The Importance of Alternative Mating Types to Speciation

Introduction Sexual selection is a powerful diversifying force leading to divergence within and among species. Sexual selection arises from competition for mates, usually from either male-male competition or female choice (Darwin 1859, 1871). Competition for mates can lead to a diversity of forms in different populations, extreme and exaggerated phenotypes within a species, and may accelerate rates of speciation (Darwin 1871; West-Eberhard 1983, Masta & Maddison 2002). Sexual selection could be a powerful mechanism for speciation given its role in shaping premating isolating mechanisms such as courtship behavior and species-specific signals (West-Eberhard 1983, Panhuis et al., 2001). A powerful way to study how sexual selection drives divergence is to study alternative mating types. Alternative mating types are distinct phenotypes found within a population that have been shaped by sexual selection to compete for mates. Such mating types are quite common in nature (West-Eberhard, 2003) and illustrate that sexual selection can lead to diversification even within populations (Sinervo & Svensson, 2002). Alternative mating types are likely to promote speciation because they maintain large amounts of adaptive genetic variation within a population, their phenotypes are often as distinct as separate species, and evolutionary change in them would lead to divergence in the mating system. Indeed, speciation is often associated with the fixation of a single alternative strategy (West-Eberhard, 1986). Despite the suggested importance of alternative mating types for speciation, a direct test as to how alternative mating types promote phenotypic divergence and reproductive isolation has been lacking. In this study I propose to: 1. Create an intraspecific phylogeny in order to test for rapid phenotypic evolution associated with the divergence of mating types across populations. 2. Conduct female choice experiments and mating trials to test whether divergence in mating types between populations can lead to reproductive isolation. Background Alternative Mating Types and Diversification

Alternative mating strategies are increasingly recognized as being common within populations (reviews in West-Eberhard 1986, 2003). For example, multiple mating types have been found in crustaceans (3 male isopod morphs, Shuster and Wade, 1991), fish (Taborsky, 1994), insects (2 male morphs, Emlen, 1997), lizards (Sinervo & Lively, 1996; Thompson & Moore, 1993), birds (Lank et al., 1995), and plants (2-3 heterostyly morphs, Barret, 1992). Such mating types often have markedly distinct behaviors, sizes, colorations, and body forms. Darwin was probably one of the first researchers to identify alternative mating types as being important for speciation. When Darwin was challenged by Thomas Huxley to show how reproductive isolation can build up within a species, he began studying heterostyly in a species of Primula. In reference to this work Darwin (1862) remarked, �My notions on hybridity are becoming considerably altered by my dimorphic work: I am now strongly inclined to believe that sterility is at first a selected quality to keep incipient species distinct.� Thus Darwin linked the buildup of mating types within a population to the possible evolution of reproductive isolation and speciation. Modern researchers studying alternative adaptations within populations have suggested that polymorphic types within a population could form the basis for new species (review in West-Eberhard, 1986; Sinervo & Svensson, 2002). West-Eberhard (1986) outlined a number of patterns that suggest that alternative adaptations are involved in speciation, including complex specialization within populations, rapid evolution following the fixation of one alternative, and recurrent speciation associated with an intraspecific alternative. Thus competition between alternative types within a population allows the

adaptation of complex, specialized phenotypes, and the fixation of such specialized types promotes rapid speciation. Alternative mating types (adaptive alternatives for sexual selection) are especially likely to be involved in speciation. Sexual selection can lead to rapid divergence and quickly generate reproductive isolation due to direct effects on traits for mate recognition (West-Eberhard 1983, Panhuis et al., 2001) and by forming coadapted gene complexes (Sinervo & Svensson, 2002; Sinervo & Clobert, 2003) thought to be involved in speciation (Reiseberg, et al. 1996). Although mating types are selected for maintaining fertility within a population, competition between mating types may drive rapid evolution and divergence between populations. Given that the divergence is in the mating system, reproductive isolation may quickly evolve. My proposed research tests this scenario in the side-blotched lizard, Uta stansburiana.

