The evolution and regulation of cooperation in the wild

Submitted by Lindsay Walker to the University of Exeter as a thesis for the degree of Doctor of Philosophy in Biological Sciences in December 2015

This thesis is available for Library use on the understanding that it is copyright material and that no quotation from the thesis may be published without proper acknowledgement.

I certify that all material in this thesis which is not my own work has been identified and that no material has previously been submitted and approved for the award of a degree by this or any other University.

Signature: ………………………………………………………….. Lindsay A. Walker

2

Summary

SUMMARY

In cooperatively breeding societies, where individuals (termed ‘helpers’) provide care to young which are not their own, group members can vary substantially in their contributions to cooperative activities. Individuals are expected to exhibit higher levels of cooperative investment if the benefit of performing that behaviour greatly outweighs the cost of performing that behaviour. This may be achieved by directing investment towards kin (thereby maximising indirect fitness benefits) and/or attaining large direct fitness benefits. In this thesis, I explore whether direct fitness benefits shape patterns of helping behaviour in the cooperatively breeding white-browed sparrow weaver (Plocepasser mahali). White-browed sparrow weavers live in year round territorial groups with high reproductive skew, comprising a dominant pair and subordinates of both sexes. Although all group members contribute to a wide range of highly conspicuous cooperative activities, there is large inter-individual variation in investment. In chapter 2, I use simulated territorial intrusions to show that sexually-selected direct benefits shape the expression of sentinel behaviour. In chapter 4, I provide evidence that the direct benefits associated with either the pay-to-stay or social prestige hypotheses are unlikely to modulate patterns of provisioning in male white-browed sparrow weavers. Evidence of marked individual differences in contributions to offspring care in cooperative societies is also generating increased interest in the proximate causes of such variation. In chapter 5, I use within-individual measurements to demonstrate that variation in provisioning effort is not directly regulated by variation in circulating levels of prolactin (a pituitary hormone). The evidence does suggest, however, that provisioning behaviour may be induced by exceeding a threshold hormone level. Individual contributions to parental behaviours (as opposed to alloparental) may be shaped by constraints associated with life-history traits. In chapter 3, I show that parents in white-browed sparrow weaver societies perform different provisioning rates yet employ similar food allocation tactics, and that these patterns are expected in tropical living species. Combined, these findings provide insights into the selection pressures that may shape individual contributions to cooperative activities.

3

4

Acknowledgements

ACKNOWLEDGEMENTS

If there is one thing that being a member of CLES has taught me, it’s that field workers are weirdos. Proper mad-hatters. And I am proud to be one of the gang.

I would like to thank my supervisory team Andy Young and Nick Royle for their invaluable insights, encouragement and support throughout my PhD research.

Andy - thanks in particular for your endless enthusiasm, and for attempting a ceilidh by head-torch in the Kalahari. The patient training and advice I received from Dom Cram (catching and bleeding sparrow weavers) and Jenny York

(behaviour and implant protocols) was invaluable - my PhD would have been so much harder without this, so massive thanks to you both. Also, extra thanks to

Dom for driving me all the way to Kimberley hospital in clapped-out Blue, and for introducing me to the epic Leslie Knope.

Nigel Bennett at the University of Pretoria kindly provided logistical support during my PhD, for which I am very grateful – this meant I could try to come face to face with a dugong. Alas, the dugong evaded me. Thanks to Simone

Meddle for conducting my hormonal analysis, and for hosting me in sunny

Edinburgh – during which, I managed to sneak a visit to the pandas. Gus van

Dyk, Dylan Smith, Cornelia le Roux, Nicolene Kotze and Jeanine Engelbrecht also deserve a massive thank you for providing invaluable logistical support in the field that enabled my research to go as smooth as is possible in the

Kalahari.

Tswalu…. My memories of you sway from dangerously dark to beautifully bright. Thank you for it all. Several people made the beautifully bright times that

5

Acknowledgements bit more sparkly, and whose generosity helped me survive the long seasons.

Dylan and Theresa Smith – thank you for sharing sensational animal stories, gin picnics, amazing food and for supporting me during the sleep deprivation.

Cornelia and Nicolene – thank you for looking out for me and for making me smile. Jo de Wilde, Adrian Bantich, Barry Peiser, Catherine Phillips, Jacques de

Vries, Peter le Roux – thank you all for the tasty braais, epic game drives and the dance moves. Pete, Jacques, Cath, Nora Weyer – thank you for sharing our converted police station home, for generously allowing me to steal your sweet bakes, and your never-ending kindness. I would also like to thank all of the

Sparrow Weaver Project assistants during my time in the Kalahari for remaining smiley even during the not-so-rainy wet seasons. Special thanks (in order of how I met you) to Jen Sturgeon, Lynda Donaldson, Cheryl Mills, Kat Wells and

Linda Tscherrin. You guys in particular ensured that fieldwork went as smooth as the harsh Kalahari would allow, that car journeys in Blue and Toby were always fun, and that non-fieldwork time was full of laughter and fantastic food - for which I am most grateful. I am also thankful to whoever installed an oven at

Dedeben, without which I would not have been able to bake my way through the hectic fieldwork.

To the creatures great and small of the ‘green’ Kalahari, thank you for poking your noses into my daily fieldwork life and for providing me with such amazing adventure stories. These awe-inspiring animal escapades kept me alive and smiling during long field seasons. Special thanks to Kagali & her stocky calf for their calm curiosity, Badgie for the rugby tackle, Zebie for joining us for dinner,

Scarback for your gentle inquisitiveness, the Rockstar group for your humorous antics, Fluffballs & their noisy pups for the playful performances, and Piglet for

6

Acknowledgements the softest nose-peck ever. Finally, thanks to LPPM for showing me that not all sparrow weavers are feisty wee thugs.

To my wonderful parents, Lewis and Pam – it’s only easy to leap forward into new adventures if you know you have a supportive base to fall back onto. An enormous thank you for your incredible encouragement over the years, for understanding my long absences, and for always being there for me - no matter what the time difference.

The following incredible people are responsible for a humungous number of spontaneous adventures, ‘knitting’ nights, sensational sing-a-longs, sea- swimming jaunts, Zumba-sweats, intentzz camping escapades, cheese overdoses and manic night-long parties that ensured my time here in Cornwall was truly amazing; Jenni McDonald, Neil McDonald, Hannah Peck, Harry

Marshall, Siobhan O’Brien, Matt Silk, Jenni Sanderson. A MASSIVE THANK

YOU for sharing your own special weirdness with me! I’m not sure who or what I would be without you – thank you all so very much for absolutely everything.

For the never ending supply of smiles, supportive chats, conference shenanigans and funny tales, thank you also to; Melanie Sullivan, Fiona

Mactaggart, Ian Cleasby, Freydis Vigfusdottir, Bea Downing, Julian Evans,

David Fisher, Chris Beirne, Alex Duff, James Duffy and Sheridan Willis. For adding gigantic sparkles to the Big London Adventure, thanks to Lucy

Robinson, Karla-Luise Herpoldt, Dan Slade, Henry Lau and Steph Osborne.

Finally, an enormous thank you to my beautiful sister Kerry. Your positivity, determination, strength and sunshine-ness inspires me every single day. I am totally blessed and humbled to have such a mind-blowingly awesome and supportive sister. Ta pal.

7

8

Table of Contents

TABLE OF CONTENTS

Summary ...... 3

Acknowledgements ...... 5

Table of Contents ...... 9

List of Figures ...... 13

List of Tables ...... 15

Author’s Declaration ...... 17

Chapter 1 ...... 19

General Introduction ...... 19 1.1 Overview ...... 21 1.2 Cooperatively breeding societies ...... 21 1.3 Kin selection: the importance of relatedness ...... 22 1.4 Potential ultimate explanations for individual contributions to cooperative activities ...... 24 1.5 Potential proximate explanations for individual contributions to cooperative activities ...... 26 1.6 Understanding contributions to parental care ...... 28 1.7 Study System ...... 30 1.8 Thesis aims and structure ...... 32 Chapter 2 ...... 35

Sexually-selected sentinels? Evidence of a role for Intra-sexual in sentinel behaviour ...... 35 2.1 Abstract ...... 37 2.2 Introduction ...... 39 2.3 Methods ...... 44 2.4 Results ...... 55 2.5 Discussion ...... 63

9

Table of Contents

Chapter 3 ...... 73

Sex differences in provisioning behaviour but not resource allocation in a subtropical cooperatively breeding bird ...... 73 3.1 Abstract ...... 75 3.2 Introduction ...... 77 3.3 Methods ...... 83 3.4 Results ...... 89 3.5 Discussion ...... 93 Chapter 4 ...... 101

No evidence of a role for social prestige or payment of rent in the helping behaviour of a cooperatively breeding bird...... 101 4.1 Abstract ...... 103 4.2 Introduction ...... 105 4.3 Methods ...... 116 4.4 Results ...... 124 4.5 Discussion ...... 135 Chapter 5 ...... 145

Prolactin and the regulation of offspring provisioning by parents and helpers in a cooperatively breeding bird ...... 145 5.1 Abstract ...... 147 5.2 Introduction ...... 149 5.3 Methods ...... 155 5.4 Results ...... 165 5.5 Discussion ...... 175 Chapter 6 ...... 183

General Discussion ...... 183 6.1 Overview ...... 185 6.2 Individual contributions to cooperative provisioning...... 188 6.3 Future research directions for the white-browed sparrow weaver system .... 194 Appendices ...... 197

A.1 Sexually-Selected Sentinels? Evidence of a role for intra-sexual competition in sentinel behaviour ...... 197 A.2 Sex differences in provisioning behaviour but not resource allocation in a subtropical cooperatively breeding bird ...... 198

10

A.3 No evidence of a role for social prestige or payment of rent in the helping behaviour of a cooperatively breeding bird ...... 200 A.4 Prolactin and the regulation of offspring provisioning by parents and helpers in a cooperatively breeding bird ...... 202 References ...... 209

11

12

List of Figures

LIST OF FIGURES

Chapter 2: Sexually-selected sentinels? Evidence of a role for intra-sexual competition in sentinel behaviour

Figure 2.1. The cooperative contributions to sentineling of each dominance and sex class 59

Figure 2.2. The sentinel activity of dominant males pre- versus post-playback of either the control or male solo song playback treatments 60

Figure 2.3. The total duet output of the resident pair in response to the playback of male solo song positively predicted the subsequent sentinel effort of the resident male 61

Figure 2.4. Latency to (a) first duet of resident pair and (b) first entry of dominant male to playback tree for dominant males exposed to male solo song in two contexts 62

Chapter 3: Sex differences in provisioning behaviour but not resource allocation in a subtropical cooperatively breeding bird

Figure 3.1. (a) Provisioning rates were significantly higher for dominant females than males. (b) Dominant females and males spent a similar duration in the nest cup per provisioning event 91

Figure 3.2. The proportion of feeds that were allocated to the nestling that was: first to beg; the smallest; positioned at the front of the nest cup 92

Chapter 4: No evidence of a role for social prestige or payment of rent in the helping behaviour of a cooperatively breeding bird

Figure 4.1. Investigating the potential for provisioning behaviour to act as a payment or signal individual quality 128

Figure 4.2. Provisioning events for dominant males were not significantly predicted by presence or absence of immigrant subordinate male within the social group 129

Figure 4.3. The extent to which males synchronise their provisioning visits with those of the dominant female or dominant male 130

13

List of Figures

Chapter 5: Prolactin and the regulation of offspring provisioning by parents and helpers in a cooperatively breeding bird

Figure 5.1. Circulating levels of prolactin were significantly predicted by (a) Class, and (b) Year 170

Figure 5.2. Individual provisioning rates were significantly predicted by (a) Class, and (b) Year 171

Figure 5.3. The relationship between two measures of provisioning effort and circulating levels of prolactin in white-browed sparrow weavers 172

Figure 5.4. Circulating prolactin levels and individual provisioning rates for the two implant categories (VIP, Sham) of dominant male white-browed sparrow weavers 174

14

List of Tables

LIST OF TABLES

Chapter 4: No evidence of a role for social prestige or payment of rent in the helping behaviour of a cooperatively breeding bird

Table 4.1. Predictions arising from pay-to-stay and social prestige hypotheses regarding different elements of provisioning in white-browed sparrow weaver societies 114

Table 4.2. Results of Bayesian GLMM investigating whether immigrant subordinate males and natal subordinate males differ in the extent to which their provisioning visits are synchronised with those of the dominant female 132

Table 4.3. Results of Bayesian GLMM investigating whether immigrant subordinate males and natal subordinate males differ in the extent to which their provisioning visits are synchronised with those of the dominant male 132

Table 4.4. Results of Bayesian GLMM investigating whether older natal subordinate males and younger natal subordinate males differ in the extent to which their provisioning visits are synchronised with those of the dominant male 133

Table 4.5. Results of Bayesian GLMM investigating whether older natal subordinate males and younger natal subordinate males differ in the extent to which their provisioning visits are synchronised with those of the dominant female 133

Table 4.6. Results of Bayesian GLMM investigating whether dominant males with or without an immigrant subordinate male in their social group differ in the extent to which their provisioning visits are synchronised with those of the dominant female 134

Chapter 5: Prolactin and the regulation of offspring provisioning by parents and helpers in a cooperatively breeding bird

Table 5.1. Table outlining all models used in this study, specifying the fixed and random effects 163

15

List of Tables

Appendix 1: Further information for “Sexually-selected sentinels? Evidence of a role for intra-sexual competition in sentinel behaviour”

Table A1.1 Model outputs of mixed-effect models as detailed in the appropriate analyses 197

Appendix 2: Further information for “Sex differences in provisioning behaviour but not resource allocation in a subtropical cooperatively breeding bird”

Table A2.1 Model outputs of mixed-effect models as detailed in the appropriate analyses 198

Appendix 3: Further information for “No evidence of a role for social prestige or payment of rent in the helping behaviour of a cooperatively breeding bird”

Table A3.1 Model outputs of mixed-effect models as detailed in the appropriate analyses 200

Appendix 4: Further information for “Prolactin and the regulation of offspring provisioning by parents and helpers in a cooperatively breeding bird”

Table A4.1 Table outlining all models detailed in the Appendix 4, specifying the fixed and random effects 203

Table A4.2 Model outputs of mixed-effect models as detailed in the appropriate analyses 205

16

Author’s Declaration

AUTHOR’S DECLARATION

The research in this thesis was supported by PhD studentship funding from the

Natural Environment Research Council (NERC), with additional support from a

BBSRC David Phillips Research Fellowship held by Dr Andrew J. Young.

All chapters in this thesis were written by Lindsay A. Walker, with comments and editing from Dr Andrew J. Young (all chapters), Dr Jennifer E. York (chapter

2) and Dr Nick J. Royle (chapter 3). Data for this thesis was collected from a wild population of white-browed sparrow weavers (Plocepasser mahali) as part of a long-term study led by Dr Andrew J. Young, and supported by Dr Jennifer

E. York until 2014.

Dedicated field volunteers assisted with the monitoring of the study population in the Kalahari Desert, South Africa, which greatly facilitated the implementation of behavioural and manipulation protocols as described in this thesis. Field assistants during the period of this thesis included Lynda Donaldson, Helen

Jones, Anthony Lowney, Cheryl Mills, Darren O’Connell, Robyn Silcock, Jenny

Sturgeon, Linda Tscherrin and Kat Wells. Chapter 4 uses long-term data collected before Lindsay A. Walker joined the research project, which were collected by numerous previous field assistants (see animalsocieties.org for details) and Dr Dominic L. Cram. Blood sample collection by Lindsay A. Walker

(chapter 5) was supported by Darren O’Connell and Linda Tscherrin.

17

Author’s Declaration

The Bayesian statistical analysis performed in chapter 4 was completed by Dr

Ian R. Cleasby. Hormonal laboratory analysis (chapter 5) was carried out by Dr

Simone L. Meddle.

The photograph used in this thesis is property of Lindsay A. Walker.

With these exceptions, I declare that the work contained in this thesis is my own and has not been submitted for any other degree or award.

18

Chapter 1

CHAPTER 1

GENERAL INTRODUCTION

19

Chapter 1

20

Chapter 1

1.1 OVERVIEW

Despite increasing interest in understanding the underlying causes for an individual to engage with cooperative activities, there still remains substantial unexplained variation in individual contributions both within and between groups. In this chapter, I first consider the importance of kinship for shaping patterns of individual cooperative efforts. I then go on to explore the potential for ultimate causes unrelated to kinship to select for cooperative behaviour in societies, with cooperation instead yielding direct fitness benefits that outweigh the cost of performing the behaviour. I also examine how proximate mechanisms may underpin variation in individual contributions to helping behaviour. Additionally, I consider how life-history traits associated with tropical living may shape contributions to parental care. Finally, I outline the aims of this thesis.

1.2 COOPERATIVELY BREEDING SOCIETIES

Cooperative breeding is defined as when some individuals within the group routinely provide care for offspring that are not their own. These individuals are termed ‘helpers’ and may assist with a wide range of care behaviours, including direct offspring care (e.g. tending eggs in the daffodil cichlid, Neolamprologus pulcher, Taborsky, 1984; Zöttl et al., 2013; provisioning behaviour in bird species, Cockburn, 2006; pup escorting in the banded mongoose, Mungos mungo, Cant et al., 2013) and indirect care towards young (e.g. sentinel duty in meerkats, Suricata suricatta, Santema and Clutton-Brock, 2013; nest defence in noisy miners, Manorina melanocephala, Arnold et al., 2005). The cooperative breeding social system has been observed across a wide-range of taxa, 21

Chapter 1 including fish (Taborsky, 2009), arachnids (Salomon and Lubin, 2007), crustaceans (Duffy and Macdonald, 2010), (Cockburn, 2006), mammals

(Lukas and Clutton-Brock, 2012) and insects (Queller and Strassmann, 1998).

Within these species, populations are usually divided into groups that are highly kin-structured; however, non-kin individuals may also be part of the social group

(e.g. chestnut crowned babblers, Pomatostomus ruficeps, Rollins et al., 2012; dwarf mongooses, Helogale parvula, Keane et al., 1994). Within these groups, high ranked or socially dominant individuals breed whereas lower ranked individuals generally do not (although in some species, subordinates do occasionally breed), thus generating a reproductive skew (Field and Cant,

2009). In the wild, reproductive skew is on a continuum, with reproduction in some species monopolised by a single breeder in a social group (Vehrencamp,

1983; Keller and Reeve, 1994). Indeed, with the exception of eusocial insects

(which show obligate sterility among workers), subordinates in cooperative societies retain the ability to breed themselves. Despite this, in many cooperatively breeding societies, subordinates perform behaviours that are costly to the individual and a benefit to others. This poses an evolutionary paradox – why help others at a cost to yourself? How can this be favoured by , where behaviours are selected to maximise individual fitness?

1.3 KIN SELECTION: THE IMPORTANCE OF RELATEDNESS

As mentioned previously, cooperatively breeding species tend to live in kin- structured groups. This finding has generated substantial research interest in the possibility that directing helping behaviour towards kin, and thereby impacting the reproductive success of relatives, could yield fitness benefits that

22

Chapter 1 offset the cost of performing the helping behaviour. Such fitness benefits arising from beneficial effects on relatives are termed indirect fitness benefits, and are separate to direct fitness benefits (which arise from effects on the reproductive success of the focal individual, Fisher, 1930). Hamilton, (1964) proposed that selection acts on the sum of these; both indirect and direct fitness, termed inclusive fitness. Under this assumption, helping behaviour may therefore be favoured by selection if r b – c > 0, where c is the cost of performing the helping behaviour, b is the fitness benefit to the recipient of the helping behaviour, and r is the relatedness between the two individuals (Hamilton, 1964). As a consequence, by preferentially directing helping behaviour towards kin (where indirect fitness benefits are greatest) an individual could increase their inclusive fitness. Termed kin selection (Maynard Smith, 1964), this inclusive fitness mechanism was the basis for an explanation of individual variation in helping behaviour.

There is marked variation in individual contributions to cooperative activities within cooperatively breeding societies, with some individuals investing substantially more than others (Komdeur, 2006). Such striking variation could be explained by the differing costs and/or benefits associated with performing a cooperative behaviour. Differences in relatedness to recipients provide one potential source of variation in these benefits received by cooperating. Indeed, several studies have shown that both the likelihood of helping at a nest and the provisioning of offspring are higher as relatedness to the breeding attempt increases (e.g. long-tailed tit, Aegithalos caudatus, Nam et al., 2010; Seychelles warbler, Acrocephalus sechellensis, Komdeur, 1994; bell miner, Manorina melanophrys, Wright et al., 2010; chestnut-crowned babbler, Browning et al., 23

Chapter 1

2012). However, a meta-analysis revealed that variation in relatedness only accounts for 10% of the variation in the likelihood to help (Griffin and West,

2003). This, coupled with the discovery that unrelated helpers also contribute to cooperative activities within a social group (e.g. meerkat, Clutton-Brock et al.,

2000; bell miner, Wright et al., 2010), suggests that direct fitness benefits, and not just indirect fitness benefits, must shape helping behaviour in at least some cooperatively breeding species.

This thesis explores how mechanisms unrelated to indirect benefits gained through traditional kin selection may select for the expression of cooperative behaviours.

1.4 POTENTIAL ULTIMATE EXPLANATIONS FOR INDIVIDUAL CONTRIBUTIONS

TO COOPERATIVE ACTIVITIES

As outlined above, studies that revealed the existence of hard-working unrelated individuals within cooperative social groups suggested that contributing to cooperative activities could also generate direct fitness benefits.

Thus, in some cases, direct fitness benefits alone could outweigh the costs of performing a helping behaviour and thereby select for the evolution of cooperative care (Clutton-Brock, 2002). Helping behaviour has been hypothesised to generate direct fitness benefits through a number of different mechanisms, including the acquisition of parenting skills and group augmentation (reviewed in Dickinson and Hatchwell, 2004). In this thesis, I combine natural observations and experimental approaches to explore the

24

Chapter 1 potential for direct benefits arising through three different routes to contribute to selection for cooperation: (i) sexually-selected direct benefits, (ii) accruing social prestige, and (iii) payment of rent.

First, I will consider the potential role of sexually-selected direct benefits. An individual may be selected to cooperate if cooperation also provides an opportunity to protect an individual’s reproductive success and/or social dominance from same-sex competitors. For example, engaging with a cooperative activity may provide a vantage point to detect intruders and/or to facilitate mate-guarding. To my knowledge, there has been no study to date that has explored the effect of intra-sexual competition on individual cooperative contributions (rather than individual competitive ability/traits, Clutton-Brock et al., 2006; Young and Bennett, 2013). Second, contributions to cooperative activities may act as an honest signal of individual quality (social prestige hypothesis; Clarke, 1989; Zahavi, 1995). As such, individuals may signal their higher quality by investing more in costly care behaviours in the presence of other group members to accrue ‘social prestige’. This may translate into an increased likelihood of being chosen as a mate or coalition partner. Evidence for this hypothesis is scare in wild-living animal populations. However, there is evidence that mate choice could have favoured the evolution of human cooperative behaviour via inter-sexual selection (explored in Phillips, 2015). For example, men gave more money to charity when observed by a member of the opposite sex rather than a member of the same sex or no observer (Iredale et al., 2008), and men’s public good contributions increased as they rated the female observer more attractive (van Vugt and Iredale, 2013). Third, where individual contributions to cooperation reduce the likelihood of eviction from a 25

Chapter 1 social group, individuals may stand to gain substantial direct benefits via the resulting access to communal resources and/or safety. An individual may therefore be selected to help if in return for contributing the individual is tolerated by the territory occupiers (pay-to-stay hypothesis; Gaston, 1978).

Evidence for this hypothesis is lacking from wild-living populations, with empirical support provided by just one species, a cooperatively breeding fish

(Bergmüller and Taborsky, 2005; Bergmüller et al., 2005).

This thesis also explores how proximate mechanisms may underpin individual variation in helping behaviour.

1.5 POTENTIAL PROXIMATE EXPLANATIONS FOR INDIVIDUAL

CONTRIBUTIONS TO COOPERATIVE ACTIVITIES

As outlined previously, there is substantial variation in individual contributions to cooperative activities within cooperatively breeding societies (e.g. Clutton-Brock et al., 1998; Clutton-Brock et al., 2002; Clutton-Brock, et al., 2004). To date, research has primarily focussed on seeking evolutionary explanations for this variation. However, the physiological mechanisms that underpin this variation are comparatively less well studied (Soares et al., 2010). Investigating the proximate mechanisms that regulate variation in behaviour can provide novel insights into the selection pressures shaping cooperation and hence improve our understanding of behavioural differences. Due to the fast-acting nature of hormones, the endocrine system presents a good candidate mechanism for regulating the expression of cooperative behaviours. Identifying the endocrine

26

Chapter 1 pathways that regulate cooperative behaviours, given the plethora of possibilities, may be best served by drawing on the extensive literature regarding the hormonal regulation of parental care in non-cooperatively breeding species.

For bi-parental species, numerous studies have now suggested that prolactin, a pituitary hormone, may regulate and/or facilitate a wide range of parental care behaviours (e.g. Buntin, 1996; Sharp et al., 1998; Chastel and Lormée, 2002;

Chastel et al., 2002; Angelier et al., 2006; Chastel et al., 2005; Sockman et al.,

2006; Christensen and Vleck, 2008; Angelier et al., 2009). In birds, the transition from sexual activity to parenting is associated with increased prolactin secretion (Sharp et al., 1998; Sockman et al., 2006), with prolactin levels at their highest during parental care behaviours (Buntin, 1996). Indeed, many avian studies report a positive correlation between the intensity/quality of care and prolactin levels in parents (e.g. Chastel and Lormée, 2002; Duckworth et al.,

2003; Van Roo et al., 2003; Angelier et al., 2007; Boos et al., 2007). Although the majority of studies investigating the role for prolactin in care have focussed on incubation-related behaviours, interest is growing in understanding if prolactin also regulates offspring provisioning, with positive correlations also observed between care and prolactin levels (Duckworth et al., 2003; Ouyang et al., 2011). Research has also been conducted on the regulation of offspring provisioning via circulating levels of prolactin in cooperatively breeding species, with mixed results (Harris’ hawks, Parabuteo unicinctus, Vleck et al., 1991;

Florida Scrub-Jays, Aphelocoma coerulescens, Schoech et al., 1996; red- cockaded , Picoides borealis, Khan et al., 2001; meerkats, Carlson et al., 2006). This highlights that further correlative and experimental studies are 27

Chapter 1 required to investigate the causal role for prolactin to influence individual contributions to cooperative care in both parents and helpers. To date, the experimental manipulation of prolactin has never been performed in a cooperatively breeding species (but see Carlson et al., 2003 for a pilot study in meerkats). In this thesis, I explore whether natural variation in circulating levels of prolactin positively predicts individual contributions to offspring provisioning in cooperative societies by using within-individual measurements of both hormone samples and behaviour observations. I also experimentally test whether hormone implants can be used to manipulate prolactin levels and behaviour.

This thesis also explores how life-history traits may modulate sex-specific patterns of care behaviours.

1.6 UNDERSTANDING CONTRIBUTIONS TO PARENTAL CARE

As a general rule, parents in cooperatively breeding societies provide care for their offspring and are assisted by other individuals (helpers; see above).

Indeed, cooperatively breeding vertebrates may commonly have evolved from monogamous pair-breeding species (Cornwallis et al., 2010; Lukas and Clutton-

Brock, 2012). Any parental behaviours that increase the growth and survival of their offspring are termed parental care; performing these behaviours are at a cost to individual survival and future reproduction (Royle et al., 2012). The most common form of parental care in birds is bi-parental care, with 90% of bird species known to display this type of care (Cockburn, 2006). In bi-parental care, conflict between parents over their individual contributions to care is expected,

28

Chapter 1 as (i) parents are typically unrelated, and (ii) parental care is costly (Trivers,

1972; Clutton-Brock, 1991; Parker et al., 2002; Lessells, 2012). Such sexual conflict over the division and amount of care is expected as each carer would benefit from the other parent performing a larger share of the work (Trivers,

1972; Houston and Davies, 1985). Additionally, males and females generally do not value current and future reproduction in the same way, due in part to extra- pair mating (Westneat et al., 1990). Although the connection between mating patterns and the evolution of care remains poorly understood (Alonzo, 2010), different mating systems can alter the balance of costs and benefits of performing care behaviours. Extra-pair paternity reduces the parentage assurance of the social mate, resulting in males valuing care less than females

(Westneat and Sherman, 1993; Kokko and Jennions, 2008). This sexual conflict over care can lead to sex-specific care patterns, which can in turn intensify sexual conflict (Andersson, 1994; Lessells, 2002; Royle et al., 2012). Indeed, with bi-parental care the two parents often display different levels and types of care (e.g. Møller and Cuervo, 2000; Trumbo, 2006; Sonerud et al., 2014). To date, studies of sexual conflict over parental investment have focussed on temperate zone bi-parental species. Few studies have considered the effect of sexual conflict on parental care patterns in tropical species, where life-history traits may impact the resolution of sexual conflict.

The impact of sexual conflict on parental investment may be modified by variation in life-history traits. For example, tropical birds generally have smaller clutch sizes than temperate zone species (Jetz et al., 2008); a pattern that is thought to reflect a constraint imposed by limited food availability and/or nest risk (Stutchbury and Morton, 2001; Martin, 2015). As such, females 29

Chapter 1 will lay clutches that reflect these selection pressures rather than producing a clutch size that maximises male investment in care (with male care becoming more valuable as clutch size increases, Davies and Hatchwell, 1992; Smith and

Härdling, 2000). Consequently, females may be limited in the amount of care they can sequester from males via an increase in clutch size (shown for bi- parental temperate species, Smith and Härdling, 2000). Additionally, the relatively shorter nestling period of tropical species, thought to be favoured due to a higher risk of nest predation (Martin, 1995), might reduce selection to preferentially feed the smaller nestling given the potential to instead advance the fledging date of the larger nestling. In this thesis, I explore the potential for life-history traits typical of tropical species to have modified the outcome of sexual conflict over parental care contributions.

1.7 STUDY SYSTEM

The white-browed sparrow weaver, a cooperatively breeding bird, lives in groups of two to twelve individuals and defends year-round territories in sub-

Saharan Africa (Collias and Collias, 1978; Wingfield and Lewis, 1993). Social groups comprise of one dominant pair and subordinates of both sexes. While both sexes frequently delay dispersal from their natal group, emigration to a non-natal group is not uncommon, resulting in groups with a mixed kin-structure

(Harrison et al. 2013a). As rain-dependent breeders, the timing of major rains dictates breeding, which may occur at any time during the Southern summer

(October to April). Within the group, reproduction is monopolised by the dominant individuals (Harrison et al., 2013a). Dominant females secure 100% of maternity in each breeding attempt (Harrison, et al., 2013a), but dominant

30

Chapter 1 males lose 12-18% of paternity to extra-group males (principally dominant males in other groups, Harrison et al. 2013a; Harrison et al. 2013b). Mixed- paternity clutches are infrequent (13% of 38 clutches from females who engaged with extra-group mating between 2007 and 2010, Harrison et al.,

2013b). Once gained, dominance positions are held by individuals for the rest of their relatively long lives. Male white-browed sparrow weavers sing dawn song throughout the breeding season (Voigt et al., 2006). This distinct song type is sung principally by the dominant male during peak reproductive periods following high rainfall (Wingfield and Lewis, 1993; York, 2012). All members of the social group engage with a wide range of cooperative activities, including provisioning nestlings, territory defence and sentinel behaviour (Collias and

Collias, 1978; Lewis, 1982; Ferguson, 1987).

White-browed sparrow weavers are an ideal study system to explore the proximate and ultimate causes of variation in helping behaviour. First, the system is highly tractable; groups maintain and defend year-round territories, birds can be reliably captured from their roost chambers at night, and individuals can safely be fitted with metal and colour rings. This permits repeated behavioural observations and biological sampling from specific individuals with known life-histories. Second, the breeding biology facilitates close observation of breeding stages; nest structures are highly distinguishable from roost structures, each group has a maximum of one active nest at any time, and clear dominance-related behaviours allow the reliable assignment of social dominance (and hence breeding status) in the field. Additionally, males and females are readily distinguished in the field from around six months of age by beak color; males have dark-brown beaks with females displaying a paler 31

Chapter 1 horn color (Leitner et al. 2009; Harrison et al. 2014). These traits, coupled with the fact that their various cooperative activities are highly conspicuous, makes the white-browed sparrow weaver a fantastic study system to explore the selection pressures for an individual to engage with cooperation and parental care.

1.8 THESIS AIMS AND STRUCTURE

The aim of this PhD thesis is to investigate potential proximate and ultimate causes of individual variation in contributions to cooperative behaviour in a social vertebrate. The research will strive to improve our understanding of the variation in individual investment to parental and alloparental activities as frequently observed in wild systems. To achieve this, I explore a diverse range of topics and structure my research as follows;

In Chapter 2, I experimentally investigate whether intra-sexual selection has shaped the expression of a putatively cooperative behaviour, sentinelling. Few studies have considered whether competition between members of the same sex influences individual contributions to cooperation. Here I explore the effect of simulated male intrusions on dominant male sentinel behaviour, thereby investigating the potential for sexually-selected direct benefits to modulate individual contributions.

In Chapter 3, I investigate whether parental care patterns are influenced by life- history traits associated with tropical living. Using behavioural observations and 32

Chapter 1 within-nest cameras, I examine if dominant individuals (typically the parents in white-browed sparrow weaver social groups) display sex-specific patterns of provisioning and food allocation tactics, as has previously been documented in studies of temperate zone bi-parental bird species.

In Chapter 4, I investigate whether pay-to-stay or social prestige, two candidate mechanisms through which helping may yield direct fitness benefits, modulate contributions to provisioning using observation data collected over three years.

The pay-to-stay hypothesis outlines that an individual may be selected to help if in return for contributing the individual has access to the group or advantages of group-living. The social prestige hypothesis postulates that contributing to cooperative activities may act as an honest signal of individual quality, accruing

‘social prestige’ which may translate into an increased likelihood of being chosen as a mate or coalition partner.

In Chapter 5, I experimentally investigate whether proximate mechanism identified in the regulation of parental care in non-cooperatively breeding species also regulate cooperative contributions to care in a cooperatively breeding species. Using a combination of natural data, hormone implants and repeated within-individual sampling, I explore whether provisioning effort

(assessed in two ways: provisioning rate and average visit duration) is regulated by circulating levels of prolactin (a pituitary hormone) in white-browed sparrow weaver societies.

33

Chapter 1

In Chapter 6, I discuss the wider implications of the combined findings of the previous chapters.

Each chapter within this thesis is intended to be an independent piece of work for publication. As such, the first page of each chapter highlights the authorship list of the intended publication, and “we” is used throughout; however, this by no measure means that any part of this thesis is not my own work (with the exceptions outlined in the author’s declaration).

Additionally, there is some repetition of methods between the chapters.

References throughout are collated under one section at the end of this thesis.

34

Chapter 2

CHAPTER 2

SEXUALLY-SELECTED SENTINELS? EVIDENCE OF A ROLE FOR

INTRA-SEXUAL COMPETITION IN SENTINEL BEHAVIOUR

Lindsay A. Walker, Jennifer E. York, Andrew J. Young

Submitted to Behavioral Ecology

35

Chapter 2

36

Chapter 2

2.1 ABSTRACT

While the evolutionary mechanisms that favour investment in cooperative behaviours have long been a focus of research, comparatively few studies have considered the role that sexual selection may play. For example, evolutionary explanations for sentinel behaviour (where one individual assumes an elevated position and scans the surroundings while other group members forage nearby) have traditionally focused on the inclusive fitness benefits arising from its effects on predation risk, while its potential role in defence against intra-sexual competitors remains largely unexplored. Here, we provide experimental evidence of a role for sentinel behaviour in intra-sexual competition, in a cooperatively breeding songbird, the white-browed sparrow weaver

(Plocepasser mahali). First, dominant males sentinel substantially more than other group members (even when controlling for variation in age and body condition), consistent with a role for sentineling in intra-sexual competition for mates and/or territory. Second, experimental playback of an unfamiliar male’s solo song elicited a marked increase in sentineling by the dominant male, and the vocal response to the playback also positively predicted his sentinel effort following the simulated intrusion. A second experiment also suggests that sentineling may facilitate mounting rapid anti-intruder responses, as responses to intruder-playback occurred significantly earlier when the dominant male was sentineling rather than foraging at playback onset. Together, our findings provide rare support for the hypothesis that sentinel behaviour plays a role in intra-sexual competition, and so highlight the potential for sexually-selected direct benefits to shape its expression in this and other social vertebrates.

