Morphological Variation and Diversification in Australian Neriid Flies

Morphological Variation and Diversification in

Australian Neriid Flies

Elizabeth Cassidy

Supervised by: Russell Bonduriansky

THESIS SUBMITTED FOR THE DEGREE OF MASTER OF PHILOSOPHY Evolution and Ecology Research Centre School of Biological, Earth and Environmental Sciences University of New South Wales October 2012

PLEASE TYPE THE UNIVERSITY OF NEW SOUTH WALES Thesis/Dissertation Sheet

Surname or Family name: Cassidy

First name: Elizabeth Other name/s: Jane Gerard

Abbreviation for degree as given in the University calendar: MPhil

School: School of Biological, Earth & Environmental Faculty: Science Sciences; Evolution & Ecology Research Centre

Title: Morphological Variation and Diversification in Australian Neriid Flies

Abstract 350 words maximum: (PLEASE TYPE) Morphological variation is a result of complex interactions between physiological constraints, selection pressures and ecological conditions. All of these factors are vital in the understanding of the evolution of morphological adaptations. In this thesis, I examine three aspects of the phenotypic plasticity and morphological variation in two species of neriid flies, Telostylinus angusticollis and Telostylinus lineolatus. Chapter one examines allometric constraints on the diversification of populations. Static allometry slope is generally thought to constrain adaptation and diversification. We examined the diversification of static allometry by manipulating larval nutrient concentration and comparing allometric slopes in sexual and non-sexual traits across populations. We found evidence of slope diversification within T. angusticollis and T. lineolatus in a sexual trait. Our results suggest the diversification of static allometry slope can be driven by sexual selection. Following this, chapter two discusses sexual selection and its impact on diversification in males and females. Using reaction norms for nutrient concentration in a range of sexual and non-sexual body shape components, we identify different patterns of morphological diversification between the sexes. In addition to this, the patterns of diversification seen in males suggest that sexual selection is acting upon male body shape as a whole, rather than specific morphological traits. We consider the ecological and selective forces contributing to the diversification of the sexes. Chapter three examines another aspect of larval ecology, group relatedness, and its benefits or disadvantages. We find that fly larvae gain an advantage from being housed with closely related individuals, and emerge larger as adults, congruent with the kin selection hypothesis. These three chapters outline some of the different factors contributing to morphological variation and highlight the importance and complexity of phenotypic plasticity.

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 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

iii

ABSTRACT

Morphological variation is a result of complex interactions between physiological constraints, selection pressures and ecological conditions. All of these factors are vital in the understanding of the evolution of morphological adaptations. In this thesis, I examine three aspects of the phenotypic plasticity and morphological variation in two species of neriid flies,

Telostylinus angusticollis and Telostylinus lineolatus. Chapter one examines allometric constraints on the diversification of populations. Static allometry slope is generally thought to constrain adaptation and diversification. We examined the diversification of static allometry by manipulating larval nutrient concentration and comparing allometric slopes in sexual and non-sexual traits across populations. We found evidence of slope diversification within T. angusticollis and T. lineolatus in a sexual trait. Our results suggest the diversification of static allometry slope can be driven by sexual selection. Following this, chapter two discusses sexual selection and its impact on diversification in males and females. Using reaction norms for nutrient concentration in a range of sexual and non-sexual body shape components, we identify different patterns of morphological diversification between the sexes. In addition to this, the patterns of diversification seen in males suggest that sexual selection is acting upon male body shape as a whole, rather than specific morphological traits. We consider the ecological and selective forces contributing to the diversification of the sexes. Chapter three examines another aspect of larval ecology, group relatedness, and its benefits or disadvantages. We find that fly larvae gain an advantage from being housed with closely related individuals, and emerge larger as adults, congruent with the kin selection hypothesis. These three chapters outline some of the different factors contributing to morphological variation and highlight the importance and complexity of phenotypic plasticity.

Contents

Introduction ...... 1 Diversification of static allometries in Australian neriid flies ...... 5 ABSTRACT ...... 6 INTRODUCTION ...... 7 METHODS ...... 10 RESULTS ...... 13 DISCUSSION ...... 18 REFERENCES ...... 23 Sex-specific patterns of diversification in reaction norms for morphological traits ... 26 ABSTRACT ...... 27 INTRODUCTION ...... 28 METHODS ...... 33 RESULTS ...... 36 DISCUSSION ...... 38 REFERENCES ...... 45 Larvae of the neriid fly Telostylinus angusticollis benefit by interacting with kin ...... 51 ABSTRACT ...... 52 INTRODUCTION ...... 53 METHODS ...... 56 RESULTS ...... 59 DISCUSSION ...... 62 REFERENCES ...... 70 Conclusion ...... 73 Acknowledgements ...... 77

Introduction

This thesis examines morphological variation and diversification in five populations of the

Australian neriid fly. Two broad areas are investigated: diversification that has occurred between populations over time, and the variation present in a population currently.

The motivation for the experiments and questions in this thesis stemmed from phenotypic plasticity present in Telostylinus angusticollis, a species collected in a nature reserve local to the University of New South Wales. Previous work investigating plasticity in response to nutrient intake has found this species to display condition-dependent sexual dimorphism

(Bonduriansky 2007a). T. angusticollis presents an interesting model species as it draws together a range of study areas including sexual selection, morphology, plasticity and condition dependence.

I was further motivated to investigate plasticity in this species following the establishment of four new laboratory populations, two additional populations of T. angusticollis and two populations of a congeneric species, T. lineolatus. Pilot studies undertaken on T. lineolatus had suggested that this species did not possess the same degree of sexual size and shape dimorphism, nor the size range present in the natural or laboratory populations of T. angusticollis.

Five populations of two similar species with divergent levels of phenotypic plasticity presented a good opportunity to consider the nature of phenotypic plasticity and its implications for evolution and diversification. The study of phenotypic plasticity has gained prominence in the past few decades. Environmental effects on phenotype, previously thought to act as a buffer to natural selection, are now being recognised as potential drivers of rapid diversification and adaptation that can promote the appearance of new phenotypes (Pigliucci 2001; Price et al.

2003; West-Eberhard 2003; Pfennig et al. 2010). I sought to contribute to the study of

1 phenotypic plasticity by utilising the laboratory populations of T. lineolatus and T. angusticollis and the contrasting levels of nutrient-dependent plasticity they possessed.

Previous work on the Sydney population of T. angusticollis has identified the importance of larval-diet quality in morphological variation. High- and low-nutrient diets have been demonstrated to have a range of complex effects on adult T. angusticollis (Bonduriansky

2007a) . To expand on this, we added another diet concentration, manipulated the diet of all five populations of Telostylinus species, and collected data on a range of morphological traits.

When examining the data from this experiment, two important features stood out: the diversification of the allometric slopes and the different patterns in diversification between the sexes. As these results were based in quite different areas of study, the results from this experiment were split into two chapters dealing with these different topics. The first discusses one particular aspect of morphological variation, static allometry, and its diversification across populations. The second chapter examines sex-specific differences in diversification, as measured by reaction norms. While the first two chapters discuss morphological variation and plasticity in response to nutrient intake, the final chapter investigates how another aspect of the larval environment – relatedness among individuals – affects phenotypic plasticity.

Chapter One discusses the differences we found in the allometric slopes across the five populations. Using an analysis targeted at detecting differences in allometric slope, we consider the biological constraints on static allometric slope, and the selective forces which act upon the diversification or maintenance of allometry.

Chapter Two looks more broadly at the patterns in diversification seen across the five populations of Telostylinus species. We consider how sexual selection acts to drive differences in diversification between the sexes by examining diversification of reaction norms for a range of morphological traits. This chapter attempts to draw together a range of issues contributing

2 to the different patterns of diversification seen between the sexes: condition dependence, sexual and viability selection, and sexual shape dimorphism. This chapter also demonstrates how the use of reaction norms (as a measure of plasticity) may be used to evaluate diversification.

Chapters One and Two deal with morphological variation in response to larval diet, which has already been shown to be a source of phenotypic plasticity. However, the variation present in genetically-similar flies when fed a nutrient-restricted diet suggested that nutrient level was not the only environmental factor which contributed to morphological differences between conspecifics. In Chapter Three, I report results of an experiment designed to examine a different aspect of the larval environment, group relatedness. Kin selection theory predicts that individuals will benefit from being housed in a group of related individuals, as the group will act more altruistically and less competitively (Hamilton 1964). Contrary to this, there have been a range of studies, including in flies (Lopez-Suarez et al. 1993), which suggest that it is more beneficial to be housed in a group of high genetic diversity (Maynard Smith 1978). This study is discussed in Chapter Three, which reports the findings of two blocks of experiments carried out to determine the advantages or disadvantages of being housed with related individuals. We consider our results within the broader literature and findings from a range of taxa, and discuss the relative roles of kin cooperation and genetic diversity.

After setting out to investigate the evolutionary implications of phenotypic plasticity, this thesis presents three chapters concerning three distinct elements of plasticity of form.

Through the discussion of these different areas, I aim to highlight the complex interactions at play in determining the morphological variation of a species.

REFERENCES

Bonduriansky, R. (2007a). "The evolution of condition-dependent sexual dimorphism."

American Naturalist 169(1): 9-19.

Hamilton, W. D. (1964). "Genetical evolution of social behaviour I." Journal of Theoretical

Biology 7(1): 1-&.

Lopez-Suarez, C., et al. (1993). "Genetic heterogeneity increases viability in competing groups

of Drosophila hydei." Evolution 47(3): 977-981.

Maynard Smith, J. (1978). The evolution of sex Cambridge [Eng.] ; New York :, Cambridge

University Press.

Pfennig, D. W., et al. (2010). "Phenotypic plasticity's impacts on diversification and speciation."

Trends in Ecology & Evolution 25(8): 459-467.

Pigliucci, M. (2001). Phenotypic plasticity: beyond nature and nurture. . Baltimore, USA, Johns

Hopkins University Press.

Price, T. D., et al. (2003). "The role of phenotypic plasticity in driving genetic evolution."

Proceedings of the Royal Society B: Biological Sciences 270(1523): 1433-1440.

West-Eberhard, M. (2003). Developmental Plasticity and Evolution, Oxford University Press,

USA.

CHAPTER ONE

Diversification of static allometries in Australian neriid flies

Elizabeth J. Cassidy1, Eleanor Bath1, Stephen F. Chenoweth2, Russell

Bonduriansky1

1Evolution & Ecology Research Centre and School of Biological, Earth and

Environmental Sciences, The University of New South Wales, Sydney, New South

Wales 2052, Australia.

2 School of Integrative Biology, University of Queensland, Brisbane, Queensland

EJC, RB and SF conceived and designed the experiment, EJC and EB carried out the experiment, EJC and

RB analysed the data, EJC wrote the manuscript, and RB provided supervision and suggestions on the

manuscript.

ABSTRACT

Static allometry describes how a trait scales with increasing body size within a population. The slope of this relationship has long been thought to be under biological constraint. In this study, we investigated divergence of allometric slope among populations in two species of Australian neriid fly, Telostylinus angusticollis and Telostylinus lineolatus. We manipulated the nutrient concentration of the larval diet within each of five populations of these two species, and then tested for divergence among populations in reduced major axis allometric slopes, and compared divergence of sexual and non-sexual traits in males and females. We found divergence in male allometric slope of a sexual trait within both species, suggesting that sexual selection is able to drive the evolution of allometric slope.

INTRODUCTION

The correlation between body size and shape is critical to the understanding of morphological evolution, divergence and adaptation. Within any natural population, adult individuals will possess a range of body sizes. Variation in overall body size is accompanied with variation of size of individual body traits. The pattern of scaling of trait size with body size among individuals at the same developmental stage (e.g. adults) within a population is referred to as static allometry. This relationship can be described by the power function Y = aX b where Y is the structure whose increase is dependent on body size (X), and b is a parameter describing how the relative size of Y changes as X increases, or the allometric slope. When b=1 the trait scales in constant proportion with body size and is referred to as isometric, if b>1 the trait displays positive allometry and is relatively larger in larger individuals, and if b<1 the trait displays negative allometry and is relatively smaller in larger individuals. The parameter a indicates the size of trait Y when X = 1, or the allometric intercept. When log-transformed, this equation becomes a linear relationship: log (Y) = log(a) + blog(X).

The role of static allometries in constraining adaptation has been a subject for debate since this hypothesis was proposed by Huxley (1924; 1932). Much of the discussion that has emerged on this subject centres on the evolvability of the allometric slope. There has been a strong argument that allometric slope is either not a trait that is acted upon by selection, or it is a trait that cannot easily be shifted by selective pressures (Gould 1966; Maynard Smith et al.

1985). Gould argued that the allometric slope seen in populations is merely a result of mechanics of growth, and that substantial modifications of relative size are a consequence of change in intercept, while slope remains constant. Canalization of form is a result of stabilizing selection leading to greater developmental stability, especially under stressful conditions.

Internal developmental processes that maintain this advantageous, uniform growth could also act to conserve allometric slope (Maynard Smith et al. 1985).

Evolvability of allometric slopes is highly relevant for studies of sexual selection, as most examples of positive allometry are found in secondary-sexual traits (Green 1992; Kodric-Brown et al. 2006). This has led to the hypothesis that directional sexual selection leads to positive allometry, which is displayed by all sexually-selected traits. Contrary to this, there are many positively allometric traits that are not sexually selected, as well as secondary sexual traits that are isometric or negatively allometric (Bonduriansky 2007b). It is, however, to be expected that under specific conditions some types of sexually-selected traits are more likely than non- sexual traits to exhibit positive allometry (Bonduriansky and Day 2003). This is corroborated by experimental evidence that sexual selection can maintain positive allometry (Cayetano et al.

2011).

Although it has been suggested that allometric slope is a trait upon which natural selection may operate (Newell 1949), more recently changes in intercept as a response to sexual selection have been identified, without any change in allometric slope (Bonduriansky 2007b).

Experimental evolution studies have imposed selection regimes to select for change in allometric slope and intercept. These studies have found support for the hypothesis that the intercept responds more readily to selection (Tobler and Nijhout 2010; Egset et al. 2012), although Tobler and Nijhout (2010) also found a decrease in the slope of wing/body allometries when selecting for smaller body size in butterflies. Work examining variation in static allometries among species and populations also supports Huxley’s conserved allometry hypothesis. Species and populations inhabiting different environments and under disparate selection pressures were generally found to have a conserved allometric slope, despite differences in intercepts (Poulin 2009; Egset et al. 2011). Poulin (2009) also identified conserved allometric slope among different traits within species, possibly suggestive of an internal growth constraint linking development of multiple traits within the individual.

