UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Morphological stasis and genetic divergence without reproductive isolation in the Rhinichthys cataractae… Ruskey, Jennifer Anne 2014

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

Item Metadata

Download

Media
24-ubc_2014_november_ruskey_jennifer.pdf [ 6.66MB ]
Metadata
JSON: 24-1.0167029.json
JSON-LD: 24-1.0167029-ld.json
RDF/XML (Pretty): 24-1.0167029-rdf.xml
RDF/JSON: 24-1.0167029-rdf.json
Turtle: 24-1.0167029-turtle.txt
N-Triples: 24-1.0167029-rdf-ntriples.txt
Original Record: 24-1.0167029-source.json
Full Text
24-1.0167029-fulltext.txt
Citation
24-1.0167029.ris

Full Text

Morphological stasis and geneticdivergence without reproductiveisolation in the Rhinichthys cataractaespecies complex: insights from a zoneof secondary contact in the lowerFraser Valley, British ColumbiabyJennifer Anne RuskeyB.Sc., Princeton University, 2007A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF SCIENCEinThe Faculty of Graduate and Postdoctoral Studies(Zoology)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)October 2014c© Jennifer Anne Ruskey 2014AbstractThe Nooksack dace (Rhinichthys cataractae putative subspecies; NSD) andlongnose dace (R. cataractae; LND) form a zone of post-glacial secondarycontact in three streams in British Columbia’s lower Fraser River valley,providing a valuable opportunity to study contact between populations sep-arated during the Pleistocene glaciations. They are morphologically cryptic,despite an estimated 2-3 million years of separation. The NSD is currentlylisted as Endangered under Canada’s Species at Risk Act (SARA) and mystudy clarifies its taxonomic and conservation status. NSD and LND havehighly divergent mitochondrial DNA types, and dace carrying each mtDNAtype have been found in roughly equal numbers in the zone of secondary con-tact. However, it was unknown whether this represented ongoing hybridiza-tion or reproductive isolation in sympatry. I conducted a morphologicalanalysis using 11 morphometric measurements and two meristic characters(N = 582, 23 sampling locations) to uncover any subtle variation betweenthe two dace, as well as to test for morphological intermediacy in the zone ofsympatry. I then employed a 10–locus microsatellite DNA assay (N = 374,12 sampling locations) to test for introgression between LND and NSD in thezone of secondary contact. I found that the two dace could not be reliablydistinguished: there was overlap in all morphological characters measured,and both morphological and microsatellite analyses showed a greater effectof location than mtDNA clade, even when restricted to allopatric popula-tions. There was no evidence of population structure within the sympatricpopulations, indicating complete admixture. The LND and NSD providean example of “ephemeral speciation”—two lineages which, despite longseparation, have developed no apparent barriers to reproduction and havecollapsed into a single interbreeding population where they come into sec-ondary contact. The zone of secondary contact should ideally be conservedfor its evolutionary significance, and is a good illustration of the complicatedpatterns of diversification caused by the Pleistocene glaciations. However,while the NSD should be protected as a distinct designatable unit, it shouldnot be considered a separate species, or even subspecies.iiPrefaceSeven of the microsatellite markers used in my study were developed specif-ically for this study and for further work on the Rhinichthys cataractaespecies complex. Work leading to the development of 25 microsatellite mark-ers for R. cataractae that distinguished the Nooksack and longnose dace wasconducted at Savannah River Ecology Laboratory by R. Beasley and Dr. S.L. Lance, with samples contributed by Dr. E. B. Taylor and myself. Thesemarkers have since been published as Beasley et al. (2014).iiiTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ixAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . xiDedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 The Rhinichthys cataractae species complex . . . . . . . . . 31.3 Secondary contact between the Nooksack dace and longnosedace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61.4 Cryptic species . . . . . . . . . . . . . . . . . . . . . . . . . . 91.5 Conservation status of the Nooksack dace . . . . . . . . . . . 101.6 Research objectives . . . . . . . . . . . . . . . . . . . . . . . 112 Morphometric Analysis of the Nooksack and Longnose dace(Cyprinidae: Rhinichthys) . . . . . . . . . . . . . . . . . . . . 122.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 122.2 Materials and methods . . . . . . . . . . . . . . . . . . . . . 142.2.1 Sampling . . . . . . . . . . . . . . . . . . . . . . . . . 142.2.2 Morphometric and meristic measurements . . . . . . 152.2.3 Statistical analysis of morphological data . . . . . . . 182.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222.3.1 Searching for new dace populations . . . . . . . . . . 222.3.2 Variation in vertebral counts . . . . . . . . . . . . . . 22ivTable of Contents2.3.3 Principal components analysis . . . . . . . . . . . . . 232.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412.4.1 Subtle morphological variation . . . . . . . . . . . . . 422.4.2 Alternative explanations for morphological variation . 442.4.3 Implications . . . . . . . . . . . . . . . . . . . . . . . 463 Genetic Analysis of Nooksack and Longnose Dace . . . . . 483.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 483.1.1 Background . . . . . . . . . . . . . . . . . . . . . . . 483.1.2 Genetic techniques . . . . . . . . . . . . . . . . . . . 503.1.3 Objectives for this chapter . . . . . . . . . . . . . . . 513.2 Materials and methods . . . . . . . . . . . . . . . . . . . . . 523.2.1 Sampling . . . . . . . . . . . . . . . . . . . . . . . . . 523.2.2 DNA extraction and amplification . . . . . . . . . . . 523.2.3 Genetic analyses—mitochondrial DNA . . . . . . . . 533.2.4 Genetic analyses—microsatellite data . . . . . . . . . 543.2.5 Admixture analyses (Structure) . . . . . . . . . . . 553.2.6 Analysis of molecular variance (AMOVA) . . . . . . . 563.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 583.3.1 Distribution of mitochondrial haplotypes . . . . . . . 583.3.2 Microsatellite analyses . . . . . . . . . . . . . . . . . 583.3.3 Admixture analysis . . . . . . . . . . . . . . . . . . . 633.3.4 AMOVA analysis . . . . . . . . . . . . . . . . . . . . 713.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 733.4.1 Similarity of sympatric R. cataractae to allopatric parentallineages . . . . . . . . . . . . . . . . . . . . . . . . . . 743.4.2 Admixture of LND and NSD in sympatric populations 753.4.3 Genetic differentiation throughout the range of R. catarac-tae . . . . . . . . . . . . . . . . . . . . . . . . . . . . 763.4.4 Implications for the taxonomic and conservation sta-tus of the Nooksack dace . . . . . . . . . . . . . . . . 784 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 794.1 Summary of findings . . . . . . . . . . . . . . . . . . . . . . . 794.2 Conservation implications . . . . . . . . . . . . . . . . . . . . 804.3 “Ephemeral” speciation . . . . . . . . . . . . . . . . . . . . . 82Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83vTable of ContentsAppendicesA Chapter 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99B Chapter 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108viList of Tables2.1 Description of morphometric and meristic measurements . . 162.2 Sampling effort and results in streams previously un-surveyedfor dace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232.3 Eigenvalues and percentage of variance explained by each axisof a principal components analysis conducted on allopatricpopulations of dace . . . . . . . . . . . . . . . . . . . . . . . . 262.4 Character loadings for the first five axes of the principal com-ponents analysis conducted on allopatric populations of dace 282.5 r2 values, F-ratios and P-values for nested ANOVAs con-ducted on the first five principal components of morphologicalsamples from allopatric populations of Rhinichthys . . . . . . 282.6 r2 values, F-ratios and P-values for nested ANOVAs con-ducted on the first five principal components of morphologicalsamples from all populations of Rhinichthys . . . . . . . . . . 292.7 F-ratios and P-values for ANOVAs conducted on the first fiveprincipal components of samples from Kanaka Creek . . . . . 302.8 Results for linear regressions comparing morphological vari-ables and environmental variables . . . . . . . . . . . . . . . . 392.9 F-ratios and P-values for ANOVAs comparing Rhinichthyssamples from east and west of the Rocky Mountains . . . . . 412.10 F-ratios and P-values for ANOVAs comparing LND samplesfrom east and west of the Rocky Mountains . . . . . . . . . . 413.1 Populations used in Structure analysis . . . . . . . . . . . 563.2 Summary of genetic data for the 12 locations used in mi-crosatellite analysis . . . . . . . . . . . . . . . . . . . . . . . . 613.3 Evanno table output from Structure Harvester for Struc-ture analysis conducted on all 12 dace populations . . . . . 643.4 Evanno table outputs from Structure Harvester for Struc-ture analysis conducted on Kanaka Creek, Coquitlam River,and Alouette River . . . . . . . . . . . . . . . . . . . . . . . . 70viiList of Tables3.5 AMOVA results comparing different groupings of allopatricand sympatric populations . . . . . . . . . . . . . . . . . . . . 72A.1 List of sampling locations for Nooksack and longnose daceused in morphological analyses . . . . . . . . . . . . . . . . . 99A.2 Raw data of vertebral counts from allopatric dace populations 103A.3 Eigenvalues and percentage of variance explained by each axisof a principal components analysis conducted on all popula-tions of dace . . . . . . . . . . . . . . . . . . . . . . . . . . . 104A.4 Character loadings for the first five axes of the principal com-ponents analysis conducted on all populations of dace . . . . 105A.5 Bayesian information criterion (BIC) values for each modeland number of clusters for the cluster analysis of principalcomponent scores from morphological analysis of samples fromall populations . . . . . . . . . . . . . . . . . . . . . . . . . . 106A.6 Bayesian information criterion (BIC) values for each modeland number of clusters for the cluster analysis of principalcomponent scores from morphological analysis of samples fromallopatric populations . . . . . . . . . . . . . . . . . . . . . . 107B.1 List of sampling locations for Nooksack and longnose dace(Rhinichthys cataractae) used in genetic analyses . . . . . . . 108B.2 Details for 10 polymorphic microsatellite loci developed forRhinichthys cataractae . . . . . . . . . . . . . . . . . . . . . . 110B.3 Observed (HE) and expected (HO) heterozygosities for 10microsatellite loci and 12 populations of Rhinichthys cataractae111B.4 Allele frequencies for 10 microsatellite loci at 12 samplinglocations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113B.5 Pairwise FST values for all sampling locations . . . . . . . . . 122B.6 Pairwise FST for all sampling locations, with each sympatricpopulation divided by mtDNA haplotype . . . . . . . . . . . 123viiiList of Figures1.1 Maximum extent of Pleistocene ice sheets in northwesternNorth America . . . . . . . . . . . . . . . . . . . . . . . . . . 41.2 Distribution of Nooksack dace (Rhinichthys cataractae puta-tive subspecies) in Canada and North America . . . . . . . . 82.1 Micro CT scan of longnose dace sample . . . . . . . . . . . . 172.2 Histogram showing distribution of vertebral counts for long-nose and Nooksack dace . . . . . . . . . . . . . . . . . . . . . 242.3 Variables factor map for the principal components analysis ofsize-transformed morphometric characters for all samples . . 272.4 Histogram showing lateral line scale count by mtDNA haplo-type for allopatric Rhinichthys samples . . . . . . . . . . . . . 312.5 Histogram showing lateral line scale count by mtDNA haplo-type for Kanaka Creek samples . . . . . . . . . . . . . . . . . 322.6 Scatter plot of dace morphological samples on the first twoprincipal components, with convex hulls enclosing each mclust-assigned cluster . . . . . . . . . . . . . . . . . . . . . . . . . . 342.7 Stacked bar chart showing the composition of each clustergenerated by the mclust analysis on all Rhinichthys samples . 352.8 Stacked bar chart showing the composition of each clustergenerated by the mclust analysis on only allopatric Rhinichthyssamples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362.9 Stacked bar chart showing the composition of each clustergenerated by the mclust analysis conducted on Rhinichthysfrom Kanaka Creek . . . . . . . . . . . . . . . . . . . . . . . . 373.1 Map of Nooksack and longnose dace cytochrome b mtDNAhaplotypes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 593.2 Factorial correspondence analysis plot based on 10 microsatel-lite loci for Rhinichthys cataractae . . . . . . . . . . . . . . . 623.3 Structure admixture analysis of Nooksack and longnosedace from all populations, K = 11 . . . . . . . . . . . . . . . 66ixList of Figures3.4 Structure admixture analysis of Nooksack and longnosedace from all populations, K = 3 . . . . . . . . . . . . . . . . 673.5 Structure admixture analysis of samples from CoquitlamRiver, K = 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 683.6 Structure admixture analysis of samples from Alouette River,K = 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69xAcknowledgementsI would first like to thank my supervisor, Dr. Eric Taylor, who has beena patient and inspiring source of guidance throughout my degree. I exitedevery meeting with you feeling like my work was more interesting and, cru-cially, more possible than I had going in! Thank you for helping me to figureout what doing science is all about. Your example as a rigorous researcher,conservation scientist and advocate, thoughtful (and occasionally enraged)Canadian citizen, gregarious community member, and full-on ichthyophilehas been inspiring from start to finish.My committee members, Darren Irwin and Peter Arcese, offered invalu-able insight and suggestions which broadened my perspective on my workand occasionally flipped my line of reasoning upside down. I only wish I hadtaken greater advantage of your advice.I owe a huge thank-you to Rebecca Seifert for her lab and field assistance.Thank you to Shannan McNally and Monica Yau for being my lab workteachers—I started out never having touched a pipette, and these two ladiesmade me a competent and enthusiastic lab rat. Thank you to Mike Pearson,Stephanie Avery–Gomm and Mike Champion for conversations about Nook-sack dace, and Don McPhail for laying the vast and invaluable groundworkthat underlies this thesis. Thank you to John Schipilow and Nancy Ford atthe UBC Center for High-Throughput Phenogenomics for helping with mymicro-CT scanning.Thank you to the members of the Taylor and Rosenfeld labs for yourhelpful ideas, discussion and support: Shannan McNally, Amanda Moreira,Monica Yau, Mike Champion, Sean Naman, Rebecca Piercey, Jill Miners,Katie Haman, Sarah Davidson, Carla Crossman, Matthew Siegle.When I was accepted to the UBC Zoology department, I had no ideajust how lucky I was to become a part of this incredible community. Aca-demically and socially, this is a wonderful place to be: rigorous, creative,collaborative, friendly, and with a commitment to science’s broader socialresponsibilities. The faculty, students, administrators, Beaty Museum staff,and incomparable IT department have all made these three years a greatexperience. Too many names to list, but I would like to especially thankxiAcknowledgementsthose involved in the Huts Skit, as well as my cohort from ZOOL502. Iwould also like to thank Rick Taylor, again, for giving me the opportunityto serve as the teaching assistant for BIOL 465; both of my semesters asDeputy Ruskey were an absolute privilege and pleasure.It’s no secret that this thesis has been extremely difficult for me tocomplete, and I considered quitting many times. Without the tremendousamount of help and support I received over the past three years I wouldnever have been able to finish. I am deeply indebted to my friends andfamily, as well as to two wise and compassionate counsellors who helped meto repair my mental health and learn an entirely new set of skills. I thinkthat the psychological struggles of graduate work are often swept under therug, despite their ubiquity, and I would like to acknowledge the people whowere particularly supportive to me during difficult times: Caitlin Kopperson,Jay Morritt, Daniel Wood, Kiri Sawyer, Ashley Waring, Andrew MacDon-ald, Chris Luft, Leanne Fessler, Laurie Kohl, Saskia Wolsak, Andrew Gillis,Gwylim Blackburn, Joel Heath, Albert Ruskey, Susan Ruskey, and FrankRuskey.Funding for this project was provided by NSERC Discovery and Equip-ment grants, Canadian Wildlife Federation Endangered Species Fund grant,and the Department of Zoology Teaching Assistantship.xiiDedicationThis thesis is dedicated to my family: Mom, Dad, and Albert.xiiiChapter 1Introduction1.1 BackgroundThe origin of new species, in its simplest formulation, involves clear dichoto-mous branching of a single lineage into two distinct lineages with separateevolutionary trajectories. However, in reality the process of speciation iscomplex and can take many different paths, with much study and debatecurrently dedicated to the roles of selection, adaptation, genetic drift, andgeographic isolation, among others (see Coyne and Orr, 2004). The pathto becoming a new species is rarely straightforward; incipient species mayremain in “limbo” for millions of years (e.g. Avise et al., 1998), may un-dergo parallel, reverse, or de-speciation (Turner, 2002; Taylor et al., 2006),and may diverge and re-merge, potentially multiple times, before ultimatelybecoming two stable species or collapsing back into one (Rosenblum et al.,2012). Episodes of recurrent vicariance and reconnection, such as thosecaused by the glacial cycles of the Pleistocene, can create complex patternsof divergence and incomplete speciation (e.g. Bernatchez and Wilson, 1998).Allopatric speciation occurs when geographic isolation disrupts gene flowand is followed by independent evolution of the separated populations, ulti-mately leading to reproductive isolation (Mayr, 1963). Divergence may bedriven by selection, or stochastic processes like genetic drift (Dobzhanskyand Dobzhansky, 1937). Molecular divergence of different populations caneasily be measured and mtDNA phylogeography has become a popular toolfor uncovering deep historical divergence between populations even when itmay be accompanied by little morphological differentiation (e.g. Bond et al.,2001; Krosch et al., 2009). However, although ecological and morphologicaldifferences and standard levels of genetic divergence have been suggested asmethods to determine whether allopatric lineages should be considered sepa-rate species (Thorpe, 1982; April et al., 2013), it is often difficult to reliablydescribe two groups as different species until they have been observed insympatry. When a population has been separated, barriers to reproductiveisolation may arise, or they may not. Some clades that differ very little inneutral molecular genetic traits may be strongly reproductively isolated (e.g.11.1. Backgroundsome African cichlid fishes, Moran and Kornfield, 1993), whereas others thatare deeply genetically diverged may still freely interbreed upon secondarycontact (e.g. ravens, Webb et al., 2011).When two lineages that have diverged in allopatry reconnect in a zone ofsecondary contact, the outcome is uncertain and provides an excellent oppor-tunity to examine the strength and causes of reproductive isolation (Bartonand Hewitt, 1985; Barton and Hewitt, 1989). The possible outcomes fallalong a spectrum: the two species may be completely reproductively iso-lated, in which case speciation can be considered complete (e.g. Pfenningerand Posada, 2002), or they may form hybrid zones of varying types, rangingfrom a bimodal hybrid zone with minimal interbreeding (e.g. Phillips etal., 2013) to a unimodal, fully introgressive hybrid swarm with nonexistentreproductive isolation (e.g. Avise and Saunders, 1984; Wiens et al., 2006).Fishes are widely considered an excellent group in which to study evo-lutionary questions and North American freshwater fishes in particular pro-vide many good opportunities in which to study hybridization and zonesof secondary contact (e.g. Hubbs, 1955; Knowlton, 1993; Bernardi, 2013).Many such zones of secondary contact formed in the wake of the Pleistoceneglaciations, which dramatically reformed the North American landscape.Ice covered most of Canada, and flora and fauna survived only in ice-freeglacial refuge areas, with previously continuous species distributions oftendivided into two or more refugia. The ice sheets reached their furthest extent∼ 24, 000 − 21, 000 years ago and began to substantially recede ∼ 14, 000years ago (Dyke et al., 2002); however, throughout the Pleistocene theyfollowed climatic cycles and receded and advanced multiple times (Imbrie,1985; Hewitt, 2004). In eastern North America, the two major refugia werethe Mississippian and Atlantic refugia—the Mississipian encompassing theupper Mississippi River basin south of the ice sheet, and the Atlantic alongthe east coast south of Long Island and east of the Appalachians (Bailey andSmith, 1981). In western North America, the three major refugia were thePacific refugium, the Missourian refugium and the Beringian refugium, withthe Pacific refuge existing along the west coast south of Vancouver Islandand west of the Rocky Mountains, the Missourian located further east inthe Missouri River basin, and the Beringian refuge encompassing the north-western corner of North America (see fig. 1.1) (Shafer et al., 2010). Theselarge refugia contained additional structure in the form of smaller refugia-within-refugia, and there also exist cryptic refugia outside of the boundariesof the major ones, which phylogeography can help identify (Stewart and Lis-ter, 2001; Provan and Bennett, 2008; Shafer et al., 2010). The distributionof refugia and the repeated cycles of glaciation dictated the movement and21.2. The Rhinichthys cataractae species complexsurvival of biota within North America, with particularly strong effects onfreshwater fishes, leading to complex phylogeography and many opportuni-ties to study recent or incomplete speciation (Lindsey and McPhail, 1986;Bernatchez and Wilson, 1998).1.2 The Rhinichthys cataractae species complexCyprinids are the largest family of vertebrates and the most diverse fresh-water fish family in North America (Dowling et al., 2002) and have been fre-quently used as study systems for evolutionary questions (Dawley and God-dard, 1988; Dowling and DeMarais, 1993). The genus Rhinichthys (or “riffledaces”) consists of eight living species that are endemic to North Americawhere they are broadly distributed in coolwater habitats (Matthews et al.,1982). Rhinichthys are relatively small bodied (maximum size of about 120mm), omnivorous fishes morphologically well-adapted to live on or near thebottom (see below). Several studies have investigated the systematics andbiogeography of blacknose dace (R. atratulus, Matthews et al., 1982; Fraseret al., 2005; Smith and Dowling, 2008; Tipton et al., 2011) and speckled dace(R. osculus, Oakey et al., 2004; Pfrender et al., 2004; Smith and Dowling,2008; Billman et al., 2010; Kinziger et al., 2011; Hoekzema and Sidlauskas,2014). R. cataractae is the most widely distributed species in the genus, andindeed the most widely distributed native North American minnow, rang-ing from the Atlantic to the Pacific coasts, and from the Arctic Circle tonorthern Mexico (McPhail and Taylor, 2009). It is likely that the longnosedace survived in at least three refugia (Mississippian, Atlantic, and Pacific)during the Pleistocene glaciations (Scott and Crossman, 1998). It occupiesvaried habitats with differing geological histories and is potentially an ex-cellent system for studying North American biogeography, and ecologicaland environmental issues from an evolutionary perspective (e.g. Girard andAngers, 2006).R. cataractae, the longnose dace, is a small benthic cyprinid that livesin shallow stream riffles with a loose gravel, cobble or boulder substrate.They prefer swift current and are morphologically adapted in a variety ofways for life at the bottom of fast-flowing streams: their body shape isstreamlined; they have large paddle-like pectoral fins which can be used ashydrofoils to maintain their position in swift current; and they have veryreduced swim bladders (Gee and Northcote, 1963). They have a long snoutwhich overhangs their mouth, and can reach 106 mm in standard length,averaging 94 mm as adults. They feed primarily upon aquatic insect larvae.31.2. The Rhinichthys cataractae species complexFigure 1.1: Extent of ice during the last glacial maxima. Ice layer is fromDyke et al. (2002). Major refugia and ice-sheets are labeled; inset mapshows the states and provinces of northwestern North America: in Canada,YT–Yukon Territory, NWT–Northwest Territory, BC–British Columbia,AB–Alberta. In the United States, WA–Washington, MT–Montana, OR–Oregon, ID–Idaho, WY–Wyoming. Figure from Shafer et al. (2010).41.2. The Rhinichthys cataractae species complexTheir most common food depends upon what is available in their location,and can include chironomid, midge, blackfly, and mayfly larvae (Scott andCrossman, 1998; McPhail and Taylor, 2009).R. cataractae spawn in the early summer, with estimates of actual spawn-ing time ranging from May to early July depending on the location (Scottand Crossman, 1998). There is little detailed information available abouttheir mating and spawning activities, but it is thought that they spawn inriffles over a gravelly bottom, and that while they do not dig a nest, theyform a territory and one parent guards the eggs. Females lay 200–2000eggs, depending on the size of the female, which hatch in 7–10 days. Theyoung dace are pelagic and live in calm waters near the shore, transitioningat approximately four months to their adult lifestyle in fast-moving water.Hybrids between R. cataractae and R. osculus and between R. cataractaeand Nocomis micropogon have been recorded (Smith, 1973; Scott and Cross-man, 1998).Although R. cataractae has a wide range and occupies diverse habitatsthroughout a continent with a complex biogeographic history, there has beenlittle population structure documented. By contrast, the blacknose dace,which occupies a small geographic range in eastern North America, hasbeen studied using both genetics and morphology and some have suggestedthat it consists of a complex of three distinct species: southern, R. obtusus;eastern, R. atratulus; and western, R. meleagris (Fraser et al., 2005; Tiptonet al., 2011). These different forms likely were isolated in separate glacialrefugia, as were lineages of R. cataractae, yet no continent-wide investigationof population structure has been done in R. cataractae. Rather, work to datein the R. cataractae complex has focused on diversity arising from “refugia-within-refugia” (Shafer et al., 2010) in the Pacific Northwest (McPhail andTaylor, 2009). The Umpqua and Millicoma dace have been identified asdistinct species within the R. cataractae species group, forming a distinctOregon coastal clade (McPhail and Taylor, 2009), and the Nooksack dace,endemic to southwest British Columbia and northwest Washington State,has been suggested as a fourth clade (McPhail, 1967, Taylor et al., unpubl.data).The Nooksack dace forms part of the Chehalis fauna, a group of animalsisolated in the Chehalis glacial refugium—south of Puget Sound and north ofthe Columbia River drainage—during the Pleistocene glaciations (McPhail,1997). It is currently found in several rivers in northwestern WashingtonState and adjacent portions of British Columbia in tributaries of the upperNooksack River and in several streams that drain to the lower Fraser River(Fig. 1.2). In Canada, the Nooksack dace is listed as Endangered under the51.3. Secondary contact between the Nooksack dace and longnose dacefederal Species at Risk Act (SARA). The Nooksack dace is morphologicallysimilar to the longnose dace; it can only be reliably distinguished genetically,although differing lateral line scale counts have been suggested to differen-tiate the two types (McPhail, 1967). Mitochondrial DNA sequence analysishas shown that Nooksack and longnose dace form reciprocally monophyleticclades that differ from each other by about 3% (cytochrome b) to 4% (ND2)mitochondrial DNA sequence divergence, implying a separation of 2-3 mya(Taylor et al., unpubl, data).1.3 Secondary contact between the Nooksackdace and longnose daceOur current understanding suggests that when the last glaciers receded,the Nooksack dace dispersed northwards to recolonize several tributariesof the lower Fraser River in British Columbia, following postglacial lakesat the edge of the receding ice (Fig 1.2) (McPhail, 1967). Streams withfish bearing only Nooksack dace mtDNA exist in four streams in BritishColumbia: Pepin Creek, Fishtrap Creek, Bertrand Creek (all tributaries ofthe upper Nooksack River), and the Brunette River (a tributary of the lowerFraser River). Additionally, fish that bear Nooksack dace mtDNA and fishthat bear longnose dace mtDNA are found in sympatry in three streamsthat drain to the north side of the Fraser River: the Coquitlam River, theAlouette River, and Kanaka Creek. These streams with sympatric Nooksackand longnose mtDNA represent a zone of postglacial secondary contact andan opportunity to study the degree of reproductive isolation between thetwo lineages. If reproductive isolation exists, the Nooksack and longnosedace may represent a case of cryptic species within Rhinichthys (see belowand McPhail and Taylor, 2009).Currently there has been no study of the streams with sympatric Nook-sack and longnose dace than a simple survey of mtDNA types, and there areseveral possibilities for the situation underlying the observed pattern. Mythesis is focused on evaluating three possible scenarios for the current statusof Nooksack and longnose dace1. Nooksack and longnose dace co-occur without interbreeding and be-have as distinct biological species2. Nooksack and longnose dace interbred completely upon secondary con-tact with extensive introgression such that there is no evidence of cur-rent reproductive isolation between types and the occurrence of two61.3. Secondary contact between the Nooksack dace and longnose dacemtDNA types represents a vestige of past isolation.3. Nooksack and longnose dace are incompletely reproductively isolatedsuch that there is ongoing hybridization and the genetic structureof each stream consists of a bimodal hybrid zone; primarily parentalforms present with some hybrids suggesting some degree of reproduc-tive isolation between Nooksack and longnose dace.Although the two lineages are thought to represent isolation in differentglacial refugia for two to three million years (Taylor et al. unpubl. data),the two fish remain morphologically very similar. Any ecological differentia-tion is unknown, and their degree of mtDNA differentiation puts them wellwithin the range of cyprinid lineages that have undergone complete specia-tion (Broughton and Gold, 2000; Dowling et al., 2002), but also lineages thathave not developed any significant reproductive barriers (Ward et al., 2005;Zemlak et al., 2009). Consequently, testing the degree of genetic differenti-ation in the nuclear genome in sympatry presents an excellent opportunityto assess the status of the Nooksack and longnose dace as distinct biologi-cal species (c.f. McPhail, 1984; Wilson and Bernatchez, 1998; Zamudio andSavage, 2003; Hey et al., 2004; Hyde et al., 2008).71.3.SecondarycontactbetweentheNooksackdaceandlongnosedaceFigure 1.2: Distribution of Nooksack dace (Rhinichthys cataractae) in Canada (left) and North America (right).