@prefix vivo: . @prefix edm: . @prefix ns0: . @prefix dcterms: . @prefix dc: . @prefix skos: . vivo:departmentOrSchool "Science, Faculty of"@en, "Zoology, Department of"@en ; edm:dataProvider "DSpace"@en ; ns0:degreeCampus "UBCV"@en ; dcterms:creator "Thompson, Claire Elizabeth"@en ; dcterms:issued "2009-01-24T00:00:00"@en, "1995"@en ; vivo:relatedDegree "Master of Science - MSc"@en ; ns0:degreeGrantor "University of British Columbia"@en ; dcterms:description """I examined lake-stream pairs of threespine stickleback {Gasterosteus aculeatus) using molecular techniques with special emphasis on the Misty Lake pair. Analysis of the lake-stream pair in Misty Lake (Keogh River system, northern Vancouver Island) revealed that the forms represent two separate gene pools as opposed to a single gene pool with some complex polymorphism. There were significant differences between the forms in the frequencies of both mitochondrial restriction fragment length polymorphism (RFLP) haplotypes and nuclear sequence types (the D3 region of 28S rRNA). As far as we know, lake-stream pairs only occur in two other drainages. I used the mitochondrial RFLP data to distinguish between two hypotheses that attempt to account for the distribution and origins of the pairs. The first hypothesis postulates a single origin of the pairs followed by dispersal. This hypothesis assumes that initial divergence of forms is allopatric subsequently followed by secondary contact. The second hypothesis postulates independent parallel evolution in at least two of the three drainages where lake-stream pairs occur. Under this hypothesis, divergence of forms could have been either allopatric or parapatric. The mitochondrial RFLP data supports the parallel evolution hypothesis; furthermore, divergence of at least the Misty Lake pair was allopatric. Lastly, unrooted neighbor-joining trees based on the mitochondrial RFLP haplotypes indicate the presence of two very divergent lineages; a Northern lineage and a Southern lineage. Other studies have identified this split, but they have never observed the Northern lineage in populations as far south as Vancouver Island."""@en ; edm:aggregatedCHO "https://circle.library.ubc.ca/rest/handle/2429/3860?expand=metadata"@en ; dcterms:extent "4751223 bytes"@en ; dc:format "application/pdf"@en ; skos:note "ORIGINS OF L A K E - S T R E A M PAIRS OF THREESPLNE S T I C K L E B A C K (Gasterosteus aculeatus) by C L A I R E E L I Z A B E T H THOMPSON B.Sc, The University of British Columbia, 1991 A THESIS SUBMITTED IN PARTIAL F U L F I L L M E N T OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE 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 Unrequired standard, UNIVERSITY OF BRITISH C O L U M B I A July 1995 © Claire Elizabeth Thompson, 1995 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 i granted by the head of my department or by his or her representatives. It is j understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of ^booloj^ The University of British Columbia Vancouver, Canada Date Tu.U| 3ijiqIST DE-6 (2/88) ABSTRACT I examined lake-stream pairs of threespine stickleback {Gasterosteus aculeatus) using molecular techniques with special emphasis on the Misty Lake pair. Analysis of the lake-stream pair in Misty Lake (Keogh River system, northern Vancouver Island) revealed that the forms represent two separate gene pools as opposed to a single gene pool with some complex polymorphism. There were significant differences between the forms in the frequencies of both mitochondrial restriction fragment length polymorphism (RFLP) haplotypes and nuclear sequence types (the D3 region of 28S rRNA). As far as we know, lake-stream pairs only occur in two other drainages. I used the mitochondrial RFLP data to distinguish between two hypotheses that attempt to account for the distribution and origins of the pairs. The first hypothesis postulates a single origin of the pairs followed by dispersal. This hypothesis assumes that initial divergence of forms is allopatric subsequently followed by secondary contact. The second hypothesis postulates independent parallel evolution in at least two of the three drainages where lake-stream pairs occur. Under this hypothesis, divergence of forms could have been either allopatric or parapatric. The mitochondrial RFLP data supports the parallel evolution hypothesis; furthermore, divergence of at least the Misty Lake pair was allopatric. Lastly, unrooted neighbor-joining trees based on the mitochondrial RFLP haplotypes indicate the presence of two very divergent lineages; a Northern lineage and a Southern lineage. Other studies have identified this split, but they have never observed the Northern lineage in populations as far south as Vancouver Island. TABLE OF CONTENTS Abstract i i Table of Contents , i i i List of Tables v List of Figures vi Acknowledgments vii INTRODUCTION 1 M A T E R I A L S A N D METHODS 9 Study Area and Sampling Sites 9 Collection and Storage of Specimens 10 Laboratory Procedures 11 Analysis of Data 16 RESULTS • --19 Mitochondrial RFLP Analysis 19 RFLP Haplotypes 19 Haplotype Variation and Distribution 19 Evolutionary Distances Between Haplotypes 20 Haplotype Diversity Within Populations and Nucleotide Divergence Among Populations Haplotype Phylogenies Haplotype Frequency Comparisons Population Phylogeny Sequence Data Sequence Types Sequence Type Frequency Comparisons 23 Main Conclusions 23 DISCUSSION 24 Two Gene Pools in the Misty Lake-Stream Pair 24 Origins of the Lake-Stream Pairs of Sticklebacks 25 Parallel Evolution in Other Fishes 27 Phylogeographic Patterns 29 Suggestions for Future Work 32 Summary 36 Literature Cited 66 Appendix A 71 iv LIST OF TABLES Table 1. Stickleback forms in the Keogh . 37 Table 2. Dates and numbers of sticklebacks collected at each site 38 Table 3. Pairwise evolutionary distances for the D3 region of 28S rRNA 39 Table 4. Mitochondrial RFLP presence-absence site matrix 40 Table 5. Composite mitochondrial haplotypes 41 Table 6. Haplotype frequencies 42 Table 7. Evolutionary distances between RFLP haplotypes 43 Table 8. Mitochondrial haplotype and nucleotide diversity 44 Table 9. Mitochondrial nucleotide divergence among populations 45 Table 10. Haplotype frequency comparisons 46 Table 11. D3 sequence type frequency comparisons ..47 v LIST OF FIGURES Figure 1. Distribution of the lake-stream pairs... 48 Figure 2. Photo of the lake and stream forms from Misty Lake 50 Figure 3. Map of the Vancouver Island study sites 52 Figure 4. Sequence alignment of the entire D3 region 54 Figure 5. Sequence alignment of the 116 base pair conserved region of the D3 region 55 Figure 6. RFLP fragment patterns 56 Figure 7. Mitochondrial haplotype phylogeny 60 Figure 8. Mitochondrial population phylogeny 62 Figure 9. D3 sequence phylogeny 64 vi Acknowledgments My supervisor, Don McPhail, allowed me to work on a project of my own choosing and he was generous with his time and with his funding. His discussions with me were invaluable in clarifying my thinking on the lake-stream pairs. My committee members; Martin Adamson, Don Moerman, and Rick Taylor were always helpful and willing to answer my questions. Also, Martin Adamson and Tracey Appleton were wonderful in helping me set up the McPhail portion of the molecular systematics lab. I have enjoyed my graduate experience in large part due to the people that I have met and worked with along the way. The graduate and summer students whom I worked with in the Adamson and Taylor labs made the work seem less like work. Sarcastic beer sessions after Carl's 527 classes helped keep things in perspective in that tough first year, and the Friday afternoon ritual of beers and then dinner was just plain fun. Many thanks go to Peter Troffe, Chris Schell, Steve Heard and Joel Sawada for their excellent help and guidance in the field. Additional thanks go to Old Dutch Potato Chips for coming up with Garlic and Onion chips which we discovered on that last rainy collecting trip and which made the trip that much more enjoyable. I am grateful to Steve Heard for editing an earlier version of this thesis and to Gordon Haas for checking in on me throughout my project to make sure that things were going well. I would like to thank Joel Sawada for calming me down when I got frantic about the thesis and for waiting for me to finish it. Finally, I would like to thank Rick Taylor for being a great mentor and friend. vii INTRODUCTION GENERAL BACKGROUND: In this thesis I investigate the phylogenetic relationship of some unusual populations of threespine sticklebacks (Gasterosteus aculeatus). Threespine sticklebacks (hereafter referred to as sticklebacks) are small fish with a wide holarctic distribution that includes marine, brackish and freshwater habitats. Their life-history,patterns are equally diverse, and there are marine-resident, estuary-resident, anadromous, stream-resident and lacustrine populations. Generally, marine and anadromous sticklebacks are viewed as morphologically uniform throughout their wide geographic distribution (Bell and Foster 1994); however, freshwater populations are notoriously variable not only in their morphology but also in their ecology and behaviour. In most cases where differences among stickleback populations in variable traits have been examined, they were found to be adaptive (Hagen and Gilbertson 1972; Moodie and Reimchen 1976; Lavin and McPhail 1986; Bentzen and McPhail 1984). EVOLUTION OF FRESHWATER STICKLEBACK: Many authors (e.g., Hagen 1967; Bell 1976; McPhail and Lindsey 1986; Bell and Foster 1994) have argued that freshwater sticklebacks are derived from either marine or anadromous sticklebacks, and that freshwater has been independently colonized many times after the retreat of the last glaciation. Support for this hypothesis comes from the phylogenetic relationships of sticklebacks: most of their near relatives are marine groups (Bell and Foster 1994). Moreover, freshwater dispersal, the alternative hypothesis to marine invasion, can not account for the presence of freshwater sticklebacks on isolated headlands and islands (McPhail 1994). The only attempt to test this multiple colonization hypothesis is an allozyme study by Withler and McPhail 1 (1985). Their results support the hypothesis, but the authors warn that their study is not a critical test of the hypothesis. Nowhere in their holarctic distribution is the divergence of freshwater forms as striking as along the glaciated west coast of North America (Moodie 1972; Reimchen et al. 1985; Reimchen 1994; McPhail 1994). Here, a remarkable concentration of divergent populations (some of which coexist) has aroused the attention of researchers interested in the process of speciation. If the multiple colonization hypothesis is correct, freshwater populations along the British Columbia coast must have diverged relatively recently, because the last glaciation only ended about 11,000 years ago. Usually, divergent freshwater populations are geographically isolated from one another and it is difficult to determine if these allopatric populations have diverged to, or are close to, the species level (McPhail 1994). Part of the problem is the lack of agreement amongst biologists as to what constitutes a biological species. McPhail (1994) has suggested that if divergent populations of stickleback can coexist despite persistent hybridization and still remain as statistically distinct gene pools, they can be considered biological species. There are, in fact, a number of cases in which divergent populations are sympatric or parapatric (McPhail 1994). These cases represent three basic divergences: parapatric pairs of anadromous and resident-stream sticklebacks; sympatric pairs of lacustrine sticklebacks with divergent foraging modes (benthics and limnetics); and parapatric pairs of lake and stream sticklebacks. In each of these parapatric, or sympatric, situations there is some resource partitioning and at least partial reproductive isolation. Thus, they are considered incipient species, and by studying these populations researchers hope to gain insights into the process of speciation and the mechanisms that maintain separate gene pools when divergent forms come into contact. 