The Role of the Environment in the Evolution of Reproductive Strategies

The role of the environment in the evolution of reproductive strategies

Zachariah Wylde

A thesis submitted in fulfillment of the requirements for the degree

Doctor of Philosophy

Evolution and Ecology Research Centre

School of Biological, Earth, and Environmental Sciences

Faculty of Science

September 2019 Surname/Family Name : Wylde Given Name/s : Zachariah Gilmer Abbreviation for degree as give in the : PhD University calendar Faculty : Science School of Biological, earth, and environmental School : sciences; Evolution and Ecology Research Centre The role of the environment in the evolution of Thesis Title : reproductive strategies

Abstract 350 words maximum: Understanding variation is a central theme in evolutionary biology. This thesis examines factors that contribute to variation in anatomy, reproductive development, lifespan and behaviour using the neriid fly, Telostylinus angusticollis (Diptera, Neriidae). It has been known for a long time that older mothers produce offspring with reduced life expectancy. My first empirical chapter provides the first direct comparison of maternal and paternal age effects, and the first investigation of the potential for both maternal and paternal age effects to accumulate over multiple generations. My second and third empirical chapters focus on the interaction between larval nutrition and the social environment at adulthood. Ejaculate (sperm and semen) traits can be under sexual selection and often exhibit a heightened sensitivity to diet. It is also well known that males transfer more sperm when facing a risk of sperm competition from other competing males. In Chapter 3, I manipulated the nutritional quality of male larval diet and perceived sperm competition and tracked competing ejaculates within the female reproductive tract using a new fluorescent sperm-labelling technique that I developed. I was able to show for the first time that increased sperm transfer in response to sperm competition risk is modulated by the larval resources. Chapter 4 examines the neriid genitalia which are highly complex, and whose function and evolution are poorly understood. This study is the first to explore how genitalic trait integration and the evolvability of these traits in both sexes compares with integration of somatic traits, and how integration is affected by juvenile nutrition. My last chapter was amongst the first studies to experimentally manipulate self- perception of social status. I found that social dominance treatment affected both males’ and females’ social status and chemical profiles, suggesting that both sexes can perceive their own status within a group and dynamically change their chemical signal. This thesis shows that factors such as age, social environment and resource availability have important effects that range from early ontogeny to post-copulatory processes during adulthood, and even ageing. This work is important in that it illustrates that selection for context- and state- dependent reproductive tactics drives the evolution of complex individual plasticity in allocation to reproduction, in both males and females.

Declaration relating to disposition of project thesis/dissertation

I hereby grant to the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or in part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all property rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.

I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstracts International (this is applicable to doctoral theses only).

…………………………………………………………… ……………………………………..……………… ……….……………………...…….… Signature Witness Signature Date The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for restriction for a period of up to 2 years must be made in writing. Requests for a longer period of restriction may be considered in exceptional circumstances and require the approval of the Dean of Graduate Research.

FOR OFFICE USE ONLY Date of completion of requirements for Award:

INCLUSION OF PUBLICATIONS STATEMENT

UNSW is supportive of candidates publishing their research results during their candidature as detailed in the UNSW Thesis Examination Procedure.

Publications can be used in their thesis in lieu of a Chapter if:

The student contributed greater than 50% of the content in the publication and is the “primary author”, ie. the student was responsible primarily for the planning, execution and preparation of the work for publication

The student has approval to include the publication in their thesis in lieu of a Chapter from their supervisor and Postgraduate Coordinator.

The publication is not subject to any obligations or contractual agreements with a third party that would constrain its inclusion in the thesis

Please indicate whether this thesis contains published material or not. ☐ This thesis contains no publications, either published or submitted for publication (if this box is checked, you may delete all the material on page 2)

Some of the work described in this thesis has been published and it has been documented

☒ in the relevant Chapters with acknowledgement (if this box is checked, you may delete all the material on page 2)

This thesis has publications (either published or submitted for publication) incorporated ☐ into it in lieu of a chapter and the details are presented below

ORIGINALITY STATEMENT

‘I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.’

Signed ……………………………………………......

Date ……………………………………………......

‘I hereby grant the University of New South Wales or its agents a non-exclusive licence to archive and to make available (including to members of the public) my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known. I acknowledge that I retain all intellectual property rights which subsist in my thesis or dissertation, such as copyright and patent rights, subject to applicable law. I also retain the right to use all or part of my thesis or dissertation in future works (such as articles or books).’

‘For any substantial portions of copyright material used in this thesis, written permission for use has been obtained, or the copyright material is removed from the final public version of the thesis.’

Signed ……………………………………………......

Date ……………………………………………......

AUTHENTICITY STATEMENT ‘I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis.’

Signed ……………………………………………......

Date ……………………………………………......

“Birth, and copulation, and death.

That’s all the facts when you come to brass tacks.”

– T. S Eliot

ABSTRACT

Understanding variation is a central theme in evolutionary biology. This thesis examines factors that contribute to variation in anatomy, reproductive development, lifespan and behaviour using the neriid fly, Telostylinus angusticollis (Diptera, Neriidae). It has been known for a long time that older mothers produce offspring with reduced life expectancy. My first empirical chapter provides the first direct comparison of maternal and paternal age effects, and the first investigation of the potential for both maternal and paternal age effects to accumulate over multiple generations. My second and third empirical chapters focus on the interaction between larval nutrition and the social environment at adulthood. Ejaculate (sperm and semen) traits can be under sexual selection and often exhibit a heightened sensitivity to diet. It is also well known that males transfer more sperm when facing a risk of sperm competition from other competing males. In Chapter 3, I manipulated the nutritional quality of male larval diet and perceived sperm competition and tracked competing ejaculates within the female reproductive tract using a new fluorescent sperm-labelling technique that I developed. I was able to show for the first time that increased sperm transfer in response to sperm competition risk is modulated by the larval resources. Chapter 4 examines the neriid genitalia which are highly complex, and whose function and evolution are poorly understood. This study is the first to explore how genitalic trait integration and the evolvability of these traits in both sexes compares with integration of somatic traits, and how integration is affected by juvenile nutrition. My last chapter was amongst the first studies to experimentally manipulate self-perception of social status. I found that social dominance treatment affected both males’ and females’ social status and chemical profiles, suggesting that both sexes can perceive their own status within a group and dynamically change their chemical signal. This thesis shows that factors such as age, social environment and resource availability have important effects that range from early ontogeny to post-copulatory processes during adulthood, and even ageing. This work is important in that it illustrates that selection for context- and state-dependent reproductive tactics drives the evolution of complex individual plasticity in allocation to reproduction, in both males and females.

ACKNOWLEDGEMENTS

I would like to dedicate this thesis to my parents whom have always fostered and appreciated my curiosity about the world and science, even if they did not understand my ramblings. Even though you probably do not realise, your confidence in me has always encouraged me to continue, particularly when I felt the tendrils of imposter syndrome wrap themselves around me! I know my mother, Christine Cook, would have been immensely proud to see the completion of this work. My father, Robert Wylde, you have always been there in any capacity that you can and that meant the world to me. Thank you, Dad, for your encouraging words and enthusiasm in my work throughout the years.

I am indebted to my wife, Cassandra Leith, for her unwavering support in all facets of my life. You fortified my confidence and make me a better human. You also endured my endless boring utterings and my stress. You picked me up when I felt unmotivated and continue to inspire me. Without your support, I am sure I would not have made this journey.

Russell Bonduriansky, my supervisor, mentor and friend. I feel so lucky to have been able to work with you on these projects. I hope you know that you are truly loved by all who have worked with you and continue to do so (sometimes I’m not sure you actually realise). Your intellectual generosity and practical support throughout this PhD have been immense and I cannot begin to even express my gratitude. I feel very humbled to have worked with someone of your calibre. Also, thank you for listening and responding to my numerous speculations and sporadic emails, humour, and lastly and mostly, your patience! Angela Crean, my lovely and super-talented co-supervisor. Your advice was always on point and with good humour. I enjoyed our conversations, and your supportive nature. Your insight and knowledge were essential to my projects and you were always happy to answer any questions I had. I also feel very proud to call you a supervisor and friend, so thank you.

I must also thank the Bonduriansky lab members throughout the years; Amy Hooper, Erin Macartney, Nathan Burke, Foteini Spagopoulou, Felix Zajitscek, and Margo Adler. I feel humbled to have had the privilege of working amongst you as my contemporaries. We all

banded close throughout the years with lots of laughter, helpful discussion, and of course cake! Thank you for the welcoming distraction or interesting intellectual discourse. I look forward to seeing you all in the future wherever that may be.

I am also indebted to my amazing advisory panel; Terry Ord, Shinichi Nakagawa, Michael Kasumovic. You always had sound advice, particularly in managing my time and mental health and were open to questions throughout this thing, so thank you. I must also just mention that the E&ERC and BEES has been an amazing work environment with so many workshops and events to attend. I truly felt I could ask anyone for help, and they would oblige with kindness. Jono and Vera, you guys rule. I must also thank Hayley bates for giving me multiple employment opportunities over the years. You helped me to grow as a teacher and also eat, so thank you!! Thank you to my family at Stanbuli restaurant for being flexible, feeding me delicious food and putting up with my strange humour over the years.

To all others who suffered my jargon filled vomitus and supported me in the many times this PhD endeavour felt particularly difficult. Not only has this process been hard work mentally, it has also been stressful in other realms of life socially, financially and emotionally. Without your unwavering support, whether it be by distraction or some words of encouragement, I am not sure I would have made it without you all, so thank you.

My colourful family. Even though you may not really understand what exactly I do, your humour and emotional support has meant so much to me. We may be a slightly dysfunctional bunch at the best of times but there is a huge amount of love there.

Finally, the star of the show. Thank you to the many neriid flies who were sacrificed throughout the making of this thesis. To most people, flies are a nuisance, but to me and many others they are of great inspiration, particularly the species used in this study that enabled me to examine so many interesting questions.

iii

PUBLICATIONS & PRESENTATIONS

Published and accepted for publication

Hooper, A. K., Spagopoulou, F., Wylde, Z., Maklakov, A. A. and Bonduriansky, R. 2017. Ontogenetic timing as a condition-dependent life history trait: high-condition males develop quickly, peak early and age fast. Evolution 71:671-685.

Wylde, Z., Adler, L., Crean, A., Bonduriansky, R. 2019. Perceived dominance status affects chemical signalling in the neriid fly Telostylinus angusticollis. Animal behaviour 158: 161- 174.

Wylde, Z., Crean, A., Bonduriansky, R. 2019. Effects of condition and sperm competition risk on sperm allocation and storage in neriid flies. Behavioural Ecology (published 07/11/19).

Wylde, Z., Spagopoulou, F., Hooper, A. K., Maklakov, A. A., and Bonduriansky, R. 2019. Parental breeding age effects on descendants’ longevity interact over 2 generations in matrilines and patrilines. PloS Biology 17(11): e3000556.

Conference presentations

Australasian Society for the Study of Evolution (2019 Sydney). “Sexual asymmetry in the condition dependence of genitalic and somatic trait integration.”Australasian Society for the Study of Animal Behaviour (2019 New Zealand). “Self-perceived dominance hierarchy effects on Cuticular hydrocarbon profiles in Telostylinus angusticollis (Diptera).”

Australasian Society for the Study of Evolution (2018 Tasmania). “Self-perceived dominance hierarchy effects on Cuticular hydrocarbon profiles in Telostylinus angusticollis (Diptera).”

Biology of spermatozoa 14 BoS (2017 Sheffield). Poster presentation: “Male condition and mating sequence effects on ejaculate storage and use in the Neriid fly, Telostylinus angusticollis.”

The Australasian Society for the study of Animal Behaviour joint conference with Behaviour (2015 Cairns). “Grandparental effects of breeding age on lifespan in the neriid fly, T. angusticollis.”

In preparation and review

Spagopoulou, F., Hooper, A. K., Wylde, Z., Bonduriansky, R., and Maklakov, A. In prep. Early-life parental diet effects on ageing depend on the sex of parents and their offspring.

Wylde, Z., & Bonduriansky ,R. In prep. Condition-dependence of phenotypic integration and the evolvability of genitalic and somatic traits in the neriid fly.

TABLE OF CONTENTS

ABSTRACT ...... i

ACKNOWLEDGEMENTS...... ii

PUBLICATIONS & PRESENTATIONS ...... iv

LIST OF FIGURES ...... x

LIST OF TABLES ...... xiv

CHAPTER 1 ...... 1

General Introduction ...... 1

References ...... 10

CHAPTER 2 ...... 18

Parental breeding age effects on descendants’ longevity interact over

two generations in matrilines and patrilines ...... 18

Abstract ...... 19 Introduction ...... 20 Results ...... 23 Lifespan ...... 23

Discussion ...... 31 Materials and Methods ...... 35 Source of experimental flies ...... 36

Larval rearing and diet manipulation ...... 36

F1 Adult housing and competitive environment ...... 36

F1 Adult male and female age-at-breeding manipulation ...... 37

F2 Adult male and female age-at-breeding manipulation ...... 37

F3 Rearing and quantification of lifespan ...... 38

Lifespan analysis ...... 40

Mortality rate analysis ...... 41

Author contributions ...... 42 Acknowledgements ...... 42 Data archiving ...... 42 References ...... 42 Supplementary information ...... 55 Effects on lifespan and mortality...... 55

Effects on body size and development time ...... 78

CHAPTER 3 ...... 85

Effects of condition and sperm competition risk on sperm allocation

and storage in neriid flies ...... 85

Abstract ...... 86 Introduction ...... 87 Materials and methods ...... 91 Experimental animals ...... 91

Larval diet manipulation ...... 92

Labelling ejaculates with rhodamine ...... 92

Mating sequence manipulation ...... 93

Sample preparation ...... 96

Quantifying amount of stored ejaculate ...... 96

Statistical analyses ...... 97

Results ...... 98 Body size ...... 98 vii

Copulatory behaviours ...... 99

Ejaculate within the spermathecae ...... 102

Discussion ...... 105 Authors’ contributions ...... 109 Acknowledgements ...... 109 Data accessibility ...... 110 References ...... 110 Supplementary information ...... 118

CHAPTER 4 ...... 123

Integration and condition-dependence of genitalic and somatic traits

in male and female neriid flies ...... 123

Abstract ...... 124 Introduction ...... 125 Materials and methods ...... 130 Fly culturing ...... 130

Dissection and sample preparation ...... 131

Morphological traits and measurement ...... 133

Statistical analyses ...... 133

Results ...... 138 Morphological integration ...... 138

Effect size of larval diet ...... 143

Effect size of G × E...... 144

Discussion ...... 145 Acknowledgements ...... 151 References ...... 151

CHAPTER 5 ...... 164

viii

Perceived dominance status affects chemical signalling in the neriid fly

Telostylinus angusticollis...... 164

Abstract ...... 165 Introduction ...... 166 Materials and methods ...... 170 Fly culturing ...... 170

Manipulation of dominance status ...... 171

Extraction of epicuticle hydrocarbons ...... 174

Chemical analyses ...... 175

Statistical analyses ...... 176

Results ...... 179 GC-MS analysis of CHC extracts ...... 180

Comparison of mean CHC abundances ...... 184

Linear discriminant function analysis of CHC blend ...... 185

Sex-limited CHCs ...... 187

Larval diet effects on CHC profiles ...... 188

Discussion ...... 190 Authors’ contributions ...... 194 Acknowledgements ...... 194 Data accessibility ...... 194 References ...... 195 Supplementary information ...... 202

CHAPTER 6 ...... 209

General conclusion ...... 209

References ...... 218

LIST OF FIGURES

CHAPTER 2 25 Figure 1

Effects of grandparental (F1) breeding age and larval diet on grand-offspring (F3) lifespan in patrilines and matrilines. 26 Figure 2 Interaction between effects of grandparental and parental breeding ages on grand-offspring lifespan in patrilines and matrilines. 27 Figure 3 Effects of grandparental larval diet and breeding age on estimated age-specific survival and mortality rates. 29 Figure 4

Effects of F1 breeding age and F2 breeding age on estimated age-specific survival and mortality rates. 38 Figure 5 Experimental design. 54 Figure S1

Combined effects of F1, F2 breeding ages and F1 larval diet quality on mean F3 life span. 62 Figure S2 Values of the KLDC for patrilines, comparing parameter posterior distributions between treatment groups. 65 Figure S3 Values of the KLDC for matrilines, comparing parameter posterior distributions between our treatment groups. 69 Figure S4 Values of the KLDC for patrilines, comparing parameter posterior distributions between treatment groups. 74 Figure S5

Values of the KLDC for matrilines, comparing parameter posterior distributions between our treatment groups. 78 Figure S6

Effects of F2 breeding age and F2 sex on F3 body size in patrilines. 80 Figure S7 Effects of F1 larval diet and age at breeding on F3 body size in patrilines and matrilines.

CHAPTER 3

78 Figure 1 Reproductive organs of T. angusticollis. 81 Figure 2 Experimental design. 82 Figure 3 Experimental workflow 86 Figure 4 The effects of larval diet quality on male thorax length. 88 Figure 5 A: Latency to mate in males reared on rich and poor larval diets. B; Copulation duration in males reared on rich or poor larval diets. 91 Figure 6 Interaction between male larval diet and mating sequence on the amount of ejaculate within each spermatheca. 92 Figure 7 Total amount of ejaculate (arbitrary fluorescence signal units, AU) within each of the three spermathecae. 109 Figure S1 Relationship between the amount of ejaculate (arbitrary fluorescence signal units, AU) and male body size (thorax length) for males mating first (black points, black line) and second (grey points, grey line) for each of the three spermatheca.

CHAPTER 4

119 Figure 1 Genitalia and reproductive traits of Telostylinus angusticollis. 123 Figure 2 Repeatability (R2) of trait measures for genitalic and somatic traits of males and females. 127 Figure 3 Principal components 1 vs. 2 for high condition (left) and low condition (right) T. angusticollis. 128 Figure 4 Principal components 1 vs. 2 for high condition (left) and low condition (right) T. angusticollis. 129 Figure 5

Comparison of integration using mean relative standard deviation of eigenvalues SDrel (λ). 130 Figure 6 Larval diet effect sizes on genitalic and somatic traits in males and females. 132 Figure 7 Conditional effect sizes including larval diet × family interaction on genitalic and somatic traits in males and females.

CHAPTER 5 157 Figure 1 Male neriid flies engaged in escalated combat (on left). On the right a male guards a female as she oviposits into a damaged area of a coral tree (Erythrina spp.). 158 Figure 2 Experimental design. 165 Figure 3 Effect of social dominance treatment on position within competitive arenas and behaviour for focal males and females.

xii

168 Figure 4 Mirrored GC chromatographic profile of a pooled male (above) and female (below) T. angusticollis (not adjusted for body-size). 169 Figure 5 Relative peak areas for each shared CHC by sex and dominance treatment. 170 Figure 6 Results of linear discriminant function analyses of cuticular hydrocarbon extracts of male and female T. angusticollis from dominant and subordinate dominance status treatments. 172 Figure 7 Scores for individual focal males (triangles) and females (circles) on the first two discriminant functions of shared CHCs with ‘dominant’ and ‘subordinate’ individuals represented by red and blue colours, respectively. 173 Figure 8 The effect of perceived dominance status on the expression of sex-limited CHCs. 174 Figure 9 PCA factor scores for the first two PCs of CHC profile for T. angusticollis males and females reared on different larval diets, and for wild-caught individuals of each sex. 187 Figure S1 The effects of larval diet quality on male (a) and female (b) thorax length.

xiii

LIST OF TABLES

CHAPTER 2 23 Table 1

Linear mixed effects models of F3 lifespan for patrilines and matrilines. Significant effects are highlighted in bold. 55 Table S1

Factorial summary of mean F3 life span, development time, and thorax length for patrilines. 56 Table S2

Factorial summary of mean F3 life span, development time, and thorax length for matrilines. 57 Table S3 Linear mixed-effects models of F3 life span for patrilines including F1 competitive environment. 58 Table S4

Linear mixed-effects models of F3 life span for patrilines and matrilines, with thorax length and development time of all focal individuals included as covariates. 60 Table S5 Model selection results based on ‘flexsurv’ package. 60 Table S6 Parameter estimates for each treatment group for the best fitting model (Gompertz with

simple shape) for grand-paternal effects of F1 larval diet × F1 breeding age. 63 Table S7 Mean KLDC values for patrilines, comparing parameter posterior distributions between F1 treatment groups. 64 Table S8 Parameter estimates for each treatment group for the best fitting model (Gompertz with

Simple shape) for grand-maternal effects of F1 larval diet × F1 breeding age. 66 Table S9

xiv

Mean KLDC values for matrilines, comparing parameter posterior distributions between F1 treatment groups. 67 Table S10 Parameter estimates for each treatment group for the best fitting model (Gompertz with

simple shape) for effects of F1 breeding age × F2 breeding age in patrilines. 70 Table S11 Mean KLDC values for patrilines comparing parameter posterior distributions between treatment groups. 72 Table S12 Parameter estimates for each treatment group for the best fitting model (Gompertz with

Simple shape) for effects of F1 breeding age × F2 breeding age in matrilines. 75 Table S13 Mean KLDC values for matrilines, comparing parameter posterior distributions between our treatment groups. 78 Table S14 Linear mixed-effects model of F3 body size. Significant effects are highlighted in bold. 81 Table S15 Linear mixed-effects model of F3 development time.

CHAPTER 3

87 Table 1 linear mixed model results for latency to mate and copulation duration (including larval diet). 90 Table 2 linear mixed models for effects of male thorax length (centred), mating sequence and copulation duration on ejaculate storage. 105 Table S1 Models ranked by AICc.

106 Table S2 linear mixed model results for latency to mate and copulation duration with male thorax length (uncentred) as a predictor instead of larval diet. 107 Table S3 Linear mixed models for effects on ejaculate storage patterns with male thorax length (uncentred) as a predictor instead of larval diet. 108 Table S4 linear mixed models for effects of male thorax length (uncentred), mating sequence and copulation duration on ejaculate storage.

CHAPTER 4

124 Table 1 Abbreviations for traits used in figures with calculated repeatability after controlling for larval diet treatment. 126 Table 2 Comparison of integration levels measured by the mean relative standard deviation of eigenvalues.

CHAPTER 5 166 Table 1 Epicuticular compounds of Telostylinus angusticollis, identified by CHC-specific ionic signatures. 171 Table 2 The success of predicting dominance status within male, female and shared CHC data sets based on LDA analysis. 188 Table S1 Summary of linear discriminant analysis (LDA) based on PCA transformation of CHC data. 189 Table S2

xvi

Results of ANOVA for CHCs within sex. We analysed each of the 20 and 24 peaks of males and females, respectively. 191 Table S3 Results of MANOVA and ANOVA for 17 CHC peaks shared between the sexes.

xvii

CHAPTER 1

General Introduction

General introduction

When I think about how I got here, I always come back to memories of being a child and conducting strange (and probably unethical) experiments on cicadas in neighbourhood trees. I was bound to become an entomologist and evolutionary ecologist, even if that is not what I intended! I also think of my first school speech which entailed a horrific history of medical practices in which I won 4th place. When I really think about it, I have always been curious about the world, science, and how things work, which is the currency that science thrives on. When it came to my teenage years, perhaps, also a little strange, I became infatuated with forensic science and from the age of 13, I must have read dozens of books and watched countless episodes of New Detectives. Coming from such a relatively humble upbringing, I must admit that I never actually thought I would end up attending university. So, for this reason, I must especially thank my high school biology teacher Mrs Woodward, who at the time appeared somewhat harsh on me, but in retrospect nurtured my abilities and curiosity giving me the confidence to attend university in the aim of becoming a forensic biologist. When it came to University, I enrolled in a biological sciences degree and was subsequently offered a MSc in forensic science, but by that time I was much too enthralled with evolutionary biology that I couldn’t even entertain that idea. Since then, I have lived and breathed evolutionary ecology and here I am today…

When it comes down to it, my thesis focuses on the role of the environment in the evolution of reproductive strategies. After all, “evolutionarily speaking, love is all about procreation” (Naskar 2016). So, I suppose in a sense, I am in love with understanding the mechanisms that drive the evolution of the diverse strategies we see animals use to gain a mate and reproduce! Reproductive strategies are so fundamental and vital to an organism, but the mechanisms that drive their evolution are inherently complex, and still not clearly understood.

Similarly complex, is trying to distil so much work into its essential bones and principles. Sometimes when I think about my PhD it appears disjunct and like a progression of moods and feelings. The theme and what is behind these emotions, or the meaning all comes later. At other times I see a central theme staring at me like a monolith, soon to disappear in a cloud of whimsical abstractions. But now, as I finally approach the end of my PhD and ascend this

General introduction

veil, I can finally see the common themes that tie this thesis together. Coming up with ideas is not the hard part, the real challenge is finding the time to actually build something meaningful with them. I feel incredibly fortuitous to have had the luxury to spend so much time in being an architect to design a house for these ideas to live, and my hope, is that this house is unique in its structure and still somewhat refined by the time you have finished reading this dissertation.

In a thesis that focuses so much on reproduction, I must first define what it is and why it is so important. Simply put, reproduction is the process of copying something, and in its biological context, this means producing offspring. Living things must evolve however, and evolution cannot occur without variation. This means that the process of reproduction or ‘copying’ must be inherently defective, in that it intrinsically produces mutations and error. Weismann (1886), eloquently described sex as ‘a source of individual variability furnishing material for the operation of natural selection,’ emphasizing the importance of reproduction and its inaccuracies on the course of evolutionary change. It is therefore of no surprise that evolutionary biologists are so interested in the variation of reproductive traits and their many peculiarities. What is the extent of reproductive trait plasticity and how does this variation influence reproductive processes? How does the plasticity in allocation strategies affect transgenerational traits such as offspring longevity to the minutiae such as genitalic traits, strategic sperm allocation and storage, and even chemical signalling?

Understanding this phenotypic plasticity is essential to a complete understanding of life- history patterns and tactics (Roff 2002). It therefore seems logical that we examine the links between environment and reproduction to increase our understanding of the mechanisms that drive its evolution. This is where my thesis sits, between reproduction and the environment. Understanding environmental factors and how sexual traits covary is of fundamental importance to understanding the way that reproductive strategies evolve (Chenoweth and Blows 2006). The definition of environment does not however exclude the social structure or social context in insects like flies that do not exhibit true sociality. The extent to which the social context has an effect in insects , is only beginning to be understood

General introduction

in terms of sexual interactions (Yun et al. 2017), the purging of deleterious mutations (Colpitts et al. 2017) and chemical communication (Gershman et al. 2014). Throughout this thesis, the environment therefore includes the social and physical contexts that can influence reproductive processes.

Often, the resources that govern reproductive allocation strategies are finite, so investment in one trait can, at least in theory, reduce or limit investment in others (Roff 1993; Stearns 2000; Emlen 2001). It is for this reason that the experimental manipulation of the nutritional environment has long been thought to be the most robust method to characterize the proximate causes of variation in reproductive traits (Ford and Seigel 1989; Du 2006). Much of my PhD research therefore hinges on the concept of resource-allocation and its proximate and downstream effects on reproductive strategies.

Classic condition-dependence theory imagines that as an individual accumulates resources, it also allocates them to either the production or upkeep of traits that help to increase reproduction or survival (Andersson, 1982,1994; Rowe & Houle, 1996). However, the condition of an individual is complex and underpinned by interdependent components of the somatic state, the genome and its epigenetic map (Hill 2011), so there is much scope for discovery within this framework. Not surprisingly, a complex array of sexually selected traits are condition-dependent and have important effects on performance in sexual competition (Bonduriansky and Rowe 2005). The most commonly studied condition-dependent traits are male secondary sexual traits that are often costly and exaggerated signals or weapons (Cotton et al. 2004), but much remains to be learnt about inter-trait variation in condition- dependence, particularly in females and non-sexual traits (Bonduriansky and Rowe 2005). Because diet quality is an important environmental determinant of adult body size and condition (Blackenhorn 2000), particularly in a holometabolous insect such as our model species T. angusticollis, herein, I use the term ‘condition’ as synonymous to larval diet quality.

The studies included in this thesis are consequently all about how the social and ecological environment affects numerous aspects of reproduction including investment in offspring longevity, sperm transfer, chemical signalling, and the development of trait morphology. I

General introduction

addressed these questions using a number of different laboratory experiments on the neriid fly, Telostylinus angusticollis.

Reproductive allocation is a an important factor in determining fitness, but parents must also balance the trade-off between fecundity and the investment in each offspring (Smith and Fretwell 1974; Rollinson and Rowe 2016). Since condition is directly related to the total amount of resources an individual has to invest in reproduction and somatic maintenance, it also makes sense to examine the relative importance of condition to other life-history traits, such as age-specific reproductive investment, and how these factors might interact to influence fitness-related offspring traits such as body size and lifespan. As a result, one might expect that low condition parents cannot afford to invest as highly in offspring when compared to high condition parents, regardless of breeding age. In my second chapter, I examine the effects of larval diet and breeding age on the body size and longevity of descendants over two generations.

In many species, offspring of older mothers have a reduced lifespan, a phenomenon known as the ‘Lansing’ effect (Lansing 1947). These types of maternal age effects have been observed in a great variety of taxa including yeast, plants, nematodes, rotifers, insects, birds and mammals (Gavrilov and Gavrilova 1997; Mousseau and Fox 1998; Priest et al. 2002; Ducatez et al. 2012). While most studies focus on offspring lifespan, some studies do show that maternal age at breeding can also affect offspring juvenile viability and adult reproductive performance (Mousseau 1991; Hercus and Hoffmann 2000; Fay et al. 2016; Lippens et al. 2017; Koch et al. 2018). A few reports have also described effects of paternal age at breeding on offspring performance (Priest et al. 2002; Ducatez et al. 2012). Although parental age effects represent a potentially important source of variation in individual mortality risk, longevity, and fitness, many aspects of these effects remain poorly understood. Where most studies only examine maternal effects on offspring lifespan, we show that both paternal and maternal ages at breeding can both contribute substantially to intra-population variation in longevity. Here, we also show that grandparental larval diet (i.e., early-life condition) has a relatively small impact on future generations’ mortality and lifespan when compared to the

General introduction

effects grandparental and parental age-specific reproductive investment. Classically, parental investment refers to any type of investment that might increase offspring fitness (Trivers, 1972). If investment in offspring declines with age because resources are running out, or the soma of the parent is deteriorating, then we would expect a negative effect of parental age on offspring. This study is the first to directly test this prediction in both sexes within the same experiment, and to show that both maternal and paternal age effects can carry over to grandoffspring. This type of parental effect could be mediated in both male traits such as sperm and seminal proteins, or female investment in egg provisioning as well as mutation accumulation in both sexes.

This brings me to the subject of sperm competition and ejaculate allocation in males. Classic theory on sperm allocation suggests that ejaculate allocation should increase as sperm competition risk increases until a threshold in sperm competition is reached, and male expenditure subsequently decreases (Parker and Begon 1986; Parker et al. 1997). However, a number of studies have failed to support this prediction (reviewed in Williams, Day, & Cameron, 2005) suggesting that other factors might be at play. Sperm competition is often seen as a powerful driving force behind the evolution of many male traits and even sexually antagonistic traits in females. Still, it seems that little attention has been paid to the environmental contexts throughout ontogeny and how these affect sperm and ejaculate allocation.

Theoretical and empirical studies have begun to reveal that both sperm and non-sperm components of the ejaculate can be similarly plastic and condition-dependent in their responses to the developmental environment (Wigby et al. 2016). Yet, how variation in male condition shapes ejaculate allocation remains poorly understood. Most studies have manipulated male condition through dietary restriction at the adult stage, but few studies have examined how the juvenile nutritional environment shapes male post-copulatory performance (Amitin and Pitnick 2007; McGraw et al. 2007; Engqvist 2008; Melo et al. 2014; Vega-Trejo et al. 2016; Dávila and Aron 2017). Furthermore, a recent meta-analysis found that, in arthropods, juvenile nutrient limitation plays a more important and larger role

General introduction

in shaping ejaculate traits when compared to adult nutrient limitation (Macartney et al. 2019). To understand ejaculate allocation strategies, it is therefore necessary to elucidate how sperm competition risk and condition jointly influence investment in ejaculate traits (Perry et al. 2013).

Large males tend to have higher mating rates than smaller males, as seen in antler flies (Bonduriansky and Brassil 2005), European vipers (Madsen et al. 1993), and rhesus macaques (Georgiev et al. 2015). High-condition males might therefore be expected to evolve prudent ejaculate allocation strategies that enable them to take advantage of frequent mating opportunities. By contrast, low-condition males might be selected to invest maximally in all matings because their probability of achieving a mating is lower, and they may lack the resources required to elevate ejaculate expenditure even further when facing high sperm competition risk. In my third chapter, I used a factorial design to examine the effects of perceived sperm competition risk on the quantity of ejaculate transferred to a female’s sperm storage organs. From this experiment, we were able to show that high-condition males initiated mating more quickly, and when mating second also transferred more ejaculate to two of the three female sperm storage organs. This is the first study to show that males allocate ejaculates strategically by incorporating variation in both condition and perceived risk of sperm competition and the results highlight the complexity of sexual interactions.

Logically, the subject that must follow are the body parts that dispense or receive and store the male ejaculate, the genitalia. This has led to the unfortunate nickname of me by my immature brothers as the ‘fly molester.’ Obviously, they have no idea how interesting these structures can be from an evolutionary standpoint, and that male genitalia vary more dramatically than any other phenotypic trait in the animal kingdom (Simmons 2014). The evolution of genitalia, particularly those of male insects, has been a subject of longstanding curiosity to evolutionary biologists and represents an important enigma (Hosken and Stockley 2004). Why should genitalia be so morphologically complex and variable? Often species look so similar that they can only be differentiated by their genitalia. If sexual selection is the driver of their evolution (which is often relatively rapid when compared to

General introduction

non-genitalic traits; (Arnqvist 1997; Hosken and Stockley 2004; Eberhard 2009; Simmons et al. 2009) then we might expect that female genitalia should also exhibit comparable variation in size and shape, because successful reproduction is an interaction between both male and female parts. However, female genitalia have been much less studied, and little is known about their variability when compared to conspicuous male structures that are often much more readily examinable (Ah-King et al. 2014; Sloan & Simmons 2019).

Under the handicap hypothesis, sexually selected traits are expected to evolve high levels of condition-dependence (Andersson 1994), so if genital traits are shaped by sexual selection then we should also see a heightened level of condition-dependence (Hosken and Stockley 2004). Moreover, male genital morphology has even been shown to diverge under experimental evolution via sexual selection (Simmons et al. 2009; Cayetano et al. 2011). Few studies have attempted to examine the condition-dependence of genitalia however, and those that have, generally show that even if some genital traits appear to function as secondary sexual traits, they do not exhibit heightened condition-dependence (House and Simmons 2007; Cayetano and Bonduriansky 2015; House et al. 2016). On the other hand, a couple of studies suggest heightened condition-dependence of some genitalic traits (Arnqvist and Thornhill 1998; Higgins et al. 2009), so it is not clear which evolutionary mechanisms we can assign a priori. Furthermore, only one study has examined the condition-dependence of a female insect’s genitalia (Cayetano and Bonduriansky 2015), most definitely reflecting a larger bias within the literature where most studies omit the female genitalia because of anatomical accessibility or persisting assumptions about their assumed passivity during sex, and relative variability (Ah-King et al. 2014).

Cheverud (1982), surmised that functionally and developmentally related traits evolve together through the selection of interdependence. The features of male and female genitalia or indeed the whole organism are not independent elements but are correlated with one another to varying degrees (Pavlicev et al. 2009). The study of morphological integration aims to describe these patterns and the amount of correlation between a set of phenotypic traits. My fourth chapter examined and compared the relative integration (i.e., level of

General introduction

morphological covariation in trait size) of genitalic and somatic traits. Based on the relative diversity of male genitalia we predicted that male genitalia would be less integrated than other traits in males, and possibly less integrated than female genitalia. We also predicted that most of the variation in genitalic traits arises from genetic sources (because of the relative rapid response of these traits to selection) and therefore expected that male genitalia would also respond relatively weakly to environmental factors when compared to somatic traits and female genitalia. It has also been postulated that variation in the condition-dependence of genital traits may not be manifested in absolute size, but rather in their fine structures (Eberhard et al. 1998; Cayetano and Bonduriansky 2015), which means that much variation may have been overlooked. We therefore asked whether integration itself is condition- dependent, such that the effects of variation in condition are reflected not just in the sizes of individual genitalic components but the degree of intercorrelation among all genitalic traits. To test these predictions we utilised a split-brood design and larval nutrient manipulation to compare trait integration of male and female genitalia to somatic traits using the neriid fly, T. angusticollis. We were able to show that genitalic traits are a lot less integrated (and therefore presumably less constrained in their evolution) than somatic traits. This finding may help to explain why insect genitalia are so quick to diverge in morphology. We also observed that males had significantly lower integration of genitalic traits when compared to females which may affect relative evolvability of these traits. Finally, we found that integration tended to decrease with increasing condition in both sexes, suggesting that different traits respond differently to resource availability. We discuss the implications of these findings for evolvability and ontogenetic constraints in genitalic diversification.

A trait that is also affected by sperm competition is the chemical profile of an individual, which is well characterised in highly social animals but less understood in species such as the neriid fly that do not exhibit eusociality. An important limitation of the existing literature on the role of cuticular hydrocarbons (CHCs) in status signalling is that much of the evidence is correlational. For example, some studies pair competing individuals in an arena to determine dominance status or to assess winner and loser effects, and report effects on chemical signals (e.g., Thomas and Simmons 2009; Rillich and Stevenson 2011). There is also some evidence 9

General introduction

in Drosophila serrata that CHCs profiles correlate with mating success but not with an individual’s ability to successfully defend a territory (White and Rundle 2014). Yet, it is not clear if differences in chemical signals between dominant and subordinate or successful and unsuccessful individuals reflect a perceived social status that can change dynamically, or whether these observed differences in both dominance and chemical profile are invariant features of adult individuals that result from genetic or environmental differences during development. My final data chapter therefore aimed to examine how the formation of a dominance hierarchy can influence chemical signalling—an effect that remains poorly known (Lin and Michener 1972; Savarit and Ferveur 2002; Grillet et al. 2006; Gershman et al. 2014). We were able to manipulate the perception of both male and female dominance status and examine their cuticular hydrocarbon (CHC) profiles. Our findings suggest that T. angusticollis males and females can alter their CHC profiles in response to their perceived social dominance status. These chemical signals could have potential downstream effects both within and between the sexes influencing mechanisms of mate attraction and the stratification of individuals within aggregation structures.

Overall, the various chapters of this thesis demonstrate the many ways in which the environment can influence the expression of reproductive strategies. My aim throughout this thesis has been to examine numerous traits, from the minutiae of genitalia and chemical signals, to the effects of breeding age on descendants’ lifespan and mortality. This body of work also raises a number of important questions around reproductive and life-history strategies that I should like to investigate in the future.

References

Ah-King, M., A. B. Barron, and M. E. Herberstein. 2014. Genital Evolution: Why Are Females Still Understudied? PLoS Biol. 12:1–8.

Amitin, E. G., and S. Pitnick. 2007. Influence of developmental environment on male- and female-mediated sperm precedence in Drosophila melanogaster. J. Evol. Biol. 20:381–391.

General introduction

Andersson, M. 1982. Sexual selection, natural selection and quality advertisement. Biol J Linn Soc 17:375–393.

Andersson, M. B. 1994. Sexual selection. Princeton University Press.

Arnqvist, G. 1997. The evolution of animal genitalia: distinguishing between hypotheses by single species studies. J. Linn. Soc 60:365–379.

Arnqvist, G., and R. Thornhill. 1998. Evolution of animal genitalia : patterns of phenotypic and genotypic variation and condition dependence of genital and non • genital morphology in water strider ( Heteroptera : Gerridae : Insecta ) Evolution of animal genitalia : patterns of phenotypic an. Genet. Res. 71:193–212.

Blackenhorn, W. U. 2000. The evolution of body size: what keeps organisms small? Q. Rev. Biol. 75:385–407.

Bonduriansky, R., and C. E. Brassil. 2005. Reproductive ageing and sexual selection on male body size in a wild population of antler flies (Protopiophila litigata). J. Evol. Biol. 18:1332– 1340.

Bonduriansky, R., and L. Rowe. 2005. Sexual selection, genetic architecture, and the condition dependence of body shape in the sexually dimorphic fly Prochyliza xanthostoma (Piophilidae). Evolution. 59:138–151.

Cayetano, L., and R. Bonduriansky. 2015. Condition dependence of male and female genital structures in the seed beetle Callosobruchus maculatus (Coleoptera: Bruchidae). J. Evol. Biol. 28:1364–1372.

Cayetano, L., A. A. Maklakov, R. C. Brooks, and R. Bonduriansky. 2011. Evolution of male and female genitalia following release from sexual selection. Evolution. 65:2171–2183.

Chenoweth, S. F., and M. W. Blows. 2006. Dissecting the complex genetic basis of mate choice. Nat Rev Genet. 7:681–692.

General introduction

Cheverud, J. M. 1982. Phenotpyic, genetic, and environmental morphological integration in the cranium. Evolution. 36:499–516.

Colpitts, J., D. Williscroft, H. S. Sekhon, and H. D. Rundle. 2017. The purging of deleterious mutations in simple and complex mating environments. Biol. Lett. 13:2–5.

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

Dávila, F., and S. Aron. 2017. Protein restriction affects sperm number but not sperm viability in male ants. J. Insect Physiol. 100:71–76. Elsevier.

Du, W. G. 2006. Phenotypic plasticity in reproductive traits induced by food availability in a lacertid lizard, Takydromus septentrionalis. Oikos 112:363–369.

Ducatez, S., M. Baguette, V. M. Stevens, D. Legrand, and H. Fréville. 2012. Complex Interactions Between Paternal And Maternal Effects: Parental Experience And Age At Reproduction Affect Fecundity And Offspring Performance In A Butterfly. Evolution. 66:3558–3569.

Eberhard, William G., B. A. Huber, Rodriguez, R L., Briceno, R D., I. Salas, and V. Rodriguez. 1998. One size fits all? Relationships between the size and degree of variation in genitalia and other body parts in twenty species of insects and spiders. Evolution. 52:415– 431.

Eberhard, W. G. 2009. Postcopulatory sexual selection: Darwin’s omission and its consequences. Proc. Natl. Acad. Sci. 106:10025–10032.

Emlen, D. J. 2001. Costs and the diversification of exaggerated animal structures. Nature 291:1534–1536.

General introduction

Engqvist, L. 2008. Genetic variance and genotype reaction norms in response to larval food manipulation for a trait important in scorpionfly sperm competition. Funct. Ecol. 22:127– 133.

Fay, R., C. Barbraud, K. Delord, and H. Weimerskirch. 2016. Paternal but not maternal age influences early-life performance of offspring in a long-lived seabird. Proc. R. Soc. B Biol. Sci. 283.

Ford, N. B., and R. A. Seigel. 1989. Phenotypic plasticity in reproductive traits: evidence from a viviparous snake. Ecology 70:1768–1774.

Gavrilov, L. A., and N. S. Gavrilova. 1997. Parental age at conception and offspring longevity. Rev. Gerontol. 7:5–12.

Georgiev, A. V., M. P. Muehlenbein, S. P. Prall, M. E. Thompson, and D. Maestripieri. 2015. Male quality, dominance rank, and mating success in free-ranging rhesus macaques. Behav. Ecol. 26:763–772.

Gershman, S. N., E. Toumishey, and H. D. Rundle. 2014. Time flies: time of day and social environment affect cuticular hydrocarbon sexual displays in Drosophila serrata. Proc. R. Soc. B Biol. Sci. 281:20140821–20140821.

Grillet, M., L. Dartevelle, and J.-F. Ferveur. 2006. A Drosophila male pheromone affects female sexual receptivity. Proc. R. Soc. B Biol. Sci. 273:315–323.

Hercus, M. J., and A. A. Hoffmann. 2000. Maternal and grandmaternal age influence offspring fitness in Drosophila. Proc. R. Soc. B Biol. Sci. 267:2105–2110.

