1995). The electric image in weakly electric ﬁsh. I. A data-based model of waveform generation in Gymnotus carapo. J. Comput. Neurosci. 2, 131–147. Carr, C. E., and Maler, L. (1986). Electroreception in gymnotiform ﬁsh. Central anatomy and physiology. In “Electroreception” (T. H. Bullock and W. Heiligenberg, Eds.), pp. 319–373. Wiley, New York. Carr, C. E., Maler, L., Heiligenberg, W., and Sas, E. (1981). Laminar organization of the afferent and efferent systems of the torus semicircularis of gymnotiform ﬁsh: Morphological substrates for parallel processing in the electrosensory system. J. Comp. Neurol. 203, 649–670. Carr, C. E., Heiligenberg, W., and Rose, G. J. (1986). A time-comparison circuit in the electric ﬁsh midbrain. I. Behavior and physiology. J. Neurosci. 6, 107–119. Coates, C. W. (1950). Electric ﬁshes. Elec. Eng. 69, 47–50.

238

PHILIP K. STODDARD

Crampton, W. G. R. (1996). “The electric ﬁsh of the Upper Amazon: Ecology and signal diversity.” Ph.D. thesis Oxford University, London. Crampton, W. G. R. (1998a). Effects of anoxia on the distribution, respiratory strategies and electric signal diversity of gymnotiform ﬁshes. J. Fish Biol. 53, 307–330. Crampton, W. G. R. (1998b). Electric signal design and habitat preferences in a species rich assemblage of gymnotiform ﬁshes from the upper Amazon basin. An. Acad. Bras. Ci. 70, 805–847. Curtis, C. C. (1999). Active female mate choice in the weakly electric ﬁsh Brachyhypopomus pinnicaudatus. M. S. thesis. Florida International University, Miami, Florida. Dulka, J. G., Maler, L., and Ellis, W. (1995). Androgen-induced changes in electrocommunicatory behavior are correlated with changes in substance P-like immunoreactivity in the brain of the electric ﬁsh Apteronotus leptorhynchus. J. Neurosci. 15, 1879–1890. Dunlap, K. D., Smith, G. T., and Yekta, A. (2000). Temperature dependence of electrocommunication signals and their underlying neural rhythms in the weakly electric ﬁsh, Apteronotus leptorhynchus. Brain Behav. Evol. 55, 152–162. Dunning, B. B. (1973). A quantitative and comparative analysis of the tonic electroreceptors of Gnathonemus, Gymnotus and Kroptypterus. Ph.D. thesis. University of Minnesota. Dusenbery, D. B. (1992). “Sensory Ecology.” Freeman, New York. Faraday, M. (1839). I. Experimental researches in electricity. Fifteenth series. Notice of the character and direction of the electric force of the Gymnotus. Philos. Trans. R. Soc. London, 129, 1–12. Farber, J., and Rahn, H. (1970). Gas exchange between air and water and the ventilation pattern in the electric eel. Respiratory Physiol. 9, 151–161. Feng, A. S. (1976). The effect of temperature on a social behavior of weakly electric ﬁsh Eigenmannia virescens. Comp. Biochem. Physiol. A 55, 99–102. Ferrari, M. B., and Zakon, H. H. (1989). The medullary pacemaker nucleus is unnecessary for electroreceptor tuning plasticity in Sternopygus. J. Neurosci. 9, 1354–1361. Ferrari, M. B., and Zakon, H. H. (1993). Conductances contributing to the action potential of Sternopygus electrocytes. J. Comp. Physiol. A 173, 281–292. Finger, T. E., Bell, C. C., and Carr, C. E. (1986). Comparisons among electroreceptive teleosts: Why are electrosensory systems so similar? In “Electroreception” (T. H. Bullock and W. Heiligenberg, Eds.), pp. 465–482. Wiley, New York. Fink, S. V., and Fink, W. L. (1981). Interrelationships of the ostariophysan ﬁshes (Teleostei). Zool. J. Linn. Soc. 72, 297–353. Franchina, C. R. (1997). Ontogeny of the electric organ and the electric organ discharge in the weakly electric ﬁsh Brachyhypopomus pinnicaudatus (Teleostei, Gymnotiformes). J. Comp. Physiol. A 181, 111–119. Franchina, C. R., and Stoddard, P. K. (1998). Plasticity of the electric organ discharge waveform of the electric ﬁsh Brachyhypopomus pinnicaudatus: I. Quantiﬁcation of day–night changes. J. Comp. Physiol. A 183, 759–768. Franchina, C. R., Salazar, V., Volmar, C. H., and Stoddard, P. K. (2001). Plasticity of the electric organ discharge waveform of male Brachyhypopomus pinnicaudatus: II. Social effects. J. Comp. Physiol. A 187, 45–52. Garey, W. F., and Rahn, H. (1970). Normal arterial gas tensions and pH and the breathing frequency of the electric eel. Respiratory Physiol. 9, 141–150. Hagedorn, M. (1995). The electric ﬁsh Hypopomus occidentalis can rapidly modulate the amplitude and duration of its electric organ discharges. Anim. Behav. 49, 1409–1413. Hagedorn, M., and Carr, C. E. (1985). Single electrocytes produce a sexually dimorphic signal in South American electric ﬁsh, Hypopomus occidentalis (Gymnotiformes, Hypopomidae). J. Comp. Physiol. 156, 522–523.

ELECTRIC SIGNALS

239

Hagedorn, M., and Heiligenberg, W. (1985). Court and spark: Electric signals in the courtship and mating of gymnotoid ﬁsh. Anim. Behav. 33, 254–265. Hagiwara, S., Szabo, T., and Enger, P. S. (1965). Physiological properties of electroreceptors in the electric eel, Electrophorus electricus. J. Neurophysiol. 28, 775–783. Hanika, S., and Kramer, B. (1999). Electric organ discharges of mormyrid ﬁsh as a possible cue for predatory catﬁsh. Naturwissenschaften 86, 286–288. Hanika, S., and Kramer, B. (2000). Electrosensory prey detection in the African sharptooth catﬁsh, Clarias gariepinus (Clariidae), of a weakly electric mormyrid ﬁsh, the bulldog (Marcusenius macrolepidotus). Behav. Ecol. Sociobiol. 48, 218–228. Heiligenberg, W. (1977). “Principles of Electrolocation and Jamming Avoidance in Electric Fish. A Neuroethological Approach.” Springer-Verlag, New York. Heiligenberg, W. (1991). “Neural Nets in Electric Fish.” MIT Press, Cambridge, MA. Heiligenberg, W., and Dye, J. (1982). Labeling of electroreceptive afferents in a gymnotid ﬁsh by intracellular injection of HRP: The mystery of multiple maps. J. Comp. Physiol. 148, 287–296. Heiligenberg, W., Keller, C. H., Metzner, W., and Kawasaki, M. (1991). Structure and function of neurons in the complex of the nucleus electrosensorius of the gymnotiform ﬁsh Eigenmannia: Detection and processing of electric signals in social communication. J. Comp. Physiol. A 169, 151–164. Heiligenberg, W., Metzner, W., Wong, C. J., and Keller, C. H. (1996). Motor control of the jamming avoidance response of Apteronotus leptorhynchus: Evolutionary changes of a behavior and its neuronal substrates. J. Comp. Physiol. A 179, 653–674. Hopkins, C. D. (1973). Lightening as background noise for communication among electric ﬁsh. Nature 242, 268–270. Hopkins, C. D. (1974a). Electric communication in the reproductive behavior of Sternopygus macrurus (Gymnotoidei). Z. Tierpsychol. 35, 518–535. Hopkins, C. D. (1974b). Electric communication: Functions in the social behavior of Eigenmannia virescens. Behaviour 50, 270–305. Hopkins, C. D. (1976). Stimulus ﬁltering and electroreception: Tuberous electroreceptors in three species of gymnotiform ﬁsh. J. Comp. Physiol. 111, 171–208. Hopkins, C. D. (1988). Neuroethology of electric communication. Ann. Rev. Neurosci. 11, 497–535. Hopkins, C. D. (1999a). Design features for electric communication. J. Exp. Biol. 202, 1217–1228. Hopkins, C. D. (1999b). Signal evolution in electric communication. In “Design of Animal Communication” (M. D. Hauser and M. Konishi, Eds.), pp. 461–491. MIT Press, Cambridge, MA. Hopkins, C. D., and Heiligenberg, W. (1978). Evolutionary designs for electric signals and electroreceptors in gymnotoid ﬁshes of Surinam. Behav. Ecol. Sociobiol. 3, 113–134. Hopkins, C. D., and Westby, G. W. (1986). Time domain processing of electric organ discharge waveforms by pulse-type electric ﬁsh. Brain Behav. Evol. 29, 77–104. Hopkins, C. D., Comfort, N. C., Bastian, J., and Bass, A. H. (1990). Functional analysis of sexual dimorphism in an electric ﬁsh, Hypopomus pinnicaudatus, order Gymnotiformes. Brain Behav. Evol. 35, 350–367. Johansen, K. (1970). Air breathing in ﬁshes. In “Fish Physiology” (W. S. Hoar and D. J. Randall, Eds.), pp. 361–413. Academic Press, New York. Juranek, J., and Metzner, W. (1998). Segregation of behavior-speciﬁc synaptic inputs to a vertebrate neuronal oscillator. J. Neurosci. 18, 9010–9019. Kawasaki, M., Rose, G., and Heiligenberg, W. (1988). Temporal hyperacuity in single neurons of electric ﬁsh. Nature 336, 173–176.

240

PHILIP K. STODDARD

Keller, C. H., Zakon, H. H., and Sanchez, D. Y. (1986). Evidence for a direct effect of androgens upon electroreceptor tuning. J. Comp. Physiol. A 158, 301–310. Kennedy, G., and Heiligenberg, W. (1994). Ultrastructural evidence of GABA-ergic inhibition and glutamatergic excitation in the pacemaker nucleus of the gymnotiform electric ﬁsh, Hypopomus. J. Comp. Physiol. A 174, 267–280. Keynes, R. D., and Martins-Ferreira, H. (1953). Membrane potentials in the electroplates of the electric eel. J. Physiol. 119, 315–351. Kirschbaum, F. (1995). Reproduction and development in mormyriform and gymnotiform ﬁshes. In “Electric Fishes History and Behavior” (P. Moller, Ed.), pp. 267–301. Chapman & Hall, London. Knudsen, E. I. (1974). Behavioral thresholds to electric signals in high-frequency ﬁsh. J. Comp. Physiol. 91, 333–353. Knudsen, E. I. (1975). Spatial aspects of the electric ﬁelds generated by weakly electric ﬁsh. J. Comp. Physiol. 99, 103–118. Kramer, B. (1995). “Electroreception and Communication in Fishes.” Georg Fischer Verlag, Stuttgart. Kramer, B., and Kuhn, B. (1993). Electric signalling and impedance matching in a variable environment. Naturwissenschaften 80, 43–46. Lauder, G. V., and Liem, K. F. (1983). The evolution and interrelationships of the actinopterygian ﬁshes. Bull. Mus. Comp. Zool. Harvard 150, 95–197. Lissmann, H. W. (1958). On the function and evolution of electric organs in ﬁsh. J. Exp. Biol. 35, 156–191. Lundberg, J. G., Lewis Jr., W. M., Saunders III, J. F., and Mago-Leccia, F. (1987). A major food web component in the Orinoco River channel: Evidence from planktivorous electric ﬁshes. Science 237, 81–83. Lundberg, J. G., Cox Fernandes, C., Albert, J. S., and Garcia, M. (1996). Magosternarchus, a new genus with two new species of electric ﬁshes (Gymnotiformes: Apteronotidae) from the Amazon River basin, South America. Copeia, 657–670. Maler, L., Sas, E. K. B., and Rogers, J. (1981). The cytology of the posterior lateral line lobe of high-frequency weakly electric ﬁsh (Gymnotidae): Dendritic differentiation and synaptic speciﬁcity in a simple cortex. J. Comp. Neurol. 195, 87–139. Marrero, C., and Taphorn, D. C. (1991). Notas sobre la historia natural y la distribuicion de los peces gymnotiformes en la cuenca del Rio Apure e otros rios de la Orinoquia. Biollan 8, 123–142. Marrero, C., and Winemiller, K. O. (1993). Tube snouted gymnotiform and mormyriform ﬁshes: Convergence of a specialized foraging mode in teleosts. Environ. Biol. Fish. 38, 299–309. Merron, G. S. (1993). Pack-hunting in two species of catﬁsh, Clarias gariepinus and C. ngamensis, in the Okavango Delta, Botswana. J. Fish Biol. 43, 575–584. Metzner, W., and Heiligenberg, W. (1991). The coding of signals in the electric communication of the gymnotiform ﬁsh Eigenmannia: From electroreceptors to neurons in the torus semicircularis of the midbrain. J. Comp. Physiol. A 169, 135–150. Metzner, W., and Juranek, J. (1997). A sensory brain map for each behavior? Proc. Natl. Acad. Sci. U.S.A. 94, 14798–14803. Meyer, J. H. (1983). Steroid inﬂuences upon the discharge frequency of a weakly electric ﬁsh. J. Comp. Physiol. A 153, 29–37. Moller, P. (1995). “Electric Fishes History and Behavior.” Chapman & Hall, London. Naruse, M., and Kawasaki, M. (1998). Possible involvement of the ampullary electroreceptor system in detection of frequency-modulated electrocommunication signals in Eigenmannia. J. Comp. Physiol. A 183, 543–552.