37

Chapter 2

38

Chapter 2

2.2 INTRODUCTION

While the evolutionary mechanisms that favour investment in cooperative behaviours have long been a focus of research, comparatively few studies have investigated the role that sexual selection may play (Reyer, 1990; Dickinson &

Hatchwell, 2004; Boomsma, 2007; DuVal, 2013). For example, contributions to sentinel behaviour (where one individual watches from an elevated position while other group members forage nearby) are widely considered to reflect inclusive fitness benefits arising from sentineling serving an anti-predator function. Indeed, a wide range of evidence supports this view: sentinels frequently alarm call at predators (Ferguson, 1987; Beynon & Rasa, 1989;

McGowan & Woolfenden, 1989; Clutton-Brock, 1999; Wright et al. 2001a), may detect predators more readily than foraging group members (Rasa, 1986;

Manser, 1999; Ridley et al. 2010), and individuals may sentinel at higher rates in response to increased predation risk (Ferguson, 1987; Clutton-Brock, 1999;

Ridley et al., 2010; Radford et al. 2011; Santema & Clutton-Brock, 2013; Kern and Radford, 2014). However, individual contributions to sentineling could also be attributable in part to fitness benefits accrued through alternative mechanisms, such as a role for sentineling in intra-sexual competition. For example, sentineling could provide a vantage point for detecting and responding to same-sex competitors, who might otherwise threaten a resident individual’s reproductive success and/or social dominance. Where this is the case, sexually selected direct benefits may play a key role in shaping contributions to sentinel behaviour.

39

Chapter 2

Consistent with a role for sentineling in intra-sexual competition over matings, studies of individual contributions to sentinel behaviour in several social species have found that males sentinel at higher rates than females (e.g. Florida scrub jay, Aphelocoma c. coerulescens: Hailman et al. 1994; vervet monkey,

Cercopithecus aethiops: Baldellou & Henzi, 1992; meerkat, Suricata suricatta:

Clutton-Brock et al. 2002; Arabian babbler, Turdoides squamiceps: Wright et al.

2001c). These findings highlight the possibility that sentinel behaviour yields differential benefits to males through a mechanism other than the mitigation of predation risk. Sentineling could enhance a male’s ability to detect, repel and advertise his presence to same-sex competitors who might otherwise contest his paternity and/or social dominance. While females too may certainly benefit from vigilance against same-sex intruders (as intra-sexual competition for dominance status in cooperatively breeding vertebrates may be at least as intense among females as males; Clutton-Brock et al., 2006; Young & Bennett,

2013), males may frequently enjoy greater benefits from doing so given the threat that even transient males may pose to a resident male’s paternity (e.g.

Young et al. 2007; Harrison et al. 2013b). As most of the species that sentinel typically live in extended family groups (e.g. Arabian babbler: Wright et al.,

2001a, 2001c; dwarf mongoose, Helogale parvula: Rasa, 1986; Florida scrub jay: McGowan and Woolfenden, 1989; white-browed sparrow weaver,

Plocepasser mahali: Harrison et al. 2013b), threats to a male’s reproductive monopoly may principally be posed by unrelated extra-group males, whether prospecting males (who may threaten both paternity and dominance; Young et al. 2007; Mares et al. 2012) or extra-group resident dominants (who may constitute the principal threat to another dominant’s paternity; Richardson et al.

2001; Harrison et al., 2013b). Indeed, observations that dominant male Arabian

40

Chapter 2 babblers sentinel at the highest rates, and emit territorial calls at higher rates than other classes when sentineling, led Wright et al. (2001c) to suggest that such males may accrue additional benefits from sentineling if it facilitates monitoring and calling to neighbouring groups. Sentineling may also facilitate the monitoring and mitigation of within-group threats to a male’s reproductive monopoly, which may be more acute in instances where more complex kin structures leave resident subordinate males unrelated to one or more resident females (Koenig and Haydock, 2004; Young, 2009). It has also been suggested that fitness costs entailed in sentineling could leave an individual’s sentinel effort an honest signal of their quality (Zahavi and Zahavi, 1997; see also

Wright, 1999). Any role for sentineling in intra-sexual competition could therefore conceivably extend beyond facilitating the detection and repulsion of same-sex competitors, to signalling the sentinel’s quality to those competitors and/or the mates for which they compete. Whether sentineling does play a role in intra-sexual competition, however, has yet to be formally tested.

Here we investigate whether sentinel behaviour plays a role in intra-sexual competition, using a combination of observational datasets and playback experiments in a wild population of the white-browed sparrow weaver, a year- round territorial cooperatively breeding songbird. White-browed sparrow weavers live in groups of 2 to 12 individuals, consisting of a dominant breeding pair and subordinates of both sexes (Collias & Collias, 1978; Lewis, 1982;

Ferguson, 1988; Harrison et al. 2013a). Sentinel behaviour in white-browed sparrow weavers, as for other ground-foraging species, is characterized by an individual assuming a raised position and scanning its surroundings while members of its group forage nearby (Ferguson, 1987). Previous work has 41

Chapter 2 attributed an anti-predator function to sentinel behaviour in this species, on the basis of sentinels emitting alarm calls when predators are detected and alarm calling at higher rates than non-vigilant birds, and individuals showing higher rates of sentinel behaviour in higher risk micro-habitats (Ferguson, 1987).

Ferguson (1987) also reported that sentineling individuals produced territorial vocalizations at significantly higher rates than perched non-vigilant birds, highlighting the possibility of an additional role for sentineling in competition with extra-group individuals over territory, group membership and/or reproductive opportunities (see also Wright et al., 2001a). Extra-group males pose the principal threats to a dominant male white-browed sparrow weaver’s reproductive monopoly and social dominance. Within-group subordinate males, including immigrant individuals, have never been known to secure paternity within their social group (Harrison et al. 2013a); dominant males do lose 12-

18% of paternity, but do so exclusively to extra-group males (principally dominant males in other groups; Harrison et al. 2013b). Likewise, dominant males appear rarely to be usurped by their resident subordinate males, with extra-group males assuming dominance in 82.7% of monitored dominance turnovers (Harrison et al. 2014). While dominant females also principally lose dominance to extra-group females (88% of monitored dominance turnovers;

Harrison et al. 2014), extra-group females may pose little immediate threat to a dominant female’s maternity, as egg-dumping has never been detected in this species; the dominant female invariably monopolizes reproduction (Harrison et al. 2013). Consequently, dominant male white-browed sparrow weavers might be predicted to benefit more from investing in the detection and repulsion of same-sex intruders than dominant females. Whether sentinel behaviour does

42

Chapter 2 play a role in intra-sexual competition, perhaps by facilitating the detection and/or repulsion of such extra-group threats, is not known.

Specifically we test the hypothesis that sentinel behaviour plays a role in intra- sexual competition, by addressing the following three specific aims. First, we use observational data to investigate whether dominant male white-browed sparrow weavers contribute more to sentinel behaviour than other group members (while controlling for variation in age and body mass), as would be predicted if it conveys advantages in competition against extra-territorial threats to the dominant’s reproductive monopoly and/or social dominance. Second, we experimentally investigate whether dominant males increase their investment in sentinel behaviour in response to the playback of an unfamiliar male’s solo song relative to paired control playbacks. White-browed sparrow weaver males sing a distinct song type during the breeding season, known as the solo song (see

Voigt et al. 2006 for a spectogram of solo song). It is sung principally by dominant males, typically at dawn but also during the day, particularly during peak reproductive periods following high rainfall (Wingfield and Lewis, 1993;

Voigt et al. 2006; York, 2012). Extra-group males are known to intrude on occupied territories, where they have been observed producing solo song, eliciting vocal and physical responses (chases) from the resident dominant male. The playback of a foreign male’s solo song is therefore likely to be indicative of the intra-sexual competitive threat posed by extra-group males.

Finally, we investigate whether sentineling may also facilitate such counter- intruder responses, by establishing whether dominant males mount faster responses to foreign male solo song playbacks when individuals are acting as sentinels than when foraging. 43

Chapter 2

2.3 METHODS

2.3.1 STUDY SPECIES AND POPULATION

Data were collected from a colour-ringed study population at Tswalu Kalahari

Reserve in the Northern Cape province of South Africa (27°16′S, 22°25′E; see

Harrison et al. 2013a for a detailed site description) between December 2012 and March 2014. All birds were fitted with a metal ring and three colour rings for identification under SAFRING license 1444. Males and females are readily distinguished in the field from around 6 months of age by beak colour; males have dark-brown beaks with females displaying a paler horn colour (Harrison et al. 2014; Leitner et al. 2009). All individuals in the study were semi-habituated to observation with telescopes at approximately 18-20 m, following five years of regular exposure to observers at this distance. The dominant bird of each sex was determined by weekly monitoring of dominance-related aggressive, displacement and reproductive behaviours (as outlined in Harrison et al. 2013a and York et al. 2014). All protocols were approved by the University of Pretoria

Ethics Committee and complied with regulations stipulated in the Guidelines for

Use of Animals in Research.

2.3.2 NATURAL SENTINEL BEHAVIOUR OBSERVATIONS

Throughout the study, individuals were defined as engaging in sentinel behaviour (following Ridley et al. 2010), if they were perched in an elevated position >1 m above the foraging group members, actively scanning the surrounding area for a duration of >30 s. Similarly to pied babblers (Turdoides bicolor, Ridley et al. 2010) white-browed sparrow weavers are predominately ground foragers (Ferguson, 1988), with individuals using their beaks to dig in 44

Chapter 2 the substrate for prey. As sentinel behaviour was conspicuous and occurred at low frequencies (as detailed in the Results section, white-browed sparrow weaver groups had a sentinel in place a mean (±SD) of 21.0% (± 7.8) of the time, and individuals did not overlap in sentinel bouts), the sentinel effort of all group members could be accurately monitored simultaneously simply by noting the start and end times of all sentinel bouts during the observation session. All sentinel observation sessions were performed between 06:45 and 10:00, during the breeding season (October – April), and at times when groups lacked nestlings (so as to minimize the impact of trade-offs between concurrent reproductive investment and sentineling on patterns of sentinel effort). On first locating each focal group, a period of at least 15 min was used to establish group composition and to allow the birds to habituate to the observer, prior to beginning the sentinel observation session.

In order to contrast the sentinel effort of the different dominance and sex classes (to address aim 1), 53 sentinel monitoring observation sessions were conducted on 25 social groups (2-6 individuals per group, mean 3.85 individuals, with social groups visited on 1 – 4 separate observation sessions) that comprised a total of 25 dominant males and females, 28 subordinate males and 11 subordinate females. Following the 15 min habituation period, sentinel observation sessions were conducted for 30 or 60 min (according to logistical constraints) and the length of the observation session (short or long) was fitted as a random effect in our statistical models to control for any effect it might have on the proportion of time spent sentineling.

To then investigate whether the higher sentinel effort of dominant males relative 45

Chapter 2 to subordinate males (as revealed by analysis of the above data set) could be driven by associated variation in body condition, the sentinel effort of dominant and subordinate males was also quantified with a separate sample of 60 minute observation sessions (n = 12; from 12 social groups, containing a total of 11 dominant males and 10 subordinate males) all collected within a three day window (mean (± SD) = 0.94 ± 1.25 days) prior to the collection of matched morphometric data for these birds (from which body condition could be calculated; see below). This second data set was collected exclusively during periods when groups were incubating (days 4-13 of incubation; invariably after clutch completion), as this is when all of our study groups are routinely captured for the collection of morphometric data (males do not incubate in this species;

Harrison et al. 2013a).

2.3.3 BODY CONDITION CALCULATION

Birds were captured at night by flushing them from their individual roost chambers into a custom capture bag, and were subsequently returned to their roost chambers the same night by hand (Cram et al. 2015). Body mass was recorded to the nearest 0.01 g (Durascale 100; MyWeigh, Phoenix, AZ, USA) and tarsus length measured ±0.1 mm using callipers; all measurements were taken by one person (LW). To investigate the effect of body condition on individual sentinel effort, we calculated the Scaled Mass Index (SMI) as a proxy for body condition (following Peig and Green, 2009), as SMI has been argued to perform better than other methods of estimating body condition (Peig and

Green, 2009). Calculations of SMI entail the estimation of a scaling exponent

(SMA) from a reduced major axis regression of the logarithm of body mass on

46

Chapter 2 the logarithm of tarsus length (in this study, SMA = 0.22). The body mass values of all birds were then scaled to the expected equivalent for a bird of the mean tarsus length in our sample (24.88 mm), using the SMA (for full details, see Peig and Green, 2009). This SMI value for each bird was then used in our analysis as a proxy for body condition.

2.3.4 PLAYBACK EXPERIMENTS

We conducted two playback experiments with similar designs. For both experiments, a paired within-individual design was utilized to control for inter- individual differences, with each focal dominant male being exposed to one of two treatments on one day and the other treatment on the subsequent day, with treatment order reversed for each successive individual. All treatments involved playbacks, which were initiated via a wireless connection with a media player

(Philips Android Connect) once specific conditions were satisfied (see details below), and conducted at an amplitude of 66 dB (measured at 10 m from the speaker with a Voltcraft SL100 digital sound level meter (Voltcraft, Barking,

UK)). Playbacks were conducted using Jawbone (Jambox) portable speakers placed at a height of 1.5 m on the main sleeping roost tree at the centre of the focal group’s territory and the observer (LW) was stationed at 20 m from the playback speaker throughout the session. All behavioural observations (details provided below) were dictated and recorded on a DM550 Olympus recorder

(ME15 Olympus microphone), and were subsequently examined using Raven

Lite 1.0 (Cornell Laboratory of Ornithology, Ithaca, NY, USA, 2006).

47

Chapter 2

The playback audio tracks were produced using CoolEditPro 2.0 (Syntrillium

Software Corporation) from solo song recordings collected from 10 different dominant males (York, 2012; York et al. 2014) using a Sennheiser ME66 directional microphone with a K6 power module (2004 Sennheiser) and a

Marantz PMD660 solid-state recorder (DandM Holdings Inc.). Each playback track consisted of either a 3 min continuous section of male solo song (male solo song treatment) or a 3 min continuous section of the ambient background sounds recorded on the same track (control treatment), resulting in 10 pairs of unique male solo song tracks and control tracks each from a different original male. Care was taken to ensure that the playback tracks included either only male white-browed sparrow weaver solo song (male solo song treatment) or no conspecific vocalizations at all (control tracks) to avoid receiver responses to other conspecific vocalizations. So as to ensure that the source-male for each playback track was unfamiliar to the resident male to whom the track was played, we ensured that the recording and playback territories for each track were at least three territories apart (mean ± SD: 858 ± 240 m).

2.3.4.1 EXPERIMENT 1: DOES AN INTRA-SEXUAL CHALLENGE IMPACT DOMINANT MALE

SENTINEL BEHAVIOUR?

To investigate whether sentinel behaviour may function in intra-sexual competition, we simulated the presence of an unseen same-sex (male) intruder by conducting a 3 min audio playback of an unfamiliar male’s solo song (a song type performed principally by dominant males during the breeding season, but that can also be sung by subordinate males and intruders; York, 2012) and contrasted the behavioural response elicited with that elicited by a control playback. We conducted the playbacks on 10 white-browed sparrow weaver 48

Chapter 2 pairs (i.e. territorial groups containing just one dominant male and one dominant female with an established pair bond) to eliminate any potential complications arising from the presence of resident subordinates (e.g. competitive or compensatory responses to the sentinel effort of subordinates). All of the playbacks were conducted during the incubation phase of breeding, but while both the male and female were foraging together away from their nest. As dominant females do not display sentinel behaviour during incubation periods

(Walker and Young, unpublished data), this approach ensured that any changes in the dominant male’s sentinel behaviour following playback could not be attributed instead to him modifying his sentinel effort in response to an effect of the playback on the sentinel effort of the dominant female. That said, this approach could conceivably elicit larger responses from the dominant male than might be anticipated were the experiment conducted on groups containing subordinates, where other individuals might also be expected to respond.

The experiment proceeded as follows. Each focal pair received one treatment per day over two successive days, with the order of presentation of the solo song and control treatments alternated for each successive pair. Two focal pairs were visited on each experimental day and they received the opposite treatments. All observation sessions began between 05:45 and 08:30. Following the completion of an initial 15 min habituation period, all sentinel activity (start and end times for all sentinel bouts) by the resident pair was recorded for 30 min. The playback was then started once the foraging pair moved in front of the portable speaker. The distance of the resident pair to the speaker when playback was initiated was estimated by first noting their location on a detailed territory map and then taking a GPS location of that point once the observation 49

Chapter 2 session was complete. There was no significant difference in the distance from the speaker to the foraging pair between the two playback treatments (mean (±

SD) for Song treatment: 51.67 (± 28.40) meters, and Control treatment: 43.11 (±

2 19.90) meters; GLM: χ 1 = 0.66, P = 0.41). While the track played (for three minutes), the following behavioural responses and the identity of the individuals involved were recorded: (1) song production (all incidences of male solo song and duets [a distinct song repertoire mainly produced by the dominant pair;

Voigt et al. 2006]); (2) territory movements: leading movements (defined as one group member leaving and being promptly followed by the other, travelling in the same direction and arriving at same destination) and approach to playback tree (defined as landing in the tree containing the speaker). Recent evidence has shown that male solo song in white-browed sparrow weavers may not function as a signal of exclusive spatial ‘ownership’, instead potentially indicating individual dominance status (York, 2012). As such, monitoring only male solo song may not provide a reliable representation of defence of territory, brood and/or dominance position. The duet song has been interpreted as the collective defence of a territory, and therefore may better represent an attempt to expel an intruder to protect territory, brood and/or dominance position. As a result, both duet song and male solo song production were monitored.Once the

3 min playback was complete, a second 30 min post-playback behavioural observation session was conducted, in which the sentinel efforts of both group members were recorded (as per the session prior to the playback).

50

Chapter 2

2.3.4.2 EXPERIMENT 2: DOES SENTINELING FACILITATE RESPONSES TO AN INTRA-

SEXUAL CHALLENGE?

To determine whether the position of the dominant male (sentineling versus foraging on the ground) affects his latency to respond to a perceived territorial threat, we simulated the presence of a same-sex intruder in the two contexts by conducting a 3 min playback of unfamiliar male solo song (according to the playback methods described above). This experiment was conducted during the breeding season, but during periods when the focal group did not have eggs or nestlings present. As for experiment 1, the playbacks were conducted on the territories of resident pairs (n = 8). Each pair received a different male solo song playback track, and the same track was utilized in both contexts for the same pair (so as to standardize across contexts any impact the specific song elements might have on the scale of the perceived threat).

All eight resident pairs were visited on one day for their first treatment and again the next day for the opposite treatment, with the pairs successively allocated alternate first treatments (male sentineling versus male foraging). All observation sessions occurred between 06:45 and 12:05. In each case, following a 15 min habituation period, the 3 min male solo song playback was begun once the dominant male was either (1) participating in sentinel behaviour for ≥30 s, or (2) foraging with his social mate for ≥30 s. The distance of the resident pair to the speaker when playback was initiated was estimated by first noting their location on a detailed territory map and then taking a GPS location of that point once the observation session was complete. There was no difference in the distance from the speaker to the dominant male between the two playback contexts (mean (± SD) for Sentinel condition: 36.7 (± 11.4) 51

Chapter 2

2 meters, and Foraging condition: 35.7 (± 11.1) meters; GLM: χ 1 = 0.029, P =

0.87). During the 3 min playback and the ensuing 10 min, behavioural responses and the identity of the individuals involved were recorded exactly as outlined for Experiment 1 above. Four of the resident pairs used in this experiment were also used in Experiment 1, but there were at least 17 days

(mean (± SD) = 49 (± 27) days) between their exposures to male solo song playback, and the male solo song tracks used for these groups were different in the two experiments.

2.3.5 STATISTICAL ANALYSES

All statistical analyses were conducted with R v 3.0.3 (R Development Core

Team, 2014). All models were checked for normality of residuals, homogeneity of variance and overdispersion. Statistical modelling utilized a stepwise model simplification approach: initially all fixed terms (detailed below) were fitted together and then the non-significant terms with the least explanatory power were sequentially removed until a minimal adequate model was reached

(retaining only those predictors whose removal now yielded a significant reduction in the explanatory power of the model). All random terms (detailed below) were retained in the model throughout. The assessments of statistical significance reported for each of the fixed terms are those calculated from the change in explanatory power on removal of that term from the minimal model (if it was in the minimal model) or following its inclusion in and subsequent removal from the minimal model (if it had not been retained in the minimal model). See

Table A1.1 in Appendix 1 for full details of all explanatory variables investigated in each model.

52

Chapter 2

2.3.5.1 NATURAL SENTINEL BEHAVIOUR: DO DOMINANT MALES INVEST

DIFFERENTIALLY IN SENTINEL BEHAVIOUR?

First, we investigated whether dominant males invest more in sentinel behaviour than other dominance/sex classes, by modelling the factors that affect individual sentinel effort (measured in all analyses as the proportion of observation time that the focal individual spent sentineling, logit transformed) using a generalized linear mixed effects model with a normal error distribution.

The fixed effects of primary interest were sex and dominance status, and to specifically test for differential sentinel effort by dominant males we also fitted the interaction between these two terms. We also fitted social group size

(number of individuals present in the social group during the observation session) in the initial maximal model to control for its potential influence on individual sentinel effort. Individual identity, observation session identity and observation session length (30 or 60 min; see above) were fitted as random factors. Second, to investigate whether the higher sentinel effort of dominant than subordinate males detected in this first analysis could be attributed instead to an effect of age (as dominant males tend to be older; Harrison et al. 2013a), a generalized linear mixed effects model was performed using the subset of available data for males of known age (again with a normal error distribution and individual identity, observation session identity and observation session length fitted as random factors) with age, dominant status and group size as fixed factors. This analysis included 85 measurements of 36 males of known hatch date from 21 social groups. Third, to investigate whether the higher sentinel effort of dominant than subordinate males could be attributable instead to an effect of body condition (using a separate sentineling data set with matched body condition measures; see description above), a generalized linear

53

Chapter 2 mixed model with a penalized quasi-likelihood estimator (glmmPQL) and a quasi-binomial error distribution was implemented in the package MASS in R

(Venables and Ripley, 2002), with observation session identity fitted as a random factor (individual identity was not fitted as there was only one measure per individual). The glmmPQL approach using quasi-binomial error distribution was favoured to account for overdispersion.

2.3.5.2 EXPERIMENT 1: DOES AN INTRA-SEXUAL CHALLENGE IMPACT DOMINANT MALE

SENTINEL BEHAVIOUR?

To investigate the effect of the treatment (simulated intruder versus control) on the sentinel effort of the resident dominant male, a generalized linear mixed model with a penalized quasi-likelihood estimator (glmmPQL) and a quasi- binomial error distribution was again implemented in the package MASS in R

(Venables and Ripley, 2002). Observation session stage (pre- or post-playback) and playback treatment (male solo song or control track) were fitted as fixed effects along with the interaction between them, with resident pair identity fitted as a random effect. We used Spearman rank correlation to test for a relationship between the duet response of the resident pair (their total duration of duet song production during the three minute playback, in seconds) to foreign male solo song and the post-playback sentinel effort of the resident dominant male for each group.

54

Chapter 2

2.3.5.3 EXPERIMENT 2: DOES SENTINELING FACILITATE RESPONSES TO AN INTRA-

SEXUAL CHALLENGE?

To investigate whether the resident pair’s latency to (a) first produce a duet and

(b) first approach the playback tree differed between the two treatment contexts

(resident male in a sentinel position versus foraging), Wilcoxon paired tests were used.

2.4 RESULTS

2.4.1 NATURAL SENTINEL BEHAVIOUR: DO DOMINANT MALES INVEST DIFFERENTIALLY

IN SENTINEL BEHAVIOUR?

During non-breeding periods (times during the breeding season when groups lacked eggs and nestlings), foraging white-browed sparrow weaver groups had a sentinel in place a mean (±SD) of 21.0% (± 7.8%; n = 53 observation sessions of 25 groups) of the time, and sentinel bouts lasted a mean (± SD) of 179 (± 60) s (n = 200 bouts from 53 observation sessions). The sentinel bouts of group members were never observed to overlap, individuals were never observed being subjected to aggression while acting as a sentinel (n = 47 observation sessions of 20 groups had subordinates present, with 17 of these groups containing a subordinate male), and we had no reason to suspect that one individual had ever interfered with the sentinel efforts of another. The proportion of time that individuals contributed to sentinel activity was determined by a

2 significant interaction between their dominance status and sex (GLMM: χ 1 =

2 4.67, P = 0.031; Figure 2.1), and unrelated to group size (χ 1 = 0.032, P = 0.86).

Dominant males displayed significantly more sentinel effort than all other

55

Chapter 2 classes of individual, with dominant individuals exhibiting higher levels of sentinel behaviour than subordinates (see Figure 2.1).

Two further analyses indicate that the elevated contributions of dominant males cannot be readily attributed to variation among males in age or body condition.

Using the subset of data for males of known age (n = 85 measurements of 36 males of known hatch date from 21 social groups), a model of male sentinel effort confirmed that dominant males spent a significantly higher proportion of

2 their time sentineling than subordinate males (GLMM: χ 1 = 7.45, P = 0.006) even while controlling for a significant positive effect of male age on sentineling

2 (GLMM: χ 1 = 4.26, P = 0.039). The mean (± SD) proportion of time spent sentineling for dominant males was 0.13 ± 0.08, and for subordinate males was

0.017 ± 0.035. Utilizing data from incubation periods, when the body masses of dominant and subordinate males were assessed, there was no significant effect of body condition (as assessed using the Scaled Mass Index; SMI) on the sentinel contributions of males (glmmPQL: t7 = -1.66, P = 0.14) and, even with

SMI retained in the model, the effect of dominance status approached significance despite the comparatively small sample size (t7 = -2.32, P = 0.053; n = 11 dominant males and 10 subordinate males of known SMI from 12 groups).

2.4.2 EXPERIMENT 1: DOES AN INTRA-SEXUAL CHALLENGE IMPACT DOMINANT MALE

SENTINEL BEHAVIOUR?

The foreign male solo song playback elicited a robust duet response from the resident pair; they produced duets for a significantly longer total duration in

56

Chapter 2 response to the foreign male solo song playback than the control playback

(Inter-Quartile Range [IQR]: Male solo song playback = 20.85 -35.48 s; Control

= 0.00 - 3.63 seconds; Wilcoxon paired test: V = 55, P = 0.002, n = 10). During the male solo song playback, the dominant male led significantly more movements around the pair’s territory than the dominant female (paired t-test: t9 = -2.68, P = 0.025), and approached the playback tree first in all of the instances in which one or both birds entered the tree during the playback period

(7 of the 10 playbacks). Male solo song production did not occur following either the male solo song playback or the control playback, therefore only duet song production was used in our analysis.The male solo song playback was then followed by a marked increase in sentinel behaviour by the dominant male once the pair had returned to foraging, while the control playback was not, as indicated by a significant interaction between playback treatment (male solo song or control) and observation session stage (pre- or post-playback; glmmPQL: t 27 = -6.52, P < 0.001; Figure 2.2). Across groups, the vocal response of the resident pair to the foreign male solo song playback (the total length of the duet song produced during the three minute playback) significantly positively predicted the dominant male’s subsequent sentineling effort

(Spearman rank correlation: rs = 0.78, N = 10, P = 0.012); Figure 2.3; trendline is for demonstration only). The dominant females did not contribute to sentineling in either treatment, before or after the playback (the playback was conducted during the incubation period, when dominant females rarely sentinel even when away from the nest). None of the dominant females engaged in incubation during the observation periods.

57

Chapter 2

2.4.3 EXPERIMENT 2: DOES SENTINELING FACILITATE RESPONSES TO AN INTRA-

SEXUAL CHALLENGE?

When the three-minute foreign male solo song playback was initiated with the dominant male in a sentinel position, the resident pair produced their first duet a median of 13.50 (Inter-Quartile Range [IQR]: 11.50- 18.25) seconds after playback initiation. When the playback was initiated while these same dominant males were foraging on the ground, the dominant males first moved to an elevated position on a nearby bush or tree 27.00 (IQR: 20.25 - 29.75) seconds after playback initiation, and the first duet was only produced a further 53.00

(IQR: 32.25 - 135.80) seconds after this movement. The resident pairs’ first duet was therefore produced with significantly shorter latencies from playback initiation if the dominant male was on sentinel at playback initiation rather than foraging (Wilcoxon paired test: n = 8 males, V = 36, P = 0.008; Figure 2.4a).

The latency for the dominant male to enter the tree that contained the playback speaker was not significantly different between the two contexts (Wilcoxon paired test: V = 7, P = 0.15; Figure 2.4b), though in all cases but one the latency to approach the speaker was lower when the playback occurred while the male was sentineling. In all eight groups and in both contexts, the dominant male entered the playback tree before the dominant female.

58

Chapter 2

Figure 2.1. The contributions to sentineling of each dominance and sex class.

Bars present means ± S.E. The data derive from 53 observation sessions on 25

groups, yielding a sample of 25 dominant males, 25 dominant females, 28

subordinate males and 11 subordinate females.

59

Chapter 2

Figure 2.2. The sentinel activity of dominant males pre- versus post-playback of either the control or male solo song playback treatments (n = 10 groups). The bars present means ± S.E.

60

Chapter 2

Figure 2.3. The total duet output of the resident pair in response to the playback of male solo song (n = 10 groups) positively predicted the subsequent sentinel effort of the resident male once the pair had returned to the foraging context.

61

Chapter 2

Figure 2.4. Latency to (a) first duet of resident pair and (b) first entry of dominant male to playback tree for dominant males exposed to male solo song in two contexts (sentineling versus foraging; n = 8 groups).

62

Chapter 2

2.5 DISCUSSION

We used a combination of natural observations and field experiments to investigate whether sentinel behaviour may play a role in male intra-sexual competition over mates and/or territory. Our findings reveal first that dominant male white-browed sparrow weavers (who face extra-group threats to both their paternity and social dominance; Harrison et al. 2013a,b) display substantially more sentinel effort than other group members and that their differential contributions cannot be readily attributed to variation in age or body condition

(SMI). Second, the playback of foreign male solo song (which would otherwise be indicative of the presence of an extra-group male), elicited a robust vocal response from the resident pair coupled with a movement response led by the dominant male, and this was followed by a marked increase in sentinel effort by the dominant male once the pair had returned to foraging, none of which were observed in paired control playbacks. Indeed, the magnitude of the vocal response to the foreign male solo song playback predicted the magnitude of the resident dominant’s subsequent sentinel effort. Finally, the resident pair also mounted significantly swifter duet responses to the foreign male solo song playback if the dominant male was acting as a sentinel at the time of playback initiation rather than foraging, suggesting that sentinel behaviour may facilitate the rapid initiation of counter-intruder responses. While research to date has primarily focused on an anti-predator function for sentinel activity, our combined results suggest that sentinel behaviour may also play a role in intra-sexual competition, and that sexually-selected direct benefits may therefore have acted in concert with other mechanisms to shape contributions to sentinel behaviour.

63

Chapter 2

Consistent with the patterns of sentineling observed in Arabian babblers (Wright et al., 2001c), dominant male white-browed sparrow weavers display higher levels of sentinel activity than any other class of individual. Several studies have now reported that heavier individuals contribute more to sentinel behaviour (e.g. meerkat; Clutton-Brock et al. 2002; Arabian babblers; Wright et al. 2001c; pied babblers; Bell et al. 2010) and have confirmed the state-dependence of sentineling contributions with feeding experiments (Florida scrub jay: Bednekoff and Woolfenden, 2003; meerkat: Clutton-Brock, 1999; Arabian babbler: Wright et al. 2001c). The elevated sentinel contributions of dominant male white- browed sparrow weavers cannot be readily attributed to variation in body condition, however, as two recent studies have examined dominance-related differences in body condition in this species, and, while both report evidence of such differences among females, no dominance-related differences in body condition were found among males in either the first or second halves of the breeding season in either study (Harrison et al. 2013b; Cram et al. 2015).

Accordingly, when variation in body condition was statistically controlled in our model of male sentinel effort, dominance status approached significance (P =

0.053) despite the greatly reduced sample size. It is difficult to rule out any role for variation in nutritional state, however, as it remains possible that rank-related differences in hunger levels (conceivably independent of body reserves) could be contributing to the patterns observed.

Several studies of sentinel behaviour have also found that older individuals contribute more (e.g. Zahavi, 1990; Hailman et al.1994; Wright et al. 2001b,c), but the elevated efforts of dominant male white-browed sparrow weavers cannot be attributed simply to age-related variation, as dominant males showed 64

Chapter 2 significantly higher sentinel effort than subordinates while statistically controlling for effects of age. The differential efforts of dominant males also cannot be readily attributed to individuals interfering with the sentinel efforts of others (as envisaged by Zahavi and Zahavi, 1997 but see Wright, 1999), as we observed no evidence of aggression between sentinels or of one individual interrupting the sentinel bouts of another. The most plausible explanation for the differential sentinel efforts of dominant males would therefore appear to be that they stand to gain differential direct benefits from sentineling as it may facilitate the detection and/or repulsion of same-sex competitors, in particular extra-group males (who intrude on territories during the day and are the principal threats to a male’s paternity and social dominance; Harrison et al. 2013a,b; Harrison et al.

2014). Indeed, it is conceivable that the sentinel position also facilitates the monitoring of the resident dominant female’s movements, further facilitating mate-guarding.

The behavioural response to the playback of an unfamiliar male solo song supports the hypothesis that sentinel behaviour plays a role in intra-sexual competition. Immediately following the male solo song playback (and not following the control playback), there was a clear duet and movement response by the resident pair, the latter being led by the dominant male. Once the resident pair had returned to foraging, the dominant male substantially increased his investment in sentinel activity relative to the period prior to the playback, a change that was only evident following the male solo song playback and not the control playback. During the 52 sentinel bouts that the 10 focal dominant males conducted in the foraging period following the male solo song playback, the focal males never engaged in duet production with their mate and 65

Chapter 2 just once produced solo song. It seems unlikely therefore that sentineling in this context simply serves as a position from which to broadcast song (as suggested by Wright et al. 2001a,b,c). Instead, our findings are consistent with the hypothesis that sentineling provides a vantage point that may facilitate the detection and monitoring of same-sex intruders. As the experiment was conducted during the incubation period, the resident male’s response could have been motivated more by the defence of his dominant position than by a risk of lost paternity. It is also worth noting that the response to the solo song playback cannot be attributed specifically to our use of solo song per se, as it could instead reflect a generalized response to the detection of an unexpected conspecific (via cues within the song; Townsend et al. 2012). The dominant male’s sentineling response to the male solo song playback cannot be attributed instead to compensatory responses (or indeed coordination responses) by the dominant male due to changes in the sentineling effort of other group members, as the only group member present was the dominant female (as the playbacks were conducted on resident pairs, to rule out the complications posed by helper responses), and the dominant female did not modify her sentinel effort in response to either playback (she contributed nothing before and after both treatments, which is not uncommon during the incubation period; Walker and Young unpublished data).

It is quite possible that the dominant male’s sentinel response to the foreign male solo song playback is a response in part to the dominant female’s own vocal response to the playback, as while the dominant male clearly led the movement response to the playback, the concomitant marked increase in duet production could have been led by either the dominant male or female (duet 66

Chapter 2 production is so synchronous that the leaders of duets cannot be readily identified in the field). While duets are frequently interpreted as cooperative vocal responses that may function in the collective defence of territory (e.g. white-browed sparrow weaver: Ferguson, 1988; Voigt et al. 2006; Wingfield and

Lewis, 1993), they may also reflect sexual conflict (Marshall-Ball et al. 2006;

Tobias and Seddon, 2009), in which one sex may advertise their presence to putative extra-pair mates, eliciting an immediate response from their social partner that may serve in mate-defence. Our finding that the duet response of the resident pair to the male solo song playback positively predicted the dominant male’s sentinel response is consistent with this view, with the dominant male potentially scaling his subsequent sentinel response according to the dominant female’s vocal response to the playback. Alternatively, this positive association could reflect a shared anti-intruder function for both the duet and sentinel responses, with the expression of both being modulated according to individual variation in the male’s expected payoffs from repelling intruders or simply according to variation in his nutritional state.

A second alternative explanation could be that the duet response of the resident pair, and the sentinel response of the dominant male, may reflect a potential role of brood defence. In our study population, breeding attempts have been observed to be destroyed during both the egg and nestling stages through means not associated with nest predators typical of the area (e.g. smashed egg shell remains, long term data unpublished). Although the cause of these infrequent destruction events remain unclear, it is not unreasonable for the playback in this experiment to represent a threat to the brood, in particular as the playback was situated in the nest tree. Consequently, the duet response of 67

Chapter 2 the resident pair may be to expel the intruder to protect the current breeding attempt. As the playbacks were performed during incubation phase, the dominant female may have been unable to engage with a concomitant sentinel response due to internal constraints (e.g. body condition), thereby leaving the dominant male as the sole responder with increased sentinel activity.

Consistent with the hypothesis that the sentinel position confers an advantage in detecting and responding to intruders, our second experiment revealed that the latency from the initiation of foreign male solo song playback to the production of the first duet by the resident pair (a song type that functions in territorial defence; Wingfield and Lewis, 1993; Voigt et al. 2006) was significantly lower if the playback was initiated while the dominant male was sentineling rather than foraging on the ground. The delay in the duet response when dominant males were foraging rather than sentineling at playback initiation could be due in part to one or more of the following mechanisms: (i) males may simply tend to be in a poorer nutritional state when foraging than when sentineling (given the likely state-dependence of sentineling; Wright et al.

