The Role of Developmental Temperature on Phenotypic Development and Evolution

Fonti Kar

A thesis submitted in fulfilment of the requirements for the degree Doctor of Philosophy

Faculty of Science School of Biological, Earth and Environmental Sciences Evolution and Ecology Research Centre

December 2020

1 THE UNIVERSITY OF NEW SOUTH WALES Thesis/Dissertation Sheet Surname or Family name: Kar First name: Fonti Other name/s: Shuk Ming Abbreviation for degree as given in the University calendar: PhD School: School of Biological, Earth, and Environmental Sciences; Evolution and Ecology Research Centre Title: The Role of Developmental Temperature on Phenotypic Development and Evolution: Metabolism to Life History Faculty: Science Abstract Animals live in an ever-changing world, but environmental perturbations are occurring at an alarming rate - threatening biodiversity and population persistence. Developmental plasticity may be an effective solution for animals to cope with environmental variation. However, it is unclear how developmental environments affect consistent phenotypic variability and shape individual responses to environmental variation later in life. Understanding these impacts of developmental environments will be important for populations living in fluctuating environments. I employed experimental and comparative approaches to investigate the impacts of incubation temperatures on phenotypic development in an Australian lizard (Lampropholis delicata). Using ‘pace-of-life’ theory as a framework, I investigated how variation in metabolic rate can result in concordant changes in life history. I used a variety of statistical tools to quantify consistent phenotypic variation of energy metabolism and growth. While development temperatures did not affect metabolic rate and its thermal plasticity, lizards reared in hot temperatures exhibited less consistent individual differences in their metabolic rate. This may be problematic in the context for global warming. However, individuals also consistently varied in their acute thermal plasticity and these consistent individual differences were robust to changes in developmental temperatures. This suggests that populations may harbour the ability to evolve suitable responses to a warming climate. Despite there being no developmental changes in metabolism, we found differences in hatching mass that persisted through to the onset of sexual maturity. Growth, and its heritability, were not affected by developmental environments. Instead, maternal effects may play an important role governing variation in growth. While metabolic rate has been purported to be a causal mechanism for variation in life history strategies, I did not find strong support for this hypothesis. Across 500 species of terrestrial ectotherms, I demonstrated that environmental factors that dictate how animals acquire and allocate resources to reproduction are major drivers to life history variation. The environment is comprised of many facets that interact to give rise to the myriad of variation we observe in nature. My thesis highlights the need to shift away from unifying theories and focus on untangling the complexities of the environment in which animals inhabit. that interact to give rise to the myriad of variation we observe in nature. My thesis highlights the need to shift away from unifying theories and focus on untangling the complexities of the environment in which animals inhabit. 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: 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 ...... COPYRIGHT STATEMENT I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or 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 proprietary 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 Abstract International (this is applicable to doctoral theses only). I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.

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. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.

Signed

......

Date ......

INCLUSION OF PUBLICATIONS STATEMENT

Acknowledgements

First and foremost, I want to express my gratitude and respect to my supervisors Daniel Noble and Shinichi Nakagawa. I’ve worked with Dan since my Masters in 2014 and it has been wonderful to watch our supervisory relationship and friendship evolve over these years. In the beginning, I was a mere naïve student that leaned on Dan a lot but through many years of collaborations, successes and failures, I hope he now considers me as more of a colleague. Dan has always provided me unwavering support. His door was always open for times when I was stuck in my head and needed someone to chat and flesh out my ideas. There were instances throughout my PhD when I was slipping, Dan made sure to check in regularly so that I knew he was there if I needed him. Dan encouraged me to take on tasks that I was perhaps a bit scared to do but these experiences have only trained me to become a more independent researcher. I distinctly remember on several occasions when I was unsure what to do, I said to myself: “What would Dan do?” and proceeded with the task ahead with more confidence. Dan is an enthusiastic teacher, he has a genuine passion for imparting knowledge with anyone who is keen to learn. It never ceases to amaze me the things he’s taught me. From designing experiments and data analysis, to writing papers and programming tricks; to presentation and job interviewing skills. For all of that and more, I am extremely grateful.

Over the years of working with Shinichi, I’ve come to find his love for science very contagious. I feel very lucky to have been taught by someone whose knowledge on evolutionary biology, research synthesis and statistical methods is so extensive and vast. Shinichi is very accommodating to my working style and always make the effort to check in despite his busy schedule. I am especially appreciative of the times when Shinichi goes out of his way to help me solve a coding problem or explain the difference between credible and confidence intervals. I’ve always found it strange when people tell me Shinichi is a machine because to me, that is just only one side of him. Shinichi is very encouraging and kind and tries to make everyone feel like they are part of the family. Meeting his Mum for lunch and spending Christmas with him and his

1 family were some of the highlights of my time in his lab. Thank you Shinichi and I hope to continue working with you in the future!

I was very lucky to be a part of a very dynamic lab at UNSW, we often had many visiting academics which made our little offices feel like a big city. I want to thank some of the constancies of the I-DEEL Lab: Losia, Rose, Joelsie, Yong and Joasia for their company, support and love for science and treats at lab meetings. I am particularly grateful for Rose’s company during this PhD, it wouldn’t have been easy without our many chats over tea and olive oil chocolate cake. I am in debt to Rose and her parents, Susan and Danny for letting me stay in their family home in Canberra during my final year of my PhD. Your generosity really helped me settle in and made me feel like a part of the family.

Thank you to my PhD advisory panel: Lisa Schwanz, Angela Moles and Russell Bounduriansky for their support and advice through all the ups and downs of my candidature. I really appreciated their confidence in me when I was not feeling at all confident in myself. I want to also extend my thanks to Will Cornwell, Mike Kasumovic and Daniel Falster for sharing their advice and personal PhD experiences with me. Hearing about the trials and tribulations of senior academics helped me see through my own struggles.

To Martin Whiting and many members of the Lizard Lab – thank you for accommodating all of my animals in your wonderful facilities at Macquarie University. None of the experiments would have been possible without your generosity. Special thanks for Christine Wilson who took care of our lizards and Joshua Cunningham, Victor Frichot and Matthieu Monserand for their enduring commitment to my project.

I want to thank the fellow PhD students at the EERC: Amy, Erin, Dax, Francesca, Justin, Stewie and Nathan who shared this journey with me. The demands of a PhD can be overwhelming at the best of times but having a group of people that share those same experiences and are keen to hang out and talk about our worries over a beer or a cocktail definitely makes those burdens feel

2 lighter. A round of ‘crushed-balls’ (a concoction of crushed apple cider and fireball whiskey) is much overdue.

Extra special shout out to Amy Hooper and Chris Setio for their friendship and support. Our Tuesday dinners were something I always looked forward to and helped take my mind off work. I really appreciated that these catch-ups continued virtually when I first relocated to Canberra as I was trying to find my feet and when COVID-19 had hit and we were all self-isolating.

To my best friends in South Africa, Julia Riley and James Baxter Gilbert and Dundee. I miss you guys lots and I am so glad that no matter the time difference, we still call each other up to practice our talks and bounce ideas off each other. Julia you are forever my science sister!

To my soul sister in Perth, Selina Tang. No time too late or too early for chats about anything. I am super grateful to have someone like you who knows my heart better than I know myself. You always rooted for me even when I was feeling a lot of self-doubt. Thank you, I wouldn’t have done it without your encouragement.

I am incredibly lucky to have the continual friendship and mentorship from Kate Umber and Chrissie Painting. Both your friendship means a lot to me. I have so much respect for you guys. The world needs more kick-ass female scientists like you! It’s wonderful that we are still in touch since my volunteering days during undergrad.

Moving to a new city was not easy, but I was very fortunate to have had Ashley and Tim immediately invite me to play board games with Lauren, Kevin Luisa and Nick that made the transition go more smoothly. I am also very grateful to have joined such a friendly cohort of RSB students at the ANU. Many thanks to Piet and Monica, Eve, Claire, Alex, Je, Leo, Ollie, Mel, Zac, Tobias, Frances and Lachy and many more for making me feel so welcomed and showing me how to picnic – Canberran style.

3 To my dear parents, Mum, Dad. Thank you for supporting me, though I know you still don’t quite understand why I dedicated 6 years of my life to study lizards, you still encouraged and supported me to do so. I am grateful for my brothers, Keith, Kenneth and Vincent who regularly checked in and kept me connected with my nieces and nephews. These moments make our family feel so much closer despite us being all over the world.

Many thanks to all the many support staff at UNSW (Jono) and at ANU (Jack, Wes and Audra) for making paperwork and working-from-home go smoothly. Big, big shout out to Duncan Smith and Martin Thompson who manages Katana at UNSW. Your patience in training me to use bash is a skill I will always have. Also, all IT staff that have been so helpful and accommodating with lending out equipment when my laptop was stolen, it kept me sane and my work going with little interruption.

Finally thank you all the little lizards for keeping me very busy for the past four years and for giving me an exciting glimpse of your world.

4 Abstract

Animals live in an ever-changing world, but environmental perturbations are occurring at an alarming rate - threatening biodiversity and population persistence. Developmental plasticity may be an effective solution for animals to cope with environmental variation. However, it is unclear how developmental environments affect consistent phenotypic variability and shape individual responses to environmental variation later in life. Understanding these impacts of developmental environments will be important for populations living in fluctuating environments.

I employed experimental and comparative approaches to investigate the impacts of incubation temperatures on phenotypic development in an Australian lizard (Lampropholis delicata). Using ‘pace-of-life’ theory as a framework, I investigated how variation in metabolic rate can result in concordant changes in life history. I used a variety of statistical tools to quantify consistent phenotypic variation of energy metabolism and growth. While development temperatures did not affect metabolic rate and its thermal plasticity, lizards reared in hot temperatures exhibited less consistent individual differences in their metabolic rate. This may be problematic in the context for global warming. However, individuals also consistently varied in their acute thermal plasticity and these consistent individual differences were robust to changes in developmental temperatures. This suggests that populations may harbour the ability to evolve suitable responses to a warming climate. Despite there being no developmental changes in metabolism, we found differences in hatching mass that persisted through to the onset of sexual maturity. Growth, and its heritability, were not affected by developmental environments. Instead, maternal effects may play an important role governing variation in growth.

While metabolic rate has been purported to be a causal mechanism for variation in life history strategies, I did not find strong support for this hypothesis. Across 500 species of terrestrial ectotherms, I demonstrated that environmental factors that dictate how animals acquire and allocate resources to reproduction are major drivers to life history variation. The environment is comprised of

5 many facets that interact to give rise to the myriad of variation we observe in nature. My thesis highlights the need to shift away from unifying theories and focus on untangling the complexities of the environment in which animals inhabit.

6 Table of Contents

Acknowledgements ...... 1 Abstract ...... 5 Table of Contents ...... 7 List of Figures ...... 10 List of Tables ...... 11 List of Supplementary Materials ...... 12

CHAPTER 1 ...... 14 General Introduction

Study system ...... 18 Statistical Arsenal ...... 19 Thesis outline ...... 21 References ...... 25

CHAPTER 2 ...... 32 Individual variation in thermal plasticity and its impact on mass-scaling

Abstract ...... 33 Introduction ...... 34 Materials and Methods ...... 37 Statistical analysis ...... 39 Results ...... 42 Discussion ...... 47 Conclusion ...... 50 Acknowledgements ...... 50 Data accessibility ...... 51 References ...... 52 Supplementary Materials ...... 58 Supplementary Materials References ...... 69

CHAPTER 3 ...... 70 Impact of developmental temperatures on the repeatability of thermal plasticity in metabolic rate 7 Abstract ...... 71 Introduction ...... 72 Materials and Methods ...... 75 Statistical Analyses ...... 78 Results ...... 82 Discussion ...... 86 Conclusion ...... 89 Data accessibility ...... 90 Acknowledgements ...... 90 References ...... 91 Supplementary Materials ...... 99

CHAPTER 4 ...... 112 Heritability and developmental plasticity of growth in an oviparous lizard

Abstract ...... 113 Introduction ...... 114 Materials and Methods ...... 118 Statistical Analyses ...... 121 Results ...... 125 Discussion ...... 131 Conclusion ...... 136 Data accessibility ...... 137 Acknowledgements ...... 137 References ...... 138 Supplementary Materials ...... 148 Supplementary Materials References ...... 160

CHAPTER 5 ...... 161 What predicts pace-of-life? Distinguishing among multiple hypotheses in terrestrial ectotherms

Abstract ...... 162 Introduction ...... 163 Methods ...... 168 Statistical analysis ...... 170

8 Results ...... 172 Discussion ...... 177 Conclusions ...... 181 Data accessibility ...... 181 Acknowledgements ...... 181 References ...... 183 Supplementary Materials ...... 192 Supplementary Materials References ...... 203

CHAPTER 6 ...... 204 Conclusions and Directions

References ...... 211

Appendix ...... 215

Presentations ...... 216 Other Research Articles ...... 217

9 List of Figures

CHAPTER 1 Figure 1 ...... 17 Figure 2 ...... 19 Figure 3 ...... 21

CHAPTER 2 Figure 1...... 43 Figure 2...... 44 Figure 3...... 45 Figure 4...... 46

CHAPTER 3 Figure 1 ...... 82 Figure 2 ...... 83 Figure 3 ...... 85

CHAPTER 4 Figure 1 ...... 125 Figure 2 ...... 126 Figure 3 ...... 127 Figure 4 ...... 129

CHAPTER 5 Figure 1 ...... 164 Figure 2 ...... 172 Figure 3 ...... 173 Figure 4 ...... 175

10 List of Tables

CHAPTER 3 Table 1 ...... 83

CHAPTER 4 Table 1 ...... 128 Table 2 ...... 129

11 List of Supplementary Materials

CHAPTER 2 Figure S1 ...... 60 Table S1 ...... 61 Figure S2 ...... 62 Table S2 ...... 62 Table S3 ...... 63 Figure S3 ...... 64 Figure S4 ...... 66 Table S4 ...... 66 Table S5 ...... 67 Figure S5 ...... 68

CHAPTER 3 Table S1 ...... 99 Table S2 ...... 99 Table S3 ...... 100 Table S4 ...... 101 Figure S1 ...... 102 Table S5 ...... 103 Figure S2 ...... 104 Table S6 ...... 104 Table S7 ...... 105 Table S8 ...... 106 Table S9 ...... 107 Table S10 ...... 108 Table S11 ...... 109 Table S12 ...... 110 Table S13 ...... 111

CHAPTER 4 Table S1 ...... 150 Table S2 ...... 150 Table S3 ...... 151

12 Table S4...... 152 Table S5 ...... 153 Figure S1 ...... 154 Table S6...... 155 Table S7 ...... 156 Table S8 ...... 157 Table S9 ...... 158 Table S10 ...... 159

CHAPTER 5 Figure S1 ...... 194 Figure S2 ...... 195 Figure S3 ...... 196 Table S1 ...... 196 Table S2 ...... 197 Table S3 ...... 197 Table S4 ...... 198 Table S5...... 199 Table S6 ...... 200 Table S7 ...... 201 Table S8 ...... 202

13 CHAPTER 1

General Introduction

14

General Introduction

Life is uncertain, especially for a developing embryo. You have limited control of your surroundings and usually have to bear the brunt of whatever the world throws at you. Over millions of years of evolutionary tweaking, embryos have evolved the ability to adjust their phenotype in response to prevailing conditions they experience during development (West-Eberhard, 2005). Like the juvenile dwarf spider (Erigone atra), if it experiences cool conditions, individuals would spin ballooning silks to disperse in search of better habitats whereas if conditions are favourable, individuals would stay put (Bonte et al., 2008). Take the larva of the taurus beetle (Onthophagus taurus), its adult body size and horn development - one of nature’s most exaggerated forms of weaponry, is strongly dictated by the quality of the brood ball it finds itself in (Moczek, 1998). Another example is starling chicks (Sturnus vulgaris), individuals brooded by stressed mothers, that go on to be more competent fliers as adults (Chin et al., 2009). Developmental plasticity is widespread across the animal kingdom and understanding how it has evolved has captivated the interests of many biologists.

Developmental plasticity is expected to evolve under conditions where environmental cues experienced by the embryo can foretell the eventual world in which the embryo will need to grow up (Bateson et al., 2014; Nettle & Bateson, 2015). An embryo can express a suite of traits that might provide it an adaptive edge later in life (Beldade et al., 2011) . For example, water fleas demonstrate adaptive developmental plasticity in response to predatory cues. Individuals develop armoured headwear in response to fish pheromones in the pond (Boersma et al., 1998), which provide a survival advantage by reducing the probability of fish predation. Unfortunately, anticipating the future is not always a fool proof strategy. Our world is constantly changing – particularly in the last century with the advent of anthropogenic climate change. Cues that used to provide reliable intel are now often misleading, resulting in environment-phenotype mismatches (Beaman et al., 2016; Gluckman et al., 2019). More than ever, it is of utmost importance for us to understand, and predict, the evolutionary consequences of experiencing different developmental environments (Dyke & Griffith, 2018) . What phenotypic traits are changed by

15

developmental cues? What is the direction and magnitude of change? Are these changes persistent or flexible to further fine-tuning? These are some of the broad questions my thesis will address.

Insight into how developmental environments influence metabolic rate is one way we can begin to probe the underlying mechanisms by which suites of morphological, life-history and fitness related traits are altered (Harman, 1956; Sæther, 1987). Life requires energy. Animals must forage to obtain energy to grow, reproduce and survive (Van Noordwijk & De Jong, 1986). At rest, metabolic rate represents the ‘idling cost of living’ which dictates how often animals need to replenish their energy reserves (Careau, Killen, et al., 2014; Ricklefs & Wikelski, 2002). For decades, researchers have observed that species with higher metabolic rates tend to have shorter lifespans, leading many to believe that metabolic rate is the ‘pacemaker’ of life (‘rate-of-living’ hypothesis, Sæther, 1987). Harman (1956) proposed that the accumulation of reactive oxygen species from the metabolic breakdown of food substrate inflicts oxidative damage to cells which accelerates aging (“free-radical theory of aging”). The functional link between oxidative stress, metabolic rate and senescence has guided the formulation of similar theories about the consequences of intraspecific variability in metabolic rate (Biro & Stamps, 2008, 2010; Careau et al., 2008) . The pace-of-life theory posits that covariation among individuals in behaviour and life history comes down to physiology (Réale et al., 2010). Individuals with slow metabolism tend to exhibit a distinct slow phenotype or ‘syndrome’ (slow to mature, timid behaviours, high investment into fewer but higher quality progeny) compared to those with a fast metabolism. By understanding interindividual variation in metabolic rate, this bottom-up framework may allow us to foresee how developmental environments influence phenotypic development.

Metabolic rate is a highly labile trait, and its plasticity means that energy expenditure of individuals is subjected to fluctuations in the environment. Plasticity in metabolic rate is in part determined by the genetic makeup of an individual as well as the developmental environment it experiences (Nussey et al., 2007). For example, metabolic rate is more plastic for mosquito fish born under cool spring conditions compared to those born in the summer (Seebacher

16

et al., 2014). Plasticity in metabolic rate can be viewed as reaction norm, whereby metabolic rate is represented as a function of an environmental variable, such as ambient temperature (‘reaction-norm approach’, Via et al., 1995). Pace-of-life theory assumes that individual variation in metabolism is consistent across contexts such that individuals with high metabolic rates will always have a fast pace of life under changing conditions. This scenario can be depicted as individual reaction norms with different elevations, but the reaction norms are parallel (Fig. 1). Physiologists have long recognised that individuals vary in their response to the environment, meaning that their reaction norms can cross over (Fig 1). But is individual plasticity consistent over time (Norin & Metcalfe, 2019; Individual x Environment, Nussey et al., 2007) ? If so, how does this impact upon life-history traits? Addressing these questions will be important for discerning how environments shape phenotypes and the evolutionary consequences this has for populations.

Figure 1. Graphical depiction of individual differences in reaction norms. On the left illustrates a scenario where individuals differ in their average trait expression but have the same level of plasticity across an environment. On the right illustrates that individuals differ in both their average trait value and plasticity across the environment which is represented by different reaction norm slopes.

Phenotypic plasticity is an effective immediate solution to changes in the environment, however genetic adaptation may be necessary to ensure the long- term survival (Sgrò & Hoffmann, 2004) . Adaptive evolution requires two ingredients: a pinch of consistent phenotypic variability among individuals that has a heritable basis (Falconer & Mackay, 2009; Wilson, 2018) , and a dash of 17

selection pressure, which shifts populations to a new adaptive fitness peak (Falconer & Mackay, 2009; Lynch & Walsh, 1998). While studies have demonstrated that metabolic rate is both repeatable and heritable, we do not yet know if this also applies to the lability of metabolic rate (Nespolo & Franco, 2007; Nilsson et al., 2009; Rønning et al., 2007). It is also unclear how developmental environments changes its repeatability (Careau, Buttemer, et al., 2014; Nussey et al., 2007) . Some work has shown that stressful developmental environments can alter underlying genetic variation which can either facilitate or dampen the potential to evolve (Hoffman & Parsons, 1991; Hoffmann & Merilä, 1999) . Whether genetic variation of metabolic rate, and the environmental factors that affect it, have cascading effects on life history is not well understood. Life history traits have been hypothesised to have low evolvabilities because they represent a composite of physiological, morphological and behavioural traits which are all are sensitive to developmental instability (Houle, 1992). However, recent work has shown that this is not always the case (Charmantier & Garant, 2005; Hoffmann & Merilä, 1999; Rowiński & Rogell, 2017) . Elucidating the complexities of developmental environments on both genetic and phenotypic variation in metabolism and life history will bring important insights as to how developmental plasticity might affect populations responses in changing environments.

Study system

I used a widespread Australian skink, the delicate skink (Lampropholis delicata) to explore the impacts of developmental environment on individual plasticity in metabolic rate and life history. The delicate skink is a small, oviparous skink with an expansive distribution along eastern Australia (Chapple, Hoskin, et al., 2011) (Fig. 2). This species is heliothermic and is found in a diverse range of moist habitats including forests, farmland and urban parks (Matthews et al., 2016; Peace, 2004) . The delicate skink is masterful in human-assisted dispersal and is the only Australian species to have successfully invaded overseas (Chapple et al., 2013, 2014; Chapple, Simmonds, et al., 2011) . The delicate skink reaches sexual maturation at approximately one year and has a lifespan between two to four years (Greer, 1990) . This skink species displays subtle sexual dimorphism, males tend to have broader and longer heads while females have larger body sizes (Chapple et al., 2014). The reproductive season of L.

18

delicata occurs between September to February with clutch sizes ranging between one to eight eggs (Chapple et al., 2014; Forsman & Shine, 1995) . Based on my own observations, females typically lay eggs between October and February. In the wild, females often produce large communal nests (Cheetham et al., 2011) . The delicate skink’s small size makes them very amenable to laboratory and field experiments. I used wild caught animals from three Sydney locations (UNSW Kensington Campus: -33.92, 151.24; Sydney Park: - 33.91, 151.18, Macquarie Park: -33.77, 151.10). These three sample locations represent the same genetic lineage of Lampropholis delicata, it is therefore unlikely that our experiments would be impacted by collection at different sampling sites (Chapple, Hoskin, et al., 2011).

Figure 2 (Left) Photograph of a female Lampropholis delicata with a large clutch of eggs. Photo credit: Dylan van Winkel (Right) Processed occurrence records of Lampropholis delicata from Tingley et al., 2016

Statistical Arsenal

Nature is organised in a nested fashion. Species are comprised of populations that are made up of individuals, governed by the interaction between their genes, development and environments they experience. Quantifying individual variation in reaction norms requires statistical tools that distinguish among different sources of variation. Mixed models allow one to separate out factors that contribute to variability in a response variable, such as metabolism, while testing the relative importance of multiple predictors (Dingemanse & 19

Dochtermann, 2013). Some factors are biologically interesting, such as shared parentage (Chapter 4, Wilson et al., 2010) or shared ancestry (Chapter 5, De Villemereuil & Nakagawa, 2014) and may allow elucidation of processes that shape variation. For example, the field of quantitative genetics relies on careful partitioning of phenotypic variance into separate components in order to calculate heritability – the evolutional potential of a given trait (Lynch & Walsh, 1998). Behavioural ecologists interested in consistent individual differences in behaviour (animal personality) require the help of mixed models to isolate variation that is truly unique to each individual (Dingemanse & Dochtermann, 2013). However, other factors, such as methodological error or temporal patterns, can add noise to the data if not accounted for (Chapter 2–3, Ponzi et al., 2018). Mixed models also provide a powerful way to measure consistent individual variation in traits and reaction norms while controlling for pseudo replication (statistical non-independence) from repeated sampling of the same subjects

Biological data is seldom perfect and is riddled with missingness. Unpredictable mishaps occur during data collection, natural deaths take place during long term studies, or some species are just very understudied. Typically, rows containing missing data are removed prior to analysis which can reduce statistical power (Nakagawa & Freckleton, 2008). Importantly, missing data may influence the conclusions we draw if our complete dataset is an unrepresentative sample of the population (Nakagawa, 2015). Data imputation techniques can rectify some of these problems by recovering missing information, however there is a slow uptake of missing data theory in the field of ecology and evolution. Missing data theory dates back to the 1980’s, and currently, there are a number of R packages that have made data imputation tools more readily accessible (Bürkner, 2017; Buuren & Groothuis-Oudshoorn, 2010; Goolsby et al., 2016). Data imputation techniques prevent users from having to resort to reduced sample sizes (Chapter 2), maximising the data that is available to them and enabling them to perform multivariate hypothesis testing (Chapter 5). This is particularly important in comparative research as researchers are limited to published databases that have, at times, poor species overlap (Chapter 5) (Pennell et al., 2016). Furthermore, data imputation can be used to our advantage if it is planned ahead of data collection (Chapter 2, Noble

20

& Nakagawa, 2018) . For example, one can randomly miss a proportion of measurements on a given individual or a subset of individuals and then recover the data using missing data techniques. Planning missing data designs can therefore alleviate handling stress on animals during data collection but also permits more efficient experimental designs (Noble & Nakagawa, 2018).

Thesis outline

My thesis explores key themes about developmental plasticity and its consequences on consistent individual variation and the evolution of plasticity in metabolism and life history (Fig. 3). More specifically, my thesis tackles four main questions: 1) How does developmental temperature influence the phenotypic development of metabolism and life history? 2) What are the consequences of developmental temperature on plasticity of metabolic rate later in life? 3) Do individuals display consistent variation in metabolic plasticity and is this influenced by developmental temperature? 4) Is metabolism the mechanistic driver of life history strategies?

Developmental Plasticity Impact on metabolism & growth Impact on plasticity, (Chapter 2 - 3) repeatability & genetic variance Evolution of (Chapter 3 - 4) Reaction Norms My thesis Repeatability of Pace-of-life reaction norms Theory (Chapter 2 - 3) Repeatability of What predicts metabolism in life history strategies? Genetic & non-genetic different environments (Chapter 5) sources of variation (Chapter 2 - 3) (Chapter 4)

Figure 3 Venn diagram illustrating the core themes of my thesis and how each chapter fits with these themes.

21

I use pace-of-life theory to make predictions on the impact of developmental temperature on integrated phenotypic traits (e.g., metabolism, growth, age at maturity). Throughout my thesis, I combine various statistical tools to test and challenge some of the core assumptions of pace-of-life theory (Chapter 2, 3, 5). This has led to the conception of new ideas for why metabolism covaries with life history at both the individual (Chapter 6 – General Discussion) and interspecific level (Chapter 5). I have come to appreciate that the influence of the environment is pervasive and works in complex ways to bring out common patterns that we see across species. I started this PhD journey thinking that the big questions in evolutionary biology have already been answered by those before me, but I now realise that those questions are complicated and, in most cases, remain unanswered. Mother nature is messy and exciting, and it seems unlikely that there is a single explanation that applies to the entirety of the animal kingdom. However, it is through questioning theoretical assumptions and considering the mechanisms that these assumptions rely on, where we find clues for the next research venture. I hope my thesis has achieved that. Below I introduce each chapter in the context of the core questions outlined above.

Chapter 2

Metabolic rate scales with body mass following a power relationship. The exponent of this relationship is surprisingly similar across broad taxonomic groups and implies that there are constraints in how metabolic rate increases with body size. However, mass-scaling changes with temperature and suggests that metabolic costs of individuals of varying sizes depend on the environment. Individuals vary in a multitude of ways which affect their metabolic rate and how they might respond to temperature. Such individual processes can influence mass scaling, as well as its temperature dependence, provided that they are consistent over time. In Chapter 2, I used an extensive study design to establish whether thermal plasticity of metabolic rate is repeatable or not and how its temperature dependence might contribute to variation in mass-scaling relationships. More broadly, this chapter establishes some foundational knowledge about the consistency in metabolism, which is important for understanding variation in pace-of-life.

22

Chapter 3

Developmental plasticity and reversible plasticity have always been considered as separate processes. However, theory predicts that developmental environments may shape plasticity later in life. In Chapter 3, I manipulate the developmental temperature of lizards to see if their metabolic reaction norms are changed. I also investigate whether repeatability of reaction norms is altered. Changes in consistent among individual differences has key implications for understanding the evolution of reaction norms under changing environments.

Chapter 4

Growth is a fundamental process that is underpinned by metabolic rate and can drive variation in life history strategies. While it is commonly observed that organisms that are reared in warmer habitats tend to be smaller compared to those in cooler habitats (temperature-size rule, Kingsolver & Huey, 2008), the developmental consequences on growth trajectories is not well established. Furthermore, the rate at which evolutionary change in growth occurs depends on its genetic variation, which can also change depending on the developmental environment. In Chapter 4, I compared the growth trajectories (another type of reaction norm) of lizards reared at different developmental temperatures. I also examined the impact of developmental temperature on genetic and non-genetic sources of variation in body mass to determine whether the evolutionary potential of growth is impacted by early life conditions.

Chapter 5

Acquisition-allocation theory is at the heart of life history evolution. Life history strategies represent a balancing act of growth, reproduction and survival, and are highly variable among species. Many hypotheses that have been proposed to explain variation in life history. Some strongly believe that metabolism underpins life history variation because it ‘sets’ the cost of living. Others believe trade-offs between current and future reproduction result in concordant changes in life history. The environment can also play a significant role in dictating how organisms acquire resources which can result in cascading changes in life history. In my final chapter, I attempt to disentangle these 23

competing arguments about the factors explaining life history variation using a phylogenetically informed analysis across squamates to understand the drivers of age at maturity and lifespan.

24

References

Bateson, P., Gluckman, P., & Hanson, M. (2014). The biology of developmental plasticity and the predictive adaptive response hypothesis. The Journal of Physiology, 592(11), 2357–2368. https://doi.org/10.1113/jphysiol.2014.271460 Beaman, J. E., White, C. R., & Seebacher, F. (2016). Evolution of plasticity: Mechanistic link between development and reversible acclimation. Trends Ecology and Evolution, 31(3), 237–249. https://doi.org/10.1016/j.tree.2016.01.004 Beldade, P., Mateus, A. R. A., & Keller, R. A. (2011). Evolution and molecular mechanisms of adaptive developmental plasticity. Molecular Ecology, 20(7), 1347–1363. https://doi.org/10.1111/j.1365-294X.2011.05016.x Biro, P. A., & Stamps, J. A. (2008). Are animal personality traits linked to life- history productivity? Trends Ecology and Evolution, 23(7), 361–368. https://doi.org/10.1016/j.tree.2008.04.003 Biro, P. A., & Stamps, J. A. (2010). Do consistent individual differences in metabolic rate promote consistent individual differences in behavior? Trends Ecology and Evolution, 25(11), 653–659. https://doi.org/10.1016/j.tree.2010.08.003 Boersma, M., Spaak, P., & De Meester, L. (1998). Predator-Mediated Plasticity in Morphology, Life History, and Behavior of Daphnia: The Uncoupling of Responses. The American Naturalist, 152(2), 237–248. https://doi.org/10.1086/286164 Bonte, D., Travis, J. M. J., Clercq, N. D., Zwertvaegher, I., & Lens, L. (2008). Thermal conditions during juvenile development affect adult dispersal in a spider. Proceedings of the National Academy of Sciences, 105(44), 17000– 17005. https://doi.org/10.1073/pnas.0806830105 Bürkner, P. C. (2017). brms: An R package for Bayesian multilevel models using Stan. Journal of Statistical Software, 80(1), 1–28. https://doi.org/10.18637/jss.v080.i01 Buuren, S. van, & Groothuis-Oudshoorn, K. (2010). MICE: Multivariate Imputation by Chained Equations in R. Journal of Statistical Software, 1– 68.

25

Careau, V., Buttemer, W. A., & Buchanan, K. L. (2014). Early-developmental stress, repeatability, and canalization in a suite of physiological and behavioral traits in female zebra finches. Integrative and Comparative Biology, 54(4), 539–554. https://doi.org/10.1093/icb/icu095 Careau, V., Killen, S. S., & Metcalfe, N. B. (2014). Adding Fuel To The “Fire Of Life”: Energy Budgets Across Levels Of Variation In Ectotherms And Endotherms. In Integrative Organismal Biology (1st ed., p. 17). Wiley- Blackwell. Careau, V., Thomas, D., Humphries, M. M., & Réale, D. (2008). Energy metabolism and animal personality. Oikos, 117(5), 641–653. https://doi.org/10.1111/j.0030-1299.2008.16513.x Chapple, D. G., Hoskin, C. J., Chapple, S. N., & Thompson, M. B. (2011). Phylogeographic divergence in the widespread delicate skink (Lampropholis delicata) corresponds to dry habitat barriers in eastern Australia. BMC Evolutionary Biology, 11(1), 191. https://doi.org/10.1186/1471-2148-11-191 Chapple, D. G., Miller, K. A., Chaplin, K., Barnett, L., Thompson, M. B., & Bray, R. D. (2014). Biology of the invasive delicate skink (Lampropholis delicata) on Lord Howe Island. Australian Journal of Zoology, 62(6), 498–506. https://doi.org/10.1071/ZO14098 Chapple, D. G., Simmonds, S. M., & Wong, B. B. M. (2011). Know when to run, know when to hide: Can behavioral differences explain the divergent invasion success of two sympatric lizards? Ecology and Evolution, 1(3), 278–289. https://doi.org/10.1002/ece3.22 Chapple, D. G., Whitaker, A. H., Chapple, S. N. J., Miller, K. A., & Thompson, M. B. (2013). Biosecurity interceptions of an invasive lizard: Origin of stowaways and human-assisted spread within New Zealand. Evolutionary Applications, 6(2), 324–339. https://doi.org/10.1111/eva.12002 Charmantier, A., & Garant, D. (2005). Environmental quality and evolutionary potential: Lessons from wild populations. Proceedings of the Royal Society B: Biological Sciences, 272(1571), 1415–1425. https://doi.org/10.1098/rspb.2005.3117 Cheetham, E., Doody, J. S., Stewart, B., & Harlow, P. (2011). Embryonic mortality as a cost of communal nesting in the delicate skink. Journal of

26

Zoology, 283(4), 234–242. https://doi.org/10.1111/j.1469- 7998.2010.00764.x Chin, E. H., Love, O. P., Verspoor, J. J., Williams, T. D., Rowley, K., & Burness, G. (2009). Juveniles exposed to embryonic corticosterone have enhanced flight performance. Proceedings of the Royal Society B: Biological Sciences, 276(1656), 499–505. https://doi.org/10.1098/rspb.2008.1294 De Villemereuil, P., & Nakagawa, S. (2014). General quantitative genetic methods for comparative biology. In Modern Phylogenetic Comparative Methods and Their Application in Evolutionary Biology (pp. 287–303). Springer Berlin Heidelberg. https://doi.org/10.1007/978-3-662-43550- 2_11 Dingemanse, N. J., & Dochtermann, N. A. (2013). Quantifying individual variation in behaviour: Mixed-effect modelling approaches. Journal of Animal Ecology, 82(1), 39–54. https://doi.org/doi: 10.1111/1365- 2656.12013 Dyke, J. U. V., & Griffith, O. W. (2018). Mechanisms of reproductive allocation as drivers of developmental plasticity in reptiles. Journal of Experimental Zoology Part A: Ecological and Integrative Physiology, 329(6–7), 275–286. https://doi.org/10.1002/jez.2165 Falconer, D. S., & Mackay, T. F. C. (2009). Introduction to Quantitative Genetics (4. ed., [16. print.]). Pearson, Prentice Hall. Forsman, A., & Shine, R. (1995). Parallel geographic variation in body shape and reproductive life history within the Australian scincid lizard Lampropholis delicata. Functional Ecology, 9(6), 818–825. https://doi.org/10.2307/2389979 Gluckman, P. D., Hanson, M. A., & Low, F. M. (2019). Evolutionary and developmental mismatches are consequences of adaptive developmental plasticity in humans and have implications for later disease risk. Philosophical Transactions of the Royal Society B: Biological Sciences, 374(1770), 20180109. https://doi.org/10.1098/rstb.2018.0109 Goolsby, E. W., Bruggeman, J., & Ané, C. (2016). Rphylopars: Fast multivariate phylogenetic comparative methods for missing data and within-species variation. Methods in Ecology, 8(1), 22–27. https://doi.org/10.1111/2041- 210X.12612

27

Greer, A. E. (1990). The biology and evolution of Australian lizards. Surrey Beatty and Sons. Harman, D. (1956). Aging: A theory based on free radical and radiation chemistry. Journal of Gerontology, 11(3), 298–300. https://doi.org/10.1093/geronj/11.3.298 Hoffman, A. A., & Parsons, P. A. (1991). Evolutionary genetics and evolutionary stress. Oxford University Press. Hoffmann, A. A., & Merilä, J. (1999). Heritable variation and evolution under favourable and unfavourable conditions. Trends in Ecology & Evolution, 14(3), 96–101. https://doi.org/10.1016/S0169-5347(99)01595-5 Houle, D. (1992). Comparing evolvability and variability of quantitative traits. Genetics, 130(1), 195–204. Kingsolver, J. G., & Huey, R. B. (2008). Size, temperature, and fitness: Three rules. Evolutionary Ecology Research, 10(2), 251–268. Lynch, M., & Walsh, B. (1998). Genetics and analysis of quantitative traits. Oxford University Press. Matthews, G., Goulet, C. T., Delhey, K., & Chapple, D. G. (2016). The effect of skin reflectance on thermal traits in a small heliothermic ectotherm. Journal of Thermal Biology, 60, 109–124. https://doi.org/10.1016/j.jtherbio.2016.06.013 Moczek, A. P. (1998). Horn polyphenism in the beetle Onthophagus taurus: Larval diet quality and plasticity in parental investment determine adult body size and male horn morphology. Behavioral Ecology, 9(6), 636–641. https://doi.org/10.1093/beheco/9.6.636 Nakagawa, S. (2015). Missing data: Mechanisms, methods and messages. In G. A. Fox, S. Negrete-Yankelevich, & V. J. Sosa (Eds.), Ecological Statistics: Contemporary theory and application (pp. 81–105). Oxford University Press. Nakagawa, S., & Freckleton, R. P. (2008). Missing inaction: The dangers of ignoring missing data. Trends Ecology and Evolution, 23(11), 592–596. https://doi.org/10.1016/j.tree.2008.06.014 Nespolo, R. F., & Franco, M. (2007). Whole-animal metabolic rate is a repeatable trait: A meta-analysis. The Journal of Experimental Biology, 210(21), 3877– 3878. https://doi.org/10.1242/jeb.013110 Nettle, D., & Bateson, M. (2015). Adaptive developmental plasticity: What is it, how can we recognize it and when can it evolve? Proceedings. Biological

28

Sciences, 282(1812), 20151005–20151009. https://doi.org/10.1098/rspb.2015.1005 Nilsson, J.-Å., Åkesson, M., & Nilsson, J. F. (2009). Heritability of resting metabolic rate in a wild population of blue tits. Journal of Evolutionary Biology, 22(9), 1867–1874. https://doi.org/10.1111/j.1420- 9101.2009.01798.x Noble, D. W. A., & Nakagawa, S. (2018). Planned missing data design: Stronger inferences, increased research efficiency and improved animal welfare in ecology and evolution. BioRxiv 247064, 8, 257–32. https://doi.org/10.1101/247064 Norin, T., & Metcalfe, N. B. (2019). Ecological and evolutionary consequences of metabolic rate plasticity in response to environmental change. Philosophical Transactions of the Royal Society B: Biological Sciences, 374(1768), 20180180–20180189. https://doi.org/10.1098/rstb.2018.0180 Nussey, D. H., Wilson, A. J., & Brommer, J. E. (2007). The evolutionary ecology of individual phenotypic plasticity in wild populations. Journal of Evolutionary Biology, 20(3), 831–844. https://doi.org/10.1111/j.1420- 9101.2007.01300.x Peace, J. E. (2004). Distribution, habitat use, breeding and behavioural ecology of rainbow skinks (Lampropholis delicata) in New Zealand [Massey University]. http://www.massey.ac.nz/ mbaling/PeaceJ/Peace_MScThesis.pdf Pennell, M. W., FitzJohn, R. G., & Cornwell, W. K. (2016). A simple approach for maximizing the overlap of phylogenetic and comparative data. Methods in Ecology, 7(6), 751–758. https://doi.org/10.1111/2041-210X.12517 Ponzi, E., Keller, L. F., Bonnet, T., & Muff, S. (2018). Heritability, selection, and the response to selection in the presence of phenotypic measurement error: Effects, cures, and the role of repeated measurements. Evolution, 72(10), 1992–2004. https://doi.org/10.1111/evo.13573 Réale, D., Garant, D., Humphries, M. M., Bergeron, P., Careau, V., & Montiglio, P.-O. (2010). Personality and the emergence of the pace-of-life syndrome concept at the population level. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 365(1560), 4051–4063. https://doi.org/10.1098/rstb.2010.0208

29

Ricklefs, R. E., & Wikelski, M. (2002). The physiology/life- history nexus. Trends Ecology and Evolution, 17(10), 1–7. https://doi.org/10.1016/S0169- 5347(02)02578-8 Rønning, B., Jensen, H., Moe, B., & Bech, C. (2007). Basal metabolic rate: Heritability and genetic correlations with morphological traits in the zebra finch. Journal of Evolutionary Biology, 20(5), 1815–1822. https://doi.org/10.1111/j.1420-9101.2007.01384.x Rowiński, P. K., & Rogell, B. (2017). Environmental stress correlates with increases in both genetic and residual variances: A meta-analysis of animal studies. Evolution, 71(5), 1339–1351. https://doi.org/10.1111/evo.13201 Sæther, B.-E. (1987). The influence of body weight on the covariation between reproductive traits in European birds. Oikos, 48(1), 79–88. https://doi.org/10.2307/3565691 Seebacher, F., Beaman, J., & Little, A. G. (2014). Regulation of thermal acclimation varies between generations of the short-lived mosquitofish that developed in different environmental conditions. Functional Ecology, 28(1), 137–148. https://doi.org/10.1111/1365-2435.12156 Sgrò, C. M., & Hoffmann, A. A. (2004). Genetic correlations, tradeoffs and environmental variation. Heredity, 93(3), 241–248. https://doi.org/10.1038/sj.hdy.6800532 Tingley, R., Thompson, M. B., Hartley, S., & Chapple, D. G. (2016). Patterns of niche filling and expansion across the invaded ranges of an Australian lizard. Ecography, 39(3), 270–280. https://doi.org/10.1111/ecog.01576 Van Noordwijk, A. J., & De Jong, G. (1986). Acquisition and allocation of resources—Their influence on variation in life-hitory tactics. The American Naturalist, 128(1), 137–142. Via, S., Gomulkiewicz, R., De Jong, G., Scheiner, S. M., Schlichting, C. D., & Van Tienderen, P. H. (1995). Adaptive phenotypic plasticity: Consensus and controversy. Trends Ecology and Evolution, 10(5), 212–217. https://doi.org/10.1016/S0169-5347(00)89061-8 West-Eberhard, M. J. (2005). Developmental plasticity and the origin of species differences. Proceedings of the National Academy of Sciences, 102(Supplement 1), 6543–6549. https://doi.org/10.1073/pnas.0501844102

30

Wilson, A. J. (2018). How should we interpret estimates of individual repeatability? Evolution Letters, 219, 631–635. https://doi.org/10.1002/evl3.40 Wilson, A. J., Reale, D., Clements, M. N., Morrissey, M. M., Postma, E., Walling, C. A., Kruuk, L. E. B., & Nussey, D. H. (2010). An ecologist’s guide to the animal model. Journal of Animal Ecology, 79(1), 13–26. https://doi.org/10.1111/j.1365-2656.2009.01639.x

31

CHAPTER 2

Individual variation in thermal plasticity and its impact on mass-scaling

Fonti Kar1, Shinichi Nakagawa1,2, Christopher R Friesen3, Daniel W.A. Noble4

1School of Biological Earth and Environmental Sciences, Ecology and Evolution Research Centre, University of New South Wales, Sydney, NSW, Australia 2Diabetes and Metabolism Division, Garvan Institute of Medical Research, 384 Victoria Street, Darlinghurst, Sydney, NSW 2010, Australia 3School of Earth, Atmospheric and Life Sciences, Faculty of Science, Medicine and Health, University of Wollongong, Wollongong, NSW, Australia 4Division of Ecology and Evolution, Research School of Biology, The Australian National University, Canberra, ACT, Australia

Currently in review in Oikos

All authors conceived the ideas and designed the study. FK and CF collected the data; FK, DN, SN analysed the data; FK wrote the first draft and all authors contributed to revising the manuscript. All authors declare no conflict of interest

32

Abstract

Physiological processes of individuals can be highly variable and accumulating evidence shows that individuals differ in their response to environmental change. Repeatability, or lack thereof, in metabolic rate across temperatures (i.e., metabolic thermal plasticity) may affect mass-scaling at the population level and has important consequences for understanding the evolution of reaction norms. Nonetheless, only a small number of studies have explicitly quantified repeatability in metabolic plasticity, and fewer have explored how it can impact mass-scaling. We repeatedly measured standard metabolic rate of forty-two delicate skinks (Lampropholis delicata) at six temperatures over the course of three months (N[observations] = 5040). Using hierarchical statistical techniques, we accounted for multi-level variation and measurement error in our data in order to quantify more precise estimates of reaction norm repeatability and mass-scaling exponents at different acute temperatures. Our results show that individual differences in metabolic thermal plasticity was consistent over time, albeit repeatability estimates were weak. After accounting for measurement error which increased steadily with temperature, we show that among individual variance remained consistent across all temperatures. Congruently, temperature specific repeatability of average metabolic rate was stable across temperatures. Cross-temperature correlations were positive but were not uniform across the reaction norm. After taking into account multiple sources of variation, our estimates for mass-scaling did not change with temperature and were in line with published values for snakes and lizards. This implies that repeatable plastic responses may contribute to thermal stability of scaling exponents. Our work contributes to our understanding of how energy expenditure scales with abiotic and biotic factors and the capacity for reaction norms to respond to selection. This is pertinent for ectotherms coping with rapid environmental change within their lifetime.