Higgins, S. L., D. J. Hosken, and N. Wedell. 2009. Phenotypic and genetic variation in male genitalia in the seedbug, Lygaeus equestris (Heteroptera). Biol. J. Linn. Soc. 98:400–405.

Hill, G. E. 2011. Condition-dependent traits as signals of the functionality of vital cellular processes. Ecol. Lett. 14:625–634.

General introduction

Hosken, D. J., and P. Stockley. 2004. Sexual selection and genital evolution. Trends Ecol. Evol. 19:87–93.

House, C. M., K. Jensen, J. Rapkin, S. Lane, K. Okada, D. J. Hosken, and J. Hunt. 2016. Macronutrient balance mediates the growth of sexually selected weapons but not genitalia in male broad-horned beetles. Funct. Ecol. 30:769–779.

House, C. M., and L. W. Simmons. 2007. No evidence for condition-dependent expression of male genitalia in the dung beetle Onthophagus taurus. J. Evol. Biol. 20:1322–1332.

Koch, R. E., J. M. Phillips, M. F. Camus, and D. K. Dowling. 2018. Maternal age effects on fecundity and offspring egg-to-adult viability are not affected by mitochondrial haplotype. Ecol. Evol. 8:10722–10732.

Lansing, A. I. 1947. A transmissible, cumulative and reversible factor in aging. Gerontology 2:228–239.

Lin, N., and C. D. Michener. 1972. Evolution of Sociality in Insects. Q. Rev. Biol. 47:131– 159.

Lippens, C., B. Faivre, C. Lechenault, and G. Sorci. 2017. Aging parasites produce offspring with poor fitness prospects. Biol. Lett. 13.

Macartney, E. L., A. J. Crean, S. Nakagawa, and R. Bonduriansky. 2019. Effects of nutrient limitation on sperm and seminal fluid: a systematic review and meta-analysis. Biol. Rev., doi: 10.1111/brv.12524.

Madsen, T., R. Shine, J. Loman, and T. Håkansson. 1993. Determinants of mating success in male adders, Vipera berus.

McGraw, L. A., A. C. Fiumera, M. Ramakrishnan, S. Madhavarapu, A. G. Clark, and M. F. Wolfner. 2007. Larval rearing environment affects several post-copulatory traits in Drosophila melanogaster. Biol. Lett. 3:607–610.

General introduction

Melo, M. C., F. R. C. L. Almeida, A. L. Caldeira-Brant, G. G. Parreira, and H. Chiarini- Garcia. 2014. Spermatogenesis recovery in protein-restricted rats subjected to a normal protein diet after weaning. Reprod. Fertil. Dev. 26:787–796.

Mousseau, T. 1991. Maternal Effects In Insect Life Histories. Annu. Rev. Entomol. 36:511– 534.

Mousseau, T., and C. W. Fox. 1998. Maternal effects as adaptations. Oxford University Press, New York.

Naskar, A. 2016. What is Mind? Neuro Cookies.

Parker, G. A., M. A. Ball, P. Stockley, and M. J. G. Gage. 1997. Sperm competition games: a prospective analysis of risk assessment. Proc. R. Soc. B Biol. Sci. 264:1793–1802.

Parker, G. A., and M. Begon. 1986. Optimal egg size and clutch size: effects of environment and maternal phenotype. Am. Nat. 128:573–592.

Pavlicev, M., J. M. Cheverud, and G. P. Wagner. 2009. Measuring morphological integration using eigenvalue variance. Evol. Biol. 36:157–170.

Perry, J. C., L. Sirot, and S. Wigby. 2013. The seminal symphony: How to compose an ejaculate. Trends Ecol. Evol. 28:414–422.

Priest, N., B. Mackowiak, and D. E. L. Promislow. 2002. The role of parental age effects on the evolution of aging. Evolution. 56:927–935.

Rillich, J., and P. A. Stevenson. 2011. Winning fights induces hyperaggression via the action of the biogenic amine octopamine in crickets. PLoS One 6.

Roff, D. A. 1993. Evolution of life histories: theory and analysis. Springer Science & Business Media.

Roff, D. A. 2002. Life history evolution. Sinauer, Massachussetts.

General introduction

Rollinson, N., and L. Rowe. 2016. The positive correlation between maternal size and offspring size: fitting pieces of a life-history puzzle. Biol. Rev. 91:1134–1148.

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

Savarit, F., and J. Ferveur. 2002. Genetic study of the production of sexually dimorphic cuticular hydrocarbons in relation with the sex- determination gene transformer in Drosophila melanogaster. Genet. Res. 79:23–40.

Simmons, L. W. 2014. Sexual selection and genital evolution. Austral Entomol. 53:1–17.

Simmons, L. W., C. M. House, J. Hunt, and F. García-González. 2009. Evolutionary Response to Sexual Selection in Male Genital Morphology. Curr. Biol. 19:1442–1446.

Sloan, N. S., and Simmons, L. W. 2019. The evolution of female genitalia. J. Evol. Biol. 32(9): 882-899.

Smith, C. C., and S. D. Fretwell. 1974. The optimal balance between size and number of offspring. Am. Nat. 108:499–506.

Stearns, S. C. 2000. Life history evolution: Successes, limitations, and prospects. Naturwissenschaften 87:476–486.

Thomas, M. L., and L. W. Simmons. 2009. Male dominance influences pheromone expression, ejaculate quality, and fertilization success in the Australian field cricket, Teleogryllus oceanicus. Behav. Ecol. 20:1118–1124.

Vega-Trejo, R., M. D. Jennions, and M. L. Head. 2016. Are sexually selected traits affected by a poor environment early in life? BMC Evol. Biol. 16:1–12.

Weismann, A. 1886. Die Bedeutung der sexuellen Fortpflanzung fur die Selektionstheorie. Jena :G. Fischer.

White, A. J., and H. D. Rundle. 2014. Territory defense as a condition-dependent component of male reproductive success in Drosophila serrata. Evolution. 69:407–418. 16

General introduction

Wigby, S., J. C. Perry, Y. H. Kim, and L. K. Sirot. 2016. Developmental environment mediates male seminal protein investment in Drosophila melanogaster. Funct. Ecol. 30:410– 419.

Williams, P. D., T. Day, and E. Cameron. 2005. The Evolution of Sperm-Allocation Strategies and the Degree of Sperm Competition. Soc. study Evol. 59:492–499.

Yun, L., P. J. Chen, A. Singh, A. F. Agrawal, and H. D. Rundle. 2017. The physical environment mediates male harm and its effect on selection in females. Proc. R. Soc. B Biol. Sci. 284.

Parental breeding age effects offspring lifespan

CHAPTER 2

Parental breeding age effects on descendants’ longevity interact over two generations in matrilines and patrilines

Zachariah Wylde1, Foteini Spagopoulou2, Amy K. Hooper1, Alexei A. Maklakov2, 3, Russell Bonduriansky1

PLoS Biology 17(11), 2019. Published 25/11/19.

1Evolution and Ecology Research Centre, School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney, 2052, Australia

2Ageing Research Group, Department of Animal Ecology, Evolutionary Biology Centre, Uppsala University, Uppsala, Sweden

3School of Biological Sciences, University of East Anglia Centre, Norwich Research Park, Norwich, United Kingdom

Parental breeding age effects offspring lifespan

Abstract

Individuals within populations vary enormously in mortality risk and longevity, but the causes of this variation remain poorly understood. A potentially important and phylogenetically widespread source of such variation is maternal age at breeding, which typically has negative effects on offspring longevity. Here, we show that paternal age can affect offspring longevity as strongly as maternal age does and that breeding age effects can interact over 2 generations in both matrilines and patrilines. We manipulated maternal and paternal ages at breeding over 2 generations in the neriid fly Telostylinus angusticollis. To determine whether breeding age effects can be modulated by the environment, we also manipulated larval diet and male competitive environment in the first generation. We found separate and interactive effects of parental and grand-parental ages at breeding on descendants’ mortality rate and life span in both matrilines and patrilines. These breeding age effects were not modulated by grand-parental larval diet quality or competitive environment. Our findings suggest that variation in maternal and paternal ages at breeding could contribute substantially to intrapopulation variation in mortality and longevity.

Keywords: Senescence, lifespan, parental age, maternal effect, paternal effect, diet, social environment

Parental breeding age effects offspring lifespan

Introduction

In many species, offspring of older mothers have a reduced mean life span, a phenomenon known as the ‘Lansing’ effect [1] or maternal age effect. Maternal age effects have been observed in a great variety of organisms, including yeast, plants, nematodes, rotifers, insects, birds, and mammals [2–6]. Although most studies have focused on offspring life span, some studies show that maternal age at breeding can also affect offspring viability and adult reproductive performance [7–11]. A few studies have also reported effects of paternal age at breeding on offspring performance [2,5,6]. Parental age effects represent a potentially important source of variation in individual mortality risk, longevity, and fitness, but many aspects of these effects remain poorly understood.

Parental age effects could be caused by the accumulation of mutations in the germline [12]. In humans, mutations accumulate at a constant rate in the male germline and at an accelerating rate in the female germline [13]. Parental age effects could also be mediated by nongenetic factors. Recent studies on mice, monkeys, and humans have shown that patterns of DNA methylation across the genome change with age—a pattern known as the ‘epigenetic clock’ [14–18], and some of these altered epigenetic factors could be transmitted across generations [19–23]. Older parents could also transmit altered microRNAs or other factors such as proteins to offspring via the gametes [24,25]. For example, in mice, the transmission of proteins in the egg cytoplasm is thought to mediate maternal age effects on offspring [26], and more recent evidence suggests a role for sperm microRNAs in paternal effects [27–31]. Although such effects are best characterised in mammals, age-related changes in gamete quality also occur in arthropods, and such effects could contribute to parental age effects. For example, in the parasitoid wasp Eupelmus vuilletti, increasing maternal age is associated with reduced egg size and altered egg composition [32]. Likewise, in Daphnia pulex, maternal age is associated with changes in egg provisioning, with effects on offspring longevity and life history [33]. The transmission of dysregulated epigenetic or cytoplasmic factors from old- breeding parents to their offspring could mediate parental age effects in many species [34].

Parental breeding age effects offspring lifespan

Maternal and paternal effects are likely to be mediated by different factors and can have distinct effects on offspring [35,36]. However, relatively few studies have tested experimentally for effects of paternal age at breeding, and even fewer studies have directly compared the effects of maternal and paternal age at breeding on offspring performance. Experimental evidence in mice shows that offspring of older fathers have a reduced life span and suggests that this effect could be mediated by epigenetic (DNA methylation) changes within sperm of gene promoters involved in evolutionarily conserved pathways of life span regulation [37]. In Drosophila melanogaster, both maternal and paternal age effects have been reported [5]. Similar effects may occur in other species (including humans), although much of the evidence is correlational. For example, in the wandering albatross, paternal but not maternal age affected juvenile survival of offspring [11]. A recent long-term study on a natural population of house sparrows showed that paternal breeding has a similar effect size on life span and reproductive success to female breeding age and that these effects are transferred to offspring in a sex-specific manner [6]. In humans, advanced paternal age at breeding is associated with reduced sperm quality and testicular functions, and such effects appear to be mediated by both epigenetic changes and genetic mutations [38]. Advanced paternal age is also associated with reduced performance on standardised tests in children, whereas the effect of maternal age was more complex [39]. Likewise, parental age, and the difference between maternal and paternal ages, are associated with risk of autism spectrum disorder [40].

Parental age effects could interact with environmental factors such as diet and stress [8,41]. For example, a restricted maternal diet mitigated the effects of advanced maternal age at breeding on offspring longevity in rotifers [42]. In mice, a fat-restricted maternal diet did not influence maternal age effects [16], but maternal age effects were mitigated by rapamycin [43]. In the butterfly Pieris brassicae, effects of parental age at breeding on offspring performance were influenced by stress [2]. However, the role of environment in modulating effects of parental age remains largely unexplored.

Parental breeding age effects offspring lifespan

Perhaps the most important gap in the understanding of parental age effects is the potential for such effects to accumulate and interact over multiple generations. In Drospohila serrata, offspring juvenile viability decreased with increasing maternal and grand-maternal ages at breeding [8], but it remains unclear whether such cumulative effects can occur in partrilines or in other species. If such multigenerational effects are widespread, they could make an important contribution to variation in mortality and longevity and, potentially, play a role in the evolution of ageing [5,34].

Here, we examined 3 aspects of parental age effects that have received little attention in previous research by (1) comparing the effects of both male and female age at breeding on descendants, (2) testing for interactions of age at breeding with key environmental factors (diet and competitive environment), and (3) investigating the potential for the effects of age at breeding to accumulate over generations. We addressed these questions in the neriid fly Telostylinus angusticollis (Enderlein), a species endemic to New South Wales and Southern Queensland, Australia. Both larval and adult nutrition affect mortality rate and life span in this species [44,45]. Larval access to dietary protein has a nonlinear effect on adult longevity [44], but high overall macronutrient (protein and carbohydrate) abundance at the larval stage accelerates larval growth and development while also promoting rapid ageing in males [46,47]. Adult protein restriction extends life [45] and can interact with larval diet to influence reproductive ageing [48]. However, effects of parental age at breeding on offspring performance have not been investigated previously in this species.

We reared individuals of the grand-parental (F1) generation on either a high-nutrient or low- nutrient larval diet and then allowed adult females and males from these larval diet treatments to breed at 15 and 35 days of age. Neriid males fight other males for access to territories and females, and such male-male interactions could affect male ageing [47]. We therefore investigated the potential for male-male interactions to affect paternal age effects by manipulating F1 male competitive environment. Female and male offspring (F2) were reared on a standard larval diet (with a nutrient concentration intermediate between the high- nutrient and low-nutrient diets) and then allowed to breed at 15-day age intervals between

Parental breeding age effects offspring lifespan

ages 15 and 60 days. We quantified the adult longevity of grand-offspring (F3) and used these data to test for effects of grand-parental ages at breeding, grand-parental environment, and parental ages at breeding on grand-offspring life span, mortality rate, and actuarial ageing rate.

Results

Lifespan

F3 individuals (grand-offspring) from both matrilines and patrilines suffered similar negative effects of F1 (grand-parental) and F2 (parental) ages at breeding on life span (Table 1; Figs 1 and 2). F3 individuals descended from old-breeding grandmothers and grandfathers had

37.8% and 39.8% shorter lifespans, respectively, than F3 individuals descended from young breeding grandmothers and grandfathers. There was no effect of F1 larval diet on F3 life span in either matrilines or patrilines, nor an F1 larval diet × F1 age interaction. There were also no main or interactive effects of F1 male competitive environment within patrilines (S3 Table).

However, we detected an F1 × F2 age interaction within both matrilines and patrilines, whereby the negative effect of F1 age at breeding was diminished as F2 age at breeding increased (Fig 2). Within matrilines, we also detected an interaction of F1 age at breeding and

F3 sex, whereby the negative effect of grandmothers’ age at breeding was stronger for F3 males than for F3 females. In patrilines, we also detected an F2 age × F2 sex interaction, such that F3 life span declined more steeply with increasing paternal (F2 male) age than with increasing maternal (F2 female) age. S1 Fig shows the combined effects of F1 and F2 breeding ages, F1 competitive environment (patrilines only), and F1 larval diet on F3 life span. Results were qualitatively similar for models including development time and body size (S4 Table). Overall, by comparison with previously published life span estimates for this species when maintained as individually housed virgin adults (e.g., male median = 37 d, female median =

36 d; [49]), the median lifespans of F3 individuals descended from young-breeding parents and grandparents are similar (male median = 25, female median = 36), whereas the median lifespans of F3 individuals descended from old-breeding parents and grandparents are substantially lower (male median = 10, female median = 15).

Parental breeding age effects offspring lifespan

Table 1. Tests of effects based on linear mixed models of F3 life span for patrilines and matrilines. Significant effects are highlighted in bold.

Negative effects of F1 and F2 age indicate that old grandparents and parents produced F3 individuals with reduced lifespans, negative effects of larval diet

indicate that low-nutrient larval diet has a negative effect on F3 life span, and negative effects of sex indicate that the life span of male descendants was lower than that of females. Effect sizes represent marginal R2. Conditional whole-model R2 values were 47.72% for the patriline model and 54.78% for the matriline model.

Effects on F3 lifespan Patrilines Matrilines

Fixed effects: Estimate SE F Χ2 P Effect size (%) Estimate SE F Χ2 P Effect size (%)

(Intercept) 81.958 6.956 - 138.809 <0.001 - 91.294 6.624 - 189.944 <0.001 -

F1 Larval diet -8.504 4.619 2.620 3.389 0.066 0.258 -6.437 4.182 1.700 2.369 0.124 2.97

F1 Age -22.325 5.247 20.227 18.106 <0.001 30.8 -20.256 4.414 25.154 21.058 <0.001 35.38

F2 Sex 8.321 5.374 1.316 2.397 0.122 5.26 1.566 4.980 0.428 0.099 0.753 0.030

F2 Age -0.948 0.155 40.983 37.404 <0.001 15.45 -1.177 0.157 46.448 56.317 <0.001 35.62

F3 Sex -17.846 4.359 16.712 16.759 <0.001 10.75 -32.761 4.551 45.070 51.818 <0.001 39.55

F1 Age  F2 Age 0.266 0.112 5.606 5.606 0.018 10.85 0.254 0.102 6.249 6.249 0.012 11.33

F1 Larval diet  F1 Age 3.482 3.460 1.013 1.013 0.314 1.31 0.793 2.518 0.099 0.099 0.753 0.0511

F1 Larval diet  F2 Sex -1.533 3.011 0.259 0.259 0.611 0.181 -1.068 2.605 0.168 0.168 0.682 0.090

F1 Age  F2 Sex -0.438 3.222 0.019 0.019 0.892 0.0151 -1.022 2.621 0.152 0.152 0.697 0.100

F2 Sex  F2 Age -0.205 0.103 3.957 3.957 0.047 4.29 -0.153 0.092 2.758 2.758 0.097 2.9

F1 Age  F3 Sex 3.796 2.732 1.931 1.931 0.165 1.55 5.361 2.362 5.150 5.150 0.023 2.99 24 24

Parental breeding age effects offspring lifespan

F2 Sex  F3 Sex -3.549 2.614 1.843 1.843 0.175 0.899 4.209 2.420 3.026 3.026 0.082 1.64

F2 Age  F3 Sex 0.120 0.079 2.351 2.351 0.125 2.44 0.425 0.080 28.587 28.587 <0.001 25.33

F1 Larval diet  F3 Sex 4.482 2.494 3.230 3.230 0.072 1.94 3.741 2.384 2.463 2.463 0.117 1.21

Mortality rate

Consistent with our results for lifespan, we found that baseline mortality rate (Gompertz bo parameter) of F3 individuals from both matrilines and

patrilines was affected positively and similarly by F1 age at breeding, but not affected by F1 larval diet (Fig 3). Individuals descended from grandparents

that bred at age 35 d had higher baseline mortality rates, regardless of F1 larval diet treatment (High condition Old = HO; Low condition Old = LO; patrilines b0 HO = -3.5, b0 LO = -3.6; matrilines b0 HO = -3.8, b0 LO = -3.7) than individuals descended from grandparents that bred at age 15 d (High

condition Young = HY; Low condition Young = LY; patrilines b0 HY = -4.4, b0 LY = -4.2; matrilines b0 HY = -4.6, b0 LY = -4.4). An effect of F1 age at breeding on the baseline mortality rate was supported by Kullback-Leibler discrepancy calibration (KLDC) values, which exceeded 0.98 for all

comparisons of b0 parameters for F3 descendants of young-breeding versus old-breeding F1 individuals within and across larval diet treatments in both patrilines and matrilines (Table S6, S8 ).

Parental breeding age effects offspring lifespan

Parental breeding age breeding effects lifespan Parental offspring

Fig 1. Effects of grandparental (F1) breeding age and larval diet on grand-offspring (F3) lifespan in patrilines and matrilines. The violin plot outline illustrates kernel probability density (width represents proportion of data located there). Within violin plots are boxplots with median and interquartile range to illustrate data distribution. Underlying data can be found in the Dryad Repository: https://doi.org/10.5061/dryad.2rbnzs7hw.

Parental breeding age effects offspring lifespan

Fig 2. Interaction between effects of grandparental and parental breeding ages on grand- offspring lifespan in patrilines and matrilines. Black lines represent the lifespans of F3 descendants of F1 individuals paired at 15 days of age, and red lines represent the lifespans of F3 descendants of F1 individuals paired at 35 days of age. Bars represent SEM. Underlying data can be found in the Dryad Repository: https://doi.org/10.5061/dryad.2rbnzs7hw.

Parental breeding age effects offspring lifespan

Fig 3. Effects of grandparental larval diet and breeding age on estimated age-specific survival and mortality rates for grand-offspring of patrilines and matrilines as fitted by the simple Gompertz mortality model. b0 is the baseline mortality rate (scale) parameter and b1 is the rate of actuarial ageing (shape) parameter. Posterior distributions are shown for b0 and b1 in the left panels. Panels on the right illustrate how these estimates translate to survival and mortality rates over time. The shaded areas in the survival plots represent 95% confidence intervals.

Parental breeding age effects offspring lifespan

Underlying data can be found in the Dryad Repository: https://doi.org/10.5061/dryad.2rbnzs7hw.

Grandparental and parental breeding ages interacted in their effects on F3 baseline mortality rates

(b0), particularly within patrilines (Fig 4). F3 individuals descended from young grandparents (F1) experienced increasingly high baseline mortality as parental (F2) age at breeding increased, and this effect was especially strong in patrilines (Table S10, S11). By contrast, for F3 individuals descended from old-breeding grandparents, there were no consistent effects of parental age at breeding.

For actuarial ageing rates (Gompertz b1 parameter), evidence of treatment effects was weaker, and patterns were less consistent. Individuals descended from grandparents that bred at age 35 days had similar actuarial ageing rates, regardless of F1 larval diet treatment (patrilines b1 HO = 0.032, b1 LO =

0.036; matrilines b1 HO = 0.031, b1 LO = 0.029), to individuals descended from grandparents that bred at age 15 d (patrilines b1 HY = 0.032, b1 LY = 0.029; matrilines b1 HY = 0.035, b1 LY = 0.034). In matrilines, KLDC values were < 0.85 for all comparisons of b1 parameters for F3 descendants of young-breeding versus old-breeding F1 females (S8 Table). In patrilines, KLDC values marginally exceeded 0.85 for some comparisons of F3 descendants of young-breeding versus old-breeding F1 males within and across larval diet treatments, but the effect of F1 age at breeding on b1 was not consistent across larval diet treatments (S7 Table). There was little evidence that grandparental and parental ages at breeding interacted in their effects on actuarial ageing rate (b1) in either matrilines or patrilines (Fig 4). Wider confidence limits for lifespan and age-dependent mortality rates for descendants of old-breeding F1 males reflect reduced sample size resulting from mortality between 15 and 35 days of age. For all other KLDC values of group comparisons refer to S2-S5 Figs and S9-

Parental breeding age effects offspring lifespan

S13 Tables.

Fig 4. Effects of F1 breeding age and F2 breeding age on estimated age-specific survival and mortality rates for grand-offspring of patrilines and matrilines as fitted by the simple

Parental breeding age effects offspring lifespan

Gompertz mortality model. b0 is the baseline mortality rate (scale) parameter and b1 is the rate of actuarial ageing (shape) parameter. Posterior distributions are shown for b0 and b1 in the left panels. Panels on the right illustrate how these estimates translate to survival and mortality rates over time. Shaded areas in the survival plots represent 95% confidence intervals. Underlying data can be found in the Dryad Repository: https://doi.org/10.5061/dryad.2rbnzs7hw.

Discussion

A recent model suggests that negative effects of parental age on offspring performance can readily evolve [50], but many aspects of such effects have received little attention in empirical research. Our results show that paternal age effects can be similar in magnitude to maternal age effects. The magnitude of the grand-maternal and grand-paternal effects detected in our study is comparable to longevity changes observed in multi-generational selection experiments in Drosophila melanogaster [51,52]. Our mortality rate analyses suggest that decreased lifespan of grand-offspring of older grandparents and parents results largely from elevated baseline mortality rather than from a higher rate of increase in mortality rate with age (i.e. actuarial ageing). Actuarial ageing could result from the accumulation of somatic damage with age [53]. Previous studies of T. angusticollis showed that males reared on a high-nutrient larval diet accumulated damage more rapidly with age than males reared on a low-nutrient larval diet [46] and exhibited more rapid actuarial and reproductive ageing [47]. Here, we show that declining offspring longevity and increasing offspring mortality rate represent additional manifestations of ageing in T. angusticollis males and females. However, breeding age effects on offspring lifespan and mortality were unaffected by grandparental larval diet. Interestingly, while we found largely similar effects of grandpaternal versus grandmaternal and paternal versus maternal ages at breeding on offspring baseline mortality rate, we also found some evidence of effects on actuarial ageing rate in patrilines but not in matrilines. These differences suggest that male and female breeding age effects could be mediated by different factors and could have different effects on offspring life history.

Our findings suggest that the effect of ancestors’ age at breeding could contribute substantially to within-population variation in longevity. However, the importance of these effects in natural populations remains unclear. T. angusticollis has a much shorter mean lifespan in the wild than in

Parental breeding age effects offspring lifespan the laboratory, and wild males also exhibit very rapid actuarial ageing [49]. The short average lifespan and rapid ageing observed in natural populations of this species is consistent with findings for other insects in the wild [54–56]. Given the very high background mortality rate experienced by T. angusticollis in the wild, it is possible that longevity of flies in natural populations is not strongly affected by parental age effects. However, it is also possible that maternal and paternal age effects are accelerated along with the overall rate of ageing in wild populations as a result of environmental stresses such as parasites and temperature fluctuations. If so, then parental age effects could have a substantial effect on fitness in natural populations, despite a short life expectancy. It is also possible that offspring of old-breeding parents or grandparents might respond by increasing their early-life reproductive effort, thereby partly mitigating the effects of reduced lifespan. For example, in Daphnia pulex, older mothers produce offspring with shortened lifespans but these offspring achieve increased early-life reproductive output [33]. We found little evidence that age at breeding effects on lifespan were mediated by body size or development time, since inclusion of these traits as covariates in lifespan models did not qualitatively alter the results.

The grandparental and parental age effects that we observed could be mediated by the accumulation of germline mutations with age. Because male and female germline cells develop differently in animals, including flies [57–59], the male and female germlines could accumulate mutations at different rates [60,61]. In particular, the rate of age-dependent mutation accumulation is likely to reflect the number of germline cell divisions, and it has long been thought that males transmit more germline mutations because the male germline undergoes a larger number of cell divisions [62]. Interestingly, however, in Drosophila, the number of germline cell divisions is larger in females than in males at young ages, but larger in males than in females at old ages [63]. This suggests that mutation-mediated maternal and paternal age effects could differ in relative magnitudes as a function of male and female age. If T. angusticollis exhibits a similar pattern of germline cell division to Drosophila, this could explain the somewhat stronger negative effect of grand-paternal age at breeding on grand-offspring lifespan, relative to the effect of grand-maternal age at breeding (Fig 1).

The rate of cell proliferation in the female germline also increases on a protein-rich diet in D. melanogaster [64], and dietary protein strongly stimulates female fecundity in T. angusticollis as well [45]. A protein-rich adult diet could therefore be expected to accentuate negative maternal breeding

Parental breeding age effects offspring lifespan age effects on offspring performance and could also accentuate paternal breeding age effects if cell division in the male germline is also enhanced on a high-protein diet. Germline mutation rate can also be affected by investment in DNA repair, and D. melanogaster reared on low nutrient food as larvae have lower rates of repair that result in increased germline mutation rate [65]. However, we found little evidence of effects of F1 larval diet on grand-offspring mortality and survival (Figs. 2, 4). Likewise, we did not detect an effect of male competitive environment (opportunity for combat interactions) or any interaction between this treatment and grand-paternal breeding age. This finding is consistent with the lack of any effect of male combat on male reproductive ageing [47] and suggests that agonistic interactions with other males do not affect the maintenance of the male germline.

A different (but non-exclusive) explanation for our findings is age-dependent transmission of epigenetic or cytoplasmic factors through the female and male germlines. DNA (cytosine) methylation contributes to the regulation of gene expression in many organisms [66], but flies have little cytosine methylation and its role in this group remains unclear [67–70]. In D. melanogaster, DNA methylation is largely limited to the early stages of embryogenesis [71,72], but two studies suggest that DNA methylation can also persist in the germline [73,74]. In mammals, DNA methylation patterns undergo changes with age throughout the genome [75,76]. Such age-related changes in methylation (known as the ‘epigenetic clock’) could mediate parental age effects, since some DNA methylation patterns can be transmitted to offspring via both sperm and eggs (for a review, see [77]). It is not known whether a DNA methylation ‘clock’ also occurs in flies.

Other epigenetic or cytoplasmic factors that change with age could also mediate the observed age-at- breeding effects. There is evidence of age-related cellular changes in the male and female germline. For example, as Drosophila males age, germline stem cells (GSCs) divide less frequently due to misorientation of centromeres [78]. Similarly, GSC division in female Drosophila declines with age, and this is accompanied by an increased rate of cell death in developing eggs [79]. RNA-mediated transmission of shortened telomeres could mediate breeding age effects in flies and other animals. Shortened telomeres are associated with cellular senescence in some taxa [80], and telomere length can be affected by non-coding telomeric repeat-containing RNAs (TERRA), which are transcriptionally active in Drosophila [81]. TERRAs are present in animal (including human)

Parental breeding age effects offspring lifespan oocytes [82] and, in female Drosophila, they affect blastoderm formation [83]. Other types of non- coding RNAs could also be involved. Flies maintain chromosome length through retro- transcription [84], which requires complex and specific chromatin structures [85]. Retrotransposon proliferation can promote mutagenesis [86]. RNA interference (RNAi) mechanisms control the silencing of retrotransposons in germline cells [87,88], and parental age effects could be mediated by the transmission of such small non-coding RNAs, with effects on chromatin states and gene expression in embryos [23]. Early development in Drosophila is thought to be governed by maternally inherited RNAs and proteins [89], but less is known about the effects of male-derived RNAs on offspring development. While T. angusticollis males do not transmit nutritional nuptial gifts during copulation [90], males probably transfer a variety of micro-RNAs in the ejaculate. The complement of seminal and egg micro-RNAs could change with male and female age and affect embryo development.

Another possibility is that flies change their investment in gametes in response to the age or mating experience of their partner. A female may decrease investment per offspring when mated to an older male, while a male may reduce the quality or quantity of accessory gland proteins or sperm produced when mated with an older female, resulting in negative effects of parental age on offspring performance. Such responses to mate quality have been reported in Drosophila and other insects [91–94] and might be mediated through cuticular hydrocarbons (CHCs) that are known to change with age in flies [95,96]. In our experiment, increasing age was also associated with increasing mating experience. Individuals of both sexes might alter their investment in offspring based on their partner’s mating experience, because previously mated males might transfer smaller or lower-quality ejaculates. For example, male mating experience was negatively correlated to nuptial gift quality and sperm number in a bush cricket [97], and female reproductive output was lower when mated with sexually experienced males than when mating with virgin males across 25 species of Lepioptera [98]. Although T. angusticollis males appear to be able to replenish their ejaculate reserves very rapidly, the effects of age and mating experience cannot be decoupled statistically in our data, and require further investigation.

We quantified effects of ancestors’ age at breeding in flies (F3) maintained as virgins in individual containers and supplied with ad libitum food and water. Housing T. angusticollis individuals in

Parental breeding age effects offspring lifespan isolation and as virgins tends to increase their longevity (e.g. [99])), whereas ad libitum availability of dietary protein tends to reduce adult longevity [45]. While our results suggest that larval diet and male competitive environment do not interact strongly with breeding age in affecting longevity of descendants, further work is required to determine whether housing, reproduction or adult diet of descendants can interact with effects of parental and grandparental ages at breeding.

Some individuals failed to produce viable offspring, or did not survive to breed at older ages, and we cannot exclude the possibility that differential mortality or reproductive success biased the composition of our treatment groups. In particular, because T. angusticollis males reared on a nutrient-rich larval diet tend to exhibit an elevated adult mortality rate relative to males reared on a nutrient-poor larval diet [47], fewer F1 focal males from the rich-diet treatment survived to breed at age 35 days, resulting in a smaller sample size for that treatment combination. This resulted in somewhat wider confidence limits for lifespan and actuarial ageing rate for the F3 descendants of those males, but we cannot exclude the possibility that the elevated F1 mortality was also associated with differential natural selection on males reared on nutrient-rich vs. nutrient-poor larval diets.

The interactive effects of grand-parental and parental ages at breeding that we observed suggest that the factors mediating these effects are stable across at least two generations. Priest et al. [5] suggested that parental age effects could play a role in the evolution of ageing by contributing to age-related decline in performance and generating selection for earlier reproduction. Bonduriansky and Day [34] argued that, if such effects can accumulate over generations, an environmental change that brings about delayed breeding or causes a more rapid decline in offspring performance with parental age could result in a progressive decline in performance over several generations, resulting in phenotypic changes that resemble the evolution of accelerated ageing. Our results support these ideas by providing experimental evidence that parental age effects can have large effects on descendants’ longevity, can occur in both matrilines and patrilines and across contrasting environments, and can be transmitted over at least two generations. Further work is needed to understand the context-dependence and fitness consequences of such effects in natural populations.

Materials and Methods

Parental breeding age effects offspring lifespan

Source of experimental flies

Experiments were performed using a lab-reared stock of T. angusticollis that originated from individuals collected from Fred Hollows Reserve, Randwick, NSW, Australia (33°54′44.04″S 151°14′52.14″E). This stock was maintained as a large, outbred population with overlapping generations, and periodically supplemented with wild-caught individuals from the same source population to maintain genetic diversity.

Larval rearing and diet manipulation

All larvae were reared in climate chambers at 25°C ± 2°C with a 12:12 photoperiod and moistened with deionised water every two days. We manipulated the quantity of resources available to larvae during development by rearing flies on either a high-nutrient, standard-nutrient or low-nutrient larval diet. Diets were based on [100] and were selected to generate considerable body size differences between treatment groups while minimising larval mortality, and to preserve the protein to carbohydrate ratio of ~ 1:3 across diets. All diets consisted of a base of 170g of coco peat moistened with 600mL of reverse osmosis-purified water. The high-nutrient larval diet consisted of 32.8g of protein (Nature’s Way soy protein isolate; Pharm-a-Care, Warriewood, Australia) and 89g of brown sugar (Woolworths Essentials Bonsucro® brand); the standard larval diet consisted of 10.9g of protein and 29.7g sugar; the low-nutrient larval diet consisted of 5.5g of protein and 14.8g sugar. These nutrients were mixed into the cocopeat and water using a hand-held blender and frozen at -

20°C until the day of use. Males and females of the F1 generation were reared on either a high or low- nutrient larval diet and standardised for larval density (40 eggs per 200g of larval food). All larvae of the F2 and F3 generations were reared on a standard larval diet (see [45] for further details). Following the first adult emergence from each larval container, adult flies were collected for 10 days, and the rest were discarded.

F1 Adult housing and competitive environment

F1 males were subjected to a “low” or “high” competition environment (Fig 5). Each adult focal male was paired with a competitor male reared on a standard larval diet inside an enclosure containing a petri dish with larval medium (which stimulates territory defence behaviours in T. angusticollis males). Males in the “high” competition environment were able to move freely around the arena and

Parental breeding age effects offspring lifespan engage in combat interactions with the competitor male, whereas males in the “low” competition environment were separated by mesh so that they could perceive the competitor’s chemical and perhaps visual cues but have no physical contact. All focal F1 females were kept in a similar housing as the “low” competitive environment males where each focal female was paired with a female reared on a standard larval diet. All housing containers had a layer of moistened cocopeat on the bottom, and dishes of oviposition medium (on which adult flies also feed) to stimulate ovary development in females.

F1 Adult male and female age-at-breeding manipulation

The age at breeding was manipulated for F1 focal individuals by pairing at ‘young’ (15 ± 1 days old) and ‘old’ (35 ± 1 days old) ages with an opposite-sex individual reared on the standard larval diet and standardized for age (15 ± 1 days old). These ages were selected because, in T. angusticollis, adults become fully reproductively mature by 10-15 days of age under laboratory conditions, while median longevity of individually housed, captive flies is 37 days for males and 36 days for females, and mortality rate begins to increase rapidly in both sexes after 30 days of age [49]. Thus, at 15 days old, both sexes are considered to be at their prime while, at 35 days old, both sexes are well past their prime. Each focal F1 adult was thus paired twice, each time with a different mate, to produce broods of F2 offspring at ‘young’ and ‘old’ ages (Fig 5). Mating pairs were kept in 60 mL glass vials under standardised light and temperature (~23°C) for one hour, and females were then placed into 250 mL enclosures with mesh coverings and a moistened cocopeat substrate and were allowed to oviposit for 96h into a petri dish containing oviposition medium. After 48h a fresh oviposition dish was provided. 20 eggs were sampled randomly from each female and transferred to 100g of standard larval medium.

F2 Adult male and female age-at-breeding manipulation

One F2 male and one F2 female focal individual was randomly sampled for breeding from each F1 larval container. Thus, where possible, each F1 focal individual contributed one F2 offspring of each sex from a reproductive bout at 15 days of age and one F2 offspring of each sex from a reproductive bout at 35 days of age. Each F2 focal individual was paired with a partner of the opposite sex (raised on a standard diet and 15 ± 1 days old on the day of pairing) at four ages (where possible): 15d, 30d,

Parental breeding age effects offspring lifespan

45d and 60d. The flies were allowed 1 hour to mate, after which eggs were collected from each female and maintained as described above.

F3 Rearing and quantification of lifespan

From each reproductive bout of each F2 individual, one male and female of the F3 generation were obtained (where possible) and housed individually in a 120mL container fitted with a feeding tube containing a sugar-yeast mixture and drinking tube containing water (with both food and water provided ad libitum), and a substrate of moistened cocoa peat to maintain humidity. F3 housing containers were maintained at ambient room temperature (23 ± 4°C) and checked daily for mortality until all individuals had died. To minimize spatial effects, containers were randomly moved to different locations every two days.

For all focal individuals, development time and body size were also recorded to investigate their possible roles in mediating treatment effects on lifespan and mortality rate (refer to S1, S2 Tables for summary statistics). All F1 and F2 focal individuals were frozen at -20°C after their final reproductive bout (or prior natural death before day 60), and all F3 individuals were frozen after natural death.

For all focal F1, F2 and F3 individuals, egg to adult development time was recorded as time from oviposition to adult emergence in days (± 1 day). Thorax length is a reliable proxy for body size in this species [101] and was measured for each F1, F2 and F3 focal individual from images taken using a Leica MS5 stereoscope equipped with a Leica DFC420 digital microscope camera. Measurements were made using FIJI open source software [102].

Parental breeding age effects offspring lifespan

Fig 5. Experimental design: Patrilines (a) consist of descendants of F1 males, while matrilines (b) consist of descendants of F1 females. F1 individuals were reared on either a high- or low-nutrient larval diet. Adult F1 males were also maintained in high- or low- competition social environments (S4 Table). F1 males and females were then mated at 15 days or 35 days of age and all offspring (F2) were reared on a standard larval diet. From each F1 breeding bout, one male and one female of the F2 generation were paired with a standard mate at 15d intervals up to 60 days of age. Grand-offspring (F3) were all reared on standard larval diet and housed individually until death. Sample sizes (number of F1 or F2 focal individuals that produced offspring, and number of F3 individuals for which longevity was quantified) are shown for each combination of treatment and sex.

Parental breeding age effects offspring lifespan

Lifespan analysis

We investigated treatment effects on F3 lifespan using R 3.3.2 [103] and the package “lme4” [104]. These analyses facilitate hypothesis testing by making it possible to test interactions within mixed- effects models. Because the lifespan of every individual was known, no censoring was required. Gaussian linear mixed models (LMM) were used, and all analyses were carried out separately for matrilines (i.e. descendants of F1 females) and patrilines (i.e. descendants of F1 males). Any effects of

F1 age at breeding, larval diet or male competitive environment therefore represent grand-maternal effects within matrilines and grand-paternal effects within patrilines. Within both matrilines and patrilines, we tested for effects of F2 age at breeding for both female parents (maternal age effects) and male parents (paternal age effects) and compared effects on F3 males and females (i.e., effect of

F3 sex). For the patriline dataset, (F1) male competitive environment and its 2-way interactions were tested by a likelihood ratio test (LRT) and were found to have no effect on any dependent variables.

The patriline models were then re-fitted without F1 competitive environment. This resulted in identical model structure for patrilines and matrilines, facilitating comparison of matrilineal and patrilineal results. Qualitatively identical results are obtained without F1 competitive environment as a predictor in the patriline models (S3 Table).

Our final models thus included F1 (grandparental) larval diet and age at breeding, F2 parental age at breeding, F2 sex and F3 sex as fixed effects. F2 breeding age was fitted as a continuous predictor, while the other factors were fitted as categorical predictors. F1 and F2 individual ID, replicate F1 larval container, and emergence date were included as random effects. We also fitted models with F1, F2 and F3 body sizes and development times as fixed covariates in order to determine whether these traits mediate treatment effects on F3 lifespan (S4 Table). Treatment effects on F3 body size and development time were also tested using similar models to those described above, and results of those analyses are shown in S6 and S7 Figs, S14 and S15 Tables, and discussed in S1 Text. Estimates and F-ratios were obtained using the packages “lme4” [104] and “lmerTest” [105], while p-values were obtained via “Type 3” likelihood ratio tests using the package “car”. To examine the relative effect size of each predictor, we also quantified marginal R2, which is variance explained by fixed factors, and conditional-whole model R2 that includes variation explained by random factors from our linear mixed models using the methods developed in [106].

Parental breeding age effects offspring lifespan

Mortality rate analysis

To gain a better understanding of treatment effects on F3 lifespan, we also investigated effects on F3 mortality rates. We used the Bayesian Survival Trajectory Analysis, implemented with the package “BaSTA” [107]. BaSTA utilises a Bayesian approach based on Markov Chain Monte Carlo (MCMC) estimation of age-specific mortality rate distributions. Our data are uncensored and the date of adult emergence is known for all individuals, allowing us to obtain reliable population estimates of the mortality distribution [108]. In order to find the mortality rate distribution that best fits our data, we first used the package “flexsurv” [109] on a combined dataset comprising both patrilines and matrilines. We compared the simple and Makeham versions of the Gompertz and Weibull models, as well as the logistic and exponential models, using the Akaike Information Criterion (AIC). This analysis showed that a simple Gompertz distribution provided the best fit to our data (S5 Table). Mortality rate was therefore modeled as:

푏0+ 푏1 푥 휇푏(푥|푏) = 푒

Survival probability was modeled as:

푒푏0 푏1푥 푆푏(푥|푏) = 푒푥푝 [ (1 − 푒 )] 푏1

The Gompertz mortality rate function includes a scale parameter, b0 (often called the “baseline mortality rate”), and a shape parameter, b1, that describes the dependency of mortality on age (푥) and is often interpreted as the rate of actuarial ageing, which reflects the rate of increase in mortality rate with age [110–113].

We used BaSTA to estimate and compare parameters of the simple Gompertz model for our experimental treatment groups. We performed four parallel BaSTA simulations, each proceeding for 2, 200, 000 iterations, with a burn-in of 200,000 chains, and took an MCMC chain sample every 4,000 iterations. Our models generated parameter estimates that converged with low serial autocorrelations (<5%) and robust posterior distributions of bo and b1 (N = 2,000) allowing for robust comparisons between treatment groups.

Parental breeding age effects offspring lifespan

We compared parameter estimates for various treatment groups based on differences between their posterior distributions, using the Kullback-Leibler divergence calibration (KLDC) implemented in BaSTA. Values near 0.5 suggest nominal differences between distributions, whereas values close to 1 indicate a sizeable divergence. KLDC thresholds can vary depending on interpretation and can range between 0.65 and 1 [114–116]. We considered a relatively conservative KLDC value > 0.85 to indicate a difference between the posterior distributions of the treatment groups being compared.

