The role of the visual train ornament in the courtship of peafowl,

Pavo cristatus

by

ROSLYN DAKIN

A thesis submitted to the Department of Biology

in conformity with the requirements for

the degree of Master of Science

Queen’s University

Kingston, Ontario, Canada

September, 2008

Copyright © Roslyn Dakin, 2008 ii

ABSTRACT

The peacock (Pavo cristatus) has long been considered the quintessential example of a

sexually selected animal, and in the last two decades, peafowl have provided widely-cited

evidence for female mate choice as well as the genetic benefits of mate preferences for

ornamented males. However, previous studies have failed to reach a consensus with

respect to the importance of various signaling modalities in peafowl courtship. In this

thesis, I repeat two previous studies of peacock train morphology and I describe the use

of light by males during their courtship displays, to clarify the role of visual signaling. I

confirm previous reports that removing a large number of eyespots decreases male

mating success, yet I find substantial variation in mating success among normal males

that cannot be explained by eyespot number. I show that these two apparently conflicting

results are not contradictory, since the removal treatment modifies males beyond the

normal range of eyespot number. Next, I describe the two display behaviours used by

males during courtship. When males perform their pre-copulatory “train-rattling” display,

they are oriented at about 45° relative to the sun on average, with females directly in front. This directional pattern suggests that train-rattling is involved in the communication of a visual signal. The “wing-shaking” display, on the other hand, is performed with females positioned behind males, and is always elicited when a model female is positioned on the shaded side of a male. The wing-shaking display may therefore allow males to control female viewing geometry. These results indicate that mate choice in peafowl is complex, and that visual signaling is important despite recent claims to the contrary. Females may avoid males missing a large number of eyespots via iii a threshold-based mechanism, while choosing among full-trained males based on some other (possibly visual) cue. iv

CO-AUTHORSHIP

This thesis conforms to the “Manuscript Format” outlined in the Department of Biology

Guide to Graduate Studies. Chapters 2 and 3 are co-authored by my supervisor, Dr.

Robert Montgomerie, who contributed to the experimental design as well as the analysis and presentation of results. v

ACKNOWLEDGEMENTS

First, I thank my supervisor Bob Montgomerie for guidance and support over the last three years – I have benefited tremendously from it. I also acknowledge my committee members, Steve Lougheed and Paul Martin, who provided a number of insights. Second,

I thank the directors and staff of the Los Angeles Arboretum (LAA), Assiniboine Park

Zoo (APZ), Toronto Zoo (TZ) and Bronx Zoo (BZ) for allowing me to chase after their peafowl. This work would not have been possible without the logistic support of Dr.

Mark Wourms (LAA), Dr. Bob Wrigley and Dr. Gord Glover (APZ), Dr. Christine

Sheppard (BZ), and Dr. Tom Mason and Dr. Graham Crawshaw (TZ). The keeper and veterinary staff from the Assiniboine, Toronto and Bronx Zoos were particularly helpful in handling the birds – especially Janice McCarthy and David McLelland. Funding for this research was provided by an NSERC Canada Graduate Scholarship. Field work was supported by NSERC grants to R Montgomerie (Discovery and equipment), and the

Society for Canadian Ornithologists (Fred Cooke award).

I am indebted to Rob Ewart and Jason Clarke for hard work (and good company) in the field – I couldn’t have done it without you. Lori Parker helped with field work in

Winnipeg and provided encouragement. Dr. Norman Johnson provided a peahen from his farm, and Vanya Rowher did a fantastic job of preparing the experimental mount. Finally,

I am grateful to my family and friends for their support, especially those in the Biology department for many inspiring discussions, and to Charlie, for help in more ways than I can list here. vi

TABLE OF CONTENTS

Abstract …………………………………………………………………….…… ii

Co-authorship ……………………………………………………………....…... iv

Acknowledgements ………………………………………………………….….. v

Table of Contents …………………………………………………………….…. vi

List of Tables ………………………………………………………………….... viii

List of Figures …………………………………………………………………... ix

CHAPTER 1. General Introduction …………………………………………….. 1

References …………………………………………………………….. 15

CHAPTER 2. Train morphology, ornamentation and mate preferences in peafowl ………………………………………………………………………….. 22

Abstract ……………………………………………………………….. 22

Introduction ………………………………………………………….... 22

Materials and Methods ………………………………………………... 26

Results ……………………………………………………………….... 30

Discussion …………………………………………………………….. 34

References …………………………………………………………….. 38

CHAPTER 3. Peacocks orient their displays towards the sun ………………….. 49

Abstract ……………………………………………………………….. 49

Introduction ………………………………………………………….... 50

Materials and Methods ………………………………………………... 54

Results ……………………………………………………………….... 60

Discussion …………………………………………………………….. 64

References …………………………………………………………….. 68 vii

CHAPTER 4. General Discussion …………………………………………….... 79

References …………………………………………………………….. 84

SUMMARY …………………………………………………………………….. 87

APPENDIX ……………………………………………………………………... 88 viii

LIST OF TABLES

Table 2.1 Previous studies of the number of eyespot feathers in peafowl ….……. 40

Table 2.2 Comparisons of the train morphology and behaviour of control males from three populations …………………………………………….….... 41

Table 2.3 Regular arrangement of eyespots in rows on the trains of four normal adult peacocks ……………………………………………………..…… 42

Table 2.4 Effect of eyespot removal on male behaviour and mating success ……. 43

Table 2.5 Relations between train morphology and mating success for control males ……………………………………………………………….…... 44

Table 3.1 Display responses of males to a model female presented under different experimental contexts ………………………………………...... 71

Table 3.2 Directionality of male display responses to a model female …………... 72

Table A.1 Sampling for model presentation experiment from Chapter 3 ……….... 88 ix

LIST OF FIGURES

Figure 1.1 The highly complex train ornament of a peacock …………………... 20

Figure 1.2 Close up view of a single eyespot feather …………………………... 21

Figure 2.1 Four specialized upper-tail covert feathers in the peacock’s train ….. 45

Figure 2.2 Train morphology of two normal adult peacocks that were not captured, and two eyespot removal treatment peacocks ………….…. 46

Figure 2.3 Effect of removing the 20 outermost eyespot feathers on male mating success ……………………………………………………………….. 47

Figure 2.4 Relations between the number of eyespots and male mating success, and between train length and male display rate ……………………... 48

Figure 3.1 Daily time budget for peacocks showing the average percent of time males devote to each activity at each level in the hierarchy ……….... 73

Figure 3.2 Diagram from above of the 60° sectors around a displaying peacock, used to identify the position of the female relative to the male, and three angles of interest ……………………………………………….. 74

Figure 3.3 Sequence of events over typical male display bouts ……………….... 75

Figure 3.4 Position of female observers relative to the male during display, and male movement during the wing-shaking and train-rattling displays .. 76

Figure 3.5 Sun-male angle when females were absent ………………………….. 77

Figure 3.6 Sun-male angles and sun-female angles during the wing-shaking and train-rattling displays ……………………………………………….... 78

1

CHAPTER 1. General Introduction

For millennia, people have marveled at the intricate complexity of the peacock’s tail

(Figure 1.1). The sheer size of the tail – or more correctly, the train – is particularly baffling. It contains roughly 200 elongated feathers, 150 of which are tipped with an iridescent multicoloured eyespot, so named because of an eye-like pattern of concentric ellipses (Figure 1.2). Many have considered peacocks “the most splendid of living birds”

(Darwin 1871), and the current distribution of feral populations throughout the western world is no doubt due to man’s fascination with their glittering train (Jackson 2006). Yet the fact that has been so obvious to generations of human admirers – that the sole purpose of the train is its beauty – is one that has posed some problems for evolutionary biology.

From Darwin (1871) asking how natural selection could allow such extravagance to more recent questions about how variation in ornamental traits could be maintained

(Williams 1975, Borgia 1979), the peacock’s train has been considered the quintessential example of a sexually selected trait. Questions about the evolution of such ornamental traits have led to a number of insights with broad relevance, such as the reasons that animal signaling systems might remain stable despite the benefits of deception (Maynard

Smith and Harper 2003). These studies have also contributed to our understanding of the causes and consequences of mate choice (Andersson 1994), and may even be relevant to the generation of biological diversity, since the main difference between many closely- related species often involves ornamental traits similar to the peacock’s train (Darwin

1871).

2

“A Taste for the Beautiful”

Peacocks were a problem for Charles Darwin. The complexity of the eyespot feathers made him sick with worry (Darwin 1860 in Burkhardt et al. 1993), especially since he was sure that the size of the train would render males “more easy prey to any prowling tiger-cat” (Darwin 1871). He eventually came up with a solution that addressed the evolution of ornamental traits in the males of many species – that purely aesthetic traits were selected through the struggle for mates rather than for survival, a process he called sexual selection (Darwin 1871). Darwin saw that the process of females choosing mates with their “individual preferences and dislikes” could select for the train of the peacock in spite of its costs; he even cited evidence that farm-bred peahens show marked preferences for certain males, in support of this theory (Darwin 1859).

Most of Darwin’s contemporaries rejected sexual selection, and the theory was largely ignored for almost 100 years (Cronin 1991). Part of the problem may have been that Darwin likened animal mate preferences to human ideals of beauty, at least in part to emphasize the continuity of man with other animals. Part of the problem may have also been a deep-seated cultural bias against explanations that attributed choice and control to females, let alone any lower animals (Harel 2001, Hiraiwa-Hasegawa 2000). However, since the mid-1970s biologists have found extensive support for Darwin’s idea, and non-random mating has been documented in many animals, from insects to mammals

(reviewed in Andersson 1994).

Indeed, females often appear to exert much control over the mating process. For example, peahens sample the courtship displays of a number of different males before choosing to mate with one or two individuals (Petrie et al. 1991, 1992). In a number of 3 celebrated cases, observational evidence for female choice has been corroborated by experimental manipulations of male ornaments (Andersson 1994). The evidence that peahens prefer males with larger train ornaments, for instance, is supported by experiments showing that reducing train size also reduces male copulation rates, as females are less receptive to their advances (Petrie et al. 1991, Petrie and Halliday 1994).

Furthermore, in model organisms amenable to laboratory testing, mate choice is often consistent across individuals in controlled experiments (e.g., pigeons, Columba livia,

Burley and Moran 1979). Although some have argued that much of this evidence could be explained by passive attraction to more detectable courtship displays rather than active choice (Kotiaho and Puurtinen 2007), Darwin’s idea that ornamental traits are selected through their effect on mating success is now widely accepted.

Benefits of Female Preference for Ornamental Traits

A major reason that Darwin’s theory of sexual selection was ignored for so long was that he failed to provide a convincing account as to why females should prefer to mate with more ornamented males (Cronin 1991). His best guess was that females preferred more beautiful males for aesthetic reasons alone, although the great variety of male ornaments suggested to him that there could be mate choice for novelty’s sake as well (Darwin

1871). But what maintains female preference, especially when a choosy female’s sons must pay the costs of bearing an extravagant ornament?

Several models have been proposed to explain the evolution of female choice, and they generally fall into two different categories: direct and indirect benefits for females.

The least controversial of the two is that females may gain direct benefits, such as 4

improved fecundity or survival for their offspring, from a preference for males with

elaborate ornamental traits (Hoelzer 1989). This could occur if, for example, more

ornamented males provide better resources or parental care for the female’s offspring.

However, in many of the most-ornamented species like peafowl, males provide no resources or parental care and contribute only genes to their offspring. Nevertheless, females can still gain direct benefits in these systems if mating preferences allow them to avoid certain costs of reproduction. For instance, Loyau et al. (2007a) have found some

evidence that peahens prefer mating with males that have brighter and more iridescent

eyespots, and they have suggested that more colourful males might be easily detected

from afar, which could reduce the cost of mate searching. Similarly, female natterjack

toads (Epidalea calamita) may be more likely to mate with males giving loud calls

simply because this preference minimizes the cost of locating males of the same species

(Arak 1988).

Another explanation involving direct benefits is that female mate preferences

might be selected through the effects of natural selection on the sensory system. Perhaps

pre-existing sensory biases among females that are useful in foraging or in other social

contexts translate into mating preferences, and thus have an impact on ornament

evolution (Ryan 1990, Basolo 1990). A likely explanation for the fact that both peafowl

and ocellated turkeys (Meleagris ocellata) have independently evolved iridescent

eyespots in their display plumage could be that females of both species are pre-disposed

to attend to bright colours and eye-like shapes. Although direct evidence for the

pleiotropic effects of natural selection on mate preferences has been limited (Fuller et al.

2005), some of the most convincing evidence comes from species where male ornaments 5

appear to mimic female prey. For example, in the water mite Neumania papillator,

courting males tremble their legs at a frequency that matches the vibrations produced by

copepod prey, and female mites respond to both courtship and prey vibrations in the same

way (Proctor 1991). In fish, morphological lures used by males in courtship often appear

to mimic prey organisms, and these lures have arisen independently in a number of species (Arnqvist and Rowe 2005). Evidence that some female fish and birds prefer males bearing novel ornaments has also been cited in support of the sensory bias hypothesis. Female zebra finches (Taeniopygia guttata) prefer males bearing coloured leg

bands (Burley 1988) and artificial crest feathers, a trait which has no previous expression

within the grassfinch taxon (Burley and Symanski 1998). These results are remarkably

consistent with Darwin’s concept of mate choice for novelty. However, they provide only

indirect evidence of the role of sensory bias in the evolution of ornaments for the simple

fact that these artificial traits have not yet evolved (Arnqvist 2006).

Models that explain female mate choice by indirect benefits are more

controversial; these claim that the benefits of choice are gained in the form of the genetic

contribution that males make to their offspring. These theories were pioneered by Fisher

(1930), who provided the first convincing account of mate preference evolution. Fisher

asserted that it is possible for a preference gene and an ornamental trait to spread

throughout a population by positive feedback, since both would entail a reproductive

advantage provided the other was common. All that is required is for the ornament and

preference to become genetically linked for Fisher’s “runaway” process to get started

(Kirkpatrick 1982). In this scenario, females gain the indirect benefits of mate choice in

the form of attractive sons. There is some evidence that Fisherian runaway selection has 6

contributed to the evolution of mate choice and ornamentation in real organisms, but

much of it is circumstantial. For example, some have argued that the positive-feedback

nature of the runaway model means that as long as mate choice is not costly, ornamental

traits will evolve readily by this mechanism in the wild (Kotiaho and Puurtinen 2007). As

plausible as this may be, the Fisherian process is difficult to demonstrate empirically

(Andersson and Simmons 2006): evidence of trait-preference coevolution is not enough, since this is also invoked in other models of sexual selection. Others have suggested that the explosive diversification of sexual display traits in groups such as the lekking manakins (Aves: Pipridae) supports a rapid positive feedback mechanism for their evolution (Prum 1997), but there appears to be no relation between sexual selection and ornament diversification in other bird groups (Bennett and Owens 2002).

An alternative indirect-benefits model is that, by preferring to mate with certain males, females gain a genetic benefit of healthier, or otherwise higher quality, offspring.