2 This thesis focuses on one such case: parapatric pairs of lake and stream stickleback. Parapatric lake-stream pairs are known from three lake systems, each in a separate central coast drainage (Figure 1). One of these systems is Misty Lake and its inlet stream on northern Vancouver Island. The other two lakes, Mayer and Drizzle, are on Graham Island in the Queen Charlotte Archipelago. In Misty Lake, contact between the lake and stream forms of Gasterosteus occurs in a swampy region where the inlet stream enters the lake. In the Misty system, the lake form is large, with a slender body, and the back and sides are black with silver counter-shading below. In contrast, the stream form is smaller, has a stockier body, and the back and sides are mottled brown (Figure 2). In addition, the lake form has shorter jaws, longer spines, and more (and longer) gill rakers than the stream form (Lavin and McPhail, 1993). Strikingly similar morphological differences are found in the lake-stream pairs associated with Mayer and Drizzle lakes (Reimchen et al. 1985; Lavin and McPhail 1993). Interestingly, the morphological differences between parapatric lake and stream forms are in the same direction, and similar to, the differences between allopatric populations of lake and stream sticklebacks. On the Queen Charlotte Islands, Stinson (1983) found evidence for positive assortative mating in the Drizzle lake-stream pair. This suggests that in the Drizzle system the two forms are at least partially reproductively isolated. Reimchen et al. (1985) noted that the Mayer and Drizzle lake-stream pairs are similar morphologically, but they argue that the pairs evolved independently in each watershed. In spite of this contention, however, they present no compelling evidence to support the suggestion. Lavin and McPhail (1993) compared the morphology of the Misty lake-stream pair with the Queen Charlotte pairs (Mayer and Drizzle) and noted some remarkable similarities in both morphology and ecology among the three systems. 3 To determine if the lake and stream forms in the Misty system represent separate gene pools, Lavin and McPhail (1993) compared their morphology, conducted breeding experiments, and examined allozyme frequencies in the two forms. They found significant differences between the forms in all of the morphological traits examined and, also, that these differences were maintained in laboratory-reared fish (i.e., crosses within the lake form produced only lake forms, and crosses within the stream form produced only stream forms). In addition, Lavin and McPhail (1993) crossed the two forms and demonstrated that the morphology of F i hybrids is intermediate. Using this morphological data, they constructed a linear discriminant function to distinguish the two forms and applied it to wild fish. They found that less than 0.4% of the adults in the lake were hybrids, and suggested that there is little, or no, gene flow between the two forms. In contrast to these morphological data, their allozyme survey suggests no divergence between the forms: 25 loci were examined and no significant frequency differences were found (Lavin and McPhail 1993). Thus, the morphological and breeding data imply separate gene pools but the allozyme data suggest a single gene pool. Still, the lack of allozyme frequency differences between the forms is troubling, but this may simply reflect the recent origin of the divergence. Another explanation is that the alleles at the polymorphic protein loci are subject to natural selection. Withler and McPhail (1985) found some evidence for selection on allozyme allele frequencies in their comparison of freshwater and anadromous populations of sticklebacks. Thus, in the case of the Misty lake-stream forms, selection may favour the same allele frequencies in both environments. This question of whether the two Misty sticklebacks represent one, or two, gene pools needs to be resolved before the more interesting question of the origins of the two forms can be examined. Since the evidence to 4 date is conflicting, I have chosen to use independent, molecular characters to try and determine if the two forms in the Misty system represent two gene pools. MODELS FOR THE ORIGINS OF THE LAKE-STREAM PAIRS: In an attempt to explain the morphological similarities among the three lake-stream pairs, and also the limited, disjunct distribution of the pairs, Lavin and McPhail (1993) proposed two models. Because testing these models is the main focus of my thesis, they are outlined in some detail below. The first model postulates a single origin of the lake-stream pairs followed by dispersal to their current disjunct distribution. This model assumes that the lake and stream forms diverged in allopatry and, thus, that contact between the forms is secondary. The observation that allopatric lake and stream stickleback differ in the same traits, and that the differences are in the same direction, as the lake-stream pairs provides some support for the single origin hypothesis; however, their disjunct distribution (the two islands are separated by approximately 400 km of ocean) is potentially a problem for the single origin model. If they evolved before (or during) the last glaciation, they had to survive either on both Vancouver and Graham islands, or they had to disperse post-glacially (the last glaciation in this region ended approximately 11,000 years ago) between the islands. There is evidence that parts of both Graham Island and northern Vancouver Island (the Brooks Peninsula) were unglaciated at the height of the last glaciation (Warner et al. 1982; Heusser 1989), and that the intervening area (Queen Charlotte Sound) that is now submerged, was also ice-free and above sea-level (Josenhans et al. 1993). Thus, the two islands probably were connected by land late in the last glaciation. If so, dispersal between the islands would be possible. 5 The second model postulates the independent evolution of lake-stream pairs at least twice: once on northern Vancouver Island, and once on Graham Island in the Queen Charlotte Archipelago. Under this model, the remarkable similarities among the pairs are explained as parallel, adaptive responses to similar selection regimes. This model applies equally to both allopatric or parapatric divergence. The allopatric situation is identical to the situation in the single origin model, except that the divergences occurr independently in the different drainage systems but, in each case, contact between the two forms is still secondary. In contrast, parapatric divergence requires disruptive selection across the transition zone (ecotone) that separates the lake and stream habitats. In the parapatric case, divergence occurs while the incipient forms are still exchanging genes (primary contact). A major problem for the parallel evolution model is that if the pairs evolved independently, but in parallel, on two islands, why not also in other areas? Over the vast holarctic range of Gasterosteus there must be hundreds of similar ecological situations (lakes with suitable inlet streams). Why, i f selection across transition zones between lakes and streams is sufficient to account for the evolution of these pairs, are there not other sites with similar pairs? Perhaps, on the British Columbia coast, the area between Vancouver Island and the Queen Charlotte Islands is not well enough sampled; however, this argument is less convincing in other areas (e.g., Europe, Japan and the accessible portions of both coasts of North America). The apparent restriction of the lake-stream pairs to the central coast of British Columbia (a tiny part of the range of G. aculeatus), argues that there is something unique about either the sticklebacks, or the geological history, of this region. Distinguishing between the two models proposed to explain the evolution of lake-stream pairs requires distinguishing between similarities due to ancestry and similarities due to parallel responses to similar selection regimes. This is not an easy task when the divergence is relatively 6 recent, and the shared morphological and behavioural traits are demonstrably adaptive responses to shared selection regimes. Perhaps, however, molecular data can discriminate between the hypotheses. In theory, mutations in D N A are effectively selectively neutral (Kimura 1968) and, therefore, shared unique D N A sequences should reflect shared ancestry as opposed to convergence. I sequenced nuclear DNA, and conducted restriction digests on mitochondrial D N A in an attempt to distinguish between the two models of Lavin and McPhail (1993). Most of my data come from the analysis of restriction fragment length polymorphisms (RFLP) of mitochondrial DNA. This approach was chosen because mitochondrial D N A is known to be variable in sticklebacks (O'Reilly et al. 1993; Orti et al. 1994), there is no recombination, and the mode of inheritance is maternal. The last two points are especially important, since they conserve any evidence of multiple colonization. Nuclear data (the D3 region of 28S rRNA) are also included to provide a molecular marker independent of the mitochondrial RFLP data. THE QUESTIONS: The main focus of this study is the Misty lake-stream pair, but a secondary focus is the evolution, and relationships, of the three known lake-stream pairs. Thus, the primary question is, are the lake and stream forms in the Misty system separate gene pools? This question is important because if the two forms in the Misty system are not separate gene pools, but some polymorphism within a single gene pool, there is no point in investigating their relationship to the Queen Charlotte pairs. Since, on this point, the morphological and breeding data of Lavin and McPhail (1993) are in conflict with their allozyme data, my molecular data provide another two independent characters for assessing this question. The null hypothesis, that the lake and the stream forms are a single gene pool, will be rejected if the haplotype frequencies between the two forms are significandy different. 7 The secondary goal of this thesis is to investigate the evolutionary relationships of the lake-stream pairs in three drainages. Lavin and McPhail (1993) suggested two hypotheses: a single origin hypothesis that postulates the evolution of one original pair followed by dispersal to their current disjunct distribution, and a multiple origins hypothesis that postulates at least two separate evolutions (once on Vancouver Island and once on Graham Island). Under the single origin hypothesis, the lake forms in the three drainages should cluster as each other's closest relatives (as inferred from the molecular data) and the stream forms also should cluster together. In contrast, under the multiple origins hypothesis, the lake forms from each of the drainages should be no more closely related to each other than they are to other stickleback populations in the same general vicinity. This also should be true for the stream forms. Finally, tied in with the question of origins of the lake-stream pairs, did the two forms in each of the pairs diverge in allopatry or in parapatry? 