(Modified from Pearson et al. (2008). 1—Brunette River, 2—Bertrand Creek, 3—Pepin Brook, 4—Fishtrap Creek.The black dots in the right panel represent all known populations of Nooksack dace; white squares indicate thelocations of major cities.81.4. Cryptic species1.4 Cryptic speciesEarly studies of the Nooksack and longnose dace uncovered subtle morpho-logical differences: a more slender caudal peduncle and fewer scales alongthe lateral line and around the caudal peduncle (Bisson and Reimers, 1977).However, it is nearly impossible to tell the two forms apart without re-sorting to scale counts or DNA analysis. Even scale counts themselves areunreliable, as there is overlap between Nooksack and longnose dace in allcounts. Their degree of morphological similarity in the presence of exten-sive mtDNA divergence places them within the category of potential crypticspecies. Cryptic species are defined as separate species that are difficult orimpossible to distinguish by morphology, and that may have been incorrectlylumped as a single species (Sa´ez and Lozano, 2005). This does not meanthat they are literally identical; these cryptic species may be morphologicallydistinguishable with the use of tools such as scanning electron microscopyand statistical morphometrics, and may rely upon non-visual cues to identifyconspecifics (e.g. Feulner et al., 2006). The key point is that there has gen-eral morphological stasis: species diversification, decoupled from ecologicaland/or morphological change (Bond et al., 2001).The rate of identification of cryptic species has increased enormouslysince the advent of PCR and affordable genetic sequencing (Bickford et al.,2007) and they are an important component of undescribed biodiversity.The discovery of cryptic species can have major conservation implications:in many cases, a species thought to have a wide distribution has been shownto be a cryptic species complex, with each species having a very small distri-bution and therefore being at much greater risk of extinction (e.g. Ravaoa-rimanana et al., 2004). Detection of cryptic species can increase diversityestimates within a group by over 100% in taxa as diverse as, for exam-ple, frogs (Funk et al., 2012), wasps (Smith and Dowling, 2008), and bats(Mayer and Helversen, 2001). Cryptic species can therefore have a hugeeffect on biodiversity estimates, making estimation of cryptic species distri-bution important for identifying global biodiversity hotspots (Trontelj andFiˇser, 2008). Recently, a population of speckled dace (Rhinichthys oscu-lus) in Oregon was proposed to consist of three cryptic species (Hoekzemaand Sidlauskas, 2014). If the Nooksack dace and longnose dace are repro-ductively isolated from one another, the fact that they are morphologicallycryptic implies that non-visual modes of mate selection are important forthese fish.Limited dispersal abilities and fragmented landscapes have frequentlybeen put forward as potential drivers of cryptic speciation (e.g. Knowlton,91.5. Conservation status of the Nooksack dace1993; Kelly et al., 2006; Leavitt et al., 2007; Bryson et al., 2011). A varietyof studies suggest a similar basic pattern: taxa with low dispersal potentialare separated in a fragmented landscape in which each fragment is eco-logically equivalent, and undergo divergence without morphological change(Knowlton, 1993). As small freshwater fish, R. cataractae have limited dis-persal ability, and their landscape is fragmented by watershed boundaries,as well as having been historically fragmented by repeated glaciation. Inthe Oregon coastal clade of R. cataractae, there are two lineages of dace:the Umpqua dace, which is located in the drainages of the Umpqua andSmith Rivers, and the Millicoma dace, which is located in the Coos River,just south of the range of the Umpqua dace (McPhail and Taylor, 2009).Both are derived from Columbia River longnose dace, which live in water-sheds north of the Smith River, but the relationships between the threewere a puzzle: the Millicoma dace is morphologically extremely similar tothe longnose dace, despite having the range of the morphologically divergentUmpqua dace located in between (McPhail and Taylor, 2009).McPhail and Taylor (2009) constructed a mtDNA phylogeny to resolvethis question and found that the Umpqua and Millicoma dace are sisterspecies, with their common ancestor having diverged from the Columbialongnose ancestor 3–4 mya, and the Millicoma dace splitting from the Umpquadace 1.5–2.0 mya. They suggested that the morphological divergence of theUmpqua dace took place after the separation of the Millicoma dace to theCoos River system, and that dace in this smaller, depauperate environmentretained ancestral characteristics. Though the range of the Nooksack dacecovers a much larger and biologically complex area than the Coos Riverwatershed, it may be that— similar to the Millicoma dace—the Nooksackdace has not been in an environment with opportunities for morphologicaldivergence.1.5 Conservation status of the Nooksack daceWithin Canada, the Nooksack dace is listed as Endangered under the Speciesat Risk Act (COSEWIC, 2007). Their population in Washington State ap-pears to be stable, and they are not listed under the American EndangeredSpecies Act (ESA); however, in Canada their extremely limited range isthreatened by habitat destruction, and they have been extirpated from twostreams: Howes Creek and Cave Creek (McPhail, 1997). While the Nook-sack dace certainly represents a distinct evolutionary lineage of R. catarac-tae and will likely continue to be assessed as a Designatable Unit (DU, a101.6. Research objectivessubspecies, variety, or genetically or geographically distinct population thatmay be assessed by COSEWIC, the Committee on the Status of Endan-gered Wildlife in Canada (COSEWIC, 2014)) regardless of how its speciesstatus is resolved, any further information about its level of reproductiveisolation from the longnose dace and level of admixture in the streams withsympatric dace will help inform decisions about whether the three streamswith sympatric dace should be considered part of the range of the NSD andprotected as such, and whether interbreeding with LND constitutes a threatto the purity of the NSD gene pool.1.6 Research objectivesThe overarching goal of my thesis research is to clarify the species status ofthe Nooksack dace by examining its relationships to the longnose dace usinggenetic and morphological data. Specifically, my objectives are:1. Determine whether the Nooksack and longnose dace can be reliablydistinguished morphologically, and whether there is morphological in-termediacy within the zone of secondary contact using robust mor-phometric and meristic analysis with samples in sympatry and fromthroughout the range.2. Provide a more rigorous assessment of the geographic distribution ofthe two clades of mtDNA.3. Test for genetic divergence between Nooksack and longnose dace insympatry using ten microsatellite markers from the nuclear genome,and to assess the degree of introgression between the Nooksack daceand longnose dace and whether or not reproductive barriers exist be-tween them.4. In light of the above, assess the status of the Nooksack dace as adesignatable unit within the R. cataractae complex.11Chapter 2Morphometric Analysis ofthe Nooksack and Longnosedace (Cyprinidae:Rhinichthys)2.1 IntroductionThe longnose and Nooksack dace are divergent forms of Rhinichthys catarac-tae. The longnose dace (LND) is found in freshwaters across North America,while the Nooksack dace (NSD) is found in several streams in southwesternBritish Columbia (BC) and adjacent portions of Washington State south tothe Columbia River (McPhail, 1967, Fig 1.2). Genetic analysis of LND andNSD indicates two deeply diverged clades, with 3-4% sequence divergenceof mitochondrial cytochrome b and ND2 mitochondrial DNA (mtDNA), in-dicating a divergence time of 2-3 mya (Taylor et al., unpubl. data). Despitethis long separation, initial morphological analysis showed little differenceexcept in lateral line and caudal peduncle scale counts, and caudal peduncledepth (McPhail, 1967). A more detailed morphological analysis may un-cover subtle differentiation, and if separate morphological forms are foundto correspond with the mtDNA clades, it will provide further support for thetwo groups being designatable units (DUs) under Canada’s Species at RiskAct and perhaps distinct species. Additionally, there are three streams insouthwestern BC in which the two mtDNA clades exist in sympatry (Taylorunpubl. data). If morphology still corresponds with mtDNA where the twoclades exist in sympatry, it would provide strong evidence that the two aredistinct species.No multivariate morphometric analysis has been carried out to date onthe R. cataractae complex, and it is an important step to fully describethe putative Nooksack dace species. If there is no consistent morphologicaldistinction between the two groups, they may be considered cryptic species:122.1. Introductionseparate species which are difficult or impossible to distinguish by morphol-ogy, and have been incorrectly lumped as a single species (Sa´ez and Lozano,2005).A multivariate morphological analysis will establish whether the Nook-sack and longnose dace are truly morphologically cryptic, and if furthergenetic analysis (Chapter 3) establishes them as separate species, they mayqualify as cryptic species. If so, it would add support to the theory thatlimited dispersal abilities and fragmented landscapes are drivers of crypticspeciation (Knowlton, 1993; Kelly et al., 2006; Leavitt et al., 2007; Bryson etal., 2011). If any consistent, fine scale morphological differentiation exists, itmay provide clues to ecological differentiation or the sources of reproductiveisolation between species (e.g. Feulner et al., 2007).There are instances of similar levels of morphological overlap betweenother dace species: for example, the eastern (Rhinichthys atratulus) andwestern (Rhinichthys obtusus) blacknose dace (Fraser et al., 2005). Theclassification of different forms in the R. atratulus complex has switchedback and forth from multiple subspecies to a single species, and a rigor-ous morphological study was needed to see if the different groups could bereliably distinguished (Fraser et al., 2005). A univariate and multivariatemorphometric analysis found that eastern and western blacknose dace couldnot be differentiated morphologically in allopatric populations, nor in sym-patric populations, where it was hypothesized that character displacementwould lead to greater morphological differentiation (Fraser et al., 2005).Perhaps the same scenario applies to the Nooksack and longnose dace.In this chapter, I report the results of a morphological analysis of 582longnose dace (LND) and Nooksack dace (NSD) from 45 sampling sites inWashington State, Michigan, Oregon, British Columbia, Alberta, Manitoba,and Ontario, comprising 25 populations. Of the sampling locations, eightare allopatric for NSD; 14 are allopatric for LND; and three are sympatricLND and NSD. The analysis consisted of multivariate statistical analysisof 11 morphometric and three meristic characters to test for differentia-tion between the two putative species, and to see if the sympatric mtDNApopulations were also morphologically distinct which would be consistentwith the existence of two species of dace. Alternatively, if the sympatricpopulations showed any sign of morphological intermediacy this would beconsistent with the idea that there is no reproductive isolation between LNDand NSD and that they readily interbreed when sympatric.I also tested for other possible explanations of observed morphologicaldifferentiation. Morphological variation along environmental gradients suchas salinity, or latitudinal gradients (e.g. McDowall, 2003; Floeter et al., 2004)132.2. Materials and methodshas been well documented, and intraspecific diversification arising from di-vergent selection has been widely studied in fishes. Differentiation arisingfrom divergent selection can be due to genetic differentiation (e.g. Schluterand McPhail, 1992; Jonsson and Jonsson, 2001) or plasticity (e.g. Wood andBain, 1995) and has been of use in defining and identifying fisheries stocks(Swain and Foote, 1999). Wood and Bain (1995) used regression analysis todemonstrate significant relationships between body morphology and micro-habitat use in a variety of cyprinid and percomorph fishes in the AlabamaRiver watershed; on a larger scale, Salini et al. (2004) showed a disjointbetween morphological variation and genetic variation in Tenualosa ilishain five countries, and concluded that the wide morphological variation wasdue to environmental differences.Many studies demonstrate that considerable intraspecific variation acrossthe species’ range is due to environmental conditions. Although R. catarac-tae does not show considerable morphological variation across its range(Scott and Crossman, 1998), I wanted to investigate whether there mightbe environmental correlates to morphological variation. While it was im-possible to gather environmental data first-hand from all sampling sites,Geographic Information System (GIS) and Remote Sensing (RS) databasesprovide a great resource for combining fish distribution data and surveyedor remotely sensed environmental data. These tools have been used exten-sively in conservation and in fisheries management and research to identifyessential habitat over large spatial scales (Valavanis et al., 2004). I used pub-licly available GIS databases to obtain environmental data for each samplinglocation.2.2 Materials and methods2.2.1 SamplingThe samples used in my study were drawn from the Beaty Biodiversity Mu-seum’s fish collections, as well as samples I collected myself in 2012. Samplesfrom the Beaty Biodiversity Museum include 532 fish from 23 streams, withcollection dates ranging from 1954 to 2011 (Appendix A, Table A.1). Aminimum of 10 fish were measured from each stream.In August and September 2012, I collected 50 fish from seven sites inthe three streams that had been previously identified as having both NSDand LND mtDNA: the Coquitlam and Alouette rivers, and Kanaka Creek.I collected fish using both electrofishing and kick-seining. All fish were sac-rificed using an overdose of MS-222 and stored in 95% ethanol. Altogether,142.2. Materials and methods582 fish from 23 locations were measured. Fish were only measured if theirstandard length was greater than 4 cm to ensure that all fish measured wereadults. Fish were collected under Ministry of Environment fish collectionpermit #SU12-81471; because the Coquitlam River, Alouette River, andKanaka Creek are not currently recognized as containing Nooksack dace,permits were for longnose dace only.During my sampling, I also tried to uncover any hitherto undiscoveredpopulations of dace that could potentially be sympatric for NSD and LNDmtDNA. I hoped to obtain finer-grained detail about the boundaries of theputative hybrid zone, and to obtain samples from streams that had not yetbeen searched for Nooksack or longnose dace. As such, I electrofished inevery stream (N = 5) that met the following criteria:• It had no previously documented dace population• It provided appropriate dace habitat: clean, shallow, fast-moving wa-ter with a rocky substrate ranging in size from gravel to boulders (Scottand Crossman, 1998).• It lay between the Brunette River in the east (the most easterly pureNSD river on the north side of the Fraser River) and Norrish Creek inthe west (the most westerly pure LND river on the north side of theFraser River.)2.2.2 Morphometric and meristic measurementsI measured 12 morphometric characters on the left side of each specimen,following Hubbs et al. (1958) and made three meristic counts: lateral linescale count, pectoral fin ray count, and vertebral count (Table 2.1). Mea-surements were made using Vernier dial calipers, and a dissecting microscopewhen necessary.To obtain vertebral counts, fish were scanned using at the UBC Cen-tre for High-Throughput Phenogenomics using a Micro-CT 100 scanner(Scanco Medical AG, Brttisellen, Switzerland). Scans were conducted withan isotropic voxel size of 49.2µm; energy of 90 kVp; intensity of 200µA;integration time of 150 ms; and filter of 0.5 mm.All raw scan data were converted into DCM (DICOM—Digital Imag-ing and Communications in Medicine) format and then imported into Mi-croView v. 2.5.0 (Parallax Innovations 2014). Scan data is stored as 2Dcross-sectional slices, so Microview was used to convert these slices into a3D image, from which vertebral counts were conducted by eye (Fig. 2.1).152.2. Materials and methodsTable 2.1: Description of morphometric (M1-M12) and meristic (Count 1-3) measurements examined in populations of Nooksack and longnose dace(Rhinichthys cataractae).Measure/ CountNumberView Description of MeasurementM1 Dorsal Distance between the eyesM2 Dorsal Width at dorsal fin originM3 Dorsal Width at origin of caudal finM4 Left side Distance from dorsal fin insertion to back side ofeyeM5 Left side Distance from dorsal fin insertion to pectoral fininsertionM6 Left side Distance from dorsal fin insertion to pelvic fininsertionM7 Left side Distance from dorsal fin insertion to anal fininsertionM8 Left side Distance from dorsal fin insertion to bottom ofcaudal fin insertionM9 Left side Distance from snout to front side of eyeM10 Left side Distance from pectoral fin insertion to anal fininsertionM11 Left side Width of caudal peduncle at caudal fin insertionM12 Left side Distance from snout to end of scales on caudalpeduncleCount 1 Left side Lateral line scale countCount 2 Left side Pectoral fin ray countCount 3 CT scan Vertebral count162.2.MaterialsandmethodsFigure 2.1: Micro CT scan of longnose dace sample CL018 (Columbia River, BC, length = 87.72 mm). Vertebralcount was conducted by eye.172.2. Materials and methods2.2.3 Statistical analysis of morphological dataThe first priority of this chapter was to determine whether NSD and LNDform different morphological groups, and whether populations from streamswith sympatric dace contain individuals that are morphologically interme-diate. Thus, analyses were first performed on allopatric populations, to ad-dress the question of differences between pure NSD and LND. Subsequently,analyses were performed on (1) all dace, and (2) the subset of samples fromstreams with sympatric dace in which mtDNA haplotype had been matchedwith morphology. Analysis of group (2) attempted to unravel whether therewas an association between morphology and mtDNA haplotype.Size-standardizationIn morphometric analyses, it is often observed that the size of body partsscale allometrically with body size; consequently, when carrying out a mor-phometric analysis of shape, it is critical to remove the effect of size variationamong samples (Reist, 1985; Albrecht et al., 1993; Lleonart et al., 2000).Reist (1985) recommended allometric scaling to a standard size, a techniquewhich has been used in many subsequent studies (Elliott et al., 1995; Fraseret al., 2005; Østbye et al., 2005), and which I have also chosen to use inmine. As such, to remove the effect of size, I standardized all morphometricmeasurements using the equation:Ms = Mo(LsLo)b (2.1)Where Ms = standardized measurement, Mo = measured characterlength, Ls = overall mean standard length for all fish, Lo = standard lengthof specimen, and b was estimated for each character using the allometricgrowth equationM = aLb (2.2)I estimated parameter b as the slope of the regression of log Mo on logLo. All samples were used to estimate b, but the intercept was allowedto vary between groups (streams allopatric for NSD, allopatric for LND,and sympatric for both NSD and LND mtDNA haplotypes.) Unless other-wise stated, all subsequent analyses utilize size-standardized morphometricmeasurements. No meristic counts were size-transformed; I tested for rela-tionships between counts and standard length but found none (r between0.02 and 0.1, all P > 0.1) These calculations, as well as all further analyses,were conducted using the R statistical environment (R Core Team, 2013).182.2. Materials and methodsThe size-standardization was carried out using the average of M12 (stan-dard length) as the size to standardize all fish to; therefore, following size-standardization, all fish had the same value for M12 and it was discardedfrom all further analyses.Principal components analysisMultivariate statistical analysis can simplify a great number of measure-ments into more useful and informative summaries across a smaller num-ber of composite variables (Dunteman, 1989). Most morphometric studieslooking to distinguish two or more groups use principal components anal-ysis (PCA) as a first step (Chapleau and Pageau, 1985; Edge et al., 1991;Østbye et al., 2005; Egge and Simons, 2006). I performed a principal compo-nents analysis on the 11 morphometric and two meristic characters (vertebralcounts were uninformative and not used—see below), and used the princi-pal components to conduct subsequent analyses. All principal componentsanalyses were performed on the correlation matrix using the FactoMineRpackage for R, v.1.24 (Husson et al., 2013). I used the Jolliffe cut-off cri-terion as a guide to deciding how many principal components to retain forsubsequent analyses, discarding any principal component whose eigenvaluewas less than one (Jolliffe, 2002).ANOVAI performed nested analysis of variance (ANOVA) on principal component(PC) scores with sampling location as a subgroup of putative species, tosee whether differences in morphology between NSD and LND persisted af-ter accounting for geographic variation within putative species. Putativespecies was a fixed factor and sampling location was a random factor. Igrouped samples according to mtDNA clades present in the stream: NSDand LND for the allopatric populations, and “both” for streams with sym-patric mtDNA types. I first performed nested ANOVAs on samples fromonly allopatric locations, to test for morphological differentiation betweenpure forms of NSD and LND. Next, I analyzed samples from all locations,to see whether sympatric dace fell into a group of their own.A subset of samples from Kanaka Creek had been analyzed for mtDNAhaplotype as well as morphology (N = 44). In order to test whether there wasa significant association between mtDNA type and morphotype in streamswith sympatric dace, I performed single-factor ANOVAs on the PCs for thesesamples, grouped by mtDNA type. Lateral line scale count has been identi-192.2. Materials and methodsfied in the past as a divergent character between LND and NSD (McPhail,1967) and I performed ANOVAs on this character separately. I performeda nested ANOVA on all samples, with population as a subgroup of puta-tive species, and a single-factor ANOVA on samples from Kanaka Creek,grouped by mtDNA type.Cluster analysisClustering is a form of data analysis that places objects into “natural” groupswithout any predefined groups being set (Fielding, 2007). I performed a clus-ter analysis on the morphological data to determine the number of clusterspresent, using R package mclust, v.4 (Fraley et al., 2012). mclust fits a seriesof different Gaussian mixture models to the data and tests the fit of eachmodel with different numbers of clusters. mclust then selects the model andnumber of clusters that maximizes the Bayesian information criterion (BIC),a model-selection criterion based on the likelihood function and accountingfor overfitting by applying a penalty for the number of model parameters(Fraley et al., 2012).In addition to the analyses on the complete dataset and the set of al-lopatric populations only, I also performed analyses on a subset that con-tained only the streams sympatric for NSD and LND mtDNA, and individualanalyses on each sympatric population, in order to test for morphologicaldifferentiation in sympatry. Again, for the Kanaka Creek samples, not onlycould I use mclust to test for the presence of two or more morphologicalgroups of dace, but I could test for an association