ELECTRIC SIGNALS

241

Northcutt, R. G. (1986). Electroreception in nonteleost bony ﬁshes. In “Electroreception” (T. H. Bullock and W. Heiligenberg, Eds.), pp. 257–285. Wiley, New York. Peters, R. C., and Buwalda, R. J. A. (1972). Frequency response of the electroreceptors (“small pit organs”) of the catﬁsh, Ictalurus nebulosus LeS. J. Comp. Physiol. 79, 29–38. Rasnow, B., and Bower, J. M. (1996). The electric organ discharges of the gymnotiform ﬁshes: I. Apteronotus leptorhynchus. J. Comp. Physiol. A 178, 383–396. Reid, S. (1983). La biologia de los bagres rayados Pseudoplatystoma fasciatum y P. tigrinum en la cuenca del rio Apure–Venezuela. Revista UNELLEZ de Ciencia y Tecnologia 1, 13–41. Rose, G., and Heiligenberg, W. (1986). Limits of phase and amplitude sensitivity in the torus semicircularis of Eigenmannia. J. Comp. Physiol. A 159, 813–822. Roth, A. (1972). The function of electroreceptors in catﬁsh. J. Comp. Physiol. 79, 113–135. Ryan, M. (1985). “The Tungara ´ Frog: A Study in Sexual Selection and Communication.” Univ. Chicago Press, Chicago. Ryan, M. J., Tuttle, M. D., and Rand, A. S. (1982). Bat predation and sexual advertisement in a neotropical frog. Am. Nat. 119, 136–139. Ryan, M. J., Tuttle, M. D., and Barclay, R. M. R. (1983). Behavioral responses of the frog-eating bat to sonic frequencies. J. Comp. Physiol. 150, 413–418. Ryan, M. J., Fox, J. H., 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, 82–96. Schaefer, J., and Zakon, H. H. (1996). Opposing actions of androgen and estrogen on in vitro ﬁring frequency of neuronal oscillators in the electromotor system. J. Neurosci. 16, 2860–2868. Shumway, C. A. (1989a). Multiple electrosensory maps in the medulla of weakly electric gymnotiform ﬁsh. I. Physiological differences. J. Neurosci. 9, 4388–4399. Shumway, C. A. (1989b). Multiple electrosensory maps in the medulla of weakly electric gymnotiform ﬁsh. II. Anatomical differences. J. Neurosci. 9, 4400–4415. Shumway, C. A., and Zelick, R. D. (1988). Sex recognition and neuronal coding of electric organ discharge waveform in the pulse-type weakly electric ﬁsh, Hypopomus occidentalis. J. Comp. Physiol. A 163, 465–478. Silva, A., Quintana, L., Galeano, M., Errandonea, P., and Macadar, O. (1999). Water temperature sensitivity of the EOD waveform in Brachyhypopomus pinnicaudatus. J. Comp. Physiol. A 185, 187–198. Sisneros, J. A., and Tricas, T. C. (2000). Androgen-induced changes in the response dynamics of ampullary electrosensory primary afferent neurons. J. Neurosci. 20, 8586–8595. Sisneros, J. A., Tricas, T. C., and Luer, C. A. (1998). Response properties and biological function of the skate electrosensory system during ontogeny. J. Comp. Physiol. A 183, 87–99. Smith, G. T., and Zakon, H. H. (2000). Pharmacological characterization of ionic currents that regulate the pacemaker rhythm in a weakly electric ﬁsh. J. Neurobiol. 42, 270–286. Stoddard, P. K. (1994). Low-frequency electric ﬁeld production and concealment in gymnotiform ﬁsh with pulsed electric organ discharges. Soc. Neurosci. Abstr. 24, 159.9. Stoddard, P. K. (1999). Predation enhances complexity in the evolution of electric ﬁsh signals. Nature 400, 254–256. Stoddard, P. K., Rasnow, B., and Assad, C. (1999). Electric organ discharges of the gymnotiform ﬁshes: III. Brachyhypopomus. J. Comp. Physiol. A 184, 609–630. Sullivan, J. P. (1997). A phylogenetic study of the Neotropical hypopomid electric ﬁshes (Gymnotiformes: Rhamphichthyoidea). Ph.D. thesis. Duke University, Durham, NC.

242

PHILIP K. STODDARD

Sullivan, J. P., Hopkins, C. D., and Comfort, N. C. (1996). Water conductivity as an invisible barrier to dispersal in South American freshwater for hypopomid electric ﬁshes. American Society for Ichthyology and Herpetology, New Orleans. Sullivan, J. P., Lavoue, S., and Hopkins, C. D. (2000). Molecular systematics of the African electric ﬁshes (Mormyroidea: teleostei) and a model for the evolution of their electric organs. J. Exp. Biol. 203, 665–683. Svensson, G. S. O. (1933). Freshwater ﬁshes from the Gambia River (British West Africa). Results of the Swedish Expedition. K. Svenska Vetensk. Akad. Handl. 12, 1–102. Szabo, T. (1974). Anatomy of the specialized lateral line organs of electroreception. In “Handbook of Sensory Physiology” (A. Fessard, Ed.), pp. 13–58. Springer-Verlag, Berlin. Szamier, R. B., and Wachtel, A. W. (1970). Special cutaneous receptor organs of ﬁsh. VI. Ampullary and tuberous organs of Hypopomus. J. Ultrastruct. Res. 30, 450–471. Tricas, T. C., Michael, S. W., and Sisneros, J. A. (1995). Electrosensory optimization to conspeciﬁc phasic signals for mating. Neurosci. Lett. 202, 129–132. Unguez, G. A., and Zakon, H. H. (1998). Phenotypic conversion of distinct muscle ﬁber populations to electrocytes in a weakly electric ﬁsh. J. Comp. Neurol. 399, 20–34. Watson, D., and Bastian, J. (1979). Frequency response characteristics of electroreceptors in the weakly electric ﬁsh Gymnotus carapo. J. Comp. Physiol. 134, 191–202. Westby, G. W. M. (1988). The ecology, discharge diversity and predatory behaviour of gymnotiforme electric ﬁsh in the coastal streams of French Guiana. Behav. Ecol. Sociobiol. 22, 341–354. Winemiller, K. O., and Kelso-Winemiller, L. C. (1996). Comparative ecology of catﬁshes of the Upper Zambezi River ﬂoodplain. J. Fish Biol. 49, 1043–1061. Wong, C. J. (1997a). Afferent and efferent connections of the diencephalic prepacemaker nucleus in the weakly electric ﬁsh, Eigenmannia virescens: Interactions between the electromotor system and the neuroendocrine axis. J. Comp. Neurol. 383, 18–41. Wong, C. J. (1997b). Connections of the basal forebrain of the weakly electric ﬁsh, Eigenmannia virescens. J. Comp. Neurol. 389, 49–64. Zahavi, A. (1977a). The cost of honesty (further remarks on the handicap principle). J. Theor. Biol. 67, 603–605. Zahavi, A. (1977b). Mate selection—Selection for a handicap. J. Theor. Biol. 53, 205–214. Zakon, H. H. (1986). The electroreceptive periphery. In “Electroreception” (T. H. Bullock, and W. Heiligenberg, Eds.), pp. 103–156. Wiley, New York. Zakon, H. H. (1993). Weakly electric ﬁsh as model systems for studying long-term steroid action on neural circuits. Brain Behav. Evol. 42, 242–251. Zakon, H. H. (1996). Hormonal modulation of communication signals in electric ﬁsh. Dev. Neurosci. 18, 115–123. Zakon, H. H. (1998). The effects of steroid hormones on electrical activity of excitable cells. Trends Neurosci. 21, 202–207. Zakon, H. H., and Unguez, G. A. (1999). Development and regeneration of the electric organ. J. Exp. Biol. 202, 1427–1434. Zakon, H. H., McAnelly, L., Smith, G. T., Dunlap, K., Lopreato, G., Oestreich, J., and Few, W. P. (1999). Plasticity of the electric organ discharge: Implications for the regulation of ionic currents. J. Exp. Biol. 202, 1409–1416.

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 31

How to Vocally Identify Kin in a Crowd: The Penguin Model THIERRY AUBIN1 AND PIERRE JOUVENTIN 2 1 NAM-CNRS URA 1491

´ PARIS-XI UNIVERSITE F-91400 ORSAY, FRANCE 2 CEFE-CNRS 34293 MONTPELLIER CEDEX 5, FRANCE

I. INTRODUCTION Sociality has several major advantages but also some disadvantages. For example, it is well known that colonial breeding improves care of offspring through communal protection (Alcock, 1972), but it is less well known that the difﬁculty of ﬁnding one’s own young is increased. Marine birds and mammals breed on land but have to forage at sea. This fact is indeed the main constraint on their social behavior and their life history: seals have large fat reserves and in some species such as fur seals the adults have to alternate feeding trips at sea, a characteristic particularly striking in seabirds also (see Jouventin and Cornet, 1980, for a comparison between pinnipeds and seabirds; Bried and Jouventin, 2001, for seabirds). A wandering albatross, Diomedea exulans, may ﬂy several thousand kilometers to ﬁnd food (Jouventin and Weimerskirch, 1990) and a king penguin, Aptenodytes patagonicus, may swim several hundred kilometers on a foraging trip. As a consequence, both sexes have to cooperate to brood and rear one chick, and this heavy breeding cost explains why all of the nearly 200 species of seabirds are monogamous (Lack, 1968). Pelagic birds have few if any predators on land, and they usually breed in colonies numbering several hundreds or thousands of pairs. After ﬁshing for several hours, days, or weeks, one parent comes back to the crowded colony, ﬁnds its mate and takes its turn at brooding while the mate forages. Later, both parents have to forage at sea to feed the growing young. The arriving parent has to be recognized at the nest by its mate to be safe and to brood in turn and later by the large chick(s) to feed it. An 243

C 2002 by Academic Press Copyright All rights of reproduction in any form reserved. 0065-3454/02 $35.00

244

THIERRY AUBIN AND PIERRE JOUVENTIN

alien bird is pecked and harassed away. The ability to invest cooperatively in their own offspring (Trivers, 1972) by identifying the members of the family is thus crucial for marine animals, as it is for many other animal species. Among seabirds where feeding trips are an ever-present feature of the breeding biology, the family of penguins is unique (1) in our knowledge of their individual recognition strategies, and (2) in the fact that they exhibit two different methods of care for eggs and chicks (some species have a nest and others do not). In a comparative study (Jouventin, 1982), we demonstrated experimentally in several species that penguins identify their mates or their chick(s) only by vocal cues, being unable to use visual cues to distinguish family members from surrounding birds: the display call is consequently the only marker used in kin recognition. Penguins are also unique because they include both nesting species as in other seabirds and nonnesting species, which walk with their egg or chick on their feet. In this group of colonial species, in which even topographical cues of nest location can be lacking (Isenmann and Jouventin, 1970), the difﬁculty of meeting up with mates and offspring is extreme, particularly with the latter, unusual way of breeding. So penguins, which are easy to observe, to record, to manipulate, and to test, constitute extraordinary models for studying individual recognition by acoustic means. There are 15 species of penguins that, as in most seabirds, breed on a nest. This nest can be made with stones, as in the antarctic Adelie ´ penguin (Pygoscelis adeliae) or the subantarctic macaroni penguin (Eudyptes chrysolophus). Nests can also be made of grass, as in the gentoo penguin (P. papua), which breeds on the same subantarctic islands as the latter species, but on plateaus. Some species nest in a burrow such as the nocturnal little blue penguin (Eudyptula minor) which breeds along the Australian and New Zealand coasts. Much more exceptional in seabirds, the two large (nearly 1-meter tall) nonnesting Aptenodytes penguins brood their single egg on their feet and usually walk around during brooding. They breed on ﬂat and inhospitable areas such as wet subantarctic beaches in the king penguin (A. patagonicus) and the antarctic sea-ice in the emperor penguin (A. forsteri). The ﬁrst of these does not move more than a few meters during the brooding phase but is much more mobile during the rearing phase when the chick waits alone for its parents to feed it (Stonehouse, 1960; Lengagne et al., 1999a). The latter species moves through all of the breeding cycle because when there are blizzards blowing, brooders and chicks have to huddle in tight groups ´ and Bourliere, ` 1957; Prevost, ´ 1961). of 10 birds/m2 to keep warm (Prevost During storms, the wind may blow at 350 km/h and the temperature drop

VOCALLY IDENTIFYING KIN IN A CROWD: THE PENGUIN MODEL

245

to −34◦ C (meteorological data from Terre Adelie, ´ where our studies on antarctic penguins were conducted). The ritualized postures and the various calls of the penguin family have been described in detail by Jouventin (1982). The biological meaning of calls and displays was determined through the observation of marked birds, often over several years, according to the ecological and social context. All genera, and for some genera several species, were studied to describe, understand, and compare their visual and vocal displays and to test by playback their use of calls in individual recognition. Using computers, more sophisticated playback experiments have been conducted to examine the coding–decoding systems of penguins. In this chapter we summarize the acoustic constraints and behavioral adaptations described by Jouventin (1982), and we synthesize playback results to give a comparative account of the acoustic systems of six different species (Aubin and Jouventin, 1998; Jouventin et al., 1999; Lengagne et al., 1999a,b,c, 2000; Aubin et al., 2000; Jouventin and Aubin, 2000, 2001).

II. LOCATIONS AND METHODS A. SUBJECTS AND LOCATION Six species of penguins have been studied in the past ﬁve years, at the following locations: r The Adelie ´ penguin and the emperor penguin were studied at the Pointe ´ Antarctica, Geologie ´ Archipelago (66◦ 40′ S, 140◦ 01′ E), Terre Adelie, r The king, the gentoo, and the macaroni penguins were studied mainly at Possession Island (46◦ 25′ S, 51◦ 45′ E), Crozet Archipelago, Indian Ocean, r The little blue penguin was studied at Phillip Island (38◦ 31′ S, 145◦ 08′ E), 60 km from Melbourne, Australia.

