UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Ecological mechanisms in species origins : divergent natural selection and the evolution of reproductive… Rundle, Howard Douglas 2001

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

Item Metadata

Download

Media
831-ubc_2001-715299.pdf [ 5.32MB ]
Metadata
JSON: 831-1.0090784.json
JSON-LD: 831-1.0090784-ld.json
RDF/XML (Pretty): 831-1.0090784-rdf.xml
RDF/JSON: 831-1.0090784-rdf.json
Turtle: 831-1.0090784-turtle.txt
N-Triples: 831-1.0090784-rdf-ntriples.txt
Original Record: 831-1.0090784-source.json
Full Text
831-1.0090784-fulltext.txt
Citation
831-1.0090784.ris

Full Text

E C O L O G I C A L M E C H A N I S M S I N S P E C I E S O R I G I N S : D I V E R G E N T N A T U R A L S E L E C T I O N A N D T H E E V O L U T I O N O F R E P R O D U C T I V E I S O L A T I O N B E T W E E N S Y M P A T R I C S T I C K L E B A C K S by H O W A R D D O U G L A S R U N D L E B . S c , University of Toronto, 1995 M . S c , University of British Columbia, 1997 A THESIS S U B M I T T E D IN P A R T I A L F U L F I L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F D O C T O R OF PHILOSOPHY in T H E F A C U L T Y OF G R A D U A T E STUDIES (Department of Zoology) We accept this thesis as conforming to the required standard T H E UNIVERSITY OF BRITISH C O L U M B I A July 2001 © Howard Douglas Rundle, 2001 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department The University of British Columbia Vancouver, Canada DE-6 (2/88) A B S T R A C T Under the ecological model of speciation, reproductive isolation arises ultimately as a result of divergent natural selection. Laboratory experiments have demonstrated the feasibility of this mechanism but evidence from nature is lacking. This thesis investigates the role of ecological mechanisms in the origin of morphologically and ecologically distinct sympatric species-pairs of threespine sticklebacks (Gasterosteus aculeatus complex) that inhabit a few post-glacial lakes in southwestern British Columbia. Ecological speciation predicts that the fitness of hybrids should depend on their ecological context. This can be tested using reciprocal transplant experiments, provided the contribution of any genetic incompatibilities is controlled. I develop a quantitative genetic model to determine whether this is possible. Analysis of the model reveals that it is, and the experimental design involves the use of both hybrid backcrosses. I employ this design in a reciprocal transplant experiment in nature using the stickleback species-pairs. Results reveal a striking pattern of ecological-dependence: in each habitat the backcross more similar to the native parent species grew at approximately twice the rate of the other. Thus ecological mechanisms have been central to the evolution of postmating isolation between these stickleback species. Ecological speciation also predicts that reproductive isolation among populations should be correlated with environmental differences. To test this prediction I conducted laboratory mating trials using the stickleback species-pairs from three lakes. Results support the prediction, demonstrating that populations that evolved under different ecological conditions show strong premating isolation, whereas populations that independently evolved under similar ecological conditions lack isolation. Such parallel evolution of premating isolation, in correlation with environmental differences, strongly implicates divergent selection in the evolution of these species-pairs. Finally, I address the mechanisms responsible for divergent natural selection in the evolution of these species-pairs, focussing on the role of ecological interactions between populations. I use a pond experiment to test the combined effect of predation and competition i i on the strength of divergent selection.on a target population. Results suggest a strong interaction: the competition treatment generated divergent selection on the target population only under high predation conditions. This suggests that the ecological mechanisms responsible for the evolution of the species-pairs include both competition and predation. / iii T A B L E O F C O N T E N T S Abstract ii Table of Contents iv List of Tables vi List of Figures vii Acknowledgements viii Preface x Chapter 1: Ecological Mechanisms in Species Origins: A n Introduction 1 A n Overview of Ecological Speciation 2 Limnetic and Benthic Threespine Sticklebacks 10 Thesis Overview 13 Chapter 2: A Genetic Interpretation of Ecologically Dependent Isolation 23 Introduction 24 Methods 25 Results 27 Discussion 29 Chapter 3: A Test of Ecologically Dependent Isolation Between Sympatric Sticklebacks . . 35 Introduction 36 Methods 39 Results 42 Discussion 43 Chapter 4: Natural Selection and Parallel Speciation in Sympatric Sticklebacks 53 Introduction 54 Methods • • • 56 Results and Discussion 57 Chapter 5: Predation and Divergent Selection During Character Displacement 65 Introduction • 66 Methods 70 Results 79 Discussion 82 General Conclusions 95 iv Literature Cited L I S T O F T A B L E S Table 2.1. Coefficients for determining mean phenotype of various cross-types using Equation (2.2) 32 Table 3.1. Mean body mass and seven other morphological traits for the fish used in the enclosure experiment 47 Table 3.2. Crossing design and number of families used in the reciprocal transplant 48 Table 4.1. Number of mating trials performed for the various combinations of populations 60 Table 5.1. Total numbers of aquatic insect predators removed from low predation ponds and added to high predation ponds 88 Table 5.2. Summary data of fish from pond experiments 89 Table 5.3. Measures of the survival and growth of target individuals as a function of their trophic morphology 90 vi L I S T O F F I G U R E S Figure 1.1. A general-purpose geographic scenario for the evolution of reproductive isolation under the ecological model of speciation 15 Figure 1.2. Strength of premating isolation between independently evolved lines of Drosophila as a function of the similarity of their environments 17 Figure 1.3. Maximum hkelihood phylogeny of freshwater and marine sticklebacks 19 Figure 1.4. Hypothesized double-invasion scenario for the evolution of Limnetics and Benthics 21 Figure 2.1. Hypothetical examples of a reduction in the fitness of hybrids 33 Figure 3.1. Mean growth rates over three weeks of Limnetic and Benthic backcrosses . . . . 49 Figure 3.2. Growth rates of individual fish when held in enclosures in the littoral zone and open water 51 Figure 4.1. Uncorrected probability of spawning in no-choice mating trials for various combinations of populations 61 Figure 4.2. Population mean probabilities of spawning as a function of shared ecomorph . . 63 Figure 5.1. Estimated distribution of gill raker numbers in surviving target individuals . . . . 91 Figure 5.2. Growth of target individuals as a function of gill raker number when in the presence of Benthics or marines 93 vii A C K N O W L E D G E M E N T S First and foremost I want to thank my supervisor, Dolph Schluter. I have spent almost six years as a student in his lab, during both my Master's and PhD, yet have little desire to leave. Looking back on my career as a graduate student, while there are some things I would do differently if starting over, having Dolph as a supervisor would not be one of them. It never fails to amaze me how someone could be so busy yet continue to invest significant amounts of their time and effort into their students. Dolph was always there when I needed guidance, wanted help, or desired feedback. Most importantly, Dolph revealed to me a joy of research and discovery that I had not previously known. Thank you Dolph. A number of people helped in the field during the course of my research, sometimes under rather grim conditions, but always with the upmost dedication. These include Shawn Anderson, Graeme Brown, Anna Gosline and Andrew Hendry. Help collecting and transporting fish was graciously provided by Jenny Boughman, Rachel Marsden, Sian Morgan, Tamara Grand, Beren Robinson, Kasia Rozalska, Steve Vamosi, and others I am sure I am missing. Aneil Agrawal, Jenny Boughman, Sian Morgan, Laura Nagel and Kasia Rozalska spent many immobile hours in the cold, watching mating trail after mating trial. Assistance with all sorts of laboratory tasks, and work at the ponds, was provided by T o m Bell, Jenny Boughman, Kayla Feldman, Anna Gosline, Rachel Marsden, Michelle Roberge, Kasia Rozalska, Dolph Schluter, Karen Tsui, and Steve Vamosi. In addition, John Pritchard took the time to show me the ropes when I first arrived, and for that I am most grateful. Many earlier versions of the following chapters were subjected to thorough scrutiny by Dolph Schluter, Mike Whitlock, Sally Otto, as well as the various members of the S.O.W. group during my time at U B C . Their efforts have greatly improved the clarity of my writing, and if I am lucky, my thinking as well. Throughout my PhD I received financial support from a number of sources. These included a Natural Sciences and Engineering Research Council of Canada (NSERC) Postgraduate Scholarship B, a predoctoral scholarship from the Izaak Walton Killam Trust (UBC), and a Craig Sandercock Memorial Scholarship (UBC). In addition, my research viii expenses, as well as travel to a number of conferences, were generously covered by various N S E R C grants to Dolph Schluter, with additional help from Mike Whitlock. I thank my committee, which included Sally Aitken, Sally Otto, Dolph Schluter, Eric Taylor, and Michael Whitlock. They provided guidance when necessary, but also allowed me to develop my thesis independently. I owe a special debit of gratitude to Mike Whitlock, who spent more time with me than many supervisors spend with their own students. Mike played a crucial role in the analyses in Chapter two and it would not have been possible without him. More importantly, Mike acted in every way as a second supervisor, spending many hours teaching me things I wanted to know and providing significant funds from his own N S E R C research grants to allow me to conduct various Drosophila experiments and to present them at meetings. I have benefited greatly from being able to divide my time between two labs, and although the majority of the work I did with him does not appear in this thesis, its contribution to shaping my views, my approach to research, and my career cannot be overestimated. Thank you Mike. Last but not least, I want to thank Natalie. Her continued support and unfailing confidence in me is the primary reason I have persevered in academia. Her sacrifices, including spending many lonely weekends and weeknights without me, have not gone unnoticed. I am coming home now. ix P R E F A C E The work in Chapter 2 is a slightly modified version of the following article, co-authored by Michael Whitlock (Department of Zoology, University of British Columbia): Rundle, H . D . and M . C . Whitlock. 2001. A genetic interpretation of ecologically dependent isolation. Evolution: 55: 198-201. As the senior author, Howard Rundle was the primary contributor to all stages of this work, including conceiving the approach, analysing the model, interpreting the results and writing the manuscript. Michael Whitlock The work in Chapter 4 is a slightly modified version of the following article, co-authored by Laura Nagel, Janette Boughman, and Dolph Schluter (Department of Zoology, University of British Columbia): Rundle, H . D . , L . Nagel, J.W. Boughman and D. Schluter. 2000. Natural selection and parallel speciation in sympatric sticklebacks. Science 287: 306-308. Data collection for this work was a collaborative project involving Howard Rundle, Laura Nagel and Janette Boughman. As senior author, Howard Rundle took the lead role in analysing and interpreting the results, and in writing the manuscript. Dolph Schluter (senior author aside from the candidate) x C H A P T E R 1: E C O L O G I C A L M E C H A N I S M S I N S P E C I E S O R I G I N S : A N I N T R O D U C T I O N 1 A N O V E R V I E W OF E C O L O G I C A L SPECIATION The notion that ecology may play a significant role in the formation of new species has existed at least as long as the modern evolutionary synthesis, with discussions in the writings of Fisher (1930), Wright (1940), Muller (1942), Dobzhansky (1951) and Mayr (1942). However, since the idea was first proposed relatively little evidence has been brought to bear on the topic. Focus has been given instead to such topics as the study of speciation by purely genetic mechanisms (i.e., drift, founder-events; see Coyne 1992; Carson and Templeton 1984). Exactly how, and to what degree, ecological processes contribute to the formation of new species remains a fundamental question in evolutionary biology. In this thesis I address this issue by focussing on a classic hypothesis for the origin of species, recently termed ecological speciation (Schluter 1996a, b, 1998, 2000a; Orr and Smith 1998). Under this hypothesis, divergent natural selection is the ultimate cause of speciation; ecological processes and the environment are central because they are the source of divergent natural selection. Throughout this thesis I adopt Mayr's (1942) biological species concept, in which species are groups of actually or potentially interbreeding natural populations that are reproductively isolated from other such groups. I broaden this definition to recognize that incomplete reproductive isolation can exist between species that nevertheless retain their distinctiveness in nature. Such species serve to highlight the processes responsible for the origin and maintenance of their divergence (e.g., see Ehrkch and Raven 1969). The advantage of this biological definition is that it makes the study of speciation feasible by reducing it to understanding the mechanisms by which reproductive isolation evolves. Ecological speciation occurs when reproductive isolation evolves as a result of divergent natural selection between populations inhabiting different environments or niches (Schluter 1996a; 2000a). The macroevolutionary phenomenon of speciation is thus the product of the same microevolutionary mechanisms that adapt populations to their local environments. Ecological speciation includes scenarios in which sexual selection is responsible for the evolution of reproductive isolation, provided the divergence in mate preferences is ultimately the result of divergent natural selection between environments. Schluter (2000a) provides a classification of the various modes of speciation involving sexual selection, 2 indicating whether or not they are included as mechanisms of ecological speciation. In contrast, non-ecological processes of speciation occur when chance events, in place of divergent natural selection, play a central role in speciation. Examples include the evolution of reproductive isolation as a result of genetic drift, founder events, polyploidization, or the fixation of alternate advantageous alleles in populations under similar selection (Schluter 1996b, 2000a). Some speciation models involving sexual selection can result in the evolution of reproductive isolation in the absence of divergent natural selection between environments (e.g., Fisher's runaway process, chase-away models; Schluter 2000a). These also constitute non-ecological models of speciation. Returning to ecological speciation, reproductive isolation may evolve under this model in two ways: indirectly, as a by-product of the adaptive divergence of other traits, and directly, as an adaptive response to natural selection (e.g., reinforcement). I address these two alternatives in detail below. Indirect (By-product) Mechanisms of Ecological Speciation. —Differences in the environments experienced by two populations may cause divergent natural selection on various traits (e.g., morphology, physiology, behaviour) and, as a side-effect of the adaptive divergence of these traits, reproductive isolation (premating and postmating) may evolve. Reproductive isolation evolves as a by-product of this adaptive divergence either because it is a pleiotropic effect of the genes under selection or as a result of gametic phase disequilibrium between alleles causing reproductive isolation and those under divergent selection. I use the term 'environment' broadly here and distinguish two situations by which divergent natural selection may arise. First, differences in the external environment experienced by the two populations may cause divergent natural selection. The two populations may be allopatric and inhabit separate environments, or they may be sympatric and occupy separate niches within a single environment (e.g., exploiting alternate resources). Second, divergent natural selection may also arise in sympatry as a result of ecological interactions between the populations such as competition or predation. For instance, competition for shared resources may cause divergent natural selection on traits related to resource acquisition (i.e., ecological or 3 divergent character displacement; Taper and Case 1992; Schluter and McPhail 1992) and the resulting divergence of these traits may cause reproductive isolation as a by-product. Ecological interactions can also cause the evolution of reproductive isolation in the presence of a continuous resource gradient (i.e,. in the absence of two discrete environments). This can occur if ecological interactions among individuals generate disruptive selection on the population, with reproductive isolation evolving as a side-effect of the divergence in ecological characters (Dieckmann and Doebeli 1999). Direct Mechanisms of Ecological Speciation. —Reproductive isolation may be adaptive and directly favoured by natural selection. This occurs when two populations are sympatric and individuals that mate heterospecifically (i.e., with members of the other population) have a reduced fitness as a result of ecological mechanisms. Under these conditions, natural selection will directly favour individuals that are less likely to mate with the other population, thus strengthening premating isolation. The reduced fitness of individuals mating heterospecifically can arise in two ways. In the first, if some postmating isolation exists (i.e., hybrids have a reduced fitness), individuals that mate heterospecifically will tend to leave fewer, or less fit, offspring. Natural selection will then favour individuals that hybridize less. This is the process known as reinforcement (Dobzhansky 1940; Blair 1955). Alternately, individuals that mate between populations may themselves suffer a reduced fitness not via the fitness of their offspring. For example, individuals mating heterospecifically may experience an increased predation risk or may be exposed to an alternate suite of parasites. Natural selection will then favour individuals that mate within their own population, strengthening reproductive isolation in a 'reinforcement-like' process. Finally, direct mechanisms of natural selection, such as reinforcement, may complete a speciation process begun by one of the non-ecological alternatives discussed earlier (e.g., genetic drift, founder events). Whether such a speciation scenario should be classified as ecological or non-ecological is an open question. Geographic context of ecological speciation. —The hypothesis of ecological speciation 4 makes no claim as to the geographic context of populations during the speciation process. While much work has focussed on the geographic context of speciation (e.g., Howard and Berlocher 1998), less attention had been given to mechanisms of speciation. The ecological model of speciation encompasses a variety of mechanisms that operate under a diversity of geographic contexts. An all-purpose geographic scenario for the evolution of reproductive isolation under the ecological model of speciation is presented in Figure 1.1. Within this scenario, the evolution of reproductive isolation is divided into an allopatric and subsequent sympatric phase. Fully allopatric and fully sympatric speciation simply represent alternate extremes in which one or the other of the phases is absent. While ecological speciation encompasses all of these possible geographic contexts, the mechanisms responsible for the evolution of reproductive isolation do vary with geography (Fig. 1.1). Thus, while complete allopatry is not required, the reduction or elimination of gene flow clearly facilitates genetic divergence, affecting the strength of divergent selection required to generate reproductive isolation (Hendry et al. 2001). Thus an important issue in ecological speciation is not the geographic context per se, but rather determining the mechanisms that are responsible for the evolution of reproductive isolation, how they depend on the geographic context, and how they interact with gene flow. Evidence for Ecological Speciation In this section I address evidence for ecological speciation, beginning with laboratory studies and then considering studies in nature. Because recent and comprehensive reviews have been conducted for various aspects of both of these topics (e.g. Howard 1993; Rice and Hostert 1993; Schluter 1996b, 2000a; Noor 1999), my review is not intended to be exhaustive but rather to highlight key studies in each area. Evidence from the Laboratory. —Laboratory studies using Drosophila have clearly demonstrated the feasibility of the by-product mechanism of ecological speciation: reproductive isolation, both premating and postmating, can evolve as a side-effect of natural selection adapting populations to their different environments or niches (e.g., Robertson 5 1966a, b; de Oliveira and Cordeiro 1980; Rice and Salt 1988, 1990; Dodd 1989; Kilias et al. 1990; reviewed in Rice and Hostert 1993). Studies of premating isolation have been conducted both in sympatry (Rice and Salt 1988, 1990) and in allopatry (Kilias et al. 1990; Dodd 1989). The design of the allopatric experiments is straightforward. Replicate populations are derived from a common ancestor and placed in one of two environments that differ in some characteristic (e.g., temperature). These populations are propagated in their respective environments for a sufficient period of time to allow them to adapt to the divergent conditions. Tests of premating isolation are then performed among the populations, with the expectation under the ecological model that premating isolation should evolve between populations adapted to different environments but not between replicate populations that have independently adapted to the same environment. Results of two of these experiments, shown in Figure 1.2, clearly demonstrate that premating isolation can evolve by such a mechanism. Laboratory studies addressing a direct role for natural selection in the evolution of reproductive isolation have focussed on the hypothesis of reinforcement. Numerous Drosophila experiments have clearly demonstrated that the requisite genetic variation is commonly present in populations to allow selection to strengthen premating isolation (reviewed in Rice and Hostert 1993). However, results of more direct tests of the hypothesis of reinforcement are less clear. Two types of experimental tests of the reinforcement model have been performed and both are relevant to ecological speciation as they rely on divergent natural selection to reduce the fitness of hybrids (these are reviewed in Rice and Hostert 1993). In the first, two populations that have adapted to different environments in allopatry are placed in sympatry to determine whether any premating isolation evolves. Results of one such study show no strong evidence for increased premating isolation following secondary contact (Wallace 1982; Ehrman et al. 1991). A second study (Robertson 1966a) differed slightly in that it permitted limited gene flow between two divergently selected populations to determine whether any premating isolation would evolve between them. Again, no premating isolation was detected, but a parallel experiment with the same stocks suggested that genetic variation for such isolation may have been lacking. 