However, a recent phylogenetic comparative study of the static allometry of 30 species of

8 stalk-eyed flies showed considerable diversification of static allometry slope across species

(Voje and Hansen 2012). This diversification was evidence that static allometry did not constrain adaptation on a macroevolutionary time scale, but the authors speculate that static allometry could still act as a constraint over shorter periods of time. Despite this research, relatively few studies have investigated the diversification of allometric slopes. Moreover, whereas allometric slopes of secondary-sexual traits can be strongly dependent on the availability of dietary resources (Bonduriansky 2007a), studies of the diversification of static allometries have not incorporated effects of variation in diet.

A broadening of the definition of static allometry to refer to any change in shape with size has led to a divergence from Huxley’s rule and studies that do therefore not conform to ‘narrow sense’ static allometry. This has resulted in the use of models of slope on an arithmetic rather than logarithmic scale. This reduces the capacity for comparison with studies of true static allometry, as on an arithmetic scale the slope does not represent a straightforward index of the scaling of relative trait size with body size (Houle et al. 2011).

In the current study, we investigated the diversification of allometric slopes in five populations belonging to two species of Australian neriid fly, Telostylinus angusticollis and T. lineolatus. T. angusticollis is a saprophagus fly found in aggregations on the rotting bark of Acacia longifolia, where male-male fights take place for resource patches and access to female conspecifics.

Variation in diet quality has been shown to have strong phenotypic consequences in the

Sydney population (Bonduriansky 2007a). Sexual size and shape dimorphism is pronounced on a high-quality diet, which produces much larger males than females. High-quality diet males also display elongated heads and antennae, secondary sexual traits that are used in combat, as well as elongated forelegs used to guard females during oviposition. Flies reared on a low- quality diet are much smaller and sexual size and shape dimorphism is minimal. Larval diet quality has been shown to affect allometric slopes of sexual traits in a Sydney population of T.

9 angusticollis (Bonduriansky 2007a). The other known Australian neriid, T. lineolatus, in which allometric slopes have not been studied previously, was used for comparison. This species is found in North Queensland on rotting fruit material. Pilot studies suggested that T. lineolatus exhibits much less phenotypic plasticity of body size and shape in response to larval diet quality.

We aimed to detect the extent of the variability of the allometric slopes among populations of these species, as well as variation among populations in the response of allometric slope to larval diet quality, in order to gauge the evolutionary lability of the allometric slope. We manipulated the larval diet quality of flies from each of our five study populations (using three diets representing 6-fold variation in nutrient concentration), and compared allometric slopes among populations for several sexual and non-sexual traits in males and females. The effects of larval diet on static allometry slope represent a reaction norm, and diversification among populations in such effects reflects the evolution of reaction norms for allometric slope. To our knowledge, this is the first study that incorporates the use of dietary manipulation in an analysis of the diversification of static allometry.

METHODS

Experimental animals

The flies used in this experiment were descendants of flies collected in 2010 from five different geographic locations spanning 20 degrees of latitude. T. lineolatus adults were collected on rotting fruit in Kuranda and Cow Bay, Queensland. T. angusticollis individuals were collected from the bark of Acacia longifolia trees in Brisbane, Queensland, and Coffs Harbour and

Sydney in New South Wales. These populations were transported to the laboratory and kept in

10 cages with moistened cocopeat (Galuku, Pty, Sydney) and sugar. Approximately 10 individuals of each sex were collected to establish the lab populations.

The diet of the parental generation of the experimental flies was standardized by rearing two hundred eggs from each population on two litres of rich larval medium (see below). At emergence, males and females were separated. Two weeks after emergence, thirty males and thirty females from each population were combined in a ten litre cage with oviposition medium and allowed to mate and lay eggs for two days. After this time, oviposition medium was removed and eggs were transferred into the dietary manipulation (see below).

Dietary manipulation

Dietary manipulation was based on the larval diet used to rear stock flies. Although the wild diet in unknown, the laboratory diet has been found to produce the same range of phenotypes seen in the wild (Bonduriansky 2006; Bonduriansky 2007a). The larval medium is made up of cocopeat, water, soy protein powder (Nature’s way, Pharm-a-Care, Warriewood, Australia), organic liquid barley malt (Spiral Foods, Leichhardt Australia) and blackstrap sugarcane molasses (Conga Foods, Preston, Australia). Each of the five populations was reared on three different concentrations of diet, rich (R), poor (P) and very poor (V). The R diet was comprised of (per litre of cocopeat) 30mL malt, 30mL of molasses, 32g of protein powder and 800mL of water. The P and V diets were made up of one-third and one-sixth dilutions, respectively, of the malt, molasses and soy protein.

For each population, twenty randomly chosen eggs from a cage with 30 males and 30 females were transferred from multiple oviposition containers into each of thirty 250mL jars containing

200mL of rich (N=10 jars), poor (N=10 jars) or very poor (N=10 jars) larval medium. The eggs in each population group were transferred into treatments over several hours, on separate days due to different emergence times of the parental generation. Within populations, the jars that

11 the eggs were taken from were switched regularly, and the order that the eggs were placed into the different diet jars was randomized to avoid bias.

Morphometric data

Five to ten days after adult emergence, flies were frozen at -20°C for later quantification of body size and shape. From each replicate jar, two males and two females (where possible) were randomly selected for measurement. Each specimen was imaged using a Leica DFC420 camera mounted on a Leica MS5 stereomicroscope. From these images, nine measurements were taken for each fly: thorax length (TL), head length, head width at the widest point across the eyes (HW), antennae length (AL), fore tibia length (FL), mid tibia length (ML), hind tibia length (HTL), left wing length from the r-m cross vein to the wing margin (LW). Measurements were made using ImageJ analysis software (Rasband 1997-2009). For additional measurement methodology see Bonduriansky (2007a)

Analysis

All measurements were log-transformed prior to analysis. Allometric slope analysis was carried out on HL, HW, FL, ML, HTL and LW (AL was omitted because this trait could not be measured on many individuals). HL and FL were considered as sexual traits, due to the use of the head capsule and forelegs during male-male combat, as well as the use of the forelegs to mate guard during female oviposition (Bonduriansky 2007a). The remaining traits were considered non-sexual. TL was used as the index of body size, based on previous morphometric analysis, which found TL to load very strongly on the first principal component (Bonduriansky 2006).

Reduced major axis (RMA) slopes were calculated for each population, diet and sex combination based on replicate means. RMA slopes were used in the analyses rather than ordinary least square (OLS) slopes because data on both the x and y axes contained

12 measurement error (Seim and Saether 1983). Each slope was tested for difference from 1

(isometry) using software (model2CI) developed by David Warton.

The matrix of 180 RMA static allometry slopes (5 populations x 2 sexes x 6 traits x 3 diets) formed the data-set used to test for diversification among populations and species. This approach, rather than a more conventional ANCOVA-based analysis (which compares ordinary least-squares regression slopes), was used in order to investigate the diversification of static allometries calculated using RMA regression, which provides unbiased estimates of the scaling of trait size with body size (Seim and Saether 1983; McArdle 1988). ANOVAs on the RMA slopes were used to test for effects of population, diet and sex, all fitted as fixed factors, in separate analyses for each of the two species. In a separate analysis, both species were combined, with species, diet and sex fitted as fixed factors. All models were re-fitted after removing interactions with p > 0.2. Statistical analysis was performed using Statistica 7.1

(StatSoft, Tulsa, OK).

RESULTS

RMA slopes for each population x diet x sex combination are shown in table 1, with significant difference from 1 highlighted in bold. Significant positive allometry is only present in male slopes of the secondary sexual traits, HL and FL, and nearly all of these are in T. angusticollis.

Results from ANOVAs on RMA slopes of traits within and among species are summarized in table 2. These results reveal diversification within and among species in allometric slopes and reaction norms for allometric slopes. Significant divergence of slopes among populations was found in both T. angusticollis and T. lineolatus. However, in both species, most evidence of diversification (population effects and interactions) was found for a sexual trait, head length

(HL).

Divergence in mean population allometric slope of HL was detected in T. angusticollis (Fig. 1).

These population effects reflect the lower slopes of the Coffs Harbour population, relative to the Brisbane and Sydney populations.

Despite no overall population effect for HL within T. lineolatus, significant population x sex (Fig.

2) and population x diet interactions (Fig. 3) were observed. Cow Bay and Kuranda populations displayed substantially different allometric slopes for HL on poor and rich diet, reflecting diversification of the reaction norm for allometric slope. The significant population x sex interaction reflects a difference among populations in the degree of sexual dimorphism for static allometry slope.

Allometry of the other sexual trait, fore leg (FL), exhibited only a significant population x sex interaction in T. lineolatus, suggesting that sexual dimorphism for this trait’s allometry differs between the two populations of this species.

Likewise, allometry of non sexual traits, head width (HW), mid leg (ML), hind leg (HTL) and left wing (LW), was less diverse across populations and species. T. angusticollis populations showed significant divergence for LW. HW showed marginally insignificant effects of population x sex in T. lineolatus. ML showed a marginally insignificant population difference in

T. angusticollis and a significant population x diet interaction in T. lineolatus. A marginally insignificant species x diet interaction was also present for ML slopes. HTL displayed a population x sex interaction in T. lineolatus, as well as a species x diet interaction. A marginally insignificant species x diet and significant species x sex interactions were present for LW also.

Table 1. RMA slopes for males and females of the five populations on each diet treatment. Morphological trait codes refer to: HL- head length, HW- head width, FL- fore tibia length, ML- mid tibia length, HTL- hind tibia length, LW- left wing. Slopes significantly different from 1 are highlighted in bold. Population Diet HL HW FL ML HTL LW Females T. angusticollis Brisbane Very Poor 0.8591 0.6726 1.0107 1.0547 1.006 0.8916 Coffs Harbour Very Poor 0.7138 0.6997 1.0566 0.9836 1.0704 1.1648 Sydney Very Poor 0.8116 0.6828 0.8390 0.8658 0.8821 0.8102 Brisbane Poor 0.8305 1.191 1.0663 1.2514 1.1694 0.9556 Coffs Harbour Poor 0.8261 0.8237 0.9633 1.0161 1.007 0.8294 Sydney Poor 0.8416 0.7434 0.8606 0.8466 0.8497 0.7338 Brisbane Rich 0.9815 0.7758 1.1481 0.8984 0.7744 0.868 Coffs Harbour Rich 0.6623 0.8307 0.8023 0.8009 0.8147 1.2183 Sydney Rich 0.6623 0.9069 0.8632 0.8994 0.89 0.9244 T. lineolatus Cow bay Very Poor 0.8131 0.8979 0.7215 0.75 0.7775 0.6806 Kuranda Very Poor 0.7647 0.8498 0.7972 0.691 0.6758 0.6853 Cow bay Poor 1.0643 1.0884 1.0964 1.1883 1.056 0.938 Kuranda Poor 0.7191 0.8812 0.8996 0.8476 0.8743 0.8064 Cow bay Rich -0.8519 -1.1741 0.8798 0.9399 1.5418 0.5787 Kuranda Rich 1.0939 0.4845 1.1094 1.2527 1.0826 1.1348 Males T. angusticollis Brisbane Ver y Poor 1.1436 0.7233 1.2894 1.2047 1.1217 0.8561 Coffs Harbour Very Poor 0.6836 0.9853 0.7124 0.7264 0.6696 0.8912 Sydney Very Poor 1.0640 0.7756 1.1925 1.129 1.0933 0.7986 Brisbane Poor 1.3489 0.6953 1.1981 1.2026 1.2126 0.8529 Coffs Harbour Poor 1.2970 0.702 1.3392 1.2486 1.2858 0.9018 Sydney Poor 1.4642 0.7196 1.1902 1.1279 1.2487 0.7776 Brisbane Rich 1.2279 0.7141 1.0591 1.051 0.8871 0.4618 Coffs Harbour Rich 1.0908 0.9656 1.0600 0.7387 0.8476 1.0508 Sydney Rich 1.0645 0.626 1.0912 1.0603 1.0022 0.8835 T. lineolatus Cow bay Very Poor 0.9248 0.8502 1.0896 0.9947 0.9503 0.6935 Kuranda Very Poor 0.9864 0.8948 1.3449 1.1978 1.1371 0.8239 Cow bay Poor 1.1411 1.2157 1.2045 1.1334 1.0067 0.9027 Kuranda Poor 0.9391 1.2283 0.9835 0.9309 1.0429 0.8561 Cow bay Rich 0.8633 -0.6336 0.9225 0.9055 1.3591 1.452 Kuranda Rich 1.1824 0.8216 2.0753 1.9635 1.4523 1.4435

Table 2. Results of ANOVAs for each trait and species on RMA slopes. Significant results are shown in bold, marginally non-significant values (<0.08) are shown in brackets. Factor Trait HL HW FL ML HTL LW T.angusticollis Population P=0.0278 NS NS (P=0.0755) NS P=0.0251 Diet P=0.0189 NS NS (P=0.0549) P=0.0289 NS Sex P=0.0002 NS P=0.0401 NS NS NS Diet*sex P=0.0350 NS NS NS NS NS Population*diet NS NS NS NS NS NS Population*sex NS NS NS NS NS NS T.lineolatus Population NS NS NS NS NS NS Diet P=0.0157 NS NS NS P=0.0027 (P=0.0777) Sex P=0.0061 NS NS NS (P=0.0571) NS Diet*sex (P=0.0645) NS NS NS NS NS Population*diet P=0.0035 NS NS P=0.0472 NS NS Population*sex P=0.0287 (P=0.0790) P=0.0061 NS P=0.03827 NS Both species Species NS NS NS NS NS NS Diet (P=0.0506) P=0.0027 NS NS P=0.0264 P=0.0477 Sex P=0.0009 NS P=0.0098 (P=0.0698) P=0.0270 NS Diet*sex NS NS NS NS NS NS Species*diet NS P=0.0023 NS (P=0.0605) P<0.0001 (P=0.0601) Species*sex NS NS NS NS NS P=0.0285

1.3

1.2

1.1

1.0

0.9 RMA slopeRMA HL

0.8

0.7 Brisbane Coffs Harbour Sydney Population

Figure 1. RMA slope for HL in three populations of T. angusticollis. Bars represent standard errors of the mean.

1.15

1.10 Cow Bay Kuranda 1.05

1.00

0.95

0.90 RMA slope HL RMAslope

0.85

0.80

0.75 F M Sex

Figure 2. Population x sex interaction plot for T. lineolatus RMA slope for HL.

1.3

Cow Bay 1.2 Kuranda

1.1

1.0

0.9 RMA RMA slope HL

0.8

0.7 Very poor Poor Rich Diet

Figure 3. Population x diet interaction plot for T. lineolatus RMA slope for HL.