B. RECORDING AND PLAYBACK PROCEDURES Calls were recorded with a Sony TCD Pro II DAT recorder (frequency response ﬂat within the range 20–20,000 Hz) and an omnidirectional Sennheiser MKH 815T microphone (frequency response 100–20,000 Hz ± 1 dB) mounted on a 3 m pole, so that birds could be approached without disturbance. The distance between the beak of the recorded bird and the microphone was approximately 1 m for all species studied. Experimental

246

THIERRY AUBIN AND PIERRE JOUVENTIN

signals were broadcast with the same tape recorder connected to a PSP-2 E.A.A. preampliﬁer and a 20 W self-powered ampliﬁer built in the laboratory, equipped with an Audax loudspeaker (frequency response 100–5600 Hz ± 2 dB). For propagation tests, signals were rerecorded by means of an omnidirectional Sennheiser MKH 815T microphone connected to another Sony TCD10 Pro II DAT. For sound pressure level measurements (SPL in dB), we used a Bruel ¨ & Kjaer Sound Level Meter type 2235 (linear scale, slow setting) equipped with a 1 in. condenser microphone type 4176 (frequency response 2.6–18,500 Hz ± 2 dB). For propagation experiments, representative calls of each species studied were chosen. These signals were broadcast repetitively (generally 10 calls) through the colony and recorded at distances of 1 m (reference), 7 m (average distance between two birds when the incoming one started the acoustic search of the mate or the chick), and 14 m (maximum distance of recognition observed in most cases). The loudspeaker and the microphone were mounted on a tripod at the height of a penguin head (0.9 m for a king or emperor penguin, 0.7 m for an Adelie ´ penguin, and 0.4 m for a little blue penguin). To quantify the screening effect of the bodies of birds, the recordings were compared with propagation records made at the same microphone and loudspeaker height and the same distances, but without any penguins present. The series of recorded calls were then examined in the amplitude-versus-times and the amplitude-versus-frequency domains. In playback experiments, both chicks and adults tested were ﬂipperbanded with a temporary plastic band to identify them. Playback experiments were conducted during clear and dry weather, with a wind speed of less than 4 m/s. The bird tested (adult or chick) was generally quiet, preening itself. The distance between the loudspeaker and the bird was on average 7 m, this corresponding to a natural calling situation between birds. Signals were played at SPL levels equivalent to those produced by the species tested. To prevent habituation, a maximum of three experimental signals per day was broadcast to any one bird. For the different birds tested, the order of presentation of the different experimental signals was randomized. In the same way, the order of presentation of signals tested was not the same for each bird from day to day. Thus, the observed responses for the whole group of individuals tested were neither the result of cumulative excitation nor dependent on playback order. For each species and for each experimental signal, from 8 to 25 individuals were tested. In natural conditions and during the absence of the parents, chicks gathered in ﬂocks, remaining silent and standing or lying quietly. An adult coming from the sea to feed its chick makes its way to the area of the colony where the nest (or the meeting place) is located and calls. Then, the corresponding chick in the ﬂock holds up its head, looks around, calls in reply, and moves

VOCALLY IDENTIFYING KIN IN A CROWD: THE PENGUIN MODEL

247

toward the parent, often running (Jouventin, 1982). The other chicks in the vicinity, resting or preening themselves, exhibit no behavioral reaction to the apparently extraneous calls. The recognition process between mates is exactly the same as this except that the brooding bird does not move toward the incoming bird, but only calls in reply until the two partners meet. On the basis of the observation of these natural meeting situations between parents and chicks or between mates a behavioral scale was used to evaluate the intensity of response of tested birds for all the different penguin species studied. The behavioral scale was ranked as follows: −, no response; +, moderate response (visually inspect the environment by head-turning and calling in reply to the signal after a delay); ++, strong response (turn to loudspeaker, immediately call in reply to the signal, and, only for the chicks, move in the direction of the loudspeaker). C. SOUND ANALYSIS AND SYNTHESIS Signals were digitized through 12- or 16-bit acquisition cards equipped with an antialiasing ﬁlter (low-pass ﬁlter, −120 dB/octave) at a sampling rate of 12–48 kHz (depending on the call studied) and stored on the hard disk of a computer. Calls were then analyzed and signals synthesized mainly using Syntana software built in the laboratory (Aubin, 1994). Signals were examined in the amplitude-versus-frequency domain by spectrum analysis (fast Fourier transform (FFT) calculation) and in the amplitude-versus-time domain by envelope analysis (analytic signal calculation). To follow the time evolution of the frequency we used the Hilbert (Papoulis, 1977) or zerocrossing calculations which provide a representation of the instantaneous frequency. Fundamental frequencies were detected and measured using the Cepstrum calculation deﬁned as the power spectrum of the logarithm of the power spectrum (Noll, 1967). Experimental signals were built either by constructive synthesis (i.e., by computer synthesis starting from scratch) or by destructive synthesis (i.e., by modifying natural calls). For the constructive synthesis, the signal to be synthesized at time t, S(t) is obtained by: S(t) =

N (h)

ωi sin[2πϕ(t)t],

i=1

where N(h) is the harmonic number, ωi is the relative amplitude of the harmonic i as determined from the power spectrum of a reference signal (a natural call taken as model). For the constructive synthesis, natural calls

248

THIERRY AUBIN AND PIERRE JOUVENTIN

were modiﬁed in the temporal, frequency, and amplitude domains. Amplitude and frequency modulations (respectively, AM and FM) were modiﬁed or removed using the Hilbert transform calculation (Bremond ´ and Aubin, 1992; Mbu-Nyamsi et el., 1994). For modiﬁcations of the harmonic structure, natural calls were ﬁltered by low-, high-, or band-pass digital ﬁlters, by applying optimal ﬁltering with FFT (Press et al., 1988; Mbu-Nyamsi et al., 1994). The window size of the FFT was 4096 points. Natural calls were also shifted up or down in frequency by picking a data record (from a natural call) through a square window, applying short-term overlapping (50%) followed by a linear shift up or down of each spectrum, and ﬁnally by a short-term overlapping inverse FFT (Randall and Tech, 1987). As before, the window size was 4096 points. To modify call or syllable durations, we truncated the sounds. To prevent spectral artifacts arising from gaps in amplitude, an envelope was applied (by multiplication) to the data set in the time domain to smooth all the edges.

III. THE CONSTRAINTS A. THE BIOLOGICAL PROBLEMS Most seabirds breed in dense colonies numbering tens, hundreds, or even thousands of pairs. In penguin species, some colonies can near one or two million pairs. It is impressive to follow a king penguin leaving the sea with its stomach full and coming directly to its brooding mate or its chick on a ﬂat area among many thousands of others in only 1 or 2 min (Lengagne et al., 1999a). Although olfaction is well developed in petrels, another family of seabirds, penguins are not known for their abilities in olfaction and seem unable to ﬁnd their partner using smell (Jouventin, 1982; Jouventin and Robin, 1984; Lequette et al., 1989; Verheyden and Jouventin, 1994). More surprisingly, penguins apparently see well, even on land, but they are unable to identify their mate or chick(s) visually. Jouventin (1982 and unpublished data) conﬁrmed its observations by experiments consisting of obstruction of the auditory ducts or of closing the birds’ beaks by means of adhesive tape. In both these situations, partners were unable to recognize each other. The lack of visual recognition is not obvious because, after recognizing their partner vocally, penguins try to follow the bird visually in the crowd. Nevertheless, if the mate or the chick disappears completely in a group, or if the visual link is lost for some minutes, the birds cannot ﬁnd each other again without calling. In fact, penguins, as with fur seals (Roux and Jouventin, 1987) and many other seabirds (Beer 1970, 1979; White and White, 1970; Charrier, 2001), ﬁnd their kin using an acoustic signal. Because individual recognition is only

VOCALLY IDENTIFYING KIN IN A CROWD: THE PENGUIN MODEL

249

vocal in penguins, and the call concerned is the display call, it was possible to experiment using playback of the partner: the mate for adults or, more often, the parent for chicks. B. THE ACOUSTIC PROBLEMS 1. The Background Noise of the Colony In all penguin species, the display call consists of a series of sound components termed syllables, separated by pronounced amplitude declines (see Figs. 1–3). The syllable is a complex signal based on harmonic series from 250 to >5000 Hz. The call is emitted at a relatively high sound pressure level (SPL) ranging from 85–90 dB for small penguins (Jouventin and Aubin, 2001, and manuscript in preparation) to >95 dB for large ones (Robisson, 1993b; Aubin and Jouventin, 1998). As penguins breed in dense colonies, calls emitted by other individuals generate a continuous background noise in the colony. In addition, other signals, such as agonistic calls and chick calls and nonbiologically signiﬁcant sounds (wind, ﬂipper ﬂap), increase the level of ambient noise so that it is particularly high inside the colony (see Fig. 4). Thus, for a king penguin colony numbering 40,000 pairs, we have measured an average value of 74 dBSPL during a 4-min recording period (Aubin and Jouventin, 1998). In these conditions, we hypothesized that penguins could only establish communication at short range, that is, some meters. Penguins are large animals that are able to call loudly, an ability that is probably useful in overcoming the sound of the sea and the wind, but which cannot prevail against the signals of their equally loud neighbors. The noise generated by birds in the colony is almost continuous, and periods of relative silence are short, infrequent, and unpredictable. Thus, in a king penguin colony, periods of relative silence represent only 15% of the time and are of short (mean: 20 ms) duration (Aubin and Jouventin, 1998). In addition, the only noises that cover exactly the same frequency band as that of a penguin call, and which would thus in theory lead to a masking effect (Scharf, 1970), are the calls produced by conspeciﬁcs. The acoustic properties of the masker and those of the signaler are similar. In these conditions of competing noise, the jamming effect is very important, from an amplitude, time, and frequency point of view, and this increases the difﬁculty for a given bird to extract the information provided by the partner. 2. The Screening Effect of Bodies In breeding areas of a penguin colony, the density of birds is high, for example, 2.2 breeders/m2, measured by Barrat (1976) in a king penguin colony. Many individuals (nonbreeding adults or chicks) also gather in groups and sometimes huddle closely together. In such places, the density of birds can

250

THIERRY AUBIN AND PIERRE JOUVENTIN

FIG. 1. Sound spectrograms and oscillograms of (a) an emperor penguin, Aptenodytes forsteri, call; and (b) a king penguin, Aptenodytes patagonicus, call. Calls correspond to a succession of syllables separated by gaps in amplitude.

VOCALLY IDENTIFYING KIN IN A CROWD: THE PENGUIN MODEL

251

FIG. 2. Sound spectrograms and oscillograms of (a) a gentoo penguin, Pygoscelis papua, call; and (b) an Adelie ´ penguin, Pygoscelis adeliae, call.

252

THIERRY AUBIN AND PIERRE JOUVENTIN

FIG. 3. Sound spectrograms and oscillograms of (a) a little blue penguin, Eudyptula minor, call; and (b) a macaroni penguin, Eudyptes chrysolophus, call.

VOCALLY IDENTIFYING KIN IN A CROWD: THE PENGUIN MODEL

253

FIG. 4. Spectra of a 4-min recording of the background noise of king and Adelie ´ colonies (for each colony, average of 937 successive FFTs, window size = 4096 data points, sampling frequency = 16 kHz, Hamming window). Adult and chick calls represent 60 and 55% of the energy (Welch calculation method) in the spectra of king and Adelie ´ colonies, respectively.

become twofold higher than it would be otherwise (Kodyman and Mullins, 1990; Lengagne et al., 1999b) and can even reach 10 birds/m2 in an emperor penguin colony during a blizzard (Prevost, ´ 1961). Then, the bodies of birds constitute an obstructed environment that impairs the propagation of signals. The screening effect of penguin bodies causes absorption and multiplescattering effects. To study the modiﬁcation of the call during propagation, we broadcast and recorded signals at different distances in penguin habitats. In “open ﬁeld” conditions, that is, without any birds between the emitter and the receiver, the measured attenuation of the signal ﬁtted well with the amplitude decrease that would be expected from spherical spreading: it diminishes with distance according to the inverse square law (−6 dB per doubling of distance). When the same measurements are performed with penguins between the emitter and the receiver, excess attenuation occurs because porous objects such as penguin bodies absorb sounds (Wiley and Richard, 1978; Dabelsteen, 1981; Dabelsteen et al., 1993). The greater the distance, and the greater the number of bodies in the way, the more the excess attenuation. According to our measurements, amplitude and frequency parameters of the signal showed strong degradation with increasing broadcast

254

THIERRY AUBIN AND PIERRE JOUVENTIN

TABLE I PEARSON PRODUCT-MOMENT CORRELATION BETWEEN REFERENCEa AND PROPAGATED SIGNALS b c FOR AVERAGED ENVELOPES AND AVERAGED SPECTRA OF THE DISPLAY CALLS ´ OF KING AND ADELIE PENGUINS King penguin

Adelie ´ penguin

Distance of propagation

Correlations for averaged envelopes

Correlations for averaged spectra

Correlations for averaged envelopes

Correlations for averaged spectra

7m

0.54∗∗∗

0.47∗∗∗

0.68∗∗∗

0.54∗∗

0.05∗

0.27∗

0.12∗

0.39∗∗

14 m

Note: ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. a For each species, the call recorded at 1 m. b N = 10,63,600 data points. c N = 10, FFT = 512 data points.

distance (see the results for two penguin species in Table I). Thus, in the penguin colonies studied, after 14 m of propagation, the attenuation is so strong that the amplitude gaps that separate syllables tended to disappear and the amplitude of the signal became equal or inferior to the background noise (signal-to-noise ratio equal to or less than 1). In the frequency domain, peaks above the wavelength corresponding to the body size of a penguin (i.e., above 350 Hz for a 0.9 m high king penguin or emperor penguin and above 500 Hz for a 0.7 m Adelie ´ penguin) are more severely attenuated and disappear in the background noise after 14 m of propagation. For little blue penguins, the same disappearance is observed, but after only 8 m of propagation, due mainly to vegetation effects (Jouventin and Aubin, 2000). For amplitude parameters, peaks tend to be embedded in the noise and to disappear in the background noise after 14 m of propagation. According to these propagation tests, communication involving individual recognition seems possible only at a short or moderate range (<16 m). Communication in a penguin colony, and generally in seabird colonies, appears poorly adapted to transmission of individual information at long range (Robisson, 1991; Aubin and Lengagne, 1997; Aubin and Jouventin, 1998; Lengagne et al., 1999b).