6 The second experimental design involves applying artificial, disruptive selection to a trait within a single population and asking whether any premating isolation evolves to prevent the formation of low fitness hybrids. In one example of this type of study, Thoday and Gibson (1962) reported near-complete premating isolation evolving in only 12 generations of disruptive selection on bristle number. However, despite numerous attempts, these remarkable results remain unreplicated (Rice and Hostert 1993). The one partial exception is a study by Barker and Karlsson (1974) in which strong disruptive selection on bristle number resulted in the evolution of sporadic premating isolation. However, subsequent analyses suggest the disruptive selection was so strong as to effectively result in complete postmating isolation between the high and low selected lines. Thus this study, and potentially that of Thoday and Gibson (1962) as well, explored the evolution of premating isolation between what were, in effect, already complete species between which gene flow was not possible (Rice and Hostert 1993). Laboratory studies such as those described above are useful for determining the feasibility of various mechanisms of speciation. Indeed, these studies tell us that ecological speciation can occur in principle, although tests of a direct role for natural selection have provided mixed results and other potential mechanisms that may directly favour the evolution of reproductive isolation in sympatry remain unexplored. What these studies do not tell us is the importance of these mechanisms in the origin of the diversity that exists in the world today. To address this question, I now turn to evidence from nature in support of ecological mechanisms of speciation. Evidence from Nature. —Evidence from nature in support of ecological speciation is extensive, but the majority of this evidence is indirect and relies on pattern-based data. Observations consistent with ecological speciation include the rapid evolution of reproductive isolation in areas of ecological opportunity (i.e., island archipelagos), the high viability and fertility of hybrids between young species (indicating the absence of strong genetic mechanisms of postmating isolation predicted by non-ecological models of speciation; see Chapter 3), the persistence of species in the face of gene flow, and the obvious ecological and 7 phenotypic differentiation of closely related species (summarized in Schluter 1996b, 2000a). Ecological speciation is also implicated when the traits responsible for reproductive isolation between populations have known adaptive significance. For example, in the yellow monkey flower, Mimulus guttatus, the allele(s) responsible for adaptation to copper-contaminated soils are the same as, or are tightly linked to, alleles causing hybrid inviability (MacNair and Christie 1983; Christie and MacNair 1984). In addition, work with a different, closely related pair of monkey flowers (Mimulus lewisii and Mimulus cardinalis) has revealed that the different floral characteristics of the two species are adaptations to alternate pollinators and these differences also form the basis of the premating isolation between them (Schemske and Bradshaw 1999). Additional examples are summarized in Schluter (2000a). Extensive indirect, pattern-based evidence from nature also supports a direct role for natural selection in the evolution of premating isolation. Studies generally focus on demonstrating reproductive character displacement, defined as the pattern of greater divergence of an isolating trait between species in areas of sympatry than allopatry (Brown and Wilson 1956; Howard 1993). While reproductive character displacement was once considered a rare phenomenon (e.g. Littlejohn 1981; Phelan and Baker 1987), more recent work suggests otherwise (e.g., Coyne and Orr 1989, 1997; Rundle and Schluter 1998; Noor 1995, 1997; see Howard 1993 and Noor 1999 for reviews). In many of these cases however, the mechanism responsible for the reduction in hybrid fitness is not known (except see Rundle and Schluter 1998; Chapter 3) and so the role of divergent natural selection and the environment in these cases is not clear. Observational evidence is thus consistent with both a direct and an indirect role for divergent natural selection in speciation in nature. A more direct, experimental approach is warranted, however, because indirect evidence is unable to exclude other potential causes of the pattern under study. Unfortunately, direct tests of ecological speciation in nature are lacking (Coyne 1992; Schluter 1996a, b; Futuyma 1998). Indeed, Schluter (1996b) noted, "I am not aware of a single study that successfully links the evolution of reproductive isolation in nature with niche-based divergent natural selection." Since that date progress has been made and below I highlight some of the best evidence for a role of ecological mechanisms in species 8 origins in nature. Ecological speciation is more directly implicated when hybrids suffer a reduced fitness as a result of ecological mechanisms (Schluter 1996a, b, 2000a). This evidence is strongest when genetic mechanisms reducing the fitness of hybrids are absent, thus ruling out the alternative scenario in which ecological divergence occurred after speciation was already complete. Few studies have tested this prediction. Nagy (1997), working with two subspecies of the annual plant Gilla capitata, demonstrated natural selection and an evolutionary response in F 2 hybrids favouring the native plant at each site. Hybrids between the apple and hawthorn races of the apple maggot fly (Rhagoletis pomonella) appear to suffer as a result of a mismatch between the timing of their diapause and the phenology of their host plant (Feder 1998; Filchak et al. 1999). Survival of first and second-generation hybrids between host-races of the gall-forming tephritid fly (Eurosta solidaginius) was found to be reduced but also variable across host plant genotypes and years, suggesting ecological mechanisms as opposed to genetic incompatibilities as the cause (Craig et al. 1997). A few additional examples of this nature are reviewed in Schluter (2000a). A test of a direct role for natural selection in the evolution of reproductive isolation is provided by Higgie et al. (2000). In the wild, geographic variation in cuticular hydrocarbons, a trait important in mate recognition, between two species of Drosophila (D. serrata and D. birchii) exhibit a pattern suggestive of reinforcement: hydrocarbon profiles of sympatric populations of the two species differ more than those of allopatric populations. Higgie et al. (2000) brought allopatric and sympatric field populations of D. serrata into the lab and exposed them to experimental sympatry with D. birchii. Hydrocarbon profiles of the allopatric D. serrata populations rapidly evolved to resemble those of the sympatric field populations, presumably resulting in strengthened reproductive isolation between them, while those of the sympatric field populations remained relatively unchanged. Their results demonstrate a direct role for natural selection in strengthening premating isolation in these laboratory populations, however they do not require that similar selection was important historically in the origin of these species. Thus we still await conclusive evidence for a direct role of natural selection in speciation in nature. 9 It is clear from this summary that direct tests of ecological speciation in the wild are limited, and thus the role of ecological mechanisms in the origin species in nature remains poorly understood. In this thesis I present evidence that ecological mechanisms may be responsible for the origin of sympatric species of threespine sticklebacks. I explore the ecological mechanisms generating divergent natural selection between these species and test predictions of the ecological model concerning the evolution of reproductive isolation between them. To begin, I present a brief natural history of this study system. L I M N E T I C A N D BENTHIC THREESPINE S T I C K L E B A C K S M y research focuses on understanding the role of ecological mechanisms in the evolution of sympatric species-pairs of threespine sticklebacks. While most lakes or rivers in southwestern British Columbia, Canada, contain a single population of sticklebacks (termed 'solitary' populations), in a few, small, low elevation lakes, pairs of species coexist (McPhail 1984, 1992, 1994; Schluter and McPhail 1992). These sympatric species-pairs are referred to as Limnetics and Benthics after the habitat in which they primarily forage. In this section I present an overview of their natural history. The introduction to each subsequent chapter contains any additional information relevant to the questions being addressed in that chapter. Limnetic and Benthic sticklebacks are morphologically and ecologically distinct (McPhail 1984, 1992, 1994; Schluter and McPhail 1992). Limnetics are characterized by relatively small, slender bodies that are silver in colour (outside of the breeding season). They have narrower gapes and possess longer, more numerous gill rakers. (Gill rakers are protuberances from the gill arch that are thought to sieve food particles or direct the movement of water through the buccal cavity [Sanderson et al. 1991].) Limnetics feed primarily on zooplankton (especially outside of the breeding season) in the open water of the lake. Benthics, on the other hand, have relatively larger, deeper bodies that are more drab green in colour. They have wider gapes and fewer, shorter gill rakers. Benthics feed in the littoral zone of the lake on invertebrates that live on or in the sediment or attached to the vegetation. The morphological differences between Limnetics and Benthics have a polygenic basis and persist for multiple generations in a common laboratory environment (McPhail 1984, 10 1992; Hatfield 1997). Limnetic and Benthic sticklebacks evolved following invasion of freshwater by the marine threespine stickleback (Gasterosteus aculeatus) after the retreat of the Pleistocene glaciers -12,000 years ago. Marine sticklebacks are zooplanktivorous and are morphologically similar to Limnetics, although larger (Schluter and McPhail 1992; Pritchard 1998). Evidence indicates that independent invasions of the marine form occurred in at least three different drainages, giving rise to Limnetic -Benthic pairs with independent evolutionary origins (see below; Taylor et al 1997; Taylor and McPhail 1999, 2000). These separate drainages are: 1) Paxton Lake on Texada Island; 2) Priest, Balkwill and Emily lakes on Texada island; and 3) Enos Lake on Vancouver Island. Evidence that the phenotypic similarities of Limnetics, and of Benthics, among these different drainages are the result of parallel evolution as opposed to shared ancestry is strong. Each drainage is dominated by unique mtDNA haplotypes which are absent from the other drainages and from the marine populations sampled to date (Taylor et al. 1997; Taylor and McPhail 1999). In addition, other haplotypes within a drainage are usually separated from common marine haplotypes by only a single restriction site, whereas no two haplotypes from different drainages are separated by such a small difference. This suggests that the haplotypes in the different drainages are derived independently from the sea and not from another drainage. A n analysis of six nuclear micro satellite loci supports these results, revealing that little of the total allelic variation among individuals is structured into groups corresponding to 'Limnetics' and 'Benthics' (2.4-4.4%, which is not significantly different from zero; Taylor and McPhail 2000). This component would be expected to be larger if Limnetics and Benthics each had a single origin. Finally, the maximum-likelihood phylogeny of all the populations tested, based on the microsatellite data, suggests a polyphyletic origin for Limnetics and for Benthics (Fig. 1.3). While this phylogeny is poorly resolved and the majority of individual nodes are uncertain, it fits the data significantly better than one enforcing monophyly of either Limnetics, or Benthics, or both (Taylor and McPhail 2000). While the independent origin of the populations in these different drainages is clear, the geographic context of Limnetic-Benthic speciation is somewhat more controversial. 11 McPhail (1993; Schluter and McPhail 1992) proposed a specific scenario for their speciation involving two separate colonization events by the marine stickleback into freshwater (Fig. 1.4). The first invasion occurred when marines colonized lakes newly formed as a result of declining sea levels following the retreat of the Pleistocene glaciers. This first invader evolved an intermediate morphology adapted to exploiting both the open water and littoral habitats, as is characteristic of most solitary freshwater populations today (Schluter and McPhail 1992). A temporary rise in sea level a few thousand years later allowed the marine form to invade a second time. Competition between these populations for resources then caused the two forms to diverge. The intermediate form derived from the first colonization event evolved into the present-day Benthic, while the second invader persisted as a zooplanktivore (the Limnetic), similar in morphology and ecology to the present-day marine stickleback (Schluter and McPhail 1992; Pritchard 1998). This double invasion scenario is supported by allozyme and microsatellite D N A evidence (McPhail 1984; 1992; Taylor and McPhail 2000), as well as measurement of salinity tolerance of various populations (Kassen et al. 1995). In contrast, mtDNA evidence is suggestive of sympatric speciation (Taylor and McPhail 1999), but this discrepancy is thought to be an artifact of gene flow following secondary contact (Taylor et al. 1997; Taylor and McPhail 2000). In any case, none of the studies in the following chapters, nor any of their conclusions, rely directly on either of these geographic scenarios. A few aspects of these stickleback species-pairs make them an excellent study system for addressing questions of speciation (Hatfield 1995). While in nature reproductive isolation between these relatively young species is strong, perhaps complete (Taylor and McPhail 2000), hybrid crosses can be made and raised in the laboratory (e.g., Hatfield and Schluter 1999). In addition, given their incomplete isolation it is reasonable to assume that the mechanisms responsible for their evolution are still in operation, maintaining and perhaps furthering their divergence. Thus my work in this thesis, and that of a number of other stickleback researchers, seeks to understand these mechanisms as they have occurred in the past and continue to function today. 12 THESIS O V E R V I E W The organization of the thesis is as follows. Chapter two adopts a general approach, using theory to determine what the nature of postmating isolation between any two species can tell us about the role of ecological mechanisms in their speciation. Two types of postmating isolation are recognized, one of which is a unique prediction of the ecological model of speciation (ecologically dependent isolation). In collaboration with Michael Whitlock, I developed a quantitative genetic model to determine whether a reciprocal transplant experiment can be designed in such a way as to permit ecologically dependent isolation to be estimated while controlling for the other type of isolation. Analysis of the model reveals that such a design exists and involves the use of hybrid backcrosses. The remaining chapters deal directly with the sticklebacks. Chapter three addresses the role of ecological mechanisms in the evolution of postmating isolation between Limnetics and Benthics. It employs the experimental design suggested in Chapter two to test for ecological-dependence in the fitness of backcross hybrids in enclosures in the lake. Chapter four addresses the role of ecological mechanisms in the evolution of premating isolation. The basic premise is that divergent natural selection is strongly implicated if it can be shown that premating isolation has independently evolved among populations in correlation with environmental differences. To obtain accurate estimates of the strength of premating isolation among populations, numerous mate choice trials were conducted by Laura Nagel, Janette Boughman, and myself. Finally, Chapter five turns away from the evolution of reproductive isolation to consider the ecological mechanisms that generate the divergent natural selection that has contributed to the evolution and the maintenance of phenotypic differences between these species, as well as their reproductive isolation. Using a pond experiment, I measure how selection on a target population of sticklebacks differs between competition treatments and how this results changes under conditions of increased and decreased predation. Following these chapters, I once again adopt a broader perspective to integrate my results, drawing conclusions about the contribution of ecological mechanisms to the origin of Limnetic and Benthic threespine sticklebacks. I conclude by addressing what my work in the sticklebacks suggests about ecological speciation in general, highlighting advances that have 13 been made and areas in need of further work. FIGURE 1.1. A general-purpose geographic scenario for the evolution of reproductive isolation under the ecological model of speciation. Reproductive isolation is initially absent on the left and evolves to completion on the right. The vertical arrow indicates the stage during speciation at which sympatry is re-established. This can occur anywhere along the line, with fully sympatric and allopatric speciation simply representing opposite extremes in which one of the two phases is absent. Various mechanisms of ecological speciation are listed above, indicating the geographic context under which they can occur. 15 D i r e c t -reinforcement / cost to heterospecific encounters Indirect Indirect -environmental -environmental / population interactions — 1 A l l o p a t r i c phase S y m p a t r i c phase S y m p a t r y none R e p r o d u c t i v e Isolation complete 16 FIGURE 1.2. Strength of premating isolation between independently evolved lines of Drosophila as a function of the similarity of their environments. Al l lines were derived from a common ancestor and then allowed to adapt to one of two environments. Open circles are from Dodd (1989) using D. pseudoobscura, solid circles from Kilias et al. (1980) using D. melanogaster. A proportion of mating of 0.5 indicates no premating isolation between the lines (i.e., random mating). 17 0 . 6 H J e 0 . 5 H a> OA-0.3H 0.2H different same E n v i r o n m e n t 18 FIGURE 1.3. Maximum likelihood phylogeny of freshwater and marine stickleback populations based on data from six microsatellite loci, redrawn from Taylor and McPhail (2000). Values at nodes represent bootstrap support greater than 50%. Solitary 1-4 represent freshwater populations occurring alone in lakes or rivers within the same geographic region as the two-species lakes. The marine samples are very similar to one another in microsatellite allele frequencies, as would be expected from high levels of gene flow among sites. Monophyly (with the exception of Marine 4) of the freshwater populations is also consistent with high levels of gene flow among marine populations and does not require a single marine population to be the common ancestor of all freshwater populations. Emily Lake is in the Priest Lake drainage, lying approximately 1 km downstream of Priest Lake. 19 R CD PH 5 6 T o C O o C O o (3 1—1 O C O <D S3 O O C O My 1^-54 o c S3 CQ c o PH (—3 a o X PH a a •= a a •3 PH 90 (3 CD t— .2 <u a -5 « 70 20 FIGURE 1.4. Hypothesized double-invasion scenario for the evolution of Limnetic and Benthic sticklebacks (Schluter and McPhail 1992; McPhail 1993). The sequence involves two separate invasions of lakes by the marine sticklebacks (indicated by the sloping arrows), separated by a few thousand years. The various populations are placed along a phenotypic axis indicative of their trophic morphology, with traits characteristic of the Benthic at the left extreme and the Limnetic at the right extreme. Populations are: ancestral marine (*); intermediate form (•); Limnetic (O); and Benthic (A) . The first invader gave rise to the present day Benthic, while the second invader remained a zooplankton specialist like its marine ancestor, evolving into the present-day Limnetic. Modified from Pritchard (1998) and Pritchard and Schluter (2001). 21 v First M a r i n e ^ Invasion GO Intermediate fa F o r m Evo lves § ^ Second Marine ^ ?* Invasion GO T3 a o CJ GO Character Displacement M o d e r n Species Pair * O o Benthivore Zooplanktivore Ecological Morphology 22 C H A P T E R 2: A G E N E T I C I N T E R P R E T A T I O N OF E C O L O G I C A L L Y D E P E N D E N T I S O L A T I O N 1 'I gratefully acknowledge the significant contributions to the work in this chapter that were made by Michael Whitlock. This chapter is a slightly modified version of a manuscript coauthored by Michael and I (Rundle and Whitlock 2001). 23 INTRODUCTION Determining the mechanisms by which new species arise is a fundamental problem in evolutionary biology. With the exception of the reinforcement of premating isolation, it is generally accepted that speciation occurs as a by-product of other factors, including divergent natural and sexual selection, polyploidization, and possibly genetic drift and founder events (Mayr 1963; Futuyma 1998). While significant advances have been made in the laboratory testing the feasibility of these various models (Rice and Hostert 1993), mechanisms of speciation in the wild remain poorly understood (Chapter 1; Coyne 1992; Schluter 1996a, b). The study of speciation is made tractable by focussing on the evolution of reproductive isolation. M y focus here is on the evolution of postzygotic isolation, of which two types have been distinguished (Rice and Hostert 1993; Coyne and Orr 1998; Schluter 1998). Ecologically dependent isolation occurs when hybrids have a reduced fitness due to an interaction between their phenotype and their environment (equivalent to Rice and Hostert's [1993] "environment-dependent" postzygotic isolation). Hybrids, if intermediate in phenotype between the two parent forms, may not be fit in the environment of either parent. Intrinsic genetic (or ecologically independent) isolation, in contrast, occurs when hybrids have reduced viability and/or fertility largely independent of their environment (termed "unconditional isolation" by Rice and Hostert [1993]). This fitness reduction is the result of incompatibilities between the parental genomes expressed when they are brought together in hybrids and/or the breakup of favourable gene combinations interacting epistatically in the parents. This classification of postzygotic isolation is valuable because not every mechanism of speciation can produce every form of isolation. For example, ecologically dependent isolation can result only from divergent natural or sexual selection; thus tests of the divergent selection speciation model (also referred to as by-product and ecological speciation, see below) focus on ecologically dependent isolation (Schluter 1988, Hatfield and Schluter 1999). On the other hand, intrinsic genetic incompatibilities can be produced by a number of speciation mechanisms (e.g., drift, founder-events, ecological speciation); thus their presence yields little information about the mechanism by which they arose. Ecological speciation occurs when reproductive isolation evolves ultimately as a result 24 of divergent natural selection. Under this hypothesis, species reside on different adaptive peaks and intermediate phenotypes suffer reduced fitness due to ecological mechanisms, in effect falling between niches in the environment (Chapter 1; Schluter 1996a, b). This prediction can be tested using reciprocal transplants in which the fitness of hybrids and parental types is evaluated in the habitats of both parents in nature (e.g., Hatfield and Schluter 1999). The use of hybrids (between divergent species or locally adapted populations) in reciprocal transplants has received much attention recently, but in general, these studies have not been designed to estimate the separate contribution of ecological mechanisms to the reduction of hybrid fitness (e.g., Nagy 1997; Emms and Arnold 1997; Craig et al. 1997; but see Hatfield and Schluter 1999). Because all mechanisms of speciation can produce intrinsic genetic incompatibilities, the key to this approach lies in determining whether any reduction in hybrid fitness has an ecological, as opposed to an intrinsic genetic, basis. Here I ask whether there is a sufficient experimental design for detecting ecologically dependent isolation. M y goal is to detect the reasons for hybrid unfitness. A model of the expected phenotype of an individual under the influence of outcrossing, originally developed by Lynch (1991a), is expanded to include two environments. I apply this model to measures of fitness of different genotypes in both ancestral environments to determine the relative strengths and weaknesses of using various cross-types in reciprocal transplants. M E T H O D S Lynch's Model The following model is a genetic interpretation of outcrossing, developed by Lynch (1991a). It describes the expected mean phenotype of various crosses between two species/populations. For further details, consult Lynch (1991a). Lynch made the following assumptions, which I will also make. Two parental populations (J5! and P 2) are in gametic -phase equilibrium, with the major loci for the characters of interest unlinked and restricted to autosomal loci. Lynch (1991a) also assumed that the phenotypes of all individuals are evaluated in a common environment. Given these conditions, two coefficients can be defined that describe the genetic composition of any individual with any combination of the two 25 parental genomes. The source index, 0 S, is a linear scale that ranges from -1 (when all an individual's genes come from P 2) to +1 (when they all come from P,). The hybridity index, 0H, is also a linear scale that ranges from -1 (when the individual contains genes from only one source population) to +1 (when the individual is crossbred at every locus). The F 2 population, derived from random crossing of Pj and P 2 and subsequent random crossing of the F,'s, is in Hardy-Weinberg equilibrium and is treated as the reference population (mean phenotype = u-0), with all gene effects present in other crosses expressed as deviations from this mean. Additive and dominance effects are represented respectively by a q and 5q, with the subscript q indicating the number of loci involved (I drop Lynch's subscript x because it is required only when the model is expanded to include inbreeding). Thus a2 represents the interaction of additive effects at two loci. The interactions among loci of additive and dominance effects are represented by terms composed of the genetic effects involved at each locus; note that these are single terms and not products. For example, {a2^i} is the single term representing the interaction of additive effects at two loci and dominance effects of another locus. Extracting the gene effects in the standard hierarchical order, the general expression for the expected phenotype of an individual, denoted ix, is n = K> + e s ( « i } + eH{8,} + 0 S 2 { « 2 } + 0 S 0 H { « A } + eH2{52} + • • • (2.1) Note that while only second-order epistatic terms are shown, the expression is easily expanded to include higher order terms (see Lynch 1991a). Two Environments and Environment Dependent Gene Effects Here I change the assumption of a single, constant environment to explore the situation of two populations or species in different environments. Let parental population P, be native to environment A and P 2 to B . I will use subscripts A and B to indicate individuals grown in these two environments. The environmental coefficient, 8E, is defined as 1 for' environment A and -1 for environment B. The environmental effect, {e}, is one-half of the difference in phenotype of the F2's in environments A and B ({e} = Vi[/x(F2 i A) 26 / i(F 2 B)]). The expanded model becomes ii = jUo + e s { a i } + 6H{61} + 0E{e} + 9 S 2 { « 2 } + 6 s e H { a A } + 6H2{52} + eseE{a1e} + eH0E{816} + -- - (2.2) where the reference ju,0 is now the average phenotype of the F 2 in both environments (/i 0 = mean [/i(F 2 A), /i(F 2 B)]). In this expanded model, {e} is the average effect of the environment on the phenotype, independent of gene action. Terms not including e represent genetic effects that can contribute to reproductive isolation between populations independent of the environment (e.g., Fig. 2.1 A). Interaction terms that include e represent gene effects on the phenotype that act in an environment-dependent manner. For instance, {axe} represents the interaction between single-locus additive effects and the environment. The presence of these components, which include both e and genetic terms, indicates ecologically dependent isolation (e.g., Fig. 2. I B ) . Again, only second-order interactions are shown in equation (2.2), but the extension to higher order terms is straightforward. Coefficients are of the form 0' s9 j H9 k E, where / and j refer to the number of additive and dominance effects involved and k is 0 or 1 depending on whether the genetic effects are environment-independent or dependent respectively. Coefficients for the common cross-types are presented in Table 2.1. Thus, for example, / i (F l i A ) = /*„ + {5,} + {e} + {82} + {5,e} + • • •. Ecologically dependent isolation is predicted by speciation via divergent selection (i.e., ecological speciation), so my goal is to estimate the extent of ecologically dependent isolation when controlling for the effects of any intrinsic genetic isolation that may exist. The primary term of interest for ecologically dependent isolation is the additive-by-environment interaction {c^e}. Using Table 2.1 and equation (2.2), examination of the mean phenotype of various crosses reveals that RESULTS I/2[/*(B1IA) - M ( B , , B ) ] = {e} + V2{axe] + V*{a2e) + • • • (2.3) 27 V2[fi(B2A) - M(B 2 B )] = {e} - V*{a,e} + V4{a2e} + • (2.4) where jU,(B, A) is the mean phenotype of backcrosses of F,'s to P, in environment A . Subtracting equation (2.3) from equation (2.4) gives an estimate of the additive-by-environment interaction: {«,€} = V2[li(BlA) - At(B 1 B) - ^(B 2 , A) + At(B2>B)] (2.5) Extending the comparison to higher order interactions, the difference between equations (2.3) and (2.4) estimates A E = £ 2 l _ q {aqe}, where q = (1,3,5,7 . . .). If the higher order terms are assumed small, then A E ~ {a,e}. The standard error of A E is S . E . ( A E ) = ' / 2 [ S . E . ( B , , A ) 2 + S . E . ( B , , B ) 2 + S . E . ( B 2 , A ) 2 + S . E . ( B 2 , B )2f , where for example B I,A is the mean phenotype of all B , individuals in environment A and S . E . ( B I . A ) is the standard error of B , , A (the standard error of B i , A is the standard deviation of B i , A divided by the square-root of the number of measurements of this particular genotype in this environment). Another second-order interaction that may be relevant to ecologically dependent isolation is the dominance-by-environment interaction. Using Table 2.1 and equation (2.2) reveals that {5,6} can be estimated by measuring the F , and F 2 in both environments. A comparison of the F,'s yields: i / 2 [ M ( F , , A ) - /x(F1 > B)] = {e} + {5,e} + {82e} + • • • . (2.6) Extending (2.6) to higher order interactions reveals that this comparison estimates {e} + D E , where D E = ^{Sj6}. In order to isolate the environment-specific dominance effects (DE), ;=] one must be able to estimate {e}. This is done by subtracting {e} = 1/2[jii(F2A) - / i ( F 2 B ) ] using F 2 data. Thus D E = ^ [ / ^ ( F , A ) - JU(F, B ) - jit(F2 A ) + |ii(F 2 B ) ] . When estimated in this manner, 28 the standard error of D E is: S . E . ( D E ) = K 2 [ S . E . ( F 1 , A ) 2 + S . E . ( F 1 , B ) 2 + S . E . ( F 2 . A ) 2 + S . E . ( F 2 , B ) 2 ] ' / I Alternately, if one is willing to assume that terms involving third-order and higher interactions are absent, equations (2.3), (2.4) and (2.6) can be used to estimate {e}, {ccxe}, and {S,e} using only both backcrosses and the F,'s. Examination of equation (2.2) also reveals that any isolation detected by a comparison of F! or F 2 hybrids to the native parental form in each habitat can involve contributions of both intrinsic genetic and environment-dependent gene effects. For instance, M ( P 1 > A ) - M(FI,A) = {<*.