The evolution of male dimorphism in arthropods

Bruno Alves Buzatto BSc (Hons) Biological Sciences, MSc Ecology

Centre for Evolutionary Biology School of Animal Biology The University of Western Australia

This thesis is presented for the degree of Doctor of Philosophy of The University of Western Australia

2012

! !

! ! Summary

Discrete morphological variation within conspecific males puzzled Darwin, and still fascinates evolutionary biologists today. Known as male dimorphism, this phenomenon reflects alternative mating tactics (AMTs) among males: the large male morphs typically guard females or reproductive territories and have more elaborate weaponry; the small male morphs normally sneak copulations and have reduced weaponry. Male dimorphism is particularly common among arthropods, and generally results from a conditional strategy in which the expression of distinct male morphs depends on the status of individuals. In this thesis I firstly review the occurrence of male dimorphism and AMTs in insects, and also provide an overview of how current theory explains their evolution. I then move on to investigate empirically some of the several unanswered questions about the genetic architecture and the evolution of male dimorphism. Firstly, using the bulb mite Rhizoghyphus echinopus as a model system, I show that there is genetic variation for the switchpoint that links male morph expression to the status of individuals, which is an important prediction of quantitative genetics models that explain the genetics of such dimorphisms. Secondly, I disentangle the sources of heritability of male morph in R. echinopus, showing that this trait is strongly influenced by a paternal effect that could either be linked to the Y chromosome of males or an indirect genetic effect that is environmentally transmitted. Next, I use the dung beetle Onthophagus taurus to investigate the importance of maternal effects for the ecology of male dimorphism. I demonstrate that female O. taurus perceive population density and respond by changing the phenotype of their offspring, through a new type of maternal effect that represents a transgenerational response of AMTs to demography. Finally, I investigate the coevolution of male dimorphism and sexual dimorphism in a group of arachnids called harvestmen (Opiliones). I show that the evolution of sexual dimorphism usually precedes the evolution of male dimorphism, and that the latter is much more evolutionarily labile than the former, providing some support for an hypothesis that states that male dimorphism mechanisms are evolutionarily co-opted from previously existing mechanisms of sexual dimorphism. My research enhances our comprehension of the different mechanisms affecting the expression of male dimorphism, from genetic, ecological, and evolutionary perspectives.

! i!

ii! ! Contents

Summary i Contents iii Acknowledgments v Publications arising from this thesis vii

Prologue 1

Chapter one, general introduction: Alternative tactics within mating systems 3 1.1 Introduction 5 1.2 Modelling the origin and evolutionary maintenance of AMTs 10 1.3 Ecological genetics and the genetic architecture of AMTs 15 1.4 Alternative mating tactics are widespread in insects 18 1.5 Variation in female mating systems 31 1.6 Conclusion 33

Chapter two: Genetic variation underlying the expression of a polyphenism 35 2.1 Abstract 37 2.2 Introduction 38 2.3 Methods 42 2.4 Results 45 2.5 Discussion 48

Chapter three: Paternal effects on the expression of a male polyphenism 55 3.1 Abstract 57 3.2 Introduction 58 3.3 Methods 60 3.4 Results 63 3.5 Discussion 67

! iii!

Chapter four: Maternal effects on male weaponry: female dung beetles produce major sons with longer horns when they perceive higher population density 77 4.1 Abstract 79 4.2 Introduction 80 4.3 Methods 83 4.4 Results 87 4.5 Discussion 94

Chapter five: Correlated evolution of sexual dimorphism and male dimorphism dimorphism in a lineage of Neotropical harvestmen 101 5.1 Abstract 103 5.2 Introduction 104 5.3 Methods 107 5.4 Results 124 5.5 Discussion 132

Epilogue 141

References 145

iv! ! Acknowledgements

The long story of this thesis started several years ago, and for unconditional love and support through all these years, I thank my parents. I also thank my sister, who put up with me for so long and who was an important part of my childhood and bringing up. In a long list of loved ones, I want to thank all my family (a lot of people), including grandmothers, aunties, uncles and cousins, all dearly missed in these long years away from home. Absolutely all of them refuelled my energies in my visits to Brazil, and without that I could not have finished this thesis. They consider me the crazy biologist of the family, for what I am really thankful.

I must also thank all the people that influenced my decision to become a biologist, through inspiration, example, and encouragement. My mother has always been the example of extreme dedication, and she is part of the reason I pursued the dream of being a biologist without hesitation. My aunt Lúcia was the first biologist I ever met, and in many ways helped me fall in love with biology. More specifically, I keep tracking my passion for arthropods back to the rocky shores of Itanhaém in southeast Brazil, where my father used to take me to watch small shore crabs. They were so amazing I just fell in love with their articulated legs and claws. I tried to convince my cousins to come with me watch these crabs day after day, and could not understand why they did not think that was the greatest way to spend a sunny day on the beach. I thank them for putting up with me, and I extend my thanks to all the family members who spent time with me at ‘tio Tella’s house’. Finally, a great friend from my childhood, Bart, once said he would grow up to be a biologist and live in a tent in the middle of the desert studying scorpions. Neither of us ever did exactly that, but that sentence never left my mind, and also played a role in what I am today.

After I chose to become a biologist, many obstacles in the way could only be surpassed with the help of amazing friends. Here I must specifically cite William, who is the closest I have to a brother (which makes Carol a step sister), Niltão to help me avoid so many “dias cinzas”, and a special group of people called ‘bio 01D’, so many of them so important to me. Being on the track to become a biologist and already in love with arthropods, I still needed to fall in love with science, and that only happened because of Glauco. Thanks for teaching me how to see past the green blur of the forest, Glauquito! And here I extend my thanks to all the others that were at the Museu de História Natural at Unicamp with me: Billy, Tais, Sam, Tiago, Egito, Fátima, Fran, Jean,

! v! Barrinho, Renata, Pato, and to the friends that I later spent time with at USP: Marie, Camila, Musgo, Rachel, Miúdo, and Roberto. Here I also want to thank the ‘Lunch club’ and all its great members: Adal, Glauco, Sam, Miúdo, Alê, Mário and Marco.

After I came to Perth, life had lots of new challenges, and I am so thankful to everyone that made my life easier and happier through the course of my years here. That includes Danilo & Steph, my greatest housemates ever, all the football guys (especially Loui, Lam and Classic V), Lalo & family, Bart, Leo & Andrea, Mariana, Danilo & Letícia, Heidi, Steve, Renee, Esther, Janine, Mike, Pete, Jane, Jen, Fran, Emma, and also all the many friends from the ODC, especially Jérôme, Clive, Will and Robin. Pretty much everyone at the Centre of Evolutionary Biology helped me somehow at some stage, especially Paco, John Fitzpatrick, and my supervisors Leigh and Joe. Also a special thanks to all the fellow postgrads that were so important: all the officemates in my three different offices (19 in total!), lab mates, fellow postgrad representatives, and fellow demonstrators. Many of them fall in two or more of these categories. I am also thankful for everyone that helped me in the lab by teaching me something or doing something for me when I was away or with a broken hand: Nalu, my parents (again!), Mariana, Bruno, Janine, Rachel, Guilherme, Fran, Marisa, and Talia. Moreover, I am very grateful to a number of experts for invaluable advice and comments on theory and statistics: Jacek Radwan, Wade Hazel, Curt Lively, James D. Fry, Damian Dowling, Martin Bader, Bill Eberhard, Michael Taborsky, H. Jane Brockmann, Mark A. Elgar, and everyone at the R group.

The last component of my thesis included field and museum work in Brazil, and I am grateful to Nalu, Musguinho, ‘Doug’ (Guilherme), Giupponi, Adrik, and Gustavo Miranda for vital assistance in the field trips; Cecilia de Faria for granting access to P.N. Serra dos Órgãos; Ricardinho and Adrik for granting access and assisting my visits to zoological collections; and Glauco again for help in all stages of this trip back home.

For funding, I thank the Australian Research Council, the University of Western Australia, Education Australia Limited, and the International Society for Behavioral Ecology.

Finally, but most importantly, I thank Nalu for being with me in the last many years, sometimes living together, sometimes a whole world apart, but always supporting and encouraging me.! vi! ! Publications arising from this thesis

This thesis is submitted as a series of manuscripts. The first manuscript has been submitted as a book chapter for a peer reviewed scientific book, and the remaining manuscripts have been submitted to or published in international journals.

I. Buzatto BA, Tomkins JL, Simmons LW, In press. Alternative tactics within mating systems. In: Shuker DM, Simmons LW, editors. The Evolution of Insect Mating Systems. Oxford, UK: Oxford University Press.

II. Buzatto BA, Simmons LW, Tomkins JL, 2012a. Genetic variation underlying the expression of a polyphenism. Journal of Evolutionary Biology 25: 748-758.

III. Buzatto BA, Simmons LW, Tomkins JL, 2012b. Paternal effects on the expression of a male polyphenism. Evolution 66: 3167-3178.

IV. Buzatto BA, Tomkins JL, Simmons LW, 2012c. Maternal effects on male weaponry: female dung beetles produce major sons with longer horns when they perceive higher population density. BMC Evolutionary Biology 12: 118.

V. Buzatto BA, Tomkins JL, Simmons LW, Machado G, In prep. Correlated evolution of sexual dimorphism and male dimorphism in a lineage of Neotropical harvestmen.

! vii! All of these manuscripts have my supervisors Leigh Simmons and Joseph Tomkins as co-authors. For the general introduction (Chapter 1), I was the first author, responsible for the literature review and writing. Leigh Simmons and Joseph Tomkins helped planning the manuscript and contributed with the writing. For the manuscripts describing new experimental data (Chapters 2-4), I was the lead author, being responsible for experimental design, data collection, statistical analysis and writing. Leigh Simmons and Joseph Tomkins guided the experimental designs, contributed with ideas for the analyses, and reviewed the writing of these chapters. The last chapter also has Glauco Machado as a co-author. In that chapter, I was the lead author, being responsible for data collection, statistical analysis and writing, whereas Leigh Simmons, Joseph Tomkins, and Glauco Machado guided the data collection, contributed with ideas for the analyses, and reviewed the writing. The contributions of each author to the manuscripts: 1 BAB=60%, LWS=20%, JLT=20%. 2 BAB=75%, LWS=10%, JLT=15%. 3 BAB=75%, LWS=10%, JLT=15%. 4 BAB=75%, LWS=15%, JLT=10%. 5 BAB=75%, LWS=5%, JLT=5%, GM=15%.

All authors have given permission for all manuscripts to be included in this thesis.

______Bruno A Buzatto Leigh W Simmons Joseph L Tomkins Glauco Machado

Candidate Coordinating Co-supervisor Co-author (Ch. 5) supervisor

viii!! Prologue

We tend to think of animal species as biological units, each of them with a set of morphological and behavioural traits. Moreover, in the great majority of animals, individuals belong to one of two sexes, males and females. It is natural to also think of each sex as a separate sub-unit, with its own set of morphological and behavioural traits. It follows that, as long as two individuals belong to the same species and sex, they should look alike and behave similarly, with some degree of variation around what would be the ‘default’ behaviour and morphology for males and females, the default phenotype for each sex. However, each sex in a species is not necessarily composed by a single set of traits. Instead, different collections of morphological and behavioural traits might coexist as distinct alternative phenotypes within a sex. This is especially common when it comes to mating systems, where sometimes there is not a single mating behaviour and associated morphology for each sex. My research focused on the evolution of the dichotomous means by which males obtain fertilizations, namely ‘alternative mating tactics’ (AMTs). More specifically, I chose to investigate the evolution of the morphological dimorphisms that are often connected to AMTs in arthropods, the most diverse group of animals on this planet. The aim of this thesis was to firstly review the subject of male dimorphism associated with AMTs in arthropods, and then to investigate some of the open questions in this area. The thesis is composed of five distinct chapters, and the goal of this prologue is to explain how these chapters are connected. The first chapter is a comprehensive review of AMTs and male dimorphisms in insects (from which most of the information about AMT in arthropods comes), covering the occurrence and diversity of this phenomenon in these animals, as well as providing a theoretical account of the genetic models that have been proposed to explain the evolution and maintenance of such dimorphisms. This chapter functions as an introduction of the topic, and it is therefore presented as the general introduction to the thesis. The second and third chapters deal with the genetic architecture of male dimorphisms, and use the male dimorphic bulb mite Rhizoglyphus echinopus as a model organism. As explained in Chapter 1, the male dimorphisms associated with AMTs in arthropods are usually the result of a conditional strategy played by males. This means that every male genotype in the population is capable of expressing either of the alternative phenotypes depending on environmental conditions (which can be the condition and/or status of the male himself). There is a value of the environmental cue

! 1! that acts as a switchpoint between the alternative phenotypes, and current quantitative genetic models consider that there is genetic variation for that switchpoint, in a way that each genotype has its own switchpoint. Chapter 2 sets out to test this hypothesis, and demonstrates that there is indeed additive genetic variation for the switchpoints that determine male morph expression in R. echinopus. Meanwhile, Chapter 3 shows that additive genetic variation is not the only source of heritability for male dimorphism in these mites, and demonstrates that there are strong paternal effects for this trait that could be either linked to the Y chromosome of males or an indirect genetic effect that is environmentally transmitted. Chapter 4 shifts from the genetic perspective toward an ecological one, and investigates the importance of maternal effects on the response of male dimorphism to population density. In this chapter I study the dung beetle Onthophagus taurus, a species for which it has been proposed that the relative success of each male morph depends on population density. Mothers in this species provide the offspring with all the resources they will consume before adulthood, and therefore have some control over the phenotype of their sons. In this chapter I show that female dung beetles are capable of perceiving population density and responding by changing the phenotype of their offspring accordingly. This finding represents a transgenerational response of male dimorphism to demography. Chapter 5 shifts to a macro-evolutionary perspective of male dimorphism, and reports a comparative approach tailored to investigate the coevolution of male dimorphism and sexual dimorphism in a group of arachnids called harvestmen (Opiliones). The results of this last study indicate that the evolution of sexual dimorphism typically precedes the evolution of male dimorphism. This supports the hypothesis that the agent behind the evolution of such dimorphisms, selection against secondary sexual traits, is stronger in females than in small males. Alternatively, sexual dimorphism might evolve before male dimorphism because the genetic architecture for sex specific expression is already provided by sex chromosomes, even in sexually monomorphic species. In conclusion, my research sheds light on the evolution of male dimorphism, improving our understanding of its genetic architecture, its responses to demographic effects, and its coevolution with sexual dimorphism. In the following chapters I will conduct the reader through these findings, explaining in detail how they came about, and elaborating on their implications.

2! !

CHAPTER ONE

Alternative tactics within mating systems

! 3! !

4! ! 1.1 Introduction

Insects were for a long time considered simple organisms with unvarying behavioural repertoires and routines, incapable of complicated behavioural responses to changing environments and/or social conditions. But nothing is further from the truth. Phenotypic plasticity is widespread in insect development, life history, physiology, and behaviour (Whitman and Ananthakrishnan 2009). Plastic responses to environmental and social conditions are actually central to the remarkable adaptability of insects, and have played a crucial role in their evolutionary histories (Moczek 2010; Simpson et al. 2011). Moreover, phenotypic plasticity in insects is not merely restricted to simple responses in metabolism or activity to abiotic factors such as temperature, but can be extremely elaborate, an illuminating example of which is the learning ability of honeybees (Giurfa 2007; Hammer and Menzel 1995; Menzel 1993; Menzel and Muller 1996). Insect mating systems are no exception to this pattern. This chapter explores the intraspecific variation in mating tactics among members of the same sex. More specifically, we focus on the evolution of the dichotomous means by which males obtain fertilizations, generally referred to as ‘alternative mating tactics’ (AMTs). The first studies to describe what we can interpret today as cases of AMTs in insects date back to at least the 1930s, but it was only in the 1970s that the number of studies reporting this phenomenon started to accumulate (Figure 1.1). In 1983, when the classic examples of AMTs in digger bees and scorpionflies were reviewed by Thornhill and Alcock (1983), approximately 50 cases of AMTs in insects were already known, a number that has now surpassed the 200 mark (Figure 1.1). Here, we review the theoretical and empirical advances that have been made in this area since Thornhill and Alcock’s volume. We start by describing two illustrative systems in detail, gryllid field crickets and onthophagine dung beetles. These two groups were chosen because their reproductive biology is well known, and because the contrasting degrees of behavioural plasticity and morphological specialisation between male tactics in these two groups illustrate the diversity of AMTs that has evolved in insects. We then discuss the genetic models that have been proposed to account for the evolution and maintenance of such dimorphisms, before reviewing the occurrence of AMTs in insects generally. Finally, we discuss the relatively limited evidence for AMTs in female insects, a somewhat new and very promising area for future research.

! 5! Studies dealing ● 400 ● with AMTs ● ● ● ● ●

300 ● ● ● ● ● Species known ● to present ●● ● ● ● ● ●● AMTs 200 ● ● ● ●● ●● ●● ●● ● ● ●● ● ● ● ● ●● ● ●● ● ●● Cumulative numbers Cumulative ●●● 100 ●● ●● ●● ●● ●● ● ●● ●●● ●●● ● ● ●● ● ● ● ● ● 0

1930 1940 1950 1960 1970 1980 1990 2000 2010 Published year

Figure 1.1 Cumulative number of studies dealing with AMTs in insects (open circles and dashed line), and cumulative number of insect species that are known to present AMTs (full circles and black line) from 1930 to 2011. The first cases of AMTs in insects were described in the 1930s, and after the 1970s the number of studies on this topic started to accumulate at an increasing rate, especially after the publication of “The Evolution of Insect Mating Systems” (Thornhill and Alcock 1983).

Behavioural plasticity in field crickets — calling versus searching for females Male gryllid crickets can be said to adopt an acoustic signalling strategy in which they remain stationary within a protected burrow or crevice and produce a species specific calling song to which females are attracted, often over considerable distances (Figure 1.2 A; Loher and Rence 1978; Zuk and Simmons 1997). Once contact is established, the male ceases calling and begins to produce a second acoustic signal, the courtship song, which stimulates the female to mount, upon which the male inserts a spermatophore into her genital opening. Sperm drain from the spermatophore into the female's reproductive tract over a period of 40-60 minutes, during which time the male guards the female in an attempt to stop her from leaving and/or removing the spermatophore before sperm have been transferred. During guarding the male also generates a fresh spermatophore, which he will offer to the female following a second bout of courtship song, or if the female has escaped his attentions, he will begin to call again to attract another female. While the calling tactic is common within natural populations of field crickets, it is by no means the only tactic that males adopt in their search for females. Males have also been shown to adopt a tactic of silently searching for females, and a satellite tactic whereby they remain close to a calling male and attempt to intercept females attracted

6! ! to that male. The tactic adopted by males does not appear to be fixed; any male can adopt any tactic, depending on current environmental and social conditions. For example, in low density populations most males call to attract females, and calling males achieve the majority of female encounters. However, when population density is high, males can have greater success in finding females by wandering silently through the habitat and encountering females by chance, rather than waiting for females to approach them (Hissmann 1990; Simmons 1986). Calling can attract unwanted attention. Male crickets are often subject to parasitization by acoustically orienting flies, Ormia ochracea (Cade 1984). The fly deposits larvae on the singing male, which burrow into his body cavity to feed on his internal organs. Parasitization is lethal, with the infested male dying when mature fly larvae leave their host to pupate in the soil. Thus, although calling to attract females can be subject to sexual selection via increased male mating success, parasitism can represent a significant selection pressure against calling (Zuk et al. 1998). Opposing sexual and natural selection can maintain genetic variation in the duration of nightly calling activity (Cade 1984). Finally, aggressive interactions among mate searching males are intense, and calling can attract the attention of nearby males who attempt to monopolise aggressively acoustic space for attracting females (Simmons 1988). Subordinate males who are unsuccessful in male contest competition are less likely to call and more likely to adopt the satellite and searching tactics than are dominant males (Burk 1983; Simmons 1986). Male and female crickets also secrete long-chain hydrocarbons or CHCs onto their cuticles that function in mate recognition and attractiveness (Thomas and Simmons 2009). Males that become subordinate after an aggressive interaction will up- regulate their CHC secretion in order to make themselves more attractive to females in the olfactory modality (Thomas et al. 2011; Thomas and Simmons 2009). It is clear from our example of these crickets, that male insects can adopt a variety of means by which to find a mating partner, and that the behavioural tactics an individual male adopts at any moment in time can depend on short-term changes in its environment and/or social pressures that affect his ability to find and compete for females. Behavioural plasticity can allow an individual cricket to make rapid adjustments to changing conditions. However, as we shall see below, for some species alternative tactics can be associated with morphological differences that constrain an individual to adopt one tactic for its entire lifespan.

! 7! Morphological dimorphisms in dung beetles — guarding versus sneaking Onthophagine dung beetles are attracted to fresh animal droppings which they use as a resource for feeding and breeding (Simmons and Ridsdill-Smith 2011). Female dung beetles excavate vertical tunnels into the ground beneath the dung. They drag fragments of dung from the surface and pack it into the blind ends of side tunnels, producing a mass of dung or "brood ball" into which they deposit a single egg. As with our example of field crickets, male dung beetles adopt alternative mate securing tactics but in this case they exhibit considerably less flexibility in terms of which tactic they adopt. Large males develop horns on the head and/or thorax, and compete for access to the tunnels within which females are provisioning offspring. These "major" males will mate with resident females and guard the entrance to their breeding tunnels, fighting with other horned males for the sole access to the breeding tunnel (Figure 2 B). Both body size and horn size are strong predictors of a male's competitive ability and ultimately his reproductive success (Hunt and Simmons 2001). Guarding males also assist females with brood provisioning, dragging fragments of dung down into the breeding tunnel and delivering it to the female to pack into her brood balls (Hunt and Simmons 2002a). Male parental care increases the size and/or number of offspring the pair can produce, and reduces the longevity costs of reproduction for females (Hunt, Simmons, and Kotiaho 2002). A second class of males are smaller in body size and have only rudimentary horns or no horns at all. These "minor" males dig independent tunnels beneath the dung that intercept breeding tunnels, allowing them to sneak into the females’ breeding chamber and copulate (Figure 1.2 B). If discovered, sneaks are quickly evicted by the resident guard who will engage in retaliatory copulations with his mate (Hunt and Simmons 2002b). Thus majors specialise in fighting and defending females while minors specialise in sperm competition. Accordingly, minors invest more heavily in sperm production than do majors. The alternative morphologies of these beetles are set by the amount of dung a beetle is provided with by its parents, with majors emerging from large brood balls and minors emerging from small brood balls. The behaviour of beetles also appears both fixed and dependent on morphology. Thus, minors never provide parental care. Majors do exhibit some behavioural plasticity insofar as they will adjust their parental care, reducing rates of brood provisioning and increasing time spent fighting and guarding when the frequency of sneaks in the population rises. However, they do not appear to adopt sneaking behaviour. Rather, guards will abandon the breeding tunnels earlier than sneaks and migrate to fresh droppings where there is less immediate competition for access to females (Hunt et al. 1999).

8! !

A

Figure 1.2 Examples of alternative mating tactics in insects. A) Male field crickets call to attract females, silently search for females, or act as satellites of calling males. This B male Gryllus campestris is calling at the entrance to its burrow. B) Males dung beetles of the genus Onthophagus exhibit a suite of behavioural and morphological traits that characterise alternative mating tactics. Females dig tunnels in the soil beneath fresh dung, and provision brood masses for their offspring. Horned major males guard these tunnels from take-overs by other major males, and assist females in brood provisioning. Small hornless males dig side tunnels, sneak into breeding chambers and copulate with females while guards are defending tunnels or collecting brood provisions (Illustration by Utako Kikutani, reproduced with permission from Natural History, The Magazine of the American Museum of Natural History).

The mating systems of field crickets and dung beetles show us that male mating tactics can vary both within individual males, and between males within a species. Moreover, they illustrate the range of traits that can vary; alternative tactics are not only characterised by phenotypic plasticity in behaviour, but can also involve specialisation in signalling modality, body morphology and/or reproductive physiology. In the next section we will examine the evolutionary basis to the origin and maintenance of alternative mating tactics.

! 9! 1.2 Modelling the origin and evolutionary maintenance of AMTs

The co-existence of alternative phenotypes within a population raise an intriguing evolutionary question: how can these alternatives persist through evolutionary time, without one eventually driving the other to extinction? Game theory, developed principally by Maynard Smith (1982), aimed to find solutions to this question of coexisting phenotypes. In the parlance of game theory, evolutionary stability (coexistence of alternative phenotypes) is sought from strategies that compete against one another. The evolutionarily stable strategy (ESS) is the strategy or set of strategies (an evolutionarily stable polymorphic state; Maynard Smith 1982, page 11) that is resistant to invasion by new mutant strategies. At this point, it is worth rehearsing the terminology of game theory before continuing: a ‘strategy’ or ‘strategy set’ is a ‘decision rule’, or set of rules, that has a genetic basis, while a tactic is the phenotypic expression of a strategy (Austad 1984). While ‘tactic’ is the game theoretic terminology some authors prefer to use ‘phenotype’ because it is by definition different from a genotype (Tomkins and Hazel 2007). Game theory has advanced three solutions to the problem of the evolutionary stability of alternative phenotypes within a population: alternative strategies, the mixed strategy and the conditional strategy (Gross 1996). While game theory provides the basis for understanding how alternative phenotypes might coexist (Gross 1996; Lively 1986); quantitative genetic models provide the possibility of understanding the effects of selection and genetic variation on the coexistence of alternative phenotypes, as well as the ecological factors that affect which strategy sets will evolve (W. Hazel and Smock 2000; Hazel et al. 1990; Hazel et al. 2004; Tomkins and Hazel 2007).

Alternative mating strategies Where reproductive competition for females is intense, selection might favour alternative phenotypes that circumvent competition. In such cases it is conceivable that the fitness pay-offs from alternative phenotypes might be negatively frequency dependent. For example a mutant male phenotype that is combative, might invade a population of males that avoid fights, while conversely, a mutant sneaker male phenotype might invade a population composed only of aggressive guarding males. The fact that strong negative frequency dependent selection can maintain genetic polymorphisms, leads to the game theoretic premise for the existence of an evolutionarily stable state where two (or more) alternative genetic strategies are maintained by negative frequency dependent selection (Gross 1996; Maynard Smith

10! ! 1982). Where there are two alternative strategies their fitness is often thought to be equal, however it is worth noting that the fitness of alternative strategies are only equal at the ESS frequency and where there are three strategies frequency dependent cycles of fitness can occur (Sinervo and Lively 1996). As a defining feature of alternative strategies, looking for equal fitness is more or less uninformative, particularly since demonstrating equality of fitness amounts to proving a null hypothesis. Furthermore, since neither equality of fitness nor negative frequency dependence, separate alternative strategies from mixed or conditional strategies (Hazel et al. 1990; Tomkins and Hazel 2007 2011) the research goal ought to be focussed on the genetic basis to the phenotypic variation in the population. The distinguishing feature of alternative strategies is their genetic architecture: traditionally, genetic polymorphisms held in balance by negative frequency dependent selection were thought to reflect a polymorphism at a single locus or few loci (Gross 1996; Maynard Smith 1982). However since alternative strategies usually involve suites of traits, the notion that ‘major genes’ literally at a single locus are responsible for most alternative strategies may be unrealistic. This is not a significant problem for the game theoretic modelling of alternative strategies however, since it is the pattern of inheritance that is important, rather than the number of loci involved. For example, so called ‘supergenes’ are tightly linked co-segregating (non-recombining) adaptive clusters of loci that tend to be inherited as-one, and have been documented in the colour polymorphisms of Batesian mimetic butterflies (Jones et al. 2012; Joron et al. 2011) and also with fitness related reproductive behaviour in male and female fire ants (Lawson et al. 2012). Supergenes seem to be a likely manner for many alternative strategies to be inherited. Hence, studies such as those of Tsubaki (Tsubaki 2003) on the male-colour polymorphic damselfly Mnais costalis, although consistent with a single locus, may also be a supergene, since the polymorphism extends to size, behaviour and survival under parasite stress (Tsubaki and Hooper 2004; Tsubaki, Hooper, and Siva- Jothy 1997). Another possible genetic architecture for alternative strategies can arise from a chromosomal inversion (Gilburn and Day 1994). Inversions have less recombination than other regions of the chromosomes (Kirkpatrick 2010) because recombination is rare to impossible in the heterozygote form. Furthermore inversions can have marked phenotypic effects (Tuttle 2003); these two features lend inversions to become co-opted into the evolution of alternative strategies. A large chromosomal inversion occurs in the seaweed fly Coelopa frigida, this inversion affects female mating preferences and male size and willingness to mate (Gilburn et al. 1996; Gilburn and

! 11! Day 1994; Gilburn et al. 1992 1993), and it seems to be implicated in the determination of alternative male phenotypes in C. nebularum (Dunn et al. 1999).

The mixed strategy Where strong negative frequency dependent selection occurs there is a frequency of alternative phenotypes which is evolutionarily stable, and to which the population is expected to return following perturbation. Theoretically a female could follow a mixed strategy and could partition her offspring into each phenotype at the ESS frequency (Alcock et al. 1977). Alternatively a male could enter a contest over a female and play each tactic according to the ESS frequency (Maynard Smith 1982). The former was thought to be an important mechanism, particularly in the hymenoptera where females have control over the sex and often the size of their offspring (Alcock 1995 1996; Alcock et al. 1977). However there appears to be very little evidence for mixed reproductive strategies neither generally (Gross 1996), nor in the solitary hymenoptera (Tomkins et al. 2001; Torchio and Tepedino 1980). The reason for this is likely to be that any adaptive tailoring (of the phenotypic alternative that is expressed) to an individual’s circumstances (its size, age or resource availability) will be favoured over a strict probabilistic decision rule.

The conditional strategy Phenotypic plasticity occurs where changes in traits arise in response to environmental variability, and is an almost ubiquitous occurrence in living organisms (West-Eberhard 1989, 2003). In the context of reproductive competition, conditional strategies occur where there is plasticity in response to a cue, that yields divergent phenotypes –aimed (evolutionarily speaking), at maximising the organism’s fitness given its circumstances. Exploring a conditional strategy (genetic decision rule) can make it clearer how conditional strategies operate. For example, frequently there are decision rules that involve competitive ability, which usually boil-down to size; so the strategy might be: on the condition that the opponent is larger, ‘play’(this is game theory) the sneak tactic and on the condition that the opponent is smaller, ‘play’ the guard tactic. First, the strategy is conditional not because of differences in condition – although there frequently are – but because the strategy contains a conditional clause (‘if x then do a; if y then do b’). Second, there is one strategy with two alternative tactics or phenotypes. As we see below, the conditionality in the decision rule reflects the expectation that individuals will adopt the tactic from which they derive the highest fitness return for their circumstances (in this case whether they are larger or smaller than a rival).

12! ! Frequency dependent selection is not an assumption of the conditional strategy, nevertheless it is a likely but rarely quantified (Simmons et al. 2004) feature of the conditionally expressed alternative phenotypes in many species. Following on from this, is that the average fitness of the alternative tactics need not be equal (Tomkins and Hazel 2007, 2011); something discussed further below. From this introduction to the conditional strategy, the game theoretic status dependent selection (SDS) model of Gross (1996) follows most intuitively, even though this was actually predated by the quantitative genetic model of the conditional strategy: the environmental threshold (ET) model (Hazel and Smock 1993; Hazel et al. 1990). Gross (1996) proposed that in most contexts where alternative tactics are selected for in males, male status is the factor that will determine the favoured tactic. Hence low status males tend to avoid aggressive interactions in favour of sneak matings, while high status males tend to fight for and guard females; this marries with the fact that in insects alternative tactics are frequently body size dependent (Hunt and Simmons 2001). Gross’s SDS model assumes that fitness can be related to status in each morph, and that these fitness functions differ for the alternative tactics (Figure 1.3 A). Where the fitness functions intersect is hypothesised to be the ESS switch point under the SDS model; the status at which it is adaptive for individuals to switch tactics (Figure 1.4).

A 120 Figure 1.3 A) The status dependent selection model proposed by Gross (1996) and B) the 100 β distribution of switch points hypothesised 80 under the Environmental Threshold model. A) Under the SDS a normal distribution of status 60 underlies tactic fitness. Low status individuals 40 α derive highest fitness from tactic α while high status individuals derive highest fitness from 20 S* Frequencyorfitness tactic β. Under the SDS the ESS switch point S* 0 is considered to be the point at which the 90 100 130 Status fitness functions cross, note that the ET B demonstrates that this is a over simplification. B) Under the ET model genotypes vary in their response to the environmental cue. Here the β sensitivities of 7 genotypes are modelled in

relation to variation in the environmental cue. At the extremes of cue strength all individuals adopt the same phenotypes, while over a range of cue strengths some individuals will switch

Phenotype but others not depending on cue strength. Status is equivalent to the environmental cue. α The variance in switch points is a key factor in determining the effects of selection on switch Environmental cue points.

! 13! The Environmental Threshold (ET) model (Hazel et al. 1990) provides a quantitative genetic model for the co-existence of alternative reproductive phenotypes. The genetic assumptions of this model are mathematically equivalent to the model proposed by Falconer (1965) to understand the heritability of dichotomous traits (Roff 1996). In Falconer’s model the population varies in a trait called ‘liability’; that proportion of individuals that exceed the threshold liability go on to develop the alternative ‘affected’ phenotype. The ET model uses the same principle but assumes that each individual in a population has a switch point, and that switch point variation is normally distributed (reflecting genetic and environmental variance; Figure 1.3 B). When the strength of the cue or cues (see Tomkins and Hazel 2011) exceeds the individual’s switch point, that individual produces the alternative phenotype (Hazel et al. 1990; Tomkins and Hazel 2007, 2011). The cue in the environmental threshold model is deemed to be environmental, and it may not, at first sight, seem appropriate to model status (for example body size) of an insect as an environmental cue. Nevertheless, as long as there is no genetic correlation between switch point and status; status can be modelled in the same way as an environmental cue, despite its environmental and genetic basis (Hazel et al. 1990; Tomkins and Hazel 2007, 2011). In other cases it is clear that the status is environmental, for example where behavioural tactics depend on relative size as in the rove beetle Leistotrophus versicolour (Forsyth and Alcock 1990). In the SDS model the ESS switch point is hypothesised to be at the status that corresponds to the intersection of the fitness functions (Gross 1996). The ET model does not make this assumption, but rather uses the fitness functions, the distribution of the cue (i.e. body size in many species), and the variation in the switch point distribution to estimate where the ESS switch point lies (Hazel et al. 1990; Tomkins and Hazel 2007, 2011). By quantifying these parameters researchers can use the ET to estimate how selection is acting on the switch point of a population, or whether there is a mismatch between the observed mean switch point and the ESS (Tomkins and Hazel 2011). Recently, evidence has been presented for the existence of tri-morphic males (Kelly 2008; Kelly and Adams 2010; Rowland and Emlen 2009). Under the conditional strategy individuals in such a population would have two switch points (see Tomkins and Hazel 2011). To have two switch points seems to require very narrow variance in the switch point distributions and very divergent fitness functions (Tomkins and Hazel 2011). Unfortunately, little is currently known about the genetic architecture of these species.

14! !

Figure 1.4 Status dependent 0.30 selection in the dung beetle Onthophagus taurus. Open circles 0.25 are minors, closed circles majors. 0.20 The fitness functions are for fertile focal males competing 0.15 against irradiated rival males or 0.10 irradiated focal males Paternity competing against fertile rivals 0.05 (Hunt and Simmons 2001). 0.00

5 3.5 4 4.5 5.5 6 Pronotum width (mm)

!

1.3 Ecological genetics and the genetic architecture of AMTs

When do we expect alternative strategies to evolve and when do we expect conditional strategies to evolve? This question goes beyond the architecture of the strategy itself and asks instead what ecological circumstances favour environmentally sensitive conditionality vs. environmentally insensitive alternative genetic strategies. Insight into this question has come from studies of the morphologically plastic defence responses of barnacles to predation by a sea-hare (Hazel et al. 2004; Lively 1986). Where an organism is not phenotypically plastic it can be said to be ‘canalised’ into one strategy, hence alternative strategies occur where mixtures of canalised strategies co- exist in a negative frequency dependent manner. Alongside canalised strategies there are also conditional strategies that respond to environmental cues (Hazel et al. 2004; Lively 1986). It turns out that whether a single canalised strategy, two alternative strategies or a conditional strategy (or a mixture) is the ESS depends — in addition to the fitness functions of the alternatives and the variance in the switch point distribution — on the reliability of the cue and the degree to which the individual’s tactic choice is tested (Figure 1.5; Hazel et al. 2004). So for example if cues are unreliable, conditional strategies do not evolve but canalised strategies (e.g. always fight or always sneak) do. Under these circumstances there is only a relatively small range of the parameter space where alternative strategies evolve to co-exist as a stable polymorphism (Figure 1.5; Hazel et al. 2004). Where cues are reliable and individual’s choices are tested by encounters with other males, the conditional strategy evolves (Figure 1.5; Hazel et al. 2004).

! 15!

! Figure 1.5 The ecological determinants of reproductive strategies (after Hazel et al. 2004). Hazel et al. (2004) showed that the conditional strategy evolves where cues are reliable and where testing of the phenotype choice is reasonably common. In contrast, where cues are unreliable and testing of phenotypes is also common (and particularly where frequency dependent selection is present) a stable polymorphism can evolve. Depending on the variance in the switch point distribution, populations can evolve where unconditional and conditional strategists are present. Which phenotypic tactic is represented by α and β is likely to depend very much on the biology of the species.

Here then is the explanation for why, despite the diversity of reproductive competition and behaviour in the insects, about 95% of male strategies are conditional strategies (Table 1.1): status is a reliable cue. The only species where alternative genetic strategies have been robustly documented is in the Odonates where aerial pursuits are contests of stamina that resolve according to energy reserves (Marden and Rollins 1994; Plaistow and SivaJothy 1996), rather than physical grapples and trials of size and strength. Much of the variation in success in territorial defence in damselflies comes down to age, so while alternative behavioural tactics in damselflies are common (e.g., Córdoba-Aguilar and Cordero-Rivera 2005), dimorphism sensu stricto based on status is unlikely, simply because size at the nymphal stage is unlikely to be a reliable cue of success as an adult. Behavioural tactics in males associated with territory holding or non territorial males appear therefore to have been canalised genetically into two alternative strategies in the rare case of Mnais (Tsubaki 2003). Finally, one way the change in genetic architecture from a conditional to a canalised strategy can be achieved is by mutations that change the sensitivity to the cue. Such major genes can simply shift the switch point beyond the range of cues, so that

16! ! the switch is always tripped, or never tripped (Hazel et al. 2004). Evidence for these kinds of major genes in insects have come from colour polymorphic tobacco horn worms (Suzuki and Nijhout 2008). Clearly if this happens mixtures of conditional and unconditional strategies can occur, and there is some evidence for this in male dimorphic acarid mites (Buzatto et al. 2012a; Chapter 2), but not to our knowledge in insects. Plaistow et al. (2004) developed a model in which the costs and limits of plasticity are considered as constraints on the extent to which populations of purely conditional strategists are likely to occur. Their modelling suggests that alternative genetic strategies will be more common than currently thought, or at least that purely conditional strategies should be very unusual. The data we have gathered for insects suggests that conditionality, as has been the long held view (West-Eberhard 1989, 2003) does in general predominate however.

Table 1.1 The cues that trigger the permanent or temporary adoption of conditional AMTs in insects can be extrinsic (such as weather conditions), intrinsic (such as age), or an interaction between both.