between morphologicalclusters and mtDNA type because these fish had mtDNA type and morphol-ogy matched by individual.Discriminant analysisDiscriminant analysis (DA) is a common classification method for multi-variate data. It uses one or more predictor variables to predict a categoricaldependent variable; unlike in cluster analysis, DA is used when groups areknown or assumed a priori. Data for which the grouping is already knowncan be used to “train” the DA, by finding the combination of the predictorvariables which maximizes the difference between predefined groups. Thiscan then be used to classify samples for which the grouping is not known.Traditionally, discriminant analysis creates a linear combination of the pre-dictor variables; however, the mclust package in R (Fraley et al., 2012) usesa model-based approach to recombine the predictor variables for maximum202.2. Materials and methodsdiscriminatory power.Because cluster analysis does not assume a particular number of clus-ters, and does not include a priori knowledge of grouping, it may uncovervariation due to location, or other factors. Performing a discriminant anal-ysis with a priori knowledge of species grouping constrains the analysis tomaximize discrimination between species groups, even if other factors maybe significant contributors to morphological variation. In this sense it is abetter test than cluster analysis of whether Nooksack and longnose dace canbe distinguished morphologically.I trained the discriminant analysis on half of the samples from allopatriclocations and tested its discriminatory ability on the other half. I also ap-plied the discriminant function obtained in the first step to sympatric sam-ples matched with mtDNA, to see if the DA could correctly differentiatebetween LND and NSD mtDNA types in sympatry.Analysis of environmental variablesTo investigate the possibility that any morphological differentiation was afunction of environment rather than lineage, I ran linear regressions of PCs1–5 for all dace populations against the following variables:1. Watershed area2. Annual mean temperature3. Maximum temperature of the warmest month4. Minimum temperature of the coolest month5. Temperature range6. Annual precipitation7. Precipitation of Wettest Quarter8. Precipitation of Driest Quarter9. Date collectedChoice of variables to test was determined by data availability as well asdemonstrated relevance in previous studies. Temperature (variables 2–5) isa factor in Baltic herring morphology (Clupea harengus, Jørgensen et al.,2008) as well as many other fishes; it may affect phenotype directly or in-directly (e.g. by affecting prey availability). Watershed area, temperature212.3. Resultsand precipitation can all affect species composition in streams, potentiallyaffecting both prey availability and predation pressure on Rhinichthys, bothof which can affect morphology (Jackson et al., 2001). Collection date wasincluded mainly to check whether there had been deterioration of older sam-ples.To obtain this data, I used raster GIS data from WorldClim (Hijmans etal., 2005, http://www.worldclim.org) and shapefiles from the United StatesGeological Survey (United States Geological Survey, 2014) and DataBC(British Columbia Provincial Government, 2014). I joined this data to mysampling site locations using the program QGIS (QGIS Development Team,2009), and ran linear regressions in R. I also performed ANOVAs compar-ing the means of PC 1–PC 5 and lateral line scale count for dace foundeast and west of the Rocky Mountains, as this is a geographical barrier of-ten associated with morphological differentiation, and has been associatedwith differences in mating behaviour in subspecies of R. cataractae (Bartnik,1970). To avoid confounding the effects of putative species and geography(as there are no NSD found east of the Rocky Mountains), I also comparedsolely allopatric LND populations from east and west of the Rocky Moun-tains.2.3 Results2.3.1 Searching for new dace populationsAfter sampling in five streams that met my criteria across a total of fourdays, I did not find any new populations of dace (Table 2.2).2.3.2 Variation in vertebral countsI scanned an initial sample of 53 fish (17 from allopatric NSD populations,36 from allopatric LND populations) in the micro-CT scanner and visualizedthem in the computer program MicroView, then counted vertebrae by eye.An ANOVA conducted on the vertebral counts, however, was non-significant(F = 0.1143, P > 0.7), indicating no difference between vertebral counts forNSD and LND, and there was almost complete overlap of vertebral countsfor fish from the two mtDNA clades (see Fig. 2.2). Given this initial sample,vertebral counts did not seem to be an informative measurement and I didnot scan any further samples. The average vertebral count (±SD) was 35.7(±0.66) and 34.7 (±0.93) for LND and NSD, respectively (see Appendix A,222.3. ResultsTable A.2 for counts by location). Vertebral counts were not used in anyfurther analyses.2.3.3 Principal components analysisFor the PCA performed on the set of all samples, the mean eigenvalue acrossall PC axes was ∼ 0.92 and served as the Jolliffe cut-off criterion, meaningthat the first five principal components were retained. This selection wasalso largely in accordance with the rule-of-thumb of discarding PCs witheigenvalues less than 1; PCs 1-4 had eigenvalues greater than 1, and PC5 had an eigenvalue of ∼ 0.95. Thus, the first five principal componentswere retained for analyses and explained a total of 58.5% of the morpho-logical/meristic variance: PC 1, 18.3%; PC 2, 13.8%; PC 3, 10.2%; PC 4,8.8%; PC 5, 7.4% (Appendix A, Table A.3). Size-standardization removedany effect of size, so the PCs represent shape differences.In PC 1, M2 (width at dorsal fin insertion) and M6 (dorsal insertion to pelvicinsertion) had the highest positive loadings, with lateral line scale count andM9 (snout to eye distance) and M10 (pectoral insertion to anal insertion)had the highest negative loadings (see Fig. 2.3). In PC 2, M7, M8 (mea-sures of distance posteriorly from dorsal insertion), M11 (caudal peduncledepth) and lateral line scale count had the highest positive loadings and M1(interorbital width) and M5 (dorsal insertion to pectoral insertion) had theTable 2.2: Sampling of previously un-surveyed streams to try and detectnew Nooksack and/or longnose dace (Rhinichthys cataractae) populations insouthwestern British Columbia. Location, survey date, latitude/longitude,sampling method, and length of stream reach sampled (m) are provided.Location SurveyDateLatitude Longitude SamplingMethodStreamReachSampled(m)Cascade Creek 10/10/2012 49.41556 -122.45305 electrofisher 50Gaudin Creek 9/9/2012 49.15681 -122.33160 electrofisher 50Lagace Creek 10/10/2012 49.21656 -122.23520 kick-seine 30WhonnockCreek10/11/2012 49.21964 -122.45782 electrofisher 50Scorey Creek 9/9/2012 49.24778 -122.25166 electrofisher 50Scorey Creek 9/20/2012 49.20019 -122.24670 electrofisher 50232.3. ResultsFigure 2.2: Histogram showing distribution of vertebral counts for longnoseand Nooksack dace (Rhinichthys cataractae) collected from six streams rang-ing from British Columbia to Ontario (N = 53.) Grey bars signify longnosedace, black bars signify Nooksack dace.242.3. Resultshighest negative loadings (Fig. 2.3). In PC 3, M3 (body width at caudalpeduncle) had the highest positive loading and M5 and M10 the lowest; PC4 was dominated by high positive loadings on M9 and M10. Finally, PC5was dominated by a high positive loading on pectoral fin ray count and anegative loading on M8 (dorsal insertion to bottom of caudal fin insertion(see Appendix A, Table A.4 for complete character loadings).For the PCA performed on allopatric populations only, the first five PCswere again retained for analysis, again according both to the Jolliffe cutoffand the convention of removing PCs with eigenvalues less than one. Thefirst five PCs accounted for 59.3% of the variance: PC 1, 19.9%; PC 2,12.7%; PC 3, 10.2%; PC 4, 8.5%; and PC 5, 7.9% (Table 2.3).In PC 1, M1 (interorbital width), M2 (width at dorsal fin origin) andM6 (dorsal fin insertion to pelvic fin insertion) had the highest loadings, andM10 (pectoral fin insertion to anal fin insertion) and lateral line scale counthad the most negative loadings. In PC 2, M7 (dorsal fin insertion to anal fininsertion) and M11 (caudal peduncle depth) had high positive loadings; inPC 3, M3 (body width at caudal peduncle) has a high positive loading, andM5 (dorsal fin insertion to pectoral fin insertion) and M10 had very negativeloadings. In PC 4, M9 (snout to eye) had a high loading and pectoral fin raywas most negative, and in PC 5, pectoral fin ray was highest and M8 (dorsalfin insertion to bottom of caudal fin insertion) was lowest (Table 2.4).Variation in body morphology and meristicsFor both datasets—only allopatric populations, and all populations—nestedANOVAs were performed on the first five principal components, with thesamples grouped by species and localities nested within species. The resultsof the nested ANOVAs conducted on allopatric populations showed thatboth putative species and sampling location had significant effects on allprincipal components, except for the effect of putative species on PC 4 (P= 0.103). Typically, location accounted for a greater percentage of the totalmorphological variation than putative species identity. For PC 1, r2 valuesshow that species accounted for 11.6% of the total variation, and locationaccounted for 39.3% (Table 2.5). For PC 2, putative species accounted for21.9% of the variation, and location accounted for 25.1%. For PC 3–PC5, putative species accounted for 0.5–1.8% of the variation, and locationaccounted for 20.0–35.3% (Table 2.5).The results of the nested ANOVA conducted on PCs for all sampling lo-cations showed that putative species and sampling location had significanteffects on all principal components, again with the latter factor typically252.3. ResultsTable 2.3: Eigenvalues and percentage of variance explained by each axisof a principal components analysis conducted on 11 size-transformed mor-phological traits and two meristic traits values of morphological samplesfor allopatric populations of Nooksack dace and longnose dace (Rhinichthyscataractae). N = 414.PC Eigenvalue Percentage of variance Cumulativepercentage of variance1 2.58 19.91 19.912 1.65 12.74 32.663 1.32 10.21 42.874 1.10 8.48 51.355 1.02 7.90 59.266 0.91 7.00 66.267 0.83 6.41 72.688 0.74 5.71 78.399 0.71 5.47 83.8710 0.66 5.10 88.9811 0.54 4.16 93.1512 0.49 3.81 96.9613 0.39 3.03 100.00262.3. ResultsFigure 2.3: Variables factor map for the principal components analysis ofsize-transformed characters for all samples of longnose dace and Nooksackdace (Rhinichthys cataractae), showing the variable loadings for the firsttwo PCs (“Dim1” and “Dim2”). The arrows represent the relative characterloadings along each axis simultaneously.272.3. ResultsTable 2.4: Character loadings for the first five principal components of thePCA conducted on 11 size-transformed morphological traits and two meris-tic traits values of morphological samples for allopatric populations of Nook-sack dace and longnose dace (Rhinichthys cataractae).Measurement PC 1 PC 2 PC 3 PC 4 PC 5M1 0.676748 −0.056530 0.356530 −0.122115 −0.002919M2 0.654676 0.114546 0.390251 0.134502 −0.038168M3 0.159775 −0.074513 0.581263 0.357333 0.167530M4 0.577557 −0.074360 −0.389468 −0.168274 0.199190M5 0.555256 −0.035977 −0.407105 0.080368 0.132872M6 0.656964 0.414734 −0.143638 −0.047379 −0.240394M7 0.149575 0.682831 −0.160072 0.052721 −0.225181M8 −0.225344 0.280736 0.287599 −0.374290 −0.567322M9 −0.187815 0.353676 0.158705 0.640619 0.121196M10 −0.383003 0.432158 −0.409583 0.318238 0.008586M11 0.270543 0.625259 0.134467 −0.097243 0.239235Lateral line scalecount−0.504469 0.326214 0.206894 −0.276134 0.140174Pectoral fin ray count −0.143405 0.320021 0.101861 −0.405594 0.647223Table 2.5: r2 value, F-ratios and P-values for nested ANOVAs conducted onthe first five principal components of morphological samples from allopatricpopulations of Rhinichthys cataractae: allopatric longnose dace (LND) andallopatric Nooksack dace (NSD). Sampling location was nested within pu-tative species. For each PC, statistics are given for putative species (firstline) and sampling location nested with putative species (second line).PC r2 F P1 Species 11.6% 92.84 < 0.0001Location 39.3% 18.57 < 0.00012 Species 21.9% 163.15 < 0.0001Location 25.1% 10.99 < 0.00013 Species 1.8% 11.07 0.000962Location 35.3% 13.02 < 0.00014 Species 0.5% 2.675 0.103Location 22.4% 6.737 < 0.00015 Species 1.3% 6.31 0.0124Location 20.0% 5.918 < 0.0001282.3. Resultsaccounting for a greater percentage of the total variation. For PC 1, speciesaccounted for 11.8% of the variation, and location accounted for 35.9% (Ta-ble 2.6). The only instance of species accounting for a greater proportionof the variance than location was for PC2: species accounted for 29.9%,location for 22.3%. For PC 3–PC 5, species accounted for 1.0–3.8% of thevariance, and location for 17.5–28.6% (Table 2.6).Table 2.6: r2 value, F-ratios and P-values for nested ANOVAs conductedon the first five principal components of morphological samples from allpopulations of Rhinichthys cataractae: allopatric longnose dace, allopatricNooksack dace, and sympatric dace. Sampling location was nested withinputative species (Nooksack dace, longnose dace, or “both” for streams withsympatric dace). For each PC, statistics are given for putative species (firstline) and sampling location nested with putative species (second line).PC r2 F P1 Species 11.8% 62.92 < 0.0001Location 35.9% 19.17 < 0.00012 Species 29.9% 175.23 < 0.0001Location 22.3% 11.67 < 0.00013 Species 3.5% 14.53 < 0.0001Location 28.6% 11.79 < 0.00014 Species 3.8% 14.645 < 0.0001Location 22.7% 8.637 < 0.00015 Species 1.0% 3.59 0.0282Location 17.5% 5.982 < 0.0001By contrast, my analysis of sympatric dace in Kanaka Creek dace showedno difference between fish carrying NSD and LND mtDNA; ANOVAs werenon-significant for all but PC 3 (F = 4.0672, P = 0.024) (Table 2.7).Variation in lateral line scale countsWhen analyzed across all populations, there was a strong difference in lateralline scale counts between NSD and LND, and ANOVA results were highlysignificant (F = 186.5, P < 0.0001) (see Fig. 2.4). In a nested ANOVA withlocation nested within species, both species and location had significanteffects (P < 0.0001), and species accounted for 28.9% of the variance whilelocation accounted for 15.5%. By contrast, when the analysis was performedon samples from Kanaka creek and separated by mtDNA haplotype, there292.3. ResultsTable 2.7: F-ratios and P-values for ANOVAs conducted on the firstfive principal components of samples of longnose dace and Nooksack dace(Rhinichthys cataractae) from Kanaka Creek which have been matched withmtDNA haplotype, testing for grouping by mtDNA type.PC F P1 0.5372 0.58842 0.3564 0.70233 4.0672 0.024474 0.6719 0.43215 0.6127 0.5468was no pattern found; the distribution of lateral line scale counts was similarbetween the LND and NSD (ANOVA: F = 0.0183, P > 0.98) (Fig. 2.5).Overall, lateral line scale counts for sympatric populations were intermediatebetween LND and NSD: mean lateral line scale count for LND was 66.26, forNSD 59.99, and for sympatric dace 62.50 (ANOVA: F = 117.6, P < 0.0001).302.3.ResultsFigure 2.4: Histogram showing lateral line scale count by mtDNA haplotype for samples from 19 allopatricNooksack and longnose dace (Rhinichthys cataractae) sampling locations. Light grey bars signify longnose dacemtDNA, dark grey bars signify longnose dace mtDNA. N = 414.312.3. ResultsFigure 2.5: Histogram showing lateral line scale count by mtDNA haplotypefor Kanaka Creek samples of longnose dace and Nooksack dace (Rhinichthyscataractae). Light grey bars signify Nooksack dace mtDNA, dark grey barssignify longnose dace mtDNA. N = 43.Morphological cluster analysisThe results for mclust analysis on all dace populations indicated that thetop three models according to the BIC criterion were VEI, 5 clusters (BIC= −8732.108); VEI, 6 clusters (BIC = −8742.650); and VVI, 4 clusters(BIC = −8761.264, see Appendix A, Table A.5 for complete BIC values foreach model and number of clusters). The model names indicate whetherthe volumes of the clusters may vary, whether the shapes of the clustersmay vary, and what the orientation of the clusters is. For example, the VEImodel indicates that the volumes of the different clusters may vary (V), theshapes of the clusters are the same (E), and the orientation of the clustersis the identity (I). The VVI model is the same except that the shapes of theclusters may vary as well. Using the five cluster model, a strong associationwas detected between morphocluster and dace type (LND, NSD or “both”),as tested with a contingency table (G = 228.7, P < 0.0001); however, eachcluster contained a mixture of samples from allopatric LND sites, allopatricNSD sites, and sympatric sites (see Fig. 2.7 for a plot of cluster by putativespecies, and Fig. 2.6 for a representation of the clusters on a scatter plot offirst two principal components). The only exception was cluster 5, which322.3. Resultswas composed entirely of NSD samples from Wynoochee River.The results for mclust analysis on allopatric dace populations indicatedthat the top three models according to the BIC criterion were VEI, 6 clus-ters (BIC = −6199.875); VEI, 3 clusters (BIC = −6224.831); and VEI, 5clusters (BIC = -6228.845, see Appendix A, Table A.6 for complete BICvalues for all models and number of clusters). Using the six cluster model,a strong association was detected between morphocluster and dace type, astested with a contingency table (G = 139.2, P < 0.0001). Each cluster wasdominated by either NSD or LND, but all clusters contained a mixture ofLND and NSD, aside from cluster 6, which consisted entirely of samplesfrom the Wynoochee River in western Washington (Fig. 2.8).The results of mclust analysis on the subset of sympatric samples in-dicated that the top three models according to the BIC criterion wereVII, 3 clusters (BIC = −2742.222), VII, 4 clusters (BIC = −2743.502),and VII, 2 clusters (−2747.559). I also analyzed each sympatric popula-tion individually: for Coquitlam River, the best model was EII, 2 clusters(BIC = −392.8082); for Alouette River, the best model was EEI, 2 clusters(−773.41116); and for Kanaka Creek, the best model was EII, 3 clusters(BIC = −1029.744). For Kanaka Creek samples which had mtDNA typematched with morphology, I performed a contingency test of the associationbetween mtDNA type and cluster. The results were non-significant: G =2.5402, P = 0.2808 (see Fig. 2.9).Predicting group membership with discriminant analysisThe mclust discriminant analysis performed on allopatric populations cor-rectly assigned 88.4% of dace to the correct group: 91.7% of LND werecorrectly assigned, and 80.6% of NSD. The model-based approach to dis-criminant analysis fits a Gaussian mixture model as a density estimate foreach category of the a priori grouping (Fraley et al., 2012), using BIC toselect the best model for each category. In my analysis, the LND group wasbest modeled by a VVI model with three components (i.e., morphologicalsubgroups), and the NSD group was best modeled by a EEI model with fourcomponents.Environmental correlates of morphological variationThe analyses performed on principal components vs. environmental vari-ables did not detect any strong associations. Though some linear modelswere significant (Table 2.8), r2 values were all < 0.10, with the exception332.3. ResultsFigure 2.6: Convex hulls enclose each mclust-assigned cluster of individ-ual dace longnose dace and Nooksack dace (Rhinichthys cataractae) basedon variation in 11 morphometric and two meristic traits, measured on allsamples from all populations (N = 582). Putative species is indicated bysymbol: open triangles indicate allopatric longnose dace, plus signs indicateallopatric Nooksack dace, and open circles indicate sympatric samples.342.3. ResultsFigure 2.7: Stacked bar chart showing the composition of each cluster gener-ated by the mclust analysis conducted on samples of Rhinichthys cataractaefrom all sampling locations (allopatric LND, allopatric NSD, and sympatric,N = 582). Each column represents one cluster (cluster number indicated atthe bottom). The colours indicate the proportions of each species that makeup the clusters: Nooksack dace (light grey), longnose dace (black), and sym-patric dace (dark grey). All clusters aside from cluster 5, which is composedentirely of Nooksack dace from Wynoochee River, WA, contain a mixtureof longnose, Nooksack, and sympatric dace.352.3. ResultsFigure 2.8: Stacked bar chart showing the composition of each clustergenerated by the mclust analysis conducted on longnose dace and Nooksackdace (Rhinichthys cataractae) from only those locations that were allopatricfor longnose or Nooksack dace (N = 414). Each column represents onecluster (cluster number indicated at the bottom). The colours indicate theproportions of each species that make up the clusters: Nooksack dace (black)and longnose dace (grey).362.3. ResultsFigure 2.9: Stacked bar chart showing the composition of each cluster gen-erated by the mclust analysis conducted on longnose dace and Nooksackdace (Rhinichthys cataractae) from Kanaka Creek (N = 44). Each col-umn represents one cluster (cluster number indicated at the bottom). Thecolours indicate the proportions of each mtDNA haplotype that makes upthe clusters: fish with Nooksack dace mtDNA (“A”, grey) and longnose dacemtDNA (“B”, black).372.3. Resultsof PC 2’s association with annual precipitation (r2 = 0.14), PC 2’s associa-tion with highest precipitation quarter (r2 = 0.14), PC 2’s association withdate collected (r2 = 0.16), and PC 2’s association with watershed area (r2= 0.25). However, for all four of these, the slope given by the model was< 0.003 (Table 2.8.) (P significant at < 0.011.)Comparisons of morphology east and west of the Rocky Mountains forall dace were significant for PC 2–PC 5, as well as lateral line scale count(Table 2.9). Fish from west of the Rocky Mountains had an average lateralline scale count of 63.1, whereas fish from the east had on average 66.8 (adifference of 3.7 scales; LND have on average 6.3 more scales than NSD).They showed the greatest differences in (transformed) values for M3 andM4, with dace east of the Rockies being larger for both.Comparisons of morphology for allopatric LND east and west of theRocky Mountains were significant for PC 2–PC4, with the comparison oflateral line scale counts being almost significant (P = 0.09) (Table 2.10).LND from west of the Rocky Mountains had on average 65.9 lateral linescales, whereas fish from the east had on average 66.8 (a difference of 0.9scales).382.3.ResultsTable 2.8: Results for linear regressions conducted on between morphological variables and environmental variablesacross 21 populations of Nooksack dace and longnose dace (Rhinichthys cataractae). Morphological variation wassummarized using principal component analysis across 11 morphological variables and two meristic traits. Statisticsprovided are slope of the linear model, adjusted r-squared value, and P-value. Significant (at the corrected valueof P < 0.0111, using the method of Benjamini and Yekutieli (2001)) tests are indicated by (*).Variable tested PC 1 PC 2 PC 3 PC 4 PC 5Watershed area−0.0002651 0.0032378 0.0005409 −0.0002586 0.0001862−0.0008472 0.2457 0.009893 0.0005729 −0.00041140.437 < 0.0001 * 0.01790 0.261 0.369−0.01019 −0.11678 −0.08763 0.05566 −0.02907−0.001411 0.05935 0.04501 0.02016 0.005392Annual meantemperature0.654 < 0.0001 * < 0.0001 * 0.000404 * 0.04403 .Maximumtemperature of hottestmonth−0.13241 0.28003 0.10925 −0.08163 0.1380.01454 0.0934 0.01792 0.01101 0.041930.002319 * < 0.0001 * 0.0008 * 0.007 * < 0.0001 *0.001248 −0.047561 −0.031099 0.02266 −0.01485−0.00173 0.06785 0.03869 0.02319 0.01106Minimum temperatureof coldest month0.885 < 0.0001 * < 0.0001 * 0.000159 * 0.00695 *Temperature range−0.005656 0.051235 0.030955 −0.022625 0.017737−0.0009039 0.09064 0.04409 0.02669 0.019170.485 < 0.0001 * < 0.0001 * < 0.0001 * 0.000548 *0.0002942 −8.42× 10−4 −2.94× 10−4 2.12× 10−4 −2.11× 10−4392.3.ResultsTable 2.8 – continued from previous pageVariable tested PC 1 PC 2 PC 3 PC 4 PC 50.01123 0.1373 0.02126 0.01213 0.01474Annual precipitation0.00657 ** < 0.0001 * 0.000288 * 0.00494 * 0.00218 *Precipitation ofwettest quarter0.0007504 −0.0018418 −0.0006103 0.0003416 −0.00037310.0164 0.1411 0.01957 0.005998 0.0093190.00129 ** <2e-16 * 0.000485 * 0.0359 * 0.0121 *−0.001981 −0.006214 −0.003094 0.0054478 −0.0039670.001978 0.04635 0.01445 0.05664 0.0353Precipitation of driestquarter0.146 < 0.0001 * 0.00238 * < 0.0001 * < 0.0001 *Date collected−4.30×10−10 −7.15×10−10 −1.50×10−10 1.49× 10−10 −8.91×10−110.0419 0.157 0.007731 0.009077 0.002883< 0.0001 * < 0.0001 * 0.0197 0.0127 0.104402.4. DiscussionTable 2.9: F-ratios and P-values for ANOVAs comparing longnose dace andNooksack dace (Rhinichthys cataractae) samples from east and west of theRocky Mountains for the first five principal components as well as lateralline scale count. All populations (allopatric and sympatric) were includedin this analysis (N = 582).F PPC 1 1.143 0.2855PC 2 4.153 0.04201PC 3 27.59 < 0.0001PC 4 36.891 < 0.0001PC 5 5.3246 0.02138Lateral line scale count 58.966 < 0.0001Table 2.10: F-ratios and P-values for ANOVAs comparing longnose dace(Rhinichthys cataractae) samples from east and west of the Rocky Mountainsfor the first five principal components as well as lateral line scale count.Only allopatric longnose dace populations were included in this analysis (N= 290.)F PPC 1 1.2895 0.2395PC 2 42.714 < 0.0001PC 3 10.239 0.001529PC 4 30.536 < 0.0001PC 5 1.4026 0.2373Lateral line scale count 2.8404 0.093012.4 DiscussionThe genus Rhinichthys has a broad North American distribution (Matthewset al., 1982), and several studies have been dedicated to the systematicsand biogeography of blacknose dace (R. atratulus, Matthews et al., 1982;Tipton et al., 2011) and speckled dace (R. osculus, Smith and Dowling,2008; Billman et al., 2010). Similar work has also been done on Rhinichthys412.4. Discussioncataractae, the longnose dace, which is the most widely distributed speciesin the genus, and indeed the most widely distributed native North Americanminnow (McPhail and Taylor, 2009); it occupies varied habitats with dif-fering geological histories and is potentially an excellent study system withsignificant undiscovered diversity (Girard and Angers, 2006). The Umpqua(R. evermanni) and Millicoma dace (R. cataractae spp.) have already beenidentified as distinct species within the R. cataractae species group (McPhailand Taylor, 2009), and the Nooksack dace, endemic to southwest BritishColumbia and northwest Washington State, has been identified as a fourthclade with mtDNA distinct from the longnose dace (McPhail and Taylor,2009).The NSD and LND mitochondrial clades co-exist in three streams in thelower Fraser Valley, presumably from postglacial secondary contact, and theconfirmation of two divergent morphological groups in sympatry would lendsupport to the idea that the Nooksack dace and longnose dace are separatespecies; however, my results show minimal differences in morphology, whichsuggests a lack of reproductive isolation between two lineages which havebeen separated for potentially millions of years.2.4.1 Subtle morphological variationMy morphometric analysis showed that there are very subtle morphologi-cal differences between the Nooksack and longnose dace. Nested ANOVAsconducted on the first five principal components of allopatric populationsshowed that both putative species and sampling location had significanteffects on all PCs; however, in each case, location accounted for a higherproportion of the total variance than putative species did.The principal components themselves were not highly informative, withPC 1 and PC 2 accounting for only 18.3% and 12.8% of the variance, re-spectively. Attempts to interpret the character