This hypothesis also has roots in ideas that were originally presented by Fisher (1915), but it was laid out more specifically by Amotz Zahavi (1975) with his theory that sexually-selected ornaments act as “marks of quality”. Zahavi’s idea was that sexual ornaments function as honest signals of male genetic quality precisely because they are handicaps on male survival: the bigger the peacock’s train, the healthier and more vigorous the male must be to bear it. Others noted that the signals used in courtship often appear to be designed to advertise health and condition; plumage and fur can provide information about external parasites, and bare patches of skin can reveal the condition of the blood (Hamilton and Zuk 1982). Subsequent theoretical modeling confirmed that as long as ornaments are necessarily honest signals of male quality, females would do well 7

to pay attention to them (Grafen 1990a, 1990b), although ornaments need not act as handicaps on male survival to function in this way.

The honest signal model for mate choice is supported by the evidence from a wide range of taxa that sexual ornaments are condition-dependent (reviewed in Andersson

1994). In peafowl, the length of the train, the number of eyespots, and eyespot size may all be related to male condition (Petrie 1992, Møller and Petrie 2002, Loyau et al. 2005a).

However, the evidence thus far has some limitations: first, experimental tests of condition-dependence are lacking. In addition, because the development of any body structure is likely to be costly, the critical test of the honest signal model involves demonstrating that ornamental traits display heightened condition-dependence relative to other body structures, and this has rarely been attempted (Cotton et al. 2004). A more conclusive line of evidence in support of the honest signal model would be to show that the offspring of more ornamented males are healthier on average. This appears to be the case in peafowl, but it has only been demonstrated in a handful of other species (Petrie

1994; reviewed in Kokko et al. 2003).

Outstanding Questions

Despite extensive theoretical and empirical work on sexual selection over the past three decades, many questions remain. We still lack a clear picture of the relative importance

and general applicability of the various models for the benefits of mate choice (Kokko et

al. 2003, Andersson and Simmons 2006). For example, the plausibility of indirect genetic

benefits of mate choice is widely accepted, but there have been very few direct tests of

the relation between male ornamentation and offspring fitness, especially in the wild 8

(Kokko et al. 2003, Kotiaho and Puurtinen 2007). This may be due in part to the difficulty of measuring offspring fitness directly in wild populations. For example, two recently published studies of collared flycatchers (Ficedula albicollis) and blue tits

(Cyanistes caeruleus) breeding in the wild found no evidence of higher fitness in the offspring of more ornamented males (Qvarnström et al. 2006, Hadfield et al. 2006), but in both cases some components of offspring fitness could not be estimated (Charmantier and

Sheldon 2006). Some authors have emphasized that the theoretical models of sexual selection are not mutually exclusive (Kokko et al. 2003); for example, one can easily imagine that sensory bias could lead to the origin of a trait that later becomes exaggerated by a Fisherian process. Nevertheless, testing for the critical distinctions between these models is important, because they differ both in terms of the signal function of ornamental traits and in the predicted outcomes of selection (Andersson and Simmons

2006). Questions about trait origins, and the kinds of traits that are under sexual selection, should be kept distinct from those addressing the processes involved in trait-preference coevolution (Fuller et al. 2005).

Another major unanswered question is how variation in ornamental traits is maintained despite strong directional selection imposed by female choice. This is especially puzzling in lekking species like peafowl. In these systems, the sole benefits of mate preferences are likely genetic ones, for which variation should be depleted rapidly

(Williams 1975). For this reason, the problem has been called the lek paradox (Borgia

1979). Numerous potential explanations have been suggested. For example, the honest signal model of condition-dependent male ornaments could resolve the lek paradox, as long as there is a constant source of genetic variation in condition (Tomkins et al. 2004), 9

which could be provided by cycles of coevolution with parasites (Hamilton and Zuk

1982). Alternatively, condition may involve many genes that provide sufficient variation through mutation (Rowe and Houle 1996). Either mechanism could maintain variation in the ornamental trait as well as the benefits of female mate preferences. Another proposed resolution to the lek paradox is that selection by female choice may not be as uniform or directional as originally supposed. Historically, few studies have addressed variation in

the strength or direction of female mate preferences; most have considered it to be the

result of errors on the part of the female or problems with the research methods (reviewed

in Jennions and Petrie 1997). Yet geographical variation in female mate preferences is

well-documented (reviewed in Widemo and Saether 1999), and recent work suggests that

in some birds, female preferences can vary over time within a population (Chaine and

Lyon 2008).

One reason for variation in female choice could be that female mate preferences

are themselves condition-dependent, an idea for which support is growing (reviewed in

Cotton et al. 2006). However, we still know very little about the costs of mate

preferences. In lekking species like pronghorn (Antilocapra americana), the energetic

costs of female mate sampling behaviour can be substantial (Byers et al. 2005). In peafowl, female-female aggression is most frequent in the territories of preferred males

(Petrie et al. 1992), but direct evidence of the costs of choice in terms of female fitness is lacking (Kotiaho and Puurtinen 2007). We also know very little about the sampling tactics and decision-making process involved in mate choice. Many species, including peafowl, possess either multiple ornamental traits or complex ornaments where any of several different aspects might be considered an independent cue. Do females use 10

multiple cues to make mate choice decisions? There is some evidence that peahens use

both the size and colour of the train ornament to discriminate between males (Loyau et al.

2007a), and multiple aspects of peacock displays may function as honest signals of male

condition (Loyau et al. 2005a). A number of other hypotheses for the function of multiple

ornaments have been proposed, but the consequences for female fitness are unclear

(Candolin 2003) – and these details of female choice may be critical to understanding

ornament evolution (Jennions and Petrie 1997).

Even though the process of mate choice may be more complex than previously

thought, the original models of sexual selection are still being examined in real

organisms, both for the mechanisms involved and their relative importance across taxa

(Andersson and Simmons 2006). These questions have broad relevance as they contribute

to our understanding of animal communication. Previously, for example, the maintenance

of signal honesty under individual selection for deception was an evolutionary paradox

(Williams 1966, Dawkins and Krebs 1978). However, due to Zahavi’s insights on mate

choice, it is now thought that many animal signals are necessarily honest (reviewed in

Maynard Smith and Harper 2003). These questions are also relevant to our understanding

of the generation of biological diversity, since different models of sexual selection predict

various outcomes with respect to diversification (Andersson 1994).

Peafowl and the Measurement of Ornamentation

Darwin left some advice for the biologists who would take up his theory of sexual

selection in the years to come. Although he emphasized the importance of documenting

variation in ornamental traits across closely-related species, to show that they could 11

evolve in “small successive steps”, he also suggested starting with “one or two strongly- characterized cases, for instance that of the peacock” for the best picture of the processes involved (Darwin 1871).

In the past few decades, people have done just that. Peafowl have provided us with some of the most widely-cited evidence for the genetic benefits of mate choice.

Even though peahens receive only sperm from males, the offspring of more-ornamented males grow substantially faster and are far more likely to survive (Petrie 1994). Peafowl

have provided support for the honest signal model of sexual selection, as several studies have suggested that the most-ornamented males are also the healthiest and most likely to

survive (Petrie 1992, Møller and Petrie 2002, Loyau et al. 2005a). They have provided

evidence for the idea that female birds are able to bias the sex ratio of their offspring in

relation to their own body condition and the attractiveness of their mate (Pike and Petrie

2005a, b). Intriguingly, there is even some evidence that peacocks prefer to display near closely-related males, suggesting a role for kin selection in the evolution of lek-based

mating systems (Petrie et al. 1999).

But the trouble with beauty is that it can be hard to measure. Modern biologists

have to make decisions about which aspects of a complex trait should be measured when

studying sexual selection and ornamentation. The perceptual differences between biologists and their study animals can make this difficult, not to mention the problem of

intercorrelations between different aspects of a single ornament. Do peahens use the

elaborate train to choose mates? If so, which component is the important one: length,

colour, the order and distribution of eyespots, or could it be more than one (Candolin

2003)? Though studies of the peacock are in no danger of invalidating materialistic 12

evolution as Darwin may have feared, the recent literature is full of problems that

illustrate the more general difficulties with studies of sexual selection, and the picture of

mate choice in this species is far from clear.

For example, a number of peafowl studies report correlations with the degree of

ornamentation of the train, but there is no consensus on what it actually means for a male to be more ornamented. The original reports of female mate preferences in peafowl used

the number of eyespot feathers in the train to measure male beauty (Petrie et al. 1991),

likely because of an earlier (and erroneous) claim that the number of eyespots increased

with age (Manning 1989; see also Petrie 1993). However, nobody has attempted to

explain how females might be assessing eyespot number. Indeed, a more recent and

extensive study found no effect of the number of eyespots on mating success (Takahashi

et al. 2008), which might seem to simplify things except for the evidence that

experimentally removing eyespots from the train does impact mating success (Petrie and

Halliday 1994).

Studies of peafowl have also been criticized for equating all kinds of different

measures with ornamentation, and either ignoring, or failing to report, other measures

(Takahashi et al. 2008). Positive correlations with mating success have been reported for

such diverse measures as train length, train mass, eyespot size, colour, the density of

eyespots, and even the number of notes in male calls. And, while it is surely important

not to focus attention too narrowly on any one trait (Candolin 2003), the selective use of

these measures in the peafowl literature only makes it more difficult to form a complete

picture. Why is train length related positively to survival (Petrie 1992) but negatively or

not at all to mating success (Loyau et al. 2005b, Petrie et al. 1991, Yasmin and Yahya 13

1996)? How do we explain the fact that eyespot size and train length are related to male

health when female preference seems to be focused on the number and arrangement of

eyespots in the train (Møller and Petrie 2002)?

There are other problems. For example, the widely-cited study demonstrating the

genetic benefits of mate choice in peafowl did not control for the effect of female

investment into eggs (Petrie 1994). Subsequent work indicates that peahens can adjust

their investment in terms of egg size, number and hormone content depending on the

attractiveness of their mate (Loyau et al. 2007b, Petrie and Williams 1993). Nevertheless,

Petrie’s (1994) study remains entrenched as a classic example of evidence for the indirect

genetic benefits of mate choice. Another peafowl study has claimed that male display rate is under sexual selection by female choice (Loyau et al. 2005b). But others have pointed out that merely having females around will cause males to display, and since preferred males are generally visited by females more often, display rate will necessarily be correlated with measures of female preference (Takahashi et al. 2008).

One reason for these difficulties may be that, despite the accessibility of feral populations of peafowl throughout the world, peafowl behaviour and morphology are surprisingly poorly-described. For example, none of the previous studies of eyespot number has offered any explanation of the source of variation among males in this measure. Until we understand the development of the train ornament, the main causes of variation among males, and which aspects are perceived by females, it will be difficult to interpret the previous work on this enigmatic species.

14

Thesis outline

In this thesis, I attempt to provide the foundation for an improved understanding of the

role of visual signaling in peafowl courtship. I use several approaches, including descriptions of basic aspects of peafowl biology that have been overlooked in the past, as well as observational and experimental studies of peafowl in several different feral and

zoo populations.

In Chapter 2, I aim to clarify some of the previous controversies in the literature

with respect to the role of eyespot number in female mate choice. I begin by describing

the geometrical arrangement of eyespot feathers in the trains of adult males. This has

implications for the meaning of variation in eyespot number, which has previously been

considered the main cue for mate choice in peafowl. I repeat an experimental

manipulation of the train ornament to verify the claim by Petrie and Halliday (1994) that

removing a number of eyespots from the peacock’s train reduces male mating success. I

also perform a correlative analysis of the effect of train morphology on mate choice

across populations, in an attempt to reconcile conflict in the previous literature (Petrie et

al. 1991, Takahashi et al. 2008).

To gain a more detailed understanding of courtship signaling in peafowl, in

Chapter 3 I describe the behaviours performed by males when displaying the visual train

ornament. Using behavioural observations of males displaying to live females as well as

an experimental model, I ask whether displaying males orient themselves relative to the

sun to enhance the impact of their eyespot plumage colours. My results point to the

importance of visual signaling, despite the fact that peahens do not count the eyespots of 15 displaying males. Moreover, they illustrate the value of understanding the biology of a species when studying ornaments as complex as the peacock’s train.

References

Andersson M (1994) Sexual selection. Princeton University Press, Princeton

Andersson M, Simmons LW (2006) Sexual selection and mate choice. TREE 21: 296- 302

Arak A (1988) Female mate selection in the natterjack toad: active choice or passive attraction? Behav Ecol Sociobiol 22: 317-327

Arnqvist G (2006) Sensory exploitation and sexual conflict. Phil Trans R Soc B 361: 375-386

Arnqvist G, Rowe L (2005) Sexual conflict. Princeton University Press, Princeton

Basolo AL (1990) Female preference predates the evolution of the sword in swordtail fish. Science 250: 808-810

Bennett PM, Owens IPF (2002) Evolutionary ecology of birds: life histories, mating systems and extinction. Oxford University Press, Oxford

Borgia G (1979) Sexual selection and the evolution of mating systems. In: Blum MS, Blum NA (eds) Sexual selection and reproductive competition in the insects. Academic Press, New York

Burkhardt F, Porter DM, Browne J, Richmond M (1993) The correspondence of Charles Darwin. Cambridge University Press, Cambridge

Burley N (1988) Wild zebra finches have band-colour preferences. Anim Behav 36: 1235-1237

Burley N, Moran N (1979) The significance of age and reproductive experience in the mate preferences of feral pigeons, Columbia livia. Anim Behav 27: 686-698

Burley NT, Symanski R (1998) “A Taste for the Beautiful”: Latent aesthetic mate preferences for white crests in two species of Australian grassfinches. Am Nat 152: 792- 802

Byers JA, Wiseman PA, Jones L, Roffe TJ (2005) A large cost of female mate sampling in pronghorn. Am Nat 166: 661-668 Candolin U (2003) The use of multiple cues in mate choice. Biol Rev 78: 575-595 16

Chaine AS, Lyon BE (2008) Adaptive plasticity in female mate choice dampens sexual selection on male ornaments in the lark bunting. Science 319: 459-462

Charmantier A, Sheldon BC (2006) Testing genetic models of mate choice evolution in the wild. TREE 21: 417-419

Cotton S, Fowler K, Pomiankowski A (2004) Do sexual ornaments demonstrate heightened condition-dependent expression as predicted by the handicap hypothesis? Proc R Soc B 271: 771-783

Cotton S, Small J, Pomiankowski A (2006) Sexual selection and condition-dependent mate preferences. Current Biol 16: 755-765

Cronin H (1991) The ant and the peacock. Cambridge University Press, Cambridge

Darwin C (1859) On the origin of species by means of natural selection, or the preservation of favoured races in the struggle for life. John Murray, London

Darwin C (1871) The descent of man, and selection in relation to sex. John Murray, London

Dawkins R, Krebs JR (1978) Animal signals: information or manipulation? In Krebs JR, Davies NB (eds) Behavioural ecology: an evolutionary approach. Blackwell, Oxford

Fisher RA (1915) The evolution of sexual preference. Eug Rev 7: 184-192

Fisher, RA (1930) The genetical theory of natural selection. Clarendon, Oxford

Fuller RC, Houle D, Travis J (2005) Sensory bias as an explanation for the evolution of mate preferences. Am Nat 166: 437-446

Grafen A (1990a) Biological signals as handicaps. J theor Biol 144: 517-546

Grafen A (1990b) Sexual selection unhandicapped by the Fisher process 144: 473-516

Hadfield JD, Burgess MD, Lord A, Phillimore AB, Clegg SM, Owens IPF (2006) Direct versus indirect sexual selection: genetic basis of colour, size and recruitment in a wild bird. Proc R Soc B 273: 1347–1353

Hamilton WD, Zuk M (1982) Heritable true fitness and bright birds: a role for parasites? Science 218: 384-387

Harel K (2001) When Darwin flopped: the rejection of sexual selection. Sex Cul 5: 29-42

17

Hiraiwa-Hasegawa (2000) The sight of the peacock’s tail makes me sick: the early arguments on sexual selection. J Biosci 25: 11-18

Hoelzer, GA (1989) The good parent process of sexual selection. Anim Behav 38: 1067- 1078.