8 MATERIALS AND METHODS STUDY AREA AND SAMPLING SITES: The main study area is Misty Lake and its inlet stream (Misty Inlet). These are part of the Keogh River drainage on the northeastern end of Vancouver Island near the city of Port Hardy (Figure 3). The Keogh River drains into the Queen Charlotte Strait. Misty Lake is a small and relatively shallow lake, and the water is stained the colour of strong tea by tannins from the sphagnum moss that is common throughout the Keogh drainage. The Misty inlet stream flows through 500 m of swamp before reaching the southeast corner of the lake. Approximately 900 m above the lake the inlet stream crosses the highway through a culvert. At low water the downstream end of the culvert is about 0.3 m above the stream surface, but at high water the culvert is passable to fish (Lavin and McPhail, 1993). Also of interest are other sites in the Keogh watershed (Figure 3) (Misty Outlet, Long Lake Outlet, Sphagnum Bog, Muir Lake, and Cub Creek), and Beaver Lake in an adjoining river system, the Waukwaas River, which drains into Rupert Inlet. These sites are included for local comparison to the Misty lake-stream pair. Drizzle and Mayer lakes also contain lake-stream pairs and are located on the northeastern end of Graham Island in the Queen Charlotte Archipelago (QCI) (Figure 1). I was only able to obtain samples from Mayer lake and Woodpile Creek (Mayer Inlet) and so I used mitochondrial RFLP data collected by O'Reilly et al (1993) on the Drizzle Lake lake forms (they did not sample the Drizzle Lake stream forms). Additional data from O'Reilly et al (1993) on Rouge Lake were used in the construction of phylogenies. Individuals in Misty inlet were sampled from two different sites (Figure 3); one site is below a culvert and the other is above the culvert. The reason for sampling above and below the culvert was the suspicion that the culvert may act as a barrier to gene flow. In addition there is a small waterfall just above the culvert which probably pre-dates the installation of the culvert and this waterfall may also have acted as a barrier to gene flow. From the sites sampled a total of four different forms of sticklebacks were collected (Table 1). The morphologies of the lake and the stream forms from Mayer Lake and Misty Lake were described briefly in the introduction, and have been described in detail in Lavin and McPhail (1993); Moodie (1972); and Stinson (1983). The two other forms of stickleback present in the Keogh are a nondescript normal form, and an unusually small form (J. D. McPhail and C. Thompson, personal observations). Males of these latter two forms have the typical nuptial colouration of stickleback: red throats with a slight bluish tinge. This is unlike the lake-stream pairs in which breeding males have black throats. COLLECTION AND STORAGE OF SPECIMENS: A l l the lake samples were collected with baited minnow traps left for a few hours or overnight. In the case of Sphagnum Bog, both regular and fine mesh traps were used because apparently the stickleback were often small enough to swim through the regular mesh. Usually, the traps were tossed in from shore and allowed to settle to the bottom of the lake, however, in Beaver Lake and Sphagnum Bog some of the traps were suspended above the bottom. Baited minnow traps were also used to trap stickleback in the streams. In some instances, trapping was not effective and sticklebacks then were collected by pole seining. Stickleback too small to yield reasonable liver samples were returned alive to their original habitat. Dates of collection, and numbers of stickleback collected at each site, are presented in Table 2. Individual sticklebacks were killed in the field and preserved in 95% ethanol (EtOH). In the May 10 1993 samples the bodies were slit from anus to throat and preserved en masse in 95% EtOH. In later collections, the livers (and sometimes testes) were removed and preserved individually in vials of 95% EtOH, and the bodies labeled to identify the fish before preserving them en masse in 95% EtOH. The samples were stored either at room temperature or at 0°C. LABORATORY PROCEDURES: The compositions of solutions are given in Appendix A. DNA Extraction: Samples were extracted using a Proteinase K protocol modified from Sambrook et al. (1989). Approximately 20 mg of liver tissue was added to 400 | i l of Proteinase K solution. In most cases the liver was first minced with forceps or ground using an epoxy pestle. The samples were incubated at 65°C until the proteins were digested (anywhere from between 4 to 12 hours). Proteins, lipids, and other contaminants were removed by extracting at least twice with 1 volume of phenohsevag (1:1) and once with 1 volume of sevag (sevag = 24:1 chloroform:isoamyl alcohol). D N A was precipitated out of solution by the addition of 2.5 volumes of cold 95% Ethanol (EtOH), rinsed twice with 2.5 volumes of 70% EtOH and finally re-suspended in 50 to 100 fil of TE buffer. Quality of the D N A was assessed by running the samples on a minigel and the quantity of D N A was determined with a U V spectrophotometer. Mitochondrial RFLP Protocols: Restriction Digests of DNA: Samples were restricted with eight different restriction enzymes chosen from a subset of enzymes used by O'Reilly et al. (1993) in their RFLP study of Queen 11 Charlotte Island sticklebacks. I chose these enzymes to facilitate comparison between my study and that of O'Reilly et al. (1993). Seven of the enzymes are 6-base enzymes {Bgl I, Eco RI, Hind III, Ksp I, Pst I, Pvu II, and Sac I) and one enzyme was a multiple-base enzyme (Hinc II). Most of the enzymes were purchased from New England Biolabs although, occasionally, Promega and Boehringer Mannheim enzymes were used. Reactions were carried out in a total volume of 25 jxl. Approximately 5 ug of D N A and 15 units of enzyme were used per sample and samples were digested either for 4 hours, or left overnight, at 37°C. Other conditions were as specified by the manufacturers. Electrophoresis: D N A fragments were electrophoresed in horizontal 1 % agarose slab gels (20 cm X 22 cm X 0.6 cm) in a solution of 0.5 X TBE buffer. Gels were run at 43 volts for 18 to 20 hours. Gels were stained with ethidium bromide, destained with distilled water and D N A bands were photographed under U V light. Southern Blotting: D N A in the electrophoresed gels was blotted onto Amersham Hybond-N membranes using a Pharmacia or BioRad vacuum blotter and vacuum pump and the Pharmacia VacuGene X L Vacuum Blotting System blotting protocol 1. After the transfer was complete, the membranes were soaked in 2 X SSC for 5 minutes with gentle shaking. Membranes were left overnight to dry at room temperature. Once dry, membranes were placed in a BioRad GS Genelinker oven and the D N A was crosslinked to the membranes with a 50 mJoule burst of UV. Labeling the Probe: Cloned Gasterosteus aculeatus mitochondrial D N A was generously supplied by Patrick O'Reilly (see O'Reilly 1991 for cloning details). The probe was labeled with non-radioactive Dig d-UTP (Boehringer Mannheim) using a random priming protocol (Feinberg and Vogelstein 1983; Hbltke et al. 1992). To improve the efficiency of the labeling reaction, the probe was digested with Eco RI to linearize the DNA. The digested probe was labeled in a total reaction volume of 100 u l At least 500 ng of probe DNA was added to the labeling reaction. Prehybridization, Hybridization, and Stringency Washes: Membranes were prehybridized individually at 58°C for 1 hour in a Robbins Scientific hybridization oven with 25 ml of pre-warmed Westneat buffer, then hybridized with 100 ul of denatured probe in 10 ml of Westneat buffer and incubated at 58°C for at least 16 hours. Following hybridization, membranes were washed once in pre-warmed (58°C) stringency solution (2 X SSC, 0.1% SDS) for 15 mins at room temperature with shaking, and once in pre-warmed (58°C) stringency solution for 20 mins at 58°C with shaking. Dig DNA Detection: The hybridized probe was detected using the protocol for detecting D N A with chemiluminescence on nylon membranes as described in the Tropix, Inc. Southern Light manual (Cat. No. SL100). Once the detection assay was completed, membranes were wrapped in Saran Wrap and were exposed to X-ray film (Kodak X O M A T - A R ) for 2 to 16 hours. DNA Size Estimation: Fragment lengths were estimated by comparing the relative mobilities of mtDNA fragments to a 1 kilobase ladder with fragments of known size (purchased from BRL). Fragment mobilities were measured directly from the autolumiographs and fragment lengths were estimated using the computer program D N A F R A G (Schaffer and Sederoff 1981; Nash 1991). Mitochondrial RFLP Site Maps: Restriction enzymes cleave double stranded D N A at specific recognition sequences, thus producing a series of fragments. The size and number of fragments 13 depend on where the recognition sites occur in the D N A molecule. A change in the number of restriction sites produces either more or fewer fragments. If a restriction site is lost, then one large fragment is generated from two smaller fragments. Conversely, if there is a gain of a restriction site then one fragment is cut into two smaller fragments. In each case, the sum of the smaller fragments should equal the size of the larger fragment. From comparisons of restriction length patterns, site gains and losses can be inferred. A problem arises when similar restriction length patterns result from different arrangements of restriction sites. In this case, the different restriction sites would not be detected and the inferred restriction site changes would underestimate the actual changes. The solution is to map the sites by performing partial and double digests. I did not perform partial and double digests, but I did use the restriction map generated by O'Reilly et al. (1993, and see O'Reilly 1991). This map, which was constructed based on partial and double digests, includes all the restriction enzymes used in my study. Thus, I was able to use their map as a guide to where restriction sites should be. To aid in correctly inferring the restriction maps, I sampled sticklebacks from one of the O'Reilly et al. (1993) sites: Mayer Lake. In the entire study I observed 18 different restriction fragment patterns, 13 of which were identical to those observed by O'Reilly et al. (1993) and I inferred the remaining five unique fragment patterns. PCR and Sequencing of the D3 Region of28S rRNA: PCR: Amplifications were performed in 100 ul reaction volumes containing 50 mM KC1, 10 mM Tris (pH 9.0), 0.1% Triton X-100, 1.5 mM M g C l 2 , 0.2 mM each of dATP, dCTP, dGTP, and dTTP, 0.5 uM of each primer, approximately 1 ng of DNA, and 1.25 units of Taq polymerase. The primers D3A and D3B (from Nunn (1992) in which he called the primers Ce26S-3A and Ce26S-3B, respectively) have the following sequences: D3A 5' G A C C C G TCT T G A A A C A C G 14 G A 3' and D3B 5' TCG G A A G G A A C C A G C T A C TA 3'. Amplification conditions were as follows: 1 cycle of denatureation at 94°C for 2 mins, annealing at 52°C for 2 mins, and extension at 72°C for 2.5 mins; 30 cycles at 94°C for 1 min, 52°C for 2 mins, and 72°C for 2.5 mins; and finally one cycle at 94°C for 30 sees, 52°C for 2 mins, and 72°C for 10 mins. Only stickleback from Misty Lake, Misty Inlet, Mayer Lake and Mayer Inlet were PCR amplified and sequenced. PCR product was purified using Promega's \"PCR 'Magic' Preps\" kit (later known as \"PCR 'Wizard' Preps\") and the Promega protocol was used. Direct Sequencing of Purified PCR Products: Double-stranded PCR products were sequenced in dideoxy sequencing reactions based on the method of Sanger et al. (1977) using the D3A and D3B primers with the USB Sequenase Version 2.0 D N A Sequencing Kit and [ 3 5S]-dATP. In all cases, both the positive and the negative strands of the amplified D3 region were sequenced. B R L model S2 sequencers were used. Sequencing reactions were run on 6% polyacrylamide/urea gels with 1 X T B E as a running buffer. Gels were run at 60 constant Watts, 1700 Volts, and pre-warmed for at least 30 mins. Short runs lasted for approximately 2.5 hours and long runs lasted for approximately 5 hours. When there were only two hours left in the run, 100 ml of 3 M Sodium Acetate was added to the bottom buffer chamber; this served to compress the sequencing bands at the bottom of the gel. Sequence gels were then transferred onto filter paper and vacuum dried using a Bio Rad Gel Dryer. Dried gels were exposed to X-ray film (Kodak X O M A T - A R ) from 24 hours up to two weeks. Sequence Alignment: Sequences were aligned by eye using the computer program ESEE (The Eyeball Sequence Editor, Version 1.09d. E. Cabot, (c) copyright 1987-1990). The region sequenced was a nuclear gene, the D3 divergent region of 28S ribosomal RNA. The total region 15 amplified was approximately 365 base pairs, however, only a 116 base pair region was alignable (Figure 4) and outside this region it was similar to matching random sequence. Regardless of whether the whole 365 base pair region, or just the 116 base pair region is used, the relationships between the different sequences remained the same. Thus, I chose to use only the 116 base pair region. Even the 116 base pair region is highly variable in sticklebacks (Figure 5) and many sequences have pairwise genetic distances greater than 16% (Table 3). This high level of variability in the D3 region reduces its usefulness in addressing problems of phylogeny. For this reason, only a few individuals were sequenced. My intent, in including the D3 sequence data, is only to determine if they are consistent with the mtDNA RFLP data. ANALYSIS OF DATA: Frequency Comparisons of Mitochondrial Haplotypes and D3 Sequence Types: To determine i f the two Misty Lake forms belong to separate