33

Introduction

All biological processes hinge on the availability of energy (Allen et al., 2005). Metabolic rate (MR) governs how much energy is available to be allocated to competing processes such as growth, reproduction and maintenance (Biro & Stamps, 2008; Brown et al., 2004; De Jong & Van Noordwijk, 1992). Metabolic rate is thought to be critical to fitness due to its functional links to morphology, behaviour and life-history promoting the integration of these traits (Biro & Stamps, 2010; Friesen et al., 2017; Malishev et al., 2017; Réale et al., 2010). For example, short-lived ecotypic garter snakes (Thamnophis elegans) tend to have much higher mass-specific metabolic rates, larger body sizes, faster growth rates and invests more heavily into reproduction compared their long lived ecotypic counterparts (Bronikowski & Vleck, 2010). The integration of these traits may be due to the close association between body mass and metabolic rate. Body mass and metabolic rate typically show a power relationship with an scaling exponent ranging from 0.64 to 0.88 (White et al., 2006). Scaling exponents less than one indicates that energy expenditure scales disproportionately with mass, such that small organisms tend to have a much energy expenditure after controlling for body mass. Metabolic scaling exponents are incredibly heterogenous among (Uyeda et al., 2017; White et al., 2006) and within taxa (Burton et al., 2011; Norin & Gamperl, 2018), yet the drivers of such variation is not well understood.

One powerful application of mass-scaling relationships is its ability to explain and predict ecological processes across levels of biological organisation (Allen et al., 2005; Barneche & Allen, 2018; Brown et al., 2004). In these theoretical studies, among and within individual variation of energy consumption is assumed to be the same; however, few empirical studies have actually tested this assumption. Indeed, individuals can vary in their relative organ mass and body composition yielding very disparate energetic demands in different environments (Scott et al., 1996; Steyermark, 2005). Additionally, variation in mitochondrial efficiency underpins stark differences in MR in fish (Salmon trutta) despite mass remaining the same (Salin et al., 2016). Ignoring individual variability in physiological processes may be problematic for comparative studies as individual effects can be erroneously absorbed into higher levels of

34

biological organisation (van de Pol & Wright, 2009). This may bias mass-scaling exponents and increase heterogeneity among studies. Furthermore, mass- scaling exponents may be susceptible to sampling variability because metabolic rate and body mass tend to be measured once per individual and then averaged across a population. Understanding the consistency of metabolism at the individual level may help explain interspecific variation in mass-scaling exponents (Uyeda et al., 2017).

Temperature fluctuates extensively within the lifetime of ectothermic organisms and this has a profound impact on metabolic rate. Numerous studies have found that scaling exponents show temperature dependence in a multitude of ways; however, the pattern is highly species-specific (Barneche et al., 2016; Glazier, 2005). For example, mass-scaling exponents increased with temperature in teleost fish (Killen et al., 2010), but decreased with temperature in crustaceans (Ivleva, 1980). In contrast, mass-scaling exponents was stable across temperatures in tegu lizards (Toledo et al., 2008). Temperature dependence of mass-scaling relationships imply that metabolic costs for individuals of varying body sizes depend on the thermal environment (Barneche et al., 2016). However, individuals can also vary in their metabolic thermal plasticity, that is, their capacity to adjust their metabolic rate in response to temperature (Individual x Temperature, Nussey et al., 2007). Individual thermal plasticity can be important for understanding temperature dependence of mass-scaling and how selection might shape these plastic responses; however, this has rarely been considered (Barneche et al., 2016; Piersma & Drent, 2003). Low consistency in individual thermal plasticity can introduce variability in metabolic rate across temperatures which can give rise to spurious patterns of temperature dependence. If individuals respond to temperature consistently though, mass-scaling is expected to be robust to temperature changes (Clarke 2004). Consistent variation in metabolic thermal plasticity is also the minimum requirement for plasticity to evolve as it represents the raw material for selection to act on (Wilson, 2018). Despite studies on a range of taxa recognising that individuals differ in their metabolic thermal plasticity, its repeatability has rarely been formally estimated (but see Briga & Verhulst, 2017; Réveillon et al., 2019)

35

Here we examine how individuals differ in energy expenditure in relation to body size and acute temperature changes and how it may impact mass-scaling exponents in male delicate skinks (Lampropholis delicata). We repeatedly measured routine metabolic rate over four months address three key questions. (1) Does metabolic thermal plasticity consistently differ among individuals? (2) How does repeatability of MR change at a given temperature? (3) Do population mass-scaling exponents change with temperature when accounting for among- and within-individual variation in MR? Unravelling the complexities of individual physiological processes will have important consequences for understanding how populations respond in warming environments.

36

Materials and Methods

Lizard collection and husbandry

Lampropholis delicata is a small oviparous, skink found in throughout Eastern Australia (Chapple et al., 2011). They have a short lifespan (2 – 4 years in the wild) and their reproductive season is from September – February (Chapple et al., 2014). Between 28 August and 8 September 2015, forty-two adult (snout- vent-length > 40mm) male L. delicata were collected from three sites near Sydney, Australia (UNSW Kensington Campus: -33.92, 151.24; Sydney Park: - 33.91, 151.18, Macquarie Park: -33.77, 151.10). Lizards were caught by hand or by mealworm fishing and were transported individually in calico bags in an ice-cooler to Macquarie University. Lizards were housed in a temperature- controlled room set at 26ºC and were provided with a thermal gradient to allow for thermoregulation (24ºC – 34ºC). Each lizard was kept individually in an opaque plastic enclosure measuring 35cm x 25cm x 15cm (L x W x H). Each enclosure was lined with newspaper and lizards were given access to a water bowl and tree bark as a refuge. Enclosures were placed under UV light (11 hours light:13 hours dark cycle) . Lizards were fed three to four small crickets (Acheta domestica) dusted with calcium powder and multi-vitamin every two days when metabolism measurements were not taking place. Animal collection was approved by the New South Wales National Parks and Wildlife Service (SL101549) and procedures were approved by the Macquarie University Ethics committee (ARA 2015/015) and University of New South Wales Animal Care and Ethics committee (ACEC 15/51A).

Measuring metabolic rate at different temperatures

Given the scale of our experiment, we used closed-system respirometry instead of intermittent-flow through respirometry. We measured routine metabolic rate (hereafter referred to as metabolic rate [MR]) as our measurements also included the energetic costs of random activity that we were not able to completely control for (Withers 1992; Mathot & Dingemanse 2015). MR was

̇ -1 measured as the volume of CO2 production per unit time (� mL min ) for animals in a post-absorptive state because CO2 production is more sensitive to change in smaller organisms, and is less susceptible to fluctuations in water

37

vapour. Our data showed that CO2 production was strongly correlated with O2 consumption nonetheless (r =0.94, p = <0.05]). Measurements took place between 26 December 2016 - 19 March 2017. Lizards were randomly assigned to one of two blocks for MR measurements (block 1: n = 23, block 2: n = 22). We used two incubators (LabWit, ZXSD-R1090) to precisely control the acute temperature at which measurements were taken (+/- 1ºC). Measurements were taken in a random order at 22ºC, 24ºC, 26ºC , 28ºC , 30ºC and 32ºC over three days (measurements at two temperatures per day). Each animal was repeatedly measured across these temperatures every 10 days (10 sampling sessions in total). We also statistically accounted for the order of temperatures animals experienced in our analyses to control for any possible carry over effects that higher temperatures may have on individuals in subsequent MR measurements (see below).

Lizards were in a post-absorptive state after fasting for 24-hour as digestion can influence MR measurements. We recorded the body temperature of each individual inside their enclosure was taken using an infrared laser gun (Stanley stht0-77365) in the morning (~06:00). Each lizard was gently encouraged into their 146mL opaque chamber and then weighed using a digital scale to the nearest 0.01g (Ohaus SP-202). Chambers were placed inside the incubators in the dark at a randomised temperature for 30 minutes. The lids of the chambers were left ajar during this time to minimise CO2 build up. After 30 minutes, each chamber was flushed with fresh air and sealed. A 3 mL ‘control/baseline’ air sample was immediately taken via a two-way valve to account for any residual

CO2 that was not flushed from the chambers. The chambers were left in the incubator at the set temperature for lizards to respire for 90 minutes. After this time, two 3mL air samples were taken from each chamber. Chambers were then reopened and flushed with fresh air before placed back into the incubator for the second measurement temperature (2 temperatures / day) following the same procedure.

All air samples were injected into the inlet line of a Sables System FMS (Las

-1 ̇ ̇ Vegas NV, USA) with the flow rate set to 200 mL min to measure � and �. Water vapour was scrubbed from the inlet air with Drierite. Output peaks were processed using the R package ‘metabR’

38

(https://github.com/daniel1noble/metabR). The rate of CO2 produced by an individual was calculated following equation 4.21 in Lighton 2008:

%�� × (� − �) �̇ mL ��� = � where %CO2 is the maximum percentage of CO2 in air sample above baseline, which was corrected by subtracting any ‘residual’ CO2 from the initial flush from the larger of the two air samples; Vchamber is the volume of the chamber (146 mL); Vlizard is the volume of the lizard, assuming that the mass of the lizard is the same as its volume, and t is the duration of time in minutes after where the chamber has been sealed and the first air sample was taken (90 minutes).

Statistical analysis

All statistical analyses were conducted using the R environment, version 3.6.1 (Core Team 2013). Details on data cleaning are presented in the electronic supplementary materials (ESM). Initial analyses showed that there were no ̇ differences in log� between statistical blocks of lizards therefore ‘block ID’ was not included in our final models (ESM). Although lizards were kept in a temperature-controlled room, there may still have been temperature differences between enclosures that had carry-over effects on metabolic rate. We tested whether the body temperature measured in the home enclosure before the first measurement or the previous measurement temperature (if MR measurements ̇ were underway) influenced log� at subsequent temperatures. We found that a model containing ‘previous temperature experience’ as a covariate was better supported compared to a model without it (∆WAIC (Full model – reduced model = -8.39), we therefore included ‘previous temperature experience’ in all subsequent analyses (Table S1). Collinearity between our predictor variables was checked using a scatterplot matrix (Fig. S1) and Pearson correlation coefficients are presented in Table S2. All data and code with which to generate our results are openly available via the Open Science Framework (see Data Accessibility).

We used Bayesian linear mixed models from either the package ‘brms’ (Bürkner, 2017) or ‘MCMCglmm' (Hadfield, 2010). For logistical reasons, we fitted the random slope model using ‘MCMCglmm’, and a multivariate response model using ‘brms’. Details on model priors and set up are presented 39

in the ESM. For every model, we pooled the posterior estimates from multiple chains and presented posterior means and their 95% credible intervals.

Measurement error and repeatability of metabolic thermal plasticity

Repeatability is a ratio of among-individual and residual variance components

(R = sA / (sA + sR) and represents the proportion of phenotypic variance attributed to among-individual differences (Nakagawa & Schielzeth 2010). The relative contribution to each variance component can shed light on the processes that promote repeatable traits (Dingemanse & Dochtermann, 2013). Measurement error; however, can bias the estimation of variance components and affect repeatability estimates (Ponzi et al., 2018). Given that we took two air samples for each MR measurement, we are able to partition measurement error among the two samples by including a nested random effect of individual ID, sampling session and temperature (Individual_ID:Session_ID:Temp, hereafter referred to as measurement error) in our models. This term partitions out variance attributed to measurement error among replicates so that the residual variance represents within individual variance. We also wanted to take into consideration that metabolic rate could change over time our study spanned over four months. We therefore fitted a nested random effect of individual ID and sampling session (ID:sampling session, hereafter referred to as series) in our models to decompose among sampling session within individual variance (see Araya-Ajoy et al., 2015 for further explanation)

We fitted the following random slope model in 'MCMCglmm’ (nobs = 4952) in order to quantify the repeatability of metabolic thermal plasticity (i.e. slopes for each individual). ̇ logV ~ Temp + zlogBodyMass + PriorTemp + (1+ Temp |

Individual_ID) + (1+ Temp | Individual_ID:Session_ID) + (1 | ID:Session_ID:Temp) ̇ ̇ where: logV is log-transformed V; Temp is the temperature in degrees Celsius; zlogBodyMass is log-transformed body mass that is then subsequently z-transformed; PriorTemp is previous temperature experienced by the lizard (enclosure temperature or the previous treatment temperature). Individual ID and series and measurement error was included as a random intercept. Temperature was included as a random slope for both individual ID and series to estimate individual slopes and among-sampling session, within individual 40

slopes. The repeatability of the slope is calculated following equation 1 in the ESM (see also Araya-Ajoy et al., 2015).

Temperature-specific Repeatability and Cross-Temperature Correlations of

Metabolic Rate

After assessing whether individuals differ in their metabolic thermal plasticity, we were interested in knowing whether consistent among-individual differences in average MR change across temperatures. To achieve this, we fitted a multivariate response model by treating MR measurements for each of the six temperatures as separate traits (nobs = 802) in a 6 x 6 response matrix: VCO VCO … VCO ⎡ ,,º ,,º ,,º ⎤ ⎢ VCO,,º VCO,,º … VCO,,º ⎥ ⎢ ⎥ ~ zlogBodyMass + ⎢ ⋮ ⋮ ⋱ ⋮ ⎥ ⎣VCO,,º VCO,,º … VCO,,º⎦ PriorTemp + (1|ID) + (1| Individual_ID:Session_ID) where, VCO,,º is the metabolic rate for individual 1 in sampling session 1 at

22ºC and VCO,,º is the metabolic rate for individual 1 in sampling session 10 at 22ºC and so forth. Similar to the random slope models, we included zlogBodyMass and PriorTemp as covariates. Note that temperature is no longer a predictor or a random slope term as temperature is now part of the response matrix. In some instances, mechanical errors occurred during air collection. Given that ‘brms’ requires complete data in the response matrix, we used the ‘mi’ function to impute the missing samples at each temperature as this prevented us to exclude 607 rows of data. We included individual ID and series were as random intercepts. In this model, series is responsible for partitioning out measurement error from the residuals. We calculated temperature specific repeatability following Equation 2 in the ESM.

We were also interested in the extent to which MR was correlated across all temperatures as this may illuminate trade-offs in physiological function at different temperatures. We obtained cross-temperature correlations at the among-individual level using the variance-covariance matrix obtained from the multivariate response model.

41

Mass-scaling exponents at different temperatures

Population estimates of scaling exponents can be affected by within and among individual variation (van de Pol & Wright 2009). We therefore wanted to partition out within individual effects in order to obtain more precise estimates of mass-scaling across temperatures. To achieve this, we calculated the mean mass across all sampling sessions for each individual (among individual effect), and subtracted an individual’s mass from its own mean to account for within individual effects (also known as within-individual centering, see van de Pol & Wright 2009). These mass effects were log-transformed and included in two models fitted in ‘brms’ (nobs = 3933). The first model (interaction model) had the following structure, ̇ logV ~ Temp * logAmongIDMass + Temp * logWithinIDMass + (1 +

logWithinIDMass|ID) + (1 | ID:sampling session :Temp) where: Temp * logAmongIDMass is the interaction term between temperature and the log transformed among individual mass effect; Temp * logWithinIDMass is the interaction term between temperature and the log transformed within individual mass effect. Individual ID was fitted a random intercept with logWithinIDMass as a random slope as individuals masses changed at different rates through the study (see Fig S3). We also included the measurement error term. The second model (main effects model) only had the main effects of temperature, the among individual mass effect and the within- individual mass effect and the same random effects structure as the interaction model. We tested whether population mass-scaling exponents (i.e. the among individual mass effects) changed with temperature by comparing different information criterion (wAIC and loo values) between model one and two. We also present in the ESM (Fig. S4, Table S5) an analysis that compared the mass scaling exponents with estimates from a model that represents the typical analysis of a metabolic scaling study from a model that did not account for the multi-level variation in the data.

Results

Repeatability of metabolic thermal plasticity

Individual slopes describing the effect of temperature on MR were weakly -3 repeatable (Rslope = 0.23, Lower CI = 1.52 ×10 , Upper CI = 0.84), suggesting 42

individuals consistently varied in how their metabolic rate changed with temperature (Fig. 1).

a) b) c) 1 5 10 −2.5 ) 1 - g 1 - −3.0 ) 2 min O 2 C V

( −3.5

e t a r

c i l −4.0 o b a t e m

−4.5 g o L

Log Metabolic Rate (VCO Rate Metabolic Log −5.0

22 24 26 28 30 32 22 24 26 28 30 32 22 24 26 28 30 32 Temperature °C

̇ Figure 1. Individual reaction norms of log mass corrected metabolic rate (� mL min-1) at six measurement temperatures at (a) sampling session one, (b) five and (c) ten. Points are predicted values from the model and reaction norms are drawn using the geom_line function. Each line represents a unique individual (n = 42).

Repeatability of metabolic rate at each temperature

We found that the repeatability of MR (i.e. individual intercepts) was stable across acute temperatures (Fig. 2). Temperature-specific repeatability was greatest at 24ºC; however, credible intervals overlapped with estimates at other temperatures (Fig. 2 ,Table S3). Upon closer inspection of the variance components at each temperature, we show that measurement error decreased steadily with increasing temperature, whereas among individual variation remained relatively consistent with temperature (Fig 2). In contrast, within individual variance showed no consistent pattern with temperature; however, it was highest in 32ºC. In other words, individuals were responding more

43

variably as 32ºC while differences among individual maintained relatively stable (Fig 2). Figure 2. (A) Posterior mean of repeatability and variance components of log

Measurement Error Among ID Within ID

0.4

0.3 Variance 0.2

0.1 22 24 26 28 30 32 22 24 26 28 30 32 22 24 26 28 30 32 Temperature °C

1.0 0.9 0.8 0.7 0.6 0.5 0.4

Repeatability 0.3 0.2 0.1 0.0 22 24 26 28 30 32 Temperature °C

̇ -1 metabolic rate (� mL min ) at six measurement temperatures estimated over four-month period across n = 42 individuals. Error bars represent 95% credible intervals.

Cross-temperature correlations in metabolic rate

Metabolic rate across temperatures were positively correlated at among- individuals (Fig. 3, Table S5). Positive correlations indicate that some individuals maintained a consistently high metabolic rate relative to other individuals, while others had a relatively low metabolic rate across all temperatures. Metabolic rate measured at neighbouring temperatures (e.g. 22ºC 44

and 24ºC) were strongly correlated, but the strength of this correlation decreased with increasing differences between the two temperatures (Fig. 3).

Among-individual correlations

1

22 0.8

0.6 0.59 24 0.4

0.2 0.63 0.71 26

0

0.49 0.67 0.79 28 −0.2

−0.4 0.51 0.64 0.79 0.84 30 −0.6

−0.8 0.46 0.51 0.69 0.8 0.75 32

−1

̇ -1 Figure 3 Cross-temperature correlations of log metabolic rate (� mL min ) at the among-individual level estimated from n = 42 individuals. Diagonal values are each measurement temperatures. Lower triangle represents posterior mean estimates of correlations. Width and colour of the ellipse in the upper triangle represents the strength of the correlation.

Temperature dependence of population mass-scaling exponents

The model containing only the main effects of temperature was better supported than a model that included the interaction terms (Main effects model: WAIC = 2133.9, loo = 2322.2, Interaction model: WAIC = 2124.90, loo = 2358.5), suggesting a lack of temperature dependence in mass scaling (Fig. 4). Across all temperatures, the average mass-scaling exponent was 0.96 (Lower CI = 0.39, Upper CI = 1.52) which is in line with values reported for squamates (Uyeda et al., 2017). Mass-scaling exponents tended to be spurious and estimated with a larger degree of error when the within individual effects and measurement error were not statistically accounted for (Fig. S4, Table S5).

45

a)

1.5

1.0 scaling exponent estimate scaling exponent

− 0.5 Mass

0.0 22 24 26 28 30 32 b) Temperature 22 24 26 28 30 32 −2 ) 1 - n i m

−3 2 O C V (

e t

a −4 r

c i l o b a t e

M −5

g o L

−6 −0.2 0.0 0.2 0.4−0.2 0.0 0.2 0.4−0.2 0.0 0.2 0.4−0.2 0.0 0.2 0.4−0.2 0.0 0.2 0.4−0.2 0.0 0.2 0.4 Log Mass (g)

Figure 4. (a) Posterior mean estimates of population mass scaling exponents ̇ -1 (i.e. among individuals) of log metabolic rate (� mL min ) across six measurement temperatures when within individual variation in mass and measurement error in metabolic rate has been statistically accounted for. The dashed line represents the mass-scaling exponent of 0.83 estimated for squamates from Uyeda (2017). Error bars represent 95% credible intervals. (b) Raw log metabolic rate plotted against log body mass for a random subset of n = 20 individuals at six measurement temperatures. Each uniquely coloured point represents one individual. Thick bold line represents the change in log metabolic rate over log body mass across all individuals (among-individual mass-scaling slope). Thin lines represent the change in log MR over log body mass within an individual (within-individual mass-scaling slopes)

46

Discussion

Our results show that metabolic thermal plasticity was weakly repeatable over the four months of study in delicate skinks. Moreover, the repeatability of average MR was also not susceptible to acute temperature changes. Cross- temperature correlations of MR were all positive at the among-individual level. However, the strength of these correlations was not uniform across all temperatures. Mass scaling exponents were not strongly affected by temperature and in line with values reported for squamates when other sources of variation were partitioned out. Below we discuss the implications of our results for understanding how plasticity may evolve, and how SMR scales at different hierarchical levels.

Consistent variation in metabolic thermal plasticity

Natural selection acts on phenotypic variation among individuals. Consistent among-individual variation is therefore a key prerequisite for any trait to evolve and sets the ‘upper limit of heritability’ (Falconer, 1952; c.f. Dohm, 2002). Our findings show that individuals differ consistently in how their MR responds to acute temperature changes over an ecologically relevant time period. Assuming that individual differences have a genetic basis and are therefore heritable, our results suggests that metabolic thermal plasticity may be capable of evolutionary change allowing shifts in population-level metabolic reaction norms (Ghalambor et al., 2007). Average metabolic rate was also repeatable and stable across temperatures and suggests that the operable range of temperature in L.delicata promotes consistency in physiological traits (Goulet et al., 2017; Matthews et al., 2016). To our surprise, measurement error declined with increasing temperature presumably because individuals were respiring at a higher rate maybe it easier to detect changes in CO2 production. Measurement error can inflate repeatability estimates if it is not accounted for statistically (Ponzi et al., 2018). Indeed, we found a significant increase in repeatability and among individual variance when we took averages between the two replicate air samples (Fig. S5). Consequently, one would mistakenly conclude that the capacity to selection to act on MR would increase at hotter temperatures. stress the importance of accounting for confounding sources of variance. We stress the importance of considering confounding sources of variances such as

47

measurement error or shared environmental effects among individuals to ascertain the potential for repeatable physiological traits to undergo selection.

Cross-temperature correlations

Metabolic rate was positively correlated across all temperatures at the among- individual level. This suggests that individuals with high MR at one temperature also tend to exhibit high MR at other temperatures (and vice versa for individuals with low MR). Individuals could vary in their acquisition or allocation of resources to their physiological system which enables certain individuals to maintain a consistently high MR across all temperatures (Angilletta Jr et al., 2003; De Jong & Van Noordwijk, 1992). Moreover, consistent individual differences in MR, irrespective of the thermal environment, may be functionally linked with other aspects of the phenotype (Biro & Stamps, 2010). Our results give precedence to ‘pace-of-life’ theory where individual differences in energetic expenditure may promote consistent differences in behaviour and life-history within the same population (Biro & Stamps, 2010; Careau et al., 2008).

The strength of cross-temperature correlations can help identify trade-offs in physiological processes across environments. Such trade-offs have been hypothesised to be important mechanisms in shaping reaction norms (Angilletta Jr et al., 2003). Generalist-specialist trade-offs occur when some individuals have enhanced physiological function in one environment but diminished function in another environment, manifesting as a negative cross environment correlation (Berger et al., 2014). We show that across different temperatures, correlations were all positive, providing no support for trade-offs between temperatures in energy expenditure. While our temperatures fell within the normal temperature range experienced by animals in the wild, trade- offs may exist in other parts of the thermal performance curve (Angilletta Jr et al., 2003). Assuming phenotypic cross-temperature correlations reflect the underlying genetic architecture of metabolic rate (Roff 1995), the strength of correlation can dictate how strongly selection acting on one component of the reaction norm will result in indirect selection on another (Via et al., 1995). This implies that response to selection would be stronger between neighbouring temperatures (e.g., 28°C vs. 32°C) compared to more distant temperatures (e.g.,

48

22°C vs. 32°C) which might be important to give rise to non-linear reaction norms (Berger et al., 2013).

Population mass scaling across different temperatures

Mass-scaling exponents were robust to acute temperature changes, which is in disagreement with a growing number of studies that show temperature dependence of mass scaling exponents (Barneche et al., 2016; Glazier, 2005, 2015; Killen et al., 2010; Price et al., 2012). Discrepancies may be due to the method with which we quantified mass scaling exponents. In our study, we sampled sexually mature males repeatedly over four months in order to estimate a static mass scaling relationship. It is important to note that the size range among mature males is limited, as such our mass-scaling exponents would be less variable compared to interspecific studies. Additionally, intraspecific studies tend to measure ontogenetic allometry (change in body mass and metabolic rate throughout development, Glazier 2009) as opposed to static allometry. The energetic demands of growth during ontogeny may be more sensitive to temperature change and therefore result in temperature- dependence in ontogenetic mass scaling exponents (Hirst, Glazier & Atkinson 2014; Barneche & Allen 2018). In support of this, a recent comparative analysis has shown that development (passing through life stages) shows stronger temperature dependence than increases in mass (Forster, Hirst & Woodward 2011).

The magnitude and precision of mass scaling exponents may be affected by processes occurring at different hierarchical levels. Genetic and developmental differences that impact the physiological system can maintain permanent differences among individuals (Dingemanse & Wolf 2013). While fluctuations in the internal environment, such as circulating hormones and body composition can affect the within individual responses (Dupoué et al., 2013; McCue, 2010; Scott et al., 1996). After accounting for within individual effects and measurement error, our mass-scaling exponent estimates were in line with values reported from a phylogenetically informed analyses in squamates (Uyeda et al., 2017). This result may have important implications for current designs of metabolic scaling studies as MR and body mass tend to only be measured once, making them sensitive to sampling error and within-individual

49

‘noise’. Theoretical studies that make use of predictive relationship between body mass and metabolism should be more aware of the different sources of variation when trying to extrapolate individual level processes to higher levels of biological organisation. Future work is warranted to investigate the degree to which intra-individual variance in MR and body mass impact scaling exponents as this has largely been neglected and yet may help elucidate why mass scaling exponents are variable at higher levels of biological organisation (Glazier, 2005; Maxwell et al., 2003; McLean & Speakman, 2000).

Conclusion

In this study, we found support that individual consistency of thermal plasticity promotes stability in mass-scaling. Our work implies that selective processes has the opportunity to shape reaction norms of metabolic rate. This may ultimately how populations respond to temperatures and allow them to persist in under warming climate. Quantitative genetic and experimental evolution studies are necessary to truly understand the evolutionary potential of metabolic thermal plasticity. Our work emphasises important methodological considerations that are often overlooked in evolutionary physiological studies. Confounding sources of variance can misconstrue our evolutionary relevance of phenotypic variability in physiological traits (Ponzi et al., 2018). Neglecting to consider individual variation, even in theoretical research may misguide predictions about ecological processes across levels of biological organisation (Botero et al., 2015).

Acknowledgements

This study would not have been possible without the support of the Australian Research Council (ARC) Discovery Early Career Research Award to D. W. A. N (DE150101774); also, S.N. was supported by an ARC Future Fellowship (FT13010026). We recognise The Office of Environment and Heritage, New South Wales for our wildlife collection permit and the animal ethics committee from University of New South Wales and Macquarie University for our animal ethics permit. We express gratitude for all the members of the Lizard Lab at Macquarie University for assistance and support throughout this study. Especially, A/Prof. Martin Whiting for the use of his facilities. We are in debt to Christine Wilson for her assistance with animal husbandry. We really 50

appreciated the help of Stephan Klopper in the construction of our metabolic chambers. We would also like to acknowledge Martin Thompson at the Division of Research, University of New South Wales for his technical aid with using the UNSW computing cluster. Finally, we thank David Mitchell, Tobias Uller insightful discussions and Rose O’Dea for her comments on an earlier draft of this manuscript.

Data accessibility

Datasets and code used to generate results of this study is accessible via Open Science Framework (DOI 10.17605/OSF.IO/TZ2H5).

51

References

Allen, A. P., Gillooly, J. F., & Brown, J. H. (2005). Linking the global carbon cycle to individual metabolism. Functional Ecology, 19(2), 202–213. https://doi.org/10.1111/j.1365-2435.2005.00952.x Angilletta Jr, M. J., Wilson, R. S., Navas, C. A., & James, R. S. (2003). Tradeoffs and the evolution of thermal reaction norms. Trends Ecology and Evolution, 18(5), 234–240. https://doi.org/10.1016/S0169-5347(03)00087-9 Araya-Ajoy, Y. G., Mathot, K. J., & Dingemanse, N. J. (2015). An approach to estimate short-term, long-term and reaction norm repeatability. Journal of Animal Ecology, 6(12), 1462–1473. https://doi.org/10.1111/2041- 210X.12430 Barneche, D. R., White, C. R., & Marshall, D. J. (2016). Temperature effects on mass-scaling exponents in colonial animals: A manipulative test. Ecology, 98(1), 103–111. https://doi.org/10.1002/ecy.1624 Berger, D., Postma, E., Blanckenhorn, W. U., & Walters, R. J. (2013). Quantitative genetic divergence and standing genetic (co)variance in thermal reaction norms along latitude. Evolution, 67(8), 2385–2399. https://doi.org/10.1111/evo.12138 Berger, D., Walters, R. J., & Blanckenhorn, W. U. (2014). Experimental evolution for generalists and specialists reveals multivariate genetic constraints on thermal reaction norms. Journal of Evolutionary Biology, 27(9), 1975–1989. https://doi.org/10.1111/jeb.12452 Biro, P. A., & Stamps, J. A. (2008). Are animal personality traits linked to life- history productivity? Trends Ecology and Evolution, 23(7), 361–368. https://doi.org/10.1016/j.tree.2008.04.003 Biro, P. A., & Stamps, J. A. (2010). Do consistent individual differences in metabolic rate promote consistent individual differences in behavior? Trends Ecology and Evolution, 25(11), 653–659. https://doi.org/10.1016/j.tree.2010.08.003 Botero, C. A., Weissing, F. J., Wright, J., & Rubenstein, D. R. (2015). Evolutionary tipping points in the capacity to adapt to environmental change. Proceedings of the National Academy of Sciences, 112(1), 184–189. https://doi.org/10.1073/pnas.1408589111

52

Briga, M., & Verhulst, S. (2017). Individual variation in metabolic reaction norms over ambient temperature causes low correlation between basal and standard metabolic rate. Journal of Experimental Biology, 220(18), 3280–3289. https://doi.org/10.1242/jeb.160069 Bronikowski, A. M., & Vleck, D. (2010). Metabolism, body size and life span: A case study in evolutionarily divergent populations of the garter snake (Thamnophis elegans). Integrative and Comparative Biology, 50(5), 880–887. https://doi.org/10.1093/icb/icq132 Brown, J. H., Gillooly, J. F., Allen, A. P., Savage, V. M., & West, G. B. (2004). Toward A Metabolic Theory Of Ecology. Ecology, 85(7), 1771–1789. https://doi.org/10.1890/03-9000 Bürkner, P. C. (2017). brms: An R package for Bayesian multilevel models using Stan. Journal of Statistical Software, 80(1). https://doi.org/10.18637/jss.v080.i01 Burton, T., Killen, S. S., Armstrong, J. D., & Metcalfe, N. B. (2011). What causes intraspecific variation in resting metabolic rate and what are its ecological consequences? Proceedings of the Royal Society of London B: Biological Sciences, 278(1724), 3465–3473. https://doi.org/10.1098/rspb.2011.1778 Careau, V., Thomas, D., Humphries, M. M., & Réale, D. (2008). Energy metabolism and animal personality. Oikos, 117(5), 641–653. https://doi.org/10.1111/j.0030-1299.2008.16513.x Chapple, D. G., Hoskin, C. J., Chapple, S. N. J., & Thompson, M. B. (2011). Phylogeographic divergence in the widespread delicate skink (Lampropholis delicata) corresponds to dry habitat barriers in eastern Australia. BMC Evolutionary Biology, 11(1), 191. https://doi.org/10.1186/1471-2148-11-191 Chapple, D. G., Miller, K. A., Chaplin, K., Barnett, L., Thompson, M. B., & Bray, R. D. (2014). Biology of the invasive delicate skink (Lampropholis delicata) on Lord Howe Island. Australian Journal of Zoology, 62(6), 498–506. https://doi.org/10.1071/ZO14098 De Jong, G., & Van Noordwijk, A. J. (1992). Acquisition and allocation of resources: Genetic (co)variances, selection, and life histories. The American Naturalist, 139(4), 749–770. https://doi.org/10.1086/285356

53

Dingemanse, N. J., & Dochtermann, N. A. (2013). Quantifying individual variation in behaviour: Mixed-effect modelling approaches. Journal of Animal Ecology, 82(1), 39–54. https://doi.org/doi: 10.1111/1365- 2656.12013 Dohm, M. R. (2002). Repeatability estimates do not always set an upper limit to heritability. Functional Ecology, 16(2), 273–280. https://doi.org/10.1046/j.1365-2435.2002.00621.x Dupoué, A. az, Brischoux, F., Lourdais, O., & Angelier, F. (2013). Influence of temperature on the corticosterone stress-response: An experiment in the Children’s python (Antaresia childreni). General and Comparative Endocrinology, 193, 178–184. https://doi.org/10.1016/j.ygcen.2013.08.004 Falconer, D. S. (1952). The problem of environment and selection. The American Naturalist, 86(830), 293–298. Friesen, C. R., Johansson, R., & Olsson, M. (2017). Morph-specific metabolic rate and the timing of reproductive senescence in a color polymorphic dragon. Journal of Experimental Zoology Part A-Ecological and Integrative Physiology, 327(7), 433–443. https://doi.org/10.1002/jez.2118 Ghalambor, C. K., McKay, J. K., Carroll, S. P., & Reznick, D. N. (2007). Adaptive versus non-adaptive phenotypic plasticity and the potential for contemporary adaptation in new environments. Functional Ecology, 21(3), 394–407. https://doi.org/10.1111/j.1365-2435.2007.01283.x Glazier, D. S. (2005). Beyond the ‘3/4-power law’: Variation in the intra- and interspecific scaling of metabolic rate in animals. Biological Reviews, 80(4), 611–662. https://doi.org/10.1017/S1464793105006834 Glazier, D. S. (2015). Is metabolic rate a universal ‘pacemaker’ for biological processes? Biological Reviews, 90(2), 377–407. https://doi.org/10.1111/brv.12115 Goulet, C. T., Thompson, M. B., Michelangeli, M., Wong, B. B. M., & Chapple, D. G. (2017). Thermal physiology: A new dimension of the pace-of-life syndrome. Journal of Animal Ecology, 86(5), 1269–1280. https://doi.org/10.1111/1365-2656.12718 Hadfield, J. D. (2010). MCMC methods for multi-response generalized linear mixed models: The MCMCglmm R package. Journal of Statistical Software, 33(2), 1–22.

54

Ivleva, I. V. (1980). The dependence of crustacean respiration rate on body mass and habitat temperature. Internationale Revue der gesamten Hydrobiologie und Hydrographie, 65(1), 1–47. https://doi.org/10.1002/iroh.19800650102 Killen, S. S., Atkinson, D., & Glazier, D. S. (2010). The intraspecific scaling of metabolic rate with body mass in fishes depends on lifestyle and temperature. Ecology Letters, 13(2), 184–193. https://doi.org/10.1111/j.1461-0248.2009.01415.x Malishev, M., Bull, C. M., & Kearney, M. R. (2017). An individual-based model of ectotherm movement integrating metabolic and microclimatic constraints. Methods in Ecology and Evolution, 9(3), 472–489. https://doi.org/10.1111/2041-210X.12909 Matthews, G., Goulet, C. T., Delhey, K., & Chapple, D. G. (2016). The effect of skin reflectance on thermal traits in a small heliothermic ectotherm. Journal of Thermal Biology, 60, 109–124. https://doi.org/10.1016/j.jtherbio.2016.06.013 Maxwell, L. K., Jacobson, E. R., & McNab, B. K. (2003). Intraspecific allometry of standard metabolic rate in green iguanas, Iguana iguana. Comparative Biochemistry and Physiology Part A: Molecular & Integrative Physiology, 136(2), 301–310. https://doi.org/10.1016/S1095-6433(03)00145-4 McCue, M. D. (2010). Starvation physiology: Reviewing the different strategies animals use to survive a common challenge. Comparative Biochemistry and Physiology Part A: Molecular & Integrative Physiology, 156(1), 1–18. https://doi.org/10.1016/j.cbpa.2010.01.002 McLean, J. A., & Speakman, J. R. (2000). Effects of body mass and reproduction on the basal metabolic rate of brown long-eared bats (Plecotus auritus). Physiological and Biochemical Zoology, 73(1), 112–121. https://doi.org/10.1086/316715 Norin, T., & Gamperl, A. K. (2018). Metabolic scaling of individuals vs. Populations: Evidence for variation in scaling exponents at different hierarchical levels. Functional Ecology, 32(2), 379–388. https://doi.org/10.1111/1365-2435.12996 Nussey, D. H., Wilson, A. J., & Brommer, J. E. (2007). The evolutionary ecology of individual phenotypic plasticity in wild populations. Journal of Evolutionary Biology, 20(3), 831–844. https://doi.org/10.1111/j.1420- 9101.2007.01300.x

55

Piersma, T., & Drent, J. (2003). Phenotypic flexibility and the evolution of organismal design. Trends Ecology and Evolution, 18(5), 228–233. https://doi.org/10.1016/S0169-5347(03)00036-3 Ponzi, E., Keller, L. F., Bonnet, T., & Muff, S. (2018). Heritability, selection, and the response to selection in the presence of phenotypic measurement error: Effects, cures, and the role of repeated measurements. Evolution, 72(10), 1992–2004. https://doi.org/10.1111/evo.13573 Price, C. A., Weitz, J. S., Savage, V. M., Stegen, J., Clarke, A., Coomes, D. A., Dodds, P. S., Etienne, R. S., Kerkhoff, A. J., McCulloh, K., Niklas, K. J., Olff, H., & Swenson, N. G. (2012). Testing the metabolic theory of ecology. Ecology Letters, 15(12), 1465–1474. https://doi.org/10.1111/j.1461-0248.2012.01860.x Réale, D., Garant, D., Humphries, M. M., Bergeron, P., Careau, V., & Montiglio, P.-O. (2010). Personality and the emergence of the pace-of-life syndrome concept at the population level. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 365(1560), 4051–4063. https://doi.org/10.1098/rstb.2010.0208 Réveillon, T., Rota, T., Chauvet, É., Lecerf, A., & Sentis, A. (2019). Repeatable inter-individual variation in the thermal sensitivity of metabolic rate. Oikos, 85, 935–938. https://doi.org/10.1111/oik.06392 Salin, K., Auer, S. K., Rudolf, A. M., Anderson, G. J., Selman, C., & Metcalfe, N. B. (2016). Variation in metabolic rate among individuals is related to tissue-specific differences in mitochondrial leak respiration. Physiological and Biochemical Zoology, 89(6), 511–523. https://doi.org/10.1086/688769 Scott, I., Mitchell, P. I., & Evans, P. R. (1996). How does variation body composition affect the basal metabolic rates of birds of birds? Functional Ecology, 10(3), 307. https://doi.org/10.2307/2390277 Steyermark, A. C. (2005). Physiological and morphological correlates of among- individual variation in standard metabolic rate in the leopard frog Rana pipiens. The Journal of Experimental Biology, 208(6), 1201–1208. https://doi.org/10.1242/jeb.01492 Toledo, L. F., Brito, S. P., Milsom, W. K., Abe, A. S., & Andrade, D. V. (2008). Effects of season, temperature, and body mass on the standard metabolic rate of tegu lizards (Tupinambis merianae). Physiological and Biochemical Zoology, 81(2), 158–164. https://doi.org/10.1086/524147

56

Uyeda, J. C., Pennell, M. W., Miller, E. T., Maia, R., & McClain, C. R. (2017). The evolution of energetic scaling across the vertebrate tree of life. The American Naturalist, 2, 185–199. https://doi.org/10.1086/692326 van de Pol, M., & Wright, J. (2009). A simple method for distinguishing within- versus between-subject effects using mixed models. Animal Behaviour, 77(3), 753–758. https://doi.org/10.1016/j.anbehav.2008.11.006 Via, S., Gomulkiewicz, R., De Jong, G., Scheiner, S. M., Schlichting, C. D., & Van Tienderen, P. H. (1995). Adaptive phenotypic plasticity: Consensus and controversy. Trends Ecology and Evolution, 10(5), 212–217. https://doi.org/10.1016/S0169-5347(00)89061-8 White, C. R., Phillips, N. F., & Seymour, R. S. (2006). The scaling and temperature dependence of vertebrate metabolism. Biology Letters, 2(1), 125–127. https://doi.org/10.1098/rsbl.2005.0378 Wilson, A. J. (2018). How should we interpret estimates of individual repeatability? Evolution Letters, 219, 631–635. https://doi.org/10.1002/evl3.40

57

Supplementary Materials

Data processing

We first identified potential outliers in mass using histograms, Cleveland plots and descriptive statistical summaries as per the recommendation of Zuur, Ieno & Elphick (2010). Using these methods, we identified 63/7056 observations across eighteen individuals (<1% of dataset) where the body mass at a given sampling session was drastically different from body mass from the other sampling session. These cases were likely due to measurement and equipment error. Given that the percent of erroneous data was so low, we used mean imputation for each individual within each sampling session to correct for these erroneous cases.