Jackson CE (2006) Peacock. Reaktion Books, London

Jennions MD, Petrie M (1997) Variation in mate choice and mating preferences: a review of the causes and consequences. Biol Rev 72: 283-327

Kirkpatrick M (1982) Sexual selection and the evolution of female choice. Evol 36: 1-12 Kokko H, Brooks R, Jennions MD, Morley J (2003) The evolution of mate choice and mating biases. Proc Roy Soc B 270: 653-664

Kotiaho JS, Puurtinen M (2007) Mate choice for indirect genetic benefits: scrutiny of the current paradigm. Func Ecol 21: 638-644

Loyau A, Saint Jalme M, Cagniant C, Sorci G (2005a) Multiple sexual advertisements honestly reflect health status in peacocks (Pavo cristatus). Behav Ecol Sociobiol 58: 552- 557

Loyau A, Saint Jalme M, Sorci G (2005b) Intra- and intersexual selection for multiple traits in the peacock (Pavo cristatus). Ethol 111: 810-820

Loyau A, Gomez D, Moureau B, Théry M, Hart NS, Saint Jalme M, Bennett ATD, Sorci G (2007a) Iridescent structurally based coloration of eyespots correlates with mating success in the peacock. Behav Ecol 18: 1123-1131

Loyau A, Saint Jalme M, Mauget R, Sorci G (2007b) Male sexual attractiveness affects the investment of maternal resources into the eggs in peafowl (Pavo cristatus). Behav Ecol Sociobiol 61: 1043-1052

Manning JT (1989) Age-advertisement and the evolution of the peacock’s train. J Evol Biol 2: 379-384

Maynard Smith J, Harper D (2003) Animal signals. Oxford University Press, Oxford

Mead LS, Arnold SJ (2004) Quantitative genetic models of sexual selection. 19: 264-271

Møller AP, Petrie M (2002) Condition dependence, multiple sexual signals and immunocompetence in peacocks. Behav Ecol 13: 248-253

Petrie M (1992) Peacocks with low mating success are more likely to suffer predation. Anim Behav 44: 585-586

18

Petrie M (1993) Do peacock’s trains advertise age? J Evol Biol 6: 443-448

Petrie M (1994) Improved growth and survival of offspring of peacocks with more elaborate trains. Nature 371: 598-599

Petrie M, Halliday T (1994) Experimental and natural changes in the peacock’s (Pavo cristatus) train can affect mating success. Behav Ecol Sociobiol 35: 213-217

Petrie M, Halliday T, Sanders C (1991) Peahens prefer peacocks with elaborate trains. Anim Behav 41: 323-331

Petrie M, Williams A (2003) Peahens lay more eggs for peacocks with larger trains. Proc Roy Soc Lon B 251: 127-131

Petrie M, Hall M, Halliday T, Budgey H, Pierpoint C (1992) Multiple mating in a lekking bird: why do peahens mate with than one male and with the same male more than once? Behav Ecol Sociobiol 31: 349-358

Petrie M, Krupa A, Burke T (1999) Peacocks lek with relatives even in the absence of social and environmental cues. Nature 401: 155-157

Pike TW, Petrie M (2005a) Maternal body condition and plasma hormones affect offspring sex ratio in peafowl. Anim Behav 70: 745-751

Pike TW, Petrie M (2005b) Offspring sex ratio is related to paternal train elaboration and yolk corticosterone in peafowl. Biol Lett 1: 204-207

Proctor HC (1991) Courtship in the water mite Neumania papillator: males capitalize on female adaptations for predation. Anim Behav 42: 589-598

Prum RO (1997) Phylogenetic tests of alternative intersexual selection mechanisms: trait macroevolution in a polygynous clade (Aves: Pipridae). Am Nat 149: 668-692

Qvarnström A, Brommer JE, Gustafsson L (2006) Testing the genetics underlying the co- evolution of mate choice and ornament in the wild. Nature 441: 84–86

Rowe L, Houle D (1996) The lek paradox and the capture of genetic variance by condition dependent traits. Proc Roy Soc B 263: 1415-1421

Ryan MJ (1990) Sexual selection, sensory systems and sensory exploitation. Oxford Surveys in Evol Biol 7: 157-195

Takahashi M, Arita H, Hiraiwa-Hasegawa M, Hasegawa T (2008) Peahens do not prefer peacocks with more elaborate trains. Anim Behav 75: 1209-1219

19

Widemo F, Saether SA (1999) Beauty is in the eye of the beholder: causes and consequences of variation in mating preferences. TREE 14: 26-31

Williams GC (1966) Adaptation and natural selection. Princeton University Press, Princeton

Williams GC (1975) Sex and evolution. Princeton University Press, Princeton

Yasmin S, Yahya HAS (1996) Correlates of mating success in Indian peafowl. Auk 113: 490-492

Zahavi A (1975) Mate selection – a selection for a handicap. J theor Biol 53: 205-214 20

Figure 1.1 The display of the peacock is highly complex. Here, a male is shown displaying to two female peafowl (peahens). 21

Figure 1.2 Close up view of a single eyespot at the tip of a train feather. There may be more than 150 of these eyespot feathers in the train of an adult male; the largest eyespots are about 5 cm across. 22

CHAPTER 2. Train morphology, ornamentation and mate preferences in peafowl

Abstract

Recent studies have reported conflicting evidence for the effect of the number of eyespots in the train on mating success in peafowl (Pavo cristatus). In this paper, we describe the geometric arrangement of eyespot feathers in the trains of adult peacocks to clarify the source of variation therein. We then evaluate the influence of eyespot number and train morphology on mate choice using both observation and experiment. Among males, the geometric pattern of eyespot feathers in the train is consistent, and variation among adult males in the number of eyespots appears to be chiefly due to feather breakage or loss.

Although we confirm previous reports that removing a large number (≥20) of feathers with eyespots decreases male reproductive success, this modification produces a train ornament that is outside the range of normal adult male morphology in all feral peafowl populations studied so far. Furthermore, we find considerable variation in the mating success of feral peacocks that cannot be explained by variation in eyespot number. Thus, mate choice in peafowl may be more complex than previously thought: the overall size of the train ornament may allow females to reject males that have experienced extensive feather loss, but females must use some other cue to choose among adult males with normal train morphology.

Introduction

The iridescent train of the lek-breeding peacock is an obvious visual display that has been thought to result from sexual selection for increased ornament size and complexity

(Darwin 1871). Studies of feral peafowl populations in the 1990s suggested that females 23

preferred to mate with males that had a greater number of eyespot feathers in their trains

(Petrie et al. 1991, Petrie and Halliday 1994), lending support to Darwin’s idea that the

elaboration of the train is due to selection in the form of female mate preferences alone.

More recently, conflicting evidence for the relation between train morphology and mating

success has led some to suggest that the train is an obsolete remnant of selection past, not

currently used by peahens to discriminate among males (Takahashi et al. 2008).

The full train of the adult peacock is highly complex, comprised of upper-tail

coverts that have been modified into four specialized feather types (Figure 2.1; Sharma

1974). First, the outermost edge of the raised train is lined by the longest “fish-tail” feathers (see Figure 1b in Manning 1989). Second, bilaterally symmetrical major eyespot feathers are distributed regularly throughout the central part of the train and form the majority of its feathers (Figure 1a in Manning 1989). Finally, the lower edges of the erect train are delimited by both curved asymmetrical sword feathers (Figure 1c in Manning

1989) and curved asymmetrical minor eyespot feathers (Sharma 1974). All four feather types are moulted annually over an 8-week period in the autumn, following the breeding season, and are regrown over an 8-12-wk period in the spring, before the breeding season

(Sharma 1974). Sharma (1974) reported that adult male peafowl moulting in captivity have about 169 eyespot feathers in total (major and minor). During the breeding season, free-range peacocks have an average of about 150 eyespots (Table 2.1, calculated using known feral individuals from previous studies; SE = 0.9, 95% CI 148-152, n = 176).

Males do not grow their full train until they reach 3-4 years of age, although some subadult males grow a highly stunted and irregular train with 5-40 eyespot feathers (R

Dakin, personal observations). 24

Manning (1989) was the first to consider total eyespot number (both major and minor eyespots) as a measure of ornamentation in peacocks. Based on a positive relation between male age and the number of eyespots, he suggested that eyespot number could be an honest advertisement of male age. However, two subsequent studies have found that the number of eyespots does not reliably increase with age (Petrie 1993, Takahashi et al. 2008), and that it often decreases (Petrie 1993). Nevertheless, since Manning (1989) proposed the age-advertisement hypothesis, eyespot number has often been used a measure of ornament elaborateness. Thus, Manning and Hartley (1991) reported that eyespot number correlates strongly with train symmetry in feral peafowl, such that males with fewer eyespots have a greater difference in eyespot number between the left and right sides of their trains. They interpreted this as a positive relation between symmetry and degree of ornamentation, arguing that it provides evidence for the idea that the symmetry of sexual ornaments signals individual quality. In addition, both observational

and experimental studies have indicated that eyespot number may be an important cue for

mate choice by peahens, since the number of eyespots in the train correlates with the

number of copulations a male achieves (Petrie et al. 1991, Loyau et al. 2005b), and

removing 20 of the outermost eyespots reduces male copulation success (Petrie and

Halliday 1994).

More recently, Loyau et al. (2005a) reported that eyespot number relates to male immunological health, based on a non-significant correlation with eyespot number and the number of heterophils in the blood as well as a significant relation between eyespot number and the rate of recovery from an immune challenge. Loyau et al. (2005a) interpreted this to mean that more ornamented males possess better immune system 25

genes. Moreover, peahens appear to bias the sex ratio of their offspring towards females

when they are mated, experimentally, with males from which 20 eyespots have been

removed (Pike and Petrie 2005), and another study has demonstrated that female

investment of resources into eggs is related to the number of eyespots on the male with

whom she has been experimentally mated (Loyau et al. 2007b).

If eyespot number signals male quality, by what mechanism does this process

work? Are higher quality males growing more or losing fewer eyespots? Recent evidence

from a long-term study suggests, in fact, that peahens may not use eyespot number to

choose a mate (Takahashi et al. 2008). Indeed, several studies have failed to find any

relations between eyespot number and measures of male quality or female reproductive

investment (e.g., Petrie et al. 1996, Møller and Petrie 2002, Petrie and Williams 1993).

In the present study, we shed light on the apparent contradictions in the literature

with respect to the importance of eyespot number. First, we describe natural variation in

the number and geometric arrangement of eyespots in three feral peafowl populations in

North America. We then evaluate the relation between eyespot number and mating success by (i) repeating the original experiment performed by Petrie and Halliday (1994) in which the number of eyespots was experimentally reduced on some males, and (ii) examining the relations between natural eyespot number and measures of mating success in three North American peafowl populations.

26

Methods and Materials

Field methods

We studied three separate populations of peafowl at (i) Assiniboine Park Zoo (APZ) in

Winnipeg, MB, Canada (April-May 2007) where about 60 peafowl are free-ranging over

50 ha of pens and woodland; (ii) Toronto Zoo (TZ) in Toronto, ON, Canada (April-June

2007), where about 30 peafowl are free ranging over 250 ha of pens and woodland; and

(iii) Los Angeles Arboretum (LAA) in Arcadia, CA, USA (February-April 2008), where

>100 peafowl live in 50 ha of parklands as well as surrounding residential areas. The

APZ and TZ populations are semi-captive, housed in large indoor pens during the coldest

winter months (December-March), whereas the LAA birds are feral year-round. All three populations are comprised of wild-type, lek-breeding individuals and thus are expected to be representative of the species in the wild.

Measurement of train morphology

Birds were caught prior to the start of the breeding season (April at APZ and TZ;

February at LAA) and marked for individual identification with a large numbered leg band. At the time of capture, we measured body mass to the nearest 0.01 kg and tarsus length to the nearest mm, as well as male train length as the length of the longest “fish- tail” feather (Petrie et al. 1991, Loyau et al. 2005). We removed 5 eyespots from each captured male for a separate study of plumage colour by cutting the rachis immediately

below the eyespot for 5 of the longest major eyespot feathers.

To quantify the number and arrangement of eyespots remaining in the train

ornament, we digitally photographed displaying males during the breeding season, 1-4 27 weeks after the initial capture. We used ImageJ (Rasband 2007) and Adobe Photoshop

(Adobe Systems 2002) to count the number of eyespot feathers, and to map the locations of eyespots throughout the train. To describe the normal arrangement of eyespot feathers in the train, we photographed 4 additional adult males at LAA that had not been captured; these birds would have lost eyespots only by natural means. Although they were not banded, we could distinguish these birds by morphological features (e.g., missing neck plumage, missing several of fish-tail train feathers) and by fidelity to specific display courts on the leks.

Eyespot removal experiment

We tested the effect of removing a large number of eyespots (following Petrie and

Halliday 1994) in the two semi-captive zoo populations (APZ and TZ). For our treatment males, we cut the rachis immediately below the eyespot for 20 of the longest major eyespot feathers (APZ: n = 6 including 1 male from which only 15 eyespots were removed; TZ: n = 1), in addition to the 5 eyespots removed from all adult males for colour analysis. We chose the treatment group haphazardly as the first males captured with a hand net in their winter holding quarters. The control males in the APZ (n = 8) and

TZ populations (n = 4) were handled the same way as treatment males, except that we removed only the 5 eyespot feathers for colour analysis. Thus, our treatment males had

20 additional eyespots removed relative to control males.

28

Behavioural observations

To measure mating success, we observed male leks during 07:00-12:00 and 16:00-18:00

local times, since these periods correspond to peak lekking activity (Petrie et al. 1991, R

Dakin personal observations). We conducted focal watches of 1-4 males at a time,

depending on visibility at different lek sites, for periods ranging from 0.5-2.5 hours. We

performed a total of 315 h of lek observations across three populations (APZ: 80 h, TZ:

115 h, LAA: 120 h). During this time, we recorded the duration of male attendance at

their display courts, the duration of male train displays (i.e., train erect), the number of

train rattling bouts performed by males (always in the presence of females), all copulation

attempts (“hoot-dashes”; Petrie et al. 1991), and successful copulations. In total, we observed 324 copulation attempts, 64 of which ended in successful copulations (34 at

APZ, 6 at TZ, 24 at LAA).

During focal watches, we counted the number of females present at 5-min intervals, noting which male each female was closest to, and estimating the distance (m) between each female visitor and that male. We used these observations to calculate two female visitation scores for each male: (i) a total visitation score as the sum of all recorded instances when a female was < 5 m from a male, and (ii) a displaying visitation score using only instances when a female was < 5 m away from a male when his train was raised.