}- 2(5,} + {a2} - K 8 , } + { « , e } - 2{8,e} + • • • (2.7) / i (P 2 , B ) - M(FI,B) = - K J " 2{5,} + M + K M + + 2{8,e} + • • • (2.8) When these cross-types alone are used, environment-independent gene effects cannot be controlled for and, if present, will contribute to any isolation detected. If these are the only data available, then one cannot infer the extent to which speciation was ecologically mediated. DISCUSSION For the ecological speciation hypothesis, it is critical to show that the fitness of hybrid genotypes is reduced relative to the parental forms in each habitat and that this reduction is the result of ecological mechanisms. An expansion of'Lynch's (1991a) model to two environments reveals that a comparison of the fitness of both backcrosses in both environments estimates environment-specific additive genetic effects. Including both the F , and F 2 in both habitats allows the additional estimation of ecological isolation attributable to dominance-by-environment genetic effects. This method is conservative as not all environment-dependent gene effects are estimated, and it does not require that the selective or genetic basis of the phenotypic differences between the species be known. It does assume that the present-day environments are representative of ancestral conditions that existed at the time of divergence and, like all reciprocal transplants, that the environments used in the transplant are representative of the true environments of the two parental forms (e.g., enclosure effects are 29 absent). Ecological speciation requires that the fitness of hybrids be reduced relative to the parental forms. Because heterosis can increase the fitness of hybrids (Lynch 1991a), the reciprocal transplant must include the native parental form in each habitat (P, in environment A , P 2 in B) to confirm that hybrid fitness is reduced. Reciprocal transplants of both parental forms alone would allow the inference of ecologically dependent selection, but data on hybrids is required to infer the basis of postzygotic isolation. The model provides a powerful design for a reciprocal transplant experiment to detect ecologically dependent postmating isolation. To summarize, both backcrosses should be included in both environments to permit the estimation of environment-dependent additive gene effects, independent of any positive or negative heterosis. Lacking previous information that hybrid fitness is reduced, the native parental form in each environment (i.e., P, in A and P 2 in B) should also be included as a benchmark. Finally, environment-specific dominance effects can also be estimated if the design permits the inclusion of the F[ and F 2 in both environments. The decision to include the parental forms, and the Fj and F 2 must be balanced against any loss in replication of the other crosses that occurs by doing so. Examination of the expanded model also reveals that any reduction in fitness detected using a comparison of the Fj or F 2 to parental forms native in each habitat (e.g., /x(P, A) -/i(F, A) or [x(P2 B ) - JM(F 2 B ) ) may result from a number of mechanisms. For instance, Fj hybrids may be unfit as a result of intrinsic genetic incompatibilities or because their intermediate phenotype is selected against ecologically. In other words, a reduction of Fj or F 2 fitness relative to parental forms in a transplant experiment is not sufficient evidence for ecological speciation. In addition to the method described above, there are at least two other ways to separate ecologically dependent from intrinsic genetic isolation. The first, used by Hatfield and Schluter (1999), is to compare the fitness of hybrids in the wild to that in a benign environment. The benign environment is one in which the environmental differences to which the divergent phenotypes are an adaptation have been removed or ameliorated. For example, if separate populations are adapted to different food resources, the benign environment may be 30 one in which an alternate food, which all individuals can easily consume, is provided in ample quantities so that differences in foraging efficiency contribute little to fitness measures. In effect, this method attempts to remove the hypothesized ecologically dependent isolation (i.e, all terms involving e) to allow the extent of the intrinsic genetic isolation to be estimated. Comparison of the two estimates can then yield the extent of ecologically dependent isolation. The second method involves artificially altering the phenotype of parental forms to resemble hybrids. Any reduction in fitness of these individuals in a reciprocal transplant, when properly controlled for the effects of the modification, can only be attributed to ecological mechanisms. An advantage of this approach is that it does not rely on the creation of hybrids and can thus be used even when reproductive isolation between parental species is complete. Each of these methods has inherent strengths and weaknesses and which of them is most useful in any situation will depend on the details of the species under study. For instance, the use of backcrosses in a reciprocal transplant experiment does not require a detailed understanding of the selective basis of the phenotypic differences between the species but does require that any intrinsic genetic isolation is not so strong as to prevent hybrids from being made. Finally, it must be noted that while useful, the separation of ecologically dependent and intrinsic genetic isolation may not be complete. The possibility exists that the fitness effect of some genetic incompatibilities may depend on the environment (e.g., along a benign to harsh axis), blurring the distinction between these two types of postzygotic isolation (see Kondrashov and Houle 1994; Szafraniec et al. 2001). For example, the fitness consequence of a specific intrinsic genetic incompatibility may depend on some aspect of the environment (e.g., its 'harshness') and this aspect may differ between the two parental environments (i.e., A and B). In this case, the fitness effect of this incompatibility is environment-dependent and would be represented in the model by a term that includes e. This is an area that clearly deserves further attention. Ecology has long been thought to play a role in speciation, but hmited progress has been made in determining its contribution to natural speciation events. We must determine unique predictions of this model and design techniques to test these predictions in the wild. Ecologically dependent isolation between young species is one such prediction. 31 T A B L E 2.1. Coefficients for determining mean phenotype of various cross-types using Equation (2.2). Coefficients for 0 S and 0 H are from Lynch (1991a). B{ and B 2 represent backcrosses of the F[ to the parental forms Fl and P 2 respectively. Cross-type 0s 0H Environment A Environment B Pi 1 -1 1 -1 P 2 -1 -1 1 -1 F , 0 1 1 -1 F n {n > 2) 0 0 1 -1 B , Vi 0 1 -1 B 2 -Yi 0 1 -1 32 FIGURE 2.1. Hypothetical examples of a reduction in the fitness of hybrids (open circles) relative to both parent species (closed circles) as a result of two alternative mechanisms. Reproductive isolation can be a function of either or both of these factors. A) Postmating isolation results only from intrinsic genetic incompatibilities between parental populations, independent of their ecological context. The relative fitness of hybrid may vary and depends on the extent of heterosis, genetic incompatibilities and the breakup of favourable gene combinations interacting epistatically in the parents. B) Intrinsic genetic incompatibilities are absent and reproductive isolation is solely the product of ecological mechanisms. Shown is the simplest case in which morphology is inherited in a purely additive fashion. Population Pl is native to environment A and P 2 to B . Relative fitness increases as genotypes approach the species native to each habitat. Lines connect the same backcrosses in the alternate environments, demonstrating the reversal in relative fitness expected under the ecological model. Non-linear patterns may result when dominance and epistatic effects contribute to the morphology of hybrids. 33 INTRINSIC G E N E T I C P. B, F, B 2 P 2 E C O L O G I C A L L Y D E P E N D E N T 1 F l B 2 P 2 P. B, F, B 2 P 2 C r o s s - t y p e 34 C H A P T E R 3: A T E S T O F E C O L O G I C A L L Y D E P E N D E N T I S O L A T I O N B E T W E E N S Y M P A T R I C S T I C K L E B A C K S 2 2This chapter is a slightly modified version of a manuscript that is under review for publication in Evolution (Rundle, H . D . In review. Ecologically dependent isolation between sympatric sticklebacks. Evolution). 35 INTRODUCTION While natural selection has long been thought to play a central role in the formation of new species, examples from nature in support of this hypothesis are scarce (Chapter 1; Coyne 1992; Schluter 1996a, b; Futuyma 1998). The study of speciation has instead focussed on other issues including the role of founder events in the evolution of reproductive isolation and the geographic context of speciation (i.e., sympatric vs allopatric). Renewed interest in mechanisms of speciation, however, has revived questions concerning selection's role in the evolution of reproductive isolation. Ecological speciation is a classic scenario for speciation in which selection plays a central role. Ecological speciation occurs when reproductive isolation evolves ultimately as a result of divergent natural selection between populations inhabiting distinct environments or exploiting alternate resources (see Chapter 1; Mayr 1942; Muller 1942; Dobzhansky 1951, 1970; Endler 1977; Rice and Hostert 1993; Schluter 1996a, 1998, 2000a). Reproductive isolation (pre- and/or postmating) builds as populations climb separate adaptive peaks under the influence of natural selection. While laboratory experiments using Drosophila have demonstrated the feasibility of this model (e.g., Fig. 1.2; reviewed in Rice and Hostert 1993), there are limited data concerning its importance in nature (MacNair and P. Christie 1983; Craig et al. 1997; Feder 1998, Nagel and Schluter 1998; Filchak et al. 1999; Hatfield and Schluter 1999; Rundle et al. 2000). Here I focus on the evolution of postmating isolation by divergent selection. Two types of postmating isolation are recognized (Chapter 2; Rice and Hostert 1993; Coyne and Orr 1998; Rundle and Whitlock 2001). The first is ecologically dependent isolation (equivalent to Rice and Hostert's [1993] 'environment-dependent' postzygotic isolation) and occurs when hybrids, because they are intermediate in phenotype, are less efficient at exploiting the dominant parental environments (or resource) and an intermediate environment (or resource) is lacking. In effect, these hybrids fall between niches and thus suffer reduced fitness. The second type of postmating isolation occurs when hybrids suffer reduced viability and/or fertility because of intrinsic genetic incompatibilities between the parental genomes independent of their ecological context (equivalent to Rice and Hostert's [1993] 36 'unconditional' isolation). While intrinsic genetic incompatibilities are an expected outcome of all mechanisms of speciation, ecologically dependent isolation is a unique expectation of ecological speciation (Fig. 2.1). I test this prediction using a recently diverged sympatric species-pair of threespine stickleback, referred to as Limnetics and Benthics (Chapter 1). Limnetics are small and fusiform and feed primarily on zooplankton taken in the open water of the lake. They have a narrow gape and long and more numerous gill rakers. In contrast, Benthics are larger and more robust, have a wider gape and fewer, shorter gill rakers, and feed on invertebrates in the littoral zone of the lake (McPhail 1984, 1992, 1994; Schluter and McPhail 1992). Premating isolation between the sympatric species is strong and previous studies suggest that it has evolved ultimately as a result of ecological mechanisms (Nagel and Schluter 1998). Despite this premating isolation, F! hybrids are found occasionally in the wild (<1% adults; McPhail 1992) and can be readily made in the laboratory using artificial crossing techniques. When raised in the laboratory, these hybrids are intermediate in morphology between the two parent species and are fully viable and fertile, showing no evidence of intrinsic genetic breakdown (McPhail 1984, 1992; Hatfield and Schluter 1999). However, tenth-generation hybrids (individuals from a population established from Fj hybrids ten generations earlier), which are also intermediate in morphology, have a foraging and growth disadvantage relative to each parent species in the habitat of that parent (Schluter 1993, 1995). Similarly, a transplant experiment in the wild revealed that Fj hybrids, despite showing no growth reduction in the laboratory, grew at 76% the rate of Limnetics in the open water and 73% the rate of Benthics in the littoral zone (Hatfield and Schluter 1999). The present study tests a further prediction of the ecological model, one that concerns the fitness of the backcrosses. If postmating isolation is ecologically dependent, hybrid fitness should depend on their phenotypic resemblance to the parent species (Fig. 2. IB). Fitness should be higher in hybrids that more closely resemble the native parent species in the habitat of that parent. The Limnetic backcross and Benthic backcross hybrids each resemble more closely the parent species from which it was mainly derived (Table 3.1; Hatfield 1997). Thus the ecological model makes a clear prediction concerning their fitnesses in the wild: the Limnetic backcross should outperform the Benthic backcross in the open water (the habitat 37 preferred by the Limnetic parent species), whereas the converse should be true in the littoral zone (Fig. 2. IB). The use of backcrosses has an additional advantage in that it allows me to discriminate between two possible hypotheses to explain the earlier F , transplant results of Hatfield and Schluter (1999). The first is that the reduced fitness of F , hybrids in the wild is the result of ecologically dependent isolation arising from their intermediate phenotype. Alternately, reduced F , hybrid fitness is the result of intrinsic genetic incompatibilities between the Limnetic and Benthic genomes but is expressed only in the stressful natural environment and not in the benign conditions of the laboratory where predation and disease are absent and food is abundant. Indeed, this possibility is suggested by Drosophila experiments demonstrating an increased difference in fitness between control and mutation accumulation lines when fitness was assayed under more harsh conditions (Kondrashov and Houle 1994; Szafraniec et al. 2001; see Hoffmann and Merlia [1999] for a review of the expression of genetic variation in favourable and unfavourable environments). Under this second hypothesis, the expression of reduced hybrid fitness is context dependent (benign vs. harsh conditions), but fitness does not conform to the 'ecological model' in which hybrids suffer because they fall between the two niches of the parent species. Backcrosses, but not F , hybrids alone, can be. used to distinguish these alternatives. Backcrosses provide a measure of ecologically dependent isolation while controlling for intrinsic genetic isolation that may be present (Chapter 2; Rundle and Whitlock 2001). This is because backcrosses are phenotypically different yet are similarly affected by genetic incompatibilities. Thus a comparison of the fitness of the backcrosses in the habitat of each parent species allows the contribution of these intrinsic incompatibilities to be removed. This expectation derives from an analysis of a quantitative genetic model of the phenotype of various individuals under the influence of outcrossing. The model, originally developed by Lynch (1991a), was expanded to included two environments and the resulting genetic effects that act in an environment dependent manner. Details are presented in Chapter 2 and in Rundle and Whitlock (2001). Analysis of this model reveals that a comparison of the fitness of both backcrosses in both environments estimates the additive effect of genes that act in an 38 environment dependent manner, independent of the contribution of any intrinsic genetic incompatibilities. This method is conservative as the comparison does not estimate all genetic effects that may contribute to ecologically dependent isolation (e.g., estimation of dominance effects that act in an environmentally-dependent manner requires additional crosses including the Fj). The current study is the first to use this method to estimate the extent of ecologically dependent isolation while controlling for any intrinsic genetic incompatibilities. In addition to the prediction concerning the relative fitness of the backcrosses, ecological speciation also requires that hybrid fitness be reduced relative to the native parental species in each habitat (i.e., selection must be divergent; Chapter 2). Thus I included both parent species in both habitats as a reference. Including the parent species also allows me to test a more general prediction of ecological speciation. In the absence of any intrinsic genetic isolation, ecological speciation predicts the rank order of the fitnesses of all cross-types. Basically, the closer in phenotype a cross-type is to the parent species native to that habitat, the higher its fitness (see Fig. 2.1). Thus the rank order of fitness in the open water of the lake is predicted to be: Limnetic > Limnetic backcross > Benthic backcross > Benthic. This order should be reversed in the littoral zone. M E T H O D S Experimental Crosses All crosses were made in the laboratory in the summer of 1999. Eight families each of both parent species (Limnetic and Benthic) and both backcrosses (F, crossed with a Limnetic or with a Benthic) were made and raised separately using the fully balanced design shown in Table 3.2. Al l Limnetic and Benthic individuals used to make these crosses were wild-caught fish recently trapped from Paxton Lake on Texada Island, British Columbia. Fj hybrids used in the crosses were lab-reared fish made the previous summer (1998) by artificially crossing wild-caught individuals from Paxton lake. To make crosses, gentle pressure was applied to the abdomen of a gravid female causing her eggs to be released into a petri dish filled with water. Fertilization was achieved by macerating the testes dissected from a single male in the same petri dish. Each clutch was then 39 hatched separately in an aerated, mesh-bottomed plastic cup suspended in one half of a divided 100L aquarium. Hatchlings were fed daily a diet of live Anemia nauplii. As they grew, their diets were supplemented with frozen bloodworms (chironomid sp.) and frozen adult brine shrimp (Anemia sp.). Approximately 2-3 months after hatching, 30 individuals from each clutch were transferred into separate, undivided 100L aquaria to equalize densities among families. Juveniles were held in the laboratory over the winter at a variable temperature (laboratory open to outside ambient temperature, 8-16°C) and a constant light regime (10L:14D). Fish were fed daily a mixed diet of live Anemia nauplii, frozen bloodworms and adult brine shrimp. To reduce the likelihood of individuals entering reproductive mode, these conditions were maintained until the fish were used in the transplant experiment. At approximately nine months of age, family sizes were reduced to 20 individuals each. When doing this, similar sized individuals were selected within each family so as to minimize body size differences among families that had arisen, likely as a result of differences in density and due to the different growth rates of the various cross-types. Despite this, small differences in body size remained among cross-types at the beginning of the transplant experiment (Table 3.1). Approximately one week prior to their introduction, the first dorsal spine of every fish was clipped to permit identification at the end of the transplant experiment. From the remaining 20 fish in each family, 12 were haphazardly selected for use in the transplant experiment. Reciprocal Transplant The design of the transplant experiment was similar to earlier experiments (Schluter 1995; Hatfield and Schluter 1999). The fitness surrogate was growth rate, measured over a three week period, of individual fish held separately in enclosures in either parental habitat (open water or littoral zone) in the lake. Past study has shown that experiments of longer duration suffer increasing colonization of the open water enclosures by benthic organisms (Schluter, pers. comm.). In May 2000, a total of 96 enclosures were placed in Paxton Lake: 48 in the littoral zone and 48 in the open water of the lake. The study was conducted at this 40 time because open water enclosures have the fewest benthic organisms colonizing them and previous studies have shown that the diet of fish inside the enclosures is similar to wild caught fish outside the enclosures at this time (Schluter 1995; Hatfield and Schluter 1999). Open water enclosures were cylinders, l m in diameter and 5m deep, made of 6-10mm knotless nylon mesh with a closed bottom and an open top encircled by a metal hoop. Eight enclosures were suspended from each of six rafts anchored in the deepest part of the lake. The littoral zone enclosures were l m x l m square, made of the same nylon mesh, with open bottoms and tops and a metal frame attached around the bottom margin. They were secured by sinking the metal frame into the sediment of the lake bottom with the top edge of each enclosure held out of the water on wooden stakes placed in each of the four corners. The enclosures were located haphazardly around the margin of the lake at a depth of approximately l m and were emptied of wild fish prior to the experiment. Prior to introduction, each fish was individually weighed to within 0.0 l g using a portable balance (Ohaus C T 200, Ohaus Corporation, Florham, New Jersey, U S A ) . On May 1, 2000, a single individual was placed in each enclosure. Twelve individuals (from 8 families) of each of the four cross-types were used in each habitat. Al l fish were approximately 11 months old when introduced to the enclosures and were large enough to prevent their escape but small enough to allow for significant further growth. Past studies have shown that calanoid copepods, a large component of the diet of wild Limnetics, tend to abandon open water enclosures (Schluter 1995). For this reason, the contents of a single plankton tow, taken in the open water of the lake at a depth of 0-2m over an approximately 100m distance, was proportioned among the open water enclosures daily. After three weeks, all fish were removed from their enclosures, weighed, anaesthetized, and then placed in 10% Formalin. Growth rate (mg-day1) of the fish from each enclosure was used as the independent observation. Fish were recovered from all 48 of the open water enclosures and from 45 of the littoral enclosures (two Limnetics and one Limnetic backcross were not retrieved). While none of the fish were in reproductive condition at the start of the experiment, a few were by the end. The weight of both males and females may change when they enter reproductive condition, so to avoid confounding effects, two gravid 41 backcrosses (one Limnetic and one Benthic) were deleted from the littoral zone analysis, leaving 43 replicates. However, results are similar when they are included. RESULTS The rank order of the fitnesses of the backcrosses was ecologically dependent and reversed in the two habitats, as predicted by ecological speciation (Fig. 3.1). In the open water, Limnetic backcrosses grew at more than twice the rate of Benthic backcrosses (10.1 vs 4.2 mg-day"1), while in the littoral zone Benthic backcrosses outperformed Limnetic backcrosses by a similar margin (16.7 