gene pools, I compared the frequencies of the mitochondrial haplotypes (and D3 sequence types) by chi-square contingency tests. Significant differences in frequencies between the two forms would imply rejection of the null hypothesis (that the two forms represent a single gene pool). A problem with molecular data, however, is that the expected frequencies in some cells often are low and this can cause bias in the chi-square value (Zar 1984). To avoid this I evaluated significance of chi-square values by randomization (Roff and Bentzen 1989) using the computer program Monte (REAP, McElroy et al. 1991). The program holds the row and column totals constant (essentially holding the expected frequencies constant) while performing repeated randomizations of the observed values. The probability of exceeding the original chi-square value is then calculated. In addition, for all statistical comparisons, I used the sequential Bonferroni technique as described by Rice (1989) to 16 compensate for escalating Type I error rates that result from performing multiple significance tests. The table-wide alpha level in each case was 0.05. Evolutionary Distances and Phytogeny Construction: Evolutionary distances from sequence data were calculated using the computer package PHYLIP version 3.5p (Felsenstein 1993). Production of presence-absence site matrices from RFLP data were accomplished using the computer package REAP (McElroy et al. 1991). Estimates of evolutionary distances (d values) between the different mitochondrial haplotypes were calculated using the program DSE from the Reap computer package (McElroy et al. 1991: estimates are based on Nei and Tajima 1981; Nei and Miller 1990 eq. 4; Nei and Tajima 1983; Nei 1987 eqs. 5.41, 5.44 & 5.51). Mitochondrial haplotype and nucleotide diversity within populations and nucleotide divergence among populations were calculated using the program D A from the REAP computer package McElroy et al. (1991: estimates are according to Nei 1987 eqs. 8.4, 8.5, 8.12, 10.19, 10.7, 10.2, 10.21; Nei and Tajima 1981). In constructing the mitochondrial haplotype phylogenies I combined my results with those of O'Reilly et al. (1993; and see O'Reilly 1991) which were focused on stickleback from the Queen Charlotte Islands. Combining the two studies required that I modify some of the O'Reilly et al. (1993) composite haplotypes because they included more enzymes in their study. I used PHYLIP version 3.5p (Felsenstein 1993) to construct unrooted Neighbor-Joining trees and bootstrapped Wagner parsimony trees (involving 100 bootstrap resamplings). To determine which of the two hypotheses concerning the evolution of the lake-stream pairs is more likely, I constructed a population phylogeny using Neighbor-Joining based upon estimates of mitochondrial pairwise nucleotide divergence between populations. Under the single origin 17 hypothesis, I expect a strong association between phenotype and mitochondrial haplotype. In other words, for the mitochondrial RFLP data, the lake forms in each of the pairs should cluster together and the inlet forms in each of the pairs also should cluster. If the lake-stream pairs evolved independently, there should be no such association between morphological phenotype and mitochondrial haplotype. 18 RESULTS MITOCHONDRIAL RFLP ANALYSIS: RFLP Haplotypes: Six of the eight restriction enzymes were polymorphic (all but Hind III and Ksp I). Most of the polymorphic enzymes produced two different restriction fragment patterns while the remaining polymorphic enzyme, Hinc II, produced six (Figure 6). The majority of the restriction fragment pattern differences are due to the loss of single restriction sites (for example see Sac I, Figure 6); however, more complex patterns of site losses and gains were observed in Hinc II (Figure 6, Table 4). Here, changes ranged from a minimum of one site change, to a maximum of five site changes. In total 36 sites were surveyed. This represents 1.26% of the entire 16.6 Kb mitochondrial genome. Length variation was observed in some fragments, however, the differences were too small to accurately measure and consistently identify among gels. Haplotype Variation and Distribution: Seven different haplotypes were detected in this study (Table 5), arbitrarily designated as haplotypes 1 through 7. Haplotypes 3 through 7 represent one to three site changes from the most common haplotype, haplotype 1. Haplotype 2 is strikingly different: there are 8 site change differences between haplotype 2 and haplotype 1 (Table 4). Haplotype 1, the most common haplotype, occurred in 97 out of 263 individuals (36.9%, Table 6). Haplotype 1 also is the most geographically widespread haplotype, and was found at six out of the 11 sites sampled (Table 6). Most of the other haplotypes are more restricted geographically: haplotype 2 was found in Long Lake outlet, Misty Lake, Misty outlet, and Misty inlet (both sites); haplotype 3 was found only in Misty Lake, Misty outlet, and Misty inlet (both sites); and haplotype 4 occurred only in Misty Lake and its outlet; haplotype 5 occurred only in 19 Mayer inlet. The most restricted haplotypes were 6 and 7. Each of these haplotypes was observed only once in individuals from Beaver Lake. Evolutionary Distances Between Haplotypes: The largest evolutionary distances (d values) were observed between haplotype 2 and the rest of the haplotypes and these values ranged from 2.6 to 3.5% (Table 7). Haplotype Diversity Within Populations and Nucleotide Divergence Among Populations: Intrapopulation haplotype diversity varied considerably among populations (Table 8). Five populations had haplotype diversity values of zero, while Misty outlet had the highest value (0.4678). In the Keogh samples, the sites closest to Misty Lake had the highest haplotype diversity values (Long Lake outlet, Misty outlet, Misty Lake, and Misty inlet). Nucleotide diversity within populations was also highest in those populations closest to Misty Lake (Table 8). Nucleotide divergence among populations ranged from zero to 0.01983 (Table 9) when only populations from this study are considered, but from zero to 0.02905 when data from Drizzle and Rouge lakes (O'Reilly et al. 1993) are included. Haplotype Phytogenies: Two haplotypes were shared between this study and the O'Reilly et al. (1993) study: haplotype 1 (denoted as haplotype A in O'Reilly et al. 1993) and haplotype 5 (denoted as haplotypes B and H by O'Reilly et al., 1993). In an attempt to determine the relationships between the different mitochondrial haplotypes I constructed an unrooted Neighbor-Joining tree (Figure 7) based upon the evolutionary distances calculated in Table 7. The most striking feature of the tree is the existence of two distinct lineages separated by a large genetic distance (approximately 3%). For reasons which will become apparent in the discussion, I refer to 20 the lineage with haplotypes 2, K, and L as the Northern lineage, and the to other lineage as the Southern lineage. Overlain on the Neighbor-Joining tree (Figure 7) are bootstrap values from a Wagner parsimony analysis involving 100 bootstrap re-samplings. Only bootstrap values greater than 65% are shown. The majority of bootstrap values are quite low; however, the bootstrap value for the Northern/Southern division is 100%. This provides strong support for the topology of this part of the tree. The only other high bootstrap value is the 92% support for the branching pattern of haplotypes 3 and 4. Haplotype Frequency Comparisons: To determine whether the lake and stream forms in the Misty system represented more than one gene pool, comparisons of haplotype frequencies were made using the contingency tests outlined in the methods section. There were significant differences between the two forms (Table lOd). Unfortunately, lake and stream samples were collected in different years (Tables 2 and 6). However, for populations with reasonable samples sizes from both years (Long outiet, Misty outlet, and Cub Creek) there were no significant differences between 1993 and 1994 haplotype frequencies (Table 10a). These results indicate that over the period 1993-1994 haplotype frequencies in the region remained fairly stable and, therefore, I assume that the haplotype frequencies of the lake and the stream populations in the Misty system were also stable over the same time period. In addition, comparisons of haplotype frequencies between Misty Outlet and Misty Lake which were both collected in 1993 give similar results to the comparison between Misty Outlet samples collected in 1994 and Misty Lake samples collected in 1993 (Tables 10b and lOe). These results indicate that the frequency of haplotypes observed in Misty inlet in 1994 probably were not significantiy different from those present in the population in 1993 and, therefore, the Misty Lake (1993) versus Misty inlet (1994) comparison is reasonable. 21 An additional comparison of frequencies was carried out on the Misty Inlet populations. Frequency comparisons of haplotypes from individuals sampled from above and below the culvert indicate that there are no significant differences (Table 10c). Consequently, individuals from above and below the culvert were pooled in all other analyses. Population Phylogeny: If we consider just the lake forms in the population phylogeny (Figure 8), there does not appear to be a strong association between morphological phenotype and mitochondrial haplotype. Still, it could be argued that a certain amount of divergence is to be expected in the time since the pairs dispersed from their point of origin. The data for the inlet forms are less equivocal: the Misty stream population is divergent from most of the other stream populations and certainly does not cluster with the Queen Charlotte Island inlet stream population included in the phylogeny (Mayer inlet). In sum, the RFLP evidence is more consistent with the multiple origins hypothesis than with the single origin hypothesis. SEQUENCE DATA: Sequence Types: Because the D3 sequence is extremely variable (Figures 4 and 5), I decided to group the sequences into different sequence \"types\" based on their similarity to each other. Sequences within a type are only 1 to 4 base pairs different from one another. Sequences between types are between 15 and 20 (13 to 17 %) base pairs different from one another. The names of the sequence types are arbitrary designations. Support for this grouping of sequence types comes from the unrooted Neighbor-Joining tree (Figure 9) using genetic distances calculated based on Kimura's 2-parameter method (Phylip 3.5, 1993 and see table 3). The star-like phylogeny of the D3 region (Figure 9) does not support the large split in lineages that was observed with the 22 mitochondrial RFLP data (Figure 7). Nor is the phylogeny useful in approaching the question of the origins of the lake-stream pairs. The frequency of the sequence types in the different populations, however, can be used to address the question of one or two gene pools in the Misty system. Sequence Type Frequency Comparisons-. There are no shared sequence types between Misty Lake and Misty inlet stream sticklebacks and there are significant differences in the frequency of sequence types between the lake and the stream forms (Table 11). The sample sizes are small, but the D3 sequence data are consistent with the mitochondrial RFLP data and also imply that the lake and inlet forms in the Misty system represent separate gene pools. MAIN CONCLUSIONS: Both the mitochondrial and nuclear data indicate the presence of more than one gene pool in the Misty lake-stream pair. In addition, the mitochondrial RFLP data indicate the presence of two divergent mitochondrial lineages (Figure 7) in the Misty Lake system, and the population phylogeny based on the mitochondrial RFLP data (Figure 8) is consistent with the hypothesis that the pairs evolved at least twice, once on Vancouver Island and once on Graham Island. 