loadings of the PCs did notyield any insight; PC 1 was dominated by two body-width measurements(M1, M2) and two diagonal body measurements (M5, M7), but only one ofthe measurements (M6) has a high enough loading to be considered “highassociation” (loadings > 0.71 or < −0.71), i.e., those characters which makelarge contributions to the PC (see Tabachnick, Fidell, et al., 2001). Indeed,among the first five principal components, only two measurements can beconsidered ”high association” traits. This suggests that morphological vari-ation between NSD and LND is subtle and involves the overall shape of thefish rather than being dominated by a few dimensions.The one morphological character identified by McPhail (1967) to differ422.4. Discussionconsistently between NSD and LND was lateral line scale count. My analysisdid indeed show that lateral line scale counts fell into two groups for NSDand LND; however, the scale counts overlap significantly and cannot be usedas a diagnostic character. Sympatric populations have intermediate valuesfor this important meristic, suggesting that they are introgressed.The cluster analysis was designed to identify any clustering within thedata. If the LND and NSD were strongly differentiated morphologically,the cluster analysis should have identified two groups. However, in thecluster analysis of all dace populations, five clusters were identified, eachcomposed of a broad mixture of samples from allopatric NSD and LND sitesas well as sympatric sites. The exception was cluster 5, which was composedsolely of Nooksack dace from the Wynoochee River. Cluster analysis of theallopatric populations revealed six clusters, with each being composed ofboth NSD and LND, with the exception of cluster 6, which was composedsolely of Nooksack dace from Wynoochee River. However, each cluster inthis analysis was clearly dominated by either NSD or LND, whereas clustersfrom the analysis of all populations were ambiguous in composition, andlikely do not represent meaningful morphological clusters. For both analyses,when the association between cluster and putative species was tested witha contingency table, it was found to be highly significant. These results,as well as the results of the nested ANOVAs, suggest that while putativespecies does have an effect on morphology, it is not strong, and samplinglocation may have a larger effect—for example, Wynoochee River sampleswere very distinctive in all morphological analyses, possibly due to uniqueenvironmental conditions, genetic drift, or a small and distinctive foundingpopulation.R. cataractae is a small, benthic freshwater fish with very specific habitatrequirements. It is a lithophilic, relatively sedentary fish, and like all fresh-water fish, it is restricted to its drainage. Thus, different populations mayremain isolated for long periods of time, and adaptations to local conditionsmay develop. Differences in stream conditions, prey availability, and preda-tor presence may all cause slight morphological adaptations; certain streams,such as Wynoochee River, may be more distinctive and cause greater diver-gence in their population. Alternately, as relatively small populations mayremain isolated for some time, genetic drift and/or founder effects may playa role, and streams such as Wynoochee River may have been isolated for alonger period of time.The cluster analysis is sensitive to all sources of variation and will formclusters “blindly.” Discriminant analysis, on the other hand, uses a prioriinformation about the grouping of interest to create a function that max-432.4. Discussionimizes the difference between these groups—in this case, putative species.Indeed, discriminant analysis was much more successful than cluster analy-sis, successfully identifying 91.7% of allopatric LND and 80.6% of allopatricNSD. This also suggests morphological differentiation between the two fish,which becomes more evident with an analysis that is able to focus on varia-tion due to putative species as opposed to location or other factors, and fallswithin the commonly-suggested 75% rule for subspecies. Again, however,given that the discriminant analysis had between 8.3% and 19.4% error, evenmultivariate analysis does not provide a truly diagnostic tool.Analysis of sympatric samples from Kanaka Creek with matched mor-phology and mtDNA revealed no association between mtDNA type and mor-phology. Lateral line scale counts for sympatric samples had a unimodaldistribution with complete overlap between NSD and LND mtDNA types.The analyses of variance on each PC for the Kanaka Creek samples werealso non-significant. Cluster analysis assigned three clusters, two with bothNSD and LND, and one with only NSD, but only two fish were contained inthis NSD exclusive grouping. A contingency test revealed no association be-tween mtDNA type and cluster, unlike the analysis on allopatric populations(see above). Finally, when the discriminant function trained on allopatricpopulations was applied to Kanaka Creek samples, it assigned only 38.7% ofsamples with NSD mtDNA to the NSD morphogroup, and 69% of sampleswith LND mtDNA to the LND morphogroup, with a total of 47.7% of fishassigned to the morphogroup that matched their mtDNA—approximatelythe same as would be achieved by chance. Compared to the overall 88.4%success rate of the discriminant function on allopatric samples, this stronglysuggests that there is no association between mtDNA and morphotype instreams with sympatric dace. Bailey et al. (1954) suggested that subspeciescriteria should be that at least 93% of the individuals from each allopatricpopulation differ from individuals from the other allopatric population; LNDand NSD do not meet this standard.The intermediacy of Kanaka Creek samples in both lateral line scalecount, and the lack of association of mtDNA with morphotype, supportsthe idea that they are an introgressed population, with NSD and LND in-terbreeding readily when they come into contact.2.4.2 Alternative explanations for morphological variationThough contingency tests of the cluster analysis did indicate that therewas an association between cluster and putative species, there was clearlyvariation that could not be explained by putative species. Morphological442.4. Discussionvariation is affected by both genetic and environmental factors, and therelatively small range of morphological variation observed in the Nooksackand longnose dace could be due to environmental variation across the widesampling range. Indeed, sampling location had significant effect on all PCs.However, there was no meaningful association between the principal compo-nent scores and any of the variables that I tested. Though there were severalsignificant linear models, the proportion of variance explained by environ-mental variables was low, and the slopes of the linear models were generallysmall, indicating relationships that are probably not biologically relevant.These slight correlations with environmental variables do not appear to beinformative, particularly after taking into consideration the relatively lowcontributions of each principal component to the total morphological vari-ability.Although my analysis showed no clear effect of environment upon mor-phology, it may be premature to draw the conclusion that dace morphologydoes not vary by environment. The environmental data I was able to usewas limited in their number and were only coarse-grained measures of envi-ronmental variation (i.e., measured at the watershed level whereas dace werecollected at the site level). R. cataractae is a habitat specialist, occupyingonly those portions of streams which suit their needs (fast-flowing shallowriffles), and it is likely that microhabitat variation plays a much greater rolethan large-scale environmental differences. Further study in this area shouldmake finer-scale measurements of dace habitat. For instance, relevant habi-tat parameters to measure at the site of capture include water depth, rateof flow, average size of substrate, as well as community composition: preda-tors and competitors, and abundance of different prey. Baltz et al. (1982)showed that speckled dace were found in numbers inversely proportionateto the number of a competitor, the riffle sculpin (Cottus gulosus), in a givenriffle, and Wood and Bain (1995) demonstrated morphological variation bymicrohabitat use in several stream fish.Comparison of Rhinichthys from east and west of the Rocky Mountainsdid yield significant results, even when restricted to allopatric LND popu-lations. R. cataractae from east of the Rocky Mountains may have beenisolated in a separate glacial refugium, and have different mating behaviour,mating during the day, while western dace mate at night (Bartnik, 1972).The morphological differentiation, different mating behaviour, and divergentmtDNA (3-8%, Taylor et al. unpubl. data) suggest that dace east of theRocky Mountains are quite divergent from those west of the Rocky Moun-tains, and merit further study particularly because these mountains havebeen shown to separate divergent lineages in a variety of taxa (e.g. wood452.4. Discussionfrogs (Rana sylvatica), Lee-Yaw et al., 2008).2.4.3 ImplicationsThe Nooksack dace was isolated in the Chehalis refugium, probably for mil-lions of years during the Pleistocene; however, only subtle morphologicaldivergence from the longnose dace occurred. There has also been little mor-phological divergence between other dace within the R. cataractae group,or across the geographical range of R. cataractae generally (Bisson andReimers, 1977). In contrast, the speckled dace (R. osculus) exhibits consid-erable morphological variation across its range (Oakey et al., 2004), yet hasstill revealed cryptic diversity within Oregon’s Great Basin (Hoekzema andSidlauskas, 2014). Like R. cataractae, R. osculus has experienced repeatedrange fragmentation and remerging during the Pleistocene glaciations. Thistype of phylogeographic history, with species of limited dispersal ability be-ing divided by a fragmented landscape, has been suggested as one mechanismfor generating genotypic differentiation without accompanying morpholog-ical differentiation. That R. osculus displays greater phenotypic variationacross its range may be reflective of its more generalist nature: though thetwo fish have similar diets, R. cataractae is adapted for specialization infast-flowing stream riffles, whereas R. osculus is found in a wide range ofhabitats, from small streams to deep lakes. Specialization is thought to con-strain phenotypic variation, and though R. cataractae is not so extreme aspecialist as fish like the African butterflyfish (Pantodon buchholzi), whichprovides one of the strongest examples of morphological stasis in extantvertebrates (Lavoue´ et al., 2011), it may still explain why its morphologyappears less geographically variable than that of other dace.With regard to the taxonomic status of Nooksack and longnose dace,there does appear to be some subtle morphological differentiation betweenthe two lineages, best displayed by the ability of the discriminant functionanalysis to separate the two groups at an average 88.7% success rate. How-ever, the magnitude of difference is small and there is considerable overlapbetween all characters. Nor was any differentiation discovered that hinted atecological separation. More importantly, sympatric dace were morphologi-cally intermediate, and there was no association between mtDNA haplotypeand morphology in streams with sympatric dace. This suggests that the twolineages are interbreeding freely where they come into contact, which arguesagainst assigning species status to the Nooksack dace.Subspecies definitions are contentious and often vague. Subspecies ofbirds have been described with discriminant function analysis success rate462.4. Discussionof less than 75% (Patten et al., 2002), but different stocks of a single fishspecies have been discriminated with up to 100% success using discriminantfunction analysis (Turan et al., 2006). Although the morphological differ-ences between the NSD and LND do put them within the range of beingseparate subspecies, I think that their status should be evaluated in light ofthe larger picture; is this morphological differentiation indicative of incipientspeciation, and/or adaptation to environmental conditions? In the case ofthe NSD and LND, I believe the answer to both questions is no, and theyshould not be classified as subspecies.47Chapter 3Genetic Analysis ofNooksack and Longnose Dace3.1 Introduction3.1.1 BackgroundVicariant events such as glaciation or mountain orogeny commonly separatepreviously continuous populations, leading to allopatric populations thatevolve independently. These different lineages may later come back intosecondary contact, as when glaciers recede and previously glaciated areasare recolonized by species from separate refugia (e.g. Tipton et al., 2011).Depending upon factors such as the length of time of separation and thedifferences in environment and selective pressures acting on the differentpopulations, as well as stochastic processes such as genetic drift, the degreeof divergence will vary and may or may not permit interbreeding upon sec-ondary contact. Even if considerable divergence has occurred, when speciesevolve in allopatry reproductive isolation may be incomplete and hybridiza-tion may occur when the species come into secondary recontact (Allendorfand Leary, 1988).The Nooksack dace (NSD) and longnose dace (LND) are representativeof two lineages within the Rhinichthys cataractae complex that were sep-arated during the Pleistocene glaciations; LND in the Columbia refugium(among others), and NSD exclusively in the much smaller Chehalis refugium(McPhail and Taylor, 2009). It is not uncommon for lineages isolated in thisway to evolve into separate species; indeed, these cycles of glaciation havebeen proposed as an “allopatric speciation pump” (April et al., 2013). How-ever, this is not always the case, and some lineages may interbreed freelyupon secondary contact (e.g. Coregonus artedi, Turgeon and Bernatchez,2001), but it is difficult to predict whether interbreeding will occur, and towhat degree, without observing the lineages in sympatry. There are otherlineages within the R. cataractae complex which have been classified as fullspecies (the Umpqua and Millicoma dace, McPhail and Taylor, 2009) but483.1. Introductionno cases of secondary contact have been documented and it is unknownwhether barriers to reproduction exist.NSD and LND mitochondrial types are sympatric in three streams in thelower Fraser valley: the Coquitlam River, the Alouette River, and KanakaCreek. This zone of secondary contact was first investigated using a diag-nostic restriction fragment length polymorphism (RFLP) in the mitochon-drial cytochrome b gene (Taylor et al., unpubl. data), finding that LNDand NSD mtDNA both were present in all three streams. Though bothmtDNA types are present, an analysis of nuclear DNA is necessary to eval-uate whether the two types have been interbreeding because mtDNA ishaploid and maternally-inherited such that distinct maternal lineages canpersist within randomly breeding population (see below).This zone of secondary contact provides a valuable opportunity to seewhether reproductive isolation is present between these putative species.Though many definitions of species have been proposed (Hausdorf, 2011),the biological species concept (Mayr, 1942, 1963) is widely accepted as thebest option for sexually reproducing organisms. The BSC defines species as“groups of actually or potentially interbreeding natural populations, whichare reproductively isolated from other such groups” (Mayr, 1942, 1963).This definition is difficult to test, however, unless the two putative speciesare sympatric in some part of their range: completely allopatric speciescan only be tested for reproductive isolation in an experimental setting,which has the potential to influence behaviour. Thus, zones of sympatrybetween otherwise allopatric species provide important opportunities to testfor reproductive isolation in a natural setting and apply the BSC. In the caseof lineages which have diverged in allopatry (as opposed to species whichhave diverged in sympatry and only subsequently separated their ranges),zones of secondary contact provide opportunities to study the mechanismsof allopatric speciation.For example, Redenbach and Taylor (2002, 2003) investigated a zoneof secondary contact between two species of Arctic Char, Dolly Varden(Salvelinus malma) and Bull Trout (Salvelinus confluentus), using mito-chondrial DNA as well as nuclear DNA markers. They uncovered bimodalhybrid zones with isolation likely being maintained by size assortative mat-ing; however, hybrid genotypes constituted 9% of the population of juvenilechar, providing the opportunity for further study of the mechanisms lead-ing to speciation (Redenbach and Taylor, 2003). Because these species eachsurvived in multiple glacial refugia, there was a unique opportunity to studyzones of sympatry of different ages: Dolly Varden and bull trout both sur-vived in the Chehalis refugium, and have thus overlapped for potentially493.1. Introductionmore than 100,000 years. The two species also have zones of secondarycontact between allopatric refugial populations that are much more recent,perhaps 14,000 years old (Redenbach and Taylor, 2002). Differing levels ofhybridization in these two zones might suggest a role for reinforcement indriving the development of reproductive isolation (Redenbach and Taylor,2002). Generally, studying a range of zones of secondary contact across ge-ographical region and taxa, can help understand the phylogeographical andevolutionary processes underlying patterns of diversity.3.1.2 Genetic techniquesMitochondrial DNA is often the first tool used by evolutionary biologiststo study variation between different populations; it is non-recombinant, hasa relatively high rate of mutation, an effective population size one-quarterthan of biparentally-inherited loci and is easy to isolate and characterize(Birky et al., 1983; Moritz et al., 1987). In fact, mtDNA has been proposedas the sole basis for species description (Avise and Walker, 1999; Hebert etal., 2003) and has been used for > 80% of phylogeographic studies (Avise,2000).Despite its usefulness, mtDNA surveys have their limitations; for someof the very reasons that make it an appropriate choice for broad popula-tion surveys (haploid, maternal inheritance) mitochondrial DNA alone isnot sufficient to detect hybridization (Scribner et al., 2000). However, a cy-tonuclear approach using both mtDNA and nuclear DNA can be a powerfultool for investigating the direction, extent, and evolutionary history of hy-bridization (Scribner et al., 2000; Avise, 2001; Toews and Brelsford, 2012).Lu et al. (2001) used mitochondrial DNA and nuclear microsatellites (shorttandem repeats) to analyze a hybrid zone between lineages of lake whitefish(Coregonus clupeaformis). While mtDNA data indicated little introgressivehybridization between glacial lineages among populations in the St. JohnRiver basin zone of secondary contact, microsatellite data revealed extensivehybridization between lineages, and identified some populations as admixedwhich had been presumed unmixed from mtDNA analysis (Lu et al., 2001).In addition to being less sensitive to admixture than microsatellites,analysis of mtDNA cannot distinguish a past hybridization event betweencurrently reproductively isolated species from ongoing hybridization betweenincompletely reproductively isolated species. For these reasons, further in-vestigation using nuclear DNA is necessary to clarify the status of the NSDand LND.Microsatellites, or short tandem repeats, are rapidly evolving nuclear503.1. Introductionmarkers that are commonly used in population genetic studies (Balloux andLugon-Moulin, 2002). They consist of a short (typically < 5 bp) repeatedsequence of DNA; the number of repeats is highly variable, with a givenlocus typically having from 1–50 alleles (Jarne and Lagoda, 1996). A set ofseveral microsatellites can be used to genotype individuals and distinguishpopulations, and can be analyzed in a variety of ways (Jarne and Lagoda,1996; Hedrick, 1999; Balloux and Lugon-Moulin, 2002). Assays of multi-locus allele frequencies can provide a powerful indirect test of reproductiveisolation between sympatric populations and their status as distinct biolog-ical species. For instance, McPhail (1984) use variation at nuclear-encodedallozyme loci to demonstrate reproductive isolation between species pairs ofthreespine stickleback, Gasterosteus aculeatus. More recently, many studiesof sympatric populations of char (Salvelinus), whitefishes, (Coregonus), andkillifish (Fundulus) have used microsatellite assays to test phylogeographictheories, population divergence, and degrees of hybridization and introgres-sion (Turgeon and Bernatchez, 2001; Redenbach and Taylor, 2002; Adamset al., 2006).3.1.3 Objectives for this chapterIn this chapter, I first expanded upon the mtDNA work of Taylor et al.(unpubl. data), who surveyed the mtDNA types of 120 NSD and LND,including 10 each from two of the streams with sympatric dace (CoquitlamRiver and Alouette River). I wanted to add samples from Kanaka Creek,and make a more extensive survey of mtDNA types in the Coquitlam Riverand Alouette River, both to have a larger set of samples matched withmorphology and to see whether any pattern emerged in the ratio of LNDto NSD haplotypes in each stream. I also wanted to confirm that allopatricpopulations have only one mtDNA type present, and expand the geographicrange and number of allopatric populations surveyed.Next, I analyzed populations of Nooksack and longnose dace from threestreams with sympatric dace to determine the degree of genetic differentia-tion between fish characterized morphologically and with respect to mtDNA.For this objective, I assayed allelic variation at 10 microsatellite loci (threefrom Girard and Angers (2006); seven from Beasley et al. (2014)) to testthe hypothesis that the two putative species were genetically distinct andto assess what level of hybridization, if any, occurs between them.513.2. Materials and methods3.2 Materials and methods3.2.1 SamplingFish samples were collected as detailed in Chapter 2. Additional samplesrepresented collections made by others (J.D. McPhail and E.B. Taylor, Uni-versity of BC, Dept of Zoology) and stored in the Beaty Biodiversity Mu-seum tissue collection (see Appendix B, Table B.1 for details of all geneticsamples). Samples for DNA analysis consisted of a small fin clip (typicallya portion of the caudal fin or one of the pelvic fins) and were stored in 95%non-denatured ethanol.3.2.2 DNA extraction and amplificationGenomic DNA for mtDNA and nuclear DNA analysis was extracted from finsamples. Extractions were performed using Qiagen DNeasy Blood & TissueKit and Qiagen QIAamp DNA Investigator Kit spin columns, and DNA wasstored at −20 ◦C.For mtDNA analysis, a 600 base pair portion of the cytochrome b genewas amplified using the polymerase chain reaction (PCR) under the follow-ing conditions: amplification was carried out in a total volume of 50µL,consisting of 5µL buffer solution, 4µL of dNTP mix, 1µL of each primer,0.3µL of Taq DNA polymerase, 1µL of DNA, and 37.7µL of sterile distilledwater. The primers used for amplification were:GluDG (5’-TGACTTGAAGAACCACCGTTG-3’; Palumbi et al.,1996)HD (5’-GGGTTGTTTGATCCTGTTTCGT-3’; Dowling et al.,2002)Amplifications were performed with an initial denaturing at 95 ◦C for 3minutes, two cycles of denaturation at 95 ◦C for 30 seconds, annealing at55 ◦C for 30 seconds, and extension at 72 ◦C for 30 seconds, followed by32 cycles of denaturation at 92 ◦C for 30 seconds, annealing at 54 ◦C for30 seconds, and extension at 72 ◦C for thirty seconds, followed by a finalextension at 72 ◦C for 10 minutes.After PCR, the 600 base pair product was subject to restriction fragmentlength polymorphism (RFLP) analysis using the restriction enzyme AvaIIwhich produced RFLP profiles diagnostic of the Nooksack dace and thelongnose dace and which were identified from fixed differences in the DNAsequences (Taylor et al., unpubl. data). Under these conditions, RFLP523.2. Materials and methodsanalysis of the ∼ 600 bp fragment produced two fragments of ∼ 500 and∼ 100 bp Nooksack dace and a single (uncut) fragment of ∼ 600 bp inlongnose dace. Some DNAs were of poor quality and resulted in inadequateamplification of the larger 600 bp fragment. Consequently, a smaller 250 bpfragment was amplified using the following primers:forward [5’—TGCCCCGTTAGCATGTATATT—3’]reverse [5’—ACGAAAAACCCACCCACTAA—3’]An annealing temperature of 54 ◦C was used, and all other conditions werethe same as for the 600 bp fragment. The RFLP analysis of the ∼ 250 bpproduct produced fragments of 157 and 93 bp in Nooksack dace, and an(uncut) 250 bp fragment in longnose dace.The resulting DNA fragments were separated on 2.5% agarose gels,stained CyberGreen (Molecular Probes, Inc.) and visualized under ultra-violet light.Nuclear DNA was analyzed using 10 microsatellite markers: Rhca15b,Rhca16, Rhca23 developed by Girard and Angers (2006), and Rhca4, Rhca5,Rhca7, Rhca36, Rhca42, Rhca43, and Rhca45 developed by Beasley et al.