UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The biogeography and conservation of rainbow trout, Oncorhynchus mykiss McCusker, Megan R. 2000

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

Item Metadata


831-ubc_2000-0123.pdf [ 5.51MB ]
JSON: 831-1.0089407.json
JSON-LD: 831-1.0089407-ld.json
RDF/XML (Pretty): 831-1.0089407-rdf.xml
RDF/JSON: 831-1.0089407-rdf.json
Turtle: 831-1.0089407-turtle.txt
N-Triples: 831-1.0089407-rdf-ntriples.txt
Original Record: 831-1.0089407-source.json
Full Text

Full Text

The Biogeography and Conservation of Rainbow Trout, Oncorhynchus mykiss by Megan R. McCusker B . A . Smith College, 1994 A THESIS S U B M I T T E D I N P A R T I A L F U L F I L L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F M A S T E R O F S C I E N C E in T H E F A C U L T Y O F G R A D U A T E S T U D I E S (Department of Zoology) We accept this thesis as conforming to the required standard T H E U N I V E R S I T Y O F B R I T I S H C O L U M B I A © Megan R. McCusker, 1999 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission Department of ^PO|Q^ The University of British Columbia Vancouver, Canada Date A/ov^^ IrTj H f f ABSTRACT Northwestern North America has been repeatedly glaciated over the past million years, with the most recent glaciation occurring between 60,000 and 10,000 years ago. Where species survived glaciation and what dispersal routes they used during recolonization most likely had a profound effect on their intraspecific genetic variation. In this study, molecular techniques were used to investigate biogeographical, taxonomic and conservation issues in the rainbow trout, Oncorhynchus mykiss. Allozyme data were consolidated from the literature to assess relationships of rainbow trout throughout its range, and a divergence between coastal and inland populations from California to Kamchatka in eastern Siberia was supported. For greater detail, a mitochondrial D N A analysis of restriction fragment length polymorphism was done, focusing mainly on British Columbia and the northern distribution of the species. Two phylogenetically distinct mitochondrial lineages were found with an average of 1.03% sequence divergence, with overlapping distributions that included both coastal and inland populations. Diversity and distributional analyses strongly suggested that both a coastal and an inland refuge were used during the last glaciation with extensive post-glacial gene flow and secondary contact. The Queen Charlotte Islands and the Columbia River were most strongly supported as refuge locations. A n additional refuge may have existed in Beringia, but post-glacial dispersal from Beringia was most likely quite limited. Sequencing analysis of mitochondrial haplotypes revealed higher diversity in California than in the northern part of the species range, indicating a more ancient presence of rainbow trout in California relative to the north. The phylogeographic divergence in British Columbia among coastal and inland groups predates adaptive variation in the species as indicated by two life history characters in this i i analysis. Genetic variation resulting from historical isolation, therefore, warrants high conservation priority. However, due to the degree of secondary contact between these groups post-glacially, subspecies designations of coastal and redband (inland) trout were not supported. Table of Contents A B S T R A C T ii Table of Contents iv List of Figures vi List of Tables vii Acknowledgements viii Ch. 1 General Introduction 1 Molecular genetics and evolutionary biology 1 Historical biogeography 1 Taxonomy and molecular systematics 2 Phylogeny and adaptive evolution 3 Phylogeny and conservation 3 Oncorhynchus mykiss: distribution and subspecies 4 Putative refugia and recolonization pathways from the last glaciation 9 Adaptive variation in Oncorhynchus mykiss 10 Conservation of Oncorhynchus mykiss 11 Organization of thesis 12 Ch.2 Allozyme Variation in Oncorhynchus mykiss 13 I N T R O D U C T I O N 13 Specific aims 13 M E T H O D S 14 R E S U L T S 17 D I S C U S S I O N 22 Ch.3 Mitochondrial D N A R F L P variation 26 I N T R O D U C T I O N 26 Specific aims 27 M E T H O D S 27 Sample collection 27 Genetic analysis 28 Phylogenetic analysis 31 Genetic diversity and hierarchical structure among watersheds 32 Historical demography 33 Taxonomy 37 Adaptive evolution 38 R E S U L T S 39 Genetic variation and relationships among haplotypes 39 Genetic diversity and cluster analysis among regions 46 Historical demography 55 Taxonomy 63 Adaptive evolution 64 D I S C U S S I O N . . . 70 Glacial history 70 Partitioning of genetic variation 74 Historical demography 75 iv Taxonomy 78 Phenotypic evolution 78 Conservation recommendations 79 Ch.4 Nuclear D N A Variation and M t D N A Sequencing 80 I N T R O D U C T I O N 80 M E T H O D S 81 Nuclear variation 81 Mitochondrial variation 82 R E S U L T S ..85 Nuclear variation : 85 Mitochondrial sequencing 85 D I S C U S S I O N 98 Ch.5 General Discussion and Synthesis 101 Inconsistencies in the data for O. mykiss 101 Impact of introductions or stocking 103 Comparative phylogeography among Pacific salmonids 104 Demographics: phylogeography, vicariance and dispersal 105 Phylogeography and taxonomy 107 Phylogeography and conservation 108 R E F E R E N C E S 110 v List of Figures Figure 1. The ranges of two putative subspecies in British Columbia and Alaska 7 Figure lb. The ranges of two putative subspecies in the western U.S 8 Figure 2. Mean frequencies of LDH-100, SOD-100 and MDH-100 for each region 16 Figure 3a. Allozyme cluster analysis of genetic distances 19 Figure 3b. Allozyme cluster analysis based on a maximum likelihood 20 Figure 4. Composite haplotype sample sites in B.C 29 Figure 5. Diagnostic sample sites in B.C 30 Figure 6. Number of each composite haplotype found in mtDNA samples 42 Figure 7. Distance-based phylogeny of mtDNA RFLP composite haplotypes 43 Figure 8a. Consensus distance-based phylogeny of mtDNA composite haplotypes 44 Figure 8b. Consensus parsimony-based phylogeny of mtDNA composite haplotypes 45 Figure 9. Distributions of mtDNA RFLP clade A and clade B haplotypes in British Columbia... 46 Figure 10. Distributions of mtDNA RFLP clade A and clade B haplotypes in the North Pacific..48 Figure 11. Diversity measures of O. mykiss mtDNA RFLP haplotypes 51 Figure 12. Population cluster analysis among watersheds using Nucleodiv 54 Figure 13. Nested network of composite haplotypes 57 Figure 14. Geographical summary of the significant results from the nested clade analysis.... 58 Figure 15. Coastal and redband putative subspecies mapped onto a population cluster analysis. 65 Figure 16a. Phenotypic and life history characters mapped on a mtDNA distance tree 68 Figure 16b. Phenotypic and life history characters mapped on a population cluster analysis 69 Figure 17. GH2D intron genotypes found in watersheds in British Columbia 86 Figure 18. Phylogeny of D-loop based on genetic distance 91 Figure 19a. Consensus distance-based phylogeny of the D-loop sequence 92 Figure 19b. Consensus parsimony-based phylogeny of the D-loop sequence 93 List of Tables Table 1. Stream populations included in each region for allozyme analysis 15 Table 2. Allozyme genetic distances among various putative rainbow trout subspecies groups. ..21 Table 3. Allozyme frequencies of various cutthroat and rainbow trout putative subspecies 25 Table 4. Predictions of the nested clade analysis 36 Table 5. Presence-absence matrix of restriction sites found for particular restriction enzymes.... 40 Table 6. The composite haplotypes found in O. mykiss samples 41 Table 7. Clade A and Clade B composite and diagnostic haplotypes within various regions 49 Table 8. Composite haplotypes of O. mykiss mtDNA found within watershed groups 53 Table 9. Partitioning of genetic variation within British Columbia 54 Table 10a. Nested analysis results for clade A 60 Table 10b. Nested analysis results for clade B 61 Table 11. A M O V A analysis for assessing the evolution of anadromy (or non-anadromy) 66 Table 12. A M O V A analysis for assessing the evolution of run-timing in steelhead 67 Table 13. GHII introns 2D samples used 83 Table 14. Haplotypes and samples chosen for D-Loop sequencing 88 Table 15. Nucleotide sequence of the right domain of the D-loop in rainbow trout 89 Table 16. Sequence haplotypes from the D-loop of cutthroat and rainbow trout 90 Table 17. ND1 samples used 95 Table 18. Sequence haplotypes from the ND1 region of cutthroat and rainbow trout 96 Table 19. Genetic divergence comparisons among different parts of the genome 97 vii Acknowledgements First and foremost, I would like to thank my advisor, Rick Taylor, for providing the support and enthusiasm to see this project through. A special thank you goes to everyone in the lab who provided more help and support than I could have expected. M y committee members, Sally Otto, Eric Parkinson, and J.D. McPhail , provided valuable input throughout the project. I want to thank the many people involved with field collections, especially regional fisheries biologists who were very receptive to my questions and requests. A special thank you goes to Sue Pollard for organizing most of the sample collection, to Gordon Haas for in many ways, and to Dave O'Brien who helped with sampling in the field and who helped with the Temple program more than he admits. Finally, I want to thank Alistair Blachford for his willingness to help with the Temple program, and Mike Stamford and Steve Latham for thorough readings of my final draft. Ch.l General Introduction Molecular genetics and evolutionary biology Phylogenetics, or, the process of inferring genealogical relationships among organisms, has diverse applications in evolutionary biology. I will describe some of these applications and how they can be used to address biogeographic and evolutionary questions in the rainbow trout, Oncorhynchus mykiss. Historical biogeography In the Northern Hemisphere, Pleistocene glaciations have had a major influence on the evolutionary history of most extant species, but they have been particularly well studied with respect to freshwater fish species (Bernatchez and Wilson, 1998). The onset of glaciation in many cases either decimated species or severely contracted their ranges, forcing them into ice-free regions called refugia (McPhail and Lindsey, 1986). Extended periods of isolation among refugial populations led to divergences within species through a combination of mutation, genetic drift, and differential natural selection. Due to the duration of glaciation, the genetic divergences resulting from the last glaciation and preceding ones define the population genetic structure of many extant species (Bernatchez and Wilson, 1998). Phylogenetic tools provide the means to assess genetic relationships among organisms and to infer historical events based on genetic divergence times. In addition, we are beginning to use neutral genetic variation to infer historical demographic processes such as range expansion, bottlenecks and allopatric fragmentation. Therefore, studies of biogeography are increasingly reliant on the combination of molecular data and known geographic distributions of organisms. 1 Taxonomy and molecular systematics The reliability of molecular data for detecting common ancestry has made it a popular tool in the field of taxonomy. Under the Biological Species Concept (BSC) , species are defined as reproductively isolated units. Due to difficulty in assessing reproductive isolation, however, phenotypic and life history characters are often used to identify species and subspecies (Avise, 1994). However, these characters are inappropriate for taxonomic use because they often reflect common environmental conditions rather than a common evolutionary origin (Ball and Avise, 1992; Avise, 1994). Molecular data, on the other hand, are presumed to be selectively neutral and are therefore not subject to local environmental conditions in the same way as. phenotypic and behavioural traits. The popularity of the Phylogenetic Species Concept (PSC) coincides with a growing reliance on molecular markers to identify true evolutionary relationships among organisms. Certain inherent problems exist with the PSC, however. In particular, taxonomic designation is subject to the gene chosen and the resolution of the marker (Avise and Bal l , 1990). The P S C also tends to rely heavily on diagnostic differences to distinguish taxa, with little regard for what those differences represent to the organism (Avise and Bal l , 1990). Avise and Bal l (1990) suggest that phylogenetically based studies should focus on "meaningful differences," i.e. those that show concordance among several markers and, therefore, reflect a true barrier to gene flow. Where the barrier is intrinsic (i.e. reproductive isolation), the groups would be distinct species, where the barrier is extrinsic (i.e. geographical barrier), they would be subspecies (Avise and Bal l , 1990). The definition of subspecies they use, and the one I use in this thesis, is "actually or potentially interbreeding populations phylogenetically distinct from, but reproductively compatible with other such groups. Importantly, the evidence for phylogenetic distinction must normally come from the 2 concordant distributions of multiple, independent, genetically based traits" (Avise and Bal l , 1990). Phylogeny and adaptive evolution Another aspect of phylogenetic studies has been to provide insight into the evolution of adaptive traits. Phylogenetic data can reveal the sequence of adaptive changes that have occurred over time or identify similar phenotypes that have evolved independently in different genetic lineages. For my purposes, phylogeny is useful for detecting convergent evolution and for inferring the time scale on which adaptive differences have evolved. This application of phylogenetic studies has been demonstrated repeatedly in the literature. To name a few examples, phylogenetic data have shown that similar developmental pathways have evolved independently in asterinid starfish (Hart et al, 1997); that similar eco-types of sticklebacks have evolved independently in lakes in southwestern British Columbia (Thompson et al., 1997); and that dwarf, normal and anadromous forms of Arctic char have evolved independently in different lineages (Hindar et al., 1986). This aspect of phylogenetic analyses will also be explored with regard to rainbow trout. Molecular data and conservation Finally, molecular tools have become integral to the field of conservation genetics (Bernatchez, 1995; Taylor and Haas, 1996). Most relevant to my study is the basic requirement in conservation biology of identifying evolutionarily independent lineages, for which neutral molecular markers are perhaps the most appropriate tools we have for detecting this. Evolutionary independence is important for conservation because it reveals populations or groups of populations that are genetically distinct from one other. In that these phylogeographic groups have distinct histories, they likely also have distinct adaptive potentials and represent the greatest amount of genetic variation available to the species. Surviving evolutionary change wil l 3 presumably depend upon genetic variation and adaptive potential of a species, so conservation should focus on what we know to be the most ancient differences in the species (Taylor and Haas, 1996). However, neutral genetic variation has limitations in conservation biology that should be acknowledged. For example, despite the ability of neutral markers to detect historical demographic processes that have lead to present population structure, they can not directly assess the quantitative trait variation upon which future selection will act. Thus although neutral markers help to determine where diversity lies, they can not directly assess that genetic variation important for adaptation. Therefore, although neutral markers can detect major phylogeographic divisions, they may miss important differences among populations which contribute substantially to intraspecific genetic variation. Phenotypic and behavioural differences, therefore, deserve an important role in conservation planning for at least two reasons. They can act as a surrogates for the otherwise difficult process of detecting quantitative trait variation among populations if we assume they represent important genetic differences among populations. They also represent extant variation upon which survival during stochastic and rapid environmental change may depend, regardless of genetic uniqueness. This thesis is, however, largely devoted to identifying the structure of neutral genetic variation in O. mykiss for purposes of identifying major phylogeographic groups in British Columbia. Oncorhynchus mykiss: distribution and subspecies The present range of O. mykiss extends from northern Mexico to the Kuskokwim River in Alaska along the eastern Pacific Ocean and from the Sea of Okhotsk to the Kamchatkan Peninsula in the western Pacific (Nielsen et ai, 1994). O. mykiss was originally thought to be a member of the trout genus, Salmo, but morphological and genetic studies later revealed its closer relationship 4 with the Pacific salmon, Oncorhynchus (Thomas et al., 1986; Stearley and Smith, 1989). Fossil evidence indicates O. mykiss most likely diverged from Pacific salmon about six million years ago and from cutthroat trout, O. clarki, about one million years ago (Smith and Stearley, 1993). Within O. mykiss, many subspecies are unofficially recognized. Four putative subspecies exist within California alone: the California golden trout, O. m. aquabonita, two Kern River subspecies, O. m. gilberti and O. m. whitei, and the McCloud River redband, O. m. stonei (Behnke, 1992). Based on the primitive characteristics of the Californian subspecies, Behnke (1992) argued for an origin of O. mykiss in California. The four putative subspecies in the northern extent of the range that will be addressed in some detail in this thesis are the Columbia River redband trout, O. m. gairdneri, the coastal trout, O. m. irideus, the Kamchatkan trout, O. m. mykiss (also coastal in distribution), and Athabascan trout, O. m. subspp. (Behnke, 1992; Carl et al, 1994; Bagley and Gall , 1998). Behnke (1992) outlined ranges from California to Alaska for the redband and coastal trout (see Figures la,b). Though these putative subspecies are unofficial, they are widely acknowledged as the major taxonomic groups in O. mykiss (Nielsen et al, 1994; Bagley and Gall , 1998). The redband trout, O. m. gairdneri, has approximately 130-170 lateral line scales, 63-66 vertebrae, and 30-50 pyloric caeca. It tends to have vestigial basibranchial teeth, yellow and orange coloration, traces of a cutthroat mark below the lower jaw, and yellow or white tips on the dorsal, anal, and pelvic fins (Wishard et al., 1984; Behnke, 1992). The coastal trout, O. m. irideus, has approximately 120-140 lateral line scales, 61-65 vertebrae, 40-70 pyloric caeca and a more elongated body type (Wishard, 1984; Behnke, 1992). The redband trout has a high tolerance for the hot, arid climates of southern Oregon and northern California, as well as the dry interior belt of B . C . (e.g. in the Kamloops area). Anadromous redband trout typically have a long 5 migration to sea. Coastal trout, on the other hand, are adapted to cooler temperatures and the anadromous form generally migrates shorter distances to sea (Behnke, 1992). Coastal trout that range from the M a d River, California, to Kodiak Island, Alaska, have been shown to differ from redband trout at two allozyme loci: lactate dehydrogenase-4 (LDH-4) and tetrazolium oxidase (TO) (Utter et al, 1980). In the Pacific Northwest and British Columbia, populations east of the Cascade Mountain range are also referred to as inland trout to distinguish them coastal trout from which they differ at several allozyme loci (Currens, pers. comm.). In particular, high frequencies of LDH-100 and low frequencies of SOD-100 characterize coastal populations. Despite these differences, evidence is mounting that redband and coastal subspecies designations are not valid. Redband trout from Oregon were found to be more closely related to Oregon coastal trout than to redband trout from the McCloud River in California (Wishard et al, 1984). Furthermore, larger genetic differences have been found using allozymes among coastal trout populations and among redband trout populations than were found between these two groups (Currens, 1997; Busby et al, 1996). The morphological and physiological features that characterize the coastal and redband putative subspecies, in other words, may be due to common selection pressure rather than common ancestry. A n analysis of relationships throughout British Columbia of coastal and inland trout, however, has yet to be done. Given the suggestion of parallel evolution in other parts of the species range, examining genetic variation in British Columbia, Alaska, and Kamchatka for a more informed discussion of the putative coastal and redband subspecies is one goal of this thesis. Another contentious issue is the glacial history and genetic divergence of the Kamchatkan trout, O. m. mykiss. Once thought to be a distinct species altogether, the Kamchatkan trout are 6 Figure 1. The ranges of two putative subspecies in British Columbia and Alaska: the coastal rainbow and the Columbia River redband trout, according to Behnke (1992). Figure lb. The ranges of two putative subspecies in the western U.S.: coastal rainbow and Columbia River redband trout, according to Behnke (1992). 