23 DISCUSSION TWO GENE POOLS IN THE MISTY LAKE-STREAM PAIR: The Misty lake and stream forms of Gasterosteus not only are morphologically different from one another (Lavin and McPhail, 1993) but also differ in molecular traits. Examination of the frequency distributions of the mitochondrial RFLP and the D3 sequence data indicate that the two forms represent different gene pools (Tables lOd and 11); however, the allozyme evidence obtained by Lavin and McPhail (1993) disagree with the morphological, breeding, and molecular evidence. Of the 25 loci they examined electrophoretically, only two were polymorphic, and the allele frequencies at these loci were the same in both forms of stickleback. Thus, the allozyme data can be interpreted as evidence that the Misty sticklebacks constitute a single gene pool. Other possible explanations for the apparent discrepancy between the allozyme data and the other evidence are that selection may favour similar allozyme allele frequencies in both environments or, as Lavin and McPhail (1993) suggested, the absence of allozyme differences may reflect gene flow between the forms at essentially neutral loci. Lavin and McPhail (1993) argued that the latter scenario is unlikely because of the rarity of hybrids in Misty Lake. In summary, although the allozyme data are equivocal, the morphological and molecular data clearly establish that the Misty lake and stream populations belong to separate gene pools. In the other locations where parapatric lake-stream pairs are known (Drizzle and Mayer lakes on Graham Island), there is also evidence that the lake and stream populations belong to separate gene pools. For example, morphological analysis of the Graham Island pairs (Reimchen et al. 1985) suggests that there is no hybridization between the forms. Moreover, Stinson (1983) provided evidence of positive assortative mating between the Drizzle lake-stream pair, and this implies that the forms are at least partially isolated reproductively. 24 ORIGINS OF THE LAKE-STREAM PAIRS OF STICKLEBACKS: The two hypotheses of Lavin and McPhail (1993) were detailed in the introduction, but briefly, one hypothesis postulates a single, allopatric, origin of the pairs, while the other hypothesis postulates at least two origins. Under the first hypothesis, contact between the pairs is secondary. Under the second hypothesis, the pairs could have evolved either in allopatry, much as in the first hypothesis (allopatric divergence followed by secondary contact), or in parapatry. This allopatric scenario envisages at least two, independent, allopatric divergences followed by contact between the forms. In contrast, the parapatric scenario envisages at least two separate episodes of divergence across the ecotones that exist wherever streams enter lakes. In parapatric divergence, the evolving populations are contiguous and, thus, divergence occurs in spite of gene flow. Endler (1977) demonstrated that parapatric divergence is a theoretical possibility, but he also suggests that patterns of divergence produced in parapatry are not different from those produced in allopatry. Indeed, this latter point has been mentioned by a number of authors (Endler 1977, Barton and Hewitt 1985, Lynch 1989), and this inability to differentiate between the end products of parapatric and allopatric divergence has dogged the concept of parapatric speciation. For example, Barton and Hewitt (1985) suggest that many well studied hybrid zones (the areas of overlap between parapatric species) originated around the end of the last glaciation, and are best explained by secondary contact (i.e., allopatric divergence). This does not exclude parapatric divergence as a speciation mechanism, however, and Barton and Hewitt (1985) suggest that in some cases a correlation between phylogeny and the distribution of the hybridizing (parapatric) groups may permit us to distinguish between primary and secondary contacts. 25 With the parapatric pairs of lake and stream sticklebacks, the results of earlier studies are ambiguous: the morphological differences could be a result of either disruptive selection across the transition zone between lake and stream habitats, or secondary contact between forms that diverged in allopatry (Lavin and McPhail 1993). Equally, the lack of allozyme differences between the Misty forms may indicate a single gene pool, or simply be a reflection of gene flow at loci that are essentially neutral. In contrast, mitochondrial D N A data indicate that the lake-stream pairs on Vancouver and Graham islands evolved independently (Figure 8). In the Misty pair, this evidence suggests genetic divergence in allopatry. The nucleotide divergence between the Misty inlet and lake populations is 1.75% (Table 9). Using a rate of mtDNA sequence divergence calculated for mammals (2% per million years, Brown et al. 1979), the lake and the stream populations diverged approximately 875 thousand years ago. This value probably represents a minimum for sticklebacks since the rate of mitochondrial D N A evolution in cold blooded animals appears to be slower than in higher vertebrates (Thomas and Beckenbach 1989). Even if this estimate of time since divergence is out by an order of magnitude, the data still argue for allopatric, not parapatric, divergence. This is because Misty Lake lies in a glaciated area, and radiocarbon dates of basal organic deposits indicate that the lake is a maximum of 12,000 years old (Walker and Mathewes 1989). Molecular data for the Mayer lake-stream pair are not as clear. Here, the nucleotide divergence between the populations is 0.016% (Table 9). Using the same rate of mtDNA sequence divergence, this represents approximately 8,000 years of divergence. Thus, for Mayer Lake, either a parapatric or allopatric divergence scenario is plausible. In summary, D N A data favour allopatric origins for the Misty lake-stream pair, but are ambiguous when applied to the origins of the Mayer lake-stream pair. Also, although the initial genetic divergence between the two forms 26 in the Misty system clearly occurred in allopatry, this does not mean that disruptive selection across the lake-stream boundary was unimportant in their divergence. In fact, the presence of the two divergent mitochondrial lineages in the Misty stream population suggests that at some point in the past there may have been considerable gene flow between the forms. This may mean that reproductive divergence was incomplete at the time of colonization. Selection across the lake-stream boundary causing further divergence in morphology may also have produced reproductive divergence. The alternative is that the lake and the stream forms were already ecologically and reproductively differentiated when they colonized the lake. This, however, does not explain why the stream population contains two extremely different mitochondrial haplotypes. Thus, the clear morphological and behavioural differences between the forms could have evolved after secondary contact. PARALLEL EVOLUTION IN OTHER FISHES: The notion of multiple origins of similar phenotypes (parallel evolution) is a repeated theme in studies on northern freshwater fish. For example, multiple, independent evolutions of similar phenotypes have been used to explain dwarf and normal lake whitefish (Bernatchez and Dodson 1990), dwarf and normal smelt (Taylor and Bentzen 1993), and sockeye and kokanee salmon (Foote et al. 1989, Taylor et al. 1995). Bernatchez and Dodson (1990) found two distinct mitochondrial lineages within several sympatric populations of dwarf and normal lake whitefish (Coregonus clupeaformis). Interestingly, the phenotypes (dwarf and normal) of the fish were not dependent on mitochondrial haplotype. This implies that the similar phenotypes are due to similar selection regimes and not to shared recent ancestry. Also, in lakes where the two forms are sympatric, the mitochondrial D N A evidence suggests that the divergence was allopatric. Thus, Bernatchez and Dodson (1990, 1991) postulated that the two divergent mitochondrial lineages 27 survived the last glaciation in different glacial refugia, and it appears that sympatric dwarf and normal populations occur only where the two mitochondrial lineages come into secondary contact. Similarly, Taylor and Bentzen (1993) examined populations of dwarf, normal and anadromous smelt (Osmerus mordax) and, again, identified two distinct mitochondrial lineages. The geographic distribution of the two lineages is similar to the distribution of whitefish mitochondrial D N A lineages (Bernatchez and Dodson 1990) and, as in the whitefish study, the two mitochondrial lineages are independent of ecotype. Again, this implies parallel evolution. Unlike the whitefish study, however, Taylor and Bentzen (1993) argue for the sympatric divergence of coexisting dwarf and normal smelts. In the same vein, Foote et al. (1989) demonstrated that sympatric populations of sockeye and kokanee (Oncorhynchus nerka) within a river system are more similar to each other genetically than they are to other sockeye or kokanee populations. Thus, kokanee appear to have evolved, independently, from sockeye salmon numerous times. These authors also argue that sockeye and kokanee diverged in sympatry, although a later study (Taylor et al. 1995) points out that the data do not provide enough resolution to confidently state that the divergence was sympatric. Given the above examples, evidence supporting parallel divergence in parapatric lake and stream sticklebacks is not, in itself, startling. What is intriguing is the fact that the different lake-stream pairs evolved from quite different genetic stocks. The estimated nucleotide divergence between the Misty stream populations and the Mayer stream populations is 1.98% (Table 9). This represents approximately 1 million years since the two populations last shared a common ancestor 28 — a value significantly higher than in any of the examples of parallel evolution given above. It is remarkable that two morphologically and ecologically similar populations can arise from ancestral populations with such divergent mitochondrial genotypes. These results argue that divergent selection pressures in the two habitats (lakes and streams), rather than ancestry, have had the strongest influence on the evolution of these lake-stream pairs. PHYLOGEOGRAPHIC PATTERNS: Most of the molecular genetic divergence between the lake and stream forms in the Misty system can be attributed to the presence of two very divergent mitochondrial haplotype lineages (Figure 7). The divergence between these lineages is approximately 3%: a level of divergence equivalent to that usually observed between different species (e.g., Thomas et al., 1986, reported interspecific sequence divergences of between 2.66% and 7.51% in Pacific salmon, Oncorhynchus spp.). Typically, large intraspecific divergences are attributed to major historical barriers to gene flow (e.g., range fragmentation by glaciation or other extreme climatic changes). This argument is especially convincing when several taxa share the same phylogeographic pattern. Avise and his colleagues (1992) have documented similar phylogenetic patterns among species of freshwater, marine, and catadromous fishes as well as crabs, oysters and sparrows in the southeastern US. In the northeastern US and southeastern Canada, roughly concordant patterns are observed between lake whitefish (Bernatchez and Dodson 1990, 1991), walleye (Billington and Hebert 1988) and smelt (Taylor and Bentzen 1993) and, in each case, the presence of two divergent mitochondrial lineages were attributed to survival in two separate refugia: the Atlantic and Mississippi refugia. On the Pacific coast of North America, recent attention has focused on the phylogeography of Pacific salmon {Oncorhynchus spp.). Chum salmon {Oncorhynchus keta, Taylor et al. 1994, 29 Cronin et al. 1993); pink salmon (O. gorbuscha, Varnavskaya and Beacham 1992), steelhead and rainbow trout (O. mykiss, Okazaki 1984, Taylor 1995); sockeye salmon (O. nerka, Bickham et al. 1995, Taylor et al. 1995), and chinook salmon (O. tshawytscha, Cronin et al. 1993) all exhibit a distinct north-south split. In addition, Bickham et al. (1995) suggest a similar north-south split in Stellar's sea lion. In most of the above mentioned cases, the north-south phylogenetic splits are attributed to allopatric divergence during the last glacial period in different refugia: the Bering refuge (Bering Sea-Yukon River valley) and Cascadian refuge (lower Columbia river). The geographic area sampled in my study is not wide enough to determine if the two mitochondrial lineages of sticklebacks represent colonization from the Beringian and Cascadian refugia. Most of British Columbia, however, was covered in ice about 12,000 years ago, and postglacial recolonization of the province was mainly from these two refugia (McPhail and Lindsey 1986, Lindsey and McPhail 1986). When my results are combined with those from other studies a broader pattern emerges. O'Reilly et al. (1993) sampled the Queen Charlotte Islands and demonstrated that two distinct mitochondrial lineages (termed Marine and Argonaut lineages) are present on the islands. Based on comparison of the restriction site patterns, their two lineages are equivalent to what I call the Southern and the Northern lineages. In the O'Reilly et al. (1993) study, their Argonaut lineage was restricted to the Argonaut Plain, Graham Island. This region of Graham Island probably was a glacial refugium (McPhail and Lindsey 1986), and radiocarbon studies at Cape Ball in the Argonaut plain region (Warner et al. 1982) indicate that the area was ice-free 16,000 years ago (i.e., during the glacial maximum on the adjacent mainland). O'Reilly et a/.