UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Speciation and the evolution of mating preferences in threespine sticklebacks (Gasterosteus aculeatus) Albert, Arianne Yvonne Kirk 2006

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2006_fall_albert_arianne.pdf [ 2.8MB ]
Metadata
JSON: 831-1.0092835.json
JSON-LD: 831-1.0092835-ld.json
RDF/XML (Pretty): 831-1.0092835-rdf.xml
RDF/JSON: 831-1.0092835-rdf.json
Turtle: 831-1.0092835-turtle.txt
N-Triples: 831-1.0092835-rdf-ntriples.txt
Original Record: 831-1.0092835-source.json
Full Text
831-1.0092835-fulltext.txt
Citation
831-1.0092835.ris

Full Text

SPECIATION AND THE EVOLUTION OF MATING PREFERENCES IN THREESPINE STICKLEBACKS (GASTEROSTEUS ACULEATUS)   by  ARIANNE YVONNE KIRK ALBERT   B.Sc., University of Victoria, 2001     A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY   in   THE FACULTY OF GRADUATE STUDIES  (Zoology)          THE UNIVERSITY OF BRITISH COLUMBIA  October, 2006  © Arianne Yvonne Kirk Albert, 2006  ii ABSTRACT My Ph.D. research has examined the evolution of mating preferences and their role in speciation. I have addressed these topics empirically, using sympatric species pairs of limnetic and benthic threespine sticklebacks, and theoretically, with multilocus population genetics. Sticklebacks are small fish that occur in lakes, streams and estuaries throughout British Columbia. Most lakes contain one type of stickleback, however, several lakes support two differentially adapted sympatric species: a large benthic form (benthic), and a smaller zooplanktivorous form (limnetic). Chapter 2 examines the role of species interactions in shaping male mating preferences. I determined that the mate preferences of the small species have shifted from preferring large females (the ancestral state) to preferring small females due either to selection against hybrids (reinforcement), or to egg predation by the larger benthic species.  Chapter 3 explores the idea that sexual imprinting may facilitate assortative mating between benthic and limnetic sticklebacks. Sexual imprinting occurs when individuals imprint on the phenotype of their parents, and subsequently prefer mates that resemble their parents. The results suggested that sexual imprinting does not contribute to assortative mating between the sympatric species pairs, implying that genetics are more important than early learning for the formation of mate preferences.  Chapter 4 focuses on differences in male breeding colour between benthics and limnetics. The results reveal that limnetic males have more intense red and blue coloration than benthic and solitary males. These differences in colour could be due to  iii reinforcement, to differences in visual sensitivity of females, or to territorial interactions between males. Chapter 5 examines the evolution of female mating preferences under different scenarios of sex-linkage, when the male display trait is sexually antagonistic. Theoretical analysis suggests that sexually antagonistic traits on the X chromosome (males XY, females XX), females will evolve to prefer mates carrying alleles beneficial to their daughters. In contrast, with a Z-linked trait (males ZZ, females ZW), females more often evolve preferences for mates carrying alleles beneficial to their sons (e.g., flashy displays). This provides an explanation for why males in ZW species have more elaborate sexual displays than males in XY species.   iv TABLE OF CONTENTS Abstract...........................................................................................................................ii Table of Contents ........................................................................................................... iv List of Tables .................................................................................................................vi List of Figures...............................................................................................................vii Acknowledgments........................................................................................................viii Dedication...................................................................................................................... ix Co-Authorship Statement ................................................................................................x CHAPTER 1: General Introduction.................................................................................1 References...................................................................................................................4  CHAPTER 2: Reproductive Character Displacement of Male Stickleback Mate Preference: Reinforcement or Direct Selection?...............................................................7 Introduction.................................................................................................................7 Methods .................................................................................................................... 12 Results ...................................................................................................................... 21 Discussion................................................................................................................. 25 References................................................................................................................. 32  CHAPTER 3: Mate Choice, Sexual Imprinting and Speciation: A Test of a One-Allele Isolating Mechanism in Sympatric Sticklebacks ............................................................ 36 Introduction............................................................................................................... 36 Methods .................................................................................................................... 39 Results ...................................................................................................................... 44 Discussion................................................................................................................. 46 References................................................................................................................. 49  CHAPTER 4: Character Displacement of Male Nuptial Colour in Threespine Sticklebacks (Gasterosteus aculeatus)........................................................................... 51 Introduction............................................................................................................... 51 Methods .................................................................................................................... 54 Results ...................................................................................................................... 61 Discussion................................................................................................................. 68 References................................................................................................................. 73      v CHAPTER 5: Sexual Selection Can Resolve Sex-linked Sexual Antagonism................ 77 Introduction............................................................................................................... 77 Methods and Results ................................................................................................. 79 Discussion................................................................................................................. 84 References................................................................................................................. 85  CHAPTER 6: Conclusions ............................................................................................ 87 Appendix 1: Sexual Selection and Sexual Antagonism Supplementary Material............ 91 The Recursions.......................................................................................................... 91 Species with male heterogamety (XY)....................................................................... 95 Species with Female Heterogamety (ZW).................................................................. 98 Autosomal Trait and Preference Loci ...................................................................... 102 References............................................................................................................... 105  Appendix 2: Animal Care Certificates ......................................................................... 106  vi LIST OF TABLES Table 2.1. Number of trials of each type with each male population. ............................. 16 Table 2.2. Behavioral data for males from mating trials (means ± 1 SE). ....................... 22 Table 4.1. Means (± 1 SE) of the colour measurements. ................................................ 68 Table 5.1. Male and female fitness components in (a) male heterogametic (XY) and (b) female heterogametic (ZW) species. ...................................................................... 80 Table A1.1. Male and female genotypes for the X-linked model.................................... 92 Table A1.2. Male and female fitness components in autosomal model......................... 102   vii LIST OF FIGURES Figure 2.1. Population means in the first principal component of body shape. ............... 14 Figure 2.2. Mean preference score of each male population from choice trials .............. 23 Figure 2.3. Relationship between male preference score and the difference in the size... 23 Figure 2.4. Ln-transformed rates of zigzagging (A) and biting (B)................................. 25 Figure 3.1. The number of females of each type that examined...................................... 45 Figure 3.2. The relationship between the absolute value of the difference in length ....... 45 Figure 4.1. Sidewelling irradiance (µmol/m2/nm) at 1m depth for the lakes................... 57 Figure 4.2. Average reflectance spectra of the throat region for each population............ 62 Figure 4.3. Average reflectance spectra of the belly region for each population. ............ 63 Figure 4.4. Component loadings for the first three principal components of the PCA..... 66 Figure 4.5. Population means for the colour measurements............................................ 69 Figure 5.1. Simulation results for the evolution of female preferences in male heterogametic species (XY).. ................................................................................. 83 Figure 5.2. Simulation results for the evolution of female preferences in female heterogametic species (ZW)................................................................................... 84    viii ACKNOWLEDGMENTS First of all I would like to thank my supervisor Dolph Schluter. His nearly ubiquitous suggestion to “make things stronger” was a driving force for the success of my PhD. Dolph taught me that one of the most important elements of experimental design is that all possible outcomes of the experiment will not only be interesting but publishable. This guidance allowed me to publish the results of an experiment that seemed initially to have failed. Dolph has also apparently never lost confidence in my ability to run experiments and learn new techniques and his confidence and enthusiasm kept me going through times when I was much more doubtful about my abilities. Thanks Dolph!  Secondly, would like to thank the members of my committee for guidance and input over the years. Particularly, Jenny Boughman who showed me the ropes when I first arrived, and who definitely contributed to the success of my first experimental season. She was also and invaluable source of information about the design of behavioural experiments. I would also like to thank Sally Otto for teaching me how to do theory, and for making me feel that math is not beyond my ability.  Thirdly, I want to thank Tim Vines for all his help in the past few years. It’s hard to give him the credit he deserves here except to say that his encouragement helped me to stay on track and stick this out until the end. I would also like to thank the rest of my family for encouragement and support. Thanks Roger, Carolyn, Marika and David!  Fourth, I want to thank the Schluter lab group for comments on papers and discussions on experimental design. The lab has changed somewhat since I started, but these are the people with whom I interacted to make this thesis possible: Jenny Boughman, Jason Weir, Howard Rundle, Anthony Waldron, Kerry Marchinko, Andrew MacColl, and Tim Vines. I also want to thank the various members of the SOWD discussion group who have patiently sat through many stickleback presentations on my behalf.  Finally, I would like to thank all of the people who have assisted me in the lab and field. Their contributions range from field trips to collect fish, to cleaning tanks and looking after my baby fish, to tips on experimental design and analysis. They are: Sandra Nicol, Karen Faller, Celia Chui, Iain Myers-Smith, Nathan Millar, Jean-Sebastien Moore, Sterling Sawaya, Deanna Yim, and Pim Edelaar. Without these people, my experiments would never have got off the ground.    ix DEDICATION   For Tim   In this thesis words and pictures will work together to explain things.   x CO-AUTHORSHIP STATEMENT  Chapter 2 is the result of a collaboration between my research supervisor Dolph Schluter and me. Dolph helped with the experimental design, and data analysis and made many extremely useful editorial comments on the manuscript. I collected and analyzed all of the data, and wrote the manuscript.  Chapter 4 resulted from a collaboration between Nathan Millar, Dolph Schluter and me. Nathan collected all of the colour data and fish for measurement. Dolph assisted in the experimental design and data analysis, and I analyzed the data, and wrote the manuscript.  Chapter 5 was the result of a collaboration between Sarah (Sally) Otto and me. We both worked on the mathematical aspects of the paper, and I was responsible for most of the writing.   1 CHAPTER 1: General Introduction  The process of species formation is ultimately responsible for the diversity of life, and it is becoming increasingly clear that speciation is often driven by adaptation to new environments (Coyne and Orr 2004; Rundle and Nosil 2005; Vines and Schluter 2006). If species are defined as reproductively isolated units, then we can ask how adaptation to new environments changes mating preferences and other mating traits to complete the speciation process. This thesis covers four aspects of this question using recently derived species pairs of threespine sticklebacks (Gasterosteus aculeatus) and mathematical modeling. First, I address how negative interactions between species that stem from adaptation to different environments can strengthen premating isolation, and specifically male mating preferences. Second, I investigate the possibility that imprinting on diverged parental phenotypes could lead to divergent female mating preferences. Third, I describe differences in male colouration between populations of sticklebacks to determine its potential role in reproductive isolation. Finally, I address a more fundamental question about the evolution of female mating preferences under sexual conflict using population genetic theory. The chapters are meant to be self-contained units and go into more background detail than is presented in this introduction. Threespine sticklebacks provide an ideal system for testing hypotheses about speciation and the evolution of mating preferences. After the retreat of the Pleistocene glaciers, marine sticklebacks colonized newly available freshwater habitats and underwent rapid adaptation and radiation in these new environments (Bell and Foster 1994). Most lakes contain one type of sticklebacks (solitaries), however, several low-lying lakes support two sympatric species (McPhail 1984; McPhail 1992; Schluter and  2 McPhail 1992). Each species pair consists of a large benthic invertebrate eating form (benthic), and a smaller streamlined zooplanktivorous form (limnetic). Benthic and limnetic sticklebacks mate assortatively (Nagel and Schluter 1998; Ridgway and McPhail 1984) but are capable of producing fertile hybrids (McPhail 1984, 1992). Sticklebacks therefore provide a system within which we can test various hypotheses about the evolution of reproductive isolation at an early stage in the speciation process. Mating preferences may diverge between allopatric populations as a byproduct of adaptation to different environments. However, they may also be strengthened by interactions between species if sympatric contact is secondarily established. The most commonly proposed method is reinforcement, where selection against hybrids indirectly leads to stronger premating isolation because those individuals that do not hybridize produce the most fit offspring (Servedio and Noor 2003). Another possibility is that interactions other than hybridization between the species may lead to increased premating isolation via direct selection on preferences (Servedio 2001). Threespine sticklebacks can be used to test these hypotheses because hybrids are less fit than parental types (Hatfield and Schluter 1999; Vamosi and Schluter 1999), and there is a possible mechanism of direct selection on male preferences via egg cannibalism (Foster 1994; Foster 1995). In Chapter 2, I examine if there is any evidence that male mating preferences of one sympatric stickleback species (Limnetic) have changed due to negative interactions with females of the other species (Benthic). Specifically, I investigate the relative roles of reinforcement and direct selection on the evolution of male mating preferences. Another role for natural selection in the speciation process is that adaptation to different environments may cause changes in the appearance and behaviour of animals.  3 Such changes may cause individuals to find members of the opposite type unattractive. The mechanisms behind such species recognition are unclear, but it has been suggested that imprinting may play a role. Imprinting is potentially important to speciation as it provides a one-allele mechanism for the evolution of reproductive isolation (Felsenstein 1981; Irwin and Price 1999). Simply put, this means that the diverging populations do not need to fix different mating preference alleles and instead the same allele (mate with someone that looks like your parents) can fix in both populations, or have been fixed in the ancestor. Male threespine sticklebacks care for the eggs and newly hatched fry (Whoriskey and FitzGerald 1994) providing a potential opportunity for females to imprint on the phenotype of their fathers. In Chapter 3, I present the results of experiments aimed at determining if imprinting influences the mating preferences of sympatric threespine stickleback females. Negative interactions between species such as competition and maladaptive hybridization may lead to the evolution of divergent mating preferences as stated above. However, they may also lead to shifts in the phenotypes used by individuals to select mates (Noor 1999). A signature of such a shift is that sympatric populations of the species will be more different from each other for the trait of interest than allopatric populations of similar age (Noor 1999). Finding such a pattern does not necessarily imply that reinforcement was the cause. It does, however, strongly suggest that the divergence in the trait has been driven by some interaction in sympatry. In Chapter 4, I present the results of a survey of male nuptial colouration in sympatric and allopatric populations of sticklebacks and address the likelihood of various interactions in the divergence (or lack thereof) of male colour.   4 The study of speciation by adaptation requires that mating preferences evolve in particular ways. However, the causes of preference evolution outside of the context of speciation are still not well understood. Generally, female preferences for male display traits are thought to enhance a female’s long-term fitness by increasing her offspring’s fitness, either directly or through genetic associations between preference and trait loci (Andersson 1994; Kokko et al. 2003). Previous models assumed that females do not initially express the male display trait, but this is only true if alleles coding for displays are always sex-limited in expression. Initially, display traits will often be sexually antagonistic, as they benefit males but reduce female fitness (Rice and Chippendale 2001). Loci coding for sexually antagonistic traits are disproportionately located on sex chromosomes and contribute substantially to genetic variance (Gibson and Chippendale 2002). In Chapter 5, I present a population genetic model that addresses the evolution of female mating preferences when the trait is expressed in both sexes and is found on the homogametic sex chromosome (X or Z).  REFERENCES Andersson, M. 1994. Sexual Selection. Princeton Univ. Press, Princeton. Bell, M. A., and S. A. Foster.  1994.  The evolutionary biology of the threespine stickleback. Oxford University Press, New York. Coyne, J. A., and H. A. Orr.  2004.  Speciation. Sinauer, Sunderland, MA. Felsenstein, J. 1981. Skepticism towards Santa Rosalia, or why are there so few kinds of animals. Evolution 35:124-138. Foster, S. A. 1994. Inference of evolutionary pattern - diversionary displays of 3-spined sticklebacks. Behav. Ecol. 5:114-121. Foster, S. A. 1995. Understanding the evolution of behavior in threespine stickleback: The value of geographic variation. Behaviour 132:1107-1129.  5 Gibson, J. R., A. K. Chippindale, and W. R. Rice. 2002. The X chromosome is a hot spot for sexually antagonistic fitness variation. Proc. R. Soc. Lond. B. 269:499-505. Hatfield, T., and D. Schluter. 1999.  Ecological speciation in sticklebacks: Environment-dependent hybrid fitness. Evolution 53:866-873. Irwin, D. E., and T. Price. 1999.  Sexual imprinting, learning and speciation. Heredity 82:347-354. Kokko, H., R. Brooks, M. D. Jennions, and J. Morley, 2003. The evolution of mate choice and mating biases. Proc. R. Soc. Lond. B. 270:653-664. McPhail, J. D. 1984. Ecology and evolution of sympatric sticklebacks (Gasterosteus) - morphological and genetic evidence for a species pair in Enos Lake, British-Columbia. Can. J. Zool. 62:1402-1408. McPhail, J. D. 1992. Ecology and evolution of sympatric sticklebacks (Gasterosteus) - evidence for a species pair in Paxton Lake, Texada-Island, British-Columbia. Can. J. Zool. 70:361-369. Nagel, L., and D. Schluter. 