The Color Morphs of Uta stansburiana

(Use NSF FastLane to view this and following figures in color)

Study System One of the most comprehensive microevolutionary studies of an alternative mating system is on

the side-blotched lizard, Uta stansburiana. Within a population at Los Baños, CA three genetically determined male mating strategies engage in male-male competition. Males with orange throats are aggressive, have high plasma testosterone, and control large territories with many females (Sinervo et al, 2000a). Blue-throated males are less aggressive, closely guard females, and cooperatively defend territory (Sinervo & Clobert 2003). Yellow-throated males mimic female behavior and sneak on to other males� territories to copulate with females (Zamudio & Sinervo 2000). Detailed pedigrees show that the heritability of throat color is close to one (Sinervo et al. 2000b, Sinervo & Svensson 2002) and color is likely to be determined by a single gene with three alleles (Sinervo et al. 2000b). The different morphs are maintained by negative frequency-dependent selection with orange males being the most fit when blue is common, yellow being the most fit when orange is common, and blue being the most fit when yellow is common. This is the first biological example of a cyclical rock-paper-scissors game (Sinervo & Lively 1996). Within the same population two different female morphs exhibit different life history strategies. Orange-throated females (r strategists) are favored at low densities: they produce large numbers of small progeny. Yellow-throated females (K strategists) are favored at high density: they produce fewer, but higher quality offspring. Population density of adult females oscillates with a two-year period which leads to the maintenance of both female morphs (Sinervo et al. 2000b). Female morphs exhibit an extraordinary degree of female choice. Female side-blotched lizards mate with multiple males and are able to take sperm from large males to produce high fitness sons and sperm from small males to produce high fitness daughters (Calsbeek & Sinervo 2002). Female choice is not limited to such cryptic choice. Female U. stansburiana will assortatively mate based on the male�s throat color, with orange females preferring orange males and yellow females preferring yellow males for their first clutch (Bleay et al, in prep.).

Phylogeographic Patterns of U. stansburiana�s Morphs

My dissertation research focuses on determining how U. stansburiana�s morphs have evolved and diversified over time. My research approach is to study intraspecific variation in the morphs and to use phylogenetics to reconstruct the evolutionary history of different populations. I have completed a survey of the throat color polymorphism across the range of U. stansburiana, which revealed significant variation in number of morphs across populations (Figure 1a & b, Corl unpub. data). Uta stansburiana elegans is polymorphic, but the polymorphism has been reduced or lost in at least 4 isolated locations: Big Sur, Anacapa Island, Pisgah Lava Flow, and Espiratu Santo. The subspecies U. s. stejnegeri is also polymorphic, but U. s. uniformis has fixed for blue in males and females (except at Wupatki, where orange persists). Finally, yellow allele has been lost in U. s. stansburiana and the orange allele is at high frequency. Substantial divergence in morphs has occurred on the islands off of Baja California, Mexico (Figure 1b). Many island endemics (U. encantadae, U. tumidarostra, U. lowei, U. squamata) are fixed for a subset of the three throat colors seen in U. s. elegans (data from Grismer, 2002). Novel colors have developed on some islands, with U. nolascensis having green throats, U. palmeri having grey throats, and U. stansburiana on Isla San Roque having red throats (Grismer, 2002). Not all islands are fixed for one color. A survey of four islands off of Baja revealed that three landbridge islands (San Jose, San Francisco, and La Partida) have the standard three-color polymorphism, while Uta on Espiratu Santo are fixed for blue. A preliminary phylogeny based on 609 base pairs of ATPase 6 reveals significant geographic structure and general agreement with previously proposed subspecies (Figure 2, Corl, unpublished data). Populations in New Mexico and Texas (U. s. stejnegeri) form a well supported clade (bootstrap=100%) that is the sister group to the remaining U. stansburiana populations. Northern populations (U. s. stansburiana) group together with populations in Nevada, Oregon, and Washington forming a well supported clade (bootstrap=99%). Populations of U. s. uniformis (Utah, Arizona, and Colorado) would be a good monophyletic group except that Dinosaur (a poorly supported branch) is placed near eastern California populations. U. s. elegans paraphyletic and includes three quite divergent lineages in central California, southern Californian (the Mojave Desert), and Baja Mexico (represented by San Jose island on the phylogeny).