(1993) interpreted their results as evidence for the survival of sticklebacks on, or near, the Argonaut Plain during the last glaciation. Additional support for a refugium in this area comes 30 from endemic plants (Ogilvie 1989, Schofield 1989), endemic beetles (Kavanaugh 1989), and mammals (Cowan 1989). Another study (Orti et al. 1994) also focused on mitochondrial D N A and surveyed threespine sticklebacks across almost their entire holarctic range. These authors sequenced part of the mitochondrial cytochrome b gene and discovered two divergent lineages which they termed Japanese and Euro-North American, respectively. The Japanese clade comprises freshwater and marine stickleback from Japan, and freshwater stickleback from British Columbia and Alaska, while the Euro-North American lineage comprises anadromous, freshwater, and marine stickleback from Alaska, British Columbia, California, New York, Quebec, Nova Scotia, Scotland, Sweden, England and France. It is important to recognize that the split between the Northern and Southern lineages is much older than the last (Fraser) glaciation: I roughly estimated the divergence time between the two lineages to be 1.5 million years, while Orti et al. (1994) estimated 0.9-1.3 mya, and O'Reilly et al. (1993) estimated 1.2 mya. Based on samples shared between the Orti et al. (1994) study and the O'Reilly et al. (1993) study, the Japanese and Euro-North American lineages are the same lineages described by O'Reilly et al. (1993) and myself. I have called them the Northern and Southern lineages because the names, Marine and Argonaut lineages, used by O'Reilly et al. (1993) and the names, Japanese and Euro-North American lineages used by Orti et al. (1994), no longer adequately capture the zoogeographic pattern. My names, Northern and Southern lineages, are based on the assumption that, when an adequate geographic area is sampled, the distribution of the two stickleback lineages in the North Pacific region will reflect the zoogeographic pattern found in other euryhaline fish. Certainly, when the data from Orti et al. (1994), O'Reilly et al. (1993) and this study are combined, a pattern similar to that of Pacific salmonids appears. It is puzzling, however, that the Northern lineage is present 31 only in freshwater sticklebacks in North America: 25 North Pacific anadromous, or marine, sticklebacks assayed from four localities by Orti et al. (1994) and O'Reilly et al. (1993) failed to uncover any Northern mitochondrial haplotypes. Orti et al. (1994) suggested two theories to account for this result. First, Orti et al. (1994) suggest that the absence of the Northern lineage haplotypes in North American marine or anadromous sticklebacks is due to inadequate sampling, and that the close affinity of Argonaut Plain, Japanese, and Alaskan freshwater sticklebacks suggests a widespread distribution of this lineage around the North Pacific, and that freshwater populations containing both lineages represent zones of secondary contact. In light of the collection of sticklebacks in the North Pacific 900 km from land (Quinn and Light 1988), they argue that gene flow from the western to the eastern Pacific is plausible. A second theory, which they favour, is that both lineages were abundant in the Northern Pacific region before the beginning of the Fraser glaciation, and that the last advance of the ice sheets displaced the Northern lineage from the eastern Pacific, except in ice-free refugia near Cook Inlet, Alaska, and the Queen Charlotte Islands. Since the Keogh River is relatively close to unglaciated areas in Queen Charlotte Sound that were above sea-level at the height of the last glaciation, either of these theories can accommodate my discovery of Northern lineage haplotypes in freshwater sticklebacks on northern Vancouver Island. SUGGESTIONS FOR FUTURE WORK: The results from this thesis suggest a number of possibilities for future studies. I have demonstrated that the Northern mitochondrial lineage, which has been identified in sticklebacks by previous authors (O'Reilly et al. 1993; Orti et al. 1994), now ranges as far south as the north end of Vancouver Island. Orti et al. (1994) have suggested two hypotheses to explain the distribution of the Northern and Southern lineages in the eastern Pacific (described above). These hypotheses 32 could be tested by conducting a broad survey of both freshwater and marine stickleback populations in the eastern Pacific. Orti et al. (1994) have developed a quick, simple and cost effective assay to discriminate between the two divergent mitochondrial lineages. Use of this assay would make a broad survey quite realistic in terms of cost and time involved in processing samples. It would be of particular interest to see how far south the Northern lineage ranges and to focus on areas in Washington State that may have been unglaciated during the last glaciation [e.g., the Chehalis River system (McPhail and Lindsey 1986)]. The phylogeographic pattern would be further strengthened by looking for congruence using additional molecular markers, particularly nuclear ones. As mentioned earlier, studies of Pacific salmon indicate a similar north-south phylogenetic split as is seen in sticklebacks. These results suggest that the pattern may be general, reflecting major historical events in the region. Although Pacific salmon have been studied extensively, none of the studies mentioned earlier have surveyed individuals from all three areas in the eastern Pacific where the Northern mitochondrial lineage has been observed in sticklebacks. If these places also represent areas of secondary contact between northern and southern lineages in Pacific salmonids, it would further strengthen the interpretation of the stickleback data. Molecular data indicate that the lake and stream forms in Misty Lake represent separate gene pools. There are no published data concerning whether lake and stream forms overlap spatially or temporally during the breeding season in Misty Lake. Collection of this type of data would be helpful in determining how the two gene pools are maintained. In addition, if nuclear markers were used in concert with mitochondrial markers they could help to detect hybrids between lake and stream forms (especially backcrosses) that might not be detectable by morphology. This 33 technique would also indicate the direction of hybridization (e.g., gene flow can be in one direction) and it might indicate how much gene flow is occurring between the forms. Furthermore, this technique would be especially interesting if applied to sticklebacks from the outlet of Misty Lake. In terms of their morphology and mitochondrial haplotype frequencies (Table 6), these fish are somewhat intermediate between the lake and the stream forms (personal observation) and may represent a hybrid zone. Mate choice tests might indicate whether sexual selection is an important factor in maintaining the lake and stream forms in Misty Lake as separate gene pools. Stinson (1983) found evidence of positive assortative mating in the Drizzle Lake pair and, based on the virtual absence of hybrids (Lavin and McPhail 1993), I would expect positive assortative mating between the forms in the Misty system. In fact, the presence of the two divergent mitochondrial lineages in the Misty inlet population suggests that at some point in the past there was substantial interbreeding between forms. As a result, one might expect a cline in the strength of reproductive isolation. For example, selection against hybridizing would, presumably, be strongest close to the lake-stream boundary where the two forms are in contact. Farther from the lake-stream boundary, selection would not be as strong since contact between the forms is less likely. A problem with this scenario is that we do not know how localized sticklebacks in this system are. If lake sticklebacks disperse throughout the lake, regardless of where in the lake they are born, any selection at the lake-stream boundary will be diluted by intra-lacustrine dispersal. With the stream populations, this is not a concern because there is a barrier part way along the stream that may be impassable to upstream movement by sticklebacks. Thus, sticklebacks above the barrier will never come into contact with the lake form. 34 It would also be of interest to perform mate choice tests within forms from different lakes. Assuming that there is positive assortative mating in each of the lakes with parapatric pairs, if given a choice, would the stream form from one lake choose to mate with the stream form from its own system more often than with the stream form from another system? The same studies could be done with the lake forms, but the stream forms would be more interesting in that we know that stream forms from the Misty and Mayer systems represent divergent lineages. These mate choice tests would get at the question of parallel speciation. Schluter and Nagel (1995) define parallel speciation as the repeated independent evolution of the same reproductive isolating mechanism. The idea is as follows: repeated evolutionary changes in similar environments (independent, parallel evolution) suggest that natural selection was the cause of these changes, since it is unlikely that genetic drift would cause concerted shifts in the same direction (Schluter and Nagel 1995, and see references therein). Parallel speciation is a special case of parallel evolution in that adaptive changes to similar environments result in reproductive isolation as a by-product. Parallel speciation emphasizes the role of natural selection in species formation. If stream forms from the Misty system show no preference when given a choice of a stream form from their system and a stream form from the Mayer system, this would indicate the evolution of parallel mate recognition systems. The parapatric lake-stream pairs of sticklebacks represent an excellent system in which to test the parallel speciation hypothesis: since the pairs are present in at least three lakes, there is some replication and we have evidence of the phylogenetic independence of the stream populations. In theory, the rest of the criteria listed by Schluter and Nagel (1995) should be relatively easy to demonstrate. 35 SUMMARY: I have demonstrated the independent, and parallel, evolution of at least the stream members of the lake-stream pairs of stickleback. In addition, I have extended the range of the northern haplotype identified by previous authors (Haglund et al. 1992, O'Reilly et al. 1993, and Orti et al. 1994) to the north end of Vancouver Island. Results from my thesis indicate that there is a wealth of interesting ecological and evolutionary questions still to be addressed about the lake-stream pairs of sticklebacks. 