IV. THE SOLUTIONS FOUND A seabird colony, and particularly a penguin colony, is an extreme environment from an acoustic point of view only partly due to the loud background noise. It presents a particularly difﬁcult problem of acoustic communication, due not only to the extraneous noises but also to propagation problems

VOCALLY IDENTIFYING KIN IN A CROWD: THE PENGUIN MODEL

255

linked to the distance between partners or between the parent and chick and to the massive screen of birds. Faced with the problem of ﬁnding a particular individual in a penguin colony among several hundreds or thousands of conspeciﬁcs, the display call alone seems at ﬁrst glance inadequate to secure communication. Nevertheless, penguins succeed, performing acoustic identiﬁcation of the partner or chick usually in a few minutes (i.e., a relatively short time). How do they manage this? A. BEHAVIORAL SOLUTIONS TO OPTIMIZE IDENTIFICATION Acoustic recognition is a critical process for the breeding success of seabirds (Brooke, 1978). The incoming bird arriving from the sea has to ﬁnd its mate or its chick in the colony on the basis of only a few calls. If it does not succeed, the bird returns to the sea. Thus, there is a possibility that the pair will fail to reunite or that the chick will die, so demonstrating the survival value of vocal recognition. To limit the high aggressiveness and the vigorous territorial defense of other breeders (via pecking and ﬂipper ﬂapping), the incoming bird must limit the time it takes in the colony to identify the mate or chick. For this purpose, penguins adopt some special strategies. 1. The Meeting Place Although colonies of penguins often number thousands of birds, an incoming adult does not have to locate its partner among all of these individuals. As described previously, the search is limited to particular meeting places (either nests or previous feeding sites for nonnesting species) that are memorized by adults and chicks. These visual cues may assist in individual recognition, but the mutual display call remains the only way to identify with certainty the partner and offspring. For the nesting species of penguins, such as macaroni, gentoo, or Adelie, ´ the nest is so important that, even when it is abandoned, large chicks stay close to it while waiting at the colony for their parents. The parent comes directly to the location of the old nest and utters the display call which the chick answers immediately. Sometimes several chicks answer and try to be fed but the parent pecks the alien ones, feeding and preening its own. On rare occasions, mates or parents and chicks meet on the nest without calling (Aubin and Jouventin, personal observations with little blue, Adelie, ´ and macaroni penguins). This occurs if no confusion is possible because of other behavioral cues and the use of the nest as a meeting place. In nonnesting penguins, the problem is more difﬁcult because the parent has to ﬁnd its partner with only vocal cues and without any landmarks. In addition, the call decreases in amplitude by half in 9 m and completely in about 15 m in this loud colony noise (Jouventin et al., 1999; Lengagne et al., 1999b). In the king penguin, during the brooding stage, the problem is not so

256

THIERRY AUBIN AND PIERRE JOUVENTIN

different from that of a nesting penguin because the brooder with its egg or small chick on its feet does not move more than 4 m (Lengagne et al., 1999a). Even if they build no nest, breeders lay in a hollow on the ground and they defend this against intruders by vigorous pecks. During the brooding stage, the mate returning from sea needed on average 5 calls and a mean of 2 min (after the ﬁrst call was emitted) to meet its moving partner in the colony (data established on the basis of observation of 28 pairs of birds, Lengagne et al., 1999a). When the incoming bird starts the acoustic search for its mate, the distance between the two birds was 8 m and 70% of the incubating birds were able to discriminate the ﬁrst call emitted by the mate. During the rearing stage, when the chick is too large to ﬁt on the adult’s feet and can walk itself, it stays in the same feeding area some 10 m across and waits among many other chicks. Both parents memorize this place, termed rendezvous site by Stonehouse (1960) and attachment zone by Barrat (1976), and come directly to the edge of the chicks’ huddles to call. The awaiting chick responds, usually within 15 s of the parent’s ﬁrst call (Stonehouse, 1960). Consequently, even without a nesting place, king penguins have a rendezvous site that assists their meeting considerably. In the emperor penguin, there is no nest, feeding place, or landmark and therefore the difﬁculty of ﬁnding the partner or the chick is greater. Birds seem to explore the colony at random, starting more often from the center of the colony and progressively expanding their search to the edge. In this species, the time necessary for an adult to ﬁnd the chick can reach more than 2 h (Robisson, 1993a). Moreover a special behavioral adaptation exists to prevent the jamming of calls: when a bird calls, a neighbor that was about to call (as we can detect by its low head position) stops and waits for the end to utter its own call. In fact, to eliminate signal jamming, two individuals less than 7 m apart abstain from calling together. This “courtesy rule” of not interrupting was proved experimentally by playing a call from a loudspeaker positioned at different distances from a calling bird (Jouventin et al., 1979; Jouventin, 1982). 2. The Signaling Posture For efﬁcient communication, the sender must maximize the propagation of the signal and the listener must optimize reception. Thus, in calling, penguins adopt particular body positions or “signaling postures.” King penguins, and in general all penguin species except emperor penguins, raise their beaks slowly to a vertical position, stretch their necks to their fullest extent, and call (Jouventin, 1982). This signaling posture limits signal-to-noise ratio reduction caused by the screening effect of the bodies of the birds gathered in dense ﬂocks. The macaroni penguin holds its head up, raising the beak, and then swings its head from side to side, beaming the signal in different

VOCALLY IDENTIFYING KIN IN A CROWD: THE PENGUIN MODEL

257

directions in the colony. These calling attitudes, with the beak from 0.4 to 1 m above the ground, maximize the signal transmission distance in the colony (Robisson, 1991, 1993; Lengagne et al., 1999b). Finally, the emperor penguin remarkably directs its beak downward during calling probably to facilitate the control of the temporal pattern of this call (Jouventin, 1982). According to the measures of Robisson (1993a), another complementary explanation is that this unusual posture directs the signal forward, so that sound energy is beamed in the direction that the bird moves during the search for its partner or chick. Concerning the receiver, different postures are observed in the colony: rising to the feet or crouching, with back bowed and head lowered between the shoulders, are characteristic of incubating birds. Finally, especially during hot weather, birds can be observed lying on the ground. This last position is not an efﬁcient one for receipt of acoustic information: we have shown in king penguin colonies (Lengagne et al., 1999b) that, when the receiver was located 10 cm above the ground, degradation of the signal was much more pronounced in both frequency and amplitude domains than when the receiver was located either 45 or 90 cm above ground. The characteristic listening postures (with head either 90 or 45 cm up) enhance the probability that the receiver will hear the call. Birds lying on the ground are generally found to be sleeping, so acoustic communication would seem to be of no importance to them. Conversely, when the incubating bird hears the call of its mate for the ﬁrst time, it leaves its incubating posture, rises to its feet, and assumes what we have shown to be the best position for signal reception. 3. The Searching Strategy Our signal propagation tests in colonies indicated that the communication system involving the mutual recognition between either mates or parents and chicks could be established in penguin species only at short range, as predicted by Falls (1982) for seabird colonies. Our playback tests with banded chicks or adults conﬁrms Falls’ hypothesis. To estimate the maximum distance at which a call can be recognized by the partner, natural calls were broadcast to the corresponding bird at different distances in the colony (Fig. 5). The experiment started at a distance from the bird to the loudspeaker of 8 to 20 m, depending on the species studied. The call was played back and the behavior of the bird was observed. Then, the distance was reduced by moving the loudspeaker closer. After a pause of 6 min, the call was played back again, until a detection process was observed. Birds detect, recognize, and localize the natural display call at a relatively short distance: an average distance of 11 m for the king penguin (number of birds tested: N = 12; Aubin and Jouventin, 1998) and 4 m for the little blue penguin (N = 7, Jouventin and Aubin, 2000).

258

THIERRY AUBIN AND PIERRE JOUVENTIN

FIG. 5. Maximal distance discrimination of the parental display call by the chick in king penguins. During experiments (12 chicks tested), the chick-to-loudspeaker distance, 20 m at the beginning, was progressively reduced by approaching the loudspeaker in steps of 1 m (data from Aubin and Jouventin, 1998, and Lengagne, 1999).

When a bird comes from the sea and makes its way into the colony to ﬁnd its partner, it calls regularly at different distances from the receiving bird. The farther from the receiving bird the acoustic search was initiated, the more time was necessary to complete the search and the greater the number of calls that were emitted by the incoming bird. On the other hand, we have shown in the king penguin that 70% of the birds started the acoustic search for their mate when the distance was less than or equal to the discrimination range (Lengagne et al., 1999b). So, the calling strategy adopted for ﬁnding the partner appears particularly efﬁcient in penguins. B. SPECIES RECOGNITION Although we have not systematically studied species recognition in penguins, but have considered them rather a unique model for individual recognition, Jouventin (1982) did compare territorial behavior between species. We found that the display call has several biological meanings, being used by a single bird or by a pair throughout the breeding cycle to indicate both the species, the sex, and the individual. The nesting penguins are the more territorial species, particularly the burrowing genera (Spheniscus and Eudyptula). The latter is nocturnal and is strongly aggressive against intruders even if it cannot see them (Waas, 1988, 1991). Consequently this species, the little blue penguin, constitutes the best model for speciﬁc recognition in penguins. We compared it with a burrowing petrel (Jouventin and

VOCALLY IDENTIFYING KIN IN A CROWD: THE PENGUIN MODEL

259

Aubin, 2000), and tested their responses to playback experiments. The response is not tuned to a precise frequency analysis since entire calls that have been shifted strongly (200 Hz) up or down still elicit territorial responses. In fact, it appears that the little blue penguin pays attention to the lower sounds, the presence of high frequencies being unnecessary to elicit territorial responses. It appears also that both exhaled and inhaled temporal syllables provide territorial information. With respect to the territorial function of the calls, the coding process appears very simple: a territorial call consists of a rhythmic succession of two sounds having a particular pattern of frequency modulation. The parameters which encode territorial information are those that are resistant to degradation (low frequencies, slow FM, gaps in frequency and amplitude). To communicate the territorial message, the information represented by the patterned arrangement of the two “binary” units of sounds is highly redundant (Brackenbury, 1978), as it is based on the repetition of identical units of information at two levels. Nevertheless, it would be enough for a territorial function where the breeder has only to know that a conspeciﬁc is approaching and merely has to reply: “Keep out, this burrow is occupied!”

C. A WELL-MATCHED ACOUSTIC CODE FOR IDENTIFICATION 1. The Cocktail-Party Effect As mentioned earlier, the display call is transmitted in a context involving the background noise generated by the colony plus the screening effect of the birds’ bodies, both reducing the signal-to-noise ratio. In addition, the signal is masked by background noise with similar spectral and temporal characteristics. To estimate the minimal discrimination threshold of the display call in a jamming situation, a series of mixed signals was broadcast to penguin chicks (Fig. 6). The parental call was combined with ﬁve extraneous adult calls with different emergence levels, the tested signal increasing in energy ratio among the extraneous noise. The superimposition results in a mixed signal with a total lack of silences and with numerous frequencies overlapping. This jamming mimics a situation frequently observed in a penguin colony. Chicks of three species were tested with signals of different emergence levels (i.e., the difference between the energy level of the tested call and that of the ﬁve extraneous calls) at a distance of 7 m. The emergence level was deﬁned as E = 20 log Ap /Ae , where E represents the emergence level in dB of the parental call of the chick tested, Ap is the absolute amplitude of the parental call, and Ae is the absolute amplitude of the mixed extraneous

260

THIERRY AUBIN AND PIERRE JOUVENTIN

FIG. 6. Test of detection of the parental call by the chick in three penguin species in a jamming situation. The parental display call is mixed with ﬁve extraneous display calls to yield different emergence levels.

calls. Our experiments indicate that the chick can detect its parental call in an extreme jamming situation. With king penguin chicks, detection is possible even when the parental call intensity is well below (−6 dB) that of the noise of simultaneous calls produced by other adults (Aubin and Jouventin, 1998). At the same distance, the Adelie ´ and emperor penguins are not as good at detecting calls as the king penguin chick. Nevertheless, Adelie ´ and emperor penguin chicks have a good ability to recognize the parental call even embedded in the noise of the colony (0 dB of emergence, Jouventin and Aubin, 2001). This capacity to perceive and extract the information from an ambient noise with similar acoustic properties to that of the signal, termed the “cocktail-party effect” in speech intelligibility tests, enhances the chick’s ability to ﬁnd its parents. This process of perception must be linked to an acoustic coding system adapted to the constraints of colonial life. 2. The Vocal Signature By playing back natural calls, it has been demonstrated that individual recognition by voice exists in all the species of penguins that have been studied (Derenne et al., 1979; Jouventin et al., 1979; Jouventin, 1982; Waas, 1988; Speirs and Davis, 1991). Previous analysis of temporal and frequency parameters have shown effectively that an individual signature can also be found in the display calls of all of these species (Jouventin and Roux, 1979; Robisson et al., 1989; Bremond ´ et al., 1990; Robisson, 1992a; Robisson et al., 1993; Lengagne et al., 1997; Jouventin and Aubin, 2000). This has mainly been done by comparing, for a given parameter, the betweenindividual variation and the within-individual variation (Jouventin, 1982;

261

VOCALLY IDENTIFYING KIN IN A CROWD: THE PENGUIN MODEL

Jouventin and Aubin, 2000). By comparing the temporal and frequency patterns in the calls of two species of penguins with nests (macaroni and Adelie) ´ with those of two species without a nests (king and emperor), a direct relationship was shown between the potential of individual coding and the difﬁculty of ﬁnding the partner in the colony (Lengagne et al., 1997). However, knowing that it is possible to distinguish the signals of individuals by analysis does not tell us whether the birds do it. The only way to investigate this process was to test the birds by playing back different kinds of experimental signals. These signals corresponded to natural display calls modiﬁed in different ways: modiﬁcations of amplitude and frequency modulations (AM and FM), modiﬁcations of frequency and temporal parameters, and modiﬁcation of the two-voice system (Table II). TABLE II RESPONSES OF PENGUINS TO EXPERIMENTAL SIGNALS CORRESPONDING TO NATURAL CALLS MODIFIED Species Signals

EP

KP

MP

AP

GP

Modulation Without AM Without FM Reversed

— + +

++ — —

— + ++

— + ++

— + ++

Frequency domain Fo Low Pass High Pass Shift ± 25 Shift ± 50 Shift ± 75 Shift ± 100

nt ++ + nt ++ ++ —

+ ++ — nt ++ + —

nt ++ ++ ++ ++ ++ +

— ++ — + — — nt

— — — ++ — — nt

Temporal domain HalfCall OneSyl HalfSyl QuarterSyl

+ — nt nt

++ ++ ++ —

++ + — nt

++ ++ — nt

nt ++ — nt

Two-voices phenomenon One voice suppressed

—

—

nt

nt

nt

Note: ++, strong response; +, moderate response; —, no response; nt, not tested. EP, KP, MP, AP, GP: respectively, emperor, king, macaroni, Adelie, ´ and gentoo penguins. (Results from: Jouventin, 1982, Aubin et al., 2000, Hildebrand, unpublished data, for the emperor penguin; Jouventin et al., 1999, Lengagne et al., 2000, for the king penguin; Jouventin and Aubin, 2001, for the Adelie ´ and gentoo penguins; Aubin and Jouventin, unpublished data, for the macaroni penguin.)