2001a,b,c; Bednekoff and Woolfenden, 2003) and so may differentially value continuing to forage relative to mounting a counter-intruder response; (ii) mounting an immediate counter-intruder response might entail an additional lost-opportunity cost for foraging males if they have to interrupt an active foraging attempt; and (iii) males that are foraging in cover may simply take longer to detect the playback than a sentineling male on an elevated perch.

While each of these mechanisms could explain why foraging dominant males take time following playback initiation to cease foraging and rise from the ground, none of them can readily explain why the delay from this point to the 68

Chapter 2 first duet produced was still significantly longer than the entire delay from playback initiation to first duet production for males that were sentineling at playback initiation (see results). This discrepancy suggests that foraging males, even after they have ceased foraging, may first take time to assess their environment and/or the location of the intruder before mounting a vocal response, while sentineling males may require little time for assessment as they have already been continuously monitoring their environment. Sentineling may therefore yield benefits in intra-sexual competition by facilitating both the efficient detection of intruders (whether visually or acoustically) and the subsequent mounting of swift responses.

While our results are consistent with sentinel behaviour simply facilitating the detection and repulsion of same-sex intruders, it is conceivable that sentinel behaviour also serves as a signal in this context (Zahavi, 1995; Zahavi and

Zahavi, 1997; Wright, 1999). Sentinel behaviour could, for example, highlight the presence of a resident dominant to passing would-be challengers, which could itself be sufficient to prevent intrusions in many cases. But where sentineling entails costs (e.g. via lost foraging time and/or exposure to risk e.g.

Ridley et al. 2013) it has also been hypothesized to serve as an honest signal of quality (e.g. Zahavi, 1995; Zahavi and Zahavi, 1997; Wright, 1999). The differential sentinel effort of dominant males and their sentinel responses to same-sex intruders could therefore also be interpreted in this light; dominant males could conceivably be signalling their quality in both cases, to same-sex competitors and/or their social mate, both of which could act in concert with improved intruder detection to further promote success in intra-sexual competition. While it has been suggested that a signalling role for sentineling 69

Chapter 2 might also ultimately lead to within-group competition over sentinel contributions

(Zahavi and Zahavi, 1997; but see Wright, 1999), we found no evidence to suggest that this was the case, consistent with the arguments of Wright (1999).

Whether sentineling does indeed serve as an honest signal of quality in this or other species remains to be investigated.

Previous research has highlighted the role that sentinel behaviour may generally play in mitigating predation risk in social groups, conceivably both for the actor and their fellow group members (e.g. Wright et al. 2001a; Ridley et al.

2010; Santema and Clutton-Brock, 2013), leading to debate over the relative roles that direct and kin-selected indirect benefits have played in the evolution of this potentially cooperative behaviour. Together, our findings support the hypothesis that sentineling may also play a role in intra-sexual competition, potentially facilitating the effective defence of both paternity and dominance status against extra-group challengers. Indeed, observations that males in a number of other species also invest differentially in sentinel behaviour (e.g.

Florida scrub jays: Hailman et al. 1994; meerkats: Clutton-Brock et al. 2002; vervet monkeys: Baldellou and Henzi 1992; Arabian babbler: Wright et al.

2001c), highlights the possibility of a more widespread role for sentinel behaviour in male-male competition. While our investigations have focused on males, females too may stand to benefit directly from investment in sentineling if it facilitates the detection and repulsion of same-sex competitors. While dominant females may rarely lose parentage to transient same-sex intruders (as dominant males frequently do), competition among females for the dominant position per se is frequently intense in cooperatively breeding species (Clutton-

Brock et al., 2006; Young & Bennett, 2013). While sentinel behaviour is 70

Chapter 2 frequently assumed to reflect an example of cooperation (specifically, a behaviour that provides a benefit to another individual (recipient), and that is selected for because of its beneficial effect on the recipient; West et al. 2007), the criterion that selection for sentineling arises because of its beneficial effect on recipients has to our knowledge yet to be conclusively demonstrated

(Clutton-Brock, 1999; Ridley et al., 2013; Santema & Clutton-Brock, 2013).

Indeed, our findings only add to the challenge of demonstrating this, as while evidence that sentinels are exposed to greater predation risk than non-sentinels might suggest that sentineling entails a direct fitness cost (Ridley et al., 2013), the possibility that sentineling also yields direct fitness benefits through mechanisms unrelated to predation (such as benefits in intra-sexual competition) leaves it more difficult to reach this conclusion. To the extent that sentinelling does indeed reflect cooperative behaviour, our findings lend new strength to the view that direct fitness benefits can shape the expression of cooperative behaviour, and highlight the wider potential for sexual selection to act in concert with more-commonly invoked mechanisms in shaping patterns of cooperation.

71

Chapter 2

72

Chapter 3

CHAPTER 3

SEX DIFFERENCES IN PROVISIONING BEHAVIOUR BUT NOT

RESOURCE ALLOCATION IN A SUBTROPICAL COOPERATIVELY

BREEDING BIRD

Lindsay A. Walker, Nick J. Royle, Andrew J. Young

73

Chapter 3

74

Chapter 3

3.1 ABSTRACT

In temperate bi-parental bird species, it is not uncommon for sex-specific patterns of provisioning behaviour to occur. These patterns, where males and females differ in the types and division of care, arise largely due to sexual conflict over the provision of care. Whether these patterns are also wide-spread in tropical species, whose slower life-histories may impact on how conflict is resolved, remains unknown. For example, the smaller clutch sizes and shorter nestling periods typical of tropical species may have important implications for the evolution of sex-specific patterns of care via impacts on the resolution of sexual conflict. Here, we test whether the dominant breeding pair (typically the parents) in a subtropical cooperatively breeding bird (Plocepasser mahali) show evidence of sex-specific patterns of provisioning behaviour. First, we reveal that dominant females provision at markedly higher rates than their social mate, yet both spend a similar length of time in the nest during a provisioning event.

Second, we show evidence that dominant females and males employ similar resource allocation tactics within the nest, preferentially feeding the nestling that begs first and is closest to the nest entrance. This latter finding, where females and males use similar food allocation rules, is unusual given the marked sex difference in provisioning rate. We argue that this finding could reflect novel outcomes of sexual conflict given the life-histories of tropical birds. Combined, our results highlight that life-history differences associated with living in the tropics may have important implications for sex-specific patterns of care, potentially via the influence on how sexual conflicts are resolved.

75

Chapter 3

76

Chapter 3

3.2 INTRODUCTION

In social systems where both parents provide care, conflict is expected to arise over the relative parental contributions to care as these carers are typically unrelated and parental effort is costly (Trivers, 1972; Clutton-Brock, 1991;

Parker et al., 2002; Lessells, 2012). With bi-parental care, the two parents often provide different levels and types of care (e.g. Møller and Cuervo, 2000;

Trumbo, 2006; Mand et al., 2013; Sonerud et al., 2014). These sex-specific care patterns can be generated by sexual conflict (see below), and can also lead to the intensification of sexual conflict as distinct sex roles can be associated with alternative reproductive tactics in the sex less constrained by care patterns (Andersson, 1994; Lessells, 2002; Royle et al., 2012). Sexual conflict over parental care has been explored extensively in bi-parental species both theoretically and empirically (reviewed in Harrison et al., 2009; Lessells,

2012). However, the empirical work to date has focussed principally on temperate zone species. Few studies to date have considered the effect of sexual conflict on parental care patterns in tropical species, whose slower life- histories (Martin, 2015) could impact the resolution of sexual conflict. For example, the typically smaller clutches of tropical species, a life-history trait that is thought to reflect a constraint imposed by environmental conditions such as nest predation risk (Martin, 2015), may have important implications for how sexual conflicts are resolved, and consequently the evolution of sex-specific care patterns. Exploring the implications of differing life history traits for patterns of parental investment would further our understanding of how sexual conflict can modulate the expression of sex-specific care patterns.

77

Chapter 3

Sexual conflict over the division and amount of care is predicted in species with bi-parental care (except with obligate monogamy, Trivers, 1972). First, sexual conflict over the division of offspring care will arise as, although parental fitness pay-offs from a given breeding attempt are positively correlated, each carer would benefit from the other performing a larger share of the work (as this would reduce individual costs yet maintain fitness benefits, Trivers, 1972;

Chase, 1980; Houston and Davies, 1985). Second, sexual conflict over the total amount of care provided is expected as males and females generally do not value current and future reproduction in the same way, due in part to extra-pair mating (Westneat et al., 1990). While the connection between mating patterns and the evolution of parental care patterns remains poorly understood (Alonzo,

2010), different mating systems can alter the balance of costs and benefits of performing care behaviours. Such costs may include ‘opportunity costs’

(Lessells, 2012), as well as physiological and non-physiological costs (reviewed in Alonso-alvarez and Velando, 2012). The variation in this balance of costs and benefits across species is suggested to explain the evolution of different parental care systems (Clutton-Brock, 1991). Given this, sex differences in the costs and/or benefits of investing in care may affect this balance, thereby determining the amount and division of care (Winkler, 1987; Kokko and

Jennions, 2008; Barta et al., 2014). For example, extra-pair mating by females may both decrease the benefits and increase the costs of care for males relative to females (as males may face an opportunity cost of caring if they could simultaneously be seeking matings elsewhere). Accordingly, extensive research on sex differences in provisioning behaviour in bi-parental species has typically documented that males contribute either less or the same as females

78

Chapter 3

(e.g. Whittingham et al., 2003; Hinde, 2006; Tanner et al., 2008; Wilkin et al.,

2009; Garcia-Navas and Sanz, 2012; Iserbyt et al., 2014).

Sex differences in parental contributions to offspring provisioning are often predicted to be accompanied by sex differences in the rules that parents use to allocate food to the offspring within a given breeding attempt, for two main reasons. First, the time and/or cognitive costs associated with accurately assessing nestling need or transferring food to smaller nestlings (so as to maximise offspring survival) may be greater for the sex that attends the nest less frequently as it has fewer interactions on which to base such decisions

(Stamps et al., 1985; Krebs et al., 1999; Lessells, 2002). Second, the sex that accrues the greater net benefit from providing a unit of care (and might therefore provision at higher rates, subsequently in contact with the nestlings more frequently) is expected to display greater sensitivity to offspring need

(Gottlander, 1987; Lessells, 2002). Of course, there are factors other than sex differences in provisioning rate that may also underpin sex differences in food allocation rules. For example, a particular sex of feeder may differentially benefit from preferentially feeding the philopatric rather than dispersing sex (e.g.

Brotherton et al., 2001). Additionally, sexual conflict over post-fledging parental investment may select for one sex to preferentially feed a particular offspring within the nest (e.g. it has been hypothesised that female blue tits, Cyanistes caeruleus, feed the smaller nestling to a greater extent than males so as to increase paternal care provision during the fledging period, Dickens and

Hartley, 2007). Evidence of sex differences in food allocation rules is widespread in bi-parental species (e.g. Leonard and Horn, 1996; Kilner, 2002;

Dickens and Hartley, 2007), with studies generally reporting that fathers (who 79

Chapter 3 often, but not always, provision at lower rates than mothers) tend to preferentially feed larger offspring, while mothers tend to favour smaller offspring (Lessells, 2002).

Studies of sex differences in both provisioning rate and food allocation tactics have been conducted principally in bi-parental bird species from the Northern temperate zone. Tropical species, however, have been largely ignored. Given that tropical bird species often show slower life-histories than temperate zone birds (Martin, 2015), latitudinal differences in the nature and outcome of sexual conflict may be expected. Recent mathematical models suggest that sex differences in life-history traits could indeed explain the origin of, and drive the transition between, different sex-specific patterns of care (e.g. maternal to bi- parental, Klug et al., 2013a, 2013b). First, the constraints on clutch size may differ between temperate and tropical species, which may in turn influence patterns of care. For bi-parental species, for example, male care and clutch size are expected to co-evolve: while the optimum clutch size for a female will depend upon the expected levels of care from her partner (as this will affect the costs and benefits of producing eggs/nestlings), the net benefit of care provision by males will depend on the size of the clutch (as clutch size will affect the benefit of providing care, Smith and Härdling, 2000). The outcome of such coevolutionary dynamics may therefore differ in tropical species, in which selection appears to have favoured the evolution of smaller clutches (Jetz et al.,

2008) due perhaps to latitudinal gradients in resource constraints and/or patterns of predation (Stutchbury and Morton, 2001; Martin, 2015). Such constraints may limit the potential for the coevolution of clutch size and male care in tropical species, as clutch size will at least in part reflect the selection by 80

Chapter 3 predation constraints rather than solely selection to maximise male investment.

As a result, more pronounced sex differences in provisioning rate might be expected in tropical species as males may be less inclined to invest in caring for small clutches, potentially obligating the female to provide higher levels of investment. Second, tropical bird species often exhibit shorter nestling periods than temperate zone species, which are thought to reflect an evolutionary response to a higher risk of nest predation in the tropics (Martin, 1995; Remes and Martin, 2002; Martin et al., 2011). Differential fitness benefits of swiftly fledging young in the tropics could conceivably reduce selection for parents to preferentially feed smaller nestlings, given the potential to instead advance the fledging date of larger nestlings. As such, while the parental sex that provisions at higher rates might normally be more likely to preferentially feed the smaller nestling (see above), the reduced benefits of doing so in tropical species might leave the sexes with similar food allocation rules (e.g. a preference to feed the nestling with the relatively greater offspring value, which may be the largest,

Klug et al., 2012) even when they differ in provisioning rate.

Here we test whether the dominant breeding pair (typically the parents) in a sub-tropical cooperatively breeding bird (the white-browed sparrow weaver;

Plocepasser mahali) show sex differences in provisioning rates and food allocation rules. White-browed sparrow weavers live in social groups of two to twelve birds, in which the dominant breeding pair and subordinates of both sexes all contribute to provisioning nestlings (Collias and Collias, 1978; Lewis,

1982; Harrison et al., 2013a). The dominant breeding pair completely monopolise within-group paternity and maternity (Harrison et al. 2013a). While there is no evidence of egg-dumping by extra-group females, the dominant 81

Chapter 3 male does lose 12-18% paternity to extra-group males (principally dominant males in other groups, Harrison et al., 2013a, 2013b). Such extra-group mating may therefore leave dominant males (relative to dominant females) experiencing both a lower benefit per unit of investment in parental care (given his reduced probably of parentage) and a higher cost (given the likely opportunity cost of being unable to secure extra-group matings while provisioning offspring). Typical of the life-histories of tropical species, white- browed sparrow weavers lay small clutches, typically comprising just two eggs

(Harrison et al. 2013a).

Specifically, we focus on the following two aims. First, we investigate whether dominant males and females differ in their nestling provisioning rates. As extra- pair mating may both decrease the benefits and increase the costs of care for males relative to females, we predict that dominant males will provision offspring at lower rates than females. This pattern may be exacerbated if the smaller clutch sizes of tropical species further reduce male commitment to parenting (see above). Second, we use within-nest cameras to examine the food allocation rules of dominant white-browed sparrow weavers. If the sexes provision offspring at different rates, we might also expect sex-specific food allocation rules (see above for logic), with the sex with the lower provisioning rate (predicted to be males) being predicted to: (i) spend less time than the female within the nest cup; and be more likely than the female to simply (ii) feed the first offspring to beg; (iii) feed the larger offspring; and (iv) feed the nestling at the front of the nest cup. However, while such sex differences in food allocation rules tend to be apparent in temperate zone species (Lessells, 2002), if life-history differences related to tropical environments are important (e.g. 82

Chapter 3 smaller clutch sizes, shorter nestling periods), sex differences in food allocation rules may be less apparent in white-browed sparrow weaver societies.

3.3 METHODS

3.3.1 STUDY SPECIES AND POPULATION

Data for this study was collected from a wild population of white-browed sparrow weavers at Tswalu Kalahari Reserve in the Northern Cape province of

South Africa (27°16′S, 22°25′E; see Harrison et al. (2013a) for a detailed site description). For this study, social group size ranged from two to five with a mean group size of 2.75 adult birds. Groups could be differentiated from each other as all group members roosted together in a single tree or cluster of trees in the centre of their territory, engaged in cooperative activities such as offspring care and sentineling, and foraged together during the day. All birds within the study population were fitted with a unique combination of a metal ring and three colour rings (SAFRING licence 1444), and were semi-habituated to observation with telescopes. Group composition was monitored at least twice per week, allowing specific courting and displacing behaviours associated with dominance status to be observed (as described in previous studies of this species, e.g.

Collias & Collias, 1978, Harrison et al. 2013b). This enabled specific individuals to be assigned as dominant or subordinate. All protocols were approved by the

University of Pretoria Ethics Committee and complied with the regulations stipulated in The Association for the Study of Animal Behaviour (ASAB)

Guidelines for Use of Animals in Research.

83

Chapter 3

3.3.2 FIELD METHODS

The contents of all nest structures within each territory were monitored on two of every three days. On the detection of at least one egg, the nest structure was monitored for consecutive days until no new eggs were detected. White-browed sparrow weavers lay one egg per day and between one and four eggs per clutch, with two eggs typical for our study population (Harrison et al. 2013a).

Monitoring of the active nest was then suspended (to avoid unnecessary disturbance) until day 14 of the incubation period, when daily nest checks were conducted until the eggs had hatched. Day 1 of the provisioning period was defined as the day on which the first egg hatched.

3.3.3 MONITORING PROVISIONING EVENTS

To identify each adult individual during the recording of nestling provisioning events (see below), birds were captured while their social group was in the incubation phase (as outlined in Cram et al., 2014) and were marked with a distinct dye-mark placed on the vent. Usually the dominant female was left unmarked to avoid any disturbance while incubating. Nestling provisioning visits made by all group members were videotaped at 16 breeding attempts (across

14 social groups) across two mornings between days 6 to 9 of the provisioning period (median: day 8, Inter-Quartile Range [IQR]: days 7-8) for a combined total of 351.7 ± 13.6 minutes (mean ± S.E.). Eight social groups were monitored in January 2013 and eight social groups monitored in January 2014 (two social groups were visited in both years). To record provisioning events, a Panasonic

SDR-S50 Camcorder attached to a tripod (approximately 0.5 meters in height) was placed underneath the nest entrance. The tripod was placed under the nest

84

Chapter 3 entrance two days before recording commenced to allow habituation to the tripod. All recordings began between 06:17 and 07:17. Video recordings were examined subsequently using VLC Media Player version 2.2. Feeder identity was recorded for each provisioning event via the distinct dye mark.

3.3.4 MONITORING PROVISIONING WITHIN THE ENCLOSED NEST

To examine whether breeders display differences in food allocation patterns, small low-light video cameras (19BSHRX from RFConcepts; attached to a

MiniDVR2 digital video recorder) were placed within the enclosed nest. The camera was positioned above and behind the bowl of the enclosed nest by creating a small hole in the woven nest structure and securing the camera to the outer structure. It was ensured that the camera lens was flush with the nest’s inner surface so as not to disrupt activities within the nest. The camera was placed two days prior to the recording of provisioning events to allow habituation. Provisioning behaviour within the nest was captured on the same days as the external-nest cameras at the same groups (see above for protocol) to allow parents to be reliably identified. Data were collected from 16 breeding attempts (across 14 social groups) on either one (n = 7) or two (n = 9; consecutive) morning(s) between days 6 to 9 of provisioning period (median: day 8, IQR: days 7-8) for a combined total of 180.4 ± 12.9 minutes (mean ±

S.E.). Not all breeding attempts had two days of internal provisioning data (as with external-nest videos) due to either equipment prioritisation or failure. All recordings began between 06:16 and 07:14. Video recordings were then examined using VLC Media Player version 2.2.

85

Chapter 3

Before recording, nestlings were weighed to ascertain which nestling was heavier. The relative size of the nestlings was based on body mass measured minutes before the recording of provisioning commenced. To identify the nestlings on camera, each was assigned a Tippex mark (one dot or two dots).

The Tippex mark was assigned based on the larger individual, alternating between each brood. For example, for one brood, the larger nestling would receive one dot marking, for the next brood monitored, the larger nestling would receive a two dot marking, and so on. This ensured that Tippex mark and size of nestling were not confounded. During each provisioning event, the behaviour of both nestlings (begging score, begging order, physical position in nest cup) was recorded directly before each feed. White-browed sparrow weavers bring a single prey item on each provisioning visit. The first nestling to be offered a prey item was considered to have been allocated that item by the adult, regardless of whether the adult then went on to feed the item to the second nestling following refusal from the first. Begging scores were measured using the following scale, as detailed in Kolliker et al., (1998): 0 = calm; 1 = weak gaping; 2 = persistent gaping; 3 = gaping, neck fully stretched; 4 = gaping, neck fully stretched, wing flapping. The duration in the nest cup per provisioning event (the time between the adult entering and leaving the nest) and any brooding behaviour were also recorded for each feed.

3.3.5 STATISTICAL ANALYSIS

All statistical analyses were conducted with R v 3.2.0 (R Core Team, 2015), and all models were checked for normality of residuals and variance homogeneity.

Statistical modelling utilized a stepwise model simplification approach. All fixed

86

Chapter 3 terms (detailed below) were fitted together initially and then the non-significant terms with the least explanatory power were sequentially removed until a minimal adequate model was reached. This model retained only those predictors whose removal now yielded a significant reduction in the explanatory power of the model. All random terms (detailed below) were retained in the model throughout the analysis. The assessments of statistical significance reported for each of the fixed terms are those calculated from the change in explanatory power on removal of that term from the minimal model (if it was in the minimal model) or following its inclusion in and subsequent removal from the minimal model (if it had not been retained in the minimal model). See Table

A2.1 in Appendix 2 for full details of all explanatory variables investigated in each model.

3.3.5.1 DO PARENTS SHOW SEX DIFFERENCES IN PROVISIONING RATE AND VISIT

DURATION?

First, to investigate the effect of parental sex on the provisioning rates of parents, a generalized linear mixed effects model was performed with a

Gaussian error distribution, with breeding attempt identity and social group identity as random factors. Provisioning rate was quantified as the number of visits to the active nest in the total observation period for each individual. Fixed effects were sex of feeder (male or female) and group size (range: 2-5 individuals). Second, to investigate the effect of parental sex on duration in the nest cup per provisioning event, a generalized linear mixed effects model was performed with a poisson error distribution. The response variable was the duration in the nest cup per provisioning event that did not also include a brood event (as dominant females are the sole brooder, this could skew the results). 87

Chapter 3

Feeder identity and breeding attempt identity were fitted as random factors. Sex of the feeder was the sole fixed effect. Brood size was not included in either of these models as all breeding attempts had a brood size of two.

3.3.5.2 DO PARENTS SHOW SEX DIFFERENCES IN FOOD ALLOCATION RULES?

The food allocation outcome of one focal nestling (of the two in the nest) was determined for each provisioning event. The nestling on the left of the nest bowl was chosen as the focal individual throughout. To investigate the effect of parental sex on food allocation patterns, a generalized linear mixed effects model with binomial error distribution was performed (1 = focal nestling was offered food first; 0 = focal nestling was not offered food first). The following fixed effects were included: relative size of the focal nestling (larger, smaller), relative position of the focal nestling (front or back), begging order of the focal nestling (first to beg or second to beg), sex of parent. Two way interactions were fitted between sex of parent and each of the other fixed effects (so as to test for sex differences in each aspect of food allocation rules). As begging score of nestling (0-4) was significantly correlated with begging order (r = -0.25, t = -7.43, p < 0.001), with the nestling showing the stronger begging score tending to beg first, begging score was not included in the statistical analysis.

The random effects were feeder identity and breeding attempt identity.

88

Chapter 3

3.4 RESULTS

3.4.1 DO PARENTS SHOW SEX-SPECIFIC PATTERNS OF PROVISIONING BEHAVIOUR?

White-browed sparrow weavers displayed sex-specific patterns of provisioning behaviour, with dominant females provisioning offspring at approximately three

2 times the rate of dominant males (GLMM: χ 1 = 23.24, P < 0.001; n = 16 breeding attempts across 14 social groups; Figure 3.1a). Provisioning rates

2 were unaffected by variation in group size (χ 1 = 2.90, P = 0.088). Dominant males and females spent a similar duration in the nest cup per provisioning

2 event (χ 1 = 1.61, P = 0.21; n = 16 breeding attempts across 14 social groups;

Figure 3.1b).

3.4.2 DO PARENTS SHOW SEX-SPECIFIC FOOD ALLOCATION RULES?

Of the 16 broods monitored, nine successfully fledged both nestlings. In one brood, both nestlings expired before fledgling (likely to be a predation event). Of the six remaining broods, one nestling expired in the nest and the other nestling survived until fledging; three of the six expired nestlings were the relatively smaller nestling. The smaller nestling was 19.0 ± 3.1% (mean ± S.E; range: 2.6

- 40.5%) lighter than the larger nestling.

There was no evidence for sex-specific food allocation patterns among parents in white-browed sparrow weavers. While both sexes were more likely to feed

2 the nestling that begged first (GLMM: χ 1 = 95.94, P < 0.001) and was at the

2 front of the nest cup (χ 1 = 37.40, P < 0.001), there was no evidence that males did so to a greater degree than females for either variable (interactions with

89

Chapter 3

2 parental sex: χ 1 ≤ 2.04, P ≥ 0.15; n = 407 feeds during 16 breeding attempts by

14 social groups; Figure 3.2). Relative nestling size did not influence food

2 allocation; large and small nestlings were fed with equal probability (χ 1 = 1.04,

P = 0.31), with no difference between the sexes (nestling size x parental sex

2 interaction: χ 1 = 1.66, P = 0.20; Figure 3.2). There was also no evidence of

‘false feeding’; in all of the provisioning visits observed (280 by 14 dominant females, 127 by 14 dominant males, and 85 by 8 subordinate helpers), the visiting bird offered their food item to the nestlings and it was ultimately consumed by the nestling.

90

Chapter 3

Figure 3.1. (a) Provisioning rates (mean ± standard error) were significantly higher for dominant females than males. (b) Dominant females and males spent a similar duration in the nest cup per provisioning event (mean ± standard error). The data derive from 16 breeding attempts of 14 social groups, yielding a sample of 14 dominant males and15 dominant females.

91

Chapter 3

Figure 3.2. The proportion of feeds (mean ± standard error) that were allocated to the nestling that was: first to beg; the smallest; positioned at the front of the nest cup. While parents of both sexes favoured the offspring that was first to beg and the offspring that was closest to the front of the nest cup, there were no significant sex differences in the food allocation rules of parents. Dotted line indicates 50% of successful feeds. The data derive from 16 breeding attempts of 14 social groups, yielding a sample of 407 provisioning events.

92

Chapter 3

3.5 DISCUSSION

This study used natural observations of a wild population of white-browed sparrow weavers to explore sex-specific patterns of parental provisioning behaviour and food allocation tactics. First, it was revealed that dominant males

(typically the father) provisioned offspring at substantially lower rates than dominant females (always the mother). Dominant males and females spent similar periods of time within the nest cup. Second, we provide evidence that once inside the nest, dominant males and females used similar cues to allocate food to a nestling. The nestling that was first to beg or at the front of the nest cup was more likely to be fed by both parents. Nestling size did not influence the decision to allocate food, despite smaller nestlings being on average 19% lighter. While the marked observed sex difference in provisioning rate might normally be predicted to be accompanied by a sex difference in food allocation rules, we found no evidence that this was the case. As might be predicted for tropical species, given selection for accelerated fledging, the sexes appeared to employ similar provisioning rules. Combined, our findings are consistent with the view that life-history constraints associated with living in the tropics may have implications for the outcomes of sexual conflict, including the nature of any sex difference in patterns of parental care.

The observed sex difference in provisioning rate is likely to reflect sex differences in the costs and benefits of providing care, as dominant male white- browed sparrow weavers lose 12-18% of paternity to extra-group males, while dominant females secure all maternity (Harrison et al., 2013a). This may impact the balance of costs and benefits to males of provide care in two main ways.

93

Chapter 3

First, the probability of extra-pair paternity reduces the likely fitness benefits of the breeding attempt for the social mate. As greater potential fitness benefits are expected to increase parental investment, reduced paternity assurance can lead to males valuing and therefore investing less in parental care relative to females (Westneat and Sherman, 1993; Kokko and Jennions, 2008). Second, the availability of extra-pair paternity prospects generates higher ‘opportunity costs’ for males to contribute to care behaviours rather than investing in seeking and capitalising on such opportunities (Lessells, 2012). While individuals with alternative mating opportunities may be less likely to care due to the

‘opportunity costs’ of missing these opportunities (Lessells, 2012, e.g.

Whittingham, 1994; Qvarnstroöm, 1997; Sanderson, 2012), other factors such as the availability of further mating opportunities and variation in the ability of males to exploit these opportunities (e.g. due to variation in male quality) will influence the magnitude of the ‘opportunity cost’ to providing care (Kokko and

Jennions, 2008; Stiver and Alonzo, 2009). In white-browed sparrow weaver societies, extra-pair paternity is secured principally by other dominant males

(only 16% of extra-group paternities is secured by extra-group subordinates,

Harrison et al., 2013b). This highlights that an individual’s paternity (both within- and extra-group) depends on maintaining dominance status once acquired.

There might therefore be intense selection pressure for males to engage with behaviours that protect dominance and thereby protect future paternity. Recent experimental evidence highlighted that sentinel behaviour (where one individual assumes an elevated position and scans the surroundings while other group members forage nearby) may play a role in intra-sexual competition in white- browed sparrow weaver societies (see Chapter 2). Dominant males displayed more sentinel behaviour than dominant females, increased sentinel activity in

94

Chapter 3 response to a male simulated intruder, and mounted quicker intruder defences when the simulated threat was detected on sentinel duty (see Chapter 2). As such, it could be expected that dominant males face a trade-off between investing in provisioning behaviour or sentinel activity. Fitness benefits to males of investing in the latter may therefore also explain in part why males contribute markedly less to provisioning behaviour than females.

The sex-specific patterns in provisioning behaviour observed in our study may be due to life-history traits typical of tropical species. The smaller clutch sizes of tropical species relative to temperate species is thought to be a response to limited food availability or high predation risk (Stutchbury and Morton, 2001;

Martin, 2015). Such constraints may limit the amount of care females can sequester from their social mate via increased clutch sizes, a phenomenon shown for bi-parental temperate species with the coevolution of clutch size and male care (Smith and Härdling, 2000). As a result, this mechanism could have contributed to the marked sex-specific patterns of provisioning behaviour that we see here. This study was performed in a cooperatively breeding species, where both parents provide care and are assisted by other group members

(‘helpers’). As cooperatively breeding species tend to live in groups of kin, indirect fitness consequences of relatedness may compress the magnitude of sex differences in patterns of parental investment (Johnstone, 2011; Savage et al., 2013), despite the potential for similar sexual conflict as with bi-parental systems. For example, a recent mathematical model showed that sex differences in the costs of parental care can give rise to sex differences in parental responses to helpers, with the parent that provides more care benefiting to a greater extent from the presence of a helper through parental 95

Chapter 3 workload-lightening (Johnstone, 2011). That we find marked sex differences in provisioning rate here could reflect the differential importance of extra-group paternity and small clutch sizes typical of tropical living species (see above), and/or the relatively small number of helpers present during the breeding attempts under study. Either way, it is clear that marked sex differences in parental contributions certainly can occur in cooperatively breeding species.

Our study provided no evidence of sex-specific food allocation tactics among dominant white-browed sparrow weavers. Instead, both parents allocated food to the nestling that begged first and was positioned at the front of the nest cup.

As described in other species (e.g. European starling, Sturnus vulgaris:

Kacelnik et al., 1995; canary, Serinus canarius: Kilner, 1995; tree ,

Tachycineta bicolor: Whittingham et al., 2003), this suggests that nestlings could be more likely to be fed due to their own behaviour rather than by independent cues not under their control (e.g. size). For example, nestlings have been shown to compete for a preferred position in the nest, particularly in cavity-nesting species (e.g. Kacelnik et al., 1995). As white-browed sparrow weavers have enclosed nests with a tube like entrance, there is only one location from which an adult can feed. This predictability may increase competition among nestlings, with the winner of these interactions at the front of the nest cup or begging first. Parents may use cues that indicate nestling competitiveness and survivorship to preferentially allocate food so to maximise nestling growth of the more competitive nestling (and hence minimising adult mortality, Martin, 2015). This strategy of preferentially feeding the nestling that may be more likely to survive may provide larger fitness benefits per unit of parental investment, in particular when food is limited (Whittingham et al., 96

Chapter 3

2003). Experimental manipulations of nestling cues would be required to understand the causal effects of particular cues in white-browed sparrow weaver societies.

Even though the smaller nestling was 19% lighter, parents did not display any preferential food allocation based on nestling size. Assuming that this nestling size difference is large enough to use as a discriminative cue, this finding is at odds with the majority of studies to date, which report preferential feeding based on nestling size (preferential feeding of both larger and smaller has been reported, reviewed in Lessells, 2002). That neither parent preferentially fed the larger nestling may be due to the fact that begging first and being at the front of the nest cup could be more reliable indicators of nestling competitiveness in white-browed sparrow weavers (see above, although this will need experimental manipulations to clarify). That neither parent preferentially fed the smallest nestling in this study may be due to the negative down-stream effects of delayed nestling growth on post-fledging survival being differentially greater in tropical living species with shorter nestling periods. For example, tree

(temperate zone species) nestlings that suffered delayed growth during the nestling period were less likely to survive and recruit post-fledging, therefore providing relatively less fitness payoffs to parents than nestlings who had not experienced delayed growth (McCarty, 2001; Whittingham et al., 2003). For tropical living species, where nestling growth strategies have been selected for to maximise flight capabilities after fledging (thereby minimising adult mortality,

Martin, 2015), this later-life handicap associated with delayed growth during nestling periods may be more severe. As such, parents of tropical species may

97

Chapter 3 be less likely to preferentially feed the smallest nestling at the expense of the larger nestling so to maximise fitness payoffs from investment in care.

Alternatively, limited food availability could explain a lack of sex differences in food allocation rules in this study. Previous studies have shown that a decrease

(Bize et al., 2006) and an experimental increase (Boland et al., 1997) in food availability can reverse food allocation patterns, with smaller nestlings only favoured in good environmental conditions (Boland et al., 1997; Davis et al.,

1999; Royle et al., 2002; Bize et al., 2006). Indeed, the strategy of preferentially feeding the needier nestling can result in complete brood failure in poorer conditions (Royle et al., 2002). Therefore, parental food allocation strategies are expected to depend on the environmental context. This study was conducted during drought conditions for the area, and so the patterns observed here may only reflect food allocation patterns during limited food availability (as prey abundance likely decreased with lower rainfall). The limited food availability might explain why parents allocated food potentially based on cues controlled by the nestling, which could signal the more competitive nestling (position in nest cup, begging order), rather than independent cues such as nestling size

(although nestling size can indicate nestling competitiveness). However, the frequency of the relatively smaller chick expiring in a breeding attempt was similar between the two year period prior to this study and the two year period of this study (in broods of two nestlings, long-term data, unpublished). This suggests that the results of this study are likely to reflect general food allocation patterns. Alternatively, the harsh environmental conditions may have increased the energy required per provisioning event. This could impact the food allocation patterns of the individual that attends the nest more frequently, as the 98

Chapter 3 increased time in the nest cup that might be required to differentially assess nestling need could incur a significant cost under such conditions. This might explain why in this study, females and males spent a similar length of time in the nest cup during a provisioning event.

Taken together, our findings highlight the possibility that the evolution of sex- specific patterns of care behaviour may be shaped by life-history traits, with constraints associated with tropical living impacting parental investment patterns. Studies of temperate bird species tend to reveal both sex-specific patterns of provisioning rate and food allocation, whereas the results of our study highlight that this subtropical species shows no sex difference in food allocation rules despite a marked sex difference in provisioning rate. Different life-history traits may impact the resolution of sexual conflict between parents, influencing the occurrence and/or magnitude of sex-specific patterns of care behaviours. Further explorations of parental care patterns in tropical species are warranted to better understand the combined impact of sexual conflict and life- history strategies on patterns of parental investment.

99

Chapter 3

100

Chapter 4

CHAPTER 4

NO EVIDENCE OF A ROLE FOR SOCIAL PRESTIGE OR PAYMENT

OF RENT IN THE HELPING BEHAVIOUR OF A COOPERATIVELY

BREEDING BIRD

Lindsay A. Walker, Ian R. Cleasby, Andrew J. Young

101

Chapter 4

102

Chapter 4

4.1 ABSTRACT

In cooperatively breeding systems, it is not uncommon for individuals unrelated to the group to provide care for offspring that are not their own. Kin selection fails to provide an adaptive explanation for this observation, as there are no indirect benefits to gain through helping. Another mechanism that may explain this finding is that individuals use their investment to act as a payment or signal to other group members, thereby generating direct benefits for contributing to cooperative activities. Such direct benefits may include staying on an occupied territory (permitted through payment of ‘rent’ via helping; pay-to-stay hypothesis) or future breeding opportunities (by signalling individual quality via helping; social prestige hypothesis). Here, we investigate whether the provisioning behaviour of males in cooperatively breeding white-browed sparrow-weaver (Plocepasser mahali) societies reflects a role for pay-to-stay or social prestige in modulating individual contributions to cooperation. Under the pay-to-stay hypothesis, immigrant males (relative to natal males) and older natal subordinate males (relative to younger natal subordinate males) might be predicted to provision offspring at higher rates and to inflate their perceived contributions by synchronising their provisioning visits with those of the dominant male or female. We found no support for any of these predictions. On the contrary, natal subordinate males were more likely to provision offspring than immigrant subordinate males and synchronised their visits with those of the dominant male to a greater extent than immigrant subordinate males. While the social prestige hypothesis would make some of the same predictions as the pay-to-stay hypothesis regarding the behaviour of immigrant subordinate males

(which were not met), it would also predict that the dominant male should contribute more to provisioning and synchronise their visits to a greater extent

103

Chapter 4 with those of the dominant female when in the presence of an immigrant subordinate male (with whom they may compete for prestige). Neither of these predictions was met either. Combined, our findings suggest that selection for helping behaviour in white-browed sparrow weaver societies does not derive from a significant role in signalling, and are instead broadly consistent instead with a central role for kin selection.