vs 7.2 mg-day"1). This difference was confirmed by a significant interaction between habitat (littoral vs. open water) and backcross type (Limnetic vs. Benthic) in a two-way A N O V A on growth rates ( F M 1 = 17.61, P < 0.0001). The main effect of habitat was also significant, indicating that overall the growth rate of backcrosses was higher in the littoral zone than in the open water ( F 1 4 1 = 6.72, P = 0.013). The other main effect of backcross was not significant, indicating that the difference in growth rates of the backcrosses was similar in the two habitats (F 1 > 4 1 = 0.91, P = 0.346). These conclusions are not altered if the data are /n-transformed (with the addition of a constant to make every growth rate positive), with the exception that the habitat term becomes only marginally significant (P = 0.057). Using the growth rates of both backcrosses and equation (2.5) gives an estimate of the additive-by-environment interaction ( A E ± S.E.) of 7.7 ± 1.8. A E is approximately equal to {c^e} if higher order interactions are assumed small (Chapter 2). The prediction concerning the rank order of fitnesses was upheld in the littoral zone (Fig. 3.2A). Cross-type (i.e., Benthic, Benthic backcross, Limnetic backcross, Limnetic) had a significant effect on growth rate ( F 3 3 9 = 9.18, P < 0.0001). Benthics grew fastest at 17.2 mg-day"1, followed very closely by Benthic backcrosses (16.7 mg-day"1), with Limnetic backcrosses growing more slowly (7.2 mg-day"1), but still outperforming Limnetics (3.7 mg-day"1). A Tukey post hoc comparison of means (Zar 1996) indicated that the growth rates of Limnetics and Limnetic backcrosses were significantly lower than that of the Benthics and Benthic backcrosses, but no other comparisons were significant. Although growth rate was again dependent on cross-type in the open water ( F 3 M = 42 3.69, P = 0.018), results were less clear (Fig. 3.2B). Benthics and Benthic backcrosses grew slowly and at similar rates (5.2 and 4.2 mg-day"1 respectively); these were the only crosses in which some individuals lost weight. Limnetic backcrosses performed better (10.1 mg-day"1) than Benthics and Benthic backcrosses, as expected. However, contrary to expectation, Limnetics performed relatively poorly, growing at the same rate as Benthics (5.2 mg-day"1). A Tukey post hoc comparison of means (Zar 1996) indicated that the only significant difference in growth rate was between Limnetic backcrosses and Benthic backcrosses. Analysis of the diet of the fish indicated that the littoral enclosures were successful at replicating the food regimes of this habitat. The gut content of fish in the littoral zone enclosures consisted primarily of benthic invertebrates such as insect larvae (33%, 38%, 36%, 41%; mean percent in diet for Limnetics, Limnetic backcrosses, Benthic backcrosses and Benthics respectively) and ostracods and chydorids (57%, 38%, 52%, 56%). For neither of these prey types do the proportions in the diet differ significantly among cross-types when tested using a one-way A N O V A (proportions were arcsine transformed prior to testing; insect larvae: F 3 3 7 = 0.11, P = 0.95; ostracods and chydorids: F 3 3 7 = 0.78, P = 0.51). In the open water enclosures, diets were similar to that of wild Limnetics in the lake, consisting mainly of zooplankton such as pelagic copepods (62%, 77%, 37%, 51%). Differences in these proportions among cross-types approached significance however (F 3 3 7 = 2.44, P = 0.08), suggesting that different cross-types may have been exploiting different food resources. These differences in diet among cross-types mirror their differences in growth rate, with Limnetic backcrosses consuming the highest fraction of zooplankton and also growing fastest, while Benthic backcrosses consumed the lowest proportion of zooplankton and also grew slowest (Fig. 3.2). Ostracods were also found in the diets of the open water fish (26%, 15%, 40%, 29%), although the proportion in the diet does not differ significantly among cross-types ( F 3 3 7 = 1.56, P = 0.21). Ostracods are a prey item generally associated with the littoral zone of the lake and not normally found in significant numbers in the diet of wild Limnetics (Schluter 1993). Their presence also suggests that the environment within these enclosures was not fully representative of the open water environment of the lake. 43 DISCUSSION Under the hypothesis of ecological speciation, reproductive isolation arises ultimately as a consequence of divergent natural selection between environments (Chapter 1). Here I have focussed on the evolution of ecologically dependent postmating isolation, a unique prediction of the ecological model, between a young, sympatric species pair of threespine stickleback. I have extended earlier work by measuring growth rate, a component of fitness, of Limnetic and Benthic backcrosses in enclosures in the wild. The two types of backcross differ considerably in morphology; the Limnetic backcross is intermediate in morphology between the Limnetic and F[ hybrid while the Benthic backcross is intermediate between the Benthic and F, hybrid (Table 3.1). M y results demonstrate a large, environmentally-dependent genetic effect in the growth rate of the backcrosses, indicating a striking pattern of ecological dependence. In the open water, Limnetic backcrosses grew at more than twice the rate of Benthic backcrosses, while in the littoral zone Benthic backcrosses grew at more than twice the rate of Limnetic backcrosses (Fig. 3.1). In addition, neither backcross showed a substantial reduction in growth relative to its parent species, supporting past work suggesting the absence of any strong intrinsic genetic isolation between the two species (Hatfield and Schluter 1999). This result adds to a growing body of evidence that postmating isolation between sympatric Limnetic and Benthic sticklebacks in Paxton Lake is ecologically dependent. Hatfield and Schluter (1999) demonstrated that, using similar enclosures in the wild, F, hybrids grew at approximately three-quarters the rate of the native parental species in each habitat, yet in the laboratory F,'s showed no reduction in any of the four fitness components measured. In addition, Schluter (1993) demonstrated that individuals from a population established from F l hybrids ten generations earlier, which are similar in phenotype to F/s , were less efficient at foraging on the main prey type in each habitat than the native parental species. M y results, in combination with these past studies, strongly argue that postmating isolation has arisen between these sympatric species as a result of divergent natural selection. Past studies have also demonstrated a role of divergent selection in the evolution of premating isolation between Limnetics and Benthics. In laboratory mating trials, hybridization was found to be dependent on body size, a trait which evidence suggests is an adaptation to 44 their alternate environments (Nagel and Schluter 1998). Premating isolation may also have been reinforced directly by selection in sympatry (Rundle and Schluter 1998). In addition, premating isolation has evolved in parallel, in correlation with differences in the environment, among independently evolved populations of Limnetics and Benthics in at least three lakes, thus implicating natural selection (Chapter 4; Schluter and Nagel 1995; Rundle et al. 2000). The conclusion, when all of these studies are taken together, is that reproductive isolation, both pre- and postmating, has evolved via ecological mechanisms and that Limnetic and Benthic sticklebacks are prime examples of ecological speciation in nature. The hypothesis of ecological speciation also makes predictions concerning the rank order of fitnesses of all four cross-types in both habitats and requires that hybrid fitness be reduced relative to the native parental species in each environment. Given the intermediate morphologies of the backcrosses (Table 3.1), the rank order in the littoral zone was predicted to be: Benthic > Benthic backcross > Limnetic backcross > Limnetic. This order should be reversed in the open water. Results in the littoral zone were as predicted, although differences in growth rate between pairs of cross-types were not all significant. Evidence suggests that, in quantitative genetic terms, the genetic basis of the morphological differences between these sticklebacks is primarily additive (with the exception of armour traits that also include a significant dominance component; Hatfield 1997). As would be expected if growth rate is a function of morphology and morphology is inherited in a roughly additive fashion, the differences in growth rate between each backcross and the parent species from which it was derived was small compared to the difference between the two backcrosses (Fig. 3.2A). Results in the open water were less clear however (Fig. 3.2B), mainly because of the low growth rate of Limnetics, the cross-type expected to perform best in that habitat. Indeed, in past studies Limnetics have grown faster than Benthics (Schluter 1995) and faster than F, hybrids (Hatfield and Schluter 1999) in the same open water enclosures in the same lake at the same time of the year. It is not known why Limnetics performed so poorly this time, although mean grow rate of all cross-types in the open water was lower than in any past experiment (6.2 mg-day"1). One possibility is that benthic species of invertebrates may colonize the sides of the open water enclosures, creating a small littoral zone-like niche. The presence of 'alien' 45 resources may favour phenotypes intermediate between Limnetics and Benthics, or may result in higher growth rates of more benthic-like phenotypes than would normally be possible in the open water environment. The extent of colonization may vary year-to-year depending on environmental conditions during the experiment, such as the amount of incident sunlight and water temperature. This possibility is supported by the diet analysis of the fish in the open water enclosures. There was near-significant heterogeneity among cross-types in the proportion of their diet consisting of zooplankton, suggesting that the different cross-types may have been exploiting alternate food resources. In addition, the presence of ostracods, a prey item generally found in the diet of Benthics in the wild (Schluter 1993), suggests that the environment in these enclosures may not have been fully representative of the open water of the lake, potentially contributing to the reduced growth rate of the Limnetics. In comparison, ostracods formed only a small component of the diet of fish in open water enclosures in past experiments (2-3% Schluter 1995; <1% Hatfield 1995). Finally, although the current study adds to our growing knowledge of the basis of reproductive isolation between Limnetic and Benthic sticklebacks, our understanding is not complete. As noted by Hatfield and Schluter (1999), only one component of fitness, growth rate, has been measured over a small portion (three weeks) of their total life span (maximum life span 1-3 yrs). In addition, individuals were confined to one or the other of the two parental habitats, not being able to move between them or to exploit any other habitat that may exist in the lake (see General Discussion below). Other components of fitness, including predator avoidance, disease resistance and fecundity, measured over more inclusive portions of their life span, still remain to be addressed. 46 T A B L E 3.1. Mean (± SE) body mass and seven other morphological traits for the fish used in the enclosure experiment. Body mass was measured prior to introduction into the enclosures, while all other morphological traits were measured at the end of the experiment. Standard length, body depth, gape width, and gill raker length were /^-transformed prior to analysis. Body depth, gape width, and gill raker length were size corrected by regression on /^(standard length). Limnetic Benthic Trait Limnetic backcross backcross Benthic Mass (g) 0.95 (0.02) 0.99 (0.03) 1.12(0.02) 1.08 (0.03) Standard length 3.80 (0.01) 3.83 (0.01) 3.86 (0.01) 3.85 (0.01) Body depth -0.050 (0.012) -0.013 (0.011) 0.028 (0.008) 0.032 (0.007) Gape width -0.10(0.02) -0.08 (0.02) 0.05 (0.02) 0.12(0.01) Gill rake number1 16.86 (0.20) 16.23 (0.21) 13.70 (0.23) 12.75 (0.14) Gill raker length 2 0.19 (0.02) 0.06 (0.02) -0.06 (0.02) -0.16(0.02) Pelvic spine length 4.18 (0.14) 3.77 (0.15) 1.54 (0.36) 0.24 (0.14) Plate number 3 6.59 (0.11) 5.18 (0.18) 2.70 (0.25) 0.42 (0.16) 1 Total number on first gill arch. 2 Length of longest raker on first gill arch 3 Includes any plate, regardless of size, measured on the right side of the individual 47 T A B L E 3.2. Crossing design and number of families used in the reciprocal transplant. First letter of a cross-type indicates the species of its mother and the second indicates the species of its father. Numbers indicated the number of separate families. Wild and lab refer to whether individuals of that cross-type were wild-caught or lab-reared fish. Male Female L L (wild) L B (lab) B L (lab) B B (wild) L L (wild) 8 2 2 L B (lab) 2 2 B L (lab) 2 2 B B (wild) 2 2 8 48 FIGURE 3.1. Mean growth rates ( ± S E ) over three weeks of Limnetic backcrosses (0) and Benthic backcrosses (•) when transplanted to enclosures in the littoral zone and open water of Paxton Lake. 49 0.5 ] 0 Littoral Zone Open Water 50 FIGURE 3.2. Growth rates of individual fish when held in enclosures in A) the littoral zone and B) the open water of Paxton Lake. Symbols with error bars to the right of each cross-type represent means (± SE), while letters (a,b) indicate significantly different means using Tukey's post hoc comparison of means (Zar 1996). 51 A) Littoral zone Benthic Benthic Limnetic Limnetic backcross backcross B) Open water Benthic Benthic Limnetic Limnetic backcross backcross 52 C H A P T E R 4: N A T U R A L S E L E C T I O N A N D P A R A L L E L S P E C I A T I O N I N S Y M P A T R I C S T I C K L E B A C K S 3 3Significant contributions to the work in this chapter were made by Janette Boughman, Laura Nagel, and Dolph Schluter. This chapter is a modified version of a published manuscript coauthored by Laura, Janette, Dolph, and I (Rundle et al. 2000). 53 INTRODUCTION In classic theories of speciation, reproductive isolation originates in part as the incidental by-product of adaptation to distinct environments (Dobzhansky 1951, 1970; Mayr 1942, 1963; Muller 1942). While laboratory experiments support this view (e.g., Kilias et al. 1980; Dodd 1989; reviewed in Rice and Hostert 1993), the role natural selection and the environment play in the origin of reproductive isolation remains contentious because evidence from nature is lacking (Chapter 1; Coyne 1992; Futuyma 1998; Schluter 1996a, b). Tests of natural selection's role in speciation have focussed instead on the reinforcement of premating isolation (Coyne and Orr 1989, 1997; Noor 1995, 1997; Rundle and Schluter 1998; Saetre et al. 1997). Yet reinforcement requires preexisting reproductive isolation in the form of reduced hybrid fitness and is generally considered a final step in the speciation process (Coyne and Orr 1998; Dobzhansky 1951). Here I present evidence that natural selection plays a fundamental role in the early stages of speciation. Parallel evolution of similar traits in populations inhabiting similar environments strongly implicates natural selection, as genetic drift is unlikely to produce concerted change, correlated with the environment, in multiple, independent lineages (Clarke 1975; Endler 1986). Parallel speciation is a special form of parallel evolution in which traits determining reproductive isolation evolve repeatedly in independent, closely related populations as a by-product of adaptation to different environments (Futuyma 1998; Schluter and Nagel 1995). The outcome is reproductive compatibility between populations inhabiting similar environments, but reproductive isolation between populations inhabiting different environments. Because reproductive isolation is more strongly correlated with differences in the environment than with geographic proximity or genetic distance, parallel speciation provides strong evidence for natural selection in the speciation process. Despite the significance of such evidence for our understanding of mechanisms of speciation in nature, there are no conclusive tests of parallel speciation (Schluter and Nagel 1995). In collaboration with three colleagues (Janette Boughman, Laura Nagel, and Dolph Schluter), I carried out a test of parallel speciation using populations of sympatric threespine sticklebacks (Gasterosteus spp.). 54 Sympatric species of threespine sticklebacks inhabit small, low-elevation lakes in coastal British Columbia, Canada (McPhail 1984, 1992). These populations are recently derived from the marine threespine stickleback (G. aculeatus) that colonized freshwater following the retreat of the glaciers at the end of the Pleistocene. One species of each sympatric pair is a larger-bodied Benthic that feeds on invertebrates in the littoral zone; the other species is a smaller, more slender Limnetic that feeds primarily on plankton in the open water (Chapter 1; McPhail 1984, 1992, 1994; Bentzen and McPhail 1984; Schluter and McPhail 1992). The Benthic and Limnetic from a given lake constitute biological species: they are reproductively isolated via strong assortative mating (Nagel and Schluter 1998; Ridgway and McPhail 1984), ecologically-based postmating isolation (Chapter 3; Schluter 1995; Hatfield and Schluter 1999), and probably sexual selection against hybrid males (Hatfield and Schluter 1996; Vamosi and Schluter 1999). Phenotypic differences between sympatric species have a genetic basis and persist over multiple generations in a common lab environment (Hatfield 1997; McPhail 1984, 1992). Both comparative (Schluter and McPhail 1992) and direct experimental evidence (Schluter 1993, 1994) indicate that divergent selection caused by competition for resources has contributed to the evolution of these phenotypic differences (see Chapter 5). The genetic evidence indicates that the Benthic-Limnetic pairs from three lakes (Priest, Paxton and Enos Lakes) are derived independently of one another (Chapter 1). Unique assemblages of mtDNA haplotypes characterize pairs from the different lakes and a hierarchical clustering analysis of mtDNA divergence estimates fails to detect any case in which populations of the same phenotype from different lakes cluster together (Taylor and McPhail 1999). Independence of these species pairs is confirmed by an analysis of six nuclear microsatellite loci (see Fig. 1.3; Taylor and McPhail 2000). Thus, neither the Benthics nor the Limnetics from different lakes are monophyletic; hence for clarity throughout this chapter I refer to the two phenotypes as ecomorphs. Independence of pairs allowed two tests of the predictions of parallel speciation. First, populations of the same ecomorph from different lakes (for instance, Benthics from Priest, Paxton and Enos Lakes) should not be reproductively isolated from one another, despite the 55 known reproductive isolation between different ecomorphs within lakes (Nagel and Schluter 1998). Second, reproductive isolation should exist between ecomorphs from different lakes (for instance between Paxton Benthics and Priest Limnetics). We tested reproductive isolation by conducting mating trials in the laboratory using wild-caught Benthics and Limnetics from these three lakes. METHODS A trial involved placing a single, gravid female into a 100-L aquarium (or one half of a divided 100-L aquarium) in which a single male had built a nest. The pair was allowed to interact directly and whether or not they spawned was recorded after 30 minutes. A detailed description of this protocol (Rundle and Schluter 1998; Nagel and Schluter 1998) and information about the populations used (Bentzen and McPhail 1984; McPhail 1994; Nagel and Schluter 1998; Schluter and McPhail 1992) are provided elsewhere. Prior to calculating the mean probability of spawning for each combination of populations, spawning probabilities were first corrected for main effects of year and male population using logistic regression and then arcsin square-root transformed. No other main effects were present. Year controls for differences in the propensity to spawn between years and male population controls for varying propensity to spawn of males from different populations. Pairwise comparisons of the mean probability of spawning between populations are not statistically independent because of the phylogenetic relatedness of pairs of populations differs and the same populations appear in different comparison. The analyses thus employed conservative paired Mests that treated each of the six populations of females as a replicate and corrected for phylogeny (see below). In each test, the pair of measurements for each female population was based on averages of the corrected spawning probabilities over all the relevant male populations. Phylogenetic correction was performed on population-level data using the method of general least squares (Draper and Smith 1981; Martins and Hansen 1997). The equivalent of a one-sample paired r-test was performed by testing the significance of the intercept in a 56 weighted regression fitted to a zero slope. Weighted regression was performed using a matrix of weights whose elements were correlations specifying the degree of phylogenetic similarity (proportion of total branch length shared from root to tip) of pairs of populations. Larger correlations resulted in lower weights. Similar methods have been applied in other studies (Lynch 1991b; Lynch and Jarrell 1993; Schluter 1997). Correlations were calculated from the phylogenetic tree constructed from microsatellite distance data (Fig. 1.3; Taylor and McPhail 2000) using the U P G M A method in P H Y L I P (Felsenstein 1995). Similar results are obtained using the Kitsch algorithm (Fitch and Margoliash 1967). RESULTS AND DISCUSSION A total of 753 mating trials were performed4 (Table 4.1); 261 of these trials involved individuals of different ecomorphs (Limnetics with Benthics), while 492 involved individuals of the same ecomorph (Limnetics with Limnetics, Benthics with Benthics). Mean probabilities of spawning for each ecomorph combination are shown in Figure 4.1. Spawning probabilities between pairs of populations depend strongly on ecomorph identity (Fig. 4.2). Reproductive isolation between ecomorphs within a lake was strong (paired Mest: t5 = 3.82, P = 0.012; Fig. 4.2, comparison A), confirming past results (Nagel and Schluter 1998, Ridgway and McPhail 1984). In accord with the first prediction of parallel speciation (that populations of the same ecomorph from different lakes should be reproductively compatible), reproductive isolation was absent among lakes within an ecomorph (t5 = 0.56, P = 0.599; Fig. 4.2, comparison B). Females were just as likely to mate with a male of the same ecomorph from a different lake as with a male of the same ecomorph from her own lake. The reduced probability of spawning between Limnetics from different lakes (Fig. 4.1) is present in two of three Limnetic populations. While suggestive, the difference is not significant when comparison B is repeated using only Limnetics (paired r-test: t2= 1.154, P = 0.368). In accord with the second prediction of parallel speciation (that the reproductive 4243 of the trials had been previously conducted by Laura Nagel as part of her M.Sc. thesis. 218 additional trials were performed by Janette Boughman and 292 by myself. 57 isolation known to exist between ecomorphs within a lake should also be present between ecomorphs from different lakes), reproductive isolation was present between ecomorphs from different lakes (r5 = 2.61, P = 0.048; Fig. 4.2, comparison C). Females from a given population spawned more frequently with males of her own ecomorph from a different lake than with males of the other ecomorph from a different lake. The probability of spawning was slightly higher between ecomorphs from different lakes than between ecomorphs from the same lake and approached statistical significance (f5 = 2.36, P = 0.065; Fig. 4.2, comparison D). Correcting for phylogeny had a negligible effect on the statistical results, confirming that parallel speciation and not shared history is responsible for the observed mating patterns. Reproductive isolation between Benthics and Limnetics within a lake remained strong (t5 = 5.26, p < 0.003; Fig. 4.2, comparison A). Females from a given population remained as likely to spawn with males of the same ecomorph from a different lake as with males of the same ecomorph from her own lake (r5 = 0.075, P = 0.943; Fig. 4.2, comparison B). The probability of spawning for populations of females with males of the other ecomorph from different lakes remained significantly lower than with males of the same ecomorph from different lakes (r5 = 3.97, P = 0.011; Fig. 4.2, comparison C). Finally, however, the small and marginally significant increase in the probability of spawning observed for populations of females with males of the other ecomorph from different lakes vs males of the other ecomorph from the same lake was significant in the absence of the phylogenetic correction (paired Mest: t5 = 3.37, P = 0.020; Fig. 4.2, comparison D). The negligible effect of phylogeny occurs because, for phylogeny to explain the observed mating patterns, populations of the same ecomorph would have to be more closely related to each other than populations of different ecomorphs. Phylogenetic trees based on mtDNA and microsatellite D N A reject this hypothesis. No significant difference in microsatellite or mtDNA divergence is detected between pairs of populations from the same versus different environments (microsatellite: A N O V A , F u o = 0.22, P = 0.65; mtDNA: A N O V A , F u o = 0.76, P = 0.40 ). Conspecific pairs (i.e., pairs in which both populations belong to the same ecomorph within a lake) were excluded from this test. The parallel evolution of reproductive isolation in these sticklebacks in nature provides 58 some of the strongest evidence yet for a role of divergent natural selection in speciation. Two studies similar to ours suggest that reproductive isolation may also have evolved in parallel: populations of stream-resident sticklebacks from Japan and North America (J. McKinnon, S. Mori and D . Schluter, pers. comm.) and populations of herbivorous leaf beatles adapted to similar host plants (Funk 1998). This suggests that parallel speciation may be widespread. Our results complement and strengthen another form of evidence in which key traits, with known adaptive significance, also form the proximate basis of reproductive isolation (Nagel and Schluter 1998; see Chapter 1). These include beak and body size in Darwin's Finches in the Galapagos (Boag and Grant 1981; Price et al. 1984; Ratcliffe and Grant 1983) and copper tolerance in Mimulus (MacNair and Christie 1983; Christie and MacNair 1984). The absence of premating isolation between independently derived stickleback populations of the same ecomorph suggests that such key traits can evolve repeatedly in similar environments, yielding parallel speciation. The trait or traits underlying parallel mate preferences in sticklebacks have not been identified, but body size is a strong candidate (Nagel and Schluter 1998; Rowland 1989a, b; Borland 1986). Reproductive isolation between these sympatric species is not just a by-product of phenotypic divergence but may have additionally involved reinforcement in sympatry (Rundle and Schluter 1998). This suggests a scenario in which premating isolation between ecomorphs arose initially as a simple by-product of divergent natural selection on key traits and was later reinforced in sympatry (see Fig. 1.1). Whether reinforcement occurred in parallel among lakes is not known. Additionally, the reduced probability of spawning between Limnetics from different lakes (Fig. 4.1) and the slight reduction in reproductive isolation between ecomorphs from different lakes (Fig. 4.2, comparison D) suggest that a small degree of independent evolution has occurred within lakes. It is not known whether this independent evolution is a product of reinforcement or a by-product of unique adaptations to each lake. Regardless, under a common selection regime speciation was repeatable. The contribution of both divergent natural selection and reinforcement to speciation may explain the high rates of phenotypic divergence that characterize adaptive radiations (Schluter 1996a). 