We report Gompertz bo parameter estimates on a log scale and refer to F1 treatment combinations as High-nutrient/Old-breeding (HO), High-nutrient/Young-breeding (HY), Low-nutrient/Old- breeding (LO), and Low-nutrient/Young-breeding (LY). Data are deposited in the Dryad repository: https://doi.org/10.5061/dryad.2rbnzs7hw [117].

Author contributions

F.S, A.H, Z.W, A.M. and R.B designed the study. Z.W, F.S. and A.H. performed the experiment. Z.W and R.B analysed the data and wrote the manuscript. F.S., A.H. and A.M contributed to data interpretation, visualisation, and editing of the manuscript.

Acknowledgements

The authors would thank E. Macartney, N. W. Burke and F. Zajitschek for their helpful comments and discussions of earlier drafts of the manuscript. This research was funded by the Australian Research Council through a Future Fellowship and Discovery Grant to RB.

Data archiving

The data will be deposited in Dryad.

References

1. Lansing AI. A transmissible, cumulative and reversible factor in aging. Gerontology. 1947;2: 228–239.

2. Ducatez S, Baguette M, Stevens VM, Legrand D, Fréville H. Complex Interactions Between

Parental breeding age effects offspring lifespan

Paternal And Maternal Effects: Parental Experience And Age At Reproduction Affect Fecundity And Offspring Performance In A Butterfly. Evolution. 2012;66: 3558–3569. doi:10.1111/j.1558-5646.2012.01704.x

3. Gavrilov LA, Gavrilova NS. Parental age at conception and offspring longevity. Rev Gerontol. 1997;7: 5–12.

4. Mousseau T, Fox CW. Maternal effects as adaptations. New York: Oxford University Press; 1998.

5. Priest N, Mackowiak B, Promislow DEL. The role of parental age effects on the evolution of aging. Evolution. 2002;56: 927–935. doi:10.1111/j.0014-3820.2002.tb01405.x

6. Schroeder J, Nakagawa S, Rees M, Mannarelli M-E, Burke T. Reduced fitness in progeny from old parents in a natural population. Proc Natl Acad Sci U S A. 2015;112: 4021–5. doi:10.1073/pnas.1422715112

7. Mousseau T. Maternal Effects In Insect Life Histories. Annu Rev Entomol. 1991;36: 511– 534. doi:10.1146/annurev.ento.36.1.511

8. Hercus MJ, Hoffmann AA. Maternal and grandmaternal age influence offspring fitness in Drosophila. Proc R Soc B Biol Sci. 2000;267: 2105–2110. doi:10.1098/rspb.2000.1256

9. Koch RE, Phillips JM, Camus MF, Dowling DK. Maternal age effects on fecundity and offspring egg-to-adult viability are not affected by mitochondrial haplotype. Ecol Evol. 2018;8: 10722–10732. doi:10.1002/ece3.4516

10. Lippens C, Faivre B, Lechenault C, Sorci G. Aging parasites produce offspring with poor fitness prospects. Biol Lett. 2017;13. doi:10.1098/rsbl.2016.0888

11. Fay R, Barbraud C, Delord K, Weimerskirch H. Paternal but not maternal age influences early-life performance of offspring in a long-lived seabird. Proc R Soc B Biol Sci. 2016;283. doi:10.1098/rspb.2015.2318

12. Kirkwood TB, Austad SN. Why do we age? Nature. 2000;408: 233–238.

Parental breeding age effects offspring lifespan

13. Wong WSW, Solomon BD, Bodian DL, Kothiyal P, Eley G, Huddleston KC, et al. New observations on maternal age effect on germline de novo mutations. Nat Commun. 2016;7: 1–10. doi:10.1038/ncomms10486

14. Horvath S. DNA methylation age of human tissues and cell types. Genome Biol. 2013;14. doi:10.1186/gb-2013-14-10-r115

15. Koch CM, Wagner W. Epigenetic-aging-signature to determine age in different tissues. Aging. 2011;3: 1018–1027. doi:100395 [pii]

16. Stubbs TM, Bonder MJ, Stark AT, Krueger F, Bolland DJ, Butcher GA, et al. Multi-tissue DNA methylation age predictor in mouse. Genome Biology. 2017.

17. Hannum G, Guinney J, Zhao L, Zhang L, Hughes G, Sadda SV, et al. Genome-wide Methylation Profiles Reveal Quantitative Views of Human Aging Rates. Mol Cell. 2013;49: 359–367. doi:10.1016/j.molcel.2012.10.016

18. Bocklandt S, Lin W, Sehl ME, Sanchez FJ, Sinsheimer JS, Horvarth S, et al. Epigenetic Predictor of Age. PLoS One. 2011;6. doi:10.1371/Citation

19. Curley JP, Mashoodh R, Champagne FA. Epigenetics and the origins of paternal effects. Horm Behav. 2011;59: 306–14. doi:10.1016/j.yhbeh.2010.06.018

20. Crews D, Gore AC, Hsu TS, Dangleben NL, Spinetta M, Schallert T, et al. Transgenerational epigenetic imprints on mate preference. Proc Natl Acad Sci U S A. 2007;104: 5942–5946. doi:10.1002/andp.19654710109

21. Franklin TB, Russig H, Weiss IC, Grff J, Linder N, Michalon A, et al. Epigenetic transmission of the impact of early stress across generations. Biol Psychiatry. 2010;68: 408– 415. doi:10.1016/j.biopsych.2010.05.036

22. Holeski LM, Jander G, Agrawal AA. Transgenerational defense induction and epigenetic inheritance in plants. Trends Ecol Evol. 2012;27: 618–626. doi:10.1016/j.tree.2012.07.011

23. Pauli A, Rinn JL, Schier AF. Non coding RNAs regulation in embryogenesis. Nat Rev

Parental breeding age effects offspring lifespan

Genet. 2011;12: 136–149. doi:10.1038/nrg2904.

24. Chen Q, Yan W, Duan E. Epigenetic inheritance of acquired traits through sperm RNAs and sperm RNA modifications. Nat Rev Genet. 2016;17: 733. Available: https://doi.org/10.1038/nrg.2016.106

25. Szyf M. Nongenetic inheritance and transgenerational epigenetics. Trends Mol Med. 2015;21: 134–144. doi:10.1016/j.molmed.2014.12.004

26. Díaz H, Esponda P. Ageing-induced changes in the cortical granules of mouse eggs. Zygote. 2004;12: 95–103. doi:10.1017/S0967199404002680

27. Rodgers AB, Morgan CP, Leu NA, Bale TL. Transgenerational epigenetic programming via sperm microRNA recapitulates effects of paternal stress. Proc Natl Acad Sci. 2015;112: 13699–13704. doi:10.1073/pnas.1508347112

28. Immler S. The sperm factor: paternal impact beyond genes. Heredity. 2018;121: 239–247. doi:10.1038/s41437-018-0111-0

29. Chen Q, Yan M, Cao Z, Li X, Zhang Y, Shi J, et al. Sperm tsRNAs contribute to intergenerational inheritance of an acquired metabolic disorder. Science (80- ). 2016;351: 397–400. doi:10.1126/science.aad7977

30. Klosin A, Lehner B. Mechanisms, timescales and principles of trans-generational epigenetic inheritance in animals. Curr Opin Genet Dev. 2016;36: 41–49. doi:10.1016/j.gde.2016.04.001

31. Blake GET, Watson ED. Unravelling the complex mechanisms of transgenerational epigenetic inheritance. Curr Opin Chem Biol. 2016;33: 101–107. doi:10.1016/j.cbpa.2016.06.008

32. Giron D, Casas J. Mothers reduce egg provisioning with age. Ecol Lett. 2003;6: 273–277. doi:10.1046/j.1461-0248.2003.00429.x

33. Plaistow SJ, Shirley C, Collin H, Cornell SJ, Harney ED. Offspring provisioning explains

Parental breeding age effects offspring lifespan

clone-specific maternal age effects on life history and life span in the water flea, Daphnia pulex. Am Nat. 2015;186: DOI: 10.1086/682277. doi:10.1086/682277

34. Bonduriansky R, Day T. Extended Heredity: A New Understanding of Inheritance and Evolution. Press PU, editor. 2018.

35. Crean A, Bonduriansky R. What is a paternal effect? Trends Ecol Evol. 2014;29: 554–559. doi:10.1016/j.tree.2014.07.009

36. Lock JE. Transgenerational effects of parent and grandparent gender on offspring development in a biparental beetle species. Biol Lett. 2012;8: 408–411. doi:10.1098/rsbl.2011.0920

37. Xie K, Ryan DP, Pearson BL, Henzel KS, Neff F, Vidal RO, et al. Epigenetic alterations in longevity regulators, reduced life span, and exacerbated aging-related pathology in old father offspring mice. Proc Natl Acad Sci. 2018;115: E2348–E2357. doi:10.1073/pnas.1707337115

38. Sharma R, Agarwal A, Rohra VK, Assidi M, Abu-Elmagd M, Turki RF. Effects of increased paternal age on sperm quality, reproductive outcome and associated epigenetic risks to offspring. Reprod Biol Endocrinol. 2015;13: 1–20. doi:10.1186/s12958-015-0028-x

39. Saha S, Barnett AG, Foldi C, Burne TH, Eyles DW, Buka SL, et al. Advanced paternal age is associated with impaired neurocognitive outcomes during infancy and childhood. PLoS Med. 2009;6: 0303–0311. doi:10.1371/journal.pmed.1000040

40. Sandin S, Schendel D, Magnusson P, Hultman C, Surén P, Susser E, et al. Autism risk associated with parental age and with increasing difference in age between the parents. Mol Psychiatry. 2016;21: 693–700. doi:10.1038/mp.2015.70

41. Lynch M, Ennis R. Resource availability, maternal effects, and longevity. Exp Gerontol. 1983;18: 147–165. doi:10.1016/0531-5565(83)90008-6

42. Gribble KE, Jarvis G, Bock M, Mark Welch DB. Maternal caloric restriction partially rescues the deleterious effects of advanced maternal age on offspring. Aging Cell. 2014;13:

Parental breeding age effects offspring lifespan

623–630. doi:10.1111/acel.12217

43. Carnes BA, Riesch R, Schlupp I. The delayed impact of parental age on offspring mortality in Mice. J Gerontol A Biol Sci Med Sci. 2012;67 A: 351–357. doi:10.1093/gerona/glr116

44. Runagall-McNaull A, Bonduriansky R, Crean AJ. Dietary protein and lifespan across the metamorphic boundary: protein-restricted larvae develop into short-lived adults. Sci Rep. 2015;5.

45. Adler MI, Cassidy EJ, Fricke C, Bonduriansky R. The lifespan-reproduction trade-off under dietary restriction is sex-specific and context-dependent. Exp Gerontol. 2013;48: 539–548. doi:10.1016/j.exger.2013.03.007

46. Adler MI, Telford M, Bonduriansky R. Phenotypes optimized for early-life reproduction exhibit faster somatic deterioration with age, revealing a latent cost of high condition. J Evol Biol. 2016;29: 2436–2446. doi:10.1111/jeb.12968

47. Hooper AK, Spagopoulou F, Wylde Z, Maklakov AA, Bonduriansky R. Ontogenetic timing as a condition-dependent life history trait: High-condition males develop quickly, peak early and age fast. Evolution. 2017; 1–15. doi:10.1111/evo.13172

48. Macartney E, Crean A, Bonduriansky R. Adult dietary protein has age- and context- dependent effects on male post-copulatory performance. J Evol Biol. 2017;

49. Kawasaki N, Brassil CE, Brooks R, Bonduriansky R. Environmental effects on the expression of lifespan and aging: An extreme contrast between wild and captive cohorts of Telostylinus angusticollis (Diptera: Neriidae). Am Nat. 2008;172: 2379–2388.

50. Moorad JA, Nussey DH. Evolution of maternal effect senescence. Proc Natl Acad Sci. 2016;113: 362–367. doi:10.1073/pnas.1520494113

51. Zwaan B, Bijlsma R, Hoekstra RE. Direct selection on life span in Drosophila melanogaster. Evolution. 1995;49: 649–659.

52. Rose MR. Laboratory evolution of postponed senescence in Drosophila melanogaster.

Parental breeding age effects offspring lifespan

Evolution. 1984;38: 1004–1010.

53. Kirkwood TBL, Holliday R. The Evolution of Ageing and Longevity. Proc R Soc B Biol Sci. 1979;205: 531–546. doi:10.1098/rspb.1979.0083

54. Bonduriansky R, Brassil CE. Rapid and costly ageing in wild male flies. Nature. 2002;420: 2002.

55. Zajitschek F, Brassil CE, Bonduriansky R, Brooks RC. Sex effects on life span and senescence in the wild when dates of birth and death are unknown. Ecology. 2009;90: 1698–1707. doi:10.1890/08-0048.1

56. Rodriguez-Munoz R, Bretman A, Slate J, Walling CA, Tregenza T. Natural and Sexual Selection in a Wild Insect Population. Science. 2010;328: 1269–1272. doi:10.1126/science.1188102

57. Danserau DA, Lasko PF. Germline Stem Cells. 2008;450. doi:10.1007/978-1-60327-214-8

58. Hennig W. Spermatogenesis in Drosophila. Int J Dev Biol. 1996;40: 167–176.

59. Bastock R, St Johnston D. Drosophila oogenesis. Curr Biol. 2008;18: 1082–1087. doi:10.1016/j.cub.2008.09.011

60. Gao J-J, Pan X-R, Hu J, Ma L, Wu J-M, Shao Y-L, et al. Highly variable recessive lethal or nearly lethal mutation rates during germ-line development of male Drosophila melanogaster. Proc Natl Acad Sci. 2011;108: 15914–15919. doi:10.1073/pnas.1100233108

61. Bachtrog D. Evidence for male-driven evolution in Drosophila. Mol Biol Evol. 2008;25: 617–619. doi:10.1093/molbev/msn020

62. Haldane JBS. The rate of spontaneous mutation of a human gene. J Genet. 1935;31: 317– 326. doi:10.1007/BF02982403

63. Bauer VL, Aquadro CF. Rates of DNA Sequence Evolution are Not Sex-Biased in Drosophila melanogaster and D. simulans. Mol Biol Evol. 1997;14: 1252–1257.

Parental breeding age effects offspring lifespan

64. Drummond-Barbosa D, Spradling AC. Stem cells and their progeny respond to nutritional changes during Drosophila oogenesis. Dev Biol. 2001;231: 265–278. doi:10.1006/dbio.2000.0135

65. Agrawal AF, Wang AD. Increased transmission of mutations by low-condition females: Evidence for condition-dependent DNA repair. PLoS Biol. 2008;6: 0389–0395. doi:10.1371/journal.pbio.0060030

66. Jung M, Pfeifer GP. Aging and DNA methylation. BMC Biol. 2015;13. doi:10.1186/s12915-015-0118-4

67. Regev A, Lamb MJ, Jablonka E. The Role of DNA Methylation in Invertebrates: Developmental Regulation\ror Genome Defense? Mol Biol Evol. 1998;15: 880–891.

68. Mandrioli M, Borsatti F. DNA methylation of fly genes and transposons. Cell Mol Life Sci. 2006;63: 1933–1936. doi:10.1007/s00018-006-6039-1

69. Field LM, Lyko F, Mandrioli M, Prantera G. DNA methylation in insects. Insect Mol Biol. 2004;13: 109–115. doi:10.1111/j.0962-1075.2004.00470.x

70. Lyko F, Ramsahoye BH, Jaenisch R. DNA methylation in Drosophila melanogaster. Nature. 2000;408.

71. Ferres-Marco D, Gutierrez-Garcia I, Vallejo DM, Bolivar J, Gutierrez-Aviño FJ, Dominguez M. Epigenetic silencers and Notch collaborate to promote malignant tumours by Rb silencing. Nature. 2006;439: 430–436. doi:10.1038/nature04376

72. Xing Y, Shi S, Le L, Lee CA, Silver-Morse L, Li WX. Evidence for transgenerational transmission of epigenetic tumor susceptibility in Drosophila. PLoS Genet. 2007;3: 1598– 1606. doi:10.1371/journal.pgen.0030151

73. Singh ND, Arndt PF, Petrov DA. Genomic heterogeneity of background substitutional patterns in Drosophila melanogaster. Genetics. 2005;169: 709–722. doi:10.1534/genetics.104.032250

Parental breeding age effects offspring lifespan

74. Keller I, Bensasson D, Nichols RA. Transition-transversion bias is not universal: A counter example from grasshopper pseudogenes. PLoS Genet. 2007;3: 0185–0191. doi:10.1371/journal.pgen.0030022

75. Horvarth S. DNA methylation age of human tissues and cell types. Genome Biol. 2013;14.

76. Law JA, Jacobsen SE. Establishing , maintaining and modifying DNA methylation patterns in plants and animals. Nat Rev Genet. 2010;11: 204–220. doi:10.1038/nrg2719

77. Ho DH, Burggren WW. Epigenetics and transgenerational transfer: a physiological perspective. J Exp Biol. 2010;213: 3–16. doi:10.1242/jeb.019752

78. Cheng J, Türkel N, Hemati N, Fuller MT, Hunt AJ, Yamashita YM. Centrosome misorientation reduces stem cell division during ageing. Nature. 2008;456: 599–604. doi:10.1038/nature07386

79. Zhao R, Xuan Y, Li X, Xi R. Age-related changes of germline stem cell activity, niche signaling activity and egg production in Drosophila. Aging Cell. 2008;7: 344–354. doi:10.1111/j.1474-9726.2008.00379.x

80. Harley CB. Telomerase is not an oncogene. Oncogene. 2002;21: 494–502. doi:10.1038/sj/onc/1205076

81. Azzalin, C.M., Reichenbach, P., Khoriauli, L., Giulotto E. Telomeric repeat containing RNA and RNA surveillance factors at mammalian chromosome ends. Science (80- ). 2007;318: 798–801.

82. Reig-Viader R, Brieño-Enríquez MA, Khoriauli L, Toran N, Cabero L, Giulotto E, et al. Telomeric repeat-containing RNA and telomerase in human fetal oocytes. Hum Reprod. 2013;28: 414–422. doi:10.1093/humrep/des363

83. Olovnikov IA, Morgunova V V, Mironova AA, Kordyukova MY, Radion EI, Olenkina OM, et al. Interaction of Telomeric Retroelement HeT A Transcripts and Their Protein Product Gag in Early Embryogenesis of Drosophila. Bioschemistry (Mosc). 2016;81. doi:10.1134/S000629791609011X

Parental breeding age effects offspring lifespan

84. Pardue M-L, DeBaryshe PG. Retrotransposons Provide an Evolutionarily Robust Non- Telomerase Mechanism to Maintain Telomeres. Annu Rev Genet. 2003;37: 485–511. doi:10.1146/annurev.genet.38.072902.093115

85. Mason JM, Frydrychova RC, Biessmann H. Drosophila telomeres: an exception providing new insights. Bioessays. 2008;30: 25–37. doi:10.3816/CLM.2009.n.003.

86. Eickbush TH, Furano A V. Fruit flies and humans respond differently to retrotransposons. Curr Opin Genet Dev. 2002;12: 669–674. doi:10.1016/S0959-437X(02)00359-3

87. Vagin V V., Sigova A, Li C, Seitz H, Gvozdev V, Zamore PD. A distinct small RNA pathway silences selfish genetic elements in the germline. Science. 2006;313: 320–324. doi:10.1126/science.1129333

88. Yin H, Lin H. An epigenetic activation role of Piwi and a Piwi-associated piRNA in Drosophila melanogaster. Nature. 2007;450: 304–308. doi:10.1038/nature06263

89. Salvany L, Aldaz S, Corsetti E, Azpiazu N. A new role for hth in the early pre-blastodermic divisions in Drosophila. Cell Cycle. 2009;8: 2748–2755. doi:10.4161/cc.8.17.9388

90. Bonduriansky R, Head M. Maternal and paternal condition effects on offspring phenotype in Telostylinus angusticollis (Diptera: Neriidae). J Evol Biol. 2007;20: 2379–2388. doi:10.1111/j.1420-9101.2007.01419.x

91. Avent TD, Price TAR, Wedell N. Age-based female preference in the fruit fly Drosophila pseudoobscura. Anim Behav. 2008;75: 1413–1421. doi:10.1016/j.anbehav.2007.09.015

92. Fox CW. The influence of maternal age and mating frequency on egg size and offspring performance in Callosobruchus maculatus (Coleoptera: Bruchidae). Oecologia. 1993;96: 139–146. doi:10.1007/BF00318042

93. Mcnamara KB, van Lieshout E, Jones TM, Simmons LW. Age-dependent trade-offs between immunity and male, but not female, reproduction. J Anim Ecol. 2013;82: 235– 244. doi:10.1111/j.1365-2656.2012.02018.x

Parental breeding age effects offspring lifespan

94. Gray DA. Intrinsic factors affecting female choice in house crickets: Time cost, female age, nutritional condition, body size, and size-relative reproductive investment. J Insect Behav. 1999;12: 691–700. doi:10.1023/A:1020983821436

95. Desena MS, Clark JM, Edman JD, Symington SB, Scott TW, Clark GG, et al. Potential for aging female Aedes aegypti (Diptera: Culicidae) by gas chromatographic analysis of cuticular hydrocarbons, including a field evaluation. J Med Entomol. 1999;36: 811–823.

96. Braga MV, Pinto ZT, de Carvalho Queiroz MM, Blomquist GJ. Effect of age on cuticular hydrocarbon profiles in adult Chrysomya putoria (Diptera: Calliphoridae). Forensic Sci Int. 2016;259: e37–e47. doi:10.1016/j.forsciint.2015.11.006

97. Wedell N, Ritchie MG. Male age, mating status and nuptial gift quality in a bushcricket. Anim Behav. 2004;67: 1059–1065. doi:10.1016/j.anbehav.2003.10.007115.

98. Torres-Vila LM, Jennions MD. Male mating history and female fecundity in the Lepidoptera: Do male virgins make better partners? Behav Ecol Sociobiol. 2005;57: 318– 326. doi:10.1007/s00265-004-0857-7

99. Adler, MI, Bonduriansky, R. The dissimilar costs of love and war: age-specific mortality as a function of the operational sex ratio. Journal of Evolutionary Biology. 2011;24: 1169-1177. doi: 10.1111/j.1420-9101.2011.02250.x

100. Sentinella AT, Crean AJ, Bonduriansky R. Dietary protein mediates a trade-off between larval survival and the development of male secondary sexual traits. Funct Ecol. 2013;27: 1134–1144.

101. Bonduriansky R. Convergent evolution of sexual shape dimorphism in Diptera. J Morphol. 2006;267: 602–611.

102. Schindelin J, Arganda-Carreras, I. Frise E, Kaynig V, Longair M, Pietzsch T, Preibisch S, et al. Fiji: an open-source platform for biological-image analysis. Nat Methods. 2012;9: 676– 682.

103. Team RC. R: A language and environment for statistical computing. R Foundation for

Parental breeding age effects offspring lifespan

Statistical Computing, Vienna, Austria. [Internet]. 2012. Available: http://www.r- project.org/

104. Bates D, Maechler M, Bolker B, Walker S. Fitting Linear Mixed-Effects Models Using lme4. Journal of Statistical Software. J Stat Softw. 2015;67: 1–48.

105. Kuznetsova A, Brockhoff PB, Christensen RHB. {lmerTest} Package: Tests in Linear Mixed Effects Models. J Stat Softw. 2017;82: 1–26. doi:10.18637/jss.v082.i13

106. Nakagawa S, Schielzeth H. A general and simple method for obtaining R2 from generalized linear mixed-effects models. Methods Ecol Evol. 2013;4: 133–142. doi:10.1111/j.2041- 210x.2012.00261.x

107. Colchero F, Jones OR, Rebke M. BaSTA: an R package for Bayesian estimation of age- specific mortality from incomplete mark-recapture/recovery data with covariates. Methods Ecol Evol. 2012; 466–470.

108. Colchero F, Clark JS. Bayesian inference on age-specific survival for censored and truncated data. J Anim Ecol. 2012;81: 139–149.

109. Jackson C. flexsurv: A Platform for Parametric Survival Modeling in R. J Stat Softw. 2016;70: 1–33. doi:10.18637/jss.v070.i08

110. Zou S, Carey JR, Liedo P, Ingram DK, Yu B. Prolongevity effects of a botanical with oregano and cranberry extracts in Mexican fruit flies: Examining interactions of diet restriction and age. Age. 2012;34: 269–279. doi:10.1007/s11357-011-9230-8

111. Carey JR, Liedo P. Sex-specific life table aging rates in large medfly cohorts. Exp Gerontol. 1995;30: 315–325.

112. Styer LM, Carey JR, Wang J-L, Scott TW. Mosquitoes do sensece: Departure from the paradigm of constant mortality. Am J Trop Med Hyg. 2007;76: 111–117. doi:10.1016/0168-5597(85)90001-2

113. Wu D, Cysper JR, Yashin AI, Johnson TE. Multiple Mild Heat-Shocks Decrease The

Parental breeding age effects offspring lifespan

Gompertz Component of Mortality in C. elegans. Exp Gerontol. 2009;44: 607–612. doi:10.1016/j.pestbp.2011.02.012.

114. Polačik M, Tomášek O, Kačer P, Methling C, Vrtílek M, Terzibasi Tozzini E, et al. Repeated intraspecific divergence in life span and aging of African annual fishes along an aridity gradient. Evolution. 2016;71: 386–402. doi:10.1111/evo.13127

115. Boonekamp JJ, Salomons M, Bouwhuis S, Dijkstra C, Verhulst S. Reproductive effort accelerates actuarial senescence in wild birds: an experimental study. Ecol Lett. 2014;17: 599–605. doi:10.1111/ele.12263

116. Zajitschek F, Jin T, Colchero F, Maklakov AA. Aging differently: Diet-and sex-dependent late-life mortality patterns in Drosophila melanogaster. J Gerontol A Biol Sci Med Sci. 2014;69: 666–674. doi:10.1093/gerona/glt158

117. Wylde Z (2019) Data from: Parental breeding age effects on descendants’ longevity interact over two generations in matrilines and patrilines. Dryad Digital Repository Openly available via https://doi.org/10.5061/dryad.2rbnzs7hw

Parental breeding age effects offspring lifespan

Supplementary information

Effects on lifespan and mortality

S1 Figure. Combined Effects of F1,F2 breeding ages and F1 larval diet quality on mean F3 lifespan. Black lines represent F3 individuals descended from young F1 individuals (15 days) and red lines represent F3 individuals descended from old (35 days) F1 individuals. In patrilines only, individuals descended from F1 males that were subjected to either a high or low competitive environment are represented by a solid or dotted line, respectively. F3 grand-offspring of F1 grandparents reared on a high-nutrient larval diet are represented by a circle, and those of F1 grandparents reared on a low-nutrient larval diet are represented by a triangle. Values on the horizontal axis represent F2 ages at breeding. All points represent means. Bars represent SEM. Underlying data can be found in the Dryad Repository: https://doi.org/10.5061/dryad.2rbnzs7hw.

Parental breeding age effects offspring lifespan

S1 Table. Factorial summary of means and standard deviations for F3 lifespan (days), development time (days) and thorax length (mm) for patrilines. NA refers to instances where there was only one observation of the treatment combination.

F1 breeding age F1 larval diet F2 sex F2 breeding Mean F3 Mean F3 Mean F3 thorax age (days) lifespan (± sd) development length (± sd) time (± sd) 15 days High-nutrient Female 15 49 ± (16.7) 28 ± (2.5) 2.7 ± (0.3) 15 days High-nutrient Female 30 31 ± (14.8) 29 ± (3.9) 2.7 ± (0.3) 15 days High-nutrient Female 45 23 ± (18.9) 27 ± (4.7) 2.7 ± (0.2) 15 days High-nutrient Female 60 16 ± (14.8) 27 ± (1.8) 2.4 ± (0.2) 15 days High-nutrient Male 15 44 ± (15.6) 28 ± (4.1) 2.7 ± (0.30) 15 days High-nutrient Male 30 33 ± (12.2) 29 ± (3.5) 2.8 ± (0.2) 15 days High-nutrient Male 45 16 ± (13.0) 30 ± (2.1) 2.8 ± (0.3) 15 days High-nutrient Male 60 7 ± (1.4) 31 ± (NA) 2.7 ± (0.2) 15 days Low-nutrient Female 15 42 ± (18.8) 29 ± (3.4) 2.7 ± (0.2) 15 days Low-nutrient Female 30 31 ± (12.9) 30 ± (4.4) 2.7 ± (0.3) 15 days Low-nutrient Female 45 20 ± (17.3) 28 ± (2.8) 2.7 ± (0.2) 15 days Low-nutrient Female 60 15 ± (9.1) 29 ± (3.6) 2.6 ± (0.2) 15 days Low-nutrient Male 15 39 ± (20.6) 28 ± (2.6) 2.7 ± (0.3) 15 days Low-nutrient Male 30 35 ± (18.9) 30 ± (2.7) 2.7 ± (0.2) 15 days Low-nutrient Male 45 11 ± (9.7) 29 ± (2.1) 2.7 ± (0.2) 35 days High-nutrient Female 15 33 ± (14.8) 31 ± (1.4) 2.6 ± (0.3) 35 days High-nutrient Female 30 20 ± (13.8) 29 ± (1.5) 2.5 ± (0.2) 35 days High-nutrient Female 45 15 ± (9.6) 31 ± (1.9) 2.6 ± (0.2) 35 days High-nutrient Female 60 6 ± (3.6) 28 ± (0.6) 2.9 ± (0.06) 35 days High-nutrient Male 15 30 ± (NA) 37 ± (NA) 3.0 ± (NA) 35 days High-nutrient Male 30 18 ± (11.2) 28 ± (4.5) 2.7 ± (0.3) 35 days Low-nutrient Female 15 31 ± (14.4) 29 ± (1.4) 2.6 ± (0.3) 35 days Low-nutrient Female 30 17 ± (11.2) 30 ± (1.8) 2.6 ± (0.3) 35 days Low-nutrient Female 45 15 ± (6.1) 29 ± (2.6) 2.5 ± (0.3) 35 days Low-nutrient Female 60 18 ± (10.7) 30 ± (1.6) 2.7 ± (0.4) 35 days Low-nutrient Male 15 26 ± (13.0) 30 ±(4.1) 2.6 ± (0.3) 35 days Low-nutrient Male 30 19 ± (15.4) 30 ± (2.4) 2.6 ± (0.2) 35 days Low-nutrient Male 45 12 ± (7.3) 30 ± (2.4) 2.7 ± (0.3) 35 days Low-nutrient Male 60 6 ± (NA) - 2.6 ± (NA)

Parental breeding age effects offspring lifespan

S2 Table. Factorial summary of means and standard deviations for F3 lifespan (days), development time (days) and thorax length (mm) for matrilines. NA refers to instances where there was only one observation of the treatment combination.

F 1 larval diet F 2 sex F2 breeding Mean F3 Mean F3 Mean F3 thorax age (days) lifespan (± sd) development length (± sd) time (± sd) 15 days High-nutrient Female 15 48 ± (17.7) 29 ± (4.0) 2.606 ± (0.3) 15 days High-nutrient Female 30 33 ± (19.6) 30 ± (5.2) 2.757 ± (0.3) 15 days High-nutrient Female 45 27 ± (17.0) 29 ± (3.4) 2.581 ± (0.3) 15 days High-nutrient Female 60 12 ± (10.7) 28 ± (3.8) 2.563 ± (0.3) 15 days High-nutrient Male 15 47 ± (18.5) 29 ± (2.6) 2.584 ± (0.3) 15 days High-nutrient Male 30 33 ± (12.8) 29 ± (2.5) 2.63 ± (0.3) 15 days High-nutrient Male 45 18 ± (15.9) 30 ± (1.9) 2.535 ± (0.2) 15 days High-nutrient Male 60 18 ± (12.3) 28 ± (3.6) 2.647 ± (0.2) 15 days Low-nutrient Female 15 49 ± (15.4) 27 ± (3.0) 2.658 ± (0.3) 15 days Low-nutrient Female 30 30 ± (10.8) 27 ± (2.9) 2.603 ± (0.3) 15 days Low-nutrient Female 45 12 ± (8.5) 28 ± (3.0) 2.629 ± (0.3) 15 days Low-nutrient Female 60 12 ± (7.5) 30 ± (5.1) 2.545 ± (0.3) 15 days Low-nutrient Male 15 46 ± (14.7) 27 ± (2.1) 2.718 ± (0.3) 15 days Low-nutrient Male 30 30 ± (14.3) 29 ± (4.5) 2.701 ± (0.2) 15 days Low-nutrient Male 45 12 ± (11.9) 28 ± (1.3) 2.71 ± (0.2) 35 days High-nutrient Female 15 32 ± (14.0) 30 ± (4.1) 2.598 ± (0.3) 35 days High-nutrient Female 30 19 ± (13.9) 29 ± (2.4) 2.769 ± (0.3) 35 days High-nutrient Female 45 16 ± (9.4) 30 ± (4.2) 2.609 ± (0.3) 35 days High-nutrient Female 60 16 ± (11.4) 31 ± (3.8) 2.712 ± (0.4) 35 days High-nutrient Male 15 36 ± (14.1) 31 ± (3.0) 2.69 ± (0.3) 35 days High-nutrient Male 30 18 ± (9.5) 29 ± (2.2) 2.604 ± (0.3) 35 days High-nutrient Male 45 14 ± (15.5) 30 ± (2.5) 2.754 ± (0.1) 35 days High-nutrient Male 60 31 ± (NA) 26 ± (NA) 2.763 ± (0.4) 35 days Low-nutrient Female 15 31 ± (16.8) 28 ± (1.9) 2.715 ± (0.2) 35 days Low-nutrient Female 30 16 ± (11.8) 30 ± (2.5) 2.583 ± (0.2) 35 days Low-nutrient Female 45 19 ± (9.3) 30 ± (1.9) 2.443 ± (0.3) 35 days Low-nutrient Female 60 16 ± (12.5) 30 ± (1.8) 2.527 ± (0.5) 35 days Low-nutrient Male 15 31 ± (13.8) 30 ± (2.6) 2.606 ± (0.4) 35 days Low-nutrient Male 30 11 ± (10.7) 29 ± (1.8) 2.54 ± (0.3)

Parental breeding age effects offspring lifespan

35 days Low-nutrient Male 45 11 ± (5.6) 29 ± (2.8) 2.508 ± (0.2)

S3 Table. Linear mixed effects models of F3 lifespan for patrilines including F1

competitive environment. Negative effects for F1 larval diet indicate that grandparents reared on a high-nutrient larval diet produced grand-offspring with a relatively longer lifespan than

descendants of grandparents reared on a low-nutrient larval diet. Negative effects of F1 and F2 age

indicate that old grandparents and parents produced F3 individuals with reduced lifespans,

negative effects of larval diet indicate that low-nutrient larval diet has a negative effect on F3 lifespan, and negative effects of sex indicate that the lifespan of male descendants was lower than that of females.. Significance codes: 0.0001 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1.

Effects on F3 lifespan

Fixed effects: Estimate SE

(Intercept) 83.072 6.989***

F1 Larval diet -8.718 4.596.

F1 Age -22.536 5.236***

F2 Sex 8.383 5.366

F2 Age -0.946 0.155***

F3 Sex -18.023 4.365***

F1 Competitive environment -1.739 1.257

F1 Age  F2 Age 0.266 0.112*

F1 Larval diet  F1 Age 3.621 3.439

F1 Larval diet  F2 Sex -1.530 3.005

F1 Age  F2 Sex -0.208 3.216

Parental breeding age effects offspring lifespan

F2 Sex  F2 Age -0.209 0.103*

F1 Age  F3 Sex 3.827 2.733

F2 Sex  F3 Sex -3.783 2.621

F2 Age  F3 Sex 0.119 0.079

F1 Larval diet  F3 Sex 4.718 2.502.

S4 Table. Linear mixed effects models of F3 lifespan for patrilines and matrilines, with thorax length and development time of all focal individuals included as covariates.

Negative effects for F1 larval diet indicate that grandparents reared on a high-nutrient larval diet produced grand-offspring with a relatively longer lifespan than descendants of grandparents reared on a low-nutrient larval diet. Negative effects of F1 and F2 age indicate that old grandparents and parents produced F3 individuals with reduced lifespans, negative effects of larval diet indicate that low-nutrient larval diet has a negative effect on F3 lifespan, and negative effects of sex indicate that the lifespan of male descendants was lower than that of females.. Significance codes: 0.0001 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1.

Effects on F3 lifespan Paternal Maternal

line line

Fixed effects: Estimate SE Estimate SE

(Intercept) 157.365 29.761*** 103.530 32.384.

F1 Larval diet -8.073 6.260 -6.017 5.544.

F1 Age -21.658 5.716*** -20.762 4.966***

F2 Sex 7.271 7.196 -0.293 6.079

Parental breeding age effects offspring lifespan

F2 Age -0.993 0.167*** -1.161 0.168***

F3 Sex -18.796 5.020* -31.215 4.942***

F1 Thorax length -4.181 4.059 -1.927 7.291

F2 Thorax length -0.056 6.252 4.286 4.833

F3 Thorax length -2.917 3.192 1.815 2.854

F1 Development time -0.856 0.660 -0.191 0.656

F2 Development time -0.596 0.576 -0.386 0.479

F3 Development time -0.521 0.205* -0.293 0.218

F1 Age  F2 Age 0.243 0.126. 0.268 0.112*

F1 Larval diet  F1 Age 3.330 4.261 -0.156 2.947

F1 Larval diet  F2 Sex 0.315 3.308 -0.343 2.873

F1 Age  F2 Sex -2.435 3.819 -0.882 3.001

F2 Sex  F2 Age -0.162 0.119 -0.139 0.101

F1 Age  F3 Sex 4.171 3.073 5.098 2.602.

F2 Sex  F3 Sex -1.285 2.905 3.931 2.637

F2 Age  F3 Sex 0.149 0.085. 0.380 0.086***

F1 Larval diet  F3 Sex 4.032 2.682 4.171 2.586

Parental breeding age effects offspring lifespan

S5 Table. Model selection results based on ‘flexsurv’ package. The simple Gompertz model provided the best fit to our data based on the Aikaike Information Criterion and was used for further analyses using BaSTA.

Distribution AIC ∆AIC

Gompertz 8629 0

Weibull 8754 125

Logistic 8920 166

Exponential 9185 265

S6 Table. Parameter estimates for each treatment group for the best fitting model

(Gompertz with simple shape) for grand-paternal effects of F1 larval diet X F1 breeding age.

Parameter Treatment group Est. Lower 95% Upper 95% Serial CI CI autocorrelation

b0 F1 High nutrient diet -4.401 -4.681 -4.121 0.003 Young breeding age (HY)

F1 High nutrient diet Old -3.503 -4.186 -2.888 -0.008 breeding age (HO)

F1 Low nutrient diet -4.170 -4.406 -3.931 -0.013 Young breeding age (LY)

F1 Low nutrient diet Old -3.560 -3.851 -3.274 0.009 breeding age (LO)

b1 F1 High nutrient diet 0.032 0.025 0.039 0.021 Young breeding age (HY)

Parental breeding age effects offspring lifespan

F1 High nutrient diet Old 0.032 0.003 0.061 0.010 breeding age (HO)

F1 Low nutrient diet 0.029 0.022 0.035 -0.032 Young breeding age (LY)

F1 Low nutrient diet Old 0.036 0.024 0.048 0.003 breeding age (LO)

Parental breeding age effects offspring lifespan

S2 Figure. Values of the KLDC for patrilines, comparing parameter posterior distributions between treatment groups. HY = High Nutrient Young Breeding treatment, HO = High Nutrient Old Breeding treatment, LY = Low Nutrient Young Breeding treatment, LO =

Parental breeding age effects offspring lifespan

Low Nutrient Old Breeding treatment. Underlying data can be found in the Dryad Repository: https://doi.org/10.5061/dryad.2rbnzs7hw.

S7 Table. Mean Kullback-Leibler discrepancy calibration values (KLDC) for patrilines, comparing parameter posterior distributions between F1 treatment groups.

Comparison b0 b1

High nutrient larval diet Old 0.996 0.897 breeding age (HO) – High nutrient larval diet Young breeding age (HY)

Low nutrient larval diet Young 0.899 0.716 breeding age (LY) – High nutrient larval diet Young breeding age (HY)

Low nutrient larval diet Young 0.981 0.913 breeding age (LY) – High nutrient larval diet Old breeding age (HO)

Low nutrient larval diet Old 1.000 0.714 breeding age (LO) – High nutrient larval diet Young breeding age (HY)

Low nutrient larval diet Old 0.749 0.789 breeding age (LO) – High nutrient larval diet Old breeding age (HO)

Low nutrient larval diet Old 0.999 0.891 breeding age (LO) – Low nutrient larval diet Young breeding age (LY)

Parental breeding age effects offspring lifespan

S8 Table. Parameter estimates for each treatment group for the best fitting model (Gompertz with

Simple shape) for grand-maternal effects of F1 larval diet X F1 breeding age.

Parameter Treatment group Est. Lower 95% Upper 95% Serial CI CI autocorrelation

b0 F1 High nutrient diet -4.556 -4.856 0.042 -0.008 Young breeding age (HY)

F1 High nutrient diet Old -3.752 -4.060 0.042 0.007 breeding age (HO)

F1 Low nutrient diet -4.362 -4.760 0.044 0.011 Young breeding age (LY)

F1 Low nutrient diet Old -3.667 -4.010 0.041 0.042 breeding age (LO)

b1 F1 High nutrient diet 0.035 0.027 0.958 0.002 Young breeding age (HY)

F1 High nutrient diet Old 0.031 0.020 0.042 -0.008 breeding age (HO)

F1 Low nutrient diet 0.034 0.022 0.042 0.007 Young breeding age (LY)

F1 Low nutrient diet Old 0.029 0.016 0.044 0.011 breeding age (LO)

Parental breeding age effects offspring lifespan

S3 Figure. Values of the KLDC for matrilines, comparing parameter posterior distributions between our treatment groups. HY = High Nutrient Young Breeding treatment, HO = High Nutrient Old Breeding treatment, LY = Low Nutrient Young Breeding treatment, LO

Parental breeding age effects offspring lifespan

= Low Nutrient Old Breeding treatment. Underlying data can be found in the Dryad Repository: https://doi.org/10.5061/dryad.2rbnzs7hw.

S9 Table. Mean Kullback-Leibler discrepancy calibration values (KLDC) for matrilines, comparing parameter posterior distributions between F1 treatment groups.

Comparison b0 b1

High nutrient larval diet Old 1.000 0.666 breeding age (HO)– High nutrient larval diet Young breeding age (HY)

Low nutrient larval diet Young 0.751 0.585 breeding age (LY) – High nutrient larval diet Young breeding age (HY)

Low nutrient larval diet Young 0.997 0.538 (LY) breeding age – High nutrient larval diet Old breeding age (HO)

Low nutrient larval diet Old 1.000 0.801 breeding age (LO) – High nutrient larval diet Young breeding age (HY)

Low nutrient larval diet Old 0.564 0.547 breeding age (LO) – High nutrient larval diet Old breeding age (HO)

Low nutrient larval diet Old 0.999 0.626 breeding age – Low nutrient larval diet Young breeding age (LY)

Parental breeding age effects offspring lifespan

S10 Table. Parameter estimates for each treatment group for the best fitting model (Gompertz with simple shape) for effects of F1 breeding age X F2 breeding age in patrilines.