Morphs and speciation U. stansburiana satisfies West-Eberhard�s (1986) criteria for alternative types promoting divergence and speciation. 1. Each of U. stansburiana�s morphs is a highly specialized adaptive phenotype. There are extensive correlations between throat color and many other traits including hormones and physiology (Sinervo et al. 2000a, Comendant et al. 2003), immune function (Svensson et al. 2001a,b, 2002), life history (Sinervo et al. 2000b), and behavior (Sinervo & Lively 1996, Sinervo & Zamudio 2002, Sinervo & Clobert 2003). 2. There is rapid evolution following the loss of one or more of the morphs. Although only recently divergent from a polymorphic ancestor (Figure 2), lizards on Anacapa island (~99% blue, 1% orange) differ in body size, life span, growth rate, and clutch size (Comendant & Sinervo, unpublished data). Also, a comparison among populations reveals a significant decrease in sexual dimorphism as the frequency of the blue allele increases to fixation (linear regression weighted by sample size, P=.028, r2=.19). Thus polymorphic populations tend to be more sexually dimorphic than monomorphic populations. This is likely to be due to intense male-male competition among morphs driving body size evolution, and the relaxation of this selection when morphs are lost. For example, while Los Baños males are 1.6 times heavier than females, a closely related (Figure 2) monomorphic population in Big Sur has no sexual dimorphism. 3. There has been recurrent speciation associated with the loss of the throat color polymorphism. The majority of insular species of Uta in Baja, Mexico have fixed for a subset of the throat colors observed in U. stansburiana (Grismer 2002). There is strong evidence to suggest that adjoining populations differing in their morphs are reproductively isolated. Lizards on the Pisgah Lava Flow are melanic and have lost the yellow morph despite being surrounded by populations of polymorphic lizards. There is also evidence for reproductive isolation in a contact zone between two clades of Uta (Fig. 1, contact zone described by McKinney, 1971; Ballinger & Tinkle 1972). Male lizards in Zion National Park belong to the northern subspecies (U.s. stansburiana) and are monomorphic, and have vibrant blue throats. Males found at Lytle Ranch (~40 miles southeast) belong to a separate clade (Figure 2), and are polymorphic with orange, blue,

and yellow throats. However, blue morphs at Lytle Ranch have throats that are faintly blue, with most of the throat white with a very pale blue cast. No other population has been observed with a similar pale blue coloration. This difference in throat color is suggestive of reproductive character displacement, with the loss of blue coloration being a mechanism to prevent interbreeding with the northern, dark-blue throated lizards. Gene flow poses a serious problem for the divergence of adjoining populations. This is especially true for the morphs of U. stansburiana. Given that the ESS state at Los Baños is 3 morphs (Sinervo & Lively, 1996) and that male morphs have a significant fitness advantage when rare, the invasion of a missing morph should be both common and rapid. Therefore the multiple independent losses of morphs across populations poses an interesting dilemma: How are populations missing morphs reproductively isolated, given that they are easily invasible by missing morphs? Determining the mechanisms and the speed at which reproductive isolation builds up between these populations is the goal of my proposed research. Proposed Research

Phenotypic Evolution and the Loss of Morphs Hybrid unfitness due to divergent evolution could explain why missing morphs cannot successfully invade. West-Eberhard (1986) hypothesized accelerated genetic and phenotypic evolution following the loss of one alternative phenotype from a population due to a release from the constraints of producing multiple specializations. Also, the coadapted gene complexes that form the lizard morphs (Sinervo & Clobert, 2003) could break down if the negative frequency dependent dynamics that maintain the gene complexes cease due to the loss of one morph.