59 TABLE 4.1. Number of mating trials performed for the various combinations of populations. Male Paxton Paxton Priest Priest Enos Enos Female Limnetic Benthic Limnetic Benthic Limnetic Benthic Paxton Limnetic 54 27 9 38 12 Paxton Benthic 37 66 19 21 12 27 Priest Limnetic 19 25 45 23 Priest Benthic 44 65 17 20 9 Enos Limnetic 37 21 43 4 Enos Benthic 14 18 6 21 60 FIGURE 4.1. Uncorrected probability of spawning in no-choice mating trials for various combinations of populations (see Table 4.1). Error bars are ± 1 S.E. and represent the amount of variation in spawning rate among the various combinations. 61 • Same Lake • Different Lakes 0.6-1 Limnet ic L imnet ic x Benthic Benthic L imnet ic Benthic 62 FIGURE 4.2. Population mean probabilities of spawning as a function of shared ecomorph. Each point is the corrected fraction of all trials in which spawning resulted when individuals from a given pair of populations were tested. Comparison A represents the test for reproductive isolation between Limnetics and Benthics within a lake. Comparisons B and C highlight the tests of the two predictions of parallel speciation. First, within an ecomorph the probability of spawning is compared for combinations of populations from the same or different lakes (comparison B). Second, the probability of spawning between populations of the same ecomorph from different lakes is compared with that between ecomorphs from different lakes (comparison C). Comparison D tests for a difference in the strength of reproductive isolation between populations of ecomorphs from the same and different lakes. Because the statistical analysis employed conservative paired r-tests that treated each population of females as a replicate (see Methods), the comparisons shown here represent the nature of the tests but do not depict the exact analyses performed. 63 S a m e L a k e o Di f ferent L a k e s Dif ferent S a m e (Limnetic x Benthic) (Limnetic x Limnetic E c o m o r p h Benthic x Benthic) 64 C H A P T E R 5: P R E D A T I O N A N D D I V E R G E N T S E L E C T I O N D U R I N G C H A R A C T E R D I S P L A C E M E N T 5 5I gratefully acknowledge the contribution of Dolph Schluter and Steve Vamosi in helping design and carry out this experiment. 65 INTRODUCTION Interactions between populations have long been thought to be central to the evolution and maintenance of phenotypic diversity (e.g., Lack 1947; Taper and Case 1992; Schluter 2000b) and are a key source of divergent selection under the ecological model of speciation (Chapter 1; Schluter 2000a). The process of phenotypic evolution in a population or species, resulting from selection pressures created by its ecological interaction with another population, is referred to as character displacement (Taper and Case 1992; Schluter 2000b). While character displacement can cause traits of the interacting populations to diverge, converge, or to shift in parallel (Abrams 1996; Schluter 2000b), here I focus on ecological interactions between populations that cause divergent character displacement. The process of divergent character displacement is traditionally thought to involve competition for resources (Schluter 2000a, b). The basic idea is that competition between morphologically similar populations results in the depletion of shared resources, generally food. This creates divergent natural selection favouring individuals within each population that are capable of exploiting more abundant resources, leading to the evolutionary divergence of the populations (Taper and Case 1992; Schluter 2000a, b). There is extensive evidence for divergent character displacement in nature, but the majority of this evidence is observational and thus indirect (Taper and Case 1992; Schluter 2000a, b). The most common form involves an exaggerated difference in traits where two populations occur together than where they occur alone (Brown and Wilson 1956; Schluter 2000a, b). In the majority of these cases, however, evidence that competition is the interaction driving the divergence is weak or lacking (but see Martin and Harding 1981; Brown and Munger 1985; Pacala and Roughgarden 1985; Abramsky et al. 1990; Schluter 1994; Gorbushin 1996; Pfennig and Murphy 2000; Pritchard and Schluter 2001). In addition, even when such evidence does exist it does not eliminate a role for other interactions between the populations in the divergence of the characters (Schluter 2000a, b). Restricting divergent character displacement to include only competition may be too restrictive. In addition to competing for food resources, two populations may interact in other ways including via their predators, by direct predation upon one another, or through 66 behavioural interference (Schluter 2000b). The contributions to character divergence of interactions other than competition have received little attention (Abrams 1996). Determining the role of various ecological interactions between populations in nature, including competition, justifies an alternate approach involving direct experimental tests. Here I address the effect of predators in a system in which evidence already suggests a role for competitive interactions between populations. The presence of predators may affect divergent selection between two prey populations in a variety of ways. For example, shared predators may cause prey populations to compete for access to 'enemy-free space'. This process, also called 'apparent competition', has been explored theoretically and, under some conditions, can result in character divergence (Holt 1977; Abrams 2000). Alternately, the presence of predators may alter the competitive interactions between prey populations, thus interacting with competition to affect divergent selection (Abrams 1996). For example, increased predation could alter the habitat use of one or both of the populations, thus increasing or decreasing their competitive interactions. Alternately, predation and competition could act antagonistically, generating opposing selection pressures on traits within the populations, decreasing the resulting character displacement. There is little theoretical basis with which to predict how competition and predation may interact in any specific case, but a meta-analysis of manipulative field experiments suggests that predation weakens the effects of competition between species (Gurevitch et al. 2000). This analysis measured population-level effects, however, and the form and strength of selection on phenotypes within a population was not studied. Determining the relative importance of competition and predation to character divergence, as well as any potential interaction between them, requires experimental tests of the ecological mechanisms generating divergent selection between populations in nature. I carried out such a test using threespine sticklebacks. While most lakes in southwestern British Columbia, Canada, contain solitary populations of sticklebacks, in a few low-elevation lakes along the coast, Limnetic and Benthic species-pairs coexist. The distinct morphology and ecology of these species-pairs (Chapter 1; McPhail 1984, 1992, 1994; Schluter and McPhail 1992), which have evolved independently in at least three separate lakes 67 (Taylor and McPhail 1999, 2000), strongly implicates divergent natural selection in their origin. In every lake Benthics are larger, deeper bodied fish that have fewer, shorter gill rakers and feed on invertebrates in the littoral zone. Limnetics, on the other hand, are smaller, more terete, with longer and more numerous gill rakers, and feed primarily on zooplankton in the open water of the lake. The morphological differences between the species have a polygenic basis and persist for multiple generations in a common laboratory environment (McPhail 1984, 1992; Hatfield 1997). Solitary populations, when inhabiting lakes similar in characteristics to these two-species lakes, tend to be intermediate in morphology and ecology and exploit both the open water and littoral habitats (Schluter and McPhail 1992). Limnetic and Benthic sticklebacks evolved following invasion of freshwater by the marine threespine stickleback (Gasterosteus aculeatus) following the retreat of the Pleistocene glaciers -12,000 years ago. The geographic context of their speciation has been somewhat controversial, but the majority of evidence currently favours the double invasion scenario of McPhail (1993; Schluter and McPhail 1992) as opposed to sympatric speciation (see Chapter 1, Fig. 1.4; Taylor et al. 1997; Taylor and McPhail 2000). Nevertheless, divergent natural selection, caused by interactions between populations, is key under either geographic scenario. M y study builds on earlier work concerning the role of divergent selection, caused by population interactions, in the evolution of these sympatric species. Schluter (1994) used a pond experiment to contrast selection on a target population, possessing an intermediate morphology, when in the presence and absence of a zooplanktivore (the Limnetic). In the presence of the Limnetic, the growth rate of individuals in the target population increased with increasing morphological distance from the Limnetic. In the absence of the Limnetic, growth rates in the target were constant across their range of morphology. This is consistent with the zooplanktivore competing more strongly with morphologically similar individuals, thus creating divergent natural selection favouring target individuals that were more Benthic-like. Pritchard and Schluter (2001) conducted a second pond experiment designed to test a prediction concerning the strength of competition between two populations: that competition should decline through time as the populations diverge. Their experiment contrasted the strength of competition on a zooplanktivorous marine population during two separate stages 68 hypothesized to have occurred at different times during the double invasion scenario: before and after character displacement (see Fig. 1.4). The effect of competition on the marine population was found to be less in the post-displacement than in the pre-displacement treatment, consistent with the expectation that the strength of competition declined through time as the two populations diverged. In addition, the diets of the marine populations were more specialized on zooplankton in the post-displacement treatment, consistent with competition for food as the interaction driving the divergence (Pritchard and Schluter 2001). While these studies strongly implicate divergent natural selection in the evolution of these sympatric species, our understanding of the ecological interactions generating this selection is incomplete. This is because, in both of the experiments discussed above, insects were always present and predation by piscivorous fish was absent. In the wild, however, both piscivorous fish (e.g., cutthroat trout) and aquatic insects (e.g., dragonfly larvae and backswimmers) are key predators of sticklebacks (Pressley 1981; Reimchen 1980, 1994) and are present in all of the Limnetic-Benthic lakes (Schluter and McPhail 1992; Vamosi 2001). Because neither piscivorous fish nor insects were manipulated, it is impossible to draw conclusions about the relative role of predation in generating divergent natural selection. Furthermore, the contribution of competition and predation by insects cannot be separated. Understanding how competition and predation affect divergent selection is necessary if we wish to understand the ecological interactions between populations that have contributed to the evolution of present-day Limnetics and Benthics. It is also of general interest for understanding the evolutionary consequences of population interactions. That predation, in addition to competition, may have contributed to the divergence of Limnetics and Benthics is suggested by their morphology. As well as being adapted to alternate feeding niches, Limnetics and Benthics also exhibit divergence in traits indicative of adaptation to alternate predation regimes (e.g., see Table 3.1; Reimchen 1980, 1994; McPhail 1984, 1992; Vamosi 2001). Limnetics, which likely face gape-limited predators in the open water of the lake (e.g., piscivorous birds and fish), commonly possess relatively long pelvic spines and first and second dorsal spines. In addition, a series of vertical, bony plates extend laterally down the sides of their bodies. At the anterior end, these plates articulate with the 69 pelvic girdle/spines, and the dorsal spines, completely encircling the body in a bony ring. In contrast, in the littoral zone of the lake Benthics likely face invertebrate predators that are thought to forage by ambushing and grasping their prey (e.g., backswimmers, dragonfly larvae; Reimchen 1980). The first dorsal spine and the pelvic spine of Benthics are either absent or reduced, and the second dorsal spine is much reduced. Benthics also have reduced lateral plates, pelvic girdles, and dorsal spines, although the degree of this armour reduction varies among lakes. For example, Paxton Lake Benthics exhibit the most extreme reduction, often lacking these traits entirely (McPhail 1992). The absence of these traits has been hypothesized to make the fish more difficult to grasp by insect predators (Reimchen 1980). Compared to these extremes, solitary populations of sticklebacks, inhabiting lakes with a similar suite of predators, tend to be intermediate between Limnetics and Benthics in lateral plate count but have longer spines relative to their body-size (Vamosi 2001). The current experiment simultaneously tested the effects of competition and predation during character displacement. The competition treatments tested whether the phenotype of a second population affected the direction of selection on a morphologically-intermediate F , hybrid population (the 'target' population). The treatments involved placing the target population in four replicate experimental ponds along with Benthic or marine sticklebacks. Adding Benthics is expected to reduce the growth rate of the most Benthic-like individuals in the target population, while adding marine sticklebacks is expected to depress the growth rate of those individuals in the target population that are most Limnetic-like. The predation treatments tested whether predator densities influenced the direction of selection on the target population between the competition treatments. Both insect and trout predators were manipulated and the treatments were reduced insect density (low predation) and enhanced insect density plus trout (high predation). METHODS The 'target' of the experiment was a morphologically-intermediate, solitary population of sticklebacks. To increase the sensitivity of my measures of selection, this population was hybridized with both Limnetic and Benthic sticklebacks, creating a F , population with 70 increased phenotypic variance (see below). Natural selection on this target population was measured as the fitness (both growth and survival) of target individuals as a function of their trophic morphology. Because of the hybrid nature of the target population, however, any relationship between fitness and trophic morphology is not necessarily indicative of selection acting directly on these traits but may alternately be the result of selection on other traits that covary with trophic morphology. The effects of competition were tested using a paired design, created by dividing each of four experimental ponds in half. The target population was placed in both sides of each pond. The Benthic species was added to one side and marine sticklebacks were added to the other. The initial density of both the target and the added population were held constant, with only the phenotype of the added population varying between pond-sides. Selection on the target population was compared between pond-sides (see below) to determine the effect of the competition treatments. Selection is expected to arise from differences in growth of target individuals because, while past experiments have demonstrated consistent effects of competition on the growth of individuals, survival results have been variable (Schluter 1994, 2000b; Pritchard and Schluter 2001). The competition treatments were nested within the predation treatments: two of the ponds had reduced densities of insect predators and lacked trout while the other two ponds had enhanced insect densities and added trout. These predators directly prey upon sticklebacks, and selection resulting from the alternate predation treatments is thus expected to arise from differences in survival of target individuals. During the course of the experiment, one of the replicate ponds in the high predation treatment (Pond 11) became covered with a thick, floating mat of algae. While sticklebacks were recovered from this pond, trout were no longer present at the end of the experiment. Because this replicate was compromised, the pond was excluded from all analyses, leaving only one pond in the high predation treatment. This compromised the power of tests and only strong experimental effects are likely to be detected. For this reason some of the analyses performed are exploratory, examining trends in the data to inform future research. 71 Fish Populations Marine sticklebacks were collected from a small tributary of the Salmon River in Fort Langley, British Columbia. This population was chosen because of its ease of trapping and because its breeding season overlapped with that of the lake sticklebacks. It is morphologically and ecologically similar to other populations of marine sticklebacks in the region (Schluter and McPhail 1992; Bell and Foster 1994; pers. obs). Limnetic sticklebacks were collected from Paxton Lake, while Benthic sticklebacks were collected from Paxton and Priest lakes. Paxton and Priest lakes are located in separate drainages on Texada Island, B . C . Benthics from these lakes differ in their degree of armour reduction, with Paxton Benthics having the least armour (McPhail 1992, 1994). The two forms are similar in other characteristics (Schluter and McPhail 1992; McPhail 1994). The solitary population used to make the F , hybrid 'target' population was collected from Cranby Lake, also on Texada Island. Cranby Lake was chosen because it is similar to Paxton Lake in size, elevation, vegetation, prey species, and it lies less than 1 km away in a separate drainage. The morphology of this population has been well characterized and is known to be intermediate between Limnetics and Benthics (Schluter and McPhail 1992). Fish used in the experiment were obtained by artificially crossing wild-caught fish from these populations (using the technique described in Chapter 3) and then transporting the eggs back to the University of British Columbia (with the exception of the Salmon River marine population, in which the adults were transported back to the university first and the crosses done subsequently). Upon hatching, young fish were fed a diet of liquid infusoria for approximately three days, then switched to a diet of live Artemia nauplii. The fish were added to the experimental ponds at approximately one month of age. The.ability to detect directional selection on target phenotypes depends strongly on the frequency of phenotypes at opposite extremes within the target. This is because these individuals are expected to show the greatest difference in fitness. Because individuals with these extreme phenotypes are inevitably rare in natural populations, I used hybridization to create a hyper-variable target population to increase the sensitivity of my measures of selection. F , hybrids were made by crossing morphologically-intermediate females from 72 Cranby Lake with Limnetic and Benthic males from Paxton Lake. The target population consisted of an equal proportion of both types of F, hybrid (i.e., Cranby x Limnetic individuals and Cranby x Benthic individuals). The same technique has been used in previous experiments (Schluter 1994, 2000b), but I made the following two modifications. Past studies included Cranby x Cranby crosses, but these were excluded from the target population of the present study to further enhance the frequency of individuals at either morphological extreme. In addition, in the current study Cranby x Benthic crosses were composed of an equal proportion of hybrids made using Benthic males from Paxton and Priest lakes (i.e., 50% Cranby x Paxton Benthics and 50% Cranby x Priest Benthics). Benthics from these two lakes were used to enhance variation among individuals in armour characteristics, permitting selection on these traits to be estimated as part of a separate study. Experimental Design The experiment was carried out in the summer of 2000 in four ponds on the campus of the University of British Columbia. Each pond measures 23m x 23m and has a limestone border and a gradually sloping sand bottom that reaches a maximum depth of approximately 3m in the centre. These ponds were constructed in 1991 and seeded with plants and invertebrates from Paxton Lake. They have been used in the past in experiments that involved the temporary introduction of sticklebacks in spring and summer months and their removal in the fall. Except for their smaller size, evidence indicates that the ponds are representative of the native lake environment of sticklebacks. In past experiments, diets of fish in the ponds were similar to wild caught fish (Schluter 1994; Pritchard and Schluter 2001). Insect predators abound in these ponds, and sporadic predation by piscivorous birds also occurs. Past pond experiments have also demonstrated that they are capable of maintaining relatively large populations of sticklebacks across years and over multiple generations, with similar life cycles to wild sticklebacks (Pritchard and Schluter 2001; D . Schluter, pers. comm.). The four replicate, paired competition treatments were created by dividing each of the four ponds in half using an impermeable, plastic barrier. The lower edge of each barrier was anchored to the bottom of the pond using a steel chain that ran its entire length and paving 73 stones that were placed at approximately 0.5m intervals. The top edge of each barrier was held afloat by styrofoam strips held in a sleeve that ran the length of the barrier. Ponds were divided just prior to beginning the experiment and the integrity of the barrier was cheeked by scuba diver. No movement of fish across the barrier was detected at the termination of the experiment. On June 27, 960 Benthic sticklebacks (from Paxton Lake) were added to one randomly chosen side of each pond. Individuals from separate laboratory aquaria were first pooled into larger groups and then subsequently allocated haphazardly to the separate ponds. Individuals were introduced at dusk to minimize initial mortality of fish naive to insect predators. On June 28, the same procedure was followed to add 960 marine sticklebacks to the opposite side of each pond. Finally, on June 29-30, 1,400 individuals of the target population were added to each side of all the ponds. This competition treatment was nested within ponds assigned to two different predation treatments. These predation treatments were created by manipulating the densities of both predatory aquatic insects and cutthroat trout (Oncorhynchus clarki) in whole ponds. In the low predation treatment (two ponds), insect densities were reduced by trapping just prior to introducing the sticklebacks (Table 5.1). Insects were captured using minnow traps and dip nets. Although densities of some insects might have recovered fairly rapidly after removal as a result of their mobility (e.g., backswimmers), observations suggest that much of the mortality caused by insect predators occurs soon after the introduction of the fish. In the high predation treatment (two ponds), insects captured from the low predation ponds were added (Table 5.1). Because it was clear from a visual survey, prior to the introduction of the sticklebacks, that the two replicate high predation ponds (Pond 6 and 11) had unequal densities of backswimmers, captured insects from the low predation ponds were allocated preferentially to Pond 11 to reduce variation in density between these ponds. However, Pond 11 was ultimately excluded from the analyses due to the death of the trout. Finally, cutthroat trout were also added to both sides of the high predation pond. Additions were made approximately one month after the introduction of the sticklebacks, on July 23-25 Four cutthroat trout were added to each pond-side. This delayed introduction allowed the 74 sticklebacks to reach a size at which they would be susceptible to predation by the trout while minimizing trout-induced mortality of the predatory aquatic insects. These trout were wild caught individuals from Placid Lake in the University of British Columbia Research Forest. Although Placid Lake does not contain native sticklebacks, previous laboratory observations indicate that these trout readily prey upon them when given the opportunity (S. Vamosi, unpubl. obs.). During the introduction, pairs of similar sized trout were placed into the opposite sides of each pond to minimize size variation between pond-sides. In addition, the size matched pairs of trout were allocated so as to minimize