1998. Body size, natural selection, and speciation in sticklebacks. Evolution 52:209-218. Noor, M. A. F. 1999. Reinforcement and other consequences of sympatry. Heredity 83:503-508. Rice, W. R., and A. K. Chippindale. 2001. Intersexual ontogenetic conflict. J. Evol. Biol. 14:685-693. Ridgway, M. S., and J. D. McPhail. 1984. Ecology and evolution of sympatric sticklebacks (Gasterosteus) - mate choice and reproductive isolation in the Enos Lake species pair. Can. J. Zool. 62:1813-1818. Rundle, H. D., and P. Nosil. 2005. Ecological speciation. Ecol. Lett. 8:336-352. Schluter, D., and J. D. McPhail. 1992. Ecological character displacement and speciation in sticklebacks. Am. Nat. 140:85-108. Servedio, M. R. 2001. Beyond reinforcement: The evolution of premating isolation by direct selection on preferences and postmating, prezygotic incompatibilities. Evolution 55:1909-1920. Servedio, M. R., and M. A. F. Noor. 2003. The role of reinforcement in speciation: Theory and data. Annual Review of Ecology Evolution and Systematics 34:339-364.  6 Vamosi, S. M., and D. Schluter. 1999. Sexual selection against hybrids between sympatric stickleback species: Evidence from a field experiment. Evolution 53:874-879. Vines, T. H., and D. Schluter. 2006. Strong assortative mating between allopatric sticklebacks as a by-product of adaptation to different environments. Proc. Roy. Soc. Lond. B 273:911-916.   7 CHAPTER 2: Reproductive Character Displacement of Male Stickleback Mate Preference: Reinforcement or Direct Selection?1  INTRODUCTION  Under the biological species concept, reproductive isolation from other populations is the defining feature of a species. Any attempt to understand the processes driving speciation must therefore also unravel the processes behind the evolution of reproductive isolation. The buildup of pre- and post-mating isolation between allopatric populations is relatively easy to imagine, although the mechanisms in nature are still unclear (Schluter 2001; Turelli et al. 2001). Less well understood is the evolution of reproductive isolation between populations that re-establish secondary contact with gene flow after some divergence in allopatry. There are several proposed mechanisms by which reproductive isolation can evolve between populations that are capable of exchanging genes. Reproductive isolation can evolve as a byproduct of adaptation to different ecological niches (Kilias et al. 1980; Dodd 1989; Rice and Hostert 1993; Rundle et al. 2000). This adaptation to different niches can result in part from competition and ecological character displacement in sympatry, which may incidentally produce increased pre-zygotic isolation if traits used in mate choice are those that are diverging through natural selection (Schluter 2000, 2001). For example, if there is ecological character displacement of body size between closely related sympatric species, and size is                                                 1 A version of this chapter has been published. Albert A.Y.K., and Schluter, D. (2004) Reproductive character displacement of male stickleback mate preference: reinforcement or direct selection? Evolution 58:1099-1107.   8 important in mate choice, there may be an increase in pre-zygotic isolation without any direct selection on mating preferences.  Alternatively, selection can cause mating preferences to diverge in sympatry by reinforcement if hybrids are less fit than parental types (Dobzhansky 1940; Noor 1995). If individuals that mate with heterospecifics produce less fit offspring, then pre-zygotic isolating mechanisms might evolve to prevent hybridization (Noor 1995). The reduction in hybrid fitness can be due to genetic incompatibilities and/or ecological or sexual selection against hybrids (Turelli et al. 2001). Regardless of the nature of selection against hybrids, reinforcement requires that linkage disequilibria are established and maintained between preference genes that reduce interspecific matings and the genes responsible for reduced hybrid fitness (Felsenstein 1981; Kirkpatrick and Servedio 1999; Servedio 2001). Therefore, selection on mating preferences during reinforcement is indirect and potentially weak (Servedio 2001; Kirkpatrick and Ravigné 2002). Finally, reproductive isolation can evolve between sympatric populations when there is direct selection on mating preferences. Direct selection occurs when individuals lacking mate discrimination face a direct decrease in fecundity or survival (Servedio 2001), which may happen if individuals that hybridize are exposed to novel parasites, or if one of the species is predatory on the other (or its offspring). Interactions between species such as predation could reduce the probability of interspecific matings without the strong selection against hybrids required by some models of reinforcement (Servedio 2001). Direct selection on mate preference is potentially more effective than indirect selection (e.g. Kirkpatrick 1996; Kirkpatrick and Barton 1997), and may consequently be  9 more important for driving the divergence of mating preferences in sympatry (Servedio 2001).  Of these three models (ecological character displacement, reinforcement and direct selection) most attention has been paid to reinforcement as the major force strengthening pre-zygotic isolation after secondary contact. Yet, the alternatives are rarely ruled out. One major problem associated with trying to determine which of these three mechanisms is occurring in natural populations is that they all predict similar outcomes: a pattern of reproductive character displacement in which greater pre-zygotic isolation develops between sympatric species pairs compared to allopatric species pairs of similar age (Coyne and Orr 1989; Noor 1999). Evidence for reinforcement comes from several empirical examples of reproductive character displacement (see Noor 1999 for review; Höbel and Gerhardt 2003) and from theoretical models demonstrating its plausibility (e.g. Liou and Price 1994; Kelly and Noor 1996; Kirkpatrick and Servedio 1999; Kirkpatrick 2001). However, the importance of direct selection relative to reinforcement is completely unknown (Servedio 2001).  In this report we test for reproductive character displacement between sympatric threespine stickleback species (Gasterosteus aculeatus) using a control for ecological character displacement, and assess the role of direct selection as a cause of the pattern. Threespine sticklebacks are small fish that occur in coastal lakes, streams and estuaries throughout British Columbia, Canada. Most lakes contain allopatric populations of sticklebacks, however, in each of several low-lying lakes a pair of sympatric species coexists. These species pairs are presumed to have formed after separately evolving allopatric populations came into secondary contact during the retreat of the Pleistocene  10 glaciers (Schluter and McPhail 1992). The “benthic” species is large and deep-bodied and forages on benthic invertebrates, while the “limnetic” species is small and streamlined and forages in the open water on zooplankton (McPhail 1984, 1992; Schluter and McPhail 1992). These species mate assortatively (Ridgway and McPhail 1984; Nagel 1994). However, they seem to produce hybrids at a low rate in the wild (McPhail 1984, 1992) and have a history of mitochondrial DNA introgression (Taylor and McPhail 2000) suggesting that hybridization has been ongoing since secondary contact. Allopatric populations of sticklebacks exhibit a range of phenotypes from limnetic-like to benthic-like (Schluter and McPhail 1992), though less extreme, and display similar differences in feeding and ecology (Lavin and McPhail 1986).  There are reasons for thinking that both reinforcement and direct selection have played a role in the evolution of pre-zygotic isolation between sympatric stickleback species since secondary contact was established. F1 hybrids have a lower growth rate than either parental type in their respective habitats (Hatfield and Schluter 1999), and F1 hybrid males have a lower mating success than parentals (Vamosi and Schluter 1999). Furthermore, a mechanism for direct selection on mating preferences has been identified. Benthic females are known to eat stickleback eggs and will raid the nests of males to eat the eggs inside (Foster 1994, 1995). Males reveal the location of their nest during courtship and will often lead females to nests that contain eggs from previous spawnings (Foster 1994, 1995). This potential for egg predation may lead to selection on limnetic males to avoid courting benthic females and thereby revealing the location of their nests (Rundle and Schluter 1998).    11  To distinguish between these alternative mechanisms we focus on male mating preferences. Male mate choice is expected in sticklebacks because males are the sole providers of parental care (Whoriskey and FitzGerald 1994). As expected, male sticklebacks tend to prefer larger and therefore more fecund females, and this tendency is probably ancestral (Sargent et al. 1986; Rowland 1989; Kraak and Bakker 1998). The presence of male mate choice leads to the possibility that males of sympatric species show reproductive character displacement of mate preference.  We tested for reproductive character displacement by comparing the preferences of one population of sympatric limnetic males with two populations of allopatric limnetic-like males. We assessed male preference by allowing males to choose between a limnetic and a benthic female. If there is reproductive character displacement, then limnetic males should prefer the smaller limnetic females, whereas the allopatric limnetic-like males should either show no preference or a preference for the larger benthic females. The use of limnetic-like allopatric populations in our test provides a control for ecological character displacement. If a preference for small limnetic females is a simple by-product of the evolution of a limnetic-like ecotype and morphology, allopatric males should display much the same preference as sympatric males. If allopatric males do not prefer limnetic females, then ecological character displacement is not enough to explain the level of reproductive isolation between sympatric limnetics and benthics.  Reproductive character displacement by itself does not distinguish reinforcement from direct selection on mate preferences. Therefore, we tested for direct selection on male preferences by examining the behavior of males toward their non-preferred female  12 type. If direct selection on limnetic males to avoid predatory benthic females has been important for the evolution of male preference, then we expect male limnetics to alter their courtship behavior towards benthic females relative to limnetic females. Limnetic males could alter their behavior by increasing the level of aggressive behavior towards benthic females, by hiding from them, or by leading them away from the nest area. Limnetic-like allopatric males would not have been exposed to benthics as nest raiders (Foster 1994, 1995), and should not alter their courtship behavior towards benthic females relative to limnetic females. Instead, we expect that allopatric males will show qualitatively the same type of courtship behavior towards benthic and limnetic females but simply display less overall to their non-preferred female type. To our knowledge, this is the first attempt explicitly to test the role of direct selection in the reproductive character displacement of mating preferences.   METHODS Collection and Maintenance of Fish  We collected male limnetic sticklebacks from Paxton Lake (a two species lake) on Texada Island, British Columbia (49º43’N, 124º30’W). Allopatric males were collected from Sproat Lake on Vancouver Island, BC (49º14’N, 124º54’W), and Sakinaw Lake on the Sunshine Coast, BC (49º42’N, 123º58’W). We used benthic and limnetic females from Priest Lake on Texada Island, BC (49º45’N, 124º34’W), another two-species lake that is in a separate drainage from Paxton Lake. This ensured that none of the females were from the same lake as any of the males.  13  The allopatric populations were chosen because they are limnetic-like in morphology (Fig. 2.1) and therefore in ecology (Lavin and McPhail 1986; Schluter and McPhail 1992). We verified this similarity using a landmark based analysis of shape (Rohlf and Marcus 1993; Ptacek 2002). We took digital photographs of preserved specimens of limnetics, benthics and both allopatric populations, and 19 landmarks were placed on each photograph using the image analysis program, tpsDig (Rohlf 2001a). We used the same morphological landmarks as Walker (1997) with an additional five that further outlined the shape of the eye and jaw. The landmarks were analyzed using the program, tpsRelw (Rohlf 2001b), which creates a consensus shape that is standardized for geometric size and rotation, and then calculates partial warp scores (Rohlf and Marcus 1993). The partial warp analysis computes the amount of energy required to “bend” the landmark configuration of each fish to the consensus shape. The relative warps produced are the principal components of the bending energy matrix. The first principal component explained 34.9% of the variation between the partial warp scores of fish from the four populations. The majority of the variation was associated with differences between the populations in body depth, eye size, pelvic girdle length and jaw shape. These morphological traits are associated with differences between limnetic and benthic ecotypes (Schluter and McPhail 1992). The second principal component explained 18.7% of the variation and described the amount of twisting and bowing of the preserved specimens. Since this was not a biologically meaningful measure, we did not use it in further analyses. Similarly, the rest of the principal components were associated with small differences between individual fish rather than differences between populations.   14  Figure 2.1. Population means in the first principal component of body shape (PC1) and body size (standard length) of males from Paxton Lake, Sakinaw Lake and Sproat Lake, as well as benthics from Priest Lake. Error bars indicate ± 1 SE.   We collected fish using minnow traps and dip nets between March and July 2002 as needed to maintain a stock of reproductive males and gravid females. The fish were held in 102 L aquaria, separated by population, at the University of British Columbia. Males from each population that exhibited reproductive coloration or behavior were placed individually in 102 L “mating tanks” containing limestone gravel and a 20 cm diameter dish of sand 2-3 cm deep for a nesting substrate. We supplied java moss (Vesicularia dubyana) to use as nesting material, and two bladderwort bundles (Utricularia sp.) to provide cover. The dish of sand was placed at one end of the aquarium, with a bundle of bladderwort on either side. The sides of the tanks were covered with black plastic to isolate the males from each other visually. All tanks were maintained on a 16:8 light:dark photoperiod at an average temperature of 18ºC. Fish were  15 fed a satiation diet of brine shrimp (Artemia sp.) and bloodworms (chironomid larvae) daily.  Mating Trials  We conducted 60 choice trials and 54 no-choice trials with 60 males. We used a choice design to evaluate the presence of reproductive character displacement and a no-choice design to further differentiate between the predictions of reinforcement and direct selection. We encouraged males to nest by stimulating them with a gravid female in a jar for 10-15 minutes each day. The females were selected from limnetic or limnetic-like populations to ensure that the allopatric males had never encountered benthic females before the trials. (Note that this design does not allow us to distinguish between learned and genetic behaviors towards benthic females. However, previous work has found that assortative mating between benthics and limnetics persists in lab-reared fish (Hatfield and Schluter 1996), suggesting a genetic component to behavioral isolation.) Once a male had built a nest and courted the stimulating female, he was considered ready for trials. We used each male in two trials: one choice and one no-choice separated by at least 24 hours. The order of the two trials was based on the availability of gravid females. After the two trials, males were anesthetized, measured, weighed, and preserved in 10% formalin. After trials, females were measured, stripped of their eggs to confirm their reproductive condition (eggs are easily released when the females are ready to spawn), and then weighed. Nineteen of the choice trials, and 15 of the no-choice trials were discarded because the male did not see both females (displayed less than five behaviors towards one or both females), or because one (or both) of the females was not yet ready to spawn (see Table 2.1 for a breakdown of the trials).   16 Table 2.1. Number of trials of each type with each male population.  Male Type Choice No-choice     Limnetic   Benthic  Paxton (sympatric) 18 9 8 Sakinaw (allopatric) 13 7 8 Sproat (allopatric) 10 7 0 Total 41 39   Choice trials  To determine the level of male mate discrimination or preference we presented sympatric and allopatric males with a choice of two gravid females: one limnetic and one benthic.  We placed the females in a clear Plexiglas box (27.5 cm x 9.5 cm x 14 cm) suspended at the top of the male’s tank, one on either side of an opaque Plexiglas divider. We randomly determined which female was on the left or the right. Opaque dividers extended 5 cm from the front and the bottom of the box so that the males could display to only one female at a time. An observer sitting 1-2m away watched male and female behavior for 10 minutes. The timing and frequency of the following male behaviors were recorded: (1) zigzag, courtship display of males that consists of one or more horizontal darts towards and away from the female; (2) bite, the male bites at the female through the Plexiglas. We also recorded several other courtship behaviors commonly used in stickleback behavioral experiments (e.g., see Rowland 1989; Kraak and Bakker 1998; Nagel and Schluter 1998; Rundle and Schluter 1998), but did not analyze them here. We  17 used females for up to two trials with different males, but we never used the same pair of females twice.   No-choice trials  We conducted no-choice trials to evaluate differences in behavior that males showed towards limnetic and benthic females when only one type was present. The female species used for each male was haphazardly selected based on the availability of gravid females of each type and each male was used in only one no-choice trial. Each female was placed in a clear Plexiglas box (17 cm x 9.5 cm x 14 cm) suspended at the top of the male's tank, allowed to rest for approximately 5 minutes and then released into the tank with the male.  An observer sitting 1-2 m away recorded male and female behavior for 20 minutes or until the female entered the male’s nest (spawned), whichever came first. We only analyzed the behavior from the first five minutes of these trials to look at the male's initial response to the female rather than his response to her later behavior (Rundle and Schluter 1998). Females were used only once in no-choice trials, and were not used in any subsequent choice trials.  Analysis Reproductive character displacement  We tested for reproductive character displacement by calculating a preference score (P), calculated as the standardized difference in the number of zigzags (N) directed towards the two females during the choice trials:   ! P =N(zigzags tolimnetic) " N(zigzags tobenthic)N(zigzags tolimnetic) + N(zigzags tobenthic).           (1)  18 A positive preference score indicates a preference for limnetic females, whereas a negative score indicates a preference for benthic females. We then compared the means of these preference scores among the male populations. The number of zigzags directed towards females is widely accepted as a good indicator of male preference (e.g. Bakker and Rowland 1995; Kraak and Bakker 1998), and all males consistently used zigzags as a part of courtship (A. A. personal observation).   We tested for differences between the male populations in mean preference score by fitting a linear model:  ! Pi = c + "Ri + ei ,               (2) where Pi is the courtship score for the ith male, c is a constant, and ei is the error term. R is an indicator variable that specifies whether male i’s population of origin was sympatric (Ri = 0), or allopatric (Ri = 1). ß is the magnitude of the difference between the male types. A significant ß term would suggest that there are differences between sympatric and allopatric populations in preference for female type. Under reproductive character displacement ß should be negative, indicating that sympatric males have a stronger preference for limnetic females than allopatric males do.  We also fitted the data to the full model, which further distinguished between the two allopatric populations:  ! Pi = c + "Ri + #Qi + ei ,              (3) where Q is an additional indicator variable (Qi = 0 for sympatric males and males from one of the two allopatric populations, and Qi = 1 for males from the remaining allopatric population). A significant γ term indicates that the preferences of the allopatric male populations differ from each other.  19  We tested for discrimination based on the size difference of the females presented in the choice trials to determine if males allocate their zigzags differently as females become increasingly different in size. This is expected because in no-choice mating trials between benthics and limnetics, the probability of hybridization increased as the individuals became closer in size (Nagel and Schluter 1998). If there is reproductive character displacement, then we expect the preference of sympatric limnetic males for the small limnetic females to become stronger as the size difference between the females becomes larger. We also expect the preference of allopatric males for the large benthic females to become stronger as the size difference between females becomes larger. We fit the following linear model to test for an effect of female size difference and male population on the courtship score for each trial:  ! Pi = c + "1zi + "2Ri + "3zi Ri + ei ,             (4) zi is the difference in size between the two females presented to male i (size of benthic female minus size of limnetic female). R is the same indicator variable as in equations (2) and (3). A significant ß2 term would indicate that the relationships for sympatric and allopatric males have different elevations. Under reproductive character displacement ß3 is predicted to be negative, indicating that the degree of preference of sympatric males for limnetic females becomes stronger with increasing difference in female size, opposite to the pattern in allopatric males.  