8 now known to be part of the North American O. mykiss lineage and closely related to North American coastal trout (Okazaki, 1984; Behnke, 1992). Allozyme data suggest that Kamchatkan trout diverged from northeastern Pacific coastal trout only 13,000 years ago and that both Kamchatkan and coastal groups diverged from redband trout about 50,000 years ago (Okazaki, 1984). Okazaki (1984) hypothesized that coastal trout may have survived the last glaciation in a northern refuge, from which Kamchatkan trout dispersed westward while coastal trout dispersed down the coast of British Columbia. On the other hand, he proposed that the inland form survived in a southern interior refuge, presumably in large glacial lakes associated with the Columbia River system (Okazaki, 1984). The validity of the subspecies classification for the Kamchatkan trout and the refugial status of the Kamchatkan region (Beringia) will also be examined in this thesis. Finally, the history and present classification of the putative Athabascan trout will also be examined. Carl et al. (1994) used allozymes to estimate that Athabascan trout diverged from coastal and redband trout 160,000 years ago and that Athabascan trout must have survived in an Athabascan refuge during the last glaciation. Meristic analysis suggests that the Athabascan trout, O. mykiss subspp., is intermediate between coastal and inland forms. It has pyloric caecal counts typical of redbands and lateral line counts typical of coastal trout (Carl et al., 1994). The ancient divergence time and the subspecies status are both worth verifying with other genetic markers as the Athabascan trout could be a very significant group for protection. Putative refugia and recolonization pathways from the last glaciation Over the last million years, North America was repeatedly glaciated and the recurrent processes of isolation and recolonization most likely had a profound effect on the population structure of native species (McPhail and Lindsey, 1986). During the last glaciation, the Cordilleran ice sheet covered the province of British Columbia and extended from Alaska south to the Olympic Peninsula from approximately 60,000-10,000 years ago (McPhail and Carveth, 1992). Several ice-free refugia have been hypothesized in and around British Columbia that may have supported O. mykiss and other species. These include the Columbia River, the Chehalis River south of Puget Sound, the Brooks Peninsula on northwestern Vancouver Island, the Queen Charlotte Islands (also known as Haida Gwaii) off the coast of central B . C . , Beringia (Yukon River Valley and the exposed portion of the Bering Strait area), the Nahanni region in the Northwest Territories and northeastern B . C . , and the Athabasca River in Alberta (Behnke, 1992; McPhail and Carveth, 1992). It remains unclear where rainbow trout survived glaciation though this information is important to our understanding of rainbow trout intraspecific variation, taxonomic groups, and conservation planning. Which of these potential refugia were used by rainbow trout during the last glaciation and how colonization of B . C . occurred are issues that will be addressed in this thesis. Adaptive variation i n Oncorhynchus mykiss O. mykiss is one of the most adaptively diverse and geographically widespread Pacific salmonids (Hershberger, 1992). The two most prominent variants are the anadromous and non-anadromous life history forms. Anadromous trout, known as steelhead, hatch in freshwater streams, but spend the majority of their lives in the ocean before returning to freshwater to spawn (Wilson et ai, 1984). Non-anadromous forms, or rainbow trout, are resident in freshwater streams and lakes throughout their lives. Freshwater and anadromous forms are common among many Pacific salmonids, including sockeye salmon (Oncorhynchus nerka), cutthroat trout (O. clarki), bull trout (Salvelinus confluentus), Dolly Varden (Salvelinus malma), and rainbow trout. Evidence suggests that life history is at least partly genetically determined, though offspring of 1 0 resident forms have been known to be anadromous, and vice-versa (Okazaki, 1984). Another prominent life history character is run-timing among steelhead trout (Withler et al., 1965). Steelhead are known to spawn at specific times during the year, generally in Apr i l , but the timing of their return to freshwater from the ocean can vary. Three different run-times are known to exist in steelhead, summer-run, winter-run, and fall-run, and evidence suggests that this return time is at least partly genetically determined (Withler et al., 1965). Other examples of prominent variation in the species include large lake piscivores such as Kootenay Lake Gerrard rainbow trout and Arrow Lake "yellow fins," monoculture rainbow trout such as Fish Lake and Khtada Lake rainbow trout, and small neotenous rainbow trout found above waterfalls (Taylor and Haas, 1996; E . Keeley and E . Parkinson, pers. comm.). I include key life history and phenotypic groups in the thesis to better assess their priority for conservation. It remains unclear how ancient these life history groups are and how much genetic variation they reflect. If they are very old and have recolonized the province separately, they will be very distinct from other adaptive groups in the province and should be given high priority for conservation. Conversely, if they have evolved relatively recently in different geographical locations, then they represent an important adaptive component of diversity, but do not reflect an enormous genetic contribution to the species. In this case, phylogeographic groups deserve the highest conservation priority. Conservation of Oncorhynchus mykiss O. mykiss is an ideal species for a phylogeographic analysis in British Columbia that focuses on intraspecific variation. For its size, B . C . has surprisingly few species, but due to a dynamic glacial history and diverse habitats in the province, intraspecific variation tends to be extensive (Taylor et al., 1999; McPhail and Taylor, unpub.). British Columbia is ecologically very 11 diverse in terms of fish habitat, the coast being characterized by hundreds of short torrential rivers, and the interior characterized by larger river systems, lakes, ponds, and swamps (McPhail and Lindsey, 1970). Climatic differences between the coast and the interior can also be extreme (McPhail and Lindsey, 1970). Rainbow and steelhead trout exist in almost every major watershed in the province, and like other species in the province, individual populations are facing an increasing risk of extinction from a variety of sources. Mining projects, hydroelectric power, forestry impacts, and construction and development are threatening many native populations. Specific conservation initiatives are in order for many native species, particularly the economically important ones, such as O. mykiss. I aim to define the biogeographical units of O. mykiss and to lay the groundwork for a conservation plan that will include both the phylogenetic and adaptive variation of O. mykiss in British Columbia. Organization of the thesis In the second chapter, I review allozyme studies in the literature in an effort to organize previously collected data within the species. This provides a comparison for the m t D N A study that follows and sheds light on subspecies designations and glacial hypotheses. In the third chapter, I present results from a m t D N A R F L P study of rainbow trout that will be used to address the major issues outlined in the introduction: the likelihood of different refugial hypotheses and recolonization routes, the validity of subspecies designations, and the genetic relationships among key life history types. The fourth chapter focuses on m t D N A sequence data and nuclear D N A variation, relating variation in B . C . to that in the southern part of the range and attempting to further resolve genetic relationships among O. mykiss within the province. I conclude with a final discussion and synthesis chapter. 12 Ch.2 Allozyme Variation in Oncorhynchus mykiss INTRODUCTION Allozymes have been assayed extensively in salmonids, particularly for identifying populations in mixed stock fisheries (Utter et al, 1980). Allozymes are the protein products of different alleles at enzyme-coding loci and are scored according to their migration rates on an electrophoretically charged gel (Utter et al, 1980). They have been used extensively to define population structure not because they have particularly high mutation rates, but because the allozymes are easily scored and isolated populations can often be distinguished by allele frequency differences. These frequency differences are generally assumed to be due to genetic drift, though neutrality has been questioned for several loci (Huzyk and Tsuyuki, 1974; Verspoor and Jordan, 1989). Specific aims The putative subspecies, coastal rainbow trout (O. m. irideus) and redband trout (O. m. gairdneri), were initially identified using allozymes (Utter et al, 1980), and this division has been further supported by additional allozyme studies (Parkinson, 1984; Currens et al, 1990). These studies (see Table 1), however, have mainly been limited to small geographic regions relative to the entire species range. One of my aims from consolidating data was to assess the consistency of genetic differences between these putative taxonomic groups throughout the species range. For example, coastal and inland divisions have been demonstrated within B . C . , among samples from Kamchatka, Vancouver Island and Columbia, and among samples from the Pacific Northwest, but a combined analysis of these watersheds has not been done to demonstrate that divisions are maintained on a broader geographic scale. 13 In addition, Athabascan and Kamchatkan trout have both been identified as distinct subspecies, O. m. subspp. and O. m. mykiss, respectively, but in both cases the comparisons were relatively limited. Both Athabascan and Kamchatkan trout were compared with only a few populations from Vancouver Island and the Columbia River. I aimed to test whether these divisions are upheld in a more thorough review of intraspecific variation. The final reason for consolidating allozyme data was to provide a comparison for the mitochondrial data that follow in the next chapter. M E T H O D S Data sets were compiled from the Kamchatkan Peninsula to California (see Table 1 for populations and references). This analysis was restricted to the only three allozyme loci that all studies had in common: L D H - B 2 equivalent to L D H - 4 (lactate dehydrogenase), M D H (malate dehydrogenase), and S O D (superoxide dismutase). Though the number of loci is low, a study using these few loci is not unheard of for assessing biogeographical relationships. Bickam et al. (1995) and Varnavskaya and Beacham (1992) used three and five loci, respectively, to compare variation throughout the ranges of pink {O. gorbuscha) and sockeye salmon (O. nerka). In addition, these three loci, particularly L D H - 4 and S O D , were among the most informative allozymes in the regional studies of rainbow trout. Regional structure was determined with cluster analyses using the P H Y L I P software package (Felsenstein, 1993). The geographically proximate populations that clustered together in previous studies were used as the regional units among which phylogenetic relationships were assessed (see Table 1). Mean frequencies of the 100 allele at each locus were constructed for each regional group of rainbow trout, and all subspecies of cutthroat trout were combined to determine average allele frequencies for the outgroup (Leary et al., 1987). Seqboot was run on 14 Table 1. Stream populations included in each regional group used for the allozyme analysis. The source used is written in parentheses. Regional groups #Pops Populations (citations) Vancouver Island/ 60 27 populations along the east coast of Vancouver Island, Lower Mainland extending from Tsulquate to Cowichan Rivers. 7 west-draining rivers, on Vancouver Island, and 28 mainland populations across the straight from the Vancouver Island populations, extending as far east as the Coquihalla and including the Nahatlach and the Stein (Parkinson, 1984) Upper Fraser 5 Nicola, Thompson, Deadman, Bridge, Chilko (Parkinson, 1984) Skeena 6 Lakelse, Zymoetz, Simpson, Deep, Kispiox, Babine (Parkinson, 1984) Kamchatka 1 Kamchatka (Okazaki, 1984) Athabasca 1 Wampus Creek (Carl et al, 1994) Dean ' 2 Dean, Atnarko (Parkinson, 1984) Upper Snake 5 Snake, Clearwater, Selway (Reisenbichler et al, 1992); Selway, Snake, (Carl et al, 1994; Schreck et al, 1986). Lower Snake 4 Little Jack, Castle, Duncan, Reynolds (Wishard et al, 1984) Owyhee 4 Jordan, Little Boulder, Cabin, Current (Wishard et al, 1984) White 8 Lower White, Lower Tygh, Upper Tygh, Little Badger, Threemile, Rock, Gate, Barlow (Currens et al, 1990) Deschutes 10 Deschutes, Bakeoven, Buck, B ig Log , Lower East Foley, Upper East Foley, Lower Nena, Middle Nena, Upper Nena, Round Butte strain (Currens et al, 1990) Oregon Coast 18 Trask, Neskowin, Slickrock, Siletz, Crooked, Lobster, M i l l , B ig , Honey, Elk, Upper Rogue, Lower Rogue, Slate Ck, Galice Ck, B ig Windy, Shasta, Saunders, Hunter (Reisenbichler et al, 1992). California 31 Sacramento (5), McCloud (7), Pit (10), Klamath (3), Eel, Gualala, Feather, B ig Ck, Devil Ck, San Luis Rey (Berg and Gall , 1988) Puget Sound 7 Quillayute, Hoh, Queets, U . Quinault, N . Chehalis, Skagit, Snow (Reisenbichler and Phelps, 1989) Cutthroat Westslope, Coastal, Lahontan, Yellowstone, Fine-spotted, Greenback, Colorado (Leary et al, 1987) 15 Figure 2. Mean allozyme frequencies of LDH-100, SOD-100 and MDH-100 for each regional group used. The line represents the coastal mountain range. Kamchatka should be located much further to the west, in eastern Siberia. 16 the matrix of mean frequencies to generate pseudoreplicate data sets for bootstrapping (N=100). Bootstrapping is a resampling technique designed to assess the robustness of phylogenetic results. If the number of resampled data sets that fit the same pattern as the original data is high, the phylogenetic tree is robust. If bootstrap support is low, the phylogenetic branching patterns are probably due to a small number of data points and are, therefore, less likely to be accurate. A phenogram was constructed on the bootstrapped data using the Cavalli-Sforza chord distance based on genetic drift without mutation with Gendist (Felsenstein, 1993). Neighbor-joining was then done with Neighbor, which does not assume a molecular clock, and Consense was used to find the best tree and to determine bootstrap support (Felsenstein, 1993). To construct a maximum likelihood phenogram, Contml was used, which has no assumption of a molecular clock and is based on genetic drift without mutation. The significance of the branch lengths was determined from the confidence intervals from Contml, although Felsenstein (1993) warns that the variance may be underestimated using this approach. Drawgram was used to construct both the distance and maximum likelihood-based phenograms. Finally, both Gendist and Neighbor were run on the original matrix to construct a tree with appropriate distance scales, and Drawgram was used to plot the phenogram. RESULTS Relationships among particular groups that had been previously uncovered at a smaller scale were also found in this study, indicating that these three loci were relatively reliable. For example, the White River and Deschutes River were grouped with coastal and inland groups, respectively, as indicated by Currens et al. (1990); Kamchatkan trout grouped more closely with Vancouver Island/ Lower Mainland trout than with Columbian populations as found by Okazaki (1984); and Athabascan trout remained somewhat distinct from both the lower Columbia and the 17 Lower Mainland as found by Carl et al. (1994). Those populations from Kamchatka to California, identified as coastal trout by Behnke (1992; Figs. la,b, C h . l ) , formed one group, whereas populations from the Owyhee drainage in Oregon to the Athabasca River, identified as redband trout by Behnke (1992), formed another group (Figs. 3a,b). Bootstrap support for the distance-based tree was 67% for the inland group and 77% for the coastal group, not including the Skeena River. The Skeena River, on the north coast of British Columbia, clustered with the coastal group with bootstrap support of only 48% (Figure 3a). The Skeena was somewhat intermediate between coastal and inland, as the high frequency of SOD-100 in the Skeena populations was similar to inland populations whereas the high frequency of LDH-100 was similar to the coastal group, though it formed part of the coastal group. Though Carl et al. (1994) claimed that Athabascan trout were highly unique, in this analysis they were quite similar to Skeena River trout. The Cavalli-Sforza chord distance between the two was 0.099, which is on the order of variation found both among coastal populations and among inland populations (Table 2). More surprisingly, however, was the similarity between Athabascan trout and cutthroat trout. The Athabascan trout was more closely related to cutthroat than to any other rainbow populations, which I will discuss further below. I do, however, recognize that inferences should be made with caution as the analysis only represents three loci. 18 51 77 77 48 Kamchatka 43 B . C . Central Coast Puget Sound Vancouver Island/ Lower Mainland Oregon Coast California 49 White River Skeena River 67 67 67 67 67 Upper Snake Athabasca Owyhee Deschutes Lower Snake - Fraser COASTAL INLAND Cutthroat 0.10 0.20 Figure 3a. Allozyme cluster analysis of genetic distances created using Cavalli-Sforza chord distance in Gendist and neighbor-joining in Neighbor. Numbers represent % support from 100 bootstrap replicates. Phylogenetic analyses were done using the PHYLIP software package (Felsenstein, 1993). 19 Skeena Vancouver Island/ Lower Mainland Puget Sound Oregon Coast California Central Coast B.C. White Kamchatka COASTAL Athabasca Lower Snake Deschutes Owyhee *Upper Snake Fraser INLAND Cutthroat 0.05 Figure 3b. Allozyme cluster analysis based on a maximum likelihood model using Contml in PHYLIP (Felsenstein, 1993). Asterisks mark significant distances (i.e. where confidence intervals do not bracket 0.00). Table 2. Allozyme genetic distances among various putative rainbow trout subspecies groups and cutthroat trout using the Cavalli-Sforza chord measure. Coastal Inland Kamchatka Athabasca Coastal 0.00-0.14 Inland 0.30-0.57 0.00-0.10 Kamchatka 0.10-0.22 0.70-0.86 0.00 Athabasca 0.01-0.35 0.17-0.28 0.48 0.00 Cutthroat 0.04-0.33 0.29-0.45 0.43 0.03 21 DISCUSSION Coastal and inland trout Allozymes have not been examined extensively throughout the northern range of O. mykiss and, therefore, taxonomic boundaries of coastal and redband subspecies in this region have been based on relatively limited allozyme data (Behnke, 1992). In this analysis, coastal and inland subdivisions were basically maintained from California to Kamchatka, but the one northern sample in British Columbia, the Skeena River trout, was relatively unusual (Figure 3a,b). The high frequencies of the 100-alleles at each of the three loci examined are suggestive of a bottleneck that brought these 100-alleles to near fixation. Whether this was caused by expansion from a small northern coastal refuge, such as one in the Queen Charlotte Islands, or due to expansion from a more distant refuge is unclear as all 100 alleles were likely present in every refugial population. Nevertheless, this analysis .weakly corroborated Behnke's (1992) "coastal" and "inland" divisions in that the Skeena trout are part of the coastal group. Further insight into the glacial origin of these trout requires a larger database as presented in subsequent chapters. Athabascan trout Carl et al. (1994) reported a very ancient divergence time for the Athabascan trout relative to coastal and inland populations in British Columbia. Their allozyme data suggested that the divergence time of Athabascan trout from other rainbow trout was about three times that of the divergence between coastal and inland trout (Carl et ai, 1994). The similarity found in this study between Athabascan and Skeena River trout, however, suggests that Athabascan trout may not be extremely different from northern rainbow trout populations in British Columbia. A n alternative explanation for the origin of the Athabascan trout, however, is that they crossed over from the Fraser or Peace rivers in British Columbia into the Athabasca River post-glacially (Behnke, 1992; 22 Nelson and Paetz, 1992). Given evidence for watershed exchanges among these rivers post-glacially (McPhail and Lindsey, 1986) and the presence of several other species common to both drainages, a cross-over event between the Fraser (or possibly the Peace) and Athabasca rivers after the last glaciation is not inconceivable (Nelson and Paetz, 1992). However, even if Athabascan trout came from northern British Columbia, the extremity of Carl et al's. (1994) findings requires further explanation. Very unusual allozyme frequencies were found in Wampus Creek trout (the Athabascan population sampled by Carl et al, 1994) including LDH-120 , PEPA-110 and PEPA-120 (dipeptidase), and MPI-94 (mannose-y-phosphate isomerase) (see Table 3). Given the similarity between cutthroat trout and Athabascan trout allozyme frequencies (Table 3), the possibility of hybridization between the two species was investigated for these other loci. Hybridization between cutthroat and rainbow trout is relatively common (e.g. Campton and Utter, 1985), and Nelson and Paetz (1992) even suggested that hybridization may have occurred in the Athabasca River. Though the westslope cutthroat is native to the Athabasca River, the only cutthroat allozyme samples available in the literature were coastal cutthroat from Puget Sound in northwestern Washington (see Fig. 4; Campton and Utter, 1987). Cutthroat trout have been introduced to Wampus Creek with unknown source locations (Scott and Crossman, 1973; Nelson and Paetz, 1992; George Sterling, pers. com.). Whether coastal cutthroat were ever introduced to the Athabasca or whether coastal cutthroat and westslope cutthroat have similar allozyme frequencies at these loci is unknown to me. However, many of the alleles found to be unique to the Wampus Creek rainbow trout were present in coastal cutthroat trout and the frequency of the PEPA-110 allele was very similar to that found in Wampus Creek rainbow trout (see Table 3). A n alternative hypothesis to that of a glacial refuge in the Athabasca, therefore, is that rainbow 23 trout gained access to the Athabasca River post-glacially due to a cross-over event from British Columbia, and further divergence of the Wampus Creek rainbow population occurred due to hybridization with cutthroat trout. This hypothesis will be further explored using mitochondrial D N A in the next chapter. Kamchatkan trout Okazaki's (1984) hypothesis regarding a Beringian origin for all coastal trout was not well supported in this analysis. Although Okazaki (1984) fournd Kamchatkan trout to be very similar to Vancouver Island and Puget Sound in allozymes, in this analysis they appear to be fairly different from North American coastal populations (see Table 2). These hypotheses, however, are based on relatively limited data and will be further explored using mitochondrial D N A in the next chapter. In conclusion, coastal and inland divisions throughout North America were weakly supported in that they grouped together in both distance and maximum likelihood cluster analyses. However, particularly in northern British Columbia, the distinction appears to weaken and little data exists north of the Skeena River. A n alternative hypothesis is proposed for the Athabascan rainbow population involving cross-over from British Columbia with subsequent hybridization with cutthroat trout. Finally, Kamchatka trout appear to be fairly different from North American coastal populations, although clearly more closely related to coastal than to inland rainbow trout (Table 2). The origin of the North American coastal populations and of Kamchatkan trout, however, remains unclear. 24 Table 3. Allozyme frequencies of rainbow trout from Athabascan rainbow (Wampus Creek) as compared to "coastal" and "inland" rainbow and to coastal cutthroat. Athabascan, "coastal" and "inland" rainbow trout frequencies are from Carl et al. (1994); and coastal cutthroat frequencies are from Leary et al, (1987) and Campton and Utter (1987). Athabascan "coastal" "inland" coastal rainbow rainbow rainbow cutthroat LDH-112 0.00 0.00 0.00 0.167 LDH-120 0.11 0.00 0.00-0.01 0.00 PEPA-110 0.74 0.00-0.03 0.00-0.29 0.794-0.966 PEPA-120 0.26 0.00 0.00 0.021-0.065 MPI-94 0.29 0.00 0.00-0.18 0.017-0.064 25 Ch.3 Mitochondrial DNA RFLP variation INTRODUCTION Allozymes have been the most commonly used markers for assessing genetic variation throughout the range of Oncorhynchus mykiss, however, evolutionary relationships among the alleles are usually not known and resolution is relatively low in comparison to mitochondrial and other nuclear loci (Avise, 1994). Furthermore, the expression of allozymes at the phenotypic level makes them more likely to be similar due to convergent evolution than many kinds of D N A (Huzyk and Tsuyuki, 1974; Verspoor and Jordan, 1989). The mitochondrial genome offers a more informative alternative for assessing evolutionary history. The mutation rate of m t D N A is high relative to allozymes, and its effective population size is one-quarter that of nuclear loci due to its maternal and haploid inheritance. The lack of recombination among m t D N A genomes makes interpretation of evolutionary relationships relatively straightforward (Avise, 1994; Wilson et al., 1985). These features make m t D N A a very appropriate and useful marker for intraspecific analysis. Increasing knowledge about the evolution of m t D N A make it an attractive marker for studying historical biogeography and molecular systematics (Avise and Ball , 1990; Crandall and Templeton, 1993; Avise, 1994; Templeton et ah, 1995). Conservation initiatives have relied on m t D N A more than any other marker (Bernatchez, 1995; Taylor and Haas, 1996). The term, Evolutionarily Significant Unit (ESU), defined by Waples (1995) as a "substantially reproductively isolated group representing an important component of the evolutionary legacy of the species," was made into a testable concept using m t D N A and nuclear loci (Moritz, 1994). Though not universally accepted, the use of ESUs in U .S . conservation plans for Pacific salmon and trout attests to the increasing reliance on molecular markers in management and conservation (Busby et al., 1996; Myers et al., 1998). For all of the above reasons, m t D N A should be an excellent marker to address the biogeographical, taxonomic, and conservation issues surrounding O. mykiss. Specific aims M y research goals were to address biogeographical hypotheses, subspecies groups, and conservation issues in O. mykiss. With respect to biogeography, I aimed to assess evidence for survival in several putative refugia in and around B . C . , namely, the Columbia River, the Chehalis River, the Queen Charlotte Islands, Beringia, and the Athabasca River. In addition, I aimed to assess post-glacial dispersal routes throughout the northern part of the species range and to determine whether significant phylogenetic structure exists. I was particularly interested in the divergence of the Athabascan trout as well as where North American coastal trout and Kamchatkan trout survived glaciation. With respect to subspecies, I aimed to assess molecular support for all the putative subspecies throughout the sample range, which included Athabascan trout, Kamchatkan trout, and North American coastal and redband trout. M y conservation goals were to determine whether any major phylogeographic divisions existed in British Columbia and how adaptive variation related to phylogenetic variation. If adaptive groups proved to represent major components of genetic variation, they should be granted high priority in conservation plans. If phylogenetic variation, however, proved to be more significant, conservation plans should be geared towards