104

Chapter 4

4.2 INTRODUCTION

In cooperatively breeding species, group members help to raise offspring that are not their own. A key challenge in evolutionary biology is to understand the mechanisms by which selection has favoured such seemingly altruistic behaviour, where individuals pay a fitness cost to boost the reproductive output of other group members. While evidence suggests that kin selection is an important force to explain animal cooperation (Komdeur, 1994; Russell &

Hatchwell, 2001; Nam, et al., 2010; McDonald & Wright, 2011; Wright, et al.,

2010), kin selection alone cannot explain contributions made by unrelated individuals to cooperative activities often observed in studies of wild populations

(e.g. Cockburn, 1998; Cornwallis, et al. 2009; Hatchwell, 2009; Wright et al.,

2010; Riehl, 2013). Indeed, kin selection may not have acted in isolation to modulate the expression of helping behaviour, and may instead have acted in concert with other selection pressures. For example, if helping behaviour acted as a payment or a signal, then direct benefits may be accrued by individuals contributing to cooperative activities. Two such hypotheses include the pay-to- stay hypothesis (Gaston, 1978) and the social prestige hypothesis (Clarke,

1989; Zahavi, 1995, 1977). In both cases helpers are envisaged to accrue direct fitness benefits arising from their perceived levels of helping (rather than from the impact of their help on the fitness of offspring and/or parents). Although a role for signalling is commonly-invoked to explain human cooperation (e.g. Van

Vugt & Iredale, 2013; Raihani & Smith, 2015) comparatively few studies of cooperative animal societies have tested these hypotheses.

The pay-to-stay hypothesis was originally proposed by Gaston (1978), stating

105

Chapter 4 that individuals (helpers) provide assistance as an act of ‘payment’. This

‘payment’ behaviour occurs in return for being tolerated on occupied territories and accessing communal resources (Mulder & Langmore, 1993; Kokko et al.

2002; Hamilton & Taborsky, 2005), and offsets the costs this incurs for breeders. Consequently, one key prediction of the pay-to-stay hypothesis is that individuals whose presence entails a greater cost to the resident dominants (the territory holders) will be under selection to work harder (i.e. offer high levels of

‘payment’; Kokko et al., 2002). For example, mature offspring could conceivably have to pay their parents in order to delay dispersal from their natal territory, and their need to do so might be predicted to increase as they age (e.g. as dispersal becomes more feasible and/or they become a greater threat to their parents). Indeed, a mathematical model highlights that when the subordinate’s dispersal prospects improve (yet are still below the threshold to disperse), the dominant individual actually benefits first from the dispersal event, rather than the subordinate, in a pay-to-stay framework (if payment of rent is the only mechanism maintaining cooperation) (Kokko et al., 2002). This may select for increased payment of rent by older subordinates than younger subordinates so to remain on the natal territory, as the dominants benefit less strongly from retaining subordinates than the subordinates from staying in a group (Kokko et al., 2002). Additionally, unrelated individuals might pose a greater cost to the resident dominants than related group members, as (i) dominants may accrue indirect fitness benefits from tolerating the presence of relatives, and (ii) unrelated birds may pose a greater threat to the dominant’s parentage and/or rank than related birds (Young, 2009; Harrison, et al., 2014). Accordingly, the helping effort required of subordinates under a pay-to-stay framework is predicted to decrease with increasing relatedness to the dominant (Kokko et al.,

106

Chapter 4

2002). This is due to the dominant individual valuing the survival and future reproduction of a related subordinate more than an unrelated one (Kokko et al.,

2002).

Central to the pay-to-stay hypothesis is the assumption that helpers will be evicted from their groups by the resident dominants if they fail to pay sufficient

‘rent’. A number of studies have sought to test this assumption experimentally.

In chestnut-crowed babblers (Pomatostomus ruficep), helpers that were temporarily removed did not receive any aggression when released, regardless of their relatedness to the dominant individuals (Nomano et al., 2015). In superb fairy-wrens (Malurus cyaneus), by contrast, male helpers that were released following temporary removal were subjected to aggression by the dominant male (Mulder and Langmore, 1993). Interestingly, the frequency of observed aggression differed according to the stage of breeding; helpers received the highest level of aggression when removed during nestling/fledging periods, when helpers contribute to provisioning young (Mulder and Langmore, 1993).

However, such aggression could instead be attributable to the removals disrupting social hierarchies that need to be re-established (entailing aggression) on the helper’s return (Bergmüller and Taborsky, 2005; Wong and

Balshine, 2011). To date, compelling empirical support for the pay-to-stay hypothesis has been provided by just one species; the cooperatively breeding cichlid Neolamprologus pulcher (Bergmüller et al. 2005; Bergmüller and

Taborsky, 2005; Heg and Taborsky, 2010). For example, Zöttl et al. (2013) found that unrelated helpers provided more alloparental care than related helpers, and that their contributions increased following the simulation of egg cannibalism (which may increase the perceived cost of a helper). Moreover, 107

Chapter 4

N. pulcher subordinates that were subjected to a forced idle period received more aggression than those that were not idle, increased their help following the idle period, and had a higher likelihood of eviction in small groups (Fischer et al.

2014). This experiment minimised the potential complications arising from hierarchy disruption (see above) as the subordinates were always present within the territory, but were restricted from participating in helping by being placed in a clear plastic cylinder.

The social prestige hypothesis postulates that contributing to cooperative activities may act as an honest signal of an individual’s quality, as investment in cooperative care is likely to entail fitness costs (e.g. Russell, et al., 2003;

Canestrari, et al., 2007), and so only higher quality individuals may be capable of contributing at higher rates. By displaying cooperative efforts in the presence of others, and hence signalling their quality, individuals are hypothesised to gain

‘social prestige’ (Clarke, 1989; Zahavi, 1995, 1977), a currency that may ultimately translate in to an increased likelihood of being chosen as a mate or coalition partner (Zahavi, 1995; Zahavi & Zahavi, 1997; Wright, 1999). This hypothesis would predict, therefore, that individuals with the most to benefit from signalling their quality (e.g. to potential mates) should invest more in helping behaviour. For example, under the social prestige hypothesis alone, individuals that are unrelated to the dominant individual of the opposite sex might generally be predicted to display higher levels of helping behaviour than related birds, so as to increase their likelihood of securing matings. Likewise, dominant males might be predicted to respond to the presence of a same-sex reproductive competitor by elevating their own contributions to care, so as to also signal their own quality to resident reproductive females. Human research 108

Chapter 4 provides the strongest support for the social prestige hypothesis. For example, strong female mate choice for altruistic traits has been observed (Phillips, et al.,

2008; Farrelly, 2013). Additionally, males contribute more to public goods when observed by a female that they consider attractive (van Vugt and Iredale, 2013), and males respond competitively to charity donations by other males when an attractive female fundraiser is present (Raihani and Smith, 2015). While there is compelling evidence from human studies, comparatively few studies have sought to test predictions of the social prestige hypothesis in cooperative animal societies (Wright, 1997; Doutrelant & Covas, 2007; McDonald et al., 2008a,b;

Nomano et al., 2013). Observations suggesting that Arabian babblers

(Turdoides squamiceps) compete with other group members to provide help inspired and initially supported the hypothesis (Carlisle and Zahavi, 1986;

Zahavi and Zahavi, 1997). However, a subsequent detailed study found little evidence of competitive interactions in this species, and questioned the utility of the social prestige hypothesis (Wright, 1997; Wright et al. 2001). In the bell miner (Manorina melanophrys), female breeders that were naturally ‘widowed’ preferentially paired with the unrelated breeding-age male that had provided the most help in previous brood (Clarke, 1989).

Both the stay-to-pay and the social prestige hypotheses assume that the benefits of helping arise from an individual’s perceived level of helping, and so both mechanisms would also be expected to favour the evolution of tactics that inflate an individual’s perceived contributions to helping (Wright, 2007; Nomano et al., 2015). Although the influence of onlooker presence on human cooperative acts has been well documented (e.g. Lotem, et al., 2003; Mas &

Moretti, 2009; Engel, 2011; Phillips, 2015), evidence from cooperative 109

Chapter 4 vertebrate societies is less clear. There has been suggestive evidence from the sociable weaver (Philetairus socius), with helpers waiting for an increase in the number of birds in the audience before feeding nestlings (Doutrelant and

Covas, 2007). Raihani et al. (2010) observed that individual pied babblers

(Turdoides bicolor) wait for other group members to arrive near the nest before provisioning, but suggested that this reflects a strategy to minimise the conspicuousness of the nest to predators, rather than to advertise an individual’s provisioning efforts to fellow group members. Also, when investigating the effect of an audience on helper contributions in bell miners, neither the presence (McDonald et al., 2008b) nor removal (McDonald et al.,

2008a) of breeders in a social group influenced the provisioning behaviour of helpers. Further, unrelated helpers of the chestnut-crowed babbler did not advertise their provisioning efforts to either dominant individual by differentially synchronising the timing of their provisioning visits to those of dominants

(Nomano et al., 2015, 2013). As such, there is little compelling evidence to date that cooperation in non-human societies functions as either a ‘payment’ (under the pay-to-stay hypothesis) or a signal of individual quality (under the social prestige hypothesis). However, this pattern could in part reflect the comparatively small number of studies that have tested these hypotheses to date, highlighting the value of further attempts to test them in new model systems.

Here, we test key predictions of the pay-to-stay and social prestige hypotheses for the evolution of helping behaviour, in the cooperatively breeding white- browed sparrow weaver (Plocepasser mahali). White-browed sparrow weavers are an ideal study system to explore the importance of these ideas. First, within- 110

Chapter 4 group paternity is always secured by the dominant male (Harrison et al.,

2013a); dominant males do lose 12-18% of paternity, but do so exclusively to extra-group males (Harrison et al. 2013b). Therefore, all within-group subordinates are helpers rather than parents of the breeding attempt. Second, white-browed sparrow weavers live in social groups with variable kin structure.

While both sexes frequently delay dispersal from their natal group until adulthood (termed ‘natal’ individuals), emigration to subordinate positions in non-natal social groups is not uncommon (termed ‘immigrants’, Harrison et al.

2013a). As such, social groups comprise of a dominant pair and related and/or unrelated subordinates. Third, all group members assist with several cooperative activities including the provisioning of nestlings born to the dominant pair (Lewis, 1982; Harrison et al., 2013a). Fourth, once an individual attains dominance, they remain dominant for the rest of their relatively long lives, which should yield intense competition for these seldom available breeding positions. Finally, resident dominant males are rarely usurped by natal subordinate males (17.2% of monitored dominance turnovers; Harrison, et al.,

2014); instead immigrant subordinate males pose the greatest threat to the resident dominant male. As white-browed sparrow weavers live in highly kin- structured societies, it is likely that kin selection has acted to some extent to shape helping behaviour in this species. If kin selection alone influenced patterns of helping, we would predict that natal subordinates should contribute substantially more than immigrant subordinates (who should not contribute at all), and neither subordinate category would be predicted to advertise their contributions to the dominants, as the benefits to be gained (by relatives) from helping would derive solely from the positive effects of helping on the offspring and/or their parents.

111

Chapter 4

If the pay to stay and/or social prestige hypothesis played a role in selection for helping in this species (conceivably in tandem with a role for kin selection) we would make the following specific predictions regarding the provisioning behaviour of males (summarised in Table 4.1). We focus on male white-browed sparrow weavers because immigrant male subordinates are more common in our study population than immigrant female subordinates (as unusually for birds, male white-browed sparrow weavers are the more dispersive sex;

Harrison et al. 2014), thereby facilitating the testing of key predictions relating the relatedness. If benefits arising from the payment of rent have shaped helping behaviour (the pay-to-stay hypothesis; Gaston, 1978) we would predict that the classes of subordinate male whose presence on the territory poses a greater cost to the dominant male or female (specifically immigrant males relative to natal males, and older natal males relative to younger natal males; see above) should: (i) provision offspring at higher rates than other classes of subordinate male (though a concomitant role for kin selection could nevertheless leave natal males contributing more than immigrant males); and

(ii) synchronise their provisioning visits more closely with those of the dominant male or female than other classes of subordinate male (so as to increased their perceived contributions). If benefits arising from signalling to accrue social prestige have shaped helping behaviour (the social prestige hypothesis; Zahavi,

1977, 1995) we would predict that males with the most to gain from signalling their quality as potential mates to the dominant female (immigrant subordinates rather than natals, and the dominant male when in the presence of a subordinate immigrant male) should: (i) provision offspring at higher rates; and

(ii) synchronise their provisioning visits more closely with those of the dominant female (so as to increase their perceived contributions). However, we do note

112

Chapter 4 that in white-browed sparrow weavers the strength of these effects may be limited as although immigrant subordinate males may conceivably pose a greater threat to the resident dominant male than natal subordinate males, within-group immigrant subordinate males rarely usurp the resident dominant male in our study population. Therefore, there may be limited scope for within- group immigrant subordinate males to pose a threat large enough to warrant either a greater payment of rent than natal subordinate males (via the pay-to- stay hypothesis) or motivate the resident dominant male to inflate perceived quality to the dominant female (via the social prestige hypothesis). Further, the prediction of older subordinates providing higher ‘payments’ to dominant individuals by provisioning offspring at higher rates/increased synchronicity than younger subordinates may not hold if dominants benefit more strongly from retaining subordinates than the subordinates benefit from staying in a group. As the magnitude of benefits associated with group living has not yet been investigated with white-browed sparrow weavers, it is worth noting that the strength of these effects may be limited.

113

Table 4.1 Predictions arising from pay-to-stay and social prestige hypotheses regarding different elements of provisioning in white- browed sparrow weaver societies.

Response If pay-to-stay true, then expect: If social prestige true, then expect: variable Effect Reason Effect Reason Nestling Immigrant SM > Immigrant SM have an associated Immigrant SM > Immigrant SM have more to gain from Provisioning Natal SM* higher cost to dominants of Natal SM* advertising quality to resident DF (as Rate remaining on territory, therefore unrelated and hence a potential mate) should offer more ‘rent’ Older Natal SM Older Natal SM have an associated DM with Immigrant SM presence threatens > younger Natal higher cost to dominants of Immigrant SM > DM, and therefore motivates DM to SM remaining on territory, therefore DM without increase perceived quality to DF should offer more ‘rent’ Immigrant SM Synchronisation Immigrant SM Immigrant SM have an associated Immigrant SM Immigrant SM have more to gain from of Feeds synchronise higher cost to dominants of synchronise to inflating perceived quality to resident with DM or DF > remaining on territory, therefore DF > Natal SMs DF (as unrelated and hence a Natal SM greater benefits from being potential mate) witnessed as ‘rent payers’ Older Natal SMs Older Natal SMs have an DM with Immigrant SM presence threatens synchronise associated higher cost to dominants Immigrant SM DM, and therefore motivates DM to with DM or DF > of remaining on territory, therefore synchronise to inflate perceived quality to DF younger Natal greater benefits from being DF > DM SMs witnessed as ‘rent payers’ without Immigrant SM ‘Effect’ summarises the key predictions about nestling provisioning rates and degrees of synchronisation of feeding visits, relating to subordinate males (SM) that differ in their relatedness to the dominants (Natal or Immigrant) or age (Older, Younger), and the impact of the presence of Immigrant SMs on the provisioning behaviour of the dominant male (DM). For the pay-to-stay hypothesis, key groups are predicted to synchronise their visits with those of either the DM or dominant female (DF), as either could control group membership. For

114

the social prestige hypothesis, key groups are predicted to synchronise their visits with those of the DF only, as she is their primary potential mate. *If pay-to-stay and/or social prestige have shaped helping behaviour alongside a role for kin selection, this prediction (immigrant male subordinates should contribute more than natal male subordinates) may not hold.

Chapter 4

4.3 METHODS

4.3.1 STUDY SPECIES AND POPULATION

This study utilised data collected in the breeding seasons of 2010-14 as part of a long-term research project monitoring white-browed sparrow weavers at

Tswalu Kalahari Reserve, South Africa (27°16S, 22°25E). As white-browed sparrow weavers are rain-dependent breeders it is the timing of major rains that dictates breeding, which may occur at any time between October to April (the

Southern summer). Each bird in our study population of ~40 groups is fitted with a unique combination of metal ring and three colour bands (SAFRING licence

1444) to allow individual identification. Dominance status within a group was assigned by monitoring key dominance-related behaviours (Harrison et al.

2013a, York et al. 2014) during at least weekly visits. Dominance assignment via this method reliably indicates the breeding pair within the social group

(Harrison et al., 2013a,b). Every breeding attempt within each study group was monitored (see below for breeding status protocols), yielding a knowledge of (i) hatch date (and therefore age data), and (ii) natal group identity, for almost every member of our study population. Dispersal events were monitored during at least weekly group visits, allowing the assignment of natal versus immigrant status for all subordinates. In this study, provisioning behaviour was monitored for 36 natal male subordinates and 19 immigrant male subordinates over the course of 68 breeding attempts by 34 groups. All protocols were approved by the University of Pretoria Ethics Committee and complied with The Association for the Study of Animal Behaviour (ASAB) guidelines for the use of animals in research.

116

Chapter 4

4.3.2 MONITORING PROVISIONING BEHAVIOUR

Breeding status was determined by monitoring the contents of all woven nest structures within the territory of each social group on two of every three days throughout the potential breeding season (October – April of each year). On the detection of at least one egg, that particular nest structure was monitored for consecutive days until clutch completion. As white-browed sparrow weavers lay one egg per day, the lay date of the first egg could be inferred from the clutch completion date and clutch size (if not already known). Daily nest checks were then suspended until day 14 of the incubation phase (as the incubation phase lasts 14-19 days; Harrison et al. 2013a), when the nest was monitored daily until all nestlings had hatched, yielding brood size and hatch date information.

The date on which the first nestling hatched was defined as day 1 of the provisioning period for the whole breeding attempt. To monitor individual provisioning rates, members of each social group were made distinguishable from each other by capturing them from their roost chambers at night and placing a individually unique dye-mark on their vent (as outlined in Cram et al.

2015) Usually, the dominant female was left unmarked to minimise disturbance during incubation. Nestling provisioning events were monitored via video recordings; a Panasonic SDR-S50 camcorder was attached to a small tripod

(approximately 0.5 meters in height) and placed under the active nest entrance.

To allow habituation to the recording equipment, the tripod was placed under the nest entrance two days prior to recording. The camera was placed on the tripod at the start of each morning recording period, with the start time standardised relative to sunrise by adjusting for the timing of sunrise (recordings began between 06:15 and 08:28). These morning provisioning recordings had a median (Inter-Quartile Range [IQR]) of 189 (range: 62 - 249) minutes, and were

117

Chapter 4 taken on days 6 – 14 of the provisioning period (median: day 11, IQR: days 10-

12). A median (IQR) of 2 (range: 1-4) mornings of video recordings were made for each breeding attempt. Video recordings were subsequently examined using

VLC Media Player version 2.2, with feeder identity determined utilising the visible vent dye-marks.

4.3.3 STATISTICAL ANALYSIS

4.3.3(A) PATTERNS OF PROVISIONING CONTRIBUTIONS

To test our predictions regarding provisioning rate we determined the total count of provisioning events by each individual during the relevant observation period.

First, we tested the predictions regarding subordinate male provisioning rates.

As the data were zero-inflated and had random effects which proved difficult to analyse using zero-inflated analysis methods, a two-step analysis approach was employed. Initially, a generalised linear mixed effects model with a binomial error distribution was conducted, with the response variable set as whether the bird was observed to provision (1) or not (0), and controlling for the total observation duration as a fixed effect predictor. Subsequently, using only those individuals that were observed to provision (i.e. all the 1s), a generalised linear mixed effects model with a normal error distribution was conducted to investigate the drivers of variation in provisioning rates (total provisioning counts divided by the total observation time per individual). To test whether natal and immigrant male subordinates differed in provisioning rate, the following fixed effects were fitted in the both the binomial and Gaussian models: male category

(immigrant or natal), brood size, nestling age category (see below) and social group size (defined as all individuals who were seen consistently foraging

118

Chapter 4 together during group composition monitoring, see above). To test if older natal subordinate males provision at higher rates, the analyses were restricted to data from subordinate natal males, and the following fixed effects were fitted in the both the binomial and Gaussian models: male category (older [> 1 year old] or younger [< 1 year old]), brood size, nestling age category (see below) and social group size. For all models, the random effects of breeding attempt identity and individual identity were retained throughout.

Second, we tested the predictions regarding dominant male provisioning rates.

To test whether dominant males showed higher provisioning rates in the presence of immigrant subordinate males, a generalised linear mixed effects model with a poisson error distribution was conducted. The response variable was the total count of provisioning behaviour over the relevant observation period; it was possible to model the provisioning rates of dominant males using a single model with Poission error structure as the data for dominant males were less zero-inflated. The following fixed terms were fitted: male category

(immigrant present, immigrant absent), brood size, nestling age category (see below) and social group size. We fitted total observation time as an offset term to control for the variable amounts of times for which breeding attempts were observed. Breeding attempt identity and individual identity were both fitted as random effects. For all models, nestling age category was assigned based on the first day of the provisioning period on which provisioning video recordings were conducted. This allowed the analysis to control for variation in the age of chicks during provisioning while still enabling the data to be ‘pooled’ per individual over multiple recordings per breeding attempt, do as to minimise the extent of zero inflation in the data. Recordings that began on provisioning days 119

Chapter 4

6-8 were assigned the nestling age category ‘young’ (n = 31 breeding attempts), with recordings occurring for a mode of 2 days (range 1-2 days). Recordings that began on provisioning day 9-10 were assigned the nestling age category

‘old’ (n = 37 breeding attempts), with recordings occurring for a mode of 3 days

(range 1-4 days).

All models outlined above were performed in R v 3.2.0 (R Development Core

Team, 2015) and statistical modelling utilized a stepwise model simplification approach. All fixed terms (detailed above) were fitted together initially and then the non-significant terms with the least explanatory power were sequentially removed until a minimal adequate model was reached. This minimal adequate model retained only those predictors whose removal now yielded a significant reduction in the explanatory power of the model. The assessments of statistical significance reported for each of the fixed terms are those calculated from the change in explanatory power on removal of that term from the minimal model (if it was in the minimal model) or following its inclusion in and subsequent removal from the minimal model (if it had not been retained in the minimal model). See

Table A3.1 in Appendix 3 for full details of all explanatory variables investigated in each model.

4.3.3(B) SYNCHRONISATION OF FEEDS WITH TARGET AUDIENCE

Individuals may increase their perceived ‘payment’ (under pay-to-stay hypothesis) or quality (under social prestige hypothesis), by timing their provisioning events to coincide with the provisioning events of their target audience. As outlined previously, under the pay-to-stay hypothesis, we define

120

Chapter 4 the target audience as the dominant male or the dominant female (we test both), and under the social prestige hypothesis, we define the target audience for provisioning males as the dominant female. To test our predictions regarding synchrony, we analysed whether the predicted factors affected the degree to which the provisioning events of the focal males were synchronous with those of their target audience. Following Nomano et al., (2013, 2015), a provisioning event by the focal male was classed as ‘synchronous’ with a provisioning event of the target audience (e.g. the dominant female for the social prestige hypothesis) if it occurred in the same ‘cluster’ of provisioning events as a provisioning event by the target audience. Clusters of provisioning events were defined as a series of successive events whose nest entry times were within

48.0 seconds of each other (the 1st quartile of the distribution of intervals between all successive provisioning events). We characterised synchrony in this way (on the basis of membership of the same provisioning cluster), as white-browed sparrow weavers use of an enclosed nest with a narrow entrance, which leaves it impossible for any given bird to enter the nest to feed the nestlings until the previous visitor has left. As such, even if multiple birds arrived in the nest tree at exactly the same time, their nest entry times (the information available from the provisioning videos) would necessarily be spaced out in time as a cluster of provisioning events with short inter-feed intervals (see also

Nomano et al., (2013) for similar logic).

Having categorised all individual provisioning events in this way, we determined a measure of the degree of synchrony between the focal male’s provisioning and that of the target individual via the calculations specified in Nomano et al.,

121

Chapter 4

(2013). Briefly, the measure of synchrony between a dominant individual D (the

target audience) and a focal male M was defined as YD,m/VDVm, where YD,m is the rate of synchronous nest visits by the focal male m and the dominant

individual D, VD is the visit rate by the dominant individual and Vm is the visit rate by the focal male (as outlined in Nomano et al., 2013, 2015). The ratio

YD,m/VD is considered as the dominant-perceived contribution by the focal

male. If males increase this perceived contribution (YD,m/VD) without increasing

their actual contribution (Vm), then YD,m/VDVm will be higher. To test the social

prestige hypothesis, YD,m/VDVm was measured and compared across (i) natal subordinate males and immigrant subordinate males, with the dominant female considered to be the target audience for advertisement contributions; and (ii) dominant males with, and those without, an immigrant subordinate male present in their social group, again with the dominant female as the target audience. To test the pay-to-stay hypothesis, YD,m/VDVm was measured and compared across (i) natal subordinate males and immigrant subordinate males, first with the dominant male as the target audience and then with the dominant female as the target audience (as either could control group membership), and (ii) old (> 1 year) natal subordinate males and young (< 1 year) natal subordinate males, again first with the dominant male as the target audience and then with the dominant female. Full predictions are outlined in Table 4.1.

One issue with dyadic data is that individuals may have specific levels of synchronicity regardless of who they interact with. Consequently, high

122

Chapter 4 synchronicity between individuals may be a by-product of the high tendency for synchronicity by one member of the dyad only. To account for this potential source of bias, we used a Bayesian generalized linear mixed-effects model in which synchrony between combinations of individuals was fitted to a Poisson distribution and problems of independence were accounted for by fitting

‘sociality’ random effects. The factors affecting the degree of synchrony of provisioning events was then modelled using Bayesian generalized linear mixed-effects models specified as (similar to Nomano et al., 2013, 2015)

푦푖푗 ~ 푃표푖푠푠표푛 (휆푖푗)

푙표푔 (휆푖푗) = 훽0 + 푙표푔 (푡푖푗) + 푙표푔 (푣푖) + 푙표푔 (푣푗) + 훽1푠푒푎푠푖푗 + 훽2푔푠푖푧푒푖푗 +

훽3푏푠푖푧푒푖푗 + 훽4푑푦푎푑푐푎푡푖푗 + 퐺푟표푢푝푎 + δi + δj for i (or j) = 1, . . . k, where k is the number of individuals and i (or j) represents the kth individual; a = 1, . . . g, where g is the number of groups and d = 1, . . . z, where z is the number of unique dyads observed. yij is the number of synchronous visits per dyad i,j, (i ≠ j). tij was the duration in hours over which a breeding attempt was observed, and vi (or vj) is the feeding visit rate/ hr by individual i (or j). β0 represents the coefficient for the model intercept, β1 is a categorical fixed effect denoting the effects of breeding season, β2 is the slope for size of the social group, β3 is the slope for brood size. β4 is the coefficient that compares the visits rate across dyad categories. The offset terms, log(tij) + log(vi) + log(vj) were added so that the synchronous visit rate (Yij = yij/tij) is evaluated relative to vivj in the model. In addition, Groupa denotes a random intercept for social group identity, and δi and δj represent sociality random effects that describe the variation in connectedness amongst members of a dyad (see: Hoff, 2005; Nomano et al., 2013). We ran five separate models (see

123

Chapter 4 above for details) in which synchrony was defined in relation to the dominant individual. Bayesian models were fitted using WinBUGS 1.4 (Lunn et al. 2000) running three separate chains for 50,000 iterations, with a thinning interval of 10 and a burn-in of 20,000. Convergence was confirmed using the Gelman-Rubin statistic (Gelman et al. 2004) and model fit was assessed using posterior predictive checks (Gelman et al., 1996).

4.4 RESULTS

4.4.1PATTERNS OF PROVISIONING CONTRIBUTIONS

If either pay-to-stay or social prestige were the sole force shaping helper contributions to offspring provisioning (i.e. in the absence of a role for kin selection), immigrant subordinate males would be predicted to provision offspring at higher rates than natal subordinate males. However, our findings reveal that natal subordinate males are more likely to provision offspring than

2 immigrants (GLMM: χ 1 = 9.43, P < 0.01; n = 82 measurements across 47 breeding attempts; Figure 4.1a), and that, among the subset of males that do provision, natal and immigrant subordinate males did not differ in their

2 provisioning rates (χ 1 = 0.31, P = 0.58; n = 52 measurements across 37 breeding attempts; mean ± S.E.; natal: 2.69 ± 0.41, immigrant: 2.22 ± 1.18).

2 Subordinate males were significantly more likely to provision larger broods (χ 1

= 5.43, P = 0.020), but those subordinate males that did provision broods did

2 not provision larger broods at significantly higher rates (χ 1 = 2.21, P = 0.14).

Neither group size nor nestling age predicted the likelihood or rate of offspring

2 provisioning by subordinate males (χ 1 ≤ 1.06, P ≥ 0.30).

124

Chapter 4

4.4.1(I) TESTING THE SPECIFIC PREDICTIONS OF THE PAY-TO-STAY HYPOTHESIS

If pay-to-stay shaped helper contributions to offspring provisioning, then we would expect older natal subordinate males to display higher levels of provisioning than younger natal subordinate males. However, younger natal subordinate males are significantly more likely to provision than older natal

2 subordinate males (χ 1 = 4.18, P = 0.041; n = 52 measurements across 30 breeding attempts; Figure 4.1b), and among the subset of subordinate natal

2 males that provisioned age class did not impact provisioning rate (χ 1 = 0.14, P

= 0.73 n = 39 measurements across 28 breeding attempts; mean ± S.E.; older:

2.56 ± 0.49, younger: 2.95 ± 0.73). Subordinate natal males were significant

2 more likely to provision larger broods (χ 1 = 5.05, P = 0.025) and those that

2 provisioning broods provisioned larger broods at higher rates (χ 1 = 4.94, P =

0.026). While subordinate natal males were not more likely to provision older

2 nestlings than younger ones (χ 1 = 0.014, P = 0.91), those that did provision

2 provisioned older nestlings at higher rates (χ 1 = 5.69, P = 0.017). Group size did not affect either the likelihood or rate of offspring provisioning by

2 subordinate natal males (χ 1 ≤ 0.36, P ≥ 0.55).

4.4.1(II) TESTING THE SPECIFIC PREDICTIONS OF THE SOCIAL PRESTIGE HYPOTHESIS

If social prestige shaped helper contributions to offspring provisioning, then the presence of an immigrant subordinate male would be expected to motivate the dominant male to provision at higher rates. However, dominant males did not provision offspring at higher rates when a subordinate immigrant male was

2 present within their group (χ 1 = 0.18, P = 0.67; n = 78 measurements across 64 breeding attempts; Figure 4.2), and their provisioning rates were unaffected by

2 group size, nestling age or brood size (χ 1 < 1.88, P > 0.17). 125

Chapter 4

4.4.2 SYNCHRONISATION OF FEEDS WITH TARGET AUDIENCE

If either pay-to-stay or social prestige shaped helper contributions to offspring provisioning, immigrant subordinate males might be predicted to increase their perceived contributions by timing their provisioning events to coincide with the provisioning events of the dominant female, to a greater extent than natal subordinate males. Our analyses revealed no evidence that the degree of synchrony with the dominant female differed for immigrant and natal subordinate males (Table 4.2; Figure 4.3a). Neither brood size nor social group size affected the degree of synchrony observed in this model (the 95% CRIs for the relevant model coefficient spanned zero; Table 4.2).

4.4.2(I) TESTING THE SPECIFIC PREDICTIONS OF THE PAY-TO-STAY HYPOTHESIS

If pay-to-stay shaped helper contributions to offspring provisioning, then we might also expect immigrant subordinate males to increase their perceived contributions by timing their provisioning events to coincide with the provisioning events of the dominant male, again to a greater extent than natal subordinate males. However, our analyses revealed the reverse pattern; natal subordinate males actually synchronised their provisioning visits with those of the dominant male to a greater extent than immigrant subordinate males (Table 4.3; Figure

4.3b). Additionally, we might expect older natal subordinate males to synchronise their provisioning visits with either the dominant male or the dominant female (we tested both separately) to a greater extent than younger natal subordinate males. Our models revealed little evidence that older natal subordinate males synchronised more than younger natal subordinate males with either the dominant male (Table 4.4, Figure 4.3c) or dominant female

126

Chapter 4

(Table 4.5, Figure 4.3d). Brood size and social group size did not affect the degree of synchrony observed in any of these three models (the 95% CRIs for the relevant model coefficients spanned zero in all cases; Tables 4.3-4.5).

4.4.2(II) TESTING THE SPECIFIC PREDICTIONS OF THE SOCIAL PRESTIGE HYPOTHESIS

If social prestige shaped helper contributions to offspring provisioning, we might expect dominant males with an immigrant subordinate male in their group to increase their perceived levels of contributions by timing their provisioning events to coincide with those of the dominant female to a greater extent than dominant males without an immigrant subordinate male in their group. However, our findings revealed no support for this prediction (Table 4.6; Figure 4.3e).

Again, brood size and social group size did not affect the degree of synchrony observed in this model (the 95% CRIs for the relevant model coefficients spanned zero; Table 4.6).

127

Chapter 4

Figure 4.1. Investigating the potential for provisioning behaviour to act as a payment or signal individual quality. (a) Natal subordinate males were significantly more likely to perform provisioning behaviour than Immigrant subordinate males. Data derives from 82 measurements across 47 breeding attempts. (b) Young (under 1 year) natal subordinate males were significantly more likely to perform provisioning behaviour than Old (over 1 year) natal subordinate males. Data derives from 52 measurements across 30 breeding attempts. For both graphs, model effect sizes plotted with the S.E. represented

128

Chapter 4

Figure 4.2. Provisioning events for dominant males were not significantly predicted by presence (Imm. present) or absence (Imm. absent) of immigrant subordinate male within the social group. Means (± standard error) of raw data plotted. Data derives from 78 measurements across 64 breeding attempts.

129

130

Chapter 4

Figure 4.3. The extent to which males synchronise their provisioning visits with those of the dominant female or dominant male. The degree of synchrony to a target individual D was calculated as YD,m/VDVm (see methods), so here we

represent the degree of synchrony to the dominant female (F) as YF,m/VFVm

and the degree of synchrony to the dominant male (M) as YM,m/VMVm. (a)

Immigrant and natal male subordinates did not differ in the extent to which they synchronise their visits with those of the dominant female (Table 4.2). (b) Natal male subordinates synchronised their visits with those of the dominant male to a greater extent than those of immigrant male subordinates (Table 4.3). (c)

Young and old subordinate natal males did not differ in the extent to which they synchronised their visits to those of the dominant female (Table 4.4). (d) Young and old subordinate natal males did not differ in the extent to which they synchronised their visits to those of the dominant male (Table 4.5). (e)

Dominant males did not synchronise their visits to those of the dominant female to a greater extent when they had a subordinate immigrant male in their group

(Table 4.6). Plots show medians and interquartile ranges of observed data.

Circles represent outliers. Numbers indicate sample sizes of unique individuals for each male category.

131

Chapter 4

Table 4.2. Results of Bayesian GLMM investigating whether immigrant subordinate males and natal subordinate males differ in the extent to which their provisioning visits are synchronised with those of the dominant female. Coefficient estimates are presented along with lower and upper Bayesian 95% Credible Intervals (CRI). Random effects are presented as standard deviations N = 53 observations, 28 subordinate male individuals, 24 groups.