We did not analyze the full model including the Q term (distinguishing the allopatric populations from each other) because it did not improve the fit of the model. Female size was calculated as the first principal component of variation in ln-transformed cube root of size and standard length, all the females combined. The first principal  20 component (body size) explained 98.9% of the variance among females. A positive body size score indicates a long and heavy female (benthic), whereas a negative score indicates a shorter and lighter female (limnetic).     The contribution of direct selection to male preference  To test for direct selection on mating preferences, we compared the behavior of males towards their preferred female type and their non-preferred female type in the no-choice trials. We used only the data of males from Sakinaw Lake and Paxton Lake in this analysis because of a lack of trials with Sproat Lake males and benthic females (Table 2.1). We compared both the rate of zigzagging towards females and the rate of biting using two-way ANOVA. A significant interaction between male type and female type would suggest that sympatric and allopatric males differ in how they behave towards their preferred and non-preferred females. Rates were calculated as the ln-transformed number of occurrences of each behavior divided by the time elapsed (5 min). We expect sympatric males to behave more aggressively towards the nest raiding benthic females than the limnetic females, or to adjust their behavior in some other way. In contrast, we expect allopatric males to show no qualitative difference in the type of behavior directed towards the females, but simply display less overall to the limnetic (non-preferred) females.    In a second test, we analyzed data from the choice trials to determine if they supported the patterns found in the no-choice trials. We calculated the ln-transformed ratio of bites to zigzags displayed by males towards preferred and non-preferred female types. We interpret the ratio of bites to zigzags as reflecting the level of aggression  21 towards each female type. Biting is part of the normal courtship sequence for most stickleback populations, but bites are generally outnumbered by zigzags (Foster 1995). However, in previous trials with limnetic males relative bite frequency rises in displays towards heterospecific females (Nagel and Schluter 1998).  RESULTS Reproductive Character Displacement  Males exhibited reproductive character displacement of preference. Sympatric limnetic males preferred the smaller limnetic females while allopatric limnetic-like males tended to prefer the larger benthic females (Table 2.2, Fig. 2.2). The preference scores of the sympatric males were significantly different from the preference scores of the allopatric males (ß = -0.461, F1,40 = 10.537, p = 0.001, one tailed test). The full model, distinguishing the two allopatric populations, did not fit the data significantly better than the reduced model (F-test, γ= -0.143, F2,39 = 0.639, p = 0.215), suggesting that there was no discernible difference in preference between the two allopatric populations.  Differences between sympatric and allopatric males in mate preference became more exaggerated the larger the difference in body size between the two presented females (Fig. 2.3). As the difference in the size of the two females presented increased, there was a trend for sympatric males to increasingly prefer the smaller (limnetic) female. For allopatric males the relationship was completely the reverse (ß3 = -0.601, F2,38 = 3.263, p = 0.04, one tailed test). In all three populations, discrimination became weaker as females became similar in size.   22 Table 2.2. Behavioral data for males from mating trials (means ± 1 SE).   Male population  Choice trials Paxton Sakinaw Sproat Number of zigzags to limnetic  104.6 ± 17.3 43.2 ± 10.9 51.8 ± 20.7 Number of zigzags to benthic 54.8 ± 10.7 74.1 ± 13.7 50.9 ± 11.7 Preference score 0.27 ± 0.12 -0.30 ± 0.13 -0.14 ± 0.15 ln(bite to zigzag ratio) to limnetic -1.68 ± 0.23 -1.16 ± 0.31 -0.98 ± 0.37 ln(bite to zigzag ratio) to benthic -0.88 ± 0.27 -1.17 ± 0.23 -0.54 ± 0.29 No-choice trials       ln(zigzag rate) to limnetic 2.25 ± 0.26 1.84 ± 0.56  2.36 ± 0.25 ln(zigzag rate) to benthic 1.95 ± 0.19 2.40 ± 0.17  N/A ln(bite rate) to limnetic 0.38 ± 0.32 0.57 ± 0.46  0.63 ± 0.31 ln(bite rate) to benthic 1.62 ± 0.19 1.19 ± 0.35 N/A     23  Figure 2.2. Mean preference score of each male population from choice trials ± 1 SE. A positive preference score indicates a preference for limnetic females, whereas a negative preference score indicates a preference for benthic females.    Figure 2.3. Relationship between male preference score and the difference in the size (PC1) of the two females presented during a choice trial. Populations are: Paxton Lake (open circles and solid line), Sakinaw Lake (grey triangles and dashed line), and Sproat Lake (black squares and dotted line).  24 The Contribution of Direct Selection to Male Preference  Sympatric and allopatric males differed qualitatively in the types of aggressive behavior directed towards their preferred and non-preferred female types in the no-choice trials (Table 2.2, Fig. 2.4). Both sympatric and allopatric males displayed a higher rate of zigzagging towards their preferred female type (limnetic females for sympatric males, benthic females for allopatric males) than to their non-preferred female type. There was therefore no significant interaction between male type and female type (preferred vs. non-preferred) on the rate of zigzagging (F2,31 = 0.181, r = 0.067, p = 0.674). In contrast, sympatric males elevated their rate of biting towards their non-preferred benthic females relative to that towards limnetic females, while allopatric males displayed a lower rate of biting towards their non-preferred limnetic females than towards benthic females. The interaction between male type and female type (preferred vs. non-preferred) on the rate of biting was significant (F2,31 = 7.622, r = 0.241, p = 0.010). Essentially, allopatric males simply displayed a higher frequency of overall courtship behavior towards their preferred female type with zigzags exceeding bites, whereas sympatric males treated their preferred and non-preferred females in qualitatively different ways, directing an excessive frequency of bites towards the non-preferred benthic females. The results from the choice trials show the same pattern of behavior. Sympatric males displayed significantly higher bite to zigzag ratios towards their non-preferred (benthic) female type than towards limnetic females (paired samples t-test, t17 = -3.24, p = 0.005), whereas allopatric males showed no difference in the bite to zigzag ratio displayed towards the different female types (t12 = 0.05, p = 0.962 for Sakinaw males, and, t9 = -1.35, p = 0.220 for Sproat males).   25  Figure 2.4. Ln-transformed rates of zigzagging (A) and biting (B) displayed by males during no-choice trials towards their preferred (solid circles) and non-preferred (open circles) female type. Points are means ± 1 SE.  DISCUSSION  Reproductive character displacement of male preference was established in this experiment by the observed difference in preference displayed by sympatric and allopatric males. Sympatric limnetic males preferred smaller limnetic females, whereas allopatric limnetic-like males preferred larger benthic females. Preference for large females is probably ancestral as it has been seen in all other allopatric and marine stickleback populations tested to date (Sargent et al. 1986; Rowland 1989; Kraak and Bakker 1998). This suggests that male traits underlying mating that decrease the potential for hybridization between benthic and limnetic sticklebacks have evolved in sympatry. Since the allopatric males were similar in phenotype to the sympatric limnetic males, ecological character displacement is not enough to explain the increased level of reproductive isolation between sympatric species pairs. However, there are at least two  26 other processes besides ecological character displacement that could produce a pattern of reproductive character displacement of male preference: direct selection on mate preferences and reinforcement.   Direct selection on mating preferences occurs when individuals displaying a particular mate preference face a direct increase or reduction in fitness or fecundity (Servedio 2001). Benthic females eat eggs in nests guarded by males (Foster 1994, 1995) and this could lead to selection on limnetic males to avoid courting large benthic females and revealing the location of their nest. In contrast, allopatric limnetic-like males have not evolved with benthic egg predation and are therefore free from selection to avoid large females. Direct selection on mate preference due to nest predation by benthics predicts that sympatric limnetic males should be more evasive or behave more aggressively towards benthic females than towards limnetic females. However, allopatric limnetic-like males should not direct more aggression or evasive behavior towards one type of female and should just display less overall towards their non-preferred female type. In agreement with this prediction, we observed differences in the level of aggressive behavior displayed by sympatric and allopatric males towards benthic and limnetic females. Sympatric males decreased their rate of zigzagging and increased their rate of biting towards benthic females compared to their behavior towards their preferred limnetic females. In contrast, allopatric males consistently displayed less overall towards their non-preferred limnetic females but did not increase their relative rate of biting.   These results suggest some influence of direct selection on male preferences in sympatry. However, processes other than direct selection can potentially explain this difference in aggressive behavior seen between males from different populations. Perhaps  27 the increased aggression of limnetic males toward benthic females is the sum of two tendencies characteristic of all populations: an increase in the rate of biting towards benthic females and an increase in the rate of biting towards their non-preferred female type. These two effects would cancel each other in allopatric males causing them to treat benthic and limnetic females similarly. Contrary to this hypothesis, allopatric males did not bite limnetic females more than did sympatric males (Fig. 4), suggesting that they were not elevating bite rates towards their non-preferred type. Although this observation lends support to the idea that an increased rate of biting by sympatric males towards benthic females is an indication of direct selection on courtship behavior, the fact remains that male behavior provides only an indirect test of selection on traits involved in assortative mating and other explanations for differences in behavior are conceivable.   The threat of egg predation by benthic females is not restricted to limnetic males. Benthic males also run the risk of having their eggs eaten by benthic females (Foster 1994). This begs the question: why do benthic males continue to mate preferentially with benthic females when non-predatory limnetic females are available? One possibility is that egg predation by benthic females imposes different costs on benthic and limnetic males. Benthic males are larger than limnetic males and are probably better able to defend their eggs against large benthic females and to control female behavior around the nest. If so, limnetic males may have more to lose than benthic males by courting benthic females. Interestingly, benthic males and limnetic females are slightly more likely to hybridize in the lab than limnetic males and benthic females (Nagel and Schluter 1998). This is concordant with the idea that there is weaker direct selection on benthic males than on limnetic males resulting from mating with the wrong female type.  28  Alternatively, male preference may experience additional, indirect selection via the cost of producing less fit offspring (reinforcement). F1 hybrids are less fit in the wild than the parental species (Hatfield and Schluter 1999; Vamosi and Schluter 1999), providing a possible source of indirect selection on mate preferences through reinforcement in addition to direct selection via egg predation. Reinforcement is potentially less effective than direct selection because it requires linkage disequilibria between alleles for mate preferences and alleles that cause reduction in hybrid fitness (Servedio 2001). However, the effectiveness of reinforcement can be increased when there is physical linkage between these types of alleles (Noor et al. 2001). We have no evidence at present to suggest such linkage is present in sticklebacks but it remains an interesting possibility for study. Reinforcement of male mate preferences, to be effective, also requires that those males preferring to mate with heterospecific females raise fewer conspecific offspring than males preferring conspecific females. This is expected if the behaviors are negatively genetically correlated (males preferring one type of female discriminate against the other type), if sperm limitation constrains the number of clutches that can be fertilized, or if nest size or oxygen requirements of eggs inhibit the number that can be raised successfully to hatching. Spermatogenesis in male stickleback is inhibited during the breeding season but males appear to manufacture enough sperm ahead of time to fertilize an excessive number of clutches (Zbinden et al. 2001). Evidence for negative genetic correlations between behaviors and for inhibitory effects of nest crowding are lacking. The important point, however, is that our evidence for direct selection on male mate preferences in sympatry does not rule out an additional contribution of reinforcement to the pattern of reproductive character displacement. It is  29 likely that both reinforcement and direct selection have been important in shaping male and female mate preferences, but teasing apart their relative contributions remains an important challenge.  One interesting question that arises from this experiment is why sympatric limnetic males still choose to court the large benthic females at all. The answer may lie in the potential benefits that could be gained from mating with a benthic female when no other options are available. Sexual selection theory suggests that males stand to lose less than females by mating with the wrong type (Andersson 1994), and may therefore be less choosy even when the stakes are high. In this case, a limnetic male might increase his fitness by mating with a large benthic female and producing lots of offspring even if those offspring are less fit hybrids. This is likely to be especially true if a male has the choice between mating with a benthic and not mating at all. A similar situation occurs between sympatric species of flycatchers where late in the breeding season females of one species face the choice between hybridization and not breeding at all. In this case females generally choose to mate with males of the other species although their offspring face a reduction in fitness relative to parental types (Veen et al. 2001). Alternatively, because the limnetic species are derived from the marine ancestral type only recently (Taylor and McPhail 2000), it is possible that courtship displays towards larger females have not yet been fully eliminated by selection.  This is not the first experiment to implicate reinforcement and/or direct selection as causes of reproductive character displacement of mate preferences in threespine sticklebacks. Similar evidence for the potential role of direct selection on male preferences comes from a previous study on marine and stream sticklebacks. In the  30 Salmon River in British Columbia, one population of stream resident sticklebacks is sympatric with marines during the breeding season, while a population that is further upstream is not (Borland 1986). Marine sticklebacks are larger than the stream residents and are known egg predators (Foster 1995). In the downstream (sympatric) population males prefer to mate with females that are smaller than themselves. Conversely, the upstream (allopatric) males prefer to mate with females that are larger than themselves, indicating that there has been reproductive character displacement of male preference. Furthermore, consistent with the predictions of direct selection, sympatric males are more aggressive towards females that are larger than themselves, whereas allopatric males show no difference in the level of aggression displayed towards females of different sizes (Borland 1986). This leads to the interesting possibility that direct selection on males to avoid nest predation may also have contributed to the evolution of reproductive isolation between stream resident and marine sticklebacks.  In the limnetic/benthic species pair system, Rundle and Schluter (1998) documented reproductive character displacement of the mating preferences of benthic females. Their test compared the spawning probabilities of sympatric benthic females, presented with either benthic or limnetic males, to spawning probabilities of benthic-like allopatric females. They determined that the sympatric benthic females were less likely to spawn with limnetic males than with benthic males, whereas allopatric benthic-like females responded similarly to both types of males (Rundle and Schluter 1998). Since their experiment also controlled for the effects of ecological character displacement on preferences by using benthic-like allopatric females, the results implicate some process in sympatry that strengthened pre-mating isolation over and above differences due to  31 adaptation to different niches. They viewed reinforcement as the most likely explanation for reproductive character displacement of female preference. However, there is still the possibility that benthic females are under direct selection to avoid mating with small limnetic males that are less able than large benthic males to defend their nests against predatory raids (Rundle and Schluter 1998).   One final explanation for reproductive character displacement we have not yet considered is biased extinction. Biased extinction occurs when allopatric populations vary in their degree of reproductive isolation from each other (Butlin 1987). When some pairs of these populations come into secondary contact, only pairs that are strongly reproductively isolated remain as separate species, while pairs with insufficient reproductive isolation collapse into hybrid swarms (Butlin 1987; Noor 1999). This produces the pattern of greater isolation between sympatric species pairs than between allopatric species pairs. Biased extinction predicts that there should be a range of reproductive isolation between randomly chosen allopatric populations (Rundle and Schluter 1998). Although we have now only tested four allopatric populations (two from Rundle and Schluter (1998) and two here), none of them seem to show anywhere near the level of discrimination displayed by both males and females from species pair lakes (Nagel and Schluter 1998). Therefore, current evidence indicates that biased extinction is unlikely to explain the pattern of reproductive character displacement seen between benthic and limnetic sticklebacks.  Although the results of these experiments support a role for direct selection in the evolution of reproductive isolation between limnetic and benthic sticklebacks, further tests are possible. This experiment provides only indirect evidence of direct selection on  32 male preference. We require some measure of the direct and indirect fitness costs associated with hybridization in order to fully understand how direct selection and reinforcement interact to strengthen reproductive isolation in sympatry. However, these results provide some preliminary evidence that direct selection on mate preferences may have been important for speciation in this system. To our knowledge this is one of the first studies to provide empirical evidence that direct selection on mating preferences in sympatry may contribute to reproductive character displacement.   REFERENCES Andersson, M. 1994. Sexual selection. Princeton Univ. Press, Princeton. Borland, M. 1986. Size-assortative mating in threespine sticklebacks from two sites on the Salmon River, British Columbia. M.Sc. thesis, University of British Columbia, Vancouver. Butlin, R. 1987. Speciation by reinforcement. Trends Ecol. Evol. 2:8-13. Coyne, J. A., and H. A. Orr. 1989. Patterns of speciation in Drosophila. Evolution 43:362-381. Dobzhansky, T. 1940. Speciation as a stage in evolutionary divergence. Am. Nat. 74:312-321. Dodd, D. M. B. 1989. Reproductive isolation as a consequence of adaptive divergence in Drosophila pseudoobscura. Evolution 43:1308-1311. Felsenstein, J. 1981. Skepticism towards Santa Rosalia, or why are there so few kinds of animals? Evolution 35:124-138. Foster, S. A. 1994. Evolution of reproductive behavior of threespine stickleback. in M. A. Bell and S. A. Foster, eds. The evolutionary biology of the threespine stickleback. Oxford Univ. Press, Oxford. ———. 1995. Understanding the evolution of behavior in threespine stickleback: the value of geographic variation. Behaviour 132:1107-1129. Hatfield, T., and D. Schluter. 1996. A test for sexual selection on hybrids of two sympatric sticklebacks. Evolution 50:2429-2434.  33 ———. 1999. Ecological speciation in sticklebacks: environment-dependent hybrid fitness. Evolution 53:866-873. Höbel, G., and H. C. Gerhardt. 2003. Reproductive character displacement in the acoustic communications system of green tree frogs (Hyla cinerea). Evolution 57:894-904. Kelly, J. K., and M. A. F. Noor. 1996. Speciation by reinforcement: A model derived from studies of Drosophila. Genetics 143:1485-1497. Kilias, G., S. N. Alahiotis, and M. Pelecanos. 1980. A multifactorial genetic investigation of speciation theory using Drosophila melanogaster. Evolution 34:730-737. Kirkpatrick, M. 1996. Good genes and direct selection in the evolution of mating preferences. Evolution 50:2125-2140. ———. 2001. Reinforcement during ecological speciation. Proc. R. Soc. Lond. B 268:1259-1263. Kirkpatrick, M., and N. H. Barton. 1997. The strength of indirect selection on female mating preferences. Proc. Natl. Acad. Sci. USA. 94:1282-1286. Kirkpatrick, M., and V. Ravigné. 