We attempted to distribute our focal watches equally among the different lek sites. The mean total time that focal males were in view was 8.3 h ±1.4 SE for APZ

(range 1.0-16.2, n = 14 males), 14.9 h ± 2.2 SE for TZ (range 10.5-22.3, n = 5 males) and

15.3 h ± 2.3 SE for LAA (range 8.3-22.2, n = 6 males); the variation here is due to 29

differences in territory attendance rather than a biased distribution of observation times at leks within populations. To account for the fact that different males were under observation for different amounts of time, we calculated individual rates for male display, total female visitation, displaying female visitation, male train-rattling bouts, copulation attempts, and copulations.

We used rates of train-rattling bouts, female visitation, copulation attempts and successful copulation as measures of male mating success. Copulation only occurs following the train-rattling display, and males only perform this display when female visitors are observing at close range, so the rate of train-rattling bouts should be correlated with female preference (Takahashi et al. 2008, R Dakin personal observations).

The rate at which males display their trains (i.e., male display rate) tends to be correlated with total female visitation rate, since the mere presence of females generally elicits males to raise their trains (Takahashi et al. 2008; this study: r = 0.63, p < 0.001, n = 25).

Thus, to assess the male tendency to display independent of female visitation rate, we calculated the residuals of display rate regressed on total female visitation rate (referred to as residual display rate).

Data analysis

We found little difference between our study populations and thus we pool data from different sites for analysis. For example, prior to removing any tail feathers, the males from all three study populations had similar train morphology (summarized in Table 2.2), and there was no difference among populations in the total number of eyespots per train.

LAA males did have significantly shorter trains (mean = 120 cm) than those at APZ and 30

TZ (mean = 146, 138 cm respectively; Tukey-Kramer HSD, p < 0.05), but LAA males were captured for train measurements in February, before train plumage growth is complete (Sharma 1974). Thus it is likely that the train feathers of LAA males were not quite fully grown at capture. There was also no difference among control males from all three sites in most behavioural measures related to mating (summarized in Table 2.2), except a significant difference in the rate of copulation attempts, with control males at

APZ attempting copulations significantly more often than those at LAA (Tukey-Kramer

HSD, p < 0.05) but not more often than at TZ (means = 1.68, 0.58, and 0.79 attempts per hour, respectively). Thus, we pool data from APZ and TZ for analyses of the eyespot removal experiment, and from all three populations when examining the relations between train morphology and mating success for control males (except train length).

We checked for equality of variance for all reported t-test and ANOVA comparisons, and where data departed considerably from that assumption and no suitable transformation could be applied, we use nonparametric analyses.

Results

Train morphology and the arrangement of eyespots

The train morphology of the peacocks that we studied was consistent with that of previous studies. The total number of eyespots on the train, prior to eyespot removal, ranged from 145-168 (mean = 156 ± 1.0 SE, n = 40), similar to other populations (Table

2.1). Train length ranged from 109-153 cm (mean 134 cm ± 2.0 SE, n = 36), also consistent with previous reports (Petrie and Cotgreave 1996). There was no relation 31

between train length and the total number of eyespots across our three populations

(Pearson correlation, r = -0.10, p = 0.55, n = 36).

The normal arrangement of eyespots in the erect trains of four males is

summarized in Table 2.3. In these males, the major eyespots were arranged in 11

bilaterally symmetrical rows. Each row of eyespots is comprised of feathers of

approximately the same length coming from a different row of feather primordia in the

upper tail coverts, such that the rows of longer feathers are behind (and thus do not obscure) the shorter feathers in front. The innermost (and shortest) 7 rows meet at the vertical midline of the erect train (Figure 2.2a-b), and these generally have an odd number of eyespots. The innermost row has just one eyespot that is sometimes obscured in head-on photos, and the number of eyespots increases in the longer posterior rows

(Table 2.3). The remaining 4 outermost (and longest) rows do not meet at the center line.

The number of eyespots in each of these rows is bilaterally symmetrical and decreases toward the periphery (Table 2.3). All four uncaptured males that we photographed had exactly 11 rows of major eyespots and nearly the same number of major eyespots (131 for 3 of 4 males; Table 2.3).

The bottom of the train is delimited by the asymmetrical minor eyespot feathers and sword feathers (Figure 2.1, 2.2), both of which are curved and arranged with their concave edges oriented downwards such that a thick fringe of green barbs borders the bottom edge of the train. Our counts of the number of minor eyespot feathers are more

variable than those of major eyespot feathers (Table 2.3), but this may due to the use of

photographs to count the eyespots, since minor eyespots can often be obscured behind

other feathers in the erect train. 32

Eyespot removal experiment

There was no difference between the treatment and control males from the APZ and TZ

populations with respect to train length (t-test, t = 0.71, p = 0.49, n = 7, 12) or the total

number of eyespots in the train (t = 0.60, p = 0.56), prior to the removal treatment. There was also no difference between these two groups in measures of general body size or condition, including body mass (t = 0.63, p = 0.54), average tarsus length (t = 0.36, p =

0.72) or body mass when controlling for average tarsus length as a measure of condition

(ANCOVA, t = 1.01, p = 0.33).

Of course, the removal treatment significantly decreased the total number of

eyespots when compared to control males (treatment mean = 128 eyespots ± 2.3 SE,

control mean = 149 ± 1.8 SE; t = 7.21, p < 0.0001) and the number of major eyespot

feathers (treatment mean = 103 ± 2.1 SE, control mean = 124 ± 1.6 SE; t = 8.28, p <

0.0001). The treatment also reduced the number of rows of major eyespots by about one

row on average (Figure 2.2; treatment = 10.0 rows ± 0.16 SE, control = 10.9 rows ± 0.12

SE; t = 4.7, p = 0.0002). The slight variation in number of rows of control males was

likely due to the removal of 5 feathers from all captured males for a separate study of

plumage colour (see Methods).

The experimental treatment did not have any effect on male display rate, male

tendency to display, or female visitation rate (Table 2.4). However, eyespot removal did

have a significant effect on indices of male mating success relative to control males from

the same populations, including the rate of train-rattling bouts, copulation attempts, and

successful copulations (Figure 2.3). These patterns were the same whether data were 33

analyzed for the pooled populations (Figure 2.3), or just for the larger APZ population

(Table 2.3).

Train morphology and mating success

The behaviour and mating success of control males, pooled across the three populations, was not generally related to natural variation in train morphology (Table 2.5; Figure 2.4).

For example, there were no relations between any of the behavioural variables and the length of the train, the number of eyespots, or the number of rows of major eyespots

(categorized as either 11 or < 11), except for a significant positive relation between male display rate and train length (Figure 2.4c; Pearson correlation, r = 0.71, p = 0.01, n = 12).

Results were similar when using ANCOVA to control for any differences between our three study populations. For eyespot number, the interaction with population was not significant in any of the models for the behavioural variables (all p > 0.30, n =

18). Removing the interaction term, eyespot number was not related to any of the behavioural variables when controlling for population (ANCOVAs, all p > 0.15, n = 18).

For train length, the interaction with population was also not significant in all of our models (all p > 0.10, n = 16). Train length was not related to the rate of female visitation, train-rattling, copulation attempts or copulations (ANCOVAs, all p > 0.55, n = 16).

Although there were significant relations between train length and both displaying rate and residual displaying rate (p = 0.05 and 0.04 respectively, n = 16), this was not surprising given that this effect was also seen within the APZ population (Table 2.5).

34

Discussion

Our results are all consistent with the seemingly contradictory results previously reported

for female mate choice in peafowl. Like Takahashi et al. (2008), we found no effect of natural variation in train morphology, including eyespot number, on male mating success.

Although we did find a significant correlation between male display rate and train length, none of our measures of female preference were correlated with male train length. We also confirmed the experimental result of Petrie and Halliday (1994) that removing a large number of eyespots reduces male mating success.

However, our data point to a possible explanation for these apparent contradictions with respect to eyespot number. The regular arrangement of major eyespots in 11 feather rows was remarkably consistent across individuals in this study.

All uncaptured males and nearly all control males had 11 rows of major eyespots. The experimental treatment of removing an additional 20 eyespots decreased the number of

rows of eyespots in the train, modifying train morphology well beyond the normal range

of variation seen in adult males (Figure 2.2). There are several reasons why females may discriminate against mating with males missing such a large number of eyespots. For example, perhaps males missing the outer rows appear to be subadult individuals.

Alternatively, those males may appear to have experienced extreme feather loss through predation or physical stress. Nevertheless, several studies have now reported substantial variation in reproductive success among feral males that cannot be explained by variation in the number of eyespots (this study, Takahashi et al. 2008, Loyau et al. 2007a). In our study, for example, 2 of the 7 removal treatment males copulated successfully, with one male achieving 5 copulations and ranking 2nd in copulation success for the removal 35 experiment. Thus this male was apparently preferred, despite having only 122 eyespots, indicating that preference must have been based on some other cue.

While it is possible that we failed to capture enough of the normal variation in eyespot number due to small sample sizes, the range we observed is consistent with that of nearly all other studies of feral peafowl including one where wild birds from India were allowed to moult in captivity (Sharma 1974; Table 2.1). Several studies have reported that a small percentage of adult males have, like our removal treatment males, <

130 eyespots. However, these individuals are relatively rare: for example, only 6% of known males from previous studies have < 130 eyespots, and 95% of known males are within the range of 118-169 eyespots (Table 2.1). It is unclear what having < 130 eyespots means for a peacock, since we have not observed any of these males. It seems unlikely that these are subadult males with an intermediate, but not yet fully-developed train as subadult males usually have < 50 eyespot feathers, and they do not maintain display territories. Another more likely explanation is that these relatively rare individuals with < 130 eyespots have experienced the extensive loss of 30-60 eyespot feathers through physical stress or predation, or have failed to grow a complete train due to poor health or condition. Again, these would all be good reasons for females to avoid mating with males that lack the full adult train.

Our description of the arrangement of eyespots in the train suggests an explanation for the primary source of variation in eyespot number. As mentioned above, one possible source could be the extent to which the rows of feathers have grown in or developed. However, this explanation is unlikely for the populations we studied, since we observed that the pattern of 11 eyespot feather rows was conserved. Furthermore, train 36

length is doubtless an index of the extent of feather growth, and we found no relation

between length and the total number of eyespot feathers despite substantial variation in

train length across our three populations.

Apart from errors in counting from photographs due to feathers being obscured, the only other possible cause of variation in eyespot number is feather breakage and loss.

This is likely the main source of variation in this study for two reasons. First, the

breakage and loss of eyespot feathers is not unusual, and males are often observed with

broken eyespot feathers that have not yet fallen out during the breeding season (R Dakin,

personal observations). Second, the variation in eyespot number among males studied

here was generally due to a few eyespots missing from specific positions within each row

rather a difference in the number of eyespot rows (Table 2.3). Thus, our observations

suggest that the rows of eyespots are formed by determinate feather growth tracts. As

with the development of the wing primaries or the rectrices, the maximum number of

dermal papillae that will eventually produce eyespot feathers may be genetically

determined in all males at the embryo stage (Lillie 1942). If this is the case, individual

variation in the growth rate of the train plumage could cause some differences with

respect to the degree to which the shortest eyespot feathers have developed and are thus

visible during the winter growing season (e.g., see Sharma 1974). However, by spring,

most of the variation in the number of eyespots should be due to loss or breakage of

individual feathers, as we observed.

If feather loss is the main source of variation in eyespot number during the

breeding season, the reports of a correlation between the number of eyespots and train

symmetry (Manning and Hartley 1991) are almost certainly spurious. Haphazard feather 37 loss would necessarily cause both number of eyespots and the symmetry of the arrangement to decrease simultaneously. In addition, Petrie et al. (2008) may have failed to find significant heritability in number of eyespots because there is no genetic variation in the pattern of feather growth, and feather loss is likely influenced primarily by environmental factors.

Interestingly, the removal of males with fewer than 130 or 140 eyespots from the analysis in Loyau et al. (2005b; see Figure 2 therein) eliminates any relation between mating success and eyespot number (males > 130 only: Pearson correlation, r = 0.24, p =

0.25, n = 24; males >140 only: r = 0.20, p = 0.38, n = 22), consistent with our study and with Takahashi et al. (2008). Thus, to date, no study with a sample size of more than 10 males has found a significant relation between eyespot number and mating success among individuals in the normal 140-160 eyespot range. The correlation originally reported by Petrie et al. (1991) between eyespot number and mating success with a relatively small sample of 10 males in this range may have therefore been spurious.

From our findings, we conclude that peafowl mate choice may be more complex than previously thought: females may first discriminate against males with less than ~130 eyespots, then choose among remaining males based on other aspects of their plumage and behaviour, though Takahashi et al. (2008) have suggested otherwise. Indeed, given the natural history of peafowl, where visual observation of the train appears to be a critical component of courtship, it seems probable that other visual aspects of the train are critical for mate choice in this puzzling species (Loyau et al. 2007a, Dakin and

Montgomerie unpublished data).

38

References

Darwin C (1871) The descent of man, and selection in relation to sex. John Murray, London.

Kimball RT, Braun EL, Ligon JD, Lucchini V, Randi E (2001). A molecular phylogeny of the peacock-pheasants (Galliformes: Polyplectron spp.) indicates loss and reduction of ornamental traits and display behaviours. Biol J Linn Soc 73: 187-198

Lillie FR (1942) On the development of feathers. Biol Rev 17: 247-266

Loyau A, Gomez D, Moureau B, Théry M, Hart NS, Saint Jalme M, Bennett ATD, Sorci G (2007a) Iridescent structurally based coloration of eyespots correlates with mating success in the peacock. Behav Ecol 18: 1123-1131

Loyau A, Saint Jalme M, Mauget R, Sorci G (2007b) Male sexual attractiveness affects the investment of maternal resources into the eggs in peafowl (Pavo cristatus). Behav Ecol Sociobiol 61: 1043-1052

Loyau A, Saint Jalme M, Cagniant C, Sorci G (2005a) Multiple sexual advertisements honestly reflect health status in peacocks (Pavo cristatus). Behav Ecol Sociobiol 58: 552- 557

Loyau A, Saint Jalme M, Sorci G (2005b) Intra- and intersexual selection for multiple traits in the peacock (Pavo cristatus). Ethol 111: 810-820

Manning JT (1989) Age-advertisement and the evolution of the peacock’s train. J Evol Biol 2: 379-384

Manning JT, Hartley MA (1991) Symmetry and ornamentation are correlated in the peacock’s train. Anim Behav 42: 1020-1021

Møller AP, Petrie M (2002) Condition dependence, multiple sexual signals and immunocompetence in peacocks. Behav Ecol 13: 248-253

Møller AP, Pomiankowski A (1993) Why have birds got multiple sexual ornaments? Behav Ecol Sociobiol 32: 167-176

Petrie M (1993) Do peacock’s trains advertise age? J Evol Biol 6: 443-448

Petrie M (1994) Improved growth and survival of offspring of peacocks with more elaborate trains. Nature 371: 598-599

Petrie M, Cotgreave P, Pike TW (2008) Variation in the peacock’s train shows a genetic component. Genetica doi: 10.1007/s10709-007-9211-0

39

Petrie M, Cotgreave P, Stewart I (1996) Variation in the train morphology of peacocks (Pavo cristatus). J Zool 238: 365-371

Petrie M, Halliday T (1994) Experimental and natural changes in the peacock’s (Pavo cristatus) train can affect mating success. Behav Ecol Sociobiol 35: 213-217

Petrie M, Halliday T, Sanders C (1991) Peahens prefer peacocks with elaborate trains. Anim Behav 41: 323-331

Petrie M, Williams A (2003) Peahens lay more eggs for peacocks with larger trains. Proc Roy Soc Lon B 251: 127-131

Pike TW, Petrie M (2005) Offspring sex ratio is related to paternal train elaboration and yolk corticosterone in peafowl. Biol Lett 1: 204-207

Rasband WS (2007) ImageJ 1.37v. US National Institutes of Health, Bethesda, Maryland. http://rsb.info.nih.gov/ij/

Sharma IK (1974) Ecological studies of the plumes of the peacock (Pavo cristatus). Condor 76: 344-346

Takahashi M, Arita H, Hiraiwa-Hasegawa M, Hasegawa T (2008) Peahens do not prefer peacocks with more elaborate trains. Anim Behav 75: 1209-1219

Table 2.1 Previous studies of the number of eyespot feathers in trains of adult peafowl. Entries are left blank where the authors did not

provide the relevant data in the text and it could not be taken from a figure.