(2014) (see Appendix B, Table B.2 for complete primer information). Poly-merase chain reactions were performed in 20uL total volumes using theQiagen Multiplex PCR Kit using the PCR conditions recommended by themanufacturer, with three multiplexes—MP 1 (Rhca5, Rhca36, Rhca42), MP2 (Rhca7, Rhca16, Rhca43), and MP 3 (Rhca4, Rhca15b, Rhca45). Rhca23was run in a separate PCR, with annealing temperature of 55 ◦C and condi-tions otherwise as above for 250 bp cytochrome b fragment. For the multi-plex PCRs, the following modifications were made to manufacturer’s instruc-tions: in MP1, initial activation was followed by two “touchdown” cycles,in which annealing temperature was 58 ◦C. This was followed by 35 cycleswith an annealing temperature of 55 ◦C. MP2 was the same as MP1. MP3had two “touchdown cycles at 60 ◦C followed by 35 cycles at 58 ◦C. Allmultiplexes used a final extension step of 30 minutes at 60 ◦C. The forwardprimer of each locus pair was fluorescently labeled to facilitate detectionand allele identification using the Beckman-Coulter CEQ 8000 automatedgenotyper.3.2.3 Genetic analyses—mitochondrial DNAI compiled mtDNA results to see if there were any populations that hadbeen independently classified as either Nooksack dace or longnose dace (us-ing morphological data, previously-collected mtDNA data, or geography)533.2. Materials and methodsthat contained longnose dace or Nooksack dace mtDNA, respectively (i.e.,if previously characterized allopatric populations contained only one kind ofmtDNA). I also used the mtDNA data to better characterize each sympatricpopulation in terms of the relative proportions of Nooksack and longnosedace mtDNA.3.2.4 Genetic analyses—microsatellite dataMicrosatellites were amplified and scored for 374 samples from 12 popula-tions. I used MICRO-CHECKER (Oosterhout et al., 2004) to check for thepresence of null alleles and PCR or scoring artifacts that could affect fur-ther analyses. I used FSTAT v2.9.3.2 (Goudet, 1995) to compile descriptivepopulation genetic statistics: number of alleles, allelic richness, and ob-served and expected heterozygosity. Using Genepop v4.2.2 (Rousset, 2012),I tested for deviations from Hardy-Weinberg Equilibrium for each combina-tion of locus and population using an exact test in which probability valueswere estimated using a Markov chain method. I tested for linkage disequi-librium for all combinations of locus pairs within each population with aMarkov chain method using Genepop default values.I used the factorial correspondence analysis in GENETIX v4.05 (Belkhiret al., 2004) as a visual assessment of the clustering of different populations.Factorial correspondence analysis is a type of ordination analysis that usesvariation in multilocus genotypes to ordinate individuals in “multiallelicspace.”I calculated pairwise FST , as estimated by Weir and Cockerham’s (1968)θ, for each pair of populations to get a general sense of the level of geneticdivergence amongst populations of R. cataractae using GENETIX v4.05(Belkhir et al., 2004). I estimated average pairwise FST for different group-ings to see whether sympatric populations were more similar to either al-lopatric NSD or LND populations, which could indicate bias in introgres-sion, or even that the streams with sympatric dace consist entirely of eitherLND or NSD in terms of the nuclear genome, which would suggest that themixed mtDNA represents an event of historical introgression. To do this, Iestimated average pairwise FST between one sympatric population and allallopatric LND populations, and average pairwise FST between that samesympatric population and all allopatric NSD populations. I performed thisestimation for each sympatric population independently, in case the patternof introgression was different in different streams.I compared average pairwise FST among allopatric NSD with averagepairwise FST among allopatric LND, to see if the overall level of differenti-543.2. Materials and methodsation was different between NSD and LND and tested for a difference usingthe permutation utility in FSTAT (Goudet, 1995).The second step was to test for association of mtDNA type with nuclearDNA in sympatric samples, which would indicate some level of reproductiveisolation. For this, I separated the fish from each sympatric populationinto two groups, those with LND mtDNA and those with NSD mtDNA. Itested whether mtDNA subgroups in each stream differed from each otherby looking at the pairwise FST between LND and NSD subsets for eachsympatric population.3.2.5 Admixture analyses (Structure)I used the model-based Bayesian clustering program Structure v2.3.4(Pritchard et al., 2000) to estimate the number of genetic populations (K).Structure uses a Bayesian algorithm to identify clusters of individualsbased on their genotypes at multiple loci, by finding the arrangement whichminimizes Hardy-Weinberg and linkage disequilibria within clusters (Pritchardet al., 2000). I used the admixture model, i.e., where individual fish canrepresent a composite or admixture of different genetic groups either fromancestral polymorphism or hybridization or both. I ran Structure fivetimes for each model of K = 1-12, for 250,000 MCMC replications after aburn-in period of 50,000. I then used Structure Harvester v0.6.93 (Earlet al., 2012) to determine the most likely K following the method of Evannoet al. (2005). Structure calculates the log probability of the observed datafor each value of K and run of the program that is conducted (Pr(X|K));however, this probability is meant only to be used as an ad hoc guide tothe most accurate number of clusters (Pritchard et al., 2000). (Evanno etal., 2005) generated genetic data, analyzed it with Structure, and testedthe efficacy of different methods in extracting the most accurate number ofclusters from the Structure output. Their study found that using ∆K,an ad hoc measurement consisting of the second order rate of change ofln[Pr(X|k)] with respect to K, was a better predictor than Pr(X|k) for allof the scenarios they tested (Evanno et al., 2005).I first analyzed the full dataset of 12 populations (see Table 3.1, andthen I examined each sympatric population independently to further testwhether there was any population structure within each stream.553.2. Materials and methodsTable 3.1: Populations of longnose dace (LND) and Nooksack dace (NSD),Rhinichthys cataractae, used in Structure analysis, showing mtDNA hap-lotypes present in each population and N for each. Populations 5–7 aresympatric, the rest are allopatric. Total N = 374.Sample No. MtDNAhaplotypespresentPopulation N1 NSD Porter Creek, WA 52 NSD Satsop River, WA 253 NSD Bertrand Creek, BC 224 NSD Brunette River, BC 115 NSD, LND Coquitlam River, BC 666 NSD, LND Alouette River, BC 997 NSD, LND Kanaka Creek, BC 928 LND Norrish Creek, BC 119 LND Fraser River, BC 610 LND Coquihalla River, BC 1311 LND Beaver Creek, BC 1612 LND Beaver River, ON 83.2.6 Analysis of molecular variance (AMOVA)I used Arlequin v.3.5.1.3 (Excoffier and Lischer, 2010) to perform Analysisof Molecular Variance (AMOVA). AMOVA takes into account pre-assignedpopulation structure and partitions variance in allele frequencies into thevariance due to (1) pre-assigned group (e.g. putative species group), (2)population within that group, and (3) individual. It can be used to assessthe importance of grouping vs. population/sampling location, as well asto compare different possible population groupings and see which group-ing explains the largest proportion of the variance (e.g. Vonlanthen et al.,2007). I used AMOVA to test the hypothesis that mtDNA types withinsympatric samples were more similar to the same mtDNA type in allopatry;i.e. that dace from streams with sympatric dace with NSD mtDNA weremore similar to allopatric NSD than to allopatric LND. To do this, I per-formed AMOVAs under the following grouping hypotheses and comparedthe proportion of variance explained by each grouping (note: “sympatricLND” refers to sympatric dace that have LND mtDNA; similar for NSD):563.2. Materials and methodsGrouping 1 Hypothesis (Putative species):1. Sympatric LND + allopatric LND2. Sympatric NSD + allopatric NSDUnder this hypothesis, if pooling samples under the groups indicatedexplains most of the variance, this would be consistent with longnose andNooksack dace being distinct genetic groups (and perhaps distinct species)in both allopatry and sympatry.Grouping 2 Hypothesis (Freely interbreeding):1. Sympatric LND + allopatric NSD2. Sympatric NSD + allopatric LNDIf the groups indicated in this hypothesis explains an equal or greater amountof the variance as Grouping 1, it would indicate that dace with LND orNSD mtDNA in sympatry are not similar to those with the same mtDNAin allopatry.I also used AMOVA to test whether sympatric samples bore a greateroverall similarity to either allopatric NSD or allopatric LND. To do this, Icompared the following two grouping hypotheses:Grouping Hypothesis 3 (Sympatric samples similar to LND):1. sympatric samples + allopatric LND2. Allopatric NSDUnder this hypothesis, if pooling the samples as per the groups above ex-plained more variance than Grouping 4, it would indicate that the dace instreams sympatric for NSD and LND mtDNA are more similar to allopatricLND, which could indicate historical introgression of NSD mtDNA into aprimarily LND population.Grouping Hypothesis 4 (Sympatric samples similar to NSD):1. sympatric samples + allopatric NSD2. Allopatric LNDIf the groups created under this hypothesis explain a greater proportionof the variance than Grouping 3, it would indicate that dace in streamssympatric for NSD and LND mtDNA are more similar to allopatric NSD.573.3. ResultsLastly, I tested a subset of only allopatric populations, to see how muchof the variance was explained by species differentiation when the potentiallycomplicating sympatric samples were not present.Grouping Hypothesis 5 (Allopatric populations more highly dif-ferentiated):1. Allopatric LND2. Allopatric NSDUnder this hypothesis, if the groups indicated above explain the largestproportion of the variance, it would indicate that without the confoundingeffect of the streams with sympatric dace, species can explain more molecularvariance than sampling location.3.3 Results3.3.1 Distribution of mitochondrial haplotypesI combined my data with previous mtDNA-based assays of Rhinichthys fromWashington State and BC (Taylor et al. unpubl. data). The results showLND and NSD mtDNA are both present in the Coquitlam River, AlouetteRiver, and Kanaka Creek, and that all other streams contain mtDNA fromonly one of the two putative species (Fig. 3.1). Taylor et al. (unpubl. data)have shown that all samples of R. cataractae east to Que´bec (includingsamples from Alberta, Manitoba, Ontario, and Que´bec) and north to mysampling area bear only longnose dace mtDNA (N = 56 from 30 localities,analyzed for cytochrome b and ND2 regions).3.3.2 Microsatellite analysesHardy-Weinberg equilibriumIn total, I performed 120 tests of Hardy-Weinberg equilibrium (10 loci *12 populations), and nine were significant after a Benjamini and Yekutieli(2001) correction for multiple comparisons (adjusted critical P-value was0.017). One population had more than one locus out of equilibrium: Alou-ette River (3). Six more populations had one locus out of equilibrium: Sat-sop River, Coquitlam River, Norrish Creek, Coquihalla River, and BeaverCreek (BC). Two loci were out of equilibrium in more than one popula-tion: Rhca4 (4) and Rhca7 (3). Two other loci were out of equilibrium in583.3. ResultsFigure 3.1: Map of Nooksack dace (black) and longnose dace (white)(Rhinichthys cataractae) cytochrome b mtDNA haplotypes. 1 = WillipaRiver (N = 5), 2 = Satsop River (N = 25), 3 = Porter Creek (N = 5), 4= Stilliguamish River (N = 5), 5 = Nooksack River (N = 5), 6 = BrunetteRiver (N = 25), 7 = Coquitlam River (N = 41), 8 = Alouette River (N =86), 9 = Bertrand Creek (N = 32), 10 = Kanaka Creek (N = 44), 11 =Norrish Creek (N = 40), 12 = Coquihalla River (N = 19).593.3. Resultsone population: Rhca42 and Rhca5 (see Appendix B, Table B.3 for testsof Hardy-Weinberg equilibrium, and observed and expected heterozygosity,for all locus-population combinations. See also Appendix B, Table B.4 forcomplete allele frequencies).Linkage disequilibriumIn total, there were 540 tests of linkage disequilibrium (12 populations with10 loci each; 45 different combinations of loci). I used the method of Ben-jamini and Yekutieli (2001) to adjust for multiple comparisons, and assignedsignificance at the adjusted P-value of 0.011. Forty tests of linkage disequi-librium were significant. Four populations had more than one locus combi-nation out of equilibrium: Satsop River (2), Bertrand Creek (2), AlouetteRiver (18), and Kanaka Creek (18). Sympatric samples (with the exceptionof samples from the Coquitlam River) accounted for the vast majority oflinkage disequilibrium.Measures of genetic diversityThe number of alleles resolved in a given locus ranged from 18 (Rhca5) to 60(Rhca7). The number of alleles resolved in a given population ranged from36 (Brunette River) to 193 (Alouette River); however, the sample sizes perpopulation varied greatly in size (N from 5 to 118) so allelic richness wasused as a more appropriate measure of diversity. The highest allelic richness(averaged over loci) was 4.45 in the Fraser River, and the lowest was 2.41in Brunette River. Allelic richness was 4.05 among allopatric LND popula-tions, 3.60 among allopatric NSD populations, and 4.05 among sympatricpopulations (Table 3.2).The highest average heterozygosity (averaged over populations) was foundat locus Rhca43 (0.92) with the lowest at Rhca5 (0.55). Averaged across loci,the observed and expected heterozygosity were highest in the CoquihallaRiver (0.86, 0.88) and lowest in Brunette River (0.47, 0.45, see Table 3.2and Appendix B, Table B.3).Factorial correspondence analysisThe factorial correspondence analysis provided an exploratory first look atthe variation in microsatellite data. There were three major outlying groups:Beaver Creek (BC), Beaver River (ON), and Satsop River (WA) (Fig. 3.2).All other samples were roughly clustered in one group.603.3.ResultsTable 3.2: Summary of genetic data for the 12 locations used in microsatellite analysis. Abbreviations: N ,number of individuals sampled; FIS : inbreeding coefficient; 95% CI: 95% confidence interval of FIS ; HO: observedheterozygosity; HE : expected heterozygosity; NA: total number of alleles across 10 loci.PopNo.Stream N FIS Latitude Longitude HO HE NA Allelicrichness1 Porter Creek, WA 5 0.143 46.9465 123.2954 0.7825 0.6809 52 3.93612 West Fork Satsop River, WA 25 -0.03 47.0598 123.5410 0.8068 0.8299 135 4.27093 Bertrand Creek, BC 22 -0.004 49.0043 122.5322 0.7744 0.7773 90 3.78244 Brunette River, BC 11 0.034 49.2414 122.8959 0.4697 0.4545 36 2.40935 Coquitlam River, BC 66 0.013 49.0416 122.7711 0.8282 0.8180 177 4.21296 Alouette River, BC 118 0.023 49.2392 122.5793 0.8192 0.8012 193 4.15817 Kanaka Creek, BC 92 0.013 49.2022 122.5413 0.7472 0.7373 159 3.76268 Norrish Creek, BC 11 0.05 49.2349 122.1336 0.7171 0.6792 67 3.54699 Fraser River, BC 6 0.041 49.3831 122.4522 0.8286 0.7966 71 4.449810 Coquihalla River, BC 13 -0.016 49.3885 121.4334 0.8610 0.8819 99 4.385811 Beaver Creek, BC 16 -0.003 49.1009 117.5572 0.7726 0.7750 90 3.883212 Beaver River, ON 8 0.182 44.3089 79.0405 0.8057 0.7143 65 3.9765613.3.ResultsFigure 3.2: Factorial correspondence analysis plot based on 10 microsatellite loci for longnose dace (LND) andNooksack dace (NSD) Rhinichthys cataractae from 12 locations. Circle 1 indicates samples from Beaver Creek(LND, BC) and circle 2 indicates samples from Beaver River (LND, ON). Red squares indicate samples fromSatsop River (NSD, WA). All others represent Porter Creek (NSD, WA), Bertrand Creek (NSD, BC), BrunetteRiver (NSD, BC), Coquitlam River (NSD and LND, BC), Alouette River (NSD and LND, BC), Kanaka Creek(NSD and LND, BC), Norrish Creek LND, (BC), Fraser River (LND, BC), and Coquihalla River (LND, BC).623.3. ResultsGenetic divergence among samplesComparisons of pairwise FST demonstrated that sympatric dace were equallydifferentiated from allopatric NSD and LND, and that allopatric popula-tions of LND and NSD were more similar to sympatric populations thanthey were to each other. Across all pairwise comparisons, FST averaged0.130 (all P < 0.05), and ranged from 0.023 to 0.346. The average pairwiseFST of allopatric LND populations compared to allopatric NSD populationswas 0.169. The average pairwise FST of Coquitlam River (sympatric site)compared to allopatric NSD populations was 0.0972, and compared to LNDpopulations was 0.0958. For Alouette River (sympatric site), average pair-wise FST was 0.1018 compared to allopatric NSD, and 0.0889 compared toLND. For Kanaka Creek (sympatric site), average pairwise FST was 0.1064compared to allopatric NSD, and 0.1276 compared to LND. Average pairwiseFST among sympatric populations was 0.041 (see Appendix B, Table B.5for all pairwise FST values.) I also compared the level of divergence amongallopatric NSD populations and allopatric LND populations using the per-mutation utility in FSTAT. With this, the average FST among allopatricNSD populations was found to be 0.159, and among allopatric LND popu-lations, 0.129, with P = 0.669. This result is not significant, and may beconfounded by differences in sampling locations; however, as the LND popu-lations cover a much larger area (ranging from British Columbia to Ontario)than the NSD populations, it seems worth noting that despite this, FST ishigher among NSD populations.There was no suggestion of population structure within sympatric popu-lations; mtDNA clades within sympatric populations were not differentiatedfrom each other with respect to microsatellite allele frequencies. For calcu-lations involving the mtDNA-divided subsets within each sympatric popu-lation, FST between dace with LND mtDNA and those with NSD mtDNAin each sympatric population was −0.001 for Coquitlam River (P = 0.509),0.003 for Alouette River (P = 0.152), and 0.001 for Kanaka Creek (P =0.421) (see Appendix B, Table B.6 for all pairwise FST values).3.3.3 Admixture analysisStructure analysis on samples from all 12 populations revealed structur-ing by location, and no clear pattern of admixture in the sympatric pop-ulations. Geographically distant populations tended to form independentgenetic clusters. Analysis of streams with sympatric dace individually indi-cated strongly that Coquitlam River and Kanaka Creek were each comprised633.3. Resultsof a single population, while analysis of Alouette River revealed a random(according to mtDNA type) admixture of three groups in each sample.The analysis by Structure (Pritchard et al., 2000) across all samplesfound K = 11 to have the highest likelihood score and K = 3 to be the mostlikely, according to the Evanno method (Evanno et al., 2005, Table 3.3).Table 3.3: Evanno table output from Structure Harvester (Earl et al.,2012) for Structure analysis conducted on all 12 longnose and Nooksackdace (Rhinichthys cataractae) populations, testing assumed number of pop-ulations K = 1–12, with 5 repetitions, and a total N = 374. K = numberof populations assumed; Reps = number of times the simulation was runfor a given K; Mean LnP (K) = mean log likelihood of K over all reps forthat K; Stdev LnP (K) = standard deviation for LnP (K) over all reps forthat K; Ln′(K) = first order rate of change of mean LnP (K), defined asLnP (K) − LnP (K − 1); |Ln′′(K)| = second order rate of change, definedas Ln′(K + 1) − Ln′(K); ∆K = |L′′(K)| divided by Stdev LnP (K). Thehighest log likelihood is indicated in bold (K = 11) and the highest ∆K ishighlighted in grey.K Reps MeanLnP (K)StdevLnP (K)Ln′(K) |Ln′′(K)| ∆K1 5 −18619.7 0.648074 — — —2 5 −17943.4 4.368066 676.3 201.02 46.0203683 5 −17468.12 4.275746 475.28 324.28 75.8417444 5 −17317.12 173.814174 151 280.14 1.6117215 5 −16885.98 62.246663 431.14 278.76 4.4783126 5 −16733.6 70.331536 152.38 20.92 0.2974487 5 −16560.3 28.745956 173.3 23.82 0.8286388 5 −16410.82 16.087013 149.48 59.36 3.6899339 5 −16320.7 16.993675 90.12 52.8 3.10703810 5 −16283.38 29.826532 37.32 25.38 0.8509211 5 −16220.68 18.157147 62.7 169.66 9.34397912 5 −16327.64 213.393025 -106.96 — —For the K = 3 model, allopatric Nooksack dace from Washington State(localities 1 - 2 in Fig. 3.3) and all allopatric longnose dace (localities 8–12) tended to form one group (dominated by grey in Fig. 3.3. Two of643.3. Resultsthe three sympatric localities (Alouette and Coquitlam), and two allopatricNSD localities (Bertrand Creek and Brunette River) tended to be dominatedby another genetic group (largely white in Fig. 3.3), and the remainingsympatric locality (Kanaka Creek) was dominated by the third genetic group(black in Fig. 3.3).The model with the highest log likelihood, K = 11, also grouped the twoWashington populations together (blue in Fig. 3.4). It assigned samples fromBertrand Creek and the Brunette River two largely exclusive genetic clus-ters (pink and red in Fig. 3.4). Norrish Creek, Fraser River, and Coquihallawere grouped together, whereas Beaver Creek and Beaver River each formedlargely exclusive clusters of their own (mauve, tan, and yellow, respectively,in Fig. 3.4). The three sympatric populations had much more admixturebetween clusters than the allopatric populations, but roughly speaking, Co-quitlam River samples were mostly one cluster (light blue); Alouette Riverwas a mixture of two (green and orange); and Kanaka Creek was a mixtureof two (green and purple, see Fig. 3.4).653.3.ResultsFigure 3.3: Output of Structure run showing admixture analysis from 10 microsatellite DNA loci in 374Nooksack and longnose dace (Rhinichthys cataractae), K = 3. Each fish is represented by a thin vertical line andeach colour represents the proportional contribution (Q) to the genome of each fish by one of three genetic groups(white, black and grey). Populations are arranged from west to east. Population name is indicated below thegraph, and putative species is indicated above.663.3.ResultsFigure 3.4: Output of Structure run showing admixture analysis from 10 microsatellite DNA loci in 374Nooksack and longnose dace (Rhinichthys cataractae), K = 11. Each fish is represented by a thin vertical lineand each colour represents the proportional contribution (Q) to the genome of each fish by one of eleven geneticgroups. Population name is indicated below the graph, and putative species is indicated above.673.3. ResultsThe Structure analysis conducted within each sympatric populationseparately revealed no evidence for more than one genetic group. For KanakaCreek the highest mean log likelihood was at K = 1 (−3221.98), with K =2 at −3564.36 as the next most likely (Table 3.4. For the Coquitlam River,the highest mean log likelihood was also K = 1 (−2747.7), with K = 2 at−2749.9 being the next most likely (Table 3.4). Bar plots for each of thesetwo localities for any K > 1 showed equal contributions of each geneticcluster to each individual in the population; i.e. each fish showed the sameproportional contribution of each hypothetical genetic group (see Fig. 3.5).For these populations, the ∆K-based method did not necessarily agree formost likely K, but this method is not appropriate in situations where amodel of K = 1 has the highest likelihood, as ∆K can only be calculatedfor K > 1 (Evanno et al., 2005).Figure 3.5: Output of Structure run showing admixture analysis with K= 3 from 10 microsatellite DNA loci in 66 R. cataractae from the CoquitlamRiver, a site sympatric for longnose and Nooksack mtDNA haplotypes. Eachfish is represented by a thin vertical line and each colour represents theproportional contribution (Q) to the genome of each fish by one of threegenetic groups (white, black and grey).Analysis of Alouette River also had K=1 with the highest mean log like-lihood (−4048.22), with K=2 as the next most likely (−4289.52) as the mostlikely (Table 3.4). Unlike Coquitlam River and Kanaka Creek, however, atK > 1, Alouette River samples showed variation in admixture between ge-netic groups. However, in the bar plot for K = 3, considered the best modelusing the Evanno method (Evanno et al., 2005)), there was no clear popu-lation structure; differences in admixture levels seem randomly distributedaccording to mtDNA type (Fig. 3.6).683.3. ResultsFigure 3.6: Output of Structure run showing admixture analysis with K= 3 from 10 microsatellite DNA loci in 99 R. cataractae from the AlouetteRiver, a site sympatric for longnose and Nooksack mtDNA haplotypes. Eachfish is represented by a thin vertical line and each colour represents theproportional contribution (Q) to the genome of each fish by one of threegenetic groups (white, black and grey). Samples are arranged by mtDNAtype, with NSD on the left (N = 25) and LND on the right (N = 74).693.3. ResultsTable 3.4: Evanno table outputs from Structure Harvester (Earl et al.2012) for Kanaka Creek (A, N = 118), Coquitlam River (B, N = 66), andAlouette River (C, N = 99). Each analysis was conducted for assumed num-ber of populations K = 1–6, with 5 repetitions. K = number of populationsassumed; Reps = number of times the simulation was run for a given K;Mean LnP(K) = mean log likelihood of K over all reps for that K; StdevLnP(K) = standard deviation for LnP(K) over all reps for that K; Ln’(K)= first order rate of change of mean LnP(K), defined as LnP(K)-LnP(K-1);—Ln”(K)— = second order rate of change, defined as Ln’(K+1)- Ln’(K);Delta K = —L”(K)— divided by Stdev LnP(K). For each stream, the high-est log likelihood is indicated in bold (K = 11) and the highest Delta K ishighlighted in grey.A K Reps MeanLnP (K)StdevLnP (K)Ln′(K) |Ln′′(K)| ∆K1 5 −3221.98 0.898332 — — —2 5 −3564.36 148.174671 −342.38 73.56 0.4964413 5 −3833.18 112.173134 −268.82 368 3.2806434 5 −3734 119.309828 99.18 50.02 0.4192455 5 −3584.8 63.632853 149.2 241.48 3.7948956 5 −3677.08 61.369756 −92.28 — —B K Reps MeanLnP (K)StdevLnP (K)Ln′(K) |Ln′′(K)| ∆K1 5 -2747.72 0.576194 — — —2 5 −2749.92 3.541469 −2.2 3.22 0.9092273 5 −2755.34 2.050122 −5.42 24.98 12.1846414 5 −2785.74 29.898043 −30.4 30.62 1.0241475 5 −2785.52 45.731302 0.22 65.3 1.4279066 5 −2850.6 111.621414 −65.08 — —C K Reps MeanLnP (K)StdevLnP (K)Ln′(K) |Ln′′(K)| ∆K1 5 −4048.22 0.870057 −−− −−− −−−2 5 −4289.52 111.020885 −241.3 50.96 0.4590133 5 −4581.78 162.750075 −292.26 166.48 1.0229184 5 −4707.56 328.296403 −125.78 195.22 0.5946465 5 −4638.12 82.209866 69.44 60.28 0.7332456 5 −4628.96 33.67704 9.16 −−− −−−703.3. Results3.3.4 AMOVA analysisIn general, all of the AMOVAs suggested that sampling location accountedfor a much greater proportion of the variance than did grouping by putativespecies (Table 3.5.) None of the AMOVA analyses suggested that NSD andLND were broadly distinct in microsatellite DNA allele frequencies afteraccounting for differences among populations within putative taxa, thoughthe comparison of groupings 3 and 4 suggested that sympatric populationsmay be more genetically similar to NSD than LND.Groupings 1 and 2 compared “matching” and “mismatched” combina-tions of mtDNA clades from sympatric populations, and allopatric popu-lations (e.g. grouping 1 consisted of sympatric LND and allopatric LNDgrouped together, and sympatric and NSD and allopatric NSD grouped to-gether, where as grouping 2 combined sympatric LND with allopatric NSD.)Both arrangements explained essentially zero percent of the variation (Ta-ble 3.5), indicating that the mtDNA clades within sympatric populationsare not more closely associated with either parental species in allopatry.Groupings 3 and 4 lumped the sympatric dace with either allopatricLND (grouping 3) or allopatric NSD (grouping 4), to test whether the sym-patric dace were more similar to either LND or NSD allopatric populations.Surprisingly, this comparison showed that when sympatric dace were com-bined with allopatric NSD, group explained ten times the variation that wasexplained by combining them with LND (Table 3.5).Grouping 5, which was conducted on allopatric populations alone, demon-strates that even in this situation, which would be expected to maximize theamount of variation explained by grouping because the allopatric NSD andLND are presumably more differentiated from each other than the sympatricdace, population still explains six times as much of the variance as putativespecies group. Interestingly, combining the sympatric dace with allopatricNSD explains a greater proportion of the variance than does Grouping 5,but these arrangements are not directly comparable because they do not usethe same number of populations.713.3.ResultsTable 3.5: AMOVA results for groupings 1–5 (see pages 59–60 for definitions). Percentage of variation and P-value are presented for all four levels: among pre-assigned groups, among populations within those groups, amongindividuals within populations, and within individuals.Grouping AmonggroupsAmongpopulationswithin groupsAmongindividuals withinpopulationsWithinindividuals1 % of variation -0.01 9.79 0.24 89.97P-value 0.21577 0 0.43973 02 % of variation -0.76 10.23 0.24 90.28P-value 0.69544 0 0.44816 03 % of variation 0.29 9.17 0.05 90.49P-value 0.17039 0 0.4933 04 % of variation 3.17 8.14 0.05 88.65P-value 0.00322 0 0.51116 05 % of variation 2.44 14.69 0 82.86P-value 0.01687 0 0.55531 0723.4. Discussion3.4 DiscussionAllopatric speciation, while widely considered to be the main form of spe-ciation (Mayr, 1963; Coyne and Orr, 2004), is difficult to study in nature.When different species are completely allopatric, it is impossible to put thebiological species concept to the test, and thus allopatric species are assessedusing a variety of different criteria: morphological divergence, ecological di-vergence, and increasingly, genetic divergence. An overall increase in theease, power, and affordability of genetic data has led to initiatives such asDNA barcoding, which uses percentage sequence divergence at a standard-ized section of the cytochrome c oxidase gene (COI) to draw preliminaryspecies boundaries based upon genetic data alone (Hebert et al., 2003; Zem-lak et al., 2009). Particularly when direct tests of reproductive isolation arenot available, assessing species status should be done using as many formsof evidence as possible (Sites and Marshall, 2004; De Queiroz, 2007), and in-terpreted with knowledge specific to the taxa, environment, and geologicalhistory. In North America, the Pleistocene glacial cycles repeatedly frag-mented the landscape, particularly for freshwater fish (April et al., 2013),splitting previously contiguous lineages into many allopatric and parapatricpopulations which are beginning to be uncovered by phylogeographic stud-ies. Studies of zones of secondary between such separated lineages give theopportunity to examine levels of reproductive isolation and can shed light onthe effects of similar