262

THIERRY AUBIN AND PIERRE JOUVENTIN

a. AM and FM. Concerning the modulations, only the king penguin among the different species studied, continues to respond after the total elimination of amplitude modulation (AM) in its natural display call. This does not imply that the AM structure is without use in this species. AM is strongly degraded during propagation through the colony (Aubin and Mathevon, 1995, Aubin and Jouventin, 1998; Lengagne et al., 1999b) and this is unlikely to carry a message at distance. Nevertheless, we cannot rule out the possibility that AM might transmit information which would allow estimating the distance of the emitter (Naguib, 1996) or locating the acoustic source (Konishi, 1973; Wiley and Richard, 1982), two helpful qualities for locating an individual in a crowd. Concerning frequency modulation (FM), all the species studied except the king penguin are able to identify the mate or parental call even if the natural FM is lacking. It is the same for signals where each syllable (and at the same time the FM) is reversed (Fig. 7). Thus FM appears to be a key parameter in individual recognition for the king penguin but not for the other species, where amplitude modulation is important. b. The Frequency Domain. In the frequency domain, all of the species studied prefer the lower part of the sound spectrum to the higher one, probably due to the fact that high frequencies attenuate strongly during propagation in the colony (Aubin and Jouventin, 1998; Lengagne et al., 1999b). Nevertheless, some species (Adelie ´ and gentoo) pay greater attention to the spectral proﬁle of the call than others (emperor, king, macaroni). For example, recognition occurs for low-pass calls in emperor, king, and macaroni penguins but not for Adelie ´ and gentoo penguins. Pitch parameters reveal the same pattern of responses. Emperor, king, and macaroni penguins tolerate more change in frequency than do Adelie ´ and gentoo penguins. To identify its partner or chick, the latter species need the right frequency values for the harmonics of the call (with an accuracy of 25 Hz), whereas the former species tolerate errors of 75–100 Hz (Fig. 8). Thus, for Adelie ´ and gentoo penguins, the coding of identity depends on the analysis of the spectral proﬁle and of the precise frequency values of the harmonics, that is, mainly on timbre analysis. c. The Temporal Domain. In the great majority of species studied, the broadcast of only one syllable is sufﬁcient to elicit recognition, and this recognition is not linked to any particular syllable in the call (Jouventin and Aubin, 2001; Jouventin et al., 1999; Lengagne et al., 2000, and unpublished data). For the king penguin, even the ﬁrst half of a syllable is sufﬁcient to elicit recognition. In this species, the basic FM structure of the syllable is always the same: an increase followed by a decrease in frequency, with the inﬂection point always in the ﬁrst half of the syllable. This inﬂection point is necessary for the recognition of the signal. The small amount of information (about 200 ms) provided by the frequency modulation shape of this half of

VOCALLY IDENTIFYING KIN IN A CROWD: THE PENGUIN MODEL

263

FIG. 7. Natural and corresponding “syllable reversed” calls for two species of penguins that do not have a nest. Only the king penguin signal with syllable reversed does not elicit positive responses during individual recognition tests.

the syllable is sufﬁcient to provide individual information (Jouventin et al., 1999; Lengagne et al., 2000). At the opposite extreme, the display call of the emperor penguin is the only one where several successive syllables (at least three; Hildebrand, personal communication) are needed to elicit recognition. Effectively, to identify the signal, emperor penguins examine the temporal succession of syllables (Jouventin, 1972, 1982; Jouventin et al., 1979). Display calls of penguins (emperor penguin calls included) appear to be highly redundant, consisting of more or less identical successive syllables with a repetition of the same information many times. This redundancy

FIG. 8. Responses of king and Adelie ´ penguin chicks to parental display calls shifted in frequency (results from Jouventin et al., 1999; Jouventin and Aubin, 2001).

VOCALLY IDENTIFYING KIN IN A CROWD: THE PENGUIN MODEL

265

FIG. 9. Enhancement of the number of syllables per call emitted by king penguins as wind speed increases: in windy conditions, birds maintain the efﬁciency of their communication by increasing the redundancy of the signal. The circle size is proportional to the number of data (ranging from 3 to 6) and the dashed line represents least-squares piecewise linear models (data replotted from Lengagne et al., 1999b).

enhances the opportunity to ﬁnd a quieter window in the continuous noisy environment of a seabird colony. In addition, in penguin colonies circumpolar winds blow strongly, generating a high level of background noise and consequently lowering the signal-to-noise ratio in the colony. In windy conditions, we have observed that birds in king penguin colonies (Lengagne et al., 1999c) maintain the efﬁciency of their communication by increasing the number of syllables per call, leading to an enhancement of the duration of the calls. From a wind speed of 8 m/s, the duration of the calls increases linearly as the wind speed increases (Fig. 9). Thus, penguins seem to take into account the constraints imposed by wind on their communication. This result conforms with predictions from information theory (Shannon and Weaver, 1949) that increased redundancy in signal improves the probability of receiving a message in a noisy channel. d. The Two-Voice System. The sound-producing structure in birds is the syrinx, usually a two-part organ located at the junction of the bronchi. As each branch of the syrinx produces sound independently, many birds have two acoustic sources. The use of the two voices was ﬁrst documented in the song of the brown thrasher, Toxostoma rufum (Potter et al., 1947), and has been actively investigated for the past ﬁfty years. Anatomical, physiological, and acoustical evidence existed for this two-voice phenomenon

266

THIERRY AUBIN AND PIERRE JOUVENTIN

(Greenewalt, 1968) but no function for it was known (Sturdy and Mooney, 2000). In songbirds, these two voices with their respective harmonics are often not activated simultaneously but two voices are obvious in the large penguins and generate a beat pattern that varies between individuals. Among the 17 species of penguins, only the Aptenodytes genus employs two frequency bands (Robisson, 1992b). Both Aptenodytes species, the emperor and the king penguins, produce a signal consisting of two simultaneous series of harmonically related bands of slightly differing frequencies (on average 65 Hz for the emperor, Aubin et al., 2000, and 25 Hz for the king, Robisson, 1992b), resulting in a two-voice call that produces audible “beats.” The two-voice system appears well suited for the environment in which it is used. We have done experiments with two voice signals broadcast through the colony and recorded at different distances (1, 8, 16 m). The recorded signals were then analyzed and the amount of degradation between the amplitude modulation of the beats and the true modulation existing in the call itself were compared (Fig. 10). We found that, although the true amplitude

FIG. 10. King penguin display call: analysis of beats generated by the “two-voice system.” Only the fundamentals of one syllable are analyzed here. The periodic amplitude ﬂuctuations [(a) oscillographic representation] coincide with the periodic frequency discontinuities [(b) spectographic + instantaneous frequency representation] and allow quantiﬁcation of the period of the beats generated by the two sources.

VOCALLY IDENTIFYING KIN IN A CROWD: THE PENGUIN MODEL

267

modulation of the call was severely compromised by propagation, the amplitude modulation of the beats produced by the two voices remained largely unchanged. Our experiments demonstrate that the beats generated by the interaction of these two frequency bands propagate well through obstacles, being robust to sound degradation through the medium of bodies in a penguin colony, but above all that they convey information about individual identity. To test the hypothesis that the two voices may play a key role in individual recognition in the Aptenodytes genus we designed a series of playback experiments

FIG. 11. Sound spectrograms (1024 points window size) of syllables of emperor (left) and king (right) penguins. (a) and (b): low-pass ﬁltered natural syllable, with only the fundamentals and the ﬁrst harmonics kept; (c): one voice removed by ﬁltration in the emperor penguin signal; (d): synthetic signal built on the model of the low-pass ﬁltered king penguin signal, but with only one voice. The low-pass signals with two voices are recognized but not the one-voice signals.

268

THIERRY AUBIN AND PIERRE JOUVENTIN

FIG. 12. Signal adaptation to recognition in the noise (>70 dB) in two categories of penguins: without nest (e.g., king penguin) and with nest (e.g., Adelie ´ penguin) species.

with signals where only one voice was kept (Aubin et al., 2000, for the emperor penguin, Lengagne et al., 2001, for the king penguin). Unfortunately, the upper frequency bands of natural calls were not spaced sufﬁciently to allow for removal, by simple ﬁltration, of one of the two voices. Therefore, a preliminary playback test was performed with the two Aptenodytes species using only the lower frequency component of the calls as stimulus (Fig. 11).

VOCALLY IDENTIFYING KIN IN A CROWD: THE PENGUIN MODEL

269

When presented with the lower frequency component of the calls, positive responses were induced in chicks and adults for both species. When one voice in the call is experimentally suppressed, either by ﬁltration (Aubin et al., 2000, for the emperor penguin) or by synthesizing a one-voice signal (Lengagne et al., 2001, for the king penguin), no response is observed for adults or chicks. Clearly, the acoustic contribution from two voices is required for call recognition in the genus Aptenodytes. This coding process, increasing the call complexity and resisting sound degradation, appears to have evolved in parallel with the loss of territoriality. e. Acoustic Solutions to Ecological Problems. If we summarize the results obtained in our playback experiments with modiﬁed display calls, two acoustic code categories emerge in the penguin family (Fig. 12). The ﬁrst one is elementary and concerns the nesting penguins, such as the Adelie ´ and gentoo. These species identify the partner by analyzing the spectral proﬁle and pitch of the call (timbre analysis). The second one is complex and concerns the two nonnesting penguins, the emperor and the king. Both these species use a vocal signature in the time domain based on an amplitude– time (AM) analysis for the emperor penguin and on a frequency–time (FM) analysis for the king penguin. This temporal analysis is complemented by another sophisticated system: the beats generated by the two-voice system. The temporal pattern of syllables associated with the two-voice system creates a huge variety of vocal signatures. This is necessary to distinguish between several thousand birds breeding without nests, that is, without visual cues. At the opposite extreme, identiﬁcation by a one-dimensional parameter such as the timbre does not offer such an impressive variety of vocal signatures, and, thus, the possibility of confusion should exist with this system. Nevertheless, in nesting penguins, the nest is used as a meeting place, even when chicks have ﬂedged, so that the probability that a bird emits the right call at the wrong place is weak. This strongly limits the possibilities for confusion.

V. PERSPECTIVES A. FUNCTIONALS The main perspective from these results is functional because these vocal signatures represent external markers of identity allowing the penguins to ﬁnd their chicks or their parents in a crowd, that is, to ﬁnd the kin-related birds with a single sound system when olfactory and, more surprisingly, visual cues are not used. During the brooding phase, the parent coming back from the sea has ﬁrst to ﬁnd its mate by its vocal signature if it is to ﬁnd the egg

270

THIERRY AUBIN AND PIERRE JOUVENTIN

or the small chick. Then the detection of the parent by the chick has an obvious survival value because usually parents feed only their own chick(s). The identiﬁcation of the chick by the parent allows it to invest in its own chick and consequently to propagate its genes. But the general pattern is not symmetrical because, although a chick may gain from getting extra food from birds other than its parents, parents have to be cautious to give food only to their own chick(s). Consequently we have observed chicks trying to obtain “extra-feeding” from adults calling differently from their parents, sometimes with success when confusion occurs, for example, after a storm or when parents cannot ﬁnd their own young (see Jouventin et al., 1995). We have also observed parents chasing away these “robbers” after hearing their call accurately. B. ACOUSTICS According to the theory, to extract a signal from the background noise penguins analyze either frequency bands or temporal patterns (AM, FM, beats) of the call. The ﬁrst coding–decoding system is used by nesting penguins such as the Adelie, ´ the gentoo, and, in the ﬁrst part of its call, the macaroni, and the second one is used by nonnesting penguins. The two codes seem not to be equivalent in efﬁciency. Frequency analysis is known to be particularly slow in a physical sense (Pimonow, 1962; Beecher, 1988) as well as physiologically (Bregman, 1978): when the duration of the analysis decreases, the uncertainty in the measurement of frequency increases. Accurate analysis of frequency is more time consuming than analysis in the time domain. The two codes seem also not to be equivalent in terms of production. Modulation in time is difﬁcult to produce (Gaunt et al., 1973; Brackenbury, 1982): frequency or amplitude modulated calls require a high degree of control of the two sound sources of a bird. Nevertheless, this acoustic signal is particularly efﬁcient in allowing an animal to locate the partner immediately in a noisy crowd on the move. Brieﬂy, frequency analysis (easy to produce but costly in terms of analysis time) is enough to solve the relatively simple problem of individual recognition in nesting birds, whereas the complex temporal analysis of modulations used by the two nonnesting penguins (quick to analyze but costly to produce) appears as an adaptation to extreme acoustic and breeding conditions. C. COMPARATIVE APPROACH The case of the macaroni penguin, a nesting penguin, is perhaps intermediate between the two systems just described. Effectively, according to our ﬁrst playback experiments, this species seems to identify the partner

VOCALLY IDENTIFYING KIN IN A CROWD: THE PENGUIN MODEL

271

on the basis of a simultaneous analysis of the spectral proﬁle and of the temporal succession of syllables. However, this last parameter, again for a nesting species, seems simple involving only a rhythm ﬁxed (= a number), whereas the emperor penguin has a more complex system using a succession of syllable-silence timings (= a “bar code”). Complementary experiments will be necessary to fully understand the coding–decoding process of the macaroni penguin and to compare with other species in this newly studied genus (Eudyptes).