variation among ponds. The experiment ran for 9-10 weeks, at which time the fish were harvested separately from each half pond. Individuals were removed over 24 hours using minnow traps, after which 0.5kg of 5% Rotenone (Syndel Laboratories, Vancouver, B.C.) was added. Sticklebacks and trout were collected using dip nets as they swam to the surface. This method collects a large portion of the fish in each pond, but some individuals that sink into the vegetation on the bottom are likely missed. Al l captured fish were first anaesthetized using tricaine methanesulfonate (MS-222, Syndel Laboratories) and then preserved in 95% ethanol. The following day after adding Rotenone, the ponds were searched again to retrieve dead individuals from the bottom that were overlooked on the first day. Data Acquisition Random subsamples of 175-200 individuals were taken from the total harvest of each half pond for data analysis. Subsamples were used because of the measurement of all the fish retrieved from each pond-side would be extremely time consuming and past analyses have shown that samples of this size are sufficient to detect selection (e.g., Schluter 1994). The fish in these subsamples were fixed in 10% formalin for two weeks, stained in a solution of 1% K O H and alazarin red for 48 hours, then preserved in 38% isopropyl alcohol. Within these subsamples, I used a different technique in the two competition treatments to identify which individuals belonged to the target. On the marine side of each pond, armour differences between marine and target individuals were sufficient to allow targets to be easily identified by visual inspection. Marine sticklebacks have 30-40 lateral plates (Hay and McPhail 1975; 75 McPhail and Hay 1983; pers. obs.), whereas the highest number of plates observed in the target population was eight. To separate target fish from Benthics in the subsamples from the Benthic side of each pond, I employed a linear discriminant function derived from the measurement of 114 laboratory-raised fish from known crosses (54 Paxton Benthics, 30 F , Cranby x Paxton Benthics, and 30 F , Cranby x Paxton Limnetics). This function separated individuals into two groups based on both their armour traits and number of gill rakers: low armour and fewer gill rakers vs. high armour and more numerous gill rakers. Hard parts of the anatomy, such as armour traits and gill rakers, were used because they are not affected by the diet of the fish, unlike some soft-bodied traits (Day et al. 1994; D . Schluter, unpubl. obs.). In addition, these traits are divergent between the groups. For example, in the laboratory-raised fish, 87% of Benthic individuals lacked pelvic spines. These Benthics also had, on average, 0.69 lateral plates and 12.6 gill rakers. In contrast, target individuals most similar to Benthics are the Cranby x Paxton Benthic F , hybrids, and only 20% of these individuals lacked pelvic spines and they averaged 8.6 lateral plates and 13.8 gill rakers. No laboratory-raised Cranby x Priest Benthic F[ hybrids were available for measurement. However, because Priest Benthics are more heavily armoured than Paxton Benthics (e.g., average lateral plate count of wild fish: Priest Benthics ~ 8.0, Paxton Benthics ~ 1.2; McPhail 1994), I assume that the discriminant function removing low-armoured Benthics will not remove any of this cross-type. To create the discriminant function, three traits were measured on each of the laboratory-raised fish: number of lateral plates, number of gill rakers, and pelvic spine ratio. Number of lateral plates was the average of the total count of any staining plate, regardless of its size, on both sides of the individual. Number of gill rakers was the total number counted on the first gill arch. The distributions of both of these traits were similar in the laboratory-raised and experimental, pond-raised fish, despite the smaller average size of the laboratory fish. Pelvic spine ratio was the length of the pelvic spine, measured on the left side of an individual, divided by standard length. This ratio was used because there was little overlap in the distribution of pelvic spine length between the laboratory and pond fish because these two groups did not overlap in body size. Dividing by standard length corrected this and was 76 preferable to using residuals from a regression of pelvic spine length on standard length because Benthics often lack pelvic spines. The resulting discriminant function classified the laboratory-raised fish with fairly high accuracy. Only one of the 30 laboratory-raised Cranby x Benthic hybrids was misclassified as a Benthic, and 3 of the 54 laboratory-raised Benthics were misclassified as Cranby x Benthics. Use of a quadratic discriminant function, which does not assume a common variance-covariance matrix for the traits in the groups being distinguished, did little to improve the classification (unpubl. obs.) so all subsequent analyses employ the linear function. The linear discriminant function was used to classify the experimental fish from the Benthic side of each pond. Al l individuals classified as Benthics were deleted, leaving only individuals identified as belonging to the target population for all subsequent analyses. The possibility that a few Benthic individuals are nevertheless mixed in with the target led me to analyse all results in two ways: first, using all individuals classified as targets by the linear discriminant function, and second, using only the individuals from the first data set that had a posterior probability of being a target greater than 95%. Results from the second analyses differed little and are not presented, with the exception of two cases involving marginally significant P-values (see Results). Measuring Selection I measured selection on trophic morphology in the target by examining both the growth and survival of individuals as a function of their number of gill rakers. Number of gill rakers (total number counted on the first gill arch) was used as an index of trophic morphology because, unlike some other trophic traits that differ between Limnetics and Benthics, it does not show a plastic developmental response to diet (Day et al. 1994). Selection on armour traits will be examined as part of a separate study. Selection resulting from differences in survival of target individuals differing in trophic morphology was measured within each pond as a 'survival differential'. The survival differential was measured as the mean number of gill rakers among survivors in the Benthic competition treatment minus the mean number in the marine competition treatment. This 77 quantity is not the difference between survivors and non-survivors within treatments (the conventional selection differential) but is identical to the difference between pond sides in these differentials. Divergent natural selection on the target between competition treatments would be indicated by a positive survival differential because under divergent selection we expect the survival of more Limnetic-like phenotypes to be highest in the presence of Benthics, whereas Benthic-like phenotypes should survive best in the presence of marines. A negative differential would indicate selection on the target favouring convergence in trophic morphology with the added population (Benthics and/or marines). Selection on the target was also examined using the slope of the relationship between growth of surviving individuals and number of gill rakers. Final standard length (measured to the nearest 0.02mm using Vernier calipers) was used as a measure of growth because fish were introduced to ponds at a small and essentially equivalent size. Within each pond, the change in strength of divergent selection on the target between competition treatments was calculated as the difference between pond-sides in slope of the relationship between growth and gill raker number. This 'growth differential' was calculated as the slope in the Benthic side minus the slope in the marine side. As with the survival differential above, this growth differential is expected to be positive if competition disproportionately reduces the growth of target individuals most similar to the second population in trophic morphology. To determine the effect of the predation treatment, the within-pond measurements of divergent natural selection (the survival and growth differentials) were compared between predation treatments. Data Analyses Survival and Growth.— I first test whether there was an overall effect of predation and competition on the mean survival and growth of the various populations. I tested differences in absolute survival of target populations between competition treatments using a paired f-test that treated ponds as replicates. Differences in absolute survival between predation treatments were tested, for target, Benthic, and marine populations, using two-sample /-tests that treated ponds as replicates. I calculated the absolute survival of target, Benthic, and marine 78 populations as the total number of individuals of that type retrieved from a pond (or a pond-side when testing for differences between competition treatments) divided by the number introduced. The total number retrieved was estimated for each population by multiplying the observed number of fish retrieved from a pond by the proportion of each type in the random subsample. Absolute survivals were arcsine transformed prior to testing in all cases. To determine whether increased predation affected the mean growth of surviving targets, I tested for differences in average standard length of targets between predation treatments using a two-sample Mest, treating ponds as replicates. To test differences in mean growth of targets between competition treatments, I used a paired-Mest that treated ponds as replicates. Measures of selection.— Significance of the survival differential was evaluated in each pond separately using a two-sample Mest between pond-sides, treating individual fish as replicates. Significance of each growth differential was analysed similarly using a Mest of the difference between two slopes. This was a one-tailed test because of the a priori expectation, deriving from past results (Schluter 1994, 2000b), that the differential should be positive. The survival differentials were compared between predation treatments using a two-sample Mest that treated ponds as replicates. The same procedure was followed to test the growth differentials between predation treatments. RESULTS Mean Survival and Growth There were few consistent effects of treatments on mean survival and growth. The average survival of targets was consistently lower in the high predation treatment than in the low predation treatment (Table 5.2), suggesting that the predators directly or indirectly reduced densities of the target populations. This difference, however, was not significant when whole ponds were treated as replicates (f, = 2.89, P = 0.21). Similarly, target survival was greater than Benthic survival in both low predation ponds, but the reverse was true in the high predation pond (Table 5.2). Again however, this difference was not significant (?, = 5.65, P = 79 0.11). The survival of Benthics and marines showed little difference between predation treatments (Benthics: r t = 0.07, P = 0.96; marines: t{ = 0.88, P = 0.54). In addition, within ponds no difference was detected in survival of target individuals between competition treatments (paired Mest: t2 = 0.53, P = 0.65). Target individuals had a higher survival in the presence of Benthics than in the presence of marines in one low predation pond, but the reverse was true in the other low predation pond and in the only high predation pond (Table 5.2). The mean growth of target individuals was consistently higher in the high predation treatment than in the low predation treatment, suggesting that the higher density of sticklebacks in the low predation ponds decreased overall resource levels. This difference, however, was not significant when whole ponds were treated as replicates (tl = 5.90, P = 0.11). Within ponds, the growth of targets was also consistently higher when in the presence of marines than Benthics (Table 5.2), suggesting that overall, competition for resources is strongest in the presence of Benthics. Again however, this difference was not significant (paired Mest: t2 = 2.91, P = 0.10). Selection on Trophic Morphology Survival.— A significant difference in selection was detected between competition treatments, but only in the high predation pond. The survival differential was significantly negative in the single high predation pond (Table 5.3), indicating selection on the target population favouring convergence with the added population. In contrast, survival differentials were positive and non-significant in both low predation ponds (Table 5.3). This difference in survival differentials between predation treatments is significant (tx = 90.64, P = 0.007), indicating selection on the target population was affected by the predators. The negative survival differential in the high predation pond reflects a shift in the distribution of gill raker counts, relative to that of the low predation ponds, in both competition treatments of the high predation pond (Table 5.3). In the presence of Benthics, mean gill raker count of target individuals was lower in the high predation pond that in the low predation ponds, although not significantly (r, = 2.48, P = 0.24). Likewise, in the presence 80 of marines, mean gill raker count of target individuals was higher in the high predation pond than in the low predation ponds, although again, not significantly (t{ = 2.93, P = 0.21). This shift of mean gill raker count in both competition treatments suggests that both Benthics and marines affected the magnitude of the survival differential. In addition, the significant survival differential in the high predation pond was caused primarily by a difference in survival of the most Limnetic-like target individuals between competition treatments (Fig. 5.1). In the presence of marines, the most Limnetic-like target individuals had high survival, approaching that of similar target phenotypes in the low predation pond. In the presence of Benthics however, survival of the most Limnetic-like target individuals was greatly reduced. In contrast, the survival of the more Benthic-like target individuals was consistently low in both competition treatments (Fig. 5.1). Growth.— A significant difference in selection was detected between competition treatments, again only in the high predation pond. In contrast to the survival differential, the growth differential was positive in the single high predation pond and approached significance (Table 5.3), suggesting divergent selection on the target population between competition treatments (Fig. 5.2). In the presence of Benthics, target individuals that were more Benthic-like in trophic morphology grew more slowly than target individuals that were more Limnetic-like. This pattern was reversed in the presence of marines, in which it was the more Limnetic-like individuals within the target population that suffered a reduction in growth compared with the more Benthic-like individuals. In contrast, the growth differential was small and non-significant in both low predation ponds (Table 5.3), indicating that selection on the target population did not differ between competition treatments. Within each pond, the growth of target individuals was similar across their range of trophic morphology in both competition treatments (Fig. 5.2). The difference in growth differentials between predation treatments is significant (tY = 53.02, P = 0.012), confirming that the presence of predators affected the direction of selection between competition treatments. This difference in growth differentials becomes marginally non-significant when tested using the subset of the data that includes only individuals with a posterior probability of being a target greater than 95% (r, = 8.63, P = 81 0.073). Comparing results from the survival and growth analyses, an important difference emerges in the high predation pond: the survival and growth differentials are opposite in sign. This indicates opposing relationships between fitness and trophic morphology when fitness is measured as growth versus survival. When present with Limnetics, target individuals that were most Limnetic-like in trophic morphology survived best but target individuals that were most Benthic-like in trophic morphology grew most. This pattern was reversed when the target population was present with Benthics. DISCUSSION The study of divergent natural selection resulting from interactions between populations or species has focussed almost exclusively on interactions arising via depletion of shared resources (i.e., primarily competition for food). Other ecological interactions have received little attention (Taper and Case 1992; Schluter 2000a; but see Abrams 2000; Doebeli and Dieckmann 2000). Populations may interact via a diversity of other mechanisms, however, including predation, direct predation upon one another, and behavioural interference (Schluter 2000b). The purpose of the current chapter was to explore the interaction between competition and predation in generating divergent natural selection between sympatric populations of sticklebacks, with the goal of understanding the ecological interactions that have contributed to the evolution of Limnetics and Benthics. I used a paired competition treatment to contrast selection on a morphologically-intermediate target population when in the presence of Benthic and marine sticklebacks. This competition treatment was nested within a predation treatment to determine how selection on the target population varied under conditions of increased and decreased predation. Two main results emerged. First, there was a strong interaction between competition and predation. In the high predation pond, selection on the target population varied between competition treatments. This was true for selection arising both from differences in survival (Fig. 5.1) and growth (Fig. 5.2) of target individuals. In contrast, in both low predation ponds there was no evidence that selection on the target population varied between competition 82 treatments, both in terms of the survival and growth of target individuals. Second, in the high predation pond, opposing selection arose from differences in survival and growth of target individuals. Mortality selection (selection arising from differences in survival of target individuals) was convergent: mean trophic morphology of the target population shifted toward the added population in both competition treatments (Fig. 5.1). This indicates that survival was highest among target individuals that most closely resembled the added population. In opposition, selection arising from differences in growth of surviving target individuals was divergent: growth was highest among target individuals that differed most in trophic morphology from the added population. When together with Benthics, target individuals that were most 'Limnetic-like' in trophic morphology grew best, while when together with marines, target individuals that were most Benthic-like grew best. What are the mechanisms by which competition and predation affected selection on the target population? The fact that growth was greatest in those individuals in which survival tended to be lowest (Table 5.2, Fig. 5.2) suggests that differences in survival (i.e., mortality selection) did not arise because some target individuals starved, but rather directly involved predation. In addition, the growth rate of target individuals in the high predation pond was high, averaging 0.48 mm/day ± 0.002 S E , whereas growth of wild sticklebacks from Cranby Lake at the same time of year averages only 0.19 mm/day ± 0.003 S E (Pritchard and Schluter 2001). For these reasons I conclude that mortality selection was largely a result of the predators and next consider the mechanism by which predation may have generated convergent selection on the target population. Enhanced survival of target individuals most similar to the added population could occur as a result of the dilution in predation risk experienced by these target individuals. The differences in survival of target individuals between competition treatments suggest that this occurred primarily as a result of the presence of marines (Fig. 5.1). When together with marines, the survival of the most Limnetic-like target individuals was remarkably high, approaching that observed in the low predation ponds. In contrast, in the presence of Benthics these Limnetic-like target individuals suffered heavy mortality whereas the more Benthic-like target individuals benefited only slightly (Fig. 5.1). That the greatest differences in mortality 83 between competition treatments occurred in the phenotypes that typically exploit the open-water habitat suggests that the cutthroat were the primary cause of this difference. Also consistent with this hypothesis, the overall survival of target individuals was twice as high in the presence of marines than Benthics (Table 5.2), suggesting that when marines were present they were the primary prey, but when absent, it was the most Limnetic-like target individuals that suffered. Consistent with this, the survival of marines was lower than the survival of Benthics in the high predation pond (Table 5.2). Nevertheless, this does not imply that predation by aquatic insects was rare; survival of the more Benthic-like target individuals was consistently low in both competition treatments. Determining the specific ecological interactions that produced convergent selection await data concerning the habitat use of various phenotypes and how predation rates by insects and trout vary in both habitats and in the presence of added populations. Enhanced growth of target individuals with increasing morphological distance from the added population (Benthic or marine) is consistent with the expectation based on stronger competition between similar phenotypes of the two populations. Competition alone, however, fails to explain why divergent natural selection was observed only under conditions of high predation. There are two possible explanations for this result. First, it may result from predation without competition. Sharing predators can generate 'apparent' competition between two prey populations (competition for access to enemy-free space), which may favour divergence between the populations that mimics competition for food resources (Holt 1977; Abrams 1996, 2000; Schluter 2000a). If shared predation led to divergent selection, however, we would expect to see it manifested in survival at least as strongly as in growth. There was no such evidence of this in the results on survival. The more likely possibility is that competition and predation interacted to affect selection on the target population. One scenario is that, by lowering population densities, predators weaken competition both within a population and between populations. A meta-analysis of field experiments supports this hypothesis, revealing that the intensity of competition tends to decline in the presence of predators (Gurevitch et al. 2000). A similar effect occurred in the current study, in which average growth rates of target individuals was 84 higher in the high predation pond than in the low predation ponds, likely as a result of changes in stickleback densities (Table 5.2). However, while the strength of competition may decline, selection does not necessarily weaken. This seemingly contradictory statement occurs because selection is not directly related to the strength of competition during character displacement. Rather, the strength of selection is determined by how much competition changes per unit change in phenotype (Doebeli 1996; Schluter 2000a). Thus, if the presence of predators increases this quantity, selection may become more effective despite a weakening of overall competition. There are two ways in which predators may alter how competition changes per increment of phenotypic change. First, predators may differentially affect the density of individuals of different phenotypes (the 'phenotypic density gradient'). For example, in the presence of marines, predators disproportionately reduced the density of more Benthic-like target phenotypes (Fig. 5.1). This may have created a steeper phenotypic density gradient, compared to that in the absence of predators, with proportionately fewer Benthic-like individuals and proportionately more Limnetic-like individuals. Under this scenario, a unit shift in phenotype of a target individual towards a more Benthic-like morphology reduces competitive interactions more than would otherwise occur in the low predation ponds. This is because the shift results in a greater reduction in relative density of phenotypicaUy-similar individuals in the high predation treatment than in the low predation treatment. Second, the presence of predators may narrow the resource spectrum available for foraging. This may alter the way in which the intensity of competition changes per increment of phenotypic change in the following way. The presence of predators may force phenotypes that were originally foraging elsewhere to a more central portion of the resource spectrum. For example, the presence of cutthroat trout may make the open water a risky habitat, forcing marines that would otherwise forage in this habitat to shift towards a more intermediate habitat, exploiting the portion of the resource spectrum used by the more Limnetic-like target individuals. This increases the density of individuals exploiting this portion of the resource spectrum, and thus as in the previous scenario, steepens the phenotypic density gradient. A unit change in phenotype towards a more Benthic-like morphology in the target population 85 would then produce a greater reduction in competition than would otherwise occur in the absence of predators. This is because of the increased density of individuals exploiting the resources at the 'Limnetic-end' of the spectrum relative to the density of individuals at the 'Benthic-end' of the spectrum. Determining the contribution of these two mechanisms to divergent selection on the target awaits further data concerning the habitat and resource use of different phenotypes under the different competition treatments. Finally, competition and predation interacted to affect selection on the target population, but it is unclear which predators were responsible. The predation treatment manipulated the densities of aquatic insects, including backswimmers and dragonfly larvae, as well as cutthroat trout. It is tempting to conclude from past results (Schluter 1994; Schluter 2000b; Pritchard and Schluter 2001) that insect predators alone are sufficient for divergent selection to result from differences in growth of target individuals between