Cue Number of species Extrinsic / Intrinsic Body size or condition / diet 77 E x I Density 36 E Time of day / season 12 E Age 10 I Territory / females availability 9 E Previous success 7 E x I Food / host availability 6 E Weather 6 E Mate competition 5 E Sex ratio 3 E Body / wing colour 2 I (E x I?) Female /copulation site 2 E Oviposition site availability 2 E Tactics frequency 1 E Total 178 12 E, 4 I

! 17! 1.4 Alternative mating tactics are widespread in insects

When reviewing any aspect of insect biology, we must keep in mind that the diversity of the group has the habit of hampering generalisations. Insects can be found in nearly every habitat, occupying most ecological niches, feeding on any form of organic matter and interacting with all other organisms in the ecosystem (Grimaldi and Engel 2005; Schowalter 2006). It is unsurprising that their mating systems are equally diverse, so that summarizing or categorizing the sexual behaviours and reproductive strategies across the whole class Insecta is a daunting challenge. Currently there are 30 extant orders of insects, and although there is still considerable uncertainty regarding the phylogenetic relationships between these orders (Ishiwata et al. 2011), most of them are clearly monophyletic (Grimaldi and Engel 2005). With the exception of Blattodea (cockroaches), Psocoptera (psocids), Mecoptera (scorpionflies), and perhaps Mantophasmatodea (gladiators), the remaining orders have clear synapomorphies and relatively homogenous basic morphology, at least when compared to the morphological heterogeneity across all orders (Grimaldi and Engel 2005; Schowalter 2006). However, we cannot overlook the fact that species richness ranges from 15 described species of Mantophasmatodea (gladiators) to over 350,000 described species of Coleoptera (beetles), and the morphological and ecological diversity within each order is in part a reflection of species number. We nevertheless attempted to review the occurrence and diversity of AMTs at the level of insect orders, as a humble step towards understanding the evolution of intrasexual diversity in the mating systems of these organisms. Our review is based on a search of the literature (through Web of Science) using the key-words ‘alternative’, ‘mating’, ‘reproductive’, ‘tactic(s)’, ‘strategy(ies)’, and combinations thereof. Even though we focused here only in mating tactics, and not on other features of insect reproduction (such as oviposition strategies, for instance), we still included ‘reproductive’ in our keywords because the usage of the term ‘alternative reproductive tactics/strategies’ in the literature often encompasses AMTs. However, some studies describe the sort of intraspecific variation in mating tactics that falls under the umbrella of AMTs, without using any of the key-words mentioned above. These studies have likely been missed. Moreover, to restrict our literature search to studies on insects, we also used the keywords ‘insect(s)’ and ‘Insecta’, and we might have missed studies that only treated their focal species by a common name. We therefore anticipate that our review is somewhat incomplete, and AMTs are probably even more widespread than currently recognized. We wouldn’t be surprised if they

18! ! were the rule, rather than the exception, in insect mating systems. The results of our literature search have been lodged in Dryad (doi: to be provided), and are summarised in Tables 1.1 and 1.2. Our review covered a total of 361 references, including journal articles and chapters in peer-reviewed books, published between 1934 and 2012. AMTs have been recorded in over 200 insect species, spread across 11 orders and all four major insect clades (Table 1.2). Within the Holometabola, AMTs can be found in a few species of butterflies and moths, flies and mosquitoes, scorpionflies, almost 40 species of beetles, and over 60 species of wasps and bees. Within the Polyneoptera, AMTs have been described for crickets, bush-crickets, mole-crickets, grasshoppers, wetas, cockroaches, and earwigs. Within the Paraneoptera, AMTs are known for planthoppers, true bugs, water striders, and thrips. Finally, within the Palaeoptera, AMTs are expressed by males of several species of dragonflies and damselflies. For an entomologist’s eye, or for the eye of any insect enthusiast, it is immediately clear that there is a link between the diversity of each insect order and the incidence of AMTs within them (Table 1.2, Figure 1.6). Moreover, the orders in which AMTs have never been described are usually either the least diverse or the least studied groups of insects (Table 1.2). These patterns strongly suggest that AMTs could indeed be ubiquitous. In fact, every new insect mating system that is investigated in detail reveals some sort of intrasexual variation in reproductive behaviour, and a broad definition of AMTs would probably encompass the great majority of well studied species. We here focus on summarizing the studies in which the authors clearly described males’ reproductive behaviours as AMTs, or studies in which the description of the mating system hints at the presence of dichotomous means of obtaining fertilizations among males. In the following sections we will illustrate the diversity of AMTs in insects, and elaborate on the continuum between plastic and fixed alternatives (sometimes involving morphological dimorphism) in the mating tactics of insects. !

! 19! Table 1.2 The incidence of male AMTs, described for over 200 species spread across 11 orders and all four major insect clades.

Spp Incidence of AMTs Clade Order number1 Families Gen Spp Conditionality2 Fixed / Reversible3 Morphology4 Dimorphism5

Holometabola Coleoptera 386,500 12 27 38 28C (7?) 18F (3?), 7R 13D (1?), 12I (2?) 9N, 25Y Diptera 155,477 12 18 22 17C (1?), 2M 2F, 17R 13D (1?), 11I 20N, 2Y Hymenoptera 116,861 16 39 67 54C (19?), 1M? 34F (6?), 30R (10?) 32D (4?), 20I (1?) 32N (9?), 34Y Lepidoptera 157,338 3 13 13 11C 1F, 11R (1?) 1D, 4I 11N, 2Y Mecoptera 757 2 2 3 3C 3R 3I 3N Megaloptera 354 0 0 0 - - - - Neuroptera 5,868 0 0 0 - - - - Raphidioptera 254 0 0 0 - - - - Siphonaptera 2,075 0 0 0 - - - - Strepsiptera 609 0 0 0 - - - - Trichoptera 14,391 0 0 0 - - - - Palaeoptera Ephemeroptera 3,240 0 0 0 - - - - Odonata 5,899 5 13 22 18C (2?), 5M (1?) 4F, 18R (4?) 2D, 5I (1?) 17N (8?), Y Paraneoptera Hemiptera 103,590 4 9 12 12C 2F, 10R 2D, 8I (4?) 10N, 2Y Thysanoptera 5,864 2 3 4 3C 1F?, 3R 1D?, 3I 3N, 1Y? Phtiraptera 5,102 0 0 0 - - - - Psocoptera 5,720 0 0 0 - - - - Polyneoptera Blattodea 4,622 1 1 1 C R Irr? N

20! ! Dermaptera 1,978 1 1 1 C R? I Y Embiodea 463 0 0 0 - - - - Grylloblattodea 34 0 0 0 - - - - Isoptera 2,692 0 0 0 - - - - Mantodea 2,400 0 0 0 - - - - Mantophasmatodea 15 0 0 0 - - - - Orthoptera 23,855 6 15 24 32C (6?), 1M 3F (1?), 21R (8?) 2D (1?), 3I 20N (5?), 4Y Phasmatodea 3,014 0 0 0 - - - - Plecoptera 3,743 0 0 0 - - - - Zoraptera 37 0 0 0 - - - - Wingless Archaeognatha 513 0 0 0 - - - - insects Zygentoma 560 0 0 0 - - - - 180C (35?), 65F (11?), 66D (8?), 126N (22?), Total 64 141 207 9M (2?) 122R (23?) 70I(8?), 1Irr? 72Y (1?)

1From Zhang (2011). 2AMTs can be conditional (C) or mendelian (M, pure genetic polymorphism); 3fixed (F) or reversible (R). 4We also included whether male morphology determines (D), influences (I) or is irrelevant (Irr) for the mating tactic employed by individuals, 5and whether there is morphological dimorphism (Y for yes, N for no) among males. ? indicates the uncertainty, whereas lack of information is omitted.

! 21!

C F ResearchResearch effort effort B

Lepidoptera B’ C’ Coleoptera

A Diptera Wingless insects! Ephemeroptera Mecoptera Diptera Palaeoptera Archaeognatha

Zygentoma Siphonaptera (B) Isoptera Polyneoptera Odonata Trichoptera Blattaria Hymenoptera Paraneoptera Lepidoptera A’ Other holometabola Mantodea Strepsiptera

Zoraptera Coleoptera (C) G Occurence of AMTs (A) Dermaptera Neuroptera Occurrence of AMTs

Plecoptera Megaloptera E Thysanoptera Diptera Lepidoptera Hemiptera Hymenoptera Orthoptera Raphidioptera (E) Phthiraptera Coleoptera

(D) Grylloblattodea Hymenoptera Wingless insects Embioptera Psocoptera D’ Palaeoptera

Mantophasmatodea Phasmatodea Untitled Tree

E’ Polyneoptera D Other holometabola Paraneoptera!

Figure 1.6 A-E) A recent phylogeny of Insecta (according to Ishiwata et al. 2011) showing the major insect clades (Holometabola in blue; Paraneoptera in yellow, Polyneoptera in brown, and Palaeoptera in red), and wingless insects in grey. The drawings illustrate five species in which male dimorphism is associated with the expression of AMTs: (A) a major and (A’) a minor of the European earwig Forficula auricularia (Dermaptera, Forficulidae); (B) a major and (B’) a minor of the damselfly Mnais costalis (Odonata, Calopterygidae); (C) a major and (C’) a minor of the dung beetle Onthophagus taurus (Coleoptera, Scarabaeidae); (D) a major and (D’) a minor of the potter wasp Synagris cornuta (Hymenoptera, Vespidae); and (E) a major and (E’) a minor of the bladder grasshopper Bullacris membracioides (Orthoptera, Pneumoridae). The pizza charts represent: (F) the proportions of papers found in Web of Science by searching the name of each order (labelled ‘research effort’); and (G) the proportion of species known to present AMTs (labelled ‘occurrence of AMTs’) in each major insect clade, with the extremely diverse Holometabola split in five categories (its four largest orders and ‘other holometabola’).

22! ! Phenotypic plasticity and reversibility between tactics Our examples of field crickets and dung beetles (1.1) are illustrative of one of the axes of variation along which AMTs can be categorized: plastic (or reversible) versus fixed (or irreversible) throughout an insect’s adult life. The behavioural plasticity of calling by male crickets illustrates how tactics can be reversed or alternated during the adult life of a male, whereas the morphological dimorphism of male dung beetles represents AMTs that are irreversible after a male reaches adulthood. There seems to be a remarkable continuum between these categories in nature, where different degrees of reversibility are possible. An example of extremely reversible AMTs is found in the butterfly Lycaena hippothoe, where males use a flexible combination of territoriality (perching) and patrol flights, the frequencies of which are correlated with a fluctuating ecological feature, favourable weather conditions (Fischer and Fiedler 2001). Such highly plastic mating tactics are to be expected whenever dynamic extrinsic factors play an important role in the relative fitness of the different alternatives (Tomkins and Hazel 2007). However, these factors need not be related to climate or weather; the social environment can exert a similar effect. High population density usually triggers the switch from a calling to a satellite tactic in crickets (Cade 1980; Cade 1981; Cade and Wyatt 1984; French and Cade 1989), bushcrickets (Bailey and Field 2000; Feaver 1983), and also Ligurotettix grasshoppers (Greenfield and Shelly 1985; Shelly and Greenfield 1985 1989). In all these cases, the relative payoffs of calling and satellite tactics seem to be so strongly influenced by fluctuating environmental conditions that, even if an intrinsic male trait (such as size or age) affects which tactic is more likely to be adopted, either tactic can be the optimal decision for any given male under the right conditions. In contrast, in some mating systems only a limited number of switches between alternatives, or only switches in one direction, are possible during an adult’s lifetime. This is illustrated by the damselfly Calopteryx maculata, whose males defend territories at the beginning of their adult life, switching to a sneaking tactic later in life when their resource holding potential declines (Forsyth and Montgomerie 1987). Similarly, in the scarlet dwarf dragonfly Nannophya pygmaea, territorial males also switch to a sneaking tactic after being displaced from their territories by younger males (Tsubaki and Ono 1986 1987). Constraints on the number or direction of switches are expected when the expression of one of the tactics is clearly more costly than the alternative, and such costs can only be paid by males that are large, young, or in good condition (hereafter called majors). As a consequence, majors can often switch between tactics, whereas smaller/older/poorer condition males (hereafter called minors) are more commonly

! 23! constrained to adopt the less costly mating tactic. But this rule is not without exceptions, and the opposite pattern is found in the damselfly Paraphlebia zoe, whose majors (black-winged) seem fixed at performing a territorial tactic, whereas minors (hyaline-winged, usually satellites) with relatively large body sizes can become territorials under certain social conditions (Munguia-Steyer et al. 2010; Romo-Beltran et al. 2009). For heuristic purposes, in Table 1.2 we have categorised insect AMTs as ‘reversible’ or ‘fixed’ during an adult insect’s life. Although we recognise that this is probably an over-simplification, it does allow us to speculate on general patterns. In particular, the category ‘reversible’ includes different degrees of reversibility. With this criterion, reversibility seems extremely common in insect AMTs, occurring to some extent in at least 65% of the reported cases. This proportion is not general across orders however; whereas full behavioural plasticity is the most common type of AMT in the Mecoptera (all three cases), Lepidoptera (at least 11 out of 13 cases), Diptera (at least 17 out of 22 cases), Orthoptera (at least 21 out of 24 cases), Hemiptera (at least 10 out of 12 cases), and Odonata (at least 18 out of 22 cases), AMTs are usually fixed in the adult stages of Coleoptera and Hymenoptera, whose adults present reversibility of tactics in only seven out of 38 cases and 30 out of 67 cases, respectively.

Costly secondary sexual traits and tactic conditionality Secondary sexual traits in insects include male weapons, such as spines, spurs, horns, and the elongation and/or thickening of legs, mandibles, antennae, and forceps; as well as ornaments, such as the expansion and/or presence of colourful patches on wings, legs and thoraxes of males. These traits often exhibit ‘exaggerated’ morphologies (sensu Emlen and Nijhout 2000), being much larger than other appendages, and can even be larger than the rest of the body. Producing and maintaining enlarged weapons and ornaments is inevitably costly (Emlen 2001; Jennions et al. 2001), and such costs cannot be paid to the same extent by all males in a population. Consequently, exaggerated traits typically display remarkable variation, with large males expressing them to a disproportionally greater extent than small males (Wilkinson and Taper 1999). Moreover, the degree of phenotypic variation in exaggerated structures is expected to be greater in species under more intense sexual selection, which has been demonstrated for the forceps of earwigs (Simmons and Tomkins 1996), the extremely long eye stalks of stalk-eyed flies (Wilkinson and Taper 1999), and the enlarged mandibles of stag beetles (Knell et al. 2004).

24! ! Heterogeneity in the ability of males to produce and maintain secondary sexual traits makes fighting and defending territories or females only profitable to the large males of a population. In contrast, small males can actually benefit from avoiding the costs of producing weapons or engaging in fights altogether (Oliveira et al. 2008). Weapons and ornaments of small males are often much reduced or completely absent, which generates dichotomous morphological variation among males, known as male dimorphism (Gadgil 1972) or intrasexual dimorphism. In such cases, one or both male phenotypes are usually specialised and constrained to employ one of the AMTs, and therefore the incidence of reversibility in insect AMTs is usually lower in orders with high frequencies of morphological dimorphism (Figure 1.7). In the Coleoptera and in the Hymenoptera, 74% and 60% of all cases of AMTs are accompanied by dimorphisms in a way that the morphology of each male phenotype determines, or at least strongly influences, the mating tactic employed by males of that phenotype. Meanwhile, in the other insect orders, from zero to 36% of AMTs occur in species that also have male dimorphisms. The exception here is the Dermaptera (earwigs), where, despite well known cases of male dimorphism, the nature of and flexibility in AMTs is not well understood. When distinct male phenotypes are determined by Mendelian inheritance, the adoption of a mating tactic is fixed, and completely insensitive to the environment. This phenomenon has traditionally been referred to as ‘alternative strategies’ (1.2; Gross 1996), and seems very rare in insects, being perhaps applicable to in one genus of seaweed flies (Dunn et al. 1999), but confirmed in one genus of damselflies (Tsubaki 2003), and one species of cricket (Tinghitella 2008). More commonly, fixed AMTs employed by morphologically distinct male phenotypes are conditionally expressed, such that environmental effects play the predominant role in tactic expression, even if a male’s tactic and morphology is irreversible after it reaches adulthood. Traditionally named ‘conditional strategies’ (1.2; Gross 1996), in these cases environmental cues determine which tactic is expressed by a given male at a given time (for reversible AMTs) or throughout his adult life (for fixed AMTs). This sort of sensitivity to the environment is central to insect development, behaviour and general life history (Whitman and Ananthakrishnan 2009), and it is no different with their mating tactics. There are over 180 species of insects in which conditionally expressed AMTs have been detected, and this number is spread across 11 orders and all four major insect clades (Holometabola, Paraneoptera, Palaeoptera, and Polyneoptera; Table 1.2). The cues that trigger the permanent or temporary adoption of a mating tactic can be almost as diverse as the mating tactics themselves (Table 1.1). Cues can be extrinsic

! 25! factors, such as density of competing males or weather conditions, or intrinsic traits, such as age, body size or colour. In many cases intrinsic and extrinsic factors interact, for example when competitive ability interacts with the density of competing males to determine the ‘previous success’ of a given male, which is then used by this male to decide which mating tactic to adopt. The most common cues used in the expression of conditional mating tactics in insects are body size and condition, both emerging as an interaction between intrinsic traits (such as foraging efficiency) and extrinsic traits (such as food availability). We refer to the cues that determine conditional tactics as ‘environmental cues’, because of the general influence of the environment on this type of AMT.

1.0

0.8 ●

● 0.6

0.4 ●

● ● ● 0.2 ● ●

●● 0.0

Incidence of male dimorphism (%) 0.0 0.2 0.4 0.6 0.8 1.0 Incidence of reversibility (%)

Figure 1.7 When AMTs are associated with male dimorphism, it is common for one, and sometimes both male phenotypes, to be specialised and therefore constrained to employ one of the possible tactics. As a result, the incidence of reversibility in insect AMTs is usually significantly lower in orders with high frequency of male dimorphism (Spearman, rs = -0.794, P < 0.01, n = 10 orders). Although this correlation is very clear, there currently are no hypothesis to explain why some orders have higher frequencies of male dimorphism than others. Dermaptera (with a single species, Forficula auricularia) is removed from this analysis because, despite the well known cases of male dimorphism, the degree of reversibility in AMTs is not well understood in the group. The two overlapping open circles represent the orders Blattodea and Mecoptera, which have the same combination of 100% reversible AMTs and 0% with male dimorphism.

26! ! Variation in AMTs and male dimorphisms Male dimorphisms have been documented for at least 8 orders, and over 70 different insect species so far. In the basal Palaeoptera, all cases of AMTs were described for the order Odonata, and in some damselflies these alternative tactics are correlated with male dimorphisms. In the genus Mnais, for instance, majors have orange wings, are territorial and guard females, whereas minors have pale wings, are non-territorial and sneak copulations (Tsubaki 2003). This dimorphism is linked to a genetic polymorphism at a single autosomal locus, and the orange wing pigmentation is inherited in a Mendelian fashion, constituting one of the rare cases of alternative strategies (sensu Gross 1996) in insects. Within the Polyneoptera, the classic cases of AMTs consist of full behavioural plasticity in the calls of different species of field and bushcrickets (see section 1.1 above), without any morphological correlates. However, there are some cases of male dimorphisms in this clade as well, most of which represent conditionally expressed dimorphisms. These conditional dimorphisms include the swollen abdomens of bladder grasshoppers (Donelson and van Staaden 2005), the enlarged mandibles of wetas (order Orthoptera; Koning and Jamieson 2001), and the elongated forceps of earwigs (order Dermaptera; Simmons and Tomkins 1996; Tomkins 1999; Tomkins and Simmons 1996). In all these cases, male phenotype expression is tightly linked to body size, which is greatly influenced by diet in the earwigs (Tomkins 1999) and in insects generally (Nylin and Gotthard 1998). But at least in bladder grasshoppers, the expression of the minor phenotype is also triggered by the crowding effect of high population density (Donelson and van Staaden 2005). Within the Paraneoptera, the dichotomy between dispersing versus fighting seems to be common and wingless/winged dimorphism in both sexes has been reported for thrips (order Thysanoptera; Crespi 1988a), planthoppers (Langellotto et al. 2000), and chinch bugs (order Hemiptera; Fujisaki 1992). In all these cases, wingless males, or males with shorter wings, attain copulations more frequently at their place of birth than their winged counterparts, generally because they are more aggressive and repel other males. Winged males are capable of dispersing and seeking copulations in other places. In contrast, in the thrips Elaphrothrips tuberculatus, all males are winged, and whereas large males are fighters and defend egg-guarding females, small males sneak copulations with females guarded by large males (Crespi 1988b). Other than male size, local sex ratio also influences the adoption and switching between fighting and sneaking tactics in this species (Crespi 1988b).

! 27! Finally, male dimorphism has been extensively reported for the Holometabola, manifested mainly in the horns and mandibles of several groups of beetles (order Coleoptera; Emlen et al. 2007), the colours of butterfly wings (order Lepidoptera; van Dyck and Wiklund 2002), as well as the wings, heads, mandibles, and body size of several families of Hymenoptera (e.g., Cook and Bean 2006; Danforth 1991; Simmons, et al. 2000). It seems that this group surpasses any other insect group in terms of the repetitive evolution of morphological dimorphism among males. However, the effect of research effort must not be overlooked, and the fact that these orders are by far the most studied (see Figure 1.6 F-G) might be the reason why male dimorphisms have more often been recorded for them. It is also important to emphasize that the degree of male dimorphism of a given species can also vary in a continuous fashion, and interestingly the whole continuum can be found within the Hymenoptera alone. In some species of sweat bees from the genus Lasioglossum (Halictidae), males can be clearly assigned to one of two phenotypes, each of them with very distinct head width and mandible size (Houston 1970; Kukuk 1996). Large headed males are flightless fighters that never leave their nest to mate, whereas small headed males fly out of the nest and disperse before mating (Kukuk and Schwarz 1987). At the other end of the continuum, the AMTs adopted by a male can depend on its body size, but the distribution of body sizes or the distribution of secondary sexual trait sizes can be unimodal and somewhat normal, with no evidence of morphological male dimorphism. In the wasp Polistes dominulus (Vespidae), males can employ a resident mating tactic that involves territoriality, aggressiveness, and site-faithfulness, or a transient mating tactic that involves larger ranges, no aggressiveness, and little site tenacity (Beani and Turillazzi 1988). Larger males are residents more frequently than small males, but some individuals switch between tactics, and the distribution of male sizes is not bimodal in the population (Beani and Turillazzi 1988). A great many species actually fall somewhere in the middle of the continuum between monomorphism and dimorphism. In an undescribed species of Philotrypesis fig wasp, for instance, males from the ‘aggressive’ phenotype are much more likely to fight than males from the ‘passive’ phenotype, but male phenotypes have overlapping distributions of both body sizes and mandible sizes (Cook and Bean 2006). The dimorphism can only be detected from the ratio of mandible size to head size (which is higher in the aggressive phenotype), constituting a phenomenon described as ‘cryptic male dimorphism’ (Cook and Bean 2006).

28! ! Insect development and the evolution of male dimorphism As in all arthropods, insect growth is constrained by their hardened exoskeletons, resulting in an punctuated growth pattern restricted to moulting events. The number of such events is usually fixed within species, and the last moulting event precedes the adult stage, when sexual maturity is reached and growth ceases in the great majority of insects (the only exceptions are silverfish and bristletails; Triplehorn and Johnson 2005). This pattern of growth holds interesting implications for the evolution of fixed AMTs. Arrested growth in adult insects means that if the adoption of AMTs is tightly connected to body size, sexually mature males cannot switch between tactics, as reversibility is constrained by the fixed adult size. For example, in the bladder grasshopper Bullacris membracioides, AMTs are tightly connected to male size and morphology, which are strikingly different between male phenotypes (VanStaaden and Romer 1997). Large males have wings and inflated abdomens, which are used to emit high intensity calls to females, whereas non-inflated small males are wingless and adopt a satellite strategy, parasitizing the sexual calls of large males (Donelson and van Staaden 2005; VanStaaden and Romer 1997). When a male reaches sexual maturity, growth ceases, and his mating tactic becomes fixed for the rest of his adult life (Donelson and van Staaden 2005). So when small male bladder grasshoppers moult into adulthood, they become permanent satellites, providing a classic example of fixed AMTs. Another implication of insect growth patterns is that punctuated growth can underlie the origin of morphological polymorphisms. If different males are capable of arresting growth and reaching maturity after different numbers of moults, adult size distribution can become multimodal. In the wellington tree weta, Hemideina crassidens, males can mature at the 8th, 9th, or 10th instar, generating a trimorphism in adult male head size. This head polymorphism is linked to AMTs in the species, although only two tactics (and not three) exist: 10th instar males defend harems and engage in intense male-male fights, whereas 8th and 9th instar males seem to represent a sneaker strategy based on acquiring mates by either furtively invading galleries defended by 10th instar males (Kelly 2008; Kelly and Adams 2010) or by mating with females in small galleries that are inaccessible to large males (Kelly 2006). In the Holometabola, development involves immature larval stages and a complete metamorphosis from the last immature larval stage to the adult stage (Triplehorn and Johnson 2005). It is tempting to speculate whether holometabolous development, by providing a window to "decide" which phenotype to become, and how to distribute resources between competing traits, could allow greater

! 29! morphological divergence between different adult male phenotypes than hemimetabolous development. By confining resource distribution in the adult stage to one moment in development, conditionality — the threshold expression of traits in relation to the environment or status — might be favoured through the greater reliability of the body size cue in holometabolous insects. An interesting prediction that arises from this idea is that genetic polymorphisms should be more common in hemimetabolous insects, and conditional dimorphisms more common in holometabolous insects. At first glance this seems to hold true, as most of the few cases of male dimorphism that are purely genetic in arthropods are in groups that do not go through complete metamorphosis, such as crickets (Tinghitella 2008), damselflies (Tsubaki 2003) and crustaceans (Shuster 2008). However, important exceptions exist, such as one case of purely genetic polymorphism in a holometabolous group, the seaweed flies (Dunn et al. 1999), and four cases of conditional morphological dimorphism among males of hemimetabolous insects: earwigs (Tomkins 1999); bladder grasshoppers (Donelson and van Staaden 2005); oriental chinch bugs (Fujisaki 1992); and planthoppers (Langellotto et al. 2000). A final implication of insect developmental patterns on the evolution of AMTs is the ability (or inability) of determining offspring sex, coupled to the importance of offspring provisioning. When parental provisioning decisions determine offspring body size — and consequently male phenotype — dimorphisms might be constrained by the parents’ inability to determine or discriminate the sex of each individual offspring. This is the case of the dung beetle O. taurus, where parents gather all the dung on which the offspring will feed throughout their entire development, and the amount of parental provisions strongly influence offspring body size and thus male phenotype (Hunt and Simmons 2000). Horn length in this species is strongly dimorphic and associated with AMTs (see session 1.1). But because dung beetle parents seem unable to discriminate the sex of their offspring (Kishi and Nishida 2008), neither the brood masses built for male offspring, nor the distribution of body sizes in those offspring are dimorphic (Buzatto et al. 2012b, Chapter 4). It is hence possible that the inability of parents to discriminate the sex of their offspring is the reason why male offspring of intermediate body size and horn length are produced, even though these males are not optimally adapted to the fighter or sneaker mating tactic. In contrast, in Dawson’s burrowing bees the amount of pollen that mothers provision their nests with will also determine the size and mating tactic of their offspring, but in this case females seem aware of the sex of the offspring they are going to produce, digging cells that are bimodal in size distribution even before they begin to provision those cells (Tomkins et

30! ! al. 2001). Unsurprisingly, in this species no intermediate sized males are produced, and male offspring body size is strongly dimorphic (Alcock 1997; Simmons et al. 2000).

1.5 Variation in female mating systems

At the time of the original publication of Thornhill and Alcock’s (1983) volume, all cases of AMTs known were restricted to males. This is somewhat unsurprising considering that mating systems theory has initially focused on the manner and the degree to which the limited sex (usually males) monopolises mates and/or the resources that their mates depend on (Emlen and Oring 1977), naturally leading to a male perspective in the study of mating systems evolution. But perspectives have shifted significantly, and intrasexual variation in the mating tactics of females is receiving increasing attention. To date female AMTs have only been recorded in the orders Hymenoptera (where the great majority of cases are in ants) and Odonata (damselflies and dragonflies). In 19 genera (and at least 31 species) of ants, female AMTs associated with female dimorphism can be found among queens, in which one queen morph is adapted to initiate new colonies solitarily, whereas the other is adapted to join colonies that are already established (Heinze and Keller 2000). Queens that start new colonies are usually heavier and possess large fat reserves and fully developed wings, as starting a new colony requires them to disperse and to rely on fat/protein reserves and histolysed body tissue, at least until workers start to emerge and forage. In contrast, queens that join already established colonies, either by invading alien colonies or by simply returning to their own colony after mating, can be smaller and lighter, lacking large fat reserves, fully developed wings and flight muscles. A similar case of female dimorphism is also found in three species of the parasitoid wasp genus Melittobia (Eulophidae; Gonzalez and Matthews 2008; Matthews et al. 2009). Here, short winged females reproduce and oviposit in the same host from which they emerged, whereas long winged females disperse to find a new host (Matthews et al. 2009). Each of these cases can be considered alternative ways of obtaining fertilizations (and hence AMTs) because dispersing or not dispersing, as well as invading an alien colony or returning to the maternal colony (for ants), could influence the number of mates that a female will have, and the relatedness between her and her mates. A very different type of female AMTs and female dimorphism is found in the Odonata, where intrasexual colour dimorphisms in females seem to have evolved as a response to male harassment (Fincke et al. 2005). Sexual conflict arises from the optimal

! 31! number of matings, which is usually much higher for males, who tend to harass females that are not sexually receptive, resulting in significant fitness costs to those females (Sirot and Brockmann 2001). In 25 genera (and at least 135 species) of damselflies and dragonflies, usually in groups where males search for mates rather than defend specific territories (Fincke 2004; Fincke et al. 2005), females can be “heteromorphs” that are distinct (in behaviour and colour) from males, or “andromorphs” that greatly resemble males in colouration and behaviour (Johnson 1975). Interestingly, in some species two heteromorphs (and no andromorph) exist, leading to the proposition of two hypothesis for the evolution of female dimorphism: the ‘learned mate recognition’ hypothesis argues that females avoid male harassment by confusing males due to variation in female signals, while the ‘male mimicry’ hypotheses argues that females avoid harassment by being similar to other males (reviewed in Fincke et al. 2005). These female dimorphisms and polymorphisms are AMTs because they implicate different tactics to avoid male harassment and optimize the number of copulations achieved. It is clear that AMTs and dimorphic morphology are not as widely reported in female insects as they are in male insects. However, intrasexual variation in female mating systems must not be overlooked, and is certainly a promising topic for future research. We emphasize that there are other sources of intrasexual variation in the reproductive biology of female insects that we have not covered in this chapter, as here we defined AMTs as dichotomous ways to obtain fertilisations (see section 1.1). Variation in how choosy females of a given species are has been described as ‘alternative female choice tactics’ (Thornhill 1984), for example. However, we refrained from covering these cases here in detail because variation in the degree of choosiness among conspecific females is a significant topic in its own right (see for instance Gray 1999; Lehmann 2007; Perry et al. 2009). Furthermore, more often than not, variation in female choosiness is continuous and does not constitute clearly dichotomous ways to obtain (or resist) fertilisations. We have also not covered alternative tactics to complete other stages of female reproduction, such as alternative ways of provisioning nests or finding oviposition sites, for instance. For a more complete review of female alternative reproductive tactics that involve all stages of reproduction, we refer the reader to Brockmann (2008).

32! ! 1.6 Conclusion

Our poor knowledge of the astonishing insect diversity is surprising: it has been estimated that the nearly one million described species of insects represent only one fifth, and maybe only one tenth, of the actual number of extant species (Grimaldi and Engel 2005). Nevertheless, the fraction of insects that have already been described reveals an enormous variety of natural histories, and provides us with model organisms for many kinds of biological research. It is hard to conceive how much more we can learn with insects, but there is little doubt that knowing a higher fraction of their extant species will not only enhance our understanding of their evolution and ecology, but also provide us with insight into many different fields of biology. Exploring the evolution of AMTs in insects can shed considerable light on the importance of phenotypic plasticity for the evolutionary origin of novel traits (West- Eberhard 2003). To illustrate this point we return to our example of behavioural plasticity in the calling behaviour of male crickets (1.1). Males decide between singing to attract females or silently searching for females based on population density and/or recent success in aggressive interactions with other males (Hissmann 1990; Simmons 1986). In the field cricket Teleogryllus oceanicus, parasitism from the acoustically- orienting parasitoid Ormia ochracea selects against calling (Zuk et al. 1998). In at least two populations of this cricket — on the Hawaiian islands of Oahu and Kauai — there has been a rapid spread of a single locus mutation that affects the morphology of male wings, rendering them incapable of calling (Tinghitella 2008; Zuk et al. 2006). Because males harbouring this mutation can not call, they always use the satellite mating strategy and avoid parasitization by flies (Tinghitella et al. 2009; Zuk et al. 2006). Pre- existing phenotypic plasticity in male mating tactics in these field crickets has thus predisposed the rapid fixation of a mutation that encodes male dimorphic morphology. Populations have shifted from adopting plastic ARTs to a mixture of plastic and fixed ARTs in just a few years. This fascinating system illustrates how promising the study of AMTs in insects can be, and we believe that investigating insects’ AMTs from behavioural, developmental and genetic perspectives offers great opportunities for expanding our understanding of evolutionary innovation, and subsequent speciation.

! 33!

34! !

CHAPTER TWO

Genetic variation underlying the expression of a polyphenism

! 35! !

36! ! 2.1 Abstract

Polyphenic traits are widespread, and represent a conditional strategy sensitive to environmental cues. The environmentally cued threshold (ET) model considers the switchpoint between alternative phenotypes as a polygenic quantitative trait with normally distributed variation. However, the genetic variation for switchpoints has rarely been explored empirically. Here we used inbred lines to investigate the genetic variation for the switchpoint in the mite Rhizoglyphus echinopus, in which males are either fighters or scramblers. The conditionality of male dimorphism varied among inbred lines, indicating that there was genetic variation for switchpoints in the base population, as predicted by the ET model. Our results also suggest a mixture between canalised and conditional strategists in R. echinopus. We propose that major genes that canalise morph expression and affect the extent to which a trait can be conditionally expressed could be a feature of the genetic architecture of threshold traits in other taxa.

! 37! 2.2 Introduction

Natural dichotomous variation occurs in a great variety of morphological and life history traits (Moran 1992). When the dichotomy is manifested between the two sexes, the phenomenon is readily recognized as sexual dimorphism, the central focus of an enormous amount of evolutionary theory (Lande 1980; Andersson 1994). Nevertheless, dichotomous traits that are not purely related to sex abound, but have received far less theoretical treatment and empirical attention. Discontinuous variation across both sexes or within a single sex has been documented for a diversity of traits, including shape or presence of morphological structures (references in Roff 1994; Roff 1996; Brockmann 2001), behaviour (Moczek and Emlen 2000), and seasonal diapause (Mousseau and Roff 1989). The game theoretic explanations for dichotomous traits that are inherited in a Mendelian manner are relatively straightforward (Maynard Smith 1982). However, a great many dichotomous traits reflect a polyphenism, where the differential expression of alternative phenotypes from a single genotype are dependent on environmental conditions (West-Eberhard 2003). Known in evolutionary game theory as the conditional evolutionarily stable strategy, polyphenic dimorphisms are thought to evolve and be maintained when individuals’ phenotypes are decided through conditional decision rules (Dawkins 1980; Hazel et al. 1990). One such example is in the context of competitive ability, where status determines the fitness of a particular phenotypic tactic (Hazel et al. 1990; Gross 1996). Where this occurs the alternative phenotypes are expected to evolve to become status dependent, where individuals with a status higher than an evolutionarily stable switchpoint benefit from adopting one phenotype, whereas individuals with status lower than this switchpoint benefit from adopting the other (Hazel et al. 1990; Gross 1996; Tomkins and Hazel 2007). Similar models have been proposed to account for threshold sex determination (see Blackmore and Charnov 1989). Under the conditional strategy the environment that an organism experiences provides an important contribution to the phenotypic outcome of development, and the genetic basis to phenotype expression becomes complicated. Noting that traits with discontinuous variation could be inherited in the same way as continuous traits, Falconer (1989) suggested that dichotomous traits actually have an underlying genetic architecture that varies continuously due to polygenic effects, coupled with a threshold mechanism that generates discontinuity in trait expression. Such traits had been named threshold traits in earlier theoretical work (Dempster and Lerner 1950), and the concept

38! ! of quantitative genetic variation underlying their expression became the central idea of the 'liability model' of quantitative genetics (Falconer 1965). The widely used notion of 'liability' assumes that a fixed threshold overlies a continuously distributed liability, which itself is influenced by both genetic and environmental factors (Falconer 1989). Many threshold traits are environmentally sensitive or environmentally cued. Where this occurs, the ‘environmentally cued threshold’ model (henceforth ET) treats liability as a heritable distribution of sensitivities to the environmental cue (Hazel et al. 1990; Tomkins and Hazel 2007). In the light of this model, the switchpoint itself can be understood as a polygenic trait with a large additive genetic component that is normally distributed and subject to selection (Hazel et al. 1990; Roff 1994; Tomkins and Hazel 2007). Thus the genetic variation for switchpoints in the ET model explains the fact that there is usually a range of cue values where both phenotypes are produced; the larger the genetic variance in switchpoint distribution, the more gradual the increase in the cumulative frequency of the alternative tactic when it is plotted against the cue (Figure 2.1). This theoretical interpretation of the manner in which populations are sensitive to environmental cues is an important facet of the ET model; with it comes the prediction that populations depauperate in genetic variance will have steeper cumulative frequency distributions (i.e. a narrow range of switchpoint values) compared to populations with much genetic variance. In the ET model, because some individuals switch earlier than others in response to the same level of cue (e.g. Tomkins 1999) there is a range of values of the cue where both phenotypes are expressed. Although under the ET model this overlap is thought to represent the switchpoint distribution, an alternative explanation, or at least an explanation for largely overlapping distributions of tactic expression, is that populations represent a mixture of strategies (Lively et al. 2000), with some genotypes expressing environmentally cued phenotypes, while others are canalised and express only one phenotype regardless of the strength of environmental cues. In this case reducing genetic variation would result in the canalised expression of phenotypes in some cases and environmental sensitivity in others. With a strategic model, Lively (1986) has demonstrated that the coexistence of conditional and canalised strategists in the same population is possible, and more recently Plaistow et al. (2004) predicted that such coexistence is actually more likely to evolve than the occurrence of only conditional strategists, as long as there is a cost for conditionality. From a theoretical standpoint that accounts for the quantitative genetics underlying the dimorphism, the potential for this mixture of strategies has been incorporated in the ET model (Hazel et

! 39! al. 2004), and empirical studies have found evidence for mixtures of conditional and pure strategists in the predator-induced defence dimorphisms of barnacles (Lively et al. 2000) and daphnids (Hammill et al. 2008).

(a) Tactic A

Tactic B T

or Fitness frequency

Status variation

(b) Mean switchpoint (c) Mean switchpoint

Cumulative Cumulative frequency of frequency of tactic A tactic A

Frequency

Switchpoint variation Switchpoint variation

Status or switchpoint Status or switchpoint

Figure 2.1 (a) The ‘status-dependent selection’ model: the bell-shaped curve is the normal distribution of males’ status, and the lines represent the fitnesses of male phenotypes as a function of status (black line for tactic A; grey line for tactic B). Because these fitness functions intersect at the threshold T, individuals of status higher than T benefit from expressing the tactic A phenotype, whereas individuals of status lower than T benefit from expressing the tactic B phenotype. Horizontal arrows indicate the proportion of males expressing each phenotype (black for tactic A; grey for tactic B), and they gradually merge at the threshold, indicating genetic variation for the switchpoint. (b) The ‘environmentally cued threshold’ model (ET): genetic variance in switchpoint distribution (the grey dotted line indicates mean switchpoint; the solid bell- shaped curve indicates switchpoint distribution) explains why there is overlap in the status of the two male morphs, generating the cumulative normal curve of tactic A expression. (c) In this case, switchpoint distribution is narrower, and hence, the cumulative normal curve of tactic A expression is steeper than in the previous case. Under the ET model, scenario C would be predicted after the establishment of inbred lines, as genetic variation for switchpoints should be very small within each inbred line.

In the present study, we investigate the importance of genetic variation in the conditional expression of alternative male phenotypes in the acarid mite Rhizoglyphus echinopus (Fumouze and Robin 1868). Two male phenotypes occur in R. echinopus: fighters possess a very thick and sharply terminated third pair of legs, while scramblers legs are all equally thin and without a sharp tip. Fighter males use their heavily armoured legs to kill rival males and monopolize females, whereas scrambler males search for unguarded females to mate with (Radwan 1993). This male dimorphism in R. echinopus is known to be environmentally-cued by body size and

40! ! male density (Radwan 2001; Tomkins et al. 2011). Indeed, in our laboratory populations, fighter expression is positively related to body size (measured as the weight of the pre- imaginal, quiescent tritonymph [QTW]), and juveniles that are reared individually are more likely to become fighters than juveniles reared in groups of 20 (Tomkins et al. 2011). It appears counterintuitive that body size is termed an ‘environmental cue’ when body size is a trait of the organism, possessing genetic variation. However, because there is no expectation for there to be a correlation between environmental quality and the genetic propensity to be large or small bodied, genes for size segregate randomly with respect to the environments that they develop in and can, in modelling terms, be considered along with the environmental variation (W. N. Hazel personal communication; Tomkins and Hazel 2007). Here we standardized male density (all males were reared in isolation, see Materials and methods below) and focused on the body size of the mite itself as the environmental cue that correlates with male status (and hence competitive ability) and therefore influences male morph outcome. There is a great deal of overlap in the sizes of males that become either fighters or scramblers in R. echinopus, raising the possibility that there is either large genetic variance in switchpoint distribution, or that some genotypes are canalised to express either the fighter or scrambler morph irrespective of environmental cues. Furthermore the dimorphism in R. robini, the sister species to R. echinopus, has been shown to be largely insensitive to environmental cues (Radwan 1995, but see Smallegange 2011) giving support to the notion that the large variance in sensitivity to the environmental cue might be due to the presence of canalised genotypes in the population. Because the environmental cue in R. echinopus is body size, variation for switchpoint distribution needs to be disentangled from variation for the environmental cue itself (body size). This distinction is possible with the ET model, because it treats the distribution of environmental cues and switchpoints independently (see Box 2 in Tomkins and Hazel 2007). Here we test the 'switchpoint variance prediction' of the ET model, using inbred lines to reveal the genetic variation for switchpoints that underlies the conditional expression of alternative male phenotypes in R. echinopus. Because inbreeding increases homozygosity across loci (Falconer 1989), variation among inbred lines that are founded from the same population is mostly genetic, and reveals the variation between different genotypes of the original population. If variation in switchpoints is due purely to additive effects on sensitivity to environmental cues we expect all inbred lines to present conditionality of male dimorphism but with different switchpoints emerging in each of the lines (e.g. Hammill et al. 2008). Moreover, we expect inbred

! 41! lines to present steeper cumulative frequency distributions of fighters as body size increases (i.e. a narrow range of switchpoint values) compared to the base population, as each inbred line only contains a fraction of the genetic variation from the base population. If on the other hand some genotypes are canalised as appears to be the case in some populations of R. robini, we expect a more complex pattern of sensitivity and insensitivity to the environment to emerge, in which some inbred lines present conditionality and others do not.