Number of Eyespots Population n Mean SE CV Range % < 130 Main result with respect to eyespot no. Approach 1974 Sharma captive 169 Description of train feather moult Observational 1989 Manning farm 16 * 142 3.7 10.3 110-161 19 % Eyespot no. related to age Observational 1991 Manning and Hartley feral 17 * 146 3.4 9.6 117-167 18 % Eyespot no. related to symmetry of eyespot arrangement Observational 1991 Petrie et al. feral 10 * 151 2.1 4.3 141-161 0 Mating success related to eyespot no. Observational 1992 Petrie feral 17 153 Probability of predation negatively related to eyespot no. Observational 1993 Petrie and Williams feral 8 144 4.3 8.5 Female egg production not related to eyespot no. Experimental 1994 Petrie and Halliday feral 24 * 146 2.0 6.8 123-167 4 % Reduction in eyespot no. decreases mating success Experimental 1996 Petrie et al. feral 17 * 160 2.3 5.9 143-177 0 Eyespot no. not related to body mass or muscle size Observational 2005a Loyau et al. feral 24 * 152 1.4 4.5 137-162 0 Eyespot no. related to immunological health Observational 2005b Loyau et al. feral 28 * 144 2.3 8.6 105-162 14 % Mating success related to eyespot no. Observational 2005 Pike and Petrie farm 21 158 2.4 7.1 Manipulating eyespot no. affects offspring sex ratio Experimental 2007 Loyau et al. feral 24 154 1.0 3.1 146-163 0 Female egg investment related to eyespot no. Experimental 2008 Petrie et al. feral 6 139 3.2 5.6 Eyespot no. not significantly heritable Observational feral – 1995 20 150 1.9 5.6 127-164 feral – 1996 20 146 2.1 6.5 118-159 feral – 1997 20 150 1.2 3.6 139-158 0 2008 Takahashi et al. feral – 1998 30 147 1.1 4.0 127-157 Mating success not related to eyespot no. Observational feral – 1999 36 148 0.9 3.6 139-158 0 feral – 2000 37 149 1.3 5.1 125-159 feral – 2001 37 151 0.8 3.2 140-160 0 This study captive and feral 40 * 156 1.0 3.9 145-168 0

ALL KNOWN INDIVIDUALS 176 150 0.9 7.6 105-177 6% * Denotes studies where data for individual males were available, used to calculate descriptive statistics (mean, SE, CV, range, % with < 130 eyespots) for a sample of known individual males. None of these studies were of the same population in the same year, although two (Loyau 2005a and Loyau 2005b) were of

the same population in two different years. 40 41

Table 2.2 Comparisons of train morphology and behaviour of control males from three populations (APZ: n = 8, TZ: n = 4, LAA: n = 6). For all ANOVAs the df are 2, 15.

ANOVA F p Train morphology Train length 24.8 < 0.0001 Number of eyespots 1.95 0.18

Mating success Display rate 0.54 0.59 Displaying female visitation rate 0.83 0.45 Rate of train-rattling bouts 0.47 0.64 Rate of copulation attempts 5.02 0.02 Copulation rate 0.87 0.44

Table 2.3 Regular arrangement of eyespots in rows on the trains of four normal adult peacocks.

Major Eyespots Minor eyespots Inner rows* Outer rows* R1 R2 R3 R4 R5 R6 R7 R8 R9 R10 R11 Total Left Right TOTAL Male A 1 5 9 13 17 20 23 16 12 9 6 131 14 12 = 157

B 1 5 11 17 21 20 19 15 10 8 4 131 16 16 = 163

C 1 5 11 17 21 19 18 13 10 8 4 127 12 10 = 149

D 1 5 9 13 19 22 19 15 12 10 6 131 11 12 = 154

* Inner refers to the shortest, central rows that meet along the vertical mid-line of the train; the longest outer rows do not meet.

42 43

Table 2.4 Effect of eyespot removal on male behaviour and mating success. Comparisons are presented using data pooled for the APZ and TZ populations, and separately for the

APZ population. All analyses are t-tests, unless otherwise noted; significant differences are highlighted in bold.

APZ and TZ APZ Only (n = 7 treatment, 12 control) (n = 6 treatment, 8 control) Test statistic p Test statistic p Display rate 1.94 0.07 1.88 0.08 Residual display rate 1.29 0.21 1.11 0.29 Female visitation rate 1.23 1 0.22 1 1.29 0.22 Rate of train-rattling 2.20 0.04 2.45 0.03 Rate of copulation attempts 2.55 0.02 2.99 0.01 Copulation rate 1.93 1 0.05 1 2.05 1 0.04 1 1 Nonparametric Wilcoxon test where data depart considerably from the assumption of equality of variance.

Table 2.5 Relations between mating success and train morphology for control males from three populations. Significant relations are highlighted in bold. Analyses for train length and total eyespots are correlations, whereas those for number of feather rows compare males with 11 rows in their train (n = 15) to those with < 11 rows (n = 3). Birds from the LAA population are excluded from analyses of train length (see Methods).

Train Length Total Eyespots No. Feather Rows (n = 12) (n = 18) (n = 18) r p r p t p Display rate 0.71 0.01 0.28 0.26 0.53 0.61 Residual displaying rate 0.50 0.10 0.33 0.19 1.15 0.27 Displaying female visitation rate 0.40 0.20 -0.04 0.86 0.14 0.89 Rate of train-rattling 0.55 0.06 -0.12 0.63 1.32 0.20 Rate of copulation attempts 0.25 0.44 -0.31 0.21 0.87 0.40 Copulation rate 0.22 0.50 -0.30 0.23 0.87 0.40

44 45

fish-tail

major eyespot

minor eyespot (asymmetrical)

sword

Figure 2.1 Four specialized types of upper-tail covert feathers comprise the peacock’s train. 46 a) b)

c) d)

Figure 2.2 Train morphology of (a-b) two normal adult peacocks, and (c-d) two males from which the 25 outermost eyespot feathers had been removed. Eyespots are circled and joined to illustrate the pattern of feather rows described in the text; the minor eyespots are grouped separately from the rows of major eyespots. 47 a) b)

3.0 1.0

2.5 0.8 2.0 0.6 1.5 0.4 1.0

Copulations (/h) 0.2 0.5 Copulation attempts (/h) 0.0 0.0

Control Treatment Control Treatment

Figure 2.3 Effect of removing the 20 outermost eyespot feathers (Treatment) on the rate of (a) male copulation attempts and (b) copulations for peacocks in the APZ and TZ populations (n = 12 for control group; n = 7 for treatment group). 48 a) b) 1.0 2.5 0.8 2.0 0.6 1.5 0.4 1.0

Copulations (/h) 0.2 0.5 0.0 Copulation attempts (/h) 0.0

140 145 150 155 160 165 140 145 150 155 160 165 Number of eyespots Number of eyespots c)

0.5

0.4

0.3

0.2

0.1

Display rate (proportion time) 0.0

130 135 140 145 150 155 Train length (cm)

Figure 2.4 Relations between (a) the number of eyespots and the rate of copulation attempts and (b) the number of eyespots and the rate of successful copulations among control males (n = 18). The significant relation between (c) train length and display rate (n = 12) is shown with the Model II regression line. Symbols denote source populations (APZ = blue crosses, TZ = red squares, LAA = green triangles). 49

CHAPTER 3. Peacocks orient their displays towards the sun

Abstract

Males of many bird species perform complex displays during courtship. These behaviours may signal athletic performance or they may serve to highlight morphological ornaments, but empirical studies of their function are rare. Here, we describe the two display behaviours of male peafowl (Pavo cristatus), focusing particularly on male orientation relative to the position of the sun to determine the importance of light during courtship. When their trains are erect, peacocks do not simply face towards the nearest female. Instead, male orientation and the type of display behaviour performed depend on the positions of both the sun and the female. During the “wing-shaking” display, females are generally positioned behind the displaying male, and male orientation is not significantly different from random. However, during the pre-copulatory “train-rattling” display, males are on average directed at about 45° to the right of the sun with the female positioned directly in front, suggesting that this behaviour is involved in the communication of a visual signal. The results of a model presentation experiment confirm that courting peacocks orient towards the sun, and that males are more likely to perform wing-shaking behaviour when the female is positioned on the shaded side of their displays. This study underscores the importance of light for visual signaling in peacock courtship, and we suggest that an angle of 45° relative to the sun might allow males to enhance the appearance of their iridescent eyespot feathers. Our results also indicate that the function of the wing-shaking display may be to move female observers into a preferred viewing geometry with respect to the sun, to maximize effectiveness of iridescent colour signals. 50

Introduction

The use of ritualized behaviour during courtship is a common feature of animal

communication (Tinbergen 1952). For birds, these courtship displays can reach extreme

and often baffling complexity: witness the deliberate footwork of blue-footed boobies

(Sula nebouxii), the co-ordinated leaps of displaying Chiroxiphia manakins, and the

varied and bizarre dances performed by male birds of paradise (genera Parotia and

Paradisaea). These displays may function in species recognition, for mate attraction, as

signals of athletic performance, or to enhance the visibility and impact of morphological

ornaments (Tinbergen 1952, Maynard Smith and Harper 2003). Courtship display

behaviours have been described in detail for a number of lek-breeding birds (e.g.,

Corapipo gutturalis, Prum 1985; Masius chrysopterus, Prum and Johnson 1987; Parotia

carolae, Scholes 2006; Parotia wahnesi, Scholes 2008), and a few studies have examined

the patterns of variation in dance repertoire size across species (Prum 1997, Johnson

2000). However, tests of the proposed function of specific behaviours employed during

courtship displays are rare (but see Kelso and Martins 2008).

Evidence that selection has favoured signal efficacy can be seen in many of the morphological and behavioural aspects of avian courtship. For example, comparative studies have shown that interspecific variation in sexually-selected plumage colours is related to variation in the light environment. Thus, among the rainforest birds of French

Guiana, males generally exhibit more conspicuous colour patterns than females in the context of the light environment in which the males display (Gomez and Théry 2004).

Similarly, among several Australasian bird families (e.g., Psittacidae, Meliphagidae,

Estrildidae, Maluridae, Petroicidae, Pachycephalinae), male plumage hues tend to be 51

highly detectable in the light environment where they display (McNaught and Owens

2002), and the male plumages of neotropical manakins (family Pipridae) contrast particularly against the background of their courtship arenas (Doucet et al. 2007).

In lek-breeding species, where males use display territories solely for courtship signaling, both the location and timing of display also appears to be selected for signal efficacy. For example, several neotropical lekking birds (Rupicola rupicola, Corapipo gutturalis, Lepidothrix serena) display at times when the contrast of male plumage against the visual background is maximized (Endler and Théry 1996). Similarly, the vertical placement of display arenas in several lekking manakins corresponds with the

locations predicted to maximize plumage contrast against the background (Heindl and

Winkler 2003). There is also some evidence from the white-throated manakin (Corapipo

gutturalis) that female visitation rates are highest at the best-lit display territories, where

males may be most conspicuous and subtle differences in their plumage colours most

easily distinguished by females (Théry and Vehrencamp 1995).

In the present study, we provide a quantitative description of some of the display

behaviour of free-ranging peacocks (Pavo cristatus) in relation to both the light

environment and the relative positions of females being courted. Readily accessible in

zoos and parks worldwide, lek-mating peacocks perform an elaborate courtship ritual

involving calls and the presentation of their train, a set of highly-elongated upper-tail

coverts, each with a single multi-coloured, iridescent eyespot. Peahens, which lack the

elaborate train and other plumage ornaments, visit a number of males during each

breeding episode and observe their train displays before copulating with one or at most

two individuals (Petrie et al. 1992, Petrie et al. 1991). Several studies have attempted to 52 measure the morphological traits favoured by peahens, and most indicate that peahens base their mate choices on some visible aspect of the male train (Petrie et al. 1991, Petrie and Halliday 1994, Loyau et al. 2005, 2007a, Dakin and Montgomerie unpublished data, but see Yasmin and Yahya 1996, Takahashi et al. 2008).

Despite all of these studies of female choice in peafowl, little attention has been paid to peacock display behaviours. Darwin (1871) described the males “strutting about, with expanded and quivering tail-feathers” and mentioned that they also “rattle their quills together”. Petrie et al. (1992) provided a more detailed description of the sequence of events during courtship: “when a female approaches a displaying male, he turns away

[from her]… showing the rear surface of his train and his orange primaries which are moved vigorously up and down… if the female follows the male’s turning movement, so that she appears in front of him, he turns to face her head-on and shivers his train.” This sequence may be repeated a number of times, after which the male will attempt to mount the female (Petrie et al. 1992, R Dakin personal observations).

We have observed peacocks with trains erect in both the presence and absence of females. When females are absent, males will often turn or walk around with their trains erect, sometimes performing a “wing-shaking display” wherein the male shakes his primaries up and down vigorously behind his erect train, sometimes for 5-10 minutes at a time. In the presence of one or more females, males often perform the wing-shaking display accompanied by rapid footwork, such as side-stepping or walking backward towards the target female – this may be what Darwin called “strutting”. Males also perform a “train-rattling display” in the presence of females, rapidly vibrating the rectrices that support their train feathers in the upright position for up to 6 minutes 53

(Takahashi et al. 2008), causing the eyespots in the train to vibrate as well. The male is generally stationary during the train-rattling display but may take a few steps towards the target female. As Petrie et al. (1992) noted, females often move around the male’s display court during courtship. However, a male usually performs his wing-shaking display when the target female is positioned behind or beside him, whereas the train-rattling display is always performed when the target female is in front of the male. The wing-shaking display often precedes the train-rattling display, and the train-rattling display always precedes copulation, though the sequence of events during courtship can somewhat variable (contra Petrie et al. 1992). For example, males may begin courtship with the train-rattling display, especially if a female approaches from directly in front of the displaying male.