competition treatments. Similar results to the current study, in which an added population reduced the growth of individuals most similar to it in trophic morphology, have been found in two past studies (Schluter 1994, 2001). Both of these experiments lacked piscivorous fish whereas insect predators were present at unmanipulated densities. Nevertheless, excluding a role for predation by piscivorous fish is premature. These experiments differed from mine in ways other than the presence of predatory fish. For example, Schluter's (1994) competition treatment involved the presence versus absence of Limnetics. In addition, both Schluter (1994 and 2001) used Limnetics instead of marines in their competition treatment, and the hybrid crosses forming the target population in both of these studies included Cranby x Cranby crosses. Finally, in the current experiment mortality selection on target individuals differed between competition treatments primarily as a result of changes in survival of the more Limnetic-like target individuals (Fig. 5.1), consistent with predation by cutthroat trout. Thus whether the strength and form of selection is additionally affected by predation by piscivorous fish will require studies that independently manipulate densities of the separate predators. In conclusion, divergent natural selection resulting from ecological interactions among populations or species has long been thought to be central to the origin and maintenance of phenotypic diversity, as well as to the evolution of reproductive isolation (Chapter 1; Lack 86 1947; Taper and Case 1992; Schluter and McPhail 1992; Schluter 1996a, 2000a). Determining what these interactions are, and how they generate divergent natural selection, is thus key to understanding the evolutionary consequences of population interactions. I used sympatric populations of sticklebacks in a direct, experimental test of how predation and competition interacted to affect the direction of selection on a target population. Results suggest a strong interaction such that selection on the target population differed between competition treatments only in the presence of increased predation. In addition, in the high predation treatment selection arising from differences in survival opposed selection caused by differences in growth. While further study is warranted to determine the repeatability of these results, they suggest that the ultimate consequence of predators to character displacement between Limnetics and Benthics is important and complex. It is clear that ecological interactions other than competition can have important evolutionary consequences and deserve more attention than they have received to date. 87 T A B L E 5.1. Total numbers of aquatic insect predators removed from low predation ponds and added to high predation ponds. Invertebrates Removed Invertebrates Added Dragonfly larvae (Odonata) Backswimmers (Notonecta) Stick insects (Ranatrd) Pond 4 Pond 10 Pond 6 Pond 11 323 608 9 222 168 9 228 100 7 317 676 8 88 T A B L E 5.2. Summary data of fish from pond experiments. Total Retrieved is the number of fish collected from each pond-side at the end of the experiment. Total number of targets retrieved is an estimate based on the classification of a subsample of 175-200 individuals from each half-pond. Survival is the absolute proportion of individuals of each type retrieved divided by the number introduced. In all cases, 1,400 target individuals were introduced to each half-pond, along with 960 Benthics or 960 marines. Final target length is the mean standard length of all target individuals in subsamples taken from each pond-side. Data from pond 11 were not analysed because no trout were retrieved. Total Retrieved Competition Treatment Pond Survival Mean Final Total Target Target Benthics / Length Trout Sticklebacks Retrieved Targets Marines ( m m ± S E ) A) Low Predation Treatment Benthic 4 1043 673 0.48 0.39 41.50 (0.30) Marine 4 696 494 0.35 0.21 41.80 (0.25) Benthic 10 465 300 0.21 0.17 40.00 (0.24) Marine 10 727 560 0.40 0.17 41.26 (0.24) B) Fligh Predation Treatment Benthic 6 4 357 110 0.08 0.26 45.38 (0.36) Marine 6 3 386 234 0.17 0.16 46.60 (0.30) Benthic 11 0 282 - - -Marine 11 0 404 - - -89 TABLE 5.3. Measures of the survival and growth of target individuals as a function of their trophic morphology (number of gill rakers). Mean number of gill rakers is the average number on the first gill arch of surviving target individuals. 'Slope' is the slope of the regression of growth (final standard length) of surviving target individuals on gill raker length. Significance of survival and growth differentials are indicated by P-values of /-tests with sample sizes as indicated (see Methods). Survival Results Growth Results Mean Number Survival Growth Competition of Gill Rakers Differential Slope Differential Sample Pond Treatment (± SE) (P-value) (± SE) (P-value) Size A) Low Predation Treatment 4 Benthic 4 Marine 13.89 (0.10) 13.79 (0.08) 0.10 (0.43) 0.33 (0.27) 0.26 (0.26) 0.07 (0.43) 271 10 Benthic 10 Marine 14.06 (0.10) 13.95 (0.08) 0.11 (0.40) -0.30 (0.21) -0.35 (0.23) 0.05 (0.43) 283 B) High Predation Treatment 6 Benthic 13.61 (0.16) -0.68 0.38 (0.32) 0.70 160 6 Marine 14.29 (0.09) (<0.0001) -0.32 (0.31) (0.061*) *This result changes little when the test is performed on the subset of the data that include only individuals with a posterior probability of being a target greater than 95% (Pond 6 growth differential = 0.76, P = 0.050). 90 FIGURE 5.1. Estimated distribution of gill raker numbers in surviving target individuals in the single high predation pond. Solid bars indicate the Benthic treatment; open bars indicate the marine treatment. The average of the four pond-sides in the low predation treatment is provided for comparison (solid line). Results from both competition treatments of both replicate ponds in the low predation treatment did not differ. For each pond-side, the total number of individuals in each gill raker class was estimated by multiplying the frequency distribution of target individuals in each gill raker class in the random subsample by the estimated total number of target individuals at the end of the experiment. 91 11 12 13 14 15 16 17 Number of Gill Rakers 92 FIGURE 5.2. Growth of target individuals as a function of gill raker number when in the presence of Benthics (solid circles, solid line) or marines (open circles, dashed line). Individuals with higher gill raker counts are more Limnetic-like in their trophic morphology, while individuals with lower counts are more Benthic-like. For clarity, gill raker counts of individuals in the two competition treatments are displaced slightly on the x-axis. 93 A) Pond 4 (low predation) 11 12 13 14 15 16 17 18 £ B) Pond 10 (low predation) E i ° c • O 3 1 ' 11 12 13 14 15 16 17 18 C) Pond 6 (high predation) o a fe) 8 • o • • — . o 8 - , B -§ 11 IL • o • § o •5 • —• "so • ^ 8 • ° • l | •a •8 o o •1 •8 . o 8 o 10 11 12 13 14 15 16 17 18 Number of Gill Rakers 94 G E N E R A L C O N C L U S I O N S In this section I integrate the results from my research to draw conclusions about ecological speciation. I begin with the sticklebacks, summarizing what they have taught us about ecological speciation, highlighting some of what we do not know, and suggesting areas that would benefit from future research. T o conclude, I take a broader perspective to evaluate more generally what my research into stickleback speciation, along with that of others, suggests about speciation and the role of ecological mechanisms in it. The Thesis Chapter two addressed the evolution of postmating isolation, recognizing two potential sources of reduced hybrid fitness. The first, intrinsic genetic isolation, results from genetic incompatibilities between parental genomes and can be produced by all models of speciation. Isolation of this type forms the basis of the majority of studies investigating the genetics of postmating isolation (e.g., Orr et al. 1997; Ting et al. 1998; but see Hollocher et al. 1997). The second source of reduced hybrid fitness, ecologically dependent isolation, is a unique prediction of the ecological model of speciation. In Chapter two I used a quantitative genetic model to determine how best to evaluate the ecological-dependence of hybrid fitness. Analysis of the model revealed that ecologically dependent isolation could be estimated, while controlling for genetic incompatibilities, by measuring the fitness of both hybrid backcrosses when transplanted to the environments of both parent species. In addition, analysis of the model revealed that the reduced fitness of F, hybrid offspring, when placed in either parental environment, can be caused by both mechanisms and thus cannot, on its own, be used as evidence for ecological speciation. While the analyses in Chapter two were conducted with the sticklebacks in mind, the model was general and the suggested experimental design could be applied to any study system in nature, provided any existing genetic incompatibilities were not so strong as to prevent first and second generation hybrids from being made. This does not appear to be a severe constraint as strong genetic incompatibilities are often lacking, at least between 95 relatively young species in nature (e.g., Coyne and Orr 1989, 1997; Grant and Grant 1996; Schluter 2000a). The use of young species, lacking strong incompatibilities, is also preferable as it excludes the alternative scenario in which speciation is begun by non-ecological processes, with adaptation to divergent environments occurring only later in the process. Such species might exhibit both ecologically dependent and intrinsic genetic isolation, but divergent selection may not have played a primary role in their origin. A potential issue in the model in Chapter two concerns the division of postmating isolation into intrinsic genetic and ecologically dependent forms. Results of a number of recent laboratory studies clearly indicate that the fitness effects of genetic incompatibilities can be environment-dependent along a benign to harsh axis. For example, a laboratory Drosophila experiment by Kondrashov and Houle (1994) demonstrated that both fecundity and viability differences between control and mutation accumulation lines increased dramatically when fitness was assayed under more harsh conditions (e.g., diluted medium, crowding). In a similar study in yeast, Szafraniec et al. (2001) demonstrated that, while there was no detectable difference under benign conditions (30°C) , the fitness of mutation accumulation lines was reduced by approximately 20% relative to control lines under more stressful conditions (38°C). The conclusion from these studies is that deleterious gene effects can be expressed to a greater degree in more harsh or stressful environments. It is unclear to what extent, if any, the increased reduction in fitness under more harsh conditions in these experiments is the result of an ecological interaction between the organisms and their environments, thus blurring the distinction between the two types of postmating isolation. This topic deserves additional attention both theoretically and empirically. In Chapter three I employed the experimental design suggested in Chapter two to test the ecological basis of hybrid fitness in the sticklebacks. Limnetic backcrosses and Benthic backcrosses, along with both parent species, were transplanted to enclosures in both habitats in Paxton Lake. Growth rates of the backcrosses in these enclosures showed a strong pattern of ecological dependence: in the open water the Limnetic backcross grew at approximately twice the rate of the Benthic backcross, while in the littoral zone the Benthic backcross grew at approximately twice the rate of the Limnetic backcross. In the littoral zone, the rank order 96 of fitnesses of the four cross-types also supported a more general prediction of ecological speciation: that fitness should decline as cross-types become more different morphologically from the native type in a habitat. Results in the open water were less clear, possibly due to the presence of a prey item in the diets of the fish that is commonly associated with littoral habitats, making these enclosures not fully representative of the open water environment and allowing some individuals to exploit a more littoral-like niche. The results of Chapter three also indicated the absence of any strong intrinsic genetic isolation between Limnetics and Benthics. In both habitats, the backcross that was closest in morphology to the native species showed no substantial reduction in growth relative to the parent species native to that habitat. This is notable as past laboratory results, while demonstrating a high fitness for F , and F 2 hybrids, revealed a slight reduction in the fitness of the backcrosses, although sample sizes were small (Hatfield and Schluter 1999). This reduction was most notable in the Benthic backcross and suggests that some intrinsic genetic incompatibilities may have accumulated between the Limnetic and Benthic genomes. However, if intrinsic genetic incompatibilities do exist, my results indicate that their fitness effects are not strong, nor do they appear to increase greatly under a more natural (and possibly more harsh) environment. This implies that the evolution of intrinsic genetic incompatibilities has not been a driving force in the speciation of Limnetics and Benthics. In these sticklebacks, the genetic basis of postmating isolation does not involve hybrid sterility or inviability, but rather the trophic traits that are adaptive to their divergent environments. While my results in Chapter three suggest that hybrids suffer reduced growth because they fall between niches in the environment, our understanding of the ecological dependence of hybrid fitness is not complete. This is because, for practical reasons, individuals in this transplant experiment were not allowed to chose a habitat to exploit but rather were confined to one or the other of the two parental habitats in the lake (open water or littoral zone). It is possible that an alternate habitat exists in the lake, to which an intermediate morphology is best adapted, that would yield higher growth rates for hybrids. The diet of solitary populations that are morphologically similar to hybrids, however, suggests that this is not the case. Their diets are composed, on average, of intermediate frequencies of prey characteristic of the diets 97 of both Benthics and Limnetics (Schluter and McPhail 1992). In addition, individuals tend to specialize, at least over the short term, on one or the other of these habitats, and their choice appears to be related to their trophic morphology (Schluter and McPhail 1992). This suggests that the open water and littoral zone constitute the two main foraging habitats of sticklebacks in the wild. Nevertheless, while evidence suggests the absence of an alternate habitat, hybrids in the wild may benefit from the ability to select which habitat to exploit over the short term, depending on factors such as prey availability. Determining how the fitness of hybrids changes if they are permitted to move between these habitats at will is a key goal for future research. Chapter four addressed the evolution of premating isolation and strongly implicated divergent natural selection and the environment in its evolution by providing one of the first conclusive demonstrations of parallel speciation. The experiment revealed that premating isolation has evolved in parallel among independent populations, in correlation with differences in the environment. Despite their independent evolutionary histories, populations of Limnetics from different lakes were shown to be reproductively compatible; the same was true for Benthics. In contrast, reproductive isolation, known to be present between the species within a lake (Nagel and Schluter 1998), was also shown to be present between the lakes. Such parallel evolution of premating isolation, in correlation with the environment, is only expected under the ecological model of speciation. The results of this study also suggest that natural selection may have directly strengthened premating isolation in sympatry, as revealed by the slightly lower probability of spawning between Limnetics and Benthics from the same lake than from different lakes (Fig. 2.2, comparison C). This possibility receives further support from a previous study I conducted during my Master's degree. Using laboratory mating trials, I investigated how the strength of premating isolation varied between sympatric and allopatric populations to test a prediction of the hypothesis of reinforcement: that interspecific mate discrimination should be stronger in sympatry than in allopatry. Sympatric Benthic females mated less readily with Limnetic than with Benthic males, whereas two different populations of allopatric (i.e, solitary) females, resembling Benthics in morphology, showed no such discrimination (Rundle and Schluter 1998). These results demonstrate the reproductive character displacement of 98 Benthic female mate preferences and are consistent with a direct role for natural selection in strengthening premating isolation in sympatry. When taken together, the results of both these studies suggest that selection has had a dual role corresponding to two stages in the evolution of premating isolation between these sticklebacks (see Fig. 1.1): premating isolation evolved initially as a by-product of divergent natural selection adapting the populations to their different environments during the first, allopatric phase, and was later directly strengthened by natural selection during a second, sympatric phase. Nevertheless, the roles of these two mechanisms in the evolution of the premating isolation are not fully understood. The results in Figure 4.2 suggest that the majority of the premating isolation is explained by the environment (i.e., same vs. different), with only a small component explained by geographic context (i.e., sympatry vs allopatry; Fig. 4.2, comparison D). It is thus tempting to conclude that the majority of the premating isolation has thus evolved as a by-product of adaptation to their alternate environments, with only a minor role for direct selection in sympatry. Alternately however, these same results could represent parallel responses in different lakes to direct selection on premating isolation in sympatry. For example, if reinforcement occurred by strengthening the dependence of mate choice on some key trait (e.g., body size; see below), and this trait evolved in parallel among lakes, parallel reinforcement would result. Such a process, in which premating isolation only evolves in sympatry, is consistent with the lack of premating isolation observed in my earlier experiments between Limnetic males and allopatric, 'Benthic-like' females (Rundle and Schluter 1998). It seems unlikely, however, that reinforcement could occur in parallel among independent lakes unless it acted to strengthen mate preferences that had already evolved in parallel as a by-product of adaptation to the environment. In addition, some premating isolation must have been present at secondary contact to prevent fusion of the two populations because postmating isolation, while present, appears too weak to prevent fusion on its own (Chapter 3; Schluter 1995; Hatfield and Schluter 1999). Determining the relative roles of direct and indirect selection, and how much reproductive isolation evolved during these separate phases, will require further studies of the mate preferences of solitary populations differing in phenotype. Conclusions concerning the relative contribution of the different mechanisms of 99 ecological speciation, and their geographic context, await these data. The parallel evolution of premating isolation among these populations also implies that similar mechanisms of mate choice have evolved repeatedly in similar environments. What these mechanisms are, and the trait or traits that form their proximate basis, are not known, although body size is a strong candidate. Premating isolation between Limnetics and Benthics is dependent on body size: in laboratory mating trials hybridization occurred only between the smallest individuals of the larger species (Benthics) and the largest individuals of the smaller species (Limnetics; Nagel and Schluter 1998). Evidence also suggests that body size differences are adaptive for foraging in their different habitats (Schluter 1993). In addition, similar evidence has demonstrated the importance of body size differences in premating isolation between anadromous and freshwater forms of sockeye salmon (Foote and Larkin 1988). Body size differences, however, may be merely correlated with differences in other traits that have not yet been investigated, so determining the independent contribution of body size will require manipulation of this trait independently of others (e.g., through the use of computer-generated animated 'fish'; McKinnon 1995). Finally, male and female breeding colours also vary among populations, being more divergent between Limnetics and Benthics in some lakes (e.g., Enos Lake) than in others. The contribution of such traits to premating isolation, and the mechanisms causing their divergence, deserve further attention. Finally, Chapter five turned away from the evolution of reproductive isolation to address the ecological interactions between populations that are responsible for generating divergent natural selection. M y experiment built on past work suggesting the importance of between-population competition for resources in their character divergence (Schluter 1994, 2000b; Pritchard and Schluter 2001). By manipulating the density of both insect and fish predators, I tested how predation and competition interacted to affect the strength and form of selection on a target population. Two main results emerged. First, there was a strong interaction between predation and competition. Selection on the target population differed between competition treatments only under conditions of high predation. Second, results also revealed opposing natural selection acting on growth versus survival of target individuals, implying that the interaction between competition and predation is complex. However, 100 survival results from past competition experiments have been highly variable (Schluter 1994, 2000b; Pritchard and Schluter 2001). In my experiment, the loss of one high predation pond left no replicates of this treatment so there exists the alternate possibility that the observed survival differences may have simply been an artifact of other differences between the ponds other than the manipulations. Only further study can address this possibility. If real however, determining the mechanism by which competition and predation interacted to affect selection on the target population is an important goal of future research. Finally, it is important to consider how these results affect our understanding of the evolution of reproductive isolation between Limnetics and Benthics. Under the ecological model of speciation, reproductive isolation can evolve as a result of divergent selection caused by ecological interactions between populations (Chapter 1). The role of ecological interactions in Limnetic-Benthic speciation, however, is not fully understood. Previous work described above demonstrated the pattern of reproductive character displacement of Benthic female mate preferences (Rundle and Schluter 1998). Such a pattern in the evolution of premating isolation could be a by-product of divergent character displacement between sympatric populations in the evolution of present-day Limnetics and Benthics. Alternately, direct selection for increased premating isolation (i.e., reinforcement; Chapter 1) could also produce this result. Conclusions concerning the role of ecological interactions in the evolution of premating isolation between Limnetics and Benthics await further work to determine the relative roles of these two processes. In addition, whether ecological interactions have contributed to the evolution of postmating isolation is an interesting possibility that has not yet been explored. Ecological Speciation: What have we learned? M y studies, as well as the work of others, seeks to understand mechanisms of speciation in nature more generally, using the threespine stickleback as a model system. Given this goal, it is important to step back to ask what lessons the investigation of stickleback speciation has taught us about speciation more generally. First and foremost, the sympatric species-pairs of Limnetic and Benthic sticklebacks 101 provide an excellent case of ecological speciation in nature. From a consideration of my results, as well as those of past studies (e.g., Schluter and McPhail 1992; Nagel and Schluter 1998; Rundle and Schluter 1998; Hatfield and Schluter 1999), it is evident that both premating and postmating isolation have evolved ultimately as a result of divergent natural selection. Such a demonstration is crucial because, as noted in the introduction, there are few direct tests of ecological speciation in nature (Coyne 1992; Schluter 1996a, b; Futuyma 1998; Orr and Smith 1998). Research into the speciation of Limnetic and Benthic threespine sticklebacks is helping to change this. While there are few conclusive demonstrations of ecological speciation in nature, this process may not be rare. This suggestion derives in part from research on the sticklebacks, which has resulted in the recognition of a previously unconsidered form of evidence for ecological speciation: the parallel evolution of reproductive isolation among independently evolved populations (Schluter and Nagel 1995). Limnetic and Benthic sticklebacks provide some of the first conclusive evidence for parallel speciation in nature, but other potential examples have also been recognized. Work with another pair of sticklebacks, the anadromous marine and stream-resident forms, indicates that premating isolation has likely evolved in parallel among independently evolved populations in Japan and western North America (J. McKinnon, pers. comm.). Unlike the recent divergence of the Limnetic-Benthic pairs, these populations have been separated on the order of one million years or more (Orti et al.1994), indicating that reproductive isolation can evolve in a repeatable manner over much longer time scales. In another study, premating isolation was also found to be correlated with the environment, and not with genetic distance, among populations of leaf beetles (Neochlamisus bebbianae) on various host-plants (Funk 1998). In addition, McPeek and Wellborn (1998) demonstrated that reproductive isolation was again correlated with the environment among independently evolved populations of freshwater amphipods (Hyalella azteca). Populations of these amphipods exhibit one of two body size morphs, with the smaller size occurring in the presence