isolation upon taxonomically or geographically relatedspecies.Nooksack (NSD) and longnose (LND) dace have a level of genetic diver-gence that establishes them well within the range of many “good” species (2–3% sequence divergence of mitochondrial cytochrome b and ND2 mtDNA,indicating divergence 2–3 mya—Taylor et al., unpubl. data; % divergencerange for fish species, April et al., 2011), but my results indicate that thereis no reproductive isolation between the two where they come into sec-ondary contact. There is also no evidence to support the hypothesis thatthe sympatric populations are the result of an event of historical introgres-sion of one mtDNA type into a largely pure population of either LND orNSD; LND and NSD are thoroughly admixed within streams throughoutthe zone of secondary contact, and there is no indication that CoquitlamRiver, Alouette River or Kanaka Creek are more similar to either putativeparental species. This represents the only study of secondary contact inthe Rhinichthys cataractae species group, and one of the only genetic stud-ies of secondary contact involving lineages preserved in the Chehalis glacialrefugium (see also Redenbach and Taylor, 2002). As such, it can provide733.4. Discussionvaluable perspective on the process of diversification and development ofreproductive isolation among taxa whose range was fragmented during thePleistocene.3.4.1 Similarity of sympatric R. cataractae to allopatricparental lineagesPartial or complete introgression of mtDNA from one taxon into anotherwithout accompanying introgression of nuclear DNA is seen frequently in na-ture and can be caused by selection, sex-biased dispersal, or chance (Toewsand Brelsford, 2012). Studies of the Carpathian newt (Lissotriton montan-doni, Zielin´ski et al., 2013) and lake trout (Salvelinus namaycush, Wilsonand Bernatchez, 1998) suggest that in both cases, a historical in situ hy-bridization event led to introgression of a second mtDNA type, which thenbecame fixed by selection or by chance. However, though a similar situationwas a possibility for R. cataractae, all of my results reject this hypothesisand instead point to admixture between NSD and LND in the sympatricpopulations.Comparisons of pairwise FST showed that sympatric populations areequally isolated from allopatric LND and NSD populations, and that al-lopatric LND and NSD are each more similar to sympatric dace than theyare to each other. Both of these results strongly suggest that the sympatricpopulations are not composed of solely LND or NSD, but rather a mixtureof the two. Structure analysis, although it did not show a clear pictureof admixture between LND and NSD, did not cluster the sympatric popula-tions with either putative parental species; instead, it tended to isolate eachsympatric population into a separate group.The AMOVA analysis did indicate, contrary to analyses of pairwise FST ,that sympatric populations were more similar to allopatric NSD than LND.The AMOVA grouping of sympatric dace + allopatric NSD explained amuch larger proportion of the variance than did grouping sympatric dacewith allopatric LND (3.44% vs. 0.29% of the total variance). Comparisonsof average pairwise FST showed no difference for comparisons of sympatricpopulations with allopatric LND or NSD (0.1045 for LND, 0.1018 for NSD).On the other hand, Structure analysis of all populations atK = 3 groupedthe Coquitlam River and Alouette River with two allopatric NSD streams:Bertrand Creek and Brunette River. These results may indicate that NSDmade a greater overall contribution to the sympatric populations’ gene pool;however, the difference appears to be slight. Any imbalance may be due toselection, drift, imbalanced initial colonization, bias in ongoing migration743.4. Discussionfrom pure streams into streams with sympatric dace, or sampling effects,but the limitations of this study do not allow further conclusions to bedrawn.3.4.2 Admixture of LND and NSD in sympatricpopulationsThere appears to be no reproductive isolation between NSD and LND wherethey come into secondary contact. My results indicated that there was nopopulation structure within any of the streams with sympatric dace, andno association of mtDNA type with nuclear DNA. Comparisons of pairwiseFST involving subsets of the sympatric samples divided by mtDNA typeindicated that the two types were completely intermixed within each streamand that mtDNA types of sympatric samples are not more closely associatedwith the nuclear genotype of either NSD or LND. Structure analysis ofeach sympatric population corroborated these results: there was no evidenceof population structure. Kanaka Creek and Coquitlam River each consist ofone population, with each individual showing completely uniform admixturelevels when a higher K is forced. In the Alouette River, K = 1 is supportedby log likelihood, while K = 3 is supported by the Evanno method, and thereis some differentiation among the individuals; however, there is no pattern tothe levels of admixture, and analysis of the Alouette population by mtDNAtype did not reveal any association of microsatellite DNA admixture levelswith mtDNA.That the individuals in the Alouette River are less uniform, accordingto Structure analysis, than those in Kanaka Creek and Coquitlam River(high variability in Q, whereas Q is nearly equal for all samples in KanakaCreek and Coquitlam River) may indicate that the Alouette River popu-lation is more recently introgressed; the Alouette River is located betweenthe Coquitlam River and Kanaka Creek, but its outlet is located approx-imately 4 km up the Pitt River, while the Coquitlam River and Kanakaempty directly into the Fraser. It may have taken dace longer to colonizethe Alouette and thus introgression may be more recent and the populationnot as homogenous.Overall it would appear that the LND and NSD have not acquired anylevel of reproductive isolation. Given that their degree of mtDNA diver-gence (2–3%) puts them within range of many well-isolated separate species(e.g. April et al., 2013), this raises questions regarding why they have notdeveloped any barriers to reproduction, and whether the length of isolationinferred from mtDNA % divergence (2–3 million years) is accurate.753.4. DiscussionSeveral studies in birds have uncovered similar situations of populationswith deeply divergent mtDNA lineages breeding without any reproductiveisolation (Webb et al., 2011; Hogner et al., 2012). Webb et al. (2011) inves-tigated cryptic lineages of the Common Raven (Corvus corax ). These birdsare widespread across the Northern Hemisphere and consist of two deep mi-tochondrial clades, Holarctic and Californian, which are over 4% divergentin mtDNA coding genes. In the western United States, the two clades forma large zone of secondary contact, and in Washington State Webb et al.(2011) conducted a comprehensive study of ecology, mate preference, mat-ing success, and offspring survival. They showed that there was no isolationwhatsoever between the two lineages, and no difference in offspring survival,concluding that the ravens were experiencing “despeciation,” in which twolong-separated lineages, possibly once-separate species or possibly mitochon-drial clades differentiated by genetic drift, remerged into one. Webb et al.(2011) have no definitive answer for why the ravens did not become repro-ductively isolated, but speculate that it may be related to their wide rangeof ecological tolerance, and/or the conservatism of signal traits in the genusCorvus.These explanations do not seem applicable to the NSD and LND; sig-nal traits and mating behaviours may not be conserved across the rangeof R. cataractae, with populations east of the Rocky Mountains displayingmating colouration and mating at a different time of day (Bartnik, 1970;Bartnik, 1972). While R. cataractae consume a wide range of prey, theyare habitat specialists, adapted in both internal and external morphologyfor life at the bottom of swift-moving water. While ecological generalism issuggested as a reason why long-separated species might experience neutralsequence divergence without functional divergence or reproductive isolation,ecological specialization has been suggested as a cause of morphological con-servatism/stasis (Trontelj and Fiˇser, 2008).3.4.3 Genetic differentiation throughout the range of R.cataractaeThe FCA showed that the two most geographically distant LND populations(Beaver Creek, BC, and Beaver River, ON) were by far the most geneti-cally distinct populations, with the other LND populations in BC clusteringclosely with the sympatric populations as well as the NSD populations. TheAMOVA analysis also indicated that sampling location explained a muchlarger proportion of the variance than did species grouping; this is similarto the results of the nested ANOVA on morphological data (Chapter 2),763.4. Discussionwhich also showed that location had a greater effect than putative species,even when sympatric populations were not included in the analysis.This is somewhat surprising given the deep divergence of mtDNA be-tween the two lineages, and suggests that isolation by distance has alsoplayed an important role. This adds further support to the idea that thereis undiscovered diversity in the R. cataractae group (Girard and Angers,2011), and populations east of the Rocky Mountains may harbour levelsof divergence equal to those western clades, particularly as their matingbehaviour is diurnal instead of nocturnal (Bartnik, 1972).Another possible explanation is that mitochondrial divergence has beendriven by selection along an ecological gradient. Though phylogeographicstudies have typically assumed that all markers used are neutral, simulationsby Irwin (2012) demonstrate that even weak selection can generate phylo-geographic breaks in uniparentally inherited markers, or that such breakscan be generated entirely by neutral processes, particularly if dispersal dis-tances are low and population sizes are small (Irwin, 2002). It seems unlikelyin the case of the NSD and LND that mitochondrial divergence has beencaused solely by selection, as the break in mitochondrial clades lines up witha known glacial refugium (the Chehalis refugium); additionally, my broadanalysis of environmental variables (Chapter 2) did not demonstrate anygradients or breaks corresponding with the ranges of NSD and LND, and inthe zone of contact both mitochondrial types are found in the same riffles.However, some combination of weak selection, geographic barriers, and neu-tral processes may be at work in maintaining the mtDNA break, and anyfurther investigation of ecological factors should consider whether they maybe causing selection on mitochondrial genes.Part of the basis for inferring a 2–3 million year separation between theLND and NSD is the congruence of the NSD’s range with the Chehalisglacial refugium. Another fish that survived in the Chehalis refugium is theSalish sucker, an endangered catostomid which forms a designatable unitwithin the longnose sucker (Catostomus catostomus, COSEWIC, 2013). Inmany regards the Salish and longnose suckers present an identical case tothe Nooksack and longnose dace: a small, specialized freshwater fish, iso-lated in separate refugia, widespread throughout North America. The Salishand longnose suckers are more morphologically divergent than the NSD andLND, but they are not known to be ecologically differentiated, and differmuch less at cytochrome b and ND2 (between 0.8 and 1.2%, McPhail andTaylor, 1999). The two suckers are not known to be sympatric, but are bothfound in the lower Fraser River and could theoretically intermix; however,as there have been no hybrids detected, it is presumed to be reproductively773.4. Discussionisolated from the longnose sucker (McPhail and Taylor, 1999). The dispar-ity in mtDNA % divergence between the Catostomus pair and Rhinichthyspair points to the complexity of unravelling times of divergence betweensister lineages. These two fish were almost certainly isolated in the samerefugia during the Pleistocene, yet one pair’s mitochondrial divergence ismuch greater. To add another twist, the more deeply diverged pair appearsto be less reproductively isolated. Among other possibilities, there mayhave been intermittent connections between the Chehalis refugium and theColumbia River drainage during the Pleistocene that allowed for differingamounts of contact between the Catostomus pair and Rhinichthys pair; or,as stated above, selection or neutral processes causing the mtDNA break inRhinichthys to be deeper than predicted by conventional molecular clocks.3.4.4 Implications for the taxonomic and conservationstatus of the Nooksack daceDespite their long separation, all of the analyses conducted indicate thatthere is no population structure in the streams with sympatric dace, andthat LND and NSD are freely interbreeding in the zone of secondary con-tact. The NSD represents a deeply divergent mitochondrial clade and a sep-arate evolutionary lineage of R. cataractae, but they are not reproductivelyisolated and cannot be considered a separate species under the biologicalspecies concept.Under the Canadian Species at Risk Act (SARA), the NSD is listed asan Endangered un-named subspecies of the LND, with a recovery strategyfocusing on habitat protection and restoration (Pearson et al., 2008). Mystudy does not necessarily impact the NSD’s status as a designatable unitunder SARA, but does argue against full species status for the NSD. Myresults also suggests that the three streams that are sympatric for NSD andLND mtDNA types should ideally be conserved as a part of the NSD’s range,and a unique evolutionary area.78Chapter 4Conclusion4.1 Summary of findingsMy study shows that despite subtle morphological differentiation betweenthe Nooksack dace (NSD) and longnose dace (LND) within the Rhinichthyscataractae group, they showed no signs of reproductive isolation in sympatry.Nooksack dace and LND show subtle morphological differentiation, butcannot be reliably discriminated, and the effect of sampling location isstronger than the effect of mtDNA clade. There is overlap between theNSD and LND for all morphometric and meristic characters measured, in-cluding the scale counts that first identified differences between NSD andLND (McPhail, 1967). Cluster analysis revealed that some allopatric NSDand LND populations are morphological outliers, but failed to separate NSDand LND from each other, either in sympatry or allopatry. In contrast, dis-criminant function analysis—which is “trained” on pre-assigned groups—was 88.4% successful in assigning fish to the correct clade (allopatric pop-ulations only; 74.9% successful for all populations), indicating that somemorphological differentiation exists between LND and NSD.My data suggest that NSD and LND freely interbred in the streams inwhich they are sympatric; there are no reproductive barriers. In all threestreams with sympatric dace there was no evidence for population structurethat was associated with mtDNA type. None of the streams with sympatricdace appear to be more similar genetically either to allopatric NSD or LND,and in a Structure analysis the streams with sympatric dace clusteredtogether. For both morphological and genetic variation, geographic locationaccounted for a greater proportion of the variation than putative species.In general, my study shows that despite potentially millions of years ofisolation in separate glacial refugia, and a 2–3% divergence in mtDNA, therehas been minimal morphological differentiation between the NSD and LND,and there are no reproductive barriers evident between the kinds of daceafter they came into secondary contact.794.2. Conservation implications4.2 Conservation implicationsThe Nooksack dace is currently listed as Endangered under Canada’s Speciesat Risk Act (SARA), but its taxonomic status remains unclear. One of themajor outcomes of my study was to determine whether the Nooksack daceshould be considered a species distinct from the longnose dace, a subspeciesof the longnose dace, or simply a geographically cohesive variant of thelongnose dace.There has long been debate among conservation biologists, policy-makers,and taxonomists (to name a few), over what the appropriate unit for conser-vation is. As the urgency to counteract anthropogenic environmental changebecomes greater, it becomes ever more important to make wise conservationdecisions; it is clear that not everything can be conserved, so where shall wedraw the line? “Evolutionarily significant units” or ESUs, loosely definedas populations that have a distinct evolutionary potential, are consideredworth conserving (Ryder, 1986; Moritz, 1994), and are distinguished usingsome combination of genetics, natural history, morphometrics, and distri-bution. Under Canada’s Species at Risk Act, the ESU concept is employedin the recognition of “designatable units” (DUs, e.g., Taylor et al., 2013).DUs are identified according to one (or more) of four criteria: being a namedsubspecies or variety; being a genetically distinct unit based on morphology,genetics, life history, or behavioural traits; being a geographically disjunctpopulation; being a biogeographically distinct unit in a different ecogeo-graphic region. The DU must also represent a subdivision within the specieswhich has a differing risk of extinction / conservation status.Under these criteria, the Nooksack dace does represent a DU regardless ofits apparent lack of reproductive isolation with the longnose dace. The NSDis morphologically differentiated (though differences are slight), and forms adistinct mtDNA clade, meeting the ”genetically distinct unit” criterion. Itsconservation status, at least within Canada, is also demonstrably differentfrom the longnose dace. In a broader sense, the Nooksack dace representsa separate evolutionary lineage, as part of the distinctive Chehalis fauna.However, the DU criteria are in large part designed to allow the listing ofpopulations that are likely to be genetically distinct, when no genetic surveyhas yet taken place. In the case of the NSD, its largely distinct geographicrange, slight morphological differences, and distinct mtDNA all suggest thatit would be genetically distinct, yet the analysis of nuclear DNA does notuphold this. In light of this, the case for continuing to list the NSD as DUis weak, and with extinction rates as high as they are right now, a triageapproach should place the NSD at a very low priority level.804.2. Conservation implicationsRegarding the taxonomic status of the Nooksack dace, its lack of repro-ductive isolation from the longnose dace argues that it should not receivespecies status. Indeed, cases like the Nooksack dace make an argumentagainst relying upon percentage divergence thresholds for declaring newspecies and conservation priorities. Despite 2–3% divergence in mtDNA,which puts it just within the estimated “species cutoff” for cyprinid fishes(April et al., 2011), the Nooksack dace exhibits no reproductive isolationfrom its sister lineage, nor does it display any apparent functional diver-gence. Particularly in sympatry, it is difficult to conceive of different adap-tations in fish bearing distinct NSD and LND mtDNA when those fish canbe caught in the same riffles (i.e., areas with 1 m2) and are morphologicallynearly identical to one another. If, as some have argued (e.g. Crandall etal., 2000), conservation efforts should prioritize the maintenance of adaptivediversity, then the Nooksack dace may not be a high priority. Future studymay uncover such functional differences, but the apparent free interbreedingbetween NSD and LND in sympatry, and the apparent health and longevityof the mtDNA–introgressed populations, does not suggest that any such dif-ferences would be significant, and that the NSD and LND are not distinctspecies.As mentioned in Chapter 2, the question of whether the NSD should beclassified as a subspecies of the LND is more contentious, in large part be-cause subspecies criteria is generally vague. The NSD certainly lies in a greyare where, by virtue of slight morphological differentiation or genetic differ-entiation, it could be classified as a subspecies, and a great many subspecieshave been defined on lesser evidence (Mayr, 1982). However, if subspeciesdifferences are taken to indicate incipient speciation and/or adaptation todiffering environmental conditions, I do not think that the NSD should re-ceive subspecies status. Rather than heading towards increased reproductiveisolation and complete speciation, the NSD and LND appear to be gradu-ally progressing towards greater intermixing and merging of their gene pools.Morphological differences do not appear to be driven by large-scale environ-mental variation, though they may be due to microhabitat differences; thispoint requires further study, but for the time being, I recommend that theNSD be classified solely as a DU and not as a subspecies.If the preservation of evolutionary processes and the network of connec-tions between populations are goals of conservation, then the three streamsthat are sympatric for NSD and LND mtDNA types should be consideredpart of the range of the Nooksack dace and protected as such. It is unlikelythat human activities caused the overlap between Nooksack and longnosedace, and thus they provide a valuable example of a natural zone of post-814.3. “Ephemeral” speciationglacial secondary contact and genetic exchange between the Nooksack andlongnose dace. However, from a more pragmatic point of view and with lim-ited resources available to conservation efforts, the NSD is only marginallydistinct from the LND, maintains a healthy population in the United States,and therefore should not be given high conservation priority.4.3 “Ephemeral” speciationThe path to speciation is rarely straightforward; for every neat dichoto-mous branching, there are many other cases of incomplete speciation, or“ephemeral speciation” in which new lineages form but do not persist (Rosen-blum et al., 2012). Dynesius and Jansson (2013) suggested that the rate ofspeciation should be considered as having three separate components: rate oflineage splitting, level of persistence of within-species lineages, and lengthof “speciation duration,” the time required to complete speciation. Theysuggested that the level of persistence as a factor of speciation rate has beenunderstudied, and that low persistence (as has been suggested by Rosen-blum et al. (2012)) is widespread and an important factor in determiningoverall rates of speciation. Additionally, while there has been a considerableamount of work on incomplete speciation and remerging of lineages whichhave arisen through ecological speciation (Taylor et al., 2006; Nosil et al.,2009; Behm et al., 2010), there has been comparatively little study of in-complete speciation in lineages which have diverged allopatrically. Studiesin zones of secondary contact are an important component of investigatingthese questions and further understanding the rate of allopatric speciation.Situations such as that of the NSD and LND are infrequently studied,in part because they are difficult to identify: the two fish are, practicallyspeaking, not morphologically differentiated, so there was no obvious markerof the zone of secondary contact until their mtDNA was studied. Discoveryof such cryptic hybrid zones is becoming more common with widespreadgenetic sampling, however, and will hopefully lead to a greater number ofstudies investigating the stages of allopatric speciation, in particular thepersistence of within-species lineages. Guidelines for determining what per-cent divergence indicates separate species is generally done by comparingthe percent divergence within and among well-studied taxa whose speciesstatus is not in doubt. Studies such as my own suggest that percent diver-gence should not be used as a reliable indicator, but deep divergences canand should be used as a pointer towards new areas of research.Further work on the R. cataractae species complex could focus on dace824.3. “Ephemeral” speciationeast of the Rocky Mountains; my morphological and genetic analyses bothindicated that populations in eastern British Columbia and in Ontario arefar outliers from northwestern American dace, and previous work by Bart-nik (1972) found that at least some of the R. cataractae east of the RockyMountains have different mating behaviour from those west of the continen-tal divide.83BibliographyAdams, Stephanie M, James B Lindmeier, and David D Duvernell (May2006). Microsatellite analysis of the phylogeography, Pleistocene historyand secondary contact hypotheses for the killifish, Fundulus heteroclitus.Molecular Ecology, 15(4), pp. 1109–23.Albrecht, Gene H, Bruce R Gelvin, and Steve E Hartman (1993). Ratiosas a size adjustment in morphometrics. American Journal of PhysicalAnthropology, 91(4), pp. 441–468.Allendorf, FW and RF Leary (1988). Conservation and Distribution of Ge-netic Variation in a Polytypic Species, the Cutthroat Trout. Conserva-tion Biology, 2(2), pp. 170–184.April, Julien, Robert H Hanner, Anne-Marie Dion-Coˆte´, and Louis Bernatchez(Jan. 2013). Glacial cycles as an allopatric speciation pump in north-eastern American freshwater fishes. Molecular Ecology, 22(2), pp. 409–22.April, Julien, Richard L Mayden, Robert H Hanner, and Louis Bernatchez(2011). Genetic calibration of species diversity among North America’sfreshwater fishes. Proceedings of the National Academy of Sciences, 108(26),pp. 10602–10607.Avise, John C et al. (1998). Pleistocene phylogeographic effects on avianpopulations and the speciation process. Proceedings of the Royal Societyof London. Series B: Biological Sciences, 265(1395), pp. 457–463.Avise, John C (2000). Phylogeography: the history and formation of species.Harvard University Press.