2002. Speciation by natural and sexual selection: models and experiments. Am. Nat. 159:S22-S35. Kirkpatrick, M., and M. R. Servedio. 1999. The reinforcement of mating preferences on an island. Genetics 151:865-884. Kraak, S. B. M., and T. C. M. Bakker. 1998. Mutual mate choice in sticklebacks: attractive males choose big females, which lay big eggs. Anim. Behav. 56:859-866. Lavin, P.A., and J. D. McPhail. 1986. Adaptive divergence of trophic phenotype among freshwater populations of the threespine stickleback (Gasterosteus aculeatus). Can. J. Fish. Aquat. Sci. 43:2455-2463. Liou, L. W., and T. D. Price. 1994. Speciation by reinforcement of premating isolation. Evolution 48:1451-1459. McPhail, J. D. 1984. Ecology and evolution of sympatric sticklebacks (Gasterosteus): morphological and genetic evidence for a species pair in Enos Lake, British Columbia. Can. J. Zool. 62:1402-1408. ———. 1992. Ecology and evolution of sympatric sticklebacks (Gasterosteus): evidence for a species-pair in Paxton Lake, Texada Island, British Columbia. Can. J. Zool. 70:361-369. Nagel, L. 1994. The parallel evolution of reproductive isolation in threespine sticklebacks. M.Sc. thesis, University of British Columbia, Vancouver.  34 Nagel, L., and D. Schluter. 1998. Body size, natural selection and speciation in sticklebacks. Evolution 52:209-218. Noor, M. A. F. 1995. Speciation driven by natural selection in Drosophila. Nature 375:674-675. ———. 1999. Reinforcement and other consequences of sympatry. Heredity 83:503-508. Noor, M. A. F., K. L. Grams, L. A. Bertucci, and J. Reiland. 2001. Chromosomal inversions and the reproductive isolation of species. Proc. Natl. Acad. Sci. USA. 98:12084-12088. Ptacek, M. B. 2002. Patterns of inheritance of mating signals in interspecific hybrids between sailfin and shortfin mollies (Poeciliidae: Poecilia: Mollienesia). Genetica 116:329-342. Rice, W. R., and E. E. Hostert. 1993. Laboratory experiments on speciation: what have we learned in 40 years? Evolution 47:1637-1653. Ridgeway, M. S. and J. D. McPhail. 1984. Ecology and evolution of sympatric sticklebacks (Gasterosteus): mate choice and reproductive isolation in the Enos Lake species pair. Can. J. Zool. 62:1813-1818. Rohlf, F. J. 2001a. tpsDig version 1.31. Distributed by the author. Department of Ecology and Evolution, State University of New York, Stony Brook, N.Y. ———. 2001b. tpsRelw version 4.10. Distributed by the author. Department of Ecology and Evolution, State University of New York, Stony Brook, N.Y. Rohlf, F. J., and L. F. Marcus. 1993. A revolution in morphometrics. Trends Ecol. Evol. 8:129-132. Rowland, W. J. 1989. The ethological basis of mate choice in male threespine sticklebacks, Gasterosteus aculeatus. Anim. Behav. 38:112-120. Rundle, H. D., and D. Schluter. 1998. Reinforcement of stickleback mate preferences: sympatry breeds contempt. Evolution 52:200-208. Rundle, H. D., L. Nagel, J. W.. Boughman, and D. Schluter. 2000. Natural selection and parallel speciation in sympatric sticklebacks. Science 287:306-308. Sargent, R. C., M. R. Gross, and E. P. van den Berghe. 1986. Male mate choice in fishes. Anim. Behav. 34:545-550. Schluter, D. 2000. The ecology of adaptive radiation. Oxford University Press, Oxford. ———. 2001. Ecology and the origin of species. Trends Ecol. Evol. 16:372-380.  35 Schluter, D., and J. D. McPhail. 1992. Ecological character displacement and speciation in sticklebacks. Am. Nat. 140:85-108. Servedio, M. R. 2001. Beyond reinforcement: the evolution of premating isolation by direct selection on preferences and postmating, prezygotic incompatibilities. Evolution 55:1909-1920. Taylor, E. B., and J. D. McPhail. 2000. Historical contingency and ecological determinism interact to prime speciation in sticklebacks, Gasterosteus. Proc. R. Soc. Lond. B 267:2375-2384. Turelli, M., N. H. Barton, and J. A. Coyne. 2001. Theory and speciation. Trends Ecol. Evol. 16:330-343. Vamosi, S. M., and D. Schluter. 1999. Sexual selection against hybrids between sympatric stickleback species: evidence from a field experiment. Evolution 53:874-879. Veen, T., T, Borge, S. C. Griffith, G. Sætre, S. Bures, L. Gustafsson, B. C. Sheldon. 2001. Hybridization and adaptive mate choice in flycatchers. Nature 411:46-50.  Walker, J. A. 1997. Ecological morphology of lacustrine threespine stickleback Gasterosteus aculeatus L. (Gasterosteidae) body shape. Biol. J. Linn. Soc. 61:3-50. Whoriskey, F. G., and G. J. FitzGerald. 1994. Ecology of the threespine stickleback on the breeding grounds. in M. A. Bell and S. A. Foster, eds. The evolutionary biology of the threespine stickleback. Oxford Univ. Press, Oxford. Zbinden, M., C. R. Largiadèr, and T. C. M. Bakker. 2001. Sperm allocation in the three-spined stickleback. Journal of Fish Biology 59: 1287-1297.  36 CHAPTER 3: Mate Choice, Sexual Imprinting and Speciation: A Test of a One-Allele Isolating Mechanism in Sympatric Sticklebacks2  INTRODUCTION A major question in speciation research is whether or not reproductive isolation between populations evolves via one or two-allele mechanisms (Maynard Smith 1966; Felsenstein 1981; Kirkpatrick and Ravigné 2002; Servedio and Noor 2003). One-allele mechanisms involve the fixation of the same allele at the same locus in diverging populations that increases reproductive isolation as a byproduct. An example is an allele that increases habitat preference when the diverging species inhabit different environments. Two-allele mechanisms involve the fixation of two different alleles at the same or different loci in diverging populations. For example, alleles controlling different male traits and alleles controlling female preferences for different male traits. One-allele mechanisms make speciation theoretically easier because they do not require linkage disequilibrium to form between alleles causing adaptation to different environments and alleles for different mating preferences. However, very few empirical studies have demonstrated the existence of one-allele mechanisms promoting the evolution of reproductive isolation in nature (Servedio and Noor 2003).   A type of one-allele mechanism that may be important in speciation is sexual imprinting (Laland 1994; Irwin and Price 1999). Sexual imprinting occurs when the mate preferences of individuals are formed during early development based on the appearance                                                 2 A version of this chapter has been published. Albert, A.Y.K. (2005) Mate choice, sexual imprinting and speciation: a test of a one-allele isolating mechanism in sympatric threespine sticklebacks. Evolution 59:927-931.  37 of their parents (Irwin and Price 1999; ten Cate and Vos 1999). This can cause reproductive isolation if an allele causing sexual imprinting (or strengthening it) spreads through two diverging populations (Laland 1994; Irwin and Price 1999; Servedio and Noor 2003). For example, if two populations meet in secondary contact after some period in allopatry, sexual imprinting may promote assortative mating if these populations differ in some trait important for mate choice such as song, or body size. This assortative mating can then be strengthened further by reinforcement and other interactions in sympatry (Irwin and Price 1999). The preceding argument illustrates the point that sexual imprinting can only strengthen assortative mating based on traits that already differ between two populations, and these differences are themselves caused by other sources of divergent selection such as competition for shared resources or the fixation of different male display traits (Irwin and Price 1999; Kirkpatrick and Ravigné 2002; Servedio and Noor 2003).  Examples of sexual imprinting within populations are widespread in birds (ten Cate and Vos 1999 and references therein), and have been documented for mammals (Kendrick et al. 1998). However, its importance for promoting assortative mating between diverging populations is less well understood. Interspecific cross-fostering experiments have had mixed results in showing that species recognition is based on sexual imprinting (e.g. Kirchhof-Glazier 1979; Clayton 1990; ten Cate and Vos 1999; Slagsvold et al. 2002). In some cases sexual imprinting is important for species discrimination, while in other cases the ability of species to distinguish con- from heterospecifics does not appear to be influenced by their rearing environment. This begs  38 the question of how important sexual imprinting is for the evolution of reproductive isolation.  Here I test for the presence of sexual imprinting in species pairs of threespine sticklebacks (Gasterosteus aculeatus spp). Threespine sticklebacks are small fish that occur in marine and freshwater habitats around the northern hemisphere. Most lakes contain one type of stickleback, but species pairs coexist in several lakes in British Columbia, Canada. These pairs consist of a large-bodied invertebrate feeding “benthic” species, and a small-bodied zooplankton feeding “limnetic” species (Schluter and McPhail 1992). Benthic and limnetic sticklebacks mate assortatively (Ridgway and McPhail 1984; Nagel and Schluter 1998), but the actual basis of assortative mating is still relatively uncertain. Although there is evidence that males display preferences for different sizes of females (Albert and Schluter 2004), and that females choose to mate assortatively by size (Hatfield and Schluter 1996, Nagel and Schluter 1998; Rundle and Schluter 1998), it is unclear how these preferences are formed within individuals.  Sexual imprinting is plausible in threespine sticklebacks as males provide parental care. Males build nests and entice female to lay eggs in them. The male then tends the eggs and newly hatched fry until they are able to swim freely (Whoriskey and FitzGerald 1994). During this period of parental care, females may imprint on the phenotype of their father causing them to prefer the correct species as adults. Sexual imprinting in this case could have facilitated divergence by increasing the probability of assortative mating upon secondary contact.  To test for effects of sexual imprinting on female preferences, I cross-fostered F1 hybrids of limnetics and benthics to males of both species. I used F1 hybrids because  39 there is no a priori expectation of what their preferences should be, and it allowed me to examine the effects of imprinting and genetics on female preferences. Another advantage of using F1 hybrids is that filial imprinting (sexual preference is based on the phenotype of siblings) is minimized as a factor influencing mate choice. F1 females were raised with other F1 fish and yet never experienced F1 males during preference trials. The use of F1 hybrids makes the assumption that the mechanism of imprinting is the same between benthics and limnetics, which is predicted for a one-allele mechanism. Although we have no direct evidence that the mechanism is the same in both species, there is no reason to expect that there has been divergent natural selection on learning via imprinting.  METHODS Imprinting Design Limnetics and benthics were collected from Paxton Lake on Texada Island between April and July 2003 both for foster-fathers and as parents for the crosses. I made F1 hybrid families in vitro from the gametes of these wild-caught individuals. All F1 crosses were made between different benthic females and limnetic males (11 families total). This provides a control for any possible maternal effects, but does not allow a test for the presence of maternal effects. Males that were foster-fathers for the crosses were placed singly in 102 L aquaria with a dish of sand and plants for nesting material. I stimulated males to build nests by showing them gravid females in jars for 15-30 min daily. Once males had built nests, they were considered ready to foster an F1 clutch. Males fostered one clutch each, and were not reused.  40  In preliminary trials I found that males ate foreign eggs that were added to their nests. I therefore placed the nest of each male containing a family of F1 eggs into a 250 ml jar covered with mesh to allow water flow. This jar was placed into the male’s tank in the location where his nest had been. I added an aerator to the jar to ensure that the eggs were properly oxygenated. Six F1 families were fostered with limnetic males and five F1 families were fostered with benthic males. One day prior to hatching, males were put into 1000 ml jars with mesh tops and aerators to prevent them from eating the fry. Once the fry hatched, they were gently shaken from their jars into the male’s tank. One month post-hatching, males were removed from the tanks entirely.  Although this setup did not allow for actual physical contact between foster-fathers and the eggs and fry, it did allow for both chemical and visual signals to pass between them. Chemical imprinting may occur if females imprint on the odour of the kidney secretions that males use to build their nests, in addition to or instead of visual cues. All fostered F1 families were raised in the laboratory until reaching sexual maturity (approximately 10 months) and were fed brine shrimp (Artemia sp) and chironomid larvae to satiation daily.   Testing F1 Female Preference  I assessed the preferences of individual fostered females using two separate no-choice trials. In one trial each female was added to an aquarium containing a nesting limnetic male. In the other trial the female was added to an aquarium containing a nesting benthic male. Wild limnetic and benthic males from Paxton Lake were caught between  41 March and June 2004 and placed singly in 102 L mating tanks. These tanks contained a dish of sand, two plastic plants for cover, and java moss (Vesicularia dubyana) for nesting material. I stimulated males to build nests as above. Once a male had built his nest he was used in up to four trials with different females. In total, 12 benthic males and 11 limnetic males were used in the trials. F1 females were tested opportunistically as they came into reproductive condition. I tested 17 females that had been raised with limnetic foster-fathers and 19 females that had been raised with benthic foster-fathers. The order of trials (limnetic or benthic male first) was alternated between females of each type to ensure that there were equal numbers starting with either type of male. I conducted 20 min no-choice trials following protocol used in previous studies (e.g., Hatfield and Schluter 1996; Rundle and Schluter 1998; Nagel and Schluter 1998; Albert and Schluter 2004). Courtship behaviors of both the male and the female were recorded for each trial by an observer sitting 1m away. At the end of a trial, I measured the length and weight of the male and female. Females were allowed to rest in a covered container for at least 30 min between the two trials.   I used female nest examination (the penultimate behavior in the stickleback courtship sequence [Hatfield and Schluter 1996]) as a measure of female preference. I used examination instead of actual spawning because very few (two) females spawned with males during the experiment, and nest examination has been used in previous experiments as a measure of female preference (e.g., Hatfield and Schluter 1996; McKinnon et al. 2004).    42 Analysis I tested for an effect of imprinting on female preference by comparing 2 x 2 contingency tables counting the number of females of both types (limnetic-reared or benthic-reared) that examined the nest of one, both, or neither males. The contingency tables were compared via a χ2 test of heterogeneity (Sokal and Rohlf 1995), which tests whether or not two or more tables have the same structure. The χ2 test of heterogeneity would be significant if the two types of female had different probabilities of examining the nests of the different males. To determine if the females of each type had an overriding preference for one type of male, I used Fisher exact tests to test for independence of the categories within each table.  As an additional test, I used logistic regression to determine if there was an effect on nest examination by individual females of foster-father type, male type in the trial, the absolute value of the difference in body length between the male and the female in the trial, and their interactions. Because the two trials with each female were not independent, I only used the first trial for each female in this analysis. In addition, the length measurements were missing for some males reducing the sample size for this test to 34. An effect of the imprinting treatment on female preference predicts a significant interaction between foster-father type and male type in the logistic regression analysis. I assessed the significance of each fixed effect by comparing the change in the deviance (ΔD) of the model caused by the addition of that factor with a Chi-square distribution (df = 1 for continuous variables and k-1 for factors, with k number of levels of the factor) (Sokal and Rohlf 1995; Hardy and Field 1998). In addition, the tests of absolute value of size difference are one-tailed because there is an a priori expectation of size assortative mating in sticklebacks (Nagel and Schluter 1998; McKinnon et al. 2004).   43 Since individual males were used multiple times (mean=3), and multiple females were used from each family (max=8, mean=4), I conducted four additional tests for effects of male identity and family on the results. In the first two tests, I added male identity or family as fixed effects in logistic regressions. However, as both had non-significant effects on nest examination (p>0.05), I did not include them further. In the next tests, I took account of the variance between individual males and families by including them as random factors in mixed-effects logistic regressions. The fit of the mixed-effects logistic regressions were assessed using the Penalized Quasi Likelihood (glmmPQL) function in R. Adding male identity and family as random effects changed the results slightly, and so these analyses were included.  In addition to the tests on occurrences of examination, I looked for differences between the two male types in the amount of time before the first nest examination by females. For this test, I used data only from females who examined at least one of the nests in her two trials. The prediction, if there was an effect of the imprinting treatment, is that limnetic-reared females would examine limnetic nests sooner than benthic nests, and vice versa for benthic-reared females. I used paired samples Wilcoxon signed rank tests on the ln-transformed time to nest examination (in seconds) for each type of female separately. If the female never examined the male’s nest, the time was set to 1200s, which was the length of the trial. The time was ln-transformed to achieve a symmetrical distribution of differences. All statistical analyses were carried out in the R language (R Development Core Team 2004).    44 RESULTS  The χ2 test of heterogeneity found no significant difference in examination frequency between the two types of females (χ21 = 1.41, p > 0.1), suggesting that they did not differ in their probability of examining the nests of the two types of males (Fig. 1). There was no significant difference in the proportion of females examining the limnetic male, the benthic male, both or neither for either type of female (Fisher exact tests: limnetic-reared females p = 0.10, benthic-reared females p = 0.65). This suggests that there was no effect of the imprinting treatment on female preference. Because of the non-significance of the χ2 test of heterogeneity, I pooled all females to see if there was an overriding preference for one type of male. Again, there was no significant difference in the proportion of females examining one, both or neither type of male (Fisher exact test: p = 0.16), suggesting that F1 females do not prefer one type of male to the other (Fig. 3.1). This is especially evident in the high proportion of females (25%) that examined the nests of both types of males. There was no significant interaction between male type and foster father type in determining the probability of nest examination (logistic regression ΔD1 = 0.97, p = 0.32), and so the interaction was removed from subsequent models. Neither male type (ΔD1 = 3.38, p = 0.07), nor foster father type (ΔD1 = 0.85, p = 0.36) were significant. In contrast, the absolute value of the difference in length (mm) between the female and the male (|female length – male length|) was a significant predictor of whether the female examined the nest (one tailed: ΔD1 = 10.05, p = 0.01). The probability of the female examining the nest decreased as the difference in body size increased (Fig 3.2).  45   Figure 3.1. The number of females of each type that examined the limnetic nest only, the benthic nest only, both types of nests, or neither nest in her two trials.   Figure 3.2. The relationship between the absolute value of the difference in length between the male and female in a trial (|flength-mlength|), and whether or not she examined the nest. The female types are shown by different symbols (circles = limnetic-reared, and triangles = benthic-reared). The line is the predicted probability of nest examination from a logistic regression of examination on the absolute value of length difference.  46 When male and family identities were included as random effects, the significance of the absolute value of the size difference was weakened (one-tailed: p = 0.04 with male id, and p = 0.08 with family id). This makes sense because variation in body size between individual test males and between female families accounted for some of the effect of size in individual trials. However, the interaction between male type and foster father type, and either of these factors singly, remained non-significant when either male identity (foster father p = 0.34, male type p = 0.13), or family identity (foster father p = 0.42, male type p = 0.09) was added to the model. This suggests that any imprinting effect that may exist is less important to assortative mating than body size differences.  Consistent with the results for occurrence of nest examination, there was no difference in the length of time to the first nest examination between benthic and limnetic males for either type of female (Wilcoxon signed rank tests, limnetic-reared females p = 0.08, benthic-reared females p = 0.91).   DISCUSSION The results of this experiment suggest that there is little or no effect of sexual imprinting on the development of female preferences in the species pairs of threespine sticklebacks. Females that were raised by benthic males were not more likely to examine the nests of benthic males than females raised by limnetic males (and vice versa). This lack of an imprinting effect has important implications for the evolution of reproductive isolation between these species.  