of a predatory sunfish. Field studies suggest that these body size differences have evolved as a result of size-based predation (Wellborn 1994), and these differences are not correlated with genetic relatedness (McPeek and Wellborn 1998). Other potential cases of 102 parallel speciation, for which the evidence is not yet conclusive, include anadromous and freshwater sockeye salmon and independently evolved populations of cave amphipods (reviewed in Schluter and Nagel 1995), suggesting that parallel speciation, and thus ecological speciation, many not be a rare phenomenon. Besides providing strong evidence that ecological speciation can occur in nature, my work with the sticklebacks also suggests a general framework for the process of ecological speciation. Under this scenario, the evolution of reproductive isolation has an initial allopatric and a subsequent sympatric phase (Fig. 1.1). Viewing the process of speciation within such a framework is useful as it places the primary focus of research on understanding mechanisms of speciation but also recognizes that the geographic context plays an important role by affecting which mechanisms are possible. While the relative roles of the different mechanisms of ecological speciation remain to be determined for the sticklebacks, results to date strongly imply that some premating isolation must have evolved during each of these phases. A role for direct selection in sympatry is implied by the reproductive character displacement of Benthic female mate preferences (Rundle and Schluter 1998). I also conclude that some premating isolation must have evolved during the allopatric phase as postmating isolation between the species does not appear to be strong enough to prevent their fusion following secondary contact (Chapter 3). In addition, in the absence of an allopatric phase, theoretical studies suggest that disruptive selection must be strong for speciation to occur in sympatry (e.g., Johnson and Gullberg 1998; Kondrashov and Kondrashov 1999; but see Dieckmann and Doebeli 1999). Taken together, the stickleback studies suggest that the by-product (i.e., indirect) mechanism of ecological speciation may commonly produce partial reproductive isolation in allopatry, but that completion of the speciation process may often require a contribution of direct selection during a subsequent, sympatric phase. A similar conclusion has been arrived at in another well-researched case study for speciation in nature: Darwin's Finches in the Galapagos islands. Long-term field studies of these birds has led Grant and Grant (1996) to propose that differences initially built in allopatry are furthered only once sympatry is reestablished. Thus both sticklebacks and finches suggest that speciation may often involve a variety of mechanisms that occur under a diversity 103 of geographic contexts, and that the debate over sympatric vs. allopatric speciation may be too simplistic. Studies in other systems are needed to determine how frequently these two phases, and what mechanisms within each, are important to the speciation process. Turning to the genetics of speciation, work on the sticklebacks may ultimately result in an altered view concerning the evolution of postmating isolation between incipient species. The lack of any strong intrinsic genetic incompatibilities between Limnetic and Benthic sticklebacks is now well established (Chapter 3; Hatfield and Schluter 1999). This suggests that when making inferences about speciation, we must consider ecological mechanisms of postmating isolation as these may evolve first (e.g., Coyne and Orr 1989, 1997). This also suggests that the genetics of speciation may commonly involve the genetics of the traits that are adaptive to the different environments of the incipient species, thus causing ecologically-dependent postmating isolation. In contrast, hybrid sterility and inviability may often arise after speciation is essentially complete and the genetic basis of these traits may not represent the genetic basis of the traits responsible for speciation but may rather represent the genetics of differences that accumulate after speciation (i.e., the genetics of species differences). Determining whether the sticklebacks, in demonstrating strong ecologically-dependent isolation in the absence of any significant intrinsic genetic incompatibilities, are an unusual case or rather represent the norm should be a priority for future speciation research. Other cases similar to the sticklebacks suggest they are not rare (e.g., Grant and Grant 1992; Campbell et al. 1997; Feder 1998). M y pond experiments in Chapter 5 suggest that the study of the ecological mechanisms generating divergent selection needs to consider alternate interactions between populations, in addition to competition for food resources. This is important for our understanding of ecological speciation, in which divergent selection plays a central role, and is also important in furthering our general understanding of the evolution of phenotypic diversity (e.g., character displacement; Taper and Case 1992; Schluter and McPhail 1992; Schluter 2000b). Little consideration has been given to the possibility that reproductive isolation could evolve indirectly in sympatry as a by-product of adaptive divergence of traits in response to interactions between populations. This possibility deserves further investigation in the 104 sticklebacks, as well as in other systems in nature. If ecological character displacement is as prevalent as recent reviews suggest (e.g, Taper and Case 1992; Schluter 2000a, b), this could be an important mechanism contributing to the evolution of reproductive isolation in nature. Finally, from a consideration of all of the above, I finish with two conclusions I have arrived at from the study of speciation in Limnetic and Benthic threespine sticklebacks. First, it has given us valuable insight into the process of speciation and the ecological mechanisms generating it. Whether these insights represent general patterns that require us to alter our traditional views of speciation awaits further work, but I suspect they will. Parallel results from other systems already exist. And the second conclusion is that there is still much remaining that they can tell us. Both of these conclusions foretell a promising future, in which sticklebacks feature prominently, in our attempts to unravel "that mystery of mysteries" (Darwin 1859). 105 LITERATURE CITED Abrams, P .A. 1996. Evolution and the consequences of species introductions and deletions. Ecology 77: 1321-1328. Abrams, P .A. 2000. Character shifts of prey species that share predators. American Naturalist 156 (suppl.): S45-S61. Abramsky, Z . , M . L . Rosenzweig, B. Pinshow, J.S. Brown, B. Kotler, and W . A . Mitchell. 1990. Habitat selection: an experimental field test with two gerbil species. Ecology 71: 2358-2369. Barker, J.S.F. and L . J . E . Karlsson. 1974. Effects of population size and selection intensity on responses to disruptive selection in Drosophila melanogaster. Genetics 78: 715-735. Bell, M . A . and S.A. Foster. 1994. Evolutionary Biology of the Threespine Stickleback. Oxford University Press, Oxford. Bentzen, P. and J.D. McPhail. 1984. Ecology and evolution of sympatric sticklebacks (Gasterosteus): specialization for alternative trophic niches in the Enos Lake species pair. Canadian Journal of Zoology 62: 2280-2286. Blair, W.F . 1955. Mating call and stage of speciation in the Microhyla Olivacea - M . Carolinensis complex. Evolution 9: 469-480. Boag, P.T. and P.R. Grant. 1981. Intense natural selection in a population of Darwin's finches (Geospizinae) in the Galapagos. Science 214: 82-85. Borland, M . 1986. Size-assortative mating in threespine sticklebacks from two sites on the 106 Salmon River, British Columbia. M.Sc. thesis, University of British Columbia, Vancouver, B . C . , Canada. Brown, J .H. and J .C. Munger. 1985 Experimental manipulation of a desert rodent community: food addition and species removal. Ecology 66: 1545-1563. Brown, W . L . andE.O. Wilson. 1956. Character displacement. Systematic Zoology 5: 49-64. Campbell, D.R. , N . M . Waser, and M . E . J . Ackerman. 1997. Analysing pollinator-mediated selection in a plant hybrid zone: hummingbird visitation patterns on three spatial scales. American Naturalist 149: 295-315. Carson, H . L . and A.R. Templeton. 1984. Genetic revolutions in relation to speciation phenomena: the founding of new populations. Annual Review of Ecology and Systematics 15: 97-131. Christie, P. and M.R. MacNair. 1984. Complementary lethal factors in two North American populations of the yellow monkey flower. Journal of Heredity 75: 510-511. Clarke, B . 1975. The contribution of ecological genetics to evolutionary theory: detecting the direct effects of natural selection on particularly polymorphic loci. Genetics 79 (suppl.): 101-113. Coyne, J.A. 1992. Genetics and speciation. Nature 355: 511-515. Coyne, J .A. and H . A . Orr. 1989. Patterns of speciation in Drosophila. Evolution 43: 362-381. Coyne, J.A. and H . A . Orr. 1997. "Patterns of speciation in Drosophila" revisited. Evolution 51: 295-303. 107 Coyne, J.A. and H . A . Orr. 1998. The evolutionary genetics of speciation. Philosophical Transactions of the Royal Society of London, Series B 353: 287-305. Craig, T.P. , J.D. Horner, and J.K. Itami. 1997. Hybridization on the host races of Eurosta solidaginis: implications for sympatric speciation. Evolution 51: 1552-1560. Day, T . , J. Pritchard and D . Schluter. 1994. Ecology and genetics of phenotypic plasticity: a comparison of two sticklebacks. Evolution 48: 1723-1734. de Oliveira, A . K . and A.R. Cordeiro. 1980. Adaptation of Drosophila willistoni experimental populations to extreme p H medium. Heredity 44: 123-130. Dieckmann, U . and M . Doebeli. 1999. On the origin of species by sympatric speciation. Nature 400: 354-357. Dobzhansky, Th. 1940. Speciation as a stage in evolutionary divergence. American Naturalist 74: 312-321. Dobzhansky, Th. 1951. Genetics and the Origin of Species, 3 r d ed. Columbia Univ. Press, N . Y . Dobzhansky, T h . 1970. Genetics of the Evolutionary Process, Columbia Univ. Press, N . Y . Dodd, D . M . B . 1989. Reproductive isolation as a consequence of adaptive divergence in Drosophila pseudoobscura. Evolution 43: 1308-1311. Doebeli, M . 1996. A n explicit model for ecological character displacement. Ecology 77: 510-520. 108 Doebeli, M . and U . Dieckmann. 2000. Evolutionary branching and sympatric speciation caused by different types of ecological interactions. American Naturalist 156 (suppl): S77-S101. Draper, N . and H . Smith. 1981. Applied Regression Analysis, 2 n d ed., Wiley, Chichester, U . K . Ehrlich, P.R. and P .H. Raven. 1969. Differentiation of populations. Science 165:1228-1232. Ehrman, L . , M . M . White and B. Wallace. 1991. A long-term study involving Drosophila melanogaster and toxic media. Pp. 175-209 in M . K . Hetch, B . Wallace, and R.J. Maclntyre, eds., Evolutionary biology, volume 25. Plenum Press, New York. Emms, S.K. and M . L . Arnold. 1997. The effect of habitat on parental and hybrid fitness: transplant experiments with Louisiana irises. Evolution 51: 1112-1119. Endler, J .A. 1977. Geographic Variation, Speciation, and Clines. Princeton University Press, N.J. Endler, J.A. 1986. Natural Selection in the Wild. Princeton University Press, Princeton, N.J. Feder, J .L. 1998. The apple maggot fly, Rhagoletis pomonella: flies in the face of conventional wisdom about speciation. Pp. 130-144 in D.J. Howard and S.H. Berlocher, eds., Endless forms: species and speciation. Oxford University Press, Oxford. Felsenstein, J. 1995. Phylogeny Inference Package, version 3.57c, University of Washington, W A . Filchak, K . E . , J .L. Feder, J.B. Roethele and U . Stolz. 1999. A field test for host-plant dependent selection on larvae of the apple maggot fly, Rhagoletis pomonella. Evolution 53: 109 187-200. Fisher, R . A . 1930. The genetical theory of natural selection. Oxford University Press, Oxford. Fitch, W . M . andE. Margoliash. 1967. Construction of phylogenetic trees. Science 155: 279-284. Foote, C J . and P.A. Larkin. 1988. The role of mate choice in the assortative mating of anadromous and non-anadromous sockeye salmon (Oncorhynchus nerkd). Behavior 106: 43-62. Funk, D.J. 1998. Isolating a role for natural selection in speciation: host adaptation and sexual selection in Neochlamisus bebbianae Leaf Beetles. Evolution 52: 1744-1759. Futuyma, D.J . 1998. Evolutionary Biology. 3 r d ed. Sinauer Associates, Sunderland, M A . Gorbushin, A . M . 1996 The enigma of mud snail shell growth: asymmetrical competition or character displacement? Oikos 77: 85-92. Grant, P.R. and B.R. Grant. 1992. Hybridization of bird species. Science 256: 193-197. Grant, P.R. and B.R. Grant. 1996. Speciation and hybridization in island birds. Philosophical Transactions of the Royal Society of London, Series B 351: 765-772. Gurevitch, J . , J.A. Morrison, and L . V . Hedges. 2000. The interaction between competition and predation: a meta-analysis of field experiments. American Naturalist 155: 435-453. Hatfield, T . 1995. Speciation in sympatric sticklebacks: hybridization, reproductive isolation and the maintenance of diversity. Ph.D. Thesis, University of British Columbia, Vancouver, 110 B . C . , Canada. Hatfield, T . 1997. Genetic divergence in adaptive characters between sympatric species of stickleback. American Naturalist 149: 1009-1029. Hatfield, T . and D . Schluter. 1996. A test for sexual selection on hybrids of two sympatric sticklebacks. Evolution 50: 2429-2434. Hatfield, T . and D . Schluter. 1999. Ecological speciation in sticklebacks: environment-dependent hybrid fitness. Evolution 53: 866-873. Hay, D . E . and J.D. McPhail 1975. Mate selection in three-spine sticklebacks (Gasterosteus). Canadian Journal of Zoology 53: 441-450. Hendry, A .P . , T . Day, and E . B . Taylor. 2001. Population mixing and the adaptive divergence of quantitative characters in discrete populations: a theoretical framework for empirical tests. Evolution 55: 459-466. Higgie, M . , S. Chenoweth, and M . W . Blows. 2000. Natural selection and the reinforcement of mate recognition. Science 290: 519-521. Hoffmann, A . A . and J. Merlia. 1999. Heritable variation and evolution under favourable and unfavourable conditions. T R E E 14: 96-101. Hollocher H . , C . - T . Ting, M . - L . Wu and C.-I. Wu. 1997. Incipient speciation by sexual isolation in Drosophila melanogaster. the genetics of the Zimbabwe race. Genetics 143: 1243-1255. Holt, R .D. 1977. Predation, apparent competition, and the structure of prey communities. I l l Theoretical Population Biology 12:197-229. Howard, D.J. 1993. Reinforcement: origin, dynamics, and fate of an evolutionary hypothesis. Pp. 46-69 in R . G . Harrison, ed., Hybrid zones and the evolutionary process. Oxford University Press, Oxford. Howard, D.J. and S.H. Berlocher. 1998. Endless Forms: Species and Speciation. Oxford University Press, Oxford. Johnson, P.A. and U . Gullberg. 1998. Theory and models of sympatric speciation. Pp. 79-98 in D.J. Howard and S.H. Berlocher, eds., Endless forms: species and speciation. Oxford University Press, Oxford. Kassen, R .D. , D . Schluter, and J.D. McPhail. 1995. Evolutionary history of threespine sticklebacks (Gasterosteus spp.) in British Columbia: insights from a physiological clock. Canadian Journal of Zoology 73: 2154-2158. Kilias, G . , S.N. Alahiotis, and M . Pelecanos. 1980. A multifactorial genetic investigation of speciation theory using Drosophila melanogaster. Evolution 34: 730-737. Kondrashov, A .S . and D . Houle. 1994. Genotype-environment interactions and the estimation of the genomic mutation rate in Drosophila melanogaster. Proceeding of the Royal Society of London, Series B 258: 221-227. Kondrashov, A .S . and F . A . Kondrashov. 1999. Interactions among quantitative traits in the course of sympatric speciation. Nature 400: 351-354. Lack, D . 1947. Darwin's Finches. Cambridge University Press, Cambridge. 112 Littlejohn, M . J . 1981. Reproductive isolation: a critical review. Pp. 298-334 in W.R Atchley and D.S. Woodruff, eds., Evolution and speciation: essays in honor of M . J . D . White. Cambridge University Press, Cambridge. Lynch, M . 1991a. The genetic interpretation of inbreeding depression and outbreeding depression. Evolution 45: 622-629. Lynch, M . 1991b. Methods for the analysis of comparative data in evolutionary biology. Evolution 45: 1065-1080. Lynch, M . and P .E . Jarrell. 1993. A method for calibrating molecular clocks and its application to animal mitochondrial D N A . Genetics 135: 1197-1208. MacNair, M . R . and P. Christie. 1983, Reproductive isolation as a pleiotropic effect of copper tolerance in Mimulus guttatusl Heredity 50: 295-302. Martin, M . M . and J. Harding. 1981 Evidence for the evolution of competition between two species of annual plants. Evolution 35: 975-987. Martins, E .P . and T . F. Hansen. 1997. Phylogenies and the comparative method: A general approach to incorporating phylogenetic information into the analysis of interspecific data. American Naturalist 149: 646-667. Mayr, E . 1942. Systematics and the Origin of Species. Columbia Univ. Press, N Y . Mayr, E . 1963. Animal Species and Evolution. Belknap Press, Cambridge, M A . McKinnon, J.S. 1995. Video mate preferences of female threespine sticklebacks from populations with divergent male coloration. Animal Behavior 50: 1645-1655. 113 McPeek, M . A . and G . A . Wellborn. 1998. Genetic variation and reproductive isolation among phenotypically divergent amphipod populations. Limnology and Oceanography 43: 1162-1169. McPhail, J.D. 1984. Ecology and evolution of sympatric sticklebacks (Gasterosteus): morphological and genetic evidence for a species pair in Enos Lake, British Columbia. Canadian Journal of Zoology 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. Canadian Journal of Zoology 70: 361-369. McPhail, J.D. 1993. Ecology and evolution of sympatric sticklebacks (Gasterosteus): origins of the species pairs. Canadian Journal of Zoology 71: 515-523. McPhail, J.D. 1994. Speciation and the evolution of reproductive isolation in the sticklebacks (Gasterosteus) of south-western British Columbia. Pp. 399-437 in M . A . Bell and S.A. Foster, eds., Evolutionary biology of the threespine stickleback. Oxford University Press, Oxford. McPhail, J.D. and D . E . Hay 1983. Differences in male courtship in freshwater and marine sticklebacks (Gasterosteus aculeatus). Canadian Journal of Zoology 61:292-297. Muller, H.J . 1942. Isolating mechanisms, evolution and temperature. Biological Symposiums 6: 71-125. Nagel, L . and D . Schluter 1998. Body size, natural selection, and speciation in sticklebacks. Evolution 52: 209-218. Nagy. E.S. 1997. Selection for native characters in hybrids between two locally adapted plant 114 subspecies. Evolution 51: 1469-1480. Noor, M . A . 1995. Speciation driven by natural selection in Drosophila. Nature 375: 674-675. Noor, M . A . F . 1997. How often does sympatry affect sexual isolation in Drosophila! American Naturalist 149: 1156-1163. Noor, M . A . F . 1999. Reinforcement and other consequences of sympatry. Heredity 83: SOS-SOS. Orr, M.R . and T . B . Smith. 1998. Ecology and speciation. T R E E 13: 502-506. Orr, H . A . , L . D . Madden, J .A. Coyne, R. Goodwin and R.S. Hawley. 1997. The developmental genetics of hybrid inviability: a mitotic defect in Drosophila hybrids. Genetics 145: 1031-1040. Ortf, G . , M . A . Bell, T . E . Reimchen and A. Meyer. 1994. Global survey or mitochondrial D N A sequences in the threespine stickleback: evidence for recent migrations. Evolution 48: 608-622. Pacala, S.W. and J. Roughgarden. 1985 Population experiments with the Anolis lizards of St. Maarten and St. Eustacius. Ecology 66: 129-141. Pfennig, D . W . and P.J. Murphy. 2000. Character displacement in polyphenic tadpoles. Evolution 54: 1738-1749. Phelan, P .L . and T . C . Baker. 1987. Evolution of male pheromones in moths: reproductive isolation through sexual selection? Science 235: 205-207. 115 Pressley, P .H. 1981. Parental effort and the evolution of nest-guarding tactics in the Threespine Stickleback, Gasterosteus aculeatus L . Evolution 35: 282-295. Price, T . D . , P.R. Grant, H . L . Gibbs, and P.T. Boag. 1984. Recurrent patterns of natural selection in a populations of Darwin's finches. Nature 309: 787-789. Pritchard, J.R. 1998. Competition and character displacement in sticklebacks. Ph.D. thesis, University of British Columbia, Vancouver, B . C . , Canada. Pritchard, J.R. and D . Schluter. 2001. Declining interspecific competition during character displacement: summoning the ghost of competition past. Evolutionary Ecology Research 3: 209-220. Ratcliffe, L . M . and P.R. Grant. 1983. Species recognition in Darwin's finches (Geospiza, Gould). I. Discrimination by morphological cues. Animal Behavior 31: 1139-1153. Reimchen, T . E . 1980. Spine deficiency and polymorphism in a population of Gasterosteus aculeatus: an adaptation to predators? Canadian Journal of Zoology 58: 1232-1244. Reimchen, T . E . 1994. Predators and morphological evolution in threespine sticklebacks. Pp. 240-276 in M . A . Bell and S.A. Foster, eds., Evolutionary biology of the threespine stickleback. Oxford University Press, Oxford. Rice, W.R. and E . E . Hostert. 1993. Laboratory experiments on speciation: what have we learned in 40 years? Evolution 47: 1637-1653. Rice, W.R. and G.W. Salt. 1988. Speciation via disruptive selection on habitat preference: experimental evidence. American Naturalist 131: 911-917. 116 Rice, W.R. and G.W. Salt. 1988. The evolution of reproductive isolation as a correlated character under sympatric conditions: experimental evidence. Evolution 44: 1140-1152. 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. Canadian Journal of Zoology 62: 1813-1818. Robertson, F . 1966a. A test of sexual isolation in Drosophila. Genetical Research 8: 181-187. Robertson, F. 1966b. The ecological genetics of growth in Drosophila. Genetical Research 8: 165-179. Rowland, W.J. 1989a. Mate choice and the supernormality effect in female sticklebacks (Gasterosteus aculeatus). Behavioural Ecology and Sociobiology 24: 433-438. Rowland, W.J. 1989b. The ethological basis of mate choice in male threespine sticklebacks, Gasterosteus aculeatus. Animal Behavior 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 . and M . C . Whitlock. 2001. A genetic interpretation of ecologically dependent isolation. Evolution 55: 198-201. Rundle, H . D . , L . Nagel, J.W. Boughman and D . Schluter 2000. Natural selection and parallel speciation in sympatric sticklebacks. Science 287: 306-308. Saetre, G.-P. , T . Mourn, S. Bures, M . Krai, M . Adamjan and J. Moreno. 1997. A sexually selected character displacement in flycatchers reinforces premating isolation. Nature 387, 589-117 592. Sanderson, S.L. , J.J. Cech and M.R. Patterson. 1991. Fluid dynamics in suspension-feeding blackfish. Science 251: 1346-1348. Schemske, D .W. and H . D . Bradshaw, Jr. 1999. Pollinator preference and the evolution of floral traits in monkeyflowers (Mimulus). Proceedings of the National Academy of Sciences, U S A 96: 11910-11915. Schluter, D . 1993. Adaptive radiation in sticklebacks: size, shape and habitat use efficiency. Ecology 74: 699-709. Schluter, D . 1994. Experimental evidence that competition promotes divergence in adaptive radiation. Science 266: 798-800. Schluter, D . 1995. Adaptive radiation in sticklebacks: trade-offs in feeding performance and growth. Ecology 76: 82-90. Schluter, D . 1996a. Ecological causes of adaptive radiation. American Naturalist 148 (suppl.): S40-S64. Schluter, D . 1996b. Ecological speciation in postglacial fishes. Philosophical Transactions of the Royal Society of London, Series B 351: 807-814. Schluter, D . 1997. Adaptive radiation along genetic lines of least resistance. Evolution 50: 1766-1774. Schluter, D . 1998. Ecological causes of speciation. Pp. 114-129 in D.J . Howard and S.H. Berlocher, eds. Endless forms: species and speciation. Oxford University Press, Oxford. 118 Schluter, D . 2000a. The Ecology of Adaptive Radiation. Oxford University Press, Oxford. Schluter, D . 2000b. Ecological character displacement in adaptive radiation. American Naturalist 156 (suppl.): S4-S16. Schluter, D . and J.D. McPhail 1992. Ecological character displacement and speciation in sticklebacks. American Naturalist 140: 85-108. Schluter, D . and L . M . Nagel 1995. Parallel speciation by natural selection. American Naturalist 146: 292-301. Szafraniec, K . , R . H . Borts and R. Korona. 2001. Environmental stress and mutational load in diploid strains of the yeast Saccharomyces cerevisiae. Proceedings of the National Academy of Sciences, U S A 98: 1107-1112. Taper, M . L . and T.J . Case. Coevolution among competitors. Oxford Surveys in Evolutionary Biology 8: 63-109. Taylor, E . B . and J.D. McPhail. 1999. Evolutionary history of an adaptive radiation in species pairs of threespine sticklebacks Gasterosteus: insights from mitochondrial D N A . Biological Journal of the Linnean Society 66: 271-291. Taylor, E . B . and J.D. McPhail. 2000. Historical contingency and ecological determinism interact to prime speciation in sticklebacks, Gasterosteus. Proceedings of the Royal Society of London, Series B 267: 2375-2385. Taylor, E . B . , J.D. McPhail and D . Schluter. 1997. History of ecological selection in sticklebacks: uniting experimental and phylogenetic approaches. Pp. 511-534 in T . J . Givnish and K.J . Sytsma, eds., Molecular evolution and adaptive radiation. Cambridge University 119 Press, Cambridge. Thoday, J . M . and J.B. Gibson. 1962. Isolation by disruptive selection. Nature 193: 1164-1166. Ting, C . - T . , S.C. Tsaur, M . - L . Wu and C.-I. Wu. 1998. A rapidly evolving homeobox at the site of a hybrid sterility gene. Science 282: 1501-1504. Vamosi, S .M. 2001. The role of predation in the evolution of sympatric stickleback species. Ph.D. thesis, University of British Columbia, Vancouver, B . C . , Canada. Vamosi, S . M . and D . Schluter. 1999. Sexual selection against hybrids between sympatric stickleback species: evidence from a field experiment. Evolution 53: 874-879. Wallace, B . 1982. Drosophila melanogaster populations selected for resistance to NaCl and C u S 0 4 in both allopatry and sympatry. Journal of Heredity 73: 35-42. Wellborn, G . A . 1994. Size-based predation and prey life histories: a comparative study of freshwater amphipod populations. Ecology 75: 2104-2117. Wright, S. 1940. The statistical consequences of Mendelian heredity in relation to speciation. Pp. 161-183 in J.S. Huxley, ed., The new systematics. Clarendon Press, Oxford. Zar, J .H. 1996. Biostatistical Analysis, 3 r d ed. Prentice Hall, NJ. 120 

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:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0090784/manifest

Comment

Related Items