VI. CONCLUSION Highly vocal and colonial, penguins communicate in a particularly constraining acoustic environment to ﬁnd their mate, parent, or chick. The screening effect of penguin bodies and the loud background noise prevent long-range communication and explain why some sophisticated adaptations have appeared, such as the highly developed “cocktail party effect” in the king penguin or the “courtesy rule” preventing jamming in the emperor penguin. These constraints explain the existence of complementary and visual behaviors optimizing vocal identiﬁcation, such as the “head up” position when hearing or calling, particularly in the king penguin, and the memory of a feeding place in nesting penguins and in king penguins. They explain also the physical properties of the calls. For species recognition, it seems that the penguins are not more sophisticated than other birds (see Becker, 1982, for a review). Thus, our playback experiments with the little blue penguin show that the acoustic code used for species identiﬁcation is elementary. But, for individual recognition, it is surprising to ﬁnd animals that are able to see nevertheless identifying their parents only by voice, in a noisy crowd, and even for some species without the help of a nest as a landmark. In the same family, natural selection has resulted in various acoustic adaptations which are all the more sophisticated given that the problem posed was difﬁcult to solve. In the nesting penguins where visual cues help vocal recognition, we found two main identiﬁcation systems based on a single parameter, frequency, in the two species of Pygoscelis and on two elementary parameters, temporal and frequency patterns, in one species of Eudyptes. The vocal solutions found in closely related species are really different and demonstrate once more that behavior can evolve fast. It will be interesting to know what systems occur in other species of nesting penguins, not yet studied, to see if the number of acoustic identiﬁcation systems is even larger. In the nonnesting penguins, once again two closely related species have two different vocal identiﬁcation systems. This dual system is much more

272

THIERRY AUBIN AND PIERRE JOUVENTIN

complex than in nesting species using topographical cues. The ﬁrst identiﬁcation system is based mainly on the variation of frequency in time for the king penguin and on the temporal syllable pattern in the emperor penguin. The second system relies on the two voices. It is common to the genus Aptenodytes, and this double-voice system is completely new because, although we have known for 30 years of the existence of this anatomical, physiological, and physical phenomenon, no function for it was known. In fact, these new ﬁndings on a single family, although exceptional by the variety of its breeding habits, are all the more interesting since we have poor knowledge of the systems of individual identiﬁcation in birds, particularly when they are not songbirds. Individual recognition playbacks are particularly difﬁcult to carry out because to test a bird we have to present a particular call to each, whereas we need only one call type in the study of species recognition. It remains to be seen whether we can extrapolate our results to other penguins and then more widely among seabirds. We can certainly learn from this penguin study that nonsongbirds can be sophisticated in their vocal adaptations and deserve more attention, being currently relatively little studied. Enlarging upon the scope of the acoustic systems, it is also possible to study individual recognition beyond birds, for example, in mammals, where identiﬁcation strategies by voice have been found in species such as fur seals (Roux and Jouventin, 1987; Charrier et al., manuscript in preparation). Further studies may reveal interesting parallels in the acoustic identiﬁcation systems between birds and mammals.

VII. SUMMARY In penguins, individual recognition is observed between mates and between parents and chick(s). During the past ﬁve years, their particular strategies of coding–decoding have been tested by playing back modiﬁed display calls to six species, in Australia (little penguin, Eudyptula minor), in Antarctica (Adelie ´ penguin, Pygoscelis adeliae; emperor penguin, Aptenodytes forsteri), and in subantarctic islands (king penguin, Aptenodytes patagonicus; macaroni penguin, Eudyptes chrysolophus; gentoo penguin, Pygoscelis papua). All species use only vocal cues to identify their partner, but in territorial species the nest is used as a meeting point. In large species, such as the king and the emperor penguins, which do not have a nest, the brooder carries the egg or the small chick on the feet, while the mate, and then the chick, has to be located in the noisy colony without any topographical cue. According to theory, to extract a signal from background calls, animals analyze either frequency bands or modulations (amplitude and frequency

VOCALLY IDENTIFYING KIN IN A CROWD: THE PENGUIN MODEL

273

modulations) of the partner’s call. The ﬁrst coding–decoding system, used by nesting penguins, is easy to produce but costly in terms of analysis time. The second one, used by nonnesting penguins, is a vocal signature which is fast to analyze but costly to produce. This acoustic signal is particularly efﬁcient as a means to locate immediately the partner on the move in a noisy crowd. Brieﬂy, frequency analysis is enough to solve the relatively easy problem of individual recognition in nesting birds, while the complex temporal analysis of modulations of the two nonnesting penguins is an adaptation to extreme acoustic and breeding conditions. The macaroni penguin, which we have begun to test, seems to use both a frequency code similar to that of the other nesting species and a temporal code close to the one of a nonnesting penguin species, but much simpler.

Acknowledgments This study was funded in antarctic and subantarctic areas by the Institut Fran¸cais pour la Recherche et la Technologie Polaires (Program N◦ 109 then N◦ 354). For the study of the little penguin in Australia, we thank the Committee of Management of the Penguin Reserve and particularly Peter Dann. We thank also Stephen Dobson, Tim Roper, Peter Slater, and Charles T. Snowdon for helpful comments and English improvements.

References Alcock, J. (1972). “Animal Behavior, an Evolutionary Approach.” Sinauer, Sunderland, MA. Aubin, T. (1994). Syntana: A software for the synthesis and analysis of animal sounds. Bioacoustics 6, 80–81. Aubin, T., and Jouventin, P. (1998). Cocktail-party effect in king penguin colonies. Proc. R. Soc. London B 265, 1665–1673. Aubin, T., and Lengagne, T. (1997). Reconnaissance du cri du parent par le poussin du manchot royal Aptenodytes patagonicus dans le milieu bruyant de la colonie. Bull. Soc. Zool. Fr. 123, 267–277. Aubin, T., and Mathevon, N. (1995). Adaptation to severe conditions of propagation: Long-distance distress calls and courtship calls of a colonial seabird. Bioacoustics 6, 153– 161. Aubin, T., Jouventin, P., and Hildebrand, C. (2000). Penguins use the two-voice system to recognise each other. Proc. R. Soc. London B 267, 1081–1087. Barrat, A. (1976). Quelques aspects de la biologie et de l’ecologie ´ du manchot royal Aptenodytes patagonicus des ˆıles Crozet. Com. Nat. Fr. Rech. Ant. 40, 107–147. Becker, P. H. (1982). The coding of species-speciﬁc characteristics in bird sounds. In “Acoustic Communication in Birds” (D. E. Kroodsma and E. H. Miller, Eds.), Vol. I, pp. 131–181. Academic Press, New York. Beecher, M. D. (1988). Spectrographic analysis of animal vocalizations. Implications of the “uncertainty principle.” Bioacoustics 1, 187–208.

274

THIERRY AUBIN AND PIERRE JOUVENTIN

Beer, C. G. (1970). Individual recognition of voice in the social behaviour of birds. Adv. Study Behav. 3, 27–74. Beer, C. G. (1979). Vocal communication between laughing gull parents and chicks. Behaviour 70, 118–146. Brackenbury, J. H. (1978). A comparison of the origin and temporal arrangement of pulsed sounds in the songs of the Grasshopper and Sedge warblers, Locustella naevia and Acrocephalus schoenobaenus. J. Zool. London 184, 187–206. Brackenbury, J. H. (1982). The structural basis of voice production and its relationship to sound characteristics. In “Acoustic Communication in Birds” (D. E. Kroodsma and E. H. Miller, Eds.), Vol. 1, pp. 53–73. Academic Press, New York. Bregman, A. S. (1978). The formation of auditory streams. In “Attention and Performance” (J. Reguin, Ed.). Erlbaum, Hillsdale, NJ. Bremond, ´ J. C., and Aubin, T. (1992). The role of amplitude modulation in distress call recognition by the black-headed gull (Larus argentatus). Ethol. Ecol. Evol. 4, 187–191. Bremond, ´ J. C., Aubin, T., Nyamsi, R. M., and Robisson, P. (1990). Le chant du manchot empereur (Aptenodytes forsteri): Recherche des parametres ` utilisables pour la reconnaissance individuelle. C. R. Acad. Sci. 311, 31–35. Bried, J., and Jouventin, P. (2001). Site and mate-choice in seabirds: An evolutionary approach to the nesting ecology. In “Biology of Marine Birds” (J. Schreiber and R. Burger, Eds.) CRC Press, Boca Raton, FL. In press. Brooke, M. L. (1978). Sexual differences in the voice and individual vocal recognition in the manx shearwater (Puffinus puffinus). Anim. Behav. 26, 622–629. Charrier, I., Jouventin, P., Mathevon, N., and Aubin, T. (2001). Acoustic communication in a black-headed gull colony: How to identify their parents? Ethology In press. Dabelsteen, T. (1981). The sound pressure level in the dawn song of the blackbird Turdus merula and a method for adjusting the level in experimental song to the level in natural song. Z. Tierpsychol. 56, 137–149. Dabelsteen, T., Larsen, O. N., and Pedersen, S. B. (1993). Habitat-induced degradation of sounds signals: Quantifying the effects of communication sounds and bird location on blur ratio, excess attenuation, and signal-to-noise ratio in blackbird song. J. Acoust. Soc. Am. 93, 2206–2220. Derenne, P., Jouventin, P., and Mougin, J. L. (1979). Le chant du manchot royal (Aptenodytes patagonicus) et sa signiﬁcation evolutive. ´ Le Gerfaut 69, 211–224. Falls, J. B. (1982). Individual recognition by sound in birds. In “Acoustic Communication in Birds” (D. E. Kroodsma and E. H. Miller, Eds.), Vol. II, pp. 237–278. Academic Press, New York. Gaunt, A. S., Stein, R. C., and Gaunt, S. L. (1973). Pressure and air ﬂow during distress calls of the starling, Sturnus vulgaris (Aves: Passeriformes). J. Exp. Zool. 183, 241–262. Greenewalt, C. H. (1968). “Bird Song: Acoustics and Physiology.” Random House (Smithsonian Inst. Press), New York. Isenmann, P., and Jouventin, P. (1970). Eco-ethologie ´ du Manchot Empereur (Aptenodytes forsteri) et comparaison avec le Manchot Adelie ´ (Pygoscelis adeliae) et le Manchot Royal (Aptenodytes patagonica). L’Oiseau R.F.O. 40, 136–159. Jouventin, P. (1972). Un nouveau systeme ` de reconnaissance acoustique chez les oiseaux. Behaviour 43, 176 –186. Jouventin, P. (1982). “Visual and Vocal Signals in Penguins, Their Evolution and Adaptive Characters” (P. Parey, Ed.). Vol. 24, of the Advances in Ethology series. Jouventin, P., and Aubin, T. (2000). Acoustic convergence between two nocturnal burrowing seabirds: Experiments with a penguin (Eudyptula minor) and a shearwater (Puffinus tenuirostris). Ibis 142, 645–656.