Using similar methods as above, we identified potential outliers in our metabolic rate data. We noticed in the histograms of the percentage change of

CO2 was strongly right skewed. 1711/7056 observations (24% of total unprocessed dataset) were below the first quantile (0.0038), which means that a lot of values were close to zero. 99% of these observations were from ‘control’ samples, which indicates there was little to no residual CO2 left in the chamber when they were flushed and we left as is. The remaining 1% of the observations

(71/7056) were first and second samples of CO2. We suspected that these samples there were lost during collection as the percent of CO2 were typically less than the control samples and as a result we set them as NA.

Finally, we corrected our two samples of CO2 for every measurement we took by subtracting the percent of CO2 found in the control samples. We used both

‘corrected’ air samples for calculating total volume of CO2 produced per min by a lizard (VCO2).

We ran our random slope model in ‘MCMCglmm’ which requires complete cases of predictors (Hadfield, 2010). The original data for the random slope model contained 5040 rows of data and <1% of the data contained NAs. We used a dataset where there are no NAs in our predictors (nobs = 4952). For our ‘MCMCglmm’ models, we ran 3 chains of 7,510,000 iterations with a burn in of

58

10000 and a thinning interval of 5000 and used uninformative, parameter expanded priors to assist with model convergence (Hadfield 2010). We ran our multivariate response model in ‘brms’ (Bürkner, 2017). Unfortunately, ‘brms’ requires complete cases in both response and predictors.

In other words, any rows containing NA in would have been excluded (nobs = 609, 73% of the dataset). 99% of missing data was in metabolic rate and was due to missing values in either first and/or second air replicates in any of the six temperatures (see provided data for more details). To avoid excluding 73% of the dataset, we used the built in function ‘mi()’, to impute the missing values in metabolic rate at each temperature which gives us a dataset of nobs = 802 (96% of the original dataset). For ‘brms’ models, we used default priors and ran 4 MCMC chains of 2000 iterations with a burn in of 1000 and a thinning interval of 1. All models were checked for proper mixing and convergence by visually inspecting trace plots and ensuring scale reduction factors were smaller than 1.1. We also checked that samples from our posterior distribution were not autocorrelated (lag < 0.1).

Testing the importance of statistical block on the relationship on metabolism and body mass

Due to logistical constraints and sample size, lizards were randomly assigned to one of two blocks for metabolism measurements throughout the course of the experiment (block 1: n = 23, block 2: n = 22). We wanted to test whether the relationship in metabolic rate and body mass differed between blocks, and if so, correctly model this variation. Exploratory plots did not show drastic differences between blocks (Fig. S1).

59

a) b)

1 2

−2

−3

−4 log.co2pmin

−5

−6

−0.2 0.0 0.2 0.4 −0.2 0.0 0.2 0.4 log.mass Figure S1 Scatter plot of log-transformed VCO2 and log-transformed body mass of 42 individuals repeated measured over 10 sampling sessions at six different temperatures. Blue line represents the linear line of best fit and the shade areas represent the confidence intervals for this line. a) are lizards assigned to block 1 (n = 23) and b) refer to lizards assigned to block 2 (n = 22).

To test for differences in blocks, we ran a GLMM using ‘brms’ with the following structure,

logVCO2 ~ logTemp + zlogBodyMass + blockID + (1+ logTemp | ID) + (1+ logTemp| seriesID) where logVCO2 log-transformed VCO2, logTemp is log temperature in degrees Celsius, zlogBodyMass is log-transformed body mass that is then subsequently z-transformed (mean of 0 and sd of 1), blockID that refers which statistical block a given lizard was randomly assigned to. We fitted individual ID and series ID as random intercepts and logTemp as random slopes. We found that slope of metabolic rate and body mass did not differ between blocks (Estimate for

60

blockID = 0, 95% credible interval = -0.11 – 0.11) We therefore did not include blockID as a covariate in subsequent analyses. The carry-over effects of previous temperature environment We investigated the effect of previous temperature environment on a lizard’s metabolic rate at subsequent temperatures. We did this by creating a ‘previous temperature’ covariate by treating body temperature in the enclosure as the ‘previous temperature’ for the first measurement temperature and the first measurement temperature as the ‘previous temperature’ for the second measurement. This ‘previous temperature’ covariate was log-transformed, and we tested whether its inclusion in our final model resulted in a better fit using information criterions (WAIC).

Table S1 Comparisons of wAIC values of a ‘brms’ model containing prior temperature (Model 1) and another model with its exclusion (modb.2). Model wAIC value SD Model.1 4211.30 150.23 Model. 2 4228.07 148.86 Model.1- Model. 2 -8.39 4.34

We fitted two models using ‘brms’, the first model had the following structure:

logVCO2 ~ Temp + zlogBodyMass + PriorTemp + (1+ Temp | ID) + (1+ Temp| seriesID) where PriorTemp is previous temperature. The second model had the same structure except that PriorTemp was excluded. WAIC values were lower in the model containing PriorTemp, suggesting that it is better supported compared to a model with PriorTemp excluded (Table S1). Based on these results, we included PriorTemp all subsequent repeatability analyses.

Checking for collinearity in predictors

We checked whether any of our predictors (Temp, PriorTemp and zlogBodyMass) were strongly collinear by using scatterplot matrices and calculating Pearson correlation coefficients (Fig.S2, Table S2). We found that PriorTemp and Temp was negatively correlated (r = -18). We calculated variance inflation factors (VIF, see ESM code) to assess whether including both variables would inflate the variance explained by the predictors.

61

We fitted two linear models to calculate VIFs, the first model had the following structure: Temp ~ logBodyMass + PriorTemp and the second model did not include PriorTemp. The VIFs for the first model was 1.06 vs. the VIF of 1.00 the second model. Following the recommendation of Zuur et a. (2010),VIF values less than 3 (a very conservative threshold) were deemed acceptable and therefore PriorTemp was included in all subsequent repeatability models.

incb_temp log.mass prior_temp 32

30 incb_temp

28

−0.01 −0.25

26

0.4

0.2 log.mass

0 0.0

−0.2

30 prior_temp

20

10

22.5 25.0 27.5 30.0 32.5 −0.2 0.0 0.2 0.4 10 20 30 Figure S2 Scatterplot matrix and correlation values between log-transformed temperature, log-transformed prior temperature

Table S2 Pairwise Pearson's correlations and the 95% confidence intervals between log-transformed temperature, log-transformed prior temperature and log- transformed-standardised body mass. Confidence intervals adjusted for multiple comparisons are also presented. Bold pairwise correlations are statistically significant from zero. T represents temperature. Variables r Lower Upper P-value T - Mass 0 -0.03 0.03 0.88

62

T - prior T -0.24 -0.27 -0.22 0 Mass - prior T 0 -0.03 0.03 0.99

Repeatability equations

All our repeatability estimates are adjusted repeatabilties (Nakagawa & Schielzeth, 2010). We calculated the repeatability of metabolic thermal plasticity from the random slope using the following equation from Araya-Ajoy, Mathot & Dingemanse (2015). Equation 1:

� � = (� + �:) where Vind1 is the individual slope and Vind:session1 series slope. We then investigated how the repeatability of average SMR (i.e. intercepts) changed with temperature. The multivariate response model treats RMR at each temperature as distinct ‘characters’ which allows us to calculate temperature-specific repeatabilities using the among-individual and the among-sampling session-within-individual variance-covariance matrices (see main text). Equation 2:

�, � = (�, + �,) where Vind0,T represents the individual intercept; Ve0,T is the residual variance component at a given temperature. Temperature-specific repeatability are presented in Table S3.

Table S3 Adjusted repeatability of log metabolic rate, among individual, measurement error and within individual variance components and their 95% credible intervals, across six measurement temperature (T) estimated using a multivariate response model. N = 42, nobs = 4952. All estimates are significantly different from zero. Repeatability Among individual variance T (ºC) Estimate Lower Upper T (ºC) Estimate Lower Upper 22 0.38 0.27 0.48 22 0.16 0.09 0.23 24 0.41 0.31 0.50 24 0.17 0.11 0.24 26 0.41 0.33 0.49 26 0.15 0.11 0.21 63

28 0.36 0.28 0.44 28 0.15 0.11 0.21 30 0.43 0.36 0.50 30 0.16 0.12 0.21 32 0.33 0.26 0.40 32 0.17 0.12 0.23 Measurement Error Residuals (Within individual variance) 22 0.39 0.36 0.44 22 0.25 0.23 0.27 24 0.35 0.32 0.40 24 0.25 0.23 0.27 26 0.32 0.29 0.36 26 0.22 0.20 0.24 28 0.33 0.29 0.36 28 0.27 0.25 0.29 30 0.30 0.27 0.33 30 0.21 0.20 0.23 32 0.26 0.22 0.31 32 0.35 0.32 0.38

Did body mass change over time?

Exploratory plot below body mass decreasing with time. It is important to note that the amount of weight loss was all less than 10% of an individual’s original body mass. Animals were closely monitored throughout the study to ensure they were active and healthy. Despite a reduction in mass and somewhat for metabolic rate over the study, individuals were still consistently different from one and another.

1.6

1.4

1.2

1.0 Mean Body Mass (g)

0.8

1 10 Sampling session

Figure S3 The mean log body mass (right) at the first and last sampling sessions. Each coloured point represent an individual, each line represents the

64

change in mass (n = 42). One individual lost its tail and therefore its mass changed drastically.

Among- and within-individual mass-scaling exponents from a multi-level model vs. a ‘typical’ metabolic scaling model

Mass-scaling exponents describe the relationship between metabolic rate and body mass across a sample of individuals. We partitioned out among- individual and within-individual sources of variation in body mass to see whether this affected the estimation of mass-scaling exponents. We compared the within- and among individual scaling exponents with a model that represents the typical statistical analysis of a metabolic scaling study model that does not account for the hierarchal structure in the data. Given that ‘typical’ studies rarely take repeated measures of metabolic rate and body mass, we randomly selected one measurement of VCO2 and body mass each individual across all sampling sessions for each measurement temperature (n = 42). We fitted a linear model with logVCO2 as a response and included an interaction term between log body mass with temperature,

logVCO2 ~ Temp * logMass We repeated this process ten times and found that mass-scaling exponents are estimated with a large degree of error (Table S4 and Fig. S4). In most cases, exponents were within the upper bounds of the within-individual exponent and the lower bounds of the among-individual exponent (Fig. S3). This suggests that when hierarchal structure in the data is not properly modelled, the among- individual or sample population mass-scaling exponent is a composite of within- and among-individual effects and are therefore less precise.

65

3.5

3.0

2.5

2.0

1.5

1.0

0.5 scaling exponent estimate Mass − scaling exponent

0.0

Among ID Typical −0.5 Within ID

22 24 26 28 30 32 Temperature Figure S4 Comparison of mass-scaling exponents across six measurement temperatures estimated by two different models. The first model properly accounts for among (black filled triangles) and within- individual (black filled squares) effects (nobs = 3933). The second model that does not account for the hierarchical nature of the data structure at the individual level (i.e. takes a single measure of each individual and averages across individuals, n = 42). Error bars from model one represents 95% credible intervals, error bars from model 2 are 95% confidence intervals. For details on how we obtained within- individual and among-individual scaling exponents, please see main text.

Table S4 Comparisons of multilevel mass-scaling exponents and their 95% credible intervals across six measurement temperatures (T). Mass-scaling exponents at among-individual, within-individual level. We also included estimates from a typical linear model where multi-level variation in not partitioned out. Bolded estimates are significantly different from zero Typical mass-scaling Among-individual Within-individual model

T (ºC) Est. Lower Upper Est. Lower Upper Est. Lower Upper 66

22 1.07 0.40 1.76 1.91 1.09 2.79 1.11 -0.63 2.85 24 0.78 0.07 1.40 2.41 1.54 3.26 1.15 -0.07 2.36 26 0.76 0.04 1.39 2.12 1.17 2.93 0.89 -0.45 2.22 28 1.12 0.46 1.76 1.30 0.44 2.13 2.75 1.55 3.95 30 0.98 0.33 1.60 1.41 0.57 2.26 0.82 -0.20 1.84 32 1.07 0.41 1.73 1.42 0.59 2.18 1.99 0.88 3.09

Table S5 Cross-temperature (T) correlations of log metabolic rate (R2)and their 95% credible intervals, estimated using a multi-response model at the among-individual level. N = 42, nobs = 802. Bolded estimates are significantly different from zero. Among-individual

Pairwise comparison Estimate Lower Upper among T (R2) 22ºC – 24ºC 0.59 0.16 0.89 22ºC – 26ºC 0.63 0.24 0.91 22ºC – 28ºC 0.49 0.05 0.85 22ºC – 30ºC 0.51 0.09 0.83 22ºC – 32ºC 0.46 0.02 0.82 24ºC – 26ºC 0.71 0.36 0.93 24ºC – 28 0.67 0.31 0.91 24ºC – 30 0.64 0.27 0.89 24ºC – 32 0.51 0.12 0.83 26ºC – 28 0.79 0.50 0.95 26ºC – 30 0.79 0.51 0.95 26ºC – 32 0.69 0.33 0.92 28ºC – 30 0.84 0.62 0.97 28ºC – 32 0.80 0.52 0.96 30ºC – 32 0.75 0.48 0.94

67

a) b) c)

Figure S5 Posterior mean of a) among individual variance, b) repeatability and ̇ c) residual individual variance components of log metabolic rate (� mL min- 1) at six measurement temperatures estimated over four-month period across n = 42 individuals. Error bars represent 95% credible intervals. Estimates are estimated from a model that did not taken into account of measurement error. As such, measurement error is attributed to the residual variance and affecting the estimates for repeatability.

68

Supplementary Materials References

Araya-Ajoy, Y. G., Mathot, K. J., & Dingemanse, N. J. (2015). An approach to estimate short-term, long-term and reaction norm repeatability. Journal of Animal Ecology, 6(12), 1462–1473. https://doi.org/10.1111/2041- 210X.12430 Bürkner, P. C. (2017). brms: An R package for Bayesian multilevel models using Stan. Journal of Statistical Software, 80(1). https://doi.org/10.18637/jss.v080.i01 Hadfield, J. D. (2010). MCMC methods for multi-response generalized linear mixed models: The MCMCglmm R package. Journal of Statistical Software, 33(2), 1–22. Nakagawa, S., & Schielzeth, H. (2010). Repeatability for Gaussian and non- Gaussian data: A practical guide for biologists. Biological Reviews, 85(4), 935–956. https://doi.org/10.1111/j.1469-185X.2010.00141.x Zuur, A. F., Ieno, E. N., & Elphick, C. S. (2010). A protocol for data exploration to avoid common statistical problems. Methods in Ecology, 1(1), 3–14. https://doi.org/10.1111/j.2041-210X.2009.00001.x

69

CHAPTER 3

Impact of developmental temperatures on the repeatability of thermal plasticity in metabolic rate

Fonti Kar1, Shinichi Nakagawa1,2, Daniel W.A. Noble3

1School of Biological Earth and Environmental Sciences, Ecology and Evolution Research Centre, University of New South Wales, Sydney, NSW, Australia 2Diabetes and Metabolism Division, Garvan Institute of Medical Research, 384 Victoria Street, Darlinghurst, Sydney, NSW 2010, Australia 3Division of Ecology and Evolution, Research School of Biology, The Australian National University, Canberra, ACT, Australia

All authors conceived the ideas and designed the study. FK and DN collected and analysed the data, FK wrote the first draft, FK, DN and SN edited the manuscript. All authors declare no conflict of interest

70

Abstract

Phenotypic plasticity is an important mechanism that allows populations to adjust to changing environments. Plastic responses induced by early life experiences can have lasting impacts on how individuals respond to environmental variation later in life (i.e., reversible plasticity). Developmental environments can also influence repeatability of plastic responses thereby altering the capacity for reaction norms to respond to selection. Here, we compared metabolic thermal reaction norms in lizards (Lampropholis delicata) that were incubated at two developmental temperatures (ncold = 26, nhot = 25). We repeatedly measured individual reaction norms across six acute temperatures 10 times over ~3.5 months (nobs = 3,818) to estimate the repeatability of average metabolic rate (intercept) and thermal plasticity (slope). The intercept and the slope of the population-level thermal reaction norm did not change with developmental temperatures. Repeatability of average metabolic rate was on average, 10% lower in hot incubated lizards and was stable across acute temperatures. The slope of the reaction norm was moderately repeatable (R = 0.44, 95% CI = 0.035 – 0.93) suggesting that individuals exhibited consistent changes in metabolism in response to acute temperature variation; however, reaction norm repeatability did not depend on early developmental temperature. Our work implies that thermal plasticity has the capacity to evolve, despite there being less consistent variation in metabolic rate under hot environments. This will be increasingly more important for terrestrial ectotherms living in changing climate.

71

Introduction

A substantial amount of variation in an individual’s phenotype is determined by formative processes experienced throughout embryonic development. Environmental perturbations during this critical period can have persistent effects on an individual’s physiology, morphology, behaviour and life history (Eyck et al., 2019; Noble et al., 2018; O’Dea et al., 2019). Developmental shifts in phenotypes may be adaptative if it allows organisms to better cope in similar environments later in life (Beldade et al., 2011). However, environment- phenotype mismatches can occur when developmental cues fail to predict later life conditions (Auld et al., 2010; Bonamour et al., 2019). A multitude of traits throughout an animal’s life are labile; reversibly responding to environmental change. Reversible plasticity in phenotypic traits allows individuals to adjust to acute changes in their surroundings (Piersma & Drent, 2003), and can broadly be classified into two categories, acclimation and phenotypic flexibility (Havird et al., 2020; Piersma & Drent, 2003). Acclimation is generally a slower form of reversible plasticity that involves remodelling of physiological systems from chronic exposure to a particular environment (Seebacher, 2005). Phenotypic flexibility, in contrast, describes short-term changes in traits that are induced by acute environmental exposure, such as changes in metabolic rate in response to acute temperature (Piersma & Drent, 2003; Piersma & Lindström, 1997).

Reversible plasticity may be able to alleviate the costs associated with phenotype mismatches induced by early life environments (Angilletta Jr et al., 2003; Ghalambor et al., 2007). When environments shift predictably, flexibility in the phenotype would be advantageous because individuals can compensate for the effects of prevailing conditions to avoid discrepancies between the environment and the phenotype (Botero et al., 2015). However, reversible plasticity can change depending on early environmental conditions and might alter phenotypic responses to environmental variation (Beaman et al. 2016). The interaction between early- and late life plasticity has been supported by a few studies that show developmental differences in plasticity for a variety of traits including mitochondrial function (Shama et al., 2014), metabolic rate (Seebacher et al., 2014) and locomotor performance (Kazerouni et al., 2016). However, these studies solely focus on the developmental effects on acclimation, whereas the

72

influence on phenotypic flexibility and variability of plastic responses is poorly known.

It has long been recognised that individuals vary in their plasticity, with some responding more flexibly than others (Dingemanse & Wolf, 2013; Nussey et al., 2007). Consistent variation among individuals may be heritable, but importantly, provides the phenotypic substrate for selective forces to act upon (Araya-Ajoy & Dingemanse, 2017; Nussey et al., 2007). Developmental environments, however, can influence variation available for selection (Sultan & Stearns, 2005). For example, zebra finches (Taeniopygia guttata) that experience nutritional stress as nestlings weigh less and have reduced growth rates contributing to increases in the repeatability of metabolism and behavioural traits (Careau, Buttemer, et al., 2014). Consistent among individual variation in plasticity has also been reported in other labile traits including aggressiveness in great tits (Parsus major) (Araya-Ajoy & Dingemanse, 2017), explorative behaviour in chickadees (Thompson et al., 2018) and metabolic rate in amphipods (Réveillon et al., 2019). Whether developmental environments affect consistent variation in plasticity per se is still poorly understood. Identifying the factors that impact repeatability is necessary for understanding the evolution of plasticity in changing environments.

Energy metabolism is a key fitness related trait that is both consistently different among individuals and highly labile within individuals (Nespolo & Franco, 2007; Norin & Metcalfe, 2019). All organisms require energy for growth, maintenance and reproduction (Careau, Killen, et al., 2014). Numerous studies have investigated the influence of various developmental environments, such as temperature (Gangloff et al., 2015; Noble et al., 2018), ultra-violet (UV) exposure (Kazerouni et al., 2016), and dietary restriction (Careau, Buttemer, et al., 2014) on metabolic rate, however, the impacts on plasticity of metabolic rate is not well established (but see Seebacher et al., 2014). Developmental cues could influence metabolic plasticity, possibly through modifications in metabolic enzymes or cellular membrane structure that influence their function in different environments (Angilletta Jr, 2016). Such changes imply that tolerance to environmental perturbations may be determined by the developmental environment a given cohort experiences. Furthermore, if

73

repeatability of metabolic plasticity is also affected, then the capacity to respond to selection might also be specific to early life conditions. Understanding how early life environments shape metabolic plasticity will be important for ectotherms where metabolic rate is closely intertwined with prevailing environmental conditions.

Here we employed a ‘reaction norm approach’ (sensu Via et al., 1995) to examine the impact of developmental temperature on plasticity of metabolic rate in an oviparous skink (Lampropholis delicata). Specifically, we were interested in testing whether developmental temperature affects the shape and repeatability of metabolic thermal reaction norms. Over 3.5 months, we repeatedly measured routine metabolic rate at six temperatures for lizards (nobs

= 3,818) that hatched from two incubation treatments (total individuals: nhot =

25, ncold = 26) to address the following key questions: (1) How does developmental temperature change the intercept and slope of the thermal reaction norm?; (2) How does the repeatability of metabolic plasticity (i.e. slope of the reaction norm) change with developmental temperature? (3) Do developmental temperature treatments differ in their repeatability of metabolic rate (intercept) at each acute temperature (i.e. temperature-specific repeatability)? Our experimental approach will provide important insights into how development environments mediate the capacity for ectotherms to respond to thermal variation during early stages of life.

74

Materials and Methods

Lizard collection and Husbandry

We established a breeding colony of adult L. delicata (nfemales = 144, nmales = 50) using wild individuals collected across three sites throughout the Sydney region between 28 August and 8 September 2015 (UNSW Kensington Campus: - 33.92, 151.24; Sydney Park: -33.91, 151.18, Macquarie Park: -33.77, 151.10). Three females were housed with a single male in opaque plastic enclosures measuring 35cm × 25cm × 15cm (L × W × H). Enclosures were kept under UV lights on a 12 hours light : 12 hours dark cycle in a temperature-controlled room set to 24ºC. Lizards had access to a heat lamp that elevated temperatures on one side of the enclosure to 32 ºC. Each enclosure was lined with newspaper and lizards had constant access to water and tree bark was used as refuge. Adult lizards were fed medium sized crickets (Acheta domestica) ad libitum dusted with calcium powder and multi-vitamin every two days. From the beginning of the egg laying season (October of each year), we replaced the newspaper lining with garden potting mix and placed an opaque plastic box (12 cm × 17.5 cm × 4.3 cm) containing moistened vermiculite in each enclosure for females to oviposit their eggs. During this time, enclosures and vermiculite boxes were sprayed gently with water every other day to maintain a relatively humid environment. From October to November, vermiculite boxes were checked every day for eggs. Animal collection was approved by the New South Wales National Parks and Wildlife Service (SL101549) and all procedures were approved by the Macquarie University Ethics committee (ARA 2015/015) and University of New South Wales Animal Care and Ethics committee (ACEC 15/51A).

Developmental Temperature Manipulations

Eggs were collected over October 2017 – March 2018. When eggs were discovered, they were weighed using a digital scale to the nearest 0.01g (Ohaus Scout SKX123). We also measured egg length (distance between the furthest points along the longest axis of the egg) and egg width (distance between the widest points along the axis perpendicular to the longest axis of the egg) using digital callipers to the nearest 0.01 mm. Following measurements, each egg was

75

placed in a plastic cup (80 ml) containing 3 g of vermiculite and 4 g of water and covered using cling wrap which was secured by an elastic band. Eggs from each clutch were pseudo-randomly assigned to one of two fluctuating incubation temperature treatments. We used two incubators to precisely control the temperature of eggs (LabWit, ZXSD-R1090). The ‘hot’ treatment was exposed to a mean temperature of 29ºC whereas the ‘cold’ treatment was exposed to a mean temperature of 23ºC. Both incubators fluctuated +/- 3ºC the mean temperature over a 24-hour period. These treatments represent the temperature extremes of natural nest sites of L. delicata (Cheetham et al., 2011). Egg cups were rotated within each incubator weekly to avoid uneven heat circulation within incubators. Incubators were also checked daily for hatchlings. On average, the incubation duration for the ‘hot’ treatment was 30 days (SD = 1.40, range = 27 - 33) days and 47.7days (SD = 5.90, range = 25 - 53) for the ‘cold’ treatment.

Planned Missing Data and Metabolic Rate at Different Temperatures

Metabolic measurements commenced in April 2018 and continued until August 2018. At the beginning of measurements, hatchlings were on average 88.68 days old (SD = 23.75, range = 26 - 131). We used closed-system respirometry instead of flow-through respirometry because it was more logistically feasible given our need to measure a large number of hatchlings at a range of temperatures. We quantified routine metabolic rate (hereafter referred to as metabolic rate [MR]) as our measurements likely included the energetic costs of random movements (Withers 1992; Mathot & Dingemanse 2015). MR was measured as

̇ -1 the volume of CO2 production per unit time (� mL min ) as CO2 production is less susceptible to fluctuations in water vapour and more feasible to detect in smaller organisms (Lighton, 2008; Tomlinson et al., 2018). Nonetheless, CO2

-16 production was strongly correlated with O2 consumption (r =0.81, p < 2.2e ) with RQ values averaged 0.77 (SD = 0.41). Due to logistical constraints, lizards were randomly assigned to one of two blocks for MR measurements (block 1: n =26, block 2: n = 25). We sampled lizards once a week for two-weeks consecutively and then allowed them to rest for one week before the next week of measurements. Each week of measurements was considered a sampling session (ten sampling sessions in total over the course of 14 weeks). We used

76

the same incubators described above to precisely control the temperature at which MR measurements were taken (+/- 1ºC).

Metabolic rate was measured at 24ºC, 26ºC, 28ºC, 30ºC, 32ºC and 34ºC in a randomised order. However, at each sampling session we intentionally missed measurements at two randomly selected temperatures using a planned missing data design (Nakagawa, 2015; Noble & Nakagawa, 2018). Missing data was imputed during analysis (see Statistical analysis). At ~06:00, lizards were gently encouraged into an opaque respiratory chamber and then weighed. After which, chambers were placed inside preheated incubators set at the randomised temperature for 30 minutes to allow body temperatures to equilibrate. The lids of the chambers were left ajar during this time to minimise

CO2 build up. After 30 minutes, each chamber was flushed with fresh air and sealed. A 3 mL ‘control/baseline’ air sample was immediately taken via a two- way valve to account for any residual CO2 that was not flushed from the chambers. The chambers were left in the incubator at the set temperature for lizards to respire for 90 minutes. After this time, two replicate air samples (3 mL) were taken from each chamber in order to estimate the change in CO2. Two samples were taken so we could explicitly estimate measurement error (see Statistical analysis, Ponzi et al., 2018). Chambers were then reopened and flushed with fresh air before being placed back into the incubator for the second measurement temperature (2 temperatures / day) following the same procedure approximately two hours later.

All air samples were injected into the inlet line of a Sables System FMS (Las

-1 ̇ ̇ Vegas NV, USA) with the flow rate set to 200 mL min to measure � and �. Water vapour was scrubbed from the inlet air with Drierite. Output peaks were processed using the R package ‘metabR’

(https://github.com/daniel1noble/metabR). The rate of CO2 produced by an individual was calculated following (Lighton, 2008):

%�� × (� − �) �̇ mL ��� = � where %CO2 is the maximum percentage of CO2 in air sample above baseline, which was corrected by subtracting any ‘residual’ CO2 from the initial flush from the larger of the two air samples; Vchamber is the volume of the chamber (70

77

mL); Vlizard is the volume of the lizard, assuming that the mass of the lizard is the same as its volume, and t is the duration of time in minutes after where the chamber has been sealed and the first air sample was taken (90 minutes).

Statistical Analyses

We fitted Bayesian linear mixed effect models in R (Core Team, 2013) using the package ‘brms’ (Bürkner, 2017). Metabolic rate was log transformed and body mass, age and temperature were z-transformed so parameter estimates of main effects and interaction terms were more interpretable (Schielzeth, 2010). Our planned missing data design resulted in random missingness across temperatures (36% missingness in MR and body mass) The package ‘brms’ is capable of performing model-based data imputation. As such, we performed imputation during model fitting in all of analyses. Model-based imputation not only retains the hierarchical structure of the dataset but also increases statistical power (P. Bürkner, personal communication 25 October 2020, Nakagawa, 2015). Sensitivity analyses suggest that models with imputed data resulted in similar conclusions to complete case analyses. However, we present results from the imputation analysis in the main text as parameter estimates were more precise (See ESM). For all models we used default priors and ran four Markov Chain Monte Carlo (MCMC) chains; taking 800 samples from the posterior distribution after discarding the first 1500 iterations. This gave a total of 3200 samples from the posterior distribution across all chains. We ensured chains were mixing by inspecting trace plots and checked that scale reduction factors were less than 1.01, suggesting all chains had converged. Throughout we report posterior means and 95% credible intervals for all parameters. All data and code to reproduce our results are provided (see Data accessibility).

To test whether developmental temperatures changed the shape of reaction norms, we fitted a full model with MR as the response and temperature, treatment and an interaction between treatment and temperature as predictors. The model also included a random intercept for lizard identity and sampling session. We wanted to account for measurement error in all our models as it may conflate parameter estimates (Ponzi et al., 2018). Using the two replicate air samples, we estimated measurement error variance by including a nested random effect of lizard identity, sampling session and temperature in all our

78

models (e.g. ID001_session1_temp24). This nested random effect (hereafter referred to as measurement error) estimates the variance attributed to differences among replicates. While we show measurement error can vary by temperature (Chapter 2), here we assumed that measurement error was constant across temperatures by fitting it as a random intercept as estimating a random slope for resulted in model convergence issues. Heterogeneous variance across temperatures can also influence parameter estimates (Careau, Buttemer, et al., 2014). However, WAIC values indicated that heterogeneous residual variance was not supported by our data, therefore homogenous variance was used in all models (Table S1). Acclimation can influence metabolic plasticity and its effects can take place throughout the course of our study. Unfortunately, it was not possible to measure MR at hatching. However, we still tested whether there were treatment differences in thermal reaction norms in the first sampling session (~2.5 months of age) where acclimation effects were likely to have the weakest effect.

We estimated adjusted repeatability of the reaction norm slope (Rslope) in each developmental temperature treatment by fitting separate models for each treatment group. MR was fitted as the response and temperature, body mass and age as predictors. We included lizard identity, measurement error and a nested random effect of individual identity and sampling session (hereafter referred to as series, Araya-Ajoy et al., 2015). Lizard identity estimates among individual variance, whereas series partitions variance within individual across all sampling sessions. A random temperature slope was estimated for lizard identity and series which allowed us to calculate slope repeatability. The repeatability of the slope is calculated as the proportion of total variance in slopes explained by among individual differences (Araya-Ajoy et al., 2015):

�, � = (�, + �,) where: �, is the among-individual variance in the temperature slope term and the �, is the within-individual variance in the temperature slope.

We estimated adjusted repeatability of average metabolic rate (i.e. intercept of the reaction norm) at each acute temperature by fitting separate models for each treatment group. Similar to above, MR was included as the response and

79

temperature, body mass and age as predictors. We included lizard identity, sampling session and measurement error as random intercepts and temperature as a random slope for lizard identity. We calculated among individual variance in metabolic rate at each temperature It following Schielzeth and Nakagawa (2020): � = � + (� . �) + (2�. ���,) where � is the among individual variance in intercepts, � is the specific temperature at which repeatability is calculated for, � is the among individual variance in slope and ���, is the covariance between the intercept and slope at the among individual level. Temperature specific repeatability (�) is then calculated as follows:

� � = (� + � + �) where: � is the variance due to sampling session and � is residual variance.

We also wanted to estimate overall repeatability of average metabolic rate across all acute temperatures. We fitted the same model as above for each treatment, but we omitted the random temperature slope for lizard identity, this estimates an average among individual variance across all acute temperatures. Similarly, we calculated repeatability as per the equation above but using just the single estimate of among individual variance.

In order to test for differences in repeatability among the two developmental temperatures, we calculated contrasts by subtracting the posterior distributions of repeatability estimates of the cold developmental treatment from the hot (Hot – Cold). To test whether the magnitude of differences among treatments were significant by chance, we calculated probabilities of direction (pd) using the package ‘bayestestR’ (Makowski, Ben-Shachar, & Lüdecke, 2019). The probability of direction is calculated relative to the posterior median and ranges from 50 -100%. The value of pd describes whether an effect is either positive or negative as it is always relative to the sign of the median (Makowski, Ben- Shachar, Chen, et al., 2019). If the median is positive, then pd describes the proportion of the posterior distribution that is also positive (Makowski, Ben-

80

Shachar, Chen, et al., 2019). A pd value of 95% can be interpreted as the effect is positive with a probability of 95%.

81

Results

−5.5 1 − 1 -

g −6.0 1 - min mL min 2 2 VCO −6.5 log log VCO

−7.0

24 26 28 30 32 34 Temperature

-1 -1 Figure 1 Predicted thermal reaction norm of metabolic rate (VCO2 min g ) for the ‘cold’ developmental temperature group (blue line, nlizards = 26) and the

‘hot’ developmental temperature group (red line, nlizards = 25) Points are raw data and are coloured according to treatment groups nobs = 3818. Dashed lines represent the upper and lower bounds of 95% credible intervals.

We found no evidence to suggest that metabolic rate or its response to acute temperature was influenced by early developmental temperature (Fig. 1, Table 1, Table S2). Congruently, there were no treatment differences in thermal reaction norms at the first sampling session when acclimation effects are likely to have the least effect (see ESM). We therefore refitted the model with just the main effects (Table S3-4). Across all models, temperature and body mass had positive effects on metabolic rate (Table 1, Table S3-4). Nonetheless, reaction norm slopes were significantly repeatable, but repeatability of slopes (Rslope) did not depend on developmental temperature treatments (Hot: Rslope = 0.42, 95% CI:

0.04 – 0.91; Cold: Rslope = 0.46, 95% CI: 0.03 – 0.95; pd = 53.5%, Fig. 2, Table S6-9). A pd value of 53.5% indicates that there is roughly equal probability that the

82

difference in Rslope is positive or negative, indicating little difference among treatment groups.

A) 1 5 10 1 1 −

- −5.5 g 1 - −6.0 min mL min 2 2 −6.5 VCO log log log VCO −7.0

24 26 28 30 32 34 24 26 28 30 32 34 24 26 28 30 32 34

B) 1 5 10 1 −

1 −5.5 - g 1 - −6.0 min mL min 2 2 −6.5 VCO

log VCO −7.0 log log

24 26 28 30 32 34 24 26 28 30 32 34 24 26 28 30 32 34 Temperature

Figure 2 Thermal reaction norms of mass-adjusted metabolic rate for lizards reared at A) ‘hot’ developmental temperatures (top, red lines, nlizards = 25) and

B) ‘cold’ developmental temperatures (bottom, blue lines, nlizards = 26) at session number one, five and ten. Each uniquely coloured line represents an individual reaction norm. There is a random subset of 10 individuals from each treatment.

Table 1 Model coefficients of the full model testing whether developmental temperature affects the elevation (intercept) and slope of the thermal reaction norm of metabolic rate. This model used an imputed dataset of nobs = 6,000, 36% of observations were imputed. The intercept is the cold developmental temperature. MR was log transformed and mass, age and temperature were z-transformed. Bolded estimates are significantly different from zero. Lower and upper bound of estimates represent 95% credible intervals. COV represents covariance. Main effects model is presented in Table S3

Parameter Estimate Lower Upper

Fixed effects

83

Intercept MR -6.292 -6.372 -6.218

Treatment 29 -0.003 -0.062 0.058

Acute Temperature 0.262 0.246 0.278

Treatment 29 × Acute -0.016 -0.039 0.007 Temperature Age -0.035 -0.079 0.006

Mass 0.128 0.105 0.151

Random Effects

Lizard Identity

Intercept 0.009 0.006 0.015

Temperature Slope 9.53e-5 1.54e-7 0.000479

COVIntercept – Slope -0.00018 -0.00122 0.000599

Sampling Session

Intercept 0.01 0.003 0.026

Measurement Error

Intercept 0.044 0.04 0.049

Residual 0.041 0.038 0.043

Overall, temperature-specific repeatability was relatively low, with the cold developmental treatment tending to have higher repeatability estimates compared to the hot developmental treatment (Fig. 3, Fig S1, Table 2). Irrespective of acute temperature, repeatability of average metabolic rate was on average 10% higher in cold incubated lizards (pd = 95.7%, Fig. 3B, C). There was a 95.7% probability that the difference in overall repeatability was negative, indicating that lizards from the cold treatment are more likely to have higher repeatability. Higher repeatability in the cold treatment was associated with significant among individual and residual variance (Fig. S2).

84

A) 0.4 B)

0.4

0.3 0.2

0.0 0.2 Cold Hot C) 96.5% Repeatability 24 96.25% 26 95.95% 0.1 28 94.25% 30 91% 32 85.95% 34 95.7% 0.0 Overall 24 26 28 30 32 34 −0.4 −0.2 0.0 0.2 Difference in Repeatability Temperature (Hot − Cold) Figure 3 (A) Temperature-specific adjusted repeatability for average metabolic rate for the ‘cold’ developmental temperature group (blue, nlizards = 26) and the

‘hot’ developmental temperature group (red, nlizards = 25). Error bars represent 95% credible intervals. (B) Violin and boxplot showing the posterior distribution of overall adjusted repeatability of each treatment group irrespective of acute temperature. (C) Posterior distribution of the difference in repeatability (Hot – Cold) overall and at each acute temperature. Point represents the median; thicker lines represents the interquartile range and thin lines represent the 95% credible intervals. The probability of direction is presented on each distribution and describes the probability that the difference in repeatability is either positive or negative. Grey regions of the distribution represent negative estimates indicating repeatability was greater in the cold treatment, whereas black regions represent positive estimates which indicates that repeatability was greater in the hot treatment. All values were calculated from an imputation model. Contrasts are presented in Table S5.

85

Discussion

Contrary to our expectations, early developmental temperature did not change the intercept or slope of the population reaction norm. Thermal plasticity of metabolic rate (slope) was significantly repeatable, but its repeatability was also unaffected by developmental temperature. In line with our prediction, temperature-specific repeatability of metabolic rate (i.e., intercept) was lower among lizards that were reared in hot developmental temperatures. Our results suggest that, while individuals displayed consistent variation in their plasticity (Individual x Environment), how metabolic rate responds to acute temperature variation later in life was robust to thermal extremes of natural nest sites. Developmental temperatures did not have an impact on average metabolic rate but rather it changed the amount of consistent individual variation in average metabolic rate. Below we discuss the implications of our results for the evolution of thermal reaction norms.

Thermal reaction norms of metabolic rate are robust to developmental temperature

Developmental environments that affect later life plasticity may alter how populations respond to environmental fluctuations (Beaman et al. 2016). Epigenetic modifications during development that influence the physiological system are likely responsible for shaping plastic responses in complex ways (Hu & Barrett, 2017; McCaw et al., 2020). However, our results suggest instead that thermal reaction norms for metabolic rate were robust to changes in incubation temperature. Results have been mixed among the few studies that have investigated the effects of pre- and post-hatching temperature on the plasticity of metabolic rate (Table 1, Beaman et al., 2016). For example, wild caught mosquitofish (Gambusia holbrooki) developing under more variable spring conditions exhibited steeper thermal reaction norms for metabolic scope compared to fish born in summer (Seebacher et al., 2014). In contrast, incubation temperature did not affect plasticity in metabolic rate of striped marsh frog tadpoles (Seebacher & Grigaltchik, 2014). Given that our lizards were reared in a common environment post hatching, the lack of difference we observed may be the result of reversible plasticity resulting from acclimation in metabolic rate to the laboratory conditions. It is possible that acclimation 86

capacities may have overwhelmed any developmental differences in thermal reaction norms. Acclimation of physiological function takes approximately 3-4 weeks to complete, so it is likely that acclimation had already taken place by the time we began the study when lizards were about ~2.5 months old (Seebacher, 2005). Nonetheless, it is clear that, regardless of whether acclimation homogenised possible effects, developmental environments have little long- term impacts on reaction norms. Future studies should employ cross factorial designs where post-hatch environments are deliberately matched and mismatched with early environmental conditions to disassociate acclimation effects (Kazerouni et al., 2016; Schnurr et al., 2014).

Stable thermal reaction norms of metabolic rate across both developmental temperatures has key evolutionary implications. Our results imply that population reaction norms may be robust to temperature variation within the thermal range of natural nests (Cheetham et al., 2011). Past thermal regimes encountered by predecessors may have canalized population responses so that they are less sensitive to fluctuations in developmental temperature (Liefting et al., 2009). Canalization may reduce the costs of phenotypic plasticity during development if environmental variation is predictable across generations (Aubret & Shine, 2010). In support of this, damselflies undergoing range expansions exhibit geographic variation in thermal reaction norms that align with past climatic conditions (Lancaster et al., 2015). Population comparisons across environmental gradients might reveal whether local adaptation shapes developmental plasticity of population reaction norms that lead to canalisation (Toftegaard et al., 2015). Developmental environments may play a stronger role in shaping population plastic responses in areas that experience greater thermal variability, such as those in temperate or high elevation regions (Bonamour et al., 2019). While our incubation treatments represent thermal extremes of natural nest sites, they may not have been severe enough to induce changes in the thermal reaction norms, particularly given that we used more realistic fluctuating nest temperatures. Developmental stress is thought to lead to the recruitment of heat shock proteins thereby changing reversible plasticity later in life (Beaman et al., 2016; Chevin & Hoffmann, 2017). Recent work has shown lizard embryos exposed to extreme heat produce higher levels of heat shock proteins and have greater thermal tolerance, however this subsequently

87

reduces thermal tolerance later in life (Gao et al., 2014). This implies there may be constraints in how phenotypic responses can be shaped by extreme developmental environments.