The primary conclusion is that the ability of females to distinguish con- from heterospecifics must be genetically controlled as opposed to learned. Comparisons can be  47 made between the behavior of the F1 females in this experiment, and females of the parental types in another experiment (Hatfield and Schluter 1996), to estimate the genetic basis of female preferences. Parental females displayed preferences for their own type of male, whereas the F1 females in this experiment did not prefer one type of male to the other. Although the details of the genetic architecture are impossible to determine with this design, the lack of preference in the F1 females suggests that there is little or no dominance of one type of preference. Admittedly, this experiment was not designed to extract quantitative genetic data about female preferences, and this remains an interesting avenue for further research.  The lack of evidence for sexual imprinting in this experiment does not necessarily rule out its existence. First, if there is a difference in the mechanism of imprinting in limnetics and benthics, an examination of cross-fostered individuals of the pure species would provide better evidence for or against a role of imprinting. However, it seems unlikely that the ability to imprint would be under divergent natural selection between the species pairs.  Second, the crossing design did not allow for an assessment of the role of maternal effects on imprinting. Although we found no evidence for sexual imprinting in one cross direction (benthic female by limnetic male), it is possible that the reciprocal cross would have exhibited imprinting. Previous experiments have failed to find any evidence for maternal effects on morphology, growth rate (McPhail 1992; Hatfield 1997), and male attractiveness to females (Hatfield and Schluter 1996), suggesting that the identity of the maternal species does not influence these traits.   48 Third, divergent female preferences that were originally shaped by imprinting may have become genetically fixed in sympatry (Irwin and Price 1999). One way to determine if this is true is to look for evidence of sexual imprinting in allopatric populations of sticklebacks that have not been under selection for greater mate discrimination due to hybridization with a sympatric species.  Fourth, females were not allowed to choose between males simultaneously in this experiment, possibly masking their preferences (Wagner 1998). However, a pilot experiment allowing for female choice in a semi-natural pond setting resulted in the same outcome as the no-choice setting. When given a choice, the fostered F1 females did not prefer their foster father type (six out of ten females made the “correct” choice), and they did not prefer one type of male to the other (six out of ten females mated with a benthic male). Remarkably, the absolute value of the difference in length between the male and female was more important than the test male species and the female’s foster-father in determining female preference. This reconfirms the importance of size-assortative mating in the evolution of reproductive isolation between differentially adapted forms of sticklebacks (Nagel and Schluter 1998; McKinnon et al. 2004). The fact that F1 hybrids between parental types still exhibit a tendency for size-assortative mating suggests that it is controlled by the same genetic mechanism in benthics and limnetics. Furthermore, McKinnon et al. (2004) documented size-assortative mating in marine and stream pairs of sticklebacks, suggesting that it may be an ancestral feature of stickleback mating behavior.   49 Ultimately, it seems clear that imprinting is not presently important for explaining assortative mating between benthic and limnetic sticklebacks, and that divergent selection on size and a mechanism for size-assortative mating are more important for the evolution of reproductive isolation.  REFERENCES Albert, A. Y. K. and D. Schluter. 2004. Reproductive character displacement of male stickleback mate preference: reinforcement or direct selection? Evolution 58:1099-1107. Clayton, N. S. 1990. The effects of cross-fostering on assortative mating between zebra finch subspecies. Anim. Behav. 40:1102-1110. Felsenstein, J. 1981. Skepticism towards Santa Rosalia, or why are there so few kinds of animals? Evolution 35:124-138. Hardy, I. C. W., and S. A. Field. 1998. Logistic analysis of animal contests. Anim. Behav. 56:787-792. Hatfield, T. 1997. Genetic divergence in adaptive characters between sympatric species of stickleback. Am. Nat. 149:1009-1029. Hatfield, T., and D. Schluter. 1996. A test for sexual selection on hybrids of two sympatric sticklebacks. Evolution 50:2429-2434. Irwin, D. E., and T. Price. 1999. Sexual imprinting, learning and speciation. Heredity 82:347-354. Kendrick, K. M., M. R. Hinton, and K. Atkins. 1998. Mothers determine sexual preference. Nature 395:229-230. Kirchhof-Glazier, D. A. 1979. Absence of sexual imprinting in house mice cross-fostered to deer mice. Physiol. Behav. 23:1073-1080. Kirkpatrick, M., and V. Ravigné. 2002. Speciation by natural and sexual selection: models and experiments. Am. Nat. 159:S22-S35. Laland, K. N. 1994. On the evolutionary consequences of sexual imprinting. Evolution 48:477-489. Maynard Smith, J. 1966. Sympatric speciation. Am. Nat. 100:637-650.  50 McKinnon, J. S., S. Mori, B. K. Blackman, L. David, D. M. Kingsley, L. Jamieson, J. Chou, and D. Schluter. 2004. Evidence for ecology’s role in speciation. Nature 429:294-298. McPhail, J. D. 1992. Ecology and evolution of sympatric sticklebacks (Gasterosteus): evidence for a species-pair in Paxton Lake, Texada Island, British Columbia. Can. J. Zool. 70:361-369. Nagel, L., and D. Schluter. 1998. Body size, natural selection and speciation in sticklebacks. Evolution 52:209-218. R Development Core Team. 2004. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-00-3, URL http://www.R-project.org. Ridgway, M. S., and J. D. McPhail. 1984. Ecology and evolution of sympatric sticklebacks (Gasterosteus): mate choice and reproductive isolation in the Enos Lake species pair. Can. J. Zool. 62:1813-1818. Rundle, H. D., and D. Schluter. 1998. Reinforcement of stickleback mate preferences: sympatry breeds contempt. Evolution 52:200–208. Schluter, D., and J. D. McPhail. 1992. Ecological character displacement and speciation in sticklebacks. Am. Nat. 140:85-108. Servedio, M. R., and M. A. F. Noor. 2003. The role of reinforcement in speciation: theory and data. Annu. Rev. Ecol. Syst. 34:339-364. Slagsvold, T., B. T. Hansen, L. E. Johannessen, and J. Lifjeld. 2002. Mate choice and imprinting in birds studied by cross-fostering in the wild. Proc. R. Soc. Lond. B 269:1449-1455. Sokal, R. R., and F. J. Rohlf. 1995. Biometry. 3rd ed. W. H. Freeman, New York. ten Cate, C., and D. R. Vos. 1999. Sexual imprinting and evolutionary processes in birds: a reassessment. Adv. Stud. Behav. 28:1-31. Wagner, W. E, Jr. 1998. Measuring female preference. Anim. Behav. 55:1029-1042. Whoriskey, F. G., and G. J. FitzGerald. 1994. Ecology of the threespine stickleback on the breeding grounds. Pp. 188-206 in M. A. Bell and S. A. Foster, eds. The evolutionary biology of the threespine stickleback. Oxford Univ. Press, Oxford.  51 CHAPTER 4: Character Displacement of Male Nuptial Colour in Threespine Sticklebacks (Gasterosteus aculeatus)3  INTRODUCTION In many taxa the predominant differences between closely related sympatric species are in mating signals such as colouration or song type. Many processes may cause such divergence, but a role for interspecific interactions in sympatry can be inferred if the differences in mating signals are greater in sympatry than when either species occurs alone (i.e. character displacement: Brown and Wilson, 1956; Schluter, 2000a, b).  Character displacement in mating signals is usually explained by the action of reinforcement, which occurs when selection against hybrid matings causes an increase in premating isolation between the species (Dobzhansky, 1940; Brown and Wilson, 1956; Servedio and Noor, 2003). Reinforcement is often measured as an increase in female mating discrimination in sympatry, but it is also implicated in cases of character displacement in mating signals (e.g., Höbel and Gerhardt, 2003; see Servedio and Noor, 2003 for review). There are two potential effects of reinforcement on signaling traits: we either expect female preferences to become more narrow and hone in on differences already present between species (no character displacement in signals), or we expect females to prefer more extreme males and drive divergence in male signaling traits (Noor, 1999).                                                 3 A version of this chapter has been accepted for publication. Albert, A.Y.K., Millar, N.P. and Schluter D. Character Displacement of Male Colour in Threespine Sticklebacks (Gasterosteus aculeatus). Accepted by Biol. J. Linn. Soc. April, 2006.  52 Despite the attraction of reinforcement as an explanation for character displacement in mating signals, other interspecific interactions can cause the same pattern (Noor, 1999; Servedio and Noor, 2003). One such interspecific interaction is competition between species for signal space, which may drive divergence of signals in sympatry. For example, the acoustic signals of different frog species may interfere with each other, causing poor communication between males and females, which may lead to dispersion of signals to avoid overlap and interference in sympatry (Chek, Bogart and Lougheed, 2003). Alternatively, species may only be able to coexist in sympatry if their signals are substantially different (species sorting). Regardless of whether there has been evolution of signals in sympatry or if displacement is due to species sorting, competition for signal space causes a pattern of character displacement.  Aggressive interactions between species may also contribute to character displacement in mating signals. Divergence of plumage colour between sympatric populations of collared and pied flycatchers is driven at least partly by the fact that only dully coloured pied flycatcher males can set up territories in the presence of collared flycatcher males without being attacked (Alatolo, Gustafsson and Lundberg, 1994). Similarly, displacement of wing spot size in the damselfly Calopteryx splendens seems to be due to aggressive territorial interactions with the sympatric C. virgo (Tynkkynen, Rantala and Suhonen, 2004). C. splendens males with large wing spots more closely resemble C. virgo males, and therefore suffer more aggression than males with smaller spots.  Competition between species for resources may lead to displacement in morphological traits that then lead to divergence in mating signals as a secondary  53 consequence. Galapagos finches are a classic example of ecological character displacement in beak morphology driven by competition for shared resources (Schluter, Price and Grant, 1985). It was recently shown that beak size and shape limit the range of songs that a species can sing, and thus divergence in beak shape has also driven divergence in song, a cue used for species recognition in finches (Podos, 2001). Ecological character displacement may also influence divergence in mating signals in a more subtle way, if displacement in habitat or diet indirectly changes the signaling environment that in turn drives changes in mating signals (Schluter, 2000b; Boughman, 2002).  Regardless of the cause, finding character displacement of mating signals suggests that signals are not diverging by arbitrary changes in female preference or genetic drift. Instead, differences are driven in predictable directions through interspecific interactions. In this report, we evaluate the evidence for character displacement of male colouration between sympatric species of threespine sticklebacks (Gasterosteus aculeatus spp.). Along the coast of British Columbia, most lakes contain one population of sticklebacks (allopatric), but several low-lying lakes support independently derived sympatric species pairs (Schluter and McPhail, 1992; Taylor and McPhail, 2000). Each pair consists of a large, benthic invertebrate feeder, the "benthic" species, and a small, streamlined, zooplanktivore, the "limnetic" species (McPhail, 1984, 1992; Schluter and McPhail, 1992).  During the breeding season males build nests, court females and then care for the eggs and newly hatched fry (Whoriskey and FitzGerald, 1994). Generally, male sticklebacks develop red throats, blue or blue-green bodies and blue irises during the  54 breeding season. Female sticklebacks display a preference for males with the most intensely red throats (e.g. McLennan and McPhail, 1990; Milinski and Bakker, 1990) and with the largest area of red colouration (Boughman, 2001). Whether or not females have any preference for the relative area or intensity of the blue body colouration remains unknown. We tested for character displacement of male colour by comparing the male colour (both red and blue) of benthics and limnetics from two species-pair lakes (sympatric) with that of males from three allopatric lakes. This is the same type of comparison previously used to detect character displacement in foraging traits (Schluter and McPhail 1992) and armour (Vamosi and Schluter, 2004), with the assumption that the allopatric populations represent the derived solitary freshwater state. Differences in red colour between sympatric sticklebacks have already been shown (McPhail, 1984; Boughman, 2001; Boughman, Rundle and Schluter 2005). However, comparison with allopatric populations in otherwise similar lakes allows us to determine whether the differences in sympatry are unusually large. They also provide a reference to determine which sympatric species departs most from the expected single-species state, and in which direction, and provide evidence that interspecific interactions were responsible for divergence.   METHODS Fish Collection We measured the nuptial colouration of males from five lakes: two species pair lakes, Paxton Lake (49°42’N, 124°31’W) and Priest Lake (49°44’N, 124°33’W), and  55 three allopatric lakes, Klein Lake (49°43’N, 123°58’W), Trout Lake (49°30’N, 123°52’W) and Cranby Lake (49°41’N, 124°30’W). We chose the allopatric populations that have the most similar ecological conditions to the species pair lakes (aside from containing two species). The lakes are small and contain cutthroat trout (Oncorhynchus clarki) as the only other fish species present. In addition, these lakes have been used in previous analyses of character displacement of other traits in the species pairs (Schluter and McPhail, 1992; Vamosi and Schluter, 2004), and are assumed to represent the expected phenotype when only one type is present in a lake. Unfortunately, we were only able to use two of the four previously described species pairs (Schluter and McPhail, 1992; McPhail, 1993). This was because one pair (Hadley Lake) has gone extinct following the introduction of brown bullhead. A second pair (Enos Lake) is in the advanced stages of collapse via hybridization for reasons that are still unclear (Kraak, Mundwiler and Hart, 2001; Taylor et al. 2006).  We collected sticklebacks using minnow traps and dip nets in May, June and July 2003. Fish were kept in 102 L mixed-sex tanks separated by population at the University of British Columbia, Vancouver, Canada. All fish were maintained at approximately 18 ºC on an 16:8 light:dark cycle and fed chironomid larvae and brine shrimp (Artemia sp.) daily to satiation. We assessed male colour in early July 2003.  Water Colour Measurement  We measured the background colour of the water within each lake using a dual channel Ocean Optics (Dunedin, Florida) SD200 spectrometer and a 200 µm UV/VIS reflectance probe attached to a CC-3-UV cosine corrector. The sidewelling irradiance  56 was recorded for depths of 10 cm, 50 cm, 1 m and 2 m (Fig. 4.1). Sidewelling irradiance provides a measure of the colour of the background against which male fish are viewed by females (McDonald and Hawryshyn, 1995). We calculated λP50 for the sidewelling light at each depth. λP50 measures the dominant wavelength of the sidewelling spectrum, and is calculated as the wavelength that halves the area under the irradiance curve. These values were used to determine if the water colour differed significantly between lakes used in this study.  The λP50 values of the sidewelling light for all of the lakes used were similar to each other and typical of blue-green lakes generally (Novales Flamirique, Hendry and Hawryshyn, 1992; McDonald and Hawryshyn, 1995). The range of λP50 values for the lakes (from depths of 10 cm to 2m) were: Priest = 562-577 nm, Paxton = 561-577 nm, Cranby = 575-597 nm, Klein = 570-576 nm, and Trout = 549-558 nm. Redshifted lakes typically have λP50 values higher than 600 nm (Novales Flamirique et al. 1992; McDonald and Hawryshyn, 1995). λP50 values increased slightly with depth from 10 cm to 2 m, however, there is no evidence to suggest that any of the lakes were particularly redshifted, which would cause a reduction in the visibility of the red throat colouration. There was also no correlation between λP50 at 1 m depth (or any other depth) and the average red score in the lakes (Pearson’s r = -0.04, p = 0.933). Since the data show no association with the minor differences between lakes in water colour, we have not included them in subsequent analyses.  57   Figure 4.1. Sidewelling irradiance (µmol/m2/nm) at 1m depth for the lakes in this study. Maximum irradiance is not identical for each lake as weather conditions differed when measurements were made. However, the shapes of the curves are comparable. A) Cranby lake, B) Klein lake, C) Trout lake, D) Paxton lake, and E) Priest lake.   Male Colour Measurement Males were chosen for measurement if they displayed nuptial colouration (red throats and blue bodies and irises). Only the most colourful males from a tank were  58 chosen for measurement at any one time. In most tanks, one male asserted himself as the dominant male (N. M. pers. obs.), an attribute most visible through behaviour and colour. The dominant male chased and bit the other males, and exhibited full nuptial colouration, whereas the other males had subdued colouration. When the dominant male was removed from the tank for a brief period (a few hours), another male became dominant and changed his behaviour and colouration. Since males were not returned to their tanks after measurement, we were able to measure all males over a period of days allowing each to develop to his maximum nuptial colouration under laboratory conditions. Male colouration is thus highly dependent on social context. For example, male sticklebacks are capable of quickly increasing the colour saturation of their throat and iris when confronted with reproductive females or with other territorial males (Rush et al., 2003). Since the males we used were not guarding nests or courting females at the time of measurement, it is possible that they were not as colourful as wild males at their peak. Nevertheless, males were able to interact with other males in dominance hierarchies as described above. Previous experiments have shown a strong correlation between the wavelength and intensity of colours that males use for both intra- and intersexual interactions (McLennan and McPhail, 1990; Rowland, 1994; Baube, 1997; Rush et al., 2003). Therefore, it seems likely that the most colourful males in the holding tanks were expressing similar levels of colour saturation as they would if courting females. In addition, all males were housed under the same conditions, and differences in colour that would be visible in the wild should still be present in the laboratory. Clearly, however, this experiment provides only a first step into understanding the dynamics of male colour evolution. Further colour measurement of free-swimming males (e.g., Rush et al., 2003)  59 at multiple stages of the nesting cycle and at different times in the season will be required to more fully describe the colour of males from different populations. Fish were individually anesthetized using carbonated water in a darkened container for 30 to 60 seconds. After the fish were anesthetized, we took photographs and reflectance measurements (described below). Fish were then either left to recover and placed in new tanks or were given an overdose of anesthetic (MS-222) and preserved in 95% ethanol. All fish care and measurement complied with the University of British Columbia animal care regulations.  Photography and Visual Assessment of Red Area We photographed the left side of males with a Nikon D1H digital SLR camera. The fish were illuminated with four halogen lights (120 V, 50 W). We mounted all of the equipment on a frame to ensure that all components (fish, camera and lights) maintained their relative spatial relationships with each other throughout. The relative extent of the red throat colour was assessed by visual examination of the digital photographs on a standard computer monitor. The photographs were labeled with random numbers and assessed in that order to prevent bias resulting from measuring all fish from one population in a block and from knowing the identity of the population. Using tpsDig (Rohlf, 2001) we measured the length of the red patch and the standard length of the fish. This allowed us to calculate the relative length of the red area on the fish by calculating the ratio of red length to total length. We measured 16 males from Klein Lake, 15 males from Trout Lake, 35 males from Cranby Lake, 30 limnetics and 39 benthics from Paxton Lake, and 25 limnetics and 28 benthics from Priest Lake.  60  Reflectance  Because measurement of colour by eye is partly subjective, we also measured the reflectance spectra of male colours to obtain a more objective comparison (Endler, 1990). Although reflectance measurements do not provide information about the area or extent of a particular colour, they do provide objective estimates of the differences in brightness (total reflectance), saturation (intensity) and hue (colour) (Endler, 1990). We used the following Ocean Optics (Dunedin, Florida) equipment: USB 2000 spectrometer, DT 1000 light source (200-1100 nm) and an R400-7-UV/VIS reflectance probe to measure the reflectance spectra of two spots (throat and belly) on each live fish. We chose these spots because the throat region is typically red and the belly is typically the bluest spot on the body even on fish with little overall blue area. Each spot was measured five times per fish to minimize measurement error. We recorded percent reflectance relative to an Ocean Optics WS-1 reflectance standard at 0.38 nm intervals from approximately 300-700 nm. The probe was inserted into a custom made black PVC probe holder to minimize the influence of ambient light on the measurements. The probe was then held against the side of the fish at a 90-degree angle, and a reflectance spectrum was stored using Ocean Optics OOIBase 32 software.   