Parameter Treatment group Est. Lower 95% Upper 95% Serial CI CI autocorrelation

b0 F1 Young breeding age F2 -5.657 -6.069 -5.255 -0.043 Young breeding age (YY)

F1 Young breeding age F2 -4.597 -4.920 -4.286 -0.008 Old breeding age (YO)

F1 Young breeding age F2 -3.106 -3.363 -2.864 -0.013 Very Old breeding age (YVO)

F1 Old breeding age F2 -4.758 -5.523 -4.040 0.009 Young breeding age (OY)

F1 Old breeding age F2 -3.437 -3.855 -3.035 0.004 Old breeding age (OO)

F1 Old breeding age F2 -3.595 -4.076 -3.135 -0.015 Very Old breeding age (OVO)

b1 F1 Young breeding age F2 0.053 0.045 0.062 -0.033 Young breeding age (YY)

F1 Young breeding age F2 0.043 0.034 0.051 -0.003 Old breeding age (YO)

F1 Young breeding age F2 0.012 0.001 0.023 -0.013 Very Old breeding age YVO)

Parental breeding age effects offspring lifespan

F1 Old breeding age F2 0.060 0.036 0.083 0.008 Young breeding age (OY)

F1 Old breeding age F2 0.031 0.012 0.048 0.007 Old breeding age (OO)

F1 old breeding age F2 0.077 0.048 0.104 -0.011 very old breeding age

Parental breeding age effects offspring lifespan

S4 Figure. Values of the KLDC for patrilines, comparing parameter posterior distributions between treatment groups. YO = Young F1 breeding age Old F2 breeding age treatment, YY = Young F1 breeding age Young F2 breeding age, YVO = Young F1 breeding age Very old F2 breeding age, OY = Old F1 breeding age Young F2 breeding age, OO = Old F1 breeding age

Old F2 breeding age. Underlying data can be found in the Dryad Repository: https://doi.org/10.5061/dryad.2rbnzs7hw.

Parental breeding age effects offspring lifespan

S11 Table. Mean Kullback-Leibler discrepancy calibration values (KLDC) for patrilines comparing parameter posterior distributions between treatment groups.

Comparison b0 b1

F1 Young breeding age F2 1.000 0.977

Old breeding age (YO) - F1

Young breeding age F2 Young breeding age (YY)

F1 Young breeding age F2 1.000 1.000 Very Old breeding age (YVO)

- F1 Young breeding age F2 Young breeding age (YY)

F1 Young breeding age F2 1.000 1.000 Very Old breeding age (YVO)

- F1 Young breeding age F2 Old breeding age (YO)

F1 Old breeding age F2 0.988 0.862

Young breeding age (OY) - F1

Young breeding age F2 Young breeding age (YY)

F1 Old breeding age F2 0.806 0.949

Young breeding age (OY) - F1

Young breeding age F2 Old breeding age (YO)

F1 Old breeding age F2 1.000 1.000

Young breeding age (OY) - F1

Young breeding age F2 Very Old breeding age (YVO)

Parental breeding age effects offspring lifespan

F1 Old breeding age F2 Old 1.000 0.992 breeding age (OO) - F1

Young breeding age F2 Young breeding age (YY)

F1 Old breeding age F2 Old 1.000 0.925 breeding age (OO) - F1

Young breeding age F2 Old breeding age (YO)

F1 Old breeding age F2 Old 0.934 0.974 breeding age (OO) - F1

Young breeding age F2 Very Old breeding age (YVO)

F1 Old breeding age F2 Old 1.000 0.985 breeding age (OO) - F1 Old breeding age F2 Young breeding age (OY)

F1 Old breeding age F2 Very 1.000 0.967

Old breeding age (OVO) - F1

Young breeding age F2 Young breeding age (YY)

F1 Old breeding age F2 Very 1.000 0.993

Old breeding age (OVO) - F1

Young breeding age F2 Old breeding age (YO)

F1 Old breeding age F2 Very 0.977 1.000

Old breeding age (OVO) - F1

Young breeding age F2 Very Old breeding age (YVO)

Parental breeding age effects offspring lifespan

F1 Old breeding age F2 Very 0.998 0.789

Old breeding age (OVO) - F1

Old breeding age F2 Young breeding age (OY)

F1 Old breeding age F2 Very 0.617 0.999

Old breeding age (OVO) - F1

Old breeding age F2 Old breeding age (OO)

S12 Table. Parameter estimates for each treatment group for the best fitting model (Gompertz with Simple shape) for effects of F1 breeding age X F2 breeding age in matrilines.

Parameter Treatment group Est. Lower 95% Upper 95% Serial CI CI autocorrelation

b0 F1 Young breeding -4.691 -5.095 -4.312 0.018

age F2 Young breeding age (YY)

F1 Young breeding -4.091 -4.521 -3.678 -0.027

age F2 Old breeding age (YO)

F1 Young breeding -3.885 -4.277 -3.529 0.000

age F2 Very Old breeding age (YVO)

F1 Old breeding age -3.921 -4.323 -3.529 -0.042

F2 Young breeding age (OY)

Parental breeding age effects offspring lifespan

F1 Old breeding age -4.057 -4.507 -3.668 0.044

F2 Old breeding age (OO)

F1 Old breeding age -3.633 -4.038 -3.242 -0.003

F2 Very Old breeding age (OVO)

b1 F1 Young breeding 0.037 0.027 0.046 0.007

age F2 Young breeding age (YY)

F1 Young breeding 0.028 0.015 0.040 -0.008

age F2 Old breeding age (YO)

F1 Young breeding 0.032 0.019 0.045 -0.032

age F2 Very Old breeding age (YVO)

F1 Old breeding age 0.027 0.014 0.040 -0.043

F2 Young breeding age (OY)

F1 Old breeding age 0.034 0.020 0.048 0.023

F2 Old breeding age (OO)

F1 Old breeding age 0.018 0.004 0.029 -0.017

F2 Very Old breeding age (OVO)

Parental breeding age effects offspring lifespan

S5 Figure. Values of the KLDC for matrilines, comparing parameter posterior distributions between our treatment groups. YO = Young F1 breeding age Old F2 breeding age treatment, YY = Young F1 breeding age Young F2 breeding age, YVO = Young F1 breeding age Very old F2 breeding age, OY = Old F1 breeding age Young F2 breeding age, OO = Old F1 breeding age

Old F2 breeding age. Underlying data can be found in the Dryad Repository:https://doi.org/10.5061/dryad.2rbnzs7hw.

Parental breeding age effects offspring lifespan

S13 Table. Mean Kullback-Leibler discrepancy calibration values (KLDC) for matrilines, comparing parameter posterior distributions between our treatment groups.

Comparison b0 b1

F1 Young breeding age F2 0.992 0.875

Old breeding age (YO) - F1

Young breeding age F2 Young breeding age (YY)

F1 Young breeding age F2 0.999 0.677 Very Old breeding age (YVO)

- F1 Young breeding age F2 Young breeding age (YY)

F1 Young breeding age F2 0.705 0.604 Very Old breeding age (YVO)

- F1 Young breeding age F2 Old breeding age (YO)

F1 Old breeding age F2 0.999 0.875

Young breeding age (OY) - F1

Young breeding age F2 Young breeding age (YY)

F1 Old breeding age F2 0.643 0.503

Young breeding age (OY) - F1

Young breeding age F2 Old breeding age (YO)

F1 Old breeding age F2 0.509 0.606

Young breeding age (OY) - F1

Young breeding age F2 Very Old breeding age (YVO)

Parental breeding age effects offspring lifespan

F1 Old breeding age F2 Old 0.995 0.607 breeding age (OO) - F1

Young breeding age F2 Young breeding age (YY)

F1 Old breeding age F2 Old 0.506 0.689 breeding age (OO) - F1

Young breeding age F2 Old breeding age (YO)

F1 Old breeding age F2 Old 0.655 0.523 breeding age (OO) - F1

Young breeding age F2 Very Old breeding age (YVO)

F1 Old breeding age F2 Old 0.597 0.686 breeding age (OO) - F1 Old breeding age F2 Young breeding age (OY)

F1 Old breeding age F2 Very 1.000 0.997

Old breeding age (OVO) - F1

Young breeding age F2 Young breeding age (YY)

F1 Old breeding age F2 Very 0.957 0.859

Old breeding age (OVO) - F1

Young breeding age F2 Old breeding age (YO)

F1 Old breeding age F2 Very 0.781 0.957

Old breeding age (OVO) - F1

Young breeding age F2 Very Old breeding age (YVO)

Parental breeding age effects offspring lifespan

F1 Old breeding age F2 Very 0.819 0.838

Old breeding age (OVO) - F1

Old breeding age F2 Young breeding age (OY)

F1 Old breeding age F2 Very 0.938 0.974

Old breeding age (OVO) - F1

Old breeding age F2 Old breeding age (OO)

Effects on body size and development time

In both matrilines and patrilines we observed significant effects of F3 sex on F3 body size (Table

S13). In the matrilineal dataset however, an interaction between F1 larval diet and F1 age was also observed, such that grand-mothers reared on a poor larval diet and bred at a ‘young’ age produced larger grand-offspring. In the patrilineal dataset we also detected an interaction of F2 age and F2 sex

(Table S13), such that F3 body size increased as a mothers’ breeding age increased, whereas F3 offspring that descended from older fathers showed a significant decrease in body size (Fig. S6).

Parental breeding age effects offspring lifespan

Figure S6. Effects of F2 breeding age and F2 sex on F3 body size in patrilines. Solid grey lines represent F3 individuals descended from F2 females and solid black lines represent F3 individuals descended from F2 males. Bars represent SEM. Underlying data can be found in the Dryad

Repository: https://doi.org/10.5061/dryad.2rbnzs7hw.

S14 Table. Linear mixed effects model of F3 body size. Significant effects are highlighted in bold.

Effects on F3 body size Patrilines Matrilines

Fixed effects: Estimate SE Estimate SE

(Intercept) -9.434 e- -0.770 0.287** 02 2.791e-01

F1 Larval diet -0.091 0.211 -4.976 1.940*e-01

F1 Age 0.384 0.296 -3.418 2.866e-01

F2 Sex -0.260 0.306 -4.127 2.749e-01

Parental breeding age effects offspring lifespan

F2 Age 0.002 0.006 -7.754 5.985e-03

F3 Sex 2.022***e- 0.942 0.192*** 9.411 01

F1 Age  F2 Age -0.011 0.006. 9.124 6.189e-03

F1 Larval diet  F1 Age 0.261 0.216 5.617 1.935**e-01

F1 Larval diet  F2 Sex -0.082 0.176 2.583 1.867e-01

F1 Age  F2 Sex -0.219 0.194 1.922 1.903e-01

F2 Sex  F2 Age 0.013 0.006* 7.124 6.433e-03

F1 Age  F3 Sex 0.316 0.128* 1.226 1.272e-01

F2 Sex  F3 Sex 0.171 0.119 2.210 1.273e-01

F2 Age  F3 Sex -0.001 0.003 2.886 4.164e-03

F1 Larval diet  F3 Sex 0.021 0.112 6.915 1.258e-01

∗P-value < 0.05, ∗∗P-value < 0.01, ∗∗∗P-value < 0.001.

Parental breeding age effects offspring lifespan

S7 Figure. Effects of F1 larval diet and age at breeding on F3 body size in patrilines and matrilines. Solid grey lines represent effects of F1 individuals reared on reared on a low-nutrient larval diet and solid black lines represent the effects of F1 individuals reared on a rich or high-nutrient larval diet. Bars represent SEM. Underlying data can be found in the Dryad Repository: https://doi.org/10.5061/dryad.2rbnzs7hw.

Parental breeding age effects offspring lifespan

For F3 development time we did not detect any significant effects apart from a marginally significant interaction between F2 sex and F2 age within patrilines (Table S14) suggesting that paternal breeding age has a direct influence on offspring developmental trajectory. Offspring descended from old fathers had a much shorter development time than offspring descended from old mothers (Fig. S6).

S15 Table. Linear mixed effects model of F3 development time. Significant effects are highlighted in bold. Solid black lines represent the development time of F3 offspring descended from F2 males and solid grey lines, individuals derived from F2 females. Bars represent SEM. SEM for descendants of F2 females at 60 days is missing due to small sample size.

Effects on F3 development Patrilines Matrilines time

Fixed effects: Estimate SE Estimate SE

(Intercept) 31.015 1.128*** 29.090 0.902***

F1 Larval diet -0.413 0.834 -0.607 0.658

F1 Age -2.279 1.177. -0.465 0.955

F2 Sex -1.803 1.225 1.383 0.943

F2 Age -0.007 0.023 0.025 0.019

F3 Sex 0.015 0.865 0.652 0.842

F1 Age  F2 Age 0.003 0.024 -0.003 0.020

F1 Larval diet  F1 Age 1.033 0.864 -0.442 0.679

F1 Larval diet  F2 Sex -0.375 0.734 -0.134 0.680

Parental breeding age effects offspring lifespan

F1 Age  F2 Sex 0.287 0.793 -0.193 0.683

F2 Sex  F2 Age 0.058 0.023* -0.023 0.021

F1 Age  F3 Sex 0.376 0.554 -0.517 0.530

F2 Sex  F3 Sex 0.881 0.538 -0.625 0.544

F2 Age  F3 Sex -0.028 0.016. -0.006 0.017

F1 Larval diet  F3 Sex 0.749 0.511 0.083 0.536

∗P-value < 0.05, ∗∗P-value < 0.01, ∗∗∗P-value < 0.001.

Previous work on T. angusticollis has shown that macronutrients in the maternal and paternal larval diets have complex effects on a number of offspring traits, including body size, juvenile viability and development time [1–3]. Here, we observed an interaction of grand- maternal age and larval diet on grand-offspring body size in matrilines, where grand-offspring body size decreased with grand-maternal age at breeding when grandmothers were reared on low-nutrient diet but increased with grand-maternal age at breeding when grandmothers were reared on high-nutrient larval diet (Fig.S4). In patrilines, we observed F2 sex × F2 age and

F1 age × F3 sex interactions for body size, indicating that body size is negatively affected by increased parental and grandparental breeding age, but this effect varies as a function of parental and offspring sex (Fig. S5). We also detected an F2 sex × F2 age interaction effect on development time (Fig. S6), indicating that offspring development time increased with maternal age at breeding but decreased with paternal age at breeding. A number of studies on other species have also found that maternal age at reproduction can affect offspring body size, and theory suggests that such patterns could reflect adaptive maternal effects [4]. However, the effects are often complex and difficult to interpret. For example, in earth mites, younger mothers produce offspring of a smaller body size (and thus lower fitness), but older mothers produce larger and better provisioned offspring [5].

Parental breeding age effects offspring lifespan

References

1. Bonduriansky R, Head M. Maternal and paternal condition effects on offspring phenotype in Telostylinus angusticollis (Diptera: Neriidae). J Evol Biol. 2007;20: 2379–2388. doi:10.1111/j.1420-9101.2007.01419.x

2. Bonduriansky R, Runagall-mcnaull A, Crean AJ. The nutritional geometry of parental effects : maternal and paternal macronutrient consumption and offspring phenotype in a neriid fl y. Funct Ecol. 2016; doi:10.1111/1365-2435.12643

3. Hooper AK, Spagopoulou F, Wylde Z, Maklakov AA, Bonduriansky R. Ontogenetic timing as a condition-dependent life history trait: High-condition males develop quickly, peak early and age fast. Evolution. 2017; 1–15. doi:10.1111/evo.13172

4. Kindsvater HK, Rosenthal GG, Alonzo SH. Survival costs of reproduction predict age-dependent variation in maternal investment. J Evol Biol. 2011;24: 2230–2240.

5. Plaistow SJ, Clair JJHS, Grant J, Benton TG. How to Put All Your Eggs in One Basket : Empirical Patterns of Offspring Provisioning throughout a Mother ’ s Lifetime. Am Nat. 2007;170: 520–529. doi:10.1086/521238

CHAPTER 3

Effects of condition and sperm competition risk on sperm allocation and storage in neriid flies

1Zachariah Wylde, 2Angela Crean & 1Russell Bonduriansky

Behavioural Ecology (2019). Published 07/11/19.

1Evolution and Ecology Research Centre, School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney, 2052, Australia

2Charles Perkins Centre, The University of Sydney, Sydney, New South Wales, Australia

Male condition and sperm competition changes ejaculate allocation

Abstract

Ejaculate traits can be sexually selected and often exhibit heightened condition-dependence. However, the influence of sperm competition risk in tandem with condition-dependent ejaculate allocation strategies is relatively unstudied. Because ejaculates are costly to produce, high-condition males may be expected to invest more in ejaculates when sperm competition risk is greater. We examined the condition-dependence of ejaculate size by manipulating nutrient concentration in the juvenile (larval) diet of the neriid fly Telostylinus angusticollis. Using a fully factorial design we also examined the effects of perceived sperm competition risk (manipulated by allowing males to mate first or second) on the quantity of ejaculate transferred and stored in the three spermathecae of the female reproductive tract. To differentiate male ejaculates, we fed males non-toxic rhodamine fluorophores (which bind to proteins in the body) prior to mating, labelling their sperm red or green. We found that high- condition males initiated mating more quickly and, when mating second, transferred more ejaculate to both of the female’s posterior spermathecae. This suggests that males allocate ejaculates strategically, with high-condition males elevating their ejaculate investment only when facing sperm competition. More broadly, our findings suggest that ejaculate allocation strategies can incorporate variation in both condition and perceived risk of sperm competition.

Keywords: Sexual selection, Ejaculate, Condition-dependence, Telostylinus angusticollis, Rhodamine, Larval diet, Sperm competition

Male condition and sperm competition changes ejaculate allocation

Introduction

When a female mates with multiple males, competing ejaculates may temporally and spatially overlap and compete for fertilisation of the female’s eggs (Parker, 1970). Selection therefore favours male traits that increase the competitive ability of sperm (Manier et al., 2012; Parker, 1970; Tommaso Pizzari & Parker, 2009; Simmons, 2001; Snook, 2005). Selection may also act on females, and favour the ability to cryptically choose (i.e., bias) fertilisation success towards a particular male’s sperm (Thornhill 1983; Birkhead and Moller 1993; Eberhard 1996). In many insects, females have multiple long-term sperm storage organs (spermathecae) within their reproductive tracts, where sperm competition is often intensified (Pascini & Martins, 2017). Sperm competition has been shown to select for traits that enhance fertilization success when multiple ejaculates co-occur within the female reproductive tract: for example, in lines with experimentally increased rates of polyandry, male and female yellow-dung flies have been shown to evolve larger testes and accessory sex glands, respectively (Hosken et al. 2001). However, how male condition influences ejaculate allocation strategies in response to perceived sperm competition remains poorly understood.

Theory on sperm allocation suggests that males are selected to transfer more sperm when detecting a risk of sperm competition (e.g., as when mating with a previously-mated female) than when mating with a virgin female (Parker, Ball, Stockley, & Gage, 1996; 1997; Parker & Begon, 1986). This prediction has been supported by empirical work (Gage and Baker 1991; Cook and Gage 1995; Fuller 1998; Aron et al. 2016). However, numerous empirical studies have also failed to support this prediction, bringing the generality of this rule into question (reviewed in Williams, Day, & Cameron, 2005). Condition-dependent variation in ejaculate allocation strategies could contribute to variable outcomes of empirical studies. If ejaculate traits are sexually selected and costly, then we should expect these traits to exhibit heightened condition-dependence when compared to most other traits (Andersson 1994; Rowe and Houle 1996; Bonduriansky and Rowe 2005). Theoretical and empirical studies have begun to reveal that both sperm and non-sperm components of the ejaculate can be similarly plastic and condition-dependent in their responses to the developmental environment (see Wigby, 87

Male condition and sperm competition changes ejaculate allocation

Perry, Kim, Sirot (2016)). Yet, how variation in male condition shapes ejaculate allocation remains poorly understood.

Condition can be defined as the amount of metabolic resources that an individual has to allocate to all fitness-related traits, which can be affected by nutrient abundance in the environment as well as genes that affect the capacity to extract and metabolise resources (Andersson, 1982; Hill, 2011; Rowe & Houle, 1996). Consistent with predictions, there is evidence that ejaculate properties can vary in response to food availability (Kahrl & Cox, 2015; Perry & Rowe, 2010; Rahman, Kelley, & Evans, 2013) and inbreeding depression (Gage et al. 2006; Zajitschek and Brooks 2010; Maximini et al. 2011), both of which likely affect the amount of resources available for a male to invest in reproduction (Mautz et al. 2013). The nature of the relationship between ejaculate quality and investment in other life- history traits is complex (Roff and Fairbairn 2007), and subject to resource allocation trade- offs (Engqvist, 2011;Simmons and Emlen 2006). Nonetheless, males that possess more resources (i.e., high-condition males) are still expected to produce larger ejaculates if they experience a lower marginal cost of ejaculate production than do low-condition males (Parker, 1990; Tazzyman, Pizzari, Seymour, & Pomiankowski, 2009). This prediction has been supported in diverse taxa, including reptiles (Kahrl and Cox 2015); fish (Rahman et al. 2013) and insects (Ferkau & Fischer, 2006; Jia, Jiang, & Sakaluk, 2000; Kaldun & Otti, 2016; Perry & Rowe, 2010; Watanabe & Hirota, 1999). High-condition males produce bigger ejaculates (Fedina and Lewis 2004; Blanco et al. 2006), transfer larger amounts of sperm (Perez-Staples et al. 2008) and can produce higher quality nuptial gifts (Jia et al. 2000). Most studies manipulate male condition through dietary restriction at the adult stage, so less is known about how the juvenile nutritional environment shapes adult post-copulatory performance, although effects of resource availability during development have been explored in several studies (Amitin & Pitnick, 2007; Dávila & Aron, 2017; Engqvist, 2008; McGraw et al., 2007; Melo, Almeida, Caldeira-Brant, Parreira, & Chiarini-Garcia, 2014; Vega-Trejo, Jennions, & Head, 2016). Most of these studies show that poor developmental nutrition has a negative effect on ejaculate size and quality (Delisle and Bouchard 1995;

Male condition and sperm competition changes ejaculate allocation

Bissoondath and Wiklund 1996; Amitin and Pitnick 2007). Juvenile nutrition is likely to play an especially important role in shaping the development of adult traits, including ejaculate traits, in holometabolous insects (Macartney et al. 2018).

To understand ejaculate allocation strategies, it is necessary to elucidate how sperm competition risk and condition jointly influence investment in ejaculate traits (Perry, Sirot, & Wigby, 2013). Males in good condition are predicted to invest more in secondary sexual traits (Andersson, 1982; Cotton, Fowler, & Pomiankowski, 2004; Pomiankowski, 1987). However, selection favours prudent ejaculate allocation strategies (Wedell et al. 2002), so high-condition males might be selected to elevate their ejaculate investment only when perceiving a risk of sperm competition. High-condition males tend to have higher mating rates than smaller males, as seen in antler flies (Bonduriansky & Brassil, 2005), Eurpean vipers

(Madsen et al. 1993), and rhesus macaques (Georgiev et al. 2015). High-condition males might therefore be expected to evolve prudent ejaculate allocation strategies that enable them to take advantage of frequent mating opportunities. By contrast, low-condition males might be selected to invest maximally in all matings because their probability of achieving a mating is lower, and they may lack the resources required to elevate ejaculate expenditure even further when facing high sperm competition risk.

Here we assess the effects of male condition and sperm competition risk (i.e., mating sequence) on the amount of ejaculate transferred by males and stored in the female spermathecae in Telostylinus angusticollis (Diptera: Neriidae), a species endemic to New South Wales and southern Queensland, Australia. Mating sequence predicts the level of sperm competition, such that males might be selected to invest more ejaculate if they can detect that they are mating second. Drosophila melanogaster males have been shown to mark females with pheromones that make the females less attractive to other males and decrease the females’ probability of re-mating (Laturney and Billeter 2016). T. angusticollis females are polyandrous, and males could utilise similar chemical cues to detect sperm competition risk. However, male responses to perceived sperm competition risk could depend on male condition. In this species, many aspects of the adult male phenotype and reproductive 89

Male condition and sperm competition changes ejaculate allocation

strategy are strongly influenced by dietary nutrients during development. Increasing the concentration of dietary nutrients during the larval stage results in increased body size and enlarged secondary sexual traits (Sentinella et al. 2013). Nutrient-rich larval diet also enhances sperm motility (Macartney et al. 2018). However, less is known about the effects of larval diet on ejaculate allocation strategies. Males of this species do not impart a nutritious nuptial gift (Bonduriansky & Head, 2007), but insect ejaculates contain seminal proteins that could influence female behaviour and physiology (e.g. stimulate egg production) and such proteins could be costly for males to produce (Perry & Rowe, 2010). The costs of sperm production can also be considerable (Dewsbury 1982; Nakatsuru and Kramer 1982; Van Voorhies 1992; Pitnick and Markow 1994; Pitnick 1996; Olsson et al. 1997; Dowling and Simmons 2012). Yet, males of this species can mate many times within a short time frame (E.L. MacCartney, unpublished data), and this ability might be associated with strategic ejaculate allocation strategies.

T. angusticollis females possess three spermathecae, two of which are joined by a common spermathecal duct (Figure 1b), and males deposit sperm directly into a spermathecal duct (Bath et al. 2012). Sperm then move in an undulating motion towards the spermathecae (Nicovich et al. 2015) and coalesce on the spermathecal equator (ZW, unpublished data). It is not known whether there is any stratification of competing sperm among these three storage organs, or if the three spermathecae have different functions. For example, some insects (such as lacebugs, Hemiptera:Tingidae) possess “pseudospermathecae” that do not store sperm but instead play a role in the transfer of secretions to eggs traveling down the oviduct (Carayon, 1958;Marchini, del Bene, & Dallai, 2010).

Male condition and sperm competition changes ejaculate allocation

Figure 1. Reproductive organs of T. angusticollis. a; Male testes and accessory glands (AG). b; Female bursa copulatrix (BC), spermathecae (S) and oviduct (O). c; Female oviscape. d; Mature spermatozoon (acrosome and midpiece shown). e; spermathecal duct valves (Va) and junction to bursa. The Image in panel b is modified from Bath, Tartarnic & Bonduriansky, 2012).

We manipulated both early-life condition (by rearing males on a nutrient-rich or nutrient- poor larval diet) and male mating sequence (by mating males 1st or 2nd to a single female) in a fully crossed design. We tracked and quantified competing male ejaculates within each of the three spermathecae using a new method of ejaculate staining with rhodamine dyes, based on previous work by Hayashi & Kamimura (2002). Differences in ejaculate storage patterns between the different spermathecae might signify a difference in spermathecal function or male ejaculate transfer strategy. We predicted that males possessing more resources (i.e., high- condition males) would transfer larger ejaculates and asked whether condition-dependent ejaculate allocation patterns are also influenced by mating sequence.

Materials and methods

Experimental animals

The flies used in these experiments were of the third and fourth generations of laboratory reared stock populations that originate from Fred Hollows Reserve, Randwick, NSW, 91

Male condition and sperm competition changes ejaculate allocation

Australia (33°54′44.04″S 151°14′52.14″E). To avoid inbreeding depression, flies in this stock were periodically replenished with wild-caught individuals from the same source population. All larvae were reared in climate chambers at 25°C ± 2°C with a 12:12 photoperiod and moistened with water every two days. The first block of this experiment was completed in March 2016 on the third-generation cohort and the second block was completed in May 2017 using the fourth-generation cohort.

Larval diet manipulation

We employed a 2X2 factorial design with manipulation of male larval diet (as a means of generating males of varying condition and body size) crossed with a manipulation of mating sequence (Figure 2). We manipulated the quantity of resources available to larvae during development by rearing flies on either a nutrient-rich or nutrient-poor diet based on Sentinella et al. (2013). All diets consisted of a base of 170g of coco peat moistened with 600mL of reverse osmosis-purified (RO) water. The “rich” larval diet consisted of 32.8g of protein (Nature’s Way soy protein isolate; Pharm-a-Care, Warriewood, Australia) and 89g of brown sugar (Woolworths Essentials Bonsucro® brand), the “standard” larval diet consisted of 10.9g of protein and 29.7g sugar, and the “poor” larval diet consisted of 5.5g of protein and 14.8g sugar. These nutrients were mixed into the cocopeat and water using a handheld blender and frozen at -20°C until the day of use. All females used in the experiments described below were reared on a “rich” larval diet. Upon adult emergence, flies were separated by sex and diet treatment and kept in populations of 30 ± 3 individuals per 8L cage with sugar/yeast and water (ad libitum) for 15 days. Females were additionally provided oviposition medium to hasten vitellogenesis.

Labelling ejaculates with rhodamine

After 15 days, focal males were transferred into individual 120 mL containers fitted with 2 X 2mL Eppendorf® tubes of RO water (gravity-fed from container lid) with a cocopeat substrate (to retain moisture) and provided with the high-nutrient medium (described above) to which either rhodamine 110 (RH110) (green) or rhodamine B (RHB) (red) had been added. 92

Male condition and sperm competition changes ejaculate allocation

The males were left to feed on these media for five days prior to being paired with a female. Most fluorescent dyes are toxic and are applied post-mortem, whereas rhodamine dyes appear to have little toxicity. A study comparing the toxicity of multiple rhodamine dyes in the house fly Musca domestica, showed no significant differences in mortality rates when flies were orally fed the dyes for up to five days (Respicio and Heitz 1981). These dyes have an affinity to proteinaceous compounds and form stable covalent bonds and so make it possible to label ejaculates and track competing sperm in the female reproductive tract without the need to genetically modify organisms to produce fluorescent probes such as Green or Red

Fluorescent Protein (GFP/RFP) (Manier et al. 2010). Artificial injection of RHB into the genital tract of a female leafhopper (Bothrogonia ferruginea) resulted in this dye being incorporated into ovarioles (Hayashi & Kamimura, 2002). In tobacco moths (Heliothis virescens), males fed rhodamine B transferred spermatophores that emitted a fluorescent signal under UV light, with no reduction in lifespan (Blanco et al., 2006). The rhodamine dyes bind to all proteinaceous components of the ejaculate so fluorescent signal intensity is proportional to the total amount of ejaculate transferred including both seminal fluid and sperm. RHB and RH110 were also chosen because of their structural similarity (almost identical apart from their attached fluorophore) to minimise any biochemical differences in their uptake/ attachment to ejaculate proteins. We observed no obvious effects of RHB or

RH110on behaviour or mortality in T. angusticollis (Z. Wylde, personal observations).

Mating sequence manipulation

We randomly assigned adult males from each larval diet treatment to a mating sequence position of either P1 or P2 (first or second to mate). Two males (previously fed different rhodamine dyes) were mated to a single rich diet female in quick succession in a fully factorial design incorporating all combinations of male condition and mating sequence (Figure 3). This enabled us to examine the main effects and interaction of mating sequence and larval diet treatments. To control for any effects of red vs. green rhodamine dye, we set up approximately equal numbers of treatment and pairings with P1red/P2green and P1green/P2red dye combinations. We also control for dye statistically (see below). 93

Male condition and sperm competition changes ejaculate allocation

Figure 2. Experimental design. Fully factorial design for mating sequence, male condition, and rhodamine dye treatment. High condition (large male symbols), Low condition (small male symbols).

Males were allowed up to 10 minutes to achieve copulation with the focal female within a scintillation vial in the presence of a food source (which stimulates sexual activity), and latency to mating and copulation duration were recorded. Latency to mate was timed from when individuals were placed within the scintillation vials (and allowed 10 seconds to adjust) until the initiation of copulation was observed. The beginning of copulation was defined when the male was observed to mount the female and lock his epandrium into position onto the female oviscape, and the end of copulation was defined as the point when the male withdrew his genitalia from the female oviscape.

Mean copulation duration of T. angusticollis males is approximately 75 seconds (Bath, Tatarnic & Bonduriansky, 2012), so all pairs that separated before 20 seconds were discarded. After the first mating was completed, the first male was immediately removed from the vial and frozen, and the second male was introduced. Following the second mating, the second male was removed and frozen and females were immediately placed in individual 120mL containers with a substrate of moistened cocopeat and a petri dish containing a sugar/yeast mixture, and 2 X 2mL Eppendorf® tubes of RO water, but without oviposition medium. T. angusticollis females do not oviposit unless oviposition medium is provided, and we wanted to prevent females from immediately using any of the sperm received from their two mates in order to quantify patterns of sperm storage in the spermathecae. The mated females were 94

Male condition and sperm competition changes ejaculate allocation

kept in these containers for 48 hours to allow for the ejaculates to reach the spermathecae. Thorax length was measured as an index for body size for all focal males and the females they were paired with.

Figure 3. Experimental workflow. Two males reared on rich or poor larval diet and fed red or green rhodamine dye were mated sequentially to a single female in a fully factorial design, as shown in Figure 2. The female reproductive tract was dissected 48 h after the second mating. Finally, spermathecae were imaged sequentially for each dye treatment using confocal microscopy to obtain fluorescent signal of competing male ejaculates. M1 and M2 indicate an example of ejaculate signal in the posterior spermathecae where the first mate

(M1) was fed RH110, and the second (M2) to mate was fed RHB. The image sequence was randomised for dye treatment filter to minimise any effects of photobleaching that can occur from confocal microscopy.

Male condition and sperm competition changes ejaculate allocation

Sample preparation

After 48 hours females were anaesthetised (but not killed) by placing them for 5 mins into a - 20°C freezer and their reproductive tracts were then carefully dissected in saline solution and mounted in Prolong® Diamond Antifade mountant (to reduce photobleaching) on 1-1.2 mm glass slides with 22 X 22 mm coverslips (Labserv®). All mounted samples were allowed 24 hours to cure at room temperature (in absence of light), stored at -80°C, and imaged within a week of preparation.

Quantifying amount of stored ejaculate

To quantify the relative amount of ejaculate from each competing male within the spermathecae, fluorescence signal intensities for each dye treatment were obtained using a Leica TCS SP5 WLL STED confocal microscope with Hybrid (HyD) detectors for enhanced sensitivity. Males that were fed RHB had ejaculates with an excitation of 540nm

(red spectrum), whereas males fed RH110 had ejaculates that showed excitation at 488nm (green spectrum). The emission of photons for both dye treatments were counted using

Hybrid detectors with spectral ranges of 560-630nm for RHB and 500-550nm for RH110 (fluorophores excited at 540/488 nm emit photons in these ranges). All images were taken at 40x with a 0.7 HCX Plan Fluotar AIR 0.4 mm objective lens. For all images, the White Light Laser (WLL) gain was set to 30 % and laser strength to 40% with a laser speed 80Hz. All images were taken in 12-bit 514X514 resolution. Sequential scans (between frames) were used for each dye treatment filter (to minimise any bleed-through) with a line accumulation of 14, frame average of 4 and a scanning speed of 100Hz to maximise signal. A maximum of 4 images were taken for each sample and randomized in their acquisition sequence for the

RH110 and RHB filter sets to minimise any effect of photobleaching.

All images were captured using Leica Application Suite Advanced fluorescence (version 2.2.3.9723; Leica Microsystems, 1997-2013). All images were subsequently analysed using ImageJ 1.51d (Schindelin et al. 2012). Signal intensity measures (integrated density) were captured by a single hand selection by tracing the perimeter of each individual spermatheca 96

Male condition and sperm competition changes ejaculate allocation

to create a region of interest (ROI) for each focal female. The same ROI was used for both

RHB and RH110 images, which were imaged in the exact same positions. This ensured consistent pixel samples within each focal sample to avoid variability that might arise from multiple hand selections of the same spermathecae. Repeatability of ejaculate signal for multiple hand traces of a single spermatheaca was 0.99 ± 0.003 (n = 20). Many insect tissues including muscles, unsclerotized body parts (Friedrich et al. 2014) and the cephaloskeleton (Grzywacz et al. 2014) will fluoresce under different wavelengths of light without staining. Although auto-fluorescence of tissue was negligible, we controlled for any background noise by taking five measurements of spermathecae from six virgin females using the same settings for each dye treatment as above. The average of the integrated density measures from these images was then deducted from each ejaculate measure according to dye treatment as a way to correct for any differences in auto-fluorescence that might occur for the different dye settings.

Statistical analyses

All missing values for thorax length (males that escaped after mating; Rich-diet n = 6, Poor- diet n = 8) were replaced with mean values for the relevant larval diet treatment (Figure S1). Fluorescence intensity data were used as a proxy for ejaculate amount. Replicates with outlier values (i.e., values > 2 standard deviations from the mean) were removed, and intensity data were then standardised (i.e., converted to z-scores) within dye treatment (red or green) X block combinations (first or second) to eliminate any mean differences in intensity between samples labelled with RHB or RH110.

We fit models with fixed effects of larval diet (LD), mating sequence (MS) and their interaction, thorax length (centred within larval diets, CMTL) to account for variation in male body size within larval diet treatments, copulation duration (CD) to determine whether increased ejaculate expenditure was a function of increased mating time or sperm transfer rate, and block (BL) to account for any variability that might occur between experimental blocks. Male thorax length values were mean-centred within larval diet

Male condition and sperm competition changes ejaculate allocation

treatments to eliminate collinearity with the categorical effect of larval diet. Body size is often correlated with female fecundity, and latency to mate (LT) could affect ejaculate transfer, so both female body size (FTL) and LT were initially included as covariates in our models. These covariates were not found to have any significant effects and were subsequently dropped from further analyses. Mating pair ID and pairing date were included as random effects in all models. In addition, because the effects of larval diet could be mediated at least in part through body size, we also modelled our data using uncentred thorax length as a predictor instead of larval diet; these analyses yielded qualitatively similar results (see Supplementary Material). Latency to mate was found to slightly violate assumptions of normality. However, we re-tested this data using a non-parametric Kruskal-Wallis test and for multiple factors, an Aligned-Rank ANOVA (Kay and Wobbrock 2019). These tests revealed a very similar result as parametric tests, reported within the supplementary information.

All statistical analyses were performed using R 3.3.2 (R Core Team, 2012) and the R package “lme4” (Bates, Maechler, Bolker, & Walker, 2015) was used to fit linear mixed models (LMM). All fixed effects in our linear mixed models were tested using the ‘lmerTest’ package (Kuznetsova, Brockhoff & Bojesen, 2017), with type III ANOVA F statistics based on Satterthwaite approximations.

Results

Body size

We found that male larval diet affected adult body size: males reared on a nutrient-poor larval diet had a smaller thorax length than did males reared on a nutrient-rich larval diet (ANOVA,

F1, 190 = 281.5, P = <0.0001) (Figure 4).

Male condition and sperm competition changes ejaculate allocation

Figure 4. The effects of larval diet quality on male thorax length. The violin plot outlines illustrate the kernel probability density i.e., the width of the outlined area represents the proportion of the data located there.

Copulatory behaviours

Male larval diet affected males’ latency to mate. Males reared on a poor larval diet took longer to start mating. There was no effect of male larval diet on copulation duration. There were no effects of mating sequence, nor the interaction of larval diet and mating sequence, for either latency to mate or copulation duration. Likewise, male body size within larval diet treatments had no effect on latency or copulation duration (Table 1, Figure 5).

Table 1. linear mixed model results for latency to mate and copulation duration (including larval diet). Effects with p < 0.05 are in bold.

Male condit Latency to mate Copulation duration

Fixed effects: Estimate SE df Estimate SE df

ion and sperm sperm ionand competition ejaculate allocationchanges

(Intercept) 178.08 29.79*** 36.45 60.259 3.781*** 49.26

Male larval diet -87.87 25.18*** 176.63 -1.542 3.598 177.75

Mating sequence -30.66 25.89 150.29 1.949 3.668 148.53

Male thorax length (centred within larval -43.80 39.51 183.62 -2.971 5.589 183.06

diets)

Block 28.65 31.22 17.71 -6.786 3.720. 20.11

Mating sequence: Male larval diet 42.04 35.85 179.05 0.403 -0.047 180.91

Significance codes: 0.0001 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1.

100

Male condition and sperm sperm Male conditionand competition ejaculate allocationchanges

Figure 5. A: Latency to mate in males reared on rich and poor larval diets. B; Copulation duration in males reared on rich or poor larval diets. Bars show mean ± SEM. The violin plot outlines illustrate the kernel probability density (i.e., the width of the outlined area represents the proportion of the data located there). Bracket and asterisk above indicate statistical difference as indicated in Table 1.

101

Male condition and sperm competition changes ejaculate allocation

Ejaculate within the spermathecae

We tested effects of larval diet, mating sequence and their interaction on the amount of ejaculate present within each spermatheca (quantified as fluorescence signal intensity from the rhodamine dyes). The larval diet and mating sequence interaction was significant for both posterior spermathecae, reflecting a relative increase in the amount of stored ejaculate from high-condition males when mating second (Table 2; Figure 6). This interaction was non-significant for the anterior spermatheca (Figure 6C) but the pattern appeared qualitatively similar to that for the posterior spermathecae (Figure 6A). The interaction between mating sequence and male larval diet persisted when copulation duration was added to the model, and became significant for the second posterior spermatheca, indicating that the effects are not mediated entirely by copulation duration. Copulation duration had a significant effect on the amount of ejaculate in both posterior spermathecae but not the anterior spermatheca (Table 2). Additionally, the total amount of ejaculate did not differ significantly among the three spermathecae (ANOVA: F2, 573 = 2.331, P = 0.098; Figure 7).

102

Table 2. linear mixed models for effects of male thorax length (centred), mating sequence and copulation duration on ejaculate storage.

Posterior Posterior Anterior sperm Male conditionand competition ejaculate allocationchanges spermatheca 1 spermatheca 2 spermatheca Fixed effects: Estimate SE df Estimate SE df Estimate SE df

(Intercept) -0.293 0.338 54.55 -0.166 0.344 42.51 0.063 0.351 44.02

Male larval diet -0.154 0.138 115.52 -0.089 0.123 101.56 -0.191 0.129 105.66

Mating sequence -0.162 0.131 106.33 -0.220 0.117. 93.56 -0.123 0.125 96.78

Copulation duration 0.008 0.003* 132.53 0.071 0.003* 113.63 0.003 0.003 121.35

Male thorax length (centred) 0.071 0.230 127.36 -0.163 0.208 110.60 0.119 0.220 117.12

Block -0.077 0.333 20.41 -0.137 0.364 20.07 -0.192 0.366 19.79

Mating sequence: Male larval diet 0.463 0.199* 120.98 0.376 0.178* 103.21 0.369 0.188. 108.49

Significance codes: 0.0001 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1.

103

Male condition and sperm sperm Male conditionand competition ejaculate allocat changes

Figure 6. Interaction between male larval diet and mating sequence on the amount of ejaculate within each spermatheca. A; ejaculate within the posterior ‘1’ spermathecal (doublet). B ejaculate within the posterior ‘2’ (doublet). C; ejaculate within the anterior spermathecal (singlet). Lines represent least squares mean ejaculate amounts measured in arbitrary units (fluorescence) ± SEM. Solid black lines represent the ejaculate from poor diet males (low

ion condition) and dashed grey represent rich diet male ejaculate (high condition). Male mate sequence is characterised as either 1 (first to mate) or 2 (second to mate). These results show that rich-diet males generally transferred more ejaculate. These patterns also provide some evidence that, when mating second, rich-diet males increased their ejaculate investment more than poor-diet males did.

104

Male condition and sperm competition changes ejaculate allocation

Figure 7. Total amount of ejaculate (arbitrary fluorescence signal units, AU) within each of the three spermathecae. The violin plot outlines illustrate the kernel probability density (i.e., the width of the outlined area represents the proportion of the data located there). Points and bars within the violin plots indicate mean + SEM.

Discussion

Our results provide evidence that male ejaculate allocation strategies in response to perceived risk of sperm competition are condition-dependent. Males reared on a rich diet had more ejaculate stored than males reared on a poor diet when mating second (i.e., under risk of sperm competition from the previous male), but not when mating first (i.e., with a virgin female).By contrast, for males reared on a poor larval diet, the effect of mating sequence on the amount of ejaculate transferred to the posterior spermathecae was weaker, as indicated by a larval diet by mating sequence interaction. A qualitatively similar pattern was observed for the anterior spermatheca, although the interaction was non-significant. Males reared on a 105 Male condition and sperm competition changes ejaculate allocation

rich larval diet were also able to initiate mating more quickly, perhaps because such males were more vigorous or more attractive to females.