Hypothesis 1: The loss of morphs will promote rapid phenotypic evolution. Test 1: Test for rapid phenotypic evolution in populations that have lost morphs. This can be tested by plotting genetic divergence vs. phenotypic divergence for pairs of sister taxa. For example, Los Baños and Big Sur have 22 base pair substitutions between their ATPase haplotypes, and they differ in sexual dimorphism by 1.6-1.0=.6. If the loss of morphs promotes rapid divergence then sister taxa that differ in morph numbers should show large amounts of phenotypic change per unit genetic change. In comparison, sisters with the same number of morphs should show relatively little phenotypic divergence per unit genetic change. Genetic change must proceed in a relatively uniform rate for this test to work and this will be verified by testing for a molecular clock using the likelihood ratio test of Huelsenbeck and Bull (1996). Test 2: Test for parallel evolution of sexual dimorphism and clutch size with loss of morphs. The phenotypes of morphs are shaped by constant intermorph competition. The loss of one or more morphs will decrease this competition and free the phenotype to evolve independent of intraspecific competition. I will test for parallel phenotypic evolution associated with the loss of morphs by mapping sexual dimorphism and clutch size on to my phylogeny. A phylogeny is necessary for these tests in order to control for historical factors. The amount of sexual dimorphism is associated with the presence/absence of morphs (above), but this test was uncorrected for phylogeny. I predict that sexual dimorphism will be lower in populations that have lost morphs. The two female morphs Los Baños differ in their clutch size, so loss of a color morph is also likely to affect clutch size. Published clutch size data exists for 7 of my sites (Tinkle et al., 1970) and Sinervo has data for 8 more of my sites. With the data that I collect in my breeding experiments (below) I will have clutch size data for 19 sites. Methods: I propose to sequence the mtDNA Atpase 6 gene (609 base pairs) for the rest of my populations. I will also sequence 555 base pairs of cytochrome b. I will sequence 3 individuals per population to control for intraspecific polymorphism. In total about 40 populations will be included in

the phylogeny. Phylogenies will be conducted using Parsimony, Likelihood, and Bayesian approaches using PAUP* 4.0 (Swofford 2002) and MrBayes 2.01 (Huelsenback & Ronquist 2001).. Ancestral character states will be estimated by using maximum likelihood methods of Schluter et al. (1997) and hierarchical Bayesian methods of Huelsenbeck & Bollback (2001); this latter method accommodates uncertainty in the estimated tree. To test for correlations among pairs of discrete traits, the methods of Pagel (1994) will be used. Correlations between continuous traits will be computed using the method of Huelsenbeck and Rannala (2003), a new Bayesian approach that accommodates uncertainty in the inferred tree.