conserving as many phylogenetically distinct groups as possible. M E T H O D S Sample collection Samples were collected from every major watershed in British Columbia and from at least two sites in each of the Kamchatkan Peninsula, Alaska, the Athabasca River, and the Pacific 27 Northwest region of the U.S . Samples were taken from several sites along each major river for the purposes of sampling a maximum amount of variation per watershed. Approximately five to ten sites were sampled per river except in the Liard River where O. mykiss only occurs in some headwaters, and at each site, no more than seven individuals were collected. In total, 292 samples from 59 sites were initially included in the composite R F L P analysis using restriction enzymes that could distinguish the major groupings (see Fig.4). Once major clades were identified, an additional 219 samples from 56 sites were included in a diagnostic R F L P analysis (see Fig. 5). The diagnostic analysis was done to increase sample sizes and to further resolve the geographical distributions of major clades. Fin clips were usually taken from the adipose fin, although muscle tissue was occasionally used. Electro shocking was the most common sampling technique, but some samples were collected by gill net or angling. Samples were identified as resident rainbow trout rather than sea-run steelhead if they were found either above barriers or in lakes. I tried to minimize the number of sites where past introductions or transplants have been documented. Genetic analysis D N A was extracted from 20-30mg of tissue using a standard phenol/ chloroform extraction procedure (Taggart et al., 1992). The regions of the m t D N A genome chosen for analysis were the D-loop, cytochrome b, and N A D H dehydrogenase subunits 5 and 6 (ND5/6) as they have all been shown to be variable in salmonid fishes (Park and Moran, 1994). D N A was amplified using the Polymerase Chain Reaction, P C R . The primers used to amplify the D-loop and cytB regions were HN20 and C-Glu (Bernatchez and Danzmann, 1993), and the primers for the ND5/6 region were C- Glu and C-Leu-3 (Park et al, 1993). The P C R conditions for both reactions were as follows: 1.5 minutes at 95°C, 1.0 minute at 55°C, and 2.5 minutes at 72°C 28 Figure 4. Composite haplotype sample sites in B.C. Vancouver Island: 1) Cowichan R., 2) Lucky Ck., 3) Elsie L., 4) Robertson Ck, 5) Quinsam R., 6) Bamfield, 7) Klootchlimmis Ck., 8.) Mahatta R., 9) Keogh R.; Lower Mainland: 10) Chehalis R., 11) Capilano R., 12) Vedder R.; upper Fraser: 13) Anderson Ck. , 14) Pennask L., 15) Fish L., 16) Bonaparte R., 17) Tzenzaicut L., 18) Cheslana L., 19) Bivouac Ck., 20) Dome Ck.; Queen Charlotte Islands: 21) Honna R., 22) B V M ; Central Coast: 23) Dean R., 24) Kitimat R.; Skeena: 25) Morice R., 26) Babine R., 27) Canyon Ck., 28) Toboggan Ck., 29) Goathorn Ck.; Nass: 30) Meziadin R.; North Coast: 31) Tahltan R., 32) Taku R.; Liard: 33) Kutcho Ck.; Yukon: 34) Mclntyre, 35) Kathleen R.; Peace: 36) Firesteel R., 37) Mesilinka R. 38) Nation R., 39) Chowade R., 40) Graham R, 41) Moose L., 42) Kiskatinaw R.; Athabasca: 43) Wampus Ck., 44) Halpenny Ck.; Columbia: 45) Tumtum Ck., 46) Hill Ck., 47) Halfway R., 48) Lardeau R., 49) Procter Ck. , 50) Bayonne Ck.; Okanagan: 51) Goatskin Ck.; Snake: 51) Selway R., 52) Dworshak R.; Puget Sound: 54) Tahuya R.; Oregon: 55) Alsea R.; Alaska: 56) Natnek, 57) Karluk R.; Kamchatka: 58) Kavachina, 59) Snotolvayam 29 Figure 5. Diagnostic sample sites in B.C. Vancouver Island: 1) Junior L., 2) Efflingham R.; Lower Mainland: 3) Squamish R., 4) Abbotsford Hatch; U. Fraser: 5) American Ck., 6) Emory Ck., 7) Bridge R., 8) Deadman R., 9) Spius Ck., 10) Coldwater R., 11) Gaspard Ck., 12) Chilko/ Chilcotin R., 13) Puntzi Ck., 14) Euchiniko, 15) Baezaeko, 16) Coglistiko, 17) Brewer's Ck., 18) Tahtsa L., 19) Hautete Ck., 20) Baptiste Ck., 21) Forfar; Queen Charlotte Islands: 22) Brent Ck., 23) B V M , 24) L. Phantom Ck.; Central Coast: 25) Ashlulm Ck., 26) Abuntlet L., 27) Union, 28) Khtada; Skeena; 29) Kispiox R., 30) Sustut R., 31) Natlan Ck.; Nass: 32) Paw L., 33) Niska Ck.; North Coast: 34) Iskut R., 35) Taku R„ 36) Disella L.; Liard: 37) Wolverine L., 38) Blue Sheep L.; Peace: 39) Firesteel R., 40) Ingenika R., 41) E l Condor 42) Sekunka R., 43) Burnt R., 44) Elizabeth Ck.; Athabasca: 45) Cabin Ck.; U. Columbia: 46) Fry Ck., 47) Summit Ck.; Okanagan: 48) Goatskin Ck., 49) Fall Ck.; Washington: 50) Bogachiel R., 51) Nooksak R., 52) Deer Ck.; L. Columbia: 53) Tucannon R., 54) Lewis R.; Alaska: 55) Nushugak R., 56) Neguthluk R. 30 (Palumbi, 1990). The reactant volume was 25ul, and the reactants and their final concentrations were: l x reaction buffer; 0.800mM dNTP mix (0.200mM each), 0 .6uM each primer, 2 Units Taq ( B R L ) , and 4 m M M g C l 2 . The P C R products from the D-loop/cytB and ND5/6 amplification reactions, approximately a total of 4.6 K B of 16.5 K B total genome size, were combined for enzymatic digestion. The composite haplotypes presented here were digested with 14 restriction enzymes (New England Biolabs, N E B ) . Twelve tetrameric restriction enzymes were used, Alul, Bfal, BstUl, Ddel, Dpnll, Haelll, Hhal, Hinfl, Mspl, Nlalll, Rsal, and TaqI, one multipentameric enzyme, A v a i l , and one multihexameric enzyme, Hindi. Diagnostic samples were digested either with Alul and Dpnll or with Alul and Haelll. Restriction sites were inferred from fragment patterns visualized on 2% agarose gels stained with ethidium bromide. Phylogenetic analysis Individual enzymatic digestion patterns and composite haplotype identities (see Tables 4 and 5 in Results) were compiled into a presence-absence matrix of restriction sites using Generate from the R E A P software package (McElroy et al., 1992). A weighted estimate of evolutionary divergence from different restriction enzyme fragment patterns was computed using D (as per Nei and Tajima 1981; Nei and Tajima, 1983; Nei and Miller, 1990, eq.4). Both parsimony and distance methods were used to construct phylogenetic trees using the P H Y L I P software package (Felsenstein, 1993). Genetic matrices were bootstrapped 100 times with Seqboot for both parsimony and distance analyses. A Dollo parsimony tree was constructed using Dollop, which allows for a higher rate of losses than gains of restriction sites (Swofford et al, 1990; Felsenstein, 1993). Phylogenetic trees based on distance were constructed in Boot (Jaarola and Tegelstrom, 1995) and Neighbor (Felsenstein, 1993) using the neighbor-joining 31 option. To find the majority-rule consensus tree from the 100 bootstrapped trees, Consense was used on both these data sets. In order to construct a phylogenetic tree with an appropriate genetic distance scale, the output of D (from the R E A P package) was run in Neighbor (in P H Y L I P ) to construct a single neighbor-joining tree of the original data. Drawgram was used to construct all phenograms. Midpoint rooting was done on all phenograms using Retree. Midpoint rooting places the root halfway between the two most divergent haplotypes. Midpoint rooting was done because the alternative of outgroup rooting required comparing rainbow trout restriction sites to cutthroat sites which proved too difficult. Genetic diversity and hierarchical structure among watersheds Diversity indices were determined using DA from the R E A P package (McElroy et al., 1992). DA estimates haplotype and nucleotide diversity within populations and computes nucleotide divergence between pairs of populations. Haplotype diversity measures the probability that any two individuals in a sample are different and is calculated according to Nei (1987; equations 8.4, 8.5 and 8.12). Nucleotide diversity is a composite statistic based on both the number of different haplotypes and the genetic divergence among them. It calculates diversity as per Nei and Tajima (1981; Nei , 1987, eqs. 10.19, 10.7). Diversity analyses were important for identifying potential refugia as former refugia would likely be more diverse than recently colonized sites (Templeton et al, 1995). Hierarchical structure of genetic variation among populations of rainbow trout was determined using the Holsinger and Mason-Gamer Nucleodiv program (Holsinger and Mason-Gamer, 1996). Nucleodiv uses pairwise g s t values (an F s t analog) to identify genetic similarity among watersheds. A cluster analysis similar to U P G M A was then used to create a dendrogram of populations. The gst statistic draws on coalescence theory such that the mean time to coalescence of populations from a single node is smaller than the mean time to coalescence of populations from different nodes (Holsinger and Mason-Gamer, 1996). In addition to the Nucleodiv program, Monte was used from the R E A P software package (Roff and Bentzen, 1989; McEl roy et al, 1992) which uses Monte Carlo simulations to test for geographic heterogeneity. Monte assesses population differentiation with haplotype frequencies alone, rather than in combination with nucleotide divergence, which was more easily interpreted. In the Monte analysis, repeated randomizations (1000 times) of the data were done to determine the probability of getting the %2 value obtained simply by chance. I tested the significance of all nodes in the Nucleodiv tree by collapsing all populations stemming to the right of a given population into one group and comparing them to the other population at that node. Partitioning of genetic variation within and among populations and regional groups was assessed using an Analysis of Molecular Variance, or A M O V A (Schneider et al, 1997). A n A M O V A computes three statistics: the proportion of variation among groups (Oct), the proportion of variation among populations within groups (Osc), and the proportion of variation within populations (1-Ost) (Excoffier et al, 1992). Diversity was based on both frequency differences and inter-haplotypic genetic distances. M y main objective was to find the biogeographic partitioning with the highest among group (Oct) variation as these groups would likely represent the major focus for conservation. Several different geographical subdivisions were used to assess the effect on genetic partitioning. Histor ical demography The historical demography of the species was assessed using the nested clade analysis of Templeton et al (1995). The test was designed to distinguish between range expansion, allopatric fragmentation and restricted gene flow as possible mechanisms for an association 33 between haplotype and geography. A n inference of range expansion would suggest post-glacial population dispersal and expansion, allopatric fragmentation would suggest past isolation events whose signature has not been completely ehminated, and restricted gene flow would suggest recurrent limited migration or dispersal among sample sites. The usefulness of the nested clade analysis to my project was that it assessed whether the distribution of genetic variation observed was mainly due to relatively recent restricted gene flow or to post-glacial range expansion. Furthermore, if range expansion was found to be the driving force, the nested clade analysis offered an objective means to assess the use of particular refugia. Of course, the predictions for restricted gene flow, range expansion, and allopatric fragmentation (outlined in Table 4) are not mutually exclusive, which complicates data interpretation. The basis for the different predictions comes from population genetics theory, computer simulations, and empirical results from the literature (Templeton, 1998). Population genetic theory predicts that older haplotypes will be more widespread and have more mutational connections to other haplotypes (Crandall and Templeton, 1993). Both simulation and empirical results support these predictions (Cann et al, 1987; Crandall and Templeton, 1993; Templeton, 1998). Restricted gene flow among populations is, therefore, more likely to be reflected by the limited distributions of derived haplotypes than ancestral haplotypes because the distributions of ancestral haplotypes may be large due to the time they have had to disperse. The specific predictions for restricted gene flow in the nested analysis are that tip (derived) haplotypes have significantly small distributions and that interior (ancestral) haplotypes have significantly large distributions. The specific predictions for range expansion are that tip haplotypes have significantly large distributions and some interior haplotypes have significantly small distributions (Templeton et al, 1995). This is based on the idea that during range expansion, some newer haplotypes wil l be part of the expanding populations and some older haplotypes will remain behind in the source populations (Templeton et al., 1995; Cann etal, 1987). Furthermore, as mutations accumulate in the newly colonized area, these newly derived haplotypes are likely to be found far away from one another (Templeton et al., 1995). The location of the significantly small interior haplotypes is likely to be the source area, or in the case of my study, glacial refugia. The predictions for allopatric fragmentation are that haplotypes will be distinct from each other by more than the average number of mutations (Templeton et al., 1995). This is a well-supported prediction both theoretically and empirically (Bagley and Gall , 1998; Templeton et al., 1995). Secondly, these distinct haplotypes, if they occur in more than one section of the cladogram due to ancestral polymorphism, should show geographic congruence indicating that the geographical isolation affected more than one genetic lineage (Templeton et al., 1995). To determine which haplotypes were ancestral and which were derived, a network of haplotypes was created. Haplotypes with more than one mutational connection to other haplotypes were termed "interior" or ancestral. These haplotypes were presumably the oldest ones that gave rise to other haplotypes in the network. Haplotypes with single mutational connections, located on the outside of the network, were termed "tip" haplotypes and were considered more derived. The haplotypes were "nested" such that haplotypes that differed from one another by one mutation were grouped into a one-step clade, and one-step clades separated from one another by a single mutation were grouped into two-step clades, and so on until all the haplotypes were grouped into a single clade. Where a haplotype could have been included in more than one clade group, it was included in the group furthest from the interior haplotypes (for example, haplotype 35 Table 4. Templeton's predictions of restricted gene flow, range expansion and allopatric fragmentation (from Templeton et al., 1995). Restricted Gene Flow Range Expansion Allopatric Fragmentation Predictions 1. More derived (general) haplotypes wil l have restricted distributions 2. Older haplotypes wil l be more widespread 1. Derived haplotypes will be located far away from their ancestral haplotypes. 2. Some ancestral haplotypes wil l have small distributions (in source location) 1. Some haplotypes will have a greater than usual number of mutations separating them from the rest of the mutational network. 2. These haplotypes will be geographically congruent with one another. Predictions 1: Several tip clades with (specific) a significantly small Dc, some interior clades with a significantly large Dc. 2. Dc(I-T) significantly large 3. Dc increases with increasing clade level 4. Above patterns hold for Dn unless some gene flow is due to long-distance dispersal events. 1. Dc and D n significantly large for several tip clades, though some may be significantly small with long-distance colonization. 2. Significantly small Dc(I-T), for contiguous range expansion and significantly small Dn(I-T) for long distance colonization. 1. Several clades, particularly at higher levels should have a significantly small Dc; this should represent a reversal of patterns at lower clade levels. 2. These clade groups should be connected by more than the average number of mutational steps. Dc = clade distance (average dispersal distance of a clade from its center of distribution) Dn = nested clade distance (average dispersal distance of a clade from the center of distribution of its nested clade, i.e. the group of clades one mutation away) Dc(I-T) = Interior minus Tip clade distances within a clade Dn(I-T) = Interior minus Tip nested clade distances within a clade 36 26 is in the group 1-1, not 3-1, Figure 13). The term "nested clade" refers to the next most inclusive grouping of a particular clade such that the nested clade of the 0-step clades (or individual haplotypes) is the 1-step clade to which they belong. The average geographic distance among samples of a particular clade is termed the "clade distance," or Dc. The average geographic distance among samples of a particular clade from the geographic midpoint of its nested clade is termed the "nested clade distance." Dc(I-T) is the clade distance of the interior haplotype within a clade minus the average clade distance of the tip haplotypes in that clade, which essentially assesses whether older haplotypes are more or less widespread than newer ones. Dn(I-.T) is the nested clade distance of the interior haplotype minus the average nested clade distance of tip haplotypes within a clade, which essentially measures whether or not older haplotypes are located further from other ancestral types than newer ones. The analysis was based on a permutation test whereby haplotypes were permuted with respect to the geographic locations that were sampled using Temple (Blachford et al., unpub.). Dc, Dn, Dc(I-T) and Dn(I-T) were all calculated, and significantly small and large distances were used to infer a pattern. Sample locations and haplotypes were permuted 100 times. As the analysis relies on consistent and thorough sampling (large geographic gaps can disrupt results, Templeton et al., 1995), only samples within B.C. or close to the B.C. border were included in the analysis. Taxonomy To assess the validity of the coastal and inland subspecies groups, a character analysis was performed using both a mitochondrial haplotype tree and the population cluster analysis using Nucleodiv. According to the definition put forward by Avise and Ball (1990), subspecies should be phylogenetically differentiated groups of populations. Subspecies may or may not be 37 reciprocally monophyletic, but they should at least be distinguishable from each other at the population level. If putative subspecies groups were neither reciprocally monophyletic nor distinguishable at the population level, the validity of the subspecies would be seriously compromised. Adaptive evolution One of the main goals of this project has been to characterize evolutionary relationships among rainbow trout populations throughout B . C . , including both phylogeographic and phenotypic variation. Phenotypic similarity may be the result of common environmental conditions or common ancestry. If phenotypic similarity is due to common environmental conditions, genetic variation should reflect biogeographical groups rather than phenotypically similar ones. If, on the other hand, phenotypic groups are similar due to common ancestry, genetic variation should be partitioned such that the greatest amount of variation exists among distinct phenotypic groups. For example, even and odd year pink salmon were found to be more divergent from each other than were populations of the same year class from the Fraser and Alaska (Aspinwall, 1974; Beacham et al., 1985; Varnavskaya and Beacham, 1992; Brykov et al., 1996). Therefore, the behavioural characteristic of whether spawning occurs in even or odd years appears to be more ancient than biogeographical subdivision caused by the last glaciation. Conversely, anadromous sockeye and non-anadromous kokanee populations (O. nerka) along the west coast of North America exhibited more variation between geographically distant populations than among different life history types that were geographically proximate to each other (Taylor et al., 1996). Consequently, the majority of the genetic variation in O. nerka exists among geographical groups, whereas in pink salmon, it exists among even and odd-year spawning runs. I tested whether geographically proximate populations or phenotypically similar 38 populations were more genetically similar to each other. A M O V A analyses were done (Schneider et al, 1997) on rainbow trout samples with respect to two life history characters, anadromy and run-timing (for samples, see Table 10 and 11 in Results). In addition, selected phenotypic groups (large-bodied piscivorous trout, above barrier trout, and sea-run steelhead) were placed onto a haplotype m t D N A tree and a population cluster analysis was performed to determine whether or not any apparent associations existed between genetic variation and phenotypic variation. If phenotypes were associated with either particular haplotypes or population clusters, this would suggest that phenotypes were relatively old and represented different genetic groups. If, however, no association existed between phenotype and genotype, I would conclude that the different phenotypes have arisen relatively recently and independently in different lineages. RESULTS Genetic variation and relationships among haplotypes Approximately 123 restriction sites and 504 base pairs were analyzed with recognition enzymes (see Table 5 for restriction enzyme patterns, see Table 6 for composite haplotype patterns). The most common haplotype (haplotype 1) was found in every major watershed and in 45% of the samples. Haplotypes 24 and 32 were also abundant, comprising 12.6% and 11% of the samples, respectively. Twenty-four of the haplotypes were found in five or fewer (<2%) of the samples (see Figure 6). Phylogenetic analyses supported the existence of two major clades, which I will call A and B . The bootstrap value separating the two clades was 87% using both parsimony and distance programs (Figures 7, 8a,b). Sequence divergence among haplotypes ranged from 0.1- 1.75%. Divergence between the two clade groups ranged from 0.71-1.75%, with an average of 1.03%. Using a molecular clock of between 0.8-2% divergence per million 39 Table 5. Presence-absence matrix of restriction sites found from different restriction enzymes in samples of O. mykiss. A "1" signifies a cutsite, a "0" signifies the absence of that cutsite. Cutsites were inferred from the restriction fragments, and are listed below in no particular order. A 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 B 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 C 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 D 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 E 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 A v a i l A 1 1 1 0 0 B 1 1 1 1 0 C 1 1 1 0 1 Bfal A 1 1 1 1 1 1 1 1 1 1 1 1 1 1 B 1 1 1 1 1 1 1 1 1 1 1 1 0 1 C 1 1 1 1 1 1 1 1 1 1 1 0 1 1 D 1 1 1 1 1 1 1 1 1 1 1 1 1 0 BstUI A 1 1 1 0 B 1 1 1 1 Ddel A 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 B 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 C 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1 Dpnll A 1 1 1 1 1 1 1 1 1 0 1 1 B 1 1 1 1 1 1 1 1 0 0 1 0 C 1 1 1 1 1 1 1 1 0 1 0 0 D 1 1 1 1 1 1 1 1 0 0 0 0 Haelll A 1 1 1 1 1 1 1 1 1 1 1 0 0 B 1 1 1 1 1 1 1 1 1 1 0 0 0 C 1 1 1 1 1 1 1 1 1 0 1 0 0 D 1 1 1 1 1 1 1 1 1 0 0 0 0 E 1 1 1 1 1 1 1 1 1 1 1 1 0 F 1 1 1 1 1 1 1 1 1 1 1 0 1 G 1 1 1 1 1 1 1 1 0 1 1 0 0 Hhal A 1 1 1 1 1 1 1 B 1 1 1 1 1 1 0 Hindi A 1 1 1 0 0 0 B 1 1 1 1 0 0 C 1 1 1 0 1 0 D 1 1 1 0 1 1 Hinfl A 1 1 1 1 1 1 1 1 1 0 1 B 1 1 1 1 1 1 1 1 1 1 1 Mspl A 1 1 1 1 1 1 0 0 B 1 1 1 1 1 1 1 0 C 1 1 1 1 1 1 0 1 Nlalll A 1 1 1 1 1 1 1 1 1 1 B 1 1 1 1 1 1 1 1 1 0 C 1 1 1 1 1 1 1 1 0 1 Rsal A 1 1 1 1 1 1 1 0 0 0 B 1 1 1 1 1 1 1 1 0 0 C 1 1 1 1 1 1 1 1 1 0 D 1 1 1 1 1 1 1 1 0 1 TaqI A 1 1 1 1 1 1 1 1 1 B 1 1 1 1 1 1 1 1 0 40 Table 6. The 32 composite haplotypes found in O. mykiss samples using 14 restriction enzymes. The symbols refer to the digestion patterns (Table 4) from the following enzymes: Alul, Avail , Bfal, BstUl, Ddel, Dpnll, Haelll, Hhal, Hindi, Hinfl, Mspl, Nlalll, Rsal, Taql 1 A A A A A A A A A A A A A A 2 A A A B A A A A A A A A A A 3 A A A A A A A A A A C A A A 4 C A A A A A A A A A A A A A 5 B A A A A A A A A A A A A A 6 A A A A A A A A A A C A A A 7 A A A A A A A D A A A A A A 8 A A A A A A A A A A B A A A 9 A A A A B A A A A A A A A A 10 A A A A A A F A A A A A A A 11 A A A A A A A A A A A A D A 12 D A A A A A A A A A A A A A 13 A A A A A B A A A A A A A A 14 A A A A A A A B A A A A A A 15 A A A A A A A A A A A B A A 16 A A A A A A A A A A A A B A 17 A A A A A A B A A A A A A A 18 A A A A A A B A A A A B A A 19 A A A A A A B A A A A A B A 20 A A A A C A A A C A A A C A 21 E A C A B A A A A A A A A A 22 A A B A A A E A A A A B A A 23 A C A A A A G A A B A B A B 24 A B A A A C C A A B A A A B 25 A B A A A C C A A B A B A B 26 A B A A A C C A B B A A A B 27 A B A A A C A A A B A A A B 28 A B A A A C D A A B A A A B 29 C B A A A C C A A B A A A B 30 E B D A A C C A A B A A A B 31 B B D A A C C A A B A A A B 32 A B A A A D C A A B A A A B 41 150 t/3 mtDNA RFLP Haplotypes Figure 6. The number of each composite haplotype found in the total sample of 292 O. mykiss mtDNA samples. 