2.3 Methods

Collecting and maintaining the base population All individuals used in this study were derived from a base colony of R. echinopus collected from an infested organic onion purchased at a health-food shop in August 2005. We kept the base colony in six Petri dishes (90 mm; part-filled with plaster-of- Paris) at 22°C (Binder KB 240 cooled incubator) in desiccators with 100 ml of KOH solution (153 g/l H2O) to maintain > 90% humidity. Each dish had a standing adult population of several hundred individuals, with occasional population peaks of a few thousand individuals, when we discarded a random half of the colony to prevent overpopulation. We periodically moved mites between the Petri dishes to maintain a genetically mixed culture, and fed the colonies with tissue paper and dried Alinson’s yeast ad libitum. Because the base population had a female biased sex ratio and tested positive for the sex ratio distorting bacterium Wolbachia sp. (J.L. Tomkins, unpublished data), we cured the laboratory populations by moistening the plaster-of-Paris in the colonies with a mixture of tetracycline and water over a period of three generations. The generation time of the mites is approximately two weeks, and experimental manipulations began after > 50 generations in the lab.

Rearing individual larvae We isolated approximately 100 larvae from the base colonies and reared them individually in cylindrical glass vials (diameter = 10 mm, height = 14 mm) with plaster-of-Paris bases 4-5mm thick that we kept damp by placing them on a piece of damp filter paper in a Petri dish. These vials will henceforth be referred to as "individual vials". We closed individual vials with a wad of non-absorbent cotton wool (BSN Medical), and provided the mites inside with food ad libitum, which consisted of a single ball of Alinson's dried yeast that is many times heavier than an adult mite, and

42! ! clearly more than an individual consumes throughout development. This procedure allowed us to obtain virgin adults from the base colony.

Pairing to obtain full siblings Using the virgins obtained from the base populations, we paired 40 couples separately in cylindrical plastic "mating tubes" made by cutting the top 25mm off a 25ml polypropylene vial (Interpath). These containers had a screw cap ventilated with 6mm holes that we covered with porous plastic (Genesee Scientific), and a plaster-of-Paris base 4-5mm thick that we kept damp by placing them on a piece of damp filter paper in a Petri dish. We sprinkled mating tubes with Alinson’s dried yeast and, five days after pairing, we started to check each of them daily for the presence of eggs. Successful pairings were capable of producing approximately 200 eggs. As soon as the eggs started to hatch, we isolated 20 larvae produced by each pair, and raised them to adulthood individually, as described above. This procedure allowed us to obtain virgin full siblings derived from 40 outbred pairs. We used these virgin full siblings to establish inbred lines (see below).

Establishing inbred lines Using the offspring from the 40 outbred pairs, we randomly selected 35 pairs of virgin full siblings and paired them in mating tubes, always providing ad libitum yeast. Five days after pairing, we isolated 20 larvae produced by each pair, and raised them to adulthood individually as described above. Throughout the experiment, approximately 14% of full sibling pairings failed to produce any offspring, rendering the lines extinct. We repeated the procedure of pairing full siblings and raising their offspring individually for six generations, after which inbred colonies were established in cylindrical plastic "colony plates" made by cutting the top 25mm off a 500ml polypropylene container (Interpath). These containers had a screw cap ventilated with 6mm holes that we covered with porous plastic (Genesee Scientific), and a plaster-of- Paris base 4-5mm thick that we kept damp by placing them on a piece of damp filter paper in a Petri dish. Between April 2008 and September 2009, we kept the inbred colonies in colony plates with ad libitum dried yeast and paper at 4ºC, hence increasing their generation time to about two months (J.L. Tomkins, unpublished data). In September 2009, we put the inbred colonies back into 22ºC incubators, and resumed inbreeding (as already described) for two further generations. Overall, inbred colonies

! 43! went through eight generations of full sibling pairings: an expected inbreeding coefficient of F = 0.826.

Assaying inbred lines for male polyphenism We successfully established eight inbred lines of R. echinopus through eight generations of full-sib crosses. These lines were assayed for the conditional expression of male phenotype. We isolated larvae from the inbred lines in individual glass vials, and raised them to adulthood at 22 ºC with ad libitum dried yeast. We assayed four lines (on average 50 larvae from each of them) at a time, and kept the vials containing all the larvae in the same incubator, totalling 200 larvae in the incubator at any given time. This process was repeated four times. Individuals cannot be sexed as nymphs, and sex ratio varied between the lines in the first two repeats. Therefore, we had to adjust the total number of larvae that we isolated from each line based on the sex ratio of each of them in the first two repeats, in order to try to obtain 50 adult males from each line. With four repeats, we managed to analyse a total of 800 larvae (100 each from lines 10, 18, 19 and 29; 142 from line 27; 62 from line 30; 66 from line 36; and 130 from line 38). During each repeat, we shuffled the position of the vials containing juveniles from each line, to avoid any effect of slightly different temperature or odours in different parts of the incubator. At least twice per day from the 6th to the 11th day after isolation, we checked the mites for the quiescent stage of the tritonymphs, when the last nymphal stage stops moving for 6-12 hours prior to its eclosion as an adult (Radwan et al. 2002). We weighed every quiescent tritonymph to the nearest 0.0001 mg on a Sartorius SE2 balance, and then returned them to their vial. Fighters loose more weight than scramblers at eclosion, hence the weight of the pre-imaginal quiescent tritonymph (QTW) is the best proxy to individual condition immediately prior to the adult phenotype expression (Radwan et al. 2002; Tomkins et al. 2004). On the day following weighing, we recorded sex and male morph for each individual. We observed 17 ’intermorphs’ where on one side the third pair of legs is thick and sharply terminated and the other is thin and without a sharp tip (like a scrambler, 11/17 with fighter right leg). The frequency of such intermediate males was positively correlated with the frequency of fighters in these populations (P < 0.05), suggesting that these males experienced problems in the developmental pathway leading to fighters. However, we excluded these males from all further analysis, as in these rare events (4.52 % of males) we could not confidently assign the males to a particular morph.

44! ! 2.4 Results

Male dimorphism in the inbred lines Inbred lines differed in the expression of male polyphenism, with fighter expression ranging from 0% to 73% (Figure 2.2). We combined the data from the eight inbred lines and modelled the probability of males developing the fighter phenotype by fitting a generalized linear model with binomial error distribution, coding fighters as one and scramblers as zero and using the weight of the pre-imaginal quiescent tritonymphs (QTW) and line as independent variables. We then performed an analysis of deviance in which we added sequentially to the null model the predictor variables in order to quantify their significance (Table 2.1). Next, in a separate analysis for each inbred line, we again fitted generalized linear models with binomial error distribution and using QTW as the independent variable. The fitted lines (Figure 2.2) represent the cumulative frequency distribution of individuals expressing the fighter phenotype as QTW increases. This analysis revealed that lines 18 and 27 showed conditionality of fighter expression (Table 2.2, Figure 2.3a), whereas all of the other lines did not. Among the lines that did not show conditionality of fighter expression, either both morphs were expressed across all values of body size (as in Lines 19, 38 and more clearly in Line 29), or fighter morph expression was suppressed (partially in Lines 10 and 36 and completely in Line 30, see Table 2.2 and Figure 2.2).

When inbred lines showed conditionality To determine whether there were shifts in the average switchpoint of the two inbred lines that showed conditionality (lines 18 and 27), we combined the data from these two lines and from the base population and used a generalized linear model approach with morph as the binomial dependent variable and QTW and line/base as independent variables. We then performed an analysis of deviance in which we added sequentially to the null model the predictor variables in order to quantify the significance of each parameter (Table 2.3). The interaction between these two variables was not included in these models because a more powerful approach was used to detect differences in the variance of switchpoint distributions between these lines (see below). The mean switchpoint (predicted fighter probability of 0.5, Hazel et al. 1990) occurred at a QTW of 0.0568 mg in line 27, significantly greater than the base population (Table 2.3; 0.0469 mg, Figure 2.3b) but not significantly greater in line 18 (0.0527 mg).

! 45! Base (n = 62) L27 (n = 47) L18 (n = 52) 1.0 1.0 1.0

1.0 62.9% 1.0 14.6% 1.0 36.5% 0.8 fighters 0.8 fighters 0.8 fighters 1.0 1.0 1.0 0.8 0.8 0.8 0.6 0.6 0.6 0.8 0.8 0.8 0.6 0.6 0.6 0.4 0.4 0.4 0.6 0.6 0.6 0.4 0.4 0.4 predicted.probability predicted.probability predicted.probability 0.2 0.2 0.2 predicted.probability predicted.probability predicted.probability 0.4 0.4 0.4 0.2 0.2 0.2 0.0 0.0 0.0 predicted.probability predicted.probability predicted.probability

0.2 0.02 0.04 0.06 0.08 0.2 0.02 0.04 0.06 0.08 0.2 0.02 0.04 0.06 0.08 0.0 0.0 0.0 0.02 0.04weight.range0.06 0.08 0.02 0.04weight.range0.06 0.08 0.02 0.04weight.range0.06 0.08 0.0 0.0 0.0 L19weight.range (n = 41) L38weight.range (n = 38) L29weight.range (n = 55) 1.0 0.02 0.04 0.06 0.08 1.0 0.02 0.04 0.06 0.08 1.0 0.02 0.04 0.06 0.08

1.0 weight.range 1.0 weight.range 1.0 weight.range 0.8 73% 0.8 32% 0.8 36% 1.0 1.0 1.0

0.8 fighters 0.8 fighters 0.8 fighters 0.6 0.6 0.6 0.8 0.8 0.8 0.6 0.6 0.6 0.4 0.4 0.4 0.6 0.6 0.6 0.4 0.4 0.4 predicted.probability predicted.probability predicted.probability 0.2 0.2 0.2 predicted.probability predicted.probability predicted.probability 0.4 0.4 0.4 0.2 0.2 0.2 0.0 0.0 0.0 predicted.probability predicted.probability predicted.probability

0.2 0.02 0.04 0.06 0.08 0.2 0.02 0.04 0.06 0.08 0.2 0.02 0.04 0.06 0.08 0.0 0.0 0.0 0.02 0.04weight.range0.06 0.08 0.02 0.04weight.range0.06 0.08 0.02 0.04weight.range0.06 0.08 0.0 0.0 0.0 weight.range weight.range weight.range 1.0 0.02 0.04 0.06 0.08 1.0 0.02 0.04 0.06 0.08 1.0 0.02 0.04 0.06 0.08

Probability of becoming a fighter L10 (n = 51) L36 (n = 47) L30 (n = 44)

1.0 weight.range 1.0 weight.range 1.0 weight.range 0.8 0.8 0.8 1.0 1.0 1.0 0.8 0.8 0.8

0.6 9.8% 0.6 2.1% 0.6 0%

0.8 fighters 0.8 fighters 0.8 fighters 0.6 0.6 0.6 0.4 0.4 0.4 0.6 0.6 0.6 0.4 0.4 0.4 predicted.probability predicted.probability predicted.probability 0.2 0.2 0.2 predicted.probability predicted.probability predicted.probability 0.4 0.4 0.4 0.2 0.2 0.2 0.0 0.0 0.0 predicted.probability predicted.probability predicted.probability

0.2 0.02 0.04 0.06 0.08 0.2 0.02 0.04 0.06 0.08 0.2 0.02 0.04 0.06 0.08 0.0 0.0 0.0 0.02 0.04 0.06 0.08 0.02 0.04 0.06 0.08 0.02 0.04 0.06 0.08 0.0 0.0 0.0 0.02 0.04 0.06 0.08 0.02 0.04 0.06 0.08 0.02 0.04 0.06 0.08

Quiescent tritonymph weight (mg)

Figure 2.2 The results of establishing inbred lines on the expression of conditional male dimorphism in the mite Rhizoglyphus echinopus. The first plot on the upper left corner shows that the probability of a juvenile becoming a fighter increases with the weight at the quiescent tritonymph stage (QTW) in the base colony. The remaining plots show the same probabilities for the eight inbred lines after eight generations of full-sib pairings. The identity of each line is given by an uppercase L followed by a number, sample sizes are given in parentheses, and the percentage of fighters is given inside each plot. Significant relationships are indicated by solid curves, whereas nonsignificant relationships are indicated by broken lines (Table 2.2 for the coefficients of each model and their respective significances). Conditionality was still significant in lines 27 and 18, although the remaining inbred lines did not show any relationship between the probability of a juvenile becoming a fighter and QTW (Table 2.2). Possible explanations for these heterogeneous results are discussed in the text.

In order to detect differences in the variance of switchpoint distributions between inbred lines 18 and 27 and the base population, we followed the analysis proposed by Tomkins and Hazel (2011), based on the ET model notion that the steepness of the

46! ! function that relates the probability of becoming a fighter to the values of body size is indicative of switchpoint distribution (Hazel et al. 1990; Tomkins and Hazel 2007; Tomkins and Hazel 2011). We estimated the standard deviation (SD) of the switchpoint distribution for the base and lines 18 and 27 by estimating the QTW at the probabilities of 0.1587 and 0.8413 of a male developing into a fighter, and calculating the range of QTW between these probabilities and the probability of 0.5 (i.e. to give 1SD above and 1SD below the mean switchpoint which we then averaged; Tomkins and Hazel 2011). We then squared these standard deviations to obtain an estimate of the variance in switchpoints for lines 18 and 27, and compared them to the base colony through variance ratio tests. The variance of the estimated distribution of switchpoints in Line 27 (1.266 x 10-4 mg2) was significantly smaller than the one obtained for the base colony

(6.101 x 10-4 mg2; F61, 46 = 4.821, P < 0.001). However, the variance of the estimated distribution of switchpoints in Line 18 (3.940 x 10-4 mg2) was not different from that of the base colony (F61, 51 = 1.548, P = 0.11).

Table 2.1 Analysis of deviance used to compare the relationship between the probability of a nymph of the bulb mite R. echinopus becoming a fighter (response variable) and body size (measured as quiescent tritonymph weight, QTW) among eight inbred lines after eight generations of full-sib pairings.

2 Model df Deviance Residual df Residual Deviance P ( χ ) Null 374 422.30 QTW 1 27.516 373 394.78 < 0.0001 € QTW + Population 7 87.021 366 307.76 < 0.0001

In order to provide an estimation of switchpoint variation in relation to the variation in the cue that determines male morph expression (i.e. body size), we divided the standard deviation of switchpoints by the standard deviation of the overall QTW distribution in each inbred line (as in Tomkins and Hazel 2011). We found that switchpoint variation in these inbred lines corresponded respectively to 211.52% and 132.82% of QTW variation. Meanwhile, the value obtained for the base colony was 250.63%. Finally, we found a positive correlation between the line average for male

QTW and the proportion of fighters in each inbred line (Pearson, r6 = 0.758, P = 0.03;

Figure 2.4), this correlation increased when the base colony was included (Pearson, r7 = 0.818, P = 0.007; Figure 2.4).

! 47! Table 2.2 The results of eight generalized linear models with binomial distribution and a logit link that we fit to the dataset of each inbred line (and the base colony), coding fighters as one and scramblers as zero in the dependent variable, and using QTW as the independent variable. The logistic curves described by the coefficients bellow for each inbred line are illustrated in Figure 2.2. The analysis was not performed for inbred line 30 because no fighters were produced in this line after inbreeding.

Population Coefficients Estimate Standard error z value P-value Base colony Intercept -3.168 1.720 -1.842 0.0654 QTW 67.579 31.485 2.146 0.0318 Line 27 Intercept -8.438 2.864 -2.946 0.00322 QTW 148.451 59.260 2.505 0.01224 Line 18 Intercept -4.430 1.715 -2.584 0.00978 QTW 84.058 35.969 2.337 0.01944 Line 19 Intercept -2.204 2.298 -0.959 0.338 QTW 61.920 44.602 1.388 0.165 Line 38 Intercept -3.825 2.184 -1.752 0.0798 QTW 63.297 44.075 1.436 0.1510 Line 29 Intercept -0.930 1.264 -0.735 0.462 QTW 7.345 24.386 0.301 0.763 Line 10 Intercept -1.464 2.623 -0.558 0.577 QTW -17.820 61.518 -0.290 0.772 Line 36 Intercept -2.316 4.795 -0.483 0.629 QTW -35.848 114.996 -0.312 0.755

2.5 Discussion

Evidence for a mixture of strategies Our approach aimed at using inbred lines to scrutinize the genetic variation in the conditional expression of alternative male phenotypes in R. echinopus. We found that the level of conditionality of male dimorphism varied greatly between the inbred lines. This variety of conditionality among these lines could indicate a natural occurrence of more than one genetic strategy in the base population – or at least variation in the genetic basis to phenotype expression in this species. In lines 27 and 18, conditional expression of male phenotypes was evident. In contrast, all the other inbred lines showed no conditionality, probably due to a high frequency of genotypes (perhaps 100% in line 30) that were only capable of expressing the scrambler phenotype (canalised strategists) under the environmental conditions they were reared in. The

48! ! coexistence of conditional strategists and canalised strategists in a single population is known (Lively 1999; Lively et al. 2000), and has received theoretical consideration in strategic (Lively 1986; Plaistow et al. 2004) and quantitative genetic models (Hazel et al. 2004). Hazel et al. (2004) proposed two genetic mechanisms for mixtures of conditional and canalised individuals: (1) the variation in switchpoints in the population is large relative to the distribution of cues and/or; (2) the population is polymorphic for an epistatic allele at a major locus that masks the expression of polygenic variation in switchpoint.

1.0 (a) Figure 2.3 (a) A comparison between the base colony and inbred lines 27 and 18 regarding the relationship

0.8 between the probability of a juvenile becoming a fighter and QTW. After eight generations of full- sib pairings, 0.6 this relationship was stronger in lines 27 (black full circles and black solid curve) and 18 (open circles and black 0.4 broken curve), than in the base colony predicted.probability27 predicted.probability18 predicted.probabilitySTK (grey full circles and grey solid curve), which did not experience any 0.2

Probability of becoming a fighter inbreeding, serving as a control. (b) A comparison between the base colony and inbred lines 27 and 18 regarding 0.0 estimated switchpoint distributions. 0.00 0.02 0.04 0.06 0.08 0.10 The arrows indicate mean 40 (b) weight.range switchpoints; the bell-shaped curves indicate switch- point distributions. After eight generations of full-sib

30 pairings, switchpoint distribution was narrower in lines 27 (black solid arrow and curve) and 18 (black broken arrow and curve), than in the base colony 20 (grey solid arrow and curve), which did not experience any inbreeding, serving as a control.

10 Frequency of switchpoints ofFrequency switchpoints

0

0.00 0.02 0.04 0.06 0.08 0.10 Quiescent tritonymph weight (mg)

These two scenarios described by Hazel et al. (2004) for the coexistence of canalised and conditional strategists in single populations are compatible with our results. The relative variation in switchpoint distribution in R. echinopus is extremely high, representing over 250 % of the variation in body size. This value is higher than all similar estimations recently calculated by Tomkins and Hazel (2011) for 18 populations

! 49! of male-dimorphic arthropods. Even in the inbred lines, the relative variation in the distribution of switchpoints in R. echinopus still represented over 130 % of the distribution of the cue (in the lines that showed conditionality). This fits the first scenario proposed by Hazel et al. (2004), in which the large distribution of switchpoints (in relation to the distribution of the cue) causes some genotypes to have their switchpoint set at a value that exceeds the natural range of the cue for that particular population (see Figure 2 in Hazel et al. 2004). In practice, these individuals are canalised to one of the male phenotypes, simply because they never experience cues strong enough to trigger the switch. This contrasts with other individuals in the same population that have switchpoints within the range of cues, and hence are still capable of conditionally expressing both phenotypes.

Table 2.3 Analysis of deviance used to compare the relationship between the probability of a nymph of the bulb mite R. echinopus becoming a fighter (response variable) and body size (measured as quiescent tritonymph weight, QTW) among the base population and the inbred lines that showed conditional male dimorphism after eight generations of full-sib pairings (Lines 27 and 18).

2 Model df Deviance Residual df Residual Deviance P ( χ ) Null 160 217.19 QTW 1 38.692 159 178.50 < 0.0001 € QTW + Population 2 6.931 157 171.56 0.03

The second scenario accounting for the coexistence of canalised and conditional strategists is the existence of an epistatic allele at a major gene that blocks the expression of one of the phenotypes (Lively et al. 2000). The presence of such major gene effects would cause the probability of becoming a fighter to reach an asymptote at less than 100%. This sub-maximal asymptote occurs because no matter how strong the cue becomes, individuals carrying the epistatic allele remain insensitive. This contrasts with the alternative hypothesis where increasing cue strength always yields increasing numbers of affected individuals. Sub-maximal asymptote seems to have occurred in line 18 and for perhaps the base colony of R. echinopus. These major gene effects would in theory be independent of the variation in switchpoint distribution, such that their presence could lead to a sub-maximal asymptote even in populations with very little genetic variation in switchpoints. Clearly when there is ample switchpoint variation their effects are harder to detect because the environmental cue may not be strong

50! ! enough to elicit 100% response, generating a similar pattern even in the absence of major gene effects. Although both scenarios for the coexistence of conditional and canalized individuals are compatible with our results, we are unable to distinguish conclusively between them. For example, in lines 10, 30 and 36, fighters were only very rarely produced. In these cases it is difficult to distinguish between whether the inbred lines contain distributions of switchpoints and distributions of body sizes (the putative cue) in a way that all individuals are smaller than their genetic switchpoints, or if these lines just contain a high frequency (100% in line 30) of a major gene that prevents the production of fighters. Evidence for the former mechanism may come from the fact that average male body size in each line was positively related to the proportion of fighters in that line (i.e. when the cue is stronger, in our case larger body sizes, more fighters are produced). Such a result is only expected if the switches are there but they simply do not get tripped.

0.8

0.6

0.4

predicted.probability

0.2 Proportion of fighter males

0.0

0.040 0.045 0.050 0.055 0.060

Average quiescentweight.range tritonymph weight (mg)

Figure 2.4 There was a positive correlation between the average quiescent tritonymph weight (QTW) of males and the proportion of fighters among the eight inbred lines and the base colony. Full circles represent inbred lines, and an open circle represents the base colony. The dashed line depicts the relationship between the probability of a juvenile becoming a fighter and QTW in the base colony (as illustrated in Figure 2.2 for all inbred lines and the base population).

The polygenic nature of switchpoints The relationship between body size and the probability of becoming a fighter was even stronger in lines 27 and 18 than in the base colony. This stronger relationship is reflected in the steepness of the function that relates the probability of becoming a

! 51! fighter to the values of QTW (Figure 2.3a). From these probability functions we estimated the switchpoint distributions in lines 27, 18 and in the base population, and demonstrated that these two inbred lines (more significantly line 27) had narrower switchpoint distributions when compared to the base colony (Figure 2.3b). These results support the prediction of the environmental threshold model that the distribution of switchpoints is genetically variable since the establishment of inbred lines caused a reduction in the variance of this trait. This study is not the first to provide evidence for genetic variation underlying the expression of polyphenisms. Sewell Wright (1934a,b), in an extensive study on the expression of extra digits in guinea pigs, found no detectable genetic variability within strongly inbred lines, but substantial variability between these lines, suggesting a great amount of genetic variation underlying this dichotomous trait. Wright's (1934a,b) study focused on analysing the proportions of the alternative phenotypes as the response variable, measuring the patterns of inheritance and inferring the sources of genetic and environmental variation, but without any measure of any putative environmental cues that influence the dimorphism. Therefore, the genetic variation for sensitivity to the cue (i.e. the switchpoint itself) is indistinguishable from the genetic variation for the cue itself, e.g. a hormone titre. This is an important distinction to be made, as in R. echinopus, male polyphenism is environmentally-cued by body size and male density (Radwan 2001; Tomkins et al. 2011), and body size itself is expected to present a strong additive genetic component. More recently, Páez et al. (2011) used a pedigree design with both maternal and paternal half-sib structure, and unveiled additive genetic variance in the switchpoint that determines the timing of sexual maturity (and consequently male polyphenism) in the Atlantic salmon. Our inbred line approach provides a novel and independent source of evidence to corroborate the idea that in cases where conditionality is isolated within the range of environmental cues, there is genetic variation for the hypothesised switchpoint distribution, as we would expect for a polygenic trait. Also consistent with our results, a study on the predator-induced morphological defences of Daphnia pulex produced reaction norms that suggest genetic variation in switchpoints (Hammill et al. 2008). However, the inducible morphological defences of D. pulex might be the result of more than one threshold trait, as Hammill et al. (2008) assigned points to different structures (pedestal and spikes), scoring individuals for different levels of induction, which were then normalized to vary between 0% and 100%. Consequently, genetic variation for these morphological defences is not necessarily related to a single threshold trait. On the other hand, in our study the discrete variation that defines a

52! ! polyphenism is absolutely clear, since there are no intermediate levels of expression of the fighter phenotype. Although we observed a few intermediate males (called intermorphs, see Materials and Methods), this phenomenon was excluded from our analysis on the basis of being rare and probably the result of development instability. Therefore, the variation between our inbred lines that were founded from the same base population and reared under common garden conditions (standardized temperature, light cycle, population density [every mite was reared in isolation] and diet [ad libitum yeast provided]), clearly reflects genetic variation underlying a single threshold trait.

Conclusion A great deal of phenotypic diversity is expressed as environmentally induced alternative phenotypes (West-Eberhard 2003). Alternative mating strategies are a particular case, with male dimorphism widespread among animal taxa (Oliveira et al. 2008). Although there are cases of alternative reproductive strategies where the phenotype adopted by a male is determined entirely by the alleles at a single locus (Shuster 2008), several cases reflect the conditional expression of male phenotypes in response to environmental cues (Emlen 2008). Despite conditional male dimorphisms being common, surprisingly few empirical studies have explored the importance of genetic variation in this type of plasticity (but see Emlen 1996; Doums et al. 1998; Tomkins and Brown 2004; Unrug et al. 2004; Páez et al. 2011). As far as we know, the present study is the first to report the effects of establishing inbred lines to directly investigate the genetic variation for the switchpoint that underlies the expression of male polyphenism. Our approach successfully detected genetic variation for the switchpoint distribution, thus we support a fundamental assumption of the environmental threshold model. Our inbred lines were all derived from a base colony collected from a single infested onion, and therefore our estimates of genetic variation could be conservative, nevertheless this population does contain enough genetic variation to evolve (Tomkins et al. 2011). We also found heterogeneous effects among our lines that suggest that our base population of R. echinopus could either be composed of a mixture of canalised and conditional strategists or that we isolated numerous lines with switchpoints beyond the range of body size cues. These results are therefore consistent with the very large amounts of genetic variation in switchpoints that we see in this species, or alternatively may indicate that major genes that canalise morph expression overlie the broad conditionality seen in the species. Radwan (1995) has shown that in R. robini, a species

! 53! that is largely insensitive to environmental cues for morph determination. This could give weight to the idea that there are major genes at play in our population. In Sancassania berlesei the male dimorphism is deemed to be conditional (Unrug et al. 2004). Even so, departures from conditionality in one population from Stirling prompted Tomkins et al. (2004) to examine offspring morph ratio patterns and the heritability of morphs; in this case the heritability was less than one and the ratios of male phenotypes in the offspring did not follow Mendelian patterns, suggesting major gene action did not predominate. How sensitive such techniques are for separating conditionality from major gene effects is unclear. For example some populations of R. robini seem much like S. berlesei and R. echinopus in their conditionality since Smallegange and Coulson (2011) estimate the heritability of male morphs in a Netherlands population to be 0.3, and Smallegange (2011) shows the male morph to be conditional on size. We anticipate that the variation in morph determination mechanisms seen in the genus Rhizoglyphus means that some populations may harbour mixtures of canalized and conditional strategists, and that further exploration of the genetics of our population may provide definitive evidence for this hypothesis. !

54! !

CHAPTER THREE

Paternal effects on the expression of a male polyphenism

! 55! !

56! ! 3.1 Abstract

Polyphenic traits are widespread, but compared to other traits, relatively few studies have explored the mechanisms that influence their inheritance. Here we investigated the relative importance of additive, nonadditive genetic, and parental sources of variation in the expression of polyphenic male dimorphism in the mite Rhizoglyphus echinopus, a species in which males are either fighters or scramblers. We established eight inbred lines through eight generations of full-sibling matings, and then crossed the inbred lines in a partial diallel design. Nymphs were isolated and raised to adulthood with ad libitum food. At adulthood, male morph was recorded for all male offspring. Using a Cockerham-Weir model we found strong paternal effects for this polyphenic trait that could be either linked to the Y chromosome of males or an indirect genetic effect that is environmentally transmitted. In additional analyses we were able to corroborate the paternal effects but also detected significant additive effects questioning the Cockerham-Weir analysis. This study reveals the potential importance of paternal effects on the expression of polyphenic traits and sheds light on the complex genetic architecture of these traits.

! 57! 3.2 Introduction

A great variety of morphological and life history traits vary dichotomously (Moran 1992), including colour (Hazel 2002), shape or presence of morphological structures (see references in Brockmann 2001; Roff 1994; Roff 1996), as well as behaviour (Moczek and Emlen 2000), and seasonal diapause (Mousseau and Roff 1989). Single locus inheritance for dichotomous traits is well understood in the light of evolutionary game theory (Maynard Smith 1982). However, the majority of dichotomous traits are phenotypically plastic and represent a polyphenism, where a single genotype is capable of expressing alternative phenotypes under specific environmental conditions (West-Eberhard 2003). Polyphenic dimorphisms have been modelled as a conditional evolutionarily stable strategy, which can evolve and be maintained when individuals’ phenotypes are decided through conditional decision rules (Dawkins 1980; Gross 1996). Unfortunately, evolutionary game theory models do not account for the complex genetics and inheritance patterns of conditional dimorphisms. The first contribution towards understanding the genetic architecture of conditional dimorphisms from a quantitative genetics perspective was the threshold trait concept introduced by Wright (1934a,b) and subsequently developed by Dempster and Lerner (1950). This concept assumes that dimorphic traits have an underlying continuous variable (named ‘liability’), coupled with a threshold mechanism that generates discontinuity on its phenotypic expression (Falconer 1989). Threshold traits are often sensitive to environmental cues, and the ‘environmentally cued threshold’ model (henceforth ET) is the current framework for understanding the evolution of conditional strategies (Hazel et al. 1990; Hazel et al. 2004; Roff 1994). This model considers that the sensitivity to the environmental cue is in itself a polygenic quantitative trait with normally distributed variation (Tomkins and Hazel 2007), reinforcing the importance of genetic variation for the expression of conditional traits. An increasingly large number of studies have demonstrated the evolutionary relevance of gene-by-environment interactions across distantly related taxa (references in Pigliucci 2005). This collection of empirical studies makes it clear that genetic variation for plastic traits is a very general pattern, and therefore the degree to which polyphenic traits can be conditionally expressed is expected to depend on underlying genetic variation (West-Eberhard 2003). In fact, the origin of polyphenisms has been shown to depend on genetic accommodation, a mechanism that starts with a mutation affecting development and allowing environmental variation to expose genetic variants, ultimately leading to an increased environmental sensitivity of a plastic

58! ! phenotype (Suzuki and Nijhout 2006; Suzuki and Nijhout 2008). This mechanism differs from the more widely accepted notion of genetic assimilation (which ultimately results in the loss of environmental sensitivity due to the canalization of the new phenotype), because genetic accommodation might increase the phenotype’s sensitivity to environmental cues (Suzuki and Nijhout 2006; West-Eberhard 2003). Several illuminating examples of the importance of genetic effects for the expression of polyphenisms come from social insects. For example, it had been long assumed that caste determination in the social Hymenoptera was purely environmentally determined, so that every single female larva was potentially capable of becoming a reproductively mature queen, or a sterile worker (Wheeler 1986). However, recent studies have challenged this assumption, showing that the evidence for different genotypes in a colony being similarly capable of developing into any caste is more apparent than real (Schwander et al. 2010). On the contrary, many cases of caste determination in the social Hymenoptera are affected by a variety of direct and indirect genetic effects (Schwander et al. 2010), including genetic compatibility between queens and their mates (Schwander and Keller 2008). A crucial next step for understanding the genetics of polyphenisms is investigating the polygenic nature of a threshold trait's sensitivity to its environmental cue. We studied this question in the acarid mite Rhizoglyphus echinopus, a species in which males are conditionally fighters or scramblers. Male dimorphism in R. echinopus is environmentally-cued by body size and male density (Radwan 2001; Tomkins et al. 2011), but there is a great deal of overlap in the sizes of males that become either fighters or scramblers, even under the same male density. This overlap suggests that there is either a large genetic variation for the switchpoint that links male morph expression to the body size cue, or alternatively some genotypes in the population are entirely canalised to expressing one of the morphs regardless of body size and male density (Buzatto et al. 2012a; Chapter 2; Lively et al. 2000). In a previous study, we established inbred lines to scrutinize the genetic variation for the switchpoint between male morphs, and we found that inbred lines had heterogeneous levels of conditionality that revealed genetic variation for the switchpoint in the base population (as predicted by the ET model), but also suggested a mixture between canalised and conditional strategists in R. echinopus (Buzatto et al. 2012a, Chapter 2). These results revealed that genes of major effect that canalise morph expression might overlay conditionality in this polyphenic trait, suggesting that the genetic architecture of this threshold trait deserves further attention.

! 59! In the present study, we investigated the genetic architecture of the male polyphenism in R. echinopus, applying a partial-diallel cross-classified design using eight inbred lines. Because our experimental design included reciprocal crosses between some of our inbred lines, we were able to partition the additive, nonadditive genetic, maternal and paternal variances. We found important paternal effects for the expression of dimorphic males in R. echinopus, but weak additive and nonadditive genetic, and no maternal effects. These paternal effects could result from direct sire genotypic effects (Y-linked genes, for example) or indirect genetic sire effects (genetically determined effects of paternal origin that are environmentally transmitted). To the best of our knowledge, this is the first time such a powerful tool from the field of quantitative genetics has been used to examine paternal effects in the genetic architecture of a dimorphic trait that is mainly environmentally cued. Our results shed light on the poorly understood interaction between environmental cues and genetic variation for conditionally expressed traits.

3.3 Methods

Model organism The mites used in this study were derived from a base colony of R. echinopus that we originally collected from an infested organic onion in August 2005. The base colony have been kept in six Petri dishes at 22°C and > 90% humidity for over 50 generations, with a standing adult population of a few thousand individuals (more details in Buzatto et al. 2012a, Chapter 2). R. echinopus present two distinct male morphs coupled to alternative mating behaviours (Radwan 2009). Fighter males possess a very thick and sharply terminated third pair of legs, which these males use to kill rivals and monopolize females. In contrast, scramblers’ legs are all equally thin and without a sharp tip, and scramblers search for unguarded females to mate with (Radwan 1993; Radwan 2009). Male dimorphism in the bulb mite R. echinopus is environmentally-cued by body size and male density (Radwan 2001; Tomkins et al. 2011).

Establishing inbred lines To initiate the inbred lines, we isolated approximately 100 larvae from the base colony and reared them individually with ad libitum food. Using the resulting virgin adults, we paired 40 couples separately. Next, we isolated 20 larvae produced by each of these 40 couples, and reared them to adulthood. Subsequently virgin full siblings were

60! ! paired and their offspring reared in the same way (more details of how we reared individual larvae and paired virgin adults can be found in Buzatto et al. 2012a, Chapter 2). Throughout inbreeding, approximately 14% of full sibling pairings failed to produce any offspring rendering the lines extinct. We successfully repeated the procedure of pairing full siblings and raising their offspring individually in eight inbred lines, for eight generations, leading to an expected inbreeding coefficient of F = 0.826.

Crossing inbred lines in a partial diallel design We paired a virgin adult female to a virgin adult male in all the combinations of inbred lines selected for our partial diallel design (Table 3.1). We attempted to perform each selected cross four times, but the number of crosses that were successful at producing offspring varied from zero to four among the different combinations of inbred lines. Our design generated 28 independent combinations of full-siblings, 4 independent combinations of reciprocal full-siblings, 36 independent combinations of maternal half- siblings, 36 independent combinations of paternal half-siblings, 88 independent combinations of reciprocal half-siblings, and 214 independent combinations of unrelated individuals. We used scramblers as sires when scramblers were more common than fighters in the sire line, and fighters as sires when fighters were more common than scramblers in the sire line. For the lines in which scramblers and sires were similarly common, we randomly used fighters or scramblers as sires in different replicates of the crosses. Successful pairings produced approximately 200 eggs. As soon as these eggs started to hatch, we isolated 50 larvae produced by each pair, and raised them to adulthood individually, as described in Chapter 2 (Buzatto et al. 2012a). At adulthood, we scored all the male offspring for morph. During the whole experiment, we kept all adults and juveniles at 22°C (Binder KB 240 cooled incubator) and > 90% humidity.

Data analysis After scoring the male offspring of all crosses for their morph, we used the Cockerham- Weir model (called 'biomodel' by Cockerham and Weir 1977; Lynch and Walsh 1998) to model the probability of an individual offspring becoming a fighter and estimate the six components of genetic and environmental variance. The equation we used was

Yijkl = µ + morphi + Ni + N j + Tij + M j + Pi + Kij + Rk(ij ) + Wl(k(ij )),

€ ! 61! in which Yijkl is the probability of becoming a fighter for the lth son from the kth replicate of the cross between sire i and dam j, µ is the mean proportion of fighters in the population, and morphi is the fixed factor of sire i morph. The remaining terms are random effects assumed to be mutually independent and normally distributed, and to have mean zero: Ni and Nj are the haploid nuclear contribution (additive effects) from lines i and j, independent of sex; Tij is the nonadditive interaction of the haploid nuclear contributions (including dominance and epistatic effects); Mj is the maternal genetic and environmental effects of line j when used as dams; Pi is the paternal genetic and environmental effects of line i when used as sires; Kij is the sum of all nuclear - extranuclear and extranuclear - extranuclear interactions (i.e., all possible interactions between maternal effects, paternal effects and nuclear effects); Rk(ij) is the effect of kth replicate cross within the sire i x dam j combination; and Wl(k(ij)) is the residual (within replicate cross) effect of individual l (as in Bilde et al. 2008; Dowling et al. 2010; Fry 2004; Ivy 2007). The Cockerham-Weir model also assumes that the variances of additive nuclear effects through sires and dams are the same, and that the reciprocal dominance effects, Tij and Tji are identical. However, the reciprocal effects of Kij and Kji are not necessarily equal, which accounts for the fact that cytoplasmatic elements contributed by sires and dams might differ (Lynch and Walsh 1998). In order to estimate these variances, we fitted the Cockerham-Weir model using the default estimation method in the GLIMMIX procedure in SAS version 9.2 (SAS Institute 2004). We used the TYPE=LIN command to model the covariance between families as linear functions of the variances (see Fry 2004 for details of how these covariances are modelled and their biological interpretations). We also used the COVTEST statement to fit reduced models in which a given covariance parameter was set to 0, and the comparison of these models with the full model (where all parameters were allowed to have positive values) provided statistical inferences about the covariance parameters through likelihood ratio tests. We performed this analysis on a covariance matrix of 406 pairwise comparisons between the 28 families derived from our successful between-line crosses (see Table 3.1). Next, we used the observational variance components (represented by sigma's) to estimate the causal variance components (represented by V's; according to Bilde et al. 2008), always assuming that our inbred lines represent a random sample from the base population, and that epistasis is small:

2 2 (1) σ N : nuclear additive variance, VA = 2σ N /F, where F is the inbreeding coefficient.

2 2 2 (2) σ T : nuclear interaction (dominance) variance, VD = σ T /F .

62! ! 2 (3) σ M : maternal effect variance VM, which can result from maternal genotype or maternal environment effects.

2 (4) σ P : paternal effect variance VP, which can result from paternal genotype or paternal environment effects.

2 (5) σ K : interaction variance VK, which can result from interactions between paternal and maternal effects, and/or interactions between nuclear and extra-nuclear effects.

2 (6) σ R : among replicate crosses variance.

2 (7) σ W : within replicate crosses variance.

2 2 Bilde et al. (2008) noted that σ R and σ W could only directly represent the environmental variance VE if the parental lines were fully inbred (F = 1). Therefore, in order to estimate VE, we first summed all the observational components of variance to obtain the total phenotypic variance (VTOT), and then subtracted all the other causal components of variance from VTOT. All data used in this study are available online in the Dryad repository (doi: 10.5061/dryad.qj8713jm).