We used principal components analysis (PCA) to assess differences between populations in reflectance spectra. PCA is a useful method for analyzing differences in the brightness and shape of spectral data when details of the visual system of the receiver are unknown (Cuthill et al., 1999; Grill and Rush, 2000). The segment analysis method of Endler (1990), which assumes trichromatic vision, gave us very similar results to those  61 presented here for the PCA on reflectance. Because stickleback colours are most likely optimized for the stickleback visual system of four cone types (Rowe et al., 2004), we present the results of the PCA instead of the segment method, because PCA does not rely on any assumptions about the number of types of cones. Generally, the first principal component of a PCA on reflectance spectra is equivalent to total reflectance (brightness) and explains > 90% of the variance among spectra (Cuthill et al., 1999; Grill and Rush, 2000). The second and third components can be more difficult to interpret (Grill and Rush, 2000). However, examination of their loadings allows for an assessment of their relationships to the original spectra. The means of the five replicate reflectance spectra for each spot and fish were used in all analyses. We reduced the amount of data in these mean spectra by calculating the median reflectance at 20 nm intervals from 310-690 nm. This provided 20 variables (wavelengths) that were used in the PCA analysis. Figures 4.2 and 4.3 show the average reflectance spectra for fish from all seven populations.  RESULTS Reflectance Analysis PCA on throat reflectance The first principal component of the PCA on the throat reflectance spectra explained the most variance (90.3%). All of the wavelengths loaded positively for PC1 (Fig. 4.4A), suggesting that it explained differences in total reflectance or brightness among the spectra. The large amount of the variance explained by the first component, and its association with brightness, has been seen in other studies using PCA on reflectance spectra (e.g. Grill and Rush, 2000).  62   Figure 4.2. Average reflectance spectra of the throat region for each population. Each point represents the median reflectance over a 20 nm interval. A) Cranby, B) Klein, C) Trout, D) Paxton benthic, E) Paxton limnetic, F) Priest benthic, and G) Priest limnetic.    63   Figure 4.3. Average reflectance spectra of the belly region for each population. Each point represents the median reflectance over a 20 nm interval. A) Cranby, B) Klein, C) Trout, D) Paxton benthic, E) Paxton limnetic, F) Priest benthic, and G) Priest limnetic.    64 The second principal component (red score) explained 5.8% of the variance and was associated with differences between the relative reflectance at short vs. long wavelengths (Fig. 4.4A). Positive values of red score represent high reflectance at long wavelengths (> 600 nm) and a lower reflectance at short wavelengths, suggesting that it is associated with variation in red intensity among males. The higher the red score, the greater the saturation (intensity) of the red colour because there is less reflectance at lower wavelengths to wash it out (Endler, 1990).  The third principal component explained 2.4% of the variance (Fig. 4.4A). A positive score for PC3 represents higher reflectance at middle wavelengths relative to short and long wavelengths. PC3 therefore represents variation in reflectance at green wavelengths. We used only the red score (PC2) in the regression analysis because the other two components were not associated with variation in red.  PCA on belly reflectance  The first principal component of the belly reflectance spectra explained 93.1% of the variance and was associated with strong positive loadings across all wavelengths (Fig. 4.4B). Therefore, it represents differences among the reflectance spectra in total reflectance or brightness, similar to the first principal component of the throat reflectance analysis.  The second principal component (blue score) explained 4.0% of the variance between reflectance spectra (Fig. 4.4B). A positive score for blue score indicates high reflectance at low to medium wavelengths and lower reflectance at long wavelengths,  65 consistent with a blue-green colour. The more positive the blue score, the more saturated (intense) the blue colour because reflectance at longer red wavelengths is reduced.  The third principal component explained 2.2% of the variance (Fig. 4.4B). A positive score for PC3 is related to greater reflectance at mid to long wavelengths. This component is difficult to interpret, but it probably reflects variation among males in the background colour of their skin beneath the blue. As with the throat reflectance, we used only the second component (blue score) in further analyses due to a lack of association between the other two components and blue colour.  Character Displacement Analysis The best test of greater differences in sympatry than in allopatry is a random effects model (e.g. Schluter and McPhail, 1992; Vamosi and Schluter, 2004). Therefore, we used a nested linear mixed-effects model to test for differences between the three types (limnetic, benthic and allopatric) in their means for the colour variables. The linear mixed effects model was constructed with type (L, B, or A) as the fixed main effect, with population (a random effect) nested within type. Character displacement was inferred if all three types had different means, or if two types had the same mean but the limnetics and benthics differed (e.g. (L = A) ≠ B, or L ≠ (B = A)). All statistical analyses were carried out in R (R Development Core Team 2004).    66  Figure 4.4. Component loadings for the first three principal components of the PCA on throat (A) and belly (B) reflectance. The solid lines are the second components (red score and blue score), the dashed lines are the first components and the dotted lines are the third components.    67 Red Red score differed significantly among types in the nested linear mixed effects model (F2, 4 = 36.79, p = 0.0027) (Fig. 4.5A, Table 4.1). To determine if the benthic and allopatric types differed, we assessed whether a model in which the benthic and allopatric types were considered as one group (F1, 5 = 31.17, p = 0.0025) fit the data better than the full model where all three types were considered separately. The full model fit the data significantly better than the reduced model (log likelihood ratio test (LLRT) = 4.45, p = 0.034), suggesting that all three types differ.  In contrast, the relative length of red did not differ significantly among types (F2, 4 = 5.09, p = 0.08: Fig. 4.5B, Table 4.1), although there is a slight trend for the allopatric populations to have less overall red area.  Blue  Blue score (F2, 4 = 4.68, p = 0.090) did not differ significantly among types (Fig. 4.5C, Table 4.1). Thus, in contrast to red there is no evidence of character displacement in blue nuptial colouration. Interestingly, blue score differed greatly between benthics and limnetics in Priest Lake (Welch’s t-test, t37 = -5.30, p < 0.001) but not Paxton Lake (Welch’s t-test, t47 = -1.20, p = 0.235). This suggests that the dynamics of male colour evolution may differ between species pairs.      68 Table 4.1. Means (± 1 SE) of the colour measurements.  Male Type  Limnetic Benthic Allopatric Red score 3.09 ± 0.41 -2.12 ± 0.38 -0.51 ± 0.40 Relative red length 0.33 ± 0.01 0.28 ± 0.01 0.23 ± 0.01 Blue score 5.84 ± 2.07 -5.74 ± 1.30 1.27 ± 1.37   DISCUSSION The results from both the reflectance measurements and the visual assessment of the males suggest that there is character displacement of male nuptial colour between sympatric species of threespine sticklebacks. Limnetics have more intense (saturated) red throats than the allopatric populations, whereas benthics lie below the allopatric populations. The parallel displacement of red intensity suggests that similar mechanisms are causing divergence in red between the two lakes. Interestingly, the relative amount of red does not differ significantly among types, suggesting that the mechanisms causing changes in red intensity and red area are different. For the blue colouration, there is evidence for divergence in one lake (Priest), but not the other, and there is weak evidence of character displacement. This suggests that the mechanisms causing divergence in blue are not parallel between independently derived species pairs.  69  Figure 4.5. Population means for the colour measurements. Paxton limnetics are represented by open circles, Priest limnetics by black circles, Paxton benthics by open squares, Priest benthics by black squares, Klein by open triangles, Cranby by black triangles, and Trout by grey triangles. A) red score, B) relative red length, C) blue score.  What interactions could drive character displacement of red in these lakes? Benthic and limnetic sticklebacks show character displacement in foraging traits driven by competition for shared resources (Schluter and McPhail, 1992; Schluter, 2000a). This ecological character displacement might influence displacement of red in several ways. First, adaptation of benthics and limnetics to different foraging environments causes a reduction in the fitness of hybrids relative to parental types (Hatfield and Schluter, 1999; Rundle, 2002). Reduced hybrid fitness could result in reinforcement and selection to  70 strengthen premating isolation. There is evidence that both male and female sticklebacks are more discriminating when sympatric with the other species than are individuals from allopatric populations (Rundle and Schluter, 1998; Albert and Schluter, 2004), a predicted outcome of reinforcement.  However, for reinforcement to lead to displacement in red between the species, the preferences of benthic and limnetic females for red would have to differ, with benthic females preferring duller red males and limnetic females preferring brighter red males. A difference in preference for red could be mediated by differences in the sensitivity of females to red wavelengths, and there is some evidence that benthic females are less sensitive to red (Boughman 2001). This observation, however, relies on measurements of females from Enos Lake where male benthics display black nuptial colouration. In lakes where males display red (all lakes in this study) females differ only slightly in red sensitivity (Boughman 2001). Furthermore, both benthic and limnetic females prefer the brightest red conspecific males, although the preferences of benthic females are slightly weaker (Boughman et al. 2005). It seems therefore unlikely that reinforcement has caused divergence in female preference for red, leading to divergence of red in males. However, this remains to be tested through a comparison of sympatric female preferences for red with the preferences of allopatric females. The second way by which ecological character displacement might influence displacement in red is through differences in diet. Differences in diet may expose benthics and limnetics to different amounts and types of carotenoids. Differences in red within a European population of sticklebacks were correlated to differences in the relative amounts of various carotenoids in the pigment cells (Wedekind et al., 1998).  71 Unfortunately, we have no data concerning the amounts and types of carotenoids available to benthics and limnetics, or if they differ in availability between the species pairs. An intermediate diet would also explain why allopatric populations are intermediate in red (Schluter and McPhail, 1992). Note, however, that differences in carotenoids are unlikely to explain differences in blue colouration in Priest Lake, as blue is a structural rather than a pigment-based colour (Rowe et al., 2004). Third, character displacement in red may be a consequence of displacement in nesting habitat use. Benthic males tend to nest in more covered and vegetated areas than limnetic males, which tend to nest in open or sparsely vegetated areas (McPhail, 1994; Hatfield and Schluter, 1996; Vamosi and Schluter, 1999). The amount of light available under cover may be much less than that available in the open, leading to a reduction in the visibility of the red signal in the benthic nesting habitat. If females are less able to see red due to poor light conditions, then the expression of the male display may be reduced relative to an open water signaling environment. A link between differences in colour between closely related species or populations and differences in the quality of light available in different signaling environments has been shown in birds (e.g., Endler and Théry, 1996; McNaught and Owens, 2002), anolis lizards (e.g., Macedonia, 2001), killifish (Fuller, 2002), and other stickleback populations (Reimchen, 1989; Boughman, 2001). There is some evidence that the colour of the water becomes redder with depth in Priest and Paxton Lakes, which would reduce the visibility of a red signal and could result in reduced expression of red in benthics if they nest at greater depths, or under cover. Further investigation is required to  72 determine the strength of any correlation between depth, habitat and the location of nests in benthic, limnetic, and allopatric populations. The difference in nest site microhabitat may be a result of male-male aggressive interactions. Benthic males may be better competitors for concealed nest locations, driving limnetic males into the open. In tests using other populations of sticklebacks, male size was strongly correlated with success at establishing territories (Rowland, 1989). Because benthic males are larger, they may force limnetic males into more open nesting locations. The body size difference between the species pairs appears to be a result of ecological character displacement in foraging ecology (Schluter and McPhail, 1992), providing another link between displacement in red and competition for shared resources. In summary, we found evidence for character displacement of red colouration between species pairs of threespine sticklebacks. The exact cause of displacement is as yet unknown, but it seems unlikely that reinforcement has played a dominant role. This implicates other types of interspecific interactions in signal displacement, such as competition for shared resources and territorial aggression. In addition, the role of differences in blue colouration between the species pairs in Priest Lake warrants further investigation. Ultimately, we would like to know if divergence in mating signals is generally driven by divergence in female preferences, or by environmental differences and other interactions. Understanding how these forces work together to shape phenotypes will provide much insight into how interspecific interactions can drive divergence between species.   73 REFERENCES Alatolo, R. V., L. Gustafsson, and A. Lundberg. 1994. Male coloration and species recognition in sympatric flycatchers. Proc. Roy. Soc. Lond. B. 256:113-188. Albert, A. Y. K., and D. Schluter. 2004. Reproductive character displacement of male stickleback mate preference: reinforcement or direct selection? Evolution 58:1099-1107. Baube, C. L. 1997. Manipulations of signaling environment affect male competitive success in three-spined sticklebacks. Anim. Behav. 53:819-833. Boughman, J. W. 2001. Divergent sexual selection enhances reproductive isolation in sticklebacks. Nature 411:944-948. — — —. 2002. How sensory drive can promote speciation. Trends in Ecol. Evol. 17:571-577. Boughman, J. W., H. D. Rundle, and D. Schluter. 2005. Parallel evolution of sexual isolation in sticklebacks. Evolution 59:361-373. Brown, W. L. Jr., and E. O. Wilson. 1956. Character displacement. Systematic Zoology 5:49-64. Chek, A. A., J. P. Bogart,  and S. C. Lougheed. 2003. Mating signal partitioning in multi-species assemblages: a null model test using frogs. Ecol. Lett. 6:235-247. Coyne, J. A., and H. A. Orr. 1989. Patterns of speciation in Drosophila. Evolution 43:362-381. Cuthill, I. C., A. T. D. Bennett, J. C. Partridge, and E. J. Maier. 1999. Plumage reflectance and the objective assessment of avian sexual dichromatism. Am. Nat. 153:183-200. Dobzhansky, T. 1940. Speciation as a stage in evolutionary divergence. Am. Nat. 74: 312-321. Endler, J. A. 1990. On the measurement and classification of colour in studies of animal colour patterns. Biol. J. Linn. Soc. 41:315-352. Endler, J. A., and M. Théry. 1996. Interacting effects of lek placement, display behavior, ambient light, and color patterns in three neotropical forest-dwelling birds. Am. Nat. 148:421-452. Fuller, R.C. 2002. Lighting environment predicts the relative abundance of male colour morphs in bluefin killifish (Lucania goddei) populations. Proc. Roy. Soc. Lond. B 269:1457-1465.  74 Grill, C.P., and V.N. Rush. 2000. Analysing spectral data: comparison and application of two techniques. Biol. J. Linn. Soc. 69:121-138. Hatfield, T., and D. Schluter. 1999. Ecological speciation in sticklebacks: environment-dependent hybrid fitness. Evolution 53:866-873. Höbel, G., and H. C. Gerhardt. 2003. Reproductive character displacement in the acoustic communication system of green tree frogs (Hyla cinerea). Evolution 57:894-904. Kraak, S. B. M., B. Mundwiler, and P. J. Hart. 2001. Increased number of hybrids between benthic and limnetic three-spined sticklebacks in Enos Lake, Canada; the collapse of a species pair? J. Fish Biol. 58:1458-1464.  Macedonia, J. M. 2001. Habitat light, colour variation, and ultraviolet reflectance in the Grand Cayman anole, Anolis conspersus. Biol. J. Linn. Soc. 73:299-320. McDonald, C. G., and C. W. Hawryshyn. 1995. Intraspecific variation of spectral sensitivity in threespine stickleback (Gasterosteus aculeatus) from different photic regimes. J. Comp. Physiol. A 176:255-260. McDonald, C. G., T. E. Reimchen, and C. W. Hawryshyn. 1995. Nuptial colour loss and signal masking in Gasterosteus: an analysis using video imaging. Behaviour 132:963-977. McLennan, D. A., and J. D. McPhail. 1990. Experimental investigations of the evolutionary significance of sexually dimorphic nuptial colouration in Gasterosteus aculeatus (L.): the relationship between male colour and female behaviour. Can. J. Zool. 68:482–492. McNaught, M. K., and I. P. F. Owens. 2002. Interspecific variation in plumage colour among birds: species recognition or light environment? J. Evol. Biol. 15:505-514. McPhail, J. D. 1984. Ecology and evolution of sympatric sticklebacks (Gasterosteus): morphological and genetic evidence for a species pair in Enos Lake, British Columbia. Can. J. Zool. 62:1402-1408. — — —. 1992. Ecology and evolution of sympatric sticklebacks (Gasterosteus): evidence for a species-pair in Paxton Lake, Texada Island, British Columbia. Can. J. Zool. 70:361-369. — — —. 1993. Ecology and evolution of sympatric sticklebacks (Gasterosteus): origin of the species pairs. Can. J. Zool. 71: 515-523. — — —. 1994. Speciation and the evolution of reproductive isolation in the sticklebacks (Gasterosteus) of south-western British Columbia. In: Bell MA, Foster SA, eds. The evolutionary biology of the threespine stickleback. Oxford: Oxford University Press, 188-206.  75 Milinski, M., and T. C. M. Bakker. 1990. Female sticklebacks use male coloration in mate choice and hence avoid parasitized males. Nature 344:330-333. Nagel, L., and D. Schluter. 1998. Body size, natural selection and speciation in sticklebacks. Evolution 52: 209-218. Noor, M. A. F. 1999. Reinforcement and other consequences of sympatry. Heredity 83:503-508. Novales Flamarique, I, A. Hendry, and C. W. Hawryshyn. 1992. The photic environment of a salmonid nursery lake. J. Exp. Biol. 169:121-141. Podos, J. 2001. Correlated evolution of morphology and vocal signal structure in Darwin’s finches. Nature 409: 185-188. R Development Core Team. 2004. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-00-3, URL: http://www.R-project.org. Reimchen, T. E. 1989. Loss of nuptial color in threespine sticklebacks (Gasterosteus aculeatus). Evolution 43:450-460. Rohlf, F. J. 2001. tpsDig. Ver. 1.31. Distributed by the author. Department of Ecology and Evolution, State University of New York, Stony Brook, NY. Rowe, M. P., C. L. Baube, E. R. Loew, and J. B. Phillips. 2004. Optimal mechanisms for finding and selecting mates: how threespine stickleback (Gasterosteus aculeatus) should encode male throat colors. J. Comp. Physiol. A. 190:241-256. Rowland, W. J. 1989. The effects of body size, aggression and nuptial coloration on competition for territories in male threespine sticklebacks, Gasterosteus aculeatus. Anim Behav. 37:282-289. — — —. 1994. Proximate determinants of stickleback behaviour: an evolutionary perspective. In: Bell MA, Foster SA, eds. The evolutionary biology of the threespine stickleback. Oxford: Oxford University Press, 297-344. Rundle, H. D. 2002. A test of ecologically dependent postmating isolation between sympatric sticklebacks. Evolution 56:322-329. Rundle, H. D, and D. Schluter. 1998. Reinforcement of stickleback mate preferences: sympatry breeds contempt. Evolution 52:200–208. Rush, V. N., J. S. McKinnon, M. A. Abney, and R. C. Sargent. 2003. Reflectance spectra from free-swimming sticklebacks (Gasterosteus): social context and eye-jaw contrast. Behaviour 140:1003-1019.  76 Schluter, D. 2000a. Ecological character displacement in adaptive radiation. Am. Nat. 156:S4-S16. — — —. 2000b. The ecology of adaptive radiation. Oxford: Oxford University Press. Schluter, D., T. D. Price, and P. R. Grant. 1985. Ecological character displacement in Darwin’s finches. Science 227:1056-1059. Schluter, D., and J. D. McPhail. 1992. Ecological character displacement and speciation in sticklebacks. Am. Nat. 140:85-108. Servedio, M. R. 2001. Beyond reinforcement: the evolution of premating isolation by direct selection on preferences and postmating, prezygotic incompatibilities. Evolution 55:1909-1920. Servedio, M. R., and M. A. F. Noor. 2003. The role of reinforcement in speciation: theory and data. Ann. Rev. Ecol. Evol. Syst. 34:339-364. Taylor, E. B., and J. D. McPhail. 