Developmental temperatures and repeatable thermal plasticity of metabolic rate

Repeatability of reaction norm slopes did not change with developmental temperature, but lizards reared in hot temperatures had reduced temperature- specific repeatability in metabolic rate (intercept). Variation in developmental time has important consequences on hatching condition and may contribute to differences in consistent variation in hatchling phenotypes. Developmental time exhibits a negative nonlinear relationship with temperature, such that development times are considerably shorter at hotter temperatures (Marshall et al., 2020; Noble et al., 2018). Consequently, eggs reared in warmer environments are expected to be more constrained in their developmental rates, thus hatching phenotypes are more likely to be less variable compared to eggs reared in cooler environments (Pettersen et al., 2019). Indeed, we found that incubation duration was short and less variable in the hot developmental treatment. Shortened development may restrict embryo yolk assimilation that is needed for growth (Oufiero & Angilletta, 2006; Storm & Angilletta, 2007). Elevated levels of proton leak at hot developmental temperatures leads to less efficient energy production and may explain why metabolic rate did not differ among treatments despite changes in repeatability (Chamberlin, 2004). Lower repeatability under hot nest temperatures may be problematic as global temperatures continue to rise (Botero et al., 2015). Provided that some of the repeatable differences in metabolic rate is heritable (Dohm, 2002; Falconer, 1952), our results suggest that the evolutionary potential of metabolic rate may be dampened for populations living in warming environments. However, populations may be able to evolve metabolic plasticity in order to persist under rising temperatures (Ghalambor et al., 2007).

We found that individuals consistently vary in how their metabolic rate changes with acute temperature. While several studies have reported significant among individual variation in thermal plasticity slopes (Briga &

88

Verhulst, 2017; Careau et al., 2014), its repeatability is rarely estimated as it requires a study design that allows partitioning of within and between individual variance of slopes (Araya-Ajoy et al., 2015). Our repeatability estimates for reaction norm slopes were consistent with another study of the same species (R = 0.23, Chapter 1). Similarly, moderate repeatability of thermal sensitivity of metabolic rate has also been observed in amphipods (R = 0.38) (Réveillon et al., 2019). Assuming that repeatable reaction norm slopes have a heritable basis (Driessen et al., 2007), our work implies that thermal plasticity can be selected upon and therefore evolve (Falconer, 1952; but see Dohm, 2002).

Consistent individual differences in metabolic rate were stable across acute temperatures. This result demonstrates that temperatures within the operable range of L. delicata maintains consistent individual differences in MR (Matthews et al., 2016). Repeatability in metabolic rate may be an important mechanism that promotes consistent variation in thermoregulation, behaviour and life history (Goulet et al., 2017; Réale et al., 2010; Sæther, 1987). Overall, our estimates for the repeatability of MR ranged from 0.09 – 0.22. Our results are in line with a meta-analysis that showed that repeatability decreases with time (White et al., 2013). The average repeatability of MR in ectotherms from studies that had a measurement interval that was equal or larger than our study (≥ 8.5 days) was R = 0.33 (SD = 0.21, n = 18). Interestingly, repeatability of average MR in wild caught adult L. delicata (R = 0.3 – 0.5, Chapter 2) was comparatively larger relative to this study. This is likely due to life stage differences in environmental effects that shape phenotypic variation. As individuals mature, their experiences in different microhabitats (diet, thermal preferences) can promote among-individual variation in traits (Kruuk & Hadfield, 2007). Such common (micro) environment effects could further increase repeatability and may explain differences between lab and wild studies (Auer et al., 2016).

Conclusion

The role of developmental temperature on phenotypic plasticity exhibited later in life is complex. At the population level, thermal plasticity of metabolic rate was robust to changes in temperature during embryonic development suggesting that thermal reaction norms may be canalised. In contrast, the

89

impact of developmental temperature manifested as a change repeatability in temperature-specific metabolic rate. This has key evolutionary implications as reduced among individual variation in hot temperatures may alter a population’s ability to respond to selection under warming climate. However, populations may still have the ability to evolve in their thermal reaction norms as individuals consistently varied in their thermal plasticity in metabolic rate. Elucidating the role of developmental environments on shaping plastic responses may require more stressful incubation conditions and cross-factorial experimental designs to disassociate the effects of acclimation from developmental plasticity.

Data accessibility

Datasets and code used to generate results of this study is accessible via Open Science Framework (https://bit.ly/38IzTsp)

Acknowledgements

We would like to thank Martin Whiting for the use of his facilities at Macquarie University. We are grateful for the assistance of numerous Lizard Lab members and interns with husbandry duties. Special thanks to Christine Wilson for her commitment to caring for our animals. We thank Timothee Bonnet for his advice on partitioning measurement error from our models.

90

References

Angilletta Jr, M. J. (2016). Thermal Adaptation: A Theoretical and Empirical Synthesis. Oxford University Press. https://doi.org/10.1093/acprof:oso/9780198570875.001.1 Angilletta Jr, M. J., Wilson, R. S., Navas, C. A., & James, R. S. (2003). Tradeoffs and the evolution of thermal reaction norms. Trends Ecology and Evolution, 18(5), 234–240. https://doi.org/10.1016/S0169-5347(03)00087-9 Araya-Ajoy, Y. G., & Dingemanse, N. J. (2017). Repeatability, heritability, and age-dependence of seasonal plasticity in aggressiveness in a wild passerine bird. Journal of Animal Ecology, 86(2), 227–238. https://doi.org/10.1111/1365-2656.12621 Araya-Ajoy, Y. G., Mathot, K. J., & Dingemanse, N. J. (2015). An approach to estimate short-term, long-term and reaction norm repeatability. Journal of Animal Ecology, 6(12), 1462–1473. https://doi.org/10.1111/2041- 210X.12430 Aubret, F., & Shine, R. (2010). Fitness costs may explain the post-colonisation erosion of phenotypic plasticity. Journal of Experimental Biology, 213(5), 735–739. https://doi.org/10.1242/jeb.040576 Auer, S. K., Bassar, R. D., Salin, K., & Metcalfe, N. B. (2016). Repeatability of metabolic rate is lower for animals living under field versus laboratory conditions. Journal of Experimental Biology, 219(5), 631–634. https://doi.org/10.1242/jeb.133678 Auld, J. R., Agrawal, A. A., & Relyea, R. A. (2010). Re-evaluating the costs and limits of adaptive phenotypic plasticity. Proceedings of the Royal Society B: Biological Sciences, 277(1681), 503–511. https://doi.org/10.1098/rspb.2009.1355 Beaman, J. E., White, C. R., & Seebacher, F. (2016). Evolution of Plasticity: Mechanistic Link between Development and Reversible Acclimation. Trends Ecol Evol, 31(3), 237–249. https://doi.org/10.1016/j.tree.2016.01.004 Beldade, P., Mateus, A. R. A., & Keller, R. A. (2011). Evolution and molecular mechanisms of adaptive developmental plasticity. Molecular Ecology, 20(7), 1347–1363. https://doi.org/10.1111/j.1365-294X.2011.05016.x

91

Bonamour, S., Chevin, L.-M., Charmantier, A., & Teplitsky, C. (2019). Phenotypic plasticity in response to climate change: The importance of cue variation. Philosophical Transactions of the Royal Society B: Biological Sciences, 374(1768), 20180178–12. https://doi.org/10.1098/rstb.2018.0178 Botero, C. A., Weissing, F. J., Wright, J., & Rubenstein, D. R. (2015). Evolutionary tipping points in the capacity to adapt to environmental change. Proceedings of the National Academy of Sciences, 112(1), 184–189. https://doi.org/10.1073/pnas.1408589111 Briga, M., & Verhulst, S. (2017). Individual variation in metabolic reaction norms over ambient temperature causes low correlation between basal and standard metabolic rate. Journal of Experimental Biology, 220(18), 3280 – 3289, jeb.160069. https://doi.org/10.1242/jeb.160069 Bürkner, P. C. (2017). brms: An R package for Bayesian multilevel models using Stan. Journal of Statistical Software, 80(1). https://doi.org/10.18637/jss.v080.i01 Careau, V., Buttemer, W. A., & Buchanan, K. L. (2014). Early-Developmental Stress, Repeatability, and Canalization in a Suite of Physiological and Behavioral Traits in Female Zebra Finches. Integrative and Comparative Biology, 54(4), 539–554. https://doi.org/10.1093/icb/icu095 Careau, V., Gifford, M. E., & Biro, P. A. (2014). Individual (co)variation in thermal reaction norms of standard and maximal metabolic rates in wild- caught slimy salamanders. Functional Ecology, 28(5), 1175–1186. https://doi.org/10.1111/1365-2435.12259 Careau, V., Killen, S. S., & Metcalfe, N. B. (2014). Adding Fuel To The “Fire Of Life”: Energy Budgets Across Levels Of Variation In Ectotherms And Endotherms. In Integrative Organismal Biology (1st ed., p. 17). Wiley- Blackwell. Chamberlin, M. E. (2004). Top-down control analysis of the effect of temperature on ectotherm oxidative phosphorylation. American Journal of Physiology-Regulatory, Integrative and Comparative Physiology, 287(4), R794–R800. https://doi.org/10.1152/ajpregu.00240.2004 Chapple, D. G., Miller, K. A., Chaplin, K., Barnett, L., Thompson, M. B., & Bray, R. D. (2014). Biology of the invasive delicate skink (Lampropholis delicata) on Lord Howe Island. Australian Journal of Zoology, 62(6), 498–506. https://doi.org/10.1071/ZO14098

92

Cheetham, E., Doody, J. S., Stewart, B., & Harlow, P. (2011). Embryonic mortality as a cost of communal nesting in the delicate skink. Journal of Zoology, 283(4), 234–242. https://doi.org/10.1111/j.1469- 7998.2010.00764.x Chevin, L.-M., & Hoffmann, A. A. (2017). Evolution of phenotypic plasticity in extreme environments. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences, 372(1723). https://doi.org/10.1098/rstb.2016.0138 Core Team, R. (2013). Team (2012). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Dingemanse, N. J., & Wolf, M. (2013). Between-individual differences in behavioural plasticity within populations: Causes and consequences. Animal Behaviour, 85(5), 1031–1039. https://doi.org/10.1016/j.anbehav.2012.12.032 Dohm, M. R. (2002). Repeatability estimates do not always set an upper limit to heritability. Functional Ecology, 16(2), 273–280. https://doi.org/10.1046/j.1365-2435.2002.00621.x Driessen, G., Ellers, J., & Van Straalen, N. M. (2007). Variation, selection and heritability of thermal reaction norms for juvenile growth in Orchesella cincta (Collembola: Entomobryidae). European Journal of Entomology, 104(1), 39–46. https://doi.org/10.14411/eje.2007.006 Eyck, H. J. F., Buchanan, K. L., Crino, O. L., & Jessop, T. S. (2019). Effects of developmental stress on animal phenotype and performance: A quantitative review. Biological Reviews, 94(3), 1143–1160. https://doi.org/10.1111/brv.12496 Falconer, D. S. (1952). The Problem of Environment and Selection. The American Naturalist, 86(830), 293–298. Gangloff, E. J., Vleck, D., & Bronikowski, A. M. (2015). Developmental and Immediate Thermal Environments Shape Energetic Trade-Offs, Growth Efficiency, and Metabolic Rate in Divergent Life-History Ecotypes of the Garter Snake Thamnophis elegans. Physiological and Biochemical Zoology, 88(5), 550–563. https://doi.org/10.1086/682239 Gao, J., Zhang, W., Dang, W., Mou, Y., Gao, Y., Sun, B.-J., & Du, W.-G. (2014). Heat shock protein expression enhances heat tolerance of reptile

93

embryos. Proceedings of the Royal Society B: Biological Sciences, 281(1791), 20141135. https://doi.org/10.1098/rspb.2014.1135 Ghalambor, C. K., McKay, J. K., Carroll, S. P., & Reznick, D. N. (2007). Adaptive versus non-adaptive phenotypic plasticity and the potential for contemporary adaptation in new environments. Functional Ecology, 21(3), 394–407. https://doi.org/10.1111/j.1365-2435.2007.01283.x Goulet, C. T., Thompson, M. B., Michelangeli, M., Wong, B. B. M., & Chapple, D. G. (2017). Thermal physiology: A new dimension of the pace-of-life syndrome. Journal of Animal Ecology, 86(5), 1269–1280. https://doi.org/10.1111/1365-2656.12718 Havird, J. C., Neuwald, J. L., Shah, A. A., Mauro, A., Marshall, C. A., & Ghalambor, C. K. (2020). Distinguishing between active plasticity due to thermal acclimation and passive plasticity due to Q10 effects: Why methodology matters. Functional Ecology, 34(5), 1015–1028. https://doi.org/10.1111/1365-2435.13534 Hu, J., & Barrett, R. D. H. (2017). Epigenetics in natural animal populations. Journal of Evolutionary Biology, 30(9), 1612–1632. https://doi.org/10.1111/jeb.13130 Kazerouni, E. G., Franklin, C. E., & Seebacher, F. (2016). UV-B radiation interacts with temperature to determine animal performance. Functional Ecology, 30(4), 584–595. https://doi.org/10.1111/1365-2435.12520 Kruuk, L. E. B., & Hadfield, J. D. (2007). How to separate genetic and environmental causes of similarity between relatives. Journal of Evolutionary Biology, 20(5), 1890–1903. https://doi.org/10.1111/j.1420- 9101.2007.01377.x Lancaster, L. T., Dudaniec, R. Y., Hansson, B., & Svensson, E. I. (2015). Latitudinal shift in thermal niche breadth results from thermal release during a climate-mediated range expansion. Journal of Biogeography, 42(10), 1953–1963. https://doi.org/10.1111/jbi.12553 Liefting, M., Hoffmann, A. A., & Ellers, J. (2009). Plasticity Versus Environmental Canalization: Population Differences in Thermal Responses Along a Latitudinal Gradient In Drosophila serrata. Evolution, 63(8), 1954–1963. https://doi.org/10.1111/j.1558-5646.2009.00683.x Lighton, J. R. B. (2008). Measuring Metabolic Rates. Oxford University Press.

94

Makowski, D., Ben-Shachar, M. S., Chen, S. H. A., & Lüdecke, D. (2019). Indices of Effect Existence and Significance in the Bayesian Framework. Frontiers in Psychology, 10. https://doi.org/10.3389/fpsyg.2019.02767 Makowski, D., Ben-Shachar, M. S., & Lüdecke, D. (2019). bayestestR: Describing Effects and their Uncertainty, Existence and Significance within the Bayesian Framework. Journal of Open Source Software, 4(40), 1541. https://doi.org/10.21105/joss.01541 Marshall, D. J., Pettersen, A. K., Bode, M., & White, C. R. (2020). Developmental cost theory predicts thermal environment and vulnerability to global warming. Nature Ecology & Evolution, 4(3), 406–411. https://doi.org/10.1038/s41559-020-1114-9 Matthews, G., Goulet, C. T., Delhey, K., & Chapple, D. G. (2016). The effect of skin reflectance on thermal traits in a small heliothermic ectotherm. Journal of Thermal Biology, 60, 109–124. https://doi.org/10.1016/j.jtherbio.2016.06.013 McCaw, B. A., Stevenson, T. J., & Lancaster, L. T. (2020). Epigenetic Responses to Temperature and Climate. Integrative and Comparative Biology, 60(6), 1– 12. https://doi.org/10.1093/icb/icaa049 Nakagawa, S. (2015). Missing data: Mechanisms, methods and messages. In G. A. Fox, S. Negrete-Yankelevich, & V. J. Sosa (Eds.), Ecological Statistics: Contemporary theory and application (pp. 81–105). Oxford University Press. Nespolo, R. F., & Franco, M. (2007). Whole-animal metabolic rate is a repeatable trait: A meta-analysis. The Journal of Experimental Biology, 210(21), 3877– 3878. https://doi.org/10.1242/jeb.013110 Noble, D. W. A., & Nakagawa, S. (2018). Planned missing data design: Stronger inferences, increased research efficiency and improved animal welfare in ecology and evolution. BioRxiv 247064, 8, 257–32. https://doi.org/10.1101/247064 Noble, D. W. A., Stenhouse, V., & Schwanz, L. E. (2018). Developmental temperatures and phenotypic plasticity in reptiles: A systematic review and meta-analysis. Biological Reviews, 93(1), 72–97. https://doi.org/10.1111/brv.12333 Norin, T., & Metcalfe, N. B. (2019). Ecological and evolutionary consequences of metabolic rate plasticity in response to environmental change.

95

Philosophical Transactions of the Royal Society B: Biological Sciences, 374(1768), 20180180–9. https://doi.org/10.1098/rstb.2018.0180 Nussey, D. H., Wilson, A. J., & Brommer, J. E. (2007). The evolutionary ecology of individual phenotypic plasticity in wild populations. Journal of Evolutionary Biology, 20(3), 831–844. https://doi.org/10.1111/j.1420- 9101.2007.01300.x O’Dea, R. E., Lagisz, M., Hendry, A. P., & Nakagawa, S. (2019). Developmental temperature affects phenotypic means and variability: A meta-analysis of fish data. Fish and Fisheries, 20(5), 1005–1022. https://doi.org/10.1111/faf.12394 Oufiero, C. E., & Angilletta, M. J. (2006). Convergent Evolution of Embryonic Growth and Development in the Eastern Fence Lizard (sceloporus Undulatus). Evolution, 60(5), 1066–1075. https://doi.org/10.1111/j.0014- 3820.2006.tb01183.x Pettersen, A. K., White, C. R., Bryson-Richardson, R. J., & Marshall, D. J. (2019). Linking life-history theory and metabolic theory explains the offspring size-temperature relationship. Ecology Letters, 22(3), 518–526. https://doi.org/10.1111/ele.13213 Piersma, T., & Drent, J. (2003). Phenotypic flexibility and the evolution of organismal design. Trends Ecology and Evolution, 18(5), 228–233. https://doi.org/10.1016/S0169-5347(03)00036-3 Piersma, T., & Lindström, Å. (1997). Rapid reversible changes in organ size as a component of adaptive behaviour. Trends Ecology and Evolution, 12(4), 134–138. https://doi.org/10.1016/S0169-5347(97)01003-3 Réale, D., Garant, D., Humphries, M. M., Bergeron, P., Careau, V., & Montiglio, P.-O. (2010). Personality and the emergence of the pace-of-life syndrome concept at the population level. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 365(1560), 4051–4063. https://doi.org/10.1098/rstb.2010.0208 Réveillon, T., Rota, T., Chauvet, É., Lecerf, A., & Sentis, A. (2019). Repeatable inter-individual variation in the thermal sensitivity of metabolic rate. Oikos, 85, 935–8. https://doi.org/10.1111/oik.06392 Rueden, C. T., Schindelin, J., Hiner, M. C., DeZonia, B. E., Walter, A. E., Arena, E. T., & Eliceiri, K. W. (2017). ImageJ2: ImageJ for the next generation of

96

scientific image data. BMC Bioinformatics, 18(1), 529. https://doi.org/10.1186/s12859-017-1934-z Sæther, B.-E. (1987). The Influence of Body Weight on the Covariation between Reproductive Traits in European Birds. Oikos, 48(1), 79–88. https://doi.org/10.2307/3565691 Schielzeth, H. (2010). Simple means to improve the interpretability of regression coefficients. Methods in Ecology, 1(2), 103–113. https://doi.org/10.1111/j.2041-210X.2010.00012.x Schielzeth, H., & Nakagawa, S. (2020). Conditional repeatability and the variance explained by reaction norm variation in random slope models. BioRxiv, 2020.03.11.987073. https://doi.org/10.1101/2020.03.11.987073 Schnurr, M. E., Yin, Y., & Scott, G. R. (2014). Temperature during embryonic development has persistent effects on metabolic enzymes in the muscle of zebrafish. Journal of Experimental Biology, 217(8), 1370–1380. https://doi.org/10.1242/jeb.094037 Seebacher, F. (2005). A review of thermoregulation and physiological performance in reptiles: What is the role of phenotypic flexibility? Journal of Comparative Physiology B, 175(7), 453–461. https://doi.org/10.1007/s00360-005-0010-6 Seebacher, F., Beaman, J., & Little, A. G. (2014). Regulation of thermal acclimation varies between generations of the short-lived mosquitofish that developed in different environmental conditions. Functional Ecology, 28(1), 137–148. https://doi.org/10.1111/1365-2435.12156 Seebacher, F., & Grigaltchik, V. S. (2014). Embryonic Developmental Temperatures Modulate Thermal Acclimation of Performance Curves in Tadpoles of the Frog Limnodynastes peronii. PLoS ONE, 9(9), e106492. https://doi.org/10.1371/journal.pone.0106492 Shama, L. N. S., Strobel, A., Mark, F. C., & Wegner, K. M. (2014). Transgenerational plasticity in marine sticklebacks: Maternal effects mediate impacts of a warming ocean. Functional Ecology, 28(6), 1482– 1493. https://doi.org/10.1111/1365-2435.12280 Storm, M. A., & Angilletta, M. J. (2007). Rapid assimilation of yolk enhances growth and development of lizard embryos from a cold environment. The Journal of Experimental Biology, 210(19), 3415–3421. https://doi.org/10.1242/jeb.005652

97

Sultan, S. E., & Stearns, S. C. (2005). Chapter 14 - Environmentally Contingent Variation: Phenotypic Plasticity and Norms of Reaction. In B. Hallgrímsson & B. K. Hall (Eds.), Variation (pp. 303–332). Academic Press. https://doi.org/10.1016/B978-012088777-4/50016-8 Thompson, M. J., Evans, J. C., Parsons, S., & Morand-Ferron, J. (2018). Urbanization and individual differences in exploration and plasticity. Behavioral Ecology, 29(6), 1415–1425. https://doi.org/10.1093/beheco/ary103 Toftegaard, T., Posledovich, D., Navarro-Cano, J. A., Wiklund, C., Gotthard, K., & Ehrlén, J. (2015). Variation in plant thermal reaction norms along a latitudinal gradient—More than adaptation to season length. Oikos, 125(5), 622–628. https://doi.org/10.1111/oik.02323 Tomlinson, S., Dalziell, E. L., Withers, P. C., Lewandrowski, W., Dixon, K. W., & Merritt, D. J. (2018). Measuring metabolic rates of small terrestrial organisms by fluorescence-based closed-system respirometry. The Journal of Experimental Biology, 221(7), jeb172874. https://doi.org/10.1242/jeb.172874 Via, S., Gomulkiewicz, R., De Jong, G., Scheiner, S. M., Schlichting, C. D., & Van Tienderen, P. H. (1995). Adaptive phenotypic plasticity: Consensus and controversy. Trends Ecology and Evolution, 10(5), 212–217. https://doi.org/10.1016/S0169-5347(00)89061-8 White, C. R., Schimpf, N. G., & Cassey, P. (2013). The repeatability of metabolic rate declines with time. The Journal of Experimental Biology, 216(10), 1763– 1765. https://doi.org/10.1242/jeb.076562

98

Supplementary Materials

Heterogenous Variance

Model with homogenous variance was best supported by WAIC values. As such, we did not explicitly model residuals in all subsequent models

Table S1 Comparisons of WAIC values for homogenous and heterogenous residuals

SE Diff Model WAIC value ELPD Diff Homogenous residuals -3.61 0 0

Heterogenous residuals -2.09 -0.76 2.08

The Influence of Developmental Temperature on the Thermal Reaction Norm of Metabolic Rate

Table S2 Model coefficients of full model testing whether developmental temperature affects the elevation and slope of the thermal reaction norm of metabolic rate. This model used a complete case dataset, n = 3818. The intercept is the cold developmental temperature. Mass and MR was log transformed and Age was z-transformed. Bolded estimates are significantly different from zero. COV represents covariance. Parameter Estimate Lower Upper

Fixed effects

Intercept -6.294 -6.364 -6.22

Treatment 29 -0.001 -0.062 0.058

Temperature 0.262 0.246 0.279

Mass 0.129 0.105 0.152

Age -0.035 -0.078 0.008

Treatment 29 × Temperature -0.016 -0.039 0.006

Random Effects

Lizard Identity

99

Intercept 0.009 0.006 0.015

Slope 9.77e-5 1.12e-7 0.001

COVIntercept – Slope -0.00018 -0.00128 0.000692

Sampling Session

Intercept 0.01 0.003 0.031

Measurement Error

Intercept 0.044 0.04 0.048

Residual 0.041 0.038 0.043

Table S3 Model coefficients of main effects model testing developmental temperature affects the elevation of the thermal reaction norm of metabolic rate. This model used an imputed dataset of n = 6000. The intercept is the cold developmental temperature. Note that the imputation model also estimates an intercept and residual variance for mass as it was also missing data. MR were log transformed and mass, age and temperature was z- transformed. Bolded estimates are significantly different from zero. COV represents covariance. Parameter Estimate Lower Upper

Fixed effects

Intercept -6.292 -6.366 -6.219

Treatment 29 -0.003 -0.062 0.055

Temperature 0.254 0.241 0.265

Age -0.034 -0.077 0.009

Mass 0.128 0.104 0.152

Random Effects

Lizard Identity

Intercept 0.009 0.006 0.014

Slope 1.01e-4 6.34-8 0.00048

COVIntercept – Slope -0.00017 -0.00122 0.000672

100

Sampling Session

Intercept 0.01 0.003 0.028

Measurement Error

Intercept 0.044 0.04 0.049

Residuals 0.041 0.038 0.044

Table S4 Model coefficients of main effects model testing developmental temperature affects the elevation of the thermal reaction norm of metabolic rate. This model used a complete case dataset, n = 3818. The intercept is the cold developmental temperature. MR were log transformed and mass, age and temperature were z-transformed. Bolded estimates are significantly different from zero. COV represents covariance. Parameter Estimate Lower Upper

Fixed effects

Intercept -6.293 -6.373 -6.218

Treatment 29 -0.003 -0.064 0.057

Temperature 0.254 0.242 0.266

Mass 0.128 0.107 0.15

Age -0.034 -0.078 0.008

Random Effects

Lizard Identity

Intercept 0.009 0.006 0.015

Slope 9.90e-5 1.24e-7 0.001

COVIntercept – Slope -0.000184 -0.00131 0.000648

Sampling Session

Intercept 0.01 0.003 0.028

Measurement Error

Intercept 0.044 0.04 0.049

101

Residuals 0.041 0.038 0.043

We expected that treatment differences in thermal reaction norms would be greatest at the beginning of the study, as such we ran the full interaction model for just the first sampling session (nobservations= 600). Similar to the overall result, we found that developmental temperatures did not affect the intercept nor the slope of the thermal reaction norm (Estimate: Treatment 29 × Temperature = 0, 95% CI [-0.02 -0.02]).

The Influence of Developmental Temperature on the Repeatability of the

Thermal Reaction Norm and Temperature Specific Repeatability of Metabolic

Rate

Figure S1 Adjusted repeatability for average metabolic rate for the ‘cold’ developmental temperature group (blue) and the ‘hot’ developmental temperature group (red). Estimates were calculated from a complete case analysis. There were no significant differences among treatment in repeatability estimates (see Table S5). Repeatability did not change with acute temperature. Error bars represent 95% credible intervals.

102

Table S5 Temperature specific, adjusted repeatability estimates of log transformed metabolic rate for lizards from two developmental temperatures (nhot = 25, ncold =

26). These values were estimated from complete case dataset, nobs = 3818. Bolded values are significantly different from zero. There were no statistical differences among treatments at each acute temperature. T represents acute temperature, R represents repeatability, L and U represents the lower and upper 95% credible intervals, pd is the probability of direction

Cold Hot

nlizards = 26 nlizards = 25

T (ºC) R L U R L U

24 0.24 0.12 0.4 0.09 0.03 0.2

26 0.23 0.12 0.38 0.09 0.03 0.19

28 0.22 0.12 0.37 0.09 0.03 0.19

30 0.21 0.11 0.36 0.1 0.04 0.2

32 0.21 0.1 0.36 0.11 0.04 0.22

34 0.2 0.09 0.35 0.12 0.04 0.25

Treatment difference (Hot - Cold)

T (ºC) Mean L U pd difference 24 -0.146 -0.310 0.015 96.5%

26 -0.137 -0.295 0.016 96.25%

28 -0.126 -0.281 0.022 95.95%

30 -0.113 -0.271 0.031 94.25%

32 -0.099 -0.259 0.053 91%

34 -0.084 -0.253 0.082 85.95%

103

a) b)

Among ID Variance Residual 0.05

0.04 2.50

0.03

2.25

0.02

0.01 2.00

0.00

Cold Hot Cold Hot

Figure S2 Violin plots of a) among individual and b) residual variance components for average metabolic rate for the ‘cold’ developmental temperature group (blue) and the ‘hot’ developmental temperature group (red) irrespective of acute temperature. pdAmong ID = 98.35% , pdResidual = 100%.

Table S6 Model coefficients of model whether body mass, temperature and age predicts variation in metabolic rate. In this model, we fitted a ‘series’ as random intercept with temperature as a random slope to estimate repeatability of the slope. See Statistical Analyses for details. This imputation model used a subset dataset of lizards in the cold developmental temperature only n = 26, nobs = 3000. MR were log transformed and mass, age and temperature were z- transformed. Bolded estimates are significantly different from zero. COV represents covariance. Parameter Estimate Lower Upper

Fixed effects

104

Intercept -6.289 -6.342 -6.237

Temperature 0.261 0.244 0.278

Age -0.012 -0.049 0.029

Mass 0.136 0.099 0.172

Random Effects

Lizard Identity

Intercept 0.015 0.007 0.028

Slope 0.00021 1.70e-7 0.001

COVIntercept – Slope -0.000579 -0.00303 0.00084

Series (Within individual)

Intercept 0.015 0.01 0.022

Slope 0.0003 3.62e-7 0.002

COVIntercept – Slope -0.00043 -0.00273 0.00136

Measurement Error

Intercept 0.037 0.03 0.043

Residuals 0.045 0.041 0.049

Table S7 Model coefficients of model whether body mass, temperature and age predict variation in metabolic rate. In this model, we fitted a ‘series’ as random intercept with temperature as a random slope to estimate repeatability of the slope. See Statistical Analyses for details. This imputation model used a subset dataset of lizards in the hot developmental temperature only n = 25, nobs = 3000. MR were log transformed and mass, age and temperature were z- transformed. Bolded estimates are significantly different from zero. COV represents covariance. Parameter Estimate Lower Upper

Fixed effects

Intercept -6.299 -6.328 -6.268

Temperature 0.245 0.229 0.262

105

Age -0.003 -0.038 0.03

Mass 0.124 0.093 0.155

Random Effects

Lizard Identity

Intercept 0.003 0 0.008

Slope 0.0003 8.50e-7 0.001

COVIntercept – Slope 0.000272 -0.000577 0.00153

Series (Within individual, among sessions)

Intercept 0.001 0 0.002

Slope 0.00053 2.38e-6 0.002

COVIntercept – Slope -0.00134 -0.00388 0.000432

Measurement Error

Intercept 0.035 0.03 0.041

Residuals 0.037 0.034 0.041

Table S8 Model coefficients of model whether body mass, temperature and age predict variation in metabolic rate. In this model, we fitted a ‘series’ as random intercept with temperature as a random slope to estimate repeatability of the slope. See Statistical Analyses for details. This model used a complete case dataset of lizards in the cold developmental temperature only n = 26, nobs = 1897. MR were log transformed and mass, age and temperature were z- transformed. Bolded estimates are significantly different from zero. COV represents covariance. Parameter Estimate Lower Upper

Fixed effects

Intercept -6.291 -6.342 -6.241

Temperature 0.261 0.244 0.278

Mass 0.136 0.098 0.172

Age -0.011 -0.05 0.028

106

Random Effects

Lizard Identity

Intercept 0.015 0.007 0.029

Slope 0.0002 3.93e-7 0.0009

COVIntercept – Slope -0.000575 -0.00289 0.000882

Series (Within individual)

Intercept 0.015 0.01 0.022

Slope 0.0003 2.79e-7 0.00015

COVIntercept – Slope -0.000417 -0.00283 0.00138

Measurement Error

Intercept 0.037 0.03 0.044

Residual 0.045 0.041 0.049

Table S9 Model coefficients of model whether body mass, temperature and age predicts variation in metabolic rate. In this model, we fitted a ‘series’ as random intercept with temperature as a random slope to estimate repeatability of the slope. See Statistical Analyses for details. This model used a complete case dataset of lizards in the hot developmental temperature only n = 25, nobs = 1921. MR were log transformed and mass, age and temperature were z- transformed. Bolded estimates are significantly different from zero. COV represents covariance. Values with * indicate very small values that are still greater than zero Parameter Estimate Lower Upper

Fixed effects

Intercept -6.298 -6.329 -6.268

Temperature 0.246 0.229 0.263

Mass 0.124 0.095 0.155

Age -0.004 -0.037 0.03

Random Effects

107

Lizard Identity

Intercept 0.003 2.5e-4 0.008

Slope 0.0003 4.3e-7 0.001

COVIntercept – Slope 0.000308 - 0.00158 0.000551 Series (Within individual, among sessions) Intercept 0.013 0.008 0.019

Slope 0.0005 1.13e-6 0.002

COVIntercept – Slope -0.00134 -0.00366 0.00037

Measurement Error

Intercept 0.035 0.03 0.042

Residual 0.037 0.034 0.041

Table S10 Model coefficients of model whether body mass, temperature and age predict variation in metabolic rate. This imputation model used a subset dataset of lizards in the cold developmental temperature only n = 26, nobs = 3000. MR were log transformed and mass, age and temperature were z- transformed. Bolded estimates are significantly different from zero. COV represents covariance. Values with * indicate very small values that are still greater than zero. Parameter Estimate Lower Upper

Fixed effects

Intercept MR -6.29 -6.368 -6.211

Temperature 0.262 0.243 0.279

Age -0.025 -0.093 0.036

Mass 0.117 0.081 0.153

Random Effects

Lizard Identity

Intercept 0.015 0.008 0.028

108

Slope 0.0002 2.1e-7 0.001

COVIntercept – Slope -0.000628 -0.00292 0.000782

Sampling Session

Intercept 0.01 0.003 0.028

Measurement Error

Intercept 0.046 0.039 0.053

Residual MR 0.045 0.041 0.049

Table S11 Model coefficients of model whether body mass, temperature and age predict variation in metabolic rate. This imputation model used a subset dataset of lizards in the hot developmental temperature only n = 25, nobs = 3000. Note that the imputation model also estimates an intercept and residual variance for mass as it was also missing data. MR were log transformed and mass, age and temperature were z-transformed. Bolded estimates are significantly different from zero. Values with * indicate very small values that are still greater than zero. Parameter Estimate Lower Upper

Fixed effects

Intercept MR -6.296 -6.365 -6.228

Temperature 0.246 0.229 0.264

Age -0.025 -0.072 0.022

Mass 0.132 0.104 0.164

Random Effects

Lizard Identity

Intercept 0.005 0.002 0.011

Slope 0.0002 6.01e-7 0.001

COVIntercept – Slope 0.000277 -0.000651 0.00152

Sampling Session

109

Intercept 0.009 0.003 0.025

Measurement Error

Intercept 0.043 0.037 0.049

Residuals 0.037 0.034 0.041

Table S12 Model coefficients of model testing whether body mass, temperature and age predict variation in metabolic rate. This model used a complete case dataset of lizards in the cold developmental temperature only n

= 26, nobs = 1897. The intercept is the cold developmental temperature. MR were log transformed and mass, age and temperature were z-transformed. Bolded estimates are significantly different from zero. Values with * indicate very small values that are still greater than zero Parameter Estimate Lower Upper

Fixed effects

Intercept -6.293 -6.376 -6.214

Temperature 0.262 0.244 0.28

Mass 0.118 0.081 0.155

Age -0.027 -0.094 0.037

Random Effects

Lizard Identity

Intercept 0.016 0.008 0.03

Slope 0.0002 3.68e-7 0.001

COVIntercept – Slope -0.000663 -0.00306 0.000776

Sampling Session

Intercept 0.011 0.003 0.034

Measurement Error

Intercept 0.046 0.039 0.053

Residual 0.045 0.041 0.049

110

Table S13 Model coefficients of model testing whether body mass, temperature and age predict variation in metabolic rate. This imputation model used a complete case dataset of lizards in the hot developmental temperature only n = 25, nobs = 1921. The intercept is the cold developmental temperature. MR were log transformed and mass, age and temperature were z- transformed. Bolded estimates are significantly different from zero. Values with * indicate very small values that are still greater than zero Parameter Estimate Lower Upper

Fixed effects

Intercept -6.293 -6.361 -6.222

Temperature 0.246 0.229 0.263

Mass 0.133 0.102 0.164

Age -0.025 -0.07 0.018

Random Effects

Lizard Identity

Intercept 0.005 0.002 0.01

Slope 0.00027 2.62e-7 0.001

COVIntercept – Slope 0.000255 -0.000688 0.00171

Sampling Session

Intercept 0.009 0.003 0.026

Measurement Error

Intercept 0.043 0.037 0.049

Residuals 0.037 0.034 0.041

111

CHAPTER 4

Heritability and developmental plasticity of growth in an oviparous lizard

Fonti Kar1, Shinichi Nakagawa1,2, Daniel W.A. Noble3

1School of Biological Earth and Environmental Sciences, Ecology and Evolution Research Centre, University of New South Wales, Sydney, NSW, Australia 2Diabetes and Metabolism Division, Garvan Institute of Medical Research, 384 Victoria Street, Darlinghurst, Sydney, NSW 2010, Australia 3Division of Ecology and Evolution, Research School of Biology, The Australian National University, Canberra, ACT, Australia

All authors conceived the ideas and designed the study. FK and DN collected and analysed the data, FK wrote the first draft, FK, DN and SN edited the manuscript. All authors declare no conflict of interest

112

Abstract

Selective processes act on phenotypic variation however the evolutionary potential of any given trait relies on the underlying heritable variation. Developmental plasticity is an important source of phenotypic variation, but it can also promote changes in genetic variation, yet we have a limited understanding on how they are both impacted. Here, we quantified the influence of developmental temperature on the growth in delicate skinks (Lampropholis delicata) and partitioned the total variance using an animal model fitted with a genomic relatedness matrix. We measured mass for 262 individuals (nhot = 126, ncold = 136) over 16 months (nobservations = 3,002) and estimated heritability and maternal effects over time. Our results show that lizards reared in cold developmental temperatures had a higher mass compared to lizards that were reared in hot developmental temperatures. We found that developmental temperature did not impact the rate of growth. On average, additive genetic variance, maternal effects and heritability were higher in hot developmental temperature treatment, however these differences were not statistically significant. Heritability increased with age, whereas maternal effects decreased upon hatching but increased again at a later age. Our work suggests that evolutionary potential of growth is complex, age dependent and not overtly affected by extremes in natural nest temperatures.

113

Introduction

Developmental plasticity plays a key role in generating phenotypic variation (Noble et al 2018; Ghalambor et al., 2007; West-Eberhard, 2003). The complex interplay between an individual’s genotype, and the developmental environment in which that genotype finds itself, means that a range of different phenotypes can give rise (Monaghan, 2008; West-Eberhard, 2003). Phenotypic changes resulting from distinct early life experiences can have persistent effects on individual fitness (Monaghan, 2008; Noble et al., 2018). Changes induced by developmental environments may result in a better match between the adult phenotype and the subsequent selective environment. In some cases; however, maladaptive phenotypes can arise if there is a mismatch between later-life environments and those experienced early in development (Beaman et al., 2016; Ghalambor et al., 2007). Regardless, phenotypic plasticity represents a promising immediate solution for threatened populations by allowing them to better track adaptive optima and persist (Beldade et al., 2011; Noble et al., 2019; West-Eberhard, 2003). Understanding the consequences of developmental environments on phenotypes and fitness is therefore critical to predict how populations will survive in stressful conditions (Botero et al., 2015; Reed et al., 2010).

A population’s capacity to evolve depends not only on the strength of selection but also on the underlying standing genetic variation (Lynch & Walsh, 1998). It has long been recognised that selection and genetic variation changes across environments (Falconer & Mackay, 1996). As such, a great deal of effort has been put towards understanding the circumstances under which genetic variation may change with the environment and the magnitude of those changes (Charmantier & Garant, 2005; Fischer et al., 2020; Hoffmann & Merilä, 1999; Noble et al., 2019; Rowiński & Rogell, 2017; Wood & Brodie, 2015). Genetic variance in novel environments may increase due to relaxation of selection pressures combined with higher mutation rates (Hoffman & Parsons, 1991; Hoffmann & Merilä, 1999). An increase in genetic variance is also expected when buffering mechanisms breakdown triggering a release of ‘cryptic genetic variation’ in stressful conditions (Paaby & Rockman, 2014). Furthermore, others mechanisms such as low cross-environment genetic

114

correlations or condition-dependence of gene expression can also affect the amount of genetic variance in different environments (Charmantier & Garant, 2005; Coltman et al., 2001). Environmental dependence of genetic variance implies that under the same selection pressure, the speed of evolutionary change would likely change thus making it difficult to predict genetic adaptation.

Comparative studies have shown that the influence of environmental stress on genetic variance during development is not straightforward (Charmantier & Garant, 2005; Hoffmann & Merilä, 1999; Rowiński & Rogell, 2017). In lab studies, elevated developmental stress has been shown to increase the heritability of morphological traits (Hoffmann & Merilä, 1999), whereas wild, non-domestic populations tend to have higher heritability in favourable environments (Charmantier & Garant, 2005). Lack of consensus may be related to increased environmental heterogeneity in wild populations, making them more difficult to compare with lab studies. It has been suggested that responses to different developmental stressors (e.g. heat shock vs. starvation) may be associated with disparate patterns of gene expression making broad comparisons more variable (Charmantier & Garant, 2005; Dahlgaard & Hoffmann, 2000). Importantly, environmental comparisons of heritability have been criticised as the ratio nature of its calculations can mask changes in the relative contributions of non-genetic and genetic variance (Rowiński & Rogell, 2017). For example, a meta-analysis found that heritability of life history traits which has been argued to be more important to fitness, did not change between control and stressful conditions (Rowiński & Rogell, 2017). The same pattern was observed for morphological traits (Fischer et al., 2020). Upon closer inspection, both additive genetic and environmental variance of life history traits increased under stressful conditions whereas the opposite was true for morphological traits (Rowiński & Rogell, 2017). The dynamics of both genetic and non-genetic sources of variation under different developmental environments can thus influence the evolutionary potential of fitness related traits.

Body size is fundamental to fitness and is both heritable and environmentally responsive (Noordwijk et al., 1988; Stillwell & Fox, 2009). Developmental

115

environments, such as temperature and nutritional stress can drive substantial variation in body size, largely through shifts in how organisms grow (Eyck et al., 2019; Noble et al., 2018). Maternal investment in offspring are also important sources of body size variation (Noble et al., 2014; Wilson & Réale, 2006). Variation among mothers in egg investment, nest site selection or timing of birth (Mitchell et al., 2018; Shine & Harlow, 1996; Uller & Olsson, 2010) are expected to contribute the most to offspring body size early in development (Mousseau & Fox, 1998). However, these effects have shown to decline with age as maternal investment subside (Krist, 2010; Wilson, Kruuk, et al., 2005). Additionally, environmental factors such as shared habitats or long-term seasonal effects can also account for a substantial proportion of variability in body size (Kruuk, 2004). For example, permanent environmental effects that varied across years explained 26% – 35% of body size variation in bighorn sheep (Ovis canadensis) (Réale et al., 1999). Similarly, 56% of variation in body mass was attributed to nest boxes shared among siblings in blue tit (Cyanistes caeruleus) chicks (Charmantier et al., 2004). As such, the various sources that influence body size variation (genetic, environmental, maternal) are predicted to vary across ontogeny and temporal approach is therefore needed in order to evaluate when evolutionary potential of body size is greatest.