Given the complexity of peacock displays, and the observation that peacocks do not necessarily orient their erect train towards the target female, our aims in this study were twofold. First, we sought to describe the directionality of peacock display behaviour and quantify the relative positions of both the male and any female observers during a courtship sequence. Second, we asked whether the direction that a male faces during train displays is related to the position of the sun, as one might expect if male behaviour functions to display the iridescent colours of the train plumage effectively to prospective mates. The orientation towards the sun of a courtship display involving an iridescent plumage signal has previously been reported in male Anna’s hummingbirds (Calypte anna). In C. anna, territorial dive displays are “oriented in the general direction of the sun” such that the “reflecting value” of the iridescent gorget plumage is maximized 54

towards the target female or male (Hamilton 1965). We interpret this to mean that the most brilliant colours are visible at that orientation.

We approached this question by observing males courting females over the normal course of the breeding season, and by conducting an experiment involving the presentation of a model female. The experiment allowed us to control the potential effect of female movement on male orientation, and to determine how the patterns of male display behaviour observed under normal conditions were influenced by directional light and the positions of female observers.

Methods and Materials

Field methods

We observed peafowl at the Assiniboine Park Zoo (Manitoba, Canada) in May 2007, at the Los Angeles Arboretum (California, USA) in March-April 2008, at the Bronx Zoo

(New York, USA) in May 2008, and at the Toronto Zoo (Toronto, Ontario) in June 2007.

The Assiniboine Park Zoo (APZ, 50 ha) and Toronto Zoo (TZ, 250 ha) have populations of approximately 60 and 30 peafowl, respectively, that range freely over the zoos’ parklands from April to October, and are kept in large indoor enclosures during the winter months. The Los Angeles Arboretum (LAA, 50 ha) and the Bronx Zoo (BZ, 100 ha) are each inhabited by feral populations of >100 peafowl that range throughout park grounds and surrounding habitat year-round. Birds at all sites were observed during their respective peak lekking seasons: March at LAA, May-June at APZ, BZ, and TZ.

55

Description of male display courts

The display courts of a total of 34 adult peacocks were studied at APZ, LAA and TZ. For

each display court, we noted the pattern of sunlight (bright sunlight, a mix of sun and

shade, or complete shade) at the two time periods when lekking activity peaks: 08:00-

10:00 and 16:00-18:00 local times (Petrie et al. 1991).

Male display orientation

Displaying peacocks were observed during the morning (07:00-12:00 local times) and

late afternoon (15:00-19:00) at APZ and LAA, on days when the sky was clear enough

for shadows to be visible on the ground, and thus when the position of the sun could be

clearly discerned. Feral peafowl spend 10-15% of their daytime time budget displaying

their erect trains during the breeding season (see Figure 3.1; see also Walther 2003), and

most of this activity occurs during the morning and late afternoon (Petrie et al. 1991,

Walther 2003). We identified males mainly by numbered leg-bands or unique morphological features, and in some cases by the location of a display court maintained

by a particular male – peacocks almost always maintain a single display court throughout

the breeding season. We observed individual males from 5-10 m away as this appeared

not to interfere in any way with their behaviour.

We studied 11 displaying males at two sites (5 males at 2 leks at APZ; 6 males at

4 leks at LAA; leks ranged in size from 1-6 males with a mean size of about 4 males),

walking between lek sites and sampling males that were displaying their trains. For each bout, we recorded the time and the identity of the male, and the following variables at

30-s intervals: display behaviour with train erect (no dance, wing-shake, train-rattle), any 56 movement (walking forwards, backwards, to the left or right, turning), the presence and number of any females within 15 m, the distance (estimated to the nearest m) to the nearest female, and the location of that female relative to the displaying male. Female location was recorded as being in one of the six 60° sectors around the male (Figure

3.2a). At each 30-s interval, we also measured the bearing (to the nearest 10°) of the displaying male as the compass bearing relative to true north of the long axis through his body (perpendicular to his erect train) from tail to head. All compass bearings taken in the field were corrected for magnetic declination (National Geophysical Data Center

2008).

At each 30-s interval, we determined the local sun azimuth using a solar angle calculator that takes into account the date, time, latitude and longitude (Gronbeck 2005).

Using the sun azimuth, the location of the target female and the bearing of the displaying male (angle A in Figure 3.2b), we then calculated additional angles of interest for further analysis. Thus, we calculated the male-sun angle as the clockwise angle between the sun’s azimuth and the bearing of the displaying male (angle B in Figure 3.2b). We also calculated the female-sun angle as the clockwise angle between the sun’s azimuth and the position of the female relative to the male (angle C in Figure 3.2b), using the median angle for each of the 6 female position sectors (i.e., 300°, 0°, 60°, 120°, 180°, or 240° for sectors 1-6 respectively in Figure 3.2a).

In total, we collected display data from 882 point samples (at 30-s intervals) for

58 display bouts performed by 11 different peacocks (6 from LAA and 5 from APZ). On average, we measured 16 point samples (range 3-66) per display bout, and 5 display bouts (range 2-10) per male. We analyzed circular data according to Zar (1999), where ā 57

is the mean angle calculated for one of the angles of interest (A-C on Figure 3.2b), and

VL (or vector length) is the length of the mean vector of that sample of angles. Thus, VL

is a measure of angular dispersion that ranges from 0 (highly dispersed, adirectional

sample) to 1 (highly directional such that all measured angles are the same). We

calculated ā and VL for each bout (mean bout angle and mean VL) using the procedure

for first-order means (Zar 1999). We then calculated ā and VL for each individual male

(mean male angle and mean male VL) using a parametric second-order analysis (Zar

1999). Finally we performed a grand mean analysis across males, using a further parametric second-order analysis on the mean male values and testing for directionality using the Hotelling procedure for second-order samples (Zar 1999). This procedure tests whether the distribution of angles differs significantly from uniform.

Most of the display bouts that we observed were encountered in progress, so we

have few data from complete bouts (i.e., from the time the male raised his train until he

lowered it). There is, however, unlikely to be any bias due to this sampling method as

male behaviours and female locations do not vary systematically during a display (Figure

3.3). For example, for the longest recorded bout from each male studied (n = 11), there

was no relation between bout length and the proportion of point samples with females

present (Pearson correlation, r = 0.27, p = 0.42, n = 11), nor was there any relation

between bout length and the proportion of samples where males were performing the

wing-shaking display (r = 0.14, p = 0.68, n = 11). There was a significant positive

relation between bout length and the proportion of samples where males were performing

the train-rattling display (r = 0.63, p = 0.04, n = 11), but this is not surprising given that

the train-rattling behaviour is relatively infrequent. There was also no difference in the 58

number of point samples of each display type observed in the first or second half of the

observation period (wing-shaking: paired t-test, t = 0.91, p = 0.38; train-rattling: t = 0.48,

p = 0.64, n = 11). Finally, there was no relation between bout length and the vector length

for the mean male bearing (Pearson correlation, r = 0.17, p = 0.62, n = 11).

Model presentation experiment

To control for the effects of female movement on male orientation, we conducted

experiments at LAA and BZ, presenting males with a taxidermically mounted peahen in

four different contexts, varying both the time of day and the position of the model female.

Thus trials were run in the morning or the afternoon (AM or PM) with the model female

facing the male on either the EAST or WEST side of his body. We selected males with erect trains (i.e., displaying, but not yet wing-shaking or train-rattling) as they were

encountered and randomly chose the initial side on which to place the model. After

successfully completing one trial for a given male, we performed the next trial at the

same period of the day with that male (i.e., AM or PM), presenting the model on the

opposite side. Our goal was to run all four trials for each male, but we succeeded with

only 2 males. We ran three trials with each of 2 other males, two trials with each of 6

males, and only one trial with each of 7 males (Table A.1). Thus, a total of 33 trials were

completed with 17 different males (AM-EAST = 9, AM-WEST = 10, PM-EAST = 8,

PM-WEST = 6 trials).

All trials were conducted using males that were displaying their trains with no

live females in view. The model female was placed about 2 m away from the displaying

male, and we observed his response from 5-10 m away to the north or south. Trials were 59

started when the male began to perform either the wing-shaking or train-rattling display

in a way that was clearly directed towards the model female (usually 0-30 s after the

model was set up) and were run for 5 min.

During all trials, we recorded the following at 30-s intervals as described above:

display behaviour of the male, any movement by the male, distance from the male to the

model female (to the nearest m), sector location of the model female relative to the male

(Figure 3.2a), and the compass bearing of the displaying male. Any trials where the male

did not respond to the model within 2 min by wing-shaking or train-rattling were

terminated and discarded.

Trials were terminated prematurely if males attempted to copulate with the model

(n = 8 trials), when male display behaviour was no longer directed towards the model due

to the arrival of additional females (n = 5 trials), or when males lost interest in the model

and stopped displaying or moved away (n = 5 trials). We have no reason to expect that

the male behaviour prior to termination was unusual in these cases, so we include all of these trials in our analyses.

We analyzed male responses in two ways. First, to examine the type of display

behaviour performed in response to the model (wing-shaking or train-rattling) we used only the first point sample from each trial, because males would often move relative to the model during the course of the trial. For these analyses, we include only the first experimental trial performed with each individual bird to avoid pseudoreplication.

Second, to examine the overall directionality of the display response, we calculated first- order mean angles for each trial, and then calculated a parametric second-order mean where more than one trial type had been conducted with a given male in the same time 60

period (AM or PM), or for the same model placement (EAST or WEST), as described

above (Zar 1999). We then used an additional parametric second-order analysis to test for directionality across individuals.

Results

Light environment at male display courts

Display courts were generally located in open areas, but most were also near the edge of a field or some other sheltering feature (e.g., trees, fences, or walls of buildings). Most courts (74%, 25/34) experienced full sun in the morning, and only 18% (6/34) were completely shaded in the morning. Similarly, most courts (62% or 21/34) experienced full sun in the afternoon, and only 18% (6/34) were shaded at that time. Only one of the

34 courts we studied was completely shaded during both morning and afternoon sampling periods.

Female position during displays

Pooling female position data across all display bouts per male (Figure 3.4a) shows that females were most often directly in front of males (sector 2, mean of 27% of point samples with a female present), or behind them to their right (sector 4, mean of 29%) or left (sector 6, mean of 21%). Interestingly, females were almost never in front of males to the right or left (sectors 1 and 3, mean of 5% for each), and were infrequently positioned directly behind the male (sector 5, mean of 13%). Thus females were significantly more often in sectors 2, 4 and 6 than in 1, 3 and 5 (Tukey post hoc test, ANOVA, F = 13.5, p 61

<0.0001, df = 5,59). When a female was visible in front of a male in sectors 1 or 3, the

male almost always turned so that the female would be in sector 2.

For males performing the wing-shaking display, females were generally positioned behind in sectors 4-6 (mean of 83% of point samples), and this mean percentage is significantly greater than 50% (one sample t-test, t = 1304, p < 0.0001, n =

11). In contrast, males only performed the train-rattling display when females were positioned directly in front (mean of 99.6% of point samples). Males often moved or turned during the wing-shaking display when females were present (mean of 33%), but they generally did not move during the train-rattling display (mean of 7%). On average, males were significantly more likely to move or turn during the wing-shaking compared to train-rattling, pooling across bouts (Figure 3.4b; t-test, t = 4.83, p < 0.0001, df = 19).

Male orientation during display bouts

When females were not present, males performed the wing-shaking display on average

34% of the time pooling across bouts for each male (n = 10 males), and otherwise simply stood or walked about with their train erect. The grand mean sun-male angle was significantly directional when females were absent (Figure 3.5; VL = 0.41, p = 0.05, n =

10), and the grand mean angle of 45° (95% CI = 308-109°) was not significantly different from 0° (i.e., facing towards the sun).

When females were present, and males performed the wing-shaking display, the grand mean sun-male angle was not directional (Figure 3.6a; VL = 0.18, p = 0.21, n =

11), though mean sun-male angle was somewhat consistent within males (mean male VL

= 0.35 ± 0.17 SD, 95% CI = 0.24-0.47, n = 11), and, when pooling display bouts for a 62

first-order Rayleigh test of directionality for each male, 7/11 males were significantly

directional. The grand mean female-sun angle was also adirectional during the wing-

shaking display (Figure 3.6c, VL = 0.15, p = 0.40, n = 11 males). However, 6/11 males

were significantly directional and interestingly, females were on average located on the

shaded side of the male for all of these significant individual means.

In contrast, during train-rattling, the grand mean sun-male angle was highly

directional (Figure 3.6b, VL = 0.52, p = 0.001, n = 10). In this context, males tended to

face towards the sun since the grand mean angle (43°) was close to, though significantly

different from 0° (95% CI = 11-95°). The mean sun-male angle during train-rattling was

also consistent within individuals (mean male VL = 0.67, 95% CI = 0.51-0.83, n = 10),

with 8/10 individual means significant by first-order Rayleigh analyses. Results are

nearly identical for the grand mean female-sun angle in this context (Figure 3.6d, VL =

0.52, p = 0.002, n = 10 males), with a grand mean angle of 44° (95% CI = 13-93°),

because females were almost always in sector 2 during this display. Thus, during the train-rattling display, males were on average facing toward the sun at approximately 43° west of the sun azimuth, and the female was standing directly in front of the male.

Model presentation experiment

The type of dance initially performed by the male after he was presented with the female

model (Table 3.1) did not depend on the time of day (AM vs. PM; Fisher Exact Test, p =

0.34) or the location of the model (EAST vs. WEST; p = 0.34). Instead, the type of dance

that the male performed depended on the position of the model relative to the sun: males

were more likely to initially perform the train-rattling dance when the model was 63

presented on the sunny side of their trains (i.e., EAST in the AM and WEST in the PM

trials), and were more likely to perform the wing-shaking dance when the model was

presented on the shaded side (Fisher Exact Test, p = 0.03). As with live females, males usually (10/12 trials where they performed this display) positioned themselves with the model female behind the train ornament when they performed the wing-shaking dance, and with the female directly in front for the train-rattling dance (all 5 trials where they performed this display).

The grand mean male compass bearing was not significantly directional when pooling EAST and WEST trials, but was significantly directional (or nearly so) when pooling trials by time of day (Table 3.2). The grand mean sun-male angle tended to be significantly directional regardless of the trial context, and did not differ significantly from 0° (i.e., facing directly toward the sun) (Table 3.2).

Comparisons of grand mean angles between different trial types also indicate that the position of the sun, rather than the model female, determined display orientation.

Using only data from the first trial performed with each individual to avoid

pseudoreplication, the grand mean male compass bearing was significantly different

between AM and PM trials (parametric two-sample second-order test, F = 11.9, p =

0.0009, n = 10, 7) but not between EAST and WEST trials (F = 0.5, p = 0.64, n = 10, 7).

The grand mean sun-male angle was not different when comparing AM and PM (F = 0.5,

p = 0.61, n = 10, 7) or EAST and WEST (F = 0.4, p = 0.66, n = 10, 7) trials.