One way that males could change allocation strategies is by increasing copulation duration. Copulation duration was found to significantly relate to the amount of ejaculate, where the longer a male copulated with a female, the more ejaculate was stored in the two posterior spermathecae. Copulation duration had no significant effect on the amount of ejaculate within the anterior spermatheca (most distal to the bursa copulatrix), suggesting that another ejaculate trait, perhaps sperm motility, or a female-mediated trait might also affect ejaculate storage in the anterior spermathecae. However, with copulation duration in the model, the interaction between mating sequence and larval diet was significant for both posterior spermathecae, suggesting that treatment effects on ejaculate size were not entirely mediated by differences in copulation duration but resulted from an increase in the rate of ejaculate transfer (seminal fluid, sperm, or both). This suggests that high-condition males allocate ejaculates strategically, elevating their rate of transfer only when perceiving a risk of sperm competition. When mating with virgin females (and thus perceiving low risk of sperm competition), this strategy might enable high-condition males to save up ejaculate resources for future matings. If high-condition males experience a high mating rate in the wild, such males might benefit by adjusting ejaculate allocation to each mating based on the perceived risk of sperm competition in order to save up ejaculate resources for subsequent matings. Conversely, if low-condition males experience a very low mating rate in the wild, such males may be selected to invest maximally in each mating. Low-condition males may have low mating rates in the wild because they perform poorly in combat and territory defence (Hooper et al., 2017), and because neriid females appear to discriminate against such males (Fricke et al. 2015).

The observed interaction of larval diet and mating sequence could be driven by a dosage mechanism (whereby high-condition males deposit larger ejaculates when mating second) and/or by condition-dependent variation in sperm quality (whereby low-condition males deposit sperm that is less likely to reach the spermathecae). High-condition T. angusticollis males transfer sperm with a higher tail-beat frequency (Macartney et al. 2018). Low-

106 Male condition and sperm competition changes ejaculate allocation

condition males may therefore transfer less competitive sperm that are less likely to enter the spermathecae when compared to sperm of high condition males. Aberrant sperm (caused by errors in spermatogenesis) with abnormal flagella and therefore lower motility have been observed in many insect species (see review Dallai, Gottardo, & Beutel, 2016). Similar results have been obtained in fish, where low-condition males produce less motile sperm (Locatello et al. 2006; Burness et al. 2008; Rahman et al. 2013; Macartney et al. 2018). Whether variation in condition resulting from our larval diet manipulation can also influence the quality and production of seminal fluid remains to be explored in T. angusticollis. Because our imaging technique was unable to differentiate between sperm and semen, we were not able to examine how the ejaculate composition may have changed in response to mating sequence.

Sperm storage patterns in the spermathecae could affect competitive fertilisation success. The insect spermatheca is thought to be involved in sperm maturation or activation and long- term storage (Klowden 2006). Once eggs have matured and been released, sperm swim back through the spermathecal ducts to the site of fertilisation, which is typically thought to be the oviduct at the junction of these ducts (Pascini & Martins, 2017). We found that ejaculate amounts deposited into all three spermathecae were very similar (Figure 7), but our results also provide tentative evidence of differential sperm storage patterns across these spermathecae. The functional consequences (if any) of such variation in sperm storage across spermathecae for cryptic female mate choice and male competitive fertilization success remain to be determined. It is not clear whether males can influence how their ejaculates are distributed among the three spermathecae, but it is possible that males might benefit by depositing equal amounts of ejaculate into all spermathecae to decrease variance in reproductive success when a female releases sperm from different spermathecae. Our behavioural assay design represents a very simplified mating environment with little opportunity for mate choice and a structurally simple arena with no refuges for females to escape male coercion, thereby limiting any effects that pre-copulatory processes may have on ejaculate storage.

107 Male condition and sperm competition changes ejaculate allocation

Our findings also suggest that males are able to assess a female’s mating history and differentiate between virgin and mated females. Cuticular hydrocarbons (CHCs) could be used by males to assess female mating history. In D. melanogaster, males transfer anti- aphrodisiac pheromones (cuticular hydrocarbons) to the female reproductive tract and cuticle, and these CHCs reduce female attractiveness to other males (Laturney and Billeter 2016). It is possible that neriid flies possess a similar mechanism that allows a male to assess a female’s reproductive status and change his ejaculate allocation strategy accordingly.

Our results suggest that the condition-dependent ejaculate budget of a T. angusticollis male affects his ability to respond to sperm competition. Non-sperm seminal fluid components appear to be especially costly and limiting in insects (Marcotte et al. 2007; Reinhardt et al. 2015), and the costs and benefits of varying ejaculate compositions have been posited to select for plasticity in the structure of ejaculates (Perry et al., 2013). Such plasticity is consistent with theory (Parker, 1990), as suggested by previous empirical findings, particularly in insects that produce nuptial gifts (spermatophores) (Jia et al. 2000; Blanco et al. 2009; Perry & Rowe, 2010). Perhaps low-condition males transferred more sperm at the expense of non-sperm components. Non-sperm components are known to make up the bulk of ejaculate composition in many species, while increased sperm numbers have been shown to contribute very little to ejaculate mass (Perry & Rowe, 2010). More costly ejaculate components are expected to be more strongly condition-dependent (Andersson 1980; Rowe & Houle 1996), but very little is known about the actual costs of seminal proteins and if their production is energetically more expensive than that of sperm.

Previous studies have shown status-dependent sperm investment (fowl; Gallus gallus) (Pizzari, Cornwallis, Lovlie, Jakobsson, & Birkhead, 2003), female influence on sperm storage and preferences for larger male sperm (dung fly; Scathophaga stercoraria) (Ward 1993), and the condition-dependence (investigated using adult diet manipulation) of ejaculate size and composition (ladybird beetle, Adalia bipunctata (Perry & Rowe, 2010)). We present the first evidence that males adjust their ejaculate allocation strategy in response to perceived sperm competition in a condition-dependent manner. Our study is also one of

108 Male condition and sperm competition changes ejaculate allocation

the first to show that larval (as opposed to adult) nutrition can affect adult sperm allocation patterns.

We used a recently-developed method that utilises rhodamine fluorophores (incorporated into adult diet) for the labelling of ejaculates. Rhodamine fluorophores appear to be a promising method for staining ejaculates because of their affinity for proteinaceous compounds, and most importantly, their non-toxicity which allows for a relatively affordable and simple labelling of insect ejaculates. This method offers an alternative to genetically altering sperm to express green fluorescent protein/ red fluorescent sperm heads (GFP/RFP) (Manier et al. 2010). The rhodamine labeling technique labels both sperm and seminal proteins. Our analysis assumes that the dyes bind with equal affinity to all ejaculates and do not interact with female tissues. The chemical structures of RHB and RH110 however are very similar (Beija et al. 2009), reducing the possibility of differences in fluorescence as a result of discrepancies in dye attachment.

Revealing how condition-dependent traits such as ejaculate amount change with social context is important in better understanding responses to sperm competition. We provide evidence that ejaculate allocation is complex and is affected by an interaction of sperm competition risk and condition. Further studies examining how male condition and social environment interact to affect ejaculate allocation and storage will contribute to understanding the mechanisms that generate differences in paternity amongst competing males.

Authors’ contributions

ZW, AC and RB conceived the ideas and designed the methodology. ZW collected the data. ZW and RB analysed the data. ZW and RB led the writing of the manuscript. All authors contributed critically to the drafts and gave final approval for publication.

Acknowledgements

109 Male condition and sperm competition changes ejaculate allocation

The authors would like to extend their thanks to the other members of the Bonduriansky lab for all support and valuable discussions during this experiment. Special thanks to the BMIF facility at UNSW for their helpful advice and training in confocal microscopy. The authors declare that there was no conflict of interest regarding the publication of this article.

Data accessibility

Data are available from the Dryad Digital Repository.

References

Amitin, E. G., and S. Pitnick. 2007. Influence of developmental environment on male- and female-mediated sperm precedence in Drosophila melanogaster. J. Evol. Biol. 20:381–391.

Andersson, M. 1982. Sexual selection, natural selection and quality advertisement. Biol J Linn Soc 17:375–393.

Andersson, M. B. 1994. Sexual selection. Princeton University Press.

Aron, S., P. Lybaert, C. Baudoux, M. Vandervelden, and D. Fournier. 2016. Sperm production characteristics vary with level of sperm competition in Cataglyphis desert ants. Funct. Ecol. 30:614–624.

Bath, E., N. Tatarnic, and R. Bonduriansky. 2012. Asymmetric reproductive isolation and interference in neriid flies: the roles of genital morphology and behaviour. Anim. Behav. 84:1331–1339.

Beija, M., C. A. M. Afonso, and J. M. G. Martinho. 2009. Synthesis and applications of Rhodamine derivatives as fluorescent probes. Chem. Soc. Rev. 38:2410–2433.

Birkhead, T. R., and A. P. Moller. 1993. Sexual election and the emporal eparation of reproductive events : sperm storage data from reptiles, birds and mammals. Biol. J. Linn. Soc. 50:295–311.

110 Male condition and sperm competition changes ejaculate allocation

Blanco, C. a, O. Perera, J. D. Ray, E. Taliercio, and L. Williams. 2006. Incorporation of rhodamine B into male tobacco budworm moths Heliothis virescens to use as a marker for mating studies. J. Insect Sci. 6:1–10.

Bonduriansky, R., and C. E. Brassil. 2005. Reproductive ageing and sexual selection on male body size in a wild population of antler flies (Protopiophila litigata). J. Evol. Biol. 18:1332– 1340.

Bonduriansky, R., and M. Head. 2007. Maternal and paternal condition effects on offspring phenotype in Telostylinus angusticollis (Diptera: Neriidae). J. Evol. Biol. 20:2379–2388.

Burness, G., A. I. Schulte-Hostedde, and R. Montgomerie. 2008. Body condition influences sperm energetics in lake whitefish ( Coregonus clupeaformis ). Can. J. Fish. Aquat. Sci. 65:615–620.

Carayon, J. 1958. Etudes sur les He´mipteres Cimicoidea. Mus Natl Hist Paris, Ser A Zool 16.

Cook, P. A., and M. J. G. Gage. 1995. Effects of risks of sperm competition on the numbers of eupyrene and apyrene sperm ejaculated by the moth Plodia interpunctella (Lepidoptera: Pyralidae). Behav. Ecol. Sociobiol. 36:261–268.

Dallai, R., M. Gottardo, and R. G. Beutel. 2016. Structure and Evolution of Insect Sperm: New Interpretations in the Age of Phylogenomics. Annu. Rev. Entomol. 61:1–23.

Dávila, F., and S. Aron. 2017. Protein restriction affects sperm number but not sperm viability in male ants. J. Insect Physiol. 100:71–76. Elsevier.

111 Male condition and sperm competition changes ejaculate allocation

Dewsbury, D. A. 1982. Ejaculate Cost and Male Choice. Am. Soc. Nat. 119:601–610.

Dowling, D. K., and L. W. Simmons. 2012. Ejaculate economics: Testing the effects of male sexual history on the trade-off between sperm and immune function in Australian crickets. PLoS One 7.

Eberhard, W. G. 1996. Female Control: Sexual Selection by Cryptic Female Choice. Princeton University Press, New Jersey.

Engqvist, L. 2008. Genetic variance and genotype reaction norms in response to larval food manipulation for a trait important in scorpionfly sperm competition. Funct. Ecol. 22:127– 133.

Engqvist, L. 2011. Male attractiveness is negatively genetically associated with investment in copulations. Behav. Ecol. 22:345–349.

Fedina, T. Y., and S. M. Lewis. 2004. Female influence over offspring paternity in the red flour beetle Tribolium castaneum. Proc. Biol. Sci. 271:1393–1399.

Ferkau, C., and K. Fischer. 2006. Costs of reproduction in male Bicyclus anynana and Pieris napi butterflies: Effects of mating history and food limitation. Ethology 112:1117–1127.

Fricke, C., M. I. Adler, R. C. Brooks, and R. Bonduriansky. 2015. The complexity of male reproductive success: effects of nutrition, morphology and experience. Behav. Ecol. 617– 624.

Fuller, R. C. 1998. Sperm competition affects male behaviour and sperm output in the rainbow darter. Proc. R. Soc. B Biol. Sci. 265:2365–2371.

Gage, M. J. G., and R. R. Baker. 1991. Ejaculate size varies with socio‐sexual situation in an insect. Ecol. Entomol. 16:331–337.

Gage, M. J. G., A. K. Surridge, J. L. Tomkins, E. Green, L. Wiskin, D. J. Bell, and G. M. Hewitt. 2006. Reduced heterozygosity depresses sperm quality in wild rabbits, Oryctolagus cuniculus. Curr. Biol. 16:612–617.

112 Male condition and sperm competition changes ejaculate allocation

Hayashi, F., and Y. Kamimura. 2002. The potential for incorporation of male derived proteins into developing eggs in the leafhopper Bothrogonia ferruginea. J. Insect Physiol. 48:153–159.

Hill, G. E. 2011. Condition-dependent traits as signals of the functionality of vital cellular processes. Ecol. Lett. 14:625–634.

Hosken, D. J., T. W. J. Garner, and P. I. Ward. 2001. Sexual conflict selects for male and female reproductive characters. Curr. Biol. 11:489–493.

Jia, Z., Z. Jiang, and S. K. Sakaluk. 2000. Nutritional condition influences investment by male katydids in nuptial food gifts. Ecol. Entomol. 25:115–118.

Kahrl, A. F., and R. M. Cox. 2015. Diet affects ejaculate traits in a lizard with condition- dependent fertilization success. Behav. Ecol. 26:1502–1511.

Kaldun, B., and O. Otti. 2016. Condition-dependent ejaculate production affects male mating behavior in the common bedbug Cimex lectularius. Ecol. Evol. 6:2548–2558.

Klowden, M. J. 2006. Switchover to the mated state by spermathecal activation in female Anopheles gambiae mosquitoes. J. Insect Physiol. 52:679–684.

Laturney, M., and J. C. Billeter. 2016. Drosophila melanogaster females restore their attractiveness after mating by removing male anti-aphrodisiac pheromones. Nat. Commun. 7:1–11. Nature Publishing Group.

Locatello, L., M. B. Rasotto, J. P. Evans, and A. Pilastro. 2006. Colourful male guppies produce faster and more viable sperm. J. Evol. Biol. 19:1595–1602.

113 Male condition and sperm competition changes ejaculate allocation

Macartney, E. L., P. R. Nicovich, R. Bonduriansky, and A. J. Crean. 2018. Developmental diet irreversibly shapes male post-copulatory traits in the neriid fly Telostylinus angusticollis. J. Evol. Biol. 31:1894–1902.

Madsen, T., R. Shine, J. Loman, and T. Håkansson. 1993. Determinants of mating success in male adders, Vipera berus.

Manier, M. K., J. M. Belote, K. S. Berben, S. Lüpold, O. Ala-Honkola, W. F. Collins, and S. Pitnick. 2012. Rapid Diversification Of Sperm Precedence Traits And Processes Among Three Sibling Drosophila Species. Evolution. 67:2348–2362.

Manier, M. K., J. M. Belote, K. S. Berben, D. Novikov, W. T. Stuart, and S. Pitnick. 2010. Resolving Mechanisms of Competitive Fertilization Success in Drosophila melanogaster. Science (80-. ). 328.

Marchini, D., G. del Bene, and R. Dallai. 2010. Functional morphology of the female reproductive apparatus of Stephanitis pyrioides (Heteroptera, Tingidae): A novel role for the pseudospermathecae. J. Morphol. 271:473–482.

Marcotte, M., J. Delisle, and J. N. McNeil. 2007. Effects of different male remating intervals on the reproductive success of Choristoneura rosaceana males and females. J. Insect Physiol. 53:139–145.

Mautz, B. S., A. P. Møller, and M. D. Jennions. 2013. Do male secondary sexual characters signal ejaculate quality? A meta-analysis. Biol. Rev. 88:669–682.

Maximini, L., B. Fuerst-Waltl, B. Gredler, and R. Baumung. 2011. Inbreeding Depression on Semen Quality in Austrian Dual-Purpose Simmental Bulls. Reprod. Domest. Anim. 46:102– 104.

114 Male condition and sperm competition changes ejaculate allocation

Nakatsuru, K., and D. L. Kramer. 1982. Is Sperm Cheap? Limited male Fertility and Female Choice in the Lemo Tetra (Pisces, Characidae). Science (80-. ). 216:753–755.

Nicovich, P. R., E. L. Macartney, R. M. Wham, and A. J. Crean. 2015. easuring sperm movement within the female reproductive tract using Fourier analysis. Microsc. Microanal. 256–263.

Olsson, M., T. Nedsen, and R. Shine. 1997. Is sperm cheap? costs of reproduction in male adders, Vipera berus. Proc. R. Soc. London, Ser. B 164:455–459.

Parker, G. A. 1990. Sperm com petition games: raffles and roles. Proc. R. Soc. B 242:120– 126.

Parker, G. A. 1970. Sperm Competition and Its Evolutionary Consequences in the Insects. Biol. Rev. 45:525–567.

Parker, G. A., M. A. Ball, P. Stockley, and M. J. G. Gage. 1997. Sperm competition games: a prospective analysis of risk assessment. Proc. R. Soc. B Biol. Sci. 264:1793–1802.

Parker, G. A., and M. Begon. 1986. Optimal egg size and clutch size: effects of environment and maternal phenotype. Am. Nat. 128:573–592.

Pascini, T. V., and G. F. Martins. 2017. The insect spermatheca: an overview. Zoology 121:56–71. Elsevier.

Perez-Staples, D., A. M. T. Harmer, S. R. Collins, and P. W. Taylor. 2008. Potential for pre- release diet supplements to increase the sexual performance and longevity of male Queensland fruit flies. Agric. For. Entomol. 10:255–262.

Perry, J. C., and L. Rowe. 2010. Condition-dependent ejaculate size and composition in a ladybird beetle. Proc. R. Soc. B Biol. Sci. 277:3639–3647.

115 Male condition and sperm competition changes ejaculate allocation

Perry, J. C., L. Sirot, and S. Wigby. 2013. The seminal symphony: How to compose an ejaculate. Trends Ecol. Evol. 28:414–422.

Pitnick, S. 1996. Investment in Testes and the Cost of Making Long Sperm in Drosophila. Am. Nat. 148:57–80.

Pitnick, S., and T. A. Markow. 1994. Large-male advantages associated with costs of sperm production in Drosophila hydei, a species with giant sperm. Proc Natl Acad Sci U S A 91:9277–9281.

Pizzari, T., C. K. Cornwallis, H. Lovlie, S. Jakobsson, and T. R. Birkhead. 2003. Sophisticated sperm allocation in male fowl. Nature 426:70–74.

Pizzari, T., and G. Parker. 2009. Sperm competition and sperm phenotype.

Pomiankowski, A. 1987. The costs of choice in sexual selection. J. Theor. Biol. 128:195–218.

Rahman, M. M., J. L. Kelley, and J. P. Evans. 2013. Condition-dependent expression of pre- and postcopulatory sexual traits in guppies. Ecol. Evol. 3:2197–2213.

Reinhardt, K., H. G. Breunig, A. Uchugonova, and K. König. 2015. Sperm metabolism is altered during storage by female insects: evidence from two-photon autofluorescence lifetime measurements in bedbugs. J. R. Soc. Interface 12:20150609.

Respicio, N. C., and J. R. Heitz. 1981. Comparative Toxicity of Rhodamine B and Rhodamine 6G to the House Fly (Musca domestica). Bull. Environ. Toxicol. 27:274–281.

Roff, D. A., and D. J. Fairbairn. 2007. The evolution of trade-offs: where are we? J. Evol. Biol. 20:443–447.

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

Schindelin, J., E. Arganda-Carreras, I. Frise, V. Kaynig, M. Longair, T. Pietzsch, S. Preibisch, C. Rueden, S. Saalfeld, B. Schmid, J. Y. Tinevez, D. J. White, V. Hartenstein, K. Eliceiri, P.

116 Male condition and sperm competition changes ejaculate allocation

Tomancak, and A. Cardona. 2012. Fiji: an open-source platform for biological-image analysis. Nat. Methods 9:676–682.

Sentinella, A. T., A. J. Crean, and R. Bonduriansky. 2013. Dietary protein mediates a trade- off between larval survival and the development of male secondary sexual traits. Funct. Ecol. 27:1134–1144.

Simmons, L. W. 2001. Sperm competition and its evolutionary consequences in the insects. Princeton University Press.

Simmons, L. W., and D. J. Emlen. 2006. Evolutionary trade-off between weapons and testes. Proc. Natl. Acad. Sci. 103:16346–16351.

Snook, R. R. 2005. Sperm in competition: Not playing by the numbers. Trends Ecol. Evol. 20:46–53.

Tazzyman, S. J., T. Pizzari, R. M. Seymour, and A. Pomiankowski. 2009. The Evolution of Continuous Variation in Ejaculate Expenditure Strategy. Am. Nat. 174:E71–E82.

Thornhill, R. 1983. Cryptic Female Choice and Its Implications in the Scorpionfly Harpobittacus nigriceps. Am. Soc. Nat. 122:765–788.

Van Voorhies, W. A. 1992. Production of sperm reduces nematode lifespan. Nature 360:456–458.

Vega-Trejo, R., M. D. Jennions, and M. L. Head. 2016. Are sexually selected traits affected by a poor environment early in life? BMC Evol. Biol. 16:1–12. BMC Evolutionary Biology.

Ward, P. I. 1993. Females Influence Sperm Storage and Use in the Yellow Dung Fly Scathophaga stercoraria. Behav. Ecol. Sociobiol. 32:313–319.

Watanabe, M., and M. Hirota. 1999. Effects of Sucrose Intake on Spermatophore Mass Produced by Male Swallowtail Butterfly Papilio xuthus. Zoolog. Sci. 16:55–61.

117 Male condition and sperm competition changes ejaculate allocation

Wigby, S., J. C. Perry, Y. H. Kim, and L. K. Sirot. 2016. Developmental environment mediates male seminal protein investment in Drosophila melanogaster. Funct. Ecol. 30:410– 419.

Williams, P. D., T. Day, and E. Cameron. 2005. The Evolution of Sperm-Allocation Strategies and the Degree of Sperm Competition. Soc. study Evol. 59:492–499.

Zajitschek, S. R. K., and R. C. Brooks. 2010. Inbreeding depression in male traits and preference for outbred males in Poecilia reticulata. Behav. Ecol. 21:884–891.

Supplementary information

Non-parametric tests revealed significant differences in latency to mate between males reared on the different larval diets (Kruskal-Wallis test; H = 11.79, df = 1, p = 0.0006). Further Aligned-Rank ANOVA, including the interaction between larval diet and mating sequence also revealed a significant difference between a males latency to mate (ART-ANOVA, F1, 188 = 12.297, p = 0.00057)

Table S1. Models ranked by AICc. Response variables: PS = Posterior spermatheca; AS = Anterior spermatheca; CD = Copulation duration; LM = Latency to mate. Covariates: CMTL = Centred male thorax length; MS = Mating sequence; LD = Larval diet.

Candidate models AICc ΔAIC Weight Ejaculate amount within PS1 PS1 ~ LD + MS + CMTL + LD*MS 466.7 0 0.97 PS1 ~ LD + MS + CMTL + CD + LD*MS 473.4 6.68 0.034 Ejaculate amount within PS2 PS2 ~ LD + MS + CMTL + LD*MS 447.6 0 0.97 PS2 ~ LD + MS + CMTL + CD + LD*MS 454.3 6.75 0.033 Ejaculate amount within AS AS ~ LD + MS + CMTL + LD*MS 453.9 0 0.996 AS ~ LD + MS + CMTL + CD + LD*MS 464.9 10.97 0.004

118

Table S2. linear mixed model results for latency to mate and copulation duration with male thorax length (uncentred) as a predictor instead of larval diet.

Latency to mate Copulation duration Male

Fixed effects: Estimate SE df Estimate SE df condition sperm and competition ejaculate allocationchanges

(Intercept) 400.56 93.68*** 186.28 57.350 13.041*** 184.24

Male thorax length -104.83 34.72** 178.50 0.761 4.884 178.85

Mating sequence -93.77 125.94 176.41 31.760 17.715. 176.48

Block 32.21 30.14 18.16 -6.840 3.661. 20.14

Mating sequence: Male thorax length 32.34 47.86 176.99 -12.671 6.739. 177.33

Significance codes: 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1.

119

Table S3. Linear mixed models for effects on ejaculate storage patterns with male thorax length (uncentred) as a predictor instead of larval diet.

Male

Posterior Posterior Anterior condition sperm and competition ejaculate allocationchanges spermatheca 1 spermatheca 2 spermatheca Fixed effects: Estimate SE df Estimate SE df Estimate SE df

(Intercept) 0.561 0.539 134.64 1.190 0.555* 119.84 0.900 0.552 118.91

Male thorax length -0.195 0.194 120.01 -0.342 0.177. 108.70 -0.294 0.187 107.20

Mating sequence -1.156 0.699 124.85 -1.751 0.632** 109.65 -1.410 0.647* 109.37

Block -0.101 0.335 19.69 -0.186 0.0361 19.69 -0.200 0.370 19.54

Male thorax length: Mating sequence 0.484 0.271. 125.83 0.677 0.241** 110.07 0.574 0.243* 110.02

Significance codes: 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1.

120

Table S4. linear mixed models for effects of male thorax length (uncentred), mating sequence and copulation duration on ejaculate storage.

Male condition and sperm sperm Male conditionand competition changes

Posterior Posterior Anterior spermatheca 1 spermatheca 2 spermatheca Fixed effects: Estimate SE df Estimate SE df Estimate SE df

(Intercept) 0.048 0.614 142.84 0.606 0.570 125.93 0.700 0.590 129.10

Male thorax length (uncentred) -0.164 0.192 118.94 -0.319 0.171. 105.75 -0.282 0.179 107.51

Mating sequence -1.240 0.689. 121.82 -1.842 0.611** 105.03 -1.433 0.641* 108.23

Copulation duration 0.007 0.003* 136.67 0.007 0.003* 116.90 0.003 0.003 122.22

Block -0.060 0.333 -0.170 -0.135 0.362 19.99 -0.178 0.369 19.75 ejaculate allocation

Male thorax length: Mating sequence 0.511 0.263. 122.74 0.704 0.233** 105.36 0.578 0.244* 108.84

Significance codes: 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1.

121

Male condition and sperm competition changes ejaculate allocation

Male condition and sperm sperm Male conditionand competition ejacu changes

Figure S1. Relationship between the amount of ejaculate (arbitrary fluorescence signal units, AU) and male body size (thorax length) for males mating first (black points, black line) and second (grey points, grey line) for each of the three spermatheca: A, posterior (doublet) spermatheca 1; B, posterior

late allocation (doublet) spermatheca 2; C, anterior (singlet) spermatheca. Lines represent linear models, and bands represent 95% confidence limits. Each point represents the ejaculate from a different focal male. These patterns suggest that ejaculate amounts increased with male body size more steeply for males mating second than for males mating first, and this pattern was supported by the significant or near-significant interactions between male thorax length and mating sequence for each of the three spermathecae. These patterns are qualitatively consistent with those reported in the main text, where we show that ejaculate amount increased more steeply when mating second in males reared on a rich larval diet than in males reared on a poor larval diet.

122

CHAPTER 4

Integration and condition-dependence of genitalic and somatic traits in male and female neriid flies

Zachariah Wylde, & Russell Bonduriansky

(2019). In prep.

123 Genitalic and somatic trait integration

Abstract

The spectacular diversity of insect male genitalia and their relative insensitivity to the environment have long puzzled evolutionary biologists and taxonomists. We asked whether the unusual evolvability of male genitalia could be associated with low morphological integration of genitalic traits, by comparison with male somatic traits and female traits. We also asked whether the degree of integration of male genitalic traits was affected by the developmental environment. To address these questions, we manipulated larval diet quality (which affects adult condition) in a split-brood design and compared levels of morphological integration of male and female genitalic and somatic traits in the neriid fly, Telostylinus angusticollis (Diptera: Neriidae). We found that male genitalic traits were substantially less integrated than male somatic traits, while female genitalic traits were only slightly less integrated than female somatic traits. Male genitalic traits also exhibited substantially lower integration than genitalic traits of females. Integration of male genitalic traits was negatively condition-dependent, with high-condition males exhibiting lower trait integration than low- condition males. Finally, genitalic traits exhibited lower genotype × environment interactions than somatic traits. These results could help to explain the unusually high evolvability of genitalic traits in insects.

Keywords: genitalia, trait integration, larval diet, condition-dependence, evolvability, variation, phenotype

124 Genitalic and somatic trait integration

Introduction

Insect genitalia evolve rapidly and in some cases genitalic traits are the only strongly divergent characters between closely related species (reviewed by Eberhard 2011). Differences in most morphological traits are typically low amongst closely related species, whereas male genitalia are often highly complex in form and function (Langerhans et al. 2016) and display a striking level of variation among species (Eberhard 1985, 1996; Edwards 1993; Birkhead 2000; Hosken and Stockley 2004). Whilst male genitalic traits appear to evolve more rapidly than female reproductive organs, there is some evidence that females also show considerable diversity in genitalic form across closely related species (e.g., Drosophila (Pitnick et al. 1999)). However, few studies have actually characterised female genitalia in much detail, most likely reflecting a larger bias in the literature where most studies omit the examination of female genitalic traits because of anatomical accessibility, or enduring assumptions about female sex roles and the relative variability of female traits (Ah-King et al. 2014; Sloan and Simmons 2019).

Explaining the diversity of male genitalia has been a longstanding problem within evolutionary biology, and the selective pressures that shape these complex traits have been the subject of much debate (Dufour 1844; Eberhard 1985; Arnqvist and Danielsson 1999; Hosken and Stockley 2004). Arnqvist (1998) was the first to take a comparative approach using multiple insect taxa, and was able to show that genital divergence is greatest in groups where females mate with multiple males. Somatic traits such as wings and legs, however, did not show any differences, strongly suggesting that post-copulatory sexual selection drives genital divergence in insects. More recently, genital morphology has been shown to diverge under experimental manipulation of sexual selection (Simmons et al. 2009). However, to add to this complexity, different aspects of both male and female genital morphology can be subject to different mechanisms of sexual selection so it is likely that no single mechanism of sexual selection operates alone (Simmons 2001; Werner and Simmons 2008). Although

125 Genitalic and somatic trait integration

sexual selection appears to drive genitalic divergence, the role of morphological integration in the rapid evolution of male genitalia seems to have been overlooked.

Olson and Miller (1958) defined morphological integration as the “summation of the totality of characters which, in their interdependency of form, produce an organism.” In their work they also surmised that the extent of interdependence in development and function amongst phenotypic traits is directly related to the extent of integration. Accordingly, we can expect functionally and developmentally related traits to evolve together through the selection of interdependence (Cheverud 1982). Within a species, some functional groups of traits may be more or less integrated, but it makes no sense to calculate an average or organism-wide measure of integration (Pigliucci 2003). The study of morphological integration therefore aims to describe the pattern and amount of correlation between a set of phenotypic traits (also referred to as a module). However, research effort has been biased towards the study of mammals, whereas plants and arthropods are not well studied (although, see Tomkins et al 2005), despite having very distinct modular body plans (for review, see Esteve-Altava 2017). Detailed within-species studies of integration offer a unique perspective that could increase our understanding of the selection mechanisms that operate on insect genitalia and somatic traits.

The level of integration of a set of traits is dependent on systemic patterns of gene interactions and not just the mode of selection (Carter et al. 2005; Hansen et al. 2006). Morphological integration has even been identified at the level of quantitative trait locus (QTL) and not just phenotypic and genetic correlations, suggesting that directional selection can cause an adaptive change in the genetic covariance matrix (Mezey et al. 2000; Leamy et al. 2002; Ehrich et al. 2003; Albertson et al. 2005). It has also been shown that selection on QTLs can result in a decrease of variational constraint caused by covariance. If one trait is under directional selection and the other is under stabilizing selection then the dissociation of these traits is favoured, facilitating independent adaptation (Pavlicev et al. 2011). Different aspects of genitalic morphology are thought to be subject to different mechanisms of selection (Simmons 2001; Werner and Simmons 2008). Thus, it seems plausible that there

126 Genitalic and somatic trait integration

are different patterns of selection on the micro-structures of genital traits in insects, leading to the coevolution of reduced integration and a rapidly evolvable diversity of form.

This potential, or what is often termed ‘evolvability’, is usually defined as the ability of trait(s) to respond to novel selective pressures (Wagner and Altenburg 1996) and is intimately related to the modularity and integration of traits (i.e., the extent to which a trait or set of traits is able to vary independently of the rest of the phenotype).

Low integration might therefore have evolved in genitalic traits if such traits function relatively independently. This would generate a lot of multivariate phenotypic variation in genitalia, and allow different components of the genitalic apparatus to respond to selection relatively independently. This could then explain the relatively high evolvability of genitalia, by comparison with somatic traits. However, for this to be true, the difference in integration between genitalic and somatic traits would need to be maintained at varying levels of larval nutrition, resulting in variation in adult condition. Insulin-like Growth Factor (IGF) and Juvenile Hormone (JH) levels are known to vary as a function of nutrition (for review see Lavine et al. 2015), so it is interesting to ask whether any differences in integration between genitalic and somatic traits are maintained in both low- and high-condition individuals.

Another problem with understanding the evolution of insect genitalic forms is understanding why male genital morphologies exhibit low or negative allometric slopes by comparison with other structures, and especially some secondary sexual traits (Huxley 1932; Gould 1974; Petrie 1988; Eberhard 2009). Traits under sexual selection are expected to have relatively high levels of phenotypic and genetic variation (Pomiankowski and Moller 1995) so genitalia might be expected to exhibit high variation as well. Like other sexually selected traits, genitalic traits are also expected to evolve heightened condition-dependence (Andersson 1994; Johnstone 1995; Rowe and Houle 1996; Arnqvist and Thornhill 1998). On the other hand, it is also not entirely clear whether genitalia in insects should be expected to evolve condition-dependence because only directional sexual selection is expected to drive the evolution of heightened condition-dependence, and the types of sexual selection that operate on the phenotypic expression of these genitalic traits is are often difficult to ascertain

127 Genitalic and somatic trait integration

(Arnqvist 1997). Studies that have examined the condition-dependence of genitalia instead often find that genitalia are nutrition-insensitive and exhibit canalized patterns of growth (Simmons and Emlen 2006; House and Simmons 2007; Simmons et al. 2009). It has been postulated that condition-dependence of genital traits may not be manifested in absolute size, but rather in their fine structures (Eberhard et al. 1998; Cayetano and Bonduriansky 2015), which means that much of this variation may have been overlooked. One way that this fine-scale variation might occur is via morphological integration. If developmental stability and canalization are costly, high-condition individuals could exhibit more stable development and higher canalization of genitalic traits, resulting in more highly integrated genitalia when compared to low-condition males. However, there is no framework to make concrete predictions so we asked instead whether differences in integration between trait types would remain consistent across both low and high condition diet treatments.

Instead of displaying levels of exaggeration and condition-dependence that are comparable to weapons or other secondary sexually selected ornaments, selection may instead confer a relatively high level of integration for traits that need to function together in a coordinated manner (i.e., wings or legs). Integration has been shown to be a facilitator of adaptation in these particular instances by reducing maladaptive variation (Gould 1977, 2002; Riedl 1978). Conversely, in traits that are functionally independent, integration has been regarded as a constraint, particularly for allometry, co-regulation of growth (Gould 1977; Firmat et al. 2014; Pélabon et al. 2014; Vallejo-Marin et al. 2014; Voje et al. 2014), and the evolvability of traits (Hansen et al. 2003). Under this framework of functional integration, we would expect that somatic traits that need to work together in a coordinated fashion, such as wings, legs, antennae etc, would show a relatively high level of integration when compared to genitalic traits. Eden (1967) mathematically posed that mutations are not random and that functionally integrated trait complexes are correlated through the epigenetic system so that a change in one trait will result in a correlated change with the functionally associated trait. His argument showed that the probability of independent mutations necessary for adaptation of a functional set of traits would be extremely small (Eden 1967). If the probability of independent mutations arising within functional sets of traits is so low, then we would not

128 Genitalic and somatic trait integration

expect to see the diversity of insect genitalia we see today if they were functionally integrated, but this could explain why we see relatively little change in the general morphology of closely related species. While we cannot rule out that parts of the genitalia are functionally integrated, especially because of a lack of knowledge surrounding the functional integration of insect genital traits (Genevcius and Schwertner 2017) and their mechanistic purpose (Simmons 2014) , some evidence does suggest that functional integration has little effect on genitalia (Richmond et al. 2016). If we then assume that genitalic parts can function independently, we might expect these types of traits exhibit relatively low integration. Low integration would generate a lot of multivariate phenotypic variation in genitalia and allow different components of the genitalic apparatus to respond to selection more independently.

Using the neriid fly Telostylinus angusticollis, we manipulated the larval nutritional environment across 19 families to examine patterns of integration (estimated as the relative standard deviation of eigenvalues (Pavlicev et al. 2009)) and their condition-dependence. Increasing the concentration of dietary nutrients during the larval stage results in increased body size and enlarged secondary sexual traits in T. angusticollis (Sentinella et al. 2013), but the condition-dependence of genitalic traits has not been investigated before in this species. Based on the high evolvability of male genitalia, we predicted that male genitalic traits would be less developmentally integrated than somatic traits in males, and less integrated than female genitalia. We also asked whether differences in integration between trait types and sexes would remain consistent across both low and high nutrient larval diet treatments. Furthermore, we used the larval diet manipulation to compare genetic, environmental and G × E effects on the expression of genitalic and somatic traits in both sexes. If these environments influence resource acquisition or the functioning of cellular processes, we would expect that more strongly condition-dependent traits should exhibit both greater overall effects of environment (E) and greater genetic variation in reaction norms (G x E) (Tomkins et al. 2004; Bussière et al. 2008). Moreover, weak environmental effects, (also known as environmental canalization (Wagner et al 1997)), could be another reason why genitalic traits tend to respond efficiently to selection (Scharloo 1991). Standing genetic variation is predicted to drive rapid evolution, particularly in novel environments because it is

129 Genitalic and somatic trait integration

available immediately at the time selective conditions change (Barrett & Schluter 2007; Hermisson & Pennings 2005).We therefore predicted that male genitalia would exhibit low E and low G × E effects when compared to somatic traits and female genitalia.

Materials and methods

Fly culturing

Flies used in this experiment were derived from lab-reared stocks of T. angusticollis (Enderlein) (Diptera; Neriidae) that originate from individuals collected in 2017 at Fred Hollows Reserve, Randwick, NSW, Australia (33°54′44.04″S 151°14′52.14″E) and were reared in the laboratory for four generations prior to this experiment. While the original number of insects was not quantified, we ensured genetic diversity by supplementing the lab- reared population from wild-caught individuals. All lab-bred individuals were reared in climate chambers at 25°C ± 2°C with a 12:12 photoperiod and provided with water every two days. Eggs from this population were randomly collected and reared using a nutrient- intermediate larval diet based on Sentinella et al. (2013). Virgin adults were collected at emergence, separated by sex and kept in 400 mL cages (maximum 30 individuals per cage) fitted with mesh stockings to allow for ventilation, a moist cocopeat substrate to provide humidity and access to sugar, yeast and water, ad libitum. We then utilised a full-sib, split- family breeding design where randomly chosen individuals from these populations were paired to create 17 mating pairs at 15 ± 2 days old. Each pair was allowed 48 hours to mate within 120 mL cages provided with a nutrient-rich oviposition medium for oviposition and access to sugar, yeast and water ad libitum. Following the 48-hour period, 40 eggs were collected randomly from the oviposition substrate of each mating pair and split equally to either a nutrient-poor or nutrient-rich larval diet, also based on Sentinella et al. (2013). All diets consisted of a base of 170g of cocopeat moistened with 600mL of reverse osmosis- purified water. The “rich” larval diet consisted of 32.8g of protein (Nature’s Way soy protein isolate; Pharm-a-Care, Warriewood, Australia) and 89g of raw brown sugar, the “standard” larval diet consisted of 10.9g of protein and 29.7g raw brown sugar, and the “poor” larval diet consisted of 5.5g of protein and 14.8g raw brown sugar. These nutrients were mixed into the

130 Genitalic and somatic trait integration

cocopeat and water using a handheld blender and frozen at -20°C until the day of use. Upon emergence, virgin adult offspring were allowed 24 hours for their exoskeletons to dry out and then frozen at -80°C for dissection and morphological measurements.

Dissection and sample preparation

The wings, head, antennae and legs were separated from the thorax of all samples before the genitalia of both sexes was carefully dissected for further measurement. All large somatic traits such as wings, antennae, head and legs were laid flat onto 1-1.2 mm microscope slides fitted with inbuilt micrometers for calibration (ISSCO®) and measurement. The heads of individuals were laid upon double sided tape mounted to a slide to enable steady manipulation of samples into a level plane in order to reduce measurement error. Genitals of both males and females were dissected and mounted on microscope slides (same as above) with 22 mm coverslips in 7.2 pH Phosphate Buffered Saline (PBS) (Figure 1). The male surstyli (proximal and distal) and internal section of the genitalia (carefully removed as one unit that included the apodeme, aedeagus and processes) were first separated from the epandrium and placed under a coverslip. All somatic (both sexes) and male genitalic traits were imaged using a Leica MZ 16 fitted with a Leica MC170 HD camera. Before dissection of spermathecae, the oviscape was imaged and its length measured as a proxy for reproductive tract length. The female reproductive tract with spermathecae was carefully removed from the oviscape, cleaned and mounted (same as described above). For all spermathecae measures, images were taken using a Zeiss Axioskop 40 compound microscope fitted with a DinoEyepiece® camera at 20X magnification.

131

Genitalic somatic integration and trait

Figure 1. Genitalia and reproductive traits of Telostylinus angusticollis. Male external and internal genitalia are depicted on the left along with testes and sperm bundles. Spermathecae of females and oviscape for females are depicted on the right. Dots represent avergae R2 values linear as indicated by red brackets, excepttestes (dotted red line mm2).

132

Genitalic and somatic trait integration

Morphological traits and measurement

We examined 93 males (six genitalic and 12 somatic traits) and 96 females (four genitalic and 11 somatic traits) in adults of T. angusticollis. Although geometric morphometrics is often more sensitive in detecting variance in trait morphology (e.g. Breno et al 2011), it is often difficult to develop and implement landmarks for traits that are already challenging to measure systematically and accurately (especially traits such as insect genitalia that are often soft and difficult to dissect). Furthermore, it is often found that traditional methods of allometry are comparable to complex morphometrics in that simpler methods capture the main axes of biological variation in functional traits (Klingenburg 2016). We therefore chose to use linear measures in most traits. However, testes size was measured in mm2 (especially considering our sample sizes which exceed most studies of this nature), All traits are abbreviated and described in Table 1. excluding . Because testes and sperm bundle are not involved directly in intromission, these traits were classified as somatic rather than genitalic traits. For both sexes the proxy for body size used was thorax length, a reliable indicator of body size in this species (Bonduriansky 2006). To minimise the loss of samples for further multivariate analyses, missing trait measure(s) were replaced with the mean value for the family and larval diet treatment group (only for samples when family × larval diet samples >3).

Statistical analyses

To gain an accurate measure of repeatability, each trait of each individual was measured twice by re-positioning each specimen on the slide, re-imaging, and re-measuring the trait in question. All analyses were carried out using R 3.5.3 (R Core Team 2019). Repeatability was calculated for all traits as the variance among individual trait means (individual-level variance VI) over the sum of individual-level and data level (residual) variance VR: R = VG/( VG + VR). We fit linear mixed models with trait as the response variable, larval diet as a fixed categorical predictor and individual I.D as a random effect and subsequently used parametric bootstrapping (1000 iterations, 500 permutations) to obtain uncertainty in estimated R2 values using the R package “rptR” (Stoffel et al. 2017). Variances in R2 differed between the sexes. We therefore compared means using Kruskal- Wallis rank sum tests and found no significant differences between male and female genitalic (H = 2.93, df = 1, p = 0.09) and somatic traits (H = 0.32, df = 1, p = 0.57) (Figure 2).

133 Genitalic and somatic trait integration

Because the data were mostly measurements of length, we stabilized variances by first log10 transforming the data. All log10 transformed data were used for model-fitting described below. Because the data varied in scale and in males, testes were measured as area mm2., for Principal

Component Analysis (PCA), we further transformed the log10 data by mean-centring and dividing each value by the standard deviation within larval diet × sex combinations. We then used PCA to ask if the sexes differ in their covariation structure of genitalic and somatic traits in how they respond to larval diet quality. We conducted four separate PCA within sex and larval diet treatments on correlation matrices.