Reproductive Isolation and the Evolution of Morphs

If divergence in morphs leads to speciation, then it is important to determine the mechanisms of reproductive isolation. At Los Baños female U. stansburiana will assortatively mate based on the male�s throat color (Bleay et al, in prep.), so female choice could lead to behavioral isolation (Test 1). There may be postzygotic isolation if the loss of a morph destabilizes coadapted gene complexes (Sinervo & Clobert, 2003; Test 2). There may also be postmating, prezygotic isolation through an ability to sort sperm as demonstrated by Calsbeek and Sinervo (2002). Sperm sorting could be a potent, but often overlooked, mechanism for assortative mating (Howard 1999, Test 3). Females may not be behaviorally isolated if many matings are by forced copulations and could show no postzygotic isolation in a no choice mating trial, but may still be reproductively isolated in the wild based on their ability to sort sperm. Hypothesis 2: Divergence in the morphs between populations will be associated with reproductive isolation. Methods: Lizards will be collected from the following pairs of neighboring populations that differ in their morphs: Big Sur (1 male morph) and Los Baños (3 male morphs), Zion Natl Park (1 morph) and Lytle Ranch (3 morphs), Pisgah Lava Flow (2 morphs) and Granite Mountains (3 morphs), Anacapa Island (2 morphs) and Stunt Ranch (3 morphs), Los Baños (3 morphs) and Granite Mountains (3 morphs). Thus there will be two comparisons between 1 morph vs 3 morph populations, two comparisons of 2 morph vs 3 morph populations, and one control that compares two populations that have the same number of morphs but are phylogenetically divergent (Figure 2). This control is important to establish that reproductive isolation between populations is due to the reduction in morphs and not just gradual divergence with time and isolation. Two pairs of populations will be sampled the first year, and the remaining three the following year. Thirty five females and twenty five males will be collected from each population. All females and males will undergo test 1. Then 20 females and 10 males will be used for test 2, and 15 females and 15 males will be used for test 3. Lizards will be collected in early spring (March-May depending on location) in order to capture prereproductive individuals. Reproductive status of females can be assessed by external palpation of the ovaries. Lizards will be housed in a greenhouse used for previous breeding experiments (Sinervo et al, 2001). Test 1: Test for behavioral isolation via female choice. Behavioral isolating mechanisms will be assessed using methods developed in the Sinervo lab to test for assortative mating between morphs (Bleay et al, in prep.). Female lizards will periodically have their ovaries externally palped to determine when they are beginning to yolk up their eggs (a sign of reproductive receptivity). A receptive female lizard and two males (one from her own population, one from the paired population) will be placed in an arena. Each male will be tethered and the two males will be separated by an opaque wall. The female will not be tethered and will be allowed to freely interact

with both males. When first placed in the arena, the female will be separated from the males by an opaque partition and the lizards will be given 1-2 days to acclimate. Then the partition will be removed and all interactions between lizards will be videotaped for one hour. Female preference will be assigned based on mean time spent with the each male, # of head bobs (head bobs are used for communication between lizards), frequency of head bobs, and whether the female assumes a mating posture. Whether a female performs a rejection display (a high frequency vibrating pushup display; see http://www.biology.ucsc.edu/~barrylab/ for an example) will also be noted. Ten females will be given a choice between two males from their own population. This will serve as a control and will assess female preference within a population. Twenty five other females will be tested for behavioral isolation between populations. Females from both the within and between population tests will be exposed to all possible male morphs (orange, blue, and yellow) that are found within the population. Test 2: Test for postzygotic isolation between populations differing in their morphs. This test is designed to measure within vs between population fertility. Following a choice trial, a female lizard will be housed for three days with one of the two males she interacted with. The ten females used for within population choice trials will be mated to a single male from their own population. Ten females used for the between population trials will be mated to a single male from outside their population. Each male will be mated to two females, one from within his population and one from the paired population. This controls for the effects of male genotype on within vs between population crosses. When females are judged to be gravid (by abdominal palpation) they will then be housed in ovipositoria (cages with moist sand that is ideal for oviposition). Once eggs are laid, they will be counted and weighed. The eggs will be incubated until hatching using husbandry methods developed in the Sinervo lab (over 15 years of experience). The number of hatchlings will be recorded, as well as hatchling size (snout-vent length), weight, and sex. The following variables will be available to test for postzygotic differences between populations: Clutch size, egg weight, mean time to hatching, egg viability, hatching success, hatchling weight, hatchling size, and hatchling sex. Test 3: Test for cryptic female choice (sperm sorting) leading to prezygotic isolation. Fifteen females and fifteen males from each population will be used to test for cryptic choice. Following a between-population choice trial, a female lizard will be housed for 3 days with one of the two males she interacted with. Then she will be housed with the second male for 3 days. Half of the females will be housed with a male from their own population first, the other females with a male from the paired population first. Each male will have the opportunity to mate with one intrapopulation female and one interpopulation female. This will control for male genotype in the experiments and will allow an assessment of individual male performance within and between populations. All fertility variables described in experiment 2 will be measured, which will allow comparisons between test 2 (no choice mating) and test 3 (choice).