3.4 Results

Single locus inheritance The proportion of fighters in the offspring of all crosses ranged from 0% to 76.2% (mean ± standard deviation = 26.3 ± 19.8 %, n = 79 crosses, Figure 3.1). In order to investigate the possibility of single-locus inheritance, we examined graphically the morph ratios of offspring sired by fighters and scramblers (as in Tomkins et al. 2004). If the sire phenotype is dominant (with possible genotypes AA and Aa), the male offspring he produces when mated to a female of unknown genotype (AA, Aa or aa) should express the same phenotype of their father with expected ratios of 1:0, 3:1 or 1:1. On the other hand, if the sire phenotype is recessive (necessarily aa), the male offspring he produces when mated to a female of unknown genotype (AA, Aa or aa) should express the same phenotype of their father with expected ratios of 0:1, 1:1 or 1:0. The hypothesis that male morph possess single-locus inheritance in R. echinopus can clearly be rejected by figure 3.2, in which the offspring of scrambler and fighter males are plotted against the expected ratios.

! 63! Table 3.1 The partial diallel design used to cross eight inbred lines and analyse the genetic architecture of morph determination in the bulb mite Rhizoglyphus echinopus. Shaded cells represent crosses with reciprocals (e.g. Sire line 2 x Dam line 6 and Sire line 6 x Dam line 2). Sample sizes are shown for the number of male offspring generated in each cross, and numbers in the same cell (separated with commas) represent replicate crosses. Numbers in bold represent offspring with fighter sires, whereas the remaining numbers represent offspring sired by scramblers. The distribution of missing crosses in our design is fairly homogenous, and these missing cells should not bias the estimation of variance components (as in Roff and Sokolovska 2004).

Dam line

Sire line 1 2 3 4 5 6 7 8 Missing

1 28 34, 40, 8 32, 9, 39 30, 30 3 36, 40, 35, 31, 24, 37, 2 0 4 37, 18 37 33, 36 18, 15, 16, 29, 27, 24, 3 24 3 18 33, 45 31 15, 18, 28, 23, 4 0 21, 17 4 17, 21 23, 3 13, 21, 24, 18, 5 16 4, 18 3 15, 17 19, 19 44, 20, 23, 27, 6 28, 19 24 3 17, 11 21 19, 17, 20, 34, 7 35, 26 0 4 26, 25 30 38, 42, 8 27, 20 0 38, 23 4 35, 35

Missing 5 3 3 3 5 3 3 3

Cockerham-Weir analysis In the Cockerham-Weir model fitted in SAS, the fixed effect of sire morph on offspring morph determination was not significant (P = 0.32). The Cockerham-Weir model also failed to detect any variance in morph determination due to additive genetic effects, maternal genotype and maternal environment effects, or to interactions between paternal and maternal effects, or between nuclear and extra-nuclear effects (Table 3.2). The variance attributable to interactions between nuclear haploid genomes (indicative of dominance or epitasis) was responsible for about 11% of the phenotypic variation in male morph, but was not statistically significant. Paternal effects on male morph in the offspring, on the other hand, were significant, accounting for over 21% of the phenotypic variation in this trait (Table 3.2). We also detected variance among replicate crosses and a large amount of variance among males within-replicate crosses (accounting for 68% of the total variance), which is expected in an environmentally- cued trait.

64! ! Figure 1

Frequency 0.80" of fighters 0.70" 1"

0.60" 2" 0.50" 3" 0.40" 4" 0.30" 0.20" 5" Sire lines 0.10" 6"

0.00" 7" 1" 2" 3" 8" 4" 5" 6" 7" Dam lines 8"

Figure 3.1 The frequency of fighters produced in the offspring of each cross in a partial diallel design in the bulb mite Rhizoglyphus echinopus. Each bar represents the average frequency of fighters in all replicates of that cross. The within-line crosses (the diagonal) was used to illustrate the frequency of fighters in each parental line, but was not included in the partial diallel analysis.

Additive effects Although the Cockerham-Weir Model did not detect significant additive effects, such effects on morph ratio can be revealed with regressions of average offspring values on midparent values (Falconer 1989). Indeed, the mean proportion of fighters in the offspring of each type of cross was positively related to the average morph ratio of the parental lines (R2 = 0.492, df = 22, P-value = 0.0001; Figure 3.3a) and to the average body size of the parental lines (measured as quiescent tritonymph weight [QTW], see Chapter 2 (Buzatto et al. 2012a) for details; R2 = 0.326, df = 22, P-value = 0.0036; Figure 3.3b). We also calculated the heritabilities of male morph using a modified formula of Falconer's proband method that takes into account the unequal variances of liability in the parental and offspring generation (equation 18.3, page 303 in Falconer 1989). This formula assumes that the affected individuals (fighters in our case) are selected as parents in the parental generation, and uses the proportion of fighters in the parental and offspring generation to calculate the heritability in the underlying liability scale. Therefore, we only calculated the heritabilities for the crosses that had fighter sires, and

! 65! we used the average between the morph ratios of parental lines in the formula. We found that mean heritability was 0.81Figure (standard 2 error: 0.08, range: 0.16-1.50, n = 25 crosses). Fighter sire Scrambler sire Fighter sire Scrambler sire 6 Fighter sire 6 Scrambler sire

6 6 5 5 5 5 4 A 4 B 4 4 3 3 3 3 2 2 2 2 1 1 Square root number of fighters 1 1 0 0

Square root number of fighters 0 0 0 1 2 3 4 5 6 0 1 2 3 4 5 6 0 1 2 3 4 5 6 0 1 2 3 4 5 6

6 6 6 6 5 C 5 D

5 5 4 4

4 4 3 3 Square rootSquare of number fighters

3 3 2 2

2 2 1 1 Square root number of fighters 1 1 0 0 Square root number of fighters

0 0 0 1 2 3 4 5 6 0 1 2 3 4 5 6

0 1 2 3 4 5 6 0 1 2 3 4 5 6 Square root number of scramblers

Figure 3.2 Square root numbers of fighters and scramblers in the offspring of fighter (A, C) and scrambler (B, D) sires. Square root transformation was used to facilitate visual interpretations of the ratios in plots in which both axes have the same scale. Dashed lines represent the expected proportions in the offspring under single gene inheritance. If fighter phenotype is dominant and scrambler phenotype is recessive (upper panels A and B), the offspring of a fighter (AA or Aa) should present fighter ratios of 1:0, 3:1 or 1:1 when mated with a female of unknown genotype (AA, Aa or aa), whereas the offspring of a scrambler (aa) should present fighter ratios of 1:0, 1:1 or 0:1. Conversely, if fighter phenotype is recessive and scrambler phenotype is dominant (lower panels C and D), the offspring of a fighter (aa) should present fighter ratios of 0:1, 1:1 or 1:0 when mated with a female of unknown genotype (AA, Aa or aa), whereas the offspring of a scrambler (AA or Aa) should present fighter ratios of 0:1, 1:3 or 1:1. Open circles represent data that are consistent with at least one of the expected ratios, and full circles represent data that are significantly different (P-value < 0.05 in Chi-square tests on untransformed values) to all expected ratios.

Paternal and maternal effects To test for paternal and maternal effects outside the Cockerham-Weir analysis, we grouped the crosses by sire line (across different dam lines) and by dam line (across different sire lines), and looked for correlations between these values and the values of paternal and maternal lines. The average morph ratios of the crosses, when grouped by sire line, were significantly positively correlated with morph ratios in the paternal lines

66! ! (R2 = 0.550, df = 6, P = 0.033; Figure 3.4a), but the average morph ratios of the crosses grouped by dam line were not significantly correlated with morph ratios in the maternal lines (R2 = 0.406, df = 6, P = 0.089; Figure 3.4c). The average body sizes of the paternal lines (again measured as QTW) were not significantly correlated with morph ratio in the offspring when the crosses were grouped by sire line (R2 = 0.294, df = 6, P- value = 0.165; Figure 3.4b). The average body sizes of the maternal lines were also not significantly correlated with morph ratio in the offspring when the crosses were grouped by dam line (R2 = 0.129, df = 6, P-value = 0.383; Figure 3.4d). Finally, the average morph ratios of the crosses grouped by sire line (across different dam lines) varied significantly more than the average morph ratios of the crosses grouped by dam line (across different sire lines; F7,7 = 7.299, P = 0.018; Figure 3.5).

3.5 Discussion

Here we have investigated the mechanisms that influence the inheritance of conditional male dimorphism in the mite R. echinopus. With a classical cross-classified design, the partial diallel (Lynch and Walsh 1998), we attempted to disentangle the relative importance of additive, nonadditive genetic, and parental sources of variation. Below we discuss how our results provide evidence for weak additive genetic variance and strong paternal effects on the expression of this polyphenism.

Additive effects Our first striking result was the lack of significant additive variance in the Cockerham- Weir analysis for morph determination in R. echinopus. This result was unexpected because offspring body size is positively related to the probability of becoming a fighter in juveniles (Tomkins et al. 2011, Buzatto et al. 2012a, Chapter 2) and therefore any additive genetic variance for body size would be indirectly detected by the Cockerham-Weir model as additive variance for morph determination. The evidence suggests that the lack of significant additive effects in the Cockerham-Weir model reflects an underestimation of these effects rather than their complete absence however. First of all, the mean frequency of fighters in the offspring of each cross was positively related to the average morph ratio (Figure 3.3a) and to the average body size of the parental lines (Figure 3.3b). Regressions of average offspring values on midparent values provide a straightforward alternative method to the Cockerham-Weir model for detecting additive variance (Falconer 1989): under purely additive effects offspring trait values should correspond to the average of parental values; departures from this

! 67! relationship indicating environmental effects, dominance, epistasis, or indirect genetic effects. Furthermore, our measures of heritability of male morph (in the underlying liability scale, Faconer 1989) using midparent values returned high values (0.81 on average). It is hence surprising that we detected this additive variance with these simple approaches, but failed to detect it with the Cockerham-Weir model.

Table 3.2 The results from fitting the Cockerham-Weir model to the data obtained with a partial diallel design to investigate the genetic architecture of morph determination in the bulb mite R. echinopus: the observational variance component estimates (with standard errors in parenthesis and P-values obtained with likelihood ratio tests; see Methods), and the causal variance component estimates. Percent represents the proportion of total phenotypic variance explained by each parameter of the Cockerham-Weir model (see Methods).

Observational Causal Variance Estimate Biological component (SE) P-value Variance interpretation Estimate Percent 2 σ N 0 - VA Nuclear Additive 0 0 2 0.1488 Nuclear σ 0.4242 VD 0.2181 10.85 T (0.1303) Dominance 2 σ M 0 - VM Maternal Effects 0 0 2 0.4237 σ 0.0089 VP Paternal Effects 0.4237 21.08 P (0.4015) 2 Interactions σ 0 - VK 0 0 K between the above 2 0.5357 σ <0.001 R (0.1512) VE Environmental 1.3685 68.07 2 0.9021 σ W (0.02921) VTOT 2.0103

The contrasting results from our regressions of average offspring values on midparent values and our Cockerham-Weir analysis probably reflect a conservative estimate of variance components by the latter. This could be a property of the model itself, or because our breeding design was a partial diallel, which generates fewer independent combinations of the different types of siblings than a full diallel and is therefore likely to be less powerful. Nevertheless, the missing cells were designed to be distributed throughout the diallel in a balanced way, such that the only possible bias would be uniform and in a downward direction for all the genetic variance estimates (Dowling et al. 2010; Roff and Sokolovska 2004). Our conclusion is that it is important to recognise the reduced power of the Cockerham-Weir model, and this means that great caution should be exercised when interpreting partial diallel designs in the absence of other analyses. We suggest that alternative methods for estimation of additive effects (such as regressions of average offspring values on midparent values)

68! ! should always be presented alongside the results of a Cockerham-Weir model, such that reliance is not placed solely on this model. Figure 3

0.7 Figure 3.3 The mean frequency A of fighters in the offspring of each cross was positively 0.6 related to the average morph ratio (A) and to the average 0.5 body size (measured as quiescent tritonymph weight

0.4 [QTW]; B) of the parental lines.

MR Dashed lines represent 95%

0.3 confidence intervals around the regression line.

0.2

Mean frequency of frequency Mean fighters

0.1

0.0

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Average frequencyMidMR of fighters of the parental lines

0.7 B

0.6

0.5

0.4 MR 0.3

0.2

Mean frequency of frequency Mean fighters 0.1

0.0 0.046 0.048 0.050 0.052 0.054 0.056 0.058 Average QTW ofMidQTW the parental lines (mg)

In conclusion, clearly there is evidence for the existence of some additive genetic variance for morph determination in R. echinopus, coming from our regressions of offspring morph ratios on midparent morph ratio and midparent body size, even though the Cockerham-Weir analysis failed to detect it. However, caution must be exercised when interpreting the magnitude of such additive effects, as offspring on midparent regressions are not capable of disentangling extra-nuclear effects from additive nuclear effects. ! 69! Figure 4

0.40 A 0.40 B 0.40 0.40 0.30 0.30 0.30 0.30 Grouped by SIRE Line 0.20 0.20 Grouped by SIRE Line 0.20 0.20 0.10 0.10 Frequency ofFrequency fighters (grouped by sire line) (grouped by sire line) 0.0 0.2 0.4 0.6 0.8 0.040 0.045 0.050 0.055 0.060 0.065 0.10 0.10

0.0 0.2 0.4 0.6 0.8 0.040 0.045 0.050 0.055 0.060 0.065

C D 0.40 0.40 0.40 0.40 0.30 0.30 0.30 0.30 Grouped by DAM Line 0.20 0.20 Grouped by DAM Line 0.20 0.20 Frequency ofFrequency fighters (grouped by dam line) (grouped by dam line) 0.10 0.10

0.0 0.2 0.4 0.6 0.8 0.040 0.045 0.050 0.055 0.060 0.065 0.10 0.10 Morph Ratio in parental line Average QTW of parental line 0.0 Morph0.2 ratio0.4 in 0.6the 0.8 0.040 Average0.045 0.050 QTW0.055 in0.060 the 0.065 Morphparental Ratio in parental line line parentalAverage QTW lineof parental (mg) line

Figure 3.4 The average morph ratios of the crosses grouped by sire line were positively correlated with morph ratios in the paternal lines (A; dashed lines represent 95% confidence intervals around the regression line), but not to the average body sizes of the paternal lines (measured as quiescent tritonymph weight [QTW]; B). The average morph ratios of the crosses grouped by dam line were not correlated either with the morph ratios in the maternal lines (C), or with the average body sizes of the maternal lines (D).

Paternal effects or sire morph effects? Even though the lack of significant additive effects indicated by the Cockerham-Weir model reflect an underestimation, it is important to note that these additive effects are probably weaker than the paternal effects detected in the same model as the strongest source of heritable variation for morph determination in R. echinopus. These results are consistent with the fact that we detected higher variation between the frequency of fighters produced by crosses grouped by sire line than between crosses grouped by dam line, which suggests that sire line predicts fighters’ frequency in the offspring better than does dam line. Likewise, we found that the average morph ratios of the crosses grouped by sire line were positively correlated with morph ratios in the paternal lines, whereas the average morph ratios of the crosses grouped by dam line were not correlated with morph ratios in the maternal lines. We interpret these results

70! ! as evidence that sire line effects are stronger than dam line effects or additive effects, pointing to the importance of the same paternal effects that we detected with the Cockerham-Weir model. Figure 5

A B 0.7 0.7 0.6 0.6 0.5 0.5 0.4 0.4 Morph ratio 0.3 0.3 0.2 0.2 Frequency ofFrequency fighters 0.1 0.1 0.0 0.0

41 23 53 46 51 68 72 87 13 24 31 42 58 65 77 86 Sire Line Dam Line Sire lines Dam lines

Figure 3.5 The frequency of fighters in the offspring produced by the crosses between inbred lines of the bulb mite Rhizoglyphus echinopus. Data were grouped by sire line (across different dam lines; A) or by dam line (across different sire lines; B). Averages and standard deviations are shown, and lines were ordered according to these averages for ease of visual inspection. Variation was significantly higher between crosses grouped by sire line than between crosses grouped by dam line (see Results), indicating that the former was a better predictor of fighters’ frequency in the offspring than the latter.

Our approach disentangles the paternal effects of sire morph from other paternal effects (such as body size, for instance) that are environmentally or genetically transmitted through the sire line. In a previous study on the closely related species R. robini (Smallegange 2011), such effects would have been confounded with sire morph effects, given that no sire traits (such as body size, for instance) were measured in that study. Therefore, any sire trait that correlates with sire morph would have been detected as a sire morph effect by Smallegange (2011). Even in the present study, if we analyse the offspring of all crosses together regardless of parental line, a t-test on arcsine square root transformed proportions in the offspring indicates that the proportion of fighters was significantly lower in offspring sired by scramblers than in offspring sired by fighters (analysis not shown). However, using the same data set and properly accounting for the genotypes of sires and dams with the Cockerham-Weir model fitted in SAS, we found no significant effect of sire morph on offspring morph determination. In that regard, our approach provided a very powerful way to conclude that sire morph is not in itself an important factor in determining offspring morph in R.

! 71! echinopus, and perhaps generally in bulb mites. It is possible that sire morph effects detected in previous studies with male-dimorphic mites could actually represent other sorts of paternal effects that might correlate with sire morph such as size.

Possible sources of paternal effects There are numerous possible explanations for the paternal effects that are clearly operating on morph determination in R. echinopus. The first potential explanation is that a major gene that influences the expression of male morph resides on the Y chromosome. Since R. echinopus is known to possess XY sex determination (Grondziel 1975), genetic variation in the Y chromosome among our inbred lines would be evident when analysing the offspring of different sire lines (across dam lines, see Figure 3.5a), but not the offspring of different dam lines (across sire lines, see Figure 3.5b), which is consistent with our results. The moving of alleles (or their regulatory elements) that are under sexually antagonistic selection to the hemizygous sex chromosome (the male Y in R. echinopus) has been seen as one of the ways to resolve intra-locus sexual conflict (Stewart et al. 2010). Investment in fighting legs would likely be costly for female R. echinopus, in terms of resources diverted away from general body size and fecundity. Indeed, the fighter phenotype in R. echinopus is sex-limited, as females never express a thick and sharply terminated third pair of legs and never fight. It is therefore possible that the morphology of the third pair of legs in this species went through a period of intra-locus sexual conflict in its evolutionary history, and this conflict was eventually resolved with the genes coding for fighter legs being moved to the Y chromosome. Similar paternal effects due to genes linked to the sex chromosomes have been detected for the wing dimorphism in the cricket Gryllus rubens (where males are XO, and females XX; Walker 1987, but see Zera and Rankin 1989). Nevertheless, paternal effects can also result from genes that reside in the X- chromosome of sires. Even though this possibility sounds counter intuitive at first, a recent study of Drosophila melanogaster found that variation on the X chromosome of sires affected the fitness of their male offspring, demonstrating that non-transmitted paternal chromosomes can influence the phenotype of their offspring (Friberg et al. 2012). In fact, epigenetic reprogramming of DNA in any chromosome carried by sperm cells would also generate the same kind of results. This has been suggested for Drosophila simulans, in which paternal photoperiodic conditions influence the development times of their offspring (Giesel 1986). Likewise, the temperature in which sires of Drosophila melanogaster are raised affects wing length in the offspring (Crill et al. 1996), and the host-plant species in which sires of the seed beetle Stator limbatus are

72! ! reared influences the relationship of host-plant species and development time in the offspring (Fox et al. 1995). In our present experiment, however, we mitigated effects of paternal environment, as the environmental conditions under which sires were reared were completely controlled (standardized temperature, light cycle, population density [every mite was reared in isolation] and diet [ad libitum yeast provided]). Therefore, the only form of epigenetics that could be responsible for the paternal effects detected in the present study is genomic imprinting, where the expression of an allele differs depending upon its parent of origin (Day and Bonduriansky 2004). Genetically determined effects of paternal origin that are environmentally transmitted represent another possible explanation for the paternal effects on morph determination in R. echinopus. In the Australian field cricket, for instance, paternal effects on offspring phenotype result from the large ejaculates or accessory gland secretions that provide resources to the female (Garcia-Gonzalez and Simmons 2007). Accessory gland secretions are also shown to enhance offspring fitness in D. melanogaster (Priest et al. 2008). As long as there is genetic variation for such accessory gland secretions among sires, this variation would generate indirect genetic effects on the offspring phenotype that would be detected as paternal effects by the Cockerham- Weir model. In bulb mites, a previous study with Rhizoglyphus robini revealed that the fecundity of females is associated with a polymorphism in the phosphogluconate dehydrogenase (Pgdh) genotype of their mates (Konior et al. 2006). If this effect of sire genotype on dam fecundity also influences other reproductive traits of females, such as egg size for instance, genetic variation for the Pgdh genotype could also generate a paternal effect on morph determination because egg size is connected to morph expression in the offspring (Smallegange 2011 and see discussion on maternal effects below).

Maternal effects At first glance, the fact that we did not detect maternal effects on morph contrasts greatly with a recent study on the congeneric species R. robini, in which there were significant maternal effects through egg size on morph determination (Smallegange 2011). However, Smallegange's (2011) results might align with ours, as the maternal effects detected by Smallegange (2011) would probably be detected as paternal effects in a diallel approach such as the one we used. The reason being that Smallegange's (2011) study also found a sire effect on the size of eggs produced by females. If a female's mate influences the size of her eggs, paternal effects could actually be the ultimate determining factor of egg size, as sires would actually be driving the

! 73! differential allocation of resources into eggs by dams (Burley 1986). In dung-beetles, for example, the size of male influences the sizes of the brood-balls produced by their mates, which ultimately affects offspring phenotype (Kotiaho et al. 2003). Similar results have been reported for butterflies (Wedell 1996), birds (Cunningham and Russell 2000; Gil et al. 1999), and fish (Kolm 2001), all indicating that males affect female resource allocation to the offspring, and pointing towards maternal effects that are driven by paternal effects. The egg size effect on offspring phenotype detected in bulb mites by Smallegange (2011) was only significant under poor dietary conditions. Given that all our mites developed on a rich yeast diet, the absence of maternal effects in our study could also come down to the fact that this egg size effect was in fact not relevant in our experimental conditions. Nevertheless we still think that the results from Smallegange's (2011) study on R. robini could shed light on the proximate mechanism through which the paternal effect that we detected operates. Investigating paternal effects on egg size in R. echinopus is hence a promising next step to understand the complex gene by environment interactions that underlie the expression of male polyphenism in this species.

Implications for the ET model By definition, polyphenisms are environmentally sensitive, and the ET model treats the sensitivity to the environmental cue as a heritable quantitative trait (Hazel et al. 1990; Tomkins and Hazel 2007). Therefore, the ET model approach is capable of identifying the value of the environmental cue that represents the mean switchpoint to which a population will evolve, which is the mean switchpoint at which the overall selection differential is zero (see Box 3 in Tomkins and Hazel 2007). Because the response to selection is a product of the selection differential and the heritability of a trait (the "breeder's equation", Lynch and Walsh 1998) the efficiency with which selection will move the switchpoint mean of a population to its equilibrium is directly proportional to the heritability of switchpoint. In R. echinopus, male dimorphism is environmentally-cued by body size and male density (Radwan 2001; Tomkins et al. 2011), and in a previous study we suggested that there is large genetic variation for the switchpoint that links male morph expression to the body size cue (Buzatto et al. 2012a, Chapter 2). However, the heritability of switchpoint variation is constrained by the fact that male dimorphism is sex-limited in its expression: genes that affect the switchpoint controlling male dimorphism are not expressed in females. The paternal effects on male dimorphism that we have detected

74! ! might mean that sons inherit the switchpoint characteristics of their father rather than their mother; this may be adaptive since it is their father’s switchpoints that are under selection while their mother’s switchpoints (if they have them) are hidden from selection (the storage effect; Reinhold 1999). This hypothesis for an adaptive shift to paternally biased inheritance could provide an explanation for the evolution of paternal effects on the expression of male polyphenism in other taxa. A different explanation might be that alleles determining the sensitivity to environmental cues have pleiotropic effects on one or more other traits that are under sexually antagonistic selection, favouring their expression on a male limited chromosome (Stewart et al. 2010). Our data alongside those of Smallegange (2011) and Kotiaho et al. (2003) suggest that parental effects are an important aspect of the phenotypic and genetic architecture of polyphenisms and worthy of further investigation.

! 75!

76! !

CHAPTER FOUR

Maternal effects on male weaponry: female dung beetles produce major sons with longer horns when they perceive higher population density

! 77! !

78! ! 4.1 Abstract

Maternal effects are environmental influences on the phenotype of one individual that are due to the expression of genes in its mother, and are expected to evolve whenever females are better capable of assessing the environmental conditions that their offspring will experience than the offspring themselves. In the dung beetle Onthophagus taurus, conditional male dimorphism is associated with alternative reproductive tactics: majors fight and guard females whereas minors sneak copulations. Furthermore, variation in dung beetle population density has different fitness consequences for each male morph, and theory predicts that higher population density might select for a higher frequency of minors and/or greater expenditure on weaponry in majors. Because adult dung beetles provide offspring with all the nutritional resources for their development, maternal effects strongly influence male phenotype. Here we tested whether female O. taurus are capable of perceiving population density, and responding by changing the phenotype of their offspring. We found that mothers who were reared with other conspecifics in their pre-mating period produced major offspring that had longer horns across a wider range of body sizes than the major offspring of females that were reared in isolation in their pre-mating period. Moreover, our results indicate that this maternal effect on male weaponry does not operate through the amount of dung provided by females to their offspring, but is rather transmitted through egg or brood mass composition. Finally, although theory predicts that females experiencing higher density might produce more minor males, we found no support for this, rather the best fitting models were equivocal as to whether fewer or the same proportions of minors were produced. Our study describes a new type of maternal effect in dung beetles, which probably allows females to respond to population density adaptively, preparing at least their major offspring for the sexual competition they will face in the future. This new type of maternal effect in dung beetles represents a novel transgenerational response of alternative reproductive tactics to population density.

! 79! 4.2 Introduction

Traditionally, maternal effects were understood to occur whenever the mother’s phenotype directly affects the phenotype of their offspring, regardless of the female’s genetic contributions to her offspring (Bernardo 1996). A more recent definition states that maternal effects are environmental influences on the phenotype of one individual that are due to the expression of genes in its mother (Wolf et al. 1998). Therefore, maternal effects represent developmental influences extended across life cycle stages, in which genetic or environmental differences in the maternal generation are expressed as phenotypic differences in the offspring (Mousseau and Dingle 1991). Maternal effects are known to act on a multitude of offspring traits, such as body size, condition, and even expression of sexually selected traits (Mousseau and Fox 1998; Mousseau et al. 2009; Qvarnstrom and Price 2001). Regarding dimorphic traits, maternal effects seem to be especially important for the expression of the “life cycle polymorphisms” (sensu Roff 1996) of insects, such as: sexual versus parthenogenic morphs in aphids, winged versus wingless morphs in the red linden bug, summer versus winter morphs in the collembolan Orchesella cincta, as well as dispersal versus sedentary morphs in the cowpea seed beetle and in several species of locusts (Mousseau and Dingle 1991). Maternal effects also control the expression of what Roff (1996) called “protective polymorphisms” in species of the crustacean genus Daphnia. An elongated structure that grows from behind the head of these animals is a threshold-trait that functions as an anti-predator defense (Parejko and Dodson 1991), and its expression is influenced by maternal effects (Tollrian 1995). Male dimorphism (Eberhard and Gutierrez 1991) usually reflects alternative reproductive tactics among males: the large (‘major’) male morph typically has more elaborate weaponry, and guards females or reproductive territories, while the small (‘minor’) male morph has reduced weaponry and sneaks copulations (Gross 1996). Male dimorphism is now known to be taxonomically widespread (Oliveira et al. 2008), having been described in taxa as distinct as molluscs (Hall and Hanlon 2002; Hanlon et al. 2005; Iwata and Sakurai 2007; Norman et al. 1999), nematodes (Ainsworth 1990), and predominantly vertebrates and arthropods (reviewed by Oliveira et al. 2008). Most cases of male dimorphism seem to be due to alternative tactics within a conditional strategy, in which the phenotype expressed by individuals is influenced by their status (Gross 1996; Hazel et al. 1990; Tomkins and Hazel 2007; but see Shuster and Wade 2003). According to the conditional strategy individual fitness varies as a function of competitive ability (i.e., status Gross 1996), for the alternative tactics to be

80! ! evolutionarily stable in the population the rate of change in fitness with status needs to differ between the alternative tactics such that the fitness functions intersect. The exact point at which selection favors a switch from one tactic to another can be calculated from knowledge of the fitness functions, the distribution of switchpoints and the variation in (in this case) status (Hazel et al. 1990; Tomkins and Hazel 2007). As a result, males with a status higher than such a switchpoint benefit from adopting the major phenotype (the primary tactic), whereas males with status lower than the switchpoint benefit from adopting the minor phenotype (the alternative tactic, Gross 1996; Hazel et al. 1990; Tomkins and Hazel 2007). In theory, the functions that relate the fitness of each male phenotype to individual body size (the most common proxy to status) may be influenced by demographic features, such as population density (Knell 2009). If only one of the male mating tactics involves fighting, for instance, it is expected that the relative costs of this tactic (compared to the alternative) will increase with population density, as encountering rival males becomes more likely. This is illustrated by male dimorphic bulb mites and soil mites, in which the average fitness of males from the fighter morph decreases with increasing numbers of rival males (Radwan 1993), and the probability that nymphs develop into the fighter morph diminishes with increasing population density (Radwan 1995). During development, nymphs of these species are capable of assessing population density through a colony pheromone, and this information seems to influence the switchpoint that nymphs must reach in order to express the fighter phenotype as adults (Radwan et al. 2002; Tomkins et al. 2004). In male dimorphic insects, population density also seems to alter the selective pressure on each morph, driving the evolution of male dimorphic horns in dung beetles (Moczek 2003; Moczek and Nijhout 2003), and male dimorphic forceps in earwigs (Tomkins and Brown 2004). The only study that has directly assessed the shape of the functions that describe how fitness varies with body size for alternative male phenotypes was conducted with the dung beetle Onthophagus taurus (Hunt and Simmons 2001), and corroborated the status-dependent selection model (Gross 1996). In this same species, other studies have suggested that an increase in the density of competing males results in a scenario in which the greater probability of encountering rival males would only allow the largest males to benefit from engaging in fights (Moczek 2003; Moczek et al. 2002). With this rationale comes the prediction that high population density would make the alternative tactic (sneaking by minor males) advantageous over a wider range of body sizes, moving the switchpoint for horn production to larger body sizes (Moczek 2003; Moczek and Nijhout 2003). Correlational data on population density and horn

! 81! allometry across several populations of O. taurus have supported this prediction (Moczek 2003; Moczek et al. 2002). A comparative analysis of 14 species of dung beetle also suggests that horns are less likely to evolve in species with greater mean crowding Pomfret and Knell 2008. However, another prediction that naturally derives from this rationale is that, among majors, selection should favor steeper horn allometries under high densities. This is because the increased probability of encountering rivals under higher population densities should cause selection on long horns to be stronger, as horn length is known to be an important predictor of fight outcome (Moczek and Emlen 2000). Importantly, this new prediction, and the traditional prediction regarding the influence of population density on the evolution of switchpoints for horn production are not mutually excluding. In other words, higher population densities could result both in more minor males (due to larger switchpoint body sizes for horn production) and a steeper horn allometry among major males. Thus, the development of juveniles in male dimorphic species should be sensitive to cues of the population density that will be faced by the adult. However, juveniles from holometabolous insects commonly develop in environments that are completely different from the environment in which they will face adult competition (Chapman 1982). In fact, Onthophagus taurus larvae develop completely isolated from conspecifics inside enclosed brood masses, and are therefore probably unable to assess cues related to adult population density. On the other hand, parental individuals are in a position to perceive the environmental factors that will influence the future fitness of their developing offspring, and may adjust their provisioning decisions accordingly. Indeed, parental dung beetles provide offspring with all resources required for their development, which is in turn the primary determinant of adult phenotype, making male morph largely determined by parental provisioning decisions (Hunt and Simmons 2000). As a result, female O. taurus could in theory be capable of perceiving population density and when the population density is high, responding adaptively by producing more minor males in their offspring and/or producing major offspring with longer horns. Here we tested this novel prediction by experimentally manipulating the social environment and the number of potential mates experienced by females of the dung beetle O. taurus, and later examining the effects of these maternal treatments on the phenotype of male offspring. The manipulation of social environment consisted of rearing females in isolation or with other females, which should affect the females’ perception of population density. The rationale behind manipulating the number of potential mates (in addition to social environment) was that females could theoretically

82! ! assess the intensity of male-male competition that will be faced by their offspring through their perceived population density and/or through the number of males that they encountered and mated with. Our approach consisted of a laboratory experiment that standardized all other environmental factors (light cycle, temperature, humidity), allowing us to test directly the influence of perceived population density (social environment) and number of potential mates on the expression of alternative male tactics in the next generation. Moreover, we analyzed the response of females on two levels, namely the amount of dung they provided to their offspring, and the body size and horn allometry of their male offspring.

4.3 Methods

Collecting and rearing dung beetles In January 2010, we collected adult Onthophagus taurus from cattle pastures in Margaret River, Western Australia. We maintained these individuals in the laboratory in single- sex populations for two weeks, with access to fresh cow dung ad libitum, in order to ensure that they were sexually mature. We then established male-female pairs in approximately 200 separate breeding chambers (PVC piping of 25 cm in length and 6 cm in diameter) that were three-quarters filled with moist sand and 250 ml of cow dung. After one week, we sieved these chambers to retrieve the brood masses, which were buried in moist sand and incubated at 25ºC until adults emerged. This is a well established protocol for breeding O. taurus (Hunt and Simmons 2000; Hunt and Simmons 2001); we repeated the procedure for two generations.

Experimental groups in the pre-mating period After emergence, second-generation offspring from the field-caught beetles were maintained in single-sex populations for two weeks, which corresponds to the pre- mating feeding period. We established these pre-mating populations in the same type of PVC piping used as breeding chambers, which were again three-quarters filled with moist sand and with dung provided ad libitum. Male chambers housed 20 males each (males were always reared in the same density), 12 of the minor morph and 8 of the major morph, which reflects the proportion (0.60) of minor males found in natural populations of O. taurus in Western Australia (Simmons et al. 1999). Meanwhile, female chambers were designated to one of two pre-mating experimental treatments that differed in social environment, and thus in the females’ perception of population density: the low-density group (L) housed only one female per chamber (n = 60

! 83! females); the high-density group (H) housed 20 females per chamber (n = 80 females in four different chambers). Natural populations of O. taurus vary in density from one to about 1,000 beetles per kilogram of dung (Moczek 2003). Therefore, the densities we used here are well within the range of conditions to which the beetles might be naturally exposed.

Experimental groups in the mating period After the pre-mating period, sexually mature males, and females from both density groups, were randomly assigned to one of two mating experimental groups: in the singly-mated group (S), one minor male was paired to one female in a breeding chamber (n = 40 chambers); in the multiply-mated group (M), ten minor males were paired to ten females in each breeding chamber (n = 8 chambers). Therefore, sex ratio in all mating groups was 1:1. Only minor males were used in this stage of the experiment, because it has been shown that females of O. taurus provide more resources to their offspring when mated with males with large-horns (Kotiaho et al. 2003), and we wanted to mitigate any paternal effects in our experiment, and focus only on the response of females to population density and number of mates. Males and females in these groups were kept together for five days, which is more than enough time for females to mate multiple times with either one minor male (S group), or potentially with ten minor males (M group). At this stage, pairs were not allowed to start building brood masses, as they were only provided with approximately 5 ml of cow dung to feed on, which is not enough to build brood masses. Females from the M group chambers in which more than one female died during the experiment were not used in the following stage of the experiment.

Brood-provisioning period After the mating period, we established females from all experimental groups individually in breeding chambers (as described above) with 250 ml of cow dung, in order to allow them to build brood masses and lay eggs. No males were included at this stage of the experiment, as we wanted females to provide their offspring alone, avoiding that any response of the females to the treatments were confounded with variation in males’ assistance to provisioning (Hunt and Simmons 2000). The sample size for this stage of the experiment was 20 females for each combination of experimental groups: high-density and singly-mated (HS); high-density and multiply- mated (HM); low-density and singly-mated (LS); and low-density and multiply-mated (LM). We did this by randomly selecting five out of ten females from each successful M

84! ! group chamber (where no more than one female died) and using all females from the S group (all of them survived and produced offspring). We weighed (to 0.01 g, using a top pan balance) and measured (to 0.01 mm, using a digital caliper) these females for pronotum width just before establishing them in the breeding chambers, and checked that females from the different combinations of experimental groups (HS, HM, LS or

LM) did not differ in either weight (F3, 76 = 1.201; P = 0.315, n = 80 females) or pronotum width (F3, 76 = 0.657; P = 0.581, n = 80 females).

Assaying female responses to the treatments After one week, we sieved breeding chambers and retrieved the brood masses produced by females of all experimental groups. We weighed each brood mass (to 0.01 g) using a top pan balance, buried them in moist sand inside individual plastic boxes (7 x 7 cm base, 5 cm height), and kept them at 25ºC until emergence. We scored newly emerged adults for sex, and then froze them. For all male offspring, we next measured pronotum width (to 0.01 mm, using a digital caliper) and horn length. In order to measure horn length, we detached a beetle's head and mounted it on a pin under a Leica MZ75 dissecting microscope. We captured images on a Leica IC D digital camera and analyzed images using the free software Image J (Rasband 1997-2011). On each image, we measured horn length starting from the middle of the horn tip, following the horn curve until reaching the horizontal line level with the lowest point at the top of the head (measurement 4 in Figure 1 of Tomkins et al. 2006).

Statistical analyses In order to investigate whether females responded to the treatments by adjusting the amount of dung they provided to each offspring, we evaluated the effects of pre- mating group (L or H), mating group (S or M), female body weight, and all possible interactions on the weight of brood masses produced. We used the library 'nlme' (Bates et al. 2011) in R (version 2.14.0) to build linear mixed-effects models, considering the weight of brood masses produced as a response variable and female's body weight as a covariate. As a first step, we built three models with different sets of random effects: (1) no random effects; (2) female identity (to account for the fact that brood masses were not independent if they were built by the same female); and (3) the identity of the replicate mating chamber (to account for the fact that each female from the multiply- mated experimental group shared the same mating chamber with other females). In this third model, female identity was nested within the identity of the replicate mating chamber. We compared these models with log-likelihood tests, and determined that

! 85! only female identity was a significant random effect, which we retained for the following step of the analysis. We used the same approach to select among different variance structures in our models, in order to fulfill the assumption of homogeneous variance of residuals across the predicted values of the models. In the second step of the analysis, we started with a null model and then added the fixed effects (pre-mating group, mating group, female body weight, and all possible interaction terms) sequentially, performing a log-likelihood ratio test at each step, to allow a quantification of the influence of each parameter. Importantly, because we were comparing models with different structures of fixed effects, we fitted the models by maximizing their log-likelihood (using the argument: method = "ML" in the function lme). Next, we repeated this analysis, but only considering the observations of brood masses from which adult offspring emerged successfully. These two analyses could return different results if the brood masses from which no offspring emerged were actually unfinished by the females. Similarly, we repeated this analysis again, but only considering the observations of brood mass weight for brood masses from which male offspring emerged. This analysis should only differ from the previous ones if brood mass provisioning by females was responding to the treatments in different ways, depending on the sex of the offspring being produced. We used the same approach to analyze the pronotum width of the male offspring from our experimental groups. However, this time we added to the models the weight of the brood mass (instead of the female's body weight) and all interaction terms with the other predictor variables. Moreover, here, the random effect of the replicate mating chamber was important, so all models had this random effect in addition to the random effect of female identity (nested within the identity of the replicate mating chamber). Regarding the effects of the treatments on the horn length of the offspring, we followed the steps proposed by (Knell 2009) for the analysis of non-linear allometries. Firstly, we pooled the offspring from all experimental groups, plotted a log-log scatterplot of horn length on pronotum width in which ∆X = ∆Y (see Tomkins and Moczek 2009 for why this is important), and concluded that there was a clear continuous relationship that was not a straight line. Importantly, we added one to the natural values of horn length before log transformation, in order to avoid zeros (due to no horns in the smallest minor males) that would result in minus infinity after transformation. Next, we selected sigmoidal nonlinear models that are appropriate to explain biological growth curves, and used the function 'nls' and the library 'grofit' (Kahm et al. 2010) in R (version 2.14.0) to fit these models (three-parameter logistic,

86! ! four-parameter logistic, Weibull growth curve, and Richards’ growth function; see Chapter 20 in Crawley 2007) and a simple linear regression to the data. Finally, we compared all these models on the basis of the Akaike information criterion (AIC), calculated as [2 × (– log likelihood) + 2 × (the number of parameters in the model)]. After finding the nonlinear model that best fitted the horn allometry of O. taurus, we again used log-likelihood tests to select among different variance structures that would fulfill the assumption of homogeneous variance of residuals across the predicted values of the model. We then built a set of models in which one (or all) of the parameters of the nonlinear model were allowed to have different values for each pre- mating or mating experimental groups, and used log-likelihood tests to compare these models with a model in which the parameters had a single value for all experimental groups. This approach allowed us to infer if the horns of the offspring from the different experimental groups differed in any of the parameters that describe their sigmoidal allometry. An advantage of this approach is that some of the parameters tested have specific and straightforward biological interpretations (see Discussion).