To measure morphological integration we used the relative standard deviation of eigenvalues for each set of traits (genitalic and somatic) and group combinations (sex and condition) using the method developed by Pavlicev et al (2009). This approach is based on the dispersion of eigenvalues, which reflects the overall correlatedness in a correlation matrix and therefore provides a measure of trait integration. Eigen value variance Var(λ) is dependent on the number of traits and is calculated as the average squared deviation of the eigenvalues from the mean eigenvalue. The sum of a correlation matrix is equal to the number of variables (N) because the trace of the sum or trace of the diagonal elements does not change under rotation of coordinates. This means that the sum of the eigenvalues is simply equal to the number of rows and columns. The eigenvalue variance is therefore:

푁 2 ∑푖=1(λ − 1) Var(λ) = 푁

Eigenvalue variance is dependent on the number of traits measured; however, the relative measure is independent of the number of traits and can thus be readily compared between datasets of

134 Genitalic and somatic trait integration

different numbers of traits. The relative standard deviation of eigenvalues SDrel(λ) is then calculated as:

√Var(λ) SD (λ) = rel √N−1

Prior to the calculation of SDrel(λ), we first ran a Jackknife procedure where we calculated each

Principal Component Analysis (PCA) using log10 transformed data with N-1 traits to obtain a distribution for the relative standard deviation of eigenvalue variances. An independent samples t- test was subsequently conducted to compare SDrel(λ) measures of integration between trait types (genitalic and somatic), and sex and larval diet group combinations. High eigenvalue variance is characteristic of highly integrated phenotypic modules whereas low eigenvalue variance is representative of traits with low integration.

Sample sizes for each family and larval diet treatment varied because of differences in offspring sex ratio, and because it was not always possible to measure the traits of all individuals (some missing due to specimen damage). The mean sample size for each family and larval diet combination was 3 individuals for all combinations of poor and rich diet males and females. It should be noted that for this type of breeding design, statistical power to detect broad-sense heritability (H2) is maximised at 5 individuals per family (Lynch and Walsh 1998). This full-sib, split-family design permitted us to partition variation amongst family and larval nutritional environment. Variation due to individual differences in development within different larval diet treatments within each family was estimated with the residual variance of our models. However, we consider that any confounding effects (variances in survivability of larvae in individual containers or inconsistencies in larval medium quality) were minimal because we standardised egg density and thoroughly homogenised and weighed larval food sources for this study. To examine the relative effect size of larval diet on each trait within sex, we quantified marginal R2 (Nakagawa and Schielzeth 2013) from linear mixed models using the package “lme4” (Bates et al. 2015), where trait length was modelled as a function of larval diet as a fixed categorical predictor and the interaction of larval diet and Family as a random effect. To calculate 95% confidence intervals of the intercept and fixed effects, we used bootstrapped distributions from 500 simulations (See supplementary tables).

2 Because variances of R and SDrel(λ) values for the genitalic and/or somatic traits were

135 Genitalic and somatic trait integration heteroscedastic, Kruskal-Wallis tests were used to compare differences in treatment group combinations. To test differences in effect sizes of G x E, we used Aligned-Rank ANOVA (referred to as ART-ANOVA) from the package “ARTool” (Kay and Wobbrock 2019) which is a non-parametric test of multiple factors and their interactions. These nonparametric tests do not assume homoscedasticity or normality and is robust to repeated measures designs. The ART analysis relies on a pre-processing step that aligns data before applying averaged ranks, which can then be analysed using a common ANOVA. This alignment step corrects for inaccuracies that classic rank transformations produce, particularly for interactive effects (Salter and Fawcett 1993; Higgins and Tashtoush 1994), and has been utilised by a number of recent studies (Yap et al. 2017; Cooper et al. 2018; Macintyre et al. 2018). Although our analysis involved multiple comparisons, our interpretation is based on the overall pattern of results rather than any particular test.

Figure 2. Repeatability (R2) of trait measures for genitalic and somatic traits of males and females. The violin plot outlines illustrate the kernel probability density i.e., the width of the outlined area represents the proportion of the data located there. The box plots within the violin plots represent the median and interquartile range. Dots represent avergae R2 values for each trait. Kruskal-Wallis tests revealed non-significant (ns) differences in R2 values for male vs. female genitalic and somatic trait.

136

Table 1. Abbreviations for traits used in figures with calculated repeatability after controlling for larval diet treatment. R represents bootstrapped repeatability estimates with 95% confidence intervals.

Abbreviation Structure description R: Males R: Females

FE Femur length (front leg in males, hind leg in females) 0.98 [0.97,0.99] 0.98 [0.97,0.99]

FT Tibia (front leg in males, hind leg in females 0.96 [0.94,0.98] 0.96 [0.94,0.97] Genitalic somatic integr and trait Fta Tarsus (front leg in males, hind leg in females) 0.95 [0.92,0.97] 0.97 [0.95,0.98] FSp Length of femur spines on male front leg 0.78 [0.67,0.86] -

WL Wing length (r4+5 wing-vein length from the r-m cross-vein to the wing margin) 0.97 [0.96,0.99] 0.99 [0.98,0.99] HW Head width (across the eyes) 0.87 [0.80,0.91] 0.86 [0.80,0.91] HL Head length to base of pedicel 0.92 [0.88,0.95] 0.85 [0.79,0.91] PP Postpedicel of antenna 0.86 [0.80,0.91] 0.79 [0.69,0.86] AR Arista length of antenna 0.80 [0.7,0.88] 0.87 [0.80,0.91] TL Thorax length 0.96 [0.94,0.98] 0.93 [0.89,0.95] OVL Oviscape length (proxy for bursa copulatrix length) - 0.93 [0.89,0.95]

PS1 Posterior spermatheca 1 width at equator - 0.93 [0.89,0.96] ation PS2 Posterior spermatheca 2 width at equator - 0.92 [0.87,0.95]

Ant Anterior spermatheca width at equator - 0.90 [0.85,0.94] Sur_P Proximal portion of surstylus 0.84 [0.75,0.90] - Sur_D Distal portion of surstylus 0.76 [0.64,0.84] - Apod Aedeagal apodeme length 0.83 [0.74,0.89] - AL Adeagus length 0.81 [0.71,0.88] - SP Short anterior processus 0.86 [0.79,0.92] - PR Processus/ epiphallus 0.94 [0.90,0.96] - TE Testes area (mm2) 0.94 [0.91,0.97] - SB Sperm bundle length 0.76 [0.64,0.85] -

137

Genitalic and somatic trait integration

Results

Morphological integration

Morphological integration was estimated by the relative standard deviation of eigenvalues SDrel (λ) from PCA performed separately on the correlation matrix for each sex × diet × trait type combination (Figures 3, 4). The higher the value of SDrel (λ), the more variance that is explained by the first few axes and therefore the higher the integration. Morphological integration tended to be higher in somatic traits for both males and females overall, although the difference was much greater in males (Figure 5,Table 2). In males, we observed significant differences in integration between larval diet treatments for both genitalic and somatic traits, whereas females only exhibited a significant difference in integration between larval diet treatments for somatic traits. Overall, male genitalia were significantly less integrated than female genitalic traits, (Kruskal-Wallis test; H = 4.8235, df = 1, p = 0.028). Likewise, when we compared integration of genitalic traits of high-condition male and females we found that males had significantly lower integration (Kruskal-Wallis test; H = 8.08, df = 1, p = < 0.004) than females. Similarly, when comparing low-condition males and females, males also had lower integration (Kruskal-Wallis test; H = 8.08, p = 0.004). Conversely, when comparing integration in the sexes across all somatic traits, males had significantly higher trait integration than females (Kruskal-Wallis test; H = 15.652, df = 1, p = < 0.0001). When comparing integration between sexes within larval diet treatments, males had more integrated somatic traits than females when reared on both the rich larval diet diet (Kruskal-Wallis test; H = 15.652, df = 1, p = < 0.0001) and poor larval diet (Kruskal-Wallis test; H = 15.652, df =

1, p = < 0.0001).

Table 2. Comparison of integration levels measured by the mean relative standard deviation of eigenvalues. The mean relative standard deviation is based on Jackknife resampling of PCA estimates with N-1, where N = number of traits. M refers to the sample size. Kruskal- Wallis (KW)-tests compared Jackknife distributions of SDrel (λ) for genitalic vs somatic traits within each sex, or rich vs poor diet samples within each sex × trait type combination.

138

Dataset M N Mean SDrel (λ) Standard error Chi-squared p-value

of SDrel (λ) H Overall differences in integration Male; Genitalic 93 6 0.461 0.044 12.611 0.0004 Male; Somatic 93 12 1.570 0.049 Genitalic somatic integration and trait Female.; Genitalic 96 4 0.818 0.200 9.375 0.022 Female; Somatic 96 11 1.469 0.019 Condition-dependence of

integration Male; High-condition; Genitalic 49 6 0.261 0.028 9.016 0.002 Male; Low-condition; Genitalic 44 6 0.354 0.038 Male; High-condition; Somatic 49 12 0.863 0.051 17.28 <0.0001 Male; Low-condition; Somatic 44 12 1.275 0.044

Female; High-condition; Genitalic 47 4 0.597 0.216 0.535 0.465 Female; Low-condition; Genitalic 49 4 0.745 0.204 Female; High-condition; Somatic 47 11 0.714 0.045 14.286 <0.0001 Female; Low-condition; Somatic 49 11 1.166 0.030 Significant effects of sex, social treatment and their interaction are shown in bold. Significance codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0

139

Genitalic somatic integration and trait

Figure 3. Principal components 1 vs. 2 for high condition (left) and low condition (right) T. angusticollis. Males. Refer to Table 1 for abbreviations of

all traits. Ellipses represent 95% confidence intervals. Dots represent individual PC scores for each trait and arrows represent eigenvectors illustrating the directionality of each trait.

140

Genitalic somati and

c integrationtrait

Figure 4. Principal components 1 vs. 2 for high condition (left) and low condition (right) T. angusticollis. females. Refer to Table 1 for abbreviations of all traits. Ellipses represent 95% confidence intervals. Dots represent individual PC scores for each trait and arrows represent eigenvectors illustrating the directionality of each trait.

141

Genitalic somatic integr and trait

Figure 5. Comparison of integration using mean relative standard deviation of eigenvalues SDrel (λ). Error bars represent standard deviation of the mean ation based on a Jackknife procedure. Significance codes equate to Kruskal-Wallis tests in Table 2 ; 0.0001 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1.

142

Genitalic and somatic trait integration

Effect size of larval diet

We found no significant difference in the marginal effect sizes of larval diet across all genitalic traits between males and females (Kruskal-Wallis; H = 1.64, df = 1, p = 0.201). We did however, observe a significant difference in effects of larval diet across all somatic traits between males and females (Kruskal-Wallis; H = 4.25, df = 1, p = 0.039). Overall, larval diet had a significantly larger effect on somatic traits when compared to genitalic traits (sexes pooled) (Kruskal-Wallis; H = 16.12, df = 1, p = <0.0001) (Figure 6).

Figure 6. Larval diet effect sizes on genitalic and somatic traits in males and females. The violin plot outlines illustrate the kernel probability density i.e., the width of the outlined area represents the proportion of the data located there. Dots represent avergae R2 values for each trait. The box plots within the violin plots represent the median and interquartile range.

143 Genitalic and somatic trait integration

Effect size of G × E

Similarly to larval diet effect sizes we found no significant difference between trait type and sex combinations for conditional effect sizes that included larval diet × family interaction terms (ART-ANOVA, F1, 27 = 1.739, p = 0.198). However, for both sexes together, we did observe a significant effect of trait type where somatic traits had larger effect sizes when compared to genital traits (ART-ANOVA, F1, 27 = 31.35, p = <0.0001) (Figure 7).To examine whether the G × E (larval diet × family) interaction term significantly increased the effect size, we conducted Kruskal-Wallis tests within sex and trait type combinations to compare marginal and conditional effect size distributions. These tests revealed that for males, somatic conditional effect sizes were 30.4% greater on average than marginal effect sizes, but this difference was not significant (Kruskal-Wallis; H= 10.65 , df = 10, p = 0.39). Similarly, the mean effect size of G × E on genitalic traits in males only increased by 10.92% and was not significantly different to marginal effect size distributions (Kruskal-Wallis; H = 5, df = 5, p = 0.42). In females, mean effect size did not significantly increase (16.45%) for somatic traits (Kruskal-Wallis; H = 8, df = 8, p = 0.43) or for genitalic traits (Kruskal-Wallis; H = 3, df = 3, p = 0.39),even with a 72.95% increase in effect size

144 Genitalic and somatic trait integration

Figure 7. Conditional effect sizes including larval diet × family interaction on genitalic and somatic traits in males and females. The violin plots outlines illustrate the kernel probability density i.e., the width of the outlined area represents the proportion of the data located there. Dots represent avergae R2 values for each trait. The box plots within the violin plots represent the median and interquartile range.

Discussion

We compared trait integration between genitalic and somatic traits in the neriid fly, T.angusticollis. Male and female genitalic traits both exhibited relatively low integration by comparison with somatic traits. Conversely, male somatic traits were more integrated than somatic traits of females. Both male and female genitalia responded relatively weakly to larval diet. While lower repeatability of genitalic traits in males could in principle result in lower estimates for integration of those traits relative to somatic traits, the observed differences in integration between genitalic and somatic traits is too large to result solely from differences in

145 Genitalic and somatic trait integration

repeatability. A number of somatic traits in males are known to be enhanced by larval diet quality during development (Sentinella et al. 2013), so it was not surprising that male somatic traits responded more strongly. Interestingly, male and female genitalic traits were similarly weakly affected by G × E, suggesting that environmental effects on the expression of these traits is relatively weak and variable across genotypes. The low integration and weak environmental sensitivity exhibited by genitalic traits could help to explain the high evolvability of insect genitalia and the considerable differences in genitalic form often observed between closely related species. Below, we discuss these findings in light of ontogeny and the evolvability of insect genitalia.

Morphological integration (Cheverud 1988) is known to influence the variation that is available to selection and thus, the independent evolvability of traits (Wagner and Altenburg 1996; Wagner et al. 2007). Low integration could enhance the evolvability of male genitalic traits. Conversely, high integration could reduce maladaptive variability in somatic traits. If less integration does indeed equate to less constraint on evolution, the nature of selection favouring genitalic diversification must have disrupted integration of these traits. Furthermore, sexually selected traits have been shown to exhibit increased evolutionary rates, regardless of selection strength (Pitchers et al 2014). We therefore propose that male genitalia evolve more rapidly than male somatic traits because of differences in evolvability, rather than the strength of selection. For example, (Parzer et al. 2018) found that genitalia shape evolves faster than somatic traits in Onthopagus beetles. We suggest that this difference in evolutionary rates is related to lower integration of genitalic traits that are weakly coupled with body size and experience less functional constraint. Our study supports the idea of Eberhard et al (1998) in that differences among species in genitalia form are driven by changes in shape and relative size of genital micro-structures. Here, we would like to extend this further by inferring a more significant role of trait integration in the evolution of genital forms. Other traits that have lower magnitudes of integration have also been shown to have higher evolutionary flexibility or evolvability, e.g., the mammalian skull (Marroig et al. 2009), primate hands and feet (Rolian 2009) and even the human hip (Grabowski et al. 2011). While evolvability captures the degree of evolutionary response under selection in some

146 Genitalic and somatic trait integration

directions of trait space, flexibility captures the alignment of this response with the selection gradient (Marroig et al. 2009). However, although it seems probable that integration influences evolutionary flexibility and evolvability, the mechanisms that underly patterns of integration and their alignment with selection are still not well understood (Arnold et al. 2008; Eroukhmanoff et al. 2009) and particularly understudied within the insects.

A few studies suggest that patterns of integration can evolve in response to natural selection (Roff and Mousseau 1999). For example, there is evidence that natural selection has favoured reduced integration in mammalian skulls and numerous human characters, and that this allowed sets of traits to respond to separate selection pressures to a greater extent than what was previously possible, promoting the reintegration of traits (Marroig et al. 2009; Porto et al. 2009; Young et al. 2010; Grabowski et al. 2011). Although this has not been tested in insect genitalic traits, perhaps natural selection has operated similarly, breaking down morphological integration and thereby facilitating subsequent rapid divergence of genitalic form. For example, a high genetic covariance between wing and leg traits might enhance performance. Conversely, perhaps past natural selection favoured low genetic covariance between male genitalic traits because high variability in genital phenotypes increased the probability of individuals successfully mating with a wider range of females that varied in body size or some other trait(s). This increased compatibility could have increased individual fitness (similar to the pleiotropy hypothesis of genital divergence (Mayr 1963)). Since male genital morphology has been shown to influence patterns of paternity in a number of insect taxa (Arnqvist and Danielsson 1999; Danielsson and Askenmo 1999; House and Simmons 2002), and since male fitness is typically more limited by mating success than is female fitness, selection for compatibility could potentially explain low integration of male genitalic traits. Low integration could enhance male genitalic evolvability and facilitate the diversification of these traits by sexual selection.

Comparative studies across insect groups have shown that monandrous taxa have relatively simple genital morphology when compared to polyandrous groups that are characterised by high levels of post-copulatory sexual selection (Arnqvist and Thornhill 1998). Selection favouring increased genitalic complexity in polyandrous groups might also have acted to

147 Genitalic and somatic trait integration

reduce the integration of formerly highly correlated and relatively simple genitalic traits. There is some evidence that the insect genitalia evolved from the modification of a primitive appendage of a common ancestor to all arthropods (Matsuda 1976). Differential selection pressures acting on different components of the genitalic apparatus, and favouring compatibility with a range of female phenotypes, might have also reduced genitalic integration as a biproduct, in turn facilitating further diversification through sexual selection (a positive feedback mechanism). Although this study does not examine integration phylogenetically, it could be fruitful to examine differences in integration on a broad scale in multiple closely related species that show varying levels of genitalic diversification.

Sexual selection can also drive the evolution of heightened condition-dependence of traits. Overall, we found that low-condition males had higher genital and somatic trait integration than did high-condition males. In females, trait integration was similarly affected by larval diet, although the effect was not significant for genitalic traits. Nonetheless, despite these effects of larval diet on trait integration, the difference in integration between genitalic and somatic traits was maintained. The negative pattern of condition-dependence observed for trait integration also shows that, contrary to our predictions, high-condition individuals do not invest their extra resources into enhanced developmental stability and canalization. Rather, it appears that differential condition-dependence of different traits results in a reduction in trait integration in high-condition individuals (Figure 5). Perhaps this suggests that high-condition individuals invest in a phenotype that elevates fitness, even if that means reduced trait integration.

Compared to genitalic traits, we also show that the effect size of larval diet is higher for somatic traits, but not significantly different between the sexes. This suggests that male and female somatic traits are similar in their sensitivity to the larval nutritional environment. We also found a trend towards male genitalic traits being more responsive to larval diet, when compared to female genitalic traits, albeit this difference was non-significant. Full-sibling designs are known to inflate maternal effects (Kruuk & Hadfield 2007). It is therefore possible that maternal effects were lower for male genitals than somatic treats to influence the pattern of lower integration for these traits. However, little is known about the sex

148 Genitalic and somatic trait integration

differences in inheritance patterns for male and female genitalia in insects, although one study on bruchid beetles was able to show a strong maternal effect on sperm length (not classified as a genital trait here) (Gay et al 2009). Furthermore, the absolute conditional effect size was found to be larger for somatic traits, suggesting that G × E has a much higher influence on somatic trait size when compared to genitalic traits. This appears consistent with studies that show that G × E is prevalent in sexual traits (Greenfield and Rodriguez 2004; Bussière et al. 2008; Ingleby et al. 2010). However, for genitalic traits conditional effect sizes were not significantly different to the marginal effects of larval diet in either sex, suggesting that the genetic component to trait plasticity is relatively small.

Surprisingly, the G× E interaction was found to have a significant effect on somatic traits in females but not males, possibly indicating a sexual dimorphism in the alleles that affect trait shape and size variance. Single alleles are often known to have pleiotropic effects on many traits and the epistatic interactions that control their expression, and can be conditional on sex, and the environment (Mackay and Anholt 2006). What this suggests is that somatic trait variance in males may be relatively low and more driven by the environment and nongenetic factors, which is possible given that paternal and maternal effects of larval diet can have strikingly different effects on offspring traits (Bonduriansky et al. 2016). Furthermore, in Drosophila, approximately 40% of genetically variable transcripts associated with complex phenotypic modules show sexual dimorphism in their genetic variation (Ayroles et al. 2009).

These findings suggest that the patterns of genitalic and somatic trait covariation in an insect can change in response to the juvenile nutritional environment. The one size fits all hypothesis predicts that genital traits would show a higher level of integration when compared to somatic traits that include weaponry such as the antennae and forelegs, foreleg spines, testes, sperm bundle length etc. Traits such as legs and wings must still enable the insect to be mobile, so, it is not entirely surprising that these traits show a high level of integration when compared to genitalia. Like the classic study on water striders (Arnqvist and Thornhill 1998), we show that genitalic traits have sizeable variation but somewhat lower levels of condition-dependence when compared to general morphological traits.

149 Genitalic and somatic trait integration

Although genitalic traits were found to exhibit relatively low condition-dependence when compared to somatic traits, it is possible that genitalic and somatic traits compete for shared and limited resources during development. Allocation trade-offs during development have been shown to exist in Onthophagus taurus where larvae with ablated genitalic precursor cells develop larger horns (Moczek and Nijhout 2004). If these different tissue types are mismatched in their capacity to use or sequester the limited available resources during development then this type of competition could have an important impact on not just the relative sizes of these traits (Nijhout and Wheeler 1996), but also trait integration. Perhaps secondary sexual traits such as the legs, head length and other somatic traits in the neriid fly are much more efficient and uniform in drawing the available metabolic resources during development resulting in the relatively low integration of genitalic traits we observed. Hormones such as Juvenile Hormone (JH), ecdysteroids, and other growth factors are known to have important roles in regulation and degree of imaginal disk formation in insects (Nijhout 1994; Miner et al. 2000; Broglio et al. 2001; Nijhout and Grunert 2002). In D. melanogaster, the genitalia are derived from a single genital disc that acquires its patterning and shape during metamorphosis (Sánchez and Guerrero 2001). The gene doublesex (dsx) is known to contribute to sexual dimorphism in trait size (Loehlin et al. 2010; Robinett et al. 2010; Tanaka et al. 2011; Kijimoto et al. 2012) by influencing downstream pattern formation genes that are responsible for the development of the genitalia (Christiansen et al. 2002; Wheeler et al. 2006; Aspiras et al. 2011). The knockdown of dsx in other insects (e.g., in stag beetles) is also known to reduce JH sensitivity in developing structures and reduce their size (Gotoh et al. 2014). Perhaps the patterning and regulation of genital trait formation is not uniformly influenced by dsx and therefore creates a mosaic of JH sensitivity reducing the integration of genitalia. Given that nutrition is known to affect JH signalling in insects (Lavine et al. 2015), it makes sense that our larval diet manipulation during development had some influence on trait integration. In fact, a study using the horned flour beetle Gnatocerus cornatus, found that JH mediates the integration between exaggerated traits (horns and mandibles) and functionally supportive traits (e. g, femur width, wing) (Okada et al. 2012), so it is possible that JH also has a pleiotropic effect on genitalia. Furthermore, JH is thought to have evolved primarily as a reproductive hormone but also plays an important role in

150 Genitalic and somatic trait integration

juvenile ontogeny and metamorphosis (Nijhout and Wheeler 1982; Nijhout 1994; Gade et al. 1997; Hall and Wake 1999; Heyland et al. 2004). In Drosophila melanogaster, JH effects many other aspects of life history (see review Flatt et al. 2005). In future studies, and to clarify the mechanism of JH-mediated integration, it would be useful to compare the amount of JH-receptive elements/genes present in somatic and genitalic traits in the neriid fly.

We carried out the first study of the morphological integration and condition-dependence of genital and somatic traits. Our study is also one of the few to examine both male and female genital structures. We show that male genitalic traits exhibit low integration by comparison with other male triats and with female traits, and argue that this could account for the high evolvability of male genitalia in insects. We also show that male genitalic traits were relatively insensitive to nutrition, although integration itself was affected by nutrition. A lot of our knowledge on trait integration derives from the study of hard tissues in rats and primate skull structures (Esteve-Altava 2017) so directing our attention to under-represented organisms and patterns represented in soft tissues will deepen our understanding of the relative influence of internal and external factors on morphological integration. Additional empirical studies examining the regulation of trait integration and their condition-dependence may help us to better understand the processes of sexual selection that influence the evolution of genitalia, particularly in the insects.

Acknowledgements

The authors would like to express their appreciation to the Bonduriansky lab for all helpful discussion. This research was funded through an Australian Research Council Future Fellowship awarded to RB. The authors declare no conflict of interest.

References

Ah-King, M., A. B. Barron, and M. E. Herberstein. 2014. Genital Evolution: Why Are Females Still Understudied? PLoS Biol. 12:1–8.

Albertson, R. C., J. T. Streelman, T. D. Kocher, and P. C. Yelick. 2005. Integration and evolution of the cichlid mandible: The molecular basis of alternate feeding strategies.

151 Genitalic and somatic trait integration

Proc. Natl. Acad. Sci. 102:16287–16292.

Andersson, M. B. 1994. Sexual selection. Princeton University Press.

Arnold, S. J., R. Burger, P. A. Hohenlohe, B. C. Ajie, and A. J. Jones. 2008. Understanding the evolution and stability of the G-matrix. Evolution. 62:2451–2461.

Arnqvist, G. 1997. The evolution of animal genitalia: distinguishing between hypotheses by single species studies. Biological Journal of the Linnean Society. 60: 365-379.

Arnqvist, G. 1998. Comparative evidence for the evolution of genitalia by sexual selection. Nature 393:784–786.

Arnqvist, G., and I. Danielsson. 1999. Copulatory behavior, genital morphology and male fertilization success in water striders. Evolution. 53:147–156.

Arnqvist, G., and R. Thornhill. 1998. Evolution of animal genitalia : patterns of phenotypic and genotypic variation and condition dependence of genital and non • genital morphology in water strider (Heteroptera : Gerridae : Insecta) Evolution of animal genitalia : patterns of phenotypic an. Genet. Res. 71:193–212.

Aspiras, A. C., F. W. Smith, and D. R. Angelini. 2011. Sex-specific gene interactions in the patterning of insect genitalia. Dev. Biol. 360:369–380. Elsevier Inc.

Ayroles, J. F., M. A. Carbone, E. A. Stone, K. W. Jordan, R. F. Lyman, M. M. Magwire, S. M. Rollmann, L. H. Duncan, F. Lawrence, R. R. H. Anholt, and T. F. C. Mackay. 2009. Systems genetics of complex traits in Drosophila melanogaster. Nat. Genet. 41:299–307.

Barrett, R. D.H., and Schluter, D. 2007. Adaptation from standing genetic variation. Trends Ecol. Evol. 23: 38-44. https://doi.org/10.1016/j.tree.2007.09.008

Bates, D., M. Maechler, B. Bolker, and S. Walker. 2015. Fitting Linear Mixed-Effects Models Using lme4. Journal of Statistical Software. J. Stat. Softw. 67:1–48.

Birkhead, T. R. 2000. Promiscuity. An evolutionary history of sperm competition. Harvard

152 Genitalic and somatic trait integration

University Press, Cambridge.

Bonduriansky, R. 2006. Convergent evolution of sexual shape dimorphism in Diptera. J. Morphol. 267:602–611.

Bonduriansky, R., A. Runagall-McNaull, and A. J. Crean. 2016. The nutritional geometry of parental effects: maternal and paternal macronutrient consumption and offspring phenotype in a neriid fly. Funct. Ecol.

Breno, M., Leirs, H., and Van Dongen, S. 2011. Traditional and geometric morphometrics for studying skull morphology during growth in Mastomys natalensis (Rodentia: Muridae). Journal of Mammology 92(6): 1395-1406.

Broglio, W. H., T. I. Stocker, F. Rintelen, R. Fernandez, and E. Hafen. 2001. An evolutionarily conserved function for the Drosophila insulin receptor and insulinlike peptides in growth control. Curr. Biol. 11:213–221.

Bussière, L. F., J. Hunt, K. N. Stölting, M. D. Jennions, and R. Brooks. 2008. Mate choice for genetic quality when environments vary: Suggestions for empirical progress. Genetica 134:69–78.

Carter, A. J. R., J. Hermisson, and T. F. Hansen. 2005. The role of epistatic gene interactions in the response to selection and the evolution of evolvability. Theor. Popul. Biol. 68:179–196.

Cheverud, J. M. 1988. A Comparison of Genetic and Phenotypic Correlations. Evolution. 42:958.

Cheverud, J. M. 1982. Phenotpyic, genetic, and environmental morphological integration in the cranium. Evolution. 36:499–516.

Christiansen, A. E., E. L. Keisman, S. M. Ahmad, and B. S. Baker. 2002. Sex comes in from the cold: the integration of sex and pattern. Trends Genet. 18:510–516.

Cooper, B. S., A. Sedghifar, W. T. Nash, A. A. Comeault, and D. R. Matute. 2018. A

153 Genitalic and somatic trait integration

Maladaptive Combination of Traits Contributes to the Maintenance of a Drosophila Hybrid Zone. Curr. Biol. 28:2940–2947.

Danielsson, I., and C. Askenmo. 1999. Male genital traits and mating interval affect male fertilization success in the water strider Gerris lacustris. Behav. Ecol. Sociobiol 46:149– 156.

Dufour, L. 1844. Anatomie generale des dipteres. Annu. Sci. Nat. I:244–264.

Eberhard, W. G. 2011. Experiments with genitalia: A commentary. Trends Ecol. Evol. 26:17–21. Elsevier Ltd.

Eberhard, W. G. 1996. Female Control: Sexual Selection by Cryptic Female Choice. Princeton University Press, New Jersey.

Eberhard, W. G. 1985. Sexual Selection and Animal Genitalia. Harvard University Press.

Eberhard, W. G. 2009. Static allometry and animal gentalia. Evolution . 63:48–66.

Eberhard, W. G., B. A. Huber, R. L. Rodriguez, R. D. Briceño, I. Salas, and V. Rodriguez. 1998. One Size Fits All ? Relationships Between the Size and Degree of Variation in Genitalia and Other Body Parts in Twenty Species of Insects and Spiders Author ( s ): William G . Eberhard , Bernhard A . Huber , Rafael Lucas Rodriguez S ., R . Daniel Briceño. Evolution. 52:415–431.

Eden, M. 1967. Inadequacies of neo-darwinian evolution as a scientific theory. P. in P. Moorhead and M. Kaplan, eds. Mathematical Challanges to the Neo-Darwinian intepretation of evolution. Wistar Inst. Press, Philadelphia.

Edwards, R. 1993. Entomological and mammalogical perspectives on genital differentiation. Trends Ecol. Evol. 8:406–409.

Ehrich, T. H., T. Y. T. Vaughn, S. F. Koreishi, R. B. Linsey, L. S. Pletscher, and J. M. Cheverud. 2003. Pleiotropic Effects on Mandibular Morphology I. Developmental Morphological Integration and Differential Dominance. J. Exp. Zool. 296B:58–79.

154 Genitalic and somatic trait integration

Eroukhmanoff, F., D. Outomuro, F. Ocharan, and E. Svensson. 2009. Patterns of phenotypic divergence in wing covariance structure of calopterygid damselflies. Evol. Biol. 36:214–224.

Esteve-Altava, B. 2017. In search of morphological modules: a systematic review. Biol. Rev. 92:1332–1347.

Firmat, C., I. Lozano-Fernandez, J. Agusti, G. H. Bolstad, G. Cuenca-Bescos, T. F. Hansen, and C. Pelabon. 2014. Walk the line: 600 000 years of molar evolution constrained by allometry in the fossil rodent Mimomys savini. Philos. Trans. R. Soc. B 369.

Flatt, T., M. P. Tu, and M. Tatar. 2005. Hormonal pleiotropy and the juvenile hormone regulation of Drosophila development and life history. BioEssays 27:999–1010.

Gade, G., K. H. Hoffman, and J. H. Spring. 1997. Hormonal regulation in insects: facts, gaps, and future directions. Physiol Rev 77:963–1032.

Gay, L., Hosken, D.J., Vasudev, R., Tregenza, T., and Eady, P. E. 2009. Sperm competition and maternal effects differentially influence testis and sperm size in Callosobruchus maculatus. J. Evol. Biol. 22: 1143-1150. doi:10.1111/j.1420-9101.2009.01724.x

Genevcius, B. C., and C. F. Schwertner. 2017. Strong functional integration among multiple parts of the complex male and female genitalia of stink bugs. Biol. J. Linn. Soc. 122:774–786.

Gotoh, H., H. Miyakawa, A. Ishikawa, Y. Ishikawa, Y. Sugime, D. J. Emlen, and L. Lavine. 2014. Developmental link between sex and nutrition; doublesex regulates sex-specific mandible growth via juvenile hormone signaling in stag beetles. PLoS Genet. 10.

Gould, S. J. 1977. Ontogeny and phylogeny. Belknap Press, Cambridge, MA.

Gould, S. J. 1974. The origin and function of “bizarre” structures: antler size and skull size in the “Irish elk.” Megaloceros giganteus. Evolution. 28:191–220.

Gould, S. J. 2002. The structure of evolutionary theory. Belknap Press, Cambridge, MA.

155 Genitalic and somatic trait integration

Grabowski, M. W., J. D. Polk, and C. C. Roseman. 2011. Divergent patterns of integration and reduced constraint in the human hip and the origins of bipedalism. Evolution. 65:1336–1356.

Greenfield, M. D., and R. L. Rodriguez. 2004. Genotype–environment interaction and the reliability of mating signals. Anim. Behav. 68:1461–1468.

Hall, B. K., and M. H. Wake. 1999. The origin and evolution of larval forms. Academic Press, San Diego.

Hansen, T. F., J. M. Alvarez-Castro, A. J. R Carter, J. Hermisson, and G. P. Wagner. 2006. Evolution of Genetic Architecture Under Directional Selection. Evolution. 60:1523– 1536.

Hansen, T. F., W. S. Armbruster, M. L. Carlson, and C. Pélabon. 2003. Evolvability and Genetic Constraint in Dalechampia Blossoms: Genetic Correlations and Conditional Evolvability. J. Exp. Zool. Part B Mol. Dev. Evol. 296:23–39.

Hermisson, J. and Pennings, P. S. 2005. Soft sweeps: Molecular population genetics of adaptation from standing genetic variation. Genetics. 169: 2335-2352. https://doi.org/10.1534/genetics.104.036947

Heyland, A., J. Hodin, and A. M. Reitzel. 2004. Hormone signaling in evolution and development: a non-model system approach. BioEssays 27:64–75.

Higgins, J. J., and S. Tashtoush. 1994. An aligned rank transform test for interaction. Nonlinear World 1:201–211.

Hosken, D. J., and P. Stockley. 2004. Sexual selection and genital evolution. Trends Ecol. Evol. 19:87–93.

House, C. M., and L. W. Simmons. 2002. Genital morphology and fertilization success in the dung beetle Onthophagus taurus: An example of sexually selected male genitalia. Proc. R. Soc. B Biol. Sci. 270:447–455.

156 Genitalic and somatic trait integration

House, C. M., and L. W. Simmons. 2007. No evidence for condition-dependent expression of male genitalia in the dung beetle Onthophagus taurus. J. Evol. Biol. 20:1322–1332.

Huxley, J. 1932. Problems of relative growth. Dover, New York.

Ingleby, F. C., J. Hunt, and D. J. Hosken. 2010. The role of genotype-by-environment interactions in sexual selection. J. Evol. Biol. 23:2031–2045.

Johnstone, R. A. 1995. Sexual selection, honest advertisement and the handicap principle: reviewing the evidence. Biol. Rev. 70:1–65.

Kay, M., and J. Wobbrock. 2019. ARTool: Aligned Rank Transform for Nonparametric Factorial ANOVAs.

Kijimoto, R., A. . Moczek, and J. Andrews. 2012. Diversification of doublesex function underlies morph-, sex-, and species-specific development of beetle horns. Proc Natl Acad Sci U S A 109:20526–31.

Klingenburg, C. P. 2016. Size, shape and form: concepts of allometry in geometric morphometrics. Dev Genes Evol. 226: 113-137. Doi: 10.1007/s00427-016-0539-2

Kruuk, L. E. B., and Hadfield, J. D. 2007. How to separate genetic and environmental causes of similarity between relatives. J. Evol. Biol. 20: 1890-1903. https://doi.org/10.1111/j.1420-9101.2007.01377.x

Langerhans, R. B., C. M. Anderson, and J. L. Heinen-Kay. 2016. Causes and Consequences of Genital Evolution. Integr. Comp. Biol. 56:741–751.

Lavine, L., H. Gotoh, C. S. Brent, I. Dworkin, and D. J. Emlen. 2015. Exaggerated Trait Growth in Insects. Annu. Rev. Entomol. 60:453–472.

Leamy, L. J., D. Pomp, E. J. Eisen, and J. M. Cheverud. 2002. Pleiotropy of quantitative trait loci for organ weights and limb bone lengths in mice. Physiol. Genomics 2002:21–29.

Loehlin, D. W., D. C. Oliveira, R. Edwards, J. D. Giebel, M. E. Clark, M. V. Cattani, L. Van de Zande, E. C. Verhulst, L. W. Beukeboom, M. Munoz-Torres, and J. H. Werren.

157 Genitalic and somatic trait integration

2010. Non-coding changes cause sex-specific wing size differences between closely related species of Nasonia. PLoS Genet. 6.

Lynch, M., and B. Walsh. 1998. Genetics and Analysis of Quantitative Traits. Sinauer, Massachussetts.

Macintyre, P. D., A. Van Niekerk, M. P. Dobrowolski, J. L. Tsakalos, and L. Mucina. 2018. Impact of ecological redundancy on the performance of machine learning classifiers in vegetation mapping. Ecol. Evol. 8:6728–6737.

Mackay, T. F. C., and R. R. H. Anholt. 2006. Of flies and man: Drosophila as a model for human complex traits. Annu. Rev. Genomics Hum. Genet. 7:339–367.

Marroig, G., L. T. Shirai, A. Porto, F. B. de Oliveira, and V. de Conto. 2009. The evolution of modularity in the mammalian skull II: Evolutionary consequences. Evol. Biol. 36:136–148.

Matsuda, R. 1976. Morphology and evolution of the insect abdomen. Pergamon Press, Oxford.

Mayr, E. 1963. Animal Species and Evolution. Harvard University Press.

Mezey, J. G., J. M. Cheverud, and P. Wagner. 2000. Is the Genotype-Phenotype Map Modular?: A Statistical Approach Using Mouse Quantitative Trait Loci Data. Genetics 156:305–311.

Miner, A. L., A. J. Rosenburg, and H. F. Nijhout. 2000. Control of growth and differentiation of the wing imaginal disk of Precis coenia (Lepidoptera: Nymphalidae). J. Insect Physiol. 46:251–258.

Moczek, A. P., and H. F. Nijhout. 2004. Trade-offs during the development of primary and secondary sexual traits in a horned beetle. Am. Nat. 163:184–191.

Nakagawa, S., and H. Schielzeth. 2013. A general and simple method for obtaining R2 from generalized linear mixed-effects models. Methods Ecol. Evol. 4:133–142.

158 Genitalic and somatic trait integration

Nijhout, H. F. 1994. Insect Hormones. Princeton University Press, Princeton.

Nijhout, H. F., and L. W. Grunert. 2002. Bombyxin is a growth factor for wing imaginal disks in Lepidoptera. Proc. Natl. Acad. Sci. USA 99:15446–15450.

Nijhout, H. F., and D. E. Wheeler. 1996. Growth Models of Complex Allometries in Holometabolous Insects. Am. Nat. 148:40–56.

Nijhout, H. F., and D. E. Wheeler. 1982. Juvenile hormone and the physiological basis of insect polymorphisms. Q. Rev. Biol. 57:109–133.

Okada, Y., H. Gotoh, T. Miura, T. Miyatake, and K. Okada. 2012. Juvenile hormone mediates developmental integration between exaggerated traits and supportive traits in the horned flour beetle Gnatocerus cornutus. Evol. Dev. 14:363–371.

Olson, E. C., and R. C. Miller. 1958. Morphological Integration. University of Chicago Press.

Parzer, H. F., P. David Polly, and A. P. Moczek. 2018. The evolution of relative trait size and shape: insights from the genitalia of dung beetles. Dev. Genes Evol. 228:83–93. Development Genes and Evolution.

Pavlicev, M., J. M. Cheverud, and G. P. Wagner. 2011. Evolution of adaptive phenotypic variation patterns by direct selection for evolvability. Proc. R. Soc. B Biol. Sci. 278:1903–1912.

Pavlicev, M., J. M. Cheverud, and G. P. Wagner. 2009. Measuring morphological integration using eigenvalue variance. Evol. Biol. 36:157–170.

Pélabon, C., C. Firmat, G. H. Bolstad, K. L. Voje, D. Houle, J. Cassara, A. Le Rouzic, and T. F. Hansen. 2014. Evolution of morphological allometry. Ann. N. Y. Acad. Sci. 1320:58–75.

Petrie, M. 1988. Intraspecific variation in structures that display competitive ability: large animals invest relatively more. Anim. Behav. 36:1174–1179.

159 Genitalic and somatic trait integration

Pigliucci, M. 2003. Phenotypic integration: Studying the ecology and evolution of complex phenotypes. Ecol. Lett. 6:265–272.

Pitchers, W., J. B. Wolf, T. Tregenza, J. Hunt, and I. Dworkin. 2014. Evolutionary rates for multivariate traits: The role of selection and genetic variation. Philos. Trans. R. Soc. B Biol. Sci. 369.

Pitnick, S., T. Markow, and G. S. Spicer. 1999. Evolution of Multiple Kinds of Female Sperm-Storage Organs in Drosophila. Evolution. 53:1804.

Pomiankowski, A., and A. P. Moller. 1995. A resolution of the lek paradox. Proc. R. Soc. London, Ser. B 260:21–29.

Porto, A., F. B. de Oliveira, L. T. Shirai, V. de Conto, and G. Marroig. 2009. The evolution of modularity in the mammalian skull I: Morphological integration patterns and magnitudes. Evol. Biol. 36:118–135.

R Core Team. 2019. R: A Language and Environment for Statistical Computing. Vienna, Austria.

Richmond, M. P., J. Park, and C. S. Henry. 2016. The function and evolution of male and female genitalia in Phyllophaga harris scarab beetles (Coleoptera: Scarabaeidae). J. Evol. Biol. 29:2276–2288.

Riedl, R. 1978. Order in living organisms: a systems analysis of evolution. Wiley, New York, NY.

Robinett, L. M., A. G. Vaughan, J. M. Knapp, and B. S. Baker. 2010. Sex and the single cell. II. There is a time and place for sex. PLoS Biol. 8.

Roff, D. A., and T. A. Mousseau. 1999. Does natural selection alter genetic architecture? An evaluation of quantitative genetic variation among populations of Allonemobius socius and A. fasciatus. J. Evol. Biol. 12:361–369.

Rolian, C. 2009. Integration and evolvability in primate hands and feet. Evol. Biol. 36:100–

160 Genitalic and somatic trait integration

117.

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

Salter, K. C., and R. F. Fawcett. 1993. The art test of interaction: A robust and powerful rank test of interaction in factorial models. Commun. Stat. Simul. Comput. 22:137– 153.

Sánchez, L., and I. Guerrero. 2001. The development of the Drosophila genital disc. BioEssays 23:698–707.

Scharloo, W. 1991. Canalization: Genetic and developmental aspects. Annu. Rev. Ecol. Syst. 22:65-93. https://doi.org/10.1146/annurev.es.22.110191.000433

Simmons, L. W. 2014. Sexual selection and genital evolution. Austral Entomol. 53:1–17.