2000. Historical contingency and ecological determinism interact to prime speciation in sticklebacks, Gasterosteus. Proc. R. Soc. Lond. B 267:2375-2384. Taylor, E. B., J. W. Boughman, M. Groenenboom, M. Sniatynski, D. Schluter, J. L. Gow. 2006. Speciation in reverse: morphological and genetic evidence of the collapse of a three-spined stickleback (Gasterosteus aculeatus) species pair. Mol. Ecol. 15: 343-355. Tynkkynen, K., M. J. Rantala, J. Suhonen. 2004. Interspecific aggression and character displacement in the damselfly Calopteryx splendens. J. Evol. Biol. 17:759-767. Vamosi, S. M., and D. Schluter. 2004. Character shifts in defensive armor of sympatric sticklebacks. Evolution 58:376-385. Wedekind, C., P. Meyer, M. Frischknecht, U. A. Niggli, and H. Pfander. 1998. Different carotenoids and potential information content of red coloration of male three-spined stickleback. J. Chem. Ecol. 24:787-801. Whoriskey FG, FitzGerald GJ. 1994. Ecology of the threespine stickleback on the breeding grounds. In: Bell MA, Foster SA, eds. The evolutionary biology of the threespine stickleback. Oxford: Oxford University Press, 188-206.  77 CHAPTER 5: Sexual Selection Can Resolve Sex-linked Sexual Antagonism4  INTRODUCTION Evolutionary biologists have long puzzled over how and why female preferences drive the evolution of exaggerated male traits. Generally, female preferences are thought to enhance a female’s long-term fitness by increasing her offspring’s fitness, either directly or through genetic associations between preference and trait loci (Andersson 1994; Kokko et al. 2003). Nearly all models assume that females do not initially express the male display trait or else assume that the fitness effects are the same in males and females (Kirkpatrick 1982). However, traits subject to sexual selection will often be sexually antagonistic with, for example, “sexy” male traits benefiting males but reducing female fitness (Rice 1984; Rice and Chippindale 2001; Seger and Trivers 1986). If a trait increases male reproductive success at the cost of female viability, females must then choose between having attractive sons (and unfit daughters) and having ugly sons (but unencumbered daughters). As long as females can detect the genotypic differences among males at sexually antagonistic loci, we expect mating preferences to evolve as described by the models explored in this paper. Theory predicts that sexually antagonistic loci are more likely to remain polymorphic on the sex chromosomes (Rice 1984; Rice and Chippindale 2001). Furthermore, recent empirical work suggests that many sexually selected traits in animals                                                 4 A version of this chapter has been published. Albert, A.Y.K., and S.P. Otto. (2005) Sexual selection can resolve sex-linked sexual antagonism. Science 310:119-121. http://dx.doi.org/10.1126/science.1115328  78 are located on the X chromosome (Reinhold 1998; Lindholm and Breden 2002) and that most polymorphic sexually antagonistic traits are located on the X chromosome in Drosophila (Rice and Chippindale 2001; Gibson et al. 2002). There is also evidence to suggest that female mate choice may result in a tradeoff in the fitness between her daughters and sons (Fedorka and Mousseau 2004). Several recent theoretical examinations of the evolution of female preferences have explored sex-linkage of the trait and/or preference (Hastings 1994; Reeve and Pfennig 2003; Andrés and Morrow 2003; Kirkpatrick and Hall 2004). However, these models, with the exception of Reeve and Pfennig, assume that sexually selected traits have male-limited expression and therefore no fitness consequences when in females, and none have addressed sexual antagonism. Here, we address the question of how female preferences evolve for traits that have contrasting fitness effects in each sex. With sexual antagonism, chromosomal location should strongly affect the evolution of female preferences. Simply put, an X-linked male trait is never passed on from an attractive father to his sons, whereas his daughters suffer the cost of carrying the display trait (Rice and Chippindale 2001; Gibson et al. 2002). Offspring in XY species therefore do not gain a fitness benefit from females preferring males with a more extreme display trait. In contrast, both males and females contribute a Z chromosome to sons in ZW species. Thus, females preferring a Z-linked display trait receive the fitness benefit of sexy sons, even though their daughters suffer a fitness cost (Rice and Chippindale 2001; Gibson et al. 2002). This cost is lessened by the fact that daughters inherit only one of their father’s Z chromosomes. With autosomal inheritance, these asymmetries in inheritance are absent.  79 METHODS AND RESULTS To verify the verbal argument laid out above, we present the results of two-locus models that follow the fate of a newly arisen preference allele p in a population that is at a polymorphic equilibrium at a trait locus. We assume that allele T is most fit in females (“female-benefit allele”) and allele t is most fit in males (“male-benefit allele”), with this tradeoff mediated by natural selection, sexual selection, or both. In the X-linked model females with a tt genotype suffer a fitness disadvantage of 1−sx relative to TT females, and the fitness of Tt females is given by 1−hsx. In the absence of sexual selection, males carrying the T allele have a relative fitness of 1−sy relative to males with the t allele (Table 5.1). Note that sy can be positive or negative allowing for natural selection either for or against the trait allele, t, in males. The mating preference of a female of genotype i at the preference locus is given by ai, which describes the relative increase (or decrease) in the female’s probability of mating with males carrying allele t (Table 5.1, see Kirkpatrick 1982). We also allow for selection against the female preference by reducing female fitness by an amount proportional to her choosiness and a cost parameter, k (see Appendix 1 for details). The fitness scheme in the Z-linked model is similar (Table 5.1). However, an additional parameter, d, is required to describe female preferences for heterozygous males when the preference locus is Z-linked. When both trait and preference loci are sex-linked, recombination occurs at a rate r between them; autosomal preference loci were also considered to examine the influence of physical linkage on the results. Finally, both autosomal preference and trait loci were modeled. The analytical solutions for the polymorphic equilibria and the invasion rate of a new preference allele, p, are approximated assuming weak selection (sx, sy, ai, and k small).  80 Table 5.1. Male and female fitness components in (a) male heterogametic (XY) and (b) female heterogametic (ZW) species.   X-linked trait Z-linked trait   Male Trait  Male Trait   T t  TT Tt tt PP 1 1+aPP P 1 1+daP 1+aP Pp 1 1+aPp Female preference: pp 1 1+app p 1 1+dap 1+ap Male viability:  1-sy 1  1-sz 1-hsz 1   Female Trait  Female trait   TT Tt tt  T t Female viability:  1 ! 1" hsx  ! 1" sx   ! 1 ! 1" sw    Female Preference  Female Preference   PP Pp pp  P p Female cost:  ! 1" a pp k  ! 1" aPp k  ! 1" a pp k   ! 1" aP k  ! 1" a p k   There are three ultimate fates for sexually antagonistic genes: i) the fixation of the allele with the higher fitness across both sexes, ii) the evolution of sex-specific expression, or iii) polymorphism (Rice 1984; Rice and Chippindale 2001). A polymorphic equilibrium for an X-linked trait is maintained by sexually antagonistic selection in the X-linked model with allele P fixed as long as   81 2hsx < (aPP + sy) < 2sx(1-h),              (1) which requires that the male-benefit allele, t, be wholly or partially recessive in females (h < 1/2). As long as (1) holds, the model allows for the maintenance of polymorphism either by sexually antagonistic natural selection (sy > 0) and/or by sexual selection opposing natural selection (app > 0). Similar criteria for maintaining a polymorphic equilibrium at a Z-linked trait locus are presented in Appendix 1. Performing a stability analysis, a new preference allele, p, invades a population that is polymorphic for an X-linked trait locus whenever it confers a stronger preference for males bearing the female-benefit allele, T (aPp < aPP). This result holds regardless of the physical location of the preference locus. While recombination breaks apart genetic associations, it also places preference alleles on the same chromosome as the trait alleles that have been preferred and is therefore critical to the development of genetic associations. These two factors balance, causing the level of genetic associations to be fairly insensitive to the recombination rate.  In contrast to the X-linked case, when both the trait and preference loci are Z-linked allele p can invade a population only when it confers a stronger preference for the male-benefit allele, t (ap > aP). Again, the recombination rate does not alter the range of conditions under which invasion occurs. However, when the trait locus is Z-linked and the preference locus autosomal, p invades only when it increases the preference for the female-benefit allele, T (aPp < aPP), as in the X-linked case. Adding a cost of female preference to both the X- and Z-linked models just slows the spread of stronger preference alleles, as long as the costs are not too strong relative to selection acting on the trait locus (Appendix 1).  82 Finally, if all of the loci are autosomal, there is no longer any selection for females to prefer traits that help only one sex, and the invasion of any particular preference allele depends crucially on the level of linkage (Appendix 1). This is consistent with a previous model (Otto 1991), which found that sexual selection could resolve a polymorphism of a male-limited trait either in favour of T or t. To understand the selective forces at work, we performed a Quasi Linkage Equilibrium (QLE) analysis (Kirkpatrick et al. 2002). When females carrying a new allele p preferentially mate with males carrying T, for example, a positive genetic association develops between the p and T alleles. In the homogametic sex (either XX females or ZZ males), both cis and trans linkage disequilibrium are present and approximately equal in magnitude, but only cis disequilibrium can be present in the heterogametic sex (either XY males or ZW females when the preference is Z-linked). Consequently, the new preference allele rises by association with T in XX females (in whom T is favoured) by twice the amount that it goes down in XY males (in whom t is favoured), explaining the spread of female preferences for the trait favoured in females among XY species. Conversely, when both the trait and preference are Z-linked, the new preference allele declines by association with T in ZZ males (in whom t is favoured) by twice the amount that it goes up in ZW females (in whom T is favoured), explaining why sexual selection does not favour preferences for the allele beneficial to daughters.  When the trait is Z-linked but the preference locus is autosomal, the situation is slightly more complex. Both sons and daughters now inherit two copies of the preference allele, but daughters inherit only one copy of the trait allele. Because daughters’ trait alleles are only inherited from their father, stronger trans disequilibrium develops  83 between the preference allele inherited from their mothers and the trait allele inherited from their fathers. The stronger genetic associations that develop in daughters than in sons again favour the spread of preferences for the female-benefit allele, T. The conditions for the invasion of a new preference allele were determined assuming weak selection. How robust are these results to stronger selection? Simulations with selection coefficients on the order of 10-30% were explored; selection coefficients in this range are not uncommon (Kingsolver et al. 2001). Invasion depends only on how the new preference allele changes female mating preferences and not on the strength of selection, confirming our analytical results (Figs. 5.1 and 5.2)   Figure 5.1. Simulation results for the evolution of female preferences in male heterogametic species (XY). All simulations were started under conditions allowing a polymorphic equilibrium at an X-linked trait locus with P fixed. A) A new allele p that prefers males carrying the female-benefit allele, T, is introduced and sweeps to fixation leaving T polymorphic at a higher frequency and improving the fitness of daughters. (aPP = 0.1, aPp = 0, app = -0.1). B) A preference allele favouring the T allele sweeps to high frequency, driving T to fixation. Sexual antagonism is thus completely resolved in favour of females. (aPP = 0.1, aPp = -0.1, app = -0.2). Both loci are X-linked with r = 0.5, h = 0.1, sx = 0.2, and sy = 0.2. The frequency of T is in thin curves, and males are shown with dashed curves. The frequency of p is in bold, with indistinguishable curves for males and females.  84  Figure 5.2. Simulation results for the evolution of female preferences in female heterogametic species (ZW). A) A new allele p that prefers males carrying the male-benefit allele, t, sweeps to fixation, driving t to high frequency and improving the fitness of sons (aP = -0.1, ap = 0.08).  B) A preference allele p favouring the t allele sweeps to high frequency, driving t to fixation (aP = -0.1, ap = 0.2). Sexual antagonism is thus completely resolved in favour of males. Both loci are Z-linked with r = 0.5, d = 0.5, h = 0.2, sw = 0.224, and sz = 0.3.  The curve types are the same as in Figure 5.1.  DISCUSSION Our results point to a potentially large effect of the sex determination mechanism on how female preferences evolve for sexually antagonistic traits. Over long timescales, evolutionary changes in female preferences will lead to the fixation of the trait alleles most fit in females in XY systems and ZW systems when the preference is autosomal, but the trait allele most fit in males in ZW systems with Z-linked preferences. Thus, sexually antagonistic selection is always resolved in favour of females in XY species (Fig. 5.1B), but in favour of males in ZW species when the preference is Z-linked (Fig. 5.2B). This process can occur very rapidly if the new allele has a strong effect on preferences.  Assuming the conditions of our model, we predict that the difference in fitness between daughters and sons should be greater for females that choose mates relative to  85 females mating at random. Choosy females should produce daughters that are more fit than sons in XY systems and vice versa in ZW systems. Another prediction of our model is that female preferences can evolve more easily for male-benefit alleles in ZW species, which is consistent with the greatly exaggerated displays observed in groups such as birds and butterflies (Reeve and Pfennig 2003). This prediction calls for phylogenetic analyses of the association between flashy displays and sex determination. Finally, our model predicts that sex-linked polymorphisms maintained by sexually antagonistic selection should disappear faster when sexual selection is present and as long as the females can detect the polymorphism. Once female preferences have resolved sexually antagonistic selection and become established within a species, they could cause the further evolution of flashy male displays at loci throughout the genome (Hastings 1994).  REFERENCES Andersson, M. 1994. Sexual Selection. Princeton Univ. Press, Princeton. Andrés, J. A., and E. H. Morrow. 2003. The origin of interlocus sexual conflict: is sex-linkage important? J. Evol. Biol. 16:219-223.  Fedorka, K. M, and T. A. Mousseau. 2004. Female mating bias results in sex-specific offspring fitness. Nature 429:65-67. Gibson, J. R., A. K. Chippindale, and W. R. Rice. 2002. The X chromosome is a hot spot for sexually antagonistic fitness variation. Proc. R. Soc. Lond. B. 269:499-505. Hastings, I. M. 1994. Manifestations of sexual selection may depend on the genetic basis of sex determination. Proc. R. Soc. Lond. B. 258:83-87.  Kingsolver, J. G., et al. 2001. The strength of phenotypic selection in natural populations. Am. Nat. 157:245-261. Kirkpatrick, M. 1982. Sexual selection and the evolution of female choice. Evolution 36:1-12.  86 Kirkpatrick, M., and D. W. Hall. 2004. Sexual selection and sex-linkage. Evolution 58:683-691.  Kirkpatrick, M., T. Johnson, and N. Barton. 2002. General models of multilocus evolution. Genetics 161:1727-1750.  Kokko, H., R. Brooks, M. D. Jennions, and J. Morley, 2003. The evolution of mate choice and mating biases. Proc. R. Soc. Lond. B. 270:653-664.  Lindholm, A., and F. Breden 2002. Sex chromosomes and sexual selection in Poeciliid Fishes. Am. Nat. 160:S214-S22. Otto, S. P. 1991. On evolution under sexual and viability selection: a two-locus diploid model. Evolution 45:1443-1457.  Reeve, H. K., and D. W. Pfennig. 2003. Genetic biases for showy males: are some genetic systems especially conducive to sexual selection? Proc. Natl. Acad. Sci. USA 100:1089-1094. Reinhold, K. 1998. Sex linkage among genes controlling sexually selected traits. Behav. Ecol. Sociobiol. 44:1-7.  Rice, W. R. 1984. Sex chromosomes and the evolution of sexual dimorphism. Evolution 38:735-742. Rice, W. R., and A. K. Chippindale. 2001. Intersexual ontogenetic conflict. J. Evol. Biol. 14:685-693.  Seger, J., and R. Trivers. 1986. Asymmetry in the evolution of female mating preferences. Nature 319:771-773.  87 CHAPTER 6: Conclusions  In Chapter 2, I investigated the potential role of direct selection on limnetic male mating preferences. I compared the mate preferences and courtship behavior of males from one sympatric limnetic population and two allopatric populations, using the limnetic-like allopatric populations to control for the effects of ecological character displacement and adaptation to different niches on mate preferences. The sympatric limnetic males preferred the small limnetic females, whereas the allopatric limnetic-like males preferred the large benthic females, suggesting that adaptation to the limnetic niche does not automatically confer a preference for small limnetic females. This reproductive character displacement of male preference is consistent with the predictions of both reinforcement and direct selection on mate preferences. To test for direct selection, I assessed a prediction of one proposed mechanism: predation by benthic females on eggs guarded by limnetic males. The allopatric males come from populations where there is no egg predation. Sympatric limnetic males were more aggressive towards benthic females than towards limnetic females, whereas the allopatric limnetic-like males did not treat the two types of females differently. The contrast in male behavior suggests that egg predation has shaped male preferences. Direct selection is potentially more effective than indirect selection via reinforcement, and it is likely that it has been important in building up reproductive isolation between limnetic and benthic sticklebacks. In Chapter 3, I tested for an effect of imprinting on the formation of female mating preferences, a potentially important one-allele mechanism of reproductive isolation. One-allele mechanisms require the fixation of the same allele in two diverging  88 populations thereby reducing the chance that hybridization and recombination will break apart associations between alleles for adaptation to different environments and alleles that produce different mating preferences. I tested the possibility that sexual imprinting promotes reproductive isolation using one sympatric species pair of threespine sticklebacks. I fostered families of F1 hybrids between the species to males of both species. Preferences of these fostered females for males of either type revealed little or no effect of sexual imprinting on assortative mating. However, F1 females showed preferences for males that were similar to themselves in length, suggesting that size assortative mating may be more important than sexual imprinting for promoting reproductive isolation between species pairs of threespine sticklebacks. In Chapter 4, I presented the results of a survey looking for character displacement in male nuptial colouration in threespine sticklebacks. Character displacement in signaling traits occurs when differences between species are greater in sympatry than where either species occurs alone. Finding character displacement in a signaling trait suggests that the trait has diverged as a result of interspecific interactions such as competition, aggression, predation, or reproductive interference. Breeding males of both the limnetic and benthic species develop red throats and blue bodies, although limnetic males appear brighter. To test for character displacement I compared the nuptial colour of benthics and limnetics from two species-pair lakes (sympatric) with that of males from three similar allopatric lakes (only one species present). I measured the intensity of blue and red colouration using reflectance spectra taken from live fish. I found that allopatric males were intermediate between limnetic and benthic males in the intensity of red colour, indicating character displacement in that trait in sympatry. In  89 contrast, I found no evidence for character displacement in blue intensity, although it differed sharply between the species pairs in one lake (Priest). The results point to interactions in sympatry causing displacement in male colour, but the mechanism remains unknown. Because of a lack of divergence in female preference for red, it seems unlikely that reinforcement is the cause. Other interspecific interactions such as competition and territorial aggression seem to be more plausible mechanisms for divergence of male colour in sympatry.  