Here we investigated the impact of developmental temperature on growth and mass in an oviparous skink (Lampropholis delicata) – two traits that are critically important to fitness. We also test how developmental environments affect evolutionary potential in these traits. Growth trajectories (nobservations = 3,002) for lizards that hatched from two incubation treatments (nhot = 126, ncold = 136), were measured over the first 16 months of life. Using 8,433 single nucleotide polymorphic (SNP) markers, we derived a genomic relatedness matrix to estimate quantitative genetic parameters. Using these data, we address two key questions: 1) How does developmental temperature affect the rate and shape of growth trajectories (initial mass, growth rate and curvature of growth trajectory)? and 2) How does developmental temperature affect genetic and non-genetic variance across age? According to the ‘temperature-size rule’, we expect lizards experiencing cold developmental temperatures to have larger initial masses and slower growth rates – possibly resulting in lizards reaching sexual maturity at a later age compared to lizards experiencing hot

116

developmental temperatures (Angilletta Jr et al., 2017). In addition, we predicted greater amount of genetic variance under higher developmental temperatures, after controlling for non-genetic sources of variance. We expected maternal effects and permanent environment effects to manifest early in development and dissipate over time.

117

Materials and Methods

Lizard collection and husbandry

From 2015 – 2017, we established a breeding colony of adult L. delicata (nfemales =

144, nmales = 50) using wild individuals collected across three sites throughout the Sydney region between August and September 2015 (UNSW Kensington Campus: -33.92, 151.24; Sydney Park: -33.91, 151.18, Macquarie Park: -33.77, 151.10). Using a half-sib breeding design, we paired three females with a single male in opaque plastic enclosures measuring 35cm × 25cm × 15cm (L × W × H). Enclosures were kept under UV lights on a 12 hours light : 12 hours dark cycle in a temperature-controlled room set to 24ºC. Lizards were given access to a heat lamp that elevated temperatures to between 28-32 ºC. Each enclosure was lined with newspaper and lizards had constant access to water. Tree bark was used as refuge. Adult lizards were fed medium sized crickets ad libitum (Acheta domestica) dusted with calcium powder and multi-vitamin every two days. From the beginning of the egg laying season (October of each year), we replaced newspaper lining with garden potting mix and placed an opaque plastic box (12 cm × 17.5 cm × 4.3 cm) containing moistened vermiculite in each enclosure for females to oviposit their eggs. During this time, enclosures were sprayed with water every second day to maintain a relatively humid environment. From October to November, egg boxes were checked every day. Tail tissue samples (~1 mm) were taken from adults that were from enclosures producing eggs for DNA extraction (see below). All tissues were stored in 70% ethanol. Animal collection was approved by the New South Wales National Parks and Wildlife Service (SL101549) and all procedures were approved by the Macquarie University Ethics committee (ARA 2015/015) and University of New South Wales Animal Care and Ethics committee (ACEC 15/51A).

Developmental Temperature Manipulations

Eggs were collected between October to March, over two reproductive seasons from 2016 and 2017. As soon as eggs were found, they were weighed using a digital scale to the nearest 0.01g (Ohaus Scout SKX123). We also measured egg length (distance between the furthest points along the longest axis of the egg) and egg width (distance between the widest points along the axis perpendicular

118

to the longest axis of the egg) using digital callipers to the nearest 0.01mm. Following measurements, each egg was placed in a plastic cup (80ml) containing three grams of vermiculite and four grams of water. Each cup was then covered using cling wrap and secured using an elastic band. We used a split-clutch design where eggs from single clutch were pseudo-randomly assigned to one of two developmental temperature treatments. We used two incubators to precisely control the temperature of eggs (LabWit, ZXSD-R1090). The ‘hot’ treatment was exposed to a mean temperature of 29ºC whereas the ‘cold’ treatment was exposed to a mean temperature of 23ºC. Both incubators fluctuated +/- 3ºC over a 24 hour period around these mean temperatures to simulate natural nest site temperature variability. These treatments represent the temperature extremes of natural nest sites for L. delicata (Cheetham et al., 2011). Egg cups were rotated within each incubator weekly to avoid uneven heat circulation within incubators. Incubators were also checked daily for hatchlings.

Quantifying Growth Rate

Newly emerged hatchlings were weighed to the nearest 0.01g and a small tail tip clipping (~2mm) was taken for genetic analyses. Ventral photographs were taken for digital measurement (Nikon Coolpix A900). For the first two months, photographs of hatchlings were taken approximately every 14 days. After which, hatchlings were photographed at approximately a 35-day interval. From six months onwards, we manually measured hatchling SVL using a clear ruler to the nearest ~0.5mm. We also recorded the mass of the individual each time photographs or SVL measurements were taken. Growth measurements continued until we had approximately 16 measures per individual (mean = 11.5 , SD = 4.71). By the end of the study, the mean age for hot incubated lizards was 335.82 (range: 0 – 711) and for cold incubated lizards it was 384.8 (range: 0 – 707) which is approximately 25 – 50% of their total lifespan (Chapple et al., 2014). From the photographs, we extracted snout-vent-length (SVL; from tip of snout to the beginning of the cloaca opening) using ImageJ software (Rueden et al., 2017). For the first initial nine months, hatchlings were housed individually in opaque plastic enclosures (32.3cm x 18.5cm x 6cm) lined with newspaper. Hatchlings were fed the same number of crickets every second day and had constant access to a tree bark refuge and water. Hatchling enclosures were

119

placed in a temperature control room under the same conditions as described above for the adult colony. For logistical reasons, at approximately nine months, hatchlings were housed in groups of five in opaque bins with the same measurements as the adult enclosures. We pseudo-randomised individuals to each shared enclosure while maintaining a similar number of individuals from each treatment.

Genomic Relatedness Matrix

We derived a genomic relatedness matrix (GRM) using single nucleotide polymorphism (SNP) genotypes for all 262 offspring with growth data (132 putative parents; nfemales = 69, nmales = 63). While our half-sib breeding design allowed us to assign parentage to derive a pedigree, high levels of sperm storage and low levels of multiple paternity (94% of females had been sired by a single male) meant our pedigree had low resolution to effectively estimate additive genetic variation. Recent studies have shown that GRM derived from SNPs have low error rates (<0.3%) and are able to reconstruct pedigree relationships when at least 200 SNP loci are used (Bérénos et al., 2014; Huisman, 2017). Moreover, both relatedness and heritability values estimated from a GRM very similar to those inferred using a pedigree (Bérénos et al., 2014; Huisman, 2017). Single nucleotide polymorphism libraries were designed and animals genotyped using DArTseq™ ( Diversity Arrays Technology) methods. For more details on DNA extraction and SNP genotyping see ESM. Prior to deriving our GRM, we filtered our SNPs using the R package dartR (Gruber et al., 2018). We filtered loci based on various metrics in the following order: 1) read depth (8 – 40); reproducibility (> 0.996); call rate by loci (> 0.97) and then by individual (> 0.80); monomorphic loci; minor allele frequencies (> 0.02); Hamming Distance among loci (> 0.25) and Hardy Weinberg Equilibrium. This clean-up process resulted in a dataset of 8,438 loci with an average call rate of 98.5% (see ESM and provided code). Using these 8,438 loci we derived a GRM, which describes the proportion of the genome that is identical by descent (VanRaden, 2008). We calculated a GRM for all hatchlings using the snpReady R package (Granato et al., 2018) following methods described by VanRaden, 2008: �� ��� = 2 ∑ �(1 − �)

120

where Z is the centered squared matrix of SNP genotypes of all individuals. This is calculated from a matrix of where heterozygote SNP genotypes (AT) were coded as 0, homozygote genotypes for the SNP allele (AA) were coded as

1 and homozygotes for the original allele (TT) were coded as -1. pi is the frequency of the second locus at locus position i. The denominator scales the GRM matrix so that the values approximate a relatedness matrix derived from a pedigree. The GRM was then inverted for modelling fitting (see ESM and provided code).

Statistical Analyses

All analyses were performed using R (Core Team, 2013). We checked the data for potential input errors using histograms, scatterplots and Cleveland plots. We fitted Bayesian linear mixed effects models (LMM) in ‘brms’ which interfaces with Stan (Bürkner, 2017; Gelman et al., 2015). Mass was log- transformed, and age was z-transformed. For all models we used noninformative priors with 4000 iterations with a burn in of 1500, sampling from the posterior distribution every fifth iteration. We ensured proper mixing by inspecting trace plots and checked that scale reduction factors were less than 1.01. We report posterior means and 95% credible intervals for all parameters throughout.

Impact of Developmental Temperature on Additive Genetic Variance and

Maternal Effects Across Age

First, we tested whether developmental temperature influenced the overall heritability of mass and the relative contributions of variance irrespective of age. For each treatment group, we fitted intercepts only in the fixed effects with random intercepts for additive genetic variance (G), maternal effects (M) and permanent environmental effects (PE) as we had repeated measures of the same individuals (Wilson et al., 2010). The model also estimated residual variance (R). We included our GRM to estimate additive genetic variation. Overall heritability (h2) of mass using this intercept (I) model was calculated as:

� ℎ = (� + � + �� + �) To then test how G, M and h2 change across age, we used model selection to determine the most appropriate random effects structure for our data as we had

121

no a priori knowledge of what (or how) variance components change with age (Wilson & Réale, 2006). We fitted seven models with varying complexity in their random effects and compared their Watanabe–Akaike Information Criterion (WAIC) values (Table S1). We fitted random intercepts and random slopes by including either a linear age term or both linear and quadratic age terms to partition variance across age. Two models were equally supported, the first included a random linear and quadratic slope for G and M and PE. (Model 3 - Table S1) and the second included a random linear and quadratic slope for G and M, respectively, and a random intercept for PE (Model 7 – Table S1). To avoid overfitting, we selected the more parsimonious model and used this random effect structure for the remaining analyses unless stated otherwise. Residual variance may be conflated with estimates of other variance components if it changes over time (heterogenous variance) and is not properly accounted for. We therefore explicitly modelled residual variance to verify if this was the case and compared homogenous and heterogenous residual variance models using WAIC. We fitted two models, both of which had the same fixed and random effects structure as Model 7 described above. The first model had homogenous residual variance whereas in the second model we modelled residual variance with a linear slope thereby allowing it to vary with age. The model with heterogenous variance was best supported (Table S2), we therefore modelled heterogenous variance in all subsequent models unless stated otherwise.

To test for treatment differences in variance components, we separate fitted an intercept-only model for each treatment group with our best supported random effect structure (Model 7) and heterogenous residual variance. We estimated a genetic variance-covariance matrix for each treatment (�), where the diagonal elements represent the additive genetic variances for the intercept (�), slope

(�) and the quadratic (�) across age. The off-diagonal elements are the additive genetic covariances between the growth curve parameters, for example, ���, is the additive genetic variance between the intercept and the quadratic slope.

� ���, ���, � = ���, � ���, ���, ���, �

122

Similarly, the variance-covariance matrix for dams (�) can be decomposed in the same manner as �.

� ���, ���, � = ���, � ���, ���, ���, � For each treatment group, we then calculated additive genetic variance at a given age � using the random slope terms and their covariances as follows: � = � + (� . �) + (� . �) + (2�. ���,) + (2� . ���,) + (2� . ���,) where � is a specific age. Age-specific maternal effect � was calculated using the same formula but with the relevant variance components from �. Age- specific heritability, ℎ, is thus a ratio of all variance components at a given age �. The proportion of variance explained by maternal effects (m2) is calculated in the same manner.

� ℎ = (� + � + �� + �) As the mean body mass increases over time, the variance may also increase concurrently due to scale effects and potentially bias estimates of quantitative genetics parameters (Wilson, Kruuk, et al., 2005). We therefore calculated coefficients of variation (CV) across age for each variance component by dividing variance by the predicted mean mass at a given age. Interpretations using CV estimates did not change our overall conclusions for additive genetic variance or maternal effects, we therefore present the raw estimates of each variance component below (See ESM).

The Influence of Developmental Temperature on Growth Trajectories

To test how developmental temperatures affect average growth trajectories, we also fitted three models that varied in their fixed effect structure to determine how developmental temperatures affect: 1) initial mass (intercept of curve), 2) linear rate of growth (linear slope) and 3) curvature of the growth trajectory (quadratic term). We also wanted to test for treatment differences in age at which lizards reach their maximum mass by solving for the maxima of quadratic regression equation. We fit mass as the response accounting for the same random effects described above. The first model included the main effect of developmental temperature and the linear and quadratic term for age (Table S2). The other two models differed in their interaction terms between 123

developmental temperature with age and age2 (Table 2, S3). We then compared WAIC values to select the best model for our data that explained changes in mass across age between the two developmental temperature treatments (Table 1).

124

Results

Over two years, we collected 3,002 observations of mass data for a total of 261 individuals (� = 125, � = 136). On average, the incubation period for the ‘hot’ treatment was 29.36 days (SD = 2.17, range = 15 - 49) days and 48.48 days (SD = 4.18, range = 25 - 56) for the ‘cold’ treatment.

V V V additive additive maternal genetic genetic 0.005 0.01 Vmaternal 0.028 V (0* – 0.016) (0* – 0.025) (0.001 – 0.056) permanent 0.006 environment (0* – 0.024) 0.008 Vpermanent (0* – 0.025) environment 0.015 (0* – 0.053)

Variable Variable Variable V_additive_g V_additive_gV_additive_g V V_maternal residual V_maternalV_maternal Vresidual V_pe 0.274 V_peV_pe 0.25 Vresidual Vresidual (0.254 – 0.294 Vresidual (0.23 – 0.271)

a) Cold incubation treatment b) Hot incubation treatment

Figure 1 Pie charts depicting the overall relative contributions of mass variance for a) hot (nlizards = 126) and b) cold (nlizards =136) incubated lizards irrespective of age. Point estimates and 95% credible intervals are presented in Table S3. There were no significant differences in variance components between developmental temperature treatments. * in indicates very small values that were above 0.

Overall, additive genetic variance, permanent environmental variance and heritability (ℎ) of growth appears to be higher in the hot developmental temperature treatment (Fig. 1). However, there were no significant differences among treatment groups (Table S3).

Treatment groups did not differ in how the relative contributions of � and � changed with age as their 95% credible intervals overlapped (Fig. 2). Additive genetic variance remained relatively low and constant upon emergence until 125

approximately nine months of age, after which it increased rapidly (Fig. 2). Maternal effects decreased sharply upon hatching and dropped to the minimum at approximately six months before it increased again (Fig. 2). There were some differences among developmental treatments in how residual variance changed with age (Fig. S1). Residual variance in cold incubated lizards had a much higher intercept compared to hot incubated lizard however their residual variance converged by eight months of age (Fig. 2).

a) G b) c) Residual G M Residual

0.4 G 0.8 M Residual

0.3 0.6 5

0.2 0.4 Variance of Mass Variance 4 0.1 0.2

0.0 0.0 0 60 120 180 240 300 360 420 480 0 60 120 180 240 300 360 420 480 0 60 120 180 240 300 360 420 480 Age Age (days)

Figure 2 Scatterplot showing how a) additive genetic variance (G), b) maternal effects (M), c) residual variance changed with age for the hot developmental

treatment (nlizards = 125, red) and the cold developmental treatment (nlizards = 136, blue). Points represent posterior means, thin lines represent the 95% credible intervals, thick lines represent the mean for each treatment group. Note that permanent environmental effects were treated as constant across

age. Vpermanent environment for the hot treatment group was 0.0047 [0.00017 –

0.0096], Vpermanent environment for the cold treatment group was 0.0047 [0.00065 – 0.0085].

We investigated whether increases in average mass over time affected variance estimates due to scaling effects between the mean and variance. However, we found that the CV of G and M followed the same pattern as the raw variance estimates suggesting that changes in variance were not the result of increasing mean body mass with age (Fig. S1). 126

After accounting for heterogenous residual variance, we found no treatment differences in heritability or the proportion of variance explained by maternal effects (�) (Fig. 3). Heritability was very low for the first year of growth in L. delicata and only began increasing at one year of age (Fig. 3). As predicted � decreased soon after hatching, however it increased slightly again from six months of age (Fig. 3). The � and � matrices for each treatment group are presented in Table S4-S5.

a) 0.100

0.075 2

h 0.050 2 h 0.025

0.000 0 60 120 180 240 300 360 420 480 b) 0.25

0.20 2

M 0.15 2 M 0.10

0.05

0.00 0 60 120 180 240 300 360 420 480 AgeAge (days)

Figure 3 (a) Heritability (h2) and (b) the proportion of total variance explained by maternal effect variance (M2) across age (days) for the hot developmental treatment (nlizards = 125, red) and the cold developmental treatment (nlizards = 136, blue). Points represent estimates generated from the posterior distribution

127

of the variance-covariance matrix, thin lines represent the 95% credible intervals, thick lines represent the mean for each treatment group.

Developmental plasticity in growth trajectories in response to temperature

While the model containing a full interaction between treatment and linear and quadratic age was best supported, the improvement in WAIC value was marginal (Table 1). Moreover, the linear growth rate (Age) and curvature of the growth trajectory (Age2) did not differ significantly between the two developmental temperature treatments in any of the models containing interactions (Table S7 - S9). Irrespective of treatment, lizard mass increased by 1.65 g for every 1 SD unit increase in age.

Table 1 Comparisons of WAIC values of four models (nobs = 2926) with different combinations of treatment interactions with age parameters. ∆ELPD represents the difference in expected log predicted density. Age measured in days was z-transformed (mean = 361.34, SD = 185.16)

Std. Err Formula of Fixed Effects WAIC ΔELPD ΔELPD

Treatment + Age + Age2 + Treatment -3301 0 0 × Age + Treatment × Age2 Treatment + Age + Age2 + Treatment × -3295 -0.62 1.182 Age Treatment + Age + Age2 + Treatment -3300 -2.798 1.375 × Age2 Treatment + Age + Age2 -3292 -4.452 1.563

Developmental temperature did, however, influence hatching mass (Table 1, Fig. 3). Lizards from the ‘cold’ treatment were on average 0.030 g (0.018g – 0.041g) heavier compared to lizards from the ‘hot’ treatment (Table. 2). Larger initial masses meant that lizards from the ‘cold’ treatment reached their maximum mass slightly earlier (382.97 days, 95% CI: 358.84 – 409.78) compared to lizards from ‘hot’ treatment (413.04 days, 95% CI: 379.70 – 452.34). G and M matrices from this model, along with other variance components, are presented in Table S6.

128

0.6

0.4 Log transformed Mass Log transformed

0.2

0 100 200 300 400 500 Age

Treatment 23 29 Figure 4 Model predictions of log-transformed mass over age from the two developmental temperatures. We randomly subset 40 lizards (20 from each treatment) to plot their individual growth curves. Points represent mean estimates for each lizard from the hot developmental treatment (hot) and the cold developmental treatment (blue). Thick lines represent average growth curve for each treatment. Faint grey lines are each individual’s growth curve. Model predictions were generated from the full model where interaction terms between treatment and both the linear component and quadratic component were included

Table 2 Coefficient estimates from full model testing the effects of developmental treatment on mass and how mass changes with age. Bolded estimates are significantly different from zero. * indicates that value is above zero prior to rounding. nobs = 2926. Age measured in days was z-transformed (mean = 361.34, SD = 185.16). G and M matrices for this model is presented in Table S6. Parameter Estimate Lower Upper

Intercept -0.991 -1.01 -0.971

Treatment -0.083 -0.114 -0.05 Age 0.5 0.476 0.526

Age2 -0.196 -0.216 -0.178 129

Treatment × Age 0.008 -0.021 0.037

Treatment × Age2 0.022 -0.007 0.052

130

Discussion

Early development at hot temperatures resulted in smaller body sizes compared to development at cold temperatures. Growth trajectories, however, were not significantly impacted by early thermal environments – lizards from both temperatures grew at the same rate despite cold animals remaining larger throughout life. Marginalising over age, we found that developmental temperature did not impact the relative contributions of additive genetic, maternal variance, permanent environment variance or residual variance. The environmental component of the phenotype (residual variance) explained most of the variability in body mass. Congruently, heritability of mass was generally low across ontogeny, increasing at one year of age. As we predicted, maternal effects on offspring mass declined in the first few months, presumably because maternal non-genetic contributions were less influential on mass over time. Unexpectedly, maternal effects increased again at approximately six months possibly from maternal genetic factors affecting mass. Upon hatching, the residual variance component of body mass was much higher in lizards that were reared at cold incubation temperatures, suggesting that aspects of development environment played a bigger role in determining their hatching mass.

Thermal developmental plasticity in growth

In ectotherms, temperature plays a pervasive role in phenotypic development (Eyck et al., 2019; Noble et al., 2018; O’Dea et al., 2019; While et al., 2018). Contrary to other reptile studies, we did not show that growth rate differed between developmental temperatures. Some researchers reported increases in growth at higher incubation temperatures (Elphick & Shine, 1999; Hare et al., 2004; Verdú‐Ricoy et al., 2014), while have others found either the opposite result or no differences at all (Andrews et al., 2000; R. M. Goodman, 2008). The directionality of change is highly variable, even among studies of the same species (e.g. Acritoscincus duperreyi, Elphick & Shine, 1998, 1999; Flatt et al., 2001; Telemeco et al., 2010). Lack of generality may be related to how growth is statistically modelled. Very few studies account for individual variation in hatching mass or growth trajectories. Indeed, if we did not account for among individual variance in our models, significant treatment differences in growth 131

can be detected (Table S10). We emphasise the importance of partitioning confounding sources of variance such as individual or clutch effects as they can misconstrue conclusions about developmental impacts on later life phenotypes. Moreover, future studies should make use of all repeated measures of mass instead of averaging across individuals as the former approach not only increases statistical power but also provide more accurate estimates of growth.

Consistent with other squamates, we found that cold incubation treatment group attained higher hatching mass compared to their hot counterparts (Dayananda et al., 2016; Downes & Shine, 1999; Flatt et al., 2001; B. A. Goodman et al., 2013). These results support the temperature-size-rule whereby organisms reared in cold temperatures tend to have larger body sizes (Angilletta Jr et al., 2017). Larger hatching size can be achieved through prolonged development at cooler temperatures during embryonic stages (Forster & Hirst, 2012). It is well known that cold developmental temperatures results in longer incubation periods in many reptiles (Booth, 2006; Dayananda et al., 2016; Downes & Shine, 1999; Elphick & Shine, 1998; R. M. Goodman, 2008). Longer developmental time may allow embryos to assimilate yolk nutrients more efficiently thus increasing mass at hatching (Storm & Angilletta, 2007). Indeed, turtle embryos exposed to high temperatures have enhanced mitochondrial metabolism and metabolic enzymic activity which constrained developmental time and reduced overall hatching size (Ji et al., 2003; Sun et al., 2015). Thermal plasticity in embryonic development may be adaptive for lizards born late in the season when nest temperatures are generally colder (Warner & Shine, 2008; While et al., 2015). Indeed, female L. delicata have an extended oviposition period (September to February in our population) and nest temperatures during this time can be highly variable in the wild (Cheetham et al., 2011). Heavier weight at emergence could mean that hatchlings are in better condition to compete with lizards that hatched earlier or have sufficient body reserves to survive harsher condtions in more seasonal environments (Downes & Shine, 1999; Gifford et al., 2017; Qualls & Shine, 2000). Understanding how body mass affects survival will be necessary to elucidate the adaptative potential of developmentally plastic responses in the wild.

132

Thermal developmental environments and the evolutionary potential of body mass

Adaptative evolutionary responses depend not only on the amount of selection operating on a trait but on also its underlying additive genetic variance (Falconer, 1952; Ghalambor et al., 2007; Hoffmann & Merilä, 1999). Stressful developmental environments are hypothesized to lead to the release of ‘cryptic’ genetic variation (Fischer et al., 2020; Noble et al., 2019; Rowiński & Rogell, 2017; Wood & Brodie, 2015), possibly increasing the evolutionary potential of a given trait. Higher genetic variation, combined with stronger selection may facilitate rapid evolutionary responses that may allow populations to adapt to novel environments (Hoffmann & Merilä, 1999; Falconer and Mackay 1996). Contrary to these hypotheses, we found no statistical differences in additive genetic variance for mass between our developmental temperature treatments. In fact, heritability for mass was overall quite low echoing heritability values for mass in various animal systems [e.g., bighorn sheep – 0.03 to 0.31 (Réale et al., 1999), macaques – 0.39 (Kimock et al., 2019) lizards – 0 to 0.54 – (Martins et al., 2019; Noble et al., 2014)]. It should be noted that decoupling additive genetic variances from other non-genetic variance such as maternal effects requires considerable paternal links in the study design and pedigree (Kruuk, 2004). Indeed, when this variance partitioning is done accordingly, heritability estimates are often low (e.g., Noble et al. 2014). In the case of our study, we found relatively low levels of multiple paternity (<1% of clutches were sired by multiple fathers), as such the number of half-sibs were generally low which may have affected our genomic relatedness matrix and estimates of quantitative genetic parameters.

The lack of difference in genetic variation between developmental temperatures environments support findings from recent meta-analyses. Fisher et al. (2020) assessed the degree to which stressful thermal environments result in the release of genetic variation. They found that these effects manifested in only a third of the studied cases – in mainly clonal organisms (Fischer et al., 2020). Furthermore, of the 25 cases where genetic variance changed across thermal environments there was no consistent direction (i.e., 11 increased and 14 decreased under thermal stress). Noble et al. (2019) also showed that the release

133

of ‘cryptic’ genetic variation depends on the study design – studies not able to partition out non-genetic sources of variation supported a release of genetic variation whereas studies that did showed the opposite pattern. As a caveat, defining an environment as stressful or novel is a difficult task which requires detailed knowledge of a given species’ past environmental exposure – information that is often unknown (Roelofs et al., 2010). While our incubation temperatures were selected based on temperature extremes of naturally occurring L. delicata nests (Cheetham et al., 2011), it is nonetheless possible they were not ‘stressful’ from an evolutionary perspective. Indeed, egg mortality did not differ across incubation treatments which suggests that lizards from both treatments experienced a similar level of thermal stress as embryos (the estimate of treatment difference: 0.80 [-0.04 -1.73]). Furthermore, treatment differences may be harder to detect under realistic fluctuating temperature regimes. As such, lizards were not exposed to extreme temperatures over extended periods which might be more important in orchestrating changes in genetic variation (Bonamour et al., 2019). Overall, our results suggest that the thermal extremes experienced by natural nest sites do not modify the evolutionary potential of mass. However this should be interpreted with caution as estimates of quantitative parameters from laboratory studies can differ from wild populations (Sgrò & Hoffmann, 2004; Weigensberg & Roff, 1996).

Ontogenetic changes in genetic and non-genetic contributions to body mass

Genetic contributions to body size are expected to vary throughout ontogeny (Lynch & Walsh, 1998). Selection pressures on body size are likely to increase at critical life stages, such as at birth or at sexual maturation, thereby reducing genetic variance at certain ages (Rollinson & Rowe, 2015). On the contrary, we found that additive genetic variance of mass was very low upon hatching but slowly increased by the end of the first year. This result parallels those seen in big horn sheep (Réale et al., 1999), soay sheep (Wilson et al., 2007) and ladybird beetles (Dmitriew et al., 2010). While the underlying cause of this pattern is not well established, it coincided with changes in the social environment (shared housing). This suggests that perhaps competition for resources (basking sites or food) may orchestrate changes in genetic variation (Dmitriew et al., 2010; Hoffmann & Merilä, 1999). Alternatively, the gradual increase in additive

134

genetic variance may be related to initial genotypic changes underpinning sexual maturation (~14 months) as L.delicata are sexually dimorphic in various morphological traits including body size (Chapple et al., 2014). Nonetheless, ontogenetic variation in genetic variance implies that potential rates of evolution varies with age (Houle, 1998), however this depends on non-genetic sources of variance as well.

Maternal non-genetic contributions to offspring body size are expected to be highest during early life stages and decline as offspring mature, particularly in precocial species (Cheverud, 1984; Wilson, Kruuk, et al., 2005). In accordance with other studies, maternal effects did in fact decline after hatching (Dmitriew et al., 2010; Lindholm et al., 2006; Pick et al., 2016; Wilson, Coltman, et al., 2005; Wilson, Kruuk, et al., 2005). Maternal investment, such as investment in clutch number or egg quality, has been shown to influence hatching size in lizards (Brown & Shine, 2009; Noble et al., 2014; Warner & Lovern, 2014), however, as predicted these effects dissipated post-hatching (Pick et al., 2016; Réale et al., 1999). Interestingly, maternal contributions increased at a later age and remained moderately low for the remainder of the study. The cause of resurgence in maternal effect variance is unclear, however this pattern may indicate other maternally inherited components such as maternal genetic effects (e.g., mitochondrial genetic variation) that promote variation in body size (Pick et al., 2016). Indeed, variation in mitochondrial function has been linked to an individual’s metabolic rate and growth – explaining as much as ~50% of the variation in food intake and growth (Salin et al., 2016, 2019). Therefore, it is likely an important driver of body size variability. Similar to additive genetic variance, resurgence of maternal effects also cooccurred with changes in the shared environment (housing conditions), suggesting that maternal effects on offspring body size is likely to be environmentally driven.

Traits under strong selection are expected to show low evolutionary potential as selection acts to remove genetic variation. While low evolutionary potential is at least in part due to reduced levels of additive genetic variance, it is also a result of larger proportions of environmental variance. In our study, the environmental component of the phenotype accounted for over 80% of variation in body mass which is in line with values reported in great tits (53 –

135

74%) and soay sheep (70 – 96%) (Noordwijk et al., 1988; Wilson et al., 2007). Interestingly, cool developmental temperatures increased the amount of environmental variance attributed to body mass at an early age. What mechanisms are comprised in this environmental component? Variation in developmental period between developmental temperatures may explain these differences. In many ectotherms, developmental time exhibits a nonlinear reaction norm with temperature (Marshall et al., 2020; Noble et al., 2018).This means that developmental time decelerates with temperature following an negative exponential function. As a result, hot incubated lizards are more constrained in their development time compared to lizards that were reared a cooler temperature. In actual fact, the cold developmental temperature treatment had much greater variance in incubation duration. With a longer incubation period, embryos can maximise the yolk resources left by their mothers which can vary considerably within clutch (Wallace et al., 2007). Our results suggest that thermodynamic effects of development time can give rise greater environmental heterogeneity in hatching mass and may affect potential for evolution at early life stages.

Conclusion

Our work illustrates the pervasive role of developmental temperature on phenotypic variation. The impact of developmental temperature on body mass manifested early and persisted through life (Monaghan, 2008). This has profound implications as developmentally induced variation in body mass may drive life history differences within populations and alter their vulnerability to environmental change (Botero et al., 2015; Marshall et al., 2020; Reed et al., 2010). In contrast, genetic variance of body mass was robust to thermal extremes experienced by natural nests and suggests that the potential to genetically adapt to warming climate may be limited. However, more stressful incubation temperatures are needed to elucidate the capacity for this species to reveal new genetic material for selection to act on. Non-genetic sources of variance were responsible for most of the variability in body mass and their dynamics with age means that effectiveness of evolution is everchanging. Understanding the complexities of adaptive evolution in response to climate change may require intensive long-term studies in wild populations.

136

Data accessibility

Datasets and code used to generate results of this study is accessible via Open Science Framework (https://bit.ly/3lIuBk8)

Acknowledgements

This study would not be possible without the assistance of many volunteers and interns from Lizard Lab. We first want to thank Martin Whiting for the use of his facilities at Macquarie University. We are grateful for Birgit Szabo and Christine Wilson who helped check and process eggs when we were not available to do so. Thank you Joshua Cunningham, Victor Frichot and Matthieu Monserand for their commitment to the projects as interns. Special thanks for Julia Riley, Carlos Pavón, Damien Esquerre, Ian Brennan and Scott Keogh for their advice with SNP data.

137

References

Andrews, R. M., Mathies, T., & Warner, D. A. (2000). Effect of incubation temperature on morphology, growth, and survival of juvenile Sceloporus undulatus. Herpetological Monographs, 14, 420–431. JSTOR. https://doi.org/10.2307/1467055 Angilletta Jr, M. J., Steury, T. D., & Sears, M. W. (2017). Temperature, growth rate, and body size in ectotherms: Fitting pieces of a life-history puzzle. Integrative and Comparative Biology, 44(6), 1–12. Beaman, J. E., White, C. R., & Seebacher, F. (2016). Evolution of plasticity: Mechanistic link between development and reversible acclimation. Trends Ecology and Evolution, 31(3), 237–249. https://doi.org/10.1016/j.tree.2016.01.004 Beldade, P., Mateus, A. R. A., & Keller, R. A. (2011). Evolution and molecular mechanisms of adaptive developmental plasticity. Molecular Ecology, 20(7), 1347–1363. https://doi.org/10.1111/j.1365-294X.2011.05016.x Bérénos, C., Ellis, P. A., Pilkington, J. G., & Pemberton, J. M. (2014). Estimating quantitative genetic parameters in wild populations: A comparison of pedigree and genomic approaches. Molecular Ecology, 23(14), 3434–3451. https://doi.org/10.1111/mec.12827 Bonamour, S., Chevin, L.-M., Charmantier, A., & Teplitsky, C. (2019). Phenotypic plasticity in response to climate change: The importance of cue variation. Philosophical Transactions of the Royal Society B: Biological Sciences, 374(1768), 20180178–12. https://doi.org/10.1098/rstb.2018.0178 Booth, D. T. (2006). Influence of incubation temperature on hatchling phenotype in reptiles. Physiological and Biochemical Zoology, 79(2), 274–281. https://doi.org/10.1086/499988 Botero, C. A., Weissing, F. J., Wright, J., & Rubenstein, D. R. (2015). Evolutionary tipping points in the capacity to adapt to environmental change. Proceedings of the National Academy of Sciences, 112(1), 184–189. https://doi.org/10.1073/pnas.1408589111 Brown, G. P., & Shine, R. (2009). Beyond size–number trade-offs: Clutch size as a maternal effect. Philosophical Transactions of the Royal Society B: Biological Sciences, 364(1520), 1097–1106. https://doi.org/10.1098/rstb.2008.0247

138

Bürkner, P. C. (2017). brms: An R package for Bayesian multilevel models using Stan. Journal of Statistical Software, 80(1), 1–28. https://doi.org/10.18637/jss.v080.i01 Chapple, D. G., Miller, K. A., Chaplin, K., Barnett, L., Thompson, M. B., & Bray, R. D. (2014). Biology of the invasive delicate skink (Lampropholis delicata) on Lord Howe Island. Australian Journal of Zoology, 62(6), 498–506. https://doi.org/10.1071/ZO14098 Charmantier, A., & Garant, D. (2005). Environmental quality and evolutionary potential: Lessons from wild populations. Proceedings of the Royal Society B: Biological Sciences, 272(1571), 1415–1425. https://doi.org/10.1098/rspb.2005.3117 Charmantier, A., Kruuk, L. E. B., Blondel, J., & Lambrechts, M. M. (2004). Testing for microevolution in body size in three blue tit populations. Journal of Evolutionary Biology, 17(4), 732–743. https://doi.org/10.1111/j.1420-9101.2004.00734.x Cheetham, E., Doody, J. S., Stewart, B., & Harlow, P. (2011). Embryonic mortality as a cost of communal nesting in the delicate skink. Journal of Zoology, 283(4), 234–242. https://doi.org/10.1111/j.1469- 7998.2010.00764.x Cheverud, J. M. (1984). Evolution by kin selection: A quantitative genetic model illustrated by maternal performance in mice. Evolution, 38(4), 766–777. https://doi.org/10.2307/2408388 Coltman, D. W., Pilkington, J., Kruuk, L. E. B., Wilson, K., & Pemberton, J. M. (2001). Positive genetic correlation between parasite resistance and body size in a free-living ungulate population. Evolution, 55(10), 2116–2125. https://doi.org/10.1111/j.0014-3820.2001.tb01326.x Core Team, R. (2013). Team (2012). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Dahlgaard, J., & Hoffmann, A. A. (2000). Stress resistance and environmental dependency of inbreeding depression in Drosophila melanogaster. Conservation Biology, 14(4), 1187–1192. https://doi.org/10.1046/j.1523- 1739.2000.99206.x Dayananda, B., Gray, S., Pike, D., & Webb, J. K. (2016). Communal nesting under climate change: Fitness consequences of higher incubation

139

temperatures for a nocturnal lizard. Global Change Biology, 22(7), 2405– 2414. https://doi.org/10.1111/gcb.13231 Dmitriew, C., Blows, M. W., & Rowe, L. (2010). Ontogenetic change in genetic variance in size depends on growth environment. The American Naturalist, 175(6), 640–649. https://doi.org/10.1086/652470 Downes, S. J., & Shine, R. (1999). Do incubation-induced changes in a lizard’s phenotype influence its vulnerability to predators? Oecologia, 120(1), 9– 18. https://doi.org/10.1007/s004420050827 Elphick, M. J., & Shine, R. (1998). Long-term effects of incubation temperatures on the morphology and locomotor performance of hatchling lizards (Bassiana duperreyi, Scincidae). Biological Journal of the Linnean Society, 63(3), 429–447. https://doi.org/10.1111/j.1095-8312.1998.tb01527.x Elphick, M. J., & Shine, R. (1999). Sex differences in optimal incubation temperatures in a scincid lizard species. Oecologia, 118(4), 431–437. https://doi.org/10.1007/s004420050745 Elshire, R. J., Glaubitz, J. C., Sun, Q., Poland, J. A., Kawamoto, K., Buckler, E. S., & Mitchell, S. E. (2011). A robust, simple genotyping-by-sequencing (gbs) approach for high diversity species. PLoS ONE, 6(5), 10. Eyck, H. J. F., Buchanan, K. L., Crino, O. L., & Jessop, T. S. (2019). Effects of developmental stress on animal phenotype and performance: A quantitative review. Biological Reviews, 94(3), 1143–1160. https://doi.org/10.1111/brv.12496 Falconer, D. S. (1952). The problem of environment and selection. The American Naturalist, 86(830), 293–298. Falconer, D. S., & Mackay, T. F. C. (1996). Introduction to Quantitative Genetics (4th ed.). Pearson Education. Fischer, K., Kreyling, J., Beaulieu, M., Beil, I., Bog, M., Bonte, D., Holm, S., Knoblauch, S., Koch, D., Muffler, L., Mouginot, P., Paulinich, M., Scheepens, J. F., Schiemann, R., Schmeddes, J., Schnittler, M., Uhl, G., van der Maaten-Theunissen, M., Weier, J. M., … Gienapp, P. (2020). Species- specific effects of thermal stress on the expression of genetic variation across a diverse group of plant and animal taxa under experimental conditions. Heredity, 126(1), 1–15. https://doi.org/10.1038/s41437-020- 0338-4

140

Flatt, T., Shine, R., Borges-landaez, P. A., & Downes, S. J. (2001). Phenotypic variation in an oviparous montane lizard (Bassiana duperreyi): The effects of thermal and hydric incubation environments. Biological Journal of the Linnean Society, 74(3), 339–350. https://doi.org/10.1006/bijl.2001.0581 Forster, J., & Hirst, A. G. (2012). The temperature-size rule emerges from ontogenetic differences between growth and development rates. Functional Ecology, 26(2), 483–492. https://doi.org/10.1111/j.1365- 2435.2011.01958.x Gelman, A., Lee, D., & Guo, J. (2015). Stan: A probabilistic programming language for Bayesian inference and optimization. Journal of Educational and Behavioral Statistics, 40(5), 530–543. https://doi.org/10.3102/1076998615606113 Ghalambor, C. K., McKay, J. K., Carroll, S. P., & Reznick, D. N. (2007). Adaptive versus non-adaptive phenotypic plasticity and the potential for contemporary adaptation in new environments. Functional Ecology, 21(3), 394–407. https://doi.org/10.1111/j.1365-2435.2007.01283.x Gifford, M. E., Robinson, C. D., & Clay, T. A. (2017). The influence of invasive fire ants on survival, space use, and patterns of natural selection in juvenile lizards. Biological Invasions, 19(5), 1461–1469. https://doi.org/10.1007/s10530-017-1370-z Goodman, B. A., Schwarzkopf, L., & Krockenberger, A. K. (2013). Phenotypic integration in response to incubation environment adaptively influences habitat choice in a tropical lizard. The American Naturalist, 182(5), 666– 673. https://doi.org/10.1086/673299 Goodman, R. M. (2008). Latent effects of egg incubation temperature on growth in the lizard Anolis carolinensis. Journal of Experimental Zoology Part A: Ecological Genetics and Physiology, 309(9), 525–533. https://doi.org/10.1002/jez.483 Granato, I. S. C., Galli, G., de Oliveira Couto, E. G., e Souza, M. B., Mendonça, L. F., & Fritsche-Neto, R. (2018). snpReady: A tool to assist breeders in genomic analysis. Molecular Breeding, 38(8), 102. https://doi.org/10.1007/s11032-018-0844-8 Gruber, B., Unmack, P. J., Berry, O. F., & Georges, A. (2018). dartr: An r package to facilitate analysis of SNP data generated from reduced representation

141

genome sequencing. Molecular Ecology Resources, 18(3), 691–699. https://doi.org/10.1111/1755-0998.12745 Hare, K. M., Longson, C. G., Pledger, S., & Daugherty, C. H. (2004). Size, growth, and survival are reduced at cool incubation temperatures in the temperate lizard Oligosoma suteri (Lacertilia: Scincidae). Copeia, 2004(2), 383–390. https://doi.org/10.1643/CP-03-084R2 Hoffman, A. A., & Parsons, P. A. (1991). Evolutionary genetics and evolutionary stress. Oxford University Press. Hoffmann, A. A., & Merilä, J. (1999). Heritable variation and evolution under favourable and unfavourable conditions. Trends in Ecology & Evolution, 14(3), 96–101. https://doi.org/10.1016/S0169-5347(99)01595-5 Houle, D. (1998). How should we explain variation in the genetic variance of traits? Genetica, 102(0), 241. https://doi.org/10.1023/A:1017034925212 Huisman, J. (2017). Pedigree reconstruction from SNP data: Parentage assignment, sibship clustering and beyond. Molecular Ecology Resources, 17(5), 1009–1024. https://doi.org/10.1111/1755-0998.12665 Ji, X., Chen, F., Du, W.-G., & Chen, H.-L. (2003). Incubation temperature affects hatchling growth but not sexual phenotype in the Chinese soft-shelled turtle, Pelodiscus sinensis (Trionychidae). Journal of Zoology, 261(4), 409– 416. https://doi.org/10.1017/S0952836903004266 Kilian, A., Wenzl, P., Huttner, E., Carling, J., Xia, L., Blois, H., Caig, V., Heller- Uszynska, K., Jaccoud, D., Hopper, C., Aschenbrenner-Kilian, M., Evers, M., Peng, K., Cayla, C., Hok, P., & Uszynski, G. (2012). Diversity Arrays Technology: A Generic Genome Profiling Technology on Open Platforms. In F. Pompanon & A. Bonin (Eds.), Data Production and Analysis in Population Genomics: Methods and Protocols (pp. 67–89). Humana Press. https://doi.org/10.1007/978-1-61779-870-2_5 Kimock, C. M., Dubuc, C., Brent, L. J. N., & Higham, J. P. (2019). Male morphological traits are heritable but do not predict reproductive success in a sexually-dimorphic primate. Scientific Reports, 9(1), 19794. https://doi.org/10.1038/s41598-019-52633-4 Krist, M. (2010). Egg size and offspring quality: A meta-analysis in birds. Biological Reviews, 86(3), 692–716. https://doi.org/10.1111/j.1469- 185X.2010.00166.x

142

Kruuk, L. E. B. (2004). Estimating genetic parameters in natural populations using the ‘animal model’. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences, 359(1446), 873–890. https://doi.org/10.1098/rstb.2003.1437 Lindholm, A. K., Hunt, J., & Brooks, R. (2006). Where do all the maternal effects go? Variation in offspring body size through ontogeny in the live- bearing fish Poecilia parae. Biology Letters, 2(4), 586–589. https://doi.org/10.1098/rsbl.2006.0546 Lynch, M., & Walsh, B. (1998). Genetics and analysis of quantitative traits. Oxford University Press. Marshall, D. J., Pettersen, A. K., Bode, M., & White, C. R. (2020). Developmental cost theory predicts thermal environment and vulnerability to global warming. Nature Ecology & Evolution, 4(3), 406–411. https://doi.org/10.1038/s41559-020-1114-9 Martins, F., Kruuk, L. E. B., Llewelyn, J., Moritz, C., & Phillips, B. (2019). Heritability of climate-relevant traits in a rainforest skink. Heredity, 122(1), 41–52. https://doi.org/10.1038/s41437-018-0085-y Mitchell, T. S., Hall, J. M., & Warner, D. A. (2018). Female investment in offspring size and number shifts seasonally in a lizard with single-egg clutches. Evolutionary Ecology, 32(2), 231–245. https://doi.org/10.1007/s10682-018-9936-5 Monaghan, P. (2008). Early growth conditions, phenotypic development and environmental change. Philosophical Transactions of the Royal Society B: Biological Sciences, 363(1497), 1635–1645. https://doi.org/10.1098/rstb.2007.0011 Mousseau, T. A., & Fox, C. W. (1998). The adaptive significance of maternal effects. Trends Ecology and Evolution, 13(10), 403–407. https://doi.org/10.1016/S0169-5347(98)01472-4 Noble, D. W. A., McFarlane, S. E., Keogh, J. S., & Whiting, M. J. (2014). Maternal and additive genetic effects contribute to variation in offspring traits in a lizard. Behavioral Ecology, 25(3), 633–640. https://doi.org/10.1093/beheco/aru032 Noble, D. W. A., Radersma, R., & Uller, T. (2019). Plastic responses to novel environments are biased towards phenotype dimensions with high

143

additive genetic variation. Proceedings of the National Academy of Sciences, 116(27), 13452–13461. https://doi.org/10.1073/pnas.1821066116 Noble, D. W. A., Stenhouse, V., & Schwanz, L. E. (2018). Developmental temperatures and phenotypic plasticity in reptiles: A systematic review and meta-analysis. Biological Reviews, 93(1), 72–97. https://doi.org/10.1111/brv.12333 Noordwijk, A. J. V., Balen, J. H. V., & Scharloo, W. (1988). Heritability of body size in a natural population of the Great Tit (Parus major) and its relation to age and environmental conditions during growth. Genetical Research, 51(2), 149–162. https://doi.org/10.1017/S0016672300024162 O’Dea, R. E., Lagisz, M., Hendry, A. P., & Nakagawa, S. (2019). Developmental temperature affects phenotypic means and variability: A meta-analysis of fish data. Fish and Fisheries, 20(5), 1005–1022. https://doi.org/10.1111/faf.12394 Paaby, A. B., & Rockman, M. V. (2014). Cryptic genetic variation: Evolution’s hidden substrate. Nature Reviews Genetics, 15(4), 247–258. https://doi.org/10.1038/nrg3688 Pick, J. L., Ebneter, C., Hutter, P., & Tschirren, B. (2016). Disentangling genetic and prenatal maternal effects on offspring size and survival. The American Naturalist, 188(6), 628–639. https://doi.org/10.1086/688918 Qualls, F. J., & Shine, R. (2000). Post-hatching environment contributes greatly to phenotypic variation between two populations of the Australian garden skink, Lampropholis guichenoti. Biological Journal of the Linnean Society, 71(2), 315–341. https://doi.org/10.1006/bijl.2000.0445 Réale, D., Festa-Bianchet, M., & Jorgenson, J. T. (1999). Heritability of body mass varies with age and season in wild bighorn sheep. Heredity, 83(5), 526–532. https://doi.org/10.1046/j.1365-2540.1999.00543.x Reed, T. E., Waples, R. S., Schindler, D. E., Hard, J. J., & Kinnison, M. T. (2010). Phenotypic plasticity and population viability: The importance of environmental predictability. Proceedings of the Royal Society of London B: Biological Sciences, 277(1699), 3391–3400. https://doi.org/10.1098/rspb.2010.0771 Roelofs, D., Morgan, J., & Stürzenbaum, S. (2010). The significance of genome- wide transcriptional regulation in the evolution of stress tolerance.