4.4 Results

During the provisioning period of one week, females produced on average 19.2 (SD = 3.2, range = 7 - 27, n = 80 females) brood masses each, and the combined weight of brood masses produced by each female averaged 68.95 mg (SD = 14.77 mg, range 24.58 - 99.19 mg, n = 1,537 brood masses). The body weight of females positively affected the weight of individual brood masses they produced (Figure 4.1a), but there was no effect of pre-mating group (L or H), mating group (S or M) nor any interaction involving these variables on the weight of brood masses (Table 4.1). The results of this analysis do not change significantly if we only consider the observations of brood masses from which male offspring or any offspring emerged successfully (analyses not shown). The weight of the brood masses affected the pronotum widths of the male offspring that emerged from them (Figure 4.1b), but there was again no effect of pre-mating group (L or H), mating group (S or M) or any interaction involving these variables on this trait (Table 4.2).

! 87! ● 6 6.0

A ● B ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ●●● ● ● ● ● ● ●● ● ● ● ●●● ● ● ● ●● ● ● ● ●●●●● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ●● ● ●●● ●● ●● 5.5 ● ● ● ●● ● ● ● ● ● ● ● ●●●●● ●●● ● ● ● ● 5 ● ● ●● ●● ●● ● ● ●● ● ●● ● ●● ●●●● ● ●● ● ● ● ● ●● ● ● ●●● ●●●●● ● ●●● ●● ●● ● ● ● ●● ● ●● ● ● ● ● ●●●● ●● ●●● ●● ● ● ● ● ● ●● ● ● ● ● ● ●●●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ●●●● ● ● ● ● ●●● ● ● ●●●●●● ●●●● ●●●●● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ●● ●● ● ●● ●●● ●●●● ●● ●●●●●●● ● ● ● ●● ● ●● ● ●● ● ●● ● ● ● ● ●●● ● ●● ●● ● ●● ● ● ● ● ●●●●●● ●● ● ●● ● ● ●● ●●● ● ● ● ● ●● ●●● ● ● ● ● ● ●● ●● ●●● ● ● ● ●● ● ● ● ●●●●●●●●●● ● ●●● ●●●●● ●● ● ● ● ●● ● ●● ●● ● ●●● ● ● ● ●● ● ●●●●● ●●● ●● ● ● ● ●●● ● ●● ●●●● ● ●● ● ● ● ●●● ●●●●●● ●● ● ● ●● ●● ● ● ● ● ● ● ● ●● ●● ●●●● ●●● ● ● ●● ● ●●●●● ●● ● ●●● ●● ● ● ● ●● ●● ●●● ●● ● ● ● ●● ● ●●● ● ● ●●● ● ● ● ● ● ●● ●● ●●●● ●● ● ●●● ● ● ●● ●● ●●● ● ● ●● ● ●● ● ●●●● ●●●●●● ●

● ● ● ● 5.0 ● ●●● ● ● ● ●●● ●● ●●● ● ● ● ● ● ● ●● ●●●● ●●● ● ●● ●● ●●● ●●● ●● ●●●●● ● ● ●●●● ● ● ●● ●● ● ● ●●●● ●●● ●● ●● ●● ●● ●● ● ● ●●● ●● ●● ●● ●●● ● ● ● ● ● ● ● ● ●●● ●●●●● ● ●● ●●● ●●●● ● ● ● ●● ● ● ● ●● ●●●● ●●

4 ● ●● ● ● ● ● ●●●●●●●●●●● ●● ●●●●●● ● ● ●●● ● ● ●● ● ●● ● ● ● ●●●● ●●●●●●● ●●● ●● ●● ● ● ● ●● ●● ● ● ● ● ●●●● ● ●●●●● ●●●●●●● ●● ●●● ● ●● ● ●●● ● ●● ●●● ●●●●●●●●● ●● ●● ● ● ● ●● ● ● ● ●●● ● ●●● ●●●● ● ●●●● ● ● ●● ● ● ●●● ●● ●●●●● ●●● ●● ● ● ●● ● ● ● ● ● ●●●●●● ●●●●●●● ● ●●● ●●●● ● ● ● ● ● ● ●● ●●●● ● ●●●●● ●● ●● ●● ●●●●● ●● ● ● ● ● ● ●●●● ●●●●●●●●● ● ●●●● ●● ● ● ●● ● ● ● ●●●●● ● ●●●●● ●●●● ●●● ●●● ● ● ● ● ● ●●●●●●● ●● ● ●●● ●●●● ● ●● ● ● ● ● ● ●●●●●● ● ●● ●●●●●●●● ●●● ●●●● ● ● ●● ●●●● ●●●●●● ●● ●● ●●●●● ●● ● ● ● ● ● ● ● ●●●● ●● ●●●●●●●●●● ●● ●●● ●●● ● ● ● ● ● ● ● ●●● ●●●●● ●●●●●●● ●● ● ● ● ● ● ● ●●●●● ●●●●●●●●● ●●● ●● ●● ● ● ● ● ●

● ●●● ●●● ● ● 4.5 ●● ●● ●●●●● ●●●●●● ●● ● ●● ●● ● ● ● ● ●● ● ● ● ●●●● ● ●●●● ●●●●●●●●●●●● ●● ●●● ● ● ● ● ● ● ● ●● ●●●●● ●●●●●●●●●● ●● ●●●●● ● ● ● ● ● ● ●● ●●●●●●● ●●●●●●● ●● ●● ● ● ● ● ●●●● ● ●● ●●●●● ● ● ●●● ● ● ● ● ● ● ●●●●●●● ●●●●●●●● ●● ●●● ● ● ● ●●● ●●●● ●●●● ●●●● ●● ● ● ● Pronotum width (mm) ● Brood mass weight (g) Brood mass weight ● ●●●● ● ● ● ● ● ● ● ● ● ●●●●●●●● ● ●●●●● ● ● ● 3 ● ● ● ● ● ● ●●●●●● ● ● ● ● ● ● ● ● ●● ● ●● ●●● ● ●● ● ● ● ● ●● ● ●● ● ● ●● ● ●●●●●● ●●● ●● ● ● ● ● ● ● ● ● ● ● ●●●●● ● ● ● ● ●● ● ●● ●● ●●●● ●●●●●●●● ● ● ● ● ● ●●● ● ●● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ●●● ●● ● ●● ● ● ● ● ● ●●● ●● ●● ● ● ● ● ● ● ● ●● ● ● ● ● 4.0 ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

2 ● ● ● ● ● ● 3.5

0.04 0.06 0.08 0.10 0.12 2 3 4 5 6 Dam body weight (g) Brood mass weight (g)

Figure 4.1 The relationships between female weight, brood mass weight, and pronotum width of their male offspring. A. The body weight of female Onthophagus taurus positively affected the weight of the brood masses they produced (the shaded area represent predicted values based on the 95% confidence intervals of the parameters of the model used). B. The weight of these brood masses positively affected the pronotum width of the male offspring that emerged from them (the shaded area represent predicted values based on the 95% confidence intervals of the parameters of the model used).

Figure 4.2 The allometry of 0.8

● ● ● horn length on pronotum ● ●● ● ●● ●● ●●●● ●● ● ● ●●●●● ●●●●● ●●● ● ●● ●●● ●●●● ●●●●●●●●●●● ●● ●●● ●●●●●●● ●●●●● ●●●● ● width of males in ●●●●●●●●●●●●●●●●●●●●●● ●● ● ●● ●●●●●●●●●●●●●● ●●●●● ● ● ●● ●●●●●●●●●●●●●● ● ●●●●● ●●●●●● ●● ● ●●●●●●● ●●●● ●●●● ●● ●●●●●●●●●● ● Onthophagus taurus. Males ●● ●●●●● ●●●●● ● ●●●● ●●●●●●●●● ●●● ● ● ●●●●● ●● ● ●● ● ● ●●● ● in the sample were the ●● ●● ●● ● ●● ●● ● 0.6 ● ●●● ● ● ● ●● ● pooled offspring produced ● ●● ● ●● ●● ● ●●●●● ● ● ● by females from all our ● ● ● ●● ● ● ● ● ● ● experimental groups (see ● ● ● ● ● ● Methods). Both axes were ● ● ● ● 0.4 ● ● ● ●●● ● transformed using natural ●●● ● ● ● ● ● ●● ● ●●●● ●●●● ●●●●● logarithms, the curve was ● ● ●●●● ● ● ● ●● ●● ● ●● ●●●●● fitted with a Richards’ ● ●●● ● ●● ●● ●●● ●●●● ●●● ● ● ● ●● ● ● ● ●● ● ● ●●● ● growth function (parameters Log [Horn length + 1] (mm) ● ● ● ●●●● ● ● ● ● ●●● ●● 0.2 ● in Table 4.3), and the shaded ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●●● ● area represents predicted ● ● ●● ● ●● ● ● ● ● ● ●● ● ● ● ● ● ● ● values based on the 95% ● ● ● ● ● ● ● ● confidence intervals of the ● ● ● ●●● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● parameters of the model 0.0 used. 0.65 0.70 0.75 0.80 0.85 Log [Pronotum width + 1] (mm)

88! ! Table 4.1 Model selection statistics for the weight of brood masses produced by females of Onthophagus taurus.

Pairwise sequential model Model comparisons Fixed effects df log Likelihood Likelihood ratio P-value

Null 4 - 1153.7

Female weight 5 - 1133.3 40.81 < 0.0001

Female weight + M group 6 - 1132.3 2.14 0.14

Female weight + PM group + M group 7 - 1132.1 0.27 0.60

Female weight + PM group + M group + 8 - 1131.6 1.07 0.30 Female weight : PM group

Female weight + PM group + M group + Female weight : PM group + 9 - 1131.5 0.10 0.75 Female weight : M group

Female weight + PM group + M group + Female weight : PM group + 10 - 1131.5 0.01 0.91 Female weight : M group + PM group : M group

Female weight + PM group + M group + Female weight : PM group + Female weight : M group + 11 - 1131.3 0.47 0.50 PM group : M group + Female weight : PM group : M group

Females were assigned to experimental groups that differed in population density during the pre-mating period (PM), and in the number of possible mates during the mating period (M). The effect of female weight was added as a covariate, female identity was added as a random effect in all models (estimated as 0.323 standard deviation in the full model), and the variance was modeled as a power function of the fitted vales with an estimated parameter value of 1.260 in the full model. Likelihood ratios were calculated as the absolute difference between the - 2 x log Likelihood of the two models being compared, and each model is being compared to the model that is one row above. Comparing these models in terms of their AIC values returns qualitatively similar results.

! 89! Table 4.2 Model selection statistics for the pronotum width of male offspring from females of Onthophagus taurus.

Pairwise sequential Model model comparisons df Fixed effects log Likelihood Likelihood ratio P-value

Null 5 - 114.19

Brood mass weight 6 - 57.72 112.93 < 0.0001

Brood mass weight + M group 7 - 56.10 3.24 0.07

Brood mass weight + M group + PM group 8 - 56.18 0.16 0.69

Brood mass weight + M group + PM group + 9 - 54.36 3.65 0.06 Brood mass weight : PM group

Brood mass weight + M group + PM group + Brood mass weight : PM group + 10 - 53.14 2.43 0.12 Brood mass weight : M group

Brood mass weight + M group + PM group + Brood mass weight : PM group + 11 - 53.13 0.03 0.87 Brood mass weight : M group + PM group : M group

Brood mass weight + M group + PM group + Brood mass weight : PM group + Brood mass weight : M group + 12 - 53.08 0.09 0.76 PM group : M group + Brood mass weight : PM group : M group

Females were assigned to experimental groups that differed in population density during the pre-mating period (PM), and in the number of possible mates during the mating period (M). The effect of brood mass weight was added as a covariate, the replicate mating chamber was added as a random effect, and female identity (nested within the replicate mating chamber) was also added as a random effect in all models. The random effects in the full model were 0.069 standard deviation for the replicate mating chamber and 0.115 standard deviation for female identity. The variance was modeled as a power function of the fitted values with an estimated parameter value of - 4.480 in the full model. Likelihood ratios were calculated as the absolute difference between the - 2 x log Likelihood of the two models being compared, and each model is being compared to the model that is one row above. Comparing these models in terms of their AIC values returns qualitatively similar results.

90! ! Regarding horn length, the model that best described the allometry between this trait and the pronotum width of males was the sigmoidal Richards’ growth function (Table 4.3, Figure 4.2). Interestingly, at least three parameters of the Richards’ growth function have biological interpretations: parameter A is an asymptote, indicating the asymptote for horn length; parameter µ is the maximum slope of the curve, indicating the biggest increase of horn length with body size; λ parameter is a lag-phase, indicating the body size after which significant horn growth occurs, and can be interpreted as the switchpoint between male morphs. Parameter ν is a shape € parameter that affects how close to the asymptote maximum growth occurs, regardless of the lag-phase, but the biological interpretation of this parameter is less € straightforward. Moreover, the model that allowed all of the parameters of the Richards’ growth function to have different values for the L and H pre-mating groups (Model 2, Table 4.4) was significantly better than the model that had the same parameters across both pre-mating groups (Model 1, Table 4.4). When looking at each specific parameter, the models that allowed only µ or only λ to vary between the L and H pre-mating groups (Models 4 and 5, Table 4.4, Figure 4.3) were also significantly better than the model that had the same parameters across both pre-mating groups € (Model 1, Table 4.4). On the other hand, the model that allowed all of the parameters of the Richards’ growth function to have different values for the S and M mating groups (Model 7, Table 4.4) was not significantly different than the model that had the same parameters across both mating groups (Model 1, Table 4.4). Therefore we concluded that Models 4 and 5, where parameters µ or λ vary between the pre-mating groups, are the minimal adequate models that fit the data best. It was not possible to model the random effect of female identity in the models € describing the horn allometry of O. taurus. The reason for this is that nine of the 80 females used in the experiment produced fewer than four male offspring. Therefore, it is impossible to estimate the effect of each female on all four parameters of the Richards’ growth function. However, to make sure that our results were robust when accounting for the number of females used, we extracted the horn length residuals from the Richards’ growth function model that had the same parameters across experimental groups (Model 1, Table 4.4). Next, we averaged the residual horn length for the male offspring of each female, and compared these averages across experimental groups using Welch's t-tests for samples of unequal variances. This approach is very conservative and avoids any possible pseudo-replication due to the non-independence among offspring from the same female. Regarding the pre-mating experimental groups, the averages of son's residual horn length per female were higher

! 91! in the high-density group than in the low-density group (t66.521 = -2.38, P = 0.02). Regarding the mating experimental groups, the averages of son's residual horn length per female were not different between the singly-mated and the multiply-mated

groups (t75.455 = -0.60, P = 0.55).

Table 4.3 Models fitted to the allometry between horn length and pronotum width of male Onthophagus taurus.

Model AIC ∆ AIC Formula and parameters Y(x) = A[1+ ν × exp(1+ ν) × exp(µ(1+ ν)(1+1/ν ) × (λ − x)/ A)](−1/ν ) Sigmoidal A (asymptote) = 0.70836 Richards’ - 983.889 0 µ (maximum slope) = 18.52390 growth € function λ (lag-phase) = 0.75852 ν (shape parameter) = 2.70687 Y(x) = A − Drop × exp(−exp(lrc) × x pwr ) € Sigmoidal A (asymptote) = 0.700975 Weibull € - 981.318 2.571 Drop (asymptote minus y intercept) = 0.684813 growth function € lrc (ln rate constant) = 13.622203 pwr (power x is raised to) = 55. 655926 Y(x) = lA + (uA − lA)/(1+ exp((xmid − x)/scal)) lA (lower asymptote) = 0. 0447065 Four- parameter - 981.127 2.762 uA (upper asymptote) = 0. 7220289 logistic € xmid (x value for inflection point) = 0. 7787523 scal (scale parameter) = 0. 0091249 Y(x) = A /1+ exp((xmid − x)/scal)) Three- A (asymptote) = 0.7306256 parameter - 977.034 6.855 logistic xmid (x value for inflection point) = 0.7774247 € scal (scale parameter) = 0.0102605 Y(x) = a + bx Linear - 721.143 262.746 a (intercept) = -6.3043 b (slope) = 8.6193 € Male offspring were pooled across females from all experimental treatments. The best model is in bold and the remaining models are sorted below by increasing values of AIC.

92! ! Table 4.4 Model selection statistics for the allometry between horn length and pronotum width in Onthophagus taurus.

Pairwise model Model comparisons log Likelihood Parameters as functions of experimental groups df P-value Likelihood ratio 1 A, µ, λ , and ν common across groups 6 524.65 Pre-mating group (PM)

2 All parameters ~ PM 10 533.39 17.47 0.002 € € 3 A ~ PM; µ, λ , and ν common across groups 7 524.68 0.06 0.81 4 µ ~ PM; A, λ , and ν common across groups 7 533.20 17.11 < 0.0001 5 λ ~ PM; A, µ, and ν common across groups 7 533.04 16.78 < 0.0001 € € 6 ν ~ PM group; A, µ, and λ common across groups 7 525.49 1.68 0.20 € € Mating group (M) € € 7 All parameters ~ M 10 525.10 0.90 0.92 € € 8 A ~ M; µ, λ , and ν common across groups 7 524.75 0.21 0.65 9 µ ~ M; A, λ , and ν common across groups 7 525.05 0.81 0.37 10 λ ~ M; A, µ, and ν common across groups 7 524.95 0.60 0.44 € € 11 ν ~ M; A, µ, and λ common across groups 7 524.79 0.29 0.59 € €

€ All models€ were fitted using the Richards’ growth function (see Table 4.3). Individuals in the sample are the male offspring produced by females from experimental groups that differed in € population€ density during the pre-mating period (PM), and in the number of possible mates during the mating period (M). The set of models being compared is composed of a full model with common parameters across experimental groups (Model 1), and models in which one (or all) of the parameters were allowed to have different values for each pre-mating or mating experimental groups (indicated by ~ PM or ~ M). The variance was modeled as an exponential function of the fitted values with an estimated parameter value of - 1.853 in the full model with common parameters across experimental groups (Model 1). Likelihood ratios were calculated as the absolute difference between the - 2 x log Likelihood of the two models being compared, and every model is being compared to the model in italic (Model 1). Comparing these models in terms of their AIC values returns qualitatively similar results.

! 93! 4.5 Discussion

Theory predicts that variations in dung beetle population density should have different fitness consequences for each male morph, so that higher population density selects for a higher frequency of minors (Moczek 2003; Moczek and Nijhout 2003; Moczek et al. 2002) and/or greater expenditure on weaponry in majors. Moreover, given that female dung beetles provide their offspring with all the resources they need to reach adulthood (Hunt and Simmons 2000), non-genetic paternal (e.g. size; Kotiaho et al. 2003) and maternal (e.g. size; Hunt and Simmons 2000) effects influence the morph adopted by male offspring. Following this rationale, in the present study we predicted that female O. taurus would be capable of responding to population density by changing the phenotype of their offspring. Our results suggest that beyond the maternal effects that arise from size variation – a seemingly adaptive effect is present whereby mothers who experienced high population density produce major offspring with longer horns. Below we build on this finding and their theoretical implications.

More minors or better majors as a response to population density? The model that best described the horn allometry of male offspring in this study was the four-parameter Richards’ growth function, which performed better than a linear model, a three-parameter logistic, a four-parameter logistic, and a Weibull growth curve (Table 4.3). Moreover, allowing two parameters (only µ or only λ ) of the Richard’s growth function to vary between the low and high density pre-mating groups resulted in significantly better fit than a model that had the same parameters € across both pre-mating groups (Table 4.4). Because the high density pre-mating group had a significantly higher value for parameter µ, the biological interpretation is that offspring from females in this group had a greater maximum increase of horn length with body size, i.e., a steeper horn length allometry (Figure 4.3a). Alternatively, because the high density pre-mating group also had a significantly lower value of parameter λ, another possible biological interpretation is that offspring of females in this group had a switchpoint for morph expression at smaller values of body size (Figure 4.3b). Both these interpretations have the same effect for major males: major € offspring of females from the high density pre-mating group had longer horns across a wider range of body sizes than major offspring of females from the low density pre- mating group (Figure 4.3). However, for minor males, differences in parameters µ and λ have different implications. If only parameter µ is different across the pre-mating groups, minors of both groups are not different, as this parameter only affects the

€ 94! ! steepness of the curve after the switchpoint (Figure 4.3a). Meanwhile, if only parameter λ is different across the pre-mating groups, the switchpoint is moved to lower values of body size under high maternal density conditions, and minors are produced over a narrower range of body sizes (Figure 4.3b), i.e. there would be fewer minors. 0.8

€ ● ● a ● ● ● ●● ● ● ●● ● ● ● ●● ● ●● ●● ●●●●● ● ● ●● ●● ● ●● ●●●● ● ●●● ● ● ●●●●●●●● ● ●● ● ●● ● ● ● ●● ●●●●● ● ● ●● ● ● ●●●●●● ● ●● ●●● ● ●●●● ● ● ● ● ●●●● ● ● ● ●●●● ●● ! ● ● ● ●●● ● ● ● ● ● ● ● ● Figure 4.3 The horn length 0.8 ● ● ● 0.8 0.6 ● ● ●● ● ● ● ● ● ● ● a ●● ●● ● ● allometry of the male offspring A ● ● ● ●● ● ● ●● ●● ● ●● ● ●●●● ●● ● ●● ●● ●● ● ●● ● ● ●● ●●●●●●●●●●● ● ● ●●● ●●● ● ● ●● ! ● ● ● ●● ●●●●● ● ● ●● ● ●● ● ●●●●●● ● ●● produced by females from ●●● ● ●●●● ● ● ● ●● ●●●● ● ● ● ●●●●● ●● ● ●● ● ●●● ! ● ● ●● ● ● different experimental groups. ● ● ●● ● ● ● ●●

0.6 ● 0.6 ●● 0.4 ● ● ● ● ● These groups differed in the ●● ● ● ● ●● ● ●● ● ●●● ● ● ● ● ● ● ●● ● ● population density experienced by ● ●● ● ● ● ● ● ●● ● ●●● ●● ● ● females during their pre-mating ! ● ● ●● ● ● ● ● ● Log [Horn length + 1] (mm) ● ● ● ● ● ● 0.4 ● ● 0.4 ●● (PM) period. Offspring produced by 0.2 ● ● ● ● ●● ● ● ● ● ● ● ●● ●● ● ●●●● ●● females from the low-density ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● experimental group are indicated by ● ●● ● ●●● ● ●

! ● ● ● ● ● Log [Horn length + 1] (mm) ● ●● ● ● ● ● ● ● ● ● ●● ● open circles and a continuous curve, ● ● ● ● ● ●● 0.2 ● 0.2 0.0 ● ● Log [Horn length + 1] (mm) length + Log [Horn ● ● ●● ● whereas offspring produced by ●●● ● ● ● ● 0.65 0.70 0.75● ● ● 0.80 0.85 ● ● ● ● ● ● females from the high-density ● Log [Pronotum width● + 1] (mm) ! ● ●● ● experimental group are indicated by ● ● ● ● ● ● ● ● 0.0 0.0 crosses, and a broken curve. Again, 0.65 0.70 0.75 0.80 0.85 both axes were transformed using ! Log [Pronotum width + 1] (mm) 0.8 natural logarithms, and the curves 0.8 ● ● ● ● b ● ● ● ● ● ●● ●● B ● ●●●● ●●●● ●● ● were fitted with a Richards’ growth ● ●● ● ●● ● ● ● ●●● ●●●● ●● ● ●● ● ● ●●●●●●● ● ! ● ● ●● ● ●● ●●●●● ● ● ●● ● ● ●●●●●● ● ●● ●●● ● ●● ● ● ●●●●●● ● ● function, but this time (A) only ●●●● ●●● ● ● ●● ●●●●●● ! ● ● ● ● ● ● ● ● ● parameter µ (Model 4), (B) or 0.8 ● ● ● 0.6

0.6 ● ● ●● ● ● ● ● ● ● b ● ● ● ● ●● ● ● ●● ●● parameter λ (Model 5) were ● ● ● ● ● ●● ● ●● ●● ●●●●● ● ● ●●● ●● ● ●● ● ●● ● ● ●●● ●●●●●●● ● ● ● ● ●● ● ●● ● ●● ● ●● ●●●●● ● ● ●● ● ●● ● ●●●●●● ● ●● allowed to vary between the ●●● ● ●●●● ● ● ● ●● ●●●● ● ● ● ●●●●● ●● ● ●● ● ●●●

! ● ● ●● ● ● ● experimental groups, whereas ● ●● ● ● ● ●● 0.6 ●●● 0.4 0.4 ● ● ● ● ● ●● ● ● ● parameters A and ν were always ●● ● ●● ● ●●● ● ● ● ● ● ● ● ● ●●● commons€ across experimental ● ● ● ● ● ● ● ● ●●● ● ●● ● ●●● ● ● groups (see Table 4.4). A. According ! ●● ● ● ● ● ● Log [Horn length + 1] (mm) ● ● ● ● ● ● 0.4 ● ●●● 0.2

0.2 to the model that allows only ● ● ● ● ● ●

Log [Horn length + 1] (mm) length + Log [Horn ● ● ● ● ● ● ●● ● ● ●●● ● ●● ● ● ●●● parameter µ to vary between ● ●● ● ● ● ● ● € ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● experimental groups (Model 4, ●●● ● ● ● ● ● ! ● ● Log [Horn length + 1] (mm) ● ●● ● ● ● ● ● ● ● ● ● ●● ● Table 4.4), the vertical dotted line ● ● ● ● ●● 0.2 ●

0.0 ● 0.0 ● ● ● ●● ● depicts a value of body size for ●●● ● ● ● ● 0.65 0.70 0.75● ● ● 0.80 0.85 0.65! 0.70! 0.75● ! ● ● 0.8! 0.85! ● ● ● which the model predicts a horn ● Log [Pronotum width● + 1] (mm) ● ●● ● Log [Pronotum● width + 1] (mm)! length 8.4% greater for offspring of ● ● ● ● ● ● ● 0.0 females that experienced high population0.65 density0.70 in their pre0.75-mating0.80 period (upper0.85 dotted horizontal line) than for offspring of females that experiencedLog [Pronotum width low + population1] (mm) density during their pre-mating period (lower dotted horizontal line). This line was chosen to depict the biggest possible difference in the horns of offspring produced by females of the two experimental groups described above.

Our original prediction was that females who experienced higher population densities could respond adaptively by producing a steeper horn allometry among major males and/or more minor males (due to larger switchpoint body sizes for horn production). Our results corroborate part of this hypothesis, consistently pointing towards longer horns in major males produced under the condition of high population density. This is because the two parameters of the Richards’ growth function that are significantly different between the pre-mating density groups have this same effect on

! 95! major males (Figure 4.3). On the other hand, our results are inconclusive regarding the effect of pre-mating population density on minor males, as one of the parameters of the Richards’ growth function that is significantly different between pre-mating density groups has no implications for the frequency of minor males produced (Figure 4.3a). If anything, females actually produced fewer minor males under high density conditions, and not more, as we originally predicted (Figure 4.3b). This result challenges the prediction that more minor males should be produced under high population density, suggesting that minor males might actually have fewer mating opportunities than major males under high population densities. Nevertheless, this result should be interpreted with care, as it was not consistently supported by the two models that best fitted our data (Models 4 and 5, Table 4.4), whereas the horn length change in major males was. The uncertainty regarding the effects of our experimental treatments on minor males was probably because we had fewer minor males than expected in the offspring of females from all experimental groups, when compared to the proportion of approximately 60% minor males commonly found in natural populations of O. taurus in Western Australia (Simmons et al. 1999). This occurred because we did not limit the reproductive activity of females by dung availability. In other words, we provided enough dung for females from all groups to produce more major offspring than they normally do in field conditions. Therefore, any differences detected in the proportion of major and minor offspring produced by females from different experimental groups could reflect adaptive allocation strategies to produce offspring that are better able to cope with the population density that they will experience, isolating effects of resource limitation for females. In conclusion, it has been argued that an increase in the density of competing males of O. taurus results in a greater probability of encountering rival males, which allows only the largest males to benefit from fights (Moczek and Nijhout 2003; Moczek 2002). Earlier studies used this rationale to predict that higher population densities make the sneaking tactic of minor males advantageous over a wider range of body sizes, moving the switchpoint for horn production to larger body sizes (Moczek 2003; Moczek and Nijhout 2003), and predicted no change in the horn allometry of major males. However, our results do not support this prediction, showing that when females experience high population densities the switchpoint is either unchanged, or moved to smaller body sizes through maternal effects. In contrast, maternal effects clearly affect the horn allometry of major males, increasing the horn length of majors produced by females experiencing high population density. Even small differences in

96! ! horn length can provide considerable advantages in male contest competition in this (Moczek and Emlen 2000) and other dung beetle species (Emlen 1997a; Pomfret and Knell 2006), and horn length in O. taurus has strong independent effects on male competitive fitness (Hunt and Simmons 2001). Females thus appear able to use cues of population density to prepare their major sons for an environment in which competition will be intense.

A new mechanism of maternal effect in O. taurus Our results indicate that females of the dung beetle Onthophagus taurus are capable of perceiving population density, and responding by changing the horn allometry of the male offspring they produce. Maternal effects on male dimorphism are well known in dung beetles, because male morph is mainly determined by the quality and quantity of dung ingested by larvae (Emlen 1994; Emlen 1997b), which in turn is collected and provisioned by the parental individuals (Hunt and Simmons 2002a). As a consequence, parental effects through offspring provisioning are chief determinants of the alternative phenotypes adopted by dung beetles (Hunt and Simmons 2000; Kotiaho et al. 2003). However, the females in our experimental treatments did not respond by changing their brood mass provisioning decisions, as we could not detect any effect of the experimental manipulations on the weight of the brood masses produced. Likewise, the body size (measured as pronotum width) of male offspring was also unaffected by our experimental manipulations. Rather, our data suggest that females seem capable of specifically influencing the horn growth of their male offspring. We suggest that the specific biological mechanism through which this maternal effect operates may be a transgenerational effect that goes beyond a maternal influence on offspring condition or body size per se. The first hypothesis that we put forward is that the maternal effect we detected may be transmitted through the eggs laid by females. Maternal effects operating through egg size are not uncommon (references in Rossiter 1996), and have been reported in male dimorphic mites (Smallegange 2011). But a straightforward effect of egg size on offspring morph would probably affect offspring body size as well (as in Smallegange 2011), which was not the case in our study. Alternatively, egg composition could in theory affect offspring horn growth without a detectable effect on offspring body size. Egg composition is known to affect offspring phenotype in birds (Groothuis and Schwabl 2008) and fish (Segers et al. 2012). In the seed beetle Stator limbatus, females change the size and probably the composition of their eggs according to the host plant they encounter while maturing eggs, and the survivorship of their

! 97! offspring is greatly affected by this plasticity in egg production (Fox and Savalli 2000). This kind of maternal effect through egg composition might also be present in male dimorphic fig wasps, in which females probably produce chemicals that influence male morph determination in the offspring by acting on the genetic cascades of their genomes (Pienaar and Greeff 2003). Not much is known about egg composition in dung beetles, but we suggest that this is a promising topic that could reveal hidden forms of maternal effects on the male offspring of these animals. Another possibility is that the maternal effect detected in this study is transmitted through a feature of the brood masses that could not be detected in our experiment, such as the composition of the saliva components used by females in the process of building brood masses. The naturalist Jean Henri Fabre (Fabre 1918) was the first to notice that the inner walls of brood masses built by females of Onthophagus dung beetles are coated with a shiny and greenish semifluid, which he suggested to be produced by the mother, probably as the result of semi-digested food. It is possible that this substance contains hormones derived from the saliva of females, which can have an effect on the development of offspring horns. By altering the hormone content of their saliva and consequently of the processed dung provided for their young, females could influence horn development of their offspring through what would be a new mechanism for maternal effects in these animals. Both a maternal adjustment to the composition of the eggs or to the composition of the brood masses could cause major phenotypic effects in their offspring, potentially through epigenetic effects, such as DNA methylation (Field et al. 2004). Epigenetics seem to play a major role in the regulation of phenotypic plasticity (Glastad et al. 2011), especially for insects that present conditional dimorphisms (Simon et al. 2011; Simpson et al. 2011), such as the well-studied plastic expression of worker and queen castes in the honeybee (Chittka and Chittka 2010; Li et al. 2010).

Conclusion As with any kind of phenotypic plasticity, male dimorphism that results from conditional strategies has an underlying genetic variation, but the adult phenotype adopted by each male is mainly determined by environmental conditions (Hazel et al. 1990). Therefore, it may seem that selection can only target conditional strategies at two levels: the genetic switchpoint that determines which alternative phenotype will be expressed in each environmental condition (Tomkins and Hazel 2007); and the physiological cascade that is involved in phenotypic expression per se (i.e., the downstream mechanisms, Emlen 2008). However, if maternal effects play a role in

98! ! offspring’s morph determination (e.g. Hunt and Simmons 2000; Tomkins et al. 2001), then the environments experienced by developing juveniles can also evolve over generations. This is because although maternal effects are environmentally transmitted, the genes that affect their expression (which reside in the mothers’ genomes) are subject to selection. To date, examples of maternal effects on male dimorphism come from a few species of fish (Taborsky 2008), fig wasps (Pienaar and Greeff 2003), burrowing bees (Tomkins et al. 2001), and also dung beetles (Hunt and Simmons 2000; Kotiaho et al. 2003). These maternal effects are expected to evolve whenever females are better capable of assessing the environmental conditions that their offspring will experience than the offspring themselves (Mousseau and Dingle 1991). Our study describes a new type of maternal effect in dung beetles, which is likely to operate through egg or brood mass composition. This maternal effect probably allows females to respond to population density adaptively, preparing at least their major offspring for the level of sexual competition they will face as adults. In Figure 4.3 it can be visualized that major offspring of the same body size can have horns up to 8.4% longer if they were produced by females that experienced high density than if they were produced by females that experienced low density. This difference is enough to significantly increase the chances of the major male with longer horns to win a fight (Moczek and Emlen 2000). Finally, this maternal effect represents a novel transgenerational response of alternative reproductive tactics to population density.

! 99! !

100!!

CHAPTER FIVE

Correlated evolution of sexual dimorphism and male dimorphism in a lineage of Neotropical harvestmen

!

! 101! !

102!! 5.1 Abstract

Under intense sexual selection, secondary sexual traits (ornaments and weapons) increase male fitness, but are generally maladaptive when expressed in females, generating intralocus sexual conflict that is ameliorated through the evolution of sexual dimorphism. Intense sexual selection on males can also favor the evolution of male dimorphism, where alternative phenotypes that avoid expenditure in secondary sexual traits achieve copulations using ‘alternative mating tactics’ (AMTs). Expressing secondary sexual traits can thus increase or decrease fitness in males, depending on which AMT they employ, generating a conflict within males that can be ameliorated by the evolution of male dimorphism. Therefore, the phenotypic optima of females and small males are similar in terms of including the suppression of such secondary sexual traits. This means that male dimorphism could coevolve with sexual dimorphism, due to the evolutionary forces acting against both intralocus sexual conflict and conflict between males employing different AMTs. Here we tested this hypothesis by investigating the evolution of sexual and male dimorphism for two secondary sexual traits in 48 species in a lineage of Neotropical harvestmen (Arachnida: Opiliones). Using a Bayesian approach with reversible-jump Markov chain Monte Carlo, we demonstrate that the two types of dimorphism present strongly correlated evolution, and that sexual dimorphism consistently precedes male dimorphism in this major arachnid group. Our findings were consistent for two different traits, and are robust to phylogenetic uncertainty. We propose that sexual dimorphism evolves earlier than male dimorphism because the genetic architecture for sex specific expression is already present even in sexually monomorphic species, due to sex chromosomes. The same is not true in the case for male dimorphism. We suggest therefore, that if a sexual trait arises first in an autosome and is expressed in all individuals, its suppression in females evolves more readily than its suppression in small males that adopt AMTs.! !

! 103! 5.2 Introduction

The term sexual dimorphism describes phenotypic differences between conspecific males and females, and is normally used for secondary sexual traits that have no direct mechanical role in insemination (Andersson 1994). The incidence and magnitude of sexual dimorphism across species is thought to reflect the extent to which the selective pressures acting on each sex differ (Lande 1980). Because the fitness of males is usually more closely correlated with the number of copulations achieved than is the fitness of females (Bateman 1948; Arnold and Duvall 1994), traits that contribute to mating success are under more intense sexual selection in males than they are in females. If traits that contribute to male mating success are costly when expressed in females, the optimal phenotype of the sexes can be pushed apart, generating intralocus sexual conflict (Rice and Chippindale 2001). Thus, the genetic basis for intralocus sexual conflict is a strong intersexual genetic correlation for a trait for which the optima differs between the sexes (Lande 1980). The resolution of this conflict is the break down of these intersexual genetic correlations and the evolution of sexual dimorphism, which can occur through a variety of proximate mechanisms (Day and Bonduriansky 2004; Stewart et al. 2010). High levels of sexual dimorphism suggest intense sex-specific sexual selection, and this has been demonstrated with comparative methods in blackbirds (Webster 1992), Anseriformes (ducks, geese, and relatives; Figuerola and Green 2000), primates (Plavcan and Vanschaik 1992; Mitani et al. 1996), and stalk-eyed flies (Wilkinson 1993). Sexual selection on male secondary sexual traits is often accompanied by the intensified expression of such traits (Emlen and Nijhout 2000), bringing about fitness costs for males (Emlen 2001; Jennions et al. 2001). As the costs of producing and maintaining weapons and ornaments are not evenly affordable for all males, the expression of these traits can be disproportionally greater in large males (Wilkinson and Taper 1999). Accordingly, the degree of phenotypic variation in secondary sexual traits correlates positively with proxies of sexual selection intensity (Simmons and Tomkins 1996; Wilkinson and Taper 1999; Knell et al. 2004). When variation in male secondary traits becomes discrete, the species is said to present male dimorphism (Gadgil 1972). Male dimorphism should also be more common in species whose males are under extreme sexual selection, because intense competition for access to receptive females can favor the evolution of alternative phenotypes that avoid aggressive competition and achieve copulations by alternative means (Brockmann and Taborsky 2008; Chapter 1). Different male morphs employ alternative mating tactics, in which

104!! large males (majors) have more elaborate weaponry and guard females or reproductive territories, while small males (minors) have reduced weaponry and sneak copulations (Gross 1996; Chapter 1). As with sexual dimorphism, male dimorphism may also result in different selective pressures, in this case acting within a sex on the different male morphs, generating bimodal trait distributions in males (often referred to as major and minor males). Because minor males refrain from displaying or fighting (Brockmann 2001; Oliveira et al. 2008), their optimum morphology is usually more similar to that of females, even more so if their tactic relies on mimicking females to gain access to harems defended by major males (Shuster 1987; Forsyth and Alcock 1990). Here an intralocus conflict can arise within males because alleles increase fitness when expressed in males in good condition or with high status (males from the major morph) but decrease fitness when expressed in males in poor condition or with low status (males from the minor morph). One resolution of this type of conflict is the evolution of mechanisms that couple the expression of a secondary trait to the status/condition of males. Male dimorphism may evolve through a reprogramming event that switches the developmental path of juvenile males (Nijhout 1999; 2003), allowing selection to act independently on male morphs (Emlen and Nijhout 2000; West-Eberhard 2003; Tomkins and Moczek 2009). Wherever male dimorphism is common, the above reasoning suggests that it should coevolve with sexual dimorphism, due to the shared evolutionary forces acting against intralocus sexual conflict and intralocus conflict between males of different status. To our knowledge, however, there is only one study that has examined the coevolution of sexual and male dimorphism, focusing on the evolution of thoracic and head horns in onthophagine dung beetles. Emlen et al. (2005) showed that the repetitive evolutionary gains and losses of these labile traits occurred in females and in minor males in a very tightly correlated fashion. In the case of dung beetles, both male dimorphism and sexual dimorphism seem to depend on an endocrine threshold mechanism that can switch horn growth on and off during development (Emlen et al. 2006; 2012; Shingleton et al. 2007), and this shared endocrine regulatory mechanism was hypothesized to be the proximate cause underlying the correlated evolution of the two types of dimorphism (Emlen et al. 2005). Even though the studies on dung beetles shed light on the developmental pathways of holometabolous insects, a fundamental question remains unanswered: are these endocrine regulatory mechanisms a prerequisite for the coevolution of sexual and male dimorphisms generally, or just one of the many ways through which this

! 105! coevolution can arise? While endocrine regulatory mechanisms may be the proximate influence behind the correlated evolution of male and sex dimorphism in dung beetles (Emlen et al. 2006), the ultimate cause for this phenomenon is the divergent selection pressures acting on major males and on minor males and females. If it holds true that both minor males and females share phenotypic optima when it comes to the reduced expression or complete absence of weapons and ornaments, the coevolution of sexual and male dimorphisms would be repeatedly achieved in animals, regardless of the specific developmental mechanisms behind them. To test this idea, we investigated the evolution of male dimorphism and sexual dimorphism in harvestmen (Arachnida: Opiliones), a group of arthropods in which there is no metamorphosis, and hence the developmental pathways from nymphs to adults is strikingly different to that of holometabolous insects (Truman and Riddiford 2002; Minelli 2003). The first goal of this study was to investigate whether male dimorphism and sexual dimorphism would still exhibit correlated evolution even in the absence of the type of endocrine regulatory mechanism that generates dimorphisms in dung beetles. We hypothesized that these dimorphisms should coevolve whenever both small males and females benefit from avoiding the costs of expressing the secondary sexual traits that are expressed by large males. Here we also asked whether male dimorphism usually follows or precedes sexual dimorphism. Assuming that a gene encoding a secondary sexual trait could originally evolve in an autosomal locus, and thus be expressed by all individuals, what should evolve first, sexual dimorphism through suppression of the trait in females or male dimorphism through suppression of the trait in small males? This question relates to whether the mechanisms of male dimorphism are co-opted from previously existing mechanisms of sexual dimorphism, or vice-versa. Emlen et al. (2005) suggested that sexual dimorphism evolving through co-option of an existing mechanism of male dimorphism is unlikely because the secondary sexual traits of males should be more costly for females than for small males due to their potential impact on female fecundity and/or longevity (Fitzpatrick et al. 1995; Martin and Badyaev 1996). Therefore, selection against the expression of such traits would be stronger in females than small males, suggesting that intralocus sexual conflict is stronger than intralocus conflict within males of different status. Moreover, whereas secondary sexual trait expression would be selected against in all females, only a fraction of males (minors) would be targeted by this selection, providing a greater opportunity for the evolution of sexual dimorphism than for the evolution of male dimorphism (West-Eberhard 2003). Finally, sexual dimorphism might evolve more readily than male dimorphism

106!! because the genetic architecture for sex specific expression is already provided by sex chromosomes, even in sexually monomorphic species. Sexual dimorphism in a new trait could thus arise from sex-specific regulation via genes on the sex chromosome, but such pre-existing genetic architecture is not available for male dimorphism. These ideas have not yet been tested, and the second goal of this study was to investigate directionality in the coevolution between sexual and male dimorphism in harvestmen. As far as we know, our approach, employing a Bayesian analysis of correlated evolution, provides the first direct evidence that sexual dimorphism precedes male dimorphism.