Parameter Coefficient Lower 95% CRI Upper 95% CRI Intercept -3.733 -5.379 -2.079

Season 2 0.0086 -0.888 1.148

Season 3 -0.0064 -1.039 1.023 Group Size 0.039 -0.124 0.199 Brood Size 0.022 -0.609 0.653 Dyad Category - 0.205 -0.592 1.011 Subordinate Natal Male σ group identity 0.654 0.078 1.214

σ sociality 0.203 0.0064 0.618

Table 4.3. Results of Bayesian GLMM investigating whether immigrant subordinate males and natal subordinate males differ in the extent to which their provisioning visits are synchronised with those of the dominant male. Coefficient estimates are presented along with lower and upper Bayesian 95% Credible Intervals (CRI). Random effects are presented as standard deviations N = 40 observations, 33 subordinate male individuals, 17 groups. Parameter Coefficient Lower 95% CRI Upper 95% CRI Intercept -4.168 -7.441 -1.043

Season 2 -0.0575 -3.127 2.409 Season 3 0.169 -2.473 3.068

Group Size -0.0523 -0.283 0.113

Brood Size -0.367 -1.246 0.491 Dyad Category - 1.955 0.820 3.344 Subordinate

Natal Male σ group identity 0.235 0.0103 0.979 σ sociality 0.145 0.0062 0.568

132

Chapter 4

Table 4.4. Results of Bayesian GLMM investigating whether older natal subordinate males and younger natal subordinate males differ in the extent to which their provisioning visits are synchronised with those of the dominant male. Coefficient estimates are presented along with lower and upper Bayesian 95% Credible Intervals (CRI). Random effects are presented as standard deviations N = 35 observations, 27 subordinate male individuals, 16 groups. Parameter Coefficient Lower 95% CRI Upper 95% CRI Intercept -2.253 -4.079 -0.125 Season 2 -0.166 -1.840 1.219 Season 3 0.334 -1.148 2.057

Group Size -0.0511 -0.273 0.113

Brood Size -0.432 -1.407 0.473 Dyad Category 0.121 -0.548 0.758 – Older subordinate natal male σ group identity 0.233 0.0095 1.010 σ sociality 0.161 0.0087 0.557 Table 4.5. Results of Bayesian GLMM investigating whether older natal subordinate males and younger natal subordinate males differ in the extent to which their provisioning visits are synchronised with those of the dominant female. Coefficient estimates are presented along with lower and upper Bayesian 95% Credible Intervals (CRI). Random effects are presented as standard deviations N = 40 observations, 29 subordinate male individuals, 17 groups. Parameter Coefficient Lower 95% CRI Upper 95% CRI Intercept -3.234 -5.791 -0.953 Season 2 0.0086 -1.597 2.411

Season 3 -0.0078 -1.792 2.190 Group Size 0.0195 -0.177 0.205 Brood Size -0.082 -0.857 0.721 Dyad Category -0.168 -0.714 0.372 – Older subordinate natal male σ group identity 0.692 0.0553 1.375

σ sociality 0.205 0.0129 0.618 133

Chapter 4

Table 4.6. Results of Bayesian GLMM investigating whether dominant males with or without an immigrant subordinate male in their social group differ in the extent to which their provisioning visits are synchronised with those of the dominant female. Coefficient estimates are presented along with lower and upper Bayesian 95% Credible Intervals (CRI). Random effects are presented as standard deviations. N = 69 observations, 45 dominant male individuals, 31 groups. Parameter Coefficient Lower 95% CRI Upper 95% CRI Intercept -3.317 -4.477 -1.628 Season 2 0.078 -1.129 1.374 Season 3 -0.279 -1.801 0.752 Group Size -0.088 -0.267 0.078 Brood Size 0.135 -0.351 0.604 Dyad Category -0.0497 -0.614 0.508 – Immigrant present σ group identity 0.417 0.042 0.805 σ sociality 0.252 0.015 0.544

134

Chapter 4

4.5 DISCUSSION

This study used natural observations to investigate whether patterns of helping behaviour in a wild cooperative bird have been shaped by direct fitness benefits arising from helping serving as a payment of rent (as envisaged by the pay-to- stay hypothesis; Gaston, 1978) or a signal of quality (as envisaged by the social prestige hypothesis; Clarke, 1989; Zahavi, 1977, 1995). We found no evidence to suggest that helping by male white-browed sparrow weavers acts either as a ‘payment’ or as a signal of individual quality. First, natal subordinate males were more likely to provision offspring than immigrant subordinate males.

This finding is not consistent with pay-to-stay and/or social prestige acting in isolation to maintain cooperation in this species (i.e. in the absence of any role for kin selection). It is, however, consistent with a role for kin selection, conceivably acting alongside a role for pay-to-stay and/or social prestige (e.g. the latter hypotheses might nevertheless explain why immigrants do sometimes help, and could still impact the contributions of natal males). Second, while under the pay-to-stay hypothesis one might predict (i) higher provisioning contributions by older subordinate natal males than younger ones, and (ii) that both immigrant subordinate males (relative to natal subordinate males) and older natal subordinate males (relative to younger ones) should synchronise their provisioning visits to a greater extent with those of either the dominant male or the dominant female, our analyses found no support for any of these predictions. Indeed, contrary to two of these predictions, we actually found that younger natal subordinate males were more likely to contribute to provisioning than older natal subordinate males, and that natal subordinate males synchronised their provisioning visits with those of the dominant male to a significantly greater extent than immigrant subordinate males. Third, while

135

Chapter 4 under the social prestige hypothesis one might predict that both immigrant subordinate males (relative to natal subordinate males) and dominant males with an immigrant subordinate male within the social group (relative to no immigrant male subordinate present) should synchronise their provisioning visits to a greater extent with those of the dominant female, our analyses found no support for either of these predictions. Combined, our analyses provide no support for a role for either pay-to-stay or social prestige in modulating male contributions to cooperative provisioning behaviour in white-browed sparrow weaver societies.

It is possible that the pay-to-stay and/or social prestige mechanisms are acting in white-browed sparrow weaver societies, but just in a manner that perhaps does not satisfy our predictions. First, the prediction that immigrant subordinate males should provision more than natal subordinate males was likely not met because of a role for kin selection in white-browed sparrow weaver societies. To overcome this, we also tested the prediction that immigrant subordinate males would maximise the opportunity for their efforts to be witnessed by coinciding provisioning visits with their target audience (thereby accounting for the lower levels of actual provisioning events by immigrant subordinate males). We found little evidence that immigrant subordinate males had higher levels of provisioning synchrony with either the dominant male or female. Second, the prediction that older natal subordinate males should provision more than younger natal subordinates may not have been met as older individuals could be (i) performing other behaviours that also provide ‘rent’ or signal individual quality (e.g. defending the territory), or (ii) having to trade-off prospecting against helping, and so are allowed to work less hard as prospecting facilitates 136

Chapter 4 dispersal. However, as above, we also tested whether individuals that would be expected to benefit most from signalling their ‘rent’/individual quality are also more likely to synchronise provisioning visits with that of the target audience

(thereby enhancing the likelihood of any contributions being witnessed). We found little evidence that older natal subordinate males coincided their visits with either the dominant male or female more than younger natal subordinate males. Third, the predictions regarding synchronisation might not be met if, for example, there is no benefit in attempting to synchronise with the visits of dominants because they can monitor the contributions of subordinates to cooperative activities accurately regardless of synchronisation. However, monitoring the behaviour of signalling individuals will have prohibitively high costs for the receiver, therefore selecting against a signalling mechanism

(Wright, 2007). Investigating the level of provisioning synchrony between the males and their target audience assumes that the costs were imposed on the signaller (the male) rather than the receiver (the dominants), averting this potential issue for the evolution of signalling (see also Doutrelant & Covas,

2007; Nomano et al., 2013). Finally, provisioning nestlings might not be the only way for subordinates to offer payment or signal individual quality. Although all individuals within a white-browed sparrow weaver social group contribute to provisioning of nestlings (Dale M Lewis, 1982), group members also contribute to several other cooperative activities (e.g. sentinel activity: Chapter 2; territorial defence: York and Young, unpublished data; weaving the nest and roost structures: personal observations). However, for white-browed sparrow weavers, provisioning behaviour is the main cooperative activity during a breeding attempt. It seems likely, therefore, that provisioning provides the

137

Chapter 4 greatest opportunity to advertise either payment or individual quality during these periods.

Our findings suggest that pay-to-stay and social prestige mechanisms have not played a key role in the evolution of nestling provisioning by helpers in white- browed sparrow weaver societies. The lack of support for these hypotheses in this species, coupled with the rarity of compelling evidence to date for either hypothesis across cooperative vertebrates, could be due in part to a number of factors acting in isolation or in concert. First, dominants may simply be unable to assess subordinate contributions with the accuracy required to enforce pay- to-stay or to identify high quality individuals, due either to constraints on their ability to monitor contributions (foraging group members are frequently out of visual contact with each other) or cognitive constraints that preclude the integration of such information over time for all subordinates (white-browed sparrow weaver groups may have up to 12 helpers). Second, it is conceivable that, even if dominants could accurately monitor the nest visits of others, that the potential for helpers to ‘cheat’ (by eating their delivered prey load themselves within the privacy of the enclosed nest) erodes the utility to dominants of making toleration and/or mate-choice decisions on the basis of subordinate’s visit rates. Indeed, that provisioning sparrow-weavers have never been observed to cheat in this way despite the opportunity to do so (see

Chapter 3) might itself suggest that there are not currently strong benefits to be accrued from higher ‘perceived’ levels of provisioning. Third, specifically with regard to pay to stay, selection may not favour the eviction of subordinates that fail to cooperate, if dominants suffer no net cost from tolerating additional subordinates (e.g. if dominants enjoy survival advantages when living in larger 138

Chapter 4 groups and/or suffer costs when attempting to evict subordinates; Bell, et al.,

2012). Fourth, specifically with regard to the social prestige hypothesis, while there is now widespread evidence to suggest that helping entails costs (e.g.

Heinsohn & Legge, 1999; Russell et al., 2003; Sanderson et al., 2015; Cram et al., 2015), it is conceivable that these costs are not sufficiently high for helping to serve as an honest signal of individual quality (Zahavi, 1977; Grafent, 1990).

Finally, it is conceivable that signalling quality via provisioning is evolutionarily unstable (see arguments in Wright, 1999). For example, if provisioning yielded social prestige, selection might favour helpers that disrupted the provisioning attempts of other helpers so as to advance their own relative social prestige

(see arguments in Zahavi & Zahavi, 1997). That we have never observed individuals interfering with the provisioning events of others suggests that such overt interference is rare in this species. Nevertheless, it remains possible that the potential for such interference erodes both the utility of attempting to signal quality in this way, and the utility to receivers of attempting to assess individual quality via provisioning (Wright 1999).

Our findings largely concord with those of other studies of cooperative provisioning in avian societies (e.g. McDonald, et al., 2008a,b; Nomano et al.,

2013, 2015) in suggesting that cooperative provisioning may rarely act as payment or as a signal. For bell miners, the finding that the absence or exclusion of dominant individuals from the social group did not influence subordinate provisioning contributions suggests that signalling to dominants may not play a central role in maintaining subordinate provisioning behaviour

(McDonald et al. 2008a, b). Akin to our findings here, there is no evidence to suggest that chestnut-crowned babbler helpers that are unrelated to the 139

Chapter 4 offspring that they’re rearing advertise their provisioning efforts to either the dominant male or female (as would be predicted if they accrued fitness benefits from their ‘perceived’ levels of cooperation; Nomano et al., 2013, 2015).

Furthermore, in acorn (Melanerpes formicivorus), the probability that a helper inherits the natal territory is unrelated to their prior feeding history, which has been interpreted as contrary to the prediction under the pay-to-stay hypothesis that more helpful subordinates would be allowed to remain in the natal territory for longer (Koenig and Walters, 2011). There is, however, some suggestive evidence of a role for pay-to-stay. For example, male helpers in superb fairy-wren societies were subjected to aggression by the dominant male following temporary removal, with higher levels of aggression observed during times when helpers could otherwise have provisioned young (Mulder and

Langmore, 1993). A study of sociable weavers has also highlighted the potential importance of perceived levels of cooperation, as subordinates were more likely to provision young in the presence of larger audiences, and held larger prey items for longer at the colony prior to delivering them to the chicks (Doutrelant and Covas, 2007). Additionally, in the pied kingfisher (Ceryle rudis) unrelated male helpers sometimes delayed provisioning the young until the breeding female was present (Reyer, 1986). However, it is difficult to rule out alternative explanations for these findings; aggression on the return of experimentally removed helpers could reflect the re-establishment of dominance hierarchies

(Wong and Balshine, 2011), and waiting for other group members to arrive before provisioning nestlings could reflect an attempt to reduce nest predation

(Raihani et al., 2010). As such, there seems yet to be any conclusive evidence of provisioning behaviour acting as payment of rent or a signal of individual quality in cooperative birds.

140

Chapter 4

To date, just one cooperatively breeding species provides compelling evidence for the pay-to-stay hypothesis. Several experimental studies of the cooperatively breeding cichlid Neolamprologous pulcher, have tested multiple predictions of the pay-to-stay hypothesis and found supporting evidence (e.g.

Zöttl et al., 2013; Fischer et al., 2014). Why only this fish system has provided evidence for pay-to-stay is currently unclear, although it is postulated that the extraordinarily high predation risk that individuals face when excluded from active territories, coupled with the comparative ease of monitoring helping effort given their small territories, may provide unusually strong selection for the enforcement of helping via pay-to-stay (Heg, et al., 2004; Fischer et al., 2014).

Further experimental studies in other systems are necessary to understand how widespread this mechanism is for shaping helper contributions.

Our finding that natal subordinate males synchronised their provisioning visits more with those of the dominant male than immigrant subordinate males was surprising, running counter to our predictions from the pay-to-stay hypothesis.

Potentially, natal subordinates may stand to benefit differentially from travelling with the dominant male. For example, dominant males have been shown to respond first to territorial intrusions by engaging with several territorial defence behaviours (York & Young, unpublished data). There might be an advantage for natal subordinate males to also respond quickly to territorial intruders. As subordinate males do not gain any within-group paternity (Harrison, et al.,

2013a) and minimal extra-group paternity (~ 15% of the 12-18% extra-group paternity lost by dominant males; Harrison, et al., 2013b), indirect fitness benefits are of high value to natal subordinates. It is conceivable that natal subordinates benefit from coordinating activities with the dominant male so to 141

Chapter 4 respond and support the defence against territorial intrusions. In doing so, the natal subordinate could help maintain the resident male’s dominance status

(and hence access to breeding opportunities), thereby protecting future indirect fitness benefits. Additionally, the subordinate may gain benefits from the experience of the dominant male with detecting intruders. Consequently, the dominant male and natal subordinate male might be more likely to provision at similar times. Alternatively, the lower provisioning synchronicity between immigrant subordinate males and dominant males may be due to the immigrant subordinate male actively avoiding the dominant male (relative to subordinate natal males). It is possible that the dominant male is less tolerant of the immigrant subordinate’s presence or proximity relative to other group members, hence resulting in lower levels of provisioning event synchrony.

Our findings provide no support for the predictions of the pay-to-stay and social prestige hypotheses for the evolution and/or maintenance of helping behaviour.

It therefore seems unlikely that signalling plays a significant role in modulating individual contributions to provisioning effort in this species. However, even if these hypotheses play no role in the regulation of provisioning by helpers, they could still conceivably play a role in the regulation of other cooperative behaviours in white-browed sparrow weavers. Our findings (in particular natal subordinate males being more likely to provision offspring than immigrant subordinate males, and the lack of evidence of attempts to modify perceived contributions via synchronisation) are broadly consistent with a central role for kin selection in shaping patterns of cooperation in this species. However, the fact that immigrant subordinate males do sometimes contribute to helping, albeit with lower probability than natal subordinate males, demands an explanation. 142

Chapter 4

Perhaps the most tractable adaptive explanation for their contributions is that they may stand to gain downstream direct benefits from their actions, by augmenting group size via improved offspring survival. Such group augmentation may yield downstream direct benefits not only because living in a larger group may enhance survival (Kokko, et al., 2001), but because dominance vacancies in this species are typically taken by immigrant males

(Harrison et al., 2014) and so immigrant subordinates that help to rear more offspring may frequently be augmenting the size of the workforce that ultimately helps to rear their own young. Support for the pay-to-stay and social prestige hypotheses will ultimately require evaluation of the direct fitness consequences of engaging with helping behaviour.

143

Chapter 4

144

Chapter 5

CHAPTER 5

PROLACTIN AND THE REGULATION OF OFFSPRING

PROVISIONING BY PARENTS AND HELPERS IN A COOPERATIVELY

BREEDING BIRD

Lindsay A. Walker, Simone L. Meddle, Linda Tschirren, Jennifer E. York, Peter J. Sharp, Andrew J. Young

145

Chapter 5

146

Chapter 5

5.1 ABSTRACT

Evidence of marked individual differences in contributions to offspring care in cooperative societies is generating increased interest in the proximate causes of such variation. The proximate mechanisms of variation in cooperative care may be similar to those that have been identified to regulate parental care in non-cooperatively breeding species. Here, we investigate whether the provisioning effort of parents and helpers in cooperatively breeding white- browed sparrow weaver (Plocepasser mahali) societies is regulated by circulating levels of the pituitary hormone prolactin. First, we report that those classes of group members that provision offspring at higher rates and with longer provisioning visits have higher circulating prolactin levels. Likewise, across all individuals, individual variation in prolactin levels positively predicted individual variation in both provisioning rate and visit duration. However, when the effects of variation in class were statistically controlled, or when data from the dominant female were removed, variation in prolactin levels no longer significantly predicted individual variation in either provisioning rate or visit duration. Moreover, within-individual changes in prolactin levels within a breeding attempt did not significantly predict within-individual changes in provisioning rates. As such, our findings provide at best mixed support for a role for prolactin in regulating provisioning in this cooperatively breeding bird. With a view to advancing our understanding of the causality or prolactin-behaviour relationships, we also tested the efficacy of subcutaneous vasoactive intestinal peptide (VIP) implants for increasing circulating prolactin levels, and monitored their effects on behaviour. However, the dosage used (based on existing validations in birds) failed to affect prolactin levels and had no impact on provisioning rates.

147

Chapter 5

148

Chapter 5

5.2 INTRODUCTION

In cooperatively breeding societies, where individuals (termed ‘helpers’) provide care to young which are not their own, group members can vary substantially in their contributions to the cooperative care of young (Komdeur, 2006; Sanderson et al., 2015). These cooperative activities that helpers engage with can include both direct offspring care (such as tending eggs: Taborsky, 1984; Zöttl et al.,

2013; and provisioning young: Cockburn, 1998; Russell, 2004) and indirect forms of care (such sentinel duty: Santema and Clutton-Brock, 2013; and stocking communal food-stores: Young et al., 2015). To date, research has primarily focussed on seeking evolutionary explanations for this variation in cooperative contributions to care, while the physiological mechanisms that underpin such variation have received comparatively little attention (Soares et al., 2010).

Attempts to identify the proximate mechanisms that regulate individual contributions to cooperative care (in cooperatively breeding species) may be well served by testing candidate mechanisms identified for the regulation of parental care (in non-cooperatively breeding species). Numerous studies have now suggested that the protein hormone prolactin may play a key role in modulating and facilitating the expression of parental behaviour in bi-parental species. In birds, the transition from sexual activity to parenting is associated with increased prolactin secretion (Sharp et al., 1998; Sockman et al., 2006), with prolactin levels at their highest during parental care behaviours (Buntin,

1996). For example, a study in zebra finches (Taeniopygia guttata) revealed that prolactin levels were the lowest in non-paired birds, increased after pair

149

Chapter 5 bonds had formed, and were highest in incubating birds (Christensen and

Vleck, 2008). Indeed, low prolactin levels have been shown to be associated with egg abandonment in the red-footed booby (Sula sula, Chastel and Lormée,

2002), and failed breeding in black-legged kittiwakes (Rissa tridactyla, Chastel et al., 2005). Experimental evidence also lends support to a causal link between prolactin and avian parental behaviours, with an experimentally induced rise in prolactin levels increasing incubation behaviour in American kestrels (Falco sparverius, Sockman et al., 2000) and an experimentally induced decrease in prolactin resulting in a lower frequency of chick kidnapping in the emperor penguin (Aptenodytes forsteri, Angelier et al., 2006). Less research, however, has sought to establish whether individual variation in prolactin levels predicts individual variation in provisioning rate, with mixed evidence from the few studies to date. Duckworth et al., (2003) found that circulating concentrations of prolactin were positively associated with paternal provisioning in the house finch

(Carpodacus mexicanus). For house sparrows (Passer domesticus), however, prolactin concentrations only correlated positively with provisioning rates when corticosterone concentrations were not included as a predictor in the analysis

(Ouyang et al., 2011). Such correlations between circulating prolactin concentrations and individual levels of offspring provisioning should be viewed cautiously, however, as associations between prolactin and provisioning rates could reflect a non-causal association, or a causal relationship in either, or indeed both, directions. Prolactin might increase as a consequence of high rates of exposure to offspring (visual and tactile stimuli, such as from the nest, eggs or young, have been shown to increase prolactin concentrations; Hall,

1987; Sharp et al., 1998), and higher levels of prolactin may induce higher levels of provisioning effort (Badyaev and Duckworth, 2005). Experimental

150

Chapter 5 manipulations of circulating prolactin levels are therefore required to advance our understanding of the hormonal mechanisms that regulate individual contributions to offspring provisioning by parents. To our knowledge, just one experimental study to date has investigated whether manipulating circulating prolactin levels impacts individual contributions to provisioning. Consistent with a causal role for prolactin, experimentally increasing the circulating prolactin levels of male house finches induced a significant increase in their offspring provisioning rates (Badyaev and Duckworth, 2005).

A number of studies have now begun to investigate the relationship between prolactin and the provisioning behaviour in cooperatively breeding species (in which parents and non-breeding helpers may care for young). In cooperative societies, group members can vary substantially in their contributions to provisioning, with some individuals contributing nothing at all, while others contribute at very high rates (e.g. Clutton-Brock et al., 2004; Arnold et al., 2005;

Komdeur, 2006; Sanderson et al., 2015). This variation in helping effort often cannot be explained by variation in relatedness between carers and recipients

(e.g. Clutton-Brock et al., 2000; Wright et al., 2010). Indeed, evidence of marked and consistent individual differences in contributions to cooperative behaviour (e.g. Komdeur, 2006, Bergmüller et al., 2010, Sanderson et al., 2015) is generating increased recent interest in the proximate mechanisms that may generate such variation. Given the likelihood that the mechanisms that regulate cooperative care have been co-opted over evolutionary time from the mechanisms that regulate parental care (as cooperatively breeding vertebrates may commonly have evolved from monogamous pair-breeding species;

Cornwallis et al., 2010; Lukas and Clutton-Brock, 2012), the limited work to date 151

Chapter 5 on the mechanisms that regulate cooperative care has focussed in particular on the potential role of prolactin (though a role for glucocorticoids has also been considered; e.g. Carlson et al., 2006a, 2006b; Sanderson et al., 2014).

However, as for bi-parental species, the evidence that individual variation in prolactin levels predicts individual variation in provisioning rates is mixed. In a study of Harris’ hawks (Parabuteo unicinctus), male helpers provisioned offspring at higher rates than parents and showed higher circulating prolactin levels (Vleck et al., 1991). For Florida scrub-jays (Aphelocoma coerulescens), the prolactin levels of male helpers were positively correlated with their nestling provisioning rates, but the same was not true of parents (Schoech et al., 1996).

In meerkat (Suricata suricatta) societies, circulating prolactin levels positively predicted individual contributions to babysitting (Carlson et al., 2006b), but only predicted individual contributions to pup-feeding when cortisol data were not included in the analysis (Carlson et al., 2006a). Furthermore, the prolactin levels of male meerkat helpers were unaffected by the playback of pup-begging calls that elicited helping behaviour; instead, such playbacks elevated individual cortisol levels (Carlson et al., 2006a). In red-cockaded woodpeckers (Picoides borealis), the circulating prolactin levels of parents and helpers were unrelated to their nestling provisioning rates (Khan et al., 2001). These mixed results highlight the need for (i) further studies investigating whether natural circulating levels of prolactin predict individual contributions to cooperative care in both parents and helpers, and (ii) experimental studies that manipulate circulating prolactin levels to test for a causal effect of circulating prolactin on cooperative contributions to provisioning. To date, such manipulations have never been conducted in a cooperatively breeding species (but see Carlson et al., 2003 for a pilot study in meerkats).

152

Chapter 5

Here, we investigate whether natural variation in circulating levels of prolactin positively predicts cooperative contributions to provisioning by both parents and helpers in cooperatively breeding white-browed sparrow-weavers (Plocepasser mahali). We also experimentally investigate whether vasoactive intestinal peptide (VIP hereafter) implants can be used to manipulate circulating prolactin levels in this species. White-browed sparrow weavers live in year-round territorial groups comprising a dominant pair and subordinates of both sexes.

Male and female subordinates have typically delayed dispersal from their natal group (Harrison et al., 2013a), but immigrant subordinates of either sex are not uncommon. Despite the presence of immigrant individuals, subordinate males have never been known to secure paternity within their social group (Harrison et al., 2013a). Among females, maternity is completely monopolised by the dominant female, and there is only one active nest within each social group’s territory at any time (Harrison et al., 2013a). All group members contribute to a wide range of cooperative activities, including the provisioning of chicks (Lewis,

1982; Harrison et al., 2013a), but the proximate mechanisms that underpin variation in contributions to offspring provisioning have never been investigated in this species.

Specifically, we aim to test the hypothesis that prolactin regulates parental and alloparental contributions to provisioning in white-browed sparrow weaver social groups by testing the following predictions. First, we predict that the patterns of variation in mean provisioning effort (assessed via provisioning rate and visit duration) among different classes of bird (dominant males, dominant females, natal subordinates, immigrant subordinates) will be mirrored by comparable patterns of variation in mean circulating prolactin levels. Second, we predict that 153

Chapter 5 individual variation in provisioning rate and duration within these classes will be positively predicted by circulating levels of prolactin. Third, we predict that within-individual changes in provisioning rate over the course of a breeding attempt will be positively predicted by the within-individual changes in prolactin levels. If there were consistent individual differences in prolactin receptor densities in the brain, then this explicitly within-individual approach may reveal a relationship between circulating prolactin and provisioning effort that might otherwise have been shrouded by among-individual variation in prolactin reception.

Last, we experimentally investigate whether VIP implants can be used to increase the circulating prolactin levels of dominant males in this wild population of cooperative birds, and monitor any associated effects on provisioning behaviour. Dominant males were chosen for the VIP manipulation as their provisioning rates are invariably lower than that of their social mate and could therefore conceivably be increased in response to a VIP-induced increase in circulating prolactin levels. In birds, prolactin is released from cells of the anterior pituitary, under the stimulatory control of VIP (Freeman et al., 2000).

Experimentally induced increases in circulating VIP (achieved via injections of the peptide), have been shown to increase circulating prolactin levels in captive zebra finches (Vleck & Patrick, 1999; Christensen & Vleck, 2008), Mexican jays

(Aphelocoma ultramarina) and blue jays (Cyanocitta cristata, Vleck and Patrick,

1999). Furthermore, subcutaneous implants that release VIP continuously also resulted in an increase in circulating prolactin levels in captive Mexican jays

(Vleck and Patrick, 1999) and free-living male house finches (Badyaev and

Duckworth, 2005). Our study will be only the second, to our knowledge, to 154

Chapter 5 investigate whether VIP implants impact circulating levels of prolactin in the wild, and will be the first to do so in a cooperative breeder.

5.3 METHODS

5.3.1 STUDY SPECIES AND POPULATION

Data were collected in the context of a long-term research project that monitors

~40 cooperative groups of white-browed sparrow weavers at Tswalu Kalahari

Reserve, South Africa (27°16S, 22°25E). Data were collected at a similar time in two separate breeding seasons (January to February 2013, and January to

March 2014). White-browed sparrow weavers are rain-dependent breeders in this semi-arid region, and may breed at any time from October through to April

(the Southern summer), depending on the timing of major rains. Each bird within our study population is fitted with a metal ring and three colour rings, providing a unique ring combination for identification (SAFRING licence 1444).

From around 6 months of age, male and female white-browed sparrow weavers can be distinguished by their beak colour; males have dark coloured beaks, females a paler horn coloured beak (Harrison et al., 2014). Social dominance was assigned based on weekly monitoring of key dominance-related behaviours, as previously described in (Collias and Collias, 1978; Voigt et al.,

2006; Harrison et al., 2013b). The dispersal status (Natal or Immigrant) of subordinate birds was determined via the continuous monitoring of the life- histories of all birds in the study population since 2007. For this study, social group size ranged from 2 to 6 individuals (mean: 3.3 birds) and was defined as the number of birds consistently seen foraging and roosting together at the time of the focal breeding attempt. Provisioning and endocrine data were collected 155

Chapter 5 from 45 breeding attempts by 30 different social groups. All protocols were approved by the University of Exeter Ethics Committee and the University of

Pretoria Ethics Committee, and complied with regulations stipulated in The

Association for the Study of Animal Behaviour (ASAB) Guidelines for Use of

Animals in Research.

5.3.2 MONITORING PROVISIONING EVENTS

To reliably identify each adult individual during the recording of nestling provisioning events (see Appendix 4 for determining breeding status of social group), each group member was captured from their roost during the breeding attempt’s incubation period (details below; see also Cram et al., 2015) and marked on the vent with a unique dye-mark. The dominant female was usually left unmarked to minimise disturbance during incubation. To record provisioning events, a Panasonic SDR-S50 camcorder attached to a tripod (approximately

0.5 meters in height) was placed directly underneath the entrance to the active nest. The tripod was placed under the nest entrance two days before recording commenced to allow habituation to the tripod. Provisioning was monitored for a combined 85 – 296 minutes per social group (mean: 189 minutes) on either one

(n = 2 breeding attempts) or two mornings (n = 43 breeding attempts), and commenced between 06:10 and 07:42 (start time was adjusted throughout the season relative to sunrise). Provisioning events were monitored by videotape at all breeding attempts (n = 45) between days 6 and 9 inclusive of the provisioning period (most frequently day 7 and day 8; henceforth referred to as

‘early’ in the provisioning period). To obtain a matched blood sample for prolactin determinations, the birds were then blood sampled (see below for

156

Chapter 5 procedure) on the evening of the second day of video recording (median: day 8,

Inter-Quartile Range [IQR]: 8-8.5 of the provisioning period). Video recordings were examined using VLC Media Player version 2.2, with feeder identity (via distinct dye mark) and duration of time the feeder spent in the nest recorded for each provisioning visit.

5.3.3 COLLECTION AND PROCESSING OF BLOOD SAMPLES

5.3.3.1 CAPTURE OF BIRDS

For the fitting of rings, dye-marking, morphometric data collection and blood sampling, birds were captured individually at night from a roost chamber within the social group’s territory (Cram et al., 2015) by flushing individuals into a custom-made capture bag. All captures, data collection and blood sampling were conducted by one person (LW). Birds were then returned to a roost within their social group’s territory to pass the remainder of the night. Tarsus length

(±0.1 mm; callipers) and body mass (±0.01 g; Durascale 100; MyWeigh,

Phoenix, AZ, USA) were collected on capture. To allow us to control for potential effects of body condition on circulating prolactin levels and provisioning rates, we calculated the Scaled Mass Index (SMI) as a proxy for body condition, as outlined in Peig & Green (2009). The estimation of a scaling exponent (SMA) from a reduced major axis regression of the logarithm of body mass on the logarithm of tarsus length was required to calculate SMI values (in this study, SMA = 0.15). The body mass values of all birds were then scaled to the expected equivalent for a bird of the mean tarsus length in our sample (24.6 mm). For full details of the SMI calculation and rationale, see Peig & Green

(2009).

157

Chapter 5

5.3.3.2 COLLECTION AND STORAGE OF BLOOD SAMPLES

Upon capture, a small blood sample (c. 140 µL) was immediately taken from the brachial vein of the bird using a 26-g needle and heparinized capillary tubes.

Time of capture was recorded to investigate any effects of time since sunset on hormone levels (i.e. circadian rhythms of hormone secretion), with captures occurring from just after dusk. The time lag between capture and blood sample completion (mean ± S.E.: 191.6 ± 4.7 seconds) was also recorded to account for any effects of capture stress on hormone levels. Blood samples were centrifuged immediately in the field (12,000 g for 3 min; Haematospin 1400;

Hawksley Medical and Laboratory Equipment, Lancing, UK) and the plasma was drawn off and stored in a cyrovial, which was initially placed on ice until it could be transferred to liquid nitrogen on return from the field (mean ± SD time lag from processing to storage on liquid nitrogen: 148 ± 63 min). At the end of the season, samples were transferred to the UK on dry ice and then stored at -

80 °C until analysis at Roslin Institute Laboratory (The University of Edinburgh,

Easter Bush, Midlothian, Scotland).

5.3.4 HORMONAL IMPLANT PROTOCOL

For 22 of the 45 breeding attempts, dominant males that were captured in the

‘early’ provisioning period for prolactin sampling (see above) were alternately assigned to one of two treatments. One treatment group received a VIP implant

(‘VIP’; n = 11 males), the other treatment group received a placebo implant

(‘Sham’; n = 11 males). The nestlings in two breeding attempts allocated the

Sham treatment expired (likely predation events) after the dominant male had been implanted but before post-implant behavioural monitoring and blood

158

Chapter 5 sampling, so only nine males are included in the analysis for Sham treatment.

The implants were a biodegradable matrix consisting of cholesterol, lactose, celluloses, phosphates, and stearates (all of synthetic origin and not derived from biologically active material). The pellets were prepared by Innovative

Research of America, USA, and either contained VIP (porcine VIP, provided by

GeneCust Europe) or were vehicle only (no VIP; placebo). The use of porcine

VIP to stimulate an increase in prolactin levels has been validated in other avian studies (e.g. Vleck and Patrick, 1999; Christensen and Vleck, 2008). Each VIP implant contained 100 µg VIP per implant, and was designed to release VIP at a steady rate over seven days. This dose was chosen based on a previous VIP implant study (that utilised the same implant design) conducted on the house finch, with 98 µg VIP per pellet over 14 days (Badyaev and Duckworth, 2005).

We used the same dosage per unit body mass (0.35 µg of VIP per day per gram of adult body mass, Badyaev and Duckworth, 2005), with an average body mass of 43 g for birds in our study population. Badyaev and Duckworth’s

(2005) study revealed a positive effect of this VIP implant on plasma prolactin levels by the fifth day following implantation. All implants (including placebos) were designed to last only to the end of the release period. Before implantation, an area of skin under the right wing was anaesthetized with a topical anaesthetic (Xylocaine, 10 mg spray). A small incision was made, and the implant was inserted subcutaneously (between the skin and the muscle). The small opening was sealed with surgical glue (3M Vetbond Tissue Adhesive). All implanted males were observed both in the days that preceded and followed the implant procedure to ensure that there were no adverse effects on the individual

(e.g. social isolation, lethargic or aggressive behaviours).

159

Chapter 5

All birds in the social group where a dominant male had been implanted (n = 20 breeding attempts, as two breeding attempts expired, see above) were captured for a second blood sample five days after the implant had been deployed

(median: day 13, IQR: 13-13.5 of provisioning period; henceforth known as ‘late’ in the provisioning period). As the implants are soluble, they were not removed on re-capture. To coincide with the blood sampling, provisioning behaviour of the social group was video recorded (see above for protocol) on the morning of the fourth and fifth day post-implant. This video monitoring of provisioning behaviour occurred between days 11 and 14 of provisioning period (most frequently days 12 and 13). The timescale of five days was chosen as these implants at this dosage per unit body mass have previously been shown to increase both prolactin levels and provisioning rates within five days post- implantation (Badyaev and Duckworth, 2005). These breeding attempts, for which both ‘early’ and ‘late’ provisioning data were available, were also used to investigate whether within-individual changes in provisioning rates between

‘early’ and ‘late’ provisioning are predicted by within-individual changes in prolactin levels over the same timescale. The implanted birds were monitored over the following weeks and months as part of the long-term research project with no adverse effects on behaviour observed.

5.3.5 RADIOIMMUNOASSAY

The hormonal assays were carried out at the Roslin Institute Laboratory (The

University of Edinburgh, Easter Bush, Midlothian, Scotland; performed by SM).

Plasma levels of prolactin were measured using a highly specific heterologous micro-radioimmunoassay of donkey anti-rabbit serum to European starling

160

Chapter 5

(Sturnus vulgaris) prolactin (Sharp antibody code 44/2). Prolactin was radiolabelled with iodine125 using chloramine-T. 168 (out of a total of 208) samples were assayed in duplicate, and the remaining 40 samples were assayed as singletons. All samples were measured in a single assay, in which the intra-assay coefficient of variation for the duplicate samples was 3.31%.

5.3.6 STATISTICAL ANALYSIS

All statistical analyses were conducted with R v 3.0.3 (R Development Core

Team, 2014). All models were general linear mixed models with Gaussian error structure, and were checked for normality of residuals and homogeneity of variance. Statistical modelling utilized a stepwise model simplification approach: initially all fixed terms (see Table 5.1 and A2.1) were fitted together and then the non-significant terms with the least explanatory power were sequentially removed until a minimal adequate model was reached (retaining only those predictors whose removal now yielded a significant reduction in the explanatory power of the model). All random terms (see Table 5.1 and A2.1) were retained in the model throughout. The assessments of statistical significance reported for each of the fixed terms are those calculated from the change in explanatory power on removal of that term from the minimal model (if it was in the minimal model) or following its inclusion in and subsequent removal from the minimal model (if it had not been retained in the minimal model). Throughout the analyses, the factor ‘Year’ was defined as either 2012-13 or 2013-14 (the two

Southern summer breeding seasons over which this study was conducted).