DISCUSSION

It has been suggested that the evolution of trait scaling with body size is constrained by highly conserved laws of growth (Gould 1966; Maynard Smith et al. 1985). By comparing RMA allometric slopes in five populations of neriid flies reared on three larval diets of varying quality, we aimed to test the capacity for allometric slopes to diverge among populations.

ANOVAs on RMA slopes provided evidence for divergence of allometric slope and reaction norms for these slopes. Interestingly, this evolutionary divergence was most clearly seen in a sexual trait (head length), which diverged across populations of both T. angusticollis and T. lineolatus. T. angusticollis populations were found to have diverged in allometric slopes for HL.

A population x diet interaction was present in T. lineolatus, suggesting that HL slope responded differently to diet manipulations (i.e., exhibited different reaction norms for allometric slope) in the two populations, and a population x sex interaction suggesting that sexual dimorphism for allometric slope also diverged for this trait. Evidence of within-species diversification for some additional, non-sexual traits (i.e., wing length allometry in T. angusticollis, and fore-tibia length allometry in T. lineolatus) suggests that sexual selection may act on overall body shape in these species (see Chapter 2).

The diversification of static allometry slope observed in this study is not congruent with the idea that there is a biological constraint on the ability of allometric slopes to respond to selection, although some traits, which did not diversify, could still be under constraint. Some studies imposing artificial selection on allometric traits have not elicited a change in slope, and allometric slopes across populations have seemed to remain constant (Tobler and Nijhout

2010; Egset et al. 2012). Other studies, however, have found results that do not support the constrained allometry hypothesis. Cayetano et al (2011) found that when sexual selection was removed, the allometric slope of a genitalic trait (aedeagal spines) declined significantly over

18 just 18 generations, supporting the potential for rapid evolution of allometry. This is consistent with our findings of diversification of a sexual trait among populations.

A number of studies have examined the evolvability of the relationship of trait size to body size on an arithmetic scale (Wilkinson 1993; Emlen 1996), or the ratio of trait size to body size

(Frankino et al. 2005). Despite the problems presented in comparing these studies with strict- sense static allometric analyses (based on regression of log trait size on log body size), support can be found within this literature for the absence of biological constraints on the evolution of trait scaling. Trait size to body size scaling variability has been found in the sexually-selected, laterally-elongated heads of the stalk-eyed flies (Diopsidae). The head-to-body ratio was altered over 10 generations of artificial selection, resulting in a change in relative and absolute eye-span length (Wilkinson 1993). In a study of ten populations of stalk-eyed flies possessing low genetic diversity among populations, it was suggested that these closely related populations share eyestalk allometry due to stabilising selection, not genetic constraints. This conclusion was drawn from the fact that one population had diverged significantly in eye stalk allometry, suggesting that if selection pressures change, the allometric slope can evolve rapidly

(Swallow et al. 2005). Likewise, horned beetles have been shown to have a variety of developmental mechanisms to ‘switch’ horn production on or off, or modify horn morphology, resulting in a range of relative horn sizes across species and populations (Emlen et al. 2007).

It is intriguing that, in contrast with our findings, allometry studies aimed at detecting evolution of slope have mainly found support for the constraint hypothesis. Lack of evidence for diversification of allometric slope in some studies may reflect a lack of statistical power.

Additionally, dietary manipulation and the comparison of the slopes across different diets and populations may have contributed to the significant variability of allometric slopes across populations observed in this study. This experimental approach allowed us to investigate the diversification of reaction norms for allometric slopes (i.e., effects of larval diet on slope),

19 rather than simply comparing “average” slopes among populations. As we show in a separate study (Chapter 2), we would have failed to detect diversification in some cases had we not examined the effects of diet on allometry.

The diversification of HL could be ascribed to rapid evolution of allometry as a response to differences among populations in regimes of sexual selection. Our findings of extensive diversification of allometric slope for HL are consistent with findings of rapid diversification in secondary sexual traits in general (Andersson 1994 ). However, mate guarding behaviour seen in T. angusticollis males, and male-male combat in both species, also implicates FL as a sexually selected trait. Although significant positive allometry was seen in FL in some population x diet combinations, no significant divergence in allometric slope among populations was seen in this trait, except for divergence in sexual dimorphism for allometric slope in T. lineolatus. The presence of positive allometry in FL and HL, and only in males, offers support for the hypothesis that sexual selection often leads to positive allometry. The locomotory functions affected by FL could impose stricter growth constraints and prevent the divergence of slopes for this trait. In contrast, sexual selection pressure without strong viability constraints may enable the amplified divergence seen in HL. However, our results are also consistent with the possibility that sexual selection acts similarly on FL in all populations, whereas sexual selection on HL varies among populations.

The greater plasticity of head length in response to diet seen in T. angusticollis relative to T. lineolatus (Chapter 2) did not lead to increased divergence of allometric slopes. If this had been the case, it would be expected that T. angusticollis would display more evidence of divergence of allometric slope, either in a higher number of traits or between sexes.

Interestingly, an alternate analysis of this data using an ANCOVA (ordinary least squares- OLS) revealed additional evidence of diversification (Chapter 2). There is a great deal of debate in the allometry literature over the most appropriate regression model to use in analysis of 20 allometric slopes (Seim and Saether 1983; Smith 2009; Hansen and Bartoszek 2012). OLS is based on the assumption that the independent variable is measured without error, and will therefore underestimate the allometric slope when the error is equal on both axes, as in this data set, although the extent of bias in the slope will be relatively small when the data cluster close to the regression line. In contrast, the RMA model is based on the assumption of equal error in both dependent and independent variables, and thus provides unbiased estimates of allometric slopes for our data. Although ANCOVA is a powerful tool for analysis of treatment effects on slope, there is no RMA-based equivalent to ANCOVA (Quinn & Keough 2002). As both types of analyses offered benefits and disadvantages, both were performed. These different approaches yielded somewhat different results. In particular, evidence of diversification of HL was obtained using the RMA-based analysis reported in this chapter, but not using OLS-based ANCOVA (see Chapter 2), perhaps because analysis of this trait is especially strongly affected by the biased slope estimates from OLS regression. This suggests that both RMA- and OLS-based analyses may need to be employed conjointly to gain a comprehensive picture of allometry diversification.

Further research in this area could expand on the allometric divergence in these two species.

Five populations were examined in this study, though due to the large geographical collection range it is certain that there are many other potential study populations. Incorporating these populations into this analysis of divergence would aid in confirming the patterns identified in this study, as well as revealing patterns which were not identifiable with the limited number of populations.

The two populations in T. lineolatus showed particularly interesting patterns of divergence.

This could be enhanced and expanded by the use of additional populations for comparison.

Expanding the number of populations in the study would increase the benefit of incorporating phylogenetic information in the analysis. Genetic information from a variety of populations

21 and species would give insight into evolutionary history and divergence time. Voje and Hansen

(2012) utilised phylogenetic information to determine the time frame in which static allometry slopes diverged in various species of stalk eyed flies. A similar analysis in these populations of

Telostylinus spp. would give a more accurate understanding of how rapidly these allometric slopes have diverged. Divergence time between these and more populations would illustrate the time taken for diversification of allometric slopes between populations to occur.

This study sheds light on the evolution of allometric slopes under natural and sexual selection.

Our study suggests that sexual selection can drive the diversification of allometric slope, although only in the confines of viability constraints. Our results are not in line with an absolute genetic or developmental constraint on allometric slope, and suggest greater evolvability in this trait than generally thought.

ACKNOWLEDGEMENTS

This research was supported by the Australian Research Council though a discovery grant to

Russell Bonduriansky. Thank you to Angela Crean, Margo Adler and Michael Garratt for helpful comments, and to Marco Telford for experimental assistance.

REFERENCES

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

Bonduriansky, R. (2006). "Convergent evolution of sexual shape dimorphism in diptera."

Journal of Morphology 267(5): 602-611.

Bonduriansky, R. (2007a). "The evolution of condition-dependent sexual dimorphism."

American Naturalist 169(1): 9-19.

Bonduriansky, R. (2007b). "Sexual selection and allometry: A critical reappraisal of the

evidence and ideas." Evolution 61(4): 838-849.

Bonduriansky, R. and T. Day (2003). "The evolution of static allometry in sexually selected

traits." Evolution 57(11): 2450-2458.

Cayetano, L., et al. (2011). "Evolution of male and female genitalia following release from

sexual selection." Evolution 65(8): 2171-2183.

Egset, C. K., et al. (2011). "Geographical variation in allometry in the guppy (Poecilia

reticulata)." Journal of Evolutionary Biology 24(12): 2631-2638.

Egset, C. K., et al. (2012). "Artificial selection on allometry: change in elevation but not slope."

Journal of Evolutionary Biology 25(5): 938-948.

Emlen, D. J. (1996). "Artificial selection on horn length body size allometry in the horned beetle

Onthophagus acuminatus (Coleoptera: Scarabaeidae)." Evolution 50(3): 1219-1230.

Emlen, D. J., et al. (2007). "On the origin and evolutionary diversification of beetle horns."

Proceedings of the National Academy of Sciences of the United States of America 104:

8661-8668.

Frankino, W. A., et al. (2005). "Natural selection and developmental constraints in the

evolution of allometries." Science 307(5710): 718-720.

Gould, S. J. (1966). "Allometry and size in ontogeny and phylogeny." Biological Reviews of the

Cambridge Philosophical Society 41(4): 587-&.

Green, A. J. (1992). "Positive allometry is likely with mate choice, competitive display and other

functions." Animal Behaviour 43(1): 170-172.

Hansen, T. F. and K. Bartoszek (2012). "Interpreting the Evolutionary Regression: The Interplay

Between Observational and Biological Errors in Phylogenetic Comparative Studies."

Systematic Biology 61(3): 413-425.

Houle, D., et al. (2011). "Measurement and meaning in biology." Quarterly Review of Biology

86(1): 3-34.

Huxley (1932). Problems of relative growth, by Sir Julian S. Huxley...With 105 illustrations. New

York, L. MacVeagh, The Dial Press.

Huxley, J. S. (1924). "Constant differential growth ratios and their significance." Nature 114:

895-896.

Kodric-Brown, A., et al. (2006). "The allometry of ornaments and weapons." Proceedings of the

National Academy of Sciences of the United States of America 103(23): 8733-8738.

Maynard Smith, J., et al. (1985). "Developmental Constraints and Evolution: A Perspective from

the Mountain Lake Conference on Development and Evolution." The Quarterly Review

of Biology 60(3): 265-287.

McArdle, B. H. (1988). "The structural relationship - regression in biology." Canadian Journal of

Zoology-Revue Canadienne De Zoologie 66(11): 2329-2339.

Newell, N. D. (1949). "Phyletic size increase, an important trend illustrated by fossil

invertebrates." Evolution 3(2): 103-124.

Poulin, R. (2009). "Interspecific allometry of morphological traits among trematode parasites:

selection and constraints." Biological Journal of the Linnean Society 96(3): 533-540.

Rasband, W. S. (1997-2009). ImageJ. Bethesda, MD, US National Instutites of Health.

Seim, E. and B. E. Saether (1983). "On rethinking allometry - which regression model to use."

Journal of Theoretical Biology 104(2): 161-168.

Smith, R. J. (2009). "Use and Misuse of the Reduced Major Axis for Line-Fitting." American

Journal of Physical Anthropology 140(3): 476-486.

Swallow, J. G., et al. (2005). "Genetic divergence does not predict change in ornament

expression among populations of stalk-eyed flies." Molecular Ecology 14(12): 3787-

3800.

Tobler, A. and H. F. Nijhout (2010). "Developmental constraints on the evolution of wing-body

allometry in Manduca sexta." Evolution & Development 12(6): 592-600.

Voje, K. L. and T. F. Hansen (2012). "Evolution of static allometries: adaptive change in

allometric slopes of eye span in stalk eyed flies." Evolution: no-no.

Wilkinson, G. S. (1993). "Artificial sexual selection alters allometry in the stalk-eyed fly

Cyrtodiopsis dalmanni (Diptera, Diopsidae)." Genetical Research 62(3): 213-222.

CHAPTER TWO

Sex-specific patterns of diversification in reaction norms for

morphological traits

Elizabeth J. Cassidy1, Eleanor Bath1, Stephen F. Chenoweth2, Russell

Bonduriansky1

1Evolution & Ecology Research Centre and School of Biological, Earth and

Environmental Sciences, The University of New South Wales, Sydney, New South

Wales 2052, Australia.

2 School of Integrative Biology, University of Queensland, Brisbane, Queensland

EJC, RB and SF conceived and designed the experiment, EJC and EB carried out the experiment, EJC and

RB analysed the data, EJC wrote the manuscript, and RB provided supervision and suggestions on the

manuscript.

ABSTRACT

It is well established that selection pressures differ between males and females. In this study, we sought to investigate sex differences in patterns of morphological diversification. As rapid diversification under sexual selection is often most apparent in males, the rate and nature of diversification is likely to vary between the sexes. To investigate this, we manipulated the diet of larvae from five populations within two species of Australian neriid fly. We measured among-population divergence of reaction norms of several sexual and non-sexual body shape components in males and females. As predicted, we found that the sexes differed in patterns of diversification. Males showed diversification of allometric slope and reaction norm for allometric slope, as well as reaction norms for mean trait size, while females only displayed diversification of reaction norms for trait means. Unexpectedly, we found no evidence that sexual traits diversify more extensively than non-sexual traits. Our findings suggest that sexual selection can drive different patterns of diversification in males and females, and have complex impacts on morphological evolution.

INTRODUCTION

The diversification and evolution of secondary sexual traits has long been recognized to occur at a faster rate than that of non-sexual traits (Andersson 1994 ; Kingsolver et al. 2001). There is a wide body of empirical work which supports this conclusion (Arnqvist 1998; Kingsolver et al.

2001), as well as many contributions to the theory through verbal arguments (West-Eberhard

1983) and modelling (Lande 1981; Kirkpatrick 1982; Iwasa and Pomiankowski 1995;

Pomiankowski and Moller 1995; Gavrilets 2000). In addition to this, sexual selection frequently acts with greater strength upon males, and it is within this sex that the expression of sexual traits is usually most pronounced (West-Eberhard 1983; Arnqvist 1998). Although sexual selection is a research area that dominates the evolutionary literature, relatively little is known about how sexual selection results in sex differences in diversification. Based on the tendency for sexual selection to drive rapid diversification in males, it could be assumed that the same diversification patterns will not occur in females of the same species. For example, sexual selection is thought to drive speciation in the East African cichlids (Teleostei, Cichlidae).