64

Discussion

Our observational and experimental results demonstrate that the direction that peacocks

face during courtship display is influenced by the position of the sun. The train-rattling display was significantly directional, with males oriented at about 45° to the right of the sun on average, and with the target female almost always positioned directly in front of the male (Figure 3.6b, 3.6d). This indicates that sunlight may be important for peafowl courtship, possibly for the communication of a plumage colour signal. Similarly, when females were absent, male displays were significantly directional relative to the sun

(Figure 3.5), though apparently not as consistently directional as they were during train- rattling. This could be because males with their trains erect will often turn around, vigilant for the approach of the next female (R Dakin, personal observations).

In contrast, during the wing-shaking display, male orientation was not directional relative to the sun (Figure 3.6a). In addition, although the grand mean female-sun angle was not significant in this context, females were usually positioned on the shaded side of the male (Figure 3.5c). Wing-shaking is generally performed while facing away from females, and the conspicuously moving orange-coloured wings could serve to attract female attention. Thus it appears that males use the wing-shaking display to encourage females into a preferred viewing configuration for subsequent train-rattling displays, and the associated male turning and “strutting” may be used to corral target females onto the sunny side of the male. This hypothesis is strongly supported by our model presentation trials – by controlling female movement, we showed that males will generally perform the wing-shaking display when the female is positioned on the shaded side of their train, versus train-rattling when the female is positioned on the sunny side. Thus, the observed 65

directionality of the train-rattling display under natural conditions is due to male

behaviour and not simply to peahens attempting to view males from the sunny side,

although it is possible that the females still prefer to do this.

While there was no consistency across males in the orientation of the

wing-shaking display, the majority of individual males showed significant directionality

(Figure 3.6a). One possible explanation is that some of the inter-male variation in display

orientation could be due to the geographical features of individual display courts.

Different male courts vary in the size and location of sheltering features and in their

placement relative to areas of preferred foraging and resting habitat and thus high female

traffic (R Dakin, personal observation). As a consequence, the most common direction of approach for females and the direction males should face most frequently when attempting to attract new visitors may vary across males.

Of the train-rattling display, Darwin said that “peacocks and birds of paradise rattle their quills together, and the vibratory movement apparently serves merely to make a noise, for it can hardly add to the beauty of their plumage” (1871). However, we believe that rapid feather movements might help to display the feathers’ iridescence. Preliminary evidence suggests that the iridescent eyespot colouration of peacock train feathers is important for mate choice (Loyau et al. 2007a; Dakin and Montgomerie, unpublished data) and it may be an honest signal of male quality (Dakin and Montgomerie, unpublished data). It is intriguing that, like iridescent hummingbirds (Hamilton 1965), peafowl orient towards the sun during this train-rattling component of their courtship.

The benefit to males of orienting relative to the sun is not yet clear. One possibility is that it improves the efficacy of the colour signal, allowing females to 66

discriminate more easily among individual males – females may prefer to visit males that

signal colour efficiently. It is also possible that males can exploit female preferences for bright plumage by orienting towards the sun, and manipulate visitors into matings that are suboptimal for the female. The idea that males can manipulate females into suboptimal matings in this way is one that requires empirical testing. However, given that peahens, like other birds, have colour constancy (Hart 2002; Vorobyev et al. 1998), this hypothesis

seems unlikely. It is also unclear why the train-rattling display was, on average, oriented at about 45° to the right of the sun. It is possible that this particular angle minimizes

specular reflectance from the eyespots. Another possibility is that this produces a

particular iridescent effect or hue that females prefer. Alternatively it may be that males

orient towards the sun but avoid facing it directly during train-rattling.

While we have focused this study on the analysis of peacock displays on sunny

days, peacocks will certainly display on overcast days and in partial or complete shade (R

Dakin, personal observation). Some birds in this study maintained highly sheltered

display courts that were in complete shade throughout the morning, and these were the males that generally deviated most from the overall directional patterns that we described. For example, the two individuals whose train-rattling displays were not

directed towards the sun both maintained display courts that were completely shaded

throughout the morning (mean sun-male angles of 123° and 159°, VL = 0.87 and 0.20

respectively, Figure 3.6b). Hamilton (1965) reported that on cloudy days, the

directionality normally seen in Anna’s hummingbird dives was no longer apparent.

Similarly, in peafowl, variation in light and shade conditions could explain some of the

dispersion in the patterns reported here. It would be an interesting natural experiment to 67 compare the patterns of male display behaviour and directionality on clear days with those of overcast days when light is diffuse.

If orienting towards the light is important, one might wonder why some peacocks would choose shaded territories at all. Display territory selection most likely involves a number of factors, including proximity to food sources where females congregate (Loyau et al. 2007b), proximity to closely-related males (Petrie et al. 1999), distance from agonistic males and predators, and shelter from the wind. These are all considerations that may be more important than the choice of light environment. For example, one male in this study maintained a highly-sheltered display court directly to the west of a tall building that was completely shaded throughout the morning. Nevertheless, he received a high rate of female visitation apparently because his territory was adjacent to a popular dust-bathing site that attracted many females, consistent with the hot-spot theory of lek formation (Bradbury and Gibson 1983). Display court selection is no doubt a complex consideration, and presumably it would not be advantageous for males to choose a well-lit territory in an area that would receive no visitors, or for males to refuse to display during periods of shade or cloud.

Previous studies have shown that features of avian courtship such as display site selection (Anciães and Prum 2008) and specialized viewing configurations (Borgia and

Presgraves 1998) are designed for signal efficacy. We show here that this is likely the case for a complex display behaviour as well. We suggest that the two different display behaviours performed by peacocks are designed not just to communicate a visual signal, but to do so efficiently by the co-ordination of female viewing geometry relative to the source of light. 68

References

Anciães M, Prum RO (2008) Manakin display and visiting behaviour: a comparative test of sensory drive. Anim Behav 75: 783-790

Borgia G, Presgraves DC (1998) Coevolution of elaborated male display traits in the spotted bowerbird: an experimental test of the threat reduction hypothesis. Anim Behav 56: 1121-1128

Bradbury JW, Gibson RM (1983) Leks and mate choice. In: Bateson P (ed) Mate Choice. Cambridge University Press, Cambridge

Darwin C (1871) The descent of man, and selection in relation to sex. John Murray, London

Doucet S, Mennill DJ, Hill GE (2007) The evolution of signal design in manakin plumage ornaments. Am Nat 169: S62-S80

Endler JA, Théry M (1996) Interacting effects of lek placement, display behavior, ambient light and color pattern in three neotropical forest-dwelling birds. Am Nat 148: 421-452

Gomez D, Théry M (2004) Influence of ambient light on the evolution of colour signals: comparative analysis of a Neotropical rainforest bird community. Ecol Lett 7: 279-284

Gronbeck C (2005) Sunangle. Retrieved May 2008 from http://susdesign.com/sunangle/.

Hamilton WJ (1965) Sun-oriented display of the Anna’s hummingbird. Wilson Bull 77: 38-44

Hart NS (2002) Vision in the peafowl (Aves: Pavo cristatus). J Exp Biol 205: 3925-3935

Heindl M, Winkler H (2003) Vertical lek placement of forest-dwelling manakin species (Aves, Pipridae) is associated with vertical gradients of ambient light. Biol J Linn Soc 80: 647-658

Johnson KP (2000) The evolution of courtship display repertoire size in the dabbling ducks (Anatini). J Evol Biol 13: 634-644

Kelso EC, Martins EP (2008) Effects of two courtship display components on female reproductive behaviour and physiology in the sagebrush lizard. Anim Behav 75: 639-646

Loyau A, Gomez D, Moureau B, Théry M, Hart NS, Saint Jalme M, Bennett ATD, Sorci G (2007a) Iridescent structurally based coloration of eyespots correlates with mating success in the peacock. Behav Ecol 18: 1123-1131

69

Loyau A, Saint Jalme M, Sorci G (2007b) Non-defendable resources affect peafowl lek organization: a male removal experiment. Behav Proc 74: 64-70

Loyau A, Saint Jalme M, Cagniant C, Sorci G (2005) Multiple sexual advertisements honestly reflect health status in peacocks (Pavo cristatus). Behav Ecol Sociobiol 58: 552- 557

Maynard Smith J, Harper D (2003) Animal Signals. Oxford University Press, Oxford

McNaught MK, Owens IPF (2002) Interspecific variation in plumage colour among birds: species recognition or light environment? J Evol Biol 15: 505-514

National Geophysical Data Center (2008) Estimated value of magnetic declination. Retrieved June 2008 from http://www.ngdc.noaa.gov/geomagmodels/struts/declZipE

Petrie M, Hall M, Halliday T, Budgey H, Pierpoint C (1992) Multiple mating in a lekking bird: why do peahens mate with more than one male and with the same male more than once? Behav Ecol Sociobiol 31: 349-358

Petrie M, Halliday T (1994) Experimental and natural changes in the peacock’s (Pavo cristatus) train can affect mating success. Behav Ecol Sociobiol 35: 213-217.

Petrie M, Halliday T, Sanders C (1991) Peahens prefer peacocks with elaborate trains. Anim Behav 41: 323-331

Petrie M, Krupa A, Burke T (1999) Peacocks lek with relatives even in the absence of social and environmental cues. Nature 401: 155-157

Prum RO (1985) The displays of the white-throated manakin (Corapipo gutturalis) in Suriname. Ibis 128: 91-102

Prum RO (1997) Phylogenetic tests of alternative intersexual selection mechanisms: trait macroevolution in a polygynous clade (Aves: Pipridae). Am Nat 149: 668-692

Prum RO, Johnson AE (1987) Display behavior, foraging ecology, and systematics of the golden-winged manakin (Masius chrysopterus) Wilson Bull 99: 521-539

Scholes E (2006) Courtship ethology of Carola’s parotia (Parotia carolae). Auk 123: 967-990

Scholes E (2008) Courtship ethology of Wahnes’ parotia Parotia wahnesi (Aves: Paradisaeidae). J Ethol 26: 79-91

Takahashi M, Arita H, Hiraiwa-Hasegawa M, Hasegawa T (2008) Peahens do not prefer peacocks with more elaborate trains. Anim Behav 75: 1209-1219

70

Théry M and Vehrencamp SL (1995) Light patterns as cues for mate choice in the lekking white-throated manakin (Corapipo gutturalis). Auk 112: 133-145

Tinbergen N (1952) "Derived" activities; their causation, biological significance, origin, and emancipation during evolution. Quart Rev Biol 27: 1-32

Vorobyev M, Osorio D, Bennett ATD, Marshall NJ, Cuthill IC (1998) Tetrachromacy, oil droplets and bird plumage colours. J Comp Phys A 183: 621-633

Walther BA (2003) Do peacocks devote maintenance time to their ornamental plumage? Time budgets of male blue peafowl Pavo cristatus. Lundiana 4: 149-154

Yasmin S, Yahya HSA (1996) Correlates of mating success in Indian peafowl. Auk 113: 490-492

Zar JH (1999) Biostatistical analysis. Prentice Hall, New Jersey.

71

Table 3.1 Initial display responses of males to a model female presented under different experimental contexts: morning or afternoon, and on the east or west side of the male.

Display response frequency Experimental context (n) Wing-shaking Train-rattling Time of day AM (10) 6 4 PM (7) 6 1 Model location (compass) EAST (10) 6 4 WEST (7) 6 1 Model location (sun) 1 Sunlit side (9) 4 5 Shaded side (8) 8 0 1 Sunlit side refers to AM-EAST and PM-WEST trials; shaded side to AM-WEST and PM-EAST.

72

Table 3.2 Directionality of male display responses to a model female by experimental

context. Statistics are for parametric second-order tests of directionality for circular data;

samples that are significantly directional are highlighted in bold. The mean angle is given

where the sample is significantly directional or nearly so. For male bearing, 0° refers to

true north, whereas for sun-male angle, 0° refers to the sun azimuth.

1 Mean angle Experimental context (n) VL F p ā 95% CI

Male bearing AM (12) 0.33 3.0 0.10 99.7° PM (9) 0.61 26.8 0.0005 277.2° 236-321° EAST (14) 0.19 0.9 0.44 WEST (13) 0.08 0.1 0.91

Sun-male angle AM (12) 0.34 3.1 0.09 357.4° PM (9) 0.59 20.7 0.001 5.8° 321-54° EAST (14) 0.36 4.4 0.04 16.1° 318-101° WEST (13) 0.53 12.6 0.001 358.6° 324-42° 1 VL is the mean vector length; ā is the mean angle in degrees where the sample is significantly directional or nearly so. The 95% confidence interval for ā can be calculated where samples are significantly directional (Zar 1999).

73 Turn/walk 33% Turn/walk Stationary 67% 7% Turn/walk Stationary 93% 32% Turn/walk Stationary 68% 47% Turn/walk Stationary 53% 45% Turn/walk Stationary 55% 26% 65% 51% 35% 23% Wing-shake Train-rattle up only Train up only Train Wing-shake 76% 24% Female(s) absent Female(s) present Stand vigilant 57% Perch 27% Preen 7% Sit 9%

77% 23% Train up Train down Train Aggression Forage Rest 70% 30% Off lek Off On lek e budget for peacocks showing the average percent of time males devote to each activity at each level in the Figure 3.1 DailyFigure tim Percentages up” are averages during calculated “Train hierarchy. from the longest display bouts males recorded at LAA for each of 11 APZ between 07:00 and 19:00 local times.and Percentages during all other activites down”) up” versus (including “Train “Train are averages calculated from where observations we sampled each over of 17 a single males once day at LAA per hour for 13 hours (07:00-19:00). 74 a)

2 1 3

6 4 5

b) N

B A

C

Figure 3.2 a) Diagram from above of the 60˚ sectors around a displaying peacock, used to identify the position of the female relative to the male. b) Angles of interest: A = male bearing, B = the sun-male angle, C = the sun-female angle. 75

1

2

3

4

5

6

7

8

9

bout 10 train-rattling wing-shaking female presence 11

0 5 10 15 20 25 30 35 Time (minutes)

Figure 3.3 Sequence of events over the longest bout recorded from each of 11 different males, based on point samples taken at 30-s intervals. Note that wing-shaking and train-rattling behaviour may have stopped between point samples, even though they are illustrated here as continuous. 76 a) b) Front Back

0.5 0.6

0.5 0.4

0.4 0.3

0.3 0.2 0.2

Mean proportion 0.1 0.1

0.0 Mean proportion moving 0.0

1 2 3 4 5 6 Wing−shaking Train−rattling Sector

Figure 3.4 Tukey box plots showing (a) the proportion of point samples where female visitors were positioned in each of the 60˚ sectors around the male (see Figure 3.2) and

(b) the proportion of point samples where males moved or turned during the wing-shaking and train-rattling displays. Data were calculated by pooling all display bouts for each of 11 different males, and taking a mean for each. Each box plot therefore shows the distribution of the 11 male mean proportions. 77

Figure 3.5 Sun-male angles when females were absent (n = 10 males). Grey vectors are the male means (one for each male) and the black vector is the grand mean across individuals. Mean vectors are drawn as solid lines when samples are significantly non-random. The 95% confidence interval (shaded) is shown for the grand mean. The outer circle has a radius of VL = 1. 78

a) b)

c) d)

Figure 3.6 Sun-male angles during (a) the wing-shaking (n = 11) and (b) train-rattling (n = 10) displays, and sun-female angles during (c) wing-shaking (n = 11) and (d) train-rattling (n = 10). Grey vectors are male means (one for each male), and the black vectors are grand means across individuals. Mean vectors are drawn as solid lines when samples are significantly non-random. The 95% confidence intervals (shaded) are shown for the significant grand means. The outer circle has a radius of VL = 1. 79

CHAPTER 4. General Discussion

The train of the peacock may be the most frequently cited example of a sexually selected

trait, but empirical studies of feral peafowl populations have failed to reach a consensus

with respect to the importance of various signaling modalities for female mate choice. Do

females choose males based on visual aspects of the train ornament (Petrie et al. 1991,

Petrie and Halliday 1994, Loyau et al. 2007), the intensity of behavioural displays (Loyau et al. 2005b), or do they perhaps base their choice on one of the auditory signals used by males during the breeding season (Yasmin and Yahya 1996, Takahashi et al. 2008)?