— (2001). Cytonuclear genetic signatures of hybridization phenomena: ra-tionale, utility, and empirical examples from fishes and other aquaticanimals. Reviews in Fish Biology and Fisheries, 10(3), pp. 253–263.84BIBLIOGRAPHYAvise, John C and Nancy C Saunders (1984). Hybridization and introgres-sion among species of sunfish (Lepomis): analysis by mitochondrial DNAand allozyme markers. Genetics, 108(1), p. 237.Avise, John C and DeEtte Walker (1999). Species realities and numbers insexual vertebrates: perspectives from an asexually transmitted genome.Proceedings of the National Academy of Sciences, 96(3), pp. 992–995.Bailey, Reeve M, Howard Elliott Winn, and C Lavett Smith (1954). Fishesfrom the Escambia River, Alabama and Florida, with ecologic and taxo-nomic notes. Proceedings of the Academy of Natural Sciences of Philadel-phia, pp. 109–164.Bailey, RM and GR Smith (1981). Origin and Geography of the Fish Faunaof the Laurentian Great Lakes Basin. Canadian Journal of Fisheries andAquatic Sciences, 38, pp. 1539–1561.Balloux, Franc¸ois and Nicolas Lugon-Moulin (Feb. 2002). The estimation ofpopulation differentiation with microsatellite markers. Molecular Ecol-ogy, 11(2), pp. 155–65.Baltz, Donald M, Peter B Moyle, and Ned J Knight (1982). Competitiveinteractions between benthic stream fishes, riffle sculpin, Cottus gulosus,and speckled dace, Rhinichthys osculus. Canadian Journal of Fisheriesand Aquatic Sciences, 39(11), pp. 1502–1511.Bartnik, V. G. (Jan. 1972). Comparison of the breeding habits of two sub-species of longnose dace, Rhinichthys cataractae. Canadian Journal ofZoology, 50(1), pp. 83–86.Bartnik, Victor G. (1970). Reproductive Isolation Between Two SympatricDace, Rhinichthys atratulus and R. cataractae, in Manitoba. Journal ofthe Fisheries Research Board of Canada, 27(12), pp. 2125–2141.Barton, NH and G M Hewitt (1989). Adaptation, speciation and hybridzones. Nature, 341(6242), pp. 497–503.Barton, Nicholas H and GM Hewitt (1985). Analysis of hybrid zones. Annualreview of Ecology and Systematics, pp. 113–148.Beasley, Rochelle, Stacey L. Lance, Jennifer A. Ruskey, and Eric B. Taylor(2014). Development and characterization of twenty-five microsatellitemarkers for the longnose dace (Cyprinidae: Rhinichthys) using paired-end Illumina shotgun sequencing. Conservation Genetics Resources, pp. 1–3.85BIBLIOGRAPHYBehm, Jocelyn E, Anthony R Ives, and Janette W Boughman (Jan. 2010).Breakdown in postmating isolation and the collapse of a species pairthrough hybridization. The American naturalist, 175(1), pp. 11–26.Belkhir, K., P Borsa, J. Goudet, L. Chikhi, and F. Bonhomme (2004).Genetix (Version 4.05) [Software].Benjamini, Y and D Yekutieli (2001). The Control of the False DiscoveryRate in Multiple Testing under Dependency. The Annals of Statistics,29(4), pp. 1165–1188.Bernardi, Giacomo (2013). Speciation in fishes. Molecular Ecology, 22(22),pp. 5487–5502.Bernatchez, L and CC Wilson (1998). Comparative phylogeography of Nearc-tic and Palearctic fishes. Molecular Ecology, 7, pp. 431–452.Bickford, David, David J Lohman, Navjot S Sodhi, Peter K L Ng, RudolfMeier, Kevin Winker, Krista K Ingram, and Indraneil Das (Mar. 2007).Cryptic species as a window on diversity and conservation. Trends inEcology & Evolution, 22(3), pp. 148–55.Billman, Eric J., Jared B. Lee, D. Owen Young, Matthew D. McKell, R. PaulEvans, and Dennis K. Shiozawa (Apr. 2010). Phylogenetic Divergence ina Desert Fish: Differentiation of Speckled Dace within the Bonneville,Lahontan, and Upper Snake River Basins. Western North American Nat-uralist, 70(1), pp. 39–47.Birky, CW, T Maruyama, and Paul Fuerst (1983). An approach to pop-ulation and evolutionary genetic theory for genes in mitochondria andchloroplasts, and some results. Genetics, 103, pp. 513–527.Bisson, Peter A and Paul E Reimers (1977). Geographic Variation amongPacific Northwest Populations of Longnose Dace, Rhinichthys cataractae.Copeia, 3, pp. 518–522.Bond, J E, M C Hedin, M G Ramirez, and B D Opell (Apr. 2001). Deepmolecular divergence in the absence of morphological and ecologicalchange in the Californian coastal dune endemic trapdoor spider Ap-tostichus simus. Molecular Ecology, 10(4), pp. 899–910.British Columbia Provincial Government (2014). DataBC Geographic Ser-vices.Broughton, RE and JR Gold (2000). Phylogenetic Relationships in the NorthAmerican Cyprinid Genus Cyprinella (Actinopterygii: Cyprinidae) Based86BIBLIOGRAPHYon Sequences of the Mitochondrial ND2 and ND4L Genes. Copeia, 2000(1),pp. 1–10.Bryson, Robert W., Robert W. Murphy, Amy Lathrop, and David Lazcano-Villareal (Apr. 2011). Evolutionary drivers of phylogeographical diver-sity in the highlands of Mexico: a case study of the Crotalus triseriatusspecies group of montane rattlesnakes. Journal of Biogeography, 38(4),pp. 697–710.Chapleau, F and G Pageau (1985). Morphological differentiation of Etheostomaolmstedi and E. nigrum (Pisces: Percidae) in Canada. Copeia, 4, pp. 855–865.COSEWIC (2007). COSEWIC assessment and update status report on theNooksack dace Rhinichthys cataractae ssp. in Canada. Ottawa: Commit-tee on the Status of Endangered Wildlife in Canada, vii + 27 pp.— (2013). COSEWIC Assessment and Status Report on the Salish SuckerCatostomus sp. cf. catostomus in Canada. Ottawa: Committee on theStatus of Endangered Wildlife in Canada.— (2014). Operations and procedures manual. Environment Canada, Cana-dian Wildlife Service. Ottawa, ON, Canada.Coyne, Jerry A and H Allen Orr (2004). Speciation. Sinauer Associates Sun-derland, MA.Crandall, Keith A, Olaf RP Bininda-Emonds, Georgina M Mace, and RobertK Wayne (2000). Considering evolutionary processes in conservation bi-ology. Trends in Ecology & Evolution, 15(7), pp. 290–295.Dawley, RM and KA Goddard (1988). Diploid-triploid mosaics among uni-sexual hybrids of the minnows Phoxinus eos and Phoxinus neogaeus.Evolution, 42(4), pp. 649–659.De Queiroz, Kevin (Dec. 2007). Species concepts and species delimitation.Systematic Biology, 56(6), pp. 879–86.Dobzhansky, T. and T.G. Dobzhansky (1937). Genetics and the Origin ofSpecies. Classics of Modern Evolution Series. Columbia University Press.Dowling, TE and BD DeMarais (1993). Evolutionary significance of intro-gressive hybridization in cyprinid fishes. Nature, 362, pp. 444–446.Dowling, TE, CA Tibbets, WL Minckley, and Gerald R Smith (2002). Evo-lutionary relationships of the plagopterins (Teleostei: Cyprinidae) fromcytochrome b sequences. Copeia, 3, pp. 665–678.87BIBLIOGRAPHYDunteman, George H (1989). Principal components analysis. Sage.Dyke, A.S., J.T. Andrews, P.U. Clark, J.H. England, G.H. Miller, J. Shaw,and J.J. Veillette (Jan. 2002). The Laurentide and Innuitian ice sheetsduring the Last Glacial Maximum. Quaternary Science Reviews, 21(1-3),pp. 9–31.Dynesius, Mats and Roland Jansson (Apr. 2013). Persistence of within-species lineages: a neglected control of speciation rates. Evolution, 68(4),pp. 923–34.Earl, Dent A et al. (2012). STRUCTURE HARVESTER: a website and pro-gram for visualizing STRUCTURE output and implementing the Evannomethod. Conservation Genetics Resources, 4(2), pp. 359–361.Edge, Thomas A, Don E McAllister, and Sami U Qadri (1991). Meristicand morphometric variation between the endangered Acadian whitefish,Coregonus huntsmani, and the lake whitefish, Coregonus clupeaformis, inthe Canadian Maritime Provinces and the State of Maine, USA. Cana-dian Journal of Fisheries and Aquatic Sciences, 48(11), pp. 2140–2151.Egge, Jacob J. D. and Andrew M. Simons (Nov. 2006). The challenge oftruly cryptic diversity: diagnosis and description of a new madtom catfish(Ictaluridae: Noturus). Zoologica Scripta, 35(6), pp. 581–595.Elliott, NG, K Haskard, and JA Koslow (1995). Morphometric analysis of or-ange roughy (Hoplostethus atlanticus) off the continental slope of south-ern Australia. Journal of Fish Biology, 46(2), pp. 202–220.Evanno, G, S Regnaut, and J Goudet (July 2005). Detecting the numberof clusters of individuals using the software Structure: a simulationstudy. Molecular Ecology, 14(8), pp. 2611–20.Excoffier, Laurent and Heidi EL Lischer (2010). Arlequin suite ver 3.5: anew series of programs to perform population genetics analyses underLinux and Windows. Molecular Ecology Resources, 10(3), pp. 564–567.Feulner, PGD, F Kirschbaum, V Mamonekene, V Ketmaier, and R Tiede-mann (2007). Adaptive radiation in African weakly electric fish (Teleostei:Mormyridae: Campylomormyrus): a combined molecular and morpho-logical approach. Journal of Evolutionary Biology, 20(1), pp. 403–414.Feulner, PGD, F Kirschbaum, C Schugardt, V Ketmaier, and R Tiedemann(2006). Electrophysiological and molecular genetic evidence for sympatri-cally occuring cryptic species in African weakly electric fishes (Teleostei:88BIBLIOGRAPHYMormyridae:Campylomormyrus). Molecular Phylogenetics and Evolution,39(1), pp. 198–208.Fielding, Alan (2007). Cluster and classification techniques for the biosciences.Cambridge, UK: Cambridge University Press.Floeter, SR, CEL Ferreira, A Dominici-Arosemena, and IR Zalmon (2004).Latitudinal gradients in Atlantic reef fish communities: trophic structureand spatial use patterns. Journal of Fish Biology, 64, pp. 1680–1699.Fraley, Chris, Adrian E. Raftery, Thomas Brendan Murphy, and Luca Scrucca(2012). mclust Version 4 for R: Normal Mixture Modeling for Model-Based Clustering, Classification, and Density Estimation. 597. TechnicalReport.Fraser, BA, NE Mandrak, and RL McLaughlin (2005). Lack of morpho-logical differentiation in eastern (Rhinichthys atratulus) and western(Rhinichthys obtusus) blacknose dace in Canada. Canadian Journal ofZoology, 83(11), pp. 1502–1509.Funk, W Chris, Marcel Caminer, and Santiago R Ron (2012). High levels ofcryptic species diversity uncovered in Amazonian frogs. Proceedings ofthe Royal Society B: Biological Sciences, 279(1734), pp. 1806–1814.Gee, JH and TG Northcote (1963). Comparative Ecology of Two Sym-patric Species of Dace (Rhinichthys) in the Fraser River System, BritishColumbia. Journal of the Fisheries Board of Canada, 20(1), pp. 105–118.Girard, Philippe and Bernard Angers (Mar. 2006). Characterization of mi-crosatellite loci in longnose dace (Rhinichthys cataractae) and interspe-cific amplification in five other Leuciscinae species. Molecular EcologyNotes, 6, pp. 69–71.— (2011). The Functional Gene Diversity in Natural Populations over Post-glacial Areas: The Shaping Mechanisms Behind Genetic Composition ofLongnose Dace (Rhinichthys cataractae) in Northeastern North Amer-ica. Journal of molecular evolution, 73(1-2), pp. 45–57.Goudet, J (1995). FSTAT (Version 1.2) A computer program to calculateF-statistics. Journal of Heredity, 86, pp. 485–486.Hausdorf, Bernhard (Apr. 2011). Progress toward a general species concept.Evolution, 65(4), pp. 923–31.Hebert, Paul D N, Sujeevan Ratnasingham, and Jeremy R DeWaard (Aug.2003). Barcoding animal life: cytochrome c oxidase subunit 1 divergences89BIBLIOGRAPHYamong closely related species. Proceedings of the Royal Society B: Bio-logical Sciences (Suppl.) 270, S96–9.Hedrick, PW (1999). Perspective: highly variable loci and their interpreta-tion in evolution and conservation. Evolution, 53(2), pp. 313–318.Hewitt, G M (Feb. 2004). Genetic consequences of climatic oscillations in theQuaternary. Philosophical transactions of the Royal Society of London.Series B, Biological sciences, 359(1442), 183–95, discussion 195.Hey, Jody, Yong-Jin Won, Arjun Sivasundar, Rasmus Nielsen, and JeffreyA. Markert (Apr. 2004). Using nuclear haplotypes with microsatellitesto study gene flow between recently separated Cichlid species. MolecularEcology, 13(4), pp. 909–919.Hijmans, Robert J, Susan E Cameron, Juan L Parra, Peter G Jones, andAndy Jarvis (2005). Very high resolution interpolated climate surfacesfor global land areas. International Journal of Climatology, 25(15), pp. 1965–1978.Hoekzema, Kendra and Brian L Sidlauskas (May 2014). Molecular phyloge-netics and microsatellite analysis reveal cryptic species of speckled dace(Cyprinidae: Rhinichthys osculus) in Oregon’s Great Basin. MolecularPhylogenetics and Evolution, 77, pp. 238–250.Hogner, Silje, Terje Laskemoen, Jan T Lifjeld, Jiri Porkert, Oddmund Kleven,Tamer Albayrak, Bekir Kabasakal, and Arild Johnsen (Dec. 2012). Deepsympatric mitochondrial divergence without reproductive isolation inthe common redstart, Phoenicurus phoenicurus. Ecology and Evolution,2(12), pp. 2974–88.Hubbs, Carl L (1955). Hybridization between Fish Species in Nature. Sys-tematic Zoology, 4(1), pp. 1–20.Hubbs, Carl L, Karl F Lagler, and G.R. Smith (1958). Fishes of the GreatLakes Region. University of Michigan Press.Husson, Francois, Julie Josse, Sebastien Le, and Jeremy Mazet (2013). Fac-toMineR: Multivariate Exploratory Data Analysis and Data Mining withR. R package version 1.25.Hyde, J R, C a Kimbrell, J E Budrick, E a Lynn, and R D Vetter (Feb.2008). Cryptic speciation in the vermilion rockfish (Sebastes miniatus)and the role of bathymetry in the speciation process. Molecular Ecology,17(4), pp. 1122–36.90BIBLIOGRAPHYImbrie, John (1985). A theoretical framework for the Pleistocene ice ages.Journal of the Geological Society of London, 142(1840), pp. 417–432.Irwin, Darren E (2002). Phylogeographic breaks without geographic barriersto gene flow. Evolution, 56(12), pp. 2383–2394.— (2012). Local adaptation along smooth ecological gradients causes phylo-geographic breaks and phenotypic clustering. The American Naturalist,180(1), pp. 35–49.Jackson, Donald A, Pedro R Peres-Neto, and Julian D Olden (Jan. 2001).What controls who is where in freshwater fish communities: the roles ofbiotic, abiotic, and spatial factors. Canadian Journal of Fisheries andAquatic Sciences, 58(1), pp. 157–170.Jarne, P and P J Lagoda (Oct. 1996). Microsatellites, from molecules topopulations and back. Trends in Ecology & Evolution, 11(10), pp. 424–9.Jolliffe, I.T. (2002). Principal Component Analysis. Springer Series in Statis-tics. Springer.Jonsson, B and N. Jonsson (Mar. 2001). Polymorphism and speciation inArctic charr. Journal of Fish Biology, 58(3), pp. 605–638.Jørgensen, Hanne BH, Cino Pertoldi, Michael M Hansen, Daniel E Ruzzante,and Volker Loeschcke (2008). Genetic and environmental correlates ofmorphological variation in a marine fish: the case of Baltic Sea herring(Clupea harengus). Canadian Journal of Fisheries and Aquatic Sciences,65(3), pp. 389–400.Kelly, D.W., H.J. MacIsaac, and D.D. Heath (2006). Vicariance and disper-sal effects on phylogeographic structure and speciation in a widespreadestuarine invertebrate. Evolution, 60(2), pp. 257–267.Kinziger, Andrew P, Rodney J Nakamoto, Eric C Anderson, and Bret CHarvey (Sept. 2011). Small founding number and low genetic diversityin an introduced species exhibiting limited invasion success (speckleddace, Rhinichthys osculus). Ecology and Evolution, 1(1), pp. 73–84.Knowlton, Nancy (1993). Sibling species in the sea. Annual Review of Ecol-ogy and Systematics, 24, pp. 189–216.Krosch, Matt N, Andrew M Baker, Brendan G Mckie, Peter B Mather, andPeter S Cranston (2009). Deeply divergent mitochondrial lineages reveal91BIBLIOGRAPHYpatterns of local endemism in chironomids of the Australian Wet Tropics.Austral Ecology, 34(3), pp. 317–328.Lavoue´, Se´bastien, Masaki Miya, Matthew E Arnegard, Peter B McIntyre,Victor Mamonekene, and Mutsumi Nishida (Apr. 2011). Remarkablemorphological stasis in an extant vertebrate despite tens of millions ofyears of divergence. Proceedings of the Royal Society B: Biological Sci-ences, 278(1708), pp. 1003–8.Leavitt, Dean H, Robert L Bezy, Keith a Crandall, and Jack W Sites (Nov.2007). Multi-locus DNA sequence data reveal a history of deep crypticvicariance and habitat-driven convergence in the desert night lizard Xan-tusia vigilis species complex (Squamata: Xantusiidae). Molecular Ecol-ogy, 16(21), pp. 4455–81.Lee-Yaw, Julie A, Jason T Irwin, and David M Green (2008). Postglacialrange expansion from northern refugia by the wood frog, Rana sylvatica.Molecular Ecology, 17(3), pp. 867–884.Lindsey, CC and JD McPhail (1986). Zoogeography of fishes of the Yukonand Mackenzie basins. The Zoogeography of North American FreshwaterFishes, pp. 639–674.Lleonart, Jordi, Jordi Salat, and Gabriel J Torres (2000). Removing allomet-ric effects of body size in morphological analysis. Journal of TheoreticalBiology, 205(1), pp. 85–93.Lu, G, DJ Basley, and L Bernatchez (2001). Contrasting patterns of mi-tochondrial DNA and microsatellite introgressive hybridization betweenlineages of lake whitefish (Coregonus clupeaformis); relevance for speci-ation. Molecular Ecology, 10(4), pp. 965–985.Matthews, William J, Robert E Jenkins, and John T Styron Jr. (1982).Systematics of Two Forms of Blacknose Dace, in a Zone of Syntopy,with a Review of the Species Group. Copeia, 4, pp. 902–920.Mayer, F and O von Helversen (Sept. 2001). Cryptic diversity in Europeanbats. Proceedings of the Royal Society of London. Series B: BiologicalSciences, 268(1478), pp. 1825–32.Mayr, Ernst (1942). Systematics and the Origin of Species, From the View-point of a Zoologist. Harvard University Press.— (1963). Animal Species and Evolution. Animal Species and Their Evo-lution.92BIBLIOGRAPHYMayr, Ernst (1982). Of what use are subspecies? The Auk, pp. 593–595.McDowall, RM (2003). Variation in vertebral number in galaxiid fishes(Teleostei: Galaxiidae): A legacy of life history, latitude and length. En-vironmental Biology of Fishes, 66, pp. 361–381.McPhail, J. D. and E. B. Taylor (June 2009). Phylogeography of the long-nose dace (Rhinichthys cataractae) species group in Northwestern NorthAmerica — the origin and evolution of the Umpqua and Millicoma dace.Canadian Journal of Zoology, 87(6), pp. 491–497.McPhail, J.D. (1967). Distribution of Freshwater Fishes in Western Wash-ington. Contribution (University of Washington. College of Fisheries).College of Fisheries, University of Washington.McPhail, JD (1984). Ecology and evolution of sympatric sticklebacks (Gas-terosteus): morphological and genetic evidence for a species pair in EnosLake, British Columbia. Canadian Journal of Zoology, 62(7), pp. 1402–1408.McPhail, J.D. JD (1997). Status of the Nooksack Dace, Rhinichthys sp., inCanada. The Canadian Field Naturalist, 111(2), pp. 258–262.McPhail, JD and EB Taylor (1999). Morphological and genetic variation innorthwestern longnose suckers, Catostomus catostomus: the Salish suckerproblem. Copeia, 4, pp. 884–893.Moran, Paul and Irv Kornfield (1993). Retention of an ancestral polymor-phism in the Mbuna species flock (Teleostei: Cichlidae) of Lake Malawi.Molecular Biology and Evolution, 10, pp. 1015–1015.Moritz, C, TE Dowling, and WM Brown (1987). Evolution of animal mito-chondrial DNA: relevance for population biology and systematics. An-nual Review of Ecology and Systematics, pp. 269–292.Moritz, Craig (1994). Defining ‘evolutionarily significant units’ for conser-vation. Trends in Ecology & Evolution, 9(10), pp. 373–375.Nosil, Patrik, Luke J Harmon, and Ole Seehausen (Mar. 2009). Ecologicalexplanations for (incomplete) speciation. Trends in Ecology & Evolution,24(3), pp. 145–56.Oakey, David D., Michael E. Douglas, and Marlis R. Douglas (May 2004).Small Fish in a Large Landscape: Diversification of Rhinichthys oscu-lus (Cyprinidae) in Western North America. Copeia, (2). Ed. by R. M.Wood, pp. 207–221.93BIBLIOGRAPHYOosterhout, Cock van, Bill Hutchinson, Derek Wills, and Peter Shipley(2004). MICRO-CHECKER: Microsatellite Data Checking Software.Østbye, K, TF Naesje, L Bernatchez, OT Sandlund, and K Hindar (2005).Morphological divergence and origin of sympatric populations of Euro-pean whitefish (Coregonus lavaretus L.) in Lake Femund, Norway. Jour-nal of Evolutionary Biology, 18(3), pp. 683–702.Patten, Michael A, Philip Unitt, and F Sheldon (2002). Diagnosability versusmean differences of Sage Sparrow subspecies. The Auk, 119(1), pp. 26–35.Pearson, M.P., T. Hatfield, J.D. McPhail, J.S. Richardson, J.S. Rosenfeld, D.Schreier, D. Schluter, D.J. Sneep, M. Stejpovic, E.B. Taylor, and P.M.Wood (2008). Recovery Strategy for the Nooksack Dace (Rhinichthyscataractae) in Canada. Species at Risk Act Recovery Stragegy Series,Fisheries and Oceans Canada, Vancouver. vi + 29 pp.Pfenninger, Markus and David Posada (2002). Phylogeographic history ofthe land snail Candidula unifasciata (Helicellinae, Stylommatophora):fragmentation, corridor migration, and secondary contact. Evolution,56(9), pp. 1776–1788.Pfrender, Michael E, Justin Hicks, and Michael Lynch (Mar. 2004). Bio-geographic patterns and current distribution of molecular-genetic varia-tion among populations of speckled dace, Rhinichthys osculus (Girard).Molecular Phylogenetics and Evolution, 30(3), pp. 490–502.Pritchard, J K, M Stephens, and P Donnelly (June 2000). Inference ofpopulation structure using multilocus genotype data. Genetics, 155(2),pp. 945–59.Provan, Jim and K D Bennett (Oct. 2008). Phylogeographic insights intocryptic glacial refugia. Trends in Ecology & Evolution, 23(10), pp. 564–71.QGIS Development Team (2009). QGIS Geographic Information System.Open Source Geospatial Foundation.R Core Team (2013). R: A Language and Environment for Statistical Com-puting. R Foundation for Statistical Computing. Vienna, Austria.Ravaoarimanana, IB, R Tiedemann, D Montagnon, and Y Rumpler (2004).Molecular and cytogenetic evidence for cryptic speciation within a rare94BIBLIOGRAPHYendemic Malagasy lemur, the Northern Sportive Lemur (Lepilemur septen-trionalis). Molecular Phylogenetics and Evolution, 31(2), pp. 440–448.Redenbach, Z. and E. B. Taylor (Nov. 2003). Evidence for bimodal hybridzones between two species of char (Pisces: Salvelinus) in northwesternNorth America. Journal of Evolutionary Biology, 16(6), pp. 1135–1148.Redenbach, Z and Eric B Taylor (May 2002). Evidence for historical in-trogression along a contact zone between two species of char (Pisces:Salmonidae) in northwestern North America. Evolution, 56(5), pp. 1021–35.Reist, James D (1985). An empirical evaluation of several univariate methodsthat adjust for size variation in morphometric data. Canadian Journalof Zoology, 63(6), pp. 1429–1439.Rosenblum, Erica Bree, Brice a J Sarver, Joseph W Brown, Simone DesRoches, Kayla M Hardwick, Tyler D Hether, Jonathan M Eastman,Matthew W Pennell, and Luke J Harmon (June 2012). Goldilocks MeetsSanta Rosalia: An Ephemeral Speciation Model Explains Patterns of Di-versification Across Time Scales. Evolutionary Biology, 39(2), pp. 255–261.Rousset, Francois (2012). Genepop (Version 4.2.2) [Software].Ryder, Oliver A (1986). Species conservation and systematics: the dilemmaof subspecies. Trends in Ecology & Evolution, 1(1), pp. 9–10.Sa´ez, Alberto G and Encarnacio´n Lozano (Jan. 2005). Body doubles. Nature,433(7022), p. 111.Salini, J.P, D.a Milton, M.J Rahman, and M.G Hussain (Jan. 2004). Al-lozyme and morphological variation throughout the geographic rangeof the tropical shad, hilsa Tenualosa ilisha. Fisheries Research, 66(1),pp. 53–69.Schluter, Dolph and JD McPhail (1992). Ecological character displacementand speciation in sticklebacks. The American Naturalist, 140(1), pp. 85–108.Scott, W.B. and E.J. Crossman (1998). Freshwater Fishes of Canada. GaltHouse Publications.Scribner, Kim T, Kevin S Page, and Meredith L Bartron (2000). Hybridiza-tion in freshwater fishes: a review of case studies and cytonuclear meth-95BIBLIOGRAPHYods of biological inference. Reviews in Fish Biology and Fisheries, 10(3),pp. 293–323.Shafer, Aaron B A, Catherine I Cullingham, Steeve D Coˆte´, and David WColtman (Nov. 2010). Of glaciers and refugia: a decade of study shedsnew light on the phylogeography of northwestern North America. Molec-ular Ecology, 19(21), pp. 4589–621.Sites, Jack W and Jonathon C Marshall (2004). Operational criteria for de-limiting species. Annual Review of Ecology, Evolution, and Systematics,35, pp. 199–227.Smith, Gerald R and Thomas E Dowling (2008). Correlating hydrographicevents and divergence times of speckled dace (Rhinichthys: Teleostei:Cyprinidae) in the Colorado River drainage. Geological Society of Amer-ica Special Papers, 439, pp. 301–317.Smith, GR (1973). Analysis of several hybrid cyprinid fishes from westernNorth America. Copeia, pp. 395–410.Stewart, John R. and Adrian M. Lister (Nov. 2001). Cryptic northern refugiaand the origins of the modern biota. Trends in Ecology & Evolution,16(11), pp. 608–613.Swain, Douglas P and Chris J Foote (1999). Stocks and chameleons: the useof phenotypic variation in stock identification. Fisheries Research, 43,pp. 113–128.Tabachnick, Barbara G, Linda S Fidell, et al. (2001). Using MultivariateStatistics. Allyn and Bacon Boston.Taylor, E B, J W Boughman, M Groenenboom, M Sniatynski, D Schluter,and J L Gow (Feb. 2006). Speciation in reverse: morphological and ge-netic evidence of the collapse of a three-spined stickleback (Gasterosteusaculeatus) species pair. Molecular Ecology, 15(2), pp. 343–55.Taylor, Eric B, Charles-A Darveau, and Patricia M Schulte (2013). SettingConservation Priorities in a Widespread Species: Phylogeographic andPhysiological Variation in the Lake Chub, Couesius plumbeus (Pisces:Cyprinidae). Diversity, 5(2), pp. 149–165.Thorpe, John P (1982). The molecular clock hypothesis: biochemical evolu-tion, genetic differentiation and systematics. Annual Review of Ecologyand Systematics, pp. 139–168.96BIBLIOGRAPHYTipton, Michelle L, Sarah Gignoux-Wolfsohn, Phoebe Stonebraker, and BarryChernoff (Nov. 2011). Postglacial recolonization of eastern BlacknoseDace, Rhinichthys atratulus (Teleostei: Cyprinidae), through the gate-way of New England. Ecology and Evolution, 1(3), pp. 343–58.Toews, David P L and Alan Brelsford (Aug. 2012). The biogeography ofmitochondrial and nuclear discordance in animals. Molecular Ecology,21(16), pp. 3907–30.Trontelj, Peter and Cene Fiˇser (Nov. 2008). Cryptic species diversity shouldnot be trivialised. Systematics and Biodiversity, 7(01), p. 1.Turan, Cemal, Mustafa Oral, Bayram O¨ztu¨rk, and Ertug˘ Du¨zgu¨nes¸ (2006).Morphometric and meristic variation between stocks of Bluefish (Po-matomus saltatrix ) in the Black, Marmara, Aegean and northeasternMediterranean Seas. Fisheries Research, 79(1), pp. 139–147.Turgeon, J and L Bernatchez (Nov. 2001). Clinal variation at microsatelliteloci reveals historical secondary intergradation between glacial races ofCoregonus artedi (Teleostei: Coregoninae). Evolution, 55(11), pp. 2274–86.Turner, George F (2002). Parallel speciation, despeciation and respeciation:implications for species definition. Fish and Fisheries, 3(3), pp. 225–229.United States Geological Survey (2014). National Hydrography Dataset.Valavanis, Vasilis D, Stratis Georgakarakos, Argyris Kapantagakis, AndreasPalialexis, and Isidora Katara (Nov. 2004). A GIS environmental mod-elling approach to essential fish habitat designation. Ecological Modelling,178(3-4), pp. 417–427.Vonlanthen, P, L Excoffier, D Bittner, H Persat, S Neuenschwander, andC R Largiade`r (Nov. 2007). Genetic analysis of potential postglacialwatershed crossings in Central Europe by the bullhead (Cottus gobioL.). Molecular ecology, 16(21), pp. 4572–84.Ward, Robert D, Tyler S Zemlak, Bronwyn H Innes, Peter R Last, and PaulD N Hebert (Oct. 2005). DNA barcoding Australia’s fish species. Philo-sophical transactions of the Royal Society of London. Series B, Biologicalsciences, 360(1462), pp. 1847–57.Webb, William C, John M Marzluff, and Kevin E Omland (June 2011).Random interbreeding between cryptic lineages of the Common Raven:evidence for speciation in reverse. Molecular Ecology, 20(11), pp. 1–13.97Wiens, John J, Tag N Engstrom, and Paul T Chippindale (Dec. 2006). Rapiddiversification, incomplete isolation, and the“speciation clock” in NorthAmerican salamanders (Genus Plethodon): testing the hybrid swarm hy-pothesis of rapid radiation. Evolution, 60(12), pp. 2585–603.Wilson, Chris and Louis Bernatchez (1998). The ghost of hybrids past: fix-ation of arctic charr (Salvelinus alpinus) mitochondrial DNA in an in-trogressed population of lake trout (S. namaycush). Molecular Ecology,7, pp. 127–132.Wood, BM and MB Bain (1995). Morphology and microhabitat use in streamfish. Canadian Journal of Fisheries and Aquatic Sciences, 52, pp. 1487–1498.Zamudio, Kelly R and Wesley K Savage (July 2003). Historical isolation,range expansion, and secondary contact of two highly divergent mi-tochondrial lineages in spotted salamanders (Ambystoma maculatum).Evolution, 57(7), pp. 1631–52.Zemlak, Tyler S, Robert D Ward, Allan D Connell, Bronwyn H Holmes, andPaul D N Hebert (May 2009). DNA barcoding reveals overlooked marinefishes. Molecular Ecology Resources, 9, pp. 237–42.Zielin´ski, P, K Nadachowska-Brzyska, B Wielstra, R Szkotak, S D Covaciu-Marcov, D Coga˘lniceanu, and W Babik (Apr. 2013). No evidence for nu-clear introgression despite complete mtDNA replacement in the Carpathiannewt (Lissotriton montandoni). Molecular Ecology, 22(7), pp. 1884–903.98Appendix AChapter 2Table A.1: List of sampling locations for Nooksack and longnose dace (Rhinichthys cataractae) used in morpho-logical analyses. Catalogue no. refers to the catalogue of the Beaty Biodiversity Museum (University of BritishColumbia), where tissue samples which have not yet been entered into the museum catalogue are indicated by“—-”.N denotes number of fish from that collection that were measured.CatalogueNo.Allopatric /SympatricSpeciesPresentLocation LocationCodeSiteNo.Latitude Longitude CollectionDateNBC80-0084 A NSD Wynoochee River, WA,USAWY 1 46.98069 123.64640 25/03/1975 13BC80-0093 A NSD Wynoochee River, WA,USAWY 2 46.98069 123.64640 25/04/1975 15BC66-0096 A NSD Willapa River, WA, USA WP 1 46.56001 123.56020 31/08/1961 13BC06-0236 A NSD Willapa River, WA, USA WP 2 46.53591 123.46143 27/07/1990 411-0539 A NSD West Fork Satsop River,WA, USASR 1 47.05983 123.54097 12/08/2007 26BC11-0544 A NSD Porter Creek, WA, USA PC 1 46.94645 123.29543 13/08/2007 511-0018 A NSD Willamette River, WA,USAWR 1 43.80578 123.05827 21/09/1994 7BC76-0038 A NSD Cowlitz Rver, WA, USA CW 2 46.35624 122.93204 18/02/1961 4Continued on next page99AppendixA.Chapter2Table A.1 – continued from previous pageCatalogueNo.Allopatric /SympatricSpeciesPresentLocation LocationCodeSiteNo.Latitude Longitude CollectionDateNBC11-0418 A NSD Cowlitz Rver, WA, USA CW 1 46.30808 122.92053 30/08/1958 1511-0019 A NSD Brunette River, BC,CanadaBR 1 49.24145 122.89594 03/06/2002 28BC11-0004 S BOTH Coquitlam River, BC,CanadaCR 3 49.28031 122.77632 08/06/2004 10BC11-0006 S BOTH Coquitlam River, BC,CanadaCR 4 49.32602 122.77153 