42 11 h i - 4 r5 9 3 • 14 6 12 21 A 10 2 87% 16 17 19 13 15 22 20 24 r - C 23 • 26 28 29 31 25 I — 32 27 •30 B 0.01 "d" Figure 7. Distance-based phylogeny of mtDNA RFLP composite haplotypes. Genetic distances were calculated in REAP's D program (McElroy et ai, 1991), using genetic distance as calculated by Nei and Tajima (1981; Nei and Miller, 1990, eq. 4) and run through PHYLIP's Neighbor (neighbor-joining) to produce a tree (Felsenstein, 1993). Genetic distances among haplotypes (see Table 6 for designations) are to scale. 43 55 54 r 23 13 " 10 " 6 2 11 14 19 57 16 12 21 18 17 15 • 22 20 7 87 32 28 26 24 29 30 31 -25 27 Figure 8a. Consensus phylogeny of mtDNA composite haplotypes (see Table 6 for haplotype designations) using distance. Boot and Neighbor (neighbor-joining) were used to produce a tree from 100 bootstraps. Only bootstrap support above 50% was reported. 44 87 51 32 72 75 28 29 30 31 27 23 51 20 " 7 18 17 14 13 9 8 11 10 - 22 " 15 12 21 24 25 26 19 16 B Figure 8b. Consensus phylogeny of mtDNA composite haplotypes (see Table 6 for numbers at branch termini) using parsimony. Dollo parsimony was used to construct the tree of R F L P haplotypes based on 100 bootstraps. Only bootstrap support above 50% was reported. 45 years (McKay et al., 1996; Bagley and Gall , 1998) and 1.03% average divergence between clades, this corresponds to an estimated divergence time between the clades of 515,000-1,288,000 years ago. The diagnostic analysis was conducted to better resolve geographic distributions of the two clades. Clade A was found in every major watershed except the upper Liard, while clade B was mainly restricted to coastal watersheds. Clade B was absent in the Columbia, the Snake, the Athabasca, and Kamchatka and was found in only two samples of 25 from Alaska (Table 7, Figs. 9, 10). Genetic diversity and cluster analysis among regions I grouped sample sites into their respective watersheds where possible to test for diversity on a regional scale (Table 7). Haplotype diversity was high in many watersheds, but it was particularly high in several regions presumed to have been ice-free in the last glaciation, including the Columbia and Snake rivers, Alaska/ Kamchatka and the Queen Charlotte Islands (see Figure 11) (McPhail and Lindsey, 1986). Interestingly, this high level of haplotype diversity was often accompanied by low nucleotide diversity (relative to nucleotide diversity in other watersheds), indicating that the many haplotypes located in these watersheds were relatively closely related to one another. In contrast, four watersheds had lower haplotype diversity than the rest (see Figure 11): the upper Fraser, Peace, and Athabasca rivers, and the North Coast/ Yukon region, as expected for recently colonized regions. In these watersheds, nucleotide diversity was relatively high, indicating few haplotypes with high divergence among them. These watersheds were all presumably glaciated until 15-10,000 years ago, and this pattern was consistent with 46 Mainland Figure 9. Distributions of mtDNA RFLP clade A and clade B haplotypes in Oncorhynchus mykiss in British Columbia. The area of the pie corresponds to the relative number of both composite and diagnostic samples taken, with the smallest sample (Okanagan) equaling seven individuals. 47 Figure 10. Distributions of mtDNA RFLP clade A and clade B haplotypes in Oncorhynchus mykiss in the North Pacific. The area of the pie corresponds to the relative number of both composite and diagnostic samples taken, with the smallest pie (Oregon) equaling 7 individuals. 48 Table 7. Clade A and Clade B composite and diagnostic haplotypes within various regions. Composite haplotypes were digested with 14 restriction enzymes whereas diagnostic haplotypes were digested with 2 enzymes (Alul and Dpnl 1 or Haelll and Dpnll) to determine clade identity. Localities without asterisks were analyzed for a composite haplotype, those with asterisks were analyzed for diagnostic haplotypes. These site groupings correspond with Figures 9 and 10. Regions (Localities) A B Regions (Localities) A B OREGON 1 LOWER FRASER 13 3 Alsea R. 1 Anderson R. 2 Emory Ck.* 1 3 PUGET SOUND 9 11 American Ck.* 5 Tahuya R. 3 2 Bridge R.* 5 Bogachiel R.* 4 Nooksak R.* 1 3 MIDDLE FRASER 35 1 Deer Ck* 3 2 Fish L. 6 Tucannon R.* 1 Tzenzaicut L. 5 Lewis R.* 1 Puntzi Ck.* 5 Euchiniko R.* 4 1 N. VANCOUVER ISLAND 6 10 Chilko/ Chilcotin R.* 5 Keogh R. 1 4 Gaspard Ck.* 4 Mahatta R. 3 2 Baezaeko* 4 Klootchlimmis Ck. 2 4 Coglistiko* 2 S. VANCOUVER ISLAND 13 20 NICOLA/ THOMPSON 18 6 Cowichan R. 3 2 Pennask L. 5 Lucky Ck. 1 4 Bonaparte R. 2 3 Elsie L. 5 Coldwater R.* 4' 1 Bamfield 1 Deadman R.* 4 Robertson Ck 3 2 Spius Ck.* 3 2 Quinsam R. 4 1 Junior L.* 3 NECHAKO R. (upper Fraser) 6 6 Efflingham R.* 2 2 Cheslatta L. 4 1 Tahtsa R. * 5 LOWER MAINLAND 10 11 Brewer's Ck.* 2 Chehalis R. 2 1 Capilano R. 1 4 UPPER FRASER 16 6 Vedder R. 1 4 Bivouac Ck. 5 Squamish R.* 1 2 Dome Ck. 5 Abbotsford* 5 Baptiste Ck.* 4 Forfar* 2 1 CENTRAL COAST 7 8 Hautete Ck.* 5 Dean R. 4 1 Ashlulm Ck.* 2 3 SKEENA RIVER 28 11 Abuntlet L.* 1 4 Babine R. 3 2 Morice R. 3 2 KITIMAT 11 2 Canyon Ck. 3 2 Kitimat R. 3 2 Goathorn Ck. 4 1 Union L.* 5 Toboggan Ck. 3 2 Khtada L.* 3 Kispiox R.* 4 1 Sustut R.* 4 1 Natlan Ck.* 4 49 (Table 7, cont'd.) Regions (Localities) A B QUEEN CHARLOTTES 4 19 Honna R. 5 B V M 2 3 Brent Ck.* 2 B V M * 3 Lower Phantom Ck.* 8 NORTH COAST 17 8 Tahltan R. 4 2 Taku R. 4 2 Disella L.* 2 3 Iskut R.* 5 Taku R.* 2 1 UPPER LIARD 0 11 Kutcho Ck. 5 Wolverine* 3 Blue Sheep L.* 3 YUKON 10 0 Mclntyre 5 Kathleen R. 5 ALASKA 23 2 Karluk R. 7 Natnek R. 6 Nushagak R.* 3 2 Neguthluk R.* 7 KAMCHATKA 11 0 Kavachina 5 Snotolvayam 6 OKANAGAN 7 0 Goatskin Ck. 2 Goatskin Ck.* 3 Fall Ck.* 2 Regions (Localities) A B NASS RIVER 5 10 Meziadin R. 5 Paw L.* 5 Niska Ck.* 5 WILLISTON RESERVOIR 9 12 Firesteel R. 3 Mesilinka R. 3 3 Nation R. 5 Ingenika R.* 1 1 El Condor* 3 Firesteel R.* 2 PEACE RIVER 20 5 Chowade R. 5 Graham R. 5 Kiskatinaw R. 5 Moose L. 5 Burnt R.* 1 Sekunka R.* 1 Elizabeth Ck.* 3 ATHABASCA RIVER 15 0 Wampus Ck. 5 Halpeny Ck. 5 Cabin Ck.* 5 KOOTENAY L A K E 23 0 Bayonne Ck. 4 Lardeau R. 5 Procter Ck. 5 Fry Ck.* 4 Summit Ck.* 5 ARROW LAKE 12 0 Halfway R. 5 HillCk 5 Tumtum Ck. 2 SNAKE RIVER 14 0 Dworshak R. 7 Selway R. 7 Total (all samples) 349 162 I Haplotype Diversity | | Nucleotide Diversity Figure 11. Diversity measures of O. mykiss mtDNA RFLP haplotypes in order of largest to smallest in haplotype diversity. Diversity was measured in watersheds throughout the region sampled using the program DA from REAP (McElroy et al., 1992). Haplotypic diversity is a probability measure reflecting the number of different haplotypes in a region whereas nucleotide diversity is a composite measure reflecting both number of different haplotypes and divergence among haplotypes. Asterisks signify putative refugia (McPhail and Carveth, 1992). Error bars represent standard error as computed by DA (McElroy et al, 1992). 51 recolonization from more than one refuge, if the refugial groups were characterized by distinct clades. Population cluster analysis was based on the population units defined in Table 8. The population cluster analysis suggested that the major geographic division was between two groups: the Queen Charlotte Islands, Vancouver Island and Lower Mainland in one, and the Columbia, Fraser, Peace, Skeena, Taku, and Stikine rivers in the other (see Fig. 12). Within these major geographic groups, however, gst distances were negative, with the exception of the Snake River population. Monte, however, resulted in significant differences at all nodes with the exceptions of Vancouver Island and the Lower Mainland and the Peace and Skeena rivers. The "0" branch lengths indicated the lack of significant differences at these (see Figure 12). The three methods of subdividing the region in the A M O V A analysis resulted in similar levels of among group (<E>ct) variation. In the first method, I partitioned watersheds into two groups, the Queen Charlotte Islands, Vancouver Island, and the Lower Mainland versus the Snake, Columbia, Fraser, Peace, Skeena, and North Coast watersheds. This was partitioned according to the major division in the population cluster analysis, Nucleodiv, results, and was referred to as Nucleodiv. In the second method, I partitioned watersheds according to Behnke's (1992) coastal and inland division, such that the North Coast and Skeena are grouped with the Queen Charlotte Islands, Vancouver Island and Lower Mainland. The inland group was composed of the Snake and Columbia rivers, the Fraser, and the Peace. This was referred to as Behnke. In the third method, I formed three groups designed to isolate the groups most likely to be divergent and one intermediate group, namely the Queen Charlotte Islands/ Vancouver Island/ Lower Mainland group, the Columbia/ Snake River group, and the Fraser/ Skeena/ Peace/ North Coast group. This was referred to as Refugia. The similarity found among the three subdivisions 52 Table 8. Composite haplotypes of O. mykiss mtDNA found within watershed groups. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 V I 9 3 2 L M F R A Q C I S K N C Y P E A A T H C O L S N A A 7 K O R E O K 21 2 32 17 17 19 10 10 15 11 3 2 3 3 1 4 2 3 2 42 38 10 40 27 34 10 26 14 24 VI = Vancouver Island LM= Lower Mainland FRA= Fraser River (above the Fraser Canyon) QCI= Queen Charlotte Islands SK= Skeena, Kitimat, Nass, Dean rivers NCY= Tahltan, Taku, Liard, Tatshenshini rivers PEA= Peace River, Williston reservoir ATH= Athabasca River COL= Kootenay and Arrow lakes SNA= Snake River A/K= Alaska/ Kamchatka ORE= Alsea River, Oregon OK= Okanagan 53 Vancouver Island (5) Lower Mainland/ Puget Sound (4) Queen Charlotte Islands (2) gst=0.165 Skeena/Nass/Dean/Kitimat (8) st=0.03 Peace River (7) Upper Fraser River (9) North Coast/Yukon (5) Columbia (6) Alaska/Kamchatka (4) Snake River (2) Figure 12. Population cluster analysis among watersheds using Nucleodiv (Holsinger and Mason-Gamer, 1996). Positive gst values, indicated at two nodes, were significant at p<0.01. Branching patterns at other nodes represent significant differences using the Monte Carlo simulation analyses (using Monte by Roff and Bentzen, 1989). All nodes were significant except for the Vancouver Island/ Lower Mainland node and the Peace River/ Skeena River node ("0" branch lengths) which were not significantly different from one another. The number of localities sampled in each watershed group is indicated in parentheses. 54 with respect to among group variation demonstrated the ambiguity in delineating a border between "coastal" and "inland." Specifically, when the north coast and the Skeena watersheds were included in "coastal," a similar level of among group variation was found as when they were included in the "inland" group (see Table 9). Despite the difficulty in delineating a clear border between coastal and inland groups, however, regional genetic structure was apparent in rainbow trout. The among group level of variation accounted for a significant amount of the total variation, and it accounted for higher levels of variation than among watersheds within group variation in all AMOVA tests (Table 5). The majority of genetic variation in the AMOVAs, however, was found within populations (1-<E>st). Even at the watershed level, where "within population variation" referred to within a sample of five to ten individual trout, the within population variation component was still the largest. These results attest to the diversity of haplotypes on a very local level. Historical demography The nested haplotype network resulted in two main clusters of haplotypes, corresponding to the "A" and "B" clades from the phylogenetic analysis (Figure 13, also Figures 7, 8a, 8b). Haplotypes 1 and 24 had the greatest number of mutational connections and were the most abundant haplotypes, comprising 45% and 12% of the samples, respectively. Both the number of mutational connections and the abundance supported their placement as interior (i.e. ancestral) haplotypes for the A and B clades, respectively (Crandall and Templeton, 1993). All distances calculated by Temple were considered significant if they were either larger than 97.5% or smaller than 2.5% of the permutation results. Due to the repeated analyses inherent in this test (112 tests of significance), 6 significant results were expected by chance alone (see Figure 14 for graphical 55 Table 9. Partitioning of genetic variation in British Columbia using an A M O V A analysis. B .C. was divided in four ways (see below) to determine how this affected the partitioning of variation. An A M O V A was performed on each of these, calculating Oct, the proportion of variation among groups; Osc, among populations within groups; and 1-Ost, the proportion of variation within populations. Structure Hierarchical Levels % Variation P-values Watersheds Among watersheds 17.48 % P=0.00 Among pops within watersheds 26.30 % P=0.00 Within populations 56.22 % P=0.00 Nucleodiv Among watersheds 20.99 % P=0.03 Among pops within watersheds 10.43 % P=0.00 Within populations 68.58 % P=0.00 Behnke Among watersheds 20.12 % P=0.02 Among pops within watersheds 9.08 % P=0.00 Within populations 70.80 % P=0.00 Refugia Among watersheds 22.04 % P=0.00 Among pops within watersheds 6.37 % P=0.00 Within populations 71.59 % P=0.00 Watersheds: (10 groups) (n=260) Vancouver Island: Keogh, Cowichan, Quinsam, Somass, Lucky, Elsie, Bamfield, Klootchlimmis, Mahatta (n=42); QCI: Honna, BVM (n=10); Lower Mainland/ Tahuya: Chehalis, Capilano, Vedder, Tahuya (n=18) Upper Fraser: Anderson, Fish, Tzenzaicut, Pennask, Bonaparte, Cheslatta, Bivouac, Dome (n=38) Skeena/Nass/Kitimat/Dean: Morice, Babine, Canyon, Goathorn, Toboggan, Meziadin, Kitimat, Dean (n=40) North Coast/Yukon: Tahltan, Taku, Kutcho, Mclntyre, Kathleen (n=27) PeaceAVilliston: Chowade, Graham, Kiskatinaw, Moose, Firesteel, Mesilinka, Nation (n=34); Athabasca: Wampus, Halpenny (n=10) Columbia: Halfway, Hill, Tumtum, Procter, Bayonne, Lardeau (n=26) Snake: Dworshak, Selway (n=14) Nucleodiv: (2 groups) Coastal: Queen Charlotte Islands, Vancouver Island, Lower Mainland/ Tahuya (n=70) Inland: Skeena/Nass/Kitimat/Dean, North Coast/Yukon, upper Fraser, PeaceAVilliston, Athabasca, Columbia, Snake (n=190) Behnke: (2 groups) Coastal: Lower Mainland/Tahuya, Vancouver Island, QCI, Skeena/Nass/Kitimat/Dean, North Coast/ Yukon (n= 137) Inland: upper Fraser, PeaceAVilliston, Athabasca, Columbia, Snake (n=123) Refugia: (3 groups) Coastal: Lower Mainland/Tahuya, Vancouver Island, Queen Charlotte Islands (n=70) Central: Skeena/Kitimat/Nass/Dean, North Coast/Yukon, upper Fraser, PeaceAVilliston, Athabasca (n=150) Inland: Columbia, Snake (n=40) 56 1-4 Figure 13. Nested network of composite haplotypes using the nesting method described in Templeton et al. (1995). Clade A haplotypes are represented by gray shaded circles and clade B haplotypes by black circles. Lines segments (i.e. either a single line or the number of line segments between small circles) indicate single mutational differences between haplotypes. The small circles, in effect, indicate missing haplotypes. The nesting method works as follows: individual haplotypes are considered 0-step clades; haplotypes separated by a single mutation are grouped into 1-step clades (1-1, 2-1, 3-1, ...10-1); 1-step clades separated by a single mutation are grouped into 2-step clades (1-2, 2-2, ... 5-2); and 2-step clades are grouped into 3-step clades (1-3 and 2-3). The last group includes all the haplotypes in a single 4-step clade (1-4). 57 Figure 14. Geographical summary of the significant results from the nested clade analysis of O. mykiss in B.C. Al l circles represent those haplotypes with significantly small distributions and are located in the approximate center of their sampled distribution. Dashed lines indicate significantly large nested distances. The overall pattern for clade A (gray circles) is that of restricted gene flow as well as some long distance dispersal into the Columbia River. The overall pattern for clade B (black circles) is that of restricted gene flow mainly along the coast. 58 summary of significant results from the permutation analysis). Some support existed for all three mechanisms leading to genetic and geographical associations, however, the results did not fit any of the predicted patterns perfectly. In clade A (Table 10a), at the 0-step clade level, significantly small clade distances were found for several individual haplotypes, including 1, 2, 3, 15 and 18 (see Table 10a). However, despite large nested clade distances for both 9-0 (haplotype 9), located in the Snake and Fraser rivers, and 7-0 (haplotype 7), located in the Yukon, no nested clade distances (Dn) were found to be significantly large so neither restricted gene flow nor range expansion could be inferred. At the 1-step clade level, however, the tip clades 2-1,5-1 and 6-1 all had significantly small clade distances and for clade 2-1, this contrasted with a significantly large nested clade distance. This is suggestive of range expansion for the 2-1 tip clade according to Templeton et al. (1995). In addition, the significantly small Dn(I-T) value in box 1-2 also supports the inference of range expansion because it suggests that tip clades have expanded significantly more than interior or "ancestral" clades. The significantly large Dc(I-T), however, suggests restricted gene flow because tip clades have restricted distributions. What is somewhat troubling about this inference is that the tip clades in box 1-2 (2-1, 3-1, 5-1, and 6-1) are predominantly located in the Columbia River, yet the more ancestral clade group 1-1 has a very widespread distribution in the province. In other words, the results suggest that restricted gene flow exists in the Columbia River and that range expansion has occurred into the Columbia River. This is an unexpected and likely erroneous inference that I will discuss later. Overall, however, the significant clade and nested clade distances in clade A failed to fit any pattern perfectly and instead reflected a combination of inferences. One difficulty with the analysis is that predictions for the different mechanisms being tested are in direct opposition to 59 CO u cd >- « u o • 2 S •S 3 13 ^ g . a T 3 O o co c -2 o f Cd CO H CO a E c Oi § CO > cd co 1 5 .ts § 2 <u o co 2 "a, b 5 « | 8 <u E c ~ 6 a ° i t i o <u _ 2 C ^ O f Id "8 CO T 3 CO CO c ^ O ^ .S >• - J2 Sj cd § -cd o > cd 5 U 1) •8 a, • CO CO o > Q *-> CO* & 8 co CO co co -a co X o 2 ^ CO cd CO * H o •° e &, o o "2 . co s ^ ~ L . >^  22 13 O ^ cd cd Q , 1 ° .22 « co - a co s i i s - -o 1 cd -it* _ h co <£> CO r -a, c cs 22 "S ^ H c CO C « Cd CO rt - £ » X S ~ <= a, 3 J= 3-- cd § § «s C ^ cd w ° " •a c cd CO CO a. CL, cd CO o o i CO cd 4= CO cd a. » x CO c c o c o > CO T 3 '•s CO T3 a « 2 • c IS a. cd cd s cd "S CO CO z cd o cd H 8 & | o 1) CO cd O U u CO "p, CO C <N e > S C L P s? co o 5 cd ^ -S o •s ° d -S a, o - s t- K O -o c cd co C ,s a M a a « ? 5 : cn o cd O cd CO o oc c o o ^ « <a feb co cd ? H 00 >~, CO OH cd C CO CJ CO CO CO 60 c cd CO cd CO o • O O CN VO Tj-© VO 1/1 •* ^ 3 N CS o — Ov • m o i/i m © 4 ° ^ © T J -' O h d — • O ON VO Ov m o m vo it. c^  n ° n ^ ^ >o © T t -J , N m T t vo n vi T ^ • o CV CN H 5 ^ i C) — f, J T t 3 u-i c i I T t fl T t T t T t T t — T t r- T t T t — Cl o T t a> _ . — =f u c c © Q Q Q Q ft E-» o c "o 'c A Q Q Q Q H tr <N Q Q Q Q ci Q Q <u o c ca TJ o Q e > - O 1 cn <u • -o a rj fl « _ i ^ - t. cn cn S a ^ CU cj s ca NO CU .52 60 C- TJ - ca •S a. _ cn s ci cU S 2 -5 a, ca <U U c O 2 c =2 •=! -a « cu ca cu a. a. ca ^ E o ca •a-s - C cU ca >, cu 60 •S M i >- .a ^ .I1 £ cn cn ca > i cn 5 « ca cj If 5 S ON to ca <u o <2 c • £ ca '33 T J u 5 Jo rt o Q CU JS H _E CU i . 1 "cu ca y cn 2 cu cn JS ca "3 T J CU ca ca ON .23 <u ^ Cd u—4 _ ° S Q 2 u •o «h r cu cn O o J '"3 -3 8 P3 c . O cu O T J 22 ca o TJ s 1 ta u o <u iS ca cn <U PQ -5 TJ cu o E & 8 _g cn § » \B ca cn •a ^ D- S ca cn 60 CU < 60 c o TJ > ca cu o Q cn Qc CU tU 60 ca l-H B W ON CU o eR 2 tN cn cu la « B f j • - 60 ca CU Z xi o CU (T^ 13 ^ — G, 4? * o CU <u CJ E ca cu to Q* cn ca cu ^ B ti u 3 c c ca C u CU ' 3 cn .2 JZ E « G <u <u 4J lu E o a o B 8 O X „ 2 1 < o o ao o O —• ts O ' ON ' ON \r} e — — 2 ^ ^ cn t/3 ^ CN ON g. OS 0 0 ' ND ao ON j^-ON CN CN c^l CN 15 ON ON V t S CN Tt tS t N a e Q Q Q O « o c u c - i O Q O Q V f u c E7 [7 Q O cu CJ C one another, yet are theoretically not mutually exclusive. Therefore, if none of the mechanisms are particularly well supported, it is unclear whether this is because the hypotheses were incorrect or because multiple mechanisms were involved. In clade B (Table 10b), at the 0-step level, several significantly small clade distances were found for tip clades, including 25-0, 26-0, 27-0, and 28-0. This in combination with the significantly large Dc(I-T) suggests that tip clades have much smaller distributions than the interior clade, so restricted gene flow is inferred in this group. Basically the same pattern exists for the 1-step clades in box 4-2, in that the ancestral clade has expanded further than the derived ones (except that the interior clade 8-1 also has a significantly small distribution). In any case, the significantly small haplotypes in box 8-1 (25, 26, 27, 28) are all coastal and three of the four are located in the Queen Charlotte Islands and Kitimat region, and one on Vancouver Island. The inference is, therefore, that restricted gene flow exists among coastal populations, the Queen Charlotte Islands, Kitimat and Vancouver Island because these derived haplotypes have restricted distributions relative to the ancestral haplotype in the group. The same inference is made in box 4-2 in that an ancestral group is more widespread than a derived one. In this case, clade 9-1 (haplotype 31) is located mainly in the center of the province, in the Skeena and Peace rivers, and its distribution is more restricted than its "ancestral" clade. The overall inference for clade B is that of restricted gene flow, largely because ancestral haplotypes have a wider distribution than derived ones, and most clade groups have restricted distributions. In some ways, this is not a fair inference because the whole of clade B has a smaller distribution than clade A (indicated by the significantly small distribution of clade 2-3, comprising all the clade B haplotypes). Clade and nested clade distances for clade B are, therefore, dwarfed in the permutation analysis by all the localities where only clade A was found. The nested clade 62 analysis is probably more appropriate for range expansion from a single refuge rather than from multiple refugia. Finally, evidence for allopatric fragmentation was inferred between the two major clade groups, A and B. Clades A and B both had a larger than average number of mutations distinguishing them based on a visual assessment of the network (Figure 13) and a visual assessment of the haplotype phylogeny (Figures 7, 8a, 8b). The size of the branch length between the two clades (Figure 7) and the high bootstrap values separating the two clades (Figures 8a, 8b) both indicate that these two clades represent the greatest largest genetic divergence among haplotypes. That clades A and B are associated with particular geographical regions also supports the case for allopatric fragmentation (Figure 9). However, not all the predictions for allopatric fragmentation were met. Templeton et al. (1995) suggested that allopatric fragmentation should result in small clade and nested clade distances particularly at higher clade levels, and instead, clade and nested clade distances tended to level out in these data. Nevertheless, the relatively large genetic divergence between the clade groups is still suggestive of past fragmentation. Taxonomy Given the presence of both A and B mitochondrial types throughout the range of O. mykiss and in the ranges of both coastal and redband putative subspecies, coastal and redband groups are clearly not monophyletic for mitochondrial DNA (see Figure la,b and haplotype list, Table 7). However, to assess whether these subspecies groups were still distinguishable from one another at the population level, redband and coastal designations were placed onto the Nucleodiv population cluster analysis. The result was that the population tree (see Figure 15) did not support genetic distinction between the subspecies. Skeena and Peace river trout were not found 63 to be significantly different from one another, despite their categorization as separate subspecies (Behnke, 1992; Figure la,b, Ch.l). Furthermore, north coast samples did not form a cohesive group with other coastal populations (Vancouver Island, Queen Charlotte Islands, Lower Mainland, Skeena) despite their classification as part of the same subspecies (Behnke, 1992). Distinguishing between "coastal" and "inland" populations with mtDNA was too ambiguous to warrant subspecies designation. Furthermore, given historical gene flow, a strong role for natural selection rather than common ancestry is suggested in maintaining the phenotypic differences between the coastal and redband. Because subspecies should reflect common ancestry rather than natural selection and because they should be distinguishable from one another at the population level, coastal and redband subspecies groups were rejected by my analyses. Adaptive evolution The AMOVA test of life history evolution resulted in evidence for parallel evolution. When populations were grouped according to life history type, among group variation was not significant, whereas among group variation based on geographical locations was (Tables 11, 12). That geographical location explained more genetic variation than life history type supports the hypothesis of parallel evolution for these life history types. Similarly, no obvious association between mtDNA variation and phenotypic variation was found when phenotypic characters were mapped onto a mtDNA phylogenetic tree and the population cluster analysis (Figures 16a,b). This analysis demonstrated that distinct phenotypes such as the large-bodied, piscivorous Gerrard trout, steelhead trout and small insectivorous rainbow trout all share some of the same mtDNA haplotypes (Figure 16a). The Gerrard rainbow trout have haplotypes that are more closely related to those from steelhead in the Snake River than to large-bodied piscivorous yellow fin trout from Arrow Lakes. Similarly, both steelhead 64 gst=0.165 Vancouver Island (5) Lower Mainland/ Puget Sound (4) Queen Charlotte Islands (2) gst=0.03 Skeena/Nass/Dean/Kitimat (8) Peace River (7) I— Upper Fraser River (9) North Coast/Yukon (5) Columbia (6) Alaska/Kamchatka (4) Snake River (2) Coastal = Redband = Figure 15. Coastal and redband putative subspecies (drawings from Behnke, 1992) are mapped onto a population cluster analysis (from Figure 12; Holsinger and Mason-Gamer, 1996). The placement of coastal (set to the right) and redband putative subspecies demonstrates that major population genetic clusters of watersheds are not reflected by present subspecies designations. 