5.3 Methods

Sampling species from natural populations and museum collections Harvestmen constitute the third largest group of arachnids, the order Opiliones, with about 6,500 extant species (Machado et al. 2007; Kury 2011 2012). Gonyleptidae is the second largest family of the order (almost 13% of all harvestmen belong to this family; Kury and Pinto-da-Rocha 2007; Kury 2012), and the bulk of its diversity is found in the Coastal Atlantic Rainforest from southeastern Brazil (Pinto-da-Rocha et al. 2005). We sampled adult males (n = 3,397) and females (n = 2,844) of 48 gonyleptid species, belonging to 10 different subfamilies and 42 genera (Table 5.1). We selected species on the basis of two criteria: (1) their phylogenetic position, in order to maximize the number of subfamilies sampled; and (2) the availability of individuals to measure, either in museum collections or in natural populations that harbor high densities of individuals. In order to examine species that were available in zoological collections, we visited two Brazilian museums between November 2010 and February 2011: Museu de Zoologia da Universidade de São Paulo (MZSP), state of São Paulo, and Museu Nacional da Universidade Federal do Rio de Janeiro (MNRJ), state of Rio de Janeiro. In order to examine the remaining species, we conducted field trips to seven natural reserves in southeastern Brazil: Parque Florestal do Itapetinga (23º10’S, 46º25’W), Parque Estadual da Serra do Mar (23º22’S, 45º01’W), Parque Estadual do Desengano (21º54’S, 41º54’W), Parque Estadual da Ilha do Cardoso (25°03’S, 47°53’W), Parque Estadual Intervales (24°14’S, 48°04’W), Parque Nacional da Serra dos Órgãos (22º26’S, 43º00’W), and Parque Nacional do Itatiaia (22º21’S, 44º44’W). We visited all reserves between October 2010 and March 2011, a period corresponds to the warm-wet season in southeastern Brazil, and includes the months when the reproductive activity of

! 107! several gonyleptid species peak (e.g., Gnaspini 1995; Machado and Oliveira 1998; Willemart and Gnaspini 2004; Buzatto and Machado 2008; Requena et al. 2012). For each species that we examined, we indicate in Table 5.1 whether they were sampled from museum collections, natural populations, or a combination of both.

Table 5.1 List of the 48 species of gonyleptid harvestmen (from 10 subfamilies) included in our study, along with samples sizes (total of 3,463 males and 2,874 females) and sources of specimen examined. “Literature” in the sample sizes columns means that only a few individuals of that gender or species were examined, and information from the literature was also consulted to diagnose sexual and male dimorphism. See Methods for an explanation of the names of museums and nature reserves visited, along with GPS coordinates for the nature reserves.

Sample size Subfamily Species Source Males Females Asarcus ingenuus 22 19 MNRJ, MZSP Bourguyiinae Bourguyia trochanteralis 66 124 P.E. Ilha do Cardoso, MNRJ, MZSP Ampheres leucopheus 29 19 P.E. Intervales, MZSP Arthrodes xanthopygus 67 20 P.N. Serra dos Órgãos, MNRJ, MZSP Caelopygus elegans 32 21 P.N. Itatiaia, MNRJ, MZSP Caelopyginae Metarthrodes pulcherrimus 30 20 MZSP Pristocnemis albimaculatus 20 10 P.N. Serra dos Órgãos, MNRJ P. pustulatus 26 20 MZSP Thereza speciosa Literature Literature MZSP Cobaniinae Cobania picea 36 27 P.N. Itatiaia, MNRJ Goniosomatinae Acutisoma longipes 48 29 P.F. Itapetinga, MZSP

108!! Goniosoma calcar 26 23 P.N. Serra dos Órgãos, MZSP G. roridum 28 16 P.N. Serra dos Órgãos G. varium 30 20 MNRJ, MZSP G. venustum 18 2 P.N. Serra dos Órgãos Heteromitobates discolor 67 33 P.E. Serra do Mar Mitogoniella taquara 68 37 P.N. Itatiaia Serracutisoma proximum 365 444 P.E. Intervales Gonyleptellus cancelatus 50 20 P.N. Serra dos Órgãos, MNRJ Gonyleptinae Gonyleptes horridus 44 20 MNRJ Neosadocus maximus 43 30 P.E. Intervales Acrogonyleptes rhinoceros 21 Lit MZSP Hernandaria una 16 6 MNRJ, MZSP Multumbo dimorphicus 46 20 P.E. Desengano Hernandariinae M. terrenus Literature Literature MZSP Piassagera brieni Literature Literature MZSP Pseudotrogulus funebris Literature Literature MZSP Magnispina neptunus 49 5 MNRJ, MZSP Heteropachylinae Pseudopucrolia mutica 68 20 MNRJ Discocyrtoides nigricans 82 30 MNRJ, MZSP Mitobatinae Encheiridium ruschii 29 24 MNRJ, MZSP Ischnoterus tenebrosus 26 15 MNRJ, MZSP

! 109! Longiperna concolor 634 1166 P.E. Intervales Metamitobates squalidus 61 25 MNRJ, MZSP Mitobates triangulus 50 16 MNRJ, MZSP Mitobatula castanea 24 17 MNRJ, MZSP Neoancistrotus gracilis 64 21 MNRJ Promitobates bellus 54 30 P.E. Serra do Mar P. ornatus 273 370 P.E. Intervales Ruschia maculata 50 20 MNRJ, MZSP Cadeadoius niger 63 39 MNRJ Deltaspidium tenue 41 31 MNRJ Heliella singularis 50 23 MNRJ, MZSP Progonyleptoidellinae Iguapeia melanocephala 21 21 MZSP Iporangaia pustulosa 514 Literature P.E. Intervales Mitopernoides variabilis Literature Literature MZSP Progonyleptoidellus striatus 66 Literature MZSP Sodreaninae Sodreana barbiellini 46 21 P.E. Serra do Mar, MNRJ

110!! Detecting sexual and male dimorphism We measured all individuals for two traits that are used in male-male fights in at least some gonyleptid species, and for the length of the dorsal scute (carapace), a commonly used proxy for body size in harvestmen (see references in Buzatto et al. 2011). The potentially sexual traits measured were: (i) the length (measured dorsally) of the lateral apophysis on the coxa of the right fourth leg (hereafter called C4A length; Figure 5.1A- F), which is known to be used in male-male fights in several gonyleptid species (Nazareth and Machado 2009; 2010; Willemart et al. 2009), and (ii) the length (measured dorsally) of the femur of the fourth leg (hereafter called F4 length; Figs. 1H-I), which is typically used in male-male fights in many gonyleptid species (Machado and Macías- Ordóñez 2007; Willemart et al. 2009, Zatz et al. 2011; BAB and GM, unpublished data). We captured dorsal images of all individuals using the macro mode of a Nikon L100 digital camera and analyzed images using the free software ImageJ (Rasband 1997- 2011). On each image, we measured C4A length starting from where the lateral apophysis of the fourth coxa becomes visible dorsally, at the retro-lateral limit of the dorsal scute, following the apophysis curve (when it did curve) until reaching its end (Figure 5.2). We then measured F4 length starting from the proximal end (where it joins the trochanter), following its curve (when applicable) until reaching the distal end (where it joins the patella). We investigated sexual dimorphism in each species by fitting general linear models with C4A and F4 lengths as dependent variables (in different models) and sex, dorsal scute length, and their interaction as independent variables. We fitted these models in R version 2.14.2 (R Development Core Team 2011), and considered C4A and F4 of a given species to be sexually dimorphic in their lengths if sex and/or sex*dorsal scute length interaction were significant at the 0.05 level in our models (Table 5.2). Next, we investigated male dimorphism in each species in our sample firstly by checking C4A and F4 lengths for bimodality through adjustment of non-parametric kernel density estimates to frequency distributions. We then parameterized these distributions as mixtures of two normal distributions, using finite mixture models implemented in the package ‘mixtools’ (Benaglia et al. 2009) for R version 2.14.2 (R Development Core Team 2011). We calculated the Bayesian Information Criterion (BIC) for the model that described a mixture of two normal distributions and for a simpler model that described one single normal distribution (parameterized through maximum likelihood optimization; Table 5.2). The formula for BIC is

, where lnL is the natural logarithm of the likelihood function evaluated at the maximum likelihood estimate for a given model, K is the number of

! 111! parameters and N is the sample size (Schwarz 1978). In our case, K = 2 (mean and variance) for models that describe one single normal distribution, and K = 5 (two means, two variances, and a value of the mixing proportions of individuals from the two distributions) for models that described a mixture of two normal distributions (Benaglia et al. 2009). Finally, we used the difference between the BIC of these two models (ΔBIC) as our basis for inference about male dimorphism. The model with the lowest BIC value was selected as the most parsimonious model (as long as ΔBIC > 2), and we only classified a species’ trait as male dimorphic when the most parsimonious model was the one that described a mixture of two normal distributions for the trait being analyzed. Our approach to detect male dimorphism is similar to the one proposed by Rowland and Qualls (2005), and recently applied to analyze male dimorphism in several families of beetles (Rowland and Qualls 2005; Rowland et al. 2005; Rowland and Emlen 2009), and in the Wellington tree weta (Kelly and Adams 2010). The main difference from these studies was that here we used BIC, instead of AIC or log- likelihood tests, as the basis for our model comparisons. This small difference is important because BIC more efficiently avoids overfitting, as the penalty term for the number of parameters in the model is larger in this metric than in AIC comparisons or in log-likelihood ratio tests (Schwarz 1978). Our classification of male dimorphism can therefore be considered conservative.

Phylogenetic relationships within the Gonyleptidae One of the 10 subfamilies of Gonyleptidae included in this study, namely Cobaniinae, has only one genus and two described species. Of the remaining nine subfamilies, seven have been through taxonomical and cladistics reviews in the last two decades. Bourguyiinae (Yamaguti and Pinto-da-Rocha 2009), Caelopyginae (Pinto-da-Rocha 2002), Goniosomatinae (DaSilva and Gnaspini 2009), Hernandariinae (DaSilva and Pinto-da-Rocha 2010), Heteropachylinae (Mendes 2011), Mitobatinae (Bragagnolo and Pinto-Da-Rocha 2012), and Sodreaninae (Pinto-da-Rocha and Bragagnolo 2010) are now known to be monophyletic, and the phylogenetic relationships among species within these subfamilies are relatively clear.

112!!

Figure 5.1 Examples of some harvestman species belonging to the family Gonyleptidae included in this study. (A) Major male, (B) minor male, and (C) female of Cobania picea (Cobaniinae). The arrows indicate the apophyses on the fourth coxa. Note that the female has only a small, blunt tubercle while the minor male has an apophysis considerably shorter than the major male. (D) Major male and (E) female of Arthrodes xanthopygus (Caelopyginae). In this species, females lack the apophysis on the fourth coxa while major males have a huge apophysis (arrows). (F) Copulating pair of Ampheres leucopheus (Caelopyginae), with the male on bottom left. Note that the female has apophyses on the fourth coxa that are about half the length of those of the male (arrows). (G) Male of Mitopernoides variabilis (Progonyleptoidellinae), which lacks the apophysis on the fourth coxa (indicated by the arrows). (H) Major male and (I) female of Promitobates ornatus (Mitobatinae). In this subfamily, the most conspicuous form of sexual dimorphism is the elongation of the fourth pair of legs in males (arrows indicate the fourth femur in both male and female).

! 113!

There are two phylogenies for the relationship among subfamilies of Gonyleptidae: one based on 74 morphological characters and 33 species, including representatives of all 16 subfamilies (Pinto-da-Rocha 2002), and another based on 63 ecological, behavioral, and chemical characters that included 28 species from 14 subfamilies (Caetano and Machado in press). Although, there are differences between the topology of these two phylogenies, both recognize a monophyletic clade called K92, which includes Caelopyginae, Gonyleptinae, Hernandariinae, Progonyleptoidellinae, and Sodreaninae. Moreover, both phylogenies show that Goniosomatinae and Mitobatinae are closely related to K92, and that Cobaniinae and Heteropachylinae are basal lineages of Gonyleptidae. Most of the other gonyleptid subfamilies were either not formally revised using a cladistic approach or their phylogenetic position within the available topologies is highly incongruent (e.g., Metasarcinae, Pachylinae, Pachylospeleinae, and Tricommatinae). For these reasons, they were not included in our sample, which focus mainly on the well-resolved clades. In order to infer the phylogenetic relationships between the 48 species in our sample, we combined the information on the phylogeny within each of the eight already reviewed subfamilies with the information on the phylogeny among these subfamilies from Caetano and Machado (in press), arriving at an almost fully resolved tree, with only one trichotomy at the base (Figs. 3A and 4A). Next we repeated this procedure, but this time using the phylogeny from Pinto-da-Rocha (2002) for the relationships among subfamilies, arriving at a fully resolved tree (Figs. 3B and 4B). We then repeated this step one last time, using a strict consensus between the phylogenies from Pinto-da-Rocha (2002) and Caetano and Machado (in press) for the relationships among subfamilies, this time arriving at a tree with two polytomies (one with six and one with five branches). Importantly, the subfamily Gonyleptinae is monophyletic under Pinto-da-Rocha’s hypothesis, but not under Caetano and Machado’s phylogeny. However, the later phylogeny included two of the three Gonyleptinae genera that were also included in our study (Gonyleptes and Neosadocus), and the position of Gonyleptellus (in ‘Gonyleptinae 3’, sister group to Sodreaninae; see Caetano and Machado in press) can be inferred based on its morphological affinity with Gonyleptes. We repeated all our comparative analyses (see below) with each of the three working phylogenies that we generated, allowing us to assess the sensitivity of our results to phylogenetic uncertainty. Given that our comparative analyses were based on an approach that is only possible with fully resolved trees, we used the three possible trees derived from resolving the trichotomy in Caetano and Machado’s (in press) phylogeny, and a sample of 1,000 trees with randomly resolved polytomies for

114!! the consensus between the phylogenetic hypotheses of Pinto-da-Rocha (2002) and Caetano and Machado (in press). The MCMC chains used in our comparative analyses (see below) copes with phylogenetic uncertainty by visiting alternative phylogenetic trees in proportion to their probability of being true given the model, priors, and data (see for example Higginson et al. 2012a).

Comparative analyses As a first step to understand the evolution of male dimorphism and sexual dimorphism in the Gonyleptidae, we reconstructed ancestral states and mapped the evolution of dimorphisms in C4A and F4 lengths in our three working phylogenies (Figs. 3-4; consensus tree not shown). We did this through a likelihood reconstruction method (Pagel 1999) in a Markov, k-status, 1-parameter model with four states, using the tool “trace character history” in Mesquite (Maddison and Maddison 2011). We firstly employed this approach for each type of dimorphism (between sexes and among males) separately, and next combining the two types of dimorphism, so that each state was a combination of sexual dimorphism (present or absent) and male dimorphism (present or absent). In this latter analysis, a given species could be sexually and male monomorphic (state 0, white in Figs. 3-4), sexually dimorphic and male monomorphic (state 1, gray in Figs. 3-4), sexually monomorphic and male dimorphic (state 2, not shown in Figs. 3-4 because no species fell in this category), or sexually and male dimorphic (state 3, black in Figs. 3-4). We always mapped trait evolution separately for C4A and for F4. We then tested for the correlated evolution between sexual dimorphism and male dimorphism for each trait with a Bayesian approach using reversible-jump Markov chain Monte Carlo (RJ MCMC) implemented in the software BayesTraits (available at www.evolution.rdg.ac.uk; Pagel and Meade 2006). This approach allows us to assess the probability that changes in sexual dimorphism preceded the evolution of male dimorphism, or vice-versa. RJ MCMC techniques have been increasingly used to investigate central topics in evolutionary biology, and recent examples include the relationships between parental care, sexual selection, and mating systems (Thomas and Szekely 2005; Gonzalez-Voyer et al. 2008), the evolution of sexual size dimorphism (Perez-Barberia et al. 2002), as well as the evolution of sperm traits (Higginson et al. 2012a) and their coevolution with female traits (Fitzpatrick et al. 2009; Higginson et al. 2012b). In our analysis, we coded both sexual and male dimorphism as discrete binary data (0 and 1), and each species could be placed in one of four categories described as

! 115! [sexual dimorphism, male dimorphism], such that category 1 = [0,0], category 2 = [0,1], category 3 = [1,0], and category 4 = [1,1]. We then used the program DISCRETE in BayesTraits that allows all possible forward and reverse transitions between the states of each binary dimorphism, assuming that transitions involving simultaneous change in both dimorphisms do not occur, and hence generating eight possible transitions between the categories that differ in only one dimorphism state (Figs. 5-6). We ran a RJ MCMC chain for 5,050,000 iterations, with a burn-in period of 50,000 iterations, after which the chain was sampled every 100th iteration. We specified exponential priors seeded from a hyperprior with a uniform distribution of 0-30. In order to achieve median acceptances between 15% and 40% of the rate parameter proposals, we used rate deviations of 12 for all C4A analyses, and 2 for all F4 analyses. We also repeated each run three times to check whether the harmonic means were stable. For each trait and working phylogeny, we ran the RJ MCMC chain with: (i) a dependent model, where transitions in male dimorphism depended on the state of sexual dimorphism (and vice-versa), and (ii) an independent model, where transitions of the two dimorphisms were mutually independent. We later compared these models on the basis of Bayes Factors (BFs), which are two times the difference in the marginal likelihoods of the best-fit and worse-fit models. These marginal likelihoods were approximated by the harmonic means from the final iteration of the RJ MCMC runs. Typically a BF > 2 supports, and a BF > 5 or more strongly supports, the best-fit model (Pagel and Meade 2006). Finally, we explored the dependent model, examining the posterior distributions of the transition parameters (named qij, for transitions from category i to category j), extracting their mean and standard deviation, and quantifying the frequency with which each of them was set to zero (Z) in the dependent model RJ MCMC chain. As a rule of thumb, transitions are probable events when Z < 10%, and improbable events otherwise (as in Higginson et al. 2012a). However, looking purely at these Z values may cause us to consider some transitions as improbable events due to low statistical power when using a small phylogeny (Fitzpatrick et al. 2009). This can be avoided by combining the two metrics (Z and qij), and considering transition parameters with Z > 10% to represent marginal events if their mean qij is higher than that of the probable event with the lowest qij.

116!! Table 5.2 Summary of our analyses to detect sexual and male dimorphism in 48 species of gonyleptid harvestmen (from 10 subfamilies). Results are for both traits analyzed (length of the apophysis on the fourth coxa, C4A; length of the fourth femur, F4). We considered a trait to be sexually dimorphic (SD) when only males expressed it, or when sex and/or the interaction between sex and dorsal scute length (proxy for body size) explained C4A or F4 length variation in a general linear model (P-values for these effects are presented, and are in bold when < 0.05). When a trait was also considered male dimorphic (see below), we compared females to major males only, in order to avoid confounding scores of male and sexual dimorphism. We considered a trait to be male dimorphic (MD) if the model that described a mixture of two normal distributions for the trait in males had a lower BIC (ΔBIC > 2 marked in bold, see Methods for a description of BIC) than the model that described one single normal distribution for the same trait. “Inference from literature” means that only a few individuals of that species were examined, and information from the literature was also consulted to diagnose sexual and male dimorphism; this usually occurred in species that clearly lack sexual and male dimorphism, and the sex of individuals can hardly be identified based on external morphology alone.

Sexual dimorphism Male dimorphism Trait Sex*Body size BIC 1 BIC 2 Species Trait presence Sex effect interaction SD mode modes ΔBIC MD Bourguyiinae

Bourguyia trochanteralis C4A ♂♀ < 0.001 0.0562 Yes 31.34 33.73 -2.39 No F4 ♂♀ < 0.001 < 0.001 Yes 371.14 379.56 -8.42 No Asarcus ingenuus C4A ♂♀ < 0.001 0.0113 Yes 14.19 22.72 -8.53 No F4 ♂♀ < 0.001 < 0.001 Yes 114.35 121.99 -7.64 No Caelopyginae

Ampheres leucopheus C4A ♂♀ < 0.001 0.6968 Yes -11.07 -7.34 -3.73 No F4 ♂♀ < 0.001 0.5990 Yes 61.11 62.78 -1.67 No Arthrodes xanthopygus C4A ♂ - - Yes 308.37 303.82 4.55 Yes F4 ♂♀ 0.1710 0.5729 No 136.68 148.53 -11.85 No Caelopygus elegans C4A ♂♀ < 0.001 0.9044 Yes 83.48 59.54 23.94 Yes

! 117! F4 ♂♀ 0.0024 0.2091 Yes 48.80 56.64 -7.84 No Metarthrodes pulcherrimus C4A ♂ - - Yes -10.45 -7.67 -2.77 No F4 ♂♀ < 0.001 0.0153 Yes 86.61 90.97 -4.36 No Pristocnemis albimaculatus C4A ♂♀ < 0.001 0.0002 Yes -2.90 1.21 -4.11 No F4 ♂♀ < 0.001 0.0079 Yes 65.42 67.95 -2.53 No Pristocnemis pustulatus C4A ♂♀ < 0.001 0.0007 Yes 15.86 22.47 -6.61 No F4 ♂♀ 0.0041 0.8744 Yes 97.39 103.91 -6.52 No Thereza speciosa C4A - - - No - - - No F4 ♂♀ Inference from literature No Inference from literature No Cobaniinae

Cobania picea C4A ♂♀ < 0.001 0.0181 Yes 289.43 278.26 11.17 Yes F4 ♂♀ < 0.001 0.0246 Yes 193.86 194.88 -1.02 No Goniosomatinae

Acutisoma longipes C4A ♂ - - Yes 70.40 75.93 -5.53 No F4 ♂♀ < 0.001 0.2890 Yes 184.41 187.80 -3.39 No Goniosoma calcar C4A ♂♀ < 0.001 < 0.001 Yes 49.79 58.37 -8.58 No F4 ♂♀ < 0.001 0.0769 No1 53.75 57.22 -3.48 No Goniosoma roridum C4A ♂♀ < 0.001 0.0586 Yes 27.23 36.14 -8.91 No F4 ♂♀ < 0.001 0.7152 No1 83.27 90.98 -7.71 No Goniosoma varium C4A ♂♀ < 0.001 0.0033 Yes 86.99 77.21 9.78 Yes

118!! F4 ♂♀ 0.0321 0.9057 No1 100.59 107.17 -6.58 No Goniosoma venustum C4A ♂ - - Yes 41.40 22.32 19.08 Yes F4 ♂♀ 0.4660 0.5798 No 53.13 53.29 -0.15 No Heteromitobates discolor C4A ♂ - - Yes 127.65 103.03 24.62 Yes F4 ♂♀ 0.0003 0.4407 Yes 131.15 130.54 0.61 No Mitogoniella taquara C4A ♂ - - Yes 44.41 51.78 -7.37 No F4 ♂♀ < 0.001 < 0.001 Yes 384.15 390.94 -6.78 No Serracutisoma proximum C4A ♂ - - Yes 47.38 53.26 -5.88 No F4 ♂♀ < 0.001 0.1957 Yes 1482.24 1431.59 50.65 Yes Gonyleptinae

Gonyleptellus cancelatus C4A ♂ - - Yes 103.32 109.51 -6.18 No F4 ♂♀ < 0.001 0.0225 Yes 252.83 247.33 5.50 Yes Gonyleptes horridus C4A ♂♀ < 0.001 < 0.001 Yes 140.77 142.98 -2.21 No F4 ♂♀ < 0.001 0.0032 Yes 118.69 119.44 -0.74 No Neosadocus maximus C4A ♂ - - Yes 86.63 78.63 7.99 Yes F4 ♂♀ 0.4905 0.7742 No 117.82 124.28 -6.46 No Hernandariinae

Acrogonyleptes rhinoceros C4A ♂ - - Yes 18.75 16.36 2.39 Yes F4 ♂♀ Inference from literature No Inference from literature No Hernandaria una C4A ♂ - - Yes 20.73 19.25 1.48 No

! 119! F4 ♂♀ 0.7205 0.9350 No 24.22 30.00 -5.78 No Multumbo dimorphicus C4A ♂ - - Yes 47.58 41.40 6.19 Yes F4 ♂♀ < 0.001 0.9542 Yes 58.32 62.89 -4.57 No Multumbo terrenus C4A - - - No - - - No F4 ♂♀ Inference from literature No Inference from literature No Piassagera brieni C4A - - - No - - - No F4 ♂♀ Inference from literature No Inference from literature No Pseudotrogulus funebris C4A - - - No - - - No F4 ♂♀ Inference from literature No Inference from literature No Heteropachylinae

Magnispina neptunus C4A ♂♀ < 0.001 0.0063 Yes -56.46 -50.55 -5.91 No F4 ♂♀ 0.0083 0.0675 No1 33.00 43.18 -10.17 No Pseudopucrolia mutica C4A ♂♀ < 0.001 0.0259 Yes -99.67 -87.94 -11.74 No F4 ♂♀ 0.8722 0.0589 No 100.19 100.86 -0.67 No Mitobatinae

Discocyrtoides nigricans C4A ♂ - - Yes 32.96 35.76 -2.80 No F4 ♂♀ < 0.001 < 0.001 Yes 458.83 464.65 -5.82 No Encheiridium ruschii C4A ♂♀ < 0.001 0.7640 Yes -53.50 -47.71 -5.79 No F4 ♂♀ < 0.001 0.0118 Yes 155.84 161.31 -5.47 No Ischnoterus tenebrosus C4A - - - No - - - No

120!! F4 ♂♀ < 0.001 0.9929 Yes 210.12 203.71 6.41 Yes Longiperna concolor C4A ♂ - - Yes 10.84 18.00 -7.16 No F4 ♂♀ < 0.001 Yes 4809.39 4742.98 66.41 Yes Metamitobates squalidus C4A - - - No - - - No F4 ♂♀ < 0.001 0.1288 Yes 436.16 437.80 -1.64 No Mitobates triangulus C4A - - - No - - - No F4 ♂♀ < 0.001 0.0006 Yes 389.73 397.41 -7.68 No Mitobatula castanea C4A - - - No - - - No F4 ♂♀ < 0.001 0.0171 Yes 180.80 187.85 -7.05 No Neoancistrotus gracilis C4A ♂♀ < 0.001 < 0.001 Yes 49.50 48.24 1.25 No F4 ♂♀ < 0.001 < 0.001 Yes 291.85 290.23 1.62 No Promitobates bellus C4A ♂ - - Yes 54.90 32.86 22.04 Yes F4 ♂♀ < 0.001 0.4793 Yes 369.37 360.38 8.99 Yes Promitobates ornatus F4 ♂♀ < 0.001 0.0082 Yes 1799.24 1743.44 55.80 Yes Ruschia maculata C4A - - - No - - - No F4 ♂♀ < 0.001 0.9799 Yes 362.08 356.52 5.57 Yes Progonyleptoidellinae

Cadeadoius niger C4A ♂ - - Yes 113.50 118.89 -5.39 No F4 ♂♀ < 0.001 0.6103 Yes 167.08 175.28 -8.19 No Deltaspidium tenue C4A ♂♀ < 0.001 0.0035 Yes 50.59 50.45 0.14 No F4 ♂♀ < 0.001 0.1766 Yes 82.96 88.54 -5.57 No

! 121! Heliella singularis C4A ♂ - - Yes 44.62 45.77 -1.15 No F4 ♂♀ 0.1216 0.8597 No 123.39 133.12 -9.73 No Iguapeia melanocephala C4A ♂♀ < 0.001 0.9431 Yes -1.87 4.78 -6.65 No F4 ♂♀ 0.5581 0.7200 No 62.69 69.05 -6.36 No Iporangaia pustulosa C4A - - - No - - - No F4 ♂♀ Inference from literature No 583.26 593.73 -10.47 No Mitopernoides variabilis C4A - - - No - - - No F4 ♂♀ Inference from literature No Inference from literature No Progonyleptoidellus striatus C4A ♂♀ Inference from literature Yes 103.05 91.29 11.76 Yes F4 ♂♀ Inference from literature Yes Inference from literature No Sodreaninae

Sodreana barbiellini C4A ♂ - - Yes 52.96 61.23 -8.28 No F4 ♂♀ < 0.001 0.0354 Yes 93.53 90.68 2.85 Yes

*Indicates species where F4 was longer in females than in males; we classified these species with such ‘reversed’ sexual dimorphism in F4 length together with the sexually monomorphic species (see Results for a comprehensive explanation of our reasons to do so).

122!!

Figure 5.2 Two examples of the allometric relationship between the length of the apophysis on the fourth coxa (C4A) and dorsal scute length (indicative of body A size). (A) Whereas there is clear

10 sexual dimorphism and male 10 ● ● ● ● ● ● dimorphism for C4A in Cobania ● ●●● ● α ● ● ● picea, (B) there is sexual ●

● ● 8 8 ● dimorphism but no male ●●● ● ● ● ● ● dimorphism for C4A in Ampheres ●● length (mm) (mm) length ● ● leucopheus (see also Table 5.2). ● ● ● 6

6 The histograms on the right of ●● ● each plot depict the distribution ESP (mm) ● ●

apophysis ● of C4A in males of each species, ● ●

●● 4 4 4 β and are overlaid by a non- ● ● ● ● ● parametric probability density

Coxae IV spine (mm) Coxae ● ● ● ●

Coxae ●● ● ● curve from a kernel density ● 2 2 ● estimator. Filled black circles ● ●● ● ●●●●● ●● ●●●●●● ●● ● ● ●● and black lines indicate males ● ● 0.00 0.05 0.10 0.15 with a probability (estimated 4 6 8 10 density 12frequency Relative with finite mixture models, see Methods) of being from the DorsalDorsal scute scute length length (mm) (mm) major morph (α) higher than 99% (or all males in A. leucopheus). Empty black circles and dashed lines indicate males with a probability of being from the minor morph (β) higher than 99%. Empty gray circles indicate

4.5 males with probabilities lower B than 99% of being from either ● morph, and filled gray circles 4.0

● ●● 4.0 ● ●● ●● ●● ● ● and dotted lines indicate females. ●● ● ● ● ● ● ● ● ● Linear models were fitted ● Y (mm) ● 3.5

● ● 3.5 through standard major axis ● ● regression, and axes are

length (mm) (mm) length isometric to show male morphs 3.0 3.0 in the most objective fashion. ! 2.5 apophysis ! 4

Coxae IV apophysis (mm) IV apophysis Coxae ● ● ! ● ! 2.0 ● ● ● Coxae ● ● ● ● ● ● ● ! ● ● ● ● ! ● ● 1.5 ! 4.0 4.5 5.0 5.5 6.0

Dorsal scuteDorsal length scute length (mm) (mm)

! 123! 5.4 Results

Sexual and male dimorphism in C4A length C4A was completely absent in both males and females of 11 species of our sample, which we therefore classified as sexually monomorphic for this trait (Table 5.2; Figure 5.1G). In another 18 species, C4A was present only in males, whereas females either completely lacked the apophysis, or only had a very rudimentary tubercle in its place (Figs. 1D-E). We classified these species as sexually dimorphic for this trait (Table 5.2). In the remaining 18 species, both sexes presented measurable C4A (Figure 5.1F), and the effect of sex and/or the interaction between sex and dorsal scute length significantly explained C4A length variation in all of them (Table 5.2). Therefore we also classified these species as sexually dimorphic for this trait. For the species that were later found to present male dimorphism in C4A length (see below), we compared females to major males only, in order to avoid confounding scores of male and sexual dimorphism (following Emlen et al. 2005). This is important because when a species is male dimorphic, it can only be sexually monomorphic if females are compared to only one of the male morphs. We therefore defined sexual dimorphism as differential expression of C4A length between females and majors, in theory allowing all combinations of monomorphisms and dimorphisms (male dimorphism without sexual dimorphism would be possible when minors avoid the production of a long C4A that is present in majors and females). However, in all cases of male dimorphism for C4A length, this trait was also significantly different between females and majors, and the species were hence classified as sexually dimorphic for C4A length (Table 5.2). As a consequence, 76.6% of the species in our sample were sexually dimorphic for C4A length, which seems to be the ancestral state for this trait (proportional likelihoods between 0.989 and > 0.999 under our different working phylogenies; Table 5.3). Moreover, losses of sexual dimorphism in C4A length have probably occurred between four and six times, and re-gains of this dimorphism either never occurred, or occurred only once or twice (Table 5.3). C4A was entirely absent in 11 out of 47 species, which we hence classified as male monomorphic for this trait (Table 5.2; Figure 5.1G). In males of the remaining 38 species, C4A length varied greatly both among males of different species and among conspecific males (Figs. 1A-B). The variances of this trait ranged from 26.7% less (in Pseudopucrolia mutica) to 2,945.1% more (in Arthrodes xanthopygus, see Figs. 1D and 5) than that of dorsal scute length (variances standardized by the means of the traits).

124!!

Figure 5.3 Mapping the evolution of sexual dimorphism and male dimorphism in the length of the apophysis on the fourth coxa in 47 species of gonyleptid harvestmen. (A) The first phylogeny was based on ecological, behavioral, and chemical characters (Caetano and Machado in press); (B) the second phylogeny was based on morphological characters (Pinto-da-Rocha 2002). For both phylogenies, the topology of Progonyleptoidellinae comes from Pinto-da-Rocha (in prep). We also used a strict consensus of these two phylogenies in our analyses (not shown). We reconstructed the evolution of sexual and male dimorphisms through a likelihood reconstruction method in a Markov, k-status, 1-parameter model with four states. Each state was a combination of sexual dimorphism (present or absent) and male dimorphism (present or absent): a species could be sexually and male monomorphic (state 0, white), sexually dimorphic and male monomorphic (state 1, gray), sexually monomorphic and male dimorphic (state 2, not shown because no species filled this category), or sexually and male dimorphic (state 3, black). Pie charts indicate the probability of the states in each node.

! 125! Moreover, when modelling the distributions of C4A length in males, we found that in 11 species the model describing a mixture of two normal distributions had a significantly lower BIC than the model describing a single normal distribution for the trait. We therefore considered these species to be male dimorphic for C4A length (23.4% of the species in our sample; Table 5.2). The ancestral state for this trait seems to be male monomorphism, but the support for this hypothesis is not evenly strong across our working phylogenies (proportional likelihoods between 0.683 and 0.989; Table 5.3). Gains of male dimorphism in C4A length may have occurred 10 or 11 independent times, whereas secondary losses of this dimorphism either never occurred, or occurred only once (Table 5.3).

Sexual and male dimorphism in F4 length Neither sex nor the interaction between sex and dorsal scute length significantly explained F4 length variation in 14 of the species we sampled, and we therefore considered them sexually monomorphic for this trait (Table 5.2). On the other hand, the interaction between sex and dorsal scute length and/or the effect of sex significantly explained F4 length variation in the remaining 34 species. However, in four of these species it was the females that had significantly longer F4s. We only considered 30 species to be significantly sexually dimorphic for F4 length (Table 5.2), and categorized the species with this ‘reversed’ sexual dimorphism in F4 length together with the sexually monomorphic species. The reasoning for this is that our main goal was to investigate the potential coevolution of male dimorphism and sexual dimorphism, which is in theory expected when a trait is under stronger sexual selection in males, and this is probably not the case for the four species with ‘reversed’ sexual dimorphism. If, on the other hand, some degree of sex role reversal and stronger sexual selection in females was responsible for the ‘reversed’ sexual dimorphism of these species, they would actually be good candidates for investigating putative female dimorphisms. However, we do not believe that to be the case for two reasons. First, in the four species with ‘reversed’ sexual dimorphism for F4 length, other forms of sexual dimorphism were present in which males had significantly longer apophyses (including C4A, Table 5.2) and other armaments, indicating that sex role reversal is improbable. Second, in these four species, F4 was evidently thicker and/or more curved in males, which probably explains why it was also shorter in males. For the species that were later found to also present male dimorphism in F4 length, we compared females to majors only, in order to avoid confounding scores of male and sexual dimorphism (for the same reasons explained above for C4A). In all

126!!

Figure 5.4 Mapping the evolution of sexual dimorphism and male dimorphism in the length of the fourth femur in 48 species of gonyleptid harvestmen. (A) The first phylogeny was based on ecological, behavioral, and chemical characters (Caetano and Machado in press); (B) the second phylogeny was based on morphological characters (Pinto-da-Rocha 2002). For both phylogenies, the topology of Progonyleptoidellinae comes from Pinto-da-Rocha (in prep). We also used a strict consensus of these two phylogenies in our analyses (not shown). We reconstructed the evolution of sexual and male dimorphisms through a likelihood reconstruction method in a Markov, k-status, 1-parameter model with four states. Each state was a combination of sexual dimorphism (present or absent) and male dimorphism (present or absent): a species could be sexually and male monomorphic (state 0, white), sexually dimorphic and male monomorphic (state 1, gray), sexually monomorphic and male dimorphic (state 2, not shown because no species filled this category), or sexually and male dimorphic (state 3, black). Pie charts indicate the probability of the states in each node.

! 127! these cases, F4 length was significantly different between females and majors, and we hence classified the species as sexually dimorphic for F4 length (Table 5.2). In summary, 62.5% of species in our sample were sexually dimorphic for F4 length, which seems to be the ancestral state for this trait (proportional likelihoods between 0.976 and > 0.999 under our different working phylogenies; Table 5.3). Losses of sexual dimorphism in this trait might have occurred between six and seven times, and re-gains of this dimorphism have probably occurred two or three times (Table 5.3). F4 length was also fairly variable among conspecific males, since the variances for this trait ranged from being 2.4% (in Hernandaria una) to 19,971.5% more (in Longiperna concolor) than that of dorsal scute length (variances standardized by the means of the traits). Furthermore, we considered seven species to be male dimorphic for F4 length because the model that described a mixture of two normal distributions for F4 length in males had a significantly lower BIC than the one that described a single normal distribution for this trait (Table 5.2). As a result, 14.6% of species in our sample presented male dimorphism in F4 length, while the ancestral state for this trait is probably male monomorphism (proportional likelihoods between 0.976 and > 0.999; Table 5.3). Repetitive gains of male dimorphism in F4 length occurred six or seven independent times, whereas secondary losses of this dimorphism either never occurred, or occurred only once (Table 5.3).

Coevolution of sexual and male dimorphism We found strong support for the correlated evolution of male dimorphism and sexual dimorphism (BFs ranging from 7.76 to 13.71; Table 5.3). This support was homogenously strong across our three different working phylogenies, and for both C4A and F4. Table 5.4 summarizes all relative transition rates between monomorphism and dimorphism in C4A and F4 lengths, as estimated by our dependent model of sexual and male dimorphism evolution. The results are presented and interpreted for our three working phylogenies. In both traits the combination of sexual dimorphism and male monomorphism seems to be the ancestral state (Table 5.3). For C4A length, in the presence of sexual dimorphism, gains of male dimorphism (q34; Z ≤ 1.6%) and secondary losses of it (q43; Z ≤ 0.5%) were always estimated to be probable. All the remaining transitions were estimated to be only marginal or improbable. However, losses of sexual dimorphism must have occurred, and the relatively greater size of transition rate q31 (along with Z ≤ 20%) when compared to transition rate q42 (along with Z ≥ 68.2%) indicates that they have probably occurred in the absence of male dimorphism (Table 5.4).