‘Class’ was defined as dominant male, dominant female, natal subordinate or immigrant subordinate. As defined above, ‘early’ in the provisioning period is

161

Chapter 5 where provisioning events (typically days 7 and 8; see above) and/or prolactin levels (typically day 8; see above) were monitored prior to implanting the dominant male, and ‘late’ in the provisioning period refers to the monitoring of provisioning events (typically days 12 and 13; see above) and/or prolactin levels

(typically day 13; see above) post-implantation. Post-hoc comparisons reported were performed using the glht function in package multcomp in R (Hothorn et al., 2008). See Table A4.2 in Appendix 4 for full details of all explanatory variables investigated in each model.

Provisioning rate was quantified as the number of visits to the active nest in the observation period for each individual within each provisioning period category

(‘early’ or ‘late’; definitions above) for each breeding attempt. For all analyses, provisioning rate refers to this calculated average. Our estimates of the provisioning rates of the birds were significantly repeatable across the two consecutive days of measurement, both for dominant individuals (n = 82 individuals; F1,63 = 6.65, r = -0.46, P < 0.001) and subordinates (n = 41 individuals; F1,34 = 7.01, r = -0.40, P < 0.001, Lessells and Boag, 1987). The duration of provisioning visits was calculated as the average of all provisioning visit durations across all days where data were available for each breeding attempt within each provisioning period category (‘early’ or ‘late’; definitions above). For all analyses, provisioning visit duration refers to this calculated average.

162

Table 5.1 Table outlining all models used in this study, specifying the fixed and random effects. ‘Early’ denotes measures obtained in the early provisioning period, ‘late’ denotes measures taken in the late provisioning period (see Methods for further details). Response Variable Model Error Fixed factors Random factors Sample distribution size (a) How do the different classes of bird differ in their circulating prolactin levels? Prolactin [‘early’] GLMM Normal Class, Year, Social group size, Brood size, Clutch identity, 84 Photoperiod of bleed date*, Time from sunset Individual identity to capture, Lag from capture to blood sample, Body mass (SMI) (b) How do the different classes of bird differ in their provisioning behaviour? Provisioning rate GLMM Normal Class, Year, Social group size, Brood size, Clutch identity, 84 [‘early’] Body mass (SMI) Individual identity Provisioning visit GLMM Normal Class, Year, Social group size, Brood size, SMI Clutch identity, 65 duration [‘early’] Individual identity (c) Does individual variation in circulating prolactin levels predict provisioning behaviour? Provisioning rate GLMM Normal Prolactin, Class, Year Clutch identity, 91 [‘early’] Individual identity Provisioning visit GLMM Normal Prolactin, Class Clutch identity, 67 duration [‘early’] Individual identity (d) Does the within-individual change in prolactin levels within a breeding attempt predict the within-individual changes in provisioning rate? Within-individual GLMM Normal Within-individual change in prolactin, Implant type** 25 change in provisioning Provisioning rate [Early], Class rate Within-individual GLMM Normal Within-individual change in prolactin, Within- Implant type** 20 change in provisioning individual change in body condition (SMI) rate (e) Does the VIP implant affect circulating prolactin levels and provisioning rates? Prolactin [‘early’ + ‘late’] GLMM Normal Implant category, Day 8 or Day 13 Individual identity, 40 Year

163

Within-individual GLMM Normal Implant category Year 20 change in prolactin Provisioning rate [‘early’ GLMM Normal Implant category, Day 8 or Day 13 Individual identity, 40 + ‘late’] Year Within-individual GLMM Normal Within-individual change in prolactin, Implant Year 20 change in provisioning category * For details of photoperiod calculation, see appendix A.2. ** The random effect of implant type was the implant category that the whole group experienced through their dominant male (VIP, Sham).

164 Chapter 5

5.4 RESULTS

5.4(A) HOW DO THE DIFFERENT CLASSES OF BIRD DIFFER IN THEIR CIRCULATING

PROLACTIN LEVELS?

Natural circulating levels of prolactin in the ‘early’ provisioning period were

2 strongly influenced by the Class of the focal bird (GLMM: χ 3 = 41.39, P <

0.001; n = 84 measurements of 75 individuals from 30 social groups; Figure

5.1a), with dominant females displaying the highest levels of prolactin (mean ±

S.E: 3.56 ± 0.28 ng/ml; dominant males: 1.81 ± 0.17 ng/ml; natal subordinates:

1.85 ± 0.28 ng/ml; immigrant subordinates: 1.05 ± 0.19 ng/ml). Post-hoc analyses revealed that while dominant females had significantly higher circulating levels of prolactin than other classes of individual (P < 0.001), there was no significant difference between any of the other classes (P ≥ 0.33).

Prolactin levels were also significantly higher in the 2013/14 breeding season

2 than the 2012/13 breeding season (Year: χ 1 = 12.42, P < 0.001; Figure 5.1b).

There were no significant effects on circulating prolactin levels of social group size, brood size, time since sunset when capture occurred, time from capture to

2 blood sample collection or photoperiod of bleed date (for all, χ 1 ≤ 2.71, P ≥

2 0.10). Additionally, SMI did not predict individual prolactin levels (χ 1 = 0.16, P =

0.69). Similar patterns of among-class variation in circulating prolactin levels were found in ‘late’ provisioning period and when data from ‘early’ and ‘late’ provisioning were combined (see Appendix 4).

165

Chapter 5

5.4(B) HOW DO THE DIFFERENT CLASSES OF BIRD DIFFER IN THEIR PROVISIONING

BEHAVIOUR?

Natural provisioning rates in the ‘early’ provisioning period were significantly

2 predicted by Class (GLMM: χ 3 = 46.11, P < 0.001; n = 84 measurements of 75 individuals from 30 social groups; Figure 5.2a), with the among-class differences in provisioning rate largely mirroring those for prolactin (Figure

5.1a); dominant females provisioned offspring at the highest rate (mean ± S.E:

7.10± 0.66), followed by dominant males (2.79± 0.31), natal subordinates

(2.37± 0.59), and finally immigrant subordinates (0.17± 0.17). Post-hoc analysis revealed that dominant females had significantly higher rates of provisioning than other classes of individual (P < 0.001). While there was no significant difference between any of the other classes (P ≥ 0.069), there was a borderline significant difference found between the provisioning rates of dominant males and immigrant subordinates (P = 0.069). There was also a significant effect of

2 Year on provisioning rates (χ 1 = 5.73, P = 0.017; Figure 5.2b), again mirroring that seen for prolactin (Figure 5.1b), with higher rates in the 2013-14 breeding season than in 2012-13. There were no significant effects of SMI, social group

2 size or brood size (for all, χ 1 ≤ 1.25, P ≥ 0.26). The average duration of a

2 provisioning event was also significantly affected by Class (χ 2 = 23.38, P <

0.001; n = 65 measurements of 57 individuals from 29 social groups). Dominant females spent the longest time in an average provisioning event (mean ± S.E:

110.59 ± 16.68 seconds), followed by dominant males (34.89 ± 8.70 seconds) then natal subordinates (31.00 ± 5.55 seconds). Immigrant subordinates were not represented in the analyses due to lack of data for this carer class (as immigrant subordinates had such low provisioning rates; Figure 5.2a). SMI,

166

Chapter 5 year, brood size and social group size did not affect the average duration of a

2 provisioning event (for all, χ 1 ≤ 0.99, P ≥ 0.32).

5.4(C) DOES INDIVIDUAL VARIATION IN CIRCULATING PROLACTIN LEVELS PREDICT

PROVISIONING BEHAVIOUR?

Individual variation in circulating prolactin levels in the early provisioning period significantly positively predicted individual variation in provisioning rate (GLMM:

2 χ 1 = 17.31, P < 0.001; n = 91 measurements of 75 individuals from 30 social groups; Figure 5.3a), in a mixed-effects model with no other fixed effect predictors (the effects of repeated measures of individuals and breeding attempts were controlled via the random effects structure). However, examining the effects of prolactin on provisioning rate in dominant males and both subordinate classes only (i.e. following removal of the dominant female data)

2 revealed no significant relationship (χ 1 = 1.08, P = 0.30; n = 69 measurements of 54 individuals from 28 social groups), suggesting that the dominant female’s higher prolactin levels coupled with higher provisioning rates (see Figures 5.1a and 5.2a) were driving the observed relationship. Indeed, after controlling for

2 2 the significant effects of Class (χ 3 = 56.71, P < 0.001) and Year (χ 1 = 5.21, P =

0.022) on provisioning rate (as detected in the analyses above), there was no longer a significant positive effect of circulating prolactin levels on individual provisioning rates when utilising the full data set, including dominant females

2 (χ 1 = 0.92, P = 0.34; Figure 5.3b). Similarly, when restricting our analyses to dominant females only, their circulating levels of prolactin did not significantly

2 predict their provisioning rates (χ 1 = 0.26, P = 0.61).

167

Chapter 5

Individual variation in circulating prolactin levels in the early provisioning period also significantly positively predicted individual variation in provisioning visit

2 durations (χ 1 = 11.50, P < 0.001; n = 67 measurements of 75 individuals from

30 social groups; Figure 5.3c), in a mixed-effects model with no other fixed effect predictors (similar to above, the effects of repeated measures of individuals and breeding attempts were controlled via the random effects structure). However, as with provisioning rate, there was no significant effect of prolactin on provisioning visit duration in dominant males and natal

2 subordinates (i.e. following removal of the dominant female data; χ 1 = 1.85, P =

0.17; n = 48 measurements of 38 individuals from 24 social groups), again suggesting that the dominant female’s higher prolactin levels coupled with longer provisioning visits (see Figure 5.1a and details above) were driving the observed relationship. Indeed, there was no longer a significant positive effect of circulating prolactin levels on individual provisioning visit durations when

2 utilising the full data set including dominant females (χ 1 = 0.95, P = 0.33;

2 Figure 5.3d), while controlling for the significant effects of Class (χ 2 = 26.51, P

< 0.001; as detected in the analysis above). Also, when utilising only data from the dominant females, circulating levels of prolactin did not significantly predict

2 their provisioning visit durations (χ 1 = 0.0021, P = 0.96).

5.4(D) DOES THE WITHIN-INDIVIDUAL CHANGE IN PROLACTIN LEVELS WITHIN A

BREEDING ATTEMPT PREDICT THE WITHIN-INDIVIDUAL CHANGES IN PROVISIONING

RATE?

The within-individual change in prolactin from the early to late provisioning periods did not significantly predict the within-individual change in provisioning

168

Chapter 5

2 rate over the same interval (GLMM: χ 1 = 0.36, P = 0.55; n = 25 individuals from

18 social groups). This within-individual change in provisioning rate was also unaffected by the individual’s provisioning rate in the early provisioning period

2 2 (χ 1 = 0.32, P = 0.57) or their Class (χ 1 = 1.96, P = 0.37). Using the subset of data for individuals for which body mass data were obtained at capture in both contexts, allowing the calculation of SMIs (n = 20 individuals from 15 social groups), the within-individual change in SMI also did not predict the within-

2 individual change in provisioning rate (χ 1 = 0.033, P = 0.86).

5.4(E) DOES THE VIP IMPLANT AFFECT CIRCULATING PROLACTIN LEVELS AND

PROVISIONING RATES?

There was no significant interaction between implant category (VIP, Sham) and stage of implant (pre-implant [‘early’ provisioning period] versus post-implant

[‘late’ provisioning period]) in our model of the prolactin levels of dominant

2 males (GLMM: χ 1 = 1.12, P = 0.29; Figure 5.4a), suggesting that the VIP implant did not influence circulating prolactin levels. There was also no significant difference in the prolactin levels of the males in the two treatment

2 groups (VIP versus Sham) either prior to implantation (‘early’ provisioning: χ 1 =

2 0.17, P = 0.68) or following implantation (‘late’ provisioning: χ 1 = 1.36, P =

0.24), nor did these experimental males (both treatment groups combined) differ significantly in their circulating prolactin levels between the early and late

2 provisioning periods (χ 1 = 2.12, P = 0.14). Furthermore, the within-individual change in prolactin level between the early and late provisioning period was not

2 significantly predicted by the implant category (VIP versus Sham; χ 1 = 1.54, P

= 0.21; Figure 5.4b).

169

Chapter 5

Similarly, there was no significant interaction between implant category (VIP,

Sham) and stage of implant (pre-implant [‘early’ provisioning period] versus post-implant [‘late’ provisioning period]) in our model of the provisioning rates of

2 dominant males (χ 1 = 1.12, P = 0.71; Figure 5.4c), suggesting that the VIP implant did not influence provisioning rate either. Provisioning rates were

2 significantly different between ‘early’ and ‘late’ provisioning (χ 1 = 4.16, P =

0.041), with dominant males (both treatment groups combined) provisioning at lower rates in the ‘early’ provisioning period (mean ± S.E.; ‘early’: 3.02 ± 0.45;

‘late’: 3.98 ± 0.67). There was no significant difference in the provisioning rates of the males in the two treatment groups (VIP versus Sham) either pre-implant

2 2 (‘early’: χ 1 = 0.99, P = 0.32) or post-implant (‘late’: χ 1 = 0.25, P = 0.62).

Furthermore, the within-individual change in provisioning rates between ‘early’

2 and ‘late’ was not affected by the implant category (VIP, Sham; χ 1 = 0.41, P =

0.52; Figure 5.4d), nor by the matched within-individual change in prolactin

2 levels (χ 1 = 0.91, P = 0.34).

170

Chapter 5

Figure 5.1. Circulating levels of prolactin were significantly predicted by (a)

Class (where DF = Dominant Female; DM = Dominant Male; N = Natal

Subordinate; I = Immigrant Subordinate), and (b) Year. The data derives from

84 prolactin measures collected in the ‘early’ provisioning period from 75 individuals (20 Dominant Females, 25 Dominant Males, 23 Natal Subordinates and 7 Immigrant Subordinates) provisioning 45 breeding attempts by 30 social groups.

Figure 5.2. Individual provisioning rates were significantly predicted by (a)

Class (where DF = Dominant Female; DM = Dominant Male; N = Natal

Subordinate; I = Immigrant Subordinate), and (b) Year. The data derives from

84 prolactin measures collected in the ‘early’ provisioning period from 75 individuals (20 Dominant Females, 25 Dominant Males, 23 Natal Subordinates and 7 Immigrant Subordinates) provisioning 45 breeding attempts by 30 social groups. 171

Chapter 5

Figure 5.3. The relationship between two measures of provisioning effort and circulating levels of prolactin in white-browed sparrow weavers. (a) Prolactin levels significantly predicted individual variation in provisioning rates when the significant effects of Class and Year on provisioning rates were not controlled.

(b) Prolactin levels did not significantly predict individual variation in provisioning rates when the significant effects of Class and Year on provisioning rates were controlled. (c) Prolactin levels significantly predicted individual variation in average provisioning visit duration when the significant effect of Class on duration was not controlled. (d) Prolactin levels did not significantly predict individual variation in provisioning visit duration when the significant effect of

172

Chapter 5

Class was controlled. For all graphs, raw data are plotted with the predicted line drawn from the model effect sizes, with the S.E. of the slope represented.

173

Chapter 5

Figure 5.4. Circulating prolactin levels and individual provisioning rates for the two implant categories (VIP, Sham) of dominant male white-browed sparrow weavers. (a) Effects of the implant types on prolactin levels pre-implant (white bars) and post-implant (grey bars). (b) Effects of the implant types on the within- individual change in prolactin levels from pre- to post-implant. (c) Effects of the implant types on provisioning rates pre-implant (white bars) and post-implant

(grey bars). (d) Effects of the implant types on the within-individual change in provisioning rate from pre- to post-implant. For all panels, data plotted from model analysis as outlined with SE represented, derived from nine Sham implant males and 11 VIP implant males.

174

Chapter 5

5.5 DISCUSSION

This study used natural observations of a wild population of white-browed sparrow weavers to investigate the role that prolactin may play in the regulation of offspring provisioning behaviour by parents and helpers. We found that significant among-class variation in circulating prolactin levels was matched by broadly parallel among-class variation in both provisioning rates and visit durations. Dominant females had the highest levels of prolactin, the highest provisioning rates and the longest provisioning visits, followed by the dominant male for all three measures, the natal subordinate, and finally the immigrant subordinates had the lowest provisioning rates and prolactin levels. There was also a congruent effect of year, with the breeding season in 2013-14 associated with both higher prolactin levels and higher provisioning rates. Indeed, individual variation in circulating levels of prolactin positively predicted individual variation in both provisioning rate and visit duration, when among-class and among-year variation were not controlled. Together, this evidence is consistent with the prediction that individual variation in provisioning behaviour is a product of variation in circulating prolactin levels. However, our finer-scale analyses did not support this prediction. First, our analyses suggest that these overall positive relationships between prolactin and provisioning behaviour are driven by the high prolactin levels of dominant females coupled with their high provisioning rates and visit durations (for which potential alternative explanations exist, see below): the relationships became non-significant when the data from the dominant female were removed, and when among-class and among-year variation in the provisioning metrics was controlled. Second, we found that within-individual changes in provisioning rate within a breeding attempt were not predicted by matched within-individual changes in prolactin levels. Combined,

175

Chapter 5 our findings provide at best mixed support for the hypothesis that individual variation in prolactin levels is a major driver of individual variation in provisioning effort (measured as both provisioning rate and average duration of provisioning visits).

One possible explanation for this balance of findings is that prolactin does causally impact provisioning effort as hypothesised, but perhaps constraints on our ability to accurately assess provisioning rates and circulating prolactin levels are leaving the relationships between prolactin and provisioning only apparent where marked differences exist (e.g. among classes and years, and when all individuals are pooled in analysis) and not in finer-scale comparisons (e.g. within classes and within individuals). As our measures of provisioning rate on subsequent days showed repeatable among-individual variation (both within subordinates and within dominants), it is not the case that provisioning rates were simply too variable over short time-scales to be meaningfully quantified using these methods. Also our findings cannot be readily attributed to among- individual variation in capture-to-bleed times (which might generate variation in prolactin levels via the prolactin stress response), as (i) individuals were typically bled within the 3 min threshold commonly considered to reflect

‘baseline’ hormone levels (mean ± SE: 191.6 ± 4.67 s); (ii) we found no significant effect of capture-to-bleed lag on prolactin levels; and (iii) prolactin did not significantly predict provisioning rates (when class and year were controlled) even after restricting our data set to samples collected within 3 min. Perhaps more plausibly, however, the temporal decoupling of provisioning monitoring (in the morning) and prolactin sampling (at dusk) in our study could have weakened an otherwise clearer prolactin-provisioning relationship. However, the 176

Chapter 5 two other studies that detected positive associations between prolactin and provisioning had comparable or greater temporal decoupling of behavioural observations and blood sampling than our study: Duckworth et al., (2003) sampled prolactin and provisioning up to 24 h apart, while Ouyang et al., (2011) sampled prolactin and provisioning one to three days apart. It is conceivable that blood sampling birds after dusk (as we did here; see also the Khan et al.,

(2001) study of prolactin dynamics in wild red-cockaded woodpeckers) exacerbates the decoupling of hormone-behaviour relationships. While this too is certainly possible, we found no evidence to suggest that prolactin levels were simply falling at night, prior to capture from the roost chamber, in the absence of stimuli from the nestlings. Specifically, there was no effect on prolactin levels of the time-lag from sunset to capture, either when the analysis included all birds or when the dominant female was excluded (which is relevant as she broods the chicks while roosting; see also below for the potential importance of this).

If, as our analyses suggest, individual variation in provisioning effort does not arise principally from individual variation in circulating prolactin levels, alternative explanations are required for the markedly elevated prolactin levels of dominant females. There are several potential explanations for these patterns. First, prolactin may not have a causal link with provisioning effort per se, but may instead regulate or be stimulated by other parental behaviours. For example, as dominant females are the sole incubator in this species (Harrison et al., 2013a), their high levels of prolactin during the provisioning period might simply reflect residual circulating prolactin from the incubation period. However, their prolactin levels were still much higher than other classes even two weeks after hatching (‘late’ provisioning period, see Appendix 4). Alternatively, the high 177

Chapter 5 prolactin levels of dominant females could play a role in the regulation of brooding behaviour, rather than provisioning behaviour. Dominant females are also the only class that broods the nestlings, leaving it conceivable that their higher levels of prolactin are either required to maintain brooding behaviour or are stimulated by daytime brooding interactions (Hall, 1987; Sharp et al., 1998), rather than provisioning per se. Indeed, it is conceivable that the prolactin levels of dominant females are more sensitive to nest or nestling stimuli than other classes, as hypothesised by Khan et al., (2001) given their observations that the prolactin levels of dominant female red-cockaded woodpeckers remained elevated during provisioning while those of males declined. This is also consistent with a study of the ring dove (Streptopelia capicola), in which exposure to nestlings stimulated a greater prolactin release in females than males (Lea and Sharp, 1991). Finally, it is also possible that the prolactin levels of dominant females in our study are elevated because she is the only class of bird to brood the chicks while roosting (and so is brooding nestlings at the point of capture for blood sampling). However, the effect of the time elapsed since sunset on the prolactin levels of dominant females did not differ from that for other classes of bird (see Appendix 4), suggesting that roosting on the nestlings did not modify the dominant female’s temporal trajectory of prolactin levels after dusk (one could expect that prolactin levels of dominant males and subordinates decreases as time without the stimulus of the nestlings increases).

A second potential explanation for the markedly elevated prolactin levels of dominant females, is that prolactin is causally impacted by offspring provisioning (as nestling begging can stimulate prolactin release, Hall, 1987;

Sharp et al., 1998), and perhaps a certain threshold level of prolactin is required 178

Chapter 5 to maintain provisioning behaviour (as has been suggested for post-hatching parental care; Angelier et al., 2006; Boos et al., 2007), but prolactin levels play no causal role in modulating variation in provisioning effort. Here, dominant females provisioned offspring at higher rates and with longer provisioning visits than other classes, and so would likely have been exposed to more offspring solicitation cues, providing a plausible explanation for their higher prolactin levels. It is conceivable that the prolactin levels of the dominant birds and the natal subordinates all exceeded a threshold level required for the expression of provisioning behaviour, but that the lower prolactin levels of immigrant birds fell below such a threshold, providing a proximate explanation for their negligible contributions to provisioning. Indeed, the findings of the one study to date to have demonstrated causal effects of prolactin on provisioning behaviour in wild birds are broadly consistent with such a view in which prolactin serves to maintain provisioning behaviour but not modulate the provisioning rates of birds that are provisioning. Badyaev & Duckworth (2005) used VIP implants to increase the prolactin levels of a particular colour morph of male house finch that rarely provisions offspring, and this caused the males provision at high rates; a change that could be interpreted as much as a transition from non- parenting to parenting as the modulation of degrees of parental effort. If interacting with nestlings does impact prolactin, then considerable noise might be expected in the relationship between provisioning effort and prolactin (as observed in this study) due not only to variation in both provisioning rates and durations (as measured here) but also to among-individual variation in the nature of their interactions with the nestlings when inside the nest cup. For example, several bird species have reported that parents differ in the way that they utilise offspring cues when allocating food within the brood (e.g. Leonard

179

Chapter 5 and Horn, 1996; Kilner, 2002; Dickens and Hartley, 2007). However, recent evidence suggests that dominant female and male white-browed sparrow weavers employ similar resource allocation tactics, at least during the period of this study (see Chapter 3). It is conceivable that other behaviours within the nest, for example internal nest maintenance or nestling faecal sac removal, may impact hormone levels.

A notable feature of white-browed sparrow weaver prolactin levels is that they are much lower than have been reported for other similar sized avian species

(white-browed sparrow weaver: 40-50 g body mass, 0-7 ng/ml prolactin; e.g. house finch: 16-27 g body mass, 15-25 ng/ml prolactin, Duckworth et al., 2003; house sparrow: 24-39.5 g body mass, 10-70 ng/ml prolactin, Ouyang et al.,

2011; red-cockaded woodpecker: 40-60g body mass, 10-35 ng/ml prolactin,

Khan et al., 2001; red-eyed vireo, Vireo olivaceus: 12-26 g body mass, 20-80 ng/ml prolactin, Van Roo et al., 2003). This could itself leave it difficult to detect a prolactin / provisioning relationship by compressing the range of observable prolactin values. The lower levels of prolactin observed in our study could reflect chronic environmental stress. During this study, the Southern Kalahari experienced a drought, which may have depressed white-browed sparrow weaver prolactin levels. Both chronically poor environmental conditions such as drought (Delehanty et al., 1997) and the experimental administration of corticosterone (Criscuolo et al., 2005) are known to suppress prolactin. Given this, the drought in the Kalahari during this study period may have impacted the levels of circulating prolactin. Consistent with this view, the drought was at its most severe in the 2012-13 breeding season, when the birds’ prolactin levels were lowest. 180

Chapter 5

Finally, our experimental manipulation provides evidence that VIP implants at the dosage used in this study do not affect either individual prolactin levels or provisioning behaviour in dominant male white-browed sparrow weavers. At least two possibilities might explain the contrast between this finding and the apparent efficacy of such implants in other species. First, differences in physiological attributes of white-browed sparrow weavers relative to other species might account for this finding. For example, the rate of turnover of specific hormone receptors in the brain may be relatively quicker or the specific hormone receptors may be down-regulated relatively quicker in response to elevated prolactin levels. Additionally, the blood-brain barrier may be less permeable at comparable circulating levels of prolactin (prolactin has been shown to modulate permeability; Rosas-Hernandez et al., 2015), with a less permeable brain barrier resulting in lower levels of prolactin secretion for a longer period of time. As a consequence of any (or indeed all) of these factors, the dosage and/or five day period between implementation and blood sampling may not have been appropriate for the attainment of elevated prolactin levels, despite apparent effects of comparable VIP implants on prolactin levels at this dose and over this timescale in another passerine bird (Badyaev and

Duckworth, 2005). Second, VIP may be incapable of elevating prolactin secretion during breeding phases. For example, the prolactin levels of breeding zebra finches did not increase following VIP injection, whereas the prolactin levels of non-breeding zebra finches did increase (Christensen and Vleck,

2008). Indeed, this might explain why similar implants to the ones used here were successful in Badyaev & Duckworth’s (2005) study but not here; that study successfully used VIP implants to elevate the prolactin levels of a certain colour morph of house finch male that rarely provisions offspring (whose endocrine

181

Chapter 5 state may therefore approximate that of a bird not engaged in parental care). A physiological maximum level of endogenous prolactin secretion (El Halawani et al., 1984) could arise through a range of mechanisms, including a negative feedback loop in which prolactin levels modulate the expression of hormone receptors that in turn constrain prolactin release (Hall et al., 1986) or different phases of the reproductive cycle affecting the half-life of plasma prolactin via different translational modifications (Young et al., 1990).

Previous research on cooperatively breeding species has provided mixed evidence for individual provisioning rates being regulated by circulating prolactin levels. Our study, consistent with other studies of cooperative breeders (e.g.

Vleck et al., 1991; Schoech et al., 1996; Khan et al., 2001; Carlson et al., 2006), provides at best mixed support for the view that individual variation in provisioning effort arises as a direct function of variation in prolactin levels. Our findings are, however, broadly consistent with the alternative hypothesis that a threshold concentration of prolactin is required to display provisioning behaviour, rather than variation in prolactin levels directly modulating and individual’s level of provisioning effort. Combined, our findings emphasise the need for more studies to probe the causality of the relationship frequently found between circulating levels of prolactin and provisioning behaviour by manipulating circulating levels of prolactin, but also highlight the potential difficulty of doing so in actively provisioning birds.

182

Chapter 6

CHAPTER 6

GENERAL DISCUSSION

183

Chapter 6

184

Chapter 6

6.1 OVERVIEW

As cooperatively breeding groups tend to comprise closely related individuals, initial research to explain why an individual would engage with helping behaviour primarily focussed on the role of kin selection. Although kin selection is certainly an important force in shaping the expression of cooperative behaviour (e.g. Komdeur, 1994; Russell and Hatchwell, 2001; Nam et al., 2010;

Wright et al., 2010; McDonald and Wright, 2011), two key discoveries altered the direction of this research. First, cooperative societies frequently contained unrelated individuals who also contributed to cooperative activities (e.g. Clutton-

Brock et al., 2000; Wright et al., 2010). Indeed, a meta-analysis revealed that variation in relatedness only accounts for 10% of variation in the likelihood of helping (Griffin and West, 2003). Second, individuals with similar relatedness performed markedly different levels of helping (e.g. Clutton-Brock et al., 2001,

2000), suggesting that individual variation could be due to factors other than relatedness. Combined, these finding suggested that variation in the direct fitness benefits (as well as variation in individual costs to engaging with cooperation), and not just indirect fitness benefits, could conceivably shape helping behaviour in some cooperatively breeding societies. This led to substantial research on exploring the impact of potential direct fitness benefits on the evolution and maintenance of cooperation. In this thesis, I have added to this growing body of research by combining natural observations and experimental approaches where possible to explore the potential for direct benefits arising through three different routes to contribute to selection for cooperation: (i) sexually-selected direct benefits, (ii) accruing social prestige, and (iii) payment of rent.

185

Chapter 6

The substantial variation in individual contributions to cooperative activities within cooperatively breeding societies (e.g. Clutton-Brock et al., 1998; Clutton-

Brock et al., 2002; Clutton-Brock, et al., 2004) may be underpinned by proximate mechanisms, regardless of the evolutionary driving force for cooperation. Research into the proximate mechanisms underlying individual variation has been comparatively less well studied (Soares et al., 2010). In this thesis, I address this shortfall in understanding by employing within-individual measurements to investigate the proximate causes of variation in provisioning behaviour in a cooperative society.

For cooperative societies, studies have largely investigated the sex differences in contributions to alloparental care (reviewed in Cockburn, 1998; Clutton-Brock et al., 2002, 2001) rather than the patterns of parental investment (although see

(Hatchwell and Russell, 1996; Clutton-Brock et al., 2004). As cooperatively breeding vertebrates are likely to have commonly evolved from monogamous pair-breeding species (Cornwallis et al., 2010; Lukas and Clutton-Brock, 2012), it is conceivable that similar selection pressures on parents in bi-parental systems are likely to be evident in cooperative societies to modulate investment in care. For example, in bi-parental care, conflict between parents over their individual contributions to care is expected, as (i) parents are typically unrelated, and (ii) parental care is costly (Trivers, 1972; Clutton-Brock, 1991;

Parker et al., 2002; Lessells, 2012). Such sexual conflict over the division and amount of care is expected as each carer would benefit from the other parent performing a larger share of the work (Trivers, 1972; Houston and Davies,

1985). This is also likely to be true of parents in the majority of cooperative societies. This sexual conflict over care can lead to sex-specific care patterns, 186

Chapter 6 which can in turn intensify sexual conflict (Andersson, 1994; Lessells, 2002;

Royle et al., 2012). However, different life-history traits might impact the resolution of such sexual conflict, potentially influencing the occurrence and/or magnitude of sex-specific patterns of care. For example, the smaller clutch sizes and shorter nestling periods typical of tropical species may have important implications for the evolution of sex-specific patterns of care via impacts on the resolution of sexual conflict. In this thesis, I have addressed this shortfall in understanding by employing behavioural observations and within-nest cameras to explore sex differences in provisioning behaviour by parents of white-browed sparrow weavers, a subtropical bird.

Within each chapter, I have discussed the implications of my results. Here, I will combine my findings of both ultimate and proximate mechanisms for parental and alloparental behaviours, and discuss areas that are ripe for future research.

As my research focus was primarily on the behavioural patterns of provisioning young within social groups, I will concentrate my discussion on this care behaviour. First, I consider the research findings for dominant individuals within white-browed sparrow weaver social groups, and subsequently explore my findings for subordinate individuals. I will then finish by suggesting areas for further study in the white-browed sparrow weaver system.

187

Chapter 6

6.2 INDIVIDUAL CONTRIBUTIONS TO COOPERATIVE PROVISIONING

6.2.1 DOMINANT INDIVIDUALS WITHIN SOCIETIES

Cooperative societies frequently exhibit reproductive division of labour either where few individuals monopolise breeding opportunities (high reproductive skew) or where breeding is distributed more evenly among group members (low reproductive skew, Vehrencamp, 1983; Keller and Reeve, 1994; Field and Cant,

2009). For white-browed sparrow weavers, reproductive skew is extremely high; within the group, reproduction is monopolised by the dominant individuals

(Harrison et al., 2013a). While dominant males do lose 12-18% of paternity to extra-group males (principally dominant males in other groups), there is no evidence of egg dumping by extra-group females (Harrison et al. 2013a;

Harrison et al. 2013b). Given this, extra-pair mating by females may both decrease the benefits and increase the costs of care for males relative to females (as males may face an opportunity cost of caring if they could simultaneously be seeking matings elsewhere). It may therefore be expected that white-browed sparrow weavers exhibit sex-specific patterns of parental investment in care.

In this thesis, I show that sex-specific patterns of parental investment are exhibited in white-browed sparrow weaver societies; dominant females provision nestlings at a higher rate than dominant males (Chapter 3, Chapter

5). These differences in provisioning behaviour were mirrored by among-class variation in prolactin (a pituitary hormone), indicating that there may be a proximate mechanism for the sex-specific patterns in provisioning behaviour.

However, on controlling for class, circulating prolactin levels no longer positively 188

Chapter 6 predicted individual variation in behaviour (Chapter 5). It is possible that the sexes differed in physiological attributes that masked any direct relationship between prolactin and provisioning behaviour, such as different densities, types and/or turnover of specific hormone receptors in the brain (Buntin et al., 1993;

Ohkubo et al., 1998a, 1998b; Freeman et al., 2000; Pitts et al., 2000) or indeed different blood-brain barrier permeability (prolactin has been shown to influence permeability; Rosas-Hernandez et al., 2015). However, although I was unable to specifically test for these possibilities, my field protocol design utilised repeated within-individual measurements thereby controlling as much as possible for any effects of between-individual variation on the relationship between hormone and behaviour. It is conceivable that instead of directly regulating provisioning behaviour, prolactin maintains the expression of this behaviour when exceeding a threshold level of hormone (see below for discussion on subordinate individuals). This threshold mechanism for maintaining care behaviour has also been postulated for post-hatching care in a precocial bird (Boos et al., 2007).

Despite the sex-specific pattern in provisioning rate, dominant males and females employ similar food allocation tactics (Chapter 3). Sex differences in provisioning rate are often predicted to lead to sex differences in food allocation tactics for two main reasons. First, the time and/or cognitive restraints associated with accurately assessing nestling need or transferring food to smaller nestlings may be greater for the sex that attends the nest less frequently (Stamps et al., 1985; Krebs et al., 1999; Lessells, 2002). Second, greater sensitivity to offspring need is expected from the sex that accrues the greater net benefit from providing a unit of care (and therefore might be 189

Chapter 6 provisioning at higher rates (Gottlander, 1987; Lessells, 2002). For dominant white-browed sparrow weavers, however, both sexes fed the nestling that was first to beg and the nestling at the front of the nest cup. Moreover, dominant males and females stayed in the nest for a similar time period during a provisioning visit (adjusted for visits that also included a brooding event, as dominant females are the sole brooder in this species; Harrison et al., 2013a).

The finding of similar resource tactics for dominant individuals was predicted for tropical zone species due to the selection pressures associated with life-history traits typical of tropical living (such as the relatively smaller clutch sizes and shorter nestling periods relative to temperate species). Due to hormone implant issues, I was not able to directly test for a causal role for prolactin to modulate food allocation rules. However, as the dominant female had substantially higher levels of circulating levels of prolactin (Chapter 5) yet employed similar resource allocation tactics (Chapter 3), there is suggestive evidence that hormone levels may not directly regulate feeding rules. It is conceivable that the drought during this study impacted care patterns. For example, previous studies have shown that a decrease (Bize et al., 2006) and an experimental increase

(Boland et al., 1997) in food availability can reverse food allocation patterns, with smaller nestlings only favoured in good environmental conditions (Boland et al., 1997; Davis et al., 1999; Royle et al., 2002; Bize et al., 2006). As such, the patterns observed here may only reflect food allocation patterns during limited food availability (as prey abundance likely decreased with lower rainfall), with any benefits to sex-specific behaviour only likely under good environmental conditions (where food availability is not restricted). However, within-brood mortality for the smaller nestling was similar for before and during this study

(long-term data, unpublished), suggesting that the results in this thesis are likely

190

Chapter 6 to reflect general food allocation patterns. Combined, my work highlights that life-history differences associated with living in the tropics may have important implications for sex-specific patterns of care, potentially via the influence on how sexual conflicts are resolved.