65 Table 11. A M O V A analysis for assessing the evolution of anadromy (or non-anadromy). The proportion of genetic variation defined by life history relative to geography was calculated using two A M O V A tests. In the first test, populations were subdivided according to life history type, i.e. whether they were anadromous or non-anadromous, whereas in the second test, populations were subdivided according to geographical locations. The percentage variation, proportion of variation and significance values were calculated for among groups (Oct), among populations within groups (Osc), and within populations (1-Ost). Subdividing according to life history does not result in significant differences among groups whereas subdividing according to geography does. Major group Hierarchical structure % Variation P-value Life history Among life history groups 0 0.39 Among pops within groups 37.00 0.00 Within populations 63.00 0.00 Geography Among geographic groups 28.39 0.00 Among pops within groups 11.69 0.00 Within populations 59.91 0.00 By Life History: Anadromous: Keogh, Cowichan, Quinsam, Somass, Chehalis, Capilano, Bonaparte, Karluk, Morice, Babine, Kitimat, Dworshak, Selway (n=69) Non-anadromous: KJootchlirnmis, Lucky, Elsie, Vedder, Fish, Tzenzaicut, Pennask, Cheslatta, Bivouac, Dome, Canyon, Goathorn, Toboggan, Natnek, Bayonne, Procter, Gerrard, Halfway, Hil l , Tumtum (n=99) By Geographic regions Vancouver Island: Somass, Quinsam, Cowichan, Keogh, Lucky, Elsie, Klootchlirnrnis (n=36) Lower Mainland: Chehalis, Capilano, Vedder (n=13) Fraser: Fish, Tzenzaicut, Pennask, Bonaparte, Cheslatta, Dome, Bivouac (n=36) Skeena: Morice, Babine, Kitimat, Canyon, Goathorn, Toboggan (n=30) Alaska: Karluk, Natnek (n=13) Columbia: Bayonne, Procter, Gerrard, Halfway, Hill , Tumtum, Dworshak, Selway (n=40) 66 Table 12. AMOVA analysis for assessing the evolution of run-timing in steelhead. The proportion of genetic variation defined by run-timing relative to geography was calculated using two AMOVA tests (identical to Table 10). Again, subdividing according to life history does not result in significant differences among groups whereas subdividing according to geography does (at p<0.05). Major group Hierarchical structure % Variation P-value Life History Among different run times 8.29 0.21 Among pops within groups 15.66 0.04 Within populations 76.05 0.00 Geography Among geographic groups 19.79 0.01 Among pops within groups 6.51 0.02 Within populations 73.70 0.00 By Life History: Summer-run: Somass (n=5) Winter-run: Cowichan, Keogh, Quinsam, Chehalis, Vedder, Kitimat (n=28) Fall-run: Babine, Morice, Dworshak, Selway (n=24) By Geographic regions VI/ LM: Somass, Cowichan, Quinsam, Keogh, Chehalis, Vedder (n=28) Sk: Morice, Babine, Kitimat (n=15) Col: Selway, Dworshak (n=14) Figure 16a. Phenotypic and life history characters, above-barrier resident rainbow, sea-run steelhead, and large-bodied piscivorous rainbow trout, were placed on a mtDNA distance tree (from Figure 7). The presence of similar life history characters in distinct phylogenetic groups, A and B, indicates the possibility of parallel evolution of these phenotypic characters. In addition, the fact that several distinct phenotypes have common haplotypes (e.g. haplotype 1) indicates that populations with similar haplotypes do not necessarily have similar adaptive traits. Photos by E. Keeley (Idaho State University). 68 !st=0.165 Vancouver Island (5) Lower Mainland/ Puget Sound ( 4 ) "™ Queen Charlotte Islands ( 2 ) gst=0.03 Skeena/Nass/D/K (8) Peace River (7) Upper Fraser River (9) North Coast/Yukon (5) Columbia (6) Alaska/Kamchatka ( 4 ) Snake River ( 2 ) Above barrier rainbow = Steelhead = Large piscivorous rainbow = Figure 16b. Phenotypic and life history characters are placed on a population cluster analysis (from Figure 12; Holsinger and Mason-Gamer, 1996). Given the overlapping distribution of clade groups from 16a, this analysis was done to demonstrate that neutral variation is not associated with phenotypic variation. The clustering of geographically proximate populations indicates a much stronger association of neutral variation with geography than with phenotype. Photos by E. Keeley (Idaho State University). 69 and neotenous trout occur in both coastal and inland populations. These results suggest convergent evolution for all three characters. The major finding from this analysis was that geographic location accounted for genetic variation far better than did phenotype which suggests relatively recent (<10,000 yrs.) convergent evolution of phenotypic groups. DISCUSSION Glacial history The effect of glaciation on molecular diversity in organisms has been studied both theoretically and empirically. Source populations are thought to have higher diversity than recently colonized areas due to the successful colonization of only a subset of the genotypes from any particular source area (Templeton et al, 1995; Hewitt, 1996; Weider and Hobaek, 1997). Empirical evidence has shown many times that low diversity is associated with formerly glaciated areas and high diversity with formerly ice-free areas (e.g. Cann et al, 1987; Jaarola and Tegelstrom, 1995; Templeton et al, 1995; Hewitt, 1996; Merila, 1997; Soltis et al, 1997; Weider and Hobaek, 1997). Queen Charlotte Islands Evidence exists for a B.C. coastal refuge in both plant and animal species. Allozyme data were used to suggest that a coastal refuge was used by sockeye salmon, (O. nerka) (Wood et al, 1994); and mitochondrial data were used as evidence that the song sparrow, black bear, pine marten and weasel survived glaciation on the Brooks Peninsula and the Queen Charlotte Islands (Byun et al, 1997). In addition, alpine plant distributions also support the refugial hypothesis for this region (Ogilvie, 1989). In O. mykiss, high diversity in the Queen Charlotte Islands, for the B clade particularly, suggests that this area may have also been used by rainbow trout as a refuge during the last 70 glaciation. The distribution of clade B haplotypes in the province, particularly in the Skeena River and upper Williston, suggests that the Queen Charlotte Island refuge may have been a major source population for post-glacial colonization into British Columbia. Columbia River The Columbia River is presumed to be another main refugial area for the province. Many species including chinook (O. tshawytscha) and sockeye salmon (O. nerka) show genetic divergences between northern and southern populations attributed to the use of the Columbia River and Beringia as refugia during the last glaciation (Bickham et al., 1995; Taylor et al., 1999; Wood et al., 1994). Bull trout, Salvelinus confluentus, also shows evidence of isolation in and expansion from the Columbia River into northern B.C. (Taylor et ah, 1999). High diversity within clade A in the Columbia drainage populations suggested that this was a likely source for many clade A haplotypes in British Columbia. Beringia As indicated above, Beringia has also been shown to have been a major refuge, particularly among Pacific salmon (Bickham et al., 1995; Taylor et al, 1999; Wood et al, 1994). Though mtDNA diversity in rainbow trout samples from Alaska and Kamchatka was suggestive of a northern refuge (Figure 11), recolonization from Beringia of British Columbia was not apparent in the data. The lack of B clades in the Beringia samples suggested that it was not a major source for coastal B.C. populations and the similarity among North Coast watersheds in B.C. (Skeena, Peace, and Fraser rivers, see Figure 12) suggests that the same source populations recolonized them all. In summary, the genetic diversity found in northwestern B.C. is not unique and can be attributed to recolonization from the Queen Charlotte Islands and the Columbia River. Southwestern British Columbia 71 Relatively high levels of diversity in both mtDNA clades (Figure 7) on Vancouver Island and the Lower Mainland suggests that this was close to a refuge and may have been colonized by more than one refuge. The Wisconsinan ice sheet extended only as far as Olympia, Washington, and rainbow trout almost surely survived the last glaciation south of the glacier (McPhail and Lindsey, 1986). The Chehalis River, south of the Olympic Peninsula, is thought to have been used by rainbow trout as a glacial refuge due to their distinct chromosome numbers in this region (Thorgaard, 1983) and given its proximity to the Queen Charlotte Islands, colonization from the north was also possible. Allozyme frequencies indicate that rainbow trout from Puget Sound, the Lower Mainland, and Vancouver Island are typical of coastal trout (Parkinson, 1984), and minisatellite data corroborate that this region belongs to a coastal group (Taylor, 1995). Mitochondrial RFLP evidence strongly supports a Vancouver Island, Queen Charlotte Island, Lower Mainland group distinct from inland populations (Figure 12). Evidence exists for bull trout as well that southern coastal and inland populations survived in distinct refugia (Taylor et al, 1999). Therefore, if mixing occurred, it was most likely among coastal populations, i.e. south from the Queen Charlotte Islands and north from either the Chehalis or more southern coastal populations. Athabasca River Mitochondrial RFLP analyses revealed no evidence of uniqueness in Athabascan populations and instead supported the hypothesis that trout crossed over from B.C. into Athabasca post-glacially (proposed in Ch. 2). No evidence existed for hybridization with cutthroat either, however, but this will be discussed later. All haplotypes found in the Athabasca were of the type most common to British Columbia and were particularly prevalent in the Peace and upper Fraser rivers (see also Wilson et al 1984). Furthermore, low diversity in the 72 Athabasca suggests that trout recently colonized the area. Nahanni River Evidence also suggests that the Nahanni River, in the Northwest Territories was ice-free in the last glaciation (Lindsey and McPhail, 1986). However, the distribution of Oncorhynchus mykiss in B.C. suggests it was not an important refuge or source of post-glacial colonization for this species. Oncorhynchus mykiss exists only in the headwaters of the Liard system and is not found elsewhere in the northeastern corner of B.C. The similarity of these haplotypes to other coastal haplotypes suggests that these headwater populations were founded by a headwater exchange with the Stikine River. McPhail and Lindsey (1970) have hypothesized a cross-over event between the upper Stikine and upper Liard, and my mtDNA data are consistent with this hypothesis. Recolonization of British Columbia Species composition analyses and geological evidence both indicate that watershed exchange occurred between the Columbia River and the Fraser River after the last glaciation (McPhail and Lindsey, 1986). Both the Kettle drainage in the Okanagan and the Thompson drainage in the central interior were potential sites for the cross-over (McPhail and Lindsey, 1986; Nelson and Paetz, 1992). That haplotypes in these populations are mainly of type 1 (clade A) indicates that fish bearing haplotype 1 likely came from the Columbia. Species composition and geological evidence also suggest that exchange occurred between the Fraser and the Peace and between the Fraser and the Skeena rivers (McPhail and Lindsey, 1986). The similarity in the Fraser, Peace and the North Coast, at least in the predominance of mtDNA clade A, suggests that they all originated from common source populations. McPhail and Lindsey (1986) documented 84% faunal similarity based on freshwater fish distributions between the Columbia and Fraser 73 River, 80% similarity between the Columbia and the Nass River along the north coast of B.C, and 71% between the Columbia and the Stikine River in northwestern B.C. This is a high level of similarity in comparison to the 35% and 40% faunal similarity between the upper Columbia and the upper Snake and Wood rivers (both tributaries to the lower Columbia), respectively (McPhail and Lindsey, 1986). The Columbia River was clearly a dominant source of fish fauna into what is called Cascadia, the region from the Columbia to the Stikine, west of the Rocky Mountains (McPhail and Lindsey, 1986). On the other hand, the Columbia and Snake rivers, and much of the central Fraser River lacked clade B haplotypes. The presence of clade B haplotypes in most of the province's rivers (Skeena, Peace, upper Fraser, Taku, Stikine) suggests that most of British Columbia was invaded by both coastal and inland populations. High diversity for the B clade and a relatively high proportion of B haplotypes in coastal populations suggest that B haplotypes nearly reached fixation in a coastal refuge and that haplotypes in clade B originated on the coast. Again, the diversity in the central part of B.C. is low relative to that on the coast for the B clade suggesting that B haplotypes came from a coastal refuge from which it colonized the Skeena, upper Fraser and Peace rivers, as well as the North Coast. Partitioning of genetic variation Most of the genetic variation in British Columbia is partitioned between the coast and the interior (Table 8). Intraspecific genetic variation is partitioned within and among these groups such that the majority of diversity occurs within populations, approximately 69-72% (see Table 8). This is common for rainbow trout and for salmonids in general. Allendorf and Leary (1988) found that 85% and Hershberger (1992) found that 92% of allozyme variation in Washington rainbow trout was due to within-population variation. 74 Given rainbow trout's history of isolation followed by extensive dispersal and gene flow among these groups, it is not surprising that large differences exist among haplotypes and that relatively weak structuring exists among regions. In the southern extent of the range, higher levels of among-population variation relative to within-population variation were found (Bagley and Gall, 1998). Specifically, Bagley and Gall (1998) found that 66% of mtDNA variation exists among populations within California and 26% within. In contrast, I found 20-22% among coastal and inland groups and 69-72% within watersheds. This likely reflects the different glacial histories of the northern and southern portions of the range of O. mykiss. Populations in the south were not directly affected by the last glaciation as intensely and, therefore, were not subject to the same massive expansion and watershed exchanges that led to large range sizes in British Columbia. Populations in the south have likely had more time to drift to different frequencies, evidenced by the higher among population variation. Historical demography Overall, the fact that range expansion did not emerge more strongly from the nested clade analysis was surprising given the recolonization of a previously glaciated landscape. In the analysis, range expansion is inferred when clade distances (Dc) or nested clade distances (Dn) for tip haplotypes are significantly large. Clade and nested clade distances will be large when derived haplotypes in the colonizing population become widespread. This occurs when some tip haplotypes are part of the colonizing population or when mutations occur in the recolonized sites such that they are located far away from other closely related derived haplotypes (Dn for these haplotypes would be large). In the event, however, that only ancestral haplotypes were part of the expanding population, or that expansion occurred so recently that not enough new mutations had yet arisen in the newly colonized region, these predictions would not hold (Templeton, 1998). 75 In a review of case studies with known historical expansions, Templeton (1998) demonstrates that his test is robust unless expansion has been very recent. In his analysis of thirteen data sets, he failed to infer range expansion in one case, that of Drosophila buzzatii, which expanded its range following its introduction from South America to Europe in the past 200 years (Templeton, 1998). Although Rossi et al. (1996) had reasonable sample sizes of both the Argentinean (source) and Iberian Peninsula (derived) populations of D. buzzatii, no population level historical events were detected because the Iberian flies had only one mitochondrial haplotype (Templeton, 1988). This haplotype happened to be an interior, or ancestral haplotype, and the most common type in Argentina (Templeton, 1998). At the 1-step clade level, where box 2-1,3-1,5-1, and 6-1 are all limited in distribution, mainly to the Columbia River, and are all relatively old given their distance from haplotype 1 in the network (Figure 13), range expansion from the Columbia River might be expected. Instead, both restricted gene flow in and range expansion into the Columbia River were supported. The large nested clade distance for 2-1 indicated long distance dispersal, but the small Des for most of these clades and the large Dc(I-T) (clade group 1-1 is much more widespread than these other clades) indicated restricted gene flow. Templeton et al.'s (1995) predictions for range expansion were based mainly on empirical evidence from Cann et al. (1987) suggesting that during range expansion some older haplotypes remain behind while some newer haplotypes expand. This probably occurred in rainbow trout (2-1,3-1,5-1, and 6-1 are old and in the Columbia River), however, due to the interior status of haplotype 1, its widespread distribution was in support of restricted gene flow, not range expansion. So, despite the fact that most of the haplotypes in the Columbia were relatively old (given their mutations), their limited distributions did not meet the predictions for range expansion because they were always considered derived relative to 76 haplotype 1. Only if haplotype 1 had remained behind while other haplotypes had undergone range expansion would the test strongly support range expansion. Clearly, however, interior haplotypes may be a part of range expansion. In summary, I think the test is an interesting one, but it has not yet been perfected. Nevertheless, it did prove to be a useful mechanism for presenting the geographical distribution of haplotypes, and it paves the way for improved methods for analyzing geographical and genetic associations. The inference of restricted gene flow in clade B was not very surprising as gene exchange is expected to be limited between the Queen Charlotte Islands, Vancouver Island, and the upper Skeena and Peace rivers. However, the inference of restricted gene flow hinted at the same phenomenon as was found from clade A, that the widespread distribution of an interior haplotype supported restricted gene flow rather than range expansion. This would only be a valid inference if the ancestral haplotype gained its widespread distribution before newer haplotypes were derived. The clustering of the derived haplotypes, however, suggests otherwise, that they existed prior to expansion of an interior haplotype, perhaps from that locality. In both clade A and clade B, restricted gene flow was most strongly inferred for sites putatively used as glacial refugia, the Queen Charlotte Islands and the Columbia River, because this was where "derived" haplotypes are geographically clustered. In both cases an interior, "ancestral" type is more widespread throughout the province. My suggestion is that they may have expanded into the province post-glacially, and that this test should indicate range expansion rather than restricted gene flow. Given the possibility that many of these haplotypes existed before recolonization, the limited distribution of haplotypes in both the Queen Charlotte Islands and the Columbia River may indicate source populations in these areas as well as restricted gene flow both before and after glaciation. 77 Taxonomy Although evidence from mtDNA exists for both coastal and inland refugia and although an AMOVA analysis (Table 9) indicated significant differences among regions when partitioned according to Behnke's (1992) subspecies boundaries, genetic similarity among coastal and inland populations argues against subspecies status. Both the distribution of mtDNA clades in the province (Figure 9) and a cluster analysis of watersheds (Figure 12) indicated that putatively coastal and putatively redband populations are not phylogenetically distinct from one another. These taxonomic designations will be further discussed in the final discussion. Phenotypic evolution Mapping phenotypes onto a mtDNA haplotypes tree and a population cluster analysis demonstrated that neutral molecular markers were not associated with any of the phenotypes in question. This suggests that these phenotypic characters evolved post-glacially, and, therefore represent recent evolutionary events. Recent divergence driven by natural selection has been documented in rainbow trout populations above and below a barrier in a stream tributary to Kootenay Lake which have been shown to have adaptive differences in directional response to current as well as having different meristic counts (Northcote, 1970, 1981 ). Therefore, that phenotypic and life history differences among populations represent adaptive differences, is certainly possible. Given the different rates at which variation is introduced into populations for quantitative traits relative to single loci (Lynch, 1996), the evolution of phenotypic and life history variation post-glacially may occur even with low molecular divergence. On average, variation is introduced at a single nuclear locus at rate of 10"8-10"5 mutations per generation, whereas variation is introduced to quantitative traits at a rate of 10"3-10"2 mutations per generation (Lynch, 1996). The level of neutral genetic variation among populations is, therefore, not an indication of increased quantitative trait variation (Lynch, 1996; Hard, 1995). In fact, bottlenecks that create low neutral variation may even increase the level of adaptive variation in populations by "releasing" additive genetic variation from epistatic variation (Goodnight et al, 1988; Hewitt, 1996; Ambruster, 1998). Therefore, despite relatively low mtDNA diversity in the central part of B.C. and despite relative molecular similarity among watersheds in the province, genetic variation at quantitative loci may be quite extensive. Conservation recommendations This study has demonstrated major phylogeographic variation in the province and that phylogenetic variation in O. mykiss precedes adaptive variation in two life history characters (see Tables 11, 12). This is not to say that demographic factors and adaptive variation should not be considered in drawing conservation plans, but that the deepest level of genetic variation in O. mykiss exists among phylogenetic groups, not among adaptively divergent groups. 