VOCALLY IDENTIFYING KIN IN A CROWD: THE PENGUIN MODEL

275

Jouventin, P., and Aubin, T. (2001). Experimental study of the kin-recognition in two penguins (Pygoscelis adeliae and papua). Acoustic systems in relation with nesting ecologies. J. Avian Biol. Submitted. Jouventin, P., and Cornet, A. (1980). The sociobiology of Pinnipeds. Adv. Study Behav. 11, 121–141. Jouventin, P., and Robin, J. P. (1984). Olfaction experiments on some antarctic birds. Emu 84, 46–48. Jouventin, P., and Roux, P. (1979). Le chant du Manchot Adelie ´ (Pygoscelis adeliae). Role ˆ dans la reconnaissance individuelle et comparaison avec le Manchot Empereur non-territorial. L’Oiseau R.F.O. 49, 31– 43. Jouventin, P., and Weimerskirch, H. (1990). Satellite tracking of wandering albatrosses. Nature 343, 746 –748. Jouventin, P., Guillotin, M., and Cornet, A. (1979). Le chant du Manchot Empereur et sa signiﬁcation adaptative. Behaviour 70, 231–250. Jouventin, P., Barbraud, C., and Rubin, M. (1995). Adoption in the emperor penguin Aptenodytes forsteri. Anim. Behav. 50, 1023 –1029. Jouventin, P., Aubin, T., and Lengagne, T. (1999). Finding a parent in a king penguin colony: The acoustic system of individual recognition. Anim. Behav. 57, 1175 –1183. Kodyman, G. L., and Mullins, J. L. (1990). Ross Sea emperor penguin breeding populations estimated by aerial photography. In “Antarctic Ecosystems: Ecological Change and Conservation” (K. R. Kerry and G. Hempel, Eds.), pp. 169 –176. Springer-Verlag, Berlin. Konishi, M. (1973). Locatable and nonlocatable acoustic signals for Barn Owls. Am. Nat. 107, 775–785. Lack, D. (1968). “Ecological Adaptations for Breeding in Birds.” Methuen, London. Lengagne, T. (1999). Communication acoustique en milieu colonial: Le reconnaissance individuelle chez le manchot royal Aptenodytes Patagonicus. Ph.D. thesis, Universite´ de Toulouse. Lengagne, T., Lauga, J., and Jouventin, P. (1997). A method of independent time and frequency decomposition of bioacoustic signals: Inter-individual recognition in four species of penguins. C. R. Acad. Sci. Paris, Sciences Vie 320, 885–891. Lengagne, T., Jouventin, P., and Aubin, T. (1999a). Finding one’s mate in a king penguin colony: Efﬁciency of acoustic communication. Behaviour 136, 533–546. Lengagne, T., Aubin, T., Jouventin, P., and Lauga, J. (1999b). Acoustic communication in a king penguin’s colony: Importance of bird location within the colony and of the body position of the listener. Polar Biol. 21, 262–268. Lengagne, T., Aubin, T., Lauga, J., and Jouventin, P. (1999c). How do king penguins Aptenodytes patagonicus apply the Mathematical Theory of Information to communicate in windy conditions? Proc. R. Soc. London Ser. B 266, 1623–1628. Lengagne, T., Aubin, T., Jouventin, P., and Lauga, J. (2000). Perceptual salience of individually distinctive features in the calls of adult king penguins. J. Acoust. Soc. Am. 107, 508–516. Lengagne, T., Lauga, J., and Aubin, T. (2001). Intra-syllabic acoustic signatures used by the king penguin in parent-chick recognition: An experimental approach. J. Exp. Biol. 204, 663–672. Lequette, B., Verheyden, C., and Jouventin, P. (1989). Olfaction in subantarctic seabirds: Its phylogenetics and ecological signiﬁcance. Condor 91, 732–735. Mbu-Nyamsi, R. G., Aubin, T., and Bremond, ´ J. C. (1994). On the extraction of some time dependent parameters of an acoustic signal by means of the analytic signal concept. Its application to animal sound study. Bioacoustics 5, 187–203. Naguib, M. (1996). Ranging by song in Carolina wrens: Effects of environmental acoustics and strength of song degradation. Behaviour 133, 541–559.

276

THIERRY AUBIN AND PIERRE JOUVENTIN

Noll, A. M. (1967). Cepstrum pitch determination. J. Acoust. Soc. Am. 41, 293–309. Papoulis, A. (1977). “Signal analysis.” McGraw-Hill, New York. Pimonow, L. (1962). “Vibrations en regime ´ transitoire: Analyse physique et physiologique.” Dunod, Paris. Potter, R. K., Kopp, G. A., and Green, H. C. (1947). “Visible Speech.” Van Nostrand, New York. Press, W. H., Flannery, B. P., Teukolsky, S. A., and Vetterling, W. T. (1988). “Numerical Recipes in C. The Art of Scientiﬁc Computing.” Cambridge Univ. Press, Cambridge, UK. Prevost, ´ J. (1961). “Ecologie du Manchot Empereur.” Hermann, Paris. Prevost, ´ J., and Bourliere, ` F. (1957). Vie sociale et thermoregulation ´ chez le manchot empereur. Alauda 25, 167–173. Randall, R. B., and Tech, B. (1987). “Frequency Analysis.” Bruel & Kjaer, Naerum. Robisson, P. (1991). Broadcast distance of the mutual display call in the emperor penguin. Behaviour 119, 302–316. Robisson, P. (1992a). Roles of pitch and duration in the discrimination of the mate’s call in the king penguin Aptenodytes patagonicus. Bioacoustics 4, 25–36. Robisson, P. (1992b). Vocalizations in Aptenodytes penguins: Application of the two-voice theory. Auk 109, 654–658. Robisson, P. (1993a). La reconnaissance individuelle chez deux especes ` jumelles, le manchot empereur Aptenodytes forsteri et le manchot royal Aptenodytes patagonicus. Ph.D. thesis, Universite´ de Rennes. Robisson, P. (1993b). Adaptation du transfert de l’information individuelle en milieu colonial chez les manchots. Terre Vie 48, 133–141. Robisson, P., Aubin, T., and Bremond, ´ J. C. (1989). La reconnaissance individuelle chez le manchot empereur (Aptenodytes forsteri): Roles ˆ respectifs du decoupage ´ temporel et de la structure syllabique du chant de cour. C. R. Acad. Sci. 309, 383–388. Robisson, P., Aubin, T., and Bremond, ´ J. C. (1993). Individuality in the voice of the emperor penguin Aptenodytes forsteri: Adaptation to a noisy environment. Ethology 94, 279– 290. Roux, J. P., and Jouventin, P. (1987). Behavioral cues to individual recognition in the subantarctic fur seal, Arctocephalus tropicalis. N.O.A.A. N.M.F.S. 51, 95–102. Scharf, B. (1970). Critical bands. In “Foundations of Modern Auditory Theory” (J. V. Tobias, Ed.), pp. 159–202. Academic Press, New York. Shannon, C. E., and Weaver, W. (1949). “The Mathematical Theory of Communication.” Univ. Illinois Press, Urbana. Speirs, E. A. H., and Davis, L. S. (1991). Discrimination by Adelie ´ penguins between the loud mutual calls of mates, neighbours and strangers. Anim. Behav. 41, 937–944. Stonehouse, B. (1960). The king penguin Aptenodytes patagonicus of South Georgia. 1. Breeding behaviour and development. Falk. Is. Dep. Surv. Sci. Rep. 23. Sturdy, C. B., and Mooney, R. (2000). Bird communication: Two voices are better than one. Curr. Biol. 10, 634–636. Trivers, R. L. (1972). Parental investment and sexual selection. In “Sexual Selection and Descent of Man” (B. Campbell, Ed.), pp. 136–179. Aldine, Chicago. Verheyden, C., and Jouventin, P. (1994). Olfactory behavior of foraging procellariiforms. Auk 111, 285–291. Waas, J. R. (1988). Acoustic displays facilitate courtship in little blue penguin Eudyptula minor. Anim. Behav. 36, 366–371. Waas, J. R. (1991). Do little blue penguins signal their intentions during aggressive interactions with strangers? Anim. Behav. 41, 375–382.

VOCALLY IDENTIFYING KIN IN A CROWD: THE PENGUIN MODEL

277

White, S. J., and White, R. E. C. (1970). Individual voice production in gannets. Behaviour 37, 41–54. Wiley, R. H., and Richards, D. B. (1978). Physical constraints on acoustic communication in the atmosphere: Implication for the evolution of animal vocalizations. Behav. Ecol. Sociobiol. 3, 69–94. Wiley, R. H., and Richards, D. G. (1982). Adaptations for acoustic communication in birds: Transmission and signal detection. In “Acoustic Communication in Birds” (D. E. Kroodsma and E. H. Miller, Eds.), Vol. I, pp. 131–181. Academic Press, New York.

This Page Intentionally Left Blank

Index

A Acoustic code, for penguin identiﬁcation, 259–269 Acoustic problems, penguin recognition, 249–254 solutions to, 269 Action-level imitation copying ﬂuid movement, 90–91 copying string of beads, 91–93 Aggressive responses birds, 139–141 ﬁsh, 143 primates, 143–144 reptiles, 142 toads, 141–142 Alarm calls eagle, by vervet monkeys, 82 by magpie, hemispheric preference in, 124–125 Amphibians predator–escape and fear responses, 125–128 prey capture, hemispheric control, 117–120 Ampullary electroreceptors, 212–216 Androgens, and female spotted hyena virilization, 14–19 Antisymmetry, differing from ﬂuctuating asymmetry, 109 Anurans auditory scene analysis, 185–186 whine-chuck courtship vocalization, 226 Approach, potential predator by ﬁsh, 130 Attention hemispheric control in birds, 133–136 ﬁsh, 136–137 primates, 138–139 rats, 137–138 role in auditory scene analysis, 168–169 selective, 183 279

Auditory objects and auditory scene, 164–165 grouping principles and stream segregation, 165–168 segregation of simultaneous birdsongs into, 170–174 Auditory scene analysis anurans, 185–186 attention role, 168–169 bottom-up theories, 192–194 chorusing among conspeciﬁcs, 182–183 dawn chorus, 186–187 ﬂocking birds and colonial species, 187–188 formal models, 194–195 as general perceptual process, 178–181 individual recognition and, 174–177 primates, 189–191 songbirds, 169–181 song timing and communication networks, 188–189 time role, 168 top-down, goal-driven theory, 191–192 undersea, 183–185

B Biological problems, penguin recognition, 248–249 Birds, see also specific species aggressive responses, 139–141 auditory scene analysis dawn chorus, 186–187 ﬂocking and colonial species, 187–188 songbirds, 169–181 song timing and communication networks, 188–189 density, screening effect of bodies, 249–257

280

INDEX

Birds (continued ) feeding, lateralization in, 112–117 foraging trips by seabirds, 243–244 male, song variation and male competitive ability, 36–40 predator–escape and fear responses, 123–125 spatial maps and attention, 133–136 Birdsong condition and song rate, 38–39 duets, 87 hierarchical organization, 98 male contest, 33–36 mate attraction, 32–33 neuromotor mechanism and vocal characteristics, 39–40 repertoires and male quality, 36–37 simultaneous, segregation into auditory objects, 170–174 song frequency and body size, 37–38 timing and communication networks, 188–189 Body size, male bird, and song frequency, 37–38 Bottom-up theories, auditory scene analysis, 192–194 Bow-coo, collared dove, 43–44 Brain, capacity and efﬁciency, lateralization effect on, 144–148

C Catﬁsh ampullary electroreceptors, 215–216 blind Amazonian, 224 predation of gymnotiformes, 217 Cats, paw preference, lateralization of, 146–147 Challenge, social, conditions among spotted hyenas, 15, 18 Challenge hypothesis, mating strategies, 19–20 Clan territories, spotted hyena, food resources within, 5–6 Cocktail party phenomenon, 164, 169, 259–260 Coding–decoding system, in penguin recognition, 270–273

Collared dove element number, variation in, 57–62 modulated elements production of, 52–53 receiver responses to, 48–51 sender characteristics, 46–48 perch-coo variability among individuals, 44–46 vocalization, 64–65 sound production, 53–57 supermodulations, 51–52 territorial behavior, 41–42 Communication networks, birdsong timing and, 188–189 Commuting system, spotted hyena access to food, 5–6 Comodulation masking release, theory of auditory scene analysis, 193–194 Competitive ability male, and song variation, 36–40 receiver responses to modulated elements and, 48–49 Conﬂict outcome, motivation role, 35 among spotted hyena over access to food, 5–8 resolution, 22–23 Conspeciﬁcs amphibian aggressive responses to, 141–142 chorusing among, 182–183 individual recognition, auditory scene analysis and, 174–177 presence, and social facilitation, 80 viewed by ﬁsh during inspection of predator, 130, 136–137 Coos two-element, 57–58 as whistles, 54–55 Coo types collared dove, 42–44 nonmodulated, responses to, 49–51 Copying, see also Imitation ﬂuid movement, 90–91 for its own sake, 86 sound patterns by birds, 89 string of beads, 91–93, 97–98 Courtship signals, male gymnotiformes, 224–227

281

INDEX

D Dawn chorus auditory scene, 186–187 as natural noise, 183 song segregation from, 173–174 Density, penguins: screening effect of bodies, 249–254 Discrimination grain–pebble, 112 learned by segregating birdsongs, 171 Disgust response, chick, 125 Distractors, individual birdsong recognition with, 175–177 Dolphins, auditory scene analysis, 184–185 Dominance, female spotted hyena, virilization and, 13–14

E Electric eels predator of gymnotiformes, 216 three pairs of electric organs 219–221 Electric organ discharge duration and amplitude, 209–211 electrostatic ﬁeld, 205–206 generated by wave and pulse ﬁsh, 208–209 pulse vs. wave, 202–204 remote, 221 sexually dimorphic, 225–226 spectrum constriction, 229 upward-shifted spectrum, 216 Electrolocation, environmental physiology of, 235 Electrolytes, serial arrangement, 206 Electroreceptors ampullary, 212–216 tuberous, 214–215, 220–221, 229 Electrosensory lateral-line lobe, 214–215, 224–225 Elements modulated production of, 52–53 receiver responses to, 48–51 sender characteristics, 46–48 supermodulations, 51–52 numbers

in perch-coos, 45–46 variation in, 57–62 Emulation deﬁned, 88 object affordances and, 89–90 Energy, gymnotiformes, oxygen and, 228–230 EOD, see Electric organ discharge Evolution imitation, 78 lateralization, 120–121 pseudo-penis, 1 signal, see Signal evolution Exclusive allocation, principle of, 165, 167 Eye use preferential, by feeding birds, 115 in tool manufacture by crows, 116–117

F Fear response amphibians, 125–128 birds, 123–125 ﬁsh, 128–130 primates, 131–133 rats, 130–131 Feeding bird, hemispheric control, 112–117 discrimination in, 111 lateralization in ﬁsh, 120–121 rodent, 121–122 Female choice, birdsong and, 32–35 Female spotted hyena dominance, virilization and, 13–14 maternal effort, 7–8 relationship with males, 21–22, 25 social hierarchy, 4–5 virilization, ﬁtness consequences, 19–20 vocal defense of resources, 6–7 Finches, auditory scene analysis, 177–178 Fish, see also Catﬁsh; Gymnotiformes aggressive responses, 143 auditory scene analysis, 184 electric, 202–204, 209 feeding response, lateralization in, 120–121 jamming avoidance response, 215 predator–escape and fear responses, 128–130