Simmons, L. W. 2001. Spem competition and its evolutionary consequences in the insects. Princeton University Press.

Simmons, L. W., and D. J. Emlen. 2006. Evolutionary trade-off between weapons and testes. Proc. Natl. Acad. Sci. 103:16346–16351.

Simmons, L. W., C. M. House, J. Hunt, and F. García-González. 2009. Evolutionary Response to Sexual Selection in Male Genital Morphology. Curr. Biol. 19:1442–1446.

Sloan, N. S., and Simmons, L. W. 2019. The evolution of female genitalia. J. Evol. Biol. 32(9): 882-889.

Stoffel, M. A., S. Nakagawa, and H. Schielzeth. 2017. rptR: repeatability estimation and variance decomposition by generalized linear mixed-effects models. Methods Ecol. Evol. 8.

161 Genitalic and somatic trait integration

Tanaka, K., O. Barmina, L. E. Sanders, M. N. Arbeitman, and A. Kopp. 2011. Evolution of sex-specific traits through changes in HOX-dependent doublesex expression. PLoS Biol. 9.

Tomkins, J. L., J. Radwan, J. S. Kotiaho, and T. Tregenza. 2004. Genic capture and resolving the lek paradox. Trends Ecol. Evol. 19:323–328.

Tomkins, J. L., Kotiaho, J. S., and LeBas, N. R. 2005. Phenotypic plasticity in the developmental integration of morphological trade-offs and secondary sexual trait compensation. Proc. R. Soc. B 272: 543-551. doi:10.1098/rspb.2004.2950

Vallejo-Marin, M., C. Walker, P. Friston-Reilly, L. Solis-Montero, and B. Igic. 2014. Recurrent modification of floral morphology in heterantherous Solanum reveals a parallel shift in reproductive strategy. Philos. Trans. R. Soc. B 369.

Voje, K. J., T. F. Hansen, C. K. Egset, G. H. Bolstad, and C. Pelabon. 2014. Allometric constraints and the evolution of allometry. Evolution. 68:866–885.

Wagner, G. P., and L. Altenburg. 1996. Complex adaptations and evolution of evolvability. Evolution. 50:967–976.

Wagner, G. P., Booth, G., and Bagheri-Chaichian, H. 1997. A population genetic theory of canalization. Evolution. 51:329-347. https://doi.org/10.1111/j.1558- 5646.1997.tb02420.x

Wagner, G. P., M. Pavlicev, and J. M. Cheverud. 2007. The road to modularity. Nat. Rev. Genet. 8:921–931.

Werner, M., and L. W. Simmons. 2008. The evolution of male genitalia: Functional integration of genital sclerites in the dung beetle Onthophagus taurus. Biol. J. Linn. Soc. 93:257–266.

Wheeler, D. E., N. Buck, and J. D. Evans. 2006. Expression of insulin pathway genes during the period of caste determination in the honey bee, Apis mellifera. Insect Mol. Biol. 15:597–602.

162 Genitalic and somatic trait integration

Yap, M. H. W., M. J. Grabowska, C. Rohrscheib, R. Jeans, M. Troup, A. C. Paulk, B. van Alphen, P. J. Shaw, and B. van Swinderen. 2017. Oscillatory brain activity in spontaneous and induced sleep stages in flies. Nat. Commun. 8:1815.

Young, N. M., G. P. Wagner, and B. Hallgrímsson. 2010. Development and the evolvability of human limbs. Proc. Natl. Acad. Sci. U. S. A. 107:3400–3405.

163

CHAPTER 5

Perceived dominance status affects chemical signalling in the neriid fly Telostylinus angusticollis.

2 3 1 Zachariah Wylde1, Lewis Adler , Angela Crean & Russell Bonduriansky

Animal Behaviour (2019) 158: 161-174

1Evolution and Ecology Research Centre, School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney, 2052, Australia

2Bioanalytical Mass Spectrometry Facility, Mark Wainwright Analytical Centre, University of New South Wales, Sydney, 2052, Australia

3Charles Perkins Centre, The University of Sydney, Sydney, New South Wales, Australia

164 Perceived dominance changes chemical signalling

Abstract

Chemical communication mediates many social interactions in insects but is still less well understood than other forms of communication. In particular, chemical signalling of social dominance is believed to play an important role in competitive interactions in both sexes, but much of the evidence is correlational. Here we manipulated social dominance and examined its effect on CHC profiles in Telostylinus angusticollis, a fly with a resource defence polygyny mating system. Focal individuals’ perception of their own dominance status was manipulated by placing them into an arena with larger or smaller competitor individuals to render them ‘subordinate’ or ‘dominant.’ We found that social dominance treatment affected males’ and females’ social status (quantified as proximity to the larval medium/oviposition dish), as well as their CHC profiles. Dominant individuals tended to have CHC profiles less similar to those of the opposite sex. Moreover, dominant females exhibited an overall elevation of all CHC expression, relative to subordinate females, whereas males that perceived themselves as subordinate exhibited a near-significant down-regulation of male-limited CHCs. Our findings suggest that T. angusticollis males and females alter their CHC profiles in response to their self-perceived social dominance status. These chemical signals could play a role in social interactions both within and between the sexes.

Keywords: Chemical communication, Cuticular hydrocarbons, Neriidae, Perceived Dominance status, Social interaction, Telostylinus angusticollis.

165 Perceived dominance changes chemical signalling

Introduction

Highly social animals have evolved complex signalling strategies that quickly respond to social status changes but may not be so obvious to the human observer (Wyatt 2003; Blomquist and Bagneres 2010). One such strategy is chemical signalling which has been well documented in animals that mark their territories with scent, where odour can be regarded as a secondary sexual trait – like antlers and bird plumage, and which often occur with other ritualised and conspicuous traits (Blaustein 1981; Dawkins 1995). Chemical signals could also potentially function in intrasexual dominance signalling, but how such signals may be utilised as a sign of status is not well known – especially in insects that do not exhibit eusociality. Moreover, since any trait that conveys information to another individual can be regarded as a signal (Dawkins 1995; Maynard Smith and Harper 1995), chemical signals could occur as nonadaptive responses (e.g. to stress) that are exploited by other individuals. Although many non-eusocial insects display aggregation behaviour (Waters 1959) where individuals show no cooperation or division of labour but gather to breed, how the social and environmental context impacts chemical cues in such insects is poorly understood - and even less is known about the roles these cues have in structuring dominance hierarchies (Gershman, Toumishey, & Rundle, 2014; Grillet, Dartevelle, & Ferveur, 2006; Lin & Michener, 1972; Savarit & Ferveur, 2002)

Cuticular hydrocarbons (CHCs) are mostly long chained non-volatile compounds that derive from fatty acid compounds found on the cuticle of various insect species (Everaerts et al. 2010). These complex chemicals have been implicated in desiccation resistance (Wigglesworth 1933) while simultaneously acting as signalling molecules in short-range chemical communication ( Gershman et al., 2014; Gibbs, 2007). Species vary greatly in the chemical compositions of their CHC signals (El-Sayed 2009), and can use CHCs to identify conspecifics. For example, D. melanogaster females use species-specific CHCs to locate egg- laying sites used by other members of their species (Duménil et al. 2016). However, a great deal of within-species variation in CHCs is also evident. For example, the expression of CHCs has been shown to be responsive to differences in age, diet, social environment, and mating history (Petfield et al. 2005; Kent et al. 2008; Yew et al. 2008; Everaerts et al. 2010;

166 Perceived dominance changes chemical signalling

Curtis et al. 2013; Gershman and Rundle 2016b,a). CHCs can play a role in female mate choice (Johansson and Jones 2007), and can evolve rapidly when natural and sexual selection pressures are altered (Blows 2002; Chenoweth and Blows 2008). There is mounting empirical evidence that sexual selection can promote the evolution of chemical traits but that the type and intensity of selection on chemical signals varies between species (for review see Steiger & Stökl, 2014).

While CHCs can convey information about an individual’s status and sex, most research has focused on signalling of fertility, reproductive status and dominance in social insects like wasps and ants (Smith et al. 2009; Izzo et al. 2010). For example, male ants (Cardiocondyla obscurior) can avoid aggression from wingless males by mimicking the chemical bouquet of virgin female queens (Cremer et al. 2002). Less is known about the role of CHCs in social signalling in non-social insects. It has been shown that male Drosophila melanogaster actively mark females during mating with anti-aphrodisiac pheromones— a form of chemical mate- guarding that functions to decrease female attractiveness (Laturney and Billeter 2016). There is also evidence that individuals can adjust their own CHC profiles in response to mating (Weddle et al. 2013).

Yet, it remains unclear whether individuals in non-eusocial species can change their CHC signalling in response to their own social status within a group. It could be advantageous to do so to intimidate potential rivals, signal dominance to potential mates, or avoid costly interactions with rivals. For example, in some species of animals (e.g. flat lizards; (Whiting et al. 2009); bluegill sunfish (Dominey 1980); giant cuttlefish (Norman et al. 1999)), subordinate or sneaker males mimic females. Likewise, in rove beetles (Aleochara curtula) young and starved males can re-gain access to a carcass by producing the female sex pheromone (Peschke 1987). This suggests that male insects that experience low social status may be able to utilise CHC profiles as a type of camouflage to avoid aggression from more dominant competitors. More generally, signals of dominance can be important, particularly in species with aggressive and damaging interactions, in allowing competitors to assess their rivals and thereby lower associated fighting costs (Maynard Smith 1982). These dominance signals may be particularly important not just in eusocial species that have strict caste

167 Perceived dominance changes chemical signalling

structures (i.e. ants and wasps), but also in non-eusocial species where individuals compete for resources such as mate-searching territories, position in dominance hierarchies, or mating opportunities (Izzo et al. 2010). It is clear that CHCs are remarkably dynamic, often changing (either as part of an adaptive strategy, or nonadaptively as a side-effect of stress or changes in other traits) within short timeframes in response to experience or fluctuations in the social environment (Ingleby 2015). Thus, CHC profiles could change to reflect individuals’ social status, and serve as an important social signal in non-eusocial insects.

An important limitation of the existing literature on the role of CHCs in status signalling is that much of the evidence is correlational. For example, some studies pair competing individuals in an arena to determine dominance status or to assess winner and loser effects, and report effects on chemical signals (e.g., Rillich & Stevenson, 2011; Thomas & Simmons, 2009). There is also some evidence in Drosophila serrata that CHCs profiles correlate with mating success but not with an individual’s ability to successfully defend a territory (White and Rundle 2014). Yet, it is not clear if differences in chemical signals between dominant and subordinate individuals reflect a perceived social status that can change dynamically, or whether these observed differences in both dominance and chemical profile are invariant features of adult individuals that result from genetic or environmental differences during development. However, to determine whether individuals can adjust their own CHC profiles dynamically in response to their social environment, experimental studies that manipulate rather than simply assess individuals’ social status are required. To our knowledge, only two experimental studies have been carried out that directly manipulated the social environment and examined the fine-scale effects on CHC signalling. Using the Australian field cricket (Teleogryllus oceanicus) Thomas, Gray and Simmons (2011) showed that males increased expression of CHCs (some of which are sexually dimorphic, and have been shown to attract females) in the absence of acoustic signals from other courting males. Another study showed that social environment can alter the circadian rhythm of CHCs associated with male attractiveness in Drosophila serrata, with the combination of CHCs that contributes to increased mating success varying over the course of a day as well as in response to social conditions (Gershman & Rundle, 2016; Gershman et al., 2014; Gershman

168 Perceived dominance changes chemical signalling

& Rundle, 2016). However, we are not aware of any previous experimental study that has examined whether an individual’s CHC profile can respond dynamically to its perception of its position in a dominance hierarchy. Examining such responses could reveal cryptic reproductive tactics. For example, subordinate males could change their CHC profiles to more closely align with female CHC profiles, facilitating sneak matings and reducing risk of attacks by dominant males. Although much less is known about competition among females, it is possible that females could also alter their CHC profiles to intimidate rivals in competition for food or egg-laying sites, or to avoid costly sexual interactions such as male harassment.

Here we investigate the effects of individuals’ perception of their own social status on CHC composition in Telostylinus angusticollis (Diptera: Neriidae). T. angusticollis forms large mating aggregations, in which females and males aggregate at oviposition sites on decaying tree bark (Kawasaki et al. 2008; Adler and Bonduriansky 2013). Individuals vary considerably in body size (Bonduriansky, 2007; Bonduriansky, 2006), and large males defend territories and frequently engage in combat while smaller males rarely fight and appear to employ non- territorial tactics (Bath et al. 2012; Hooper et al. 2017). Combat success is strongly related to body size (Bonduriansky & Head, 2007; Hooper et al., 2017). Small and subordinate males also exhibit increased mating duration (Fricke et al. 2015). Females vary considerably in body size as well and have been observed to interact aggressively at oviposition sites (ZW and RB, personal observations). However, it is not known whether T. angusticollis males or females employ chemical signalling tactics or adjust their CHC profiles in response to their social status within an aggregation.

We manipulated the self-perceived dominance status of focal males and females by placing them into social environments consisting of same-sex competitors of either larger or smaller body size than the focal individual. This experiment enabled us to examine (1) whether CHC profiles of males and females are responsive to perceived dominance status; (2) whether subordinate individuals exhibit CHC profiles resembling those of the opposite sex; and (3) whether the sexes respond differently to cues of dominance status in their social environment.

169 Perceived dominance changes chemical signalling

Materials and methods

Fly culturing

Flies for use in the chemical identification and quantitation of epicuticular compounds were derived from lab-reared stocks of T. angusticollis (Enderlein) (Diptera; Neriidae) that originate from individuals collected in 2017 at Fred Hollows Reserve, Randwick, NSW, Australia (33°54′44.04″S 151°14′52.14″E) and were reared in the laboratory for four generations prior to this experiment. While the original number of insects was not quantified, we ensured genetic diversity by supplementing the lab-reared population from wild-caught individuals. All lab-bred individuals were reared in climate chambers at 25°C ± 2°C with a 12:12 photoperiod and provided with water every two days. We manipulated the adult body size of individuals used in the experiment by rearing larvae on either a nutrient- rich, nutrient-intermediate (henceforth, “standard”) or nutrient-poor larval diet. Diets were based on Sentinella et al. (2013) and were selected to generate considerable body size differences between competitor flies used in dominance treatments. All diets consisted of a base of 170g of cocopeat moistened with 600mL of reverse osmosis-purified water. The “rich” larval diet consisted of 32.8g of protein (Nature’s Way soy protein isolate; Pharm-a- Care, Warriewood, Australia) and 89g of raw brown sugar, the “standard” larval diet consisted of 10.9g of protein and 29.7g raw brown sugar, and the “poor” larval diet consisted of 5.5g of protein and 14.8g raw brown sugar. These nutrients were mixed into the cocopeat and water using a handheld blender and frozen at -20°C until the day of use.

Virgin adults were collected at emergence and separated by sex, larval diet treatment, and emergence date to control for age (± 2 days) across all treatments. Age is known to effect CHC profiles in flies (Gershman & Rundle, 2016). All adult flies were allowed to mature in individual 120 mL containers fitted with a feeding tubes containing a sugar-yeast mixture and a drinking tube containing water ad libitum, and a substrate of moistened cocopeat. All males were kept in these containers until 5 ± 2 days of age (when males are fully reproductively mature), whereas females were kept until 12 ± 2 days of age (the typical age of full ovary development), prior to dominance treatments and assays. All adults were housed

170 Perceived dominance changes chemical signalling

and dominance treatments applied in a controlled-temperature room set at 25°C and 60% humidity and a 14:10 hour light/ dark cycle.

Manipulation of dominance status

Males of this species engage in escalated combat interactions (foreleg strikes and headlock, ‘chest’ impacts; Figure 1) with rivals of similar body size, but a male challenged by a larger rival usually withdraws and displays submissive behaviours (Bath et al. 2012). Large, dominant males defend oviposition sites or females by chasing away or flicking their wings at males that attempt takeover. Females do not show the same aggression behaviours as males but can ‘wing flick’ or engage in brief bouts of foreleg boxing with other females (Figure S1B). Thus, the mean body size of rivals with which a T. angusticollis individual interacts could determine its perception of its own place in the dominance hierarchy. To examine the plasticity of CHC profiles in response to the social environment we placed focal individuals (all reared on a standard larval diet) into an arena for 48 hours with three competitor individuals of the same sex that were either reared on nutrient-rich larval diet (such that competitors were larger than the focal) or nutrient-poor larval diet (such that competitors were smaller than the focal). These social environment treatments enabled us to directly manipulate the focal individual’s self-perception of its social dominance so as to render it ‘subordinate’ or ‘dominant’ within its social environment (Figure 2).

171 Perceived dominance changes chemical signalling

Figure 1. Male neriid flies engaged in escalated combat (on left). On the right a male guards a female as she oviposits into a damaged area of a coral tree (Erythrina spp.). Photo courtesy of Russell Bonduriansky.

Focal individuals were randomly assigned to ‘dominant’ or ‘subordinate’ treatment groups, and placed individually into competitive arenas for 48 hours, which is a sufficient length of time for neriid flies to establish a dominance hierarchy (Bonduriansky & Head, 2007). The arenas consisted of 1L cylindrical containers covered with mesh stockings. Each arena contained a layer of moistened cocopeat and a 3cm diameter petri dish containing oviposition medium in the centre.

The oviposition medium stimulated males and females to engage in reproductive behaviour similar to those observed at natural oviposition sites in the wild (Figure 1). At the end of the 48-hour period, each focal individual was observed for 10 mins and any aggressive interactions with competitor individuals were recorded (Figure 3). The focal individual’s distance from the oviposition site was also estimated in body lengths every 2.5 mins during the behavioural observation time (Figure A1A). The mean distance of the focal individual was then used for further analyses. These observations enabled us to assess focal individuals’ dominance status relative to competitors, in order to determine whether the treatments were 172 Perceived dominance changes chemical signalling

successful. Immediately following the 10-min observation period, entire arenas were placed in a -80°C freezer to anaesthetise flies quickly and minimise any effects of stress that might affect CHC profiles. All focal individuals were then stored in Eppendorf tubes at -80°C until chemical analysis. Additional flies, reared on each of the three larval diets, were kept individually (without competitors) in identical arenas to test for an effect of larval diet on CHC profile and determine whether dominance treatment effects could be explained simply as effects of competitors’ CHCs transferring onto focal individuals.

Figure 2. Experimental design. (a) Dominant treatment: focal male or female (blue) surrounded by three competitors reared on a poor larval diet, and therefore smaller than the 173 Perceived dominance changes chemical signalling

focal individual (yellow). (b) Subordinate treatment: focal male or female (blue) surrounded by three competitors reared on a rich larval diet and therefore larger than the focal individual (red). All individuals were kept in these social environments for 48 hours. Brown circle in the centre of arenas symbolises the oviposition food used to elicit competitive behaviours. Only the CHCs of focal individuals were extracted.

Extraction of epicuticle hydrocarbons

Single fly extractions

Focal flies were thawed for 15 mins at room temperature before hydrocarbons were extracted by immersing single flies in 100 μl of hexane (Sigma-Aldrich, Australia product # 650552) spiked with a 10 μl/mL of hexacosane (Sigma-Aldrich, product # 241687) internal standard. Individual flies were placed in 2 mL autosampler vials (Agilent, Australia) and were immersed in hexane for 3 mins and vortexed for 1 min before the fly was removed. Water was removed from each extraction by filtering the elution through a glass pasteur pipette packed with silane treated glass wool (Alltech, Australia) and a small amount of anhydrous sodium sulphate (Ajax Finechem, Australia, 503-500). Extracts were stored at -20°C until analysis. Following the CHC extraction all flies were frozen at -20°C for subsequent morphometric analysis. Thorax length is a reliable proxy for body size in this species (Bonduriansky, 2006) and was measured from images taken using a Leica MS5 stereoscope equipped with a Leica DFC420 digital microscope camera. Measurements were made using FIJI open source software (Schindelin et al., 2012).

Identification of CHC compounds

To aid in the identification of individual compounds present in the cuticle of T. angusticollis, CHCs from pooled samples (6 flies) were extracted together in a single vial containing 400 μl of hexane (same extraction protocol as above). Individuals used in these extractions were pooled by larval diet (Rich, Poor and standard) and sex. For comparison, we also extracted CHCs from pooled wild-caught individuals of unknown mating status/age by sex that were trapped at Fred Hollows Reserve in early January 2018.

174 Perceived dominance changes chemical signalling

Chemical analyses

Compound abundances

Chemical analysis of all single fly extracts was carried out on a 6890 Gas Chromatograph (GC) combined with an Agilent 5973 Mass Selective Detector (MSD) (Agilent technologies, US). An Agilent 7673 autosampler fitted with a 10 μL syringe injected a sample volume of 2 μL, with the split/splitless injector set to 290°C. Separations were carried out using a TRACE 260R-154P capillary column (60 m X 0.25mm ID, 0.25 μm film thickness, Thermo scientific, Australia). Helium was used as the carrier gas at a flow rate of 1.0ml/min with a splitless injection. The Gas Chromatography Mass Spectrometry (GC-MS) data were processed with Agilent Chemstation software. The temperature program began at 150°C increasing to 300°C at a rate of 30°C/min, holding at 300°C for 0 mins, then increasing to a final temperature of 330°C at a rate of 3°C/min. The run time for this method was based on Curtis et al. (2013) and was optimised to maximise efficiency and keep the column clean. From these data, individual profiles of the CHC compounds were determined by integration of the area under peaks (20 peaks for males; 24 peaks for females (not all identified in Table 1). CHC values were converted to relative proportions by dividing the area under each peak by the area under the peak for the hexacosane internal standard present in each sample. This enabled us to correct for technical errors associated with GC-MS and any intrinsic changes to column integrity that might occur from high-throughput analysis.

To process the data from our samples pooled by diet, sex and wild-caught individuals, we utilised Progenesis ® QI Informatics software (Paglia et al. 2014). Each GC-MS run was imported as an ion-intensity map including m/z and retention time. These ion maps were then aligned to the retention times. From the aligned runs, an aggregate run was constructed and compared with all runs so that the same ions were detected in each run. Isotope and adduct deconvolution was then used to reduce the number of features identified. All data were then normalised to total ion intensity and then extracted for multivariate analysis.

175 Perceived dominance changes chemical signalling

Compound identification

Unambiguous identification of the chemical compounds present in the CHC extracts of T. angusticollis was not possible using regular GC-MS methods and comparison with the available NIST 11/ Wiley 275 databases.

To increase the resolution and obtain more precise chemical identification, pooled extractions were analysed using a Thermo Trace 1310 GC with a Thermo QExactive-GC orbitrap high resolution MS (Thermo Scientific, Germany). A Thermo TriPlus RSH autosampler fitted with a 10 μL syringe injected a sample volume of 2 μL with a split/splitless injector at 270°C, using helium as a carrier gas (flow rate 1.0 mL/min) and a TG5silMS capillary column (30m X 0.25mm ID, 0.25 μm film thickness) for separation (Thermo Scientific, Australia). The temperature program consisted of three ramps starting at 50°C increasing at a rate of 30°C/min until 90°C and then increasing at 10°C/min until 180°C and then increasing at 7°C/min until a final temperature of 330°C that was held for 12 mins. Data was acquired initially in EI and PCI mode at a resolution of 60,000, and subsequently in PCI mode at 120,000. These analyses were conducted by the Central Analytical Research Facility at Queensland University of Technology on samples that were dried and reconstituted in GC grade hexane. The identities of CHCs were ascertained from the presence of molecular ions in their chromatographic peaks and identified using mass spectral fragmentation patterns from the PCI data. A 100 μg/mL sample of C7-C40 saturated alkane mixture (Sigma-Aldrich, Australia product #49452) was used for identification of branching, using the same methods above but with a split injection with a split ratio of 20. Most CHCs did not co-elute with the C7-C40 saturated alkane mixture demonstrating a relatively high abundance of branched to straight chained alkanes (Carlson et al. 1998; Katritzky et al. 2000).

Statistical analyses

All statistical analyses were carried out using R 3.4.3 (R Core Team 2017) using the Modern Applied Statistics with S package “MASS” (Venables & Ripley, 2002) and the Classification and Regression Training package “caret.” (Kuhn, 2009) CHC signalling involves complex

176 Perceived dominance changes chemical signalling

blends of compounds that probably function as a whole, so we did not analyse specific compounds or subsets of compounds (Bonduriansky et al., 2015; Everaerts et al., 2010). Instead, we utilised a multivariate approach to analyse differences in CHC profiles across our social treatment groups and between the sexes. First the retention time was used to differentiate each peak, with multiple samples overlaid to ensure peaks were consistent between samples. The standardised peak areas (calculated as the area of each peak divided by the internal standard) at each retention time were then analysed by multivariate analysis of variance (MANOVA) to test for effects of social treatment group (“dominant” or “subordinate”), sex, and their interaction.

Data transformation and pre-processing

All data were log10 transformed prior to analysis. Multicollinearity can yield solutions that are numerically unstable or overfitted, impacting the generalizability of results, particularly in linear methods (Næs and Mevik 2001). To avoid these problems, we constructed a correlation matrix among peaks and removed the columns that contributed a mean absolute correlation of >0.75 from the data. We then performed recursive feature elimination (Breiman 2001) to eliminate redundant features and improve our models’ predictive accuracies using the random forest selection function with cross-validation (repeated 10 times). This algorithm is known to identify strong predictors in smaller data sets and produce optimal subsets of features that yield high classification accuracy (Darst et al. 2018). Near- zero variance predictors can also cause the classifier to fail when training models (Kuhn and Johnson 2013) so all predictors that had near zero variance were also removed. Subsequently all data were centred, scaled and finally transformed using Principal Component analysis (PCA) to reduce dimensionality of the data.

Model training

We used Linear Discriminant analysis on PC scores (see above) to investigate which combinations of CHCs discriminate focal individuals (within sex) into their social treatment groups. Subsequently, CHCs that are shared between the sexes (based on diagnostic ion and retention time) were analysed in the same way to investigate how shared CHCs discriminate

177 Perceived dominance changes chemical signalling

between sexes as well as ‘dominant’ and ‘subordinate’ groups within-sex, resulting in four sex × treatment combinations. We interpreted factor loadings >0.25 to have a significant contribution to the axes of variation for each discriminant function (Weddle et al. 2013).

LDAs were limited to principal components that accounted for 99% of the total variation in the data. We used a 10-fold repeated cross-validation to assess each model’s accuracy. This method partitions the data into 10 subsets but maintains the proportionality of each treatment representation (Valetta et al., 2018). The models were trained on 9 of the subsets and the remaining subset was used to assess its accuracy. This process was repeated until all of the subsets had been utilized as train and test sets. Because this method can sometimes overestimate accuracy, we used 90% of the data for cross-validation (as described above), and the remaining 10% of each original dataset to test the accuracy of our final models. Furthermore, to determine whether the resulting LDA models performed better than random, we re-run the LDA analysis on 1000 randomly generated training sets, each consisting of the actual data with randomly assigned group labels (similar to the method used by Nehring, Evison, Santorelli, d’Ettorre, & Hughes (2011)). This genearated a null distribution of LDA model accuracy values for comparison with our actual LDA results.

LDA yields nclasses – 1 discriminant functions. Therefore, analyses of within-sex differences only yielded one discriminant function while analyses of both sex and treatment yielded three discriminant functions.

To determine whether the degree of male-female similarity in CHC profile was affected by perceived dominance status, we utilised the “bayesboot” package (Baath 2016) to calculate the posterior differences in bootstrapped LD1 mean scores between treatment groups, and to calculate confidence intervals for these posterior differences. We calculated the mean difference (δ) between LD1 scores of individuals from each dominance treatment and individuals of the opposite sex and bootstrapped the posterior difference (posterior draws = 10,000) separately for each sex as δ = (DI-OS) – (SI-OS), where DI is the bootstrapped dominant LD1 mean of the focal sex, OS is the opposite sex LD1 mean (treatments pooled), and SI is the bootstrapped subordinate LD1 mean of the focal sex.

178 Perceived dominance changes chemical signalling

We analysed the effects of sex, social environment and their interaction on mean distance to food source and the mean number of aggressive behaviours using a one-way ANOVA test. Sex-limited CHCs provide the most unambiguous signals of sex and could therefore be especially important in signalling of sex and status. We therefore also examined whether our social treatments affected the relative expression of sex-limited CHCs by MANOVA. To investigate whether the sexes and dominance treatments differed in total CHC expression, all peak areas were summed per individual and compared between dominance status and sex using two-way ANOVA.

For comparison, we also examined the effects of rich, standard and poor larval diets on CHC profile using Principal Component Analysis (PCA) of relative peak areas (with data pooled across sexes). PCA was also carried out on samples of wild-caught females and males. CHC profiles from individuals reared on rich and poor larval diets were used to gauge whether dominance treatment effects could plausibly be explained as a simple transfer of CHCs from competitors to focal individuals (see Discussion).

Results

Social dominance treatment was found to affect the mean distance of a focal male from the oviposition site (Figure 3A) where ‘dominant’ individuals were significantly closer than their

‘subordinate’ counterparts (ANOVA, F1, 84 = 62.26, P = <0.001). Females, however, did not show a significant difference in mean distance to the oviposition site (ANOVA, F1, 61 = 2.252, P = 0.139). Likewise, social dominance treatment affected the mean number of aggressive behaviours performed by males (ANOVA, F1, 84 = 20.11, P = <0.001) (Figure 3B), but did not affect aggression in females (ANOVA, F1, 61 = .251, P = 0.618).

179 Perceived dominance changes chemical signalling

Figure 3. Effect of social dominance treatment on position within competitive arenas and behaviour for focal males and females. a: mean distance of a focal individual measured as approximate body lengths from the petri dish containing a food/oviposition resource; b: mean number of aggressive behaviours performed by the focal, including ‘wing flicks’, ‘chasing away a competitor’ and higher ‘escalated’ combat interactions (headlock, foreleg strikes, chest impacts). Asterix indicate significant differences (P = <0.001) by one-way ANOVA models. In both plots, bars show the mean ± standard error of the mean (SEM).

GC-MS analysis of CHC extracts

We analysed cuticular hexane extracts for 84 males and 61 females and identified a total of 30 CHCs. Of these, we definitively identified the structural formula of a total of 17 CHCs (Table 1). Some peaks could not be identified due to their relatively weak signal or complex mixture of molecular ions but were consistently present in samples and were thus included in semi-quantitative analyses of CHC profiles below. Unidentified CHCs 1 and 2 were consistently present in female samples but detected in only some male samples, and these compounds were thus excluded from semi-quantitative analyses in males. Peaks 16 and 19 were not consistently present in either male or female samples and were therefore excluded from quantitative analyses in both sexes. Mass spectrometry revealed five female-limited and

180 Perceived dominance changes chemical signalling

three male-limited CHCs with a diverse range of branched alkanes ranging from 27 to 36 carbons in length (Table 1).

Table 1. Epicuticular compounds of Telostylinus angusticollis, identified by CHC-specific ionic signatures. Compounds and their branching were identified by their GC Orbitrap-MS spectral fragmentation patterns and compared to a saturated alkane standard. Our samples also included compounds that were consistently present in male and/or female samples, but for which we were unable to determine the molecular formula and branch locations because of low signal, co-elution, or chain length >40. The analyses below include all sex-specific compounds, but only those shared compounds that were consistently present in all replicates for both sexes (indicated by the superscript “S”). Shared compounds that could not be detected in some samples were excluded from the analyses.

Peak ID Identification Molecular Formula Diagnostic ion Sex

1 Unidentified CHC 1 - - ♂♀

2 Unidentified CHC 2 - - ♂♀

3S Unidentified CHC 3 - - ♂♀

4S Unidentified CHC 4 - - ♂♀

S 5 Heptacosane C27 H56 381 ♂♀

6 3-Methylheptacosane C28 H58 394 ♂

7 3-Methyloctacosane C29 H60 408 ♂

S 8 2-Methylheptacosane C29 H60 408 ♂♀

S 9 2-Methylnonacosane C30 H62 422 ♂♀

S 10 3-Methylnonacosane C30 H62 422 ♂♀

11S Unidentified CHC 5 - - ♂♀

12S Unidentified CHC 6 - - ♂♀

S 13 3-Methyltriacontane C31 H64 436 ♂♀

S 14 Hentriacontane C31 H64 436 ♂♀

181 Perceived dominance changes chemical signalling

# 15 2-Methyltriacontane C31 H64 436 ♂♀

S 16 2-Methylhentriacontane C32 H66 450 ♂♀

S 17 Dotriacontane C32 H66 450 ♂♀

# 18 3-Methyldotriacontane C33 H68 464 ♂♀

S 19 3-Methylhentriacontane C32 H66 450 ♂♀

S 20 2,3-Dimethylhentriacontane C33 H68 464 ♂♀

21 4-Methyltritriacontane C33 H68 478 ♀

22S Unidentified CHC 7 - - ♂♀

S 23 2-Methyltritriacontane C34 H70 478 ♂♀

24 2-Methylpentatriacontane C36 H74 506 ♂

25S Unidentified CHC 8 - - ♂♀

26 Unidentified CHC 9 - - ♀

27 Unidentified CHC 10 - - ♀

28 Unidentified CHC 11 - - ♀

29 Unidentified CHC 12 - - ♀

Of compounds present in both sexes, the most abundant was identified as 3- MethylHentriacontane. The most abundant sex-limited compounds were 3- Methylheptacosane and 4-Methyltritriacontane in males and females, respectively.

Hydrocarbon peaks clustered particularly from C27-C36 with an increasingly complex

number of peaks around C31-C33. Figure 4 represents a typical gas chromatographic GC profile of CHC extracts of male and female T. angusticollis individuals. Females tended to express a lower overall abundance of CHCs when compared to males, a difference that may simply reflect males’ larger mean body size (Figure S1).

182 Perceived dominance changes chemical signalling

Figure 4. Mirrored GC chromatographic profile of a pooled male (above) and female (below) T. angusticollis (not adjusted for body-size). The x-axis shows retention time (min) and the y-axis shows MS signal strength (Arbitrary Units) of a total ion chromatogram (TIC). Peaks one and two were excluded from quantitative analyses in males because of inconsistent presence between samples. Peaks 16 and 19 were also excluded from quantitative analyses because of inconsistency, but in both sexes. Un-numbered peaks in chromatograms represent analytical artefacts or non-CHC compounds based on their ionic signatures.

183 Perceived dominance changes chemical signalling

Comparison of mean CHC abundances

We tested effects of treatment and sex using only those CHCs that were consistently present in all male and/or female samples (Table 1). GC-MS analysis revealed a total of 17 CHC peaks that were present in all female and male samples (i.e., shared between sexes) (Figure 5). For these shared CHCs we found a significant effect of sex (MANOVA, Pillai’s trace = 0.91,

F1, 140 = 73.1, P < 0.0001), social treatment (MANOVA, Pillai’s trace = 0.27, F1, 140 = 2.72, P < 0.0001) and a significant interaction between sex and social treatment (MANOVA, Pillai’s trace = 0.24, F1, 140 = 2.3, P < 0.001) on relative peak area. Of the 17 shared CHCs, 13 were found to be significantly sexually dimorphic, 5 peaks were found to be affected by social treatment, and 6 peaks were found to be affected by an interaction of sex and social treatment (Table S3). CHCs shared between the sexes did not differ in mean expression level

(t = 0.89137, P = 0.38). However, ‘dominant’ females showed significantly higher mean expression levels of shared CHCs than ‘subordinate’ females (t = 2.2757, P = 0.027), whereas males showed no such difference (t = 0.2865, P = 0.78).

Figure 5. Relative peak areas for each shared CHC by sex and dominance treatment. Mean (± SEM) relative peak area (proportional to internal standard) for each shared CHC peak detected by GC-MS is shown.

184 Perceived dominance changes chemical signalling

Linear discriminant function analysis of CHC blend

The cuticular profiles differed between our treatment groups, and individuals could be classified by sex and social treatment using Linear Discriminant Analysis (LDA). LDA on CHC profiles within each sex (including shared and sex-limited CHCs) yielded one discriminant function that explained 99 % of CHC variation (nclasses – 1) separately for each sex (Figure 6). Model accuracies were high for both male and female LDAs, and although 95% confidence intervals were wide, model accuracies for both sexes substantially exceeded the null expectation.

Figure 6. Results of linear discriminant function analyses of cuticular hydrocarbon extracts of male and female T. angusticollis from dominant and subordinate dominance status treatments. (a) Female and (b) Male show CHC profiles, where blue represents ‘subordinate’ and red ‘dominant.’ Density refers to the kernel density estimate, or proportion of data located there.

185 Perceived dominance changes chemical signalling

We also carried out an LDA based on the shared CHCs, yielding three discriminant functions that explained 89.79 %, 5.47 % and 4.74 % of variation in CHCs between the four sex x treatment combinations. Although this LDA model had somewhat lower accuracy than the sex-specific LDAs, its accuracy substantially exceeded the null expectation (Table 2, Figure 7). As might be expected when comparing fine scale variation of CHCs within sex, there was considerable overlap in hydrocarbon profiles between ‘dominant’ and ‘subordinate’ individuals (Figure 6), and only some peaks contributed significantly to social treatment group separation (Table S1).

Table 2. The success of predicting dominance status within male, female and shared CHC data sets based on LDA analysis.

% correctly assigned Dataset Number of Overall % Permutation Dominant Subordinate PCs model accuracy test results % (95% CI) accuracy (95% CI) All CHCs by sex: Males 6 85.7 (0.4213, 58.8 (0.5860, 50 100 0.9964) 0.5901) Females 6 70.8 (0.4891, 50.3 (0.4992, 100 68.2 0.8738) 0.5074) Shared CHCs: Overall 9 69.23 (0.3857, 30.3 (0.3009, 0.9091) 0.3045) Males 50 100 Females 33 80

Based on individual scores for the main linear discriminant function (LD1) for shared CHCs (Figure 7), if subordinate males are more similar than dominant males to mean female CHC profiles, mean δ value will be positive (i.e., > 0). Conversely, if subordinate females are more similar than dominant females to mean male CHC profiles, mean δ value will be negative (i.e., < 0). Bayesian bootstrap analysis supported these predictions for both males (δ = 0.169, 95% CI = 0.162, 0.176) and females (δ = -0.608, 95% CI =-0.618, -0.597). Thus, although there is considerable overlap in CHC profiles between dominance treatment groups and

186 Perceived dominance changes chemical signalling

sexes, these results suggest that subordinate males and females tend to be more similar to each other in their CHC profile than dominant males and females are (Figure 7).

Figure 7. shows scores for individual focal males (triangles) and females (circles) on the first two discriminant functions of shared CHCs with ‘dominant’ and ‘subordinate’ individuals represented by red and blue colours, respectively. Black points represent the centroids of each group. The dashed arrow indicates the relative distance (δ) between the centroids of dominant males and females and the solid arrow indicates the relative distance between centroids of subordinate males and females. Ellipses represent 95% confidence intervals.

Sex-limited CHCs

Social treatment significantly affected the expression of sex-limited CHCs in males

(MANOVA, Pillai’s trace = 0.117, F1, 79 = 73.1, P = 0.022) but not in females (MANOVA,

Pillai’s trace = 0.105, F1, 61 = 61, P = 0.264) (Figure 8).

187 Perceived dominance changes chemical signalling

Figure 8. The effect of perceived dominance status on the expression of sex-limited CHCs. Bars show mean ± SEM of the mean relative peak area for sex-limited CHCs. Male and female sex-limited expression patterns are presented in (a) and (b), respectively.* Denotes significant differences between treatments (Table S2).

Larval diet effects on CHC profiles

For comparison, we also examined CHC profiles of males and females reared on rich, standard and poor larval diets, and individuals collected from the wild, by means of PCA on pooled samples (Figure 9). PC1 and PC2 collectively described 90.9% of variation in CHC profiles. Males and females clustered by larval diet and lab vs wild origin along PC1, while PC2 separated lab-reared from wild-caught flies.

188

Perceived dominance changes chemicalPerceived dominance changes signalling

Figure 9. PCA factor scores for the first two PCs of CHC profile for T. angusticollis males and females reared on different larval diets, and for wild- caught individuals of each sex. Each point represents six individuals pooled by sex, diet and origin (lab, field). Males are represented by triangles and females by circle.

189

Perceived dominance changes chemical signalling

Discussion

Our results suggest that CHC profiles are sensitive to an individual’s perceived dominance status within a group. Focal males and females placed with same-sex competitors of either larger or smaller body size achieved ‘dominant’ or ‘subordinate’ status within the group, and this treatment effect on social status was associated with differences in CHC profile. Our data thus suggest that an individual’s performance in interactions with same-sex rivals (which is determined in part by its body size relative to rivals’ body sizes) influences the formation of an individual’s own CHC profile, allowing an individual to signal his or her place in the dominance hierarchy. Our findings are consistent with work on chemical signalling in D. melanogaster by Kent et al., (2008) that showed individual CHC profiles to be modulated by their neighbours’ genotypes and previous social context. An insect’s ability to outcompete conspecifics can be determined by developmental resources such as larval diet (Amitin and Pitnick 2007), which can affect body size and secondary sexual trait expression, and therefore influence social dominance (Bonduriansky, 2007; Moczek, 2002; Nijhout & Emlen, 1998). By rearing all focal individuals on the same larval diet and then placing them in contrasting adult competitive environments, we were able to control for effects of genetic variation and variation in developmental environment, to ask whether CHC profile can change in response perceived dominance status. Our experiment is one of the first to attempt to directly manipulate this social context and our results suggest that males and females both adjust their chemical profiles, depending on whether they perceive themselves to be dominant or subordinate.

Our experimental treatments successfully manipulated focal individuals’ social status. After 48h in their social treatments, focal males in the ‘dominant’ treatment group were found to be closer on average to the oviposition site (petri dish with larval medium) than focal males in the ‘subordinate’ treatment group. Similar results (albeit weaker) were obtained for females. We therefore believe that the differences in CHC profile between social treatment groups within sexes are most plausibly attributed to dynamic changes in chemical signalling associated with self-perception of social dominance within the group. A different (but non- exclusive) explanation is that differences in CHC profiles resulted from differential access to

190 Perceived dominance changes chemical signalling

food in our social treatments: ‘subordinate’ individuals may have been excluded by competitors from the petri dish of larval medium and altered their CHC profiles as a result of nutrient limitation. While we cannot exclude this possibility, our observations suggest that all flies had some access to the petri dish (and all flies had ad libitum access to water), and we believe that focals are therefore unlikely to have suffered substantial nutrient limitation. Moreover, if food limitation played a role in treatment effects on CHC signalling, this effect would represent a mechanism mediating the effects of social dominance on CHC signalling. T. angusticollis individuals in natural populations aggregate at oviposition sites, and adults also feed at these sites (Adler and Bonduriansky 2013). If subordinate individuals suffer food limitation as a result of being excluded from these sites by dominant individuals, then the effects of nutrient limitation on individuals’ CHC profiles could result in reliable chemical signalling of social dominance status. Such nutritional stress-mediated effects on CHC signalling could be nonadaptive (e.g. reducing individuals’ sexual attractiveness or ability to intimidate same-sex rivals), but could still play important roles in social interactions via the information that such signals provide to other individuals. For example, nutrient-limited individuals could be rejected as potential mates, or subjected to increased attacks by rivals.

We found some evidence that subordinate male CHC profiles were more female-like. Sex- limited CHCs in subordinate males showed expression levels that were significantly lower than those of dominant males. Subordinate males also exhibited a shared CHC profile that more closely resembled the CHC profile of a female. One plausible interpretation of this result is that subordinate T. angusticollis males employ a form of chemical mimicry by adopting a female-like CHC profile. By chemically mimicking a female, a male that perceives itself to be subordinate might reduce its risk of being damaged or threatened by dominant rivals. Equally, it could be advantageous for a dominant male to signal his dominance to help intimidate rival males and/or to make themselves more attractive to females. Indeed, previous work showed that T. angusticollis males that previously won fights against rival males appeared to be preferred by females (Fricke et al. 2015). ‘Subordinate’ females were also observed to have CHC profiles more closely resembling male CHC profiles. This could affect female-female competition for prime oviposition sites, perhaps by deterring rival

191 Perceived dominance changes chemical signalling

females through the threat of male harassment. ‘Subordinate’ females may also chemically mimic males to avoid male harassment themselves, enabling them to spend more time feeding. A similar effect has been observed in the context of mating, where mated female D. melanogaster release a male-limited CHC that acts as an anti-aphrodisiac (Scott 1986). A plausible alternative explanation of our findings is that subordinate males and females both exhibit CHC profiles that primarily serve viability-related functions (such as desiccation resistance) while dominant males and females adopt CHC profiles that signal their sex and dominance status. A similar, viability-related role of CHCs in subordinate males and females might explain why their CHC profiles are more similar than those of dominant males and females.