144

Evolutionary Ecology, 24(3), 527–539. https://doi.org/10.1007/s10682-009- 9345-x Rollinson, N., & Rowe, L. (2015). Persistent directional selection on body size and a resolution to the paradox of stasis. Evolution, 69(9), 2441–2451. https://doi.org/10.1111/evo.12753 Rowiński, P. K., & Rogell, B. (2017). Environmental stress correlates with increases in both genetic and residual variances: A meta-analysis of animal studies. Evolution, 71(5), 1339–1351. https://doi.org/10.1111/evo.13201 Rueden, C. T., Schindelin, J., Hiner, M. C., DeZonia, B. E., Walter, A. E., Arena, E. T., & Eliceiri, K. W. (2017). ImageJ2: ImageJ for the next generation of scientific image data. BMC Bioinformatics, 18(1), 529. https://doi.org/10.1186/s12859-017-1934-z Salin, K., Auer, S. K., Anderson, G. J., Selman, C., & Metcalfe, N. B. (2016). Inadequate food intake at high temperatures is related to depressed mitochondrial respiratory capacity. The Journal of Experimental Biology, 219(9), 1356–1362. https://doi.org/10.1242/jeb.133025 Salin, K., Villasevil, E. M., Anderson, G. J., Lamarre, S. G., Melanson, C. A., McCarthy, I., Selman, C., & Metcalfe, N. B. (2019). Differences in mitochondrial efficiency explain individual variation in growth performance. Proceedings of the Royal Society B: Biological Sciences, 286(1909), 1–8. https://doi.org/10.1098/rspb.2019.1466 Sgrò, C. M., & Hoffmann, A. A. (2004). Genetic correlations, tradeoffs and environmental variation. Heredity, 93(3), 241–248. https://doi.org/10.1038/sj.hdy.6800532 Shine, R., & Harlow, P. S. (1996). Maternal manipulation of offspring phenotypes via nest-site selection in an oviparous lizard. Ecology, 77(6), 1808–1817. https://doi.org/10.2307/2265785 Stillwell, R. C., & Fox, C. W. (2009). Geographic variation in body size, sexual size dimorphism and fitness components of a seed beetle: Local adaptation versus phenotypic plasticity. Oikos, 118(5), 703–712. https://doi.org/10.1111/j.1600-0706.2008.17327.x Storm, M. A., & Angilletta, M. J. (2007). Rapid assimilation of yolk enhances growth and development of lizard embryos from a cold environment.

145

The Journal of Experimental Biology, 210(19), 3415–3421. https://doi.org/10.1242/jeb.005652 Sun, B.-J., Li, T., Gao, J., Ma, L., & Du, W.-G. (2015). High incubation temperatures enhance mitochondrial energy metabolism in reptile embryos. Scientific Reports, 5(1), 8861. https://doi.org/10.1038/srep08861 Telemeco, R. S., Radder, R. S., Baird, T. A., & Shine, R. (2010). Thermal effects on reptile reproduction: Adaptation and phenotypic plasticity in a montane lizard. Biological Journal of the Linnean Society, 100(3), 642–655. https://doi.org/10.1111/j.1095-8312.2010.01439.x Uller, T., & Olsson, M. (2010). Offspring size and timing of hatching determine survival and reproductive output in a lizard. Oecologia, 162(3), 663–671. https://doi.org/10.1007/s00442-009-1503-x VanRaden, P. M. (2008). Efficient methods to compute genomic predictions. Journal of Dairy Science, 91(11), 4414–4423. https://doi.org/10.3168/jds.2007-0980 Verdú-Ricoy, J., Iraeta, P., Salvador, A., & Díaz, J. A. (2014). Phenotypic responses to incubation conditions in ecologically distinct populations of a lacertid lizard: A tale of two phylogeographic lineages. Journal of Zoology, 292(3), 184–191. https://doi.org/10.1111/jzo.12091 Wallace, B. P., Sotherland, P. R., Santidrian Tomillo, P., Reina, R. D., Spotila, J. R., & Paladino, F. V. (2007). Maternal investment in reproduction and its consequences in leatherback turtles. Oecologia, 152(1), 37–47. https://doi.org/10.1007/s00442-006-0641-7 Warner, D. A., & Lovern, M. B. (2014). The maternal environment affects offspring viability via an indirect effect of yolk investment on offspring size. Physiological and Biochemical Zoology, 87(2), 276–287. https://doi.org/10.1086/674454 Warner, D. A., & Shine, R. (2008). Determinants of dispersal distance in free- ranging juvenile lizards. Ethology, 114(4), 361–368. https://doi.org/10.1111/j.1439-0310.2008.01475.x Weigensberg, I., & Roff, D. A. (1996). Natural heritabilities: Can they be reliably estimated in the laboratory? Evolution, 50(6), 2149–2157. https://doi.org/10.1111/j.1558-5646.1996.tb03605.x West-Eberhard, M. J. (2003). Developmental Plasticity and Evolution. Oxford University Press.

146

While, G. M., Noble, D. W. A., Uller, T., Warner, D. A., Riley, J. L., Du, W.-G., & Schwanz, L. E. (2018). Patterns of developmental plasticity in response to incubation temperature in reptiles. Journal of Experimental Zoology Part A: Ecological and Integrative Physiology, 329(4–5), 162–176. https://doi.org/10.1002/jez.2181 While, G. M., Williamson, J., Prescott, G., Horvathova, T., Fresnillo, B., Beeton, N. J., Halliwell, B., Michaelides, S., & Uller, T. (2015). Adaptive responses to cool climate promotes persistence of a non-native lizard. Proceedings of the Royal Society of London B: Biological Sciences, 282(1803), 20142638– 20142638. https://doi.org/10.1098/rspb.2014.2638 Wilson, A. J., Coltman, D. W., Pemberton, J. M., Overall, A. D. J., Byrne, K. A., & Kruuk, L. E. B. (2005). Maternal genetic effects set the potential for evolution in a free-living vertebrate population. Journal of Evolutionary Biology, 18(2), 405–414. https://doi.org/10.1111/j.1420-9101.2004.00824.x Wilson, A. J., Kruuk, L. E. B., & Coltman, D. W. (2005). Ontogenetic patterns in heritable variation for body size: Using random regression models in a wild ungulate population. The American Naturalist, 166(6), E177–E192. https://doi.org/10.1086/497441 Wilson, A. J., Pemberston, J. M., Pilkington, J. G., Clutton-Brock, T. H., Coltman, D. W., & Kruuk, L. E. B. (2007). Quantitative genetics of growth and cryptic evolution of body size in an island population. Evolutionary Ecology, 21(3), 337–356. https://doi.org/10.1007/s10682-006-9106-z Wilson, A. J., & Réale, D. (2006). Ontogeny of Additive and Maternal Genetic Effects: Lessons from Domestic Mammals. The American Naturalist, 167(1), E23–E38. https://doi.org/10.1086/498138 Wilson, A. J., Reale, D., Clements, M. N., Morrissey, M. M., Postma, E., Walling, C. A., Kruuk, L. E. B., & Nussey, D. H. (2010). An ecologist’s guide to the animal model. Journal of Animal Ecology, 79(1), 13–26. https://doi.org/10.1111/j.1365-2656.2009.01639.x Wood, C. W., & Brodie, E. D. (2015). Environmental effects on the structure of the G-matrix. Evolution, 69(11), 2927–2940. https://doi.org/10.1111/evo.12795

147

Supplementary Materials

Pedigree and genomic relatedness

We submitted a total of 437 tissues samples, five samples experienced problems during extraction and sequencing and were therefore excluded from the final dataset (n = 432). DNA was extracted from tissue samples using a Qiagen DNeasy Blood and Tissue Kits following the manufacturer’s instructions. Diversity Arrays Technology (DArT) combines next generation sequencing platforms and genome complexity reduction methods to select the most appropriate method for L.delicata. Four methods of complexity reduction were tested in our study species (data not presented) and the PstI-HpaII method was selected. DNA samples were processed in digestion/ligation reactions principally following the methods of (Kilian et al., 2012), but replacing a single PstI-compatible adaptor with two different adaptors each corresponding to different Restriction Enzyme (RE) overhangs. The PstI-compatible adapter was designed to include Illumina flowcell attachment sequence, sequencing primer sequence and “staggered”, varying length barcode region similar to the sequence reported by (Elshire et al., 2011). The reverse adapter contained a flowcell attachment region and HpaII-compatible overhang sequence. Only “mixed fragments” (PstI-HpaII) were effectively amplified in 30 rounds of PCR using the following reaction conditions: 1. 94̊ C for 1 min; 2. 30 cycles of 94̊ C for 20 sec 58̊ C for 30 sec 72̊ C for 45 sec; 3. 72̊ C for 7 min. After PCR equimolar amounts of amplification products from each sample of the 96-well microtiter plate were bulked and applied to c-Bot (Illumina) bridge PCR followed by sequencing on Illumina Hiseq2500. The sequencing (single read) was run for 77 cycles. (EBPCRI primer: 5’- AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTC TTCCGATCT - 3’ and EBHpaIIpcr primer: 5’- CAAGCAGAAGACGGCATACGAGATCGGTCTCGGCATTCCTGCTGAACC GC TCTTCC GATCTCGG - 3’).

Sequences from all lanes were then processed using DArT specific pipelines. The main pipeline filtered out poor quality sequences. In that way the assignments of the sequences to specific samples carried in the “barcode split”

148

step are very reliable. Approximately 2500000 (+/- 7%) sequences per barcode/sample were used in marker calling in routine DArTseq assay but we applied a more cost effective version of the assay using half of the normal tag number (average of 1.3 million per sample). Finally, identical sequences were collapsed into “fastqcall files”. These files were used in the secondary pipeline for DArT PL’s proprietary SNP and SilicoDArT (presence/absence of restriction fragments in representation) calling algorithms (DArTsoft14). For our samples, this filtering process by DArT resulted in a total of 185,963 SNPs. One individual was excluded from the dataset and possibly due to contamination as this individual appeared to be unrelated to any other samples in the dataset.

Model fitting and selection of random effects structure

We fitted seven different models to investigate what random effects structure was best suited for our dataset. Only the intercept was included as fixed effects in these models and lizard identity was included twice to partition out permanent environmental effects (PE) given we had repeated measures of the same individuals (Wilson et al., 2010). Age was z-transformed. In model 1 we assume that additive genetic variance, maternal effect variance and permanent environmental variance was constant across age by fitting a random intercept for lizard identity (G), dam identity (M) and for permanent environmental effects (PE). In model 2, we assumed all variance components varied across age following a linear relationship and included random linear slopes for G, M, PE. In model 3, we assumed that all variance components changed with age in a quadratic fashion by including random quadratic slopes for G, M, PE. In model 4, we included random intercept for G and random linear slopes for M and PE. In model 5, we included random intercept for M and random linear slopes for G, PE. In model 6, we included random intercept for PE and random linear slopes for G and PE. Finally, in model 7, we included a random intercept for PE and random quadratic slopes for G and M. WAIC values are presented in Table S1. Model 1 has the highest WAIC value indicating that it is the least supported model. Model 2 was had the lowest WAIC value Model 1 has the highest WAIC value indicating that it is the least supported model. Model 3 was had the lowest WAIC value indicating that it is the best supported model. Between model 5 -6, model 6 had the highest ∆DIC, indicating that PE should be included as a random intercept only. Model 7 was

149

the second-best supported model and improvement in WAIC value was marginal by including a quadratic slope for PE. To avoid overfitting, To avoid overfitting, we selected the more parsimonious model (Model 7) and used this random effect structure for the remaining analysis unless stated otherwise. Table S1 WAIC of six intercept models with different combinations of random effects structures for additive genetic variance, maternal variance and permanent environmental variance. ∆DIC was calculated by subtracting the DIC value of each model from the DIC value of model 1. G M, PE represents additive genetic variance, maternal variance and permanent environment variance, respectively. Std. Error WAIC ∆ELPD Model ∆ELPD Model 3 - quadratic slopes for G, M, -3258 0 0 PE Model 7 - intercept for PE, quadratic -3261 -1.122 3.245 slopes for G, M Model 4 - intercept for G, linear slopes -1382 -939.4 38.91 for M, PE Model 2 - linear slopes for G, M, PE -1381 -939.5 39.03 Model 6 - intercept for PE, linear -1382 -940 38.94 slopes for G, M Model 5 - intercept for M, linear slopes -1370 -945.4 40.3 for G, PE Model 1 - intercepts for G, M, PE 4550 -3905 49.91

Does our data have heterogenous residual variance?

Residual variance may conflate with estimates of other variance components if it changes over time (heterogenous variance) and is not properly accounted for. We therefore explicitly modelled residual variance to verify if this was the case using WAIC values. We fitted two models, both of which had the same fixed and random effects structure as Model 7 described above. The first model had homogenous residual variance whereas in the second model we modelled residual variance with a linear slope thereby allowing it to vary with age. The model with heterogenous variance was best supported (Table S2), we therefore modelled heterogenous variance in all subsequent models unless stated otherwise Table S2 Comparisons of expected log predictive density values for WAIC values to test the importance of heterogenous and homogenous residual 150

variance. Note that difference in values are calculated by subtracting values from the model with lowest LOO and WAIC values i.e. the heterogenous variance model Std. Error Model WAIC ∆ELPD ∆ELPD Model 7 with heterogenous -3280 0 0 variance Model 7 with homogenous -3261 -9.783 6.429 variance

The influence of developmental temperature on genetic and non-genetic variance across age

We fitted random intercepts for �, �, �� and � to estimate the overall estimate across age. In other words, the average variance across all age classes. We found that additive genetic variance, permanent environmental variance and heritability of growth appears to be higher in the hot developmental temperature treatment however, there were no significant differences among treatment groups (Table S3).

Table S3 Treatment comparisons of additive genetic variance, maternal variance, permanent environmental variance, residual variance and heritability. These are estimated from a model where random intercepts were fitted for all variance components. Bolded estimates are significantly different from zero. Hot treatment group nobs = 1892, cold treatment group nobs = 2036. Values with * indicate very small values that were above zero. V represents variance. Hot developmental Cold developmental temperature temperature (nlizards = 125) (nlizards = 136) Estimate Lower Upper Estimate Lower Upper Vadditive genetic 0.028 0.001 0.056 0.01 0* 0.025

Vmaternal 0.006 0* 0.024 0.005 0* 0.016

Vpermanent environment 0.015 0* 0.053 0.008 0* 0.025

Vresidual 0.25 0.23 0.271 0.274 0.254 0.294 Heritability (h2) 0.041 0.002 0.083 0.013 0* 0.033

To test how variance components and heritability change with age in each treatment group, we fitted an intercept in our fixed effects and used the best 151

supported random effect structure (Model 7) with heterogenous residual variance. There were no differences among treatment groups (Fig. 3). The � and � matrices for each treatment group are presented in Table S3-S4.

Table S4. G and M variance-covariance / correlation matrices between growth trajectory parameters (intercept, linear slope and quadratic slope) for lizards from the hot developmental temperature treatment group (nlizards =

125, nobs = 1330). PE variance is also presented. Variances are represented along the diagonal, covariances are represented in the upper triangle and correlations are represented in the lower triangle. Bolded estimates are significantly different from zero. Values in the brackets represent the 95% credible intervals. Note that residual variance slope is in SD units. G Intercept Linear Slope Quadratic Slope 0.00163 0.008 -0.0817 Intercept (-0.0023 to (0.002 to 0.018) (-0.134 to -0.0447) 0.00611) Linear 0.151 0.013 -0.101 Slope (-0.287 to 0.534) (0.008 to 0.02) (-0.146 to -0.07)

Quadratic -0.516 0.636 0.011 Slope (-0.815 to -0.141) (0.356 to 0.819) (0.007 to 0.017)

M Intercept Linear Slope Quadratic Slope 0.156 0.19 -0.00738 Intercept (0.052 to 0.306) (0.102 to 0.305) (-0.0185 to -0.00205) Linear 0.976 0.252 -0.0113 Slope (0.932 to 0.997) (0.179 to 0.362) (-0.0225 to -0.00545) -0.927 Quadratic -0.973 0.047 (-0.992 to - Slope (-0.997 to -0.924) (0.031 to 0.071) 0.818) PE Residual Residual Slope (SD)

4.64 0.00463 (4.48 to -0.0237 (0.000792 to 0.00796) 0.021) (-0.06676 to -0.021)

152

Table S5. G and M variance-covariance / correlation matrices between growth trajectory parameters (intercept, linear slope and quadratic slope) for lizards from the cold developmental temperature treatment group (nlizards = 136, nobs = 1596). PE variance is also presented. Variances are represented along the diagonal, covariances are represented in the upper triangle and correlations are represented in the lower triangle. Bolded estimates are significantly different from zero. Values in the brackets represent the 95% credible intervals. Note that residual variance slope is in SD units. G Intercept Linear Slope Quadratic Slope 0.008 0.00132 -0.00488 Intercept (0.001 to (-0.00331 to (-0.0104 to -0.000589) 0.018) 0.00704) Linear 0.123 0.014 0.00422 Slope (-0.454 to 0.62) (0.008 to 0.023) (0.000231 to 0.00917) -0.646 Quadratic 0.411 0.008 (-0.931 to - Slope (0.0231 to 0.731) (0.004 to 0.013) 0.193) M Intercept Linear Slope Quadratic Slope 0.032 0.0718 -0.0227 Intercept (0 to 0.12) (-0.00275 to 0.183) (-0.0585 to 0.000568) 0.78 Linear 0.267 -0.078 (-0.327 to Slope (0.185 to 0.382) (-0.114 to -0.0517) 0.991) -0.794 Quadratic -0.933 0.026 (-0.992 to Slope (-0.994 to -0.811) (0.016 to 0.041) 0.244) PE Residual Residual Slope (SD) 0.065 4.15 -0.135 (6.55e-05 to 0.00938) (3.97 to 4.34) (-0.184 to -0.075)

Accounting for scale-effects using coefficients of variation

As the mean body mass increases over time, the variance may also increase concurrently due to scale effects and potentially bias estimates of quantitative genetics parameters (Wilson, Kruuk, et al., 2005). We therefore calculated coefficients of variation (CV) across age for each variance component using the following equation:

153

. 100 × � �� = � where �� is the CV for a given variance component e.g. �� is the CV of maternal effects. The CV of maternal effects (��) and �� showed the same pattern as the raw variance estimates (Fig. S1-2). There were no treatment differences in ��, �� (Fig. S2). Both of which followed a quadratic relationship like the raw variance estimates in Fig. S1. We therefore conclude changes in mean mass did not affect estimation of variance components.

a) b) G M 160 80

120 60

80 40 CV of Mass

20 40

0 0 60 120 180 240 300 360 420 480 0 60 120 180 240 300 360 420 480 Age

Figure S1 Scatterplot showing the relationship of the how the coefficient of variance of a) additive genetic variance (G) and b) maternal effects (M) changed with age for the hot developmental treatment (n = 125, red) and the cold developmental treatment (n = 136, blue). Points represent posterior means, thin lines represent the 95% credible intervals, thick lines represent the mean for each treatment group.

Do growth trajectories differ among incubation treatments?

To test for treatment differences in growth trajectories, we fitted four models with different combinations of treatment interactions with the linear and quadratic age parameters and compared their WAIC values (Table 1). The best supported model was the full model (Table S5), its G and M matrix is presented in Table S6 below. Model coefficients for the other three models are presented in Table S7-S9)

154

Table S6. G and M variance-covariance / correlation matrices between growth trajectory parameters (intercept, linear slope and quadratic slope) for the overall population (nlizards = 261, nobs = 2926). PE and Residual variance are also presented. Variances are represented along the diagonal, covariances are represented in the upper triangle and correlations are represented in the lower triangle. Bolded estimates are significantly different from zero. Values in the brackets represent the 95% credible intervals. Note that residual variance slope is in SD units. G Intercept Linear Slope Quadratic Slope 0.012 5.82e-05 -0.00724 Intercept (0.007 to (-0.00309 to (-0.0108 to -0.00416) 0.019) 0.00311) 0.00484 Linear 0.014 0.0096 (-0.237 to Slope (0.01 to 0.019) (0.00645 to 0.0139) 0.238) -0.559 Quadratic 0.689 0.014 (-0.751 to - Slope (0.549 to 0.81) (0.01 to 0.019) 0.362) M Intercept Linear Slope Quadratic Slope -5.11e-05 0.001 -0.000196 Intercept (-0.00175 to (0 to 0.004) (-0.00169 to 0.000277) 0.00165) -0.0555 Linear 0.005 5.78e-05 (-0.844 to Slope (0.002 to 0.01) (-0.00114 to 0.00176) 0.773) -0.181 Quadratic -0.0121 0 (-0.937 to Slope (-0.865 to 0.845) (0 to 0.002) 0.799) PE Residual Residual Slope (SD) 0.00333 4.36 -0.0483 (0.000317 to 0.00653) (4.23 to 4.48) (-0.0852 to -0.0105)

155

Table S7 Estimates from model with interaction between treatment and age (linear growth rate) only. Bolded estimates are significantly different from zero. * indicates that value is above zero prior to rounding. nobs = 2926. Age measured in days was z-transformed (mean = 361.34, SD = 185.16)

Parameter Estimate Lower Upper

Fixed effects

Intercept -0.995 -1.014 -0.977

Treatment -0.072 -0.101 -0.045

Age 0.507 0.482 0.531

Age2 -0.186 -0.199 -0.173

Treatment × Age -0.005 -0.028 0.018

Random effects

Maternal variance

Intercept 0.024 0.001 0.064

Slope 0.069 0.04 0.099

Quadratic 0.016 0.001 0.044

Cor intercept - slope -0.058 -0.845 0.776

Cor intercept - quadratic -0.157 -0.921 0.849

Cor slope - quadratic -0.06 -0.877 0.815

Permanent Environment Variance 0.055 0.017 0.08

Additive Genetic Variance

Intercept 0.109 0.08 0.137

Slope 0.116 0.1 0.135

Quadratic 0.117 0.101 0.137

Cor intercept - slope 0.014 -0.233 0.242

Cor intercept - quadratic -0.549 -0.742 -0.356

156

Cor slope - quadratic 0.686 0.542 0.804

Residual -2.087 -2.118 -2.057

Residual x Age -0.048 -0.084 -0.011

Table S8 Estimates from model with interaction between treatment and quadratic age parameter only. Bolded estimates are significantly different from zero. * indicates that value is above zero prior to rounding. nobs = 2926. Age measured in days was z-transformed (mean = 361.34, SD = 185.16). Cor represents correlation

Parameter Estimate Lower Upper

Fixed effects

Intercept -0.99 -1.009 -0.97

Age 0.504 0.482 0.525

Treatment -0.083 -0.116 -0.053

Age2 -0.194 -0.212 -0.177

Treatment x Age2 0.017 -0.007 0.042

Random effects

Maternal variance

Intercept 0.025 0.001 0.064

Slope 0.069 0.041 0.098

Quadratic 0.016 0.001 0.047

Cor intercept - slope -0.091 -0.868 0.765

Cor intercept - quadratic -0.175 -0.929 0.826

Cor slope - quadratic -0.047 -0.902 0.835

Permanent Environment Variance 0.056 0.02 0.081

Additive Genetic Variance

Intercept 0.109 0.077 0.135

Slope 0.116 0.099 0.135

157

Quadratic 0.118 0.102 0.135

Cor intercept - slope 0.006 -0.233 0.249

Cor intercept - quadratic -0.559 -0.755 -0.352

Cor slope - quadratic 0.687 0.544 0.8

Residual -2.087 -2.119 -2.055

Residual x Age -0.047 -0.084 -0.01

Table S9 Estimates from model with main effects of treatment and age only. Bolded estimates are significantly different from zero. * indicates that value is above zero prior to rounding. nobs = 2926. G is the additive genetic variance, Age measured in days was z-transformed (mean = 361.34, SD = 185.16).

Parameter Estimate Lower Upper

Fixed effects

Intercept -0.997 -1.015 -0.979

Treatment -0.069 -0.095 -0.046

Age 0.504 0.481 0.526

Age2 -0.186 -0.198 -0.172

Random effects

Maternal variance

Intercept 0.025 0.001 0.064

Slope 0.07 0.042 0.099

Quadratic 0.017 0.001 0.047

Cor intercept - slope -0.057 -0.871 0.811

Cor intercept - quadratic -0.163 -0.924 0.84

Cor slope - quadratic -0.02 -0.865 0.814

Permanent Environment Variance 0.055 0.017 0.081

Additive Genetic Variance

158

Intercept 0.109 0.077 0.136

Slope 0.115 0.098 0.134

Quadratic 0.117 0.101 0.134

Cor intercept - slope 0.021 -0.218 0.265

Cor intercept - quadratic -0.552 -0.737 -0.357

Cor slope - quadratic 0.68 0.529 0.794

Residual -2.087 -2.118 -2.057

Residual x Age -0.048 -0.085 -0.013

Table S10 Estimates from model testing the effects of incubation treatment on changes in mass over age. Note that individual variation is not accounted for in random effects, only variation among different mothers to account for non-independence among siblings. Bolded estimates are significantly different from zero. * indicates that value is above zero prior to rounding. nobs = 2926. Age measured in days was z- transformed (mean = 361.34, SD = 185.16). * indicates that value is very small but does not overlap zero prior to rounding

Parameter Estimate Lower Upper

Fixed effects

Intercept -0.77 -0.85 -0.68

Treatment -0.01 -0.01 -0.01

Age 0.63 0.57 0.7

Age2 -0.05 -0.1 0.01

Treatment × Age -0* -0.01 -0*

Treatment × Age2 -0* -0* -0*

Random effects

Maternal variance

Intercept 0.08 0.06 0.09

159

Residual 0.18 0.18 0.19

Supplementary Materials References

Elshire, R. J., Glaubitz, J. C., Sun, Q., Poland, J. A., Kawamoto, K., Buckler, E. S., & Mitchell, S. E. (2011). A Robust, Simple Genotyping-by-Sequencing (GBS) Approach for High Diversity Species. PLoS ONE, 6(5), 10. Kilian, A., Wenzl, P., Huttner, E., Carling, J., Xia, L., Blois, H., Caig, V., Heller- Uszynska, K., Jaccoud, D., Hopper, C., Aschenbrenner-Kilian, M., Evers, M., Peng, K., Cayla, C., Hok, P., & Uszynski, G. (2012). Diversity Arrays Technology: A Generic Genome Profiling Technology on Open Platforms. In F. Pompanon & A. Bonin (Eds.), Data Production and Analysis in Population Genomics: Methods and Protocols (pp. 67–89). Humana Press. https://doi.org/10.1007/978-1-61779-870-2_5 Wilson, A. J., Reale, D., Clements, M. N., Morrissey, M. M., Postma, E., Walling, C. A., Kruuk, L. E. B., & Nussey, D. H. (2010). An ecologist’s guide to the animal model. Journal of Animal Ecology, 79(1), 13–26. https://doi.org/10.1111/j.1365-2656.2009.01639.x

160

CHAPTER 5

What predicts pace-of-life? Distinguishing among multiple hypotheses in terrestrial ectotherms

Fonti Kar1, Shinichi Nakagawa1,2, Shai Meiri3,4, Daniel W.A. Noble5

1 School of Biological Earth and Environmental Sciences, Ecology and Evolution Research Centre, University of New South Wales, Sydney, NSW, Australia 2 Diabetes and Metabolism Division, Garvan Institute of Medical Research, 384 Victoria Street, Darlinghurst, Sydney, NSW 2010, Australia 3 School of Zoology, Tel Aviv University, Tel Aviv, Israel 4 Steinhardt Museum of Natural History, Tel Aviv University, Tel Aviv, Israel 5 Division of Ecology and Evolution, Research School of Biology, The Australian National University, Canberra, ACT, Australia

All authors conceived the ideas and designed the study; FK and SM collected the data; FK, DN, SN analysed the data; FK wrote the first draft and all authors contributed to revising the manuscript. All authors declare no conflict of interest 161

Abstract

Life history strategies are incredibly diverse across taxa, yet traits covary predictably along a fast-slow continuum such that species exhibit a distinct pace-of-life. Theory puts acquisition-allocation trade-offs at the core of life history evolution; however, reproductive investment in species with different reproductive modes can modify acquisition-allocation trade-offs which ultimately alters their life history. While metabolic rate has been touted as a key explanation for pace-of-life across diverse taxa some consider trade-offs between current and future reproduction as being more influential. Other evidence points to environmental variation playing an overarching role. No comparative study has attempted to disentangle these competing explanations. Here, we integrate physiological, reproductive and ecological data across more than 500 terrestrial ectotherms (snakes, lizards and the tuatara) where viviparity has arisen independently over 100 times throughout their evolutionary history. Applying a single multivariate phylogenetic framework, we test how metabolic rate, the environment and reproductive investment contribute in explaining the (co)variation in age at maturity and life span. We also determine if the importance of these factors vary according to reproductive mode (oviparous and viviparous taxa). Overall, while metabolic rate had a negative impact on age at maturity and lifespan, there was high heterogeneity among taxa in its relative role. Age at maturity and lifespan were strongly influenced by the environment, possibly through its effect on mediating activity patterns. Relative reproductive investment was also moderately important in predicting variation in age at maturity but not lifespan, particularly in oviparous species. Our results exemplify the complex nature of life history evolution. We highlight environmental drivers and its effect on patterns of growth, resource availability and mortality as likely sources of life history variation.

162

Introduction

Life history variation among species tends to fall along a fast-slow continuum (Sæther, 1987), with predictable covariation between certain traits often referred to as ‘pace-of-life’ (sensu Réale et al., 2010). Species with a fast pace-of-life are characterised by high growth rates, early ages at maturation, greater investment in current reproduction and shorter lifespan compared to species with a slow pace-of-life (Bakewell et al., 2019; Healy et al., 2019; Pettersen et al., 2016). Consistent covariation among life history traits imply that constraints play a defining role in life history evolution, yet the mechanisms that underpin these constraints are poorly understood (Ricklefs & Wikelski, 2002; S C Stearns, 1989; Van Noordwijk & De Jong, 1986). Theoretical models suggest that variation in energy acquisition and allocation are fundamental in explaining why life- history traits covary in certain contexts (Araya-Ajoy et al., 2018; Salzman et al., 2018; Van Noordwijk & De Jong, 1986). This line of inquiry has led many to predict that metabolic rate is the ‘pacemaker’ of life (Biro & Stamps, 2008; Careau et al., 2008; Glazier, 2015; Ricklefs & Wikelski, 2002). Energy metabolism is essential to all biological processes and is functionally linked with growth and senescence (Monaghan et al., 2009; Speakman, 2003). As such, metabolic rate has been viewed as a conceptually appealing mechanism explaining life history diversity across the animal kingdom (Auer et al., 2017; Healy et al., 2019; Stark et al., 2020; Wikelski et al., 2003).

Despite the appeal, interspecific evidence for metabolic rate as the driver in pace-of-life has been mixed and mainly focused on endotherms (Stark et al. 2020; Wiersma et al. 2007; Bech et al. 2016). In insectivorous mammals, lower metabolic rate is associated with longer lifespans, and gestations lengths (Symonds, 1999), but in mammals more broadly, there is limited evidence that metabolic rate is related to longevity (Stark et al. 2020). A comparative analysis across 69 bird species has shown that tropical bird species with slow paces-of- life (i.e., long-lived, develop slowly and mature late) have lower metabolic rates than temperate species (Wiersma et al. 2007). Comparative work in Australian passerines; however, casts doubt on the generality of this conclusion, suggesting that slow paces-of-life may be more related to environmental conditions and thermoregulatory demands of different clades rather than

163

metabolic rate per se (Bech et al. 2016). Similar results have been echoed in a broader phylogenetically controlled analysis in birds (Stark et al. 2020). Weak links between metabolism and life-history have also been observed across ectotherms suggesting that the endothermy-ectothermy divide is not a major factor driving life-history variation in terrestrial vertebrates (Stark et al. 2020). While these represent significant advances in testing the relationship between metabolic rate and life history, they have neglected to integrate other factors that might impact energy acquisition and allocation in explaining life-history variation (Van Noordwijk and De Jong, 1986).

(D) Environment Temperature Precipitation (A)Acquisition Predation Activity Population density Foraging Mate search Territory defense

(B) Metabolism Aerobic metabolism ATP production Oxidative stress (C) Allocation Current vs. Future reproduction Nutrient provisioning Offspring size/number Maintenance Growth

Figure 1 Conceptual diagram depicting acquisition and allocation of energy and how these processes relate to our hypotheses of why different paces-of-life exist. Resources are acquired (A) from the environment and then metabolised

(B) in order to produce energy. Energy is then allocated (C) to various biological processes. The environment (D) imposes constraints on acquisition 164

and allocation. The ‘activity constraints hypothesis’ posits that the environment specifically constrains opportunities to acquire resources through limiting activity times. Species that have more time to acquire resources are able to sustain fast lifestyles. The ‘rate of living hypothesis’ states that metabolic rate sets pace of life. Species with high a metabolic rate also have a fast lifestyle, characterised by high growth rates, early maturity and short lifespans. The

‘reproduction allocation hypothesis’ postulates that resources allocated to reproduction is traded off with other vital processes. Organisms with high reproductive investment must compromise their maintenance which results in reduced lifespans. The ‘environmental constraints hypothesis’ recognises the all-encompassing role of the environment and its impact on limiting acquisition, metabolism as well as allocation which could affect the (co)variation of life history traits.

Stronger conceptual links between theory and empirical data are needed to refine our understanding of the evolution of pace-of life. Metabolic rate as a causal mechanism for life-history variation merely captures one aspect of energy acquisition (Fig. 1). A multitude of factors beyond metabolic rate, such as resource availability and foraging time, are also crucial to energy acquisition and can lead to similar covariance between life-history traits (Angilletta, 2001; Glazier, 2015; Healy et al., 2019). At a broad scale, latitudinal variation in temperature and precipitation can restrict the available time for gathering resources thus constraining variation in life history (“activity constraints hypothesis”, Fig. 1) (Dunham et al., 1989). For example, climate variability associated with latitude can explain divergent strategies in growth and reproductive investment among populations of eastern fence lizards (Sceloporus undulatus) (Angilletta, 2001; Niewiarowski, 1995; Niewiarowski & Roosenburg, 1993). Warmer and more stable seasons at lower latitudes provide lizards with longer foraging times and better thermoregulatory opportunities which increase energy budgets by ~60% compared to high latitude populations (Angilletta, 2001). Environmental effects on energy acquisition and assimilation are likely to be an important components of life history evolution across species

165

(Fig. 1), however this also depends on how different species allocate energy to competing processes.

Reproductive investment can also influence life history variation observed across species (Fig. 1). Resources allocated to current reproduction are expected to compromise future reproduction and survival (S C Stearns, 1989; Stephen C. Stearns, 1976; Williams, 1966). As such, species investing heavily into reproduction are expected have a faster pace-of-life. Reproductive allocation can also depend on the reproductive mode exhibited by a species. In most species, females reproduce by laying eggs in the external environment (oviparity), while a smaller proportion of species have evolved to give birth to live young (viviparity) (Blackburn, 1999). These reproductive strategies are associated with distinct energetic costs/benefits that can influence patterns of acquisition-allocation trade-offs (Zhang et al., 2018). In oviparous taxa, egg development occurs in the external environment. As such, eggs are exposed to higher rates of predation, parasitism and unfavourable environmental conditions which can reduce offspring survival (Blackburn, 1999). In contrast, viviparous mothers can buffer their offspring and mitigate these risks. Indeed, viviparity has several disadvantages too (Hoyle & Ezard, 2012). Given gestation lengths tend to be longer in viviparous species, pregnant mothers themselves incur higher predation risks and decreased food intake due to reduced locomotor performance (Blackburn, 1999; Shine, 2005). Furthermore, viviparity is often associated with the evolution of complex forms of parental care (Royle et al., 2012). Mothers or parents must therefore invest more heavily in their current offspring at the cost of their own fitness (Blackburn, 1999; Hamel et al., 2011). Taken together, differential patterns of reproductive investment between viviparous and oviparous species are likely to influence energy allocation to other processes, such as growth and maintenance thereby shaping life-history.

Lepidosaurs (a clade including the tuatara, snakes, amphisbaenians and lizards) are an ideal system for and evaluating alternative hypotheses for life-history traits falls along a fast-slow continuum. Unlike mammals and birds, viviparity has evolved over 100 times allowing direct comparisons between closely related species that differ in reproductive mode and reproductive costs (Blackburn, 2000; Pyron & Burbrink, 2014). Moreover, lepdiosaurs have precocial young

166

and lack complex forms of parental care which means that resources invested into offspring are acquired prior to birth/oviposition. Here, using a multivariate statistical approach we test between alternative hypotheses underlying life history (co)variation between age at maturity and lifespan in over 500 squamates (Fig. 1). First, the ‘rate of living hypothesis’ posits that interspecific variation of metabolic rate underlies age at maturity and lifespan so that high metabolic rates causes in ‘fast’ lifestyles (Careau & Garland Jr, 2015; Raymond, 1928). If this hypothesis is supported, then metabolic rate should explain variation in age at maturity and lifespan independent of other causal drivers. In contrast, the ‘activity constraints hypothesis’ suggests that the amount of time available for energy acquisition is limited by the environment – species that are active all year around are predicted to have more opportunities to forage, grow and reproduce and therefore have a ‘fast’ pace-of-life (Angilletta, 2001). Support for this hypothesis would come from activity time predicting variation in pace-of-life across taxa. More generally, abiotic and biotic factors associated with latitude could also lead to different paces-of-life (‘environmental constraints hypothesis’). Latitudinal patterns in resource availability (Mueller & Diamond, 2001), parasite prevalence (Merino et al., 2008) and population density (Johnson, 1998; Wright et al., 2018) can limit organisms in ways other than patterns of activity. Finally, energy allocation processes may instead be more important in governing pace-of-life. The ‘reproductive allocation hypothesis’ postulates that species allocating more energy into current reproduction are expected to reproduce early and die young (Stephen C. Stearns, 1976). While these hypotheses are not mutually exclusive, multivariate phylogenetic methods allow us to directly compare their effect sizes and tease apart their relative importance to better understand the evolution of life history strategies.