In Chapter 5, I used population genetic modeling to investigate the effects of sexual conflict and sex-determination on the evolution of female mating preferences. Most previous models of the evolution of female preferences assume that the display trait is sex-limited in expression. This seems unlikely especially in the early stages of preference and trait evolution. Sexually antagonistic traits, where alternative forms are favoured in each sex, seem to congregate on the sex chromosomes. Therefore, I modeled the evolution of female mating preferences with sex-linked sexually antagonistic traits. The results of the model depend crucially on the mechanism of sex determination. When the display trait is on the X chromosome in XY species, females evolve preferences for the trait allele that increases the fitness of daughters over sons. Alternatively, when the display trait is on the Z chromosome in WZ species, females can evolve preferences for the allele that increases the fitness of sons over daughters. This suggests that WZ species are more likely to evolve female preferences for exaggerated male traits than XY species. This prediction needs to be tested in a phylogenetic framework. In summary, I have investigated the evolution of reproductive isolation and mating preferences from many different angles. The important theme is that adaptation to  90 different environments may cause complex changes in the behaviour and appearance of animals that can contribute to reproductive isolation if two such populations meet after some time in allopatry. In addition, further interactions between the species, stemming either directly or indirectly from their divergence, can lead to the evolution of further differences in appearance and behaviour thus strengthening the level of reproductive isolation between them. Ultimately, we want to know whether the former or the latter phases are more important in the process of speciation. In the case of the sticklebacks, it seems likely that reproductive isolation was relatively strong on secondary contact. However, the relative importance of these phases will depend crucially on the level of divergence between species at contact, the genetic basis of the adaptive differences between them, and the type of mating system and strength of mating preferences. These are all questions currently being addressed in the stickleback system.    91 APPENDIX 1: Sexual Selection and Sexual Antagonism Supplementary Material THE RECURSIONS  The model is designed to examine the conditions under which a new female preference allele can invade a population polymorphic at a trait locus subject to sexually antagonistic selection. At the trait locus, allele T is favoured in females and allele t is favoured in males. The life cycle proceeds as follows. We census among the male and female zygotes, after which natural (viability) selection occurs, followed by sexual selection, recombination and fusion of gametes to form the starting genotype frequencies of the next generation. All diploid genotypes are tracked separately because there is nonrandom mating. Each individual has a two-locus genotype with a trait locus affecting viability (T/t) and a preference locus affecting mating probabilities (P/p).  In the following, we describe the development of the recursions assuming that both the trait and preference loci are X-linked and that the two loci recombine at rate r. Among the zygotes in the population, the genotype frequencies are given in Table A1.1. When the preference locus is autosomal, the number of possible male genotypes increases to six (y1 = TP/P, y2 = tP/P, etc.), and when both loci are autosomal the number of male and female genotypes are equal at ten. In female heterogametic species, a Z-linked trait locus was instead modeled using similar notation to Table A1.1 but with the x’s and y’s reversed.     92   Table A1.1. Male and female genotypes for the X-linked model. Female genotypes Male genotypes x1 = TP/TP y1 = TP x2 = TP/tP y2 = tP x3 = tP/tP y3 = Tp x4 = TP/Tp y4 = tp x5 = TP/tp  x6 = tP/Tp  x7 = tP/tp  x8 = Tp/Tp  x9 = Tp/tp  x10 = tp/tp    Females display mating biases depending on their genotype at the P locus. PP females prefer to mate with t males by an amount aPP relative to T males, Pp females prefer t males by an amount aPp, and pp females prefer t males by an amount app (Table 5.1). This follows the fixed relative preference scheme laid out by Kirkpatrick 1982.  In the following, we illustrate the process by which the recursions were developed. Following viability selection, the frequency of the TP/TP genotype in females becomes ! x1s= x1/V f , where   93 ! V f = x1 + (1" hsx )x2 + (1" sx )x3 + x4 + (1" hsx )x5 + (1" hsx )x6 + (1" sx )x7 + x8+ (1" hsx )x9 + (1" sx )x10      (1a) represents the mean viability in females and ensures that the frequencies of each genotype among females sums to one. Similarly, the frequency of the TP/tP genotype in females becomes ! x2s = 1" hsx( )x2 /V f . Among males, the frequency of the TP genotype after viability selection becomes ! y1s= (1" sy )y1 /V m , and the frequency of the tP genotype becomes ! y2s= y2/V m  where  ! V m = (1" sy )y1 + y2 + (1" sy )y3 + y4             (1b) Among the surviving adults, females choose mates according to the preference scheme in Table 1, after which point gamete production (including recombination) and union of gametes occurs. For example, the probability that a female of genotype TP/TP mates with a male of genotype tP equals ! x1s (1+ aPP )y2sa PP, which accounts for the fact that females of genotype PP prefer t males by an amount aPP. The frequency of a mating pair involving a female of genotype i is divided by the average strength of the mating preferences of that female, ! a i : ! a PP =1+ (y2s+ y4s) aPP  ! a Pp =1+ (y2s+ y4s) aPp  ! a pp =1+ (y2s+ y4s) app                 (2) This assumes that all females are able to mate and that mate choice does not reduce female fitness. The results of relaxing this assumption are discussed later. The frequency of any particular zygote in the next generation is then calculated over all possible mating pairs that can produce that zygote. Female zygotes of the  94 genotype TP/TP are produced by the following matings: TP/TP! "TP, TP/tP! "TP, TP/Tp! "TP, TP/tp! "TP, tP/Tp! "TP. Accounting for the probability that each of these matings produces a TP/TP daughter, we then have the genotype frequency of TP/TP after one generation (including viability and sexual selection), ! " x 1, given by: ! " x 1 =x1sy1sa PP+x2sy1s2a PP+x4sy1s2a Pp+(x5s(1# r) + x6sr) y1s2a Pp.            (3) Similarly, female zygotes of genotype TP/tP are produced by the following matings: TP/TP! "tP, TP/tP! "TP, TP/tP! "tP, tP/tP! "TP, TP/Tp! "tP, TP/tp! "tP, tP/Tp! "tP, tP/Tp! "TP, TP/tp! "TP, tP/tp! "TP. We then have the genotype frequency of TP/tP after one generation (including natural and sexual selection), ! " x 2, given by: ! " x 2 =x1sy2s(1+ aPP )a PP+x2sy1s2a PP+x2sy2s(1+ aPP )2a PP+x3sy1sa PP+x4sy2s(1+ aPp )2a Pp+(x5s(1# r) + x6sr) y2s(1+ aPp )2a Pp+(x6s(1# r) + x5sr) y1s2a Pp+x7sy1s2a Pp          (4) The male recursions are put together similarly with the genotype frequency of ! y1 (TP) after one generation of viability and sexual selection (including recombination), ! " y 1, given by: ! " y 1 = x1s+1/2x2s+1/2x4s+1/2(x5s(1# r) + x6sr) .             (5) Because males inherit their X chromosomes solely from their mother, the genotype frequencies of the adult males do not appear in the final recursions describing the male zygote frequencies in the next generation. The genotype frequency of ! y2 (tP) after one generation, ! " y 2 is given by: ! " y 2 =1/2x2s+ x3s+1/2(x6s(1# r) + x5sr) +1/2x7s.             (6)  95 The recursions for the remaining eight female genotypes and two male genotypes listed in Table A1.1 are derived similarly and are available upon request.  SPECIES WITH MALE HETEROGAMETY (XY) Polymorphic equilibrium at an X-linked trait locus We investigated the conditions required to maintain a polymorphic equilibrium for T when allele P is fixed. We assumed that selection is weak and that s, z, and a are all small terms. To leading order in the selection coefficients, the equilibrium genotype frequencies are:  ! male frequency of T =2sx (1" h) " (aPP + sy )2(1" 2h)sxfemale frequency of TT =(aPP " 2sx + 2hsx + sy )24(1" 2h)2 sx2female frequency of Tt =2sx (1" h) " (aPP + sy )(aPP " 2hsx + sy )2(1" 2h)2sx2         (7) For these equilibrium frequencies to lie between zero and one and for the equilibrium to be stable before the introduction of p requires that condition (1) in the text is met.   Evolution of mating preferences with an X-linked trait locus We then investigated the conditions under which a new preference allele p could invade a population at the polymorphic equilibrium (7). First, we consider the case of an X-linked trait locus and an X-linked preference locus. Assuming weak selection, the leading eigenvalue of the stability matrix obtained from the recursions under the assumption that p is rare is: ! " =1+1/6(aPP # aPp )Freq(T)Freq(t)(aPP + sy ),            (8)  96 where Freq(T) Freq(t) are the equilibrium frequencies of T and t, respectively (to leading order in the selection coefficients, these are the same in males and females). Because (aPP + ! sy ) must be positive according to condition (1) in the text, the eigenvalue will be greater than one if aPp < aPP, and less than one if aPp > aPP. This means that the polymorphic equilibrium will be unstable if aPp < aPP, allowing p to invade when it confers a stronger preference for the trait allele favoured in females, T. The above assumes that selection is weak but that the recombination rate is not small. When r = 0 (complete linkage between trait and preference), there are two contenders for the leading eigenvalue. One describes the rate of spread of the Tp haplotype: ! " =1+1/3(aPP # aPp )Freq(T)Freq(t)hsx ,           (9a) and one describes the rate of spread of the tp haplotype: ! " =1+1/3(aPP # aPp )Freq(T)Freq(t)(1# h)sx .          (9b) Equations (9a) and (9b) are both greater than one if aPp < aPP and are both less than one if aPp > aPP , indicating that invasion occurs under the same conditions as (8). Note, however, that the rate of invasion is proportional to the maximum of 1/3 hsx and 1/3 (1 – h)sx, the largest of which will always be greater than or equal to 1/6 (aPP + sy) according to condition (1) in the Chapter 5 for the maintenance of a polymorphism. Thus whether the leading eigenvalue is greater or less than one does not change with r, although the maximum eigenvalue is always greater with r = 0 than with r >> 0, implying that the new preference alleles spread faster when tightly linked.  When the preference allele is autosomal, the leading eigenvalue becomes: ! " =1+1/10(aPP # aPp )Freq(T)Freq(t)(aPP + sy ).          (10)  97 This is identical to (8) except for the constant factor of 1/10 in place of 1/6, suggesting that the invasion of p is slower with an autosomal preference locus but occurs under the same conditions. That invasion should be slower when the preference locus is autosomal is expected because the preference alleles spend less time in females and more time in males, where they do not act. In summary, in XY species with a sexually antagonistic trait that is X-linked, female preferences evolve to favour males bearing the female-benefit allele, T. This condition holds whether the preference and trait loci are linked or unlinked and regardless of the chromosome on which the preference locus resides.  Adding a cost of female preference We next investigated the impact of costs to sexual selection where females that are choosier suffer a fitness cost relative to females with no preference. Following Pomiankowski 1987, we model costs by assuming that preferences decrease fitness by an amount ! ai times k, where ! ai measures choosiness (ai = 0 corresponds to random mating) and k measures the fitness costs. After selection, the frequency of the TP/TP genotype in females becomes ! x1s= x1(1" aPPk) /V k, where  ! V k = (1" aPP k)x1 + (1" hsx )(1" aPP k)x2 + (1" sx )(1" aPP k)x3 + (1" aPp k)x4+ (1" hsx )(1" aPp k)x5 + (1" hsx )(1" aPp k)x6 + (1" sx )(1" aPp k)x7+ (1" app k)x8 + (1" hsx )(1" app k)x9 + (1" sx )(1" app k)x10      (11) represents the average female fitness including viability selection and the costs of choosiness. The absolute value of the preference is used in these equations to allow for  98 selection against any preference allele, either one that confers a preference for T (aPp < aPP) or for t (aPp > aPP). The male recursions remain unchanged.  With costs added to the model with X-linked trait and preference loci, equation (8) becomes: ! " =1+1/6(aPP # aPp )Freq(T)Freq(t)(aPP + sy ) + 2 /3( aPP # aPp )k .        (12) Overall, costs drive the population towards aPP = aPp = 0, but this selective pressure can be overwhelmed if ! Freq(T)Freq(t)(aPP + sy )  is large enough relative to k.   SPECIES WITH FEMALE HETEROGAMETY (ZW) Polymorphic equilibrium at a Z-linked trait locus To leading order in the selection coefficients, the genotypic frequencies at the polymorphic equilibrium for a Z-linked trait locus are: ! female frequency of T =2aPd + sw " 2aP " 2hsz2(2aPd + sz " aP " 2hsz )male frequency of TT =(2aPd + sw " 2aP " 2hsz )24(2aPd + sz " aP " 2hsz )2male frequency of Tt =(2aPd " 2(h "1)sz " sw )(2aP (d "1) " 2hsz + sw )2(aP (2d "1) + sz " 2hsz )2       (13) For the equilibrium frequencies to lie between zero and one and for the equilibrium to be stable before the introduction of p requires that: ! 2aPd + sw " 2aP " 2hsz > 02aPd + sz " aP " 2hsz > 02aPd + 2sz " 2hsz " sw > 0.             (14)  99 Conditions (14) can be simplified by combining the fitness effects of natural and sexual selection on male genotypes, and standardizing by the frequency of the most fit male genotype tt:      TT Tt tt Sexual selection: 1 ! 1+ daP ! 1+ aP  Natural selection: ! 1" sz ! 1" hsz 1 Product: ! 1" sz ! (1+ daP)(1" hsz) ! 1+ aP Fitness relative to tt: (! 1" sz)/(! 1+ aP) ! (1+ daP)(1" hsz) /(1+ aP) 1  The overall fitness of TT and Tt in males can then be defined as ! 1" Sz and ! 1"HSz, respectively, where! Sz= aP+ sz and! H = (hsz+ aP" aPd) /(aP+ sz)  to leading order in the selection coefficients. Using these equations, conditions (14) can be rewritten as: 2HSz < sw < 2Sz (1-H).       (15) Only when condition (15) is met will there be a stable polymorphic equilibrium for the trait allele. In particular, the deleterious allele in males, T, must be wholly or partially recessive (H < 1/2) for 2HSz to be less than 2Sz(1-H).   Evolution of mating preferences with a Z-linked trait locus  100 We then investigated the conditions under which a new preference allele p can invade a population at the polymorphic equilibrium (11). First, we consider the case of a Z-linked trait locus and a Z-linked preference locus. Assuming weak selection, the leading eigenvalue of the stability matrix obtained from the recursions under the assumption that p is rare is: ! " =1#(aP # ap )Freq(T)Freq(t)[Freq(t)(1# d) + Freq(T)(d)]sw6.        (16) Therefore, this eigenvalue is less than one when aP > ap, and greater than one when aP < ap. This result indicates that the new preference allele p could invade only when aP < ap, such that it confers a stronger preference for males carrying the trait allele t.  The above assumes that selection is weak but that the recombination rate is not small. When r = 0 (complete linkage between trait and preference), there are again two contenders for the leading eigenvalue. One describes the rate of spread of the tp haplotype: ! " =1#(aP # ap )Freq(T)Freq(t)[Freq(t)(1# d) + Freq(T)(d)](aP # aPd + hsz)3,    (17a) and one describes the spread of the Tp haplotype: ! " =1#(aP # ap )Freq(T)Freq(t)[Freq(t)(1# d) + Freq(T)(d)](aPd + sz # hsz)3.    (17b) Under the equilibrium condition (15), equations (17a) and (17b) are both greater than one if aP < ap and are both less than one if aP > ap, indicating that invasion occurs under the same conditions as (14). As in the X-linked model, r does not change the conditions of invasion but does alter the rate. Specifically, the maximum of (17) is greater than (16) as  101 long as ! min aP 1" d( ) + hsz , aPd + sz 1" h( )( ) <sw2. As this condition can be rewritten as ! min 2HSz ,2Sz (1"H)( ) < sw , it is guaranteed by (15). When the preference allele is autosomal, the leading eigenvalue becomes: ! " =1+(aPP # aPp )Freq(T)Freq(t)[Freq(t)(1# d) + Freq(T)(d)]sw10,         (18) which is identical to (16) except for the constant is now 1/10 instead of 1/6 and the sign after the one is reversed. Now, this eigenvalue is greater than one when aPP > aPp, indicating that invasion occurs when p confers a stronger preference for males carrying the female-benefit allele, T.   Adding a cost of female preference The result of adding a cost of female preference in the Z-linked model is qualitatively very similar to the X-linked model. The only difference is that the constant multiplying k is now 1/3 instead of 2/3. This makes sense as 1/3 of the Z chromosomes are carried by females. In summary, in ZW species with a sexually antagonistic trait locus and preference locus that are Z-linked, female preferences evolve to favour males bearing the trait allele, t, which increases fitness in sons. In contrast, when the preference locus is autosomal, the results are the same as those for the XY model, and female preferences evolve to favour males carrying the trait allele, T, which increases fitness in daughters.   102 AUTOSOMAL TRAIT AND PREFERENCE LOCI We constructed a model with autosomal trait and preference loci to compare the results to a previous model by Otto (1991). In the Otto 1991 model, the evolution of mating preferences was considered near a polymorphism maintained by overdominant selection in males only. We extended this model to consider overdominance induced by selection for allele T in females balanced by selection for allele t in males (Table A1.2). As overdominance is present in both models, we predicted that the results would be similar.  Table A1.2. Male and female fitness components in autosomal model.    Male Trait   TT Tt tt PP 1 1+daPP 1+aPP Pp 1 1+daPp 1+aPp Female preference: pp 1 1+dapp 1+app Male viability:  1-sm 1-hm sm 1   Female Trait   TT Tt tt Female viability:  1 1-hf sf 1-sf   103 As in the Otto 1991 model, we found that the fate of a new preference allele depends critically on the rate of recombination. When r is small, any new preference allele can invade, allowing for the resolution of sexual antagonism in favor of either sex depending on the initial conditions. Specifically, when r = 0, there are two candidates for the leading eigenvalue to leading order in the selection coefficients. One describes the rate of spread of the tp haplotype: ! " =1+ (aPP # aPp )DFreq(T)2Freq(t)$W1 + $W22% & ' ( ) * ,       (19a)  and one describes the rate of spread of the Tp haplotype: ! " =1# (aPP # aPp )DFreq(T)Freq(t)2 $W1 + $W22% & ' ( ) *  .         (19b) In these equations, ! D = Freq(T)d +Freq(t) 1" d( )  measures the average degree of dominance at the trait locus with respect to sexual selection and, for the sake of discussion, is assumed positive. ! "W1 is the difference in fitness between Tt and TT individuals, averaged across sexes. Similarly, ! "W2  is the difference in fitness between Tt and tt individuals, averaged across sexes. For there to be a stable polymorphic equilibrium, both ! "W1 and ! "W2  must be positive. Thus, when the new preference allele favors the male-benefit allele (! aPp > aPP ), (19a) is greater than one, and the new preference allele invades. Conversely, when the new preference allele favors the female-benefit allele (! aPp < aPP ), (19b) is greater than one, and the preference allele still invades. The same results were observed and explained qualitatively by Otto 1991. Essentially, the haplotype Tp can spread when rare if females carrying this haplotype prefer males bearing the t allele, as such parents are more likely to produce heterozygous offspring, whose fitness is, on average, higher.  104 Interestingly, Otto 1991 found that the above scenario reverses when linkage is loose. In the absence of selection in females (sf = 0), no new preference alleles can invade when recombination rates are high. Qualitatively, this occurs because recombination breaks apart the positive association between the new preference allele and the trait allele that ensures offspring are more likely to be heterozygous, eliminating the advantage of having a different mating preference. With selection in females, the leading eigenvalue is more complicated, however:  ! " =1# (aPP # aPp )2+ 2(aPP # aPp ) sfKD$ % & ' ( ) Freq(T)2Freq(t)2D2 *W1 + *W24+ , - . / 0 ,       (20) which is written to leading order in the selection coefficients. Here, ! K = Freq(T)hf +Freq(t) 1" hf( )  measures the average degree of dominance at the trait locus with respect to natural selection in females and is assumed positive. When sf is near zero, we recover the results of Otto 1991, and no new preference allele can invade. When the trait and preference loci are loosely linked, a new preference allele can spread only if the term in braces is negative. This requires that the new preference allele favor the male-benefit allele, t (aPp > aPP). It also requires that selection in females be strong enough and that the difference in preferences be small enough that the second term within the braces dominates the first term. This second term reflects the fact that natural selection in females causes adult females to carry the T allele slightly more often than expected. Consequently, females that prefer t-bearing males are more likely to have Tt heterozygous offspring, driving the spread of a preference for t. Regardless, the strength of this force is very weak, as can be seen by the fact that ! " #1 in (20) is of third order in the selection coefficients (sf, sm, and a) compared to second order as found under tight linkage (19) or with sex-linked modifiers.  105  The critical value of the recombination rate below which any new preference allele can invade is: ! rc ="W1 "W2"W1 + "W211+2sf K(aPP # aPp )D,            (21) to leading order in the selection coefficients. Equation (21) is of the order of the selection coefficients, indicating that new preference alleles can invade across a larger portion of the genome when selection is stronger. In summary, visible autosomal polymorphisms maintained by sexual antagonism will also be resolved over the long term, because tightly linked preferences can invade that tip the balance in favor of one or the other allele. This process is less directed than sex-linked loci, as preferences in either direction can invade only when linkage is tight. With autosomal loci, there is a slight tendency for sexual antagonism to be resolved in favor of the males (i.e., preferences evolve to favor the t allele) at loosely linked preference loci. However, the strength of this selection is weak.   REFERENCES Kirkpatrick, M. 1982. Sexual selection and the evolution of female choice. Evolution 36:1-12.  Otto, S. P. 1991. On evolution under sexual and viability selection: a two-locus diploid model. Evolution 45:1443-1457.  Pomiankowski, A. 1987. The costs of choice in sexual selection. J. Theor. Biol. 128:195-218.    106 APPENDIX 2: Animal Care Certificates   107   

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0092835/manifest

Comment

Related Items