36 Site Form Misty Lk. lake Misty Inlet stream Misty Outlet large stream-like Long Lk. Outlet lake-like Sphagnum Bog small Muir Lk. normal Cub Cr. stream-like Table 1. Forms of stickleback collected at each site in the Keogh. See text for details. Location Collected # Sampled Beaver Lake Oct. '93 30 Cub Creek Oct. '93 12 Cub Creek May '93 5 Cub Creek May '94 13 Long Outlet Oct '93 9 Long Outlet May '94 33 Misty Inlet (above) May '93 3 Misty Inlet (above) May '94 26 Misty Inlet (below) Oct. '93 3 Misty Inlet (below) May '94 24 Misty Lake Feb. '93 4 Misty Lake Mar. '93 4 Misty Lake Oct. '93 9 Misty Lake ?? '93 8 Misty Outlet May '93 3 Misty Outlet Oct. '93 12 Misty Outlet May '94 30 Mayer Lake June '91 3 Mayer Lake Aug. '91 1 Muir Lake May '93 8 Muir Lake Oct. '93 3 Sphagnum Lake Oct. '93 12 Mayer Inlet Summer '93 8 Table 2. Dates and Sites of Collection. Sequence 2 0.0000 Sequence 5 0.0174 0.0000 Sequence 8 0.0176 0.0000 0.0000 Sequence 4 0.1738 0.1528 0.1543 0.0000 Sequence 3 0.1977 0.1760 0.1777 0.1408 0.0000 Sequence 6 0.1795 0.1795 0.1813 0.2074 0.1538 0.0000 Sequence 7 0.1801 0.1801 0.1708 0.2081 0.1543 0.0268 0.0000 SequencelO 0.1675 0.1675 0.1691 0.1952 0.1425 0.0087 0.0177 0.0000 Table 3. Above: Pairwise evolutionary distances of the D3 sequence based on Kimura's two paramater model. Sequence names are arbitrary. Below: Frequencies of each sequence in the different populations sampled. Population Sequence # Misty Lk. 3, 10, 7, 4 Misty In. 2, 5, 8 Mayer Lk. 2, 6 Mayer In. 3 39 Bgl I EcoRl HindUl Kspl Pstl PVMII Sacl HincTL Haplol 010101 11 11 11 1110 01101 1 LI 11110100111001 Haplo2 100101 10 11 11 1111 01101 1 LO 10100100111000 Haplo3 010101 11 11 11 1110 01101 1 LI 11110111111001 Haplo4 010101 11 11 11 1110 11101 1 LI 11110111111001 Haplo5 010101 11 11 11 1110 01101 1 LI 11110100110001 Haplo6 010101 11 11 11 1110 01101 1 LI 11010100111001 Haplo7 010101 11 11 11 1110 01101 1 11 11110100011001 Table 4. Mitochondrial RFLP presence/absence site matrix. This study This study O'Reilly etal. (1993) O'Reilly etal. (1993) Composite Haplo Composite Haplo aaaaaaaa 1 aaaaaaaa A ?aaaaaaa 1 a?a?aaaa A ?aaaa?aa 1 aa??aaaa A a?aa?aaa 1 aaa?aaaa A aaaaaa?a 1 aaaa?a?a 1 aaaaa?aa 1 bbaababd 2 aaaaaaac 3 aa?aaaac 3 aaaaaa?c 3 aBaaaaac 3 aaaaabac 4 aaaaaaae 5 aaaaaaab(b=e) B/H aa?aaaab(b=e) B/H aaaa?aab(b=e) B/H aaaaaaaf 6 aaaaaaab 7 aaaaaaac*(c*=h) C daa?aaaa D aaaaaaae*(e*=j) E aaaaaaaf*(f*=k) F eaaabaaa G represents 2 mitochondria aaaaacab(b=e) I aabaacad*(d*=i) J baaabb*ag(b*=d ;g=d) K baaabb*bg(b*=d;g=d) L baa?bb*bg(b*=d*;g=d) L Table 5. Composite restriction fragment patterns for this study and modified patterns from O'Reilly etal. (1993). Lowercase letters represent restriction fragment patterns produced by the enzymes Bgl I, Eco Rl, Hind III, Ksp I (Sac II, Sst II), Pst I, Pvu II, Sac I (Sst I), and Hinc II. Question marks (?) indicate missing fragment profiles. Underlined letters indicate patterns from O'Reilly et al. (1993) having the same letter as patterns from this study, but referring to a different shared pattern. Letters accompanied by an asterisk (*) indicate enzyme patterns from O'Reilly etal. (1993) which have the same letter as patterns from this study, but which refer to unique patterns which are not shared by the two studies. 41 Population Year H1 H2 H3 H4 H5 H6 H 7 Total Beaver Lk 1993 28 1 1 30 Cub Cr 1993 17 17 CubCr 1994 13 13 Long Out 1993 4 5 9 Long Out 1994 8 25 33 Misty In A 1993 3 3 Misty In A 1994 21 5 26 Misty In B 1993 1 2 3 Misty In B 1994 18 6 24 Misty Lk 1993 1 21 3 25 Misty Out 1993 4 9 2 15 Misty Out 1994 6 21 3 30 Mayer Lk 1991 4 4 Muir Lk 1993 11 11 Sphagnum Lk 1993 12 12 Woodpile Cr 1993 8 8 6, Haplotype frequencies. H1 to H7 represent haplotypes 1 through 7. Haplol Haplo7 0.002906 HapIo6 0.002906 Haplo2 0.025629 Haplo3 0.005685 Haplo4 0.008279 Haplo5 0.002906 0.011743 0.011743 0.018952 0.006116 0.030138 0.030138 0.008937 0.008937 0.011516 0.011516 0.006116 0.006116 0.009215 0.014756 0.009477 0.020264 0.009477 0.020264 0.009428 0.032227 0.034883 0.002611 0.030138 0.008937 0.003672 0.011743 0.003727 0.018952 0.003727 0.018952 0.010238 0.009477 0.003792 0.020264 0.003727 0.011516 Table 7. Estimates of evolutionary distance from restriction site data are reported below the major diagonal. Standarad errors are reported above the major diagonal. Underlined values emphasize the large amounts of evolutionary distance between Haplo 2 and the rest of the haplotypes (from 2.6% to 3.5%). 43 Population Haplotype Diversity Nucleotide Diversity BeaverL 0.1288 +/- 0.05801 0.000388 CubCr 0.0000 +/- 0.00000 0.000000 LongOut 0.3730 +/- 0.05492 0.009708 Misty In 0.3467 +/- 0.04665 0.011286 Misty Lk 0.2841 +/- 0.07711 0.003153 MistyO 0.4678 +/- 0.06402 0.011156 MayerLk 0.4156 +/- 0.09018 0.001268 MuirLk 0.0000 +/- 0.00000 0.000000 Sphagnum 0.0000 +/- 0.00000 0.000000 Mayerln 0.0000 +/- 0.00000 0.000000 DrizzleL 0.0000 +/- 0.00000 0.000000 RougeLk 0.0000 +/- 0.00000 0.000000 Average 0.1680 +/- 0.00310 0.003080+/-0.0000018 Table 8. Haplotype and Nucleotide Diversity within populations +/- SE. (Populations estimates are based on haplotypes from '94 samples except for Beaver Lk., Misty Lk., Muir Lk., and Sphagnum Lk. which were sampled in '93. Mayer Lk., Drizzle Lk., and Rouge Lk. estimates are based on haplotypes presented in O'Reilly etal. (1993) and are from '88, '89, and '88 respectively.) 44 B L 0.000 C 0.000 0.000 L 1.464 1.456 0.000 Miln 1.569 1.560 0.002 0.000 M i L 0.524 0.522 1.890 1.751 0.000 MiO 0.439 0.435 1.171 1.057 0.058 0.000 M L 0.149 0.148 1.692 1.804 0.698 0.627 0.000 Mu 0.000 0.000 1.456 1.560 0.522 0.435 0.148 0.000 S 0.000 0.000 1.456 1.560 0.522 0.435 0.148 0.000 0.000 M i n 0.293 0.291 1.868 1.983 0.852 0.786 0.016 0.291 0.291 0.000 D L 0.000 0.000 1.456 1.560 0.522 0.435 0.148 0.000 0.000 0.291 0.000 R L 2.480 2.471 0.586 0.607 2.885 2.082 2.723 2.471 2.471 2.905 2.471 0.000 Table 9. Nucleotide divergence among populations * 10' 2. Abreviations are as follows: BL = Beaver Lk., C = Cub Cr., L = Long Outlet, Miln = Misty Inlet, MiL = Misty Lk., MiO = Misty Outlet, ML = Mayer Lk., Mu = Muir Lk., S = Sphagnum Bog, Min = Mayer Inlet, DL = Drizzle Lk., and RL = Rouge Lk. (Estimates were based on samples from the same years as those presented in Table 6.) 45 RFLP DATA Tablewise alpha = 0.05 Between years, same site: Population p monte +/- SE rank test alpha Long 0 '93 vs '94 .233+/- .0134 4 NS MiO '93 vs '94 .815 +/- .0123 6 NS Cub Cr. '93 vs '94 NA 7 NS Between sites, same year: Population p monte +/- SE rank test alpha MiO '93 vs MiL '93 .122+/- .0103 2 0.008 NS MilnA '94 vs MilnB '94 .479+/-.0158 5 NS Between sites, different years: Population p monte +/- SE rank test alpha MilnC '94 vs MiL '93 <0.001 1 0.007 * MiO '94 vs MiL '93 .218+/-.0131 3 NS Table 10. Results from multiple chi square contingency tests. The sequential bonferoni adjustment is being used with a tablewise alpha value of 0.05. A star (*) refers to a statistically significant test. Site abreviations are as follows: Long O = Long Outlet, MiO = Misty Outlet, MiL = Misty Lake, Miln A = Misty Inlet above the culvert, Miln B = Misty Inlet below the culvert, Miln C = Misty Inlet above and below combined. 46 D3 SEQUENCE DATA Population p monte MilnC vs MiL <.001 Table 11. Results from a chi square contingency test comparing the frequency of D3 sequence types between Misty Inlet and Misty Lake. A star (*) refers to a statistically significant test. Figure 1. Distribution of the lake-stream pairs. 49 Figure 2. Photograph of the lake and stream form of the Misty pair. Top: Male, stream type (57.4 mm standard length). Bottom: Male, lake type (63.9 mm standard length). Figure 3. Study sites on Vancouver Island. Arrows indicate the areas sampled. M i L 2 1 KNGGAAGGAA CCAGCTACTA GATGGTTCGA TTAGTCTTTC GCCCCTATAC ML34 MiL3 M i L 2 1 ML3 4 M i L 3 M i L 2 1 ML3 4 M i L 3 M i L 2 1 ML3 4 M i L 3 M i L 2 1 ML3 4 M i L 3 TCACGTTTGA CGATCGATTT GCACGTCAGA ATCGCTGCGG TCCTCCACCA C . . A . .CA C T. . . A.T ... C . . A .A C A G T. . GAGTTTCCTC TGGCTTCAAC T. • • Ai • • • TGO • TCTTATCGTG • * * C • G • • • At m m C • • • • * CACACCGAAA GCT.TG.TGC .CAG.GATGC CACGCTCCTG TG TA. T ••••••••* TGCACGCCTT GCTG.AGTGG . C T C . A . . A A CTACGCAAGC ..G.T..G.. • • • • T • • ^ 3 • • I CTCCGCCCCG TA . . . A A T . A T A TCCAACGACC C . G T . A C C A . . A T C . T C T . G ATAGTTCACC ATCTTTCGGG ATGACGCGCC T . . C A A G C A A G . C T . A A T A G ACATCATGGG . . . G A T C C C A CGTGG.GAT. GGCTCCAACG CAAAGACGG. ACTCAACTG. ACGGGCCGGT . A T C A A . T . C . . . C T A T T T A 50 18 14 100 68 64 150 118 114 200 168 164 250 218 214 M i L 2 1 GATGCGACCA CTGCTCTTGT GCCTAGTGCT AGGCGAGGTG CGAGAAACAG 3 00 ML34 A C . A A A G . A . T C T T . A C . T . C A T . T C G C . TTAG.TTTAC . A C C C . . T G A 268 M i L 3 C T . T . . T T A C G C C A . A G G . . T T T . . A . A . A . . C A A . A . C . TATCCCCTT. 264 M i L 2 1 TGGCATCTCA CCACATCTCG AGAACAAGTC GAGACTTCAC TTTCACTGCG 350 ML3 4 CTCGCG.AT. TGCT. 2 83 M i L 3 ACT.GNG.AT 274 M i L 2 1 CCTTTGGGTT TCGATGGACC CATTGACTCG CGTGCATGAT AA 392 Figure 4. Sequence alignment of the D3 region of 28S rRNA (from 5' to 3'). Nucleotide positions indicated by a dot (.) are identical to that on the first line. The 116 conserved base pair region which this study focuses on is represented in bold; the beginning and end of this region are indicated by the arrows. The names Mil_21, Mil_3, and ML34 refer to individual stickleback from Misty Lake, Misty Lake, and Mayer Lake respectively. The conserved regions of MiL21, MiL3, and ML34 are the same as Seq 10, Seq 3, and Seq 5 respectively. 54 Seq 2 AGACGATCGA TTTGCACGTC AGAACCGCTT CGGACTTCCA 40 Seq 5 C . . . 40 Seq 8 N C . . . . 40 Seq 4 C A T . . . . T . . . . A . . A G . C . . . . 40 Seq 3 T A G . . A G . C . . . . 40 Seq 6 T T . . . G G . . . T . C . . . . 40 Seq 7 T GG T . . . . G . . . N . C . . . . 40 Seq 10 T T . . . . G . . . T . C . . . . 40 Seq 2 CCAGAGTTTC CTCTGGCTTC ATCCTGCTCA GGCATAGTTC 80 Seq 5 80 Seq 8 80 Seq 4 T A C . . . A T 80 Seq 3 T A . . . T G C . . . A 80 Seq 6 A . . . A . G . . A 80 Seq 7 A . . . A . G . . A 80 Seq 10 A . . . A . G . . A 80 Seq Seq Seq Seq Seq Seq Seq 2 ACCATCTTTC GGGTCTCAGC GTATGCGCTC TACCTC 116 5 8 116 116 4 A . . . A . T A . . . 116 3 C T . T . . C . . A C T G . . . 116 6 T . T . . . G C A C T G . . . 116 7 T . T . . . G C A C T G . . . 116 10 T . T . . . G C A C T G . . . 116 Figure 5. Sequence alignment of the relatively conserved 116 bp region of the D3 region of 28S rRNA (from 5' to 3'). Nucleotide positions indicated by a dot (.) are identical to that on the first line (Seq 2). 55 Figure 6. Restriction fragment length polymorphisms of threespine stickleback from this study. Shown are the approximate molecular weights of observed restriction fragments in number of kilobase pairs for Bgl I, Eco RI, Hind III, Ksp I, Pst I, Sac I, Pvu II, and Hinc II. Extra bands represent doublets (see Pst I, pattern b and Hinc U, patterns c and d). 56 Bgl I a b 16 6 8.7 7 9 Eco Rl a b Hind III a Ksp\\ a 14 7 8.5 96 5.7 4 1 Pst\\ a b Sac I a b 98 • 6.2 3 65-2 58-87 7.0 5 0 20 1 6 1.2 3 6 3 4 3 1 7 3 2 7 2.9 2 3 2 4 5 2.05 19 1 6 1 3 1 14 10 Figure 7. Mitochondrial haplotype phylogeny. Unrooted Neighbor-Joining tree; percent values are bootstrap values from 100 bootstrap resamplings of a Wagner parsimony analysis (only values greater than 65% are shown). Black ovals represent site gains relative to the most common haplotype, 1/A. Open ovals represent site losses relative to haplotype 1/A. Ovals with the same lower case letter indicate site losses or gains at the same site. Haplotypes with a number code are from this study, while haplotypes with a letter code are from O'Reilly et al. (1993). 60 0 0.5 % Scale 61 Figure 8. Neighbor joining population phylogeny from mitochondrial RFLP data. The lake and stream forms from each pair are indicated by an L and an / symbol respectively. 62 CubCr RougeLk Mi sty In [I LongOut r MistyLk (T MistyO MuirLk DrizzleL (A MayerLk (J L Mayerln Li Sphagnum BeaverLk 0 0.5 % Scale 63 Figure 9. Unrooted neighbor joining phylogeny of the D3 region of 28S rRNA. Sequence types are indicated by circles and the letters A, B, C, and D. 64 65 LITERATURE CITED Avise, J.C. 1992. Molecular population structure and the biogeographic history of a regional fauna: a case history with lessons for conservation biology. Oikos, 63: 62-76. Barton, N.H. , and Hewitt, G . M . 1985. Analysis of hybrid zones. Ann. Rev. Ecol. Syst., 16: 113-148. Bell, M . A . 