79 Ch.4 Nuclear DNA Variation and MtDNA Sequencing INTRODUCTION Using multiple, independent loci for biogeographical inference is increasingly advocated (Bagley and Gall, 1998). Although mtDNA is one of the more accurate markers for intraspecific and interspecific analysis (Moore, 1994), it only represents a single locus and only the maternal lineage (Avise, 1994). Both Bagley and Gall (1998) and Nielsen et al. (1994) have reported significant differences between mitochondrial and nuclear markers in O. mykiss populations, and they both suggest that the two genomes taken together were more informative than either one alone. Allozyme variation (Ch.2) provides a comparison to mtDNA variation, but allozymes were assessed in a fairly limited portion of the species range, with little sampling in British Columbia. In an attempt to further examine nuclear variation as a comparison to mitochondrial variation, additional nuclear loci were analyzed. The growth hormone II intron D (GH2D intron) was chosen because steelhead populations in Washington and Idaho were shown to be polymorphic at this locus (Moran et al., 1997). In addition to the GH2D intron, two other nuclear loci were sequenced in a small number of samples to search for variable markers in British Columbia: intron 2 of the immunoglobulin heavy chain (IgH) and the first internal transcribed spacer of ribosomal PvNA (ITS-1). The IgH region was chosen because it was also found to be variable in Washington steelhead (Moran et al., 1997). The ribosomal ITS-1 region was chosen due to its documented sensitivity to intraspecific variation (Hillis et al, 1990; Phillips et al, 1995). In addition to assessing nuclear variation, sequencing analysis of the mtDNA genome was done for several reasons. As early as 1958, Neave proposed that rainbow and cutthroat trout originated in the southern part of North America where their ranges extend the furthest inland 80 (Neave, 1958). Though Behnke (1992) favored the south as the origin of the species, particularly because of the primitive morphology of southern trout relative to northern trout, he remained inconclusive on this point. In general, conclusive evidence regarding the origin of the species has been lacking. I aimed to test Neave's hypothesis by assessing divergence levels among California steelhead relative to those found in Kamchatka and throughout British Columbia. The right domain of the D-Loop region was sequenced to compare variation found in British Columbia, Alaska, and Kamchatka to variation in California steelhead for the same locus published by Nielsen et al. (1994). The D-Loop is one of the most highly variable regions in the mitochondrial DNA, and the right domain is particularly so (Digby et al., 1992). Therefore, this short sequence was thought to be informative for biogeographical purposes. Sequencing analysis was also done on the ND1 region from several samples for the purposes of assessing variation at another region of the mitochondrial genome and finding further resolution among haplotypes. Divergence estimates are known to be affected by variation in mutation rates across genes and the presence of non-random mutational clusters (Bernatchez and Danzmann, 1993; Cummings et al., 1995), therefore, sequencing was done to determine whether divergence estimates from the RFLP analysis and sequencing analyses were fairly consistent. M E T H O D S Nuclear variation I sequenced 413 base pairs from the nuclear ribosomal ITS-1 region in six samples representing genetic and geographic differences. Two samples were from the Klootchlimmis River on northern Vancouver Island (mtDNA haplotypes 1 and 24 from clades A and B, respectively). Two were from the Taku River in northern B.C. (mtDNA haplotypes 1 and 24 from clades A and B, respectively), and two were from the Columbia River: one Gerrard rainbow 81 from the Lardeau River and one yellow fin rainbow from the Halfway River, upper Arrow Lake (mtDNA haplotypes 18 and 22, both in clade A). The primers used in the PCR amplification were KP-2 and ITS-R (Phillips et al, 1995). Three samples from four geographic regions were sequenced for 317 base pairs in the IgH intron in. The samples were from the Queen Charlotte Islands (haplotype 27), Snotolvayam, Kamchatka (haplotype 12), and the Lardeau River in the upper Columbia River (haplotype 22). The primers used for this analysis were IgH-Fl and IgH-Rl (Moran et al, 1997). The sequencing protocol, including PCR annealing temperatures for both these genes, was identical to that used for the mtDNA sequencing, (see below). In addition, the GH2D intron was amplified in 101 samples from across British Columbia (see Table 13) using the GTH2(3-F1 and GTH2p-Rl primers from Moran et al (1997). All 101 samples were digested with BsrGl and Dral, two enzymes listed as polymorphic in Washington steelhead from Moran et al. (1997). In an attempt to identify further variation addition, 79 of these samples were also digested with Hindi. Mitochondrial variation The 188 bp region of the D-Loop was sequenced in the most divergent RFLP haplotypes possible, which included 13 RFLP clade A haplotypes, five RFLP clade B haplotypes, and two coastal cutthroat samples as outgroups (see Table 14). These samples were sequenced for the same region previously sequenced by Nielsen et al. (1994), and all 14 steelhead sequences published by Nielsen et al (1994) were included in the analysis. To find further resolution among haplotypes, two samples of RFLP haplotype 1 were included from Kamchatka and the Athabasca River, and two samples of RFLP haplotype 3 were included from Alaska and Vancouver Island. For the ND1 analysis, I sequenced 468 base pairs from 13 samples, eight from RFLP clade A, 82 Table 13. GHII intron 2D samples used Watershed Sample Site (number sampled) VANCOUVER ISLAND Cowichan (5), Elsie Lake (3), Klootchlimmis (6) QUEEN CHARLOTTE ISLANDS BVM (4) FRASER RIVER SKEENA/ NASS NORTH COAST Capilano (5), Fish Lake (5), Dome Ck (5), Pennask (1), Tzenzaicut (5) Babine (4), Morice (5), Canyon (5), Goathorn Ck (5), Toboggan Ck (5), Meziadin Ck (3) Tahltan (3), Mclntyre Ck, Yukon (3) PEACE/ ATHABASCA Kiskatinaw (5), Moose (5), Chowade (1), Wampus (1) COLUMBIA RIVER Procter (5), Gerrard (3), Hill Creek (5) KAMCHATKA Kavachina (3), Snotolvayam (6) Total Number Sampled 101 83 four from RFLP clade B, and one cutthroat sample as an outgroup (see Table 17). S-phe and P2 (Nielsen et al, 1994) were used as primers for the D-Loop amplification; and NDB and NDC (Redenbach and Taylor, 1999) were used as primers for the ND1 amplification. PCR concentrations and conditions were the same as in the RFLP analysis, with the exception of increased volume (150ul). The concentrations used for the PCR amplification were as follows: 6ul whole DNA, lx Taq Buffer, 0.8 mM dNTP mix, 0.6 uM of each primer, 9 U Taq Polymerase, and 2mM MgC12. PCR conditions for the D-Loop and ND1 sequences were as follows: 1 cycle of 2:00 minutes at 95°C, 1:30 minutes at 55°C, 2:00 minutes at 72°C, followed by 35 cycles of 1 minute at 95°C, 1 minute at 55°C, 2 minutes at 72°C, then finally 1 cycle of 2 minutes at 72°C. PCR products from both the D-Loop and the ND 1 region were purified and sequenced as per instructions in the Qiaquick PCR Purification Kit. The termination PCR reaction was done using FS-Taq, which adds ddNTPs at all stages of the PCR. The reactants and concentrations used were 8ul Taq premix (FS-Taq), lul DMSO, 3.2 picomoles primer, 90-120ng DNA template, and enough H 2 0 to reach a total volume of 20 ul. The PCR conditions used were 1 cycle of 4 minutes at 95°C, 25 cycles of 30 second denaturation at 95°C, 15 second annealing at 55°C, and 4 minute extensions at 60°C. Maximum storage time was overnight at 4°C. Centri-Sep purification columns were used to remove extra ddNTPs and primers, after which the PCR products were sequenced by an ABI automated sequencer. The sequences were aligned and formatted using the program Clustal in the PCGENE software package (Bairoch, 1991). Phylogenetic trees were constructed for D-Loop sequences in PHYLIP (Felsenstein, 1993), using Seqboot to create 100 bootstrapped data sets, Dnapars and Consense for parsimony-based trees, and Dnadist, Neighbor (neighbor-joining) and Consense for 84 distance-based trees. The Kimura 2-parameter distance method was used in Dnadist as this assumed both equal base frequencies and different rates of transitions versus transversions. A ratio of 2:1 transitions to transversions was used, as transitions are thought to be more common than transversions. Drawgram was used in all cases to create the phenogram. RESULTS Nuclear variation No variation was found among any of the six samples analyzed for the ITS locus or the four samples analyzed for the IgM intron. In the 101 samples analyzed for the GH2D intron, all but four were monomorphic for the B allele, the most common allele in Mo ran et al. (1997). The only other allele found, identical to C from Moran et al. (1997), was found in four samples at only one locality, Tzenzaicut Lake in the Fraser River watershed (see Figure 17). Three of the samples from Tzenzaicut Lake were homozygous and one was heterozygous (see Figure 17). That the only variability found at this locus was in the Fraser River was consistent with colonization of the Fraser from the Columbia River (from Moran et al, 1997) although no variation was found in the upper Columbia River in B.C in this study. No further variation was found with the Hindi digests. Mitochondrial sequencing D-loop sequencing, including Nielsen et a/.'s (1994) steelhead data, indicated 26 point mutations and three insertion or deletion events (indels) in the 188 base pair sequence (Table 16). The three indel events and 17 of the 26 mutation events occurred strictly between rainbow and cutthroat trout. The transition to transversion ratio overall was 20:6, with a ratio of 12:5 between rainbow and cutthroat, and 8:1 among rainbow trout. These ratios suggest that the 2:1 ratio used in the divergence estimates was an underestimate of transitions, and therefore, the calculated 85 Figure 17. GH2D intron genotypes in Oncorhynchus mykiss samples from a variety of watersheds. The genotypes are identical to those published in Moran et al. (1997) for O. mykiss in Washington State. 86 divergence estimates of cutthroat and rainbow trout may be underestimated. For my purposes, however, which were mainly to estimate the relationships among rainbow trout sequences, a 2:lratio was adequate. Five D-Loop haplotypes were found in the rainbow trout samples from British Columbia and Kamchatka, four of which (ST1, ST3, ST9, and ST12) had previously been identified by Nielsen et al. (1994), and one of which was new to this study (BC1). Three haplotypes were found in clade A samples (ST1, ST3, and ST9) and two in clade B samples (BC1 and ST12). ST1 was the most common haplotype found in the clade A samples, ST3 was found in the Dean River (haplotype 23), and ST9 was found in the Wampus Creek sample, from the Athabasca River (RFLP haplotype 1) and in the Tumtum sample from the Columbia River (RFLP haplotype 19). ST 12 was found in most of the clade B haplotypes, and BC1 was found in Vedder Creek in southwestern B.C. (RFLP haplotype 31) (see Table 16). Phylogenetic analysis of cutthroat trout and the 15 rainbow trout sequence haplotypes revealed relatively low bootstrap support (Figure 19a,b). Based on the pattern of mutations (Table 16), low bootstrap support was probably due to few polymorphic sites and the presence of homoplastic mutations. Nevertheless, the trees did indicate a divergence between previously designated RFLP clade A and clade B haplotypes. ' ND1 sequencing indicated 25 mutations in 468 base pairs, with 21 of them occurring strictly between rainbow and cutthroat trout (Table 17). The overall transition to transversion ratio in ND1 was 20:5 (4:1), with a ratio of 16:5 between rainbow and cutthroat trout, and a ratio of 4:0 among rainbow trout samples. The ND1 transition: transversion ratio was very similar to that estimated for the D-loop. Overall, four haplotypes (RBI-4) were found for the ND1 region, two from the RFLP clade A and two from the RFLP clade B. Most RFLP 87 Table 14. Haplotypes and samples chosen for D-Loop sequencing. RFLP haplotypes chosen for D-Loop sequencing are listed on the far left. The particular sample chosen for the analysis is listed under sample number with the location in parentheses. The sequence haplotype was the resulting sequence found for D-Loop from my analysis (see Table 16 for corresponding sequences). D-Loop RFLP Haplotype Sample No. (Sample Site) Sequence Haplotype 1 Kavachina 6, (Kamchatka Peninsula) ST1 1 Wampus 3 (Athabasca) ST9 3 Natnek 1 (Alaska) ST1 3 Quinsam 6 (Vancouver Island) ST1 6 Karluk 1 (Alaska) ST1 8 Karluk 5 (Alaska) ST1 12 Snotolvayam 1 (Kamchatka) ST1 18 Halfway 1 (upper Columbia) ST1 19 Tumtum 1 (upper Columbia) ST9 20 Robertson Ck 3 (Vancouver Island) ST1 21 Selway 5 (Idaho) ST1 22 Gerrard 3 (upper Columbia ST1 23 Dean 3 (Cariboo) ST3 27 B V M 3 (Queen Charlotte Islands) ST12 28 Kitimat 4 (central coast, B.C.) ST12 30 Cheslatta 2 (upper Fraser) ST12; 31 Vedder 2 (Lower Mainland, B.C.) BC1 32 Klootchlimmis 1 (Vancouver Island) ST12 Northern California ST1 Northern California ST2 central, southern California ST3 central California ST4 southern, central California ST5 southern California ST6 northern ST7 California ST8 California ST9 southern California ST10 B.C. ST11 northern California ST12 southern California ST13 southern California ST14 coastal cutthroat 34 (B.C.) CUT1 coastal cutthroat 36 (B.C.) CUT2 88 Table 15. Nucleotide sequence of the right domain of the D-loop (188 bp) in rainbow trout (from Digby et al., 1992). The nucleotide at the 5' end coincides with number 962-1149 from Digby et al. (1992). This sequence is identical to the ST10 found by Nielsen et al. (1994). GTCTCAAATC TTTAGCATTT A G T A T A C A T T AACGCTGTTA AGCAATATTT GGCACCGACA A A T A A A C T T T TCAATGCCAT T T T T A T A T A C GCGCTGTAAT TCGATCCACT TTCCACGCAC A T T A A T A A A C GCGTACACTT TTGTAGCACC AGCCCGCC TTTTATGCAC C C A T A A A T A A TAGCACCAAC 89 Table 16. The 15 haplotypes from sequencing the right domain of the D-loop in two coastal cutthroat samples, 16 0. mykiss samples and Nielsen et a/.'s (1994) steelhead from California. Only mutational sites are indicated, the sequence corresponding to ST 10 is listed in Table 15. Nielsen et al. (1994) found 14 different haplotypes among steelhead in California (ST1-14), and in my study, one new haplotype was found (BC1). Insertion-deletion (indel) events are indicated by an absence of nucleotides in either O. mykiss or O. clarki. Nucleotides are listed according to the numbers previously assigned to them (see Nielsen et al, 1994 or Digby et al, 1992). indel indel indel 964 967 984 988,9 1005/6 1014/5 1021 1032 1039 1040 1044 1050 1052 1058 CUT1 A G G G AT c A G G G T T T CUT2 A G G G AT T A G G G T T T BC1 C A T TA c G A A G T T A ST1 C A T TA T G A A G T T A ST2 c A T TA c G A A G T T A ST3 c A T TA T G A A G T T A ST4 c A T TA T G A A G T T A ST5 c A T TA T G A A G T T A ST6 c A T TA T G A A G c T A ST7 c A T TA T G A A G T T A ST8 c A T TA T G A A G T T A ST9 c A T TA T G A A G T T A ST10 c A T TA T G A A G T c A ST11 c A T TA T G A A A T T A ST12 c A T TA T G A A G T T A ST13 c A T TA T G A A G T T A ST14 c A T TA c G A A G T T A 1085 1086 1089 1093 1100 1103 1106 1109 1115 1119 1123 1127 1143 1147 1149 CUT1 G C C C T G A G A A T A A A T CUT2 G C C C T G A G A A T A A A T BC1 A C A T C A C G G T C G G G C ST1 A T A T C A A G G T C G G G C ST2 A T A T C A A G G T C G G G C ST3 A T A T C A A A G T C G G G C ST4 A C A T C G A G G T C G G G C ST5 A C A T C G C G G T C G G A C ST6 A C A T C G C G G T C G G A C ST7 A C A T C A A G G T C G G A C ST8 A C A T C A c G G T C G G A C ST9 A T A T C A A G G T C G G A C ST10 A T A T C A A A G T C G G G C ST11 A T A T C A A G G T C G G G C ST12 A C A T C A C G G T C G G G C ST13 A C A T C G C G G T C G G G C ST14 A T A T C A A A G T C G G G ' C 90 ST4 ST1 (A) L- ST11 r ST2 ST3 (A) L ST10 S T H ST9 (A) LST7 - BC1 (B) " ST12(B) ST13 ST5 " ST6 L ST8 r CUT2 - CUT1 i—l o.oi Figure 18. Phylogeny of D-loop using the Kimura 2-parameter distance method in Dnadist from PHYLIP (Felsenstein, 993). Neighbor was used to construct the tree. Genetic distances are drawn to scale according to Kimura's 2-parameter distance (A=RFLP clade A, B=RFLP clade B). 91 100 42 56 BC1 (B) ST12(B) ST8 54 66 ST1 (A) ST11 — S T H ST10 ST3 (A) ST2 ST9 (A) — ST7 ST13 — ST6 — ST5 ST4 CUT 2 CUT 1 Figure 19a. Consensus distance phylogeny of the D-loop sequence bootstrapped 100 times in Seqboot. The Kimura 2-parameter method in Dnadist was used to construct the distance matrix and the neighbor-joining algorithm from Neighbor was used to create the tree. All programs were taken from the PHYLIP software package (Felsenstein, 1993). Only bootstrap values above 40% are reported (A=RFLP clade A, B=RFLP clade B). 92 " ST12(B) " B C l (B) ST8 — ST9(A) ST10 43] 43 ST3 (A) - S T 1 4 " ST1 (A) - ST2 ST11 ST7 ioo 41 ST13 ST4 ST6 ST5 CUT2 CUT1 Figure 19b. Consensus parsimony phylogeny of the D-loop sequence bootstrapped 100 times in Seqboot. Dollo parsimony, using Dnapars, was used to create the trees. Again, phylogenetic programs were from PHYLIP (Felsenstein, 1993), and only bootstrap values above 40% are reported (A=RFLP clade A, B=RFLP clade B). 93 clade A haplotypes were of the type RB 1, though one sample from the Dean River, the Cariboo region (haplotype 23) had the RB2 sequence. Similarly, most RFLP clade B haplotypes were of the type RB3, though one sample from Kutcho Creek in the upper Liard (haplotype 24) had the RB4 sequence. The previously designated RFLP clade A and clade B samples could be distinguished by a single mutation despite low variability in general for this gene (see Table 18). Further resolution was found among similar RFLP haplotypes in the D-Loop but not in the ND1 sequences. The D-Loop sequences distinguished RFLP haplotype 1 samples from the Athabasca River (Wampus Creek 1) and Kamchatka (Kavachina 6), but whether this is consistent among other Athabasca and Kamchatka samples is unknown. Another sample, Tumtum from the Columbia River (RFLP haplotype 19) was identical to the Wampus Creek sample (ST9), whereas almost all other RFLP clade A samples were identical to the Kamchatka sample (ST1). Sequence divergence estimates revealed that the divergence rate of the D-Loop is about three times greater than that of the ND1 region (see Table 19). A robust result of the mitochondrial sequencing analysis was that deeper divergences were found in California (Nielsen et al., 1994) than were found in any of my samples from British Columbia, Alaska and Kamchatka. Not only were more haplotypes found in California despite sequencing the most divergent samples from the RFLP analysis, but the California sequences were more divergent from one another than were any of the northern sequences (see Figure 18, Table 19). Deeper divergences are indicative of a longer history of O. mykiss in California than in the northern part of the species range. The fact that Californian haplotypes were found to be more closely related to both "A" and "B" haplotypes than "A" and "B" haplotypes were to each other (Figure 19) suggests that both of these clade groups existed in California and subsequently spread north. The distributional differences of the two clades in 94 Table 17. ND1 samples used, with the RFLP haplotype (from Ch.3), the specific location and sample number. The sequence haplotype refers to variation at the ND1 locus. RFLP Sample No. (Sample Site) Sequence Haplotype Haplotype 1 Kavachina 6 (Kamchatkan Peninsula) RBI 1 Wampus 1 (Athabasca) RBI 2 Klootchlimmis 2 (Vancouver Island) RBI 6 Alsea 1 (Oregon) RBI 20 Robertson Ck 3 (Vancouver Island) RBI 21 Selway 5 (Idaho) RBI 22 Gerrard 3 (upper Columbia) RBI 23 Dean 3 (Cariboo) RB2 24 Kutcho 2 (Liard River) RB4 26 Honna 2 (Queen Charlotte Islands) RB3 27 B V M 3 (Queen Charlotte Islands) RB3 28 Kitimat 4 (central coast, B.C.) RB3 36 coastal cutthroat (B.C.) CUT1 95 Table 18. Four ND1 haplotypes found among rainbow trout samples (RB1-4) and one cutthroat haplotype. Only mutations are listed in the table, with the entire 468 bp sequence of RB 1 listed below. The sequenced is numbered from 1-468, starting at the 5' end with NDB as the primer. 60 123 134 135 138 159 201 219 231 252 276 296 309 CUT1 C A A C G c C G G C A T G RBI T G G T G T T A C T G G A RB2 T G G T A T T A C T G G A RB3 T G G T G T T A c c G G A RB4 T G G T G T T A c T G G A 342 366 381 384 390 399 411 414 423 435 441 465 CUT1 C C G C T G T T C C T C RBI T T A T T T C T T T G G RB2 T T A T T T C T T T G G RB3 T T A T T T C C T T G G RB4 T T A T C T C C T T G G TCTCTTGAGG ATGGGTTCGA GCCCCTTCTC TCTAGAATTG AGGGGACTTG A A C C C C T A T T AGCCACGCTA TCAAGGTGGT CCTTAAGCAT TCAGGCACAA TTCCGGGCTA AAGCTGAGGG GGGAGGCCTG CTAGTGCGAT GGGAAGTGCT A A A T G T C A T A GTACAAGGGC CAGAGTCAGG GGTAGGAAGC T T T T C C A A A C TAAATGTATG AGTTGATCAT A C C G A A A T C G CGGGTAGGAA GCTCGTACTC A T A A A A A T A C AACGGAGAGG AGGGCGGCTT T C G T T A T T A G GTTTAGGGCA GTTAGTTCAG GGAAAGCAGG GATGTGGGAT GCGCCTAAAA ATAGGACGGC TGAGAGCGTA TTTATTAGAA GGATATTAGC ATATTCGGCT AGGAAAAATA GGGCAAAGGG CCCTCCAGCA T A T T C T A C A T TGAATCCGGA G A C T A A T T C T GACTCTCCTT CTGTGAGG 96 Table 19. Range of genetic distances in different parts of the genome. The number of base pairs analyzed for the RFLP and D-Loop sequencing analyses are approximate. For the RFLP analysis, this is because 504 is the average number of base pairs examined per sample, and for the D-Loop, insertion and deletion events between O. mykiss and O. clarki slightly altered the total number of base pairs assessed. D-Loop/cyt B, ND5/6 D-Loop sequencing ND1 sequencing RFLP (-504 bp) (188 bp) (468 bp) O. mykiss- O. mykiss 0.10-1.75% 0.53-2.16% 0.21-0.64% (Within B.C.) O. mykiss- O. mykiss 0.53-3.79% (Within California) O. mykiss-O. clarki 12.09-14.69% 4.86-5.31% 97 British Columbia, therefore, are more likely due to genetic drift among pre-existing clades during glaciation than to the emergence of new clade groups. In addition, however, the fact that a widespread haplotype throughout British Columbia (haplotype 31) had a sequence not yet documented in California, suggests that it may have originated in British Columbia and be unique to the region. DISCUSSION Nuclear sequence variation was low for both the ITS-1 and IgM sequences and for the growth hormone intron. Very few samples were used for the ITS-1 and IgM sequences, which may partly explain the lack of variation, though the samples were chosen from some of the most geographically distant populations in British Columbia. More samples were used for the growth hormone intron analysis, but, unfortunately, little diversity existed for this marker even in Washington. British Columbia may have had lower diversity than Washington due to its more recent glaciation, but diversity appears to be low at this locus throughout the Pacific Northwest. Minisatellites would likely provide a better tool for further study as variation has been documented distinguishing Skeena River, Vancouver Island, and Fraser River populations (Taylor, 1995). Overall, mtDNA sequencing corroborated the divergence between the two major mtDNA RFLP clades. Evidence for divergence between the two clade groups existed in both the ND1 and the D-Loop. Nevertheless, bootstrap support was low, given the paucity of polymorphic sites. Differences in the rate of sequence divergence between the two mtDNA regions were apparent in the divergence between rainbow and cutthroat trout (Table 19). At a minimum, D-loop sequences were 12% divergent between the two species while ND1 sequences were 5% divergent. Using mutation rates of 2% for the D-Loop and 0.8% for ND1 (Shedlock et al., 1992; 98 McKay et al, 1996; Bagley and Gall, 1998), these estimates indicated a minimum divergence time of approximately 6 and 6.25 million years ago, respectively. These estimates are comparable to other estimates of divergence between these two species such as 5.5 mya (Thomas and Beckenbach, 1989), 3.5-8 mya (McKay etal, 1996) and 2.6-6.3 mya (Shedlock etal, 1992). The single mutation difference between the two RFLP haplotype 1 s in the study, Wampus Creek, Athabasca (sequencing haplotype ST9) and Kavachina River, Kamchatka (sequencing haplotype ST1), contrasted with the lack of resolution among haplotypes previously found to be distinct using RFLPs (mostly of sequencing haplotype ST1). ST1 was the most common haplotype found in the McCloud River and in Sheepheaven Creek in.California, and exists at low frequency throughout southern California and Mexico (Nielsen et al, 1994), suggesting it may be very old. ST9 may also be old, having been found in both Alaska and California (Nielsen et al, 1994) and having one more mutation in common with cutthroat trout than STL However, the lack of differences between the ND1 sequences for the Wampus Creek and Kavachina samples suggests that these two samples did not diverge very long ago. Comparing levels of diversity among mitochondrial DNA from British Columbia and California revealed much higher diversity in California (1.6% maximum divergence versus 3.8% maximum divergence). One caveat to this is that the other postulated origin for the species, Beringia, may have more variation than I have yet detected. Nielsen et al. (1994) included ten samples from Alaska in their study of genetic variation, with haplotypes ST1, 4 and 9. Furthermore, Nielsen et al. (1994) found no equivalent haplotypes to a divergent sequence published by Shedlock et al. (1992) from an Alaskan sample. Nevertheless, the diversity apparent in Nielsen et al.'s (1994) still appears to be greater in the south than the north. Busby et al. (1996), Nielsen et al. (1997), and Bagley and Gall (1998) all found high levels of diversity among 99 coastal steelhead trout in California, but this diversity had not yet been compared to variation in the northern part of the species range. My study shows that California steelhead have more haplotypes and have greater genetic divergence among haplotypes than do northern populations of rainbow and steelhead, suggesting that the south is a likely origin for the species. One confounding factor in this search for an origin of the species is that the northern part of the range is known to have been more recently glaciated. Less diversity would be expected where rainbow trout recolonized, therefore, regardless of where the species originally evolved. Nevertheless, multiple factors indicate that the south has been inhabited for a long period of time. The extensive range of O. mykiss in California (Behnke, 1992), the primitive morphology of some rainbow trout populations (Behnke, 1992), and the fact that both more genetic variation exists in California and genetic variation is geographically more tightly structured all attest to an ancient presence (Nielsen et al, 1994; Bagley and Gall, 1998). Bagley and Gall (1998) found a higher percentage of total variation (65% of mtDNA variation) among 10 groups within California than was found among the coastal and inland groups in all of British Columbia (approximately 33% for mtDNA). Allozymes corroborated higher diversity among California populations than elsewhere (Busby et al, 1996). In conclusion, given the history of glaciation, the origin of the species is almost impossible to definitively locate, but molecular and morphological data point to California as the center of the most ancient divergences in the species. 