128!! Table 5.3 Summary of our comparative analyses for the evolution of sexual dimorphism (SD), male dimorphism (MD) and their coevolution in gonyleptid harvestmen. We analysed two morphological traits (length of the apophysis on the fourth coxa, C4A; length of the fourth femur, F4), and show results for three working phylogenies: one based on ecological, behavioral, and chemical characters; one based on morphological characters; and a strict consensus of the previous two. We determined the most likely ancestral states through likelihood reconstruction (proportional likelihoods in parentheses), and the range in numbers of losses and gains through parsimony. For the analyses of correlated evolution of SD and MD, we present the harmonic means (from the final iteration of the RJ MCMC runs) for the dependent model, where transitions in MD depended on the state of SD (and vice-versa); and for the independent model, where transitions of SD and MD were mutually independent. We compared these models through Bayes Factors (BFs): BFs shown here (all ≥ 5) strongly support the co-evolution of SD and MD.

Working phylogenies Type of Trait Inferences from phylogenies Ecological, behavioral dimorphism Morphological Consensus & chemical Most likely ancestral state Dimorphism (0.999) Dimorphism (0.989) Dimorphism (> 0.999) SD Number of losses of SD 4 – 6 4 – 6 4 – 6 Number of re-gains of SD 0 – 2 0 – 2 0 – 2 Coxae 4 Most likely ancestral state Monomorphism (0.683) Monomorphism (0.855) Monomorphism (0.989) apophysis MD Number of gains of MD 10 – 11 / 10 – 11 10 – 11 / 10 – 11 10 – 11 / 10 – 11 (C4A) Number of losses of MD 0 – 1 / 0 – 1 0 – 1 / 0 – 1 0 – 1 / 0 – 1 Harmonic mean - independent -54.932 -53.672 -55.255 Correlated Harmonic mean - dependent -49.920 -49.793 -49.730 evolution Bayes factor 10.03 7.76 11.05 Most likely ancestral state Dimorphism (0.717) Dimorphism (0.942) Dimorphism (0.997) SD Number of losses of SD 6 – 7 / 6 – 7 6 – 7 / 6 – 7 6 – 7 / 6 – 7 Number of re-gains of SD 2 – 3 / 2 – 3 2 – 3 / 2 – 3 2 – 3 / 2 – 3 Most likely ancestral state Monomorphism (0.999) Monomorphism (0.976) Monomorphism (> 0.999) Femur 4 MD Number of gains of MD 6 – 7 / 6 – 7 7 / 7 7 / 7 (F4) Number of losses of MD 0 – 1 / 0 – 1 0 / 0 0 / 0 Harmonic mean - independent -55.659 -55.725 -55.773 Correlated Harmonic mean - dependent -48.802 -50.349 -49.475 evolution Bayes factor 13.71 10.75 12.6

! 129!

Figure 5.5 Illustration of the evolutionary transitions of sexual dimorphism (SD) and male dimorphism (MD) in the length of the apophysis on the fourth coxa (C4A; shaded) of gonyleptid harvestmen, inferred from the posterior distributions of probabilities estimated (with RJ MCMC) by our dependent model of SD and MD evolution. Probable transitions (Z < 10%) are depicted by thick black arrows, marginal transitions (Z > 10%, but qij higher than that of the lowest probable event) are depicted by thin black arrows, and improbable transitions (10% ≤ Z ≤ 70%) are depicted by thin gray arrows. Dotted arrows represent transitions for which the probability estimates were not consistent across our three working phylogenies (see Table 5.4). Very improbable (Z ≥ 70%) transitions were removed from the figure. Upper center: Thereza speciosa (upper pair) and a hypothetical species (lower pair) illustrate the lack of SD and MD; center left: Metarthrodes pulcherrimus (upper pair) and Ampheres leucopheus (lower pair) illustrate SD without MD; lower center: Arthrodes xanthopygus (upper trio) and Cobania picea (lower trio) illustrate SD with MD; center right: a hypothetical species illustrates MD without SD. Sample sizes (number of species) are given for each of these combinations of SD and MD.

130!!

Figure 5.6 Illustration of the evolutionary transitions of sexual dimorphism (SD) and male dimorphism (MD) in the length of the fourth femur (F4; shaded) of gonyleptid harvestmen, inferred from the posterior distributions of probabilities estimated (with RJ MCMC) by our dependent model of SD and MD evolution. Probable transitions (Z < 10%) are depicted by thick black arrows, marginal transitions (Z > 10%, but qij higher than that of the lowest probable event) are depicted by thin black arrows, and improbable transitions (10% ≤ Z ≤ 70%) are depicted by thin gray arrows. Dotted arrows represent transitions for which the probability estimates were not consistent across our three working phylogenies (see Table 5.4). Very improbable (Z ≥ 70%) transitions were removed from the figure. Upper center: Pseudopucrolia mutica illustrates the lack of SD and MD; center left: Metarthrodes pulcherrimus illustrates SD without MD; lower center: Promitobates bellus illustrates SD with MD; center right: a hypothetical species illustrates MD without SD. Sample sizes (number of species) are given for each of these combinations of SD and MD.

! 131! For F4 length, gains of male dimorphism in the presence of sexual dimorphism (q34; 8.5% < Z < 16.5%) are more probable than in its absence (q12; Z ≥ 89.5%). Secondary losses of male dimorphism were also more probable in the presence of sexual dimorphism (q43; Z ≤ 4.8%) than in its absence (q21; 19.2% < Z < 22.2%). Losses and secondary re-gains of sexual dimorphism were more probable in the absence of male dimorphism (q31; 5.1% < Z < 19.9% and q13; Z ≤ 2.5%, respectively), than in its presence (q42; Z ≥ 76.9% and q24; 22.2% < Z < 24.2%, respectively).

5.5 Discussion

Our study demonstrates correlated evolution of sexual dimorphism and male dimorphism in gonyleptid harvestmen, and also indicates that sexual dimorphism consistently precedes male dimorphism in this major arachnid group. These findings apply to two different morphological traits usually employed as weapons in male-male fights, and are robust to phylogenetic uncertainty. Below we build upon these results, suggesting that the coevolution of sexual and male dimorphism is probably a consequence of a similar selection regime against secondary sexual traits in small males and females, and that this phenomenon may be widespread across many animal groups.

Table 5.4 Summary of the relative rates of all evolutionary transitions of sexual dimorphism (SD) and male dimorphism (MD) in gonyleptid harvestmen. For each transition, we present the mean ± standard deviation of the posterior probabilities estimated (with RJ MCMC) by our dependent model of SD and MD evolution. We also present the frequency with which each of them was set to zero (Z) in the RJ MCMC chain (percentages in parentheses). Following Fitzpatrick et al. (2009), we considered transitions to be probable events when they had Z < 10%, marginal events when their mean qij was higher than that of the lowest probable event (that had Z < 10%), and improbable events otherwise (very improbable events had Z ≥ 70%). We performed the analyses for two morphological traits (length of the apophysis on the fourth coxa, C4A; length of the fourth femur, F4), and show results for our three working phylogenies: one based on ecological, behavioral, and chemical characters; another one based on morphological characters; and a strict consensus of these two phylogenies. In the last column we present the inferences about how probable each transition can be, with the number of phylogenies that support that inference in parentheses.

132!! Working phylogenies Transition Evolutionary transition Ecological, behavioral & Morphological Consensus likelihood inference chemical Coxae 4 apophysis Gain of SD in absence of MD (q13) 9.81 ± 17.85 (17.9%) 10.25 ± 17.74 (17.4%) 11.31 ± 18.52 (15.4%) Improbable (3) Gain of SD in presence of MD (q24) 10.79 ± 17.98 (20.1%) 10.60 ± 16.14 (20.6%) 11.66 ± 17.30 (20.0%) Improbable (3) Loss of SD in absence of MD (q31) 6.28 ± 12.55 (20.0%) 6.45 ± 12.17 (19.2%) 7.07 ± 12.55 (15.9%) Improbable (3) Loss of SD in presence of MD (q42) 0.31 ± 1.83 (68.2%) 0.23 ± 1.48 (70.1%) 0.27 ± 2.37 (72.8%) Very improbable (3) Gain of MD in absence of SD (q12) 0.31 ± 2.17 (71.6%) 0.19 ± 1.40 (74.5%) 0.18 ± 1.21 (76.1%) Very improbable (3) Gain of MD in presence of SD (q34) 10.85 ± 14.78 (1.6%) 12.16 ± 15.51 (0.2%) 13.00 ± 15.54 (0.1%) Probable (3) Loss of MD in absence of SD (q21) 11.57 ± 17.72 (19.7%) 11.67 ± 17.08 (19.5%) 11.59 ± 16.87 (20.9) Marginal (1) Loss of MD in presence of SD (q43) 14.68 ± 18.79 (0.5%) 16.26 ± 19.10 (0.0%) 17.62 ± 19.63 (0.0%) Probable (3) Femur 4 Gain of SD in absence of MD (q13) 2.03 ± 8.58 (1.5%) 7.32 ± 17.42 (2.5%) 4.34 ± 15.48 (1.5%) Probable (3) Gain of SD in presence of MD (q24) 2.66 ± 9.01 (23.0%) 6.05 ± 15.30 (22.2%) 4.27 ± 13.82 (24.2%) Marginal (1) Loss of SD in absence of MD (q31) 1.96 ± 8.54 (19.9%) 7.26 ± 17.31 (5.1%) 4.26 ± 15.42 (17.0%) Probable (1) Loss of SD in presence of MD (q42) 0.14 ± 1.17 (76.9%) 0.25 ± 2.27 (82.0%) 0.23 ± 2.38 (77.0%) Very improbable (3) Gain of MD in absence of SD (q12) 0.04 ± 0.37 (91.6%) 0.11 ± 0.96 (91.4%) 0.06 ± 0.61 (89.5%) Very improbable (3) Gain of MD in presence of SD (q34) 1.81 ± 6.64 (8.5%) 4.61 ± 11.14 (16.5%) 3.75 ± 11.48 (11.4%) Probable (1) Loss of MD in absence of SD (q21) 3.07 ± 9.20 (22.2%) 6.91 ± 15.82 (19.2%) 5.03 ± 14.35 (22.2%) Marginal (2) Loss of MD in presence of SD (q43) 2.98 ± 10.04 (4.8%) 8.32 ± 18.30 (3.2%) 5.97 ± 17.22 (4.0%) Probable (3)

! 133! Weaponry diversity in gonyleptid harvestmen The apophysis of the fourth coxa (C4A) in gonyleptids can vary among species in presence, relative size, general shape, and orientation (Figs. 1 and 5). This apophysis grows from the first leg segment (the coxa), but its length is not necessarily correlated with the length of the fourth pair of legs in general. Nevertheless, the fourth pair of legs in gonyleptids can also present varying degrees of elongation, which is usually very evident in the femur (F4). This trait can vary in its relative size, curvature, and degree of armature (Figs. 1 and 6). Here, for heuristic purposes, we focused in investigating sexual and male dimorphisms simply in the length of C4A and F4. Although C4A and F4 are employed as weapons in male-male fights, their functions in these fights are fundamentally different. C4A is always used in battles of strength, nipping the adversary and functioning as pliers (e.g., Nazareth and Machado 2009; 2010; Willemart et al. 2009). F4 can also be used in this manner in some species where this segment bears spines and is thickened and curved, which is clearly the case of Neosadocus maximus (Willemart et al. 2009) and Magnispina neptunus (Nazareth and Machado 2010). Indeed, the curving and thickening of F4, as well as the presence of spines, probably causes the length of this trait per se to lose its importance in a fight, explaining why males of N. maximus and M. neptunus do not have longer F4s than females. This whole study could be repeated focusing on the curvature of F4 and the presence of spines. However, only nine species in our sample present such modifications, and an analysis focused on them would be less powerful. The femur of the fourth leg (F4), however, is not always used as a weapon of strength in harvestmen fights. In all representatives of the subfamily Mitobatinae included in our sample, this trait is hugely elongated, but does not vary in thickness, curvature, or presence of spines, either between or within the sexes (Figs. 1H-I). Behavioral observations show that these straight elongated F4s are used as whips when fighting males turn their backs to each other, keep the fourth pair of legs widely opened (with the right leg forming a 180º angle with the left), and strike each other with the tips of the fourth legs (Machado and Macías-Ordóñez 2007; Zatz et al. 2011). In these cases, rivals seem to measure each other, using the length of this structure as an indication of size/status, and settling the outcome of fights based on differences in these structures between rivals. Thus, in Mitobatinae harvestmen F4 probably functions more as a signal of male size and therefore strength and fighting ability than as an actual weapon, similarly to the long eye stalks of stalk-eyed flies (David et al. 1998; Panhuis and Wilkinson 1999). This never seems to be the case with C4A, but despite this difference, our results on the correlated evolution of sexual and male

134!! dimorphism are consistent for C4A and F4.

Correlated evolution between sexual and male dimorphism As far as we know, our study with gonyleptid harvestmen demonstrates for the first time that male dimorphism can be more evolutionarily labile than sexual dimorphism. In C4A length, gains of male dimorphism (ancestrally monomorphic) occurred approximately twice as many times as losses of sexual dimorphism, which is the ancestral condition in the family (Table 5.3). On the other hand, gains of male dimorphism in F4 length (ancestrally monomorphic) and losses of sexual dimorphism (ancestrally dimorphic) probably occurred a similar number of times. Our analyses of correlated evolution agree with these results. For C4A evolution, the loss of sexual dimorphism (in the absence of male dimorphism, when it was more probable) was set to zero in 18.4% of the iterations on average. Meanwhile, the gain of male dimorphism (in the presence of sexual dimorphism, when it was more probable) was only set to zero in 0.6% of the iterations on average, meaning that it changed much more frequently than sexual dimorphism. This trend was similar for F4 length, but in a less pronounced fashion: the loss of sexual dimorphism (in the absence of male dimorphism) was set to zero in 14.0% of the iterations on average, whereas the gain of male dimorphism (in the presence of sexual dimorphism) was set to zero in 12.1% of the iterations on average. The higher lability of male dimorphism, when compared to sexual dimorphism, suggests that the former evolves faster than the latter, but it can not be taken as evidence of their correlated evolution or of which one evolved earlier in the phylogenies. On the other hand, the fact that male dimorphism never occurred in sexually monomorphic species for either C4A or F4 points towards a correlation between these two types of dimorphism. At first, this may seem obvious because females of male dimorphic species could not be morphologically similar to males of both morphs at the same time, unless females were also dimorphic, which is rare generally and not yet observed in arachnids. However, as explained in the methods, we compared females of male dimorphic species only with majors, allowing species to fall in the category that combined male dimorphism with sexual monomorphism. In comparison, this combination existed for the horns of dung beetles, where the combination that did not occur was that of sexual dimorphism and male monomorphism (Emlen et al. 2005). Male dimorphism and sexual dimorphism are so strongly correlated in dung beetles that these dimorphisms were either both present or both absent in 97% of all dung beetle species studied (Emlen et al. 2005). This very tight correlation between the types

! 135! of dimorphisms in dung beetles probably results from the endocrine threshold mechanism that can switch horn growth on and off during development (Emlen et al. 2006; 2012; Shingleton et al. 2007), and that seems to be present in both small males and in females of all sizes (Emlen et al. 2005). In gonyleptid harvestmen, we found strong support for the correlated evolution of sexual and male dimorphism, as indicated by the Bayes Factors > 5 for all comparisons between our independent and dependent models of sexual dimorphism and male dimorphism evolution. Nevertheless whereas male dimorphism always occurred with sexual dimorphism, sexual dimorphism occurred both with and without male dimorphism in both traits that we analyzed. Indeed, the combination of sexual dimorphism and male monomorphism, non-existent in Emlen et al.’s (2005) study of dung beetles, was the most common state in our study, containing 53% and 45% of species for C4A and F4, respectively. This might reflect a difference in the proximate mechanisms behind male and sexual dimorphisms in Harvestmen that contrasts with dung beetles. If this is the case, similarity between the proximate mechanisms of sexual and male dimorphism is not necessary for their correlated evolution (although it may favor the process). Hence, correlated evolution of sexual and male dimorphisms is likely the result of selection acting against secondary sexual traits in females and small males, generating intralocus sexual conflict (Rice and Chippindale 2001) and a similar intralocus conflict among males of different status (West-Eberhard 2003) at the same time, and selecting for sexual and male dimorphisms through different proximate mechanisms. Thus, investigating these proximate mechanisms in small males and in females of sexually and male dimorphic harvestmen is a promising avenue for better understanding the coevolution of such dimorphisms in non-holometabolous arthropods.

Directionality of the correlation between sexual and male dimorphism What about the directionality of the correlation between sexual and male dimorphism; is male dimorphism following or preceding sexual dimorphism? Are the mechanisms of male dimorphism co-opted from previously existing mechanisms of sexual dimorphism, or vice versa? Even if such a co-option does not exist in harvestmen (as we suggest above), and both types of dimorphism are only correlated because they respond to similar selection forces in females and small males (but evolve through different proximate mechanisms), which one is responding to selection first? The simple fact that male dimorphism is more evolutionarily labile than sexual dimorphism does not answer this question, as evolutionary lability does not tell us

136!! how deep in the phylogeny transitions between each type of dimorphism occur. In Emlen et al.’s (2005) study of dung beetle horns, the analysis employed (concentrated changes test; Maddison 1990) assessed whether changes in male dimorphism were concentrated on branches of the phylogeny with or without sexual dimorphism (and vice-versa), allowing simultaneous transitions in both dimorphisms. In fact, the transitions between the categories [sexual monomorphism, male monomorphism] and [sexual dimorphism, male dimorphism] were responsible for all but one of the transitions detected for the dimorphisms in dung beetle horns (see Figure 10 in Emlen et al. 2005). This type of transition in both dimorphisms at the same time is biologically unrealistic though, and probably means that a transition in one dimorphism was followed by a transition in the other relatively quickly in an evolutionary time scale, and therefore these transitions map together in the phylogenetic trees. The approach that we employed here (making use of RJ MCMC chains) models correlated evolution by assuming that transitions involving simultaneous change in both dimorphisms do not occur. According to how the states of each dimorphism are dispersed in the phylogeny, the RJ MCMC chain generates the posterior probability distributions for the transitions that are biologically more realistic, i.e., the ones between the states of one type of dimorphism at a time. With this analysis we found that gains and losses of male dimorphism were the most frequent transitions in C4A length (Figure 5.5), and that they were all between one or two orders of magnitude more probable in the presence than in the absence of sexual dimorphism (Table 5.4). Meanwhile, transitions in sexual dimorphism for this trait were always less probable than transitions in male dimorphism, and were much more probable in the absence than in the presence of male dimorphism (Table 5.4, Figure 5.5). For F4 length, gains and losses of male dimorphism were also among the most frequent transitions (Figure 5.6), being from five to 10 times more probable in the presence than in the absence of sexual dimorphism (Table 5.4). However, the most probable of all transitions in this trait was the gain of sexual dimorphism, which was estimated to be from eight to 16 times more probable in the absence than in the presence of male dimorphism (Table 5.4, Figure 5.6). Taken together, these results strongly suggest that sexual dimorphism precedes the evolution of male dimorphism. One possible reason for this is that the expression of secondary sexual traits is more costly for females than for small males (Fitzpatrick et al. 1995; Martin and Badyaev 1996). Therefore, females respond to the selection against secondary sexual traits earlier than do males. Alternatively, the earlier response of

! 137! females (when compared to small males) to selection against enlarged secondary traits can also result from the fact that females represent half the genetic population exposed to selection (Fisher's condition or principle; Carvalho et al. 1998; Kokko and Jennions 2008), whereas small males necessarily represent a smaller or much smaller fraction of the population. Moreover, sexual dimorphism could evolve earlier than male dimorphism simply because, even in sexually monomorphic species, the genetic architecture for sex specific expression is provided by sex chromosomes, whereas the genetic architecture for male dimorphism does not exist. If any (or all) of these hypotheses are true, regardless of the complex and potentially diverse genetic and physiological mechanisms governing sexual and male dimorphism, their correlated evolution should always be expected.

Losing a dimorphism or losing a structure? The two structures that we chose to investigate also differ in the fact that F4 always exists in both sexes of all species because it is a leg segment, whereas C4A can be absent in males, females or both, because it is just an apophysis on the fourth coxa. In comparison, the horns of the dung beetles studied by Emlen et al. (2005) are structures that can be absent in all individuals as well, and in this respect they resemble C4A in gonyleptid harvestmen. The complete loss of this apophysis in some species of harvestmen can somewhat blur the interpretation of the results. This is because a direct transition in our models between complete monomorphism (category 1, [0,0]) and complete dimorphism (category 4, [1,1]) is assumed to be impossible as it would require two steps (loss/gain of male dimorphism + loss/gain of sexual dimorphism). However, the complete loss of the apophysis in all individuals, regardless of sex or male morph (or size) does not necessarily involve more than one step. It can be argued that a species that is both sexually dimorphic and male dimorphic can become entirely monomorphic due to a single mutation that shuts off the expression of C4A, and whose expression is not dependent on sex or size/condition. If this is true, that means that a species can transit between categories 1 [0,0] and 4 [1,1] without going through categories 2 [1,0] or 3 [0,1]. This is perhaps the reason why transitions q21 (loss of male dimorphism in absence of sexual dimorphism) and q24 (gain of sexual dimorphism in presence of male dimorphism), representing transitions out of category 2, sometimes turned out to be marginal in our analyses, even when category 2 had no species, both for C4A and F4. This may occur because the use of RJ MCMC chains to detect correlated evolution was originally proposed to analyze the evolution of two traits that are potentially

138!! independent under the null hypothesis (Pagel and Meade 2006). While we believe this can be true for male dimorphism and sexual dimorphism, there is no question that mutations in major genes that affect the trait of interest in both sexes can in theory affect both types of dimorphism simultaneously. Fortunately however, this is not an issue for F4, a trait that is always present in both sexes. In this case, losses and gains of dimorphisms in F4 length should always reflect the evolution of proximate mechanisms for the dimorphisms, and not individual mutations that shut off the expression of the trait altogether in all individuals at the same time. Therefore, the fact that our results for C4A were similar to our results for F4 suggests that such mutations have probably not occurred in F4 expression.

Concluding remarks We found that the evolution of sexual dimorphism was strongly correlated with the evolution of male dimorphism, and that sexual dimorphism preceded male dimorphism in gonyleptid harvestmen. Moreover, our findings are relatively similar for both traits analyzed, C4A and F4, despite their relatively different functions in male-male fights, and their diversified shapes and sizes. Our results were also surprisingly consistent across the working phylogenies that we used. This is probably due to the fact that our phylogenies differed from each other in the level of relationships among subfamilies, whereas most of the transitions in dimorphisms (especially in male dimorphism) occurred within subfamilies, rather than in the common ancestor of two or more subfamilies. We believe that the major force behind the relationship between sexual and male dimorphism is the similarity in selection against intralocus sexual conflict and against intralocus conflict among genes that encode different male tactics. Additionally, sexual dimorphism might evolve more readily than male dimorphism because the genetic architecture for sex specific expression is already present even in sexually monomorphic species, due to sex chromosomes, whereas the same is not true for male dimorphism. In conclusion, if a sexual trait arises first on an autosome and is expressed in all individuals, it seems that its suppression in females might evolve earlier than its suppression in small males that adopt AMTs. !

! 139! !

140!! Epilogue

This thesis provided insight into the evolution of male dimorphism associated with alternative mating tactics (AMTs), particularly in arthropods. I investigated questions related to the genetic architecture and the ecology of male dimorphism, as well as its coevolution with sexual dimorphism. Each chapter in this thesis raises new questions that might guide future research on the evolution of male dimorphism and AMTs. Here I emphasize some of the questions raised, and speculate on the possible findings that might arise from investigating them. AMTs linked with male dimorphism are widespread in insects, can take a great variety of forms (Figure 1.6), and usually result from conditional strategies played by males (section 1.2 of Chapter 1). Conditional male dimorphism is the result of a conditional strategy played by males, where every male in the population is, in theory, capable of expressing either of the alternative phenotypes, and the phenotype that is expressed by each individual will depend on environmental conditions (often the condition and/or status of the individual). This seems to be the prevalent type of dimorphism (instead of purely genetic polymorphisms) associated with AMTs in animals (Chapter 1), but its genetic architecture remains poorly understood. The genetics of conditional dimorphisms can be complex, and even within a single species of bulb mite we can find additive genetic variance for switchpoints (Chapter 2) and also strong paternal effects (Chapter 3). The results of Chapter 2 suggest that the phenotypic plasticity behind the expression of male dimorphism could be constrained by genes of large effect, in a way that some males in the male dimorphic population are actually canalized in expressing one of the morphs. This is an hypothesis that deserves further investigation. If indeed such mixtures of conditional and non-conditional individuals are common, this blurs the dichotomy between conditional strategies (purely phenotypically plastic) and alternative strategies (purely genetically determined; Chapter 1). Genes of major effect that canalize male morph expression could potentially represent a constraint for the evolution of conditionality and phenotypic plasticity generally, and this is a promising avenue for future research. Additionally, the paternal effects that I discovered for male dimorphism in R. echinopus (Chapter 3) mean that sons inherit the switchpoint characteristics of their father rather than their mother, and this could be adaptive because the fathers’ switchpoints are under selection. Meanwhile, the genes that affect the switchpoint being carried by the mother are hidden from selection. Such an adaptive shift to paternally biased inheritance could result from intralocus sexual conflict and explain

! 141! the evolution of paternal effects on male dimorphism in other taxa. This hypothesis remains to be tested. It would be interesting to compare the occurrence of paternal effects in conditional dimorphisms that are sex limited and conditional dimorphisms that are not. I predict that paternal effects should be prevalent in conditional dimorphisms that are only expressed in males, and this encompasses the great majority of dimorphisms associated with AMTs. Conversely, dimorphisms that occur in both sexes, such as the ‘winged versus wingless’ dimorphisms that are known in Heteroptera and Thysanoptera, are probably not affected by paternal effects because it would be adaptive for both males and females to express the alternative phenotypes, and intralocus sexual conflict would not be expected. In dung beetles, maternal effects have been shown to be important for male morph determination, and now we know that mothers can specifically alter the horn allometry (in addition to general body size and morph) of their major offspring (Chapter 4). Rearing females in isolation or in dense populations, I showed that mothers that were reared with other conspecifics produced major offspring with longer horns across a wider range of body sizes than the major offspring of females that were reared in isolation. However, this maternal effect did not operate through the amount of dung provided by females, as this was controlled for in my analyses. So how does it operate? Potential mechanisms might include egg or brood mass composition (instead of just its mass). Egg composition affects offspring phenotype in birds (Groothuis and Schwabl 2008), fish (Segers et al. 2012), and seed beetles (Fox and Savalli 2000). This is certainly a promising avenue to investigate hidden forms of maternal effects in dung beetles, but the techniques to successfully manipulate their eggs (such as cross fostering experiments) have not yet been developed. Regarding the composition of brood masses, the inner walls of brood masses of Onthophagus dung beetles are coated with a "shiny and greenish semifluid" (Fabre 1918) that could contain hormones derived from the saliva of females. It would be valuable to investigate whether this substance does contain insect hormones, and whether it could be responsible for the kind of maternal effects that I demonstrated in these animals. The composition of eggs and brood masses in dung beetles are a promising topic to investigate the vehicles though which females influence their offspring phenotypes. However, the proximate mechanisms that allow these influences reside in the offspring themselves, and they constitute yet another interesting area of research in these animals. Maternal adjustments to the composition of eggs or brood masses could trigger epigenetic effects in the offspring, which in turn could be the ultimate mechanisms behind the conditionality of horn length in dung beetles. Epigenetics are

142!! known to be crucial for the regulation of phenotypic plasticity generally (Glastad et al. 2011; Simon et al. 2011; Simpson et al. 2011), and explain the plastic expression of castes in honeybees (Chittka and Chittka 2010; Li et al. 2010). It is still not known how widespread these mechanisms are, but conditional dimorphisms in insects that present AMTs are a promising place to look. Finally, I also found that male dimorphism generally follows the evolution of sexual dimorphism in harvestmen (Chapter 5). This possibly means that when a sexual trait arises first on an autosome and is expressed in all individuals, its suppression in females (sexual dimorphism) will evolve before its suppression in small males that adopt AMTs (male dimorphism). It would be very valuable to investigate the proximate mechanisms of these dimorphisms in order to discover how this coevolution comes about. I suggested that sexual dimorphism evolves more readily than male dimorphism because the genetic architecture for sex specific expression is provided by sex chromosomes, even in sexually monomorphic species, whereas this pre-existing genetic architecture is unavailable for the evolution of male dimorphism. Nevertheless, there is still much to learn about the connection of sex limited expression with sex chromosomes. For example, it has recently been shown that variation on the X chromosome of male Drosophila melanogaster affects the fitness of their male offspring, so that a non-transmitted paternal chromosome can influence the phenotype of the offspring (Friberg et al. 2012). Much research still needs to be done on the role of sex chromosomes in the evolution of sexual dimorphism, which might also shed light on the coevolution of sexual and male dimorphism. !

! 143! !

144!! References

Ainsworth R, 1990. Male dimorphism in 2 new species of nematode (Pharyngodonidae, Oxyurida) from New Zealand lizards. Journal of Parasitology 76:812-822. Alcock J, 1995. Persistent size variation in the anthophorine bee Centris-pallida (Apidae) despite a large male mating advantage. Ecological Entomology 20:1-4. Alcock J, 1996. Provisional rejection of three alternative hypotheses on the maintenance of a size dichotomy in Dawson's burrowing bee, Amegilla dawsoni (Apidae, Apinae, Anthophorini). Behavioral Ecology and Sociobiology 39:181-188. Alcock J, 1997. Competition from large males and the alternative mating tactics of small males of Dawson's burrowing bee (Amegilla dawsoni) (Apidae, Apinae, Anthophorini). Journal of Insect Behavior 10:99-113. Alcock J, Jones CE, Buchmann SL, 1977. Male mating strategies in bee Centris-pallida Fox (Anthophoridae-Hymenoptera). American Naturalist 111:145-155. Andersson M, 1994. Sexual Selection. New Jersey: Princeton University Press. Arnold SJ, Duvall D, 1994. Animal mating systems - a synthesis based on selection theory. American Naturalist 143:317-348. Austad SN, 1984. A classification of alternative reproductive behaviors and methods for field-testing ESS models. American Zoologist 24:309-319. Bailey WJ, Field G, 2000. Acoustic satellite behaviour in the Australian bushcricket Elephantodeta nobilis (Phaneropterinae, Tettigoniidae, Orthoptera). Animal Behaviour 59:361-369. Bateman AJ, 1948. Intra-sexual selection in Drosophila. Heredity 2:349-368. Bates D, Maechler M, Bolker B, 2012. lme4: Linear mixed-effects models using S4 classes. R package version 0.999999-0 http://CRAN.R- project.org/package=lme4 Beani L, Turillazzi S, 1988. Alternative mating tactics in males of Polistes-dominulus (Hymenoptera, Vespidae). Behavioral Ecology and Sociobiology 22:257-264. Benaglia T, Chauveau D, Hunter DR, Young D, 2009. mixtools: An R Package for Analyzing Finite Mixture Models. Journal of Statistical Software 32:1-29. Bernardo J, 1996. Maternal effects in animal ecology. American Zoologist 36:83-105. Bilde T, Friberg U, Maklakov AA, Fry JD, Arnqvist G, 2008. The genetic architecture of fitness in a seed beetle: assessing the potential for indirect genetic benefits of female choice. BMC Evolutionary Biology 8:11. Blackmore MS, Charnov, 1989. Adaptive variation in environmental sex determination in a nematode. American Naturalist 134:817-823.

! 145! Bragagnolo C, Pinto-Da-Rocha R, 2012. Systematic review of Promitobates Roewer, 1913 and cladistic analysis of Mitobatinae Simon, 1879 (Arachnida: Opiliones: Gonyleptidae). Zootaxa 3308:1-48. Brockmann HJ, 2001. The evolution of alternative strategies and tactics. Advances in the Study of Behavior 30:1-51. Brockmann HJ, 2008. Alternative reproductive tactics in insects. In: Oliveira RF, Taborsky M, Brockmann HJ, editors. Alternative Reproductive Tactics: an Integrative Approach. Cambridge: Cambridge University Press. p. 177-223. Brockmann HJ, Taborsky M, 2008. Alternative reproductive tactics and the evolution of alternative allocation phenotypes. In: Oliveira RF, Taborsky M, Brockmann HJ, editors. Alternative Reproductive Tactics: an Integrative Approach. Cambridge: Cambridge University Press. p. 25-51. Burk T, 1983. Male aggression and female choice in a field cricket (Teleogryllus oceanicus): the importance of courtship song. In: Gwynne DT, Morris GK, editors. Orthopteran mating systems: sexual selection in a diverse group of insects. Boulder, CO: Westview Press. p. 97-119. Buzatto BA, Machado G, 2008. Resource defense polygyny shifts to female defense polygyny over the course of the reproductive season of a Neotropical harvestman. Behavioral Ecology and Sociobiology 63:85-94. Buzatto BA, Requena GS, Lourenço RS, Munguía-Steyer R, Machado G, 2011. Conditional male dimorphism and alternative reproductive tactics in a Neotropical arachnid (Opiliones). Evolutionary Ecology 25:331-349. Buzatto BA, Simmons LW, Tomkins JL, 2012a. Genetic variation underlying the expression of a polyphenism. Journal of Evolutionary Biology 25:748-758. Buzatto BA, Simmons LW, Tomkins JL, 2012b. Maternal effects on male weaponry: female dung beetles produce major sons with longer horns when they perceive higher population density. BMC Evolutionary Biology 12:118. Buzatto BA, Tomkins JL, Simmons LW, in press. Alternative tactics within mating systems. In: Shuker DM, Simmons LW, editors. The Evolution of Insect Mating Systems. Oxford: Oxford University Press. Cade W, 1980. Alternative male reproductive behaviors. Florida Entomologist 63:30-45. Cade WH, 1981. Alternative male strategies - genetic-differences in crickets. Science 212:563-564. Cade WH, 1984. Genetic variation underlying sexual behavior and reproduction. American Zoologist 24:355-366.

146!! Cade WH, Wyatt DR, 1984. Factors affecting calling behavior in field crickets, Teleogryllus and Gryllus (age, weight, density, and parasites). Behaviour 88:61- 75. Carvalho AB, Sampaio MC, Varandas FR, Klaczko LB, 1998. An experimental demonstration of Fisher's principle: Evolution of sexual proportion by natural selection. Genetics 148:719-731. Chapman RF, 1982. The Insects: Structure and Function. London: Hodder and Stoughton. Chittka A, Chittka L, 2010. Epigenetics of royalty. PloS Biology 8: e1000532. Cockerham CC, Weir BS, 1977. Quadratic analysis of reciprocal crosses. Biometrics 33:187-203. Cook JM, Bean D, 2006. Cryptic male dimorphism and fighting in a fig wasp. Animal Behaviour 71:1095-1101. Córdoba-Aguilar A, Cordero-Rivera A, 2005. Evolution and ecology of Calopterygidae (Zygoptera: Odonata): status of knowledge and research perspectives. Neotropical Entomology 34:861-879. Crawley MJ, 2007. The R Book. Chichester: John Wiley & Sons Ltd. Crespi BJ, 1988a. Adaptation, compromise, and constraint - the development, morphometrics, and behavioral basis of a fighter-flier polymorphism in male Hoplothrips-karnyi (Insecta, Thysanoptera). Behavioral Ecology and Sociobiology 23:93-104. Crespi BJ, 1988b. Alternative male mating tactics in a thrips - effects of sex-ratio variation and body size. American Midland Naturalist 119:83-92. Crill WD, Huey RB, Gilchrist GW, 1996. Within- and between-generation effects of temperature on the morphology and physiology of Drosophila melanogaster. Evolution 50:1205-1218. Cunningham EJA, Russell AF, 2000. Egg investment is influenced by male attractiveness in the mallard. Nature 404:74-77. Danforth BN, 1991. The morphology and behavior of dimorphic males in Perdita- portalis (Hymenoptera, Andrenidae). Behavioral Ecology and Sociobiology 29:235-247. DaSilva MB, Gnaspini P, 2009. A systematic revision of Goniosomatinae (Arachnida : Opiliones : Gonyleptidae), with a cladistic analysis and biogeographical notes. Invertebrate Systematics 23:530-624. DaSilva MB, Pinto-da-Rocha R, 2010. Systematic review and cladistic analysis of the Hernandariinae (Opiliones: Gonyleptidae). Zoologia 27:577-642.

! 147! David P, Hingle A, Greig D, Rutherford A, Pomiankowski A, Fowler K, 1998. Male sexual ornament size but not asymmetry reflects condition in stalk-eyed flies. Proceedings of the Royal Society B-Biological Sciences 265:2211-2216. Dawkins, R, 1980. Good strategy or evolutionary stable strategy? In: Barlow GW, Silverberg J, editors. Sociobiology: Beyond Nature/Nurture? Boulder, Colorado: Westview. p. 331-367. Day T, Bonduriansky R, 2004. Intralocus sexual conflict can drive the evolution of genomic imprinting. Genetics 167:1537-1546. Dempster ER, Lerner IM, 1950. Heritability of threshold characters. Genetics 35:212-236. Donelson NC, van Staaden MJ, 2005. Alternate tactics in male bladder grasshoppers Bullacris membracioides (Orthoptera : Pneumoridae). Behaviour 142:761-778. Doums C, Viard F, Jarne P, 1998. The evolution of phally polymorphism. Biological Journal of the Linnean Society 64:273-296. Dowling DK, Nystrand M, Simmons LW, 2010. Maternal effects, but no good or compatible genes for sperm competitiveness in Australian crickets. Evolution 64:1257-1266. Dunn DW, Crean CS, Wilson CL, Gilburn AS, 1999. Male choice, willingness to mate and body size in seaweed flies (Diptera : Coelopidae). Animal Behaviour 57:847-853. Eberhard WG, Gutierrez EE, 1991. Male dimorphisms in beetles and earwigs and the question of developmental constraints. Evolution 45:18-28. Emlen DJ, 1994. Environmental-control of horn length dimorphism in the beetle Onthophagus acuminatus (Coleoptera, Scarabaeidae). Proceedings of the Royal Society of London Series B-Biological Sciences 256:131-136. Emlen DJ, 1996. Artificial selection on horn length body size allometry in the horned beetle Onthophagus acuminatus (Coleoptera: Scarabaeidae). Evolution 50:1219- 1230. Emlen DJ, 1997a. Alternative reproductive tactics and male dimorphism in the horned beetle Onthophagus acuminatus. Behavioral Ecology and Sociobiology 41:355-341. Emlen DJ, 1997b. Diet alters male horn allometry in the beetle Onthophagus acuminatus (Coleoptera: Scarabaeidae). Proceedings of the Royal Society of London Series B-Biological Sciences 264:567-574. Emlen DJ, 2001. Costs and the diversification of exaggerated animal structures. Science 291:1534-1536. Emlen DJ, 2008. The roles of genes and the environment in the expression and evolution of alternative tactics. In: Oliveira RF, Taborsky M, Brockmann HJ,

148!! editors. Alternative Reproductive Tactics: an Integrative Approach. Cambridge: Cambridge University Press. p. 85-108. Emlen DJ, Hunt J, Simmons LW, 2005. Evolution of sexual dimorphism and male dimorphism in the expression of beetle horns: Phylogenetic evidence for modularity, evolutionary lability, and constraint. American Naturalist 166:S42- S68. Emlen DJ, Lavine LC, Ewen-Campen B, 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. Emlen DJ, Nijhout HF, 2000. The development and evolution of exaggerated morphologies in insects. Annual Review of Entomology 45:661-708. Emlen DJ, Szafran Q, Corley LS, Dworkin I, 2006. Insulin signaling and limb- patterning: candidate pathways for the origin and evolutionary diversification of beetle 'horns'. Heredity 97:179-191. Emlen DJ, Warren IA, Johns A, Dworkin I, Lavine LC, 2012. A mechanism of extreme growth and reliable signaling in sexually selected ornaments and weapons. Science 337:860-864. Emlen ST, Oring LW, 1977. Ecology, sexual selection, and evolution of mating systems. Science 197:215-223. Fabre JH, 1918. The Sacred Beetle and Others. New York: Dodd, Mead and Company. Falconer DS, 1965. Inheritance of liability to certain diseases estimated from incidence among relatives. Annals of Human Genetics 29:51-76. Falconer DS, 1989. Introduction to Quantitative Genetics. New York: John Wiley and Sons. Feaver M, 1983. Pair formation in the katydid Orchelimum nigripes (Orthoptera: Tettigoniidae). In: Gwynne DT, Morris GK, editors. Orthopteran Mating Systems: Sexual Competition in a Diverse Group of Insects. Boulder, CO: Westview Press. p. 205-239. Field LM, Lyko F, Mandrioli M, Prantera G, 2004. DNA methylation in insects. Insect Molecular Biology 13:109-115. Figuerola J, Green AJ, 2000. The evolution of sexual dimorphism in relation to mating patterns, cavity nesting, insularity and sympatry in the Anseriformes. Functional Ecology 14:701-710. Fincke OM, 2004. Polymorphic signals of harassed female odonates and the males that learn them support a novel frequency-dependent model. Animal Behaviour 67:833-845.