Sympatric groups of this speciose family maintain reproductive isolation between brightly coloured males and cryptically coloured females of different species by mate choice alone

(Seehausen et al. 1997; Seehausen et al. 1999). This speciation is an example of how sexual selection impacts the way the sexes diversify.

Studies of diversification have almost exclusively focused on diversification of phenotypic means (Cuervo and Moller 1999; Gonzalez-Voyer and Kolm 2011). However, most traits exhibit some level of plasticity, and secondary sexual traits are known to be especially plastic (Cotton et al. 2004), so it can be more biologically meaningful to look at the reaction norms of traits. A reaction norm characterizes phenotypic responses as a function of an environmental variable, and represents the nature and degree of phenotypic plasticity (West-Eberhard 2003). The incorporation of plasticity into studies of diversification is particularly important due to the

28 increasing interest in the role of phenotypic plasticity in adaptation and evolution. Despite the long held assumption that plasticity would dull the force of natural selection and slow genetic evolution, a plastic phenotype is now thought by many researchers to play a considerable role in promoting rapid diversification, speciation, and the appearance of novel phenotypes

(Pigliucci 2001; Price et al. 2003; West-Eberhard 2003; Pfennig et al. 2010).

Despite the clear evidence of rapid diversification of sexual traits, very little is known about how the reaction norms of sexual traits diversify. Studies comparing divergence of sexual and non-sexual traits have typically compared population-specific or species-specific trait means, and have not examined differences in response to environmental variation (i.e. reaction norms) (Cuervo and Moller 1999; Gonzalez-Voyer and Kolm 2011). There has been some reference to the plasticity of the sexual phenotype in studies of genotype x environment interactions, which have identified genetic reaction norm variants for sexual traits within populations (Zhou et al. 2008). However, these studies have not compared divergence of reaction norms across populations. Thus, it remains unclear whether secondary sexual traits exhibit high rates of diversification in reaction norms.

Several studies have reported variation among populations in the reaction norms of sexual and non-sexual traits, as well as evidence of divergent reaction norms between sexes. Hoverflies

(Eristalis arbustorum) have been shown to have divergent reaction norms for ambient temperature during pupal development in colour pattern, development time, wing length and thorax length across four populations (Ottenheim et al. 1998). Guppies (Poecilia reticulata) from neighbouring populations display divergent reaction norms for temperature in life-history traits such as brood size and courtship behaviour (Rodd et al. 1997). Two populations of

Drosophila melanogaster showed variation for abdominal pigmentation and its sexual dimorphism in response to temperature (Gibert et al. 2009). The sex differences between these two populations caused the authors to speculate on the possible consequences of

29 differing selection pressures on males and females. Thus, while these papers do not specifically compare the reaction norms of sexual traits to non-sexual traits, there is some indication that different selection regimes could impact the way the sexes diversify, and lead to sex-specific variation in reaction norms across populations.

The reaction norms of sexual traits may be particularly responsive to environmental perturbation as sexual traits have been found to be highly condition dependent (Rowe and

Houle 1996; David et al. 2000; Cotton et al. 2004). Secondary sexual traits are expected to be expressed in relation to the condition of the bearer (Rowe and Houle 1996). Thus, sexual selection theory also predicts that condition dependence will coevolve with sexual dimorphism

(Bonduriansky 2007a). The rapid diversification that is known to occur under sexual selection, along with the high level of plasticity that has been found to occur in sexual traits, could be an important driver of diversification patterns between males and females and contribute to sexual dimorphism.

Variation between males and females in the degree of condition dependence and degree of investment in secondary sexual traits could also be reflected in sex differences in allometric slopes. Secondary sexual traits are often found to exhibit positive static allometry slopes, where a larger individual will possess a relatively larger trait size (Green 1992; Kodric-Brown et al. 2006). This may occur because larger males are expected to have a greater pool of resources and can therefore allocate more to fitness-enhancing traits such as sexual signals and weapons, or because larger males are better able to bear the viability costs associated with large secondary sexual traits (Rowe and Houle 1996). Larger males may also benefit more from expressing relatively larger secondary sexual traits (Bonduriansky 2007b). When males are under sexual selection and females are not, this could lead to the sexually selected trait of the male having a steeper allometric slope than the female homolog.

The condition dependence of male secondary sexual traits also has implications for diversification of such traits. Both the viability costs and sexual benefits of secondary sexual trait expression are likely to covary with body size (Bonduriansky 2007b). However, the scaling of costs and benefits with body size is likely to vary with local ambient and social conditions.

For example, populations could experience different levels of predation risk as a function of body size and secondary sexual trait size (Burk 1982). Conversely, differences among populations in operational sex ratio could result in differential benefits of investing in enlarged secondary sexual traits (Jirotkul 1999). Thus, male secondary sexual traits may be expected to diversify in their expression in relation to body size, reflected in inter-population variation in static allometry slope. Moreover, costs and benefits need not scale as a linear function of body size, and are also likely to depend on other condition-dependent traits, such as fat stores.

Thus, the static allometry slope for such traits may vary in relation to nutrient abundance

(Bonduriansky 2007a), and the effect of nutrient abundance on allometric slope may vary across populations, reflected in inter-population variation in reaction norms for static allometry slope.

In this study, we investigated differences between the sexes in the nature and degree of diversification in reaction norms for body size and body shape components in order to gauge the impact of sexual selection on reaction norm evolution. We manipulated the quality of the larval diet (nutrient concentration) of larvae from five populations of two species of Australian neriid fly, Telostylinius angusticollis and T. lineolatus. These two species are found along the east coast of Australia, spanning roughly 20 degrees of latitude. T. angusticollis are found in aggregations on the beetle-damaged bark of Acacia longifolia. Males defend territories and search for mates near damaged areas of bark, where females feed and oviposit. As in all holometabolous insects, adult body size and shape of neriid flies are determined during the larval feeding and development phase. T. angusticollis displays a high level of phenotypic

31 plasticity in response to larval nutrient intake (Bonduriansky 2007a). Males and females reared on a low-quality diet are small and virtually indistinguishable from each other, whereas individuals reared on a high-quality diet are larger in size and show sexual dimorphism. The size effect seen in high-quality individuals is enhanced in males, who also display elongated heads and antennae, secondary sexual traits used in combat with other males and possibly for courtship. These larger males also utilize their long limbs to guard ovipositing females

(Bonduriansky 2006). Pilot studies suggest that this degree of sexual dimorphism and phenotypic plasticity is not present in T. lineolatus, which is found on rotting fruit in North

Queensland. Despite the morphological differences, similar combat behaviours have been observed in both species (Bath et al. 2012).

We evaluated among-population diversification in several ways. First, we asked whether populations differ in mean body shape overall (i.e., across all larval diets). In the analysis, this is reflected in the main effect of population on trait size, with body size included in the model as a covariate, across all diets. Second, we asked whether populations differ in the response of body shape to larval diet quality (henceforth, “reaction norm for body shape”). This is reflected in a population x diet interaction for trait size, with body size included as a covariate. Third, we asked whether populations differ in the response of the static allometry slope to larval diet quality (henceforth, “reaction norm for allometric slope”). This is reflected in a population x diet x body size interaction.

In T. angusticollis, male body size and shape are under sexual selection (Bonduriansky 2006, C.

Fricke, M. Adler, R. Brooks and R. Bonduriansky, submitted), while female body size and shape do not appear to be subject to sexual selection. We therefore hypothesised that male and female reaction norms would differ in how they diversify across populations. In particular, we expected to detect greater diversification of allometric slopes in males than females. We also expected to find more evidence of diversification in secondary sexual traits (i.e., head length

32 and foretibia length) than in other traits. T. lineolatus displays similar morphology and behaviour to T. angusticollis (Bath et al. 2012), suggesting that male body size and shape are also sexually selected in this species. However, because sexual dimorphism in body size and shape are less pronounced in T. lineolatus, sex differences in patterns of selection on these traits are likely to be smaller than in T. angusticollis. We thus expected that, in T. lineolatus, sex-differences in diversification would also be less pronounced than in T. angusticollis.

METHODS

Experimental animals

The flies used in this experiment were descendants of flies collected in 2010. T. lineolatus adults were collected on rotting fruit in Kuranda and Cow Bay, Queensland. T. angusticollis individuals were collected from the bark of Acacia longifolia trees in Brisbane, Queensland, and Coffs Harbour and Sydney in New South Wales. About 10 individuals of each sex were used to found the lab colony for each population except Sydney, which was founded with about 30 individuals of each sex.

Dietary manipulation

Individuals of each of the five populations were reared on three different nutrient- concentration diets: rich (R), poor (P, containing one-third the nutrients of the rich diet) and very poor (V, one-sixth the nutrients of the rich diet). For each population, 30 males and 30 females were allowed to mate and lay eggs, from which twenty randomly-chosen eggs were transferred into each of the 150 replicate containers. The 250mL replicate containers were filled with 200mL of rich (N=10 jars per population), poor (N=10 jars per population) or very poor (N=10 jars per population) larval medium. The jars were maintained in a temperature-

33 controlled room and kept moist. Five to ten days after adult emergence flies were frozen for measurement (see Chapter 1 for further details).

Morphometric data

From each replicate jar, two adult males and two adult females (where possible) were imaged, and nine measurements were taken: thorax length (TL), head length (HL), head width at the widest point across the eyes (HW), antennae length (AL), fore tibia length (FL), mid tibia length

(ML), hind tibia length (HTL), left wing length (LW) from the r-m cross vein to the wing margin.

Measurements were made using ImageJ analysis software (Rasband 1997-2009). For additional details see Chapter 1.

Analyses

All data were log-transformed prior to analysis. AL was removed from the analysis, as we were not able to obtain a measurement for this trait in a significant number of replicates. HL and FL were classified as sexual traits based on their use in combat behaviour between males and male-female interactions (Bonduriansky 2006). The remaining traits were classified as non- sexual traits. TL was used as the index of body size (See Bonduriansky 2007a). Each sex was analysed separately. For body size (TL), we constructed a model with population and diet as fixed, categorical predictors. For each of the other traits, we first used analysis of covariance

(ANCOVA) to test for differences among groups in allometric slope. We constructed a general linear model with population and diet as fixed, categorical predictors and TL as covariate.

Inclusion of TL in the models allowed us to examine effects on relative trait sizes (i.e., shape components). Population × TL, diet × TL and population × diet × TL interactions were initially included in all models. Significant interactions with TL indicate an effect of a categorical predictor (population or diet) or interaction on allometric slope. Where the P-value for all interactions involving TL was non-significant, indicating that slopes were not significantly

34 different across categorical treatment groups, interactions with TL were removed from the model, and ANCOVA (with TL as covariate) was used to test for effects of population and diet on mean trait size. All analyses were carried out on replicate means, and a separate analysis was carried out for each species and sex. These analyses were performed using Statistica v. 7

When slopes were found to be heterogeneous, we tested for population, diet and population x diet effects on trait means using the Wilcox test, which is a modification of the Johnson-

Neyman procedure (Quinn and Keough 2002). This analysis allows for a pairwise comparison of groups that reveals covariate ranges (if any) where group means differ significantly, with P- values adjusted to account for the number of comparisons. Population and diet effects on mean trait sizes were inferred when any significant differences between groups (i.e., populations across all diets or within a particular diet) were identified. Wilcox tests were carried out using the program WILCOX.exe written by Andrew Constable (available from

[http://www.zoology.unimelb.edu.au/qkstats/software.html]).

The matrix of 24 results for population effects on static allometric slope (Population x TL or

Population x diet x TL) in males and females provided a data set to test for differences in diversification between the sexes using Fishers exact tests. The frequency of significant and near-significant (P < 0.06) results was compared with that of non- significant values. If both interactions were significant for a particular trait, only one of these was counted. One-tailed tests were used to test directional predictions. A similar analysis was used to test for sex- differences in diversification in mean trait size (Population main effect or Population x diet interaction).

RESULTS

Results are summarized in Table 1. Divergence in reaction norm for allometric slope occurred for several traits in male T. angusticollis (Population x Diet x TL interactions for FL, ML, HTL). In

T. lineolatus males, one marginally significant divergence in reaction norm of allometric slope was also found (Population x Diet x TL interactions for ML), while several other traits exhibited evidence of divergence in mean allometric slope between populations (Population x TL interaction for FL, ML, HTL). Only one female trait, HL in T. lineolatus, showed a marginally significant divergence in reaction norm of allometric slope. Allometric slopes based on ordinary least-squares regression for each trait in each population and on each of the three larval diets are listed in the appendix (Table 2).

The females of both species showed significant diversification among populations for reaction norm for body shape (Population x Diet interactions) in a range of traits (T. angusticollis - HL,

HW, FL, HTL, LW, T. lineolatus - HW, ML, LW). Males showed reduced diversification of reaction norm for body shape, as indicated by the Wilcox test results, though some diversification was present (T. angusticollis - FL, ML, HTL, T. lineolatus - ML).

Across both species, males showed significantly more diversification in reaction norm of static allometric slope (Fishers exact test, one-tailed P = 0.049), and females showed more diversification for reaction norms of trait means (Fishers exact test, one-tailed P = 0.030).