Chemical signals can be transferred between mates. For example, male butterflies (Danaus glippus) directly transfer crystals of danaidone (a sex pheromone) onto the antennae of a female during courtship to promote mating (Eisner and Meinwald 1995). Similarly, D. melanogaster males have been shown to transfer cis-Vaccenyl Acetate (cVA) to females during copulation, perhaps to elicit aggressive (rather than sexual) reactions to the mated female from competitor males (Jallon 1984). To our knowledge, there is no evidence that CHCs are passively transferred between individuals during aggressive interactions. Nonetheless, it is possible that competitor individuals directly transferred CHCs to our focal individuals—an effect that could confound the effect of perceived dominance on CHC profiles. However, we believe that this is unlikely for several reasons. First, T. angusticollis males rarely engage in aggressive behaviours unless closely matched for body size (Adler and Bonduriansky 2013). For this reason, we would expect individuals that are mismatched in body size to have little direct physical contact. Secondly, direct transfer of CHCs from competitors is not consistent with effects of larval diet on CHC profiles (Figure 5, S1). If the observed effect of social environment was driven by transfer of CHCs from competitors, then we would expect that dominant individuals should resemble the CHC profiles of their poor larval diet competitors, and vice versa. Since males and females reared on poor larval diet clustered by CHC profile on PC1 (Figure 7), this might be expected to result in similar CHC profiles in dominant females and males. Instead, we observed greater

192 Perceived dominance changes chemical signalling

similarity between subordinate males and females than between dominant males and females – a pattern that cannot be readily explained by CHC transfer from competitors. Rather, our findings are more consistent with focal individuals actively changing CHC production in response to their social environment, a response that could be mediated by the visual and olfactory perception of competitors, similar to the feedback mechanisms utilised by D. melanogaster males to recognise conspecific competitors (Fernandez et al. 2010).

The clustering of pooled samples by dietary treatment aligns with the substantial body of evidence suggesting that CHCs are costly traits that can be influenced by dietary manipulations that affect condition (Blows 2002; Ferveur 2005; Delcourt and Rundle 2011; Bonduriansky et al. 2015; Ingleby 2015). Furthermore, in D. melanogaster, some dietary hydrocarbons have even been shown to be incorporated directly into the CHC profile (Blomquist 2010). To our knowledge, no studies have examined the degree of sexual dimorphism in CHC profile as a function of the larval dietary environment, and this remains an interesting area for future research.

The CHCs extracted from T. angusticollis ranged from 26-36 in chain length. Such long molecular chains compounds tend to be more stable than short-chain CHCs, which are often volatile and involved in defensive secretions in insects (Blum 1981). Because of this stability, at least some of these compounds are likely to be involved in tactile chemical communication that contributes to social interactions in this species. On the other hand, because of the dual function of CHCs (Chung and Carroll 2015), it is difficult to know which CHCs are more important for communication compared to desiccation avoidance. The functions of CHC signalling have not been investigated previously in neriid flies, and the fitness consequences of the observed changes in CHC profile in response to perceived dominance status remain to be determined.

In summary, current knowledge of communicative complexity in sub-social and non-social insects is limited, particularly in relation to chemical signaling (Nehring and Steiger 2018). Our findings suggest that T. angusticollis individuals of both sexes adjust their chemical displays in response to self-perception of dominance status within a group. To our

193 Perceived dominance changes chemical signalling

knowledge, this is the first evidence in a species of non-social insect that an individual’s perception of its own status within a group can affect its CHC signalling. We also find qualitatively similar responses in both sexes, with subordinate males and females both showing greater resemblance than dominant individuals to the CHC profile of the opposite sex. These changes in CHC profile could affect performance in inter- and intra-sexual interactions.

Authors’ contributions

ZW, AC and RB conceived the ideas and designed the methodology. ZW ran the experiments and conducted the chemical analyses alongside LA. ZW and RB led the writing of the manuscript. All authors contributed critically to the drafts and gave final approval for publication.

Acknowledgements

We thank Sonia Bustamante, Andrew Jenner, and Russell Pickford from the Bioanalytical Mass Spectrometry Facility at the Mark Wainwright Analytical Centre for all of their support in practical considerations and training for this project. We are also grateful to Howard D. Rundle and Joanne Yew for their advice and helpful discussions on sample preparation and analysis. A special thanks to Joe Brophy for his help with analysis and interpretation of hydrocarbon spectra. This research was funded through an Australian Research Council Future Fellowship awarded to RB. The authors declare no conflict of interest.

Data accessibility

Data will be available on the Dryad database.

194 Perceived dominance changes chemical signalling

References

Adler, M. I., and R. Bonduriansky. 2013. Paternal Effects on Offspring Fitness Reflect Father’s Social Environment. Evol. Biol. 40:288–292.

Amitin, E. G., and S. Pitnick. 2007. Influence of developmental environment on male- and female-mediated sperm precedence in Drosophila melanogaster. J. Evol. Biol. 20:381–391.

Baath, R. 2016. bayesboot: An Implementation of Rubin’s (1981) Bayesian Bootstrap.

Bath, E., N. Tatarnic, and R. Bonduriansky. 2012. Asymmetric reproductive isolation and interference in neriid flies: the roles of genital morphology and behaviour. Anim. Behav. 84:1331–1339.

Blaustein, A. R. 1981. Sexual Selection and Mammalian Olfaction Author. Am. Nat. 117:1006–1010.

Blomquist, G. J. 2010. Biosynthesis of cuticular hydrocarbons. Pp. 35–51 in A. G. Bagneres, ed. Insect Hydrocarbons. Cambidge University Press, Cambridge.

Blomquist, G. J., and A. G. Bagneres. 2010. Insect Hydrocarbons Biology, Biochemistry, and Chemical Ecology. Cambridge University Press.

Blows, M. W. 2002. Interaction between natural and sexual selection during the evolution of mate recognition. Proc. R. Soc. B Biol. Sci. 269:1113–1118.

Blum, M. S. 1981. Hydrocarbons. Pp. 41–86 in Chemical Defenses of Arthropods.

Bonduriansky, R. 2006. Convergent evolution of sexual shape dimorphism in Diptera. J. Morphol. 267:602–611.

Bonduriansky, R. 2007. The evolution of condition dependent sexual dimorphism. Am. Nat. 169:9–19.

Bonduriansky, R., and M. Head. 2007. Maternal and paternal condition effects on offspring phenotype in Telostylinus angusticollis (Diptera: Neriidae). J. Evol. Biol. 20:2379–2388.

195 Perceived dominance changes chemical signalling

Bonduriansky, R., M. A. Mallet, D. Arbuthnott, V. Pawlowsky-Glahn, J. J. Egozcue, and H. D. Rundle. 2015. Differential effects of genetic vs. environmental quality in Drosophila melanogaster suggest multiple forms of condition dependence. Ecol. Lett. 18:317–326.

Breiman, L. 2001. “Random Forests.” Mach. Learn. 45:5–32.

Carlson, D. A., U. R. Bernier, and B. D. Sutton. 1998. Elution patterns from capillary GC for methyl-branched alkanes. J. Chem. Ecol. 24:1845–1865.

Chenoweth, S. F., and M. W. Blows. 2008. QST meets the G matrix: The dimensionality of adaptive divergence in multiple correlated quantitative traits. Evolution. 62:1437–1449.

Chung, H., and S. B. Carroll. 2015. Wax, sex and the origin of species: Dual roles of insect cuticular hydrocarbons in adaptation and mating. BioEssays 37:822–830.

Cremer, S., M. F. Sledge, and J. Heinze. 2002. Male ants disguised by the queen ’ s bouquet. Nature 419:897.

Curtis, S., J. L. Sztepanacz, B. E. White, K. A. Dyer, H. D. Rundle, and P. Mayer. 2013. Epicuticular Compounds of Drosophila subquinaria and D. recens: Identification, Quantification, and Their Role in Female Mate Choice. J. Chem. Ecol. 39:579–590.

Darst, B. F., K. C. Malecki, and C. D. Engelman. 2018. Using recursive feature elimination in random forest to account for correlated variables in high dimensional data. BMC Genet. 19:1–6.

Dawkins, M. S. 1995. Unravelling Animal Behaviour. Longman, Harlow.

Delcourt, M., and H. D. Rundle. 2011. Condition Dependence of a Multicomponent Sexual Display Trait in Drosophila serrata . Am. Nat. 177:812–823.

Dominey, W. J. 1980. Female mimicry in male bluegill sunfish - A genetic polymorphism?

Duménil, C., D. Woud, F. Pinto, J. T. Alkema, I. Jansen, A. M. Van Der Geest, S. Roessingh, and J. C. Billeter. 2016. Pheromonal Cues Deposited by Mated Females Convey Social

196 Perceived dominance changes chemical signalling

Information about Egg-Laying Sites in Drosophila melanogaster. J. Chem. Ecol. 42:259–269. Journal of Chemical Ecology.

Eisner, T., and J. Meinwald. 1995. The chemistry of sexual selection. Proc. Natl. Acad. Sci. 92:50–55.

El-Sayed, A. M. 2009. The Pherobase: Database of Insect Pheromones and Semiochemicals.

Everaerts, C., J. P. Farine, M. Cobb, and J. F. Ferveur. 2010. Drosophila cuticular hydrocarbons revisited: Mating status alters cuticular profiles. PLoS One 5:1–12.

Fernandez, M. de la P., Y. B. Chan, J. Y. Yew, J. C. Billeter, K. Dreisewerd, J. D. Levine, and E. A. Kravitz. 2010. Pheromonal and behavioral cues trigger male-to-female aggression in Drosophila. PLoS Biol. 8.

Ferveur, J. F. 2005. Cuticular hydrocarbons: Their evolution and roles in Drosophila pheromonal communication. Behav. Genet. 35:279–295.

Gershman, S. N., and H. D. Rundle. 2016a. Crowd control: sex ratio affects sexually selected cuticular hydrocarbons in male Drosophila serrata. J. Evol. Biol. 30:583–590.

Gershman, S. N., and H. D. Rundle. 2016b. Level up: The expression of male sexually selected cuticular hydrocarbons is mediated by sexual experience. Anim. Behav. 112:169– 177. Elsevier Ltd.

Gibbs, A. G. 2007. Waterproof cockroaches: the early work of JA Ramsay. J. Exp. Biol. 210:921.

Grillet, M., L. Dartevelle, and J.-F. Ferveur. 2006. A Drosophila male pheromone affects female sexual receptivity. Proc. R. Soc. B Biol. Sci. 273:315–323.

197 Perceived dominance changes chemical signalling

Hooper, A. K., F. Spagopoulou, Z. Wylde, A. A. Maklakov, and R. Bonduriansky. 2017. Ontogenetic timing as a condition-dependent life history trait: High-condition males develop quickly, peak early and age fast. Evolution. 1–15.

Ingleby, F. C. 2015. Insect cuticular hydrocarbons as dynamic traits in sexual communication. Insects 6:732–742.

Izzo, A., M. Wells, Z. Huang, and E. Tibbetts. 2010. Cuticular hydrocarbons correlate with fertility, not dominance, in a paper wasp, Polistes dominulus. Behav. Ecol. Sociobiol. 64:857–864.

Jallon, J.-M. 1984. A few chemical words exchanged by Drosophila during courtship and mating. Behav. Genet. 14:441–478.

Johansson, B. G., and T. M. Jones. 2007. The role of chemical communication in mate choice. Biol. Rev. 82:265–289.

Katritzky, A. R., K. Chen, U. Maran, and D. A. Carlson. 2000. QSPR correlation and predictions of GC retention indexes for methyl- branched hydrocarbons produced by insects. Anal. Chem. 72:101–109.

Kawasaki, N., C. E. Brassil, R. C. Brooks, and R. Bonduriansky. 2008. Environmental effects on the expression of life span and aging: an extreme contrast between wild and captive cohorts of Telostylinus angusticollis (Diptera: Neriidae). Am. Nat. 172:346–357.

Kent, C., R. Azanchi, B. Smith, A. Formosa, and J. D. Levine. 2008. Social Context Influences Chemical Communication in D. melanogaster Males. Curr. Biol. 18:1384–1389.

Kuhn, M., and K. Johnson. 2013. Applied predictive modeling.

198 Perceived dominance changes chemical signalling

Lin, N., and C. D. Michener. 1972. Evolution of Sociality in Insects. Q. Rev. Biol. 47:131– 159.

Maynard Smith, J. 1982. Do animals convey information about their intentions? J. Theor. Biol. 97:1–5.

Maynard Smith, J., and D. G. C. Harper. 1995. Animal Signals: Models and Terminology. J. Theor. Biol. 177:305–311.

Moczek A P. 2002. Allometric Plasticity in a Polyphenic Beetle. Ecol. Entomol. 27:58–67.

Næs, T., and B. H. Mevik. 2001. Understanding the collinearity problem in regression and discriminant analysis. J. Chemom. 15:413–426.

Nehring, V., S. E. F. Evison, L. A. Santorelli, P. d’Ettorre, and W. O. H. Hughes. 2011. Kin- informative recognition cues in ants. Proc. R. Soc. B Biol. Sci. 278:1942–1948.

Nehring, V., and S. Steiger. 2018. Sociality and communicative complexity: insights from the other insect societies. Curr. Opin. Insect Sci. 28:19–25. Elsevier Inc.

Nijhout, H. F., and D. J. Emlen. 1998. Competition among body parts in the development and evolution of insect morphology. Proc. Natl. Acad. Sci. 95:3685–3689.

Norman, N. D., J. Finn, and T. Tregenza. 1999. Female impersonation as an alternative reproductive strategy in giant cuttlefish. Proc Biol Sci 266:1347–1349.

Paglia, G., J. P. Williams, L. Menikarachchi, J. W. Thompson, R. Tyldesley-Worster, S. Halldórsson, O. Rolfsson, A. Moseley, D. Grant, J. Langridge, B. O. Palsson, and G. Astarita. 2014. Ion Mobility Derived Collision Cross Sections to Support Metabolomics Applications. Anal. Chem. 86:3985–3993. American Chemical Society.

Peschke, K. 1987. Male Aggression, Female Mimicry and Female Choice in the Rove Beetle, Aleochara curtula (Coleoptera, Staphylinidae). Ethology 75:265–284.

199 Perceived dominance changes chemical signalling

Petfield, D., S. F. Chenoweth, H. D. Rundle, M. W. Blows, and J. C. Avise. 2005. Genetic variance in female condition predicts indirect genetic variance in male sexual display traits. Proc. Natl. Acad. Sci. U. S. A. 102:6045–6050.

R Core Team. 2017. R: A Language and Environment for Statistical Computing. Vienna, Austria.

Rillich, J., and P. A. Stevenson. 2011. Winning fights induces hyperaggression via the action of the biogenic amine octopamine in crickets. PLoS One 6.

Scott, D. 1986. Sexual mimicry regulates the attractiveness of mated Drosophila melanogaster females. Proc. Natl. Acad. Sci. U. S. A. 83:8429–8433.

Smith, A. A., B. Hölldober, and J. Liebig. 2009. Cuticular Hydrocarbons Reliably Identify Cheaters and Allow Enforcement of Altruism in a Social Insect. Curr. Biol. 19:78–81.

Steiger, S., and J. Stökl. 2014. The Role of Sexual Selection in the Evolution of Chemical Signals in Insects. Insects 5:423–438.

Thomas, M. L., B. Gray, and L. W. Simmons. 2011. Male crickets alter the relative expression of cuticular hydrocarbons when exposed to different acoustic environments. Anim. Behav. 82:49–53. Elsevier Ltd.

Waters, W. E. 1959. A Quantitative Measure of Aggregation in Insects. J. Econ. Entomol. 52:1180–1184.

200 Perceived dominance changes chemical signalling

Weddle, C. B., S. Steiger, C. G. Hamaker, G. D. Ower, C. Mitchell, S. K. Sakaluk, and J. Hunt. 2013. Cuticular hydrocarbons as a basis for chemosensory self-referencing in crickets: A potentially universal mechanism facilitating polyandry in insects. Ecol. Lett. 16:346–353.

White, A. J., and H. D. Rundle. 2014. Territory defense as a condition-dependent component of male reproductive success in Drosophila serrata. Evolution. 69:407–418.

Whiting, M. J., J. K. Webb, and J. S. Keogh. 2009. Flat lizard female mimics use sexual deception in visual but not chemical signals. Proc. R. Soc. B Biol. Sci. 276:1585–1591.

Wigglesworth, V. B. 1933. The physiology of the cuticle and the ecdysis in Rhodnius prolixus (Triatomadae, Hemiptera); with special reference to the function of the oenocytes and of the dermal glands. Q. J. Microsc. Sci. 76:269–318.

Wyatt, T. D. 2003. No Pheromones and animal behaviour: Communication by smell and taste. Cambidge University Press.

Yew, J. Y., R. B. Cody, and E. A. Kravitz. 2008. Cuticular hydrocarbon analysis of an awake behaving fly using direct analysis in real-time time-of-flight mass spectrometry. Proc. Natl. Acad. Sci. 105:7135–7140.

201 Perceived dominance changes chemical signalling

Supplementary information

Figure S1. The effects of larval diet quality on male (A) and female (B) thorax length. A subsample of 50 competitor individuals per larval diet treatment were used to compare to our focals. The violin plot outlines illustrate the kernel probability density (the width of the outlined area represents the proportion of the data located there). Within violin plots are boxplots with median and interquartile range to illustrate data distribution. There was a significant difference in body size between all larval diet treatments within sex (indicated with *) (males: Chi square = 126.72; females: Chi square = 106.62) and a significant difference in mean body size between the sexes (ANOVA, F1, 347 = 26.15, P = <0.0001).

202 Perceived dominance changes chemical signalling

Table S1. Summary of linear discriminant analysis (LDA) based on PCA transformation of CHC data.

Male Female Combined

Principal Component LD 1 LD 1 LD1 LD2 LD3

% variation 100 100 89.79 5.47 4.74

PC1 0.557 0.582 -1.069 0.062 0.086

PC2 -0.438 -0.358 0.451 0.290 0.128

PC3 -0.523 -0.139 -0.169 0.290 -0.890

PC4 -0.438 -0.769 -0.362 0.702 0.371

PC5 -0.113 0.175 -0.393 -0.278 -0.078

PC6 -0.118 0.231 0.315 0.129 0.031

PC7 - - -0.064 1.095 0.099

PC8 - - 0.013 0.012 0.062

PC9 - - -0.242 0.417 -0.530

The analysis yielded one function that discriminates the two treatment groups within each sex

(based on both shared and sex-limited CHCs), and three functions that discriminate the four sex x treatment combinations (based on shared CHCs only). Loadings >0.25 (i.e. ± 8 % overlapping variance) were interpreted as contributing significantly to the axis of variation

(bold). The number of Principal Components accounts for 99% of variation in all datasets.

Table S2. Results of ANOVA for CHCs within sex. We analysed each of the 20 and 24 peaks of males and females, respectively.

203 Perceived dominance changes chemical signalling

Male Female

Peak I.D P P

Overall difference 0.049* 0.017*

1 Unidentified CHC 1 - 0.002**

2 Unidentified CHC 2 - 0.001**

3s Unidentified CHC 3 0.800 0.303

4 s Unidentified CHC 4 0.984 0.336

5s Heptacosane 0.900 0.716

6 3-Methylheptacosane 0.008** -

7 3-Methyloctacosane -0.003** -

8s 2-Methylheptacosane 0.010* 0.792

9s 2-Methylnonacosane 0.001** 0.547

10s 3-Methylnonacosane 0.296 0.622

11s Unidentified CHC 5 0.652 0.073

12s Unidentified CHC 6 0.152 0.384

13s 3-Methyltriacontane 0.702 0.056

204 Perceived dominance changes chemical signalling

14s Hentriacontane 0.290 0.002**

16s 2-Methylhentriacontane 0.873 0.080

17s Dotriacontane 0.544 0.654

19s 3-Methylhentriacontane 0.708 0.578

20s 2,3-Dimethylhentriacontane 0.489 0.820

21 4-Methyltritriacontane - 0.074

22s Unidentified CHC 7 0.009** 0.002**

23s 2-Methyltritriacontane 0.990 0.006**

24 2-Methylpentatriacontane 0.746 -

25s Unidentified CHC 8 0.096 0.565

26 Unidentified CHC 9 - 0.039*

27 Unidentified CHC 10 - 0.668

28 Unidentified CHC 11 - 0.345

29 Unidentified CHC 12 - 0.274

Significant effects of perceived social status are shown in bold. Significance codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1.

205

Table S3. Results of MANOVA and ANOVA for 17 CHC peaks shared between the sexes.

Sex Social treatment Sex*Social treatment

Peak I.D P P P

Perceived dominance changes chemicalPerceived dominance changes signalling

Overall difference 2.2e-16*** 0.0008*** 0.0048**

(MANOVA)

(Pillai’s trace = 0.909, F1, 140 = 73.065) (Pillai’s trace = 0.271, F1, 140 = 2.716) (Pillai’s trace = 0.239, F1, 140 = 2.293)

3S SP1 8.487e-6*** 0.2698 0.2161

4S SP2 1.487e-8*** 0.5591 0.6778

5S SP3 7.573e-7*** 0.01741* 0.00930**

8S SP4 0.9834 0.01594* 0.04740*

206

9S SP5 4.661e-11*** 0.0021** 0.0283*

10S SP6 1.383e-11*** 0.3425 0.9911

11S SP7 2e-16*** 0.9451 0.3525

Perceived dominance changes Perceived dominance changes 12S SP8 2e-16*** 0.0992. 0.4308

13S SP9 2e-16*** 0.0634. 0.1079

14S SP10 5.191e-5*** 0.0008*** 0.0114*

16S SP11 2e-16*** 0.0951. 0.0951. chemical signalling

17S SP12 2e-16*** 0.4950 0.8467

19 SP13 0.2951 0.9153 0.5005

207

20S SP14 4.944e-15*** 0.4717 0.6109

22S SP15 0.7113 0.1998 0.0024**

23S SP16 5.09e-5*** 0.0409* 0.0276*

25S SP17 0.0983 0.0526. 0.1107

Significant effects of sex, social treatment and their interaction are shown in bold. Significance codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0

208

CHAPTER 6

General conclusion

209 General conclusion

In studying reproduction and life-history, I wanted to explore and reveal some of the relatively unknown and often subtle variation that can arise through complex environmental interactions. In the context of life-history theory, I mainly investigated the effects of age- specific parental effects on senescence in offspring, juvenile development, and growth patterns in reproductive traits. Consequently, I also thought it was important to examine female traits where possible. Throughout this thesis, I was often interested in how these life- history traits interacted with the ecological and social environment to affect reproductive allocation strategies. Often, sexual traits are bizarre and striking when compared to other traits, so researchers commonly study those characters in isolation from the rest of the organism. However, when it comes to reproduction, the evolution of sexual traits is often specific to the social and ecological environment of breeding. This thesis has focused on this intersection between environment and reproduction by viewing reproductive investment from an entire organism approach, where individuals are subject to life-history trade-offs as well as the effects of the social and ecological environment in which breeding occurs.

In chapter two, I examined the transgenerational effects of parental breeding age on offspring lifespan and mortality rate. Historically, the negative effect of parental breeding age on offspring longevity has mostly been attributed to maternal sources (also known as the ‘Lansing effect’) (Lansing 1947), although see Radwan (2003) for an analysis of male age and germline mutations (also see Radwan et al 2003; Mallet et al 2011). The results of this experiment however, revealed that contrary to popular belief, paternal effects of breeding age can have similarly marked and negative impacts on offspring mortality and lifespan. The overall effect in both patrilines and matrilines was mediated through both F1 and F2 breeding age, but we also detected an interaction between these two factors suggesting a mechanism that works differently to mutation accumulation. Understanding the proximate mechanism(s) that drive this pattern would be exceptionally informative and shed light on age-dependent transmission of non-genetic factors to offspring. While only a small proportion of flies in F2 survived to 45 and 60 days old, it is unlikely that the observed effects of F3 longevity reflect genotype sampling effects because this would require that descendants of long-lived flies to have short lifespans. We also found that the juvenile nutritional and

210 General conclusion

adult competitive environments experienced by grandparents had little impact on regulating negative breeding age effects in offspring. Additionally, we found some evidence of effects on actuarial ageing rate in patrilines but not matrilines. The male Y-chromosome in Drosophila is known to influence the regulation of hundreds of X and autosomal genes throughout the genome (Lemos et al 2010; Jian et al 2010) and is thought to be associated with variation in the Y-chromosome’s non-coding heterochromatin (Lemos et al 2010; Paredes et al 2011; Francisco & Lemos 2014). A recent study (Kutch & Fedorka 2015) was able to show that the Y-chromosome in Drosophila can influence baseline immune gene expression and bacterial defence. One possible explanation is that in the neriid flies this Y-chromosomal effect is similar, where male breeding age influences ageing rate of descendants through some kind of epigenetic modification of this Y-linked modifier region. This could explain why we see no age-dependent effects in matrilines, but evidence of breeding age effects on actuarial ageing within patrilines and would be an interesting endeavour to explore in the future. These differences suggest that male and female breeding age effects could be mediated by different factors and could have slightly different effects on offspring life history. Overall, these findings suggest that the effect of an ancestors’ age at breeding could contribute substantially to within-population variance in longevity.

In terms of the role of parental age effects in the evolution of ageing, I suspect much insight will come from understanding the proximate mechanism(s) that mediate these effects, particularly non-coding RNAs that are present in male ejaculates and female eggs and that could potentially vary with age. Historically, analyses of inter-generational effects of age have primarily focused on maternal effects, but from our study, it seems likely that breeding age affects offspring relatively similarly in both sexes, at least in terms of effects on ageing and lifespan. Likewise, a recent study using mice showed that the tissue of offspring from aged fathers shared epigenetic signatures of longevity-related genes (Xie et al. 2018). Currently, little is known as to what systems are mediating these changes in lifespan and mortality. If we also look from a broader perspective, it remains unclear what principal mechanisms drive the evolution of senescence, or how extrinsic mortality rates and other demographic factors might influence this process. The enormous variation in lifespan and senescence in nature

211 General conclusion

also makes me question the carefully controlled lab experiments that have been used to test evolutionary theories of ageing. High background mortality in the wild could mean that parental effects of old age are weaker because most offspring arise from grandparents and parents that are young. In the future, it would be interesting to determine whether breeding age affects offspring performance similarly in the wild. This could be done by mating old and young wild flies to laboratory reared stock in selection lines. However, I do believe that parental age effects may play a currently underappreciated role in driving the diversity of senescence rates across the tree of life (see Jones et al 2014 for a survey on the diversity of ageing rates in multiple taxa).

In chapter three, I examined the effects of juvenile larval nutrition and how this interacts with sperm competition to influence male ejaculate allocation strategy and female storage patterns. Theoretical and empirical studies have only just begun to reveal that ejaculate and sperm traits can be similarly condition-dependent and plastic when compared to secondary sexual traits in their response to the developmental environment, particularly within the insects (Wigby et al. 2016). How this variation in male condition influences male ejaculate allocation strategies is poorly understood. We found that high-condition males mate more quickly, and when mated second, transfer more ejaculate to both of the female’s posterior spermathecae. This suggests that males are prudent to avoid sperm depletion when sperm competiton is low, but only high-condition males can afford to increase their allocation in the face of high male-male competition. Low-condition males could also be expected to evolve prudent allocation strategies, but our data suggest that low-condition males invest heavily in all matings, regardless of sperm competition. However, it is difficult to know if this might be driven more by the lack of resources, or primarily by the fact that smaller, low- condition males are less able to attain mates in a natural setting (because of lower mating or resource holding potential). Our findings from this study however, show for the first time, that ejaculate allocation strategies can incorporate variation in both condition and perceived sperm competition risk.

Rhodamine-B fluorescent dye has been shown to be incorporated into developing eggs upon injection into the female bursa (Hayashi and Kamimura 2002). Rhodamine-B dye forms

212 General conclusion

stable covalent bonds with proteins and has also been shown to be incorporated from male spermatophores into oocytes (van der Reijden et al. 1997). During the study on ejaculate allocation, we further developed this tracking method by feeding rhodamine-B and rhodamine-110 to competing males to track and quantify their ejaculates within the female reproductive tract. While we found that high-condition males can invest relatively more ejaculate when mated second, we did not extend this study to examine any fitness effects. These context-dependent effects of male condition on ejaculate allocation are not well known, particularly when extended to offspring. To better understand these subtle effects of sperm competition, I would also like to investigate how female storage patterns and male condition influence offspring number and viability by further optimizing this labelling method. This technique is promising and could also be a tool to examine whether competing male ejaculates are differentially incorporated into eggs. The techniques required to examine these effects are often expensive, time consuming and/or limited to particular taxa. The rhodamine dyes could therefore be utilised as a relatively cheap and widely applicable alternative to the green and red fluorescent probes developed for Drosophila (Manier et al. 2010). Furthermore, there is the possibility that we could further calibrate this technique to see if greater sperm numbers are expected to result in higher fluorescence values.

This study has also generated many other follow-up questions I would like to address. Is it possible that when mated second, high-condition males fertilise less eggs per female but increase their fertilisation rate by decreasing female re-mating rate? A study in ladybird beetles found that the ingestion of spermatophores decreases the propensity of a female to re- mate and accelerates egg production (Perry and Rowe 2008), and a more recent study found that high-condition males transfer larger ejaculates but less sperm when compared to low- condition males (Perry and Rowe 2010). When a high-condition neriid male is second to mate and increases ejaculate allocation, perhaps females are also slower to re-mate because of an increased dose of seminal proteins that influence female behaviour and increases egg production? This could increase the probability that high-condition males obtain successful fertilisations even when their low-condition competitors (that cannot afford to increase ejaculate production) transfer higher numbers of sperm. I also wonder whether high-

213 General conclusion

condition and low-condition ejaculates have different lifespans within the female sperm storage organs themselves. Is it possible that males of high condition produce less sperm, but that their sperm have longer lifespan (maybe because of substances transferred in the ejaculate) within the female reproductive tract? Alternatively, low-condition males could produce low quality sperm but larger numbers to displace a high-condition male’s sperm within the female storage organs? In many insects, sperm competition within the female reproductive tract and organs conforms to a simple lottery system (Parker 1990), where more sperm equates to more offspring. In some flies, second male sperm precedence also occurs (Hosken & Ward 2000; Amitin & Pitnick 2007). However, the sperm competition system of neriid flies has yet to be characterized and would enable us to better link sperm storage in females with male reproductive success that selection acts upon. In this study, we also found that copulation duration had no effect on the amount of ejaculate stored in the most distant (to the bursa) spermeatheca suggesting that another ejaculate trait such as sperm motility or a female-mediated trait could be playing a role. In fact, sperm tail beat frequency has been shown in T. angusticollis to decrease in males reared on a nutrient-poor larval diet (Macartney et al. 2018). The study of condition-dependent variation in the responses of sperm and semen is relatively new and complex. It is therefore likely that other biological factors (i.e., epigenetic factors such as non-coding RNAs and methylation, metabolic rate) and environmental factors (population density and sex ratio, predation, temperature etc) also are at play.

It was only natural that my next chapter compared the morphological integration of genitalia and somatic traits. Although insect genitalia have been shown to evolve via sexual selection (Simmons et al. 2009) and twice as fast as somatic traits (Arnqvist 1998), explaining the reasons for the unusually rapid evolution of these traits has been the subject of much debate (Dufour 1844; Eberhard 1985; Arnqvist and Danielsson 1999; Hosken and Stockley 2004; Simmons and Fitzpatrick 2019). In this study I wanted to ask whether the unusual evolvability of male genital traits is associated with low morphological integration of genitalia and whether the degree of integration can be affected by the developmental environment. We found that male genitalic traits were much less integrated than female and somatic traits.

214 General conclusion

Interestingly, we also found that the integration of male genital traits and somatic traits in both sexes was condition-dependent, whereby high-condition males exhibited less trait integration than low-condition males and females. Jacob (1977) argued that evolution is not an optimization process, but often proceeds through disordered re-assembly of whatever components are available. We therefore suggest that this relatively low integration in genitalia has enhanced the evolvability of these traits, allowing them to diversify relatively independently and rapidly over time.

This study on genitalic and somatic trait integration also raises further questions regarding developmental and functional constraints on trait variability and size. I am especially interested in examining the downstream effects of hormones on genitalic integration. Larval diet quality is known to affect hormonal pathways in insects, particularly juvenile growth hormone (JH) which is known to respond to nutritional state (Lavine et al. 2015). We found that high-condition individuals (particularly males) had less integrated genitalia than their low-condition counterparts. If integration does indeed act as a constraint on evolution, then I might predict that the expression of genitalic traits in high-condition individuals would evolve faster under experimental sexual selection lines when compared to the expression of those traits in low-condition individuals that have more highly integrated genitalia. Moreover, genitalic trait integration in insects has yet to be investigated phylogenetically. Perhaps groups that have relatively invariant genitalia across species also have more integration in their genitalic traits when compared to groups that have highly variable genitalia? By focusing attention on singular traits, we can miss variation. This makes it difficult to examine why particular traits remain relatively unchanged or, conversely, become highly modified over evolutionary timescales as seen in insect genitalia. To gain a deeper understanding into why structures like Drosophila wing characters have not changed for ~ 50 million years (Hansen and Houle 2004), or why genitalic traits in insects have so rapidly diverged, I think the area of morphological integration could be highly insightful.

My final chapter was a slight departure from studying reproductive traits and allocation strategies. However, chemical signalling in insects is known to influence attractiveness (Laturney and Billeter 2016), mating status (Everaerts et al. 2010; Weddle et al. 2013) and

215 General conclusion

dominance (Smith et al. 2009), which are all tightly linked with reproduction and fitness. In neriid flies, body size is likely a good indicator of dominance status as males have been observed to size match before a battle commences, and even small differences in body size have been shown to determine dominance status (Bonduriansky & Head 2007). If one male is substantially smaller then the interaction does not escalate to full combat. I instead, wanted to directly manipulate male and female self-perception (not body size) of their dominance status by changing group structure to ask if their chemical profiles (cuticular hydrocarbons, or CHCs) change as a result. I found that ‘dominant’ individuals tended to have CHC profiles less similar to those of the opposite sex. Moreover, we also found that ‘dominant’ males exhibited an overall elevation of male specific CHCs, and that ‘subordinate’ males exhibited a slight decrease in CHC expression. The findings from this study are the first to suggest, in a non-social species of insect, that individuals can change their CHC profiles dynamically in response to self-perceived dominance status. It is well known in a number of organisms that reproductive bias often results from dominance interactions (reviewed in Berglund et al. 1996). Whether or not the CHC changes we observed in the neriid flies could result in an increase in mating probability or other fitness-related interactions would be of utmost interest for future studies. A study in crickets experimentally perfumed males and females with CHCs, revealing that individuals bearing similar chemical profiles were less preferred as mating partners (Weddle et al. 2013). It would be interesting to mimic this technique in the neriid flies and examine how the transfer of ‘dominant’ and ‘subordinate’ CHCs influences intra-sexual aggression, mating success, the number of offspring produced and viability. To extend this study in this way, and also rule out physical transfer of CHCs, a non-invasive technique (such as Solid Phase Micro Extraction) would be useful to measure focal individuals before and after interactions.

Overall, these investigations demonstrate how studying the connections between the environment and life-history can yield important insights into the factors that drive the evolution of reproductive allocation strategies. All of these responses are tightly linked to strategies that are known to directly and indirectly influence reproductive processes in nature. In exploring the many facets of secondary and primary sexual traits, particularly

216 General conclusion

within the framework of condition-dependence, this thesis shows how environmental effects on reproductive allocation strategies are often context-dependent and more complex than hitherto appreciated. This thesis illustrates that individual plasticity in allocation to reproduction in both males and females is context- and state-dependent. To better understand the mechanisms underlying the operation of sexual selection and the evolution of sexual traits and behaviours, this thesis highlights the importance of widening our approach to the study of reproductive phenotypes that incorporates morphology, behaviour, and demographic factors in both sexes.

217 General conclusion

References

Amitin, E. G., and Pitnick. S. 2007. 2007. Influence of developmental environment on male- and female-mediated sperm precedence in Drosophila melanogaster. Journal of Evolutionary Biology. 20: 381-391.

Arnqvist, G. 1998. Comparative evidence for the evolution of genitalia by sexual selection. Nature 393:784–786.

Arnqvist, G., and I. Danielsson. 1999. Copulatory behavior, genital morphology and male fertilization success in water striders. Evolution. 53:147–156.

Berglund, A., A. Bisazza, and A. Pilastro. 1996. Armaments and ornaments: An evolutionary explanation of traits of dual utility. Biol. J. Linn. Soc. 58:385–399.

Bonduriansky, R., and Head, M. 2007. Maternal and paternal condition effects on offspring phenotype in Telostylinus angusticollis (Diptera: Neriidae). J. Evol. Biol 20: 2379-2388.

Dufour, L. 1844. Anatomie generale des dipteres. Annu. Sci. Nat. I:244–264.

Eberhard, W. G. 1985. Sexual Selection and Animal Genitalia. Harvard University Press.

Everaerts, C., J. P. Farine, M. Cobb, and J. F. Ferveur. 2010. Drosophila cuticular hydrocarbons revisited: Mating status alters cuticular profiles. PLoS One 5:1–12.

Francisco, FO., and Lemos, B. 2014. How do y-chromosomes modulate genome-wide epigenetic States: genome folding, chromatin sinks, and gene expression. J. Genomics 2, 94– 103. doi:10.7150/jgen.8043

Hansen, T. F., and D. Houle. 2004. Evolvability, Stabilizing Selection, and the Problem of Stasis. P. in M. Pigliucci and K. Preston, eds. Phenotypic Integration: Studying the Ecology and Evolution of Complex Phenotypes. Oxford University Press.

Hayashi, F., and Y. Kamimura. 2002. The potential for incorporation of male derived proteins into developing eggs in the leafhopper Bothrogonia ferruginea. J. Insect Physiol. 48:153–159.

218 General conclusion

Hosken, D. J., and P. Stockley. 2004. Sexual selection and genital evolution. Trends Ecol. Evol. 19:87–93.

Hosken, D. J., and Ward, P. I. 2000. Copula in yellow dung flies (Scathophaga stercoraria): investigating sperm competition models by histological observation. J. Insect. Physiol. 1: 1355-1363.

Jacob, F. 1977. Evolution and Tinkering. Science. 196:1161–1166.

Jiang PP., Hartl, DL., and Lemos, B. 2010. Y not a dead end: epistatic interactions between Y-linked regulatory polymorphisms and genetic background affect global gene expression in Drosophila melanogaster. Genetics 186, U109–U221. doi:10.1534/genetics.110.118109

Jones, O.R., Scheuerlein, A., Salguero-Gómez, R., Camarda, C.G., Schaible, R., Casper, B.B., Dahlgren, J.P., Ehrlén, J., García, M.B., Menges, E.S. and Quintana-Ascencio, P.F., 2014. Diversity of ageing across the tree of life. Nature, 505(7482), p.169.Kutch, IC., and Fedorka, KM. 2015. Y-linked variation for autosomal immune gene regulation has the potential to shape sexually dimorphic immunity. Proc. R. Soc. B Biol. 282(1820). https://doi.org/10.1098/rspb.2015.1301

Lansing, A. I. 1947. A transmissible, cumulative and reversible factor in aging. Gerontology 2:228–239.

Lavine, L., H. Gotoh, C. S. Brent, I. Dworkin, and D. J. Emlen. 2015. Exaggerated Trait Growth in Insects. Annu. Rev. Entomol. 60:453–472.

Lemos, B., Branco, AT., and Hartl, DL. 2010. Epigenetic effects of polymorphic Y chromosomes modulate chromatin components, immune response, and sexual conflict. Proc. Natl Acad. Sci. USA 107, 15 826–15 831. doi:10.1073/pnas.1010383107

219 General conclusion

Mallet, M. A., Bouchard, J. M., Kimber, C. M., and Chippendale, A. K. 2011. Experimental mutation-accumulation on the X chromosome of Drosophila melanogaster reveals stronger selection on males than females. BMC Evolutionary Biology. 11(156). doi:10.1186/1471- 2148-11-156

Manier, M. K., J. M. Belote, K. S. Berben, D. Novikov, W. T. Stuart, and S. Pitnick. 2010. Resolving Mechanisms of Competitive Fertilization Success in Drosophila melanogaster. Science. 328.

Paredes, S., Branco, AT., Hartl, DL., Maggert, KA., and Lemos, B. 2011. Ribosomal DNA deletions modulate genome-wide gene expression: ‘rDNA-sensitive’ genes and natural variation. PLoS Genet. 7, e1001376. doi:10.1371/journal.pgen.1001376

Parker, G. A. 1990. Sperm competition games: raffles and roles. Proc. R. Soc. Lond. B. 242: 120-126.

Perry, J. C., and L. Rowe. 2010. Condition-dependent ejaculate size and composition in a ladybird beetle. Proc. R. Soc. B Biol. Sci. 277:3639–3647.

Perry, J. C., and L. Rowe. 2008. Ingested spermatophores accelerate reproduction and increase mating resistance but are not a source of sexual conflict. Anim. Behav. 76:993–1000.

Radwan, J. 2003. Male age, germline mutations and the benefits of polyandry. Ecol. Letters 6(7): 581-586.

Radwan, J., Unrug, J., Snigorska, K., and Gawronska, K. 2003. Effectiveness of sexual selection in preventing fitness deterioration in bulb mite populations under relaxed natural selection. J. Evol. Biol. 17: 94-99. doi: 10.1046/j.1420-9101.2003.00646.x

Simmons, L. W., C. M. House, J. Hunt, and F. García-González. 2009. Evolutionary Response to Sexual Selection in Male Genital Morphology. Curr. Biol. 19:1442–1446.

220 General conclusion

Simmoms, L. W., and Fitzpatrick, J. L. 2019. Female genitalia can evolve more rapidly and divergently than male genitalia. Nat commun 10, 1312. doi:10.1038/s41467-019-09353-0

Smith, A. A., B. Hölldober, and J. Liebig. 2009. Cuticular Hydrocarbons Reliably Identify Cheaters and Allow Enforcement of Altruism in a Social Insect. Curr. Biol. 19:78–81. van der Reijden, E. D., J. D. Monchamp, and S. M. Lewis. 1997. The formation, transfer, and fate of spermatophores in Photinus fireflies (Coleoptera: Lampyridae). Can. J. Zool. 75:1202–1207.

Wigby, S., J. C. Perry, Y. H. Kim, and L. K. Sirot. 2016. Developmental environment mediates male seminal protein investment in Drosophila melanogaster. Funct. Ecol. 30:410– 419.

Xie, K., D. P. Ryan, B. L. Pearson, K. S. Henzel, F. Neff, R. O. Vidal, M. Hennion, I. Lehmann, M. Schleif, S. Schröder, T. Adler, B. Rathkolb, J. Rozman, A.-L. Schütz, C. Prehn, M. E. Mickael, M. Weiergräber, J. Adamski, D. H. Busch, G. Ehninger, A. Matynia, W. S. Jackson, E. Wolf, H. Fuchs, V. Gailus-Durner, S. Bonn, M. Hrabě de Angelis, and D. Ehninger. 2018. Epigenetic alterations in longevity regulators, reduced life span, and exacerbated aging-related pathology in old father offspring mice. Proc. Natl. Acad. Sci. 115:E2348–E2357.

221