! 149! Fincke OM, Jödicke R, Paulson D, Schultz TD, 2005. The evolution and frequency of female color morphs in Holarctic Odonata: why are male-like morphs typically the minority? International Journal of Odonatology 8:183-212. Fischer K, Fiedler K, 2001. Resource-based territoriality in the butterfly Lycaena hippothoe and environmentally induced behavioural shifts. Animal Behaviour 61:723-732. Fitzpatrick JL, Montgomerie R, Desjardins JK, Stiver KA, Kolm N, Balshine S, 2009. Female promiscuity promotes the evolution of faster sperm in cichlid fishes. Proceedings of the National Academy of Sciences of the United States of America 106:1128-1132. Fitzpatrick S, Berglund A, Rosenqvist G, 1995. Ornaments or offspring - costs to reproductive success restrict sexual selection processes. Biological Journal of the Linnean Society 55:251-260. Forsyth A, Alcock J, 1990. Female mimicry and resource defense polygyny by males of a tropical rove beetle, Leistotrophus-versicolor (Coleoptera, Staphylinidae). Behavioral Ecology and Sociobiology 26:325-330. Forsyth A, Montgomerie RD, 1987. Alternative reproductive tactics in the territorial damselfly Calopteryx-maculata - sneaking by older males. Behavioral Ecology and Sociobiology 21:73-81. Fox CW, Savalli UM, 2000. Maternal effects mediate host expansion in a seed-feeding beetle. Ecology 81:3-7. Fox CW, Waddell KJ, Mousseau TA, 1995. Parental host-plant affects offspring life- histories in a seed beetle. Ecology 76:402-411. French BW, Cade WH, 1989. Sexual selection at varying population-densities in male field crickets, Gryllus-veletis and Gryllus-pennsylvanicus. Journal of Insect Behavior 2:105-121. Friberg U, Stewart AD, Rice WR, 2012. X- and Y-chromosome linked paternal effects on a life-history trait. Biology Letters 8:71-73. Fry JD, 2004. Estimation of genetic variances and covariances by restricted maximum likelihood using PROC MIXED. In: Saxton A, editor. Genetic Analysis of Complex Traits Using SAS. Cary, NC: SAS Institute Inc. p. 11-34. Fujisaki K, 1992. A male fitness advantage to wing reduction in the oriental chinch bug, Cavelerius-saccharivorus Okajima (Heteroptera, Lygaeidae). Researches on Population Ecology 34:173-183. Gadgil M, 1972. Male dimorphism as a consequence of sexual selection. American Naturalist 106:574-580.

150!! Garcia-Gonzalez F, Simmons LW, 2007. Paternal indirect genetic effects on offspring viability and the benefits of polyandry. Current Biology 17:32-36. Giesel JT, 1986. Effects of parental photoperiod regime on progeny development time in Drosophila-simulans. Evolution 40:649-651. Gil D, Graves J, Hazon N, Wells A, 1999. Male attractiveness and differential testosterone investment in zebra finch eggs. Science 286:126-128. Gilburn AS, Crean CS, Day TH, 1996. Sexual selection in natural populations of seaweed flies: variation in the offspring fitness of females carrying different inversion karyotypes. Proceedings of the Royal Society B-Biological Sciences 263:249-256. Gilburn AS, Day TH, 1994. Sexual dimorphism, sexual selection and the αβ chromosomal inversion polymorphism in the seaweed fly, Coelopa frigida. Proceedings of the Royal Society B-Biological Sciences 257:303-309. Gilburn AS, Foster SP, Day TH, 1992. Female mating preference for large size in Coelopa frigida (seaweed fly). Heredity 69:209-212. Gilburn AS, Foster SP, Day TH, 1993. Genetic correlation between a female mating preference and the preferred male character in seaweed flies (Coelopa frigida). Evolution 47:1788-1795. Giurfa M, 2007. Behavioral and neural analysis of associative learning in the honeybee: a taste from the magic well. Journal of Comparative Physiology a- Neuroethology Sensory Neural and Behavioral Physiology 193:801-824. Glastad KM, Hunt BG, Yi SV, Goodisman MAD, 2011. DNA methylation in insects: on the brink of the epigenomic era. Insect Molecular Biology 20:553-565. Gnaspini P, 1995. Reproduction and postembryonic development of Goniosoma- spelaeum, a cavernicolous harvestman from southeastern Brazil (Arachnida, Opiliones, Gonyleptidae). Invertebrate Reproduction & Development 28:137- 151. Gonzalez JM, Matthews RW, 2008. Female and male polymorphism in two species of Melittobia parasitoid wasps (Hymenoptera : Eulophidae). Florida Entomologist 91:162-169. Gonzalez-Voyer A, Fitzpatrick JL, Kolm N, 2008. Sexual selection determines parental care patterns in cichlid fishes. Evolution 62:2015-2026. Gray DA, 1999. Intrinsic factors affecting female choice in house crickets: Time cost, female age, nutritional condition, body size, and size-relative reproductive investment. Journal of Insect Behavior 12:691-700.

! 151! Greenfield MD, Shelly TE, 1985. Alternative mating strategies in a desert grasshopper - evidence of density-dependence. Animal Behaviour 33:1192-1210. Grimaldi D, Engel MS, 2005. Evolution of the Insects: Cambridge University Press. Grondziel E, 1975. Sex chromosomes in Rhizoglyphus echinopus (F, & R.) (Acarina, Acaridae). Folia Biologica - Krakow 23:69-72. Groothuis TGG, Schwabl H, 2008. Hormone-mediated maternal effects in birds: mechanisms matter but what do we know of them? Philosophical Transactions of the Royal Society B-Biological Sciences 363:1647-1661. Gross MR, 1996. Alternative reproductive strategies and tactics: Diversity within sexes. Trends in Ecology & Evolution 11:92-98. Hall KC, Hanlon RT, 2002. Principal features of the mating system of a large spawning aggregation of the giant Australian cuttlefish Sepia apama (Mollusca : Cephalopoda). Marine Biology 140:533-545. Hammer M, Menzel R, 1995. Learning and memory in the honeybee. Journal of Neuroscience 15:1617-1630. Hammill E, Rogers A, Beckerman AP, 2008. Costs, benefits and the evolution of inducible defences: a case study with Daphnia pulex. Journal of Evolutionary Biology 21:705-715. Hanlon RT, Naud MJ, Shaw PW, Havenhand JN, 2005. Behavioural ecology - Transient sexual mimicry leads to fertilization. Nature 433:212-212. Hazel W, Smock R, 2000. Inheritance in the conditional strategy revisited. Journal of Theoretical Biology 204:307-309. Hazel WN, 2002. The environmental and genetic control of seasonal polyphenism in larval color and its adaptive significance in a swallowtail butterfly. Evolution 56:342-348. Hazel WN, Smock R, 1993. Modeling selection on conditional strategies in stochastic environments. In: Yoshimura J, Clark CW, editors. Adaptation in stochastic environments. Berlin: Springer Verlag. p. 147-154. Hazel WN, Smock R, Johnson MD, 1990. A polygenic model for the evolution and maintenance of conditional strategies. Proceedings of the Royal Society of London Series B-Biological Sciences 242:181-187. Hazel WN, Smock R, Lively CM, 2004. The ecological genetics of conditional strategies. American Naturalist 163:888-900. Heinze J, Keller L, 2000. Alternative reproductive strategies: a queen perspective in ants. Trends in Ecology & Evolution 15:508-512.

152!! Higginson DM, Miller KB, Segraves KA, Pitnick S, 2012a. Convergence, recurrence and diversification of complex sperm traits in diving beetles (Dytiscidae). Evolution 66:1650-1661. Higginson DM, Miller KB, Segraves KA, Pitnick S, 2012b. Female reproductive tract form drives the evolution of complex sperm morphology. Proceedings of the National Academy of Sciences of the United States of America 109:4538-4543. Hissmann K, 1990. Strategies of mate finding in the European field cricket (Gryllus- campestris) at different population-densities - a field-study. Ecological Entomology 15:281-291. Houston T, 1970. Discovery of an apparent male soldier caste in a nest of a halictine bee (Hymenoptera : Halictidae), with notes on the nest. Australian Journal of Zoology 18:345-351. Hunt J, Kotiaho JS, Tomkins JL, 1999. Dung pad residence time covaries with male morphology in the dung beetle Onthophagus taurus. Ecological Entomology 24:174-180. Hunt J, Simmons LW, 2000. Maternal and paternal effects on offspring phenotype in the dung beetle Onthophagus taurus. Evolution 54:936-941. Hunt J, Simmons LW, 2001. Status-dependent selection in the dimorphic beetle Onthophagus taurus. Proceedings of the Royal Society of London Series B- Biological Sciences 268:2409-2414. Hunt J, Simmons LW, 2002a. Behavioural dynamics of biparental care in the dung beetle Onthophagus taurus. Animal Behaviour 64:65-75. Hunt J, Simmons LW, 2002b. Confidence of paternity and paternal care: covariation revealed through the experimental manipulation of the mating system in the beetle Onthophagus taurus. Journal of Evolutionary Biology 15:784-795. Hunt J, Simmons LW, Kotiaho JS, 2002. A cost of maternal care in the dung beetle Onthophagus taurus? Journal of Evolutionary Biology 15:57-64. Ishiwata K, Sasaki G, Ogawa J, Miyata T, Su ZH, 2011. Phylogenetic relationships among insect orders based on three nuclear protein-coding gene sequences. Molecular Phylogenetics and Evolution 58:169-180. Ivy TM, 2007. Good genes, genetic compatibility and the evolution of polyandry: use of the diallel cross to address competing hypotheses. Journal of Evolutionary Biology 20:479-487. Iwata Y, Sakurai Y, 2007. Threshold dimorphism in ejaculate characteristics in the squid Loligo bleekeri. Marine Ecology-Progress Series 345:141-146.

! 153! Jennions MD, Moller AP, Petrie M, 2001. Sexually selected traits and adult survival: A meta-analysis. Quarterly Review of Biology 76:3-36. Johnson C, 1975. Polymorphism and natural selection in ischnuran damselflies. Evolutionary Theory 1:81-90. Jones RT, Salazar PA, Jiggins CD, Joron M, 2012. Evolution of a mimicry supergene from a multilocus architecture. Proceedings of the Royal Society B: Biological Sciences 279:316-325. Joron M, Frezal L, Jones RT, Chamberlain NL, Lee SF, Haag CR, Whibley A, Becuwe M, Baxter SW, Ferguson L, 2011. Chromosomal rearrangements maintain a polymorphic supergene controlling butterfly mimicry. Nature 477:203-206. Kahm M, Hasenbrink G, Lichtenberg-Frate H, Ludwig J, Kschischo M, 2010. grofit: Fitting Biological Growth Curves with R. Journal of Statistical Software 33:1-21. Kelly CD, 2006. The relationship between resource control, association with females and male weapon size in a male dominance insect. Ethology 112:362-369. Kelly CD, 2008. Sperm investment in relation to weapon size in a male trimorphic insect? Behavioral Ecology 19:1018-1024. Kelly CD, Adams DC, 2010. Sexual selection, ontogenetic acceleration, and hypermorphosis generates male trimorphism in Wellington tree weta. Evolutionary Biology 37:200-209. Kirkpatrick M, 2010. How and why chromosome inversions evolve. PloS Biology 8. Kishi S, Nishida T, 2008. Optimal investment in sons and daughters when parents do not know the sex of their offspring. Behavioral Ecology and Sociobiology 62:607-615. Knell RJ, 2009. On the analysis of non-linear allometries. Ecological Entomology 34:1-11. Knell RJ, Pomfret JC, Tomkins JL, 2004. The limits of elaboration: curved allometries reveal the constraints on mandible size in stag beetles. Proceedings of the Royal Society of London Series B-Biological Sciences 271:523-528. Kokko H, Jennions MD, 2008. Parental investment, sexual selection and sex ratios. Journal of Evolutionary Biology 21:919-948. Kokko H, Rankin DJ, 2006. Lonely hearts or sex in the city? Density-dependent effects in mating systems. Philosophical Transactions of the Royal Society B-Biological Sciences 361:319-334. Kolm N, 2001. Females produce larger eggs for large males in a paternal mouthbrooding fish. Proceedings of the Royal Society of London Series B- Biological Sciences 268:2229-2234.

154!! Koning JW, Jamieson IG, 2001. Variation in size of male weaponry in a harem-defence polygynous insect, the mountain stone weta Hemideina maori (Orthoptera : Anostostomatidae). New Zealand Journal of Zoology 28:109-117. Konior M, Radwan J, Kolodziejczyk M, Keller L, 2006. Strong association between a single gene and fertilization efficiency of males and fecundity of their mates in the bulb mite. Proceedings of the Royal Society of London Series B-Biological Sciences 273:309-314. Kotiaho JS, Simmons LW, Hunt J, Tomkins JL, 2003. Males influence maternal effects that promote sexual selection: a quantitative genetic experiment with dung beetles Onthophagus taurus. American Naturalist 161:852-859. Kukuk PF, 1996. Male dimorphism in Lasioglossum (Chilalictus) hemichalceum: The role of larval nutrition. Journal of the Kansas Entomological Society 69:147-157. Kukuk PF, Schwarz M, 1987. Intranest behavior of the communal sweat bee Lasioglossum-(Chilalictus) erythrurum (Hymenoptera, Halictidae). Journal of the Kansas Entomological Society 60:58-64. Kury AB, 2011. Order Opiliones Sundevall, 1833. In: Zhang, Z-Q, editor. Animal biodiversity: An outline of higher-level classification and survey of taxonomic richness. Zootaxa 3148:112-114. Kury AB, 2012. A synopsis of catalogs and checklists of harvestmen (Arachnida, Opiliones). Zootaxa 3184:35-58. Kury AB, Pinto-da-Rocha R, 2007. Gonyleptidae Sundevall, 1833. In: Pinto-da-Rocha R, Machado G, Giribet G, editors. Harvestmen: The Biology of Opiliones. Cambridge, MA: Harvard University Press. p. 196-203. Lande R, 1980. Sexual dimorphism, sexual selection, and adaptation in polygenic characters. Evolution 34:292-305. Langellotto GA, Denno RF, Ott JR, 2000. A trade-off between flight capability and reproduction in males of a wing-dimorphic insect. Ecology 81:865-875. Lawson LP, Vander Meer RK, Shoemaker DW, 2012. Male reproductive fitness and queen polyandry are linked to variation in the supergene Gp-9 in the fire ant Solenopsis invicta. Proceedings of the Royal Society B: Biological Sciences 279:3217-3222. Lehmann GUC, 2007. Density-dependent plasticity of sequential mate choice in a bushcricket (Orthoptera : Tettigoniidae). Australian Journal of Zoology 55:123- 130.

! 155! Li J, Wu J, Rundassa DB, Song F, Zheng A, Fang Y, 2010. Differential protein expression in honeybee (Apis mellifera L.) larvae: underlying caste differentiation. Plos One 5:e13455. Lively CM, 1986. Canalization versus developmental conversion in a spatially-variable environment. American Naturalist 128:561-572. Lively CM, 1999. Developmental strategies in spatially variable environments: barnacle shell dimorphism and strategic models of selection. In: Tollrian R, Harvell CD, editors. The Ecology and Evolution of Inducible Defences. Princeton, New Jersey: Princeton University Press. p. 245-258. Lively CM, Hazel WN, Schellenberger MJ, Michelson KS, 2000. Predator-induced defense: variation for inducibility in an intertidal barnacle. Ecology 81:1240- 1247. Loher W, Rence B, 1978. Mating-behavior of Teleogryllus-commodus-(Walker) and its central and peripheral control. Zeitschrift Fur Tierpsychologie-Journal of Comparative Ethology 46:225-259. Lynch M, Walsh B, 1998. Genetics and analysis of quantitative traits. Sunderland: Sinauer Associates Inc. Machado G, Macías-Ordóñez R, 2007. Reproduction. In: Pinto-da-Rocha R, Machado G, Giribet G, editors. Harvestmen: the Biology of Opiliones. Cambridge: Harvard University Press. p. 414-454. Machado G, Oliveira PS, 1998. Reproductive biology of the neotropical harvestman Goniosoma longipes (Arachnida, Opiliones : Gonyleptidae): mating and oviposition behaviour, brood mortality, and parental care. Journal of Zoology 246:359-367. Machado G, Pinto-da-Rocha R, Giribet G, 2007. What are harvestmen? In: Pinto-da- Rocha R, Machado G, Giribet G, editors. Harvestmen: the Biology of Opiliones. Cambridge: Harvard University Press. p. 1-13. Maddison WP, 1990. A method for testing the correlated evolution of 2 binary characters - are gains or losses concentrated on certain branches of a phylogenetic tree? Evolution 44:539-557. Maddison WP, Maddison DR, 2011. Mesquite: a modular system for evolutionary analysis. Version 2.75 http://mesquiteproject.org Marden JH, Rollins RA, 1994. Assessment of energy reserves by damselflies engaged in aerial contests for mating territories. Animal Behaviour 48:1023-1030. Martin TE, Badyaev AV, 1996. Sexual dichromatism in birds: Importance of nest predation and nest location for females versus males. Evolution 50:2454-2460.

156!! Matthews RW, Gonzalez JM, Matthews JR, Deyrup LD, 2009. Biology of the parasitoid Melittobia (Hymenoptera: Eulophidae). Annual Review of Entomology. p. 251- 266. Maynard-Smith J, 1982. Evolution and the Theory of Games. Cambridge: Cambridge University Press. Mendes AC, 2011. Phylogeny and taxonomic revision of Heteropachylinae (Opiliones: Laniatores: Gonyleptidae). Zoological Journal of the Linnean Society 163:437- 483. Menzel R, 1993. Associative learning in honey-bees. Apidologie 24:157-168. Menzel R, Muller U, 1996. Learning and memory in honeybees: from behavior to neural substrates. Annual Review of Neuroscience 19:379-404. Minelli A, 2003. The Development of Animal Form. Cambridge: Cambridge University Press. Mitani JC, GrosLouis J, Richards AF, 1996. Sexual dimorphism, the operational sex ratio, and the intensity of male competition in polygynous primates. American Naturalist 147:966-980. Moczek AP, 2002. Allometric plasticity in a polyphenic beetle. Ecological Entomology 27:58-67. Moczek AP, 2003. The behavioral ecology of threshold evolution in a polyphenic beetle. Behavioral Ecology 14:841-854. Moczek AP, 2010. Phenotypic plasticity and diversity in insects. Philosophical Transactions of the Royal Society B-Biological Sciences 365:593-603. Moczek AP, Emlen DJ, 2000. Male horn dimorphism in the scarab beetle, Onthophagus taurus: do alternative reproductive tactics favour alternative phenotypes? Animal Behaviour 59:459-466. Moczek AP, Hunt J, Emlen DJ, Simmons LW, 2002. Threshold evolution in exotic populations of a polyphenic beetle. Evolutionary Ecology Research 4:587-601. Moczek AP, Nijhout HF, 2003. Rapid evolution of a polyphenic threshold. Evolution & Development 5:259-268. Moran NA, 1992. The evolutionary maintenance of alternative phenotypes. American Naturalist 139:971-989. Mousseau TA, Dingle H, 1991. Maternal effects in insect life histories. Annual Review of Entomology 36:511-534. Mousseau TA, Fox CW, 1998. The adaptive significance of maternal effects. Trends in Ecology & Evolution 13:403-407.

! 157! Mousseau TA, Roff DA, 1989. Adaptation to seasonality in a cricket - patterns of phenotypic and genotypic variation in body size and diapause expression along a cline in season length. Evolution 43:1483-1496. Mousseau TA, Uller T, Wapstra E, Badyaev AV, 2009. Evolution of maternal effects: past and present. Philosophical Transactions of the Royal Society B-Biological Sciences 364:1035-1038. Munguia-Steyer R, Cordoba-Aguilar A, Romo-Beltran A, 2010. Do individuals in better condition survive for longer? Field survival estimates according to male alternative reproductive tactics and sex. Journal of Evolutionary Biology 23:175- 184. Nazareth TM, Machado G, 2009. Reproductive behavior of Chavesincola inexpectabilis (Opiliones, Gonyleptidae) with description of a new and independently evolved case of paternal care in harvestmen. Journal of Arachnology 37:127-134. Nazareth TM, Machado G, 2010. Mating system and exclusive postzygotic paternal care in a Neotropical harvestman (Arachnida: Opiliones). Animal Behaviour 79:547-554. Nijhout HF, 1999. Control mechanisms of polyphenic development in insects. Bioscience 49:181-192. Nijhout HF, 2003. Development and evolution of adaptive polyphenisms. Evolution & Development 5:9-18. Norman MD, Finn J, Tregenza T, 1999. Female impersonation as an alternative reproductive strategy in giant cuttlefish. Proceedings of the Royal Society of London Series B-Biological Sciences 266:1347-1349. Nylin S, Gotthard K, 1998. Plasticity in life-history traits. Annual Review of Entomology 43:63-83. Oliveira RF, Taborsky M, Brockmann HJ, 2008. Alternative Reproductive Tactics: an Integrative Approach. Cambridge: Cambridge University Press. Páez DJ, Bernatchez J, Dodson JJ, 2011. Alternative life histories in the Atlantic salmon: genetic covariances within the sneaker sexual tactic in males. Proceedings of the Royal Society of London Series B-Biological Sciences 278:2150-2158. Pagel M, 1999. The maximum likelihood approach to reconstructing ancestral character states of discrete characters on phylogenies. Systematic Biology 48:612-622. Pagel M, Meade A, 2006. Bayesian analysis of correlated evolution of discrete characters by reversible-jump Markov chain Monte Carlo. American Naturalist 167:808-825.

158!! Panhuis TM, Wilkinson GS, 1999. Exaggerated male eye span influences contest outcome in stalk-eyed flies (Diopsidae). Behavioral Ecology and Sociobiology 46:221-227. Parejko K, Dodson SI, 1991. The evolutionary ecology of an antipredator reaction norm - Daphnia-pulex and Chaoborus-americanus. Evolution 45:1665-1674. Perez-Barberia FJ, Gordon IJ, Pagel M, 2002. The origins of sexual dimorphism in body size in ungulates. Evolution 56:1276-1285. Perry JC, Sharpe DMT, Rowe L, 2009. Condition-dependent female remating resistance generates sexual selection on male size in a ladybird beetle. Animal Behaviour 77:743-748. Pienaar J, Greeff JM, 2003. Maternal control of offspring sex and male morphology in the Otitesella fig wasps. Journal of Evolutionary Biology 16:244-253. Pigliucci M, 2005. Evolution of phenotypic plasticity: where are we going now? Trends in Ecology and Evolution 20:481-486. Pinto-da-Rocha R, 2002. Systematic review and cladistic analysis of the Brazilian subfamily Caelopyginae (Opiliones: Gonyleptidae). Arquivos de Zoologia 36:357-464. Pinto-da-Rocha R, Bragagnolo C, 2010. Systematic revision and cladistic analysis of the Brazilian subfamily Sodreaninae (Opiliones: Gonyleptidae). Invertebrate Systematics 24:509-538. Pinto-da-Rocha R, da Silva MB, Bragagnolo C, 2005. Faunistic similarity and historic biogeography of the harvestmen of southern and southeastern Atlantic Rain Forest of Brazil. Journal of Arachnology 33:290-299. Plaistow S, SivaJothy MT, 1996. Energetic constraints and male mate-securing tactics in the damselfly Calopteryx splendens xanthostoma (Charpentier). Proceedings of the Royal Society of London Series B-Biological Sciences 263:1233-1239. Plaistow SJ, Johnstone RA, Colegrave N, Spencer M, 2004. Evolution of alternative mating tactics: conditional versus mixed strategies. Behavioral Ecology 15: 534- 542. Plavcan JM, Vanschaik CP, 1992. Intrasexual competition and canine dimorphism in anthropoid primates. American Journal of Physical Anthropology 87:461-477. Pomfret JC, Knell RJ, 2006. Sexual selection and horn allometry in the dung beetle Euoniticellus intermedius. Animal Behaviour 71:567-576. Pomfret JC, Knell RJ, 2008. Crowding, sex ratio and horn evolution in a South African beetle community. Proceedings of the Royal Society B-Biological Sciences 275:315-321.

! 159! Priest NK, Roach DA, Galloway LF, 2008. Cross-generational fitness benefits of mating and male seminal fluid. Biology Letters. 4:6-8. Qvarnstrom A, Price TD, 2001. Maternal effects, paternal effects and sexual selection. Trends in Ecology & Evolution 16:95-100. R Core Team, 2012. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. ISBN 3-900051-07-0 http://www.R-project.org/ Radwan J, 1993. The adaptive significance of male polymorphism in the acarid mite Caloglyphus-berlesei. Behavioral Ecology and Sociobiology 33:201-208. Radwan J, 1995. Male morph determination in 2 species of acarid mites. Heredity 74:669-673. Radwan J, 2001. Male morph determination in Rhizoglyphus echinopus (Acaridae). Experimental and Applied Acarology 25:143-149. Radwan J, 2009. Alternative mating tactics in acarid mites. Advances in the Study of Behavior 39:185-208. Radwan J, Unrug J, Tomkins JL, 2002. Status-dependence and morphological trade-offs in the expression of a sexually selected character in the mite, Sancassania berlesei. Journal of Evolutionary Biology 15:744-752. Rasband WS, 1997-2011. ImageJ. Bethesda, USA: U. S. National Institutes of Health. Reinhold K, 1999. Evolutionary genetics of sex-limited traits under fluctuating selection. Journal of Evolutionary Biology 12:897-902. Requena GS, Buzatto BA, Martins EG, Machado G, 2012. Paternal care decreases foraging activity and body condition, but does not impose survival costs to caring males in a Neotropical arachnid. Plos One 7:e46701. Rice WR, Chippindale AK, 2001. Intersexual ontogenetic conflict. Journal of Evolutionary Biology 14:685-693. Roff DA, 1994. The evolution of dimorphic traits - predicting the genetic correlation between environments. Genetics 136:395-401. Roff DA, 1996. The evolution of threshold traits in animals. Quarterly Review of Biology 71:3-35. Roff DA, Sokolovska N, 2004. Extra-nuclear effects on growth and development in the sand cricket Gryllus firmus. Journal of Evolutionary Biology 17:663-671. Romo-Beltran A, Macias-Ordonez R, Cordoba-Aguilar A, 2009. Male dimorphism, territoriality and mating success in the tropical damselfly, Paraphlebia zoe Selys (Odonata: Megapodagrionidae). Evolutionary Ecology 23:699-709.

160!! Rossiter MC, 1996. Incidence and consequences of inherited environmental effects. Annual Review of Ecology and Systematics 27:451-476. Rowland JM, Emlen DJ, 2009. Two thresholds, three male forms result in facultative male trimorphism in beetles. Science 323:773-776. Rowland JM, Qualls CR, 2005. Likelihood models for discriminating alternative phenotypes in morphologically dimorphic species. Evolutionary Ecology Research 7:421-434. Rowland JM, Qualls CR, Beaudoin-Ollivier L, 2005. Discrimination of alternative male phenotypes in Scapanes australis (Boisduval) (Coleoptera : Scarabaeidae : Dynastinae). Australian Journal of Entomology 44:22-28. Schowalter TD, 2006. Insect Ecology: An Ecosystem Approach. Burlington, MA: Elsevier. Schwander T, Keller L, 2008. Genetic compatibility affects queen and worker caste determination. Science 322:552-552. Schwander T, Lo N, Beekman M, Oldroyd BP, Keller L, 2010. Nature versus nurture in social insect caste differentiation. Trends in Ecology and Evolution 25:275-282. Schwarz G, 1978. Estimating the dimension of a model. The Annals of Statistics 6:461- 464. Segers F, Berishvili G, Taborsky B, 2012. Egg size-dependent expression of growth hormone receptor accompanies compensatory growth in fish. Proceedings of the Royal Society B-Biological Sciences 279:592-600. Shelly TE, Greenfield MD, 1985. Alternative mating strategies in a desert grasshopper - a transitional analysis. Animal Behaviour 33:1211-1222. Shelly TE, Greenfield MD, 1989. Satellites and transients - ecological constraints on alternative mating tactics in male grasshoppers. Behaviour 109:200-221. Shingleton AW, Frankino WA, Flatt T, Nijhout HF, Emlen DJ, 2007. Size and shape: the developmental regulation of static allometry in insects. BioEssays 29:536-548. Shuster SM, 1987. Alternative reproductive behaviors - 3 discrete male morphs in Paracerceis-sculpta, an intertidal isopod from the northern gulf of California. Journal of Crustacean Biology 7:318-327. Shuster SM, Wade MJ, 2003. Mating Systems and Strategies. New Jersey: Princeton University Press. Shuster, S.M. 2008. The expression of crustacean mating strategies. In: Oliveira RF, Taborsky M, Brockmann HJ, editors. Alternative Reproductive Tactics: an Integrative Approach. Cambridge: Cambridge University Press. p. 224-250.

! 161! Simmons LW, 1986. Intermale competition and mating success in the field cricket, Gryllus-bimaculatus (Degeer). Animal Behaviour 34:567-579. Simmons LW, 1988. The calling song of the field cricket, Gryllus-bimaculatus (Degeer) - constraints on transmission and its role in intermale competition and female choice. Animal Behaviour 36:380-394. Simmons LW, Beveridge M, Krauss S, 2004. Genetic analysis of parentage within experimental populations of a male dimorphic beetle, Onthophagus taurus, using amplified fragment length polymorphism. Behavioral Ecology and Sociobiology 57:164-173. Simmons LW, Ridsdill-Smith TJ, 2011. Ecology and Evolution of Dung Beetles. Oxford, UK: Wiley-Blackwell. Simmons LW, Tomkins JL, 1996. Sexual selection and the allometry of earwig forceps. Evolutionary Ecology 10:97-104. Simmons LW, Tomkins JL, Alcock J, 2000. Can minor males of Dawson's burrowing bee, Amegilla dawsoni (Hymenoptera : Anthophorini) compensate for reduced access to virgin females through sperm competition? Behavioral Ecology 11:319-325. Simmons LW, Tomkins JL, Hunt J, 1999. Sperm competition games played by dimorphic male beetles. Proceedings of the Royal Society of London Series B- Biological Sciences 266:145-150. Simon JC, Pfrender ME, Tollrian R, Tagu D, Colbourne JK, 2011. Genomics of environmentally induced phenotypes in 2 extremely plastic arthropods. Journal of Heredity 102:512-525. Simpson SJ, Sword GA, Lo N, 2011. Polyphenism in insects. Current Biology 21:R738– R749. Sinervo B, Lively CM, 1996. The rock-paper-scissors game and the evolution of alternative male strategies. Nature 380:240-243. Sirot LK, Brockmann HJ, 2001. Costs of sexual interactions to females in Rambur's forktail damselfly, Ischnura ramburi (Zygoptera: Coenagrionidae). Animal Behaviour 61:415-424. Smallegange IM, 2011. Complex environmental effects on the expression of alternative reproductive phenotypes in the bulb mite. Evolutionary Ecology 25:857–873. Smallegange IM, Coulson T, 2011. The stochastic demography of two coexisting male morphs. Ecology 92:755-764. Stewart AD, Pischedda A, Rice WR, 2010. Resolving intralocus sexual conflict: genetic mechanisms and time frame. Journal of Heredity 101:S94-S99.

162!! Suzuki Y, Nijhout HF, 2006. Evolution of a polyphenism by genetic accommodation. Science 311:650-652. Suzuki Y, Nijhout HF, 2008. Genetic basis of adaptive evolution of a polyphenism by genetic accommodation. Journal of Evolutionary Biology 21:57-66. Taborsky M, 2008. Alternative reproductive tactics in fish. In: Oliveira RF, Taborsky M, Brockmann HJ, editors. Alternative Reproductive Tactics: an Integrative Approach. Cambridge: Cambridge University Press. p. 251-299. Thomas GH, Szekely T, 2005. Evolutionary pathways in shorebird breeding systems: Sexual conflict, parental care, and chick development. Evolution 59:2222-2230. Thomas ML, Gray B, Simmons LW, 2011. Male crickets alter the relative expression of cuticular hydrocarbons when exposed to different acoustic environments. Animal Behaviour 82:49-53. Thomas ML, Simmons LW, 2009. Sexual selection on cuticular hydrocarbons in the Australian field cricket, Teleogryllus oceanicus. BMC Evolutionary Biology 9:162. Thornhill R, 1984. Alternative female choice tactics in the scorpionfly Hylobittacus- apicalis (Mecoptera) and their implications. American Zoologist 24:367-383. Thornhill R, Alcock J, 1983. The Evolution of Insect Mating Systems. Cambridge, MA: Harvard University Press. Tinghitella RM, 2008. Rapid evolutionary change in a sexual signal: genetic control of the mutation 'flatwing' that renders male field crickets (Teleogryllus oceanicus) mute. Heredity 100:261-267. Tinghitella RM, Wang JM, Zuk M, 2009. Preexisting behavior renders a mutation adaptive: flexibility in male phonotaxis behavior and the loss of singing ability in the field cricket Teleogryllus oceanicus. Behavioral Ecology 20:722-728. Tollrian R, 1995. Predator-induced morphological defenses - costs, life-history shifts, and maternal effects in Daphnia-pulex. Ecology 76:1691-1705. Tomkins JL, 1999. Environmental and genetic determinants of the male forceps length dimorphism in the European earwig Forficula auricularia L. Behavioral Ecology and Sociobiology 47:1-8. Tomkins JL, Brown GS, 2004. Population density drives the local evolution of a threshold dimorphism. Nature 431:1099-1103. Tomkins JL, Hazel WN, 2007. The status of the conditional evolutionarily stable strategy. Trends in Ecology and Evolution 22:522-528. Tomkins JL, Hazel WN, 2011. Explaining phenotypic diversity: The conditional strategy and threshold trait expression. In: Simmons LW, Ridsdill-Smith TJ,

! 163! editors. Dung Beetle Ecology and Evolution. Oxford, UK: Wiley-Blackwell. p. 107-125. Tomkins JL, Hazel WN, Penrose MA, Radwan J, LeBas NR, 2011. Habitat complexity drives experimental evolution of a conditionally expressed secondary sexual trait. Current Biology 21:569-573. Tomkins JL, Kotiaho JS, LeBas NR, 2006. Major differences in minor allometries: a reply to Moczek. American Naturalist 167:612-618. Tomkins JL, LeBas NR, Unrug J, Radwan J, 2004. Testing the status-dependent ESS model: population variation in fighter expression in the mite Sancassania berlesei. Journal of Evolutionary Biology 17:1377-1388. Tomkins JL, Moczek AP, 2009. Patterns of threshold evolution in polyphenic insects under different developmental models. Evolution 63:459-468. Tomkins JL, Simmons LW, 1996. Dimorphisms and fluctuating asymmetry in the forceps of male earwigs. Journal of Evolutionary Biology 9:753-770. Tomkins JL, Simmons LW, Alcock J, 2001. Brood-provisioning strategies in Dawson's burrowing bee, Amegilla dawsoni (Hymenoptera : Anthophorini). Behavioral Ecology and Sociobiology 50:81-89. Torchio PF, Tepedino VJ, 1980. Sex ratio, body size and seasonality in a solitary bee, Osmia lignaria propinqua Cresson (Hymenoptera: Megachelidae). Evolution 34:993-1003. Triplehorn CA, Johnson NF, 2005. Borror and DeLong's Introduction to the Study of Insects. Belmont, CA: Thomson Brooks/Cole. Truman JW, Riddiford LM, 2002. Endocrine insights into the evolution of metamorphosis in insects. Annual Review of Entomology 47:467-500. Tsubaki Y, 2003. The genetic polymorphism linked to mate-securing strategies in the male damselfly Mnais costalis Selys (Odonata : Calopterygidae). Popul Ecol 45:263-266. Tsubaki Y, Hooper RE, 2004. Effects of eugregarine parasites on adult longevity in the polymorphic damselfly Mnais costalis Selys. Ecological Entomology 29:361-366. Tsubaki Y, Hooper RE, Siva-Jothy MT, 1997. Differences in adult and reproductive lifespan in the two male forms of Mnais pruinosa costalis Selys (Odonata : Calopterygidae). Researches on Population Ecology 39:149-155. Tsubaki Y, Ono T, 1986. Competition for territorial sites and alternative mating tactics in the dragonfly, Nannophya pygmaea Rambur (Odonata, Libellulidae). Behaviour 97:234-252.

164!! Tsubaki Y, Ono T, 1987. Effects of age and body size on the male territorial system of the dragonfly, Nannophya-pygmaea Rambur (Odonata, Libellulidae). Animal Behaviour 35:518-525. Tuttle EM, 2003. Alternative reproductive strategies in the white-throated sparrow: behavioral and genetic evidence. Behavioral Ecology 14:425-432. Unrug J, Tomkins JL, Radwan J, 2004. Alternative phenotypes and sexual selection: can dichotomous handicaps honestly signal quality? Proceedings of the Royal Society of London Series B-Biological Sciences 271:1401-1406. van Dyck H, Wiklund C, 2002. Seasonal butterfly design: morphological plasticity among three developmental pathways relative to sex, flight and thermoregulation. Journal of Evolutionary Biology 15:216-225. van Staaden MJ, Romer H, 1997. Sexual signalling in bladder grasshoppers: tactical design for maximizing calling range. Journal of Experimental Biology 200:2597- 2608. Walker TJ, 1987. Wing dimorphism in Gryllus-rubens (Orthoptera, Gryllidae). Annals of the Entomological Society of America 80(5):547-560. Webster MS, 1992. Sexual dimorphism, mating system and body size in New-World blackbirds (Icterinae). Evolution 46:1621-1641. Wedell N, 1996. Mate quality affects reproductive effort in a paternally investing species. American Naturalist 148:1075-1088. West-Eberhard MJ, 1989. Phenotypic plasticity and the origins of diversity. Annual Review of Ecology and Systematics 20:249-278. West-Eberhard MJ, 2003. Developmental Plasticity and Evolution. Oxford: Oxford University Press. Wheeler DE, 1986. Developmental and physiological determinants of caste in social Hymenoptera - evolutionary implications. American Naturalist 128:13-34. Whitman DW, Ananthakrishnan TN, 2009. Phenotypic Plasticity of Insects: Mechanisms and Consequences. ^ eUSA USA: Science Publishers. Wilkinson GS, 1993. Artificial sexual selection alters allometry in the stalk-eyed fly Cyrtodiopsis-dalmanni (Diptera, Diopsidae). Genetical Research 62:213-222. Wilkinson GS, Taper M, 1999. Evolution of genetic variation for condition-dependent traits in stalk-eyed flies. Proceedings of the Royal Society of London Series B- Biological Sciences 266:1685-1690. Willemart RH, Gnaspini P, 2004. Spatial distribution, mobility, gregariousness, and defensive behaviour in a Brazilian cave harvestman Goniosoma albiscriptum (Arachnida, Opiliones, Gonyleptidae). Animal Biology 54:221-235.

! 165! Willemart RH, Osses F, Chelini MC, Macias-Ordonez R, Machado G, 2009. Sexually dimorphic legs in a neotropical harvestman (Arachnida, Opiliones): ornament or weapon? Behavioural Processes 80:51-59. Wolf JB, Brodie ED, Cheverud JM, Moore AJ, Wade MJ, 1998. Evolutionary consequences of indirect genetic effects. Trends in Ecology & Evolution 13:64-69. Wright S, 1934a. An analysis of variability in number of digits in an inbred strain of guinea pigs. Genetics. 19:0506-0536. Wright S, 1934b. The results of crosses between inbred strains of guinea pigs, differing in number of digits. Genetics 19:0537-0551. Yamaguti HY, Pinto-da-Rocha R, 2009. Taxonomic review of Bourguyiinae, cladistic analysis, and a new hypothesis of biogeographic relationships of the Brazilian Atlantic Rainforest (Arachnida: Opiliones, Gonyleptidae). Zoological Journal of the Linnean Society 156:319-362. Zatz C, Werneck RM, Macias-Ordonez R, Machado G, 2011. Alternative mating tactics in dimorphic males of the harvestman Longiperna concolor (Arachnida: Opiliones). Behavioral Ecology and Sociobiology 65:995-1005. Zera AJ, Rankin MA, 1989. Wing dimorphism in Gryllus-rubens - genetic-basis of morph determination and fertility differences between morphs. Oecologia 80:249-255. Zuk M, Rotenberry JT, Simmons LW, 1998. Calling songs of field crickets (Teleogryllus oceanicus) with and without phonotactic parasitoid infection. Evolution 52:166- 171. Zuk M, Rotenberry JT, Tinghitella RM, 2006. Silent night: adaptive disappearance of a sexual signal in a parasitized population of field crickets. Biology Letters 2:521- 524. Zuk M, Simmons LW, 1997. Reproductive strategies of the crickets (Orthoptera: Gryllidae). In: Choe JC, Crespi BJ, editors. The Evolution of Mating Systems in Insects and Arachnids. Cambridge: Cambridge University Press. p. 89-109.

166!!