I will test for cryptic choice by genotyping all sires and hatchlings to determine male parentage. The Sinervo lab has 9 microsatellites that are routinely used for assigning paternity. However, given the need to only distinguish between two possible fathers, paternity for the majority of hatchlings will be easily assigned using only three of the most variable microsatellites scored in a single multiplex group. Paternity assignment will allow me to test for sperm precedence and cryptic female choice as possible mechanisms for reproductive isolation. Evidence for cryptic choice would be sperm precedence of intrapopulation males.

Predictions: The null hypothesis is that divergence in mating types does not promote

reproductive isolation. This hypothesis will be accepted if within vs. between population comparisons show no differences in female preference or fertility. The null hypothesis will be rejected if evidence for reproductive isolation is found between populations differing in their morphs. If all such populations show reproductive isolation, a strong role for mating types driving speciation will be inferred. Such a result would be consistent with West-Eberhard�s (1986) hypothesis that rapid evolution follows the fixation of one alternative strategy. The rate at which reproductive isolation builds up will be judged relative to the control crosses between the Granite Mountains and Los Baños. These 3 morph x 3 morph crosses will provide a relative scale of how much reproductive isolation builds up due to gradual divergence over time. Different pairs of populations are likely to show varying degrees of reproductive isolation, and this will give clues about the factors that act in combination with divergence in mating types to give rise to reproductive isolation. If only sympatric populations (Pisgah & Granites, Zion & Lytle) show reproductive isolation, then a role for reinforcing selection in creating reproductive isolation may be inferred. If only allopatric populations (Los Baños & Big Creek, Anacapa Island & Stunt Ranch) show reproductive isolation, then some geographic isolation may also be necessary. If only more divergent populations (Los Baños & Big Creek, Zion & Lytle; divergence= ATPase sequence divergence) are isolated, then genetic divergence in combination with mating system divergence is needed for reproductive isolation. Intellectual Merits: Much work on speciation has focused on the geography of speciation (Mayr, 1963). While this work has been instrumental in shaping our view of the speciation process, other modes of speciation have been largely understudied. The evolution of alternative types within a population provides a potentially powerful mechanism for speciation across many geographic contexts, including sympatric speciation. A test of this model of evolution would further a recent trend in research favoring the importance of selection over geography in understanding divergence between species (Hatfield and Schluter, 1999; Schneider et al., 1999). While others have investigated the role of ecological selection in driving divergence, my focus on alternative mating types tests the role of sexual selection in generating and promoting reproductive isolation. My study has a number of unique features. First, I have documented divergence in alternative mating types across populations, and have shown that mating types have been lost independently multiple times. Second, I will test for cryptic female choice, which is a mechanism of assortative mating that has rarely been tested. Finally, I use both comparative and experimental methods to test how reproductive isolation may form when alternative mating types diverge. While evidence exists that mating type divergence may promote speciation, a direct test of the isolating mechanisms between populations divergent in mating types would represent a significant advance in our understanding of the process of speciation. Broader Significance: Working with undergraduates is integral to my research. I am currently training two undergraduates in molecular biology and phylogenetic techniques. Nicole Chaney (a woman minority) is researching whether paedomorphic, aquatic cave salamanders are a separate species from salamanders in the surrounding watershed that metamorphose into a terrestrial form. Frasier Haney is working on his senior thesis project to date the independent losses of morphs in U. stansburiana using a molecular clock calibrated by the geologic separation times of insular lizards off of Baja, Mexico. Both students intend to apply to graduate school. I have previously trained three other undergraduates in field methods and herpetology. At least three other undergraduates will gain experience in molecular biology (DNA extraction, PCR, microsatellite analysis and sequencing), phylogenetics, field methods, behavioral observation, and experimental design over the course of my proposed research.