Table 1. Results of ANCOVAs and Wilcox tests for effects of population, diet and their interaction on static allometry slopes and mean trait sizes. Significant effects are highlighted in bold, and near-significant effects (P<0.06) are shown in parenthesis. Population Diet Pop. x diet Pop. x TL Diet x TL Pop. x diet x TL T. ang., male Thorax P<0.0001 P<0.0001 NS NA NA NA Head length NS P=0.0014 NS NS P=0.0012 NS Head width NS P=0.02 NS NS NS NS Foretibia NS NS Wilcox P<0.05 NS NS P<0.0001 Midtibia - - Wilcox P<0.05 NS NS P=0.0012 Hindtibia - - Wilcox P<0.05 NS P=0.0258 P=0.0114 Left wing NS NS NS NS NS NS T. ang., female Thorax P<0.0001 P<0.0001 NS NA NA NA Head length P<0.0001 NS P=0.01326 NS NS NS Head width NS P=0.0048 P=0.0444 NS NS NS Foretibia P<0.0001 (P=0.0527) P=0.03647 NS NS NS Midtibia P<0.0001 NS NS NS NS NS Hindtibia P<0.0001 P=0.0341 P=0.0059 NS NS NS Left wing NS NS P=0.0022 NS NS NS T. lin., male Thorax NS P<0.0001 NS NA NA NA Head length P=0.0018 NS NS NS NS NS Head width - - - NS P=0.008 NS Foretibia NS NS Wilcox NS (P=0.0548) NS NS Midtibia - - Wilcox P<0.05 P=0.0153 NS (P=0.0590) Hindtibia - - Wilcox NS P=0.0379 NS NS Left wing NS NS NS NS NS NS T. lin., female Thorax NS P<0.0001 NS NA NA NA Head length - - Wilcox NS NS NS P=0.04994 Head width P=0.0005 P=0.0001 P=0.0021 NS NS NS Foretibia P<0.0001 NS (P=0.0666) NS NS NS Midtibia P<0.0001 NS P=0.0180 NS NS NS Hindtibia P=0.0012 NS NS NS NS NS Left wing P=0.0009 NS P<0.0001 NS NS NS

DISCUSSION

We investigated patterns of trait diversification across five different populations of neriid flies belonging to two species. When sex-specific differences are observed between populations and species, this is likely to be a consequence of different selection pressures acting on males and females. Sexual selection, in particular, is expected to drive changes in males that increase their ability to compete for mates (Darwin 1871). In support of this, our analysis revealed considerable differences in the manner of diversification between males and females. Males and females of T. angusticollis diversified across populations in body size, but did not show divergent reaction norms for body size (Population x diet interaction). T. lineolatus did not show any diversification across populations for body size. For sizes of other traits in relation to body size (i.e., shape components), we found that females mainly diversified in reaction norms for mean trait size, while males diversified in static allometry slope and reaction norm for allometric slope as well as mean trait size. Diversification of the allometric slope in males is consistent with the expectation that male body shape traits impose viability costs and will therefore be expressed in a condition-dependent manner. Given that body size reflects condition to some extent, and that the costs and benefits of secondary sexual trait expression are likely to scale differently with body size in different ambient and social environments, the optimum static allometry slope for such traits is likely to vary among populations, as well as among nutritional environments. Our findings provide some of the first evidence supporting this prediction. However, we also predicted, but did not see, greater evidence of diversification in traits that have secondary sexual functions in males.

As predicted, populations of both species diversified in allometric slope for several body shape components in males. In T. angusticollis, this occurred as variation of the reaction norms of allometric slope, representing a change across populations in how traits scaled with body size in adults reared on different larval diets. In contrast, in T. lineolatus, the mean allometric slope

38 of the two populations diverged, whereas both populations responded similarly to larval diet manipulation.

Static allometry slope represents the rate of increase in trait size per unit increase of body size among individuals at a given developmental stage. Differing allometric responses to varying diets suggest that optimal static allometry slope, or the investment in a given trait per unit of body size, varies depending on larval environment or nutrient abundances. Our results provide evidence that the costs and benefits of trait investment under the same nutritional conditions have diverged between populations. For example, on the very poor (V) diet, the Brisbane population showed a steep increase in investment in FL length per unit of body size (OLS slope

1.28) while the Coffs Harbour population decreased investment per unit of body size in these low-nutrient conditions (OLS slope 0.56). When reared on the poor (P) diet, the Brisbane population still increased investment in FL per unit of body size, though not as greatly as on the very poor diet, (OLS slope 1.11), while Coffs dramatically increased investment in FL length per unit of body size. On the rich larval diet, both populations showed a lower investment rate as body size increased, compared to the other diets (Brisbane OLS slope = 0.92, Coffs Harbour

OLS slope = 0.69). As FL is directly involved in sexual competition, populations may vary in the degree to which they benefit from trait size increase in relation to body size, which may reflect changes in sexual selection pressures. Previous work supports the variability of sexual selection regimes across populations of the same species (Iwasa and Pomiankowski 1995; Fairbairn and

Preziosi 1996).

The inter-population diversification of reaction norms of allometric slope, which was seen in T. angusticollis but not T. lineolatus, is likely a reflection of the greater degree of plasticity and condition dependence in T. angusticollis males, as indicated by effects of diet (Table 1). These differences may stem from the variation in nutrient availability and environmental heterogeneity in the habitats of the two species. The natural larval environment of T.

39 angusticollis, rotting tree bark, is likely to vary considerably in nutrients across a small spatial gradient and over time, resulting in selection for a high degree of phenotypic plasticity in response to developmental nutrient availability. This would allow T. angusticollis to take advantage of a high level of nutrients if available, but still develop normally under low resources. This leads to the evolution of condition dependence, as condition can be defined as the amount of resources available to allocate to fitness-enhancing traits (Tomkins et al. 2004), and previous work has found condition to be linked with body size in T. angusticollis

(Bonduriansky 2007a). T. angusticollis males in high condition (with access to plentiful nutrients in development) are able to utilise the abundance of resources by enlarging sexual signals and weapons, winning dominance over smaller males in combat, and producing larger offspring via a paternal effect (Bonduriansky 2006; Bonduriansky and Head 2007). Larger males are expected to derive a greater net benefit from expressing these costly secondary sexual traits, as the viability costs of sexual trait expression are likely to diminish with increasing body size. This may be because smaller individuals are under greater biomechanical constraints in terms of wielding a large weapon in combat, or displaying an exaggerated signal trait, and are more likely to pay viability costs (Bonduriansky 2007b). These differing costs and benefits of sexual trait investment across the range of body sizes of T. angusticollis males likely drive the diversification of the reaction norms of allometric slope.

T. lineolatus develops in rotting fruit, which is likely to have a high but ephemeral nutrient concentration. This probably represents a less variable larval diet than that of T. angusticollis, and has likely selected for a low level of phenotypic plasticity, as phenotypic plasticity itself is assumed to be costly (DeWitt et al. 1998). The resulting lower level of condition dependence and reduced variability of body size likely diminishes the contrast in net benefit of trait investment. Selection for divergence of reaction norm of allometric slope is therefore weaker.

The divergence of mean allometric slope within T. lineolatus likely signifies diversification of sexual selection pressures across populations.

Contrary to predictions, we did not see greater evidence of divergence in the traits thought to be sexual traits, HL or FL, relative to the other traits. In males, diversification of allometric slope was seen in both sexual and non-sexual traits. This suggests that sexual selection is acting upon male body shape as a whole, rather than specific traits: even though some traits

(like the head and forelegs) are more directly involved in male-male and male-female sexual interactions than other structures, and may therefore be the most direct targets of sexual selection, other aspects of body shape may also experience sexual selection, or may undergo correlated or correlational responses to sexual selection on the head and forelegs as a result of genetic correlations or functional interactions between traits. Geometric morphometric analysis of stalk-eyed flies (Teleopsis dalmanni) illustrates complex patterns of shape variation with increase in size, which the authors argue is evidence for entire suites of characters evolving to maximise the whole organism performance (Worthington et al. 2012). Although this is mainly attributed to viability constraints, Worthington et al. (2012) also consider the hypothesis that shape patterns could be attributed to female preference. Nonetheless, analysis of the same traits based on RMA regression yielded evidence of diversification in male head length (Chapter 1), suggesting that we may have failed to detect diversification of this trait in ANCOVA-based analysis because of a lack of statistical power or estimation bias associated with ordinary least-squares slopes.

While females gain a viability and reproductive benefit from increased body size (Bonduriansky and Head 2007), female body shape is expected to approximate the viability optimum, and selection is likely stabilising (Darwin 1871; Andersson 1994 ). Female body shape is therefore not expected to exhibit strongly body-size-dependent expression. Consistent with this expectation, we found very little evidence of diversification of reaction norms for allometric

41 slope in females. Despite the absence of allometric slope diversification, females displayed significant diversification of reaction norms for body shape in both species. Female diversification in reaction norms was observed in female homologs of male sexual traits, as well as in non-sexual traits, suggesting that this diversification was not due to genetic correlation with male sexual traits. There was no diversification of reaction norm for body size

(TL), indicating that variation in shape was not merely a response to variation in size. The considerable level of diversification of female reaction norms could suggest that these changes are adaptive, although it is not clear why strong selection would be acting on body shape in females, or why selection on female body shape would vary among populations.

Heterogeneity of allometric slopes complicated the comparison of trait means among species for males. However, the results suggested some of the traits exhibiting divergence in reaction norms of allometric slope also showed some divergence of reaction norm for mean trait size.

This suggests that the sexual selection presumably acting upon males can lead to diversification of reaction norms for allometric slope, as well as some diversification of reaction norms for mean trait size.

The patterns of differences between populations of T. angusticollis and T. lineolatus are in support of our hypothesis that the sexes diversify in different ways. The extensive literature detailing the diversification of sexual traits has not previously incorporated the evolution of reaction norms of these traits. By examining the diversification of reaction norms of multiple sexual and non-sexual traits we have been able to identify sex-specific patterns of diversification. In these results, there is some support for the hypothesis that sexual selection facilitates rapid evolution, as sexual selection appears to drive different patterns of diversification in males and females. Although both males and females exhibited considerable diversification, compelling evidence of diversification of allometric slope was obtained only for

42 males. As allometric slope is thought to be biologically constrained, this could be a reflection of the strong selective pressure on secondary sexual selection.

We attributed sex-specific differences to the presence or absence of sexual selection, due to behaviours previously identified within T. angusticollis, such as male combat and mate guarding. However, further work would be required to estimate selection acting upon traits in both sexes and therefore confirm the inferred sex differences in selection pressures. A previous study found that sexually-dimorphic traits are not necessarily subject to strong sexual selection in males (Fairbairn and Preziosi 1996). The authors suggested reasons for this discrepancy, such as conflicting selection pressures and genetic correlation between traits.

Although our study suggests that sexual selection pressures differ among populations,

Fairbairn and Preziosi (1996) demonstrate that the ways in which patterns of trait expression and sexual dimorphism relate to contemporary selection pressures on the sexes need not be straightforward.

We employed ANCOVA based on ordinary least squares (OLS) regression. However, there is a great deal of debate over which is the most appropriate analysis to use in the study of static allometry (Seim and Saether 1983; Smith 2009; Hansen and Bartoszek 2012). OLS analysis does not incorporate error on the X-axis, and can therefore underestimate the slope. Reduced major axis (RMA) analysis incorporates error on both the X- and Y-axis, and therefore provides an unbiased slope estimate. An RMA analyses was also carried out on the same data, which revealed contrasting results (Chapter 1). We infer that the differences in results of the two analyses reflect differing goodness of fit of the regression model across different traits. The present OLS-based ANCOVA revealed diversification of allometric slopes where there was a high coefficient of determination, but otherwise did not have the power to detect diversification of slopes. As the ANOVA of RMA slopes did not incorporate the goodness of fit of the regression model, only slope values which were vastly different were identified as

43 divergent. Slopes values which were more similar and identified as significantly different in the

ANCOVA analysis due to the high coefficient of determination were overlooked in the RMA analysis. We suggest that RMA-based analysis on allometric slope estimates (Chapter 1) can provide a useful check on results of more conventional ANCOVA-based analysis. In particular, the ANCOVA-based analysis reported in this chapter suggests that head length did not diversify in T. angusticollis males, whereas RMA-based analysis (Chapter 1) provided evidence of diversification for this trait.

Further experimental work is also required to evaluate the experimental diets against the natural diet of Telostylinus spp. There would be difficulty in identifying the diet of these flies in the wild, but a comparison of the morphological characters and allometric slopes of a range of wild-caught flies with the current data could be an indication of the impacts of the laboratory environment. Further understanding of ecology and behaviour in the wild would also be required to make further inferences from these results. In particular, the diversification of female body shape calls for further investigation of selection pressures on female phenotype.

Behavioural and life-history traits could also be measured for comparison with differences in morphology. The incorporation of other populations within these species would also help to evaluate patterns identified in this study. This work could be extended with the use of a phylogeny to determine time of divergence of populations, thereby giving an estimate of the rate of diversification.

ACKNOWLEDGEMENTS

This research was supported by the Australian Research Council though a discovery grant to

Russell Bonduriansky. Thank you to Angela Crean, Margo Adler and Michael Garratt for helpful comments, and to Marco Telford for experimental assistance.

REFERENCES

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

Arnqvist, G. (1998). "Comparative evidence for the evolution of genitalia by sexual selection."

Nature 393(6687): 784-786.

Bath, E., et al. (2012). "Asymmetric reproductive isolation and interference in neriid flies: the

roles of genital morphology and behaviour." Animal Behaviour (in press).

Bonduriansky, R. (2006). "Convergent evolution of sexual shape dimorphism in diptera."

Journal of Morphology 267(5): 602-611.

Bonduriansky, R. (2007a). "The evolution of condition-dependent sexual dimorphism."

American Naturalist 169(1): 9-19.

Bonduriansky, R. (2007b). "Sexual selection and allometry: A critical reappraisal of the

evidence and ideas." Evolution 61(4): 838-849.

Bonduriansky, R. and M. Head (2007). "Maternal and paternal condition effects on offspring

phenotype in Telostylinus angusticollis (Diptera : Neriidae)." Journal of Evolutionary

Biology 20(6): 2379-2388.

Burk, T. (1982). "Evolution of significance of predation on sexually signalling males." Florida

Entomologist 65(1): 90-104.

Cotton, S., et al. (2004). "Do sexual ornaments demonstrate heightened condition-dependent

expression as predicted by the handicap hypothesis?" Proceedings of the Royal Society

B-Biological Sciences 271(1541): 771-783.

Cuervo, J. J. and A. P. Moller (1999). "Evolutionary rates of secondary sexual and non-sexual

characters among birds." Evolutionary Ecology 13(3): 283-303.

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

Albemarle Street.

David, P., et al. (2000). "Condition-dependent signalling of genetic variation in stalk-eyes flies."

Nature 406(6792): 186-188.

DeWitt, T. J., et al. (1998). "Costs and limits of phenotypic plasticity." Trends in Ecology &

Evolution 13(2): 77-81.

Fairbairn, D. J. and R. F. Preziosi (1996). "Sexual selection and the evolution of sexual size

dimorphism in the water strider, Aquarius remigis." Evolution 50(4): 1549-1559.

Gavrilets, S. (2000). "Rapid evolution of reproductive barriers driven by sexual conflict." Nature

403(6772): 886-889.

Gibert, P., et al. (2009). "Phenotypic plasticity of abdomen pigmentation in two geographic

populations of Drosophila melanogaster: male-female comparison and sexual

dimorphism." Genetica 135(3): 403-413.

Gonzalez-Voyer, A. and N. Kolm (2011). "Rates of phenotypic evolution of ecological characters

and sexual traits during the Tanganyikan cichlid adaptive radiation." Journal of

Evolutionary Biology 24(11): 2378-2388.

Green, A. J. (1992). "Positive allometry is likely with mate choice, competitive display and other

functions." Animal Behaviour 43(1): 170-172.

Hansen, T. F. and K. Bartoszek (2012). "Interpreting the Evolutionary Regression: The Interplay

Between Observational and Biological Errors in Phylogenetic Comparative Studies."

Systematic Biology 61(3): 413-425.