100 Ch.5 General Discussion and Synthesis Inconsistencies in the data for O. mykiss Origin of Kamchatkan trout In rainbow and steelhead, allozyme and mtDNA provided conflicting results regarding the evolutionary placement of rainbow trout from Kamchatka. Allozyme data suggested a relatively recent divergence between Kamchatkan and coastal trout, particularly from Vancouver Island (Ch.2, Okazaki, 1984). Using 37 allozyme loci, Okazaki (1984) estimated a divergence time between Kamchatkan trout and coastal North American trout (Vancouver Island, Puget Sound, and the lower Columbia River) of approximately 13,000 years ago. MtDNA from the two regions, however, were quite different. Nearly half of Vancouver Island mtDNA samples had clade B haplotypes, but Kamchatka trout were composed of strictly clade A haplotypes. Sampling may have been too limited in Kamchatka to observe clade B haplotypes as only 13 samples were taken from two Kamchatkan populations. However, in a more substantial sample size (n=25) from Alaska, only two clade B haplotypes were revealed. Based on his allozyme data, Okazaki (1984) suggested that coastal trout from North America survived glaciation in a Beringian refuge. MtDNA, however, revealed that deep divergences exist in coastal North American trout that are not often observed in Alaskan and Kamchatkan trout, indicating that trout from the Bering did not recolonize coastal British Columbia. Nevertheless, the Bering region may well have been used as a refuge by rainbow trout in the last glaciation. A Beringian refuge was proposed by McPhail and Lindsey (1970) to explain the northern distribution of rainbow and steelhead trout, though they suggested that little post-glacial dispersal reached beyond that region. The mitochondrial data corroborate this in that some unusual RFLP haplotypes as well as sequencing haplotypes were found in Alaska and Kamchatka 101 (Ch. 3, 4) that were not abundant elsewhere on the B.C. coast. Allozyme frequencies may simply have remained more stable in Kamchatka for longer than mitochondrial haplotype frequencies, as mutation is less common and allozymes are less subject to genetic drift (Avise, 1994). The weight of the evidence, therefore, is in favor of a relatively small Beringian refuge for O. mykiss during the last glaciation, with little post-glacial dispersal. Origin of Athabascan trout Another discrepancy among the allozyme and mtDNA data sets exists with respect to Athabascan trout. Carl et al. (1994) suggested an ancient divergence of this group using allozymes, but the mtDNA haplotypes sampled in my study in Athabasca were identical to the most common haplotype in British Columbia, particularly in the upper Fraser and Peace rivers. The variation in the Wampus Creek D-Loop sequence was intriguing, but was also found in a Columbia River sample (though with a different RFLP haplotype). The D-loop haplotype, ST9, was also found in Alaska and California and is therefore not unique to the Athabasca population (Nielsen et al, 1994). The lack of further variation found in the ND1 sequence supports the hypothesis that the Athabascan trout are not extremely ancient and probably colonized the area from the Fraser River post-glacially. One hypothesis to account for the discrepancy between the allozymes (from Carl et al, 1994) and mtDNA is hybridization between rainbow and cutthroat trout in Wampus Creek (see Ch.2). If mating occurred mainly between female rainbow trout and male cutthroat trout, mtDNA would be unlikely to reveal evidence of the hybridization event. Some evidence suggests that hybrid mating between rainbow and cutthroat trout might be sex-specific. Hawkins and Foote (1998) found that fish emerged earlier and had an abundance of yolk if the hybrid pair had a steelhead dam, whereas fish emerged late and had little yolk with a cutthroat dam. As 102 mitochondria are maternally inherited and allozymes are biparentally inherited, if hybridization occurred strictly between cutthroat males and rainbow trout females, hybrids would have in rainbow trout mitochondria and a mixture between rainbow trout and cutthroat trout allozyme frequencies. Therefore, historical hybridization between the two species is still certainly possible given the data. Coastal-inland division Another discrepancy among the data is that population cluster analysis of mtDNA did not distinguish coastal from inland populations whereas allozyme analyses did (Ch. 2, Utter et al, 1980). The easiest explanation for this is that allozymes have not been extensively sampled, particularly in the northern part of the province which was where the mitochondrial differences were least clear. However, mitochondrial DNA is also more susceptible to introgression than nuclear DNA and, therefore, may be more likely to reflect complex post-glacial dispersal patterns. The detail provided by mitochondrial data may simply provide a clearer indication of post-glacial events, and reflect complex patterns where allozymes do not. The repeated glaciations compounded with high dispersal capabilities of the species most likely created complex phylogeographic boundaries that are, in reality, quite unclear. Furthermore, given a complex history, it would be unlikely for multiple molecular markers to provide identical biogeographical patterns. Both marker systems at least reflected significant genetic divergences in the province most likely dating back to events during the last glaciation. Impact of introductions or stocking The well known stocking and illegal transplants of rainbow trout throughout most of its native range need to be addressed as they obviously could undermine phylogenetic patterns found in the species. For several reasons, I think the general phylogeographic patterns I have observed 103 should not be discounted. First, the native distribution of rainbow trout is very widespread and where native populations of rainbow trout already exist, introductions generally do not work (E. Parkinson, pers. comm.). Second, most stocking occurs on a local scale that would be unlikely to affect a phylogeographic study very extensively. Third, the distribution of haplotypes observed in this study is consistent across a very large geographic range and among sites known to be pristine, indicating that stocking has not had a huge impact on the molecular results. For instance, sites known to be pure and unstocked, such as Goatskin Creek in the Okanagan, the Iskut and Stikine rivers, Graham and Halfway rivers in the Peace River drainage, and Pennask Lake are characteristic of other populations sampled in their regions. Fourth, significant geographical structure exists in this study, particularly in the B clade, further suggesting that if stocking has occurred, it has been limited or on a local scale. Finally, the phylogeographic patterns fit those expected by glacial theory and hypotheses regarding putative refugia. Diversity is higher in the Columbia River, the Queen Charlotte Islands, and in the Vancouver Island/ Lower Mainland region than elsewhere, which is consistent with expectations regarding source populations and recently colonized areas. For all the above reasons, I would argue that these phylogeographic findings are valid reflections of the biogeographical history of rainbow trout. Comparative phylogeography among Pacific salmonids That Beringia does not appear to have been a major Wisconsinan refuge for rainbow and steelhead trout stands in stark contrast to many other Pacific salmonids, such as sockeye (O. nerka), chum (O. keta), pink (O. gorbuscha), chinook (O. tshawytscha), and coho salmon (O. kisutch) (Varnavskaya and Beacham, 1992; Cronin et al, 1993; Taylor et al, 1994; Taylor et al, 1996; Carney et al, 1997; Seeb and Crane, 1999). In these studies, the majority of genetic variation in the species is generally in the north, whereas the majority of variation in cutthroat and 104 rainbow trout, two Pacific trout, is in southern North America (Allendorf and Leary, 1988; Behnke, 1992). Neave (1958) proposed that rainbow and cutthroat trout originated in southern North America (i.e. in and around southwestern California), from which they dispersed north and eventually gave rise to Oncorhynchus in Sea of Japan (Neave, 1958). He proposed that Oncorhynchus then recolonized much of the former range of its ancestors, the Pacific trout, but never reached as far south as rainbow and cutthroat trout lineages did. Patterns of intraspecific variation do seem to fit his hypothesis, although the way he envisions speciation in Pacific salmon is not necessarily accurate. Deep phylogenetic divisions among cutthroat subspecies in North America (Allendorf and Leary, 1988) and the large sequence divergence among rainbow trout populations found in this study and in California (Nielsen et al, 1994; Bagley and Gall, 1998) contrast with the relatively low intraspecific variation and more northern distributions of many Pacific salmon. For example, Cronin et al. (1993) found variation in mitochondrial DNA sequence divergence on the order of 0.03-0.4% for chinook salmon and 0.06-0.62% for chum salmon. The contrasting distributions of genetic variation between rainbow trout and Pacific salmon reinforce a southern origin for rainbow trout with a relatively recent colonization of the extreme northern parts of its range. Demographics: phylogeography, vicariance and dispersal My data support the hypothesis that Pleistocene glaciations have been one of the most significant influences on the present genetic variation in O. mykiss. Theory predicts that species directly affected by glaciation will have broader distributions due to the dispersal potential from large proglacial lakes (despite less time to disperse) and lower diversity due to range contraction and smaller population sizes in glacial refugia, than species in non-glaciated regions (Bernatchez 105 and Wilson, 1998). In general, species strongly affected by glaciation will appear younger than they are because intraspecific coalescent times will be more recent. This is certainly true of many Pacific salmonids and is true of rainbow trout, at least within clades in the northern extent of its range. Theory also predicts that the population genetic structure of species affected by glaciation will fit into zoogeographic provinces, in which a variety of species and taxa show similar phylogeographic structure (Bernatchez and Wilson, 1998). Bernatchez and Wilson (1998) tested some of these patterns and found lower diversity among species affected by glaciation than among ones that were not. In contrast to popular concepts, however, they also found that the phylogeographic structure of species affected by glaciation is not always clear. Though the phylogeographic patterns of species in the northern hemisphere are sometimes characterized by monophyletic groups in discrete geographic units show, such as lake whitefish (Coregonus clupeaformis) and lake trout {Salvelinus namaycush), many species do not have such simple structure (Bernatchez and Wilson, 1998). Therefore, the strong indication of gene flow in rainbow and steelhead trout during recolonization of the province is not uncommon, despite the evidence that rainbow and steelhead survived glaciation in multiple refugia. Another contention of Bernatchez and Wilson (1998) was that zoogeographical provinces should exist which reflect common biogeographical histories of a variety of species and taxa. The different phylogeographic histories even among freshwater fishes in British Columbia (Taylor et al, 1996; Taylor et al, 1999), calls this concept of zoogeographic provinces into question. It remains unclear what the major factors are in determining the effect of glaciation on a given species, though presumably both intrinsic and extrinsic factors (natural dispersal abilities, ecological requirements, and ecological conditions) play a large role. Though zoogeographic 106 provinces still may be a valuable concept, this study has certainly shown that even species with very similar ranges do not necessarily have similar phylogenetic structures, and there remains much more to discover about why such large biogeographical differences exist. Phylogeography and taxonomy According to the subspecies definition put forward by Avise and Ball (1990), phylogenetic distinction should reflect distinct evolutionary histories and be demonstrated by concordant differences using several molecular markers. Monophyletic differences are not necessary as long as a genetic distinction is apparent. This study shows that post-glacial mixing and survival in multiple refugia during the last glaciation has most likely blurred the genetic distinction between coastal and inland populations (Figure 15, Table 9). As such, genetic differences are not clearly defined and subspecies status is not supported by the genetic data. Several hypotheses exist to explain the inconsistency between Behnke's subspecies groups and the genetic data. Firstly, the morphological differences pointed to by Behnke may not be very clear. Behnke (1992) does not specify how many morphological samples he collected or where they were taken from, and morphological distinctions may simply not be as apparent in northern British Columbia as his subspecies designation would imply. A second possibility is that morphological differences are real and have evolved only once but that events during post-glacial recolonization have obscured their genetic differentiation. As mitochondrial DNA is presumably selectively neutral and because it recombines independently of any selected nuclear DNA, it may introgress more easily than nuclear DNA, resulting in incorrect inferences from genetic data. Despite these alternative explanations, however, the extent to which many of the watersheds in British Columbia share the same mitochondrial haplotypes argues against the introgression of the mtDNA alone. The extent to which post-glacial gene flow appears to have 107 occurred suggests that selection for coastal and redband phenotypes would have had to be quite strong to maintain distinct morphologies, or that the phenotypes are plastic. In either case, subspecies designations would not be warranted at least according to the definition put forward by Avise and Ball (1990). That is, in either case, coastal and redband trout do not demonstrate independent evolutionary histories and, therefore, should not be regarded as valid subspecies. The role of natural selection in driving morphological and physiological differences between coastal and redband populations has previously been demonstrated in the southern part of the species range (Wishard et al, 1984). Similar to my study, the authors used their results to argue against the validity of subspecies designations for these groups. Phylogeography and conservation Though distinction among coastal and inland populations is not sufficient for taxonomic status due to extensive secondary contact and gene flow, the intraspecific genetic variation resulting from the history of glaciation warrants recognition and protection. Multiple marker systems have shown differences (albeit below subspecies status in some cases) between coastal and inland rainbow trout, including mtDNA (this study, see Figure 12), allozymes (Parkinson, 1984; this study, Figures 3a,b), minisatellites (Taylor, 1995) and morphology (Behnke, 1992). These studies reinforce the significance of historical isolation of coastal and inland populations in British Columbia. Utter et al. (1980) cite allozyme variation between inland and coastal populations extending as far north as Alaska and Kamchatka but sampling was relatively limited in northern populations. Parkinson (1984) also demonstrated differences between coastal and inland populations in British Columbia, although unusual frequencies found in the Skeena River failed to place it clearly with either coastal or inland populations. Taylor (1995) demonstrated coastal and 108 inland differences using minisatellites between the Skeena and Vancouver Island rainbow trout, on the one hand, and the Fraser River, on the other. However, sampling did not extend further north or inland. Though phylogenetic boundaries are difficult to define precisely, they almost certainly reflect the most ancient genetic divisions within the province and likely predate most life history and phenotypic variation in the province (Tables 11, 12). Therefore, I would suggest maintaining the coastal-inland division as previously identified by morphology and allozyme loci (Behnke, 1992) as the major units of conservation in the province. Populations located in or near former refugia should be granted the highest conservation priority not only because they are likely to contain the most distinct genetic resources for the species, but also because these localities are more likely to be the refugia of future as well. In summary, the genetic divergence at neutral loci between coastal and inland populations strongly suggests that they have been isolated from each other for a substantial portion of their histories, and therefore, they represent genetic variation worthy of our greatest conservation priority. However, surviving rapid environmental change may well depend on the variation of phenotypic and life history traits among populations. Particularly if differences are genetically based, these groups represent important sources of adaptive potential to the species (studies are underway to assess the genetic basis of several phenotypes in B.C., E.B. Taylor, E. Keeley). My study, however, relates to the glacial history of the province and its effect on rainbow and steelhead genetic variation. My results attest to the strong role glaciation has had in shaping genetic diversity in British Columbia and throughout much of the range of O. mykiss and they strongly suggest that phylogeography should play a major role in conservation planning for the species. 109 REFERENCES Adams, N.S., Spearman, W.J., Burger, C.V., Currens, K.P., Schreck, C.B., and H.W. Li , 1994. Variation in mitochondrial DNA and allozymes discriminates early and late forms of chinook salmon (Oncorhynchus tshawytscha) in the Kenai and Kasilof Rivers, Alaska. Canadian Journal of Fisheries and Aquatic Sciences, 51(Suppl.l): 172-181. Allendorf, F.W., Bayles, D., Bottom, D.L, Currens, K.P., Frissell, C.A., Hankin, D., Lichatowich. J.A., Nehlsen, W., Trotter, P.C, and T.H. Williams, 1997. Prioritizing Pacific salmon stocks for conservation. Conservation Biology, 11(1): 140-152. Allendorf, F.W. and R.F. Leary, 1988. Conservation and distribution of genetic variation in a polytypic species, the cutthroat trout. Conservation Biology, 2(2): 170-184. Angers, B. Bernatchez, L., Angers, A. and L. Desgroseillers, 1995. Specific microsatellite loci for brook charr reveal strong population subdivision on a microgeographic scale. Journal of Fish Biology, 47 (Suppl. A): 177-185. Armbruster, P., Bradshaw, W.E. and C M . Holzapfel, 1998. Effects of postglacial expansion on allozyme and quantitative genetic variation of the pitcher-plant mosquito, Wyeomyia smithii. Evolution, 52(6): 1697-1704. Aspinwall, N. , 1974. Genetic analysis of North American populations of the pink salmon, Oncorynchus gorbuscha, possible evidence for the neutral mutation-random drift hypothesis. Evolution, 28: 295-305. Avise, J.C, 1994. Molecular Markers, Natural History and Evolution. Chapman & Hall, New York. Avise, J.C. and R.M. Ball, JR, 1990. Principles of genealogical concordance in species concepts and biological taxonomy. Oxford Survey of Evolutionary Biology, 7:45-67. Avise, J.C, Arnold, J., Ball, R.M., Bermingham, E., Lamb, T., Neigel, J.E., Reeb, C.A., and N . C Saunders, 1987. Intraspecific phylogeography: the mitochondrial DNA bridge between population genetics and systematics. Annual Review of Ecology and Systematics, 18: 489-522. Bagley, M.J., and G.A.E. Gall, 1998. Mitochondrial and nuclear DNA sequence variability among populations of rainbow trout (Oncorhynchus mykiss). Molecular Ecology, 7: 945-961. Bairoch, A., 1991. PC/ Gene: the nucleic acid and protein sequence analysis software system, v. 6.60. IntelliGenetics Inc. University of Geneva, Switzerland. Ball, R.M., Jr. and J.C. Avise, 1992. Mitochondrial DNA phylogeographic differentiation among 110 avian populations and the evolutionary significance of subspecies. The Auk, 109(3): 626-636. Ballard, J.W.O. and M. Kreitman, 1995. Is mitochondrial DNA a strictly neutral marker? Trends in Ecology and Evolution, 10(12): 485-488. Beacham, T.D. and C B . Murray, 1986. The effect of spawning time and incubation temperature on meristic variation in chum salmon (Oncorhynchus keta). Canadian Journal of Zoology, 64: 45-48. Beacham, T.D., Withler, R.E. and A.P. Gould, 1985. Biochemical genetic stock identification of pink salmon (Oncorhynchus gorbuscha) in southern British Columbia and Puget Sound. Canadian Journal of Fisheries and Aquatic Sciences, 42: 1474-1483. Beckenbach, A.T., Thomas, W.K., and H. Sohrabi, 1990. Intraspecific sequence variation in the mitochondrial genome of rainbow trout (Oncorhynchus mykiss). Genome, 33: 13-15 Behnke, R.J., 1992. Native Trout of Western North America. American Fisheries Society Monograph 6. Berg, W.J., and G.A.E. Gall, 1988. Gene flow and genetic differentiation among California coastal rainbow trout populations. Canadian Journal of Fisheries and Aquatic Sciences, 45: 122-131. Bernatchez, L. 1995. A role for molecular systematics in defining evolutionarily significant units in fishes. In Evolution and the Aquatic Ecosystem, ed. J.L. Nielson. American Fisheries Society Symposium, 17: pp. 114-132 Bernatchez, L. 1996. Genetic diversity of steelhead trout (Oncorhynchus mykiss) within the Skeena River and among four river systems of British Columbia assessed by PCR-RFLP analysis of mitochondrial DNA variation. Final Report for Susan Pollard, B.C. Ministry of Fisheries, Victoria, B.C. Bernatchez, L. and Danzmann, R.G., 1993. Congruence in control-region sequence and restriction-site variation in mitochondrial DNA of brook charr (Salvelinus fontinalis