167

Methods

Metabolism data

Species means for body mass and standard metabolic rate (hereafter MR) were collated from Uyeda et al., (2017), which makes use of a processed dataset compiled by White et al., (2006) (n[species] = 140). To supplement our dataset further, we performed a forward search on references that cited White et al., (2006) and extracted MR, the temperature at which MR was measured at and body mass data from a further eight empirical papers (n[observations] = 61, n[species] = 25). All MR data were standardised to VO2 ml-1 hr-1 and temperature-corrected to 20ºC (see ESM for more details).

Life History and Reproductive Investment

Age at maturity and lifespan were collected from Scharf et al., (2015) and an updated version of Meiri (2018). Lifespan represents the maximum value in years reported for each species and age at maturity represents the mean number of months until the first reproductive event (Meiri, 2018; Scharf et al., 2015). All other life history traits were collected from an updated version of Meiri (2018). Parity mode (1 = viviparous, 0 = oviparous) was treated as a binary factor. Ovoviviparous species were categorised as viviparous as mothers carry eggs until full term (n[species] = 5). For species that were reported to show both forms of reproduction (n[species] = 6), we went back to the original article of where the metabolism data was collected from to determine the parity mode of population that was sampled (n[species] = 2/6). We were unable to determine the parity mode for four out of six species. As such, these species were assigned the parity mode reported in reputable databases such The Reptile database and The Animal Diversity Web (Meyers et al., 2015; Uetz & Etzold, 1996). These assignments were also verified using literature searches in Google Scholar.

We calculated reproductive investment following Meiri et al., (2020) as the mean number of clutches/litters per year multiplied by mean hatchling mass divided by female mass. We took the mid value of the number of clutches/litters per year as the mean. Female mass were calculated using female snout-vent-length using clade-specific mass equations (Feldman et al., 2016). 168

Estimates of key life history traits tend to be more accurate for well- studied species as researchers may be more persistent in tracking individuals. We therefore included a proxy for ‘research effort’ in all our analyses (Valcu et al., 2014). Research effort is the number of documented results found in a Scopus Search for each species (see ESM for more details).

Activity and Spatial Data

Environmental conditions, such as temperature and seasonality, are predicted to constrain species to a multitude of ways. We quantified the total number of active months per year for each species as a measure of how the environments specifically imposes limits on activity (‘activity constraints hypothesis’) (Meiri, 2020, unpublished data). We also collected the average absolute latitude for each species global map distribution from Meiri (2018) (see also Roll et al., 2017) to account for other ways through which the environment may exert limitations on energy acquisition and allocation ('environmental limitation hypothesis’).

Incorporating Species Trait Data to Deal with Missing Data

Phylogenetic comparative analyses using multiple datasets can be challenging given the lack of overlap in species trait data across datasets (Nakagawa, 2015; Pennell et al., 2016). To deal with this problem, we took advantage of an existing squamate trait database with over 6000 species (Meiri, 2018). One major benefit of this database is its extensive coverage of body size (snout-vent-length, SVL) data. Many phenotypic traits, such as body mass, MR and life history are strongly correlated with SVL (see ESM for correlations), and thus provide a powerful way to maximize the retention of data when used in conjunction with missing data imputation techniques (see Statistical analyses below). We retained species from Meiri (2018) if they had both data for SVL and either age at maturity or lifespan (but ideally both). The variables we selected from the Meiri (2018) database were: maximum SVL (hereafter referred to as just SVL) and mean clutch size. Using these criteria, we increased our species list from n[species] = 156 to n[species] = 512. Once we incorporated the relevant data from Meiri (2018) to our dataset, we performed targeted searches for MR, age at maturity and lifespan for each species (see ESM for more details) We also collated data for all other variables in the same manner described above.

169

Phylogeny

We used a molecular phylogeny compiled by Zheng and Weins (2016) for our analysis. For species that did not initially match to the tree tip names (n[species] = 72), we matched 20/72 species using their synonyms (see ESM for more details). We made small amendments to the tree tips to accommodate four species for which we had MR data (Phymaturus spurcus, P. excelus and P. tenebrosus, P.dorsimaculatus). The overall topology of the tree did not change (see ESM for more details). Our final tree was transformed into an ultrametric tree using the ‘chronopl’ function from the package ‘ape’ (Paradis & Schliep, 2019).

Statistical analysis

All statistical analyses were conducted in R (Core Team, 2013). Data were checked for outliers/errors and normality using density plots and scatter plots. All variables were log10-transformed and then z-transformed. Absolute latitude was only z-transformed. Data standardisation allowed us to make direct comparisons between predictor variables which represent our four competing hypotheses. Data and code to generate our results are available via the Open Science Framework (see Data Accessibility).

Our goal was to impute missing data in order to maximise coverage across the phylogeny. Both MR and body mass had 66% of missing data (N = 512), whereas age at maturity and lifespan had 8.4% and 10.35% missing, respectively. We used predictive mean matching in the package ‘mice’ to generate twenty imputed datasets (Buuren & Groothuis-Oudshoorn, 2010). Variables used during imputation include: MR, body mass, age at maturity, lifespan, absolute latitude, months active, SVL and traits that were used to calculate reproductive investment. We then used these datasets to fit our Bayesian phylogenetic mixed models in the package ‘MCMCglmm’ (Jarrod D Hadfield, 2010). To account for phylogenetic relatedness, we calculated the inverse of phylogenetic variance-covariance matrix using the ‘inverseA’ function ‘MCMCglmm’ and incorporated this matrix into the models. For each dataset, we fitted a bivariate response model and suppressed the estimation of the intercept. To test alternative drivers of life-history covariation, we included metabolic rate, reproductive investment, active months and absolute latitude as

170

fixed effects in a single multivariate model. Body mass and research effort were included as covariates. We ran separate models for each parity mode and estimated an unstructured phylogenetic variance-covariance matrix (for more details see Data accessibility).

Uninformative priors were used for all models. For each dataset, we ran one chain consisting of 930,000 iterations, a burn-in of 30,000 and a thinning interval of 600. Each model run was checked to ensure proper mixing and that posterior samples were strongly auto-correlated using diagnostic plots and the ‘autocorr’ function in the package ‘coda’ (Plummer et al., 2006). The posterior distribution for each model parameter was pooled across models for all 20 imputed datasets before we calculated posterior means and 95% credible intervals (Nakagawa & De Villemereuil, 2019). Phylogenetic signal (�) for age at maturity and lifespan was calculated as phylogenetic heritability (sensu J D Hadfield & Nakagawa, 2010) which represents the proportion of variance explained by the phylogeny relative to the total variance of our data.

As a sensitivity analysis, we used the package ‘Rphylopars’ to impute our missing data (Goolsby et al., 2016). This method incorporates phylogenetic relatedness during imputation; however it only generates one dataset using maximum likelihood algorithms (n[species] = 512). The same models were ran as described above but with three chains instead of one as we only have one dataset per analysis. The results were less conservative because it was not possible to account for imputation error; however, there were generally congruent to that of multiple imputation analysis (Fig. S3, Table S6-8). We also tested whether reproductive investment differed between reproductive modes. For each of the twenty imputed datasets, we ran a univariate phylogenetic mixed model and included reproductive mode as a fixed effect and absolute latitude as a covariate.

171

Results

roviparous = 0.89 (0.69 – 0.98)

rviviparous = 0.78 (-0.42 – 1)

Figure 2 Relationship between age at maturity (months) and lifespan (years) for oviparous (yellow points, n = 398) and viviparous (green points, n = 114) taxa.

Values are predicted estimates from a multiple imputation Bayesian phylogenetic mixed model. We plotted the line of best fit through the predicted values using the function ‘geom_line’. All other predictor variables are set to their mean on the log10 scale. The correlation (r) between age at maturity and lifespan are estimated from the same model. Values in brackets are 95% credible intervals. The small viviparous group in amongst the oviparous taxa (Age at maturity ~ 30-40 months, lifespan ~20 years) are all geckos from New Zealand known for their slow life history strategies.

Viviparous species tended to mature at a later age; however, there were no significant differences in age at maturity or lifespan between reproductive

172

modes (Fig. 2, Table S1). There was high phylogenetic heritability for age at maturity and lifespan across both oviparous and viviparous species (age at

maturity: �oviparous =: 0.70 [0.46 – 0.88], �viviparous = 0.62 [0.05 – 0.93]; lifespan:

�oviparous =: 0.66 [0.43 – 0.84], �viviparous = 0.61 [0.11-0.94]) (Table S2). Age at maturity and longevity were significantly positively correlated in oviparous species (Fig. 2). Viviparous species on the other hand, followed a similar pattern; however, there were a few taxonomic groups that diverged from other groups which may have contributed to credible intervals overlapping zero (Fig. 2). Pooling data from both reproductive modes, the overall correlation was positive and significantly different from zero (r = 0.90 [0.75 – 0.98]).

Age at maturity (months)Age at maturity (months)Lifespan (years) Lifespan (years)

-0.18 (-0.77 – 0.42) -0.37 (-0.96 – 0.28) Metabolic Acquisition Metabolic rate Metabolic rate -0.05 (-0.38 – 0.24) -0.18 (-0.45 – 0.07) Hypothesis -0.11 (-0.43 – 0.18) -0.22 (-0.52 – 0.06)

-0.02 (-0.31 – 0.27) Reproduction Allocation -0.04 (-0.34 – 0.24) Hypothesis Reproductive InvestmentReproductive* Investment -0.11 (-0.23 – -0.01) -0.01 (-0.12 – 0.10) * -0.13 (-0.25 – -0.02) -0.03 (-0.13 – 0.08)

-0.14 (-0.41 – 0.14) -0.05 (-0.33 – 0.23) Activity Constraints Active months per year Active months* per year -0.22 (-0.39 – -0.03) -0.08 (-0.27 – 0.13) Hypothesis . -0.18 (-0.37 – 0) -0.06 (-0.26 – 0.15)

0.24 (-0.04 – 0.53) 0.19 (-0.11 – 0.49) Environmental Constraints Absolute Latitude Absolute Latitude 0.22 (0.07 – 0.40) Hypothesis * 0.13 (-0.03 – 0.31) * 0.25 (0.10 – 0.42) * 0.15 (0.01 – 0.34)

−1.0 −0.5 0.0 −1.00.5 −0.51.0 0.01.5 −1.00.5 −0.51.0 0.0 −1.00.5 −0.51.0 0.01.5 0.5 1.0 Estimate Estimate Estimate Estimate

Overall Oviparous OverallViviparousOviparous Viviparous

Figure 3 Pooled estimates from Bayesian phylogenetic mixed model using

multiple imputed datasets (m = 20). We tested for the importance of metabolic

rate, reproductive investment, active months per year and absolute latitude in

independently predicting (left panel) age at maturity and (right panel) life span

in oviparous (yellow points, n = 398) and viviparous (green points, n = 114)

lepidosaurs. Black points represent estimates from a model where data from

both reproductive modes were combined (N = 512). This model controls for

173

the effects of body mass and research effort. For more details see Statistical

Analyses and ESM. Points represent posterior means and error bars represent

95% credible intervals. Dashed vertical line represents estimate values of 0 (i.e. no effect on response variables). * represent effects that are significantly different from zero. Numeric values adjacent to the points are point estimates and their credible intervals.

After controlling for body mass, research effort and phylogeny, metabolic rate did not predict variation in age at maturity or lifespan for both reproductive modes (Fig. 3, Table S3, 4), although effect sizes were generally comparable or larger in magnitude to other predictors. Accounting for latitudinal effects, there was also no evidence to suggest that reproductive modes differ in their reproductive investment (Estimate: -1.08 [-3.55 – 1.51]). Reproductive investment, the number of months active and absolute latitude were all significant predictors for age at maturity in oviparous species, but not viviparous species (Fig. 3). With all else held constant, egg-laying species with higher reproductive investment tended to mature earlier (Fig. 3, Table S3). In oviparous species, the effect sizes for the number of months active and absolute latitude were the same but were in opposing directions (Fig. 3, Table S3). On average, species with a greater number of months active, matured more quickly compare to species with fewer number of months active at higher latitudes (Fig. 4). Viviparous species followed a similar trend for these effects; however, they were not significant (Fig. 3, Table S4).

174

0.03 0.03

A) 0.02 B) 0.02 0.03

Probability density Probability 0.01

Probability density Probability 0.01 0.02

0.00

0.00 0 20 density Probability 0.01 40 Absolute Latitude 0 20 40 Absolute Latitude

0.03 0.00 0.03 0.2 0.2 0 20 40 0.2 Absolute Latitude 0.2 0.02 C) 0.02

0.1 0.1 0.2

Number of species 0.1 Number of species 0.01 0.1 density Probability Number of species 0.01 Number of species Probability density Probability

0.00 0.0 0.0 0.1 0.00 0 20 40 0.0 6 8 10 12 6 8 density Probability 10 12 0.0 Absolute Latitude Active months 0 20 40 Active months 6 8 10 12 6 8 10 12 ReproductiveAbsolute Latitude Mode 0Oviparous1 ReproductiveActive months Mode 0 1 Active monthsReproductive Mode Oviparous Viviparous Reproductive Mode 0 1Viviparous Reproductive Mode 0 1 Reproductive Mode Oviparous Viviparous 0.0 4.54.5 Monthstrait value active 12 6 8 10 12 Active months Months active length=0.5 Reproductive Mode Oviparous Viviparous Figure 4 A) Phylogeny based on Zheng and Wiens (2016) of 512 squamates

used in our study. Colour of branches represent the predicted values for the

number of months active per year. The grey bars surrounding the tree

represents the predicted values for mean age at maturity (months). Yellow

represent oviparous species (n = 398), green represents viviparous species (n =

114). B) Probability densities for absolute latitude of species distributions

showing that viviparous species (green) tend to occupy higher latitudes

whereas oviparous species (yellow) are more often found in lower latitudes C)

Probability densities for number of months active depicting that a larger

proportion of oviparous species (yellow) have more months active compared to

viviparous species (green).

Irrespective of reproductive mode, the overall effect for metabolic rate in predicting age at maturity and lifespan remained non-significant. Across reproductive modes, reproductive investment and absolute latitude also had overall significant effects on age at maturity (Fig. 3, Table S5). The number of active months per year tended to be negatively associated with age at maturity; however, the credible intervals slightly overlapped zero (Fig. 3, Table S5). Only, 175

absolute latitude had positive effects on lifespan indicating that species occupying in higher latitudes tend to live longer (Fig. 3, Table S5)

176

Discussion

Here we tested alternative hypotheses that explains life-history variation across 500 species of lepidosaurs using a multivariate phylogenetic statistical framework. Our understanding of why life history covaries along a fast-slow continuum has often focused on the idea that metabolic rate is the ‘pacemaker of life’ (Londono et al., 2015; Stark et al., 2020; Wiersma et al., 2007). Despite this, we found that metabolic rate did not significantly explain variation in both age at maturity or lifespan; although effect sizes were still relatively large. Lifespan was best explained by the ‘environmental constraints hypothesis’, the other three hypotheses were weakly supported. In contrast, both the ‘activity’ and ‘environmental constraints’ hypotheses could independently explain variation in age at maturity. Species that were observed more often throughout the year reached sexual maturity earlier. Additionally, environmental conditions associated with higher latitudes (e.g. reduced temperatures, resource availability and possibly lower juvenile mortality rates) increased the time to reach sexual maturity. Age at maturity, to a lesser extent, was also explained by changes in reproductive investment across species, also supporting the ‘reproductive allocation hypothesis’. Species investing more into reproduction reached sexual maturity earlier, but this did not trade-off with longevity as predicted. Interestingly, these results appear to be independent of reproductive mode (oviparity vs. viviparity). Overall, our results suggest that variation in ‘pace-of-life’ is driven by a multitude of intrinsic and extrinsic factors across terrestrial ectotherms.

Interspecific life-history patterns are expected to be strongly influenced by both the environmental conditions under which species evolve. As such, it comes as no surprise that both latitude and patterns of activity both explained significant variation in pace-of-life – particularly age at maturity. Our results resonate with many studies that show abiotic and biotic factors such as rainfall, temperature, competition and predation risk moderate life history variation (Cabezas-Cartes et al., 2018; Scharf et al., 2015; Stark et al., 2020). Theoretical work has also highlighted that interactions between the environment and demography (e.g., density-dependent selection) as being important drivers of life history

177

evolution (Healy et al., 2019; Sæther, 1997; Williams, 1957; Wright et al., 2018). High population densities often increases levels of juvenile mortality (reviewed in Sæther, 1997), with age-dependent mortality being key to establishing patterns of life-history covariation (Healy et al., 2019; Moorad et al., 2020). Of course, the reciprocal feedback between environmental and demographic effects mean that they are not easily disentangled in practice. As such, most comparative studies (including ours) often resort to broad and easily accessible data that captures the milieu of environmental variation (Cabezas-Cartes et al., 2018; Scharf et al., 2015; Stark et al., 2020). Latitude is one such surrogate as that it has been shown to covary with seasonality, temperature, predation and competition (Barnes, 2002; Guralnick, 2006; McKinnon et al., 2010; Reynolds et al., 2018). While activity periods and latitude are themselves partially correlated, our multivariate approach using standardised effect sizes allowed us to begin addressing their independent contributions to pace-of-life. Our results suggest that activity patterns were as influential as other sources of environmental variation in explaining age at maturity. This result may not be surprising because thermal constraints on activity are known to affect growth and sexual maturation by limiting opportunities to assimilate resources (Angilletta Jr et al., 2017; Gunderson & Leal, 2016; Niewiarowski & Roosenburg, 1993) (Sinervo & Adolph, 1994). Although our study was unable to discern what specific aspect of the environment beyond activity periods affect life history, there are clearly other significant contributors to consider (Healy et al. 2019; Scharf et al. 2015; Stark et al. 2020). It will be powerful to conduct a comparative study that attempts to unify population dynamics (e.g., see Healy et al. 2019), thermal ecology and additional sources of environmental variation to begin decomposing how the environment shapes life history evolution across species.

Irrespective of reproductive mode, species with higher reproductive investment matured faster but did not live shorter lives. Trade-offs between reproduction and survival are at the heart of life-history theory (Stephen C. Stearns, 1992). Higher reproductive investment is expected to increase reproductive costs leading to reduced survival and future reproduction (Stearns, 1989); however, such trade-offs do not always exist within and across species (Culina et al., 2019; Scharf et al., 2015). The lack of relationship between reproductive

178

investment and lifespan observed in our data suggests that reproductive costs are not as widespread than previously thought. Rather, it is more likely that variation in the environments or reproductive strategies in lepidosaurs enable species to offset potential reproductive costs (Van Noordwijk & De Jong, 1986). Indeed, our results imply that the environment is one of the key factors explaining variation in lifespan. Environmental effects on lifespan are supported by results from Scharf et al. 2015 and Stark et al. 2020, who found that diet and temperature, respectively, explained a significant amount of variation in lifespan. A growing body of literature points to complex dietary effects on lifespan (Adler et al., 2013; Runagall-McNaull et al., 2015; Senior et al., 2015) which suggests that exploring more fine-scale measures of a species’ diet, in the context of their niche, may provide substantial insights into variation in life history across species.

Interestingly, reproductive investment affected the timing of sexual maturation only. Moreover, this effect of seemed to be pertinent for oviparous species only, despite no differences in reproductive investment between parity modes (see also Meiri et al., 2020). This is likely due to higher statistical power as we had more oviparous species in our dataset and may not reflect true biological differences between parity modes. Greater investment into offspring may have facilitated growth permitting offspring to reach sexual maturity faster. Reproductive investment may be modified by differential investment in eggs or embryos which increases the rate of offspring development and sexual maturation (Dyke & Griffith, 2018). For example, egg investment has been shown to significantly shorten incubation duration by speeding up developmental rate (Aubret et al., 2017). This may allow offspring to hatch earlier in the season, possibly allowing them more opportunities to grow and to reach sexual maturity earlier. Such effects are likely to be species specific as reduced reproductive investment has also been shown to have negative consequences on offspring growth in brown anole lizards (Anolis sagrei) (Warner & Lovern, 2014). Mothers can also alter hormone levels embryos are exposed to (e.g., corticosterone and testosterone), which in turn influences offspring growth, and ultimately sexual maturation (T. Uller et al., 2007; Tobias Uller et al., 2009). Mothers can also increase levels of reproductive investment by reducing the total number of offspring per litter thereby allocating more

179

resources per offspring (offspring size-number tradeoff, Lim et al., 2014). Indeed, larger hatching mass facilitates growth and promote larger offspring size in some reptile species (Janzen, 1993; Webb et al., 2006; but see Warner & Shine, 2007).

Contrary to long-held views, we found weak support for metabolic rate as the ‘pacemaker’ of life. This result is consistent with comparative studies on a diverse range of vertebrates (Harvey et al., 1991; Stark et al., 2020). Holding all other factors constant including body mass, metabolic rate did not explain significant variation in either age at maturity or lifespan. Regardless, it is important to recognize that effect size for metabolic rate was in the predicted direction of the ‘metabolism hypothesis’. Furthermore, the effect size for metabolic rate was as large as the other competing predictors. Thus, our findings suggest that metabolism plays some role on life history, but that role is highly variable across species (White & Seymour, 2004). It is possible that oxygen production as a proxy for metabolic rate may not truly reflect energy production and fails to capture the elevated oxidative costs that is responsible for ‘fast’ life histories (Salin et al., 2015). Alternatively, correlated selection on body size and other extrinsic mortality risks may explain why the effects of metabolic rate are so heterogenous. Given that body size scales with age at maturity and longevity in tetrapods (Stark et al. 2020 and our results), the weak relationship between metabolic rate, age at maturity and longevity may stem from indirect selection on body size that drives ‘pace-of-life’. In support of this, both age at maturity and longevity have strong phylogenetic signals suggesting that evolutionary processes have substantially shaped variation we see across species. Furthermore, body mass was strongly related to both age at maturity and longevity in our models (Table S5, Estimate for age at maturity: 0.62 [0.3 – 0.96]; Estimate for lifespan: 0.81 [0.45 – 1.09]), confirming that it is a more important contributor to life history variation. Body size variation in ectotherms could contribute to age or size specific mortality which can influence selection on body size and metabolic rate and promote different paces-of-life (Healy et al. 2019). Additonally, external factors (e.g. predation, drought, thermal stress, food shortages) could possibly lead to size dependent mortality on ectotherms, and influence how energy is allocated to reproduction and somatic

180

maintenance thus further altering species life history (Olalla-Tárraga et al., 2006).

Conclusions

For decades, the immense diversity of life history strategies among different species have captivated the interest of biologists. Here, using over 500 Lepidosauria taxa exhibiting divergent parity modes, we demonstrate that the environment, selection on body size and how they both mediate extrinsic mortality, are major sources of variation in age at maturity and lifespan. Overall, we found weak evidence that metabolic rate was related to pace-of-life, whereas reproductive investment contributed to variation in age at maturity. Our field should begin probing specific aspects of the environments to better understand the interactions between resource acquisition, allocation and life history. Future work on how biotic and abiotic factors interact, resulting in distinct life history strategies would be invaluable. Our work highlights the need to draw more concrete connections from theoretical studies and comparative tools. While it can be difficult to find suitable data across a wide range of species, we believe this step is crucial to establish generality of the causes of life history variation. The continual advancement of statistical techniques will enable researchers to not only expand their scope but also tackle more specific hypotheses that will elucidate the mechanisms through which selection generates life history variation.

Data accessibility

Datasets and code used to generate results of this study is accessible via Open Science Framework ( https://bit.ly/3lIuBk8)

Acknowledgements

We express gratitude to Christopher Belter for advice on developing the Scopus API function used in this study. Matt Pennell, Alejandro Gonzales Voyer for their useful discussions. Will Cornwell for his useful advice on resolving polytomies and editing trees. Simone Bloomberg for advice on data imputation. Howard Hsu for troubleshooting access to Biology of Reptilia. DN was 181

supported by an Australian Research Council (ARC) Discovery Early Career Research Award to (DE150101774) and SN an ARC Future Fellowship (FT13010026).

182

References

Adler, M. I., Cassidy, E. J., Fricke, C., & Bonduriansky, R. (2013). The lifespan- reproduction trade-off under dietary restriction is sex-specific and context-dependent. Experimental Gerontology, 48(6), 539–548. https://doi.org/10.1016/j.exger.2013.03.007 Angilletta Jr, M. J., Steury, T. D., & Sears, M. W. (2017). Temperature, growth rate, and body size in ectotherms: Fitting pieces of a life-history puzzle. Integrative and Comparative Biology, 44(6), 1–12. Angilletta, M. J. J. (2001). Thermal and physiological constraints on energy assimilation in a widespread lizard (Sceloporus undulatus). Ecology, 82(11), 3044–3056. Araya-Ajoy, Y. G., Bolstad, G. H., Brommer, J., Careau, V., Dingemanse, N. J., & Wright, J. (2018). Demographic measures of an individual’s “pace of life”: Fecundity rate, lifespan, generation time, or a composite variable? 72(5), 1–14. https://doi.org/10.1007/s00265-018-2477-7 Aubret, F., Bignon, F., Bouffet-Halle, A., Blanvillain, G., Kok, P. J. R., & Souchet, J. (2017). Yolk removal generates hatching asynchrony in snake eggs. Scientific Reports, 7(1), 3041. https://doi.org/10.1038/s41598-017-03355-y Auer, S. K., Dick, C. A., Metcalfe, N. B., & Reznick, D. N. (2017). Metabolic rate evolves rapidly and in parallel with the pace of life history. Nature Communications, 9(1), 1–6. https://doi.org/10.1038/s41467-017-02514-z Bakewell, A. T., Davis, K. E., Freckleton, R. P., Isaac, N. J. B., & Mayhew, P. J. (2019). Comparing life histories across taxonomic groups in multiple dimensions: How mammal-like are insects? The American Naturalist, 195(1), 706195–48. https://doi.org/10.1086/706195 Barnes, D. K. A. (2002). Polarization of competition increases with latitude. Proceedings of the Royal Society of London. Series B: Biological Sciences, 269(1504), 2061–2069. https://doi.org/10.1098/rspb.2002.2105 Biro, P. A., & Stamps, J. A. (2008). Are animal personality traits linked to life- history productivity? Trends Ecology and Evolution, 23(7), 361–368. https://doi.org/10.1016/j.tree.2008.04.003 Blackburn, D. G. (1999). Viviparity and Oviparity: Evolution and Reproductive Strategies. Encyclopedia of Reproduction, 4.

183

Blackburn, D. G. (2000). Reptilian viviparity: Past research, future directions, and appropriate models. Comparative Biochemistry and Physiology Part A: Molecular & Integrative Physiology, 127(4), 391–409. https://doi.org/10.1016/S1095-6433(00)00272-5 Buuren, S. van, & Groothuis-Oudshoorn, K. (2010). MICE: Multivariate Imputation by Chained Equations in R. Journal of Statistical Software, 1– 68. Cabezas-Cartes, F., Boretto, J. M., & Ibargüengoytía, N. R. (2018). Effects of Climate and Latitude on Age at Maturity and Longevity of Lizards Studied by Skeletochronology. Integrative and Comparative Biology, 58(6), 1086–1097. https://doi.org/10.1093/icb/icy119 Careau, V., & Garland Jr, T. (2015). Performance, personality, and energetics: Correlation, causation, and mechanism. Physiological and Biochemical Zoology, 85(6), 543–571. https://doi.org/10.1086/666970 Careau, V., Thomas, D., Humphries, M. M., & Réale, D. (2008). Energy metabolism and animal personality. Oikos, 117(5), 641–653. https://doi.org/10.1111/j.0030-1299.2008.16513.x Core Team, R. (2013). Team (2012). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Culina, A., Linton, D. M., Pradel, R., Bouwhuis, S., & Macdonald, D. W. (2019). Live fast, don’t die young: Survival–reproduction trade-offs in long-lived income breeders. Journal of Animal Ecology, 88(5), 746–756. https://doi.org/10.1111/1365-2656.12957 Dunham, A. E., Grant, B. W., & Overall, K. L. (1989). Interfaces between Biophysical and Physiological Ecology and the Population Ecology of Terrestrial Vertebrate Ectotherms. Physiological Zoology, 62(2), 335–355. Dyke, J. U. V., & Griffith, O. W. (2018). Mechanisms of reproductive allocation as drivers of developmental plasticity in reptiles. Journal of Experimental Zoology Part A: Ecological and Integrative Physiology, 329(6–7), 275–286. https://doi.org/10.1002/jez.2165 Feldman, A., Sabath, N., Pyron, R. A., Mayrose, I., & Meiri, S. (2016). Body sizes and diversification rates of lizards, snakes, amphisbaenians and the tuatara: Lepidosaur body sizes. Global Ecology and Biogeography, 25(2), 187–197. https://doi.org/10.1111/geb.12398

184

Glazier, D. S. (2015). Is metabolic rate a universal ‘pacemaker’ for biological processes? Biological Reviews, 90(2), 377–407. https://doi.org/10.1111/brv.12115 Goolsby, E. W., Bruggeman, J., & Ané, C. (2016). Rphylopars: Fast multivariate phylogenetic comparative methods for missing data and within-species variation. Methods in Ecology, 8(1), 22–27. https://doi.org/10.1111/2041- 210X.12612 Gunderson, A. R., & Leal, M. (2016). A conceptual framework for understanding thermal constraints on ectotherm activity with implications for predicting responses to global change. Ecology Letters, 19(2), 111–120. https://doi.org/10.1111/ele.12552 Guralnick, R. (2006). The legacy of past climate and landscape change on species’ current experienced climate and elevation ranges across latitude: A multispecies study utilizing mammals in western North America. Global Ecology and Biogeography, 15(5), 505–518. https://doi.org/10.1111/j.1466-822X.2006.00231.x Hadfield, J D, & Nakagawa, S. (2010). General quantitative genetic methods for comparative biology: Phylogenies, taxonomies and multi-trait models for continuous and categorical characters. Journal of Evolutionary Biology, 23(3), 494–508. https://doi.org/10.1111/j.1420-9101.2009.01915.x Hadfield, Jarrod D. (2010). MCMC methods for multi-response generalized linear mixed models: The MCMCglmm R package. Journal of Statistical Software, 33(2), 1–22. Hamel, S., Côté, S. D., & Festa-Bianchet, M. (2011). Tradeoff between offspring mass and subsequent reproduction in a highly iteroparous mammal. Oikos, 120(5), 690–695. https://doi.org/10.1111/j.1600-0706.2011.19382.x Harvey, P. H., Pagel, M. D., & Rees, J. A. (1991). Mammalian Metabolism and Life Histories. The American Naturalist, 137(4), 556–566. https://doi.org/10.1086/285183 Healy, K., Ezard, T. H. G., Jones, O. R., mez, R. S.-G. x000F3, & Buckley, Y. M. (2019). Animal life history is shaped by the pace of life and the distribution of age-specific mortality and reproduction. Nature Ecology & Evolution, 1–10. https://doi.org/10.1038/s41559-019-0938-7

185

Hoyle, R. B., & Ezard, T. H. G. (2012). The benefits of maternal effects in novel and in stable environments. Journal of The Royal Society Interface, 9(75), 2403–2413. https://doi.org/10.1098/rsif.2012.0183 Janzen, F. J. (1993). An experimental analysis of natural selection on body size of hatchling turtles. Ecology, 74(2), 332–341. https://doi.org/10.2307/1939296 Johnson, C. N. (1998). Rarity in the tropics: Latitudinal gradients in distribution and abundance in Australian mammals. Journal of Animal Ecology, 67(5), 689–698. https://doi.org/10.1046/j.1365-2656.1998.00232.x Lim, J. N., Senior, A. M., & Nakagawa, S. (2014). Heterogeneity in individual quality and reproductive trade-offs within species. Evolution, 68(8), 2306– 2318. https://doi.org/10.1111/evo.12446 Londono, G. A., Chappell, M. A., & Castañeda, M. R. (2015). Basal metabolism in tropical birds: Latitude, altitude, and the ‘pace of life’. Functional Ecology. https://doi.org/10.1111/1365-2435.12348/full McKinnon, L., Smith, P. A., Nol, E., Martin, J. L., Doyle, F. I., Abraham, K. F., Gilchrist, H. G., Morrison, R. I. G., & Bety, J. (2010). Lower predation risk for migratory birds at high latitudes. Science, 327(5963), 326–327. https://doi.org/10.1126/science.1183010 Meiri, S. (2018). Traits of lizards of the world: Variation around a successful evolutionary design. Global Ecology and Biogeography, 27(10), 1168–1172. https://doi.org/10.1111/geb.12773 Meiri, S., Feldman, A., Schwarz, R., & Shine, R. (2020). Viviparity does not affect the numbers and sizes of reptile offspring. Journal of Animal Ecology, 89(2), 360–369. https://doi.org/10.1111/1365-2656.13125 Merino, S., Moreno, J., Vásquez, R. A., Martínez, J., Sánchez-Monsálvez, I., Estades, C. F., Ippi, S., Sabat, P., Rozzi, R., & Mcgehee, S. (2008). Haematozoa in forest birds from southern Chile: Latitudinal gradients in prevalence and parasite lineage richness. Austral Ecology, 33(3), 329–340. https://doi.org/10.1111/j.1442-9993.2008.01820.x Meyers, P., Espinosa, R., Parr, C. S., Jones, T., Hammond, G. S., & Dewey, T. A. (2015). The Animal Diversity Web. https://animaldiversity.org. Monaghan, P., Metcalfe, N. B., & Torres, R. (2009). Oxidative stress as a mediator of life history trade-offs: Mechanisms, measurements and

186

interpretation. Ecology Letters, 12(1), 75–92. https://doi.org/10.1111/j.1461-0248.2008.01258.x Moorad, J., Promislow, D., & Silvertown, J. (2020). George C. Williams’ Problematic Model of Selection and Senescence: Time to Move on. Trends Ecology and Evolution, 35(4), 303–305. https://doi.org/10.1016/j.tree.2020.02.001 Mueller, P., & Diamond, J. (2001). Metabolic rate and environmental productivity: Well-provisioned animals evolved to run and idle fast. Proceedings of the National Academy of Sciences, 98(22), 12550–12554. https://doi.org/10.1073/pnas.221456698 Nakagawa, S. (2015). Missing data: Mechanisms, methods and messages. In G. A. Fox, S. Negrete-Yankelevich, & V. J. Sosa (Eds.), Ecological Statistics: Contemporary theory and application (pp. 81–105). Oxford University Press. Niewiarowski, P. H. (1995). Effects of supplemental feeding and thermal environment on growth rates of eastern fence lizards, Sceloporus undulatus. Herpetologica, 51(4), 487–496. Niewiarowski, P. H., & Roosenburg, W. (1993). Reciprocal transplant reveals sources of variation in growth rates of the lizard Sceloporus Undulatus. Ecology, 74(7), 1992–2002. https://doi.org/10.2307/1940842 Olalla-Tárraga, M. Á., Rodríguez, M. Á., & Hawkins, B. A. (2006). Broad-scale patterns of body size in squamate reptiles of Europe and North America. Journal of Biogeography, 33(5), 781–793. https://doi.org/10.1111/j.1365- 2699.2006.01435.x Paradis, E., & Schliep, K. (2019). ape 5.0: An environment for modern phylogenetics and evolutionary analyses in R. Bioinformatics, 35(3), 526– 528. https://doi.org/10.1093/bioinformatics/bty633 Pennell, M. W., FitzJohn, R. G., & Cornwell, W. K. (2016). A simple approach for maximizing the overlap of phylogenetic and comparative data. Methods in Ecology, 7(6), 751–758. https://doi.org/10.1111/2041-210X.12517 Pettersen, A. K., White, C. R., & Marshall, D. J. (2016). Metabolic rate covaries with fitness and the pace of the life history in the field. Proceedings of the Royal Society of London B: Biological Sciences, 283(1831), 20160323. https://doi.org/10.1098/rspb.2016.0323 Plummer, M., Best, N., Cowles, K., & Vines, K. (2006). CODA: Convergence Diagnosis and Output Analysis for MCMC. R News, 6, 7–11.

187

Pyron, R. A., & Burbrink, F. T. (2014). Early origin of viviparity and multiple reversions to oviparity in squamate reptiles. Ecology Letters, 17, 13–21. Raymond, P. (1928). The rate of living. University of London Press. Réale, D., Garant, D., Humphries, M. M., Bergeron, P., Careau, V., & Montiglio, P.-O. (2010). Personality and the emergence of the pace-of-life syndrome concept at the population level. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 365(1560), 4051–4063. https://doi.org/10.1098/rstb.2010.0208 Reynolds, P. L., Stachowicz, J. J., Hovel, K., Boström, C., Boyer, K., Cusson, M., Eklöf, J. S., Engel, F. G., Engelen, A. H., Eriksson, B. K., Fodrie, F. J., Griffin, J. N., Hereu, C. M., Hori, M., Hanley, T. C., Ivanov, M., Jorgensen, P., Kruschel, C., Lee, K.-S., … Duffy, J. E. (2018). Latitude, temperature, and habitat complexity predict predation pressure in eelgrass beds across the Northern Hemisphere. Ecology, 99(1), 29–35. https://doi.org/10.1002/ecy.2064 Ricklefs, R. E., & Wikelski, M. (2002). The physiology/life- history nexus. Trends Ecology and Evolution, 17(10), 1–7. https://doi.org/10.1016/S0169- 5347(02)02578-8 Roll, U., Feldman, A., Novosolov, M., Allison, A., Bauer, A. M., Bernard, R., Böhm, M., Castro-Herrera, F., Chirio, L., Collen, B., Colli, G. R., Dabool, L., Das, I., Doan, T. M., Grismer, L. L., Hoogmoed, M., Itescu, Y., Kraus, F., LeBreton, M., … Meiri, S. (2017). The global distribution of tetrapods reveals a need for targeted reptile conservation. Nature Ecology & Evolution, 1–6. https://doi.org/10.1038/s41559-017-0332-2 Royle, N. J., Smiseth, P. T., & Kölliker, M. (2012). The Evolution of Parental Care. Oxford University Press. Runagall-McNaull, A., Bonduriansky, R., & Crean, A. J. (2015). Dietary protein and lifespan across the metamorphic boundary: Protein-restricted larvae develop into short-lived adults. Scientific Reports, 5(1), 11783. https://doi.org/10.1038/srep11783 Sæther, B.-E. (1987). The influence of body weight on the covariation between reproductive traits in European birds. Oikos, 48(1), 79–88. https://doi.org/10.2307/3565691

188

Sæther, B.-E. (1997). Environmental stochasticity and population dynamics of large herbivores: A search for mechanisms. Trends Ecology and Evolution, 12(4), 143–149. https://doi.org/10.1016/S0169-5347(96)10068-9 Salin, K., Auer, S. K., Rudolf, A. M., Anderson, G. J., Cairns, A. G., Mullen, W., Hartley, R. C., Selman, C., & Metcalfe, N. B. (2015). Individuals with higher metabolic rates have lower levels of reactive oxygen species in vivo. Biology, 11(9), 1–4. https://doi.org/10.1098/rsbl.2015.0538 Salzman, T. C., McLaughlin, A. L., Westneat, D. F., & Crowley, P. H. (2018). Energetic trade-offs and feedbacks between behavior and metabolism influence correlations between pace-of-life attributes. Behavioral Ecology and Sociobiology, 72(3), 54. https://doi.org/10.1007/s00265-018-2460-3 Scharf, I., Feldman, A., Novosolov, M., Pincheira-Donoso, D., Das, I., Böhm, M., Uetz, P., Torres-Carvajal, O., Bauer, A., Roll, U., & Meiri, S. (2015). Late bloomers and baby boomers: Ecological drivers of longevity in squamates and the tuatara. Global Ecology and Biogeography, 24(4), 396– 405. https://doi.org/10.1111/geb.12244 Senior, A. M., Nakagawa, S., Lihoreau, M., Simpson, S. J., & Raubenheimer, D. (2015). An overlooked consequence of dietary mixing: A varied diet reduces interindividual variance in fitness. The American Naturalist, 186(5), 649–659. https://doi.org/10.1086/683182 Shine, R. (2005). Life-history evolution in reptiles. Annual Review of Ecology, Evolution, and Systematics, 36(1), 23–46. https://doi.org/10.1146/annurev.ecolsys.36.102003.152631 Sinervo, B., & Adolph, S. C. (1994). Growth plasticity and thermal opportunity in Sceloporus lizards. Ecology, 75(3), 776–790. https://doi.org/10.2307/1941734 Speakman, J. R. (2003). Oxidative phosphorylation, mitochondrial proton cycling, free-radical production and aging. In Advances in Cell Aging and Gerontology (Vol. 14, pp. 35–68). Elsevier. https://doi.org/10.1016/S1566-3124(03)14003-5 Stark, G., Pincheira-Donoso, D., & Meiri, S. (2020). No evidence for the ‘rate-of- living’ theory across the tetrapod tree of life. Global Ecology and Biogeography, 29(5), 1–28. https://doi.org/10.1111/geb.13069 Stearns, S C. (1989). Trade-offs in life-history evolution. Functional Ecology, 3(3), 259. https://doi.org/10.2307/2389364

189

Stearns, Stephen C. (1976). Life-History Tactics: A Review of the Ideas. The Quarterly Review of Biology, 51(1), 3–47. https://doi.org/10.1086/409052 Stearns, Stephen C. (1992). The Evolution of Life Histories. Oxford University Press UK. Uetz, P., & Etzold, T. (1996). The EMBL/EBI Reptile Database. Herpetological Review, 27(4), 174–175. Uller, T., Astheimer, L., & Olsson, M. (2007). Consequences of maternal yolk testosterone for offspring development and survival: Experimental test in a lizard. Functional Ecology, 21(3), 544–551. Uller, Tobias, Hollander, J., Astheimer, L., & Olsson, M. (2009). Sex-specific developmental plasticity in response to yolk corticosterone in an oviparous lizard. Journal of Experimental Biology, 212(8), 1087–1091. https://doi.org/10.1242/jeb.024257 Uyeda, J. C., Pennell, M. W., Miller, E. T., Maia, R., & McClain, C. R. (2017). The evolution of energetic scaling across the vertebrate tree of life. The American Naturalist, 190(2), 185–199. https://doi.org/10.1086/692326 Valcu, M., Dale, J., Griesser, M., Nakagawa, S., & Kempenaers, B. (2014). Global gradients of avian longevity support the classic evolutionary theory of ageing. Ecography, 37(10), 930–938. https://doi.org/10.1111/ecog.00929 Van Noordwijk, A. J., & De Jong, G. (1986). Acquisition and Allocation of Resources—Their Influence on Variation in Life-History Tactics. The American Naturalist, 128(1), 137–142. Warner, D. A., & Lovern, M. B. (2014). The maternal environment affects offspring viability via an indirect effect of yolk investment on offspring size. Physiological and Biochemical Zoology, 87(2), 276–287. https://doi.org/10.1086/674454 Warner, D. A., & Shine, R. (2007). Fitness of juvenile lizards depends on seasonal timing of hatching, not offspring body size. Oecologia, 154(1), 65–73. https://doi.org/10.1007/s00442-007-0809-9 Webb, J. K., Shine, R., & Christian, K. A. (2006). The adaptive significance of reptilian viviparity in the tropics: Testing the maternal manipulation hypothesis. Evolution, 60(1), 115–122. https://doi.org/10.1111/j.0014- 3820.2006.tb01087.x

190

White, C. R., Phillips, N. F., & Seymour, R. S. (2006). The scaling and temperature dependence of vertebrate metabolism. Biology Letters, 2(1), 125–127. https://doi.org/10.1098/rsbl.2005.0378 White, C. R., & Seymour, R. S. (2004). Does basal metabolic rate contain a useful signal? Mammalian BMR allometry and correlations with a selection of physiological, ecological, and life-history variables. Physiological and Biochemical Zoology, 77(6), 929–941. https://doi.org/10.1086/425186 Wiersma, P., Munoz-Garcia, A., Walker, A., & Williams, J. B. (2007). Tropical birds have a slow pace of life. Proceedings of the National Academy of Sciences, 104(22), 9340–9345. https://doi.org/10.1073/pnas.0702212104 Wikelski, M., Spinney, L., Schelsky, W., Scheuerlein, A., & Gwinner, E. (2003). Slow pace of life in tropical sedentary birds: A common-garden experiment on four stonechat populations from different latitudes. Proceedings of the Royal Society of London B: Biological Sciences, 270(1531), 2383–2388. https://doi.org/10.1098/rspb.2003.2500 Williams, G. C. (1957). Pleiotropy, natural selection, and the evolution of senescence. Evolution, 11(4), 398–411. https://doi.org/10.2307/2406060 Williams, G. C. (1966). Natural selection, the costs of reproduction, and a refinement of Lack’s principle. The American Naturalist, 100(916), 687–690. https://doi.org/10.1086/282461 Wright, J., Bolstad, G. H., Araya-Ajoy, Y. G., & Dingemanse, N. J. (2018). Life- history evolution under fluctuating density-dependent selection and the adaptive alignment of pace-of-life syndromes. Biological Reviews, 94(1), 857–18. https://doi.org/10.1111/brv.12451 Zhang, L., Guo, K., Zhang, G.-Z., Lin, L.-H., & Ji, X. (2018). Evolutionary transitions in body plan and reproductive mode alter maintenance metabolism in squamates. BMC Evolutionary Biology, 18(1), 45. https://doi.org/10.1186/s12862-018-1166-5 Zheng, Y., & Wiens, J. J. (2016). Combining phylogenomic and supermatrix approaches, and a time-calibrated phylogeny for squamate reptiles (lizards and snakes) based on 52 genes and 4162 species. Molecular Phylogenetics and Evolution, 94, 537–547. https://doi.org/10.1016/j.ympev.2015.10.009

191

Supplementary Materials

Data cleaning

Data for body mass and standard metabolic rate (hereafter MR) was collected from Uyeda et al., (2017), which makes use of a processed dataset compiled by White et al., (2006) (n[species] = 140). We found an entry error for Sphaerodactylus cinereus and re-extracted the data from the original publication.