Iwasa, Y. and A. Pomiankowski (1995). "Continual change in mate preferences " Nature

377(6548): 420-422. 46

Jirotkul, M. (1999). "Operational sex ratio influences female preference and male-male

competition in guppies." Animal Behaviour 58: 287-294.

Kingsolver, J. G., et al. (2001). "The strength of phenotypic selection in natural populations."

American Naturalist 157(3): 245-261.

Kirkpatrick, M. (1982). "Sexual selection and the evolution of female choice." Evolution 36(1):

1-12.

Kodric-Brown, A., et al. (2006). "The allometry of ornaments and weapons." Proceedings of the

National Academy of Sciences of the United States of America 103(23): 8733-8738.

Lande, R. (1981). "Models of speciation by sexual selection on polygenic traits." Proceedings of

the National Academy of Sciences of the United States of America-Biological Sciences

78(6): 3721-3725.

Ottenheim, M. M., et al. (1998). "Geographic variation in plasticity in Eristalis arbustorum."

Biological Journal of the Linnean Society 65(2): 215-229.

Pfennig, D. W., et al. (2010). "Phenotypic plasticity's impacts on diversification and speciation."

Trends in Ecology & Evolution 25(8): 459-467.

Pigliucci, M. (2001). Phenotypic plasticity: beyond nature and nurture. . Baltimore, USA, Johns

Hopkins University Press.

Pomiankowski, A. and A. P. Moller (1995). "A resolution of the lek paradox." Proceedings of the

Royal Society of London Series B-Biological Sciences 260(1357): 21-29.

Price, T. D., et al. (2003). "The role of phenotypic plasticity in driving genetic evolution."

Proceedings of the Royal Society B: Biological Sciences 270(1523): 1433-1440.

Quinn, G. P. and M. J. Keough (2002). Experimental design and data analysis for biologists /

Gerry P. Quinn, Michael J. Keough. Cambridge, UK ; New York :, Cambridge University

Press.

Rasband, W. S. (1997-2009). ImageJ. Bethesda, MD, US National Instutites of Health.

Rodd, F. H., et al. (1997). "Phenotypic plasticity in the life history traits of guppies: Responses

to social environment." Ecology 78(2): 419-433.

Rowe, L. and D. Houle (1996). "The lek paradox and the capture of genetic variance by

condition dependent traits." Proceedings of the Royal Society of London Series B-

Biological Sciences 263(1375): 1415-1421.

Seehausen, O., et al. (1999). "Evolution of colour patterns in East African cichlid fish." Journal

of Evolutionary Biology 12(3): 514-534.

Seehausen, O., et al. (1997). "Cichlid fish diversity threatened by eutrophication that curbs

sexual selection." Science 277(5333): 1808-1811.

Seim, E. and B. E. Saether (1983). "On rethinking allometry - which regression model to use."

Journal of Theoretical Biology 104(2): 161-168.

Smith, R. J. (2009). "Use and Misuse of the Reduced Major Axis for Line-Fitting." American

Journal of Physical Anthropology 140(3): 476-486.

Tomkins, J. L., et al. (2004). "Genic capture and resolving the lek paradox." Trends in Ecology &

Evolution 19(6): 323-328.

West-Eberhard, M. (2003). Developmental Plasticity and Evolution, Oxford University Press,

USA.

West-Eberhard, M. J. (1983). "Sexual selection, social competition, and speciation." Quarterly

Review of Biology 58(2): 155-183.

Worthington, A. M., et al. (2012). "Size matters, but so does shape: quantifying complex shape

changes in a sexually selected trait in stalk-eyed flies (Diptera: Diopsidae)." Biological

Journal of the Linnean Society 106(1): 104-113.

Zhou, Y. H., et al. (2008). "Reaction norm variants for male calling song in populations of

Achroia grisella (Lepidoptera : Pyralidae): Toward a resolution of the lek paradox."

Evolution 62(6): 1317-1334.

APPENDIX

Table 2. Ordinary least squares allometric slopes for each trait Population Diet HL HW FL ML HTL LW T. angusticollis

Brisbane Very Poor 1.1224 0.7051 1.2768 1.1882 1.0977 0.8332 Coffs Harbour Very Poor 0.6073 0.3044 0.5557 0.5358 0.5665 0.7568 Sydney Very Poor 1.0145 0.6996 1.1355 1.0715 1.0339 0.812 Brisbane Poor 1.3027 0.64 1.1166 1.1166 1.1289 0.6783 Coffs Harbour Poor 1.2465 0.6442 1.2986 1.2144 1.2542 0.8345 Sydney Poor 1.374 0.6409 1.0821 1.0557 1.134 0.5277 Brisbane Rich 1.011 0.5138 0.924 0.878 0.7739 0.323 Coffs Harbour Rich 0.6778 0.7027 0.6945 0.5422 0.4432 0.9137 Sydney Rich 0.9348 0.3806 0.9983 0.9846 0.8482 0.6833 T. lineolatus

Cow bay Very Poor 0.8785 0.7709 1.0434 0.9524 0.8958 0.6544 Kuranda Very Poor 0.6713 0.6387 0.8367 0.7569 0.8126 0.5874 Cow bay Poor 0.9424 0.8469 1.0537 0.8298 0.9035 0.8462 Kuranda Poor 0.8849 1.1127 0.9023 0.8556 0.9506 0.7771 Cow bay Rich 0.1863 -0.0644 0.2415 0.3073 0.3797 0.3196 Kuranda Rich 0.6071 0.3678 0.9769 0.868 0.7182 0.4495 Female

T. angusticollis

Brisbane Very Poor 0.8233 0.6394 0.9715 0.948 0.9764 0.8292 Coffs Harbour Very Poor 0.6109 0.5815 0.8252 0.9169 0.9558 1.0622 Sydney Very Poor 0.797 0.6659 0.8334 0.879 0.8751 0.8059 Brisbane Poor 0.7173 0.9644 0.8246 1.1809 0.9722 0.935 Coffs Harbour Poor 0.7831 0.7397 0.9105 0.991 0.96014 0.7431 Sydney Poor 0.7748 0.7118 0.768 0.7179 0.7893 0.7016 Brisbane Rich 0.7162 0.5182 0.6493 0.4183 0.4528 0.1799 Coffs Harbour Rich 0.3866 0.6776 0.5059 0.5638 0.5449 0.7357 Sydney Rich 0.7946 0.8161 0.6354 1.1215 0.7233 0.9905 T. lineolatus

Cow bay Very Poor 0.6966 0.8063 0.6033 0.6871 0.6445 0.6075 Kuranda Very Poor 0.6729 0.7376 0.5261 0.6288 0.5852 0.5193 Cow bay Poor 0.6517 0.5136 0.5119 0.4389 0.057 0.1989 Kuranda Poor 0.4194 0.7155 0.6803 0.6136 0.6722 0.7071 Cow bay Rich -0.0155 -0.0841 0.355 0.4557 1.0637 0.3569 Kuranda Rich 0.8012 0.1671 0.4812 0.2534 1.0285 1.0263

CHAPTER THREE

Larvae of the neriid fly Telostylinus angusticollis benefit by interacting with kin

Elizabeth J. Cassidy, Russell Bonduriansky

Evolution & Ecology Research Centre and School of Biological, Earth and Environmental

Sciences, The University of New South Wales, Sydney, New South Wales 2052,

Australia.

EJC and RB conceived and designed the experiment, EJC carried out the experiment, EJC and RB

analysed the data, EJC wrote the manuscript, and RB provided supervision and suggestions on the

manuscript.

ABSTRACT

A range of studies have evaluated the costs and benefits of high relatedness of competing individuals during development, but results are mixed. Two contrasting hypotheses have been proposed: the kin cooperation hypothesis predicts that individuals will benefit by interacting with kin, whereas the genetic diversity hypothesis predicts that dissimilar/unrelated competitors will perform better on average. In this study we examined the performance of larvae of the Australian neriid fly, Telostylinus angusticollis, reared either in sib or non-sib groups. We simultaneously manipulated nutrient abundance in a full-factorial design, and measured egg to adult viability, egg to adult development time and adult body size.

We found that when reared on restricted nutrients, members of sib-groups were significantly larger than members of non-sib groups, but we detected no effects of relatedness on egg-to-adult viability or development time. Larvae developing with their sibs therefore appear to benefit from cooperation or from some ecological benefit of genetic similarity. Our results, with past evidence, suggest that complex interactions of prevailing ecological conditions determine whether individuals will achieve higher fitness when competing with related or unrelated individuals.

INTRODUCTION

Kin selection theory predicts that related individuals will act more cooperatively and less competitively towards each other due to a high proportion of shared genes (Hamilton 1964).

This theory (henceforth, ‘kin cooperation hypothesis’) seemed to provide an explanation for some instances of altruism seen in nature. The most well known examples appear in the eusocial systems of bees and ants, but kin cooperation has also been identified in the study of siblings and their interactions in other taxa. In passerine birds, begging intensity has been found to be reduced in full-sib versus half-sib broods (Briskie et al. 1994). Interestingly, a variety of plants modify growth strategy depending on the relatedness of nearby individuals

(Dudley and File 2007; File et al. 2012). Full-sib offspring of live bearing fish (Heterandria formosa) have been found to be larger at birth than half-sib offspring, suggesting larval fish cooperate to extract resources from the mother (Ala-Honkola et al. 2011). Flour beetles

(Tribolium castaneum and T. confusum) display a shorter larval development time when reared with sibs than with unrelated individuals (Jasienski et al. 1988).

Conversely, it is also hypothesised that individuals will benefit from being reared in genetically heterogeneous groups (henceforth, ‘genetic diversity hypothesis’). The most frequently cited reason for this is the niche partitioning hypothesis, in which groups comprised of genetically varied individuals will have greater variation in resource use and therefore exploit a greater range of resources, resulting in higher mean fitness of individuals within such groups (Maynard

Smith 1978). Other explanations have included greater susceptibility to parasites of genetically homogeneous groups (diversity/disease hypothesis) or increased interference mechanisms such as conflict, resulting from greater competition over resources (Lopez-Suarez et al. 1993).

In this way an unrelated brood would be less ecologically competitive and more successful than a related brood (Maynard Smith 1978). Among the empirical support for this hypothesis is work carried out on Drosophila hydei. Non-sib groups of Drosophila larvae were found to

53 have a higher viability and shorter mean development time than groups comprised of full sibs

(Martin et al. 1988; Lopez-Suarez et al. 1993). Likewise, as relatedness of a population of solitary ascidians (Ciona intestinalis) decreased, metamorphic success, post-metamorphic survival and post-metamorphic size all increased (Aguirre and Marshall 2012).

Some studies have attempted to incorporate both of these hypotheses, and found evidence that both mechanisms could operate simultaneously. In the pygmy grasshopper (Tetrix subulata) survival rate was measured across groups of differing relatedness as well as groups of increasing diversity of colour morph. Colour morph is genetically determined and represents different eco-morphs which differ in a range of life-history and behavioural features.

Interestingly, an increase in both factors saw an increase in survival rate, suggesting conflicting selection pressures, as high relatedness and high diversity were beneficial (Caesar et al. 2010).

Ala-Honkola et al (2011) also attempted to find costs and benefits for homogeneity in broods of poeciliid fish, but obtained mixed results: probability of pregnancy was greater when females were mated to multiple males, but maturation time was shorter and offspring size was greater when females were mated to only one male.

In this study we sought to determine the impact on larval development and performance of competing with related versus unrelated individuals in neriid flies (Telostylinus angusticollis).

We quantified size at adult emergence, egg-to-adult development time and egg-to-adult viability. There have been very few studies which investigate kinship interactions in insect larvae. Previous studies on addressing this subject in dipterans have examined life-history traits (viability, development time) without incorporating impacts on adult body size, and there have been few studies on holometabolous insects in general.

The Australian neriid fly, T. angusticollis, is found in aggregations on beetle-damaged bark of

Acacia longifolia. Females feed and oviposit on nutrient rich areas of bark, while males compete amongst each other for access to these territories and females (Bonduriansky 2006; 54

Bonduriansky 2007a). As with many holometabolous insects, this species exhibits a high level of phenotypic plasticity in response to variation in nutrient abundance in the larval diet

(Bonduriansky 2007a). Flies reared on a nutrient-rich diet are large and display sexual dimorphism for size and shape: males are larger than females, and exhibit an elongated head and antennae. These secondary sexual traits are utilised for combat with other males, and elongated limbs are used to guard mates. This sexual dimorphism is not present when flies are reared on poor-quality larval diet, which produces much smaller flies (Bonduriansky 2007a).

Although a low-nutrient diet produces a smaller average adult body size, flies reared on low nutrients also display a greater range of body sizes than those reared on a rich diet

(Bonduriansky 2007a). The size range exhibited by genetically similar individuals on a low nutrient diet suggests size or shape plasticity may also occur in response to other environmental factors, which could include relatedness with competitors. A broad range of size phenotypes are also found in the wild (Bonduriansky 2006).

We manipulated the larval kinship environment by placing larvae in sib or non-sib groups.

Because effects on larval performance may only be detected under conditions of nutrient restriction, we also manipulated nutrient abundance in the larval medium in a full-factorial design. Nutrients were restricted to stimulate competition and enhance the effects of costs or benefits of the sibling manipulation. Increased density has been found to expose the effects of sib groups when effects are not apparent in low density (Caesar et al. 2010).

We expected to see differences in development time, viability and size due to the larval environment. As a high degree of size and shape variability has previously been identified in T. angusticollis, both in natural populations (Bonduriansky 2006) and in the lab (Bonduriansky

2007a), we anticipated that this variability could be responsive to the advantages or disadvantages of relatedness. If larvae benefit from interacting with kin, we expected to see higher viability, shorter development time and increased adult body size in flies reared in full-

55 sib groups. Fly larvae could benefit from relatedness by cooperation in breaking down the larval medium or tunnelling, or reduced stress from less conflict or competition. Alternatively, if there was a greater benefit to genetic heterogeneity, we expected to see higher performance in the unrelated groups. Heterogeneous groups could benefit by exploiting a wider range of available nutrients, microhabitats or foraging strategies and thus experience reduced competition for resources in comparison with sib-groups. To our knowledge, this is the first study to investigate the impact of relatedness on adult body size, as well as development time and viability, in holometabolous insects.

METHODS

Experimental animals

The flies used in this experiment were descendants of approximately 10 flies of each sex collected in 2010 in Brisbane, Queensland. This population was transported to the laboratory and kept in cages with moistened cocopeat (Galuku, Pty, Sydney) and sugar. Approximately 3 generations of this population were reared in laboratory prior to this experiment.