Mitchill). Molecular Biology and Evolution, 10(5): 1002-1014. Bernatchez, L, and A. Osinov, 1995. Genetic diversity of trout (genus Salmo) from its most eastern native range based on mitochondrial DNA and nuclear gene variation. Molecular Ecology, 4: 285-297. Bernatchez, L, and C.C. Wilson, 1998. Comparative phylogeography of Nearctic and Palearctic fishes. Molecular Ecology, 7:431-452. Bickham, J.W., Wood, C.C, and J.C Patton, 1995. Biogeographic implications of cytochrome b 111 sequences and allozymes in sockeye (Oncorhynchus nerka). Journal of Heredity, 86: 40-144. Blachford, A., O'Brien, D., and M.R. McCusker, unpublished. Clades program for Templeton analysis. Dept. Zoology, U.B.C., Vancouver, B.C. Brykov, VL.A., Polyakova, N. , Skurikhina, L.A. and A.D. Kukhlevsky, 1996. Geographical and temporal mitochondrial DNA variability in populations of pink salmon. Journal of Fish Biology, 48: 899-909. Busby, P.J., Wainwright, T.C., Bryant, G.J., Leirheimer, L.J., Waples, R.S., Waknitz, F.W., and I.V. Lagomarsino, 1996. Status review of west coast steelhead from Washington, Oregon, and California. NOAA-NWFSC Technical Memorandum 27. Byun, S.A., Koop, B.F., and T.E. Reimchen, 1997. North American black bear mtDNA phylogeography: Implications for morphology and the Haida Gwaii glacial refugium controversy. Evolution, 51(5): 1647-53. Campton, D.E. and F.M. Utter, 1985. Natural hybridization between steelhead trout (Salmo gairdneri) and coastal cutthroat trout (Salmo clarki clarki) in two Puget Sound streams. Canadian Journal of Fisheries and Aquatic Sciences, 42: 110-119. Campton, D.E. and F.M. Utter, 1987. Genetic structure of anadromous cutthroat trout (Salmo clarki clarki) populations in the Puget Sound area: Evidence for restricted gene flow-Canadian Journal of Fisheries and Aquatic Sciences, 44: 573-582. Cann, R.L., Stoneking, M. , and A.C. Wilson, 1987. Mitochondrial DNA and human evolution. Nature, 325: 31-36. Carl, L.M., Hunt, C , and Ihssen, P.E., 1994. Rainbow trout of the Athabasca River, Alberta: a unique population. Transactions of the American Fisheries Society, 123(2): 129-140. Carney, B.L., Gray, A.K. and A.J. Gharrett, 1997. Mitochondrial DNA restriction site variation within and among five populations of Alaskan coho salmon (Oncorhynchus kisutch). Canadian Journal of Fisheries and Aquatic Sciences, 54: 940-949. Carson, H.L., 1990. Increased genetic variance after a population bottleneck. Trends in Ecology and Evolution, 5(7): 228-231. Carvalho, G.R., 1993. Evolutionary aspects of fish distribution: genetic variability and adaptation. Journal of Fish Biology, 43(Suppl A.): 53-73. Crandall, K.A. and A.R. Templeton, 1993. Empirical tests of some predictions from coalescent theory with applications to intraspecific phylogeny reconstruction. Genetics, 134: 959-969 112 Cronin, M.A., Spearman, W.J., Wilmot, R.L., Patton, J.C. and J.W. Bickham, 1993. Mitochondrial DNA variation in chinook (Oncorhynchus tshawytscha) and chum salmon (O. keta) detected by restriction enzyme analysis of polymerase chain reaction (PCR) products. Canadian Journal of Fisheries and Aquatic Sciences, 50: 708-715. Cummings, M.P., Otto, S.P., and J. Wakeley, 1995. Sampling properties of DNA sequence data in phylogenetic analysis. Molecular Biology and Evolution, 12(5): 814-822. Currens, K.P., Schreck, C.B., and H.W. Li , 1990. Allozyme and morphological divergence of rainbow trout (Oncorhynchus mykiss) above and below waterfalls in the Deschutes River, Oregon. Copeia, 3: 730-746. Currens, K.P., 1997. Evolutionary ecology of redband trout. In Evolutionary and Risk in Conservation of Pacific Salmon. Ph.D. Dissertation. Oregon State University, Corvallis, Oregon, pp. 4-51. Danzmann, R.G., Ferguson, M.M. and S.K.A. Arndt, 1993. Mitochondrial DNA variability in Ontario and New York rainbow trout (Oncorhynchus mykiss). Canadian Journal of Zoology, 71: 1923-1933. Deagle, B.E., Reimchen, T.E. and D.B. Levin, 1996. Origins of endemic stickleback from the Queen Charlotte Islands: mitochondrial and morphological evidence. Canadian Journal of Zoology, 74: 1045-1056. .y Digby, T.J., Gray, M.W., and C B . Lazier, 1992. Rainbow trout mitochondrial DNA: sequence sequence and structural characteristics of the non-coding region and flanking tRNA genes. Gene, 113: 197-204. Doyle, J.J., 1997. Trees within trees: genes and species, molecules and morphology. Systematic Biology, 46(3): 537-553. Excofiier, L., Smouse, P.E., and J.M.Quattro, 1992. Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics, 131:479-491. Felsenstein, J., 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution, 39(4): 783-791. Felsenstein, J., 1988. Phylogenies from molecular sequences: inference and reliability. Annual Review of Genetics, 22: 521-565. Felsenstein, J., 1993. PHYLIP — Phylogeny inference package (Version 3.5c). Distributed by author. Dept. Genetics, University of Washington, Seattle. 113 Foote, C.J., Clayton, J.W., Lindsey, C C , and R.A. Bodaly, 1992. Evolution of lake whitefish (Coregonus clupeaformis) in North America during the Pleistocene: evidence for a Nahanni glacial refuge race in the Northern Cordillera region. Canadian Journal of Fisheries and Aquatic Sciences, 49: 760-768. Foote. C.J., Mayer, I., Wood, C C , Clarke, C , and J. Blackburn, 1994. On the developmental pathway to nonanadromy in sockeye salmon, Oncorhynchus nerka. Canadian Journal of Zoology, 72: 397-405. Gharrett, A.J., Shirley, S.M. and G.R. Tromble, 1987. Genetic relationships among populations of Alaskan chinook salmon (Oncorhynchus tshawytscha). Canadian Journal of Fisheries and Aquatic Sciences, 44:165-11 A. Gharrett, A.J., Smoot, C , McGregor, A.J. and P.B. Holmes, 1988. Genetic relationships of even-year northwestern Alaskan pink salmon. Transactions of the American Fisheries Society, 117: 536-545. Goodnight, CJ . , 1988. Epistasis and the effect of founder events on the additive genetic variance. Evolution, 42(3): 441-454. Hart, M.W., Byrne, M. and M.J. Smith, 1997. Molecular phylogenetic analysis of life-history evolution in asterinid starfish. Evolution, 51(6): 1848-1861. Hawkins, D.K. and'C.J. Foote, 1998. Early survival and development of coastal cutthroat trout (Oncorhynchus clarki clarki), steelhead (Oncorhynchus mykiss), and reciprocal hybrids. Canadian Journal of Fisheries and Aquatic Sciences, 55: 2097-2104. Hewitt, G.M., 1996. Some genetic consequences of ice ages, and their role in divergence and speciation. Biological Journal of the Linnean Society, 58: 247-276. Hershberger, 1992. Genetic variability in rainbow trout populations. Aquaculture, 100:51-71. Hillis, D.M. and M.T. Dixon, 1991. Ribosomal DNA: molecular evolution and phylogenetic inference. Quarterly Review of Biology, 66(4): 411-453. Hindar, K, Ryman, N. and G. Stahl, 1986. Genetic differentiation among local populations and morphotypes of Arctic charr, Salvelinus alpinus. Biological Journal of the Linnean Society, 27: 269-285. Holsinger, K.E. and R.J. Mason-Gamer, 1996. Hierarchical analysis of nucleotide diversity in geographically structured populations. Genetics, 142: 629-639. Huelsenbeck, J.P., Bull, J.J. and CW. Cunningham, 1996. Combining data in phylogenetic analysis. Trends in Ecology and Evolution, 11(4): 152-158. 114 Huzyk, L., and H. Tsuyuki, 1974. Distribution of LDH-B" gene in resident and anadromous rainbow trout (Salmo gairkneri) from streams in British Columbia. Journal of the Fisheries Research Board of Canada, 31: 106-108. Jaarola, M. and H. Tegelstrom, 1995. Colonization history of north European field voles (Microtus agrestis) revealed by mitochondrial DNA. Molecular Ecology, 4: 299-310. Jaarola, M . and H. Tegelstrom, 1996. Mitochondrial DNA variation in the field vole {Microtus agrestis): regional population structure and colonization history. Evolution, 50(5): 2073-2085. Kondzela, C M , Guthrie, C M . , Hawkins, S.L., Russell, CD. , Helle, J.H. and A.J. Gharrett, 1994. Genetic relationships among chum salmon populations in southeast Alaska and northern British Columbia. Canadian Journal of Fisheries and Aquatic Sciences, 51(Suppl. 1): 50-64. Leary, R.F., Allendorf, F.W., Phelps, S.R., and K.L. Knudsen, 1987. Genetic divergence and identification of seven cutthroat trout subspecies and rainbow trout. Transactions of the American Fisheries Society, 116: 580-587. Lindsey, C.C. and J.D. McPhail, 1986. Zoogeography of fishes of the Yukon and Mackenzie Basins. In Hocutt, C.H. and E.O. Wiley, The Zoogeography of North American Freshwater Fishes. John Wiley & Sons. New York, Ch. 17. pp. 634-674. > t Losos, J.B., 1992. The evolution of convergent structure in Caribbean Anolis communities. Systematic Biology, 41(4): 403-420. Lynch, M. , Pfrender, M. , Spitze, K., Lehman, N., Hicks, J., Allen, D., Latta, L., Ottene, M. , Bogue, F., and J. Colbourne, 1999. The quantitative and molecular genetic architecture of a subdivided species. Evolution, 53(1): 100-110. Maddison, W.P., 1997. Gene trees in species trees. Systematic Biology, 46(3): 523-536. Martin, A.P. and S.R. Palumbi, 1993. Body size, metabolic rate, generation time, and the molecular clock. Proclamations of the National Academy of Science, 90: 4087-4091. McElroy, D., Moran, P., Bermingham, E. and I. Kornfield, 1992. REAP: an integrated environment for the manipulation and phylogenetic analysis of restriction data. Journal of Heredity, 83(2): 157-158. McKay, S.J., Devlin, R.H., and M.J. Smith, 1996. Phylogeny of Pacific salmon and trout based on growth hormone type-2 and mitochondrial NADH dehydrogenase subunit 3 DNA sequences. Canadian Journal of Fisheries and Aquatic Sciences, 53: 1165-1176. McPhail, J.D. and R. Carveth, 1992. A foundation for Conservation: the nature and origin of the 115 freshwater fish fauna of British Columbia. U.B.C. Fish Museum, Dept. of Zoology, U.B.C., Vancouver, B.C. McPhail, J.D. and C.C. Lindsey, 1970. Freshwater Fishes of Northwestern Canada and Alaska. Bulletin of the Fisheries Research Board of Canada, no. 173, Ottawa. McPhail, J.D. and C.C. Lindsey, 1986. Zoogeography of the freshwater fishes of Cascadia (the Columbia system and rivers north to the Stikine). In Ffocutt, C.H. and E.O. Wiley, The Zoogeography of North American Freshwater Fishes. John Wiley & Sons. New York, Ch. 16. pp. 615-637. Merila, J., Bjorklund, M . and A.J. Baker, 1997. Historical demography and present day population structure of the greenfinch, Carduelis Moris- an analysis of mtDNA control-region sequences. Evolution, 51(3): 946-956. Moore, W.S., 1995. Inferring phylogenies from mtDNA variation: mitochondrial-gene trees versus nuclear-gene trees. Evolution, 49(4): 718-726. Moran, P., Dightman, D.A., Waples, R.S., and L.K. Park, 1997. PCR-RFLP analysis reveals substantial population-level variation in the introns of Pacific salmon (Oncorhynchus spp.). Molecular Marine Biology and Biotechnology, 6(4): 315-327. Moritz, C , 1994. Applications of mitochondrial DNA analysis in conservation: a critical review. Molecular Ecology, 3: 401 -411. Moritz, C , 1995. Uses of molecular phylogenies for conservation. Philosophical Transactions of the Royal Society of London. Series B, 349: 113-118. Moritz, C , 1994. Defining 'Evolutionarily Significant Units' for conservation. Trends in Ecology and Evolution, 9(10): 373-375. Moritz. C , and D.P. Faith, 1998. Comparative phylogeography and the identification of genetically divergent areas for conservation. Molecular Ecology, 7:419-429. Myers, J.M., Kope, R.G., Bryant, G.J., Teel, D., Lierheimer, L.J., Wainwright, T.C., Grant, W.S., Waknitz, F.W., Neely, K., Lindley, S.T., and R.S. Waples, 1998. Status review of chinook salmon from Washington, Idaho, Oregon, and California. US. Dept. Commer. NOAA Technical Memorandum NMFS-NWFSC-35, 443 p. Neave, F., 1958. The origin and speciation of Oncorhynchus. Transactions of the Royal Society of Canada, 12 (series 3, section 5): 25-39. Nei, M. 1987. Molecular Evolutionary Genetics. Columbia Univ. Press, N.Y. Nei, M. and J.C. Miller, 1990. A simple method for estimating average number of nucleotide 116 substitutions within and between populations from restriction data. Genetics, 125: 873-879. Nei, M. and F. Tajima, 1981. DNA polymorphism detectable by restriction endonucleases. Genetics, 97: 145-163. Nei, M . and F. Tajima, 1983. Maximum likelihood estimation of the number of nucleotide substitutions from restriction sites data. Genetics, 105: 207-217. Nelson, J.S. and M.J. Paetz, 1992. The Fishes of Alberta. 2nd edition. Edmonton, AB, Univ. Ab. Press. Nielsen, J.L, Fountain, M.C., and J.M. Wright, 1997. Biogeographic analysis of Pacific trout (Oncorhynchus mykiss) in California and Mexico based on mitochondrial DNA and nuclear microsatellites. In: Molecular Systematics of Fishes, eds. Kocher, T.D., Stepien C.A., Academic Press, San Diego, CA. pp.53-73. Nielsen, J.L, Gan, C.A., Wright, J.M., Morris, D.B., and W.K. Thomas, 1994. Biogeographic distributions of mitochondrial and nuclear markers for southern steelhead. Molecular Marine Biology and Biotechnology, 3(5): 281-293. Ogilvie, R.T., 1989. Disjunct vascular flora of northwestern Vancouver Island in relation to Queen Charlotte Islands' endemism and Pacific Coast refugia, pp. 127-130. In: The Outer Shores, eds. G.G.E. Scudder & N. Gessler, Queen Charlotte Islands Museum,< Queen Charlotte City, B.C. Ogilvie, R.T. and A. Ceska, 1984. Alpine plants of phytogeographic interest on northwestern Vancouver Island. Canadian Journal of Botany, 62: 2356-2362. Okazaki. T., 1984. Genetic divergence and its zoogeographic implications in closely related species Salmo gairdenri and Salmo mykiss. Japanese Journal of Ichthyology, 31(3): 297-310. Orti, G., Bell, M.A., Reimchen, T.E. and A. Meyer, 1994. Global survey of mitochondrial DNA sequences in the threespine stickleback: evidence for recent migrations. Evolution, 48(3) 608-622. Osentoski, M.F. and T. Lamb, 1995. Intraspecific phylogeography of the gopher tortoise Gopherus polyphemus: RFLP analysis of amplified mtDNA segments. Molecular Ecology, 4: 709-718. Palumbi, S.R., 1990. Nucleic acids II: the polymerase chain reaction. In Hillis, D.M. and C. Moritz, eds. Molecular Systematics. Sinauer Associates, Sunderland, Mass., Ch. 7. Palumbi, S.R., and CS . Baker, 1994. Contrasting population structure from nuclear intron 1 sequences and mtDNA of humpback whales. Molecular Biology and Evolution, 11(3): 426-435. Park, L.K. and P. Moran, 1994. Developments in molecular genetic techniques in fisheries. Reviews in Fish Biology and Fisheries, 4: 272-299. Park, L.K., Brainard, M.A. and D.A. Dightman, et al., 1993. Low levels of intraspecific variation in mitochondrial DNA of chum salmon (Oncorhynchus keta). Molecular Marine Biology and Biotechnology, 2: 362-370. Parkinson, E., Behnke, R.J., and W. Pollard, 1984. A morphological and electrophoretic comparison of rainbow trout (Salmo gairdneri) above and below barriers on five streams on Vancouver Island, B.C. Fisheries Management Report, 83. Ministry of Environment, Province of British Columbia. Parkinson, E.A., 1984. Genetic variation in populations of steelhead trout (Salmo gairdneri) in British Columbia. Canadian Journal of Fisheries and Aquatic Sciences, 41: 1412-1420. Parkinson, E.A. and J. Rosenfeld, 1996. Quantifying Stock Diversity. Fisheries Project Report, No. RD 53. Ministry of Environment, Lands and Parks. Province of British Columbia. Phillips, R.B., Sajdak, S.L. and M.J. Domanico, 1995. Relationships among charrs based on DNA sequences. Nordic Journal of Freshwater Research, 71: 378-391. Redenbach, Z. and E.B. Taylor, 1999. Zoogeographical implications of variation in mitochondrial DNA of Arctic grayling (Thymallus arcticus). Molecular Ecology, 8: 23-35. Reisenbichler, R.R., Mclntyre, J.D., Solazzi, M.F., and S.W. Landino, 1992. Genetic variation in steelhead of Oregon and Northern California. Transactions of the American Fisheries Society, 121: 158-169. Reisenbichler, R.R. and S.R. Phelps, 1989. Genetic variation in steelhead (Salmo gairdneri) from the north coast of Washington. Canadian Journal of Fisheries and Aquatic Sciences, 46: 66-73. Roff, D.A. and P. Bentzen, 1989. The statistical analysis of mitochondrial DNA polymorphisms: X 2 and the problem of small samples. Molecular Biology and Evolution, 6: 539-545. Schneider, S., Kueffer, J-M, Roessli, D., and L. Excoffier, 1997. Arlequin: a software package for population genetics. Genetics and Biometry Lab, Dept. of Anthropology, University of Geneva. Scott, W.B. and E.J. Crossman, 1973. Freshwater Fishes of Canada. Bulletin of the Fisheries Resesarch Board of Canada, no. 184, Ottawa. 118 Seeb, L.W. and P.A. Crane, 1999. High genetic heterogeneity in chum salmon in western Alaska, the contact zone between northern and southern lineages. Transactions of the American Fisheries Society, 128: 58-87. Shed'ko, S.V., 1991. On systematics of Kamchatka trout Salmo mykiss (Salmoniformes, Salmonidae): restriction endonuclease analysis of mitochondrial DNA. Zhurnal Zoologistica, 31(7): 107-112. Shedlock, A.M. , Parker, J.E., Crispin, D.A., Pietsch, T.W., and G.C. Burmer, 1992. Evolution of the salmonid mitochondrial control region. Molecular Phylogenetics and Evolution, 1: 179-192. Shiozawa, D.K., Evans, R.P., Squires, D. and R.N. Williams, 1993. The genetic relatedness of rainbow trout populations from the Columbia River system. RL&L Environmental Services Ltd. Report. Dept. Zoology, Brigham Yound University Slatkin, M. and R.R. Hudson, 1991. Pairwise comparisons of mitochondrial DNA sequences in stable and exponentially growing populations. Genetics, 129: 555-562. Smith, G.R. and R.F. Stearley, 1989. The classification and scientific names of rainbow and cutthroat trouts. Fisheries, 14(1): 4-10. Soltis, D.E., Gitzendanner, M.A., Strenge, D.D., 1997. Chloroplast DNA intraspecific phylogeography of plants from the Pacific Northwest of North America. Plant Systematics and Evolution, 206: 353-373. Stearley, R.F. and G.R. Smith, 1993. Phylogeny of the Pacific trouts and salmons (Oncorhynchus) and genera of the family Salmonidae. Transactions of the American Fisheries Society, 122: 1-33. Stewart, C-B, 1993. The powers and pitfalls of parsimony. Nature, 361: 603-607. Swofford, D.L., Olsen, G.J., Waddell, P.J. and D.M. Hillis, 1990. Phylogenetic inference. In Hillis, D.M. and C. Moritz, eds. Molecular Systematics. Sinauer Associates, Sunderland, Mass., Ch. 11. Taggart, J.B., Hynes, R.A., and A. Ferguson, 1992. A simplified protocol for routine total DNA isolation from salmonid fishes. Journal of Fish Biology, 40: 963-965. Taylor, E.B., 1991. A review of local adaptation in Salmonidae, with particular reference to Pacific and Atlantic salmon. Aquaculture, 98: 185-207. Taylor, E.B., 1995. Genetic variation at minisatellite DNA loci among North Pacific populations of steelhead and rainbow trout (Oncorhynchus mykiss). Journal of Heredity, 86: 354-363. 119 Taylor, E.B., Pollard, S. and D. Louie, 1999. Mitochondrial DNA variation in bull trout (Salvelinus confluentus) from northwestern North America: implications for zoogeography and conservation. Molecular Ecology, 8: In press. Taylor, E.B., Beacham, T.D. and M . Kaeriyama, 1994. Population structure and identification of North Pacific ocean chum salmon (Oncorhynchus keta) revealed by an analysis if minisatellite DNA variation. Canadian Journal of Fisheries and Aquatic Sciences, 51; 1430-1442. Taylor, E.B., Foote, C.J. and C C . Wood, 1996. Molecular genetic evidence for parallel life-history evolution within a Pacific salmon (sockeye salmon and kokanee, Oncorhynchus nerka). Evolution, 50(1): 401-416. Taylor, E.B. and G.R. Haas, 1996. Maintaining a species tapestry: a plan for the conservation of genetic diversity in British Columbia rainbow trout (Oncorhynchus mykiss). Fisheries Technical Circular No. 99. Ministry of Environment, Lands and Parks, Province of British Columbia, 51 p. Templeton, A.R., 1993. The "Eve" hypothesis: a genetic critique and reanalysis. American Anthropologist, 95(1): 51-72. Templeton, A.R., Routman, E. and C A . Phillips, 1995. Separating population structure from population history: a cladistic analysis of the geographical distribution of mitochondrial DNA haplotypes in the tiger salamander, Ambystoma tigrinum. Genetics, 140: 767-782. Templeton, A.R., 1998. Nested clade analyses of phylogeographic data testing hypotheses about gene flow and population history. Molecular Ecology, 7: 381-397. Thomas, W.K. and A.T. Beckenbach, 1989. Variation in salmonid mitochondrial DNA: evolutionary constraints and mechanisms of substitution. Journal of Molecular Evolution, 29:233-245. Thomas, W.K., Withler, R.E. and A.T. Beckenbach, 1986. Mitochondrial DNA analysis of Pacific salmonid evolution. Canadian Journal of Zoology, 64: 1058-1064. Thompson, C.E., Taylor, E.B. and J.D. McPhail, 1997. Parallel evolution of lake-stream pairs of threespine sticklebacks (Gasterosteus) inferred from mitochondrial DNA variation. Evolution, 51(6): 1955-1965. Thorgaard, G.H., 1983. Chromosomal differences among rainbow trout populations. Copeia, 3: 650-662. Utter, F.M., and F.W. Allendorf, 1994. Phylogenetic relationships among species of Oncorhynchus: a consensus view. Conservation Biology, 8(3): 864-867. 120 Utter, F.M., Seeb, J.E., and L.W. Seeb, 1993. Complementary uses of ecological and biochemical genetic data in identifying and conserving salmon populations. Fisheries Research, 18: 59-76. Utter, F.M., 1981. Biological criteria for definition of species and distinct intraspecific populations of anadromous salmonids under the U.S. Endangered Species Act of 1973. Canadian Journal of Fisheries and Aquatic Sciences, 38: 1626-1635. Utter, F.M., Campton, D., Grant, S., Milner, G., Seeb, J. and L. Wishard, 1980. Population structures of indigenous salmonid species of the Pacific Northwest. In McNeil, W. and D. Himsworth, Salmonid Ecosystems of the North Pacific, pp. 285-304. Varnavskaya, N.V. and T.D. Beacham, 1992. Biochemical genetic variation in odd-year pink salmon (Oncorhynchus gorbuscha) from Kamchatka. Canadian Journal of Zoology, 70: 2115-2120. Varnavskaya, N.V., Wood, C.C, and R.J. Everett, 1994. Genetic variation in sockeye salmon (Oncorhynchus nerka) populations of Asia and North America. Canadian Journal of Fisheries and Aquatic Sciences, 51(Suppl. 1): 132-146. Varnavskaya, N.V., Wood, C.C, Everett, R.J., Wilmot, R.L., Varnavskaya, V.S., Midanaya, V.V. and T.P. Quinn, 1994. Genetic differentiation of subpopulations of sockeye salmon (Oncorhynchus nerka) within lakes of Alaska, British Columbia, and Kamchatka, Russia. Canadian Journal of Fisheries and Aquatic Sciences, 51(Suppl. 1): 147-157. Verspoor, E., and W.C Jordan, 1989. Genetic variation at the Me-2 locus in the Atlantic salmon within and between rivers: evidence for its selective maintenance. Journal of Fish Biology, 35(Suppl. A), 205-213. Vuorinen, J., and O.K.Berg, 1989. Genetic divergence of anadromous and nonanadromous Atlantic Salmon (Salmo Salar) in the River Namsen, Norway. Canadian Journal of Fisheries and Aquatic Sciences, 46: 406-409. Waples, R.S., 1995. Evolutionarily significant units and the conservation of biological diversity under the Enadangered Species Act. American Fisheries Society Symposium, 17: 8-27. Warner, B.G., Mathewes, R.W. and J.J. Clague, 1982. Ice-free conditions on the Queen Charlotte Islands, British Columbia, at the height of late Wisconsin glaciation. Science, 218: 675-677. Weider, L.J., and A. Hobaek, 1997. Postglacial dispersal, glacial refugia, and clonal structure in Russian/ Siberian populations of the arctic Daphnia pulex complex. Heredity, 78: 363-372. Wilmot, R.L., Everett, R.J., Spearman, W.J., Baccus, R., Varnavskaya, N.V. and S.V. Putivkin, 121 1994. Genetic stock structure of western Alaska chum salmon and a comparison with Russian far east stocks. Canadian Journal of Fisheries and Aquatic Sciences, 51 (Suppl 1): 84-94. Wilson, A.C., Cann, R.L., Carr, S.M., George, M , Gyllensten, U.B., Helm-Bychowski, K.M. , Higuchi, R.G., Palumbi, S.R., Prager, E.M., Sage, R.D., and M . Stoneking, 1985. Mitochondrial DNA and two perspectives on evolutionary genetics. Biological Journal of the Linnean Society, 26: 375-400. Wilson, G.M., Thomas, W.K. and A.T. Beckenbach, 1984. Intra- and inter-specific mitochondrial DNA sequence divergence in Salmo: rainbow, steelhead, and cutthroat trouts. Canadian Journal of Zoology, 63: 2088-2094. Wishard, L.N., Seeb, J.E., Utter, F.M., and D. Stefan, 1984. A genetic investigation of suspected redband trout populations. Copeia, 1984(1): 120-132. Withler, I.L., 1965. Variability in life history characteristics of steelhead trout (Salmo gairdneri) along the Pacific coast of North America. Journal of the Fisheries Research Board of Canada, 23(3): 365-393. Wood, C C , Riddell, B.E., Rutherford, D.T., and R.E. Withler, 1994. Biochemical genetic survey of sockeye salmon (Oncorhynchus nerka) in Canada. Canadian Journal of • Fisheries and Aquatic Sciences, 51 (Suppl. 1): 114-131. i v 122 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items