There are several possible explanations for the inconsistencies in the peafowl literature: perhaps female choice in peafowl varies in strength or direction between study populations (Endler and Houde 1995), or perhaps it varies within populations over time

(Chaine and Lyon 2008). Another possibility is that mate choice in peafowl is complex and based on multiple cues. Despite extensive study, this enigmatic species remains relatively poorly described, and a lack of background information on the natural history of peafowl has made it difficult to interpret seemingly contradictory results.

One of the aims of this thesis was to address this lack of background information.

In both Chapters 2 and 3, I described in detail two aspects of visual signalling in peafowl courtship that have previously been ignored. In Chapter 2, I described the geometric arrangement of eyespots in the train ornament of adult males. This is relevant to our understanding of variation in eyespot number, a measure that has produced some of the most recent conflicting results with respect to mate choice in peafowl. In Chapter 3, I described the behaviours performed by males during the display of the train ornament, asking how they might function in the context of visual signalling. 80

Major Findings

My results resolve the previously controversial relationship between train morphology

and mating success. I repeated Petrie and Halliday’s (1994) experimental test of the effect

of manipulating the train ornament on mating success, something rarely done in

behavioural ecology yet critical to understanding the generality and relevance of a

particular result (Palmer 2000). Although I confirmed that removing a large number of

eyespots reduces male mating success, I failed to find any effect of normal variation in

train morphology on mating success across three study populations. This is consistent

with the most extensive peafowl study to date (Takahashi et al. 2008) but is in sharp

contrast with some earlier ones (Petrie et al. 1991, Loyau et al. 2005b).

This apparent contradiction can be resolved by a more detailed understanding of variation in eyespot number. First, I showed that the arrangement of eyespot feathers in the train is highly consistent, indicating that variation in eyespot number is primarily the result of feather loss rather than differences in feather growth. Second, I showed that the experimental eyespot removal treatment used by Petrie and Halliday (1994) modifies the train well beyond the normal range of variation for feral males. Although I found substantial variation in mating success among full-trained adult males, that variation is not correlated with eyespot number (see also Takahashi et al. 2008). Thus, mate choice in peafowl may be more complex than previously thought: females may simply avoid males missing a large number of train feathers, and instead choose among full-trained males using some other cue. These results demonstrate that exaggeration in one aspect of a sexually-selected trait (such as length) does not necessarily imply that sexual selection is currently acting on that particular aspect (Møller and Pomiankowski 1993). Additionally, 81

my results highlight the fact that inconsistencies in the previous literature are not due to

local differences in peahen mate preferences. In fact, the replication of Petrie and

Halliday’s (1994) results implies that peahen mate choice may be consistent across

disparate populations, possibly due to a genetic basis for mate choice (Boake 1989).

My results also point to some new ideas about the function of courtship display

behaviour in visual signaling. Based on the observation that displaying peacocks do not

always orient towards the nearest female, I tested the possibility that males face towards

the sun to enhance the reflectance of their iridescent train plumage. Interestingly, I found

that directionality depended on the type of display behaviour: the train-rattling display

was highly directional relative to the sun, whereas the wing-shaking display was more

variable and was performed when females were positioned behind the male. This allowed

me to formulate some hypotheses about the functions of these two behaviours: namely,

that the wing-shaking display functions to move females into a preferred viewing geometry for visual communication during the train-rattling display. An experimental test using a model female supported this, since the wing-shaking display was elicited more frequently when the model was positioned on the shaded side of the male. Thus, the

ultimate function of these two different display behaviours may be to convey a plumage

colour signal and to do so efficiently, something that has previously been shown for other

aspects of avian courtship (Endler and Théry 1996, Borgia and Presgraves 1998, Doucet

and Montgomerie 2003a) but not for display behaviour specifically.

In summary, the results presented here highlight the problems with interpreting

measures of ornamentation without some understanding of the biology of the trait in

question. The geometrical arrangement of eyespot feathers in the train was relatively 82 simple to measure. In fact, the precise arrangement of these feathers was originally noted by the artist Ernest Thompson Seton (1896) in an instructional guide to drawing animals, though it escaped the view of a number of scientists studying eyespot number (e.g.,

Manning and Hartley 1991, Petrie et al. 1991). None of these authors considered the source of variation in eyespot number despite its relevance to the interpretation of their results. Furthermore, this thesis points to the fact that, despite suggestions to the contrary

(Takahashi et al. 2008), mate choice in peafowl is likely based predominantly on visual signals. Although the number of eyespots may not be the main cue used by females, this does not preclude the importance of the visual train ornament altogether. The results of

Chapter 3 strongly implicate visual communication in peafowl courtship. Indeed, the most dramatic component of the peacock’s display, the pre-copulatory train-rattling display, is highly directional with respect to light.

Future research

It is intriguing that the train-rattling display involves the rapid movement of the eyespot feathers. Theorists from Darwin (1871) to John Maynard Smith (Harper 2006) have suspected that the auditory component of this display could be important. However, the rapid feather movement might also allow females to perceive the degree of iridescence of the eyespots, or change in colour with changing viewing geometry. Interestingly, a recent study has reported correlations between both the static and dynamic aspects of eyespot colouration and mating success in peafowl (Loyau et al. 2007).

The colours of the peacock’s eyespot feathers are produced structurally rather than by pigmentation. Each feather barbule contains a highly-complex nanostructural 83

array of melanin, keratin and air, the precise arrangement of which is responsible for the

resulting colour (Zi et al. 2003). Nanostructural colours are widespread throughout the

animal kingdom, having arisen several times independently in birds, insects and fish

(Prum 2006); they are among the most conspicuous and sexually dimorphic animal

colours we know of. Evidence is growing that species as widely diverse as starlings and

butterflies use nanostructural colour signals in mate choice (Bennett et al. 1997, Kemp

2007). In some species, these colours may be condition-dependent, and thus a good signal

of individual quality (Kemp and Rutowski 2007, Doucet and Montgomerie 2003b). We

currently know very little about the development of these colour-producing

nanostructures. However, evidence that individual variation in nanostructural architecture

is responsible for variation in colour suggests that the signal content of these colours may

be recorded during development (Shawkey et al. 2003, Doucet et al. 2006, Kemp et al.

2006). Production of highly-ordered structures may be energetically expensive, so that

only the individuals in the best condition can afford it, or it may simply be prone to

disruption during growth.

With respect to peafowl, the question of whether the train-rattling display enhances the iridescent component of the signal remains open. Further experimental work is also required to establish that the correlation between eyespot colour and mating success reported by Loyau et al. (2007) is due to female perception of colour, and not some correlated trait. If eyespot colour does prove to be important, what mechanism do females use to assess this trait? Is it a threshold-based mechanism similar to that described here for eyespot number, or do females rank males on the basis of colour traits?

Experimental manipulations of multiple aspects of the train ornament could provide some 84

insight into the cognitive basis for mate choice in peafowl in addition to the function of

multiple sexual ornaments.

More work is also needed to clarify the signal content of the peacock’s train.

Although previous research has indicated that variation in the number of eyespots is

condition-dependent (Loyau et al. 2005a), it is not clear whether this relation holds within the normal range of 140-160 eyespots among feral peacocks. It is also unclear whether this is the same range of eyespot number found among wild peafowl, which may experience harsher conditions than feral birds. What are the factors involved when males have less than 130 eyespots in their trains – is this caused by feather loss or stunted growth? Another area for further study is the signal content of the iridescent eyespot colouration. If eyespot colour is condition-dependent, how does it relate to the growth or maintenance of a large number of eyespot feathers in the train? An understanding of the signal function of eyespot colouration could also clarify the results of Chapter 3. Does orienting towards the sun improve the efficacy of an honest visual signal? Does it allow some males to “cheat”, manipulating females into suboptimal matings by enhancing eyespot colouration? Either way, the fact that peafowl are readily accessible and amenable to experimental approaches will doubtless prove useful in addressing fundamental questions about the design and content of courtship signals.

References

Bennett ATD, Cuthill IC, Partridge JC, Lunau K (1997) Ultraviolet plumage colors predict mate preferences in starlings. PNAS 94: 8618-8621

Boake CRB (1989) Repeatability: its role in evolutionary studies of mating behavior. Ev Ecol 3: 173-182

85

Borgia G, Presgraves DC (1998) Coevolution of elaborated male display traits in the spotted bowerbird: an experimental test of the threat reduction hypothesis. Anim Behav 56: 1121-1128

Chaine AS, Lyon BE (2008) Adaptive plasticity in female mate choice dampens sexual selection on male ornaments in the lark bunting. Science 319: 459-462

Darwin C (1871) The descent of man, and selection in relation to sex. John Murray, London

Doucet SM, Montgomerie R (2003a) Bower location and orientation in satin bowerbirds: optimizing the conspicuousness of male display? Emu 103: 105-109

Doucet SM, Montgomerie R (2003b) Structural plumage colour and parasites in satin bowerbirds Ptilonorhynchus violaceus: implications for sexual selection. J Avian Biol 34: 237-242

Doucet SM, Shawkey MD, Hill GE, Montgomerie R (2006) Iridescent plumage in satin bowerbirds: structure, mechanisms and nanostructural predictors of individual variation in colour. J Exp Biol 209: 380-390

Endler JA, Houde AE (1995) Geographic variation in female preferences for male traits in Poecilia reticulata Evolution: 49: 456-468

Endler JA, Théry M (1996) Interacting effects of lek placement, display behavior, ambient light and color pattern in three neotropical forest-dwelling birds. Am Nat 148: 421-452

Harper DGC (2006) Maynard Smith: amplifying the reasons for signal reliability. J theor Biol 239: 203-209

Kemp DJ (2007) Female butterflies prefer males bearing bright iridescent ornamentation. Proc Roy Soc B 274: 1043-1047

Kemp DJ, Rutowski RL (2007) Condition dependence, quantitative genetics and the potential signal content of iridescent ultraviolet butterfly coloration. Evolution 61: 168- 183

Kemp DJ, Vukusic P, Rutowski RL (2006) Stress-mediated covariance between nano- structural architecture and ultraviolet butterfly coloration. Func Ecol 20: 282-289

Loyau A, Saint Jalme M, Cagniant C, Sorci G (2005a) Multiple sexual advertisements honestly reflect health status in peacocks (Pavo cristatus). Behav Ecol Sociobiol 58: 552- 557

86

Loyau A, Saint Jalme M, Sorci G (2005b) Intra- and intersexual selection for multiple traits in the peacock (Pavo cristatus). Ethol 111: 810-820

Loyau A, Gomez D, Moureau B, Théry M, Hart NS, Saint Jalme M, Bennett ATD, Sorci G (2007) Iridescent structurally based coloration of eyespots correlates with mating success in the peacock. Behav Ecol 18: 1123-1131

Manning JT (1989) Age-advertisement and the evolution of the peacock’s train. J Evol Biol 2: 379-384

Manning JT, Hartley MA (1991) Symmetry and ornamentation are correlated in the peacock’s train. Anim Behav 42: 1020-1021

Møller AP, Pomiankowski A (1993) Why have birds got multiple sexual ornaments? Behav Ecol Sociobiol 32: 167-176

Palmer AR (2000) Quasi-replication and the contract of error: lessons from sex ratios, heritabilities and fluctuating asymmetry. Ann Rev Ecol Sys 31: 441-480

Petrie M, Halliday T (1994) Experimental and natural changes in the peacock’s (Pavo cristatus) train can affect mating success. Behav Ecol Sociobiol 35: 213-217

Petrie M, Halliday T, Sanders C (1991) Peahens prefer peacocks with elaborate trains. Anim Behav 41: 323-331

Prum RO (2006) Anatomy, physics and evolution of avian structural colors. In: Hill GE, McGraw KJ (eds) Bird coloration, vol. 1. Mechanisms and measurement. Harvard University Press, Cambridge

Seton ET (1896) Studies in the art anatomy of animals: being a brief analysis of the visible forms of the more familiar mammals and birds. Macmillan, New York.

Shawkey MD, Estes AM, Siefferman LM, Hill GE (2003) Nanostructure predicts intraspecific variation in ultraviolet-blue plumage colour. Proc Roy Soc B 270: 1455- 1460

Takahashi M, Arita H, Hiraiwa-Hasegawa M, Hasegawa T (2008) Peahens do not prefer peacocks with more elaborate trains. Anim Behav 75: 1209-1219

Yasmin S, Yahya HAS (1996) Correlates of mating success in Indian peafowl. Auk 113: 490-492

Zi J, Yu X, Li Y, Hu X, Xu C, Wang X, Liu X, Fu R (2003) Coloration strategies in peacock feathers. PNAS 100: 12576-12578

87

SUMMARY

1. Peafowl have provided widely-cited evidence for female choice and the genetic benefits of mating with more ornamented males. However, several studies have failed to reach a consensus with respect to which male trait and which signaling modality females primarily use when assessing mates.

2. Removing a large number of eyespot feathers decreases male mating success, yet there is substantial variation in mating success among full-trained males that cannot be explained by eyespot number. I confirmed these two previous results and showed that they are not contradictory, since the removal treatment modifies males beyond the normal range of eyespot number. Mate choice in peafowl may thus be based on multiple cues.

Females may avoid males missing a large number of eyespots via a threshold-based mechanism, while choosing among full-trained males based on some other cue.

3. The wing-shaking display is performed with females positioned behind the male, and it is elicited most often when a female is positioned on his shaded side. The pre-copulatory train-rattling display is performed with females directly in front, and males oriented at about 45° relative to the sun. Train-rattling may therefore function to enhance a plumage colour signal. The wing-shaking display may allow males to move females into a specific viewing geometry relative to the sun for communication of this colour signal.

4. Despite recent controversy, visual signaling is likely important in peafowl courtship.

The sun orientation of the train-rattling display points to the potential importance of the iridescent eyespot colouration as a courtship signal. Future studies of the signal content of eyespot colour and its use in the context of multiple cues will help clarify the function of complex sexual ornaments. 88

APPENDIX

Table A.1 Sampling for model presentation experiment from Chapter 3, with the types of trials completed listed by individual male (n=17). I completed 33 trials in total, but was unable to perform all four trial types with each individual.

AM PM

Male EAST WEST EAST WEST

1 9 9 9 9

2 9 9 9 9

3 9 9 9

4 9 9

5 9 9

6 9 9

7 9 9

8 9 9 9

9 9 9

10 9 9

11 9

12 9

13 9

14 9

15 9

16 9

17 9