The diet of the parental generation of the experimental flies was standardized by rearing two hundred eggs on two litres of rich larval medium (see below). Once these flies had emerged, they were separated into same-sex groups for two weeks. After this time, males and females were paired in 75mL jars containing 25mL of oviposition medium and allowed to mate and lay eggs for approx. 48 hours. The eggs from these individuals were used in the experiment described below.

Dietary manipulation

The larval medium was comprised of (per litre of dry cocopeat) 30mL organic liquid barley malt

(Spiral Foods, Leichhardt Australia), 30mL of blackstrap sugarcane molasses (Conga Foods,

Preston, Australia), 32g of soy protein powder (Nature’s way, Pharm-a-Care, Warriewood,

Australia) and 800mL of water.

The experiment was performed in two blocks. In Block 1, parental males and females were paired in 75mL jars with 25mL of oviposition medium. After 24 h, eggs were collected from twenty mating pairs and transferred into the following treatments: 150mL larval medium (high food treatment) and 75mL larval medium (low food treatment). From each of five mating pairs, 20 eggs were transferred to a full-sib high food treatment replicate and another 20 eggs were transferred to a full-sib low food treatment replicate. From these ten pairs, plus an additional ten pairs, eggs were also transferred to make up the non-sib high food treatment replicates (N = 5 replicates) and non-sib low food treatment replicates (N = 5 replicates). One egg from each family was placed in each replicate container.

In Block 2, eggs were collected from 10 mating pairs only. From for each of the 10 pairs, 10 eggs were transferred into a full-sib high food treatment replicate and a full-sib low food treatment replicate. Two additional eggs were then transferred from each pair to make up groups of 10 unrelated individuals in the high and low diet treatments. The amount of food per larva was reduced relative to Block 1: the high-food treatment consisted of 60mL larval medium per replicate container and the low-food treatment consisted of 20mL of larval medium per replicate container. Eggs were transferred in alternating order to treatments from the oviposition medium to avoid any bias associated with oviposition order. A graphical representation of Blocks 1 and 2 is provided in figure 1.

Morphometric data

Five to ten days after adult emergence, flies were frozen at -20°C for later quantification of body size. All emerged flies were imaged using a Leica DFC420 camera mounted on a Leica

MS5 stereomicroscope. From the images, thorax length (an index of total body size) was measured using imageJ image analysis software (Rasband 1997-2009).

Analysis

All analyses were carried out on replicate (container) means as observational units. Body size

(thorax length) data were standardized within block so as to preserve differences between sexes but eliminate inter-block differences in variance. Because standardization resulted in identical means for the two blocks, the main effect of block was removed from all models, but interactions of other factors with block were retained. A separate analysis of variance was carried out for each response variable (egg-to-adult viability, development time, adult body size). For body size, relatedness (sib/non-sib), food (high/low) and block were fitted as fixed among-subjects factors, and sex as a within-subjects factor. Block was fitted as a fixed factor because the number of eggs per replicate and the amount of food in the high and low food treatments were adjusted between blocks. Sex was fitted as a within-subjects factor because both sexes were obtained from each replicate jar. All data conformed reasonably well to the assumptions of parametric testing.

M/F pair x5 M/F pair x5 M/F pair M/F pair ... (x20) (including 10 pairs from sib treatments)

20 20 1 1 1 1

Sib high Sib low Non-sib Non-sib 150mL 75mL high low 150mL 75mL

M/F pair M/F pair ... (x10)

10 10 1 1 1 1

Sib high Sib low Non-sib Non-sib 60mL 20mL high low 60mL 20mL

Figure 1. Experimental design, block one (top) and two (bottom). Arrows and boxed numbers refer to eggs transferred from each pair to a replicate.

RESULTS

Body size

Results from ANOVA for body size are summarized in Table 1. Predictably, effects of diet, sex and sex x diet were present: as in previous studies (Bonduriansky 2007a), body size was smaller on low food, and especially so in males. There was an overall effect of relatedness, and a relatedness x diet interaction (Fig. 2). On the high-food treatment there was no difference between sib and non-sib groups. On low food, sibs were significantly larger than non-sibs.

There was also a significant relatedness x diet x block interaction, reflecting a stronger effect of relatedness and stronger relatedness x diet interaction in block one. Although larvae

59 developing with sibs were larger than larvae developing with non-sibs in both blocks, this 3- way interaction suggests that the interaction of relatedness and food abundance is non-linear, such that the substantially smaller amounts of food provided in block two weakened the effect of relatedness on larval interactions.

Egg-to-adult viability

Results from ANOVA for egg-to-adult viability are summarized in Table 2. There was an overall effect of diet on viability (Fig. 4). Viability was higher on low food, and this was consistent across related and unrelated individuals. There was no effect of block or relatedness, or interactions between block, relatedness and diet on viability.

Development time

Results from ANOVA for development time are summarized in Table 3. There was an effect of block on development time. Block one had a much shorter number of days to first emergence than block two. There was no effect of relatedness or diet on development time, and no interactions between block, diet or relatedness.

Table 1: Results from ANOVA for adult body size Effect* DF F P Relatedness 1 4.91 0.0313 Diet 1 33.45 <0.0001 Relatedness x diet 1 5.73 0.0204 Relatedness x block 1 1.53 0.2214 Diet x block 1 0.30 0.5895 Relatedness x diet x block 1 7.13 0.0101 Sex 1 322.30 <0.0001 Sex x relatedness 1 1.63 0.2078 Sex x food 1 9.40 0.0035 Sex x relatedness x food 1 0.44 0.5110 Sex x relatedness x block 1 1.26 0.2662 Sex x diet x block 1 0.11 0.7397 Sex x relatedness x diet x block 1 1.82 0.1839 * Error df=51

1.2

1.0

0.8

0.6

0.4

0.2

0.0

-0.2 Thorax length

-0.4

-0.6

-0.8

-1.0 Sib Non sib

Figure 2: Relatedness x diet interaction for body size (thorax length). Solid lines indicate high food and dashed lines indicate low food.

1.5

1.0

0.5

0.0

length -0.5

Thorax -1.0

-1.5

-2.0 Sib Non sib Sib Non sib Block: 1 Block: 2 Figure 3: Relatedness x diet x block interaction for body size. Solid lines indicate high food and dashed lines indicates low food

Table 2: Results from ANOVA for egg-to-adult viability (total proportion emerged) Effect* DF F P Block 1 0.004 0.9496 Relatedness 1 0.20 0.6584 Diet 1 10.83 0.0018 Relatedness x diet 1 0.42 0.5199 Relatedness x block 1 1.17 0.2853 Diet x block 1 0.02 0.8894 Relatedness x diet x block 1 0.02 0.8894 * Error df=51

0.85 0.80 0.75 0.70 0.65 0.60 Viability 0.55 0.50 0.45 High Low Diet Figure 4: Effect of diet on egg-to-adult viability.

Table 3: Results from ANOVA for development time (days to first emergence). Effect* DF F P Block 1 137.25 <0.0001 Relatedness 1 0.02 0.8934 Diet 1 0.45 0.5038 Relatedness x diet 1 0.89 0.3503 Relatedness x block 1 0.45 0.5038 Diet x block 1 0.02 0.8934 Relatedness x diet x block 1 0.45 0.5038 * Error df=51

DISCUSSION

Previous studies have presented a range of results on the impacts of relatedness among competing individuals during development. Across these results, support has been garnered for both the genetic diversity hypothesis and the kin cooperation hypothesis. In this study, we

62 reared flies in sib and non-sib groups and measured viability, development time and adult body size to examine the costs or benefits of being housed with related individuals.

There were no significant differences between sibs and non-sibs in viability and development time on either food treatment. There was an overall effect of diet on viability, whereby larvae reared on low food achieved higher survival rate than larvae reared on high food. Recent findings from our lab shed light on this result: egg-to-adult viability declines strongly with increasing protein concentration in the larval diet, while body size increases with protein content (A. Sentinella, A.J. Crean and R. Bonduriansky, manuscript). Our high food treatment provided more protein for developing larvae, and resulted in reduced viability but increased adult body size. Studies using Drosophila species have found higher viability with increased larval density, as the ‘churning up’ of the larval medium modifies its consistency to an optimal level (Sang 1949; Lewontin 1955). As the Telostylinius species were cultured on a medium using coco-peat as a base, rather than agar as used with Drosophila species, we feel this is less likely to be a factor impacting viability in this experiment, due to the soil-like texture of cocopeat likely increasing ease of tunnelling. Between the two blocks, development time was more rapid in the first block. This may be a result of other environmental factors such as temperature or humidity, or the impacts of being housed in large verses small groups

(discussed below).

Despite relatedness not impacting viability or development time, there was an interaction effect of relatedness x diet on adult body size (Fig. 2). On the low-food treatment, flies reared with sibs had a larger adult body size than flies reared with non-sibs, while on the high-food treatment there was no difference between sibs and non-sibs in adult body size. The body size difference between sibs and non-sibs on low food was much more pronounced in the first block (Fig. 3). This result is congruent with the kin cooperation hypothesis, as being large can confer a number of advantages in this species. Sexual selection favours large males, and both

63 large males and females have been found to pass on favourable characteristics to offspring

(Bonduriansky 2006; Bonduriansky and Head 2007). Although very little support for the kin cooperation hypothesis has been found in holometabolous insects, studies in other taxa have suggested benefits of developing in a group of related individuals. Full sib guppies (Poecilia reticulata) have been found to spend more time shoaling together than with half sibs, likely contributing to group predator avoidance (Evans and Kelley 2008). Evidence of brood cooperation during development has been found in the least killifish (Heterandria formosa).

Offspring from a full sib brood have a shorter maturation time and larger size at birth than those from a half sib brood, suggesting cooperation to extract resources from the mother (Ala-

Honkola et al. 2011). A similar tendency has been predicted to occur in placental species where offspring gain fitness from being in a full-sib litter as they are able to cooperate in the production of placental hormones to extract maternal resources (Haig 1996). Among the arthropods, evidence of cooperation has been identified in the sub-social spider Stegodyphus tentoriicola, where feeding efficiency is shown to be greater in sib groups than non-sib groups

(Ruch et al. 2009).

In contrast to our results and previous studies suggesting an advantage of high relatedness in broods or groups, past work in flies was more consistent with the genetic diversity hypothesis.

D. hydei were found to have a greater viability when housed with non-sibs than with sibs, for which a range of potential mechanisms were suggested (Lopez-Suarez et al. 1993). Genetic determination and variability in the larvae of D. melanogaster has been identified in a range of behavioural characters in relation to larval resource use, including foraging strategy (de Belle and Sokolowski 1987), digging activity (Godoy-Herrera 1986) and tolerance to a range of environmental factors (Hoffmann and Parsons 1989). This research offers support for the importance of the resource partitioning hypothesis in the sister species D. hydei as genetic variability is likely to increase the range of larval foraging strategies (Lopez-Suarez et al. 1993).

Other potential mechanisms for the success of genetically heterogeneous groups in D. hydei relate to the possible absence of interference mechanisms (such as the secretion of metabolites into the larval medium, or conflict among larvae), a reduction in vulnerability to pathogens due to genetic diversity, or a benefit from a larger range of pupation sizes and times

(Lopez-Suarez et al. 1993).

It is not clear why our results differed from the abovementioned study, although it is also worth noting that the D. hydei experiment did not measure adult size (Lopez-Suarez et al.

1993). We did not examine the mechanism which allows T. angusticollis to recognise kin or related larvae to grow larger than unrelated groups, which would presumably give more insight into the differing results. Social insects detect kin and nest mates from chemical cues

(Gadagkar 1985; Holmes 2004; Lize et al. 2006). A range of mechanisms for the origin of this ability has been suggested, such as learning cues from the environment, ‘phenotype matching’, where an individual compares its own phenotype to another to determine kin, or recognition alleles (the ‘green beard’ effect) (Gadagkar 1985; Holmes 2004). Although it has been found that social insects learn chemical cues from the environment (Gadagkar 1985), solitary insects have also displayed kin recognition (Lize et al. 2006). The larvae of the parasitoid Aleochara bilineata will burrow into a fly puparium host, and will then ‘plug’ the hole with a viscous secretion. When offered an already parasitized puparium, A. bilineata larvae were found to detect relatedness of the individual by the pupal ‘plug’, and were less likely to parasitize pupae which contained a related individual. As the larvae interacting in these experiments were reared in isolation, the authors concluded that kin recognition could only occur by self referent phenotype matching or recognition alleles (Lize et al. 2006). This study supports the idea that insect larvae are capable of kin recognition without prior interaction (Lize et al. 2006).

The larval medium used in the rearing of T. angusticollis uses cocopeat as a base, containing a mixture of fine particles and fibrous coconut husks. For this reason, the nutrient content and availability in the larval medium could be more spatially heterogeneous than that of the more homogeneous medium which Drosophila spp. are usually cultured on. Due to the nature of this medium it is conceivable that the sib group could cooperate to extract nutrients, either by the breakdown of the medium, or cooperative tunnel making. Alternatives for the success of the kin group could be the absence of interference mechanisms, such as conflict or aggression, or the absence of competitive interactions in related individuals. However, this hypothesis would not offer any insight into the differing results between T. angusticollis and D. hydei.

Although there is often emphasis on cooperation and altruism in the study of kin interactions, it is also possible that beneficial impacts are a result of the ecological advantages of similar strategies, resulting from genetic homogeneity, without the presence of kin recognition mechanisms or a modification of behaviour (Jasienski et al. 1988). In T. angusticollis, this could present as similar foraging strategies among kin, especially if differing strategies resulted in intragroup interference.

Care must be taken in reviewing evidence for the genetic diversity hypothesis as some studies have been confounded by effects of pre- or post-copulatory sexual selection or maternal provisioning. This can occur when females are permitted to mate freely, to either a group of males or a single male. Results can be confounded as the low relatedness group (females allowed to mate to more than one male) have the benefit of female pre-mating preference, cryptic female choice (Evans and Magurran 2000; Evans and Kelley 2008), and the incidence of genetic incompatibility is reduced (Jennions and Petrie 2000). Additionally, multiple matings allow sperm competition to take place, potentially enhancing the success of the mixed-brood offspring (Jennions and Petrie 2000). These problems can be overcome with the use of artificial

66 insemination in viviparous species (Evans and Kelley 2008), or direct egg manipulation in the case of oviparity, as used in the current study.