282

INDEX

Fish (continued ) pulse, see Pulse ﬁsh schooling, associated with population bias, 149 spatial maps and attention, 136–137 wave, see Wave ﬁsh Fitness consequences, female hyena virilization, 19–20 Fluctuating asymmetry, antisymmetry differing from, 109 Food access to, conﬂict among spotted hyenas, 5–8 buried, chicks searching for, 134–135 cached by birds left hemisphere in, 116 recovery, 135 Foraging options for female spotted hyena, 7–8 trips, by seabirds, 243–244 Formant transitions, Mynah birds, 89 Frequency domain, penguin species preferences, 262 Frequency modulation, perch-coo, 45

G Galloping sequence, auditory objects organized into, 178–181 Genetic indicator hypothesis, 32–33 Goal-driven theory, auditory scene analysis, 191–192 Greeting ceremony, spotted hyena, 23 Grouping principles, auditory objects and, 165–168 Gymnotiformes electric mate choice, neuroethology, 234 electric sense, 212–215 electrolocation, 204–205 electromotor system, 205–212 mormyrid wave ﬁsh, 232–233 oxygen and energy, 228–230 predation of, 215–224 electrosensory, 234 sex vs. predators, 224–228 male quality, 228 sensory bias, 227 thermal traps, 230–231

H Handling, rats, effect on lateralization, 137–138 Hand preferences apes, 122–123 associated with immune responses, 150–151 monkeys, 147 primates, 131–133 Harmonics, EODs, 209 Hemispheric specialization model, 110–111 predator–escape and fear responses, 123–133 Hierarchical organization, modules as subroutines, 94–96 Hyenas, see Spotted hyena

I Identiﬁcation, penguin acoustic code for, 259–269 optimization, 255–258 Imitation, see also Copying contextual, 82–83 evolutionary origins, 78 production learning by, 87–97 limits to, 97–98 social learning without, 79–81 social mimicry and picking up methods, 85–87 Immune responses, lateralization and, 150–151 Impedance matching, gymnotiformes, 232 Inertia, phylogenetic, 227 Intrusion pressure, spotted hyena, 6–7 Invariant features, in string of elements, 95 Isochronous sequences, auditory objects organized into, 178–181

L Lateralization aggressive responses, 139–144 effects of handling on, 137–138

283

INDEX

enhanced cognitive capacity and efﬁciency due to, 144–148 feeding responses in birds, 112–117 immune responses and, 150–151 individual and population, 107–110 population, in social situations, 148–149 Learning production, see Production learning social, without imitation, 79–81 when and how, 81–85 Light, chick embryos exposed to, 140–141, 145–149 Lizards, leftward agonistic encounters, 142

M Male contest signal evolution in context of, 33–36 vocal variability and, 62–64 Male quality, gymnotiformes, 227 Male spotted hyena androgen levels, 14–15 dominance hierarchy, 20–21 mating success, 19–20 natal postdispersal, 22 relationship with females, 21–22, 25 submission, 4 twin litters, siblicide and, 11 Mate attraction, birdsong and, 32–33 Mate choice, electric, neuroethology of, 234 Maternal care female spotted hyena, 7–8 siblicide and, 9–10 Mating strategies, spotted hyena, 19–20 Mauthner cell, driving ballistic escape movement, 128 Meeting places, penguin species, 255–256 Melody, interleaved, 167 Memorization, acoustic features of birdsongs, 171–173 Mimicry, social, 100 and picking-up methods, 85–87 Mirror neurons, 84, 91–92, 96 Modulation, AM and FM, penguin recognition and, 262 Monkeys, see also Primates copying by human-reared apes, 86 great ape, program-level imitation, 98

hand preference, 147 macaque, dialects among, 80 vervet, eagle alarm calls, 82 Motor behavior, contextual learning, 83–84

N Natal clan, spotted hyena, 4–5 Nest-coo, collared dove, 43–44 Neuroethology, of electric mate choice, 234 Neuromotor mechanisms, bird vocal characteristics and, 39–40 Neuromuscular exhaustion hypothesis, 58–59 Noise background, penguin colony, 249 environmental, acoustic ﬁlters, 188 natural, 182–183

O Oxygen consumption, gymnotiform energy and, 228–230

P Pacemaker neurons, 208 Paw preference associated with immune responses, 150–151 cat, lateralization of, 146–147 as individual lateralization, 108–110 Penguins acoustic communication, 254–255 coding–decoding system, 270–273 identiﬁcation behavioral solutions, 255–258 well-matched acoustic code for, 259–269 nonnesting brooding, 244–245 vocal identiﬁcation systems, 271–272 recognition acoustic problems, 249–254 biological problems, 248–249

284 Penguins (continued ) recording and playback procedures, 245–247 sound analysis and synthesis, 247–248 species recognition, 258–259 vocal signatures, 269–270 Perceptual opacity, 83 Perch-coo modulated elements, 46–57 sexual selection and, 44 three-element, 64–65 variability among individuals, 44–46 Peripheral channeling, theory of auditory scene analysis, 192 Phylogenetic inertia, gymnotiformes, 227 Playback experiments electric, to resting electric eel, 221 penguins, 245–247 two-voice system, 267–268 responses to coo variants, 48–51, 60–62 Predation avoidance, electric signal role, 204 electrosensory, 234 of gymnotiformes, 215–224 Predator–escape response amphibians, 125–128 birds, 123–125 ﬁsh, 128–130 primates, 131–133 rats, 130–131 Prey capture by kookaburras, eye dominance in, 135 by toads, hemispheric control, 117–120 Primates, see also Monkeys aggressive responses, 143–144 auditory scene analysis, 189–191 feeding, handedness in, 122–123 predator–escape and fear responses, 131–133 spatial maps and attention, 138–139 Primitive segregation, distinct from schema-based segregation, 167 Production learning, by imitation, 87–97 limits to, 97–98 Program-level imitation, 92–99 Propagation experiments, penguins, 246 Pseudo-penis evolution, 1 female spotted hyena, 13–14 Pulse ﬁsh

INDEX

courtship signals, 225–226 dyadic interactions, 211 EOD trains of, 208–209 thermal range restrictions, 231

Q Queuing, for social status, by male spotted hyena, 20–21

R Rats feeding responses, lateralization of, 121–122 predator–escape and fear responses, 130–131 spatial maps and attention, 137–138 Recognition individual, auditory scene analysis and, 174–177 penguin parents and chicks, 246–247 penguin species, 258–259 Redundancy, as adaptive trait, 63 Relationships between male and female spotted hyena, 21–22, 25 signal evolution with habitat, 201 Repetition rate, birdsong, 39–40 Reptiles, aggressive responses, 142 Resonance, acoustic, collared dove, 55 Rivalry, spotted hyena siblings, 8–10 Runaway model, mate attraction, 32–33

S Scars predator-inﬂicted, preponderance on left side of ﬁsh, 129–130 reﬂecting lateralized intraspecies interactions, 142 Schooling, associated with population bias, 149 Screening effect, penguin bodies, 249–257 Searching strategy, penguin species, 257–258 Selection

285

INDEX

intra- and intersexual, 34–35, 51–52 sexual, perch-coo and, 44 Sensory bias, gymnotiform courtship signal, 227 Sexual dimorphism, EOD characters, 209–211, 225–226 Shadowing, by immigrant male spotted hyena, 21 Siblicide habitual, 24 spotted hyena, 8–10 twin litters: sex composition biases, 10–11 Sickness response, left hemispheric control, 121 Signal evolution in context of male contest, 33–36 relationship with habitat, 201 supermodulations and, 51 Signaling posture, penguin species, 256–257 Silence, replacing part of song by, 60–62 Skewed bias, rightward, at population level, 108–109 Social challenge, effect on spotted hyenas, 15, 18 Social facilitation, 80 Social learning, without imitation, 79–81 Social mimicry, 85–87, 100 Social organization, spotted hyena, 4–5 Social situations, population lateralization in, 148–149 Social status, queuing by male spotted hyena, 20–21 Sound analysis, penguin calls, 247–248 Sound production in collared dove, 53–57 coupled with respiration, 59 Spatial maps birds, 133–136 ﬁsh, 136–137 primates, 138–139 rats, 137–138 Specialization, hemispheric, model, 110–111 Species recognition, penguin, 258–259 Species speciﬁcity birdsong, acoustic tests with, 173 responses to coo variants, 61–62 territorial birdsong, 34–35 Spotted hyena commuting system, 5–6

conﬂict resolution, 22–23 female, see Female spotted hyena male, see Male spotted hyena maternal care, 7–8 sibling rivalry, 8–10 social organization, 4–5 territorial defense and intrusion pressure, 6–7 twin litters: sex composition biases, 10–11 virilization and androgens, 14–19 Starlings, individual recognition and auditory scene analysis, 174–177 Statistical sketch, 94, 96 Stream segregation based on acoustic dimensions, 167–168 in formal model of auditory scene analysis, 194–195 as general perceptual process, 178–181 Striking response, prey, of toads, 117–120 String parsing model, 92–97, 99–100 Structural asymmetry, thalamofugal visual projections of chick, 114 Submission, by male spotted hyena, 4, 24 Syrinx dove, sound production, 53–57 penguin species, 265–269

T Temporal domain, penguin species preferences, 262–265 Territorial defense birdsong in, 33–36 male collared dove, 41–42 spotted hyena, 6–7 Thermal traps, gymnotiformes, 230–231 Time, role in auditory scene analysis, 168 Toads aggressive responses, 141–142 predator–escape and fear responses, 125–128 prey capture, hemispheric control, 117–120 Tools, manufactured by birds, eye use during, 116–117 True imitation, 81–84 Tuberous electroreceptors, 214–215, 220–221, 229

286

INDEX

Twin litters, hyena, siblicide and, 10–11 Two-voice system, penguin species, 265–269

as multicomponent signals, 63–64 Streptopelia doves, 40 Vocal signature, penguin recognition, 260–269

U Undersea life, auditory scene analysis, 183–185

V Virilization female spotted hyena androgens and, 14–19 dominance and, 13–14 ﬁtness consequences of, 19–20 Vocal characteristics neuromotor mechanisms and, 39–40 variability, and male contest, 62–64 Vocalization alarm calling by magpies, 124–125 bird species, body size role, 37–38 collared dove, 42–44 eagle alarm calls by vervet monkeys, 82 by female spotted hyena, in defense of resources, 6–7 frog whine-chuck courtship, 226

W Water ﬂow rate, affecting gymnotiformes, 212, 235 Wave ﬁsh adapted to fast waters, 212 chirps, 226 EODs generated by, 208 habitat restriction, 233 mormyrid, paucity of, 232–233 multiphasic waveforms, 216–217 Waveform dc-balanced, 217 EOD, 209–210 Weight, male bird correlation with modulated elements, 47–48 song rate and, 38–39 Whales, auditory scene analysis, 184–185 Whistle model, collared dove sound production, 54–56

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 Cicadian 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

Social Behavior and Organization in the Macropodoidea PETER J. JARMAN

Kin Recognition: Problems, Prospects, and the Evolution of Discrimination Systems C. J. BARNARD

The t Complex: A Story of Genes, Behavior, and Population SARAH LENINGTON

Maternal Responsiveness in Humans: Emotional, Cognitive, and Biological Factors CARL M. CORTER AND ALISON S. FLEMING

The Ergonomics of Worker Behavior in Social Hymenoptera PAUL SCHMID-HEMPEL 287

288

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 MØLLER, 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 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 in Warning Coloration in Avian Predators W. SCHULER AND T. J. ROPER Analysis and Interpretation of Orb Spider Exploration and Web-Building Behavior FRITZ VOLLRATH 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

Proximate and Developmental Aspects of Antipredator Behavior E. CURIO Newborn Lambs and Their Dams: The Interaction That Leads to Sucking MARGARET A. VINCE The Ontogeny of Social Displays: Form Development, Form Fixation, and Change in Context T. G. GROOTHUIS

Volume 23 Sneakers, Satellites, and Helpers: Parasitic and Cooperative Behavior in Fish Reproduction MICHAEL TABORSKY Behavioral Ecology and Levels of Selection: Dissolving the Group Selection Controversy LEE ALAN DUGATKIN AND HUDSON KERN REEVE 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

289

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

Sociality, Group Size, and Reproductive Suppression among Carnivores SCOTT CREEL AND DAVID MACDONALD

Maternal Behavior in Rabbits: A Historical and Multidisciplinary Perspective ´ GABRIELA GONZALEZ-MARISCAL AND JAY S. ROSENBLATT

Development and Relationships: A Dynamic Model of Communication ALAN FOGEL

Parental Behavior in Voles ZUOXIN WANG AND THOMAS R. INSEL

Why Do Females Mate with Multiple Males? The Sexually Selected Sperm Hypothesis LAURENT KELLER AND HUDSON K. REEVE Cognition in Cephalopods JENNIFER A. MATHER

Volume 25 Parental Care in Invertebrates STEPHEN T. TRUMBO Cause and Effect of Parental Care in Fishes: An Epigenetic Perspective STEPHEN S. CRAWFORD AND EUGENE K. BALON Parental Care among the Amphibia MARTHA L. CRUMP

Physiological, Sensory, and Experiential Factors of Parental Care in Sheep ´ F. LEVY, K. M. KENDRICK, E. B. KEVERNE, R. H. PORTER, AND A. ROMEYER Socialization, Hormones, and the Regulation of Maternal Behavior in Nonhuman Simian Primates CHRISTOPHER R. PRYCE Field Studies of Parental Care in Birds: New Data Focus Questions on Variation among Females PATRICIA ADAIR GOWATY Parental Investment in Pinnipeds FRITZ TRILLMICH Individual Differences in Maternal Style: Causes and Consequences of Mothers and Offspring LYNN A. FAIRBANKS

290

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 Seawood Flies THOMAS H. DAY AND ´ S. GILBURN ANDRE 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 H. MOORE 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 MØLLER 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 Ethiological 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

291

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

This Page Intentionally Left Blank

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

Recommend Documents