1976. Evolution of phenotypic diversity in Gasterosteus aculeatus superspecies on the Pacific coast of North America. Systematic Zoology, 25: 211-227. Bell, M.A . , and Foster, S.A. 1994. Introduction to the evolutionary biology of the threespine stickleback. In The evolutionary biology of the threespine stickleback. Edited by M . A . Bell and S.A. Foster. Oxford University Press, Oxford. Bentzen, P., and McPhail, J.D. 1984. Ecology and evolution of sympatric sticklebacks (Gasterosteus): specialization for alternative trophic niches in the Enos Lake species pair. Can. J. Zool., 62: 2280-2286. Bernatchez, L. , and Dodson, J.J. 1991. Phylogeographic structure in mitochondrial D N A of the lake whitefish (Coregonus clupeaformis) and its relation to Pleistocene glaciations. Evol., 45: 1016-1035. Bernatchez, L . , and Dodson, J.J. 1990. Allopatric origin of sympatric populations of lake white fish (Coregonus clupeaformis) as revealed by mitochondrial D N A Restriction analysis. Evol., 44: 1263-1271. Bickham, J.W., Wood, C . C , and Patton, J.C. 1995. Biogeographic implications of cytochrome b sequences and allozymes in sockeye (Oncorhynchus nerka). Journal of Heredity, 86: 140-144. Billington, N . , and Hebert, P.D.N. 1988. Mitochondrial D N A variation in Great Lakes walleye (Stizostedion vitreum) populations. CJFAS, 45: 643-654. Brown, W.M. , George, J., M , and Wilson, A.C. 1979. Rapid evolution of animal mitochondrial DNA. Proc. Natl. Acad. Sci. USA, 76: 1969-1971. Cowan, I.M. 1989. Birds and mammals on the Queen Charlotte Islands. In The Outer Shores. Edited by G.G.E. Scudder and N . Gessler. Queen Charlotte Islands Museum Press, Skidegate, Queen Charlotte, British Columbia. Cracraft, J. 1982. Geographic differentiation, cladistics, and vicariance biogeography: reconstructing the tempo and mode of evolution. Amer. Zool., 22: 411-424. Cronin, M.A . , Spearman, W.J., Wilmot, R.L., Patton, J .C , and Bickham, J.W. 1993. Mitochondrial D N A variation in chinook and chum salmon detected by restriction enzyme analysis of PCR products. CJFAS, 50: 708-715. Endler, J.A. 1977. Geographic Variation, Speciation, and Clines. Monographs in Population Biology. 10. Feinberg, A.P., and Vogelstein, B. 1983. A technique for radiolabelling D N A restriction endonuclease fragments to high specific activity. Analytical Biochemistry, 132: 6-13. 66 Felsenstein, J. 1993. PHYLIP (Phylogeny Inference Package) version 3.5p. Distributed by the author. Department of Genetics, University of Washington, Seattle. Foote, C.J., Wood, C.C., and Withler, R.E. 1989. Biochemical genetic comparison of sockeye salmon and kokanee, the anadromous and nonanadromous forms of Oncorhynchus nerka. CJFAS,46: 149-158. Hagen, D.W. 1967. Isolating mechanisms in threespine sticklebacks (Gasterosteus). Journal of the Fisheries Research Board of Canada, 24: 1637-1692. Hagen, D.W., and Gilbertson, L .G. 1972. Geographic variation and environmental selection in Gasterosteus aculeatus L . in the Pacific Northwest, America. Evol., 26: 32-51. Haglund, T.R., Buth, D.G., and Lawson, R. 1992. Allozyme variation and phylogenetic Relationships of Asian, North American, and European populations of threespine stickleback, Gasterosteus aculeatus. Copeia, 2: 432-442. Heusser, C.J. 1989. North Pacific coastal refugia-the Queen Charlotte Islands in perspective. In The Outer Shores. Edited by G.G.E. Scudder and N . Gessler. Queen Charlotte Islands Museum Press, Skidegate, Queen Charlotte, British Columbia. Holtke, H.J., Sagner, G., Kessler, C , and Schmitz, G. 1992. Sensitive chemiluminescent detection of digoxigenin-labeled nucleic acids: a fast and simple protocol and its applications. BioTechniques, 12: 104-113. Josenhans, H.W., Barrie, J.V., Conway, K.W., Patterson, R.T., Mathewes, R.W., and Woodsworth, G.J. 1993. Surficial geology of the Queen Charlotte Basin: evidence of submerged proglacial lakes at 170 m on the continental shelf of Western Canada. Geological Survey of Canada, Paper 93-1A: 119-127. Kavanaugh, D.H. 1989. The ground-beetle (Coleoptera: Carabidae) fauna of the Queen Charlotte Islands. Its composition, affinities, and origins. In The Outer Shores. Edited by G.G.E. Scudder and N . Gessler. Queen Charlotte Islands Museum Press, Skidegate, Queen Charlotte, British Columbia. Kimura, M . 1968. Evolutionary rate at the molecular level. Nature, 217: 624-626. Lavin, P.A., and McPhail, J.D. 1993. Parapatric lake and stream sticklebacks on northern Vancouver Island: disjunct distribution or parallel evolution? Can. J. Zool., 71: 11-17. Lavin, P.A., and McPhail, J.D. 1986. Adaptive divergence of trophic phenotype among freshwater populations of the threespine stickleback (Gasterosteus aculeatus). CJFAS, 43: 2455-2463. Lindsey, C.C., and McPhail, J.D. 1986. Zoogeography of fishes of the Yukon and Mackenzie basins. In The zoogeography of North American freshwater fishes. Edited by C .H. Horcutt and E.O. Wiley. John Wiley and Sons, Inc. Lynch, J.D. 1989. The gauge of speciation: on the frequencies and modes of speciation. In Speciation and its Consequences. Edited by D. Otte and J.A. Endler. Sinauer Associates, Inc., Sunderland, Massachusetts. McElroy, D., Moran, P., Bermingham, E., and Kornfield, I. 1992. REAP: An integrated environment for the manipulation and phylogenetic analysis of restriction data. Journal of Heredity, 83: 157-158. 67 McPhail, J.D. 1994. Speciation and the evolution of reproductive isolation in the sticklebacks (Gasterosteus) of south-western British Columbia. In The evolutionary biology of the threespine stickleback. Edited by M . A . Bell and S.A. Foster. Oxford University Press, Oxford. McPhail, J.D., and Lindsey, C.C. 1986. Zoogeography of the freshwater fishes of Cascadia (the Columbia system and rivers north to the Stikine). In Zoogeography of North American freshwater fishes. Edited by C.H. Horcutt and E.O. Wiley. John Wiley and Sons, Inc. Moodie, G.E.E. 1972. Morphology, life history, and ecology of an unusual stickleback (Gasterosteus aculeatus) in the Queen Charlotte Islands, Canada. Can. J. Zool., 50: 721-732. Moodie, G.E.E., and Reimchen, T.E. 1976. Phenetic variation and habitat differences in Gasterosteus populations of the Queen Charlotte Islands. Syst. Zool., 25: 49-61. Nei, M . 1987. Molecular Evolutionary Genetics. Columbia University Press, New York. Nei, M . and Miller, J.C. 1990. A simple method for estimating average number of nucleotide substitutions within and between populations from restriction data. Genetics, 125: 873-879. Nei, M . , and Tajima, F. 1983. Maximum likelihood estimation of the number of nucleotide substitutions from restriction site data. Genetics, 105: 207-217. Nei, M . , and Tajima, F. 1981. D N A polymorphism detectable by restriction endonucleases. Genetics, 97: 145-163. Nunn, G.B. 1992. An investigation of evolutionary patterns among nematodes based upon D N A sequences. Ph.D. Dissertation, University of Nottingham, Nottingham England. Ogilvie, R.T. 1989. Disjunct Vascular Flora of Nothwestern Vancouver Island in Relation to Queen Charlotte Islands' Endemism and Pacific Coast Refugia. In The Outer Shores. Edited by G.G.E. Scudder and N . Gessler. Queen Charlotte Islands Museum Press, Queen Charlotte City, BC, Canada. Okazaki, T. 1984. Genetic divergence and its zoogeographic implications in closely related species Salmo gairdneri and Salmo mykiss. Japanese Journal of Ichthyology, 31: 279-311. O'Reilly, P. 1991. The phylogeny and distribution of mitochondrial D N A in threespine stickleback (Gasterosteus aculeatus) on the Queen Charlotte Islands: evidence for Pleistocene glacial refugia. M.Sc. Thesis, Department of Zoology, University of Alberta, Edmonton. O'Reilly, P.O., Reimchen, T.E., Beech, R., and Strobeck, C. 1993. Mitochondrial D N A in Gasterosteus and Pleistocene Glacial Refugium on the Queen Charlotte Islands, British Columbia. Evol., 47: 678-684. Orti, G., Bell, M.A. , Reimchen, T.E., and Meyer, A . 1994. Global survey of mitochondrial D N A sequences in the threespine stickleback: evidence for recent migrations. Evol., 48: 608-622. 68 Quinn, T.P., and Light, J.T. 1989. Occurrence of threespine sticklebacks (Gasterosteus aculeatus) in the open North Pacific Ocean: migration or drift? Can. J. Zool., 67: 2850-2852. Reimchen, T.E. 1994. Predators and morphological evolution in threespine stickleback. In The Evolutionary Biology of the Threespine Stickleback. Edited by M . A . Bell and S.A. Foster. Oxford University Press, Oxford. Reimchen, T.E., Stinson, E .M. , and Nelson, J.S. 1985. Multivariate differentiation of parapatric and allopatric populations of threespine stickleback in the Sangan River watershed, Queen Charlotte Islands. Can. J. Zool., 63: 2944-2951. Rice, W.R. 1989. Analyzing tables of statistical tests. Evol., 43: 223-225. Roff, D.A., and Bentzen, P. 1989. The statistical analysis of mitochondrial D N A polymorphisms: chi-square %2 and the problem of small samples. Molecular Biol. Evol., 6: 539-545. Sambrook, J., Fritsch, E.F., and Maniatis, T. 1989. Molecular Cloning a Laboratory Manual. Cold Spring Harbour Laboratory Press. Sanger, F., Nicklen, S., and Coulson, A.R. 1977. DNA sequencing with chain-terminating inhibitors. Proc. Natl. Acad. Sci. USA, 74: 5463-5467. Schaffer, H.E., and Sederoff, R.R. 1981. Improved estimation of D N A fragment lengths from agarose gels. Analytical Biochemistry, 115: 113-122. Schluter, D., and Nagel, L . M . 1995. Parallel speciation by natural selection. Am. Nat., in press. Schofield, W.B. 1989. Structure and affinities of the bryoflora of the Queen Charlotte Islands. In The Outer Shores. Edited by G.G.E. Scudder and N . Gessler. Queen Charlotte Islands Museum Press, Skidegate, Queen Charlotte Island, British Columbia. Stinson, E . M . 1983. Threespine sticklebacks (Gasterosteus aculeatus) in Drizzle Lake and its inlet, Queen Charlotte Islands: ecological and behavioural relationships and their relevance to reproductive isolation. M.Sc. Thesis, Department of Zoology, University of Alberta, Edmonton. Taylor, E.B. 1995. Genetic variation at minisatellite D N A loci among North Pacific populations of steelhead and rainbow trout (Oncorhynchus mykiss). J. Heredity, in press. Taylor, E.B., Foote, C.J., and Wood, C.C. 1995. Molecular genetic evidence for parallel life history evolution within a Pacific salmon (sockeye salmon and kokanee, Oncorhynchus nerka). Evol., 49, in press. Taylor, E.B., Beacham, T.D., and Kaeriyama, M . 1994. Population structure and identification of North Pacific Ocean chum salmon (Oncorhynchus keta) revealed by an analysis of minisatellite D N A variation. CJFAS, 51: 1430-1442. Taylor, E.B., and Bentzen, P. 1993. Evidence for multiple origins and sympatric divergence of trophic ecotypes of smelt (Osmerus) in northeastern North America^ Evol., 47: 813-832. Thomas, W.K., and Beckenbach, A.T. 1989. Variation in salmonid mitochondrial DNA: evolutionary constraints and mechanisms of substitution. J. Moi. Evol., 29: 233-245. 69 Thomas, W.K., Withler, R.E., and Beckenbach, A T . 1986. Mitochondrial D N A analysis of Pacific salmonid evolution. Can. J. Zool., 64: 1058-1064. Varnavskaya, N.V. , and Beacham, T.D. 1992. Biochemical genetic variation in odd-year pink salmon (Oncorhynchus gorbuscha) from Kamchatka. Can. J. Zool., 70: 2115-2120. Walker, I.R., and Mathewes, R.W. 1989. Early postglacial chironomid succession in southwestern British Columbia, Canada, and its paleoenvironmental significance. J. Paleolimnol., 2: 1-14. Warner, B.G. , Mathewes, R.W., and Clague, J.J. 1982. Ice-free conditions on the Queen Charlotte Islands, British Columbia, at the height of late Wisconsin Glaciation. Science, 218: 675-677. Withler, R.E., and McPhail, J.D. 1985. Genetic variability in freshwater and anadromous sticklebacks (Gasterosteus aculeatus) of southern British Columbia. Can. J. Zool., 63: 528-533. Zar, J.H. 1984. Biostatistical analysis. Prentice-Hall, Inc., Englewood Cliffs, New Jersey. 70 APPENDIX A SOLUTIONS USED: TE: lOmM Tr i sCl (pH 8.0) and ImM EDTA (pH 8.0) 0.5 X TBE: 0.045M Tris-Borate, and 0.001M EDTA. 20 X SSC (stock solution): 175.3 g NaCl, and 88.2g NaCitrate in 1L total volume. Westneat Buffer: 1% BSA, 7% SDS, ImM EDTA, and 0.263M Na 2 HP0 4 . 6% polyacrylamide/urea gel: 15 ml 40% (w/v) acrylamide: bisacrylamide (19:1) 50 g Urea 10 ml 10 X T B E 35 ml H 2 0 Compositions of the other solutions I used are described in the accompanying references in Materials and Methods section. "@en ; edm:hasType "Thesis/Dissertation"@en ; vivo:dateIssued "1995-05"@en ; edm:isShownAt "10.14288/1.0086855"@en ; dcterms:language "eng"@en ; ns0:degreeDiscipline "Zoology"@en ; edm:provider "Vancouver : University of British Columbia Library"@en ; dcterms:publisher "University of British Columbia"@en ; dcterms:rights "For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use."@en ; ns0:scholarLevel "Graduate"@en ; dcterms:title "Origins of lake-stream pairs of threespine stickleback (Gasterosteus aculeatus)"@en ; dcterms:type "Text"@en ; ns0:identifierURI "http://hdl.handle.net/2429/3860"@en .