Such differences in behavioural patterns between the sexes may in part reflect an investment trade-off. In this thesis, I provided the first experimental evidence that sexually-selected direct benefits may modulate the expression of a putatively cooperative behaviour (Chapter 2). Focussing on dominant males, I show that they perform higher sentinel effort relative to all other group members, their sentinel activity increases following a simulated intrusion, and that sentinel activity may facilitate a response to intruders (Chapter 2). As within- and extra-group paternity is secured principally by dominant males (only

16% of extra-group and zero within-group paternities is secured by subordinates, Harrison et al., 2013b), an individual’s paternity depends on maintaining dominance status once acquired. There might therefore be intense selection pressure for males to engage with behaviours that protect dominance and thereby protect future paternity. It therefore seems conceivable that engaging with sentinel behaviour secures future reproductive output for the resident male. As such, it could be expected that dominant males face a trade- off between investing in provisioning behaviour or sentinel activity. Fitness benefits to males of investing in the latter may therefore also explain in part why males contribute markedly less to provisioning behaviour than females

(Chapter 3, Chapter 5).

191

Chapter 6

6.2.2 SUBORDINATE INDIVIDUALS WITHIN SOCIETIES

As outlined above, cooperative societies typically have some level of reproductive skew. Within these groups, high ranked or socially dominant individuals breed whereas lower ranked individuals generally do not (although in some species, subordinates do occasionally breed, Field and Cant, 2009). For white-browed sparrow weavers, social groups comprise of a dominant pair and subordinates of both sexes. Male and female subordinates have typically delayed dispersal from their natal group (Harrison et al., 2013a), but immigrant subordinates of either sex are not uncommon. Additionally, as stated above, dominant individuals secure all within-group parentage (Harrison et al., 2013a); therefore, all within-group subordinates are helpers rather than parents of the breeding attempt. As such, natal subordinates are related to the nestlings they are helping their parent(s) to raise, whereas immigrant subordinates are unlikely to be related to the offspring of the dominant pair. If kin selection alone influenced patterns of helping in white-browed sparrow weavers, we would expect immigrant individuals not to contribute to cooperative activities.

However, all group members contribute to provisioning nestlings in white- browed sparrow weaver societies (although there is large inter-individual variation in contributions, Lewis, 1982; Harrison et al., 2013a). As such, it seems unlikely that indirect fitness benefits are the sole explanation for the variation in individual contributions to cooperative activities. Consequently, it could be expected that direct fitness benefits from cooperating modulate the expression of individual contributions.

192

Chapter 6

In this thesis, I reveal that natal subordinates are more likely to perform provisioning behaviour and perform higher rates of provisioning than immigrant subordinates (Chapter 4, Chapter 5). I also reveal that these patterns in subordinate provisioning behaviour are not directly regulated as a function of prolactin. The among-class variation in provisioning rate was reflected in among-class variation in prolactin levels; however, on controlling for the different classes it was found that prolactin did not positively predict provisioning behaviour (Chapter 5). Instead, it is conceivable that the prolactin levels of the dominant individuals (see above) and the natal subordinate all exceeded a threshold level required for the expression of provisioning behaviour, and the lower prolactin levels of immigrant subordinates fell below such a threshold thereby explaining the negligible contributions to provisioning behaviour. I also show that contributions to provisioning by immigrant subordinate males are unlikely to be maintained by direct fitness benefits such as staying on an occupied territory/advantages of group-living (permitted through payment of

‘rent’ via helping; pay-to-stay hypothesis; Gaston, 1978) or future breeding opportunities (by signalling individual quality via helping; social prestige hypothesis; Zahavi, 1995; Zahavi and Zahavi, 1997). Immigrant male subordinates did not inflate their perceived contributions by synchronising their provisioning visits with those of the dominant male or female, providing little evidence for either hypothesis. It is worth noting that these direct benefits may have been important with initially promoting cooperative contributions, but perhaps are no longer the driving force in extant systems. The higher likelihood and actual rates of provisioning behaviour by natal subordinates suggests that indirect fitness benefits could potentially be impacting helper effort in this species. However, this still does not explain why immigrant subordinates do

193

Chapter 6 provision nestlings that are not their own (albeit at low levels). Potentially, immigrant subordinates stand to gain downstream direct benefits from their actions by augmenting group size via improved offspring survival, as (i) dominance vacancies in this species are typically taken by immigrant males

(Harrison et al., 2014), and (ii) living in a larger group may enhance survival

(Kokko, et al., 2001). However, further studies are required to strengthen this statement. Combined, my work highlights that selection for helping behaviour in white-browed sparrow weaver societies does not derive from a significant role in signalling, and are instead broadly consistent instead with a central role for kin selection.

6.3 FUTURE RESEARCH DIRECTIONS FOR THE WHITE-BROWED SPARROW

WEAVER SYSTEM

The tractability of this study species provides fantastic opportunities to examine the mechanisms underpinning behaviour. By employing targeted repeated within-individual measurements and observations, a myriad of possibilities exist.

Building on the research outlined in this thesis, a key development would be to further investigate the potential for sexually-selected direct benefits to modulate

(putatively) cooperate contributions to sentinel activity. First, it would be beneficial to understand how variation in female fertility/receptiveness impacts the sentinel response by the dominant male to a simulated territorial intrusion. It could be expected that a higher female receptiveness (i.e. following rains but prior to egg laying) could prompt a greater response by the resident dominant male. Second, it would be interesting to vary the scale of threat (via more substantial simulated territorial intrusions, or using ‘neighbour’ versus ‘stranger’

194

Chapter 6 playbacks) and monitor the influence of this variation on the sentinel response by the dominant male. Third, using repeated within-individual behavioural measures and matched participants, the downstream effects of varying magnitudes of perceived threat to paternity (via simulated territory intrusions) on contributions to provisioning behaviour could be investigated. This would reveal whether intra-sexual competition at specific points in the breeding cycle affects investment in the current brood (via provisioning behaviour). Finally, this thesis investigated the effect of intra-sexual competition in males on contributions to sentinel behaviour. It is conceivable that sentinel behaviour may also be important in female-female competition. As female white-browed sparrow weavers do not lose any maternity to within- or extra-group females via egg- dumping (Harrison et al. 2013a; Harrison et al. 2013b), access to any or indeed all breeding opportunities are via the acquisition of social dominance and territory ownership. Indeed, intra-sexual competition for dominance status in cooperatively breeding vertebrates may be at least as intense among females as males (Clutton-Brock et al., 2006; Young & Bennett, 2013). As such, it would be interesting to investigate whether intra-sexual competition also shapes sentinel behaviour of females in cooperative societies.

Data collection for this thesis coincided with a severe drought in the Kalahari

Desert, with a sudden ~40% decline in our study population. As such, the patterns of behaviour reported here may only be apparent when individuals have limited resources and are under environmental (and therefore likely physiological) stress. The Sparrow Weaver Project has been running continuously since 2007; therefore, there is a fantastic opportunity to study the environmental effects on several aspects of cooperative behaviour. With a focus 195

Chapter 6 on repeated within-individual measurements over a considerable time frame, investigations could highlight conditions where higher levels of cooperation by subordinates are (i) differentially important via improved individual/offspring survival, and (ii) more likely to occur, thereby revealing environmental aspects that may have been crucial for cooperation to first evolve.

196

Appendices

APPENDICES

A.1 SEXUALLY-SELECTED SENTINELS? EVIDENCE OF A ROLE

FOR INTRA-SEXUAL COMPETITION IN SENTINEL BEHAVIOUR

Table A1.1 Model outputs of mixed-effect models as detailed in the appropriate analyses.

Model terms Estimate SE df P

Natural sentinel data: Do dominant males invest differentially in sentinel behaviour?

Sex × Dominance Figure 2.1 1 0.031 Social group size 0.0095 0.064 1 0.86

Natural sentinel data: Is the dominance effects among males attributable to age?

Age -0.00062 0.00030 1 0.039* Dominance 0.87 0.32 1 0.0063* Social group size -0.013 0.086 1 0.87

Natural sentinel data: Is the dominance effect among males attributable to body condition?†

Body condition < 0.00001 < 0.00001 7 0.14 Dominance < 0.00001 < 0.00001 7 0.053

Experiment 1: Does an intra-sexual challenge impact dominant male sentinel behaviour?†

Observation session stage × Figure 2.2 27 < 0.001 Playback treatment

The Estimate and SE values are reported at the point where the fixed term was either removed from the model (with P > 0.05) or retained (with P < 0.05) under stepwise model simplification. *The significance of these terms was derived when dropping these terms from the minimum adequate model, with all other significant terms retained. †Values reported from a full glmmPQL model (for details, see methods section).

197

Appendices

A.2 SEX DIFFERENCES IN PROVISIONING BEHAVIOUR BUT NOT

RESOURCE ALLOCATION IN A SUBTROPICAL COOPERATIVELY

BREEDING BIRD

Table A2.1 Model outputs of mixed-effect models as detailed in the appropriate analyses.

Model terms – fixed effects Estimate SE df P

Do parents show sex differences in provisioning rate?

Intercept* 7.47 0.65

Sex of parent -5.19 0.92 1 < 0.001 Social group size -0.72 0.43 1 0.088

Do parents show sex differences in provisioning visit duration?

Intercept* 3.37 0.13

Sex of parent -0.19 0.14 1 0.2043

Do parents show sex differences in food allocation rules?

Intercept* 2.64 0.45

Sex of parent × Relative size of nestling 0.73 0.57 1 0.094

Sex of parent × Relative position of 0.78 0.55 1 0.15 nestling

Sex of parent × Begging order -0.38 0.59 1 0.51

Relative size of nestling -0.28 0.28 1 0.31

Relative position of nestling 1.50 0.25 1 < 0.001 Begging order -2.32 0.26 1 < 0.001

Sex of parent 0.15 0.28 1 0.58

198

Appendices

The Estimate and SE values are reported at the point where the fixed term was either removed from the model (with P > 0.05) or retained (with P < 0.05) under stepwise model simplification. *The intercept values were derived from the minimum adequate model, with all other significant terms retained.

199

Appendices

A.3 NO EVIDENCE OF A ROLE FOR SOCIAL PRESTIGE OR

PAYMENT OF RENT IN THE HELPING BEHAVIOUR OF A

COOPERATIVELY BREEDING BIRD

Table A3.1 Model outputs of mixed-effect models as detailed in the appropriate analyses.

Model terms – fixed effects Estimate SE df P

Patterns of provisioning contributions: natal and immigrant male subordinates

(a) Binomial model

Intercept* -1.83 1.23

Nestling age -0.011 0.73 1 0.99

Social group size 0.12 0.12 1 0.30

Clutch size 1.26 0.55 1 0.020

Class - Natal 1.65 0.67 1 0.0021

(b) Guassian model

Intercept* 2.55 0.47

Nestling age 0.42 1.09 1 0.63

Social group size -0.13 0.23 1 0.56

Clutch size 1.50 1.02 1 0.14

Class - Natal 0.57 0.98 1 0.58

Patterns of provisioning contributions: older natal and younger natal male subordinates

(a) Binomial model

Intercept* 0.19 1.49

Nestling age -0.13 1.10 1 0.91

200

Appendices

Social group size -0.030 0.17 1 0.86

Clutch size 1.65 0.78 1 0.025

Class - Old 1.92 1.12 1 0.041

(b) Guassian model

Intercept* -2.30 1.52

Nestling age 2.03 0.83 1 0.017

Social group size -0.12 0.21 1 0.55

Clutch size 2.02 0.78 1 0.026

Class - Old 0.37 0.80 1 0.70

Patterns of provisioning contributions: dominant males with and without an immigrant male present

Intercept* -3.26 0.30

Nestling age 0.12 0.27 1 0.65

Social group size -0.089 0.066 1 0.17

Clutch size 0.22 0.24 1 0.39

Class – Imm. present 0.12 0.28 1 0.67

The Estimate and SE values are reported at the point where the fixed term was either removed from the model (with P > 0.05) or retained (with P < 0.05) under stepwise model simplification. *The intercept values were derived from the minimum adequate model, with all other significant terms retained.

201

Appendices

A.4 PROLACTIN AND THE REGULATION OF OFFSPRING

PROVISIONING BY PARENTS AND HELPERS IN A COOPERATIVELY

BREEDING BIRD

A.4.1 DETERMINING BREEDING STATUS OF SOCIAL GROUP

The breeding status of the social group was required to enable the observation of provisioning behaviour of group members. This was achieved by monitoring the contents of all woven nest structures within each social group territory on the afternoons of two of every three days throughout the potential breeding season (October – March). When one or more eggs were detected, the social group was said to have entered the ‘Incubation’ period. This active nest was then monitored daily in the afternoons until no new eggs were detected. White- browed sparrow weavers lay one egg per day in the morning, and lay clutches of just one to four eggs (Harrison et al., 2013a). Daily monitoring of the active rest resumed on day 14 of incubation phase, as instructed by the lay date of the first egg (incubation lasts 14-19 days, Harrison et al., (2013a). This method reliably determined the hatch date of the first nestling, which was then termed

Day 1 of the provisioning period for the breeding attempt.

A.4.2 PHOTOPERIOD CALCULATION

As prolactin has been shown to increase over a breeding season as a result of photo stimulation (Dawson & Goldsmith, 1982; reviewed in Sockman et al.,

2006) photoperiod data were collected so to investigate and control for any effects in white-browed sparrow weavers. Data on sunrise and sunset was obtained from the Navy Oceanography Portal (Naval Meteorology and 202

Appendices

Oceanography Command, 1100 Balch Blvd, Stennis Space Center, MS) using the coordinates of the study site (27°160S, 22°250E). From this, photoperiod was calculated.

Table A4.1 Table outlining all models detailed in the appendix 4, specifying the fixed and random effects.

Response Model Error Fixed factors Random Sample Variable distribution factors Size (a) How do the different classes of bird differ in their circulating prolactin levels? Prolactin GLMM Normal Class, Year, Social Clutch 34 [‘late’] group size identity, Individual identity, Implant type* Prolactin GLMM Normal Photoperiod of Clutch 34 [‘late’] bleed date, Time identity, from sunset to Individual capture, Lag from identity, capture to blood Implant sample type* Prolactin GLMM Normal Class, Year, Social Clutch 82 [‘early’ + group size, Brood identity, ‘late’] size, Early or Late in Individual provisioning period, identity Photoperiod of bleed date, Time from sunset to capture, Lag from capture to blood sample * The random effect of implant type was the implant category that the whole group experienced through their dominant male (VIP, Sham).

A.4.3 HOW DO THE DIFFERENT CLASSES OF BIRD DIFFER IN THEIR CIRCULATING

PROLACTIN LEVELS DURING THE ‘LATE’ PROVISIONING PERIOD?

Natural circulating levels of prolactin in the ‘late’ provisioning period (assessed only for dominant females and subordinates, as dominant males were typically

203

Appendices implanted on the evening of day 8) were also found to be significantly affected

2 by class (χ 2 = 29.86, P < 0.001), with dominant females showing the highest levels of prolactin (mean ± S.E.: 3.35 ± 0.22 µg/ml) . Neither year nor social

2 group size affected prolactin levels in the ‘late’ provisioning period (χ 1 ≤ 0.81, P

≥ 0.37).Time of capture since sunset, time from capture to blood sample and

2 photoperiod did not affect circulating levels of prolactin (χ 1 ≤ 2.29, P ≥ 0.13)

(ascertained in a separate analysis). Repeating this analysis utilising both the

‘early’ and ‘late’ prolactin data combined (for dominant females and

2 subordinates only), it was found that class (χ 2 = 55.68, P < 0.001) significantly affected circulating prolactin levels. There was a borderline significant effect of

2 year (χ 1 = 3.50, P = 0.06). There was no significant difference between the

2 prolactin levels in either the ‘early’ or ‘late’ provisioning period (χ 1 = 2.24, P =

2 0.13). There was also no significant effect of brood size or social group size (χ 1

≤ 2.38, P ≥ 0.12). Again, time of capture since sunset, time from capture to

2 blood sample and photoperiod did not affect circulating levels of prolactin (χ 1 ≤

1.82, P ≥ 0.18).

A.4.4 INVESTIGATING THE EFFECT OF THE TIME ELAPSED SINCE SUNSET ON THE

PROLACTIN LEVELS OF DOMINANT FEMALES

For this analysis only, Class is defined as Dominant Female or Not Dominant

Female (i.e. dominant males, natal and immigrant subordinates). There was no

2 significant interaction between Class and time from sunset to capture (χ 1 =

0.54, P = 0.46) on circulating levels of prolactin. This suggests that the prolactin levels of individuals who were not roosting the nestlings upon capture for blood

204

Appendices

sampling (i.e. dominant males, natal and immigrant subordinates who roost

individually) did not decline as time away from the nestlings increased.

A.4.5 FULL TABLES FOR ALL MULTIFACTORIAL MODELS EMPLOYED

Table A4.2 Model outputs of mixed-effect models as detailed in the appropriate

analyses.

Model terms – fixed effects Estimate SE df P a) How do the different classes of bird differ in their circulating prolactin levels?

Intercept* 3.15 0.24

Class - DM -1.76 0.32 3 < 0.001

Class - Natal -1.63 0.27 3 < 0.001

Class - Imm. -2.44 0.49 3 < 0.001

Year 0.75 0.14 1 < 0.001

Social group size -0.041 0.096 1 0.61

Brood size 0.29 0.19 1 0.098

Photoperiod of bleed date -0.0048 0.004 1 0.25

6

Time from sunset to capture -0.0012 0.001 1 0.28

5

Lag from capture to blood sample 0.00027 0.002 1 0.84

1

Body mass (SMI) 0.021 0.19 1 0.62

(b) (i) How do the different classes of bird differ in their provisioning rate?

Intercept* 6.52 0.58

Class - DM -4.39 0.67 3 < 0.001

205

Appendices

Class - Natal -6.81 1.04 3 < 0.001

Class - Imm. -4.70 0.74 3 < 0.001

Year 1.09 0.51 1 0.029

Social group size 0.023 0.27 1 0.93

Brood size -0.011 0.47 1 0.98

Body mass (SMI) -0.11 0.11 1 0.26 b) (ii) How do the different classes of bird differ in their provisioning duration?

Intercept* 109.78 11.76

Class - DM -72.50 12.80 2 < 0.001

Class - Natal -75.95 16.01 2 < 0.001

Year -2.96 19.35 1 0.88

Social group size -6.61 8.05 1 0.40

Brood size -15.06 15.71 1 0.32

Body mass (SMI) 1.22 3.30 1 0.70 c) (i) Does individual variation in circulating prolactin levels predict provisioning rate?

Intercept* 6.49 0.57 1

Prolactin -0.24 0.25 1 0.34

Class - DM -4.49 0.65 3 < 0.001

Class - Natal -4.65 0.72 3 < 0.001

Class - Imm. -6.82 0.97 3 < 0.001

Year 1.14 0.48 1 0.022 c) (ii) Does individual variation in circulating prolactin levels predict provisioning duration?

Intercept* 110.60 11.78

Prolactin 5.64 5.96 1 0.33

206

Appendices

Class - DM -76.40 14.96 2 < 0.001

Class - Natal -79.49 17.14 2 < 0.001

Year -3.01 15.05 1 0.92

(d) (i) Does the within-individual change in prolactin levels within a breeding attempt predict the within-individual changes in provisioning rate?[with prior provisioning rate]

Intercept* 0.76 1.38

Within-individual change in prolactin -0.77 0.72 1 0.55

Provisioning rate 0.17 0. 33 1 0.57

Class - Natal -1.85 2.07 2 0.37

Class - Imm. 1.71 2.62 2 0.37 d) (ii) Does the within-individual change in prolactin levels within a breeding attempt predict the within-individual changes in provisioning rate?[with body condition]

Intercept* 0.71 2.17

Within-individual change in prolactin -1.55 0.89 1 0.30

Within-individual change in body -0.081 0.81 1 0.86 condition (SMI) e) (i) Does the VIP implant affect circulating prolactin levels?

Intercept* 1.30 0.49

Implant category × Pre or Post implant -0.72 0.60 1 0.29

Implant category 0.24 0.32 1 0.55

Pre or Post implant 0.43 0.30 1 0.15 e) (ii) Does the VIP implant affect change in circulating prolactin levels?

Intercept* -0.83 0.44

Implant category 0.72 0.60 1 0.21

207

Appendices e) (iii) Does the VIP implant affect provisioning behaviour?

Intercept* 4.20 0.79

Implant category × Pre or Post implant 0.33 0.94 1 0.71

Implant category -0.35 0.91 1 0.74

Pre or Post implant -0.95 0.46 1 0.041 e) (iv) Does the VIP implant affect change in provisioning behaviour?

Intercept* 1.39 0.72

Within-individual change in prolactin 0.31 0.35 1 0.34

Implant category -0.55 0.92 1 0.52

The Estimate and SE values are reported at the point where the fixed term was

either removed from the model (with P > 0.05) or retained (with P < 0.05) under

stepwise model simplification. *The intercept values were derived from the

minimum adequate model, with all other significant terms retained.

208

References

REFERENCES

Alonso-alvarez, C., Velando, A., 2012. Benefits and costs of parental care, in: Royle, N.J., Smiseth, P.T., Kölliker, M. (Eds.), The Evolution of Parental Care. Oxford University Press, Oxford, UK., pp. 40–54. doi:10.1093/acprof:oso/9780199692576.003.0003

Alonzo, S.H., 2010. Social and coevolutionary feedbacks between mating and parental investment. Trends Ecol. Evol. 25, 99–108. doi:10.1016/j.tree.2009.07.012

Andersson, M.B., 1994. Sexual selection. Princeton University Press, Princeton.

Angelier, F., Barbraud, C., Lormée, H., Prud’homme, F., Chastel, O., 2006. Kidnapping of chicks in emperor penguins: a hormonal by-product? J. Exp. Biol. 209, 1413–1420. doi:10.1242/jeb.02138

Angelier, F., Clement-Chastel, C., Welcker, J., Gabrielsen, G.W., Chastel, O., 2009. How does corticosterone affect parental behaviour and reproductive success? A study of prolactin in black-legged kittiwakes. Funct. Ecol. 23, 784–793. doi:10.1111/j.1365-2435.2009.01545.x

Angelier, F., Moe, B., Weimerskirch, H., Chastel, O., 2007. Age-specific reproductive success in a long-lived bird: Do older parents resist stress better? J. Anim. Ecol. 76, 1181–1191. doi:10.1111/j.1365- 2656.2007.01295.x

Arnold, K., Goldizen, A., Owens, I., 2005. Division of labour within cooperatively breeding groups. Behaviour 142, 1577–1590. doi:10.1163/156853905774831927

Badyaev, A. V., Duckworth, R.A., 2005. Evolution of plasticity in hormonally- integrated parental tactcis: An example with the house finch, in: Dawson, A., Sharp, P.J. (Eds.), Functional Avian Endocrinology. Narosa Publishing House, New Delhi, India, pp. 375–386.

Baldellou, M., Peter Henzi, S., 1992. Vigilance, predator detection and the presence of supernumerary males in vervet monkey troops. Anim. Behav. 43, 451–461. doi:10.1016/S0003-3472(05)80104-6

Barta, Z., Székely, T., Liker, A., Harrison, F., 2014. Social role specialization promotes cooperation between parents. Am. Nat. 183, 747–761. doi:10.1086/676014

Bednekoff, P.A., Woolfenden, G.E., 2003. Florida scrub-says (Aphelocoma coerulescens) are sentinels more when well-fed (even with no kin nearby). Ethology 109, 895–903.

Bell, M.B. V, Radford, A.N., Smith, R.A., Thompson, A.M., Ridley, A.R., 2010. Bargaining babblers: vocal negotiation of cooperative behaviour in a social bird. Proc. R. Soc. B Biol. Sci. 277, 3223–3228. doi:10.1098/rspb.2010.0643 209

References

Bell, M.B. V., Nichols, H.J., Gilchrist, J.S., Cant, M.A., Hodge, S.J., 2012. The cost of dominance: suppressing subordinate reproduction affects the reproductive success of dominant female banded mongooses. Proc. R. Soc. B Biol. Sci. 279, 619–624. doi:10.1098/rspb.2011.1093

Bergmüller, R., Heg, D., Taborsky, M., 2005. Helpers in a cooperatively breeding cichlid stay and pay or disperse and breed, depending on ecological constraints. Proc. R. Soc. B Biol. Sci. 272, 325–331. doi:10.1098/rspb.2004.2960

Bergmüller, R., Schürch, R., Hamilton, I.M., 2010. Evolutionary causes and consequences of consistent individual variation in cooperative behaviour. Philos. Trans. R. Soc. B Biol. Sci. 365, 2751–2764. doi:10.1098/rstb.2010.0124

Bergmüller, R., Taborsky, M., 2005. Experimental manipulation of helping in a cooperative breeder: Helpers “pay to stay” by pre-emptive appeasement. Anim. Behav. 69, 19–28. doi:10.1016/j.anbehav.2004.05.009

Beynon, P., Rasa, O., 1989. Do dwarf mongooses have a language? Warning vocalizations transmit complex information. S. Afr. J. Sci. 85, 447–450.

Bize, P., Piault, R., Moureau, B., Heeb, P., 2006. A UV signal of offspring condition mediates context-dependent parental favouritism. Proc. R. Soc. B Biol. Sci. 273, 2063–2068. doi:10.1098/rspb.2006.3546

Boland, C.R.J., Heinsohn, R., Cockburn, A., 1997. Experimental manipulation of brood reduction and parental care in cooperatively breeding white-winged choughs. J. Anim. Ecol. 66, 683–691. doi:10.2307/5921

Boomsma, J.J., 2007. Kin selection versus sexual selection: why the ends do not meet. Curr. Biol. 17, 673–683. doi:10.1016/j.cub.2007.06.033

Boos, M., Zimmer, C., Carriere, A., Robin, J.P., Petit, O., 2007. Post-hatching parental care behaviour and hormonal status in a precocial bird. Behav. Processes 76, 206–214. doi:10.1016/j.beproc.2007.05.003

Brotherton, P.N.M., Clutton-Brock, T.H., Riain, M.J.O., Gaynor, D., Sharpe, L., Kansky, R., 2001. Offspring food allocation by parents and helpers in a cooperative mammal. Behav. Ecol. 12, 590–599.

Browning, L.E., Patrick, S.C., Rollins, L. a., Griffith, S.C., Russell, a. F., 2012. Kin selection, not group augmentation, predicts helping in an obligate cooperatively breeding bird. Proc. R. Soc. B Biol. Sci. 279, 3861–3869. doi:10.1098/rspb.2012.1080

Buntin, J.D., 1996. Neural and hormonal control of parental behaviour in birds, in: Rosenblatt, J.S., Snowdon, C.. (Eds.), Advances in the Study of Behavior Vol. 25. Academic Press, New York, pp. 161–213.

Buntin, J.D., Ruzycki, E., Witebsky, J., 1993. Prolactin receptors in dove brain: autoradiographic analysis of binding characteristics in discrete brain regions and accessibility to blood-borne prolactin. Neuroendocrinology 57, 738–750. 210

References

Canestrari, D., Marcos, J.M., Baglione, V., 2007. Costs of chick provisioning in cooperatively breeding crows: an experimental study. Anim. Behav. 73, 349–357. doi:10.1016/j.anbehav.2006.04.013

Cant, M.A., Vitikainen, E., Nichols, H.J., 2013. Demography and social evolution of banded mongooses, in: Advances in the Study of Behavior. pp. 407– 445. doi:10.1016/B978-0-12-407186-5.00006-9

Carlisle, T.R., Zahavi, A., 1986. Helping at the nest, allofeeding and social status in immature arabian babblers. Behav. Ecol. Sociobiol. 18, 339–351. doi:10.1007/BF00299665

Carlson, A.., Manser, M.B., Young, A.J., Russell, A.F., Jordan, N.R., McNeilly, A.S., Clutton-Brock, T.H., 2006a. Cortisol levels are positively associated with pup-feeding rates in male meerkats. Proc. R. Soc. B Biol. Sci. 273, 571–577. doi:10.1098/rspb.2005.3087

Carlson, A.., Russell, A.F., Young, A.J., Jordan, N.R., McNeilly, A.S., Parlow, A.F., Clutton-Brock, T.H., 2006b. Elevated prolactin levels immediately precede decisions to babysit by male meerkat helpers. Horm. Behav. 50, 94–100. doi:10.1016/j.yhbeh.2006.01.009

Carlson, A.A., Nicol, L., Young, A.J., Parlow, A.F., McNeilly, A.S., 2003. Radioimmunoassay of prolactin for the meerkat (Suricata suricatta), a cooperatively breeding carnivore. Gen. Comp. Endocrinol. 130, 148–156. doi:doi:10.1016/S0016-6480(02)00610-X

Chase, I.D., 1980. Cooperative and noncooperative behavior in animals. Am. Nat. 115, 827–857.

Chastel, O., Gabrielsen, G.W., Maho, Y. Le, 2002. Factors affecting plasma concentrations of prolactin in the common eider Somateria mollissima Franc. Gen. Comp. Endocrinol. 409, 399–409. doi:10.1006/gcen.2001.7767

Chastel, O., Lacroix, A., Weimerskirch, H., Gabrielsen, G.W., 2005. Modulation of prolactin but not corticosterone responses to stress in relation to parental effort in a long-lived bird. Horm. Behav. 47, 459–466. doi:10.1016/j.yhbeh.2004.10.009

Chastel, O., Lormée, H., 2002. Patterns of prolactin secretion in relation to incubation failure in a tropical seabird, the red-footed booby. Condor 104, 873–876.

Christensen, D., Vleck, C.M., 2008. Prolactin release and response to vasoactive intestinal peptide in an opportunistic breeder, the zebra finch (Taeniopygia guttata). Gen. Comp. Endocrinol. 157, 91–98. doi:10.1016/j.ygcen.2008.04.013

Clarke, M.F., 1989. The pattern of helping in the bell miner (Manorina melanophrys). Ethology 80, 292–306.

Clutton-Brock, T.H., 2002. Breeding together: kin selection and mutualism in cooperative vertebrates. Science. 296, 69–72.

211

References

Clutton-Brock, T.H., 1999. Selfish sentinels in cooperative mammals. Science. 284, 1640–1644. doi:10.1126/science.284.5420.1640

Clutton-Brock, T.H., 1991. The Evolution of Parental Care. Princeton University Press, Princeton, New Jersey.

Clutton-Brock, T.H., Brotherton, P.N., O’Riain, M.J., Griffin, A.S., Gaynor, D., Sharpe, L., Kansky, R., Manser, M.B., McIlrath, G.M., 2000. Individual contributions to babysitting in a cooperative mongoose, Suricata suricatta. Proc. Biol. Sci. 267, 301–305. doi:10.1098/rspb.2000.1000

Clutton-Brock, T.H., Brotherton, P.N.M., O’Riain, M.J., Griffin, A.S., Gaynor, D., Kansky, R., Sharpe, L., McIlrath, G.M., 2001. Contributions to cooperative rearing in meerkats. Anim. Behav. 61, 705–710. doi:10.1006/anbe.2000.1631

Clutton-Brock, T.H., Gaynor, D., Kansky, R., Maccoll, A.D.C., Mcilrath, G., Chadwick, P., Brotherton, P.N.M., O’Riain, J.M., Manser, M.B., Skinner, J.D., 1998. Costs of cooperative behaviour in suricates (Suricata suricatta). Proc. R. Soc. B Biol. Sci. 265, 185–190.

Clutton-Brock, T.H., Hodge, S.J., Spong, G., Russell, A.F., Jordan, N.R., Bennett, N.C., Sharpe, L.L., Manser, M.B., 2006. Intrasexual competition and sexual selection in cooperative mammals. Nature 444, 1065–1068. doi:10.1038/nature05386

Clutton-Brock, T.H., Russell, A.F., Sharpe, L.L., 2004. Behavioural tactics of breeders in cooperative meerkats. Anim. Behav. 68, 1029–1040. doi:10.1016/j.anbehav.2003.10.024

Clutton-Brock, T.H., Russell, A.F., Sharpe, L.L., Young, A.J., Balmforth, Z., McIlrath, G.M., 2002. Evolution and development of sex differences in cooperative behavior in meerkats. Science. 297, 253–256. doi:10.1126/science.1071412

Cockburn, A., 2006. Prevalence of different modes of parental care in birds. Proc. R. Soc. B Biol. Sci. 273, 1375–1383. doi:10.1098/rspb.2005.3458

Cockburn, A., 1998. Evolution of helping in cooperatively breeding birds. Annu. Rev. Ecol. Syst. 29, 141–177.

Collias, E., Collias, E.C., 1978. Cooperative breeding behaviour in the white- browed sparrow weaver. Auk 95, 472–484.

Cornwallis, C.K., West, S.A., Davis, K.E., Griffin, A.S., 2010. Promiscuity and the evolutionary transition to complex societies. Nature 466, 969–972. doi:10.1038/nature09335

Cornwallis, C.K., West, S.A., Griffin, A.S., 2009. Routes to indirect fitness in cooperatively breeding vertebrates: kin discrimination and limited dispersal. J. Evol. Biol. 22, 2445–2457. doi:10.1111/j.1420-9101.2009.01853.x

Cram, D.L., Blount, J.D., Young, A.J., 2015. Oxidative status and social dominance in a wild cooperative breeder. Funct. Ecol. doi:10.1111/1365- 212

References

2435.12317

Criscuolo, F., Chastel, O., Bertile, F., Gabrielsen, G.W., Le Maho, Y., Raclot, T., 2005. Corticosterone alone does not trigger a short term behavioural shift in incubating female common eiders Somateria mollissima, but does modify long term reproductive success. J. Avian Biol. 36, 306–312. doi:10.1111/j.0908-8857.2005.03371.x

Davies, N.B., Hatchwell, B.J., 1992. The value of male parental care and its influence on reproductive allocation by male and female dunnocks. J. Anim. Ecol. 61, 259–272.

Davis, J.N., Todd, P.M., Bullock, S., 1999. Environment quality predicts parental provisioning decisions. Proc. R. Soc. B Biol. Sci. 266, 1791–1797. doi:10.1098/rspb.1999.0848

Dawson, A., Goldsmith, A.R., 1982. Prolactin and gonadotrophin secretion in wild starlings (Sturnus vulgaris) during the annual cycle and in relation to nesting, incubation, and rearing young. Gen. Comp. Endocrinol. 48, 213– 221. doi:10.1016/0016-6480(82)90019-3

Delehanty, D.J., Oring, J.W., Fivizzani, A.J., El Halawani, M.E., 1997. Circulating prolactin of incubating male Wilson’s phalaropes corresponds to clutch size and environmental stress. Condor 99, 397–405.

Dickens, M., Hartley, I.R., 2007. Differences in parental food allocation rules: evidence for sexual conflict in the blue tit? Behav. Ecol. 18, 674–679. doi:10.1093/beheco/arm029

Dickinson, J.L., Hatchwell, B.J., 2004. Fitness consequences of helping, in: Koenig, W.D., Dickinson, J.L. (Eds.), Ecology and Evolution of Cooperative Breeding in Birds. Cambridge University Press, pp. 48–66. doi:dx.doi.org/10.1017/CBO9780511606816.004

Doutrelant, C., Covas, R., 2007. Helping has signalling characteristics in a cooperatively breeding bird. Anim. Behav. 74, 739–747. doi:10.1016/j.anbehav.2006.11.033

Duckworth, R.A., Badyaev, A. V., Parlow, A.F., 2003. Elaborately ornamented males avoid costly parental care in the house finch (Carpodacus mexicanus): a proximate perspective. Behav. Ecol. Sociobiol. 55, 176–183. doi:10.1007/s00265-003-0671-7

Duffy, J.E., Macdonald, K.S., 2010. Kin structure, ecology and the evolution of social organization in shrimp: a comparative analysis. Proc. R. Soc. B Biol. Sci. 277, 575–584. doi:10.1098/rspb.2009.1483

DuVal, E.H., 2013. Does cooperation increase helpers’ later success as breeders? A test of the skills hypothesis in the cooperatively displaying lance-tailed manakin. J. Anim. Ecol. 82, 884–893. doi:10.1111/1365- 2656.12057

El Halawani, M.E., Burke, W.H., Millam, J.R., Fehrer, S.C., Hargis, B.M., 1984. Regulation of prolactin and its role in gallinaceous bird reproduction. J. Exp. 213

References

Zool. 232, 521–529. doi:10.1002/jez.1402320319

Engel, C., 2011. Dictator games: a meta study. Exp. Econ. 14, 583–610. doi:10.1007/s10683-011-9283-7

Farrelly, D., 2013. Altruism as an indicator of good parenting quality in long term relationships: further investigations using the mate preferences towards altruistic traits scale. J. Soc. Psychol. 153, 395–398. doi:10.1080/00224545.2013.768595

Ferguson, J.W.H., 1988. Comparison of communication and signalling patterns of white-browed sparrow weavers and other gregarious ploceid weavers. Ostrich 59, 54–62.

Ferguson, J.W.H., 1987. Vigilance behaviour in white-browed sparrow-weavers Plocepasser mahali. Ethology 76, 223–235.

Field, J., Cant, M.A., 2009. Social stability and helping in small animal societies. Philos. Trans. R. Soc. B Biol. Sci. 364, 3181–3189. doi:10.1098/rstb.2009.0110

Fischer, S., Zöttl, M., Groenewoud, F., Taborsky, B., 2014. Group-size- dependent punishment of idle subordinates in a cooperative breeder where helpers pay to stay. Proc. R. Soc. B Biol. Sci. 281, 1–9. doi:10.1098/rspb.2014.0184

Fisher, R.A., 1930. The genetical theory of natural selection. Clarendon Press, Oxford.