Despite this, some studies show clear support for the genetic diversity hypothesis. In a study of solitary ascidians (Aguirre and Marshall 2012), the greater metamorphic success, post metamorphic survival and post metamorphic size found in genetically heterogeneous groups was attributed to genetically diverse organisms being able to fill a wider range of ecological niches, although other mechanisms were not ruled out, such as diversity-disease effects or interference mechanisms. The niche partitioning hypothesis was also suggested to explain the greater population biomass and individual fish condition in unrelated groups of juvenile salmon (Salmo salar) compared to kin groups (Griffiths and Armstrong 2001). It was suggested that salmonid families specialise in different microhabitats, resulting in greater success for genetically heterogeneous groups (Griffiths and Armstrong 2001).

In the light of these studies it can seem difficult to reconcile the kin cooperation and genetic diversity hypotheses. However it is possible that both selective pressures could be occurring simultaneously. A study in pygmy grasshoppers (Tetrix subulata) (Caesar et al. 2010) identified conflicting selection pressures by showing two peaks in grasshopper survival, when housed with kin, as well as when being housed with a diversity of colour morphs (an indicator of genetic diversity). This illustrates the importance of context in determining where the emphasis lies.

A study in tadpoles (Rana cascadae) found any kin advantage was dependent on a range of interactions with ecological factors such as density of the group and access to substrate, and the authors hypothesise that kin related behaviour is determined by environmental conditions

(Hokit and Blaustein 1997). This conjecture is strengthened by studies in other taxa including salmonids (Griffiths and Armstrong 2000), and amphibians (Pakkasmaa and Aikio 2003). The benefits of sib broods in our study were only obvious on low food, which is presumably due to 67 the presence of ad libitum nutrients available for all individuals on high food, and less so on the context dependency of sib interactions. However, the difference between blocks could be indicative of importance of brood size on kin interactions. The brood size was larger in block one (N=20) than block two (N=10), suggesting that higher population numbers are required to obtain the cooperative or genetic similarity benefits in a kin group.

It is likely there is no overarching mechanism which regulates kin group interactions.

Presumably, if a benefit is to be gained from group cooperation, this cooperation would have the greatest benefits for members of the most closely related group. This would be dependent on the capacity of the offspring to cooperate to extract additional resources, and the relative costs and benefits of a homogeneous group (such as increased susceptibility to disease) compared to individuals filling different niches, for example. Similarly, if kin groups benefit from being genetically or phenotypically similar (without kin recognition), this is likely to be the case under only particular environmental conditions.

If niche partitioning within a group is more productive for the individual, presumably the capacity for kin recognition would be lost. As seen in the pygmy crickets T. subulata, genetic heterogeneity as well as relatedness can both act to increase survival (Caesar et al. 2010). This could reflect some degree of behavioural plasticity depending on the environmental conditions, and supports the idea that kin relations can be strongly affected by the ecological setting. The net benefits of relatedness vs. diversity could also influence selection on mating behaviour. In species which gain a greater benefit from genetic diversity, such as the ascidian

C. intestinalis (Aguirre and Marshall 2012), there could be increased selection on polyandry

(Griffiths and Armstrong 2001). Further research into the mechanisms which increase fitness in related or unrelated groups could enhance understanding of the effects of kin cooperation and genetic diversity.

In T. angusticollis further work looking at larval interactions could contribute to the understanding of these mechanisms. Evidence of group tunnel making or resource extraction could imply cooperation can occur in fly larvae, whereas incidence of conflict or aggression could be suggestive of interference mechanisms. The absence of any change in behaviour depending on the proportion of genes shared with surrounding individuals could be suggestive that some non-behavioural effect of genetic similarity which conveys an advantage to kin groups. This further research could help to determine if cooperative behaviours occur in flies, and when these behaviours are likely to be selected to evolve.

ACKNOWLEDGEMENTS

This research was supported by the Australian Research Council though a discovery grant to

Russell Bonduriansky. Thank you to Margo Adler, Angela Crean, Jessica Roe and Michael

Garratt for experimental assistance and helpful comments.

REFERENCES

Aguirre, J. D. and D. J. Marshall (2012). "Does genetic diversity reduce sibling competition?"

Evolution 66(1): 94-102.

Ala-Honkola, O., et al. (2011). "Costs and benefits of polyandry in a placental poeciliid fish

Heterandria formosa are in accordance with the parent-offspring conflict theory of

placentation." Journal of Evolutionary Biology 24(12): 2600-2610.

Bonduriansky, R. (2006). "Convergent evolution of sexual shape dimorphism in diptera."

Journal of Morphology 267(5): 602-611.

Bonduriansky, R. (2007a). "The evolution of condition-dependent sexual dimorphism."

American Naturalist 169(1): 9-19.

Bonduriansky, R. and M. Head (2007). "Maternal and paternal condition effects on offspring

phenotype in Telostylinus angusticollis (Diptera : Neriidae)." Journal of Evolutionary

Biology 20(6): 2379-2388.

Briskie, J. V., et al. (1994). "Begging intensity of nestling birds varies wtih sibling relatedness."

Proceedings of the Royal Society of London Series B-Biological Sciences 258(1351): 73-

78.

Caesar, S., et al. (2010). "Diversity and Relatedness Enhance Survival in Colour Polymorphic

Grasshoppers." PloS one 5(5). de Belle, S. J. and M. B. Sokolowski (1987). "Heredity of rover//sitter: Alternative foraging

strategies of Drosophila melanogaster larvae." Heredity 59(1): 73-83.

Dudley, S. A. and A. L. File (2007). "Kin recognition in an annual plant." Biology Letters 3(4):

435-438.

Evans, J. P. and J. L. Kelley (2008). "Implications of multiple mating for offspring relatedness

and shoaling behaviour in juvenile guppies." Biology Letters 4(6): 623-626.

Evans, J. P. and A. E. Magurran (2000). "Multiple benefits of multiple mating in guppies."

Proceedings of the National Academy of Sciences of the United States of America

97(18): 10074-10076.

File, A. L., et al. (2012). "Fitness consequences of plants growing with siblings: reconciling kin

selection, niche partitioning and competitive ability." Proceedings of the Royal Society

B-Biological Sciences 279(1727): 209-218.

Gadagkar, R. (1985). "Kin recognition in social insects and other animals - A review of recent

findings and a consideration of their relevance for the theory of kin selection."

Proceedings of the Indian Academy of Sciences-Animal Sciences 94(6): 587-621.

Godoy-Herrera, R. (1986). "The development and genetics of digging behaviour in Drosophila

larvae." Heredity 56: 33-41.

Griffiths, S. W. and J. D. Armstrong (2000). "Differential responses of kin and nonkin salmon to

patterns of water flow: does recirculation influence aggression?" Animal Behaviour 59:

1019-1023.

Griffiths, S. W. and J. D. Armstrong (2001). "The benefits of genetic diversity outweigh those of

kin association in a territorial animal." Proceedings of the Royal Society of London

Series B-Biological Sciences 268(1473): 1293-1296.

Haig, D. (1996). "Placental hormones, genomic imprinting, and maternal-fetal

communication." Journal of Evolutionary Biology 9(3): 357-380.

Hamilton, W. D. (1964). "Genetical evolution of social behaviour I." Journal of Theoretical

Biology 7(1): 1-&.

Hoffmann, A. A. and P. A. Parsons (1989). "Selection for increased desiccation resistance in

Drosophila melanogaster - additive genetic control and correlated responses for other

stresses." Genetics 122(4): 837-845.

Hokit, D. G. and A. R. Blaustein (1997). "Effects of kinship on interactions between tadpoles of

Rana cascadae." Ecology 78(6): 1722-1735.

Holmes, W. G. (2004). "The early history of Hamiltonian-based research on kin recognition."

Annales Zoologici Fennici 41(6): 691-711.

Jasienski, M., et al. (1988). "Ecology of kin and nonkin larval interactions in tribolium beetles."

Behavioral Ecology and Sociobiology 22(4): 277-284.

Jennions, M. D. and M. Petrie (2000). "Why do females mate multiply? A review of the genetic

benefits." Biological Reviews 75(1): 21-64.

Lewontin, R. C. (1955). "THE EFFECTS OF POPULATION DENSITY AND COMPOSITION ON

VIABILITY IN DROSOPHILA-MELANOGASTER." Evolution 9(1): 27-41.

Lize, A., et al. (2006). "Kin discrimination and altruism in the larvae of a solitary insect."

Proceedings of the Royal Society B-Biological Sciences 273(1599): 2381-2386.

Lopez-Suarez, C., et al. (1993). "Genetic heterogeneity increases viability in competing groups

of Drosophila hydei." Evolution 47(3): 977-981.

Martin, M. J., et al. (1988). "Competition and genotypic variability in Drosophila

melanogaster." Heredity 60: 119-123.

Maynard Smith, J. (1978). The evolution of sex Cambridge [Eng.] ; New York :, Cambridge

University Press.

Pakkasmaa, S. and S. Aikio (2003). "Relatedness and competitive asymmetry - the growth and

development of common frog tadpoles." Oikos 100(1): 55-64.

Rasband, W. S. (1997-2009). ImageJ. Bethesda, MD, US National Instutites of Health.

Ruch, J., et al. (2009). "Relatedness facilitates cooperation in the subsocial spider, Stegodyphus

tentoriicola." Bmc Evolutionary Biology 9.

Sang, J. H. (1949). "THE ECOLOGICAL DETERMINANTS OF POPULATION GROWTH IN A

DROSOPHILA CULTURE .3. LARVAL AND PUPAL SURVIVAL." Physiological Zoology 22(3):

183-202.

Conclusion

In this thesis I examined three different subjects, broadly related by the concept of phenotypic plasticity. Each of these chapters helps to illustrate the nature of morphological variation within and across populations, and highlights factors contributing to this variation. After initially seeking to investigate differences in phenotypic plasticity between populations, experimental work led to three more specific areas, diversification of static allometry slopes, differences in diversification between males and females, and kin cooperation amongst holometabolous insects.

Chapter One considered the diversification of allometric slopes found across populations of two Telostylinius species. Static allometry slopes have been thought to act as a constraint on morphological variation and adaptation (Gould 1966; Maynard Smith et al. 1985). In this way static allometry could be thought to constrain the level of plasticity present in and across populations. We manipulated nutrient concentrations and compared allometric slopes of a range of traits across five populations. This close examination revealed the diversification of static allometry slopes in a sexual trait, suggesting sexual selection can drive the diversification of slope within a species, contrary to previous opinion. This research contributes to the understanding of allometry, which is vital in conceptualising morphological variation, plasticity and adaptation.

As well as analysing static allometry slopes across a range of traits in the five populations, we also examined the diversification of reaction norm for mean trait size. Looking more broadly at this matrix of results, patterns clearly emerged in how the sexes diversified. Chapter Two discussed these results and how sexual selection can drive differences in diversification between the sexes. We found males display diversification of static allometry slope and reaction norm for allometric slope, while females diversify in reaction norms of mean trait size.

In addition to this, we also found that the patterns of diversification in males suggest that sexual selection is acting on body shape as a whole, rather than specific traits. Although there is a wealth of literature detailing rapid diversification under sexual selection, this is the first study, to our knowledge, to demonstrate sex-specific patterns of diversification in reaction norms. This chapter illustrates complex interacting factors, such as condition dependence, allometry and sexual selection, contribute to morphological variation, diversification and phenotypic plasticity.

Chapters One and Two also contribute to the debate surrounding allometric regression analyses. Two different analyses were used across the two chapters, with differing results. This comparison is useful to demonstrate the varying benefits of the two analyses, and suggests that it is important to consider the results of both methods of analyses.

The first two chapters of this thesis examined morphological variation in response to nutrient intake. Our results here highlighted the presence of additional factors contributing to morphological variation. This led to the experimental work discussed in Chapter Three, kin selection in a holometabolous insect. Previous work investigating kin selection amongst the

Diptera found greater support for the benefits of genetic diversity (Lopez-Suarez et al. 1993), which is also supported by studies in a range of other taxa (Griffiths and Armstrong 2001;

Aguirre and Marshall 2012). Interestingly, we found that related fly larvae emerge larger as adults, suggesting that individuals of this species benefit from kin selection rather than genetic heterogeneity. This finding is discussed in the framework of other relevant literature, and we suggest that our result is a consequence of the ecological context, which could also apply in other taxa. To our knowledge, this study is the first to show evidence of kin selection in larval interactions in the Diptera. Further research is required to establish the mechanisms whereby rearing with sibs enhances the adult body size of Telostylinus angusticollis individuals.

The findings from these three chapters illustrate the range of areas of study which are encompassed by the broad term of phenotypic plasticity. The presence of morphological variation is a result of a complex interaction of past and present selection pressures, constraints, and ecological factors. This thesis focuses, in particular, on the impacts of larval ecology, which we demonstrate to be varied and complex in its effects on adult phenotype.

REFERENCES

Aguirre, J. D. and D. J. Marshall (2012). "Does genetic diversity reduce sibling competition?"

Evolution 66(1): 94-102.

Gould, S. J. (1966). "Allometry and size in ontogeny and phylogeny." Biological Reviews of the

Cambridge Philosophical Society 41(4): 587-&.

Griffiths, S. W. and J. D. Armstrong (2001). "The benefits of genetic diversity outweigh those of

kin association in a territorial animal." Proceedings of the Royal Society of London

Series B-Biological Sciences 268(1473): 1293-1296.

Lopez-Suarez, C., et al. (1993). "Genetic heterogeneity increases viability in competing groups

of Drosophila hydei." Evolution 47(3): 977-981.

Maynard Smith, J., et al. (1985). "Developmental Constraints and Evolution: A Perspective from

the Mountain Lake Conference on Development and Evolution." The Quarterly Review

of Biology 60(3): 265-287.

Acknowledgements

Thank you to everyone who has helped me over the last year two years.

Russell, you have been such a wonderful supervisor. Thank you for welcoming me into your lab as an undergraduate, it was such a big deal for me and gave me the confidence to think I could keep studying biology. You have been a wonderfully patient and inspiring teacher and an immensely encouraging and supportive supervisor.

Margo, thank you for including me in your experiments, making me feel at home in the lab, and for encouraging me and giving me the confidence to do my masters. You have been a great friend and lab mate! Thank you to Ange for all your support and guidance, and giving me the insider tips on how to be a scientist. Luis, you are a great lab mate and friend, thanks for the encouragement and laughs. Thanks to everyone else in the lab, Lara, Ellie, Alex and Bart for making it a great place to study and work.

Thanks to the Brooks lab for all the fun times and for letting me tag along to Townsville.

Thanks to the staff in BEES, particularly Jono and Matt, for all your help and kindness over the years. Thank you to Jess Roe for being a wonderful friend and office-mate, thank you Gabby,

Eddie, Maddie, Anna, Kylie, Cam, Courtney and Jorge. Thank you Mike for all your support and encouragement.

Thank you Mum, Dad, Merry and Ben for your support, patience, encouragement and everything else over all the years.