Metabolic rate from the forward search needed to be converted to VO2 ml-1 hr- 1g-1 (hereafter VO2). As such, MR data reported in mW (n[species] = 1) was

-1 converted following Lighton (2008, pp 109) whereas data reported in VCO2 ml hr-1 was converted to VO2 values by assuming a respiratory quotient of 0.80 (n[species] = 2). In some cases, there were multiple observations of the same species from different populations or studies (n[species] = 9, n[observations] = 27). In these cases, we took the weighted mean by weighting data by the sample size. Finally, all MR data was temperature corrected using the Arrhenius equation following Gillooly et al., (2001).

Research effort

Research effort is calculated by first obtaining a list of synonyms of each species from The Open Tree of Life using the ‘synonym’ function in the ‘rotl’ package (Michonneau et al., 2016). On the 22nd of August 2017, we performed an exact phrase query in titles, abstracts and keywords of each species name and all its associated synonyms using the Scopus Search API and extracted the number of published documents for each search (see ESM for code). The number of document results found in each name search was summed for each species and used as a measure of research effort. Prior to log10-transformation, we added 0.5 to all the research effort data because some species had no document results (i.e. research effort = 0).

Targeted searches

Once we incorporated the relevant data from Meiri (2018) to our dataset, we performed targeted searches for each species on the Web of Science on the 1st of August 2019 to recover any additional data on MR, age at maturity and

192

lifespan. We searched across all years in the Science Citation, Conference Proceedings, Book Citation and emerging sources indices. For species lacking MR and body mass data (n[species] = 264), we used the following search string format: "species name" AND ( "standard" OR "resting" OR "basal" ) AND ( "metabolic rate" ) AND ( "mass" ). For species lacking age at maturity data (n[species] = 54), we used the following search string format: “species name” AND ( “age at matur*” OR “age at sex* matur*” ). For species lacking lifespan data (n[species] = 48), we used the following search string format: “species name” AND ( “longev*” OR “life span” OR “life expect*"). Data from figures were extracted using the R package ‘metadigitise’ (Pick et al. 2019). If minimum and maximum values were provided for age at maturity, we took the average of the two. When a female and male estimate was reported for lifespan, we took the midpoint of the two values. If a study had multiple comparisons (e.g. between localities), weighted means were computed as per above. All MR data were converted to VO2 and temperature-corrected in the same way as above. Overall, we recovered MR and body mass data for n[species] = 10, and one species each for age at maturity and life span using targeted searches.

Matching species names to phylogeny

For species that did not initially match to the tree tip names (n[species] = 72), we searched for their synonyms in R package ‘rotl’ and rematched them to the tree using their synonyms. Overall, 20/72 species were able to be matched using synonyms.

Four species were not found in the phylogeny (n = 4, Phymaturus spurcus, P. excelus and P. tenebrosus, P.dorsimaculatus). Based on a recently published morphological and molecular tree of the Phymaturus clade (Lobo et al., 2018), we replaced P. tenebrosus with P. somuncurensis and P.dorsimaculatus with P. vociferator in the same clade so the topology of the clade is retained . When added P. spurcus and P. excelsus as sister taxa with minimum branch lengths in place of P. patagonicus (a sister to both P.spectabilis and P. excelsus). Our final tree was transformed into an ultrametric tree using the ‘chronopl’ function from the R package ‘ape’ (Paradis & Schliep, 2019).

193

Correlations between traits for data imputation

We wanted to expand our dataset using a published dataset (n[species > 6000]) of squamate data (Meiri 2018). Data imputation would be effective under this circumstance because many phenotypic traits are correlated with snout-vent- length (SVL), a measure of body length (Figure S1).

logSMR logMass logAgemat logLongev logResEffort logmaxSVL logFemSVL logHatchSVL logCS logBroods logActivem abslatmean_active_months

0.5 logSMR 0.4 0.3 Corr: Corr: Corr: Corr: Corr: Corr: Corr: Corr: Corr: Corr: Corr: Corr: 0.2 0.1 0.915 0.385 0.53 0.243 0.814 0.791 0.733 0.592 −0.229 0.0326 −0.323 0.00521 0.0 4 logMass 3 Corr: Corr: Corr: Corr: Corr: Corr: Corr: Corr: Corr: Corr: Corr: 2 1 0.488 0.645 0.23 0.912 0.913 0.842 0.622 −0.36 −0.0268 −0.39 −0.0554 0 logAgemat 2.0 1.5 Corr: Corr: Corr: Corr: Corr: Corr: Corr: Corr: Corr: Corr: 1.0 0.642 0.115 0.434 0.494 0.55 0.216 −0.558 −0.373 0.283 −0.39 0.5 2.0 logLongev 1.5 Corr: Corr: Corr: Corr: Corr: Corr: Corr: Corr: Corr: 1.0 0.5 0.289 0.563 0.598 0.639 0.185 −0.344 −0.171 0.094 −0.18

0.0 logResEffort 3 2 Corr: Corr: Corr: Corr: Corr: Corr: Corr: Corr: 1 0.275 0.216 0.114 0.269 −0.0037 −0.0267 0.168 −0.039 0 logmaxSVL 3.0 2.5 Corr: Corr: Corr: Corr: Corr: Corr: Corr: 2.0 0.972 0.896 0.598 −0.348 −0.0018 −0.178 −0.018 1.5 logFemSVL 3.0 2.5 Corr: Corr: Corr: Corr: Corr: Corr: 2.0 0.928 0.578 −0.406 −0.0137 −0.17 −0.0302 1.5 logHatchSVL

2.0 Corr: Corr: Corr: Corr: Corr: 1.6 1.2 0.368 −0.421 −0.0577 −0.122 −0.0719

1.5 logCS 1.0 Corr: Corr: Corr: Corr: 0.5 −0.402 −0.228 0.0154 −0.262 0.0 1.5 logBroods 1.0 Corr: Corr: Corr: 0.5 0.0 0.388 −0.285 0.413 −0.5 1.1 logActivem 1.0 0.9 Corr: Corr: 0.8 −0.646 0.993 0.7 abslat 40 Corr: 20 −0.673 mean_active_months 0 12 10 8 6 0 1 2 0 1 2 3 4 0.51.01.52.0 0.00.51.01.52.0 0 1 2 3 1.52.02.53.0 1.52.02.53.0 1.21.62.0 0.00.51.01.5 −0.50.00.51.01.50.70.80.91.01.10 20 40 6 8 1012 Figure S1 Scatterplot matrix of phenotypic traits used in imputation. Upper triangle are estimates of Pearson correlations, lower triangle are the scatterplots of raw data. Note that NAs have been excluded to generate this plot.

194

Results: What predicts variation in age at maturity and lifespan? Multiple imputation model

Age at maturity (months)Age at maturity (months)Longevity (years) Longevity (years)

Trait mean Trait mean

Metabolic rate Metabolic rate

Body Mass * Body Mass * * * *

Reproductive Investment Reproductive Investment

Active months per year Active months* per year *

Absolute Latitude Absolute* Latitude * *

Research effort Research effort * * −1.5 −1.0 −0.5 0.0 0.5−1.51.0−1.01.5−0.5 0.0 −1.50.5 −1.01.0 −0.51.5 0.0 0.5−1.51.0−1.01.5−0.5 0.0 0.5 1.0 1.5 Estimate Estimate Estimate Estimate

Overall Oviparous ViviparousOverall Oviparous Viviparous

Figure S2 Pooled estimates from a multiple imputation (20 datasets) Bayesian phylogenetic mixed model. We tested for the importance of metabolic rate, body mass, reproductive investment, active months per year and absolute latitude and research effort in predicting (left panel) age at maturity and (right panel) life span in oviparous (yellow points, n = 398) and viviparous (green points, n = 114) squamates. Black points represent estimates from a model where data from both reproductive modes were combined (N = 512). Points represent posterior means and error bars represent 95% credible intervals. Dashed vertical line represents estimate values of 0 (i.e. no effect on response variables). * represent effects that are significantly different from zero. For point values and credible intervals see Table S3-5.

Sensitivity analysis using Rphylopar: What predicts variation in age at maturity and lifespan?

195

Figure S3 Pooled estimates from a Bayesian phylogenetic mixed model using a maximum likelihood algorithm which incorporates phylogenetic relatedness. We tested for the importance of metabolic rate, body mass, reproductive

Age at maturity (months)Age at maturity (months)Longevity (years)Longevity (years)

Trait mean Trait mean

Metabolic rate Metabolic rate

Body Mass Body Mass * * * * *

Reproductive InvestmentReproductive Investment

* Active months per year Active months* per year * * *

Absolute Latitude Absolute Latitude

* Research effort Research effort * * −1.5−1.0−0.5 0.0 0.5 −1.01.5−1.51.0−2.00.5 2.50.0 0.5 1.0−1.51.5−1.02.0−0.52.50.0 0.5−1.51.0−1.01.5−0.52.00.0 0.5 1.0 1.5 2.0 Estimate Estimate Estimate Estimate

Overall Oviparous OverallViviparousOviparous Viviparous investment, active months per year and absolute latitude and research effort in predicting (left panel) age at maturity and (right panel) life span in oviparous (yellow points, n = 398) and viviparous (green points, n = 114) squamates. Black points represent estimates from a model where data from both reproductive modes were combined (N = 512). Points represent posterior means and error bars represent 95% credible intervals. Dashed vertical line represents estimate values of 0 (i.e. no effect on response variables). * represent effects that are significantly different from zero. For point values and credible intervals see Table S6-8.

Table S1 Pooled posterior mean estimates for age at maturity (months) and lifespan (years) in oviparous and viviparous squamates (N = 514). Lower and upper bounds represent 95% credible intervals. Oviparous species Viviparous species 196

(n = 398) (n = 114) Estimat Lowe Uppe Estimat Lowe Uppe e r r e r r

Age at maturity (months) 27.07 14.16 49.03 31.67 12.85 66.6 Lifespan (years) 14.12 6.75 27.32 14.44 5.19 33.43

Table S2 Pooled posterior mean estimates for phylogenetic heritability of age at maturity (months) and lifespan (years) in oviparous and viviparous squamates (N = 514). Lower and upper bounds represent 95% credible intervals. Oviparous species Viviparous species (n = 398) (n = 114) � Lower Upper � Lower Upper

Age at maturity (months) 0.70 0.46 0.88 0.62 0.05 0.93 Lifespan (years) 0.66 0.43 0.84 0.61 0.11 0.94

Table S3 Pooled model coefficients from twenty Bayesian bivariate generalised mixed models using twenty imputed datasets. This model tests the relative importance of metabolic rate, reproductive investment, active months per year and latitude in promoting covariation in age at maturity and longevity in oviparous squamates (n = 398). Body mass and research effort were included as covariates. All variables were log10 transformed and then z- transformed, except absolute latitude which was only z-transformed. Bolded estimates represent coefficients that where their credible intervals do not overlap zero. Note that estimation of an intercept was suppressed therefore trait coefficients represent means. For more details see main text. Age at maturity Estimate Lower Upper

Mean age at maturity (months) 0.38 -0.45 1.27 Metabolic rate -0.05 -0.38 0.24 Body mass -0.18 -0.45 0.07 Reproductive investment -0.11 -0.23 -0.01 Active months per year -0.22 -0.39 -0.03 Absolute latitude 0.22 0.07 0.4 Research effort -0.04 -0.16 0.06

197

Longevity Estimate Lower Upper

Mean lifespan (years) 0.43 -0.36 1.28 Metabolic rate -0.18 -0.45 0.07 Body mass 0.6 0.26 0.94 Reproductive investment -0.01 -0.12 0.1 Active months per year -0.08 -0.27 0.13 Absolute latitude 0.13 -0.03 0.31 Research effort 0.15 0.06 0.24

Table S4 Pooled model coefficients from twenty Bayesian bivariate generalised mixed models using twenty imputed datasets. This model tests the relative importance of metabolic rate, reproductive investment, active months per year and latitude in promoting covariation in age at maturity and longevity in viviparous squamates and the tuatara (n = 114). Body mass and research effort were included as covariates. All variables were log10 transformed and then z-transformed, except absolute latitude which was only z-transformed. Bolded estimates represent coefficients that where their credible intervals do not overlap zero. Note that estimation of an intercept was suppressed therefore trait coefficients represent means. For more details see main text. Age at maturity Estimate Lower Upper

Mean age at maturity (months) 0.55 -0.58 1.69 Metabolic rate -0.18 -0.77 0.42 Body mass 0.62 -0.05 1.26 Reproductive investment -0.04 -0.34 0.24 Active months per year -0.14 -0.41 0.14 Absolute latitude 0.24 -0.04 0.53 Research effort -0.07 -0.28 0.12 Longevity Estimate Lower Upper

Mean lifespan (years) 0.4 -0.67 1.52 Metabolic rate -0.37 -0.96 0.28 Body mass 0.94 0.22 1.54 Reproductive investment -0.02 -0.31 0.27

198

Active months per year -0.05 -0.33 0.23 Absolute latitude 0.19 -0.11 0.49 Research effort 0.13 -0.08 0.32

Table S5. Pooled model coefficients from twenty Bayesian bivariate generalised mixed models using twenty imputed datasets. This model tests the relative importance of metabolic rate, reproductive investment, active months per year and latitude in promoting covariation in age at maturity and longevity in lepidosaurs, irrespective of reproductive mode (n = 512). Body mass and research effort were included as covariates. All variables were log10 transformed and then z-transformed, except absolute latitude which was only z-transformed. Bolded estimates represent coefficients that where their credible intervals do not overlap zero. Note that estimation of an intercept was suppressed therefore trait coefficients represent means. For more details see main text. Age at maturity Estimate Lower Upper Mean age at maturity (months) 0.47 -0.44 1.41 Metabolic rate -0.11 -0.43 0.18 Body mass 0.62 0.3 0.96 Reproductive investment -0.13 -0.25 -0.02 Active months per year -0.18 -0.37 0 Absolute latitude 0.25 0.1 0.42 Research effort -0.05 -0.16 0.04 Longevity Estimate Lower Upper Mean lifespan (years) 0.51 -0.32 1.38 Metabolic rate -0.22 -0.52 0.06 Body mass 0.81 0.45 1.09 Reproductive investment -0.03 -0.13 0.08 Active months per year -0.06 -0.26 0.15 Absolute latitude 0.15 0.01 0.34 Research effort 0.15 0.06 0.23

199

Table S6 Pooled model coefficients a Bayesian phylogenetic mixed model using a maximum likelihood algorithm which incorporates phylogenetic relatedness. This model tests the relative importance of metabolic rate, reproductive investment, active months per year and latitude in promoting covariation in age at maturity and longevity in oviparous squamates (n = 398). Body mass and research effort were included as covariates. All variables were log10 transformed and then z-transformed, except absolute latitude which was only z-transformed. Bolded estimates represent coefficients that where their credible intervals do not overlap zero. Note that estimation of an intercept was suppressed therefore trait coefficients represent means. For more details see main text. Age at maturity Estimate Lower Upper

Mean age at maturity (months) 0.61 -0.46 1.7 Metabolic rate 0.08 -0.21 0.38 Body mass 0.73 0.42 1.04 Reproductive investment -0.06 -0.14 0.02 Active months per year -0.48 -0.61 -0.36 Absolute latitude 0 -0.12 0.11 Research effort -0.01 -0.07 0.06 Longevity Estimate Lower Upper

Mean lifespan (years) 0.61 -0.22 1.42 Metabolic rate -0.21 -0.51 0.08 Body mass 1.04 0.73 1.35 Reproductive investment 0.04 -0.04 0.12 Active months per year -0.2 -0.33 -0.07 Absolute latitude -0.01 -0.13 0.1 Research effort 0.2 0.13 0.27

200

Table S7 Pooled model coefficients a Bayesian phylogenetic mixed model using a maximum likelihood algorithm which incorporates phylogenetic relatedness. This model tests the relative importance of metabolic rate, reproductive investment, active months per year and latitude in promoting covariation in age at maturity and longevity in viviparous squamates and the tuatara (n = 114). Body mass and research effort were included as covariates. All variables were log10 transformed and then z-transformed, except absolute latitude which was only z-transformed. Bolded estimates represent coefficients that where their credible intervals do not overlap zero. Note that estimation of an intercept was suppressed therefore trait coefficients represent means. For more details see main text. Age at maturity Estimate Lower Upper

Mean age at maturity (months) 0.84 -0.88 2.63 Metabolic rate 0.2 -0.41 0.79 Body mass 0.4 -0.22 1.02 Reproductive investment 0.04 -0.27 0.35 Active months per year -0.48 -0.79 -0.17 Absolute latitude -0.12 -0.36 0.12 Research effort 0.02 -0.1 0.13 Longevity Estimate Lower Upper

Mean lifespan (years) 0.65 -0.65 2 Metabolic rate -0.15 -0.82 0.54 Body mass 0.85 0.15 1.52 Reproductive investment 0.03 -0.26 0.34 Active months per year -0.16 -0.47 0.15 Absolute latitude -0.02 -0.26 0.23 Research effort 0.23 0.09 0.37

201

Table S8 Pooled model coefficients a Bayesian phylogenetic mixed model using a maximum likelihood algorithm which incorporates phylogenetic relatedness. This model tests the relative importance of metabolic rate, reproductive investment, active months per year and latitude in promoting covariation in age at maturity and longevity in lepidosaurs, irrespective of reproductive mode (n = 514). Body mass and research effort were included as covariates. All variables were log10 transformed and then z-transformed, except absolute latitude which was only z-transformed. Bolded estimates represent coefficients that where their credible intervals do not overlap zero. Note that estimation of an intercept was suppressed therefore trait coefficients represent means. For more details see main text. Age at maturity Estimate Lower Upper

Mean age at maturity (months) 0.69 -0.48 1.88 Metabolic rate 0.09 -0.18 0.36 Body mass 0.65 0.37 0.93 Reproductive investment -0.08 -0.16 0.01 Active months per year -0.46 -0.58 -0.34 Absolute latitude 0 -0.1 0.1 Research effort -0.01 -0.07 0.05 Longevity Estimate Lower Upper

Mean lifespan (years) 0.68 -0.2 1.55 Metabolic rate -0.18 -0.45 0.09 Body mass 0.98 0.7 1.26 Reproductive investment 0.03 -0.05 0.11 Active months per year -0.18 -0.3 -0.06 Absolute latitude 0.01 -0.09 0.11 Research effort 0.2 0.14 0.27

202

Supplementary Materials References

Gillooly, J. F., Brown, J. H., West, G. B., Savage, V. M., & Charnov, E. L. (2001). Effects of size and temperature on metabolic rate. Science, 293(5538), 2248–2251. https://doi.org/10.1126/science.1061967 Lighton, J. R. B. (2008). Measuring Metabolic Rates. Oxford University Press. Lobo, F., Barrasso, D. A., Paz, M., & Basso, N. G. (2018). Phylogenetic relationships within a Patagonian clade of reptiles (Liolaemidae: Phymaturus) based on DNA sequences and morphology. Journal of Zoological Systematics and Evolutionary Research, 56(4), 549–569. https://doi.org/10.1111/jzs.12221 Meiri, S. (2018). Traits of lizards of the world: Variation around a successful evolutionary design. Global Ecology and Biogeography, 27(10), 1168–1172. https://doi.org/10.1111/geb.12773 Michonneau, F., Brown, J. W., & Winter, D. J. (2016). rotl: An R package to interact with the Open Tree of Life data. Methods in Ecology, 7(12), 1476– 1481. https://doi.org/10.1111/2041-210X.12593 Paradis, E., & Schliep, K. (2019). ape 5.0: An environment for modern phylogenetics and evolutionary analyses in R. Bioinformatics, 35(3), 526– 528. https://doi.org/10.1093/bioinformatics/bty633 Uyeda, J. C., Pennell, M. W., Miller, E. T., Maia, R., & McClain, C. R. (2017). The evolution of energetic scaling across the vertebrate tree of life. The American Naturalist, 190(2), 185–199. https://doi.org/10.1086/692326 White, C. R., Phillips, N. F., & Seymour, R. S. (2006). The scaling and temperature dependence of vertebrate metabolism. Biology Letters, 2(1), 125–127. https://doi.org/10.1098/rsbl.2005.0378

203

CHAPTER 6

Conclusions and Directions

204

Conclusions and Directions

Understanding the evolutionary consequences of early life experiences on phenotypic development is a significant task (Monaghan, 2008; West-Eberhard, 2005). The goal of my thesis was to gain insight into the impacts of developmental temperature on consistent inter-individual variation and phenotypic plasticity of metabolism and life history. Pace-of-life theory predicts that metabolic rate covaries with morphology, behaviour and life history (Biro & Stamps, 2010; Ricklefs & Wikelski, 2002). As such, focusing on how developmental temperature influences metabolism, may enable us to better predict the cascading effects it has other aspects of the phenotype, such as growth, age at maturity and lifespan. Furthermore, the effects of developmental temperature on repeatability and genetic variation can tell us how much substrate is available for selection to work with, and by extension, the evolutionary potential of traits in different developmental environments (Falconer, 1952; Lynch & Walsh, 1998).

My work demonstrated that plasticity in metabolism and growth at both the phenotypic and genotypic level is impervious to changes in developmental temperatures (Chapter 2, 3). Experimentally manipulating the developmental temperature of lizard embryos had no impact on the thermal reaction norm of metabolic rate nor the repeatability of reaction norm slopes (Chapter 2). This suggests that in Lampropholis delicata, developmental effects on plastic responses are likely to be canalised under naturalistic nest temperatures, but may play a bigger role under more evolutionary novel conditions (Crispo, 2007; Ghalambor et al., 2007). Nonetheless, individual metabolic plasticity was repeatable in L. delicata implying that thermal reaction norms can still be moulded by selection (Chapter 1, 2) (Falconer, 1952; Wilson, 2018). The impacts of developmental temperature manifested in subtle ways. Temperature effects on developmental rate resulted in considerable differences in incubation duration (Chapter 2, 3), such that prolonged development under cool temperatures induced more consistent individual variation in metabolic rate (Chapter 2) and larger hatching masses (Chapter 3). While ‘hot’ embryos were thermally restricted to hatch early (Chapter 2, 3), ‘cold’ embryos may be more constrained by the amount of yolk deposited by their mothers. Indeed, incubation duration among reptiles is a positively related to the amount of yolk found in eggs (Aubret et al., 2017; Warner & Shine, 2007). This suggests that

205

maternal investment into eggs, coupled with thermodynamic constraints of physiological rates during embryonic development, may be an important source of persistent phenotypic variability in L. delicata (Angilletta et al., 2000; Salin et al., 2019). After hatching, the environment can still exert many challenges on animals that can play a role in shaping their life history strategies. In Chapter 5, I found using data across 500 species of snakes, lizards and the tuatara, that metabolic rate had weak explanatory power in describing patterns of interspecific variation in age at maturity and lifespan. My results strongly suggest that environmental constraints on activity patterns, energy acquisition and allocation in reproductive investment may be key drivers of life history variation.

The environment influenced metabolism and life history in ways that I did not expect. Placing embryos at different incubation temperatures did not set them on divergent developmental pathways where suites of correlated traits are altered (West-Eberhard, 2003). Nonetheless, it seems that thermodynamic effects on embryonic metabolism and rates of development play a pivotal role in development plasticity of body size in L. delicata. Indeed, comparative studies have shown that development (passing through embryonic stages) is more thermally sensitive compared to growth (increases in body size) which can explain why we did not observe the developmental effects on growth post-hatching but found distinctive patterns of mass variation known as the temperature-size-rule (Forster et al., 2011; Forster & Hirst, 2012). Presumably, hot incubated embryos depleted yolk resources more quickly in order to sustain their elevated rates of energy expenditure (Booth et al., 2000). Furthermore, yolk assimilation and energy production have been shown to be operate more efficiently in cold temperatures which may have contributed to larger hatching masses (Salin, Auer, Anderson, et al., 2016; Storm & Angilletta, 2007). Given that body mass is an important fitness related trait, it is possible that the long-lasting impacts of developmental temperature on body mass can influence life history and fitness traits such as age at sexual maturity (Chapter 5), fecundity (Roitberg et al., 2015) and resource holding potential (Kar et al., 2016).

Identifying subtle changes, like the treatment difference in hatching mass, can provide important clues on finding the mechanistic underpinnings that orchestrate phenotypic changes during development. Predicting how populations change and respond under fluctuating environments is a formidable challenge which requires

206

examination across a wide range of taxa. It will also require a deeper investigation into the specific cellular and molecular mechanisms underlying such phenotypic change. Generally speaking, situations where biology does not fit our expectations are where we are likely to find some interesting answers and new research directions. Below I have put forward some ideas for future research more broadly as well as some new ideas in L. delicata that seem fruitful to address in the future. I hope these directions will further our understanding of how the complexities of the environment can shape phenotypic development.

Beyond Metabolic Rate: Other Mechanisms at Play

Energetic constraints and oxidative costs are central themes for pace-of-life theory and life history evolution (Careau et al., 2008; Finkel & Holbrook, 2000; Ricklefs & Wikelski, 2002). Metabolic rate has been assumed to be a reliable correlate for a number of biochemical processes that dictate energy production and oxidative stress. However, growing evidence suggests that metabolic rate is an imperfect proxy and may explain why the relationship between metabolism and life history is so variable across species. For example, high metabolic rates do not necessarily equate to high ATP production due to variation in mitochondrial efficiency (Hood et al., 2018). In other words, two individuals that have the same level of energy expenditure can differ in the amount of ATP produced per O2 molecule consumed (Salin, Auer, Rudolf, et al., 2016). Furthermore, high rates of energy expenditure may not be associated with elevated oxidative costs that is predicted constrain life history (Brand, 2000). Individuals with high metabolic rates can also display increased levels of uncoupling which can reduce cellular membrane potential and lower levels of reactive oxygen species (Salin et al., 2015). In order for us to truly understand how energy acquisition-allocation trade-offs govern phenotypic variation in life history traits across environments, we will require measurements of specific biochemical processes in conjunction with metabolic rate. Identifying these mechanisms and their relationship with metabolism enables more targeted experimental manipulations in L. delicata and other model systems. For instance, metabolism, ATP production and levels of reactive oxygen species are predicted to increase at high developmental temperatures, what would the phenotypic consequences if one supplies antioxidants to individuals at hot temperatures? Does that compensate or mitigate the effects of potential increased oxidative stress? Decoupling the complex relationships between

207

biochemical processes can help us better understand how phenotypic variation along the slow-fast continuum.

The Role of Maternal Investment

From Chapter 4, maternal effects appear to play a dynamic role in shaping offspring body size. Mothers can influence offspring phenotype by trading-off investment in offspring size or in the number of offspring she has across the season (Warne & Charnov, 2008). L. delicata has an impressive reproductive season (Chapple et al., 2014). Oviposition takes place from September to March in the Sydney population I was sampling from. This poses an interesting opportunity to investigate the role of maternal investment and its interaction with seasonal variation in nest temperatures. Mothers that mate early in the season have the potential to double clutch in later months when ambient temperatures are likely to be different. What are the factors that prompt mothers to have a second clutch? Thermal variability might be an important cue that can inform mothers of potential thermal conditions later in the season (Bonamour et al., 2019). The risk for mothers to spread her investment across two clutches is likely to increase if temperatures become more variable across the season. However, mothers may be able to offset harsher or more variable conditions later in the season by increasing investment more into her eggs. Furthermore, eggs laid at different times of the season are likely to experience contrasting early environments that are predicted to shape reversible plasticity later in life (Beaman et al., 2016). Understanding these complex interactions will require knowledge about how maternal investment in egg mass or yolk composition changes across the season which is currently unknown in L.delicata.

Pace of life and Phenotypic Integration

Traditionally, the slow-fast continuum was primarily used to understand variation in metabolism and life-history traits (Ricklefs & Wikelski, 2002; Sæther, 1987). However, many aspects of the phenotype are also limited by energy metabolism and are predicted to also vary along the slow-fast continuum. Recently, pace-of-life syndromes have been extended to include consistent individual variation in behaviour (‘animal personality’, Réale et al., 2010), thermoregulatory traits (Goulet et al., 2017) and cognitive traits (Sih & Del Giudice, 2012). For example, in L.delicata, fast-paced individuals have a preference for lower temperatures; tend to be more

208

active, explorative, social and bold; and learning at a slower rate compared to slow individuals (Goulet et al., 2017, 2018). Expansion of the pace-of-life framework permits more opportunities to understand how energy acquisition-allocation trade- offs influence phenotypic integration in different environments. Indeed, the magnitude of behavioural correlations vary across wild populations of L.delicata which suggests they may be environmentally sensitive (Michelangeli et al., 2019). Lampropholis delicata is a very amenable system to explore the impacts of developmental environments on phenotypic integration. Early life environments, such as incubation temperature or maternal investment into eggs, could influence how physiological and neurological systems are formed and maintained which could lead to cascading effects on thermoregulatory traits and behaviour. Elucidating the environmental factors that dampen or strengthen trait correlations will not only be important for understanding the phenotype variation in natural settings but also predicting how integrated traits respond to environmental perturbations.

Down to the Details: Environmental Correlates of Latitude

A major conclusion I reached in my comparative analysis was that the environment plays an overarching role in shaping life history variation. However, latitude is too coarse of an environmental variable to elucidate what specific factors limit energy acquisition at the species level. For example, temperature, rainfall and net primary productivity are all correlated with latitude. From this point forward, it would be valuable to obtain more specific environmental data to more rigorously test the ‘environmental constraints’ hypothesis using a comparative approach. Insight from a more comprehensive comparative analysis can also inform on experimental designs that aim to identify the molecular, cellular or physiological mechanisms that are environmentally sensitive and responsible for creating the diversity in life history in the animal kingdom.

Conclusions

My thesis examines the impacts of developmental temperature on individual variation and phenotypic plasticity of metabolism and life history. The environment is inextricably dynamic and complex; not only does it generate phenotypic variation and at times, genotypic variation but it is also responsible for eroding some of the variation away. Phenotypic plasticity can manifest at different stages of ontogeny

209

and may enable individuals to cope and adapt to changing conditions. Pace-of-life theory is a simple and intuitive framework to help us predict how developmental environments might affect aspects of the phenotype via the effects on metabolism. However, in reality the factors that govern phenotypic integration are likely far more elaborate and will involve both abiotic and biotic factors. Regardless, I hope my work has inspired new research avenues in the field of evolutionary biology to better understand the intricacies of how the environment influences phenotypic development. An interdisciplinary approach that combines research synthesis, manipulative experiments under naturalistic conditions and molecular tools will be necessary to make sense of the eco-evolutionary consequences of developing under fluctuating environments.

210

References

Angilletta, M. J., Winters, R. S., & Dunham, A. E. (2000). Thermal effects on the energetics of lizard embryos: Implications for hatchling phenotypes. Ecology, 81(11), 2957–2968. https://doi.org/10.1890/0012- 9658(2000)081%5B2957:TEOTEO%5D2.0.CO;2/full Aubret, F., Bignon, F., Bouffet-Halle, A., Blanvillain, G., Kok, P. J. R., & Souchet, J. (2017). Yolk removal generates hatching asynchrony in snake eggs. Scientific Reports, 7(1), 3041. https://doi.org/10.1038/s41598-017-03355-y Beaman, J. E., White, C. R., & Seebacher, F. (2016). Evolution of plasticity: Mechanistic link between development and reversible acclimation. Trends Ecology and Evolution, 31(3), 237–249. https://doi.org/10.1016/j.tree.2016.01.004 Biro, P. A., & Stamps, J. A. (2010). Do consistent individual differences in metabolic rate promote consistent individual differences in behavior? Trends Ecology and Evolution, 25(11), 653–659. https://doi.org/10.1016/j.tree.2010.08.003 Bonamour, S., Chevin, L.-M., Charmantier, A., & Teplitsky, C. (2019). Phenotypic plasticity in response to climate change: The importance of cue variation. Philosophical Transactions of the Royal Society B: Biological Sciences, 374(1768), 20180178–12. https://doi.org/10.1098/rstb.2018.0178 Booth, D. T., Thompson, M. B., & Herring, S. (2000). How incubation temperature influences the physiology and growth of embryonic lizards. Journal of Comparative Physiology B, 170(4), 269–276. https://doi.org/10.1007/s003600000097 Brand, M. D. (2000). Uncoupling to survive? The role of mitochondrial inefficiency in ageing. Experimental Gerontology, 35(6), 811–820. https://doi.org/10.1016/S0531-5565(00)00135-2 Careau, V., Thomas, D., Humphries, M. M., & Réale, D. (2008). Energy metabolism and animal personality. Oikos, 117(5), 641–653. https://doi.org/10.1111/j.0030-1299.2008.16513.x Chapple, D. G., Miller, K. A., Chaplin, K., Barnett, L., Thompson, M. B., & Bray, R. D. (2014). Biology of the invasive delicate skink (Lampropholis delicata) on Lord Howe Island. Australian Journal of Zoology, 62(6), 498–506. https://doi.org/10.1071/ZO14098

211

Crispo, E. (2007). The Baldwin effect and genetic assimilation: Revisiting two mechanisms of evolutionary change mediated by phenotypic plasticity. Evolution, 61(11), 2469–2479. https://doi.org/10.1111/j.1558- 5646.2007.00203.x Falconer, D. S. (1952). The problem of environment and selection. The American Naturalist, 86(830), 293–298. Finkel, T., & Holbrook, N. J. (2000). Oxidants, oxidative stress and the biology of ageing. Nature, 408(6809), 239–247. https://doi.org/10.1038/35041687 Forster, J., & Hirst, A. G. (2012). The temperature-size rule emerges from ontogenetic differences between growth and development rates. Functional Ecology, 26(2), 483–492. https://doi.org/10.1111/j.1365- 2435.2011.01958.x Forster, J., Hirst, A. G., & Woodward, G. (2011). Growth and Development Rates Have Different Thermal Responses. The American Naturalist, 178(5), 668–678. https://doi.org/10.1086/662174 Ghalambor, C. K., McKay, J. K., Carroll, S. P., & Reznick, D. N. (2007). Adaptive versus non-adaptive phenotypic plasticity and the potential for contemporary adaptation in new environments. Functional Ecology, 21(3), 394–407. https://doi.org/10.1111/j.1365-2435.2007.01283.x Goulet, C. T., Michelangeli, M., Chung, M., Riley, J. L., Wong, B. B. M., Thompson, M. B., & Chapple, D. G. (2018). Evaluating cognition and thermal physiology as components of the pace-of-life syndrome. Evolutionary Ecology, 32(5), 469–488. https://doi.org/10.1007/s10682-018- 9948-1 Goulet, C. T., Thompson, M. B., Michelangeli, M., Wong, B. B. M., & Chapple, D. G. (2017). Thermal physiology: A new dimension of the pace-of-life syndrome. Journal of Animal Ecology, 86(5), 1269–1280. https://doi.org/10.1111/1365-2656.12718 Hood, W. R., Austad, S. N., Bize, P., Jimenez, A. G., Montooth, K. L., Schulte, P. M., Scott, G. R., Sokolova, I., Treberg, J. R., & Salin, K. (2018). The mitochondrial contribution to animal performance, adaptation, and life- history variation. Integrative and Comparative Biology, 58(3), 480–485. https://doi.org/10.1093/icb/icy089 Kar, F., Whiting, M. J., & Noble, D. W. A. (2016). Influence of prior contest experience and level of escalation on contest outcome. Behavioral Ecology

212

and Sociobiology, 70(10), 1679–1687. https://doi.org/10.1007/s00265-016- 2173-4 Lynch, M., & Walsh, B. (1998). Genetics and analysis of quantitative traits. Oxford University Press. Michelangeli, M., Chapple, D. G., Goulet, C. T., Bertram, M. G., & Wong, B. B. M. (2019). Behavioral syndromes vary among geographically distinct populations in a reptile. Behavioral Ecology, 30(2), 393–401. https://doi.org/10.1093/beheco/ary178 Monaghan, P. (2008). Early growth conditions, phenotypic development and environmental change. Philosophical Transactions of the Royal Society B: Biological Sciences, 363(1497), 1635–1645. https://doi.org/10.1098/rstb.2007.0011 Réale, D., Garant, D., Humphries, M. M., Bergeron, P., Careau, V., & Montiglio, P.-O. (2010). Personality and the emergence of the pace-of-life syndrome concept at the population level. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 365(1560), 4051–4063. https://doi.org/10.1098/rstb.2010.0208 Ricklefs, R. E., & Wikelski, M. (2002). The physiology/life- history nexus. Trends Ecology and Evolution, 17(10), 1–7. https://doi.org/10.1016/S0169- 5347(02)02578-8 Roitberg, E. S., Eplanova, G. V., Kotenko, T. I., Amat, F., Carretero, M. A., Kuranova, V. N., Bulakhova, N. A., Zinenko, O. I., & Yakovlev, V. A. (2015). Geographic variation of life-history traits in the sand lizard, Lacerta agilis: Testing Darwin’s fecundity-advantage hypothesis. Journal of Evolutionary Biology, 28(3), 613–629. https://doi.org/10.1111/jeb.12594 Sæther, B.-E. (1987). The influence of body weight on the covariation between reproductive traits in European birds. Oikos, 48(1), 79–88. https://doi.org/10.2307/3565691 Salin, K., Auer, S. K., Anderson, G. J., Selman, C., & Metcalfe, N. B. (2016). Inadequate food intake at high temperatures is related to depressed mitochondrial respiratory capacity. The Journal of Experimental Biology, 219(9), 1356–1362. https://doi.org/10.1242/jeb.133025 Salin, K., Auer, S. K., Rudolf, A. M., Anderson, G. J., Cairns, A. G., Mullen, W., Hartley, R. C., Selman, C., & Metcalfe, N. B. (2015). Individuals with

213

higher metabolic rates have lower levels of reactive oxygen species in vivo. Biology, 11(9), 1–4. https://doi.org/10.1098/rsbl.2015.0538 Salin, K., Auer, S. K., Rudolf, A. M., Anderson, G. J., Selman, C., & Metcalfe, N. B. (2016). Variation in metabolic rate among individuals is related to tissue-specific differences in mitochondrial leak respiration. Physiological and Biochemical Zoology, 89(6), 511–523. https://doi.org/10.1086/688769 Salin, K., Villasevil, E. M., Anderson, G. J., Lamarre, S. G., Melanson, C. A., McCarthy, I., Selman, C., & Metcalfe, N. B. (2019). Differences in mitochondrial efficiency explain individual variation in growth performance. Proceedings of the Royal Society B: Biological Sciences, 286(1909), 1–8. https://doi.org/10.1098/rspb.2019.1466 Sih, A., & Del Giudice, M. (2012). Linking behavioural syndromes and cognition: A behavioural ecology perspective. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences, 367(1603), 2762– 2772. https://doi.org/10.1098/rstb.2012.0216 Storm, M. A., & Angilletta, M. J. (2007). Rapid assimilation of yolk enhances growth and development of lizard embryos from a cold environment. The Journal of Experimental Biology, 210(19), 3415–3421. https://doi.org/10.1242/jeb.005652 Warne, R. W., & Charnov, E. L. (2008). Reproductive Allometry and the Size- Number Trade-Off for Lizards. The American Naturalist, 172(3), 80–98. https://doi.org/10.1086/589880 Warner, D. A., & Shine, R. (2007). Fitness of juvenile lizards depends on seasonal timing of hatching, not offspring body size. Oecologia, 154(1), 65–73. https://doi.org/10.1007/s00442-007-0809-9 West-Eberhard, M. J. (2003). Developmental Plasticity and Evolution. Oxford University Press. West-Eberhard, M. J. (2005). Developmental plasticity and the origin of species differences. Proceedings of the National Academy of Sciences, 102(Supplement 1), 6543–6549. https://doi.org/10.1073/pnas.0501844102 Wilson, A. J. (2018). How should we interpret estimates of individual repeatability? Evolution Letters, 219, 631–635. https://doi.org/10.1002/evl3.40

214

Appendix

The following section contains a list of presentations I gave throughout my candidature, as well as research articles I contributed to that are either published or accepted

215

Presentations

Slides for each presentation can be found at https://fontikar.wordpress.com/presentations/

2017 Individual variation in thermal plasticity. Oral presentation for Australasian Evolution Society, Tasmania, 2017 *Best speed talk award

Live fast die young: The role of developmental temperature on the evolution of pace-of-life syndromes. Oral presentation for UNSW Postgraduate Research Forum, 2017 *Outstanding Ecology and Evolution talk award

2018 What does your athletic ability say about your fitness? Invited guest lecture for BIOL260 Science of Sex at Macquarie University, 2018

Learning from lizards Invited public lecture for the Australian Herpetological Society, 2018

Life in cold blood: does metabolic rate predict age at maturity and longevity? Poster presentation at Joint Congress of Evolutionary Biology, Montpellier, 2018

Individual variation in thermal plasticity. Oral presentation for UNSW Postgraduate Research Forum, 2018 *Runner up for Outstanding Ecology and Evolution talk award

2019 Life in cold blood. Oral presentation for UNSW Postgraduate Research Forum, 2019 *Winner of the 3-minute thesis competition

Life in cold blood: correlation of metabolic rate and life history. Oral presentation for Australasian Evolution Society, 2019 *Honourable mention for best short talk

Life in cold blood: covariation of metabolic rate and life history. Oral presentation for ANU, Research School of Biology HDR conference, 2019 *Runner up for best student talk

How smart can they be? Invited public outreach for Pint of Science, Nags Head Hotel, Glebe, Sydney 2019

216

Other Research Articles

217