Freeman, M.E., Kanyicska, B., Lerant, A., Nagy, G., 2000. Prolactin: structure, function, and regulation of secretion. Physiol. Rev. 80, 1524–1585. doi:10.1016/B978-012515400-0/50037-3

Garcia-Navas, V., Sanz, J.J., 2012. Environmental and within-nest factors influencing nestling-feeding patterns of mediterranean blue tits (Cyanistes caeruleus). Condor 114, 612–621. doi:10.1525/cond.2012.110120

Gaston, A.J., 1978. The evolution of group territorial behavior and cooperative breeding. Am. Nat. 112, 1091–1100. doi:10.2307/2678832

Gelman, A., Carlin, J.B., Stern, H.S., Rubin, D.B., 2004. Bayesian data analysis. Chapman and Hall, Florida, Texts in Statistical Science.

Gelman, A., Gelman, A., Meng, X.., Meng, X.., Stern, H., Stern, H., 1996. Posterior predictive assessment of model fitness via realized discrepancies. Vol.6, No.4. Stat. Sin. 6, 733–807. doi:10.1.1.142.9951

Gottlander, K., 1987. Parental feeding behaviour and sibling competition in the pied flycatcher Ficedula hypoleuca. Ornis Scand. 18, 269–276. doi:10.2307/3676895

Grafent, A., 1990. Biological signals as handicaps. J. Theor. Biol. 144, 517–546.

Griffin, A.S., West, S.A., 2003. Kin discrimination and the benefit breeding 214

References

vertebrates. Science. 302, 634–636.

Hailman, J.P., McGowan, K.J., Woolfenden, G.E., 1994. Role of Helpers in the Sentinel Behaviour of the Florida Scrub Jay (Aphelocoma c. coerulescens). Ethology 97, 119–140. doi:10.1111/j.1439-0310.1994.tb01034.x

Hall, M.R., 1987. External stimuli affecting incubation behavior and prolactin secretion in the duck (Anas platyrhynchos). Horm. Behav. 21, 269–287.

Hall, T.R., Harvey, S., Chadwick, A., 1986. Control of prolactin secretion in birds: a review. Gen. Comp. Endocrinol. 62, 171–184. doi:http://dx.doi.org/10.1016/0016-6480(86)90107-3

Hamilton, I.M., Taborsky, M., 2005. Unrelated helpers will not fully compensate for costs imposed on breeders when they pay to stay. Proc. R. Soc. B Biol. Sci. 272, 445–454. doi:10.1098/rspb.2004.2961

Hamilton, W.D., 1964. The genetical evolution of social behaviour. I. J. Am. Stat. Assoc. 7, 1–16.

Harrison, F., Barta, Z., Cuthill, I., Székely, T., 2009. How is sexual conflict over parental care resolved? A meta-analysis. J. Evol. Biol. 22, 1800–1812. doi:10.1111/j.1420-9101.2009.01792.x

Harrison, X.A., York, J.E., Cram, D.L., Hares, M.C., Young, A.J., 2013a. Complete reproductive skew within white-browed sparrow weaver groups despite outbreeding opportunities for subordinates of both sexes. Behav. Ecol. Sociobiol. 67, 1915–1929. doi:10.1007/s00265-013-1599-1

Harrison, X.A., York, J.E., Cram, D.L., Young, A.J., 2013b. Extra-group mating increases inbreeding risk in a cooperatively breeding bird. Mol. Ecol. 22, 5700–5715. doi:10.1111/mec.12505

Harrison, X.A., York, J.E., Young, A.J., 2014. Population genetic structure and direct observations reveal sex-reversed patterns of dispersal in a cooperative bird. Mol. Ecol. 5740–5755. doi:10.1111/mec.12978

Hatchwell, B.J., 2009. The evolution of cooperative breeding in birds: kinship, dispersal and life history. Philos. Trans. R. Soc. B Biol. Sci. 364, 3217– 3227. doi:10.1098/rstb.2009.0109

Hatchwell, B.J., Russell, A.F., 1996. Provisioning rules in cooperatively breeding long-tailed tits Aegithalos caudatus: an experimental study. Proc. R. Soc. B Biol. Sci. 263, 83–88.

Heg, D., Bachar, Z., Brouwer, L., Taborsky, M., 2004. Predation risk is an ecological constraint for helper dispersal in a cooperatively breeding cichlid. Proc. R. Soc. B Biol. Sci. 271, 2367–2374. doi:10.1098/rspb.2004.2855

Heg, D., Taborsky, M., 2010. Helper response to experimentally manipulated predation risk in the cooperatively breeding cichlid Neolamprologus pulcher. PLoS One 5, 1–11. doi:10.1371/journal.pone.0010784

Heinsohn, R., Legge, S., 1999. The cost of helping. Trends Ecol. Evol. 14, 53– 215

References

57.

Hinde, C.A., 2006. Negotiation over offspring care ?— a positive response to partner-provisioning rate in great tits. Behav. Ecol. 17, 6–12. doi:10.1093/beheco/ari092

Hoff, P.D., 2005. Bilinear Mixed-Effects Models for Dyadic Data. J. Am. Stat. Assoc. 100, 286–295. doi:10.1198/016214504000001015

Hothorn, T., Bretz, F., Westfall, P., 2008. Simultaneous Inference in General Parametric Models. Biometrical J. 50, 346–363.

Houston, A.I., Davies, N.B., 1985. The evolution of cooperation and life history in the dunnock, Prunella modularis, in: Sibly, R., Smith, R.. (Eds.), Behavioural Ecology: Ecological Consequences of Adaptive Behaviour. Oxford: Blackwell Scientific, pp. 471–487.

Iredale, W., Vugt, M. Van, Dunbar, R., 2008. Showing off in humans: male generosity as a mating signal. Evol. Psychol. 6, 386–392.

Iserbyt, A., Farrell, S., Eens, M., Müller, W., 2014. Sex-specific negotiation rules in a costly conflict over parental care. Proc. R. Soc. B Biol. Sci. 100, submitted. doi:10.1016/j.anbehav.2014.11.014

Jetz, W., Sekercioglu, C.H., Bohning-Gase, K., 2008. The worldwise variation in avian clutch size across species and space. PLoS Biol. 6, 2650–2657. doi:10.1371/journal.Citation

Johnstone, R.A., 2011. Load lightening and negotiation over offspring care in cooperative breeders. Behav. Ecol. 22, 436–444. doi:10.1093/beheco/arq190

Kacelnik, A., Cotton, P.A., Stirling, L., Wright, J., 1995. Food allocation among nestling starlings: sibling competition and scope of parental choice. Proc. R. Soc. B Biol. Sci. 259, 259–263.

Keane, B., Waser, P.M., Creel, S.R., Creel, N.M., Elliott, L.F., Minchella, D.J., 1994. Subordinate reproduction in dwarf mongooses. Anim. Behav. 47, 65– 75.

Keller, L., Reeve, H.K., 1994. Partitioning of reproduction in animal societies. Trends Ecol. Evol. 9, 98–102.

Kern, J.M., Radford, A.N., 2014. Sentinel dwarf mongooses, Helogale parvula, exhibit flexible decision making in relation to predation risk. Anim. Behav. 98, 185–192. doi:10.1016/j.anbehav.2014.10.012

Khan, M.Z., McNabb, F.M., Walters, J.R., Sharp, P.J., 2001. Patterns of testosterone and prolactin concentrations and reproductive behavior of helpers and breeders in the cooperatively breeding red-cockaded woodpecker (Picoides borealis). Horm. Behav. 40, 1–13. doi:10.1006/hbeh.2001.1658

Kilner, R., 2002. Sex differences in canary (Serinus canaria) provisioning rules. 216

References

Behav. Ecol. Sociobiol. 52, 400–407. doi:10.1007/s00265-002-0533-8

Kilner, R., 1995. When do canary parents respond to nestling signals of need? Proc. R. Soc. B Biol. Sci. 260, 343–348. doi:10.1098/rspb.1995.0102

Klug, H., Alonzo, S.H., Bonsall, M.B., 2012. Theoretical foundations of parental care, in: Royle, N.J., Smiseth, P.T., Kölliker, M. (Eds.), The Evolution of Parental Care. Oxford University Press, Oxford, UK., pp. 21–36.

Klug, H., Bonsall, M.B., Alonzo, S.H., 2013a. Sex differences in life history drive evolutionary transitions among maternal, paternal, and bi-parental care. Ecol. Evol. 3, 792–806. doi:10.1002/ece3.494

Klug, H., Bonsall, M.B., Alonzo, S.H., 2013b. The origin of parental care in relation to male and female life history. Ecol. Evol. 3, 779–791. doi:10.1002/ece3.493

Koenig, W., Haydock, J., 2004. Incest and incest avoidance, in: Koenig, W., Dickinson, J.L. (Eds.), Ecology and Evolution of Cooperative Breeding in Birds. Cambridge University Press, Cambridge, pp. 142–156.

Koenig, W.D., Walters, E.L., 2011. Age-related provisioning behaviour in the cooperatively breeding acorn woodpecker: Testing the skills and the pay- to-stay hypotheses. Anim. Behav. 82, 437–444. doi:10.1016/j.anbehav.2011.05.028

Kokko, H., Jennions, M.D., 2008. Parental investment, sexual selection and sex ratios. J. Evol. Biol. 21, 919–948. doi:10.1111/j.1420-9101.2008.01540.x

Kokko, H., Johnstone, R.A., Clutton-Brock, T.H., 2001. The evolution of cooperative breeding through group augmentation. Proc. R. Soc. B Biol. Sci. 268, 187–196. doi:10.1098/rspb.2000.1349

Kokko, H., Johnstone, R.A., Wright, J., 2002. The evolution of parental and alloparental effort in cooperatively breeding groups: when should helpers pay to stay? Behav. Ecol. 13, 291–300. doi:10.1093/beheco/13.3.291

Kolliker, M., Richner, H., Werner, I., Heeb, P., 1998. Begging signals and biparental care: nestling choice between parental feeding locations. Anim. Behav. 55, 215–22. doi:10.1006/anbe.1997.0571

Komdeur, J., 2006. Variation in individual investment strategies among social animals. Ethology 112, 729–747. doi:10.1111/j.1439-0310.2006.01243.x

Komdeur, J., 1994. The Effect of Kinship on Helping in the Cooperative Breeding Seychelles Warbler (Acrocephalus sechellensis). Proc. R. Soc. B Biol. Sci. 256, 47–52. doi:10.1098/rspb.1994.0047

Krebs, E.A., Cunningham, R., Donnelly, C.F., 1999. Complex patterns of food allocation in asynchronously hatching broods of crimson rosellas. Anim. Behav. 57, 753–763. doi:10.1006/anbe.1998.1029

Lea, R.W., Sharp, P.J., 1991. Effects of presence of squabs upon plasma concentrations of prolactin and LH and length of time of incubation in 217

References

ringdoves on “extended” incubatory patterns. Horm.Behav. 25, 275–282.

Leitner, S., Mundy, P., Voigt, C., 2009. Morphometrics of white-browed sparrow-weavers Plocepasser mahali in south western Zimbabwe. Ostrich 80, 99–102.

Leonard, M., Horn, A., 1996. Provisioning rules in tree swallows. Behav. Ecol. Sociobiol. 38, 341–347. doi:10.1007/s002650050250

Lessells, C.M., 2012. Sexual conflict, in: Royle, N.J., Smiseth, P.T., Kölliker, M. (Eds.), The Evolution of Parental Care. Oxford University Press, Oxford, UK., pp. 150–169.

Lessells, C.M., 2002. Parentally biased favouritism: why should parents specialize in caring for different offspring? Philos. Trans. R. Soc. B Biol. Sci. 357, 381–403. doi:10.1098/rstb.2001.0928

Lessells, C.M., Boag, P.T., 1987. Unrepeatable repeatabilities: a common mistake. Auk 104, 116–121.

Lewis, D.M., 1982. Dispersal in a population of white-browed sparrow weavers. Condor 84, 306. doi:10.2307/1367373

Lewis, D.M., 1982. Cooperative breeding in a population of white-browed sparrow weavers Plocepasser mahali. Ibis (Lond. 1859). 124, 511–522. doi:10.2307/1367373

Lotem, A., Fishman, M.A., Stone, L., 2003. From reciprocity to unconditional altruism through signalling benefits. Proc. R. Soc. B Biol. Sci. 270, 199– 205. doi:10.1098/rspb.2002.2225

Lukas, D., Clutton-Brock, T.H., 2012. Cooperative breeding and monogamy in mammalian societies. Proc. R. Soc. B Biol. Sci. 279, 2151–2156. doi:10.1098/rspb.2011.2468

Lunn, D.J., Thomas, A., Best, N., Spiegelhalter, D., 2000. WinBUGS – A Bayesian modelling framework: Concepts, structure, and extensibility. Stat. Comput. 10, 325–337. doi:10.1023/A:1008929526011

Mand, R., Rasmann, E., Magi, M., 2013. When a male changes his ways: sex differences in feeding behavior in the pied flycatcher. Behav. Ecol. 24, 853– 858. doi:10.1093/beheco/art025

Manser, M.B., 1999. Response of foraging group members to sentinel calls in suricates, Suricata suricatta. Proc. R. Soc. B Biol. Sci. 266, 1013 – 1019. doi:10.1098/rspb.1999.0737

Mares, R., Young, A.J., Clutton-Brock, T.H., 2012. Individual contributions to territory defence in a cooperative breeder: weighing up the benefits and costs. Proc. R. Soc. B Biol. Sci. 279, 3989–3995. doi:10.1098/rspb.2012.1071

Marshall-Ball, L., Mann, N., Slater, P.J.B., 2006. Multiple functions to duet singing: hidden conflicts and apparent cooperation. Anim. Behav. 71, 823– 218

References

831. doi:10.1016/j.anbehav.2005.05.021

Martin, T.E., 2015. Age-related mortality explains life history strategies of tropical and temperate songbirds. Science. 349, 966–970.

Martin, T.E., 1995. Avian life history evolution in relation to nest sites, nest predation, and food. Ecol. Monogr. 65, 101–127.

Martin, T.E., Lloyd, P., Bosque, C., Barton, D.C., Biancucci, A.L., Cheng, Y.-R., Ton, R., 2011. Growth rate variation among passerine species in tropical and temperate sites: an antagonistic interaction between parental food provisioning and nest predation risk. Evolution. 65, 1607–1622. doi:10.1111/j.1558-5646.2011.01227.x

Mas, A., Moretti, E., 2009. Peers at work. Am. Econ. Rev. 99, 112–145. doi:10.1257/aer.99.1.112

Maynard Smith, J., 1964. and kin selection. Nature 201, 1145– 1147.

McCarty, J.P., 2001. Variation in growth of nestling tree swallows across multiple temporal and spatial scales. Auk 118, 176–190.

McDonald, P.G., Kazem, A.J.N., Clarke, M.F., Wright, J., 2008a. Helping as a signal: Does removal of potential audiences alter helper behavior in the bell miner? Behav. Ecol. 19, 1047–1055. doi:10.1093/beheco/arn062

McDonald, P.G., te Marvelde, L., Kazem, A.J.N., Wright, J., 2008b. Helping as a signal and the effect of a potential audience during provisioning visits in a cooperative bird. Anim. Behav. 75, 1319–1330. doi:10.1016/j.anbehav.2007.09.005

McDonald, P.G., Wright, J., 2011. Bell miner provisioning calls are more similar among relatives and are used by helpers at the nest to bias their effort towards kin. Proc. R. Soc. B Biol. Sci. 278, 3403–3411. doi:10.1098/rspb.2011.0307

McGowan, K.J., Woolfenden, G.E., 1989. A sentinel system in the Florida scrub jay. Anim. Behav. 37, 1000–1006. doi:10.1016/0003-3472(89)90144-9

Møller, A.P., Cuervo, J.J., 2000. The evolution of paternity and paternal care in birds. Behav. Ecol. 11, 472–485. doi:10.1093/beheco/11.5.472

Mulder, R.A., Langmore, N.E., 1993. Dominant males punish helpers for temporary defection in superb fairy-wrens. Anim. Behav. 45, 830–833.

Nam, K.B., Simeoni, M., Sharp, S.P., Hatchwell, B.J., 2010. Kinship affects investment by helpers in a cooperatively breeding bird. Proc. R. Soc. B Biol. Sci. 277, 3299–3306. doi:10.1098/rspb.2010.0737

Nomano, F.Y., Browning, L.E., Rollins, L.A., Nakagawa, S., Griffith, S.C., Russell, A.F., 2013. Feeding nestlings does not function as a signal of social prestige in cooperatively breeding chestnut-crowned babblers. Anim. Behav. 86, 277–289. doi:10.1016/j.anbehav.2013.05.015 219

References

Nomano, F.Y., Browning, L.E., Savage, J.L., Rollins, L.A., Griffith, S.C., Russell, A.F., 2015. Unrelated helpers neither signal contributions nor suffer retribution in chestnut-crowed babblers. Behav. Ecol. 26, 1–10. doi:10.1093/beheco/arv023

Ohkubo, T., Tanaka, M., Nakashima, K., Sharp, P.J., 1998a. Relationship between prolactin receptor mRNA in the anterior pituitary gland and hypothalamus and reproductive state in male and female bantams (Gallus domesticus). Gen. Comp. Endocrinol. 111, 167–76. doi:10.1006/gcen.1998.7099

Ohkubo, T., Tanaka, M., Nakashima, K., Talbot, R.T., Sharp, P.J., 1998b. Prolactin receptor gene expression in the brain and peripheral tissues in broody and nonbroody breeds of domestic hen. Gen. Comp. Endocrinol. 109, 60–8. doi:10.1006/gcen.1997.7008

Ouyang, J.Q., Sharp, P.J., Dawson, A., Quetting, M., Hau, M., 2011. Hormone levels predict individual differences in reproductive success in a passerine bird. Proc. R. Soc. B Biol. Sci. 278, 2537–45. doi:10.1098/rspb.2010.2490

Parker, G.A., Royle, N.J., Hartley, I.R., 2002. Intrafamilial conflict and parental investment: a synthesis. Philos. Trans. R. Soc. B Biol. Sci. 357, 295–307. doi:10.1098/rstb.2001.0950

Peig, J., Green, A.J., 2009. New perspectives for estimating body condition from mass/length data: The scaled mass index as an alternative method. Oikos 118, 1883–1891. doi:10.1111/j.1600-0706.2009.17643.x

Phillips, T., 2015. Human altruism and cooperation explainable as adaptations to past environments no longer fully evident in the modern world. Q. Rev. Biol. 90, 295–314. doi:10.1086/682589

Phillips, T., Barnard, C., Ferguson, E., Reader, T., 2008. Do humans prefer altruistic mates? Testing a link between sexual selection and altruism towards non-relatives. Br. J. Psychol. 99, 555–572. doi:10.1348/000712608X298467

Pitts, G.R., You, S., Foster, D.N., El-Halawani, M.E., 2000. Evidence for multiple prolactin receptor transcripts in the turkey. Poult Sci 79, 355–62.

Queller, D.C., Strassmann, J.E., 1998. Kin selection and social insects. Bioscience 48, 165–175.

Qvarnstroöm, A., 1997. Experimentally increased badge size increases male competition and reduces male parental care in the collared flycatcher. Proc. R. Soc. B Biol. Sci. 264, 1225–1231. doi:10.1098/rspb.1997.0169

R Development Core Team, 2015. R: A language and environment for statistical computing. R Foundation for Statistical Computing.

R Development Core Team, 2014. R: A Language and Environment for Statistical Computing. R Found. Stat. Comput.

Radford, A.N., Bell, M.B. V, Hollén, L.I., Ridley, A.R., 2011. Singing for your 220

References

supper: sentinel calling by kleptoparasites can mitigate the cost to victims. Evolution. 65, 900–906. doi:10.1111/j.1558-5646.2010.01180.x

Raihani, N.J., Nelson-Flower, M.J., Moyes, K., Browning, L.E., Ridley, A.R., 2010. Synchronous provisioning increases brood survival in cooperatively breeding pied babblers. J. Anim. Ecol. 79, 44–52. doi:10.1111/j.1365- 2656.2009.01606.x

Raihani, N.J., Smith, S., 2015. Competitive helping in online giving. Curr. Biol. 25, 1183–1186. doi:10.1016/j.cub.2015.02.042

Rasa, O., 1986. Coordinated vigilance in dwarf mongoose family groups: The “Watchman’s Song” hypothesis and the costs of guarding. Ethology 71, 340–344. doi:10.1111/j.1439-0310.1986.tb00598.x

Remes, V., Martin, T.E., 2002. Environmental influences on the evolution of growth and developmental rates in . Evolution. 56, 2505–2518. doi:10.1111/j.0014-3820.2002.tb00175.x

Reyer, H.U., 1990. Pied Kingfishers: ecological causes and reproductive consequences of cooperative breeding, in: Stacey, P., Koenig, W. (Eds.), Cooperative Breeding in Birds: Long-Term Studies of Ecology and Behavior. Cambridge University Press, pp. 529–557.

Reyer, H.U., 1986. Breeder-helper-interactions in the pied kingfisher reflect the costs and benefits of cooperative breeding. Behaviour 96, 277–302. doi:10.1163/156853986X00522

Richardson, D.S., Jury, F.L., Blaakmeer, K., Komdeur, J., Burke, T., 2001. Parentage assignment and extra-group paternity in a cooperative breeder: The Seychelles warbler (Acrocephalus sechellensis). Mol. Ecol. 10, 2263– 2273. doi:10.1046/j.0962-1083.2001.01355.x

Ridley, A.R., Nelson-Flower, M.J., Thompson, A.M., 2013. Is sentinel behaviour safe? An experimental investigation. Anim. Behav. 85, 137–142. doi:10.1016/j.anbehav.2012.10.017

Ridley, A.R., Raihani, N.J., Bell, M.B. V, 2010. Experimental evidence that sentinel behaviour is affected by risk. Biol. Lett. 6, 445–448. doi:10.1098/rsbl.2010.0023

Riehl, C., 2013. Evolutionary routes to non-kin cooperative breeding in birds. Proc. R. Soc. B Biol. Sci. 280, 1–7.

Rollins, L.., Browning, L.E., Holleley, C.., Savage, J.L., Russell, A.F., Griffith, S.C., 2012. Building genetic networks using relatedness information: a novel approach for the estimation of dispersal and characterization of group structure in social animals. Mol. Ecol. 21, 1727–1740. doi:10.1111/j.1365- 294X.2012.05492.x

Rosas-Hernandez, H., Ramirez, M., Ramirez-Lee, M.A., Ali, S.F., Gonzalez, C., 2015. Inhibition of prolactin with bromocriptine for 28 days increases blood- brain barrier permeability in the rat. Neuroscience 301, 61–70. doi:10.1016/j.neuroscience.2015.05.066 221

References

Royle, N.J., Hartley, I.R., Parker, G.A., 2002. Begging for control: when are offspring solicitation behaviours honest? Trends Ecol. Evol. 17, 434–440. doi:10.1016/S0169-5347(02)02565-X

Royle, N.J., Smiseth, P.T., Kölliker, M., (eds), 2012. The evolution of parental care. Oxford University Press.

Russell, A.F., 2004. Mammals: comparisons and contrasts, in: Koenig, W.D., Dickinson, J.L. (Eds.), Ecology and Evolution of Cooperative Breeding in Birds. Cambridge University Press, pp. 210– 227.

Russell, A.F., Hatchwell, B.J., 2001. Experimental evidence for kin-biased helping in a cooperatively breeding vertebrate. Proc. R. Soc. B Biol. Sci. 268, 2169–2174. doi:10.1098/rspb.2001.1790

Russell, A.F., Sharpe, L.L., Brotherton, P.N.M., Clutton-Brock, T.H., 2003. Cost minimization by helpers in cooperative vertebrates. Proc. Natl. Acad. Sci. U. S. A. 100, 3333–3338. doi:10.1073/pnas.0636503100

Salomon, M., Lubin, Y., 2007. Cooperative breeding increases reproductive success in the social spider Stegodyphus dumicola. Behav. Ecol. Sociobiol. 61, 1743–1750. doi:10.1007/s00265-007-0406-2

Sanderson, J., 2012. Conflict and cooperation in vertebrate societies (PhD Thesis). University of Exeter.

Sanderson, J.L., Stott, I., Young, A.J., Vitikainen, E.I.K., Hodge, S.J., Cant, M.A., 2015. The origins of consistent individual differences in cooperation in wild banded mongooses, Mungos mungo. Anim. Behav. 107, 193–200. doi:10.1016/j.anbehav.2015.06.022

Sanderson, J.L., Young, A.J., Hodge, S.J., Kyabulima, S., Walker, S.L., Cant, M.A., 2014. Hormonal mediation of a carry-over effect in a wild cooperative mammal. Funct. Ecol. 28, 1377–1386. doi:10.1111/1365-2435.12307

Santema, P., Clutton-Brock, T.H., 2013. Meerkat helpers increase sentinel behaviour and bipedal vigilance in the presence of pups. Anim. Behav. 85, 655–661. doi:10.1016/j.anbehav.2012.12.029

Savage, J.L., Russell, A.F., Johnstone, R.A., 2013. Intra-group relatedness affects parental and helper investment rules in offspring care. Behav. Ecol. Sociobiol. 67, 1855–1865. doi:10.1007/s00265-013-1595-5

Schoech, S.J., Mumme, R.L., Wingfield, J.C., 1996. Prolactin and helping behaviour in the cooperatively breeding Florida scrub-jay , Aphelocoma c . coerulescens. Proc. R. Soc. B Biol. Sci. 445–456.

Sharp, P.J., Dawson, A., Lea, R.W., 1998. Control of luteinizing hormone and prolactin secretion in birds. Comp. Biochem. Physiol. C. Pharmacol. Toxicol. Endocrinol. 119, 275–282. doi:10.1016/S0742-8413(98)00016-4

Smith, H.G., Härdling, R., 2000. Clutch size evolution under sexual conflict enhances the stability of mating systems. Proc. R. Soc. B - Biol. Sci. 267, 2163–2170. doi:10.1098/rspb.2000.1264 222

References

Soares, M.C., Bshary, R., Fusani, L., Goymann, W., Hau, M., Hirschenhauser, K., Oliveira, R.F., 2010. Hormonal mechanisms of cooperative behaviour. Philos. Trans. R. Soc. B Biol. Sci. 365, 2737–2750. doi:10.1098/rstb.2010.0151

Sockman, K.W., Schwabl, H., Sharp, P.J., 2000. The role of prolactin in the regulation of clutch size and onset of incubation behavior in the American kestrel. Horm. Behav. 38, 168–176. doi:10.1006/hbeh.2000.1616

Sockman, K.W., Sharp, P.J., Schwabl, H., 2006. Orchestration of avian reproductive effort: an integration of the ultimate and proximate bases for flexibility in clutch size, incubation behaviour, and yolk androgen deposition. Biol. Rev. Camb. Philos. Soc. 81, 629–666. doi:10.1017/S1464793106007147

Sonerud, G., Steen, R., Løw, L.M., Røed, L.T., Skar, K., Selås, V., Slagsvold, T., 2014. Evolution of parental roles in raptors: prey type determines role asymmetry in the Eurasian kestrel. Anim. Behav. 96, 31–38. doi:10.1016/j.anbehav.2014.07.011

Stamps, J., Clark, A., Arrowood, P., Kus, B., 1985. Parent-offspring conflict in budgerigars. Behaviour 94, 1–40.

Stiver, K.A., Alonzo, S.H., 2009. Parental and mating effort: is there necessarily a trade-off? Ethology 115, 1101–1126. doi:10.1111/j.1439- 0310.2009.01707.x

Stutchbury, B.J.M., Morton, E.S., 2001. Life history traits, in: Behavioral Ecology of Tropical Birds. Academic Press, London, UK, pp. 23–38.

Taborsky, M., 2009. Reproductive skew in cooperative fish groups: virtue and limitations of alternative modeling approaches, in: Hager, R., Jones, C. (Eds.), Reproductive Skew in Vertebrate Societies: Proximate and Ultimate Causes. Cambridge University Press, pp. 265–304.

Taborsky, M., 1984. Broodcare helpers in the cichlid fish Lamprologus brichardi: Their costs and benefits. Anim. Behav. 32, 1236–1252. doi:10.1016/S0003- 3472(84)80241-9

Tanner, M., Kölliker, M., Richner, H., 2008. Differential food allocation by male and female , Parus major, parents: are parents or offspring in control? Anim. Behav. 75, 1563–1569. doi:10.1016/j.anbehav.2007.10.010

Tobias, J.A., Seddon, N., 2009. Signal jamming mediates sexual conflict in a duetting bird. Curr. Biol. 19, 577–582. doi:10.1016/j.cub.2009.02.036

Townsend, S.W., Allen, C., Manser, M.B., 2012. A simple test of vocal individual recognition in wild meerkats. Biol. Lett. 8, 179–182. doi:10.1098/rsbl.2011.0844

Trivers, R.L., 1972. Parental investment and sexual selection, in: Campbell, B. (Ed.), Sexual Selection and the Descent of Man 1871 -1971. Aldine Publishing Company, Chicago, Illinois, pp. 136–179.

223

References

Trumbo, S.T., 2006. Infanticide, sexual selection and task specialization in a biparental burying beetle. Anim. Behav. 72, 1159–1167. doi:10.1016/j.anbehav.2006.05.004

Van Roo, B., Ketterson, E., Sharp, P.J., 2003. Testosterone and prolactin in two songbirds that differ in paternal care: the blue-headed vireo and the red- eyed vireo. Horm. Behav. 44, 435–441. doi:10.1016/j.yhbeh.2003.07.001 van Vugt, M., Iredale, W., 2013. Men behaving nicely: public goods as peacock tails. Br. J. Psychol. 104, 3–13. doi:10.1111/j.2044-8295.2011.02093.x

Vehrencamp, S.L., 1983. A model for the evolution of despotic versus egalitarian societies. Anim. Behav. 31, 667–682. doi:10.1016/S0003- 3472(83)80222-X

Venables, W.N., Ripley, B.D., 2002. Modern Applied Statistics with S. Fourth Edition.

Vleck, C.M., Mays, N.A., Dawson, J.D., Goldsmith, A.R., 1991. Hormonal Correlates of parental and helping behaviour in cooperatively breeding Harris’ Hawks (Parabuteo unicinctus). Auk 108, 638–648.

Vleck, C.M., Patrick, D.J., 1999. Effects of vasoactive intestinal peptide on prolactin secretion in three species of passerine birds. Gen. Comp. Endocrinol. 113, 146–54. doi:10.1006/gcen.1998.7191

Voigt, C., Leitner, S., Gahr, M., 2006. Repertoire and structure of duet and solo songs in cooperatively breeding white-browed sparrow weavers. Behaviour 143, 159–182.

West, S.A., Griffin, A.S., Gardner, A., 2007. Social semantics: altruism, cooperation, mutualism, strong reciprocity and group selection. J. Evol. Biol. 20, 415–432. doi:10.1111/j.1420-9101.2006.01258.x

Westneat, D.F., Sherman, P.W., 1993. Parentage and the evolution of paternal care. Behav. Ecol. 4, 66–77. doi:10.1093/beheco/4.1.66

Westneat, D.F., Sherman, P.W., Morton, M.L., 1990. The ecology and evolution of extra-pair copulations in birds, in: Power, D.M. (Ed.), Current Ornithology. Plenum Press, New York, pp. 331–369.

Whittingham, L.A., 1994. Additional mating opportunities and male parental care in red-winged blacbirds. Anim. Behav. 875–883.

Whittingham, L.A., Dunn, P.O., Clotfelter, E.D., 2003. Parental allocation of food to nestling tree swallows: the influence of nestling behaviour, sex and paternity. Anim. Behav. 65, 1203–1210. doi:10.1006/anbe.2003.2178

Wilkin, T.A., King, L.E., Sheldon, B.C., 2009. Habitat quality, nestling diet, and provisioning behaviour in great tits Parus major. J. Avian Biol. 40, 135–145. doi:10.1111/j.1600-048X.2009.04362.x

Wingfield, J.C., Lewis, D.M., 1993. Hormonal and behavioural responses to simulated territorial intrusion in the cooperatively breeding white-browed 224

References

sparrow weaver, Plocepasser mahali. Anim. Behav. 45, 1–11.

Winkler, D.W., 1987. A general model for parental care. Am. Nat. 130, 526–543.

Wong, M., Balshine, S., 2011. The evolution of cooperative breeding in the African cichlid fish, Neolamprologus pulcher. Biol. Rev. 86, 511–530. doi:10.1111/j.1469-185X.2010.00158.x

Wright, J., 2007. Cooperation theory meets cooperative breeding: exposing some ugly truths about social prestige, reciprocity and group augmentation. Behav. Processes 76, 142–148. doi:10.1016/j.beproc.2007.01.017

Wright, J., 1999. Altruism as a Signal: Zahavi’s Alternative to Kin Selection and Reciprocity. J. Avian Biol. 30, 108–115.

Wright, J., 1997. Helping-at-the-nest in Arabian babblers: signalling social status or sensible investment in chicks? Anim. Behav. 54, 1439–1448. doi:10.1006/anbe.1997.0563

Wright, J., Berg, E., De Kort, S., Khazin, V., Marklakoz, A., 2001. Cooperative sentinel behaviour in the Arabian babbler. Anim. Behav. 62, 973–979.

Wright, J., Berg, E., De Kort, S., Khazin, V., Marklakoz, A., 2001. Safe selfish sentinels in a cooperative birds. J. Anim. Ecol. 70, 1070–1079.

Wright, J., Maklakovi, A.A., Khazin, V., 2001. State dependent sentinels in arabian babblers. Proc. R. Soc. B Biol. Sci. 268, 821–826.

Wright, J., McDonald, P.G., te Marvelde, L., Kazem, A.J.N., Bishop, C.M., 2010. Helping effort increases with relatedness in bell miners, but “unrelated” helpers of both sexes still provide substantial care. Proc. R. Soc. B Biol. Sci. 277, 437–45. doi:10.1098/rspb.2009.1360

York, J., Young, A., Radford, A., 2014. Singing in the moonlight: dawn song performance of a diurnal bird varies with lunar phase. Biol. Lett. 10, 1–4. doi:10.1098/rsbl.2013.0970

York, J.E., 2012. The evolution and regulation of dawn song performance in white-browed sparrow weaver societies (PhD Thesis).

Young, A.J., 2009. The causes of physiological suppression in vertebrate societies: a synthesis, in: Hager, R., Jones, C. (Eds.), Reproductive Skew in Vertebrate Societies: Proximate and Ultimate Causes. Cambridge University Press, Cambridge, pp. 397–437.

Young, A.J., Bennett, N.C., 2013. Intra-sexual selection in cooperative mammals and birds: why are females not bigger and better armed? Philos. Trans. R. Soc. B Biol. Sci. 368, 20130075. doi:10.1098/rstb.2013.0075

Young, A.J., Jarvis, J.U.M., Barnaville, J., Bennett, N.C., 2015. Workforce Effects and the Evolution of Complex Sociality in Wild Damaraland Mole Rats. Am. Nat. 186, 302–311. doi:10.1086/682048

Young, A.J., Spong, G., Clutton-Brock, T.H., 2007. Subordinate male meerkats 225

References

prospect for extra-group paternity: alternative reproductive tactics in a cooperative mammal. Proc. R. Soc. B Biol. Sci. 274, 1603–1609. doi:10.1098/rspb.2007.0316

Young, K.H., Linzer, D., Buhi, W.C., Horseman, N., Davis, J., Kraeling, R., Linzer, D., Bazer, F.W., 1990. Biological activities of glycosylated and nonglycosylated porcine prolactin. Mol. Cell. Endocrinol. 71, 155–162.

Zahavi, A., 1995. Altruism as a handicap: the limitations of kin selection and reciprocity. J. Avian Biol. 26, 1–3.

Zahavi, A., 1990. Arabian babblers: the quest for social status in a cooperative breeder, in: Stacey, P.B., Koenig, W. (Eds.), Cooperative Breeding in Birds: Long-TermStudies of Ecology and Behavior. Cambridge University Press, Cambridge, pp. 103–130.

Zahavi, A., 1977. Reliability in communication systems and the evolution of altruism, in: Stonehouse, B., Perrins, C. (Eds.), Evolutionary Ecology. Baltimore (MA): University Park Press., pp. 253–259.

Zahavi, A., 1974. Communal nesting by the Arabian babbler: a case of individual selection. Ibis (Lond. 1859). 116, 84–87.

Zahavi, A., Zahavi, A., 1997. Babblers, competition for prestige and the evolution of altruism, in: Zahavi, A., Zahavi, A., Zahavi-Ely, N., Ely, M. (Eds.), The Handicap Principle. The Missing Piece of Darwin’s Puzzle. Oxford University Press, Oxford, pp. 125 – 150.

Zöttl, M., Heg, D., Chervet, N., Taborsky, M., 2013. Kinship reduces alloparental care in cooperative cichlids where helpers pay-to-stay. Nat. Commun. 4, 1– 9. doi:10.1038/ncomms2344

226