08/08/2004 1— S BOTH Coquitlam River, BC,CanadaCR 4 49.32602 122.77153 31/08/2008 2— S BOTH Coquitlam River, BC,CanadaCR 5 49.32602 122.77153 8BC11-0007 S BOTH Coquitlam River, BC,CanadaCR 3 49.04155 122.77107 09/07/2004 6BC11-0010 S BOTH Coquitlam River, BC,CanadaCR 2A 49.33616 122.76821 07/06/2004 5BC57-0269 A LND Wright Creek, BC, Canada WC 1 54.09764 122.68121 13/06/1953 30BC11-0005 S BOTH Alouette River North, BC,CanadaAN 3 49.24242 122.60108 15/08/2004 7— S BOTH Alouette River North, BC,CanadaAN 1 49.24285 122.57974 27/09/2012 19BC06-0128 S BOTH Alouette River, BC,CanadaAR 4 49.24159 122.57978 18/05/1994 27BC11-0008 S BOTH Alouette River, BC,CanadaAR 3 49.23439 122.60097 18/07/2004 5BC82-0012 S BOTH Alouette River, BC,CanadaAR 4 49.23918 122.57932 21/05/1976 1BC11-0009 S BOTH Alouette River, BC,CanadaAR 4 49.23918 122.57932 09/07/2004 5Continued on next page100AppendixA.Chapter2Table A.1 – continued from previous pageCatalogueNo.Allopatric /SympatricSpeciesPresentLocation LocationCodeSiteNo.Latitude Longitude CollectionDateN11-0541 S BOTH Kanaka Creek, BC,CanadaKC 1 49.20216 122.54129 17/08/2007 3311-0540 S BOTH Kanaka Creek, BC,CanadaKC 2 49.20727 122.53630 17/08/2007 12— S BOTH Kanaka Creek, BC,CanadaKC 2 49.20727 122.53630 2008/01/08 2— S BOTH Kanaka Creek, BC,CanadaKC 4 49.21558 122.52399 31/08/2008 11— S BOTH Kanaka Creek, BC,CanadaKC 3 49.21210 122.50855 31/08/2008 6— S BOTH Kanaka Creek, BC,CanadaKC 5 49.22612 122.46117 31/08/2008 8BC11-0542 A NSD Bertrand Creek, BC,CanadaBT 1 49.00432 122.53217 19/07/2007 2011-0020 A LND Norrish Creek, BC, Canada NC 2 49.23485 122.13359 11/06/1992 7BC06-0215 A LND Norrish Creek, BC, Canada NC 3 49.23485 122.13359 18/05/1994 1859-0600 A LND Norrish Creek, BC, Canada NC 1 49.23581 122.13261 27/08/1955 6BC59-0002 A LND Hope Coquihalla River,BC, CanadaCQ 1 49.38851 121.43344 02/07/1952 32BC11-0003 A LND Petit Creek, BC, Canada PT 1 50.15396 121.03088 15/08/1994 20BC56-0558 A LND Columbia River, Big Bendregion, BC, CanadaCL 1 51.55186 118.53917 07/08/1952 3011-0517 A LND Beaver Creek, BC, Canada BC 1 49.06946 117.60099 04/08/2007 111-0516 A LND Beaver Creek, BC, Canada BC 2 49.10086 117.55722 04/08/2007 9BC57-0361 A LND High Wood River, AB,CanadaHW 1 50.80708 113.78779 15/06/1953 16Continued on next page101AppendixA.Chapter2Table A.1 – continued from previous pageCatalogueNo.Allopatric /SympatricSpeciesPresentLocation LocationCodeSiteNo.Latitude Longitude CollectionDateNBC57-0362 A LND Junction of Blindman &Red Deer Rivers, AB,CanadaRD 1 51.35493 113.75697 16/06/1953 30BC57-0357 A LND Milk River, AB, Canada MR 1 49.07399 110.77100 14/06/1953 19BC59-0563 A LND Pine River, LakeWinnipegosis, MB, CanadaPR 1 52.47925 100.03532 16/10/1950 30BC58-0047 A BND Schoolcraft Country, MI,USABN 1 46.07371 86.19049 06/10/1957 10102AppendixA.Chapter2Table A.2: Raw data of vertebral counts made from CT scans for allopatric samples of Nooksack (NSD) andlongnose dace (LND, Rhinichthys cataractae) sampled from British Columbia (BC), Ontario (ON), andWashingtonState (WA) . N = 53.Vertebral CountLocation Species 34 35 36 37Beaver Creek, BC LND 0 1 4 3Beaver River, ON LND 0 5 3 0Columbia River, BC LND 0 4 6 0Norrish Creek, BC NSD 0 4 5 1Satsop River, WA NSD 0 1 4 2Wynoochee River, WA NSD 2 4 3 1103AppendixA.Chapter2Table A.3: Eigenvalues and percentage of variance explained by each axis of a principal components (PC) analysisconducted on 11 size-transformed morphological values and two meristic traits in all samples of Nooksack andlongnose dace (Rhinichthys cataractae).PC Eigenvalue Percentage variance Cumulativepercentage variance1 2.38 18.28 18.282 1.80 13.84 32.123 1.32 10.21 42.334 1.14 8.80 51.145 0.95 7.36 58.506 0.90 6.99 65.507 0.82 6.38 71.898 0.72 5.55 77.459 0.69 5.35 82.8010 0.65 5.00 87.8111 0.62 4.80 92.6112 0.53 4.14 96.7613 0.42 3.23 100.0014 0.00 0.00 100.00104AppendixA.Chapter2Table A.4: Character loadings for the first five axes of the principal components analysis of 11 size-transformedmorphological traits and two meristic traits. Principal components analysis was conducted on 582 longnose daceand Nooksack dace (Rhinichthys cataractae).Measurement PC 1 PC 2 PC 3 PC 4 PC 5M1 0.5439 −0.2775 0.3319 0.0785 −0.1842M2 0.6771 −0.0322 0.3298 0.2085 −0.0250M3 0.2003 −0.0638 0.6246 0.2065 0.3013M4 0.5440 −0.1216 −0.3152 -0.2379 0.2689M5 0.5242 −0.1583 −0.4185 -0.0294 0.3064M6 0.7130 0.2040 −0.2647 0.1237 −0.2361M7 0.3562 0.5658 −0.2208 0.2264 −0.2344M8 0.0647 0.5473 0.1996 −0.2424 −0.4586M9 −0.1981 0.1125 0.0852 0.7703 0.1549M10 −0.2819 0.3631 −0.4387 0.4145 0.0871M11 0.4005 0.5723 0.1834 −0.0320 0.2461Lateral line scale count −0.3023 0.5296 0.2150 −0.175 0.0647Pectoral fin ray count 0.0154 0.5136 0.0750 −0.2699 0.4987105AppendixA.Chapter2Table A.5: Bayesian information criterion (BIC) values for each model and number of clusters for the clusteranalysis of principal component scores from morphological analysis of all samples of Nooksack and longnose dace(Rhinichthys cataractae). The top three models based on the BIC criterion are in boldface type.No. ofclus-tersEII VII EEI VEI EVI VVI EEE EEV VEV VVV1 −9517.3 −9517.3 −9386.3 −9386.3 −9386.3 −9386.3 −9449.9 −9449.9 −9449.9 −9449.92 −9320.4 −9001.1 −9284.2 −8863.0 −9168.4 −8862.2 −9321.8 −9186.6 −8895.4 −8915.03 −9213.0 −8941.6 −9131.1 −8871.0 −9088.2 −8787.4 −9125.8 −9122.5 −8846.6 −8957.94 −9182.7 −8915.7 −9104.9 −8792.4 −8935.6 -8761.3 −9162.9 −9128.5 −8828.8 −8891.05 −9181.0 −8768.0 −9096.8 -8732.1 −8930.3 −8784.3 −9137.6 −9009.5 −8861.1 −8932.96 −9110.1 −8787.6 −8913.8 -8742.7 −8932.0 −8818.1 −8955.8 −9168.8 −8939.4 −9018.87 −9031.9 −8781.2 −8808.1 −8764.7 −8960.4 −8857.4 −9065.7 −9087.5 −8988.3 −9102.58 −9127.3 −8792.1 −8952.5 −8768.3 −9033.2 −8914.1 −9057.8 −9142.6 −9071.1 NA9 −9060.5 −8801.2 −9007.2 −8770.6 −9042.7 −8949.0 −8981.6 −9224.4 −9130.9 NA106AppendixA.Chapter2Table A.6: Bayesian information criterion (BIC) values for each model and number of clusters for the clusteranalysis for the cluster analysis of principal component scores from morphological analysis of samples of Nooksackand longnose dace (Rhinichthys cataractae) from allopatric populations. The top three models based on the BICcriterion are in boldface type.No. ofclus-tersEII VII EEI VEI EVI VVI EEE EEV VEV VVV1 −6805.5 −6805.5 −6706.8 −6706.8 −6706.8 −6706.8 −6767.0 −6767.0 −6767.0 −6767.02 −6634.7 −6449.2 −6627.9 −6375.7 −6570.2 −6368.1 −6665.4 −6537.9 −6403.0 −6418.13 −6650.3 −6429.2 −6500.7 -6224.8 −6442.9 −6291.8 −6699.2 −6575.3 −6453.2 −6494.04 −6524.2 −6419.1 −6426.5 −6319.9 −6460.0 −6273.1 −6478.5 −6564.2 −6344.7 −6390.95 −6521.4 −6281.1 −6455.3 -6228.8 −6366.2 −6283.9 −6464.7 −6530.8 −6454.8 NA6 −6543.9 −6276.2 −6450.5 -6199.9 −6442.5 −6276.4 −6474.4 −6565.0 −6447.6 NA7 −6492.9 −6273.4 −6477.4 −6233.1 −6477.2 −6330.0 −6509.7 −6571.4 −6499.3 NA8 −6428.7 −6294.6 −6514.6 −6256.9 −6521.5 −6362.0 −6507.8 −6627.9 −6556.1 NA9 −9060.5 −8801.2 −9007.2 −8770.6 −9042.7 −8949.0 −8981.6 −9224.4 −9130.9 NA107Appendix BChapter 3Table B.1: List of sampling locations for Nooksack and longnose dace (Rhinichthys cataractae) used in geneticanalyses. N denotes number of fish from that collection that were measured.Allopatric /SympatricSpeciesPresentLocation LocationCodeSiteNo.Latitude Longitude Collection Date NA NSD Wynoochee River, WA, USA WY 1 46.98069 123.64640 25/03/1975 4A NSD Wynoochee River, WA, USA WY 2 46.98069 123.64640 25/04/1975 4A NSD West Fork Satsop River, WA, USA SR 1 47.05983 123.54097 12/08/2007 25A NSD Porter Creek, WA, USA PC 1 46.94645 123.29543 13/08/2007 5A NSD Brunette River, BC, Canada BR 1 49.24145 122.89594 03/06/2002 30S BOTH Coquitlam River, BC, Canada CR 3 49.28031 122.77632 08/06/2004 27S BOTH Coquitlam River, BC, Canada CR 4 49.32602 122.77153 08/08/2004 12S BOTH Coquitlam River, BC, Canada CR 3 49.04155 122.77107 09/07/2004 10S BOTH Coquitlam River, BC, Canada CR 2A 49.33616 122.76821 07/06/2004 5S BOTH Coquitlam River, BC, Canada CR X 49.23417 122.75389 2004 14S BOTH Alouette River, BC, Canada AR 3 49.23439 122.60097 18/07/2004 22S BOTH Alouette River, BC, Canada AR 1 49.22278 122.60028 2004 32S BOTH Alouette River, BC, Canada AR 4 49.24159 122.57978 18/05/1994 11S BOTH Alouette River North, BC, Canada AN 1 49.24285 122.57974 27/09/2012 19S BOTH Alouette River, BC, Canada AR 2 49.23918 122.57932 09/07/2004 24S BOTH Alouette River, BC, Canada AR 5 49.23503 122.56014 27/09/2012 2Continued on next page108AppendixB.Chapter3Table B.1 – continued from previous pageAllopatric /SympatricSpeciesPresentLocation LocationCodeSiteNo.Latitude Longitude Collection Date NS BOTH Alouette River, BC, Canada AR 6 49.24443 122.53481 27/09/2012 1A NSD Bertrand Creek, BC, Canada BT 1 49.00432 122.53217 19/07/2007 22S BOTH Kanaka Creek, BC, Canada KC 1 49.20216 122.54129 17/08/2007 49S BOTH Kanaka Creek, BC, Canada KC 2 49.20727 122.53630 17/08/2007 14S BOTH Kanaka Creek, BC, Canada KC 4 49.21558 122.52399 31/08/2008 11S BOTH Kanaka Creek, BC, Canada KC 3 49.21210 122.50855 31/08/2008 6S BOTH Kanaka Creek, BC, Canada KC 5 49.22612 122.46117 31/08/2008 8A LND Fraser River, BC, Canada FR 1 49.38306 122.45222 27/09/2012 7A LND Norrish Creek, BC, Canada NC 1 49.23581 122.13261 27/08/1955 34A LND Hope Coquihalla River, BC, Canada CQ 1 49.38851 121.43344 02/07/1952 19A LND Beaver Creek, BC, Canada BC 1 49.06946 117.60099 04/08/2007 28A LND Beaver River, ON, Canada BV 1 52.47925 100.03532 16/10/1950 10109AppendixB.Chapter3Table B.2: Details for 10 polymorphic microsatellite loci developed for Rhinichthys cataractae. The size indicatesthe range of observed alleles in base pairs and includes the length of the CAG tag; K is number of alleles observed;dye indicates the fluorescent tag used to label the locus.Locus Sequence Forward, 5’—3’Sequence Reverse,5’—-3’Repeat Motif Size (bp) K Dye SourceRhca15b CTCACAGACTAC-CTGCCCCAGAGGT-CAAACAGTAG-TAGG(CTAT)12(CTATCATAT)8(CTAT)3125-341 16TYE665Girard andAngers 2006Rhca16 GAGAACGAGTG-GACATCCAGTGAGTG-GTTGAGTAGG(GA)12GCGT(GT)12110-128 10TYE705Girard andAngers 2006Rhca23 TTCGTCCATATC-TAGAGGTCATGAAT-GCAGTACTGG(CA)7(CT) 4CACT(CA)8(CT)3220-262 8TYE705Girard andAngers 2006Rhca4 ATGCATCCACC-CAAACCCAAGCTTTCAAG-TAATCAGAT-GAGGCATGG 147-163 5TYE705Beasley et al.2014Rhca5 TGTTTCCTTGCC-TACAGCCCAATGTTGAAAT-GTTGCACAAACCATATT 180-210 7TYE665Beasley et al.2014Rhca7 ATCCTGCTGGATC-CGTTGCATAATAGCCAT-GAGGGAGTCCGAAAG 217-313 16TYE705Beasley et al.2014Rhca36 GATACACCTTCT-GCGCTTGCTGAGATGCTGCT-GTGAATGCAAAG 215-275 13WellRedD2Beasley et al.2014Rhca42 GCTGCCTGGTG-TAAATTATGTCCTTACGTTACTCAT-GCGTTTACCCATCT 196-240 9TYE705Beasley et al.2014Rhca43 CAAATCCCTTTG-GAAACAAACCTTACTCTAGACCT-GATCTGTGAGGCAAAG 244-384 19WellRedD2Beasley et al.2014Rhca45 TGCTTTAT-TACTTTCATC-CACCAGGTTCATGC-TAGTTGTTCAACC-TATGGATCT 167-307 19WellRedD2Beasley et al.2014110AppendixB.Chapter3Table B.3: Observed (HE) and expected (HO) heterozygosities for each locus/population combination, for 10microsatellite loci and 12 populations of Rhinichthys cataractae. P indicates the P-value for a test of Hardy-Weinberg equilibrium performed in Genepop v4.2.2 (Rousset 2008). Significant departures from Hardy-Weinbergequilibrium are indicated in bold and with (*).Locus Rhca44 Rhca4 Rhca15b Rhca36 Rhca42 Rhca5 Rhca45 Rhca16 Rhca7 Rhca23PopulationPorter Creek,WAHE 0.7143 0.6 0.9556 0.9111 0.5333 0.3778 0.9556 0.6444 0.9778 0.6857HO 0.8 0.2 0.8 0.8 0.4 0.4 1 0.2 1 0.8P 1 0.0857 0.204 0.3627 0.619 1 1 0.0476 1 1HE 0.6077 0.8065 0.8290 0.9396 0.5657 0.4669 0.9600 0.7698 0.9429 0.8898HO 0.64 1 0.76 0.96 0.52 0.4 1 0.76 1 0.96Satsop River,WAP 1 1 0.204 0.3627 0.619 1 1 0.0476 1 1BertrandCreek, BCHE 0.9144 0.7833 0.9165 0.7992 0.7051 0.5877 0.8118 0.6089 0.9144 0.7030HO 0.8636 0.7273 0.8636 0.7727 0.8182 0.5909 0.8636 0.5909 0.8182 0.8636P 0.2098 0.3025 0.2447 0.4184 0.1195 0.7755 0.7414 0.5119 0.0911 0.7563HE 0.7662 0.2597 0.7446 0.6190 0.5671 0.6104 0.3680 0 0.7619 0HO 0.7273 0.2727 0.6364 0.5455 0.7273 0.4545 0.4545 0 0.7273 0BrunetteRiver, BCP 0.345 1 0.3324 0.4645 0.9173 0.16 1 NA 0.415 NACoquitlamRiver, BCHE 0.8950 0.7865 0.8624 0.8455 0.7363 0.7267 0.9021 0.7289 0.9438 0.6415HO 0.8939 0.8788 0.7879 0.8636 0.7273 0.7424 0.8788 0.8182 0.8182 0.5606P 0.5673 0.608 0.0527 0.5199 0.0242 0.3741 0.3227 0.9927 0* 0.0429HE 0.9286 0.6645 0.8930 0.8396 0.7849 0.6873 0.9168 0.6429 0.9194 0.7577HO 0.9394 0.6667 0.9091 0.8283 0.7576 0.6869 0.9394 0.7576 0.6768 0.6970AlouetteRiver, BCP 0.6382 0* 0.4348 0.469 0.0852 0* 0.9555 0.9741 0* 0.1125Continued on next page111AppendixB.Chapter3Table B.3 – continued from previous pageLocus Rhca44 Rhca4 Rhca15b Rhca36 Rhca42 Rhca5 Rhca45 Rhca16 Rhca7 Rhca23PopulationKanakaCreek, BCHE 0.9073 0.7246 0.8144 0.7642 0.5709 0.4437 0.9529 0.6781 0.8858 0.6165HO 0.9022 0.7717 0.8261 0.6848 0.6087 0.4239 0.9348 0.6413 0.8804 0.5870P 0.12280.0161*0.8506 0.136 0.9502 0.0223 0.2769 0.6128 0.2512 0.0893HE 0.9567 0.6061 0.5152 0.6450 0.7359 0.1775 0.9177 0.8009 0.8571 0.6986HO 1 0.3636 0.5455 0.6364 0.8182 0.1818 0.9091 0.8182 0.4545 0.8182NorrishCreek, BCP 1 0.0817 0.7533 0.4333 0.8639 1 0.2847 0.3823 0* 0.9669Fraser River,BCHE 0.9546 0.7424 0.9546 0.9243 0.8485 0.3182 0.9697 0.7424 0.8148 0.8788HO 1 0.6667 0.8333 1 0.8333 0.3333 0.8333 0.6667 0.8333 0.8333P 1 0.4717 0.3172 1 0.6718 1 0.1581 0.1706 1 0.5322HE 0.9292 0.5992 0.8215 0.9108 0.8800 0.7908 0.9077 0.7415 0.9662 0.8646HO 1 0.3846 0.9231 1 0.8462 1 0.7692 0.8462 1 0.8462CoquihallaRiver, BCP 10.0024*0.9674 1 0.3704 1 0.035 0.9078 1 0.4886BeaverCreek, BCHE 0.7762 0.7964 0.8629 0.7077 0.8629 0.8206 0.9496 0.6351 0.8972 0.4173HO 0.8125 0.75 1 0.75 0.875 0.8125 1 0.625 0.875 0.25P 0.77860.0164*1 0.5653 0.7301 0.4535 1 0.5465 0.1035 0.0315HE 0.3 0.9333 0.6731 0.9417 0.7083 0.4904 0.6111 0.7583 0.8417 0.7917HO 0 0.75 0.625 0.875 0.625 0.625 0.625 0.5 0.625 1Beaver River,ONP 0.0667 0.1087 0.5257 0.421 0.3497 1 1 0.0488 0.0826 1112AppendixB.Chapter3Table B.4: Allele frequencies for 10 microsatellite loci for Nooksack and longnose dace at 12 sampling locations.LocationLocus Allele PorterCreek,WASatsopRiver,WABertrandCreek,BCBrunetteRiver,BCCoquit-lamRiver,BCAlou-etteRiver,BCKanakaCreek,BCNorrishCreek,BCFraserRiver,BCCoqui-hallaRiver,BCBeaverCreek,BCBeaverRiver,BCRhca43 N 4 16 22 11 63 98 89 11 6 13 16 3136 0 0 0 0 0.008 0 0 0 0 0 0 0148 0 0 0 0 0.008 0 0 0 0 0 0 0164 0 0 0 0 0 0 0 0 0 0 0 0.333232 0 0 0 0 0 0 0 0 0 0.038 0 0236 0 0 0 0 0 0 0 0 0 0.115 0 0240 0 0 0 0 0 0 0 0 0.083 0 0 0244 0 0 0 0 0 0 0 0 0.167 0 0 0252 0 0 0 0 0 0 0 0.091 0 0 0 0256 0 0 0.045 0 0 0 0 0 0 0 0 0264 0 0 0 0 0 0 0.006 0 0 0 0.031 0268 0 0.031 0.091 0 0 0.005 0 0.045 0 0.115 0.031 0272 0.375 0.063 0 0 0 0 0.017 0.045 0 0 0 0276 0 0.125 0 0.045 0 0.015 0.011 0 0 0.115 0.031 0280 0 0.063 0 0 0.016 0 0.022 0 0.083 0 0.031 0284 0.125 0.031 0 0 0.008 0.036 0.051 0 0.083 0.038 0.063 0.333288 0.125 0.031 0.136 0 0.079 0.077 0.107 0 0 0.192 0 0.333292 0.125 0.063 0.114 0.091 0.095 0.082 0.056 0 0 0.038 0.063 0296 0 0.031 0.136 0 0.056 0.071 0.067 0.091 0.167 0.115 0.063 0300 0.125 0.094 0.159 0.182 0.087 0.107 0.101 0.045 0.083 0.038 0.031 0304 0 0.094 0.068 0 0.095 0.077 0.056 0.136 0.083 0 0.094 0308 0 0.125 0.068 0.045 0.032 0.071 0.062 0.091 0 0 0.031 0312 0 0.094 0.068 0.091 0.056 0.107 0.112 0.045 0 0 0.031 0316 0 0.031 0 0 0.127 0.026 0.067 0.045 0.083 0 0.469 0320 0 0.063 0 0 0.024 0.066 0.039 0.136 0.167 0.077 0.031 0324 0.125 0 0 0.455 0.04 0.051 0.034 0.045 0 0.038 0 0328 0 0 0.068 0.091 0.048 0.046 0.028 0.045 0 0.038 0 0332 0 0 0.045 0 0.024 0.015 0.017 0 0 0.038 0 0336 0 0.031 0 0 0.016 0.005 0.006 0 0 0 0 0340 0 0.031 0 0 0 0.015 0 0 0 0 0 0344 0 0 0 0 0 0 0.011 0 0 0 0 0348 0 0 0 0 0.016 0 0.011 0.045 0 0 0 0352 0 0 0 0 0.008 0 0 0 0 0 0 0356 0 0 0 0 0.048 0.041 0 0 0 0 0 0360 0 0 0 0 0 0.026 0 0 0 0 0 0364 0 0 0 0 0.071 0.005 0.051 0 0 0 0 0368 0 0 0 0 0.032 0.02 0.062 0 0 0 0 0372 0 0 0 0 0 0 0.006 0.091 0 0 0 0380 0 0 0 0 0 0.015 0 0 0 0 0 0384 0 0 0 0 0.008 0.02 0 0 0 0 0 0Continued on next page113AppendixB.Chapter3Table B.4 – continued from previous pageLocationLocus Allele PorterCreek,WASatsopRiver,WABertrandCreek,BCBrunetteRiver,BCCoquit-lamRiver,BCAlou-etteRiver,BCKanakaCreek,BCNorrishCreek,BCFraserRiver,BCCoqui-hallaRiver,BCBeaverCreek,BCBeaverRiver,BCRhca4 N 4 25 22 11 65 93 89 11 6 10 16 899 0 0 0 0 0 0 0.011 0 0 0 0 0111 0 0 0.045 0 0.008 0 0 0 0 0 0 0115 0 0 0 0 0 0.005 0.006 0 0 0 0 0119 0 0.06 0.091 0 0 0 0.011 0 0 0 0 0123 0 0.28 0.068 0 0.1 0 0.062 0.045 0 0.05 0.063 0127 0 0.02 0.136 0 0.123 0.005 0.18 0 0.083 0 0.344 0131 0 0.32 0.091 0 0.046 0 0 0 0 0.05 0.063 0135 0.25 0.02 0.023 0 0.092 0.038 0.039 0.5 0.167 0.15 0.156 0.188139 0 0.02 0.068 0.864 0.185 0.065 0.084 0 0.083 0.1 0 0143 0.375 0.08 0.432 0 0.369 0.489 0.455 0.045 0.5 0.25 0.25 0147 0.375 0 0 0 0.008 0.194 0.022 0.409 0.167 0.4 0.125 0.063151 0 0.04 0 0 0.008 0 0.017 0 0 0 0 0.063155 0 0.12 0 0 0 0 0.011 0 0 0 0 0163 0 0.04 0 0 0 0 0 0 0 0 0 0.125167 0 0 0.023 0 0 0 0 0 0 0 0 0.188171 0 0 0 0.045 0 0 0 0 0 0 0 0175 0 0 0 0 0.031 0.005 0 0 0 0 0 0179 0 0 0 0 0.008 0.005 0.017 0 0 0 0 0.125183 0 0 0 0 0.008 0 0.011 0 0 0 0 0187 0 0 0 0 0 0.005 0 0 0 0 0 0.063195 0 0 0 0 0 0.005 0.006 0 0 0 0 0199 0 0 0 0 0 0 0.006 0 0 0 0 0211 0 0 0 0 0 0 0 0 0 0 0 0.063235 0 0 0 0 0.008 0.011 0.028 0 0 0 0 0239 0 0 0 0 0 0 0 0 0 0 0 0.063247 0 0 0 0.045 0 0 0 0 0 0 0 0251 0 0 0 0 0 0 0 0 0 0 0 0.063255 0 0 0 0.045 0 0 0 0 0 0 0 0259 0 0 0.023 0 0 0 0 0 0 0 0 0271 0 0 0 0 0 0.027 0.006 0 0 0 0 0275 0 0 0 0 0.008 0.108 0.028 0 0 0 0 0279 0 0 0 0 0 0.038 0 0 0 0 0 0Rhca15bN 5 25 22 11 63 99 91 8 6 13 16 7109 0 0 0.091 0 0 0 0 0 0 0 0 0117 0 0.02 0.114 0 0 0 0.011 0 0 0 0 0119 0 0 0 0 0 0.005 0.005 0 0 0 0 0123 0.2 0.38 0.068 0 0.095 0.005 0.077 0 0 0 0 0125 0.1 0.06 0.159 0 0.135 0.136 0.357 0 0 0 0 0.429127 0 0 0 0 0 0 0 0.063 0 0.077 0.125 0129 0 0 0 0 0 0.005 0 0 0 0 0 0Continued on next page114AppendixB.Chapter3Table B.4 – continued from previous pageLocationLocus Allele PorterCreek,WASatsopRiver,WABertrandCreek,BCBrunetteRiver,BCCoquit-lamRiver,BCAlou-etteRiver,BCKanakaCreek,BCNorrishCreek,BCFraserRiver,BCCoqui-hallaRiver,BCBeaverCreek,BCBeaverRiver,BC131 0 0.02 0 0 0 0.015 0 0 0 0 0 0133 0 0.02 0.023 0 0 0 0 0 0 0.038 0 0135 0 0 0 0 0 0 0 0 0 0 0.125 0137 0 0 0 0 0.008 0.01 0.005 0 0.083 0.038 0 0139 0 0.04 0.091 0 0.159 0.187 0.06 0.5 0.083 0.385 0 0.071141 0 0 0 0 0 0 0.005 0 0 0 0 0145 0 0 0 0 0 0 0 0 0 0 0 0.286149 0.1 0.04 0 0 0 0 0 0 0 0 0 0151 0.1 0.02 0 0 0 0 0 0 0 0 0 0153 0.1 0.14 0 0 0 0 0 0 0 0 0 0155 0 0.02 0 0 0 0 0 0 0 0 0 0157 0.1 0.02 0 0 0 0 0 0 0 0 0.063 0161 0.2 0.1 0 0 0 0 0 0 0 0 0.031 0165 0.1 0.02 0.045 0 0 0 0 0 0.083 0 0 0167 0 0 0 0 0 0 0 0.125 0 0 0 0169 0 0 0 0 0 0 0 0 0 0 0.125 0173 0 0 0 0.136 0 0 0 0.25 0.083 0.038 0 0175 0 0 0 0 0.048 0.025 0.005 0 0.083 0.154 0 0177 0 0 0 0 0.032 0 0 0 0 0.154 0 0179 0 0 0 0 0 0 0 0 0 0.038 0 0183 0 0 0 0 0 0 0 0 0 0.038 0 0185 0 0 0 0 0 0.005 0 0 0 0 0 0211 0 0 0 0 0 0 0.005 0 0 0 0 0217 0 0 0 0 0 0 0 0 0 0 0.313 0225 0 0 0 0 0 0.01 0.005 0 0 0.038 0.031 0231 0 0 0 0 0.016 0.005 0 0 0 0 0.094 0233 0 0 0 0 0.008 0 0 0 0 0 0 0237 0 0 0 0 0 0 0 0 0.083 0 0.031 0239 0 0 0.023 0 0.016 0.076 0.137 0 0.25 0 0.063 0241 0 0 0 0 0 0.005 0 0 0 0 0 0243 0 0 0 0 0 0 0.005 0 0 0 0 0245 0 0 0 0.045 0.04 0.015 0.071 0 0.083 0 0 0247 0 0 0.023 0.227 0.095 0.101 0.038 0.063 0 0 0 0253 0 0 0.136 0.045 0.087 0.152 0.11 0 0 0 0 0255 0 0.06 0.045 0.455 0.135 0.096 0 0 0 0 0 0259 0 0 0 0 0 0.02 0 0 0.083 0 0 0261 0 0.02 0.136 0.091 0.103 0.086 0 0 0 0 0 0263 0 0 0.045 0 0.008 0 0 0 0 0 0 0.071267 0 0 0 0 0.008 0.02 0.016 0 0.083 0 0 0269 0 0.02 0 0 0.008 0.01 0.005 0 0 0 0 0.071275 0 0 0 0 0 0 0.077 0 0 0 0 0281 0 0 0 0 0 0.01 0 0 0 0 0 0303 0 0 0 0 0 0 0 0 0 0 0 0.071Continued on next page115AppendixB.Chapter3Table B.4 – continued from previous pageLocationLocus Allele PorterCreek,WASatsopRiver,WABertrandCreek,BCBrunetteRiver,BCCoquit-lamRiver,BCAlou-etteRiver,BCKanakaCreek,BCNorrishCreek,BCFraserRiver,BCCoqui-hallaRiver,BCBeaverCreek,BCBeaverRiver,BCRhca36 N 5 25 22 11 66 97 92 11 6 13 16 8119 0 0 0 0 0 0.005 0 0 0 0 0 0135 0 0 0 0 0 0.005 0 0 0 0 0 0179 0 0 0 0 0 0 0.005 0 0 0 0 0191 0 0 0 0 0.008 0 0 0 0 0 0 0195 0 0 0 0 0.136 0.103 0.038 0 0 0.038 0 0199 0 0.04 0 0 0.008 0.005 0 0 0 0 0 0203 0 0.14 0.25 0 0 0 0 0.318 0.083 0 0.094 0207 0.2 0.12 0.341 0.591 0.212 0.201 0.38 0.5 0 0.154 0.125 0.063211 0 0.06 0.114 0.091 0.015 0.026 0.234 0 0.083 0.115 0.5 0215 0 0.08 0.136 0 0.015 0.098 0.022 0 0.167 0.077 0 0.063219 0.2 0.06 0 0 0.098 0.191 0.043 0 0.083 0.192 0 0.063223 0 0.04 0.045 0.136 0.25 0.196 0.174 0 0.25 0.077 0 0.188227 0.2 0.1 0 0.182 0.144 0.098 0.076 0.182 0.167 0.038 0.188 0.188231 0.1 0.06 0.068 0 0.053 0.031 0.027 0 0 0.077 0.063 0.063235 0 0.04 0 0 0 0 0 0 0 0.154 0 0239 0 0.06 0.045 0 0.015 0 0 0 0 0 0 0.063243 0 0.02 0 0 0.015 0.021 0 0 0.083 0.077 0 0.125247 0.2 0.04 0 0 0 0.021 0 0 0.083 0 0 0.063251 0 0.06 0 0 0.023 0 0 0 0 0 0 0255 0 0.06 0 0 0 0 0 0 0 0 0.031 0259 0 0 0 0 0 0 0 0 0 0 0 0.063263 0.1 0 0 0 0 0 0 0 0 0 0 0.063267 0 0.02 0 0 0 0 0 0 0 0 0 0279 0 0 0 0 0.008 0 0 0 0 0 0 0Rhca42 N 5 25 22 11 66 99 92 11 6 13 16 820 0 0 0 0 0 0 0 0.045 0 0 0 0164 0 0.04 0 0 0 0 0 0 0 0 0 0168 0 0.02 0.045 0 0 0.005 0.005 0 0 0 0 0172 0 0 0 0.045 0.015 0 0 0 0 0 0 0176 0 0 0 0 0 0 0 0 0 0.077 0 0180 0.6 0.64 0.455 0 0.28 0.187 0.609 0.091 0 0.038 0 0.125184 0 0 0.068 0.455 0.091 0.136 0.082 0.136 0.083 0.231 0 0188 0 0 0 0 0.023 0.071 0 0 0.083 0.154 0.094 0192 0 0 0.091 0 0.114 0.131 0.038 0 0.25 0 0 0.25196 0 0 0.295 0.5 0.409 0.374 0.228 0.409 0.167 0.192 0 0.5200 0.4 0.16 0.045 0 0.015 0.02 0 0 0 0 0.063 0.063204 0 0.08 0 0 0.008 0 0 0 0 0.038 0 0.063208 0 0.02 0 0 0.015 0.056 0.038 0.318 0.333 0.077 0.313 0212 0 0 0 0 0 0.005 0 0 0 0 0.031 0216 0 0.04 0 0 0.03 0 0 0 0 0 0.063 0Continued on next page116AppendixB.Chapter3Table B.4 – continued from previous pageLocationLocus Allele PorterCreek,WASatsopRiver,WABertrandCreek,BCBrunetteRiver,BCCoquit-lamRiver,BCAlou-etteRiver,BCKanakaCreek,BCNorrishCreek,BCFraserRiver,BCCoqui-hallaRiver,BCBeaverCreek,BCBeaverRiver,BC220 0 0 0 0 0 0 0 0 0 0.154 0.125 0224 0 0 0 0 0 0.005 0 0 0 0.038 0 0228 0 0 0 0 0 0 0 0 0.083 0 0.125 0232 0 0 0 0 0 0 0 0 0 0 0.125 0236 0 0 0 0 0 0 0 0 0 0 0.031 0244 0 0 0 0 0 0 0 0 0 0 0.031 0380 0 0 0 0 0 0.005 0 0 0 0 0 0420 0 0 0 0 0 0.005 0 0 0 0 0 0Rhca5 N 5 25 22 11 66 99 92 11 6 13 16 7140 0 0 0 0 0 0.005 0 0 0 0 0 0150 0 0 0 0 0.008 0 0 0 0 0 0 0160 0 0 0 0 0 0 0 0 0 0 0 0.643165 0.1 0.04 0.25 0 0 0.015 0.016 0 0 0 0.031 0170 0.8 0.72 0.591 0.182 0.379 0.53 0.739 0.909 0.833 0.385 0.281 0.214175 0 0.02 0.023 0.591 0.023 0.025 0.027 0 0.083 0.077 0 0180 0 0 0 0 0 0.005 0 0 0 0.077 0.188 0.143185 0.1 0.14 0 0.182 0.341 0.096 0.087 0.045 0 0.192 0.25 0190 0 0.02 0 0.045 0.045 0.015 0 0 0 0.077 0.031 0195 0 0.02 0 0 0.068 0.061 0.033 0 0 0.192 0.156 0200 0 0 0 0 0 0.01 0 0 0 0 0.031 0205 0 0 0 0 0.106 0.131 0.033 0.045 0 0 0 0210 0 0 0.114 0 0.023 0.04 0.038 0 0 0 0.031 0215 0 0.04 0 0 0.008 0.04 0.005 0 0 0 0 0225 0 0 0.023 0 0 0.015 0.011 0 0.083 0 0 0235 0 0 0 0 0 0 0.011 0 0 0 0 0405 0 0 0 0 0 0.005 0 0 0 0 0 0435 0 0 0 0 0 0.005 0 0 0 0 0 0Rhca45 N 5 25 22 11 63 99 92 11 6 13 16 5119 0 0 0 0 0 0 0 0 0 0 0.031 0139 0 0 0 0 0 0.005 0 0 0 0 0 0143 0 0 0 0.773 0.048 0.01 0.011 0 0 0 0 0147 0 0 0 0 0 0.025 0.005 0 0 0 0 0151 0 0 0.023 0 0.008 0.03 0.033 0 0 0 0 0.2155 0 0 0.364 0.227 0.087 0.051 0.033 0 0 0 0 0159 0 0 0.205 0 0.119 0.116 0.011 0.045 0 0 0 0163 0 0 0 0 0.087 0.111 0.011 0.182 0 0 0 0.1167 0 0 0.136 0 0.048 0.005 0.005 0.227 0 0 0 0171 0 0 0 0 0.111 0.04 0.076 0.091 0.083 0 0 0175 0 0 0 0 0.04 0 0 0 0 0 0 0179 0 0 0 0 0 0 0.033 0 0 0 0 0183 0 0 0 0 0.008 0 0 0 0 0 0 0Continued on next page117AppendixB.Chapter3Table B.4 – continued from previous pageLocationLocus Allele PorterCreek,WASatsopRiver,WABertrandCreek,BCBrunetteRiver,BCCoquit-lamRiver,BCAlou-etteRiver,BCKanakaCreek,BCNorrishCreek,BCFraserRiver,BCCoqui-hallaRiver,BCBeaverCreek,BCBeaverRiver,BC187 0 0.02 0 0 0 0 0 0 0 0 0 0191 0 0 0 0 0 0 0.043 0.091 0.083 0 0 0195 0 0 0 0 0.024 0 0.005 0 0 0.231 0.156 0199 0 0 0 0 0.048 0.01 0 0 0.167 0.115 0.063 0203 0.1 0 0 0 0.016 0.005 0.06 0.091 0.083 0.192 0.031 0207 0 0.02 0.091 0 0.04 0.01 0.005 0 0 0.038 0.063 0211 0 0.06 0 0 0.071 0.005 0 0.045 0.083 0.038 0.094 0215 0 0.06 0 0 0.032 0.02 0.005 0.045 0 0.038 0.094 0219 0 0.02 0 0 0.016 0.005 0.011 0 0.083 0 0 0.1223 0 0.04 0.023 0 0.024 0.01 0 0 0 0 0.063 0227 0 0.02 0.023 0 0.016 0.015 0 0 0 0 0.031 0.1231 0 0.06 0 0 0 0.015 0 0 0 0 0.031 0235 0 0.1 0 0 0 0 0 0 0 0 0 0239 0 0 0.023 0 0 0.01 0.005 0 0 0 0.063 0243 0.1 0.1 0.023 0 0 0.01 0.038 0 0.083 0 0 0.1247 0 0 0 0 0 0 0.082 0 0 0 0 0.1251 0.1 0.08 0 0 0 0.01 0.033 0 0 0 0 0.1255 0.2 0.02 0 0 0 0.015 0.011 0.045 0.083 0 0 0.1259 0 0.06 0 0 0.008 0.01 0.076 0.045 0 0 0 0263 0 0.06 0 0 0 0.01 0.043 0.045 0 0 0 0267 0.1 0.06 0 0 0 0.005 0.011 0 0 0.038 0 0271 0 0 0 0 0 0 0.022 0 0 0.077 0 0275 0 0 0 0 0.008 0 0.038 0 0 0.038 0.031 0.1279 0.1 0 0 0 0 0.03 0.022 0 0 0.077 0 0283 0.2 0.02 0 0 0.016 0.091 0.038 0 0 0 0 0287 0 0.02 0 0 0.016 0.202 0.076 0 0.167 0.038 0 0291 0 0.06 0 0 0.024 0.056 0.082 0.045 0.083 0.038 0.031 0295 0 0.02 0 0 0.008 0.025 0.06 0 0 0 0 0299 0 0 0 0 0.016 0.015 0.016 0 0 0 0 0303 0 0.04 0.045 0 0.008 0.005 0 0 0 0 0 0307 0 0 0 0 0.024 0.005 0 0 0 0.038 0 0311 0 0 0.023 0 0.016 0.005 0 0 0 0 0 0315 0 0.02 0 0 0.016 0.005 0 0 0 0 0.063 0323 0 0.02 0 0 0 0 0 0 0 0 0 0335 0 0 0 0 0 0 0 0 0 0 0.094 0343 0.1 0 0 0 0 0 0 0 0 0 0 0351 0 0.02 0 0 0 0 0 0 0 0 0 0355 0 0 0 0 0 0 0 0 0 0 0.063 0395 0 0 0.023 0 0 0 0 0 0 0 0 0Rhca16 N 5 25 22 11 64 96 92 11 6 13 16 880 0 0 0 0 0 0 0 0 0 0 0.031 088 0 0 0 0 0 0.005 0 0 0 0 0 0Continued on next page118AppendixB.Chapter3Table B.4 – continued from previous pageLocationLocus Allele PorterCreek,WASatsopRiver,WABertrandCreek,BCBrunetteRiver,BCCoquit-lamRiver,BCAlou-etteRiver,BCKanakaCreek,BCNorrishCreek,BCFraserRiver,BCCoqui-hallaRiver,BCBeaverCreek,BCBeaverRiver,BC90 0 0 0 0 0 0.005 0 0 0 0 0 096 0 0 0.068 0 0 0 0 0 0 0.038 0 0100 0 0.02 0 0 0.039 0 0.049 0 0 0 0.125 0102 0 0.04 0 0 0 0 0 0.136 0 0 0 0104 0.5 0.4 0 0 0 0 0 0 0 0 0 0108 0.4 0.08 0 0 0 0 0 0 0 0 0 0.438110 0 0.04 0.386 0 0.148 0.057 0.315 0.136 0.167 0.077 0.063 0112 0.1 0.24 0.5 1 0.336 0.495 0.457 0.409 0.083 0.192 0.219 0114 0 0.12 0.023 0 0.313 0.146 0.109 0.136 0.167 0.385 0 0116 0 0.06 0 0 0 0 0 0 0 0 0.563 0118 0 0 0 0 0 0.021 0 0 0 0 0 0120 0 0 0 0 0 0 0 0.045 0 0 0 0122 0 0 0 0 0 0 0 0 0 0 0 0.063124 0 0 0 0 0 0 0.005 0 0 0 0 0.125126 0 0 0 0 0 0 0 0.045 0 0 0 0.25128 0 0 0 0 0 0 0 0 0 0 0 0.125130 0 0 0 0 0 0 0 0 0.083 0 0 0132 0 0 0.023 0 0.141 0.266 0.06 0.091 0.5 0.308 0 0134 0 0 0 0 0.008 0 0 0 0 0 0 0136 0 0 0 0 0 0.005 0 0 0 0 0 0144 0 0 0 0 0.016 0 0 0 0 0 0 0170 0 0 0 0 0 0 0.005 0 0 0 0 0Rhca7 N 5 25 22 11 64 95 91 11 5 13 16 821 0 0 0 0 0 0 0.005 0 0 0 0 0137 0 0 0 0 0.008 0 0 0 0 0 0 0181 0 0 0 0 0.008 0 0 0 0 0 0 0189 0 0 0.045 0 0.039 0 0 0 0 0.038 0 0193 0 0 0 0 0 0 0.044 0 0 0.038 0 0197 0 0 0 0 0.023 0.011 0.027 0 0 0 0 0201 0 0.04 0 0 0.016 0.042 0 0 0 0 0 0205 0 0.02 0 0 0.008 0.037 0 0 0 0.077 0.094 0209 0 0.08 0.227 0 0.008 0.011 0.154 0 0.1 0.077 0.219 0.063213 0 0.06 0.023 0 0.016 0.011 0.159 0 0 0 0.188 0.125217 0.1 0.02 0.045 0 0.063 0.079 0.071 0.273 0.1 0.077 0.156 0.375221 0.1 0.12 0.045 0.318 0.016 0.084 0.198 0.091 0 0.038 0.031 0.063225 0 0.08 0.045 0 0.055 0.068 0 0.045 0.2 0.038 0.031 0229 0.1 0.12 0.114 0.136 0.078 0.026 0.016 0.227 0 0.038 0 0233 0 0.02 0.114 0.227 0.063 0.037 0.088 0.091 0 0 0 0.188237 0.1 0.1 0.114 0.318 0.023 0.058 0.005 0 0.1 0 0 0241 0 0.02 0.045 0 0 0.047 0.033 0 0.1 0 0.031 0245 0.1 0.04 0.023 0 0.016 0.037 0.005 0 0 0.038 0.031 0249 0 0 0.023 0 0.047 0.084 0.005 0 0.1 0.077 0.063 0.063Continued on next page119AppendixB.Chapter3Table B.4 – continued from previous pageLocationLocus Allele PorterCreek,WASatsopRiver,WABertrandCreek,BCBrunetteRiver,BCCoquit-lamRiver,BCAlou-etteRiver,BCKanakaCreek,BCNorrishCreek,BCFraserRiver,BCCoqui-hallaRiver,BCBeaverCreek,BCBeaverRiver,BC253 0 0.02 0 0 0.016 0.047 0.033 0.182 0 0 0 0257 0.2 0.1 0 0 0.031 0.042 0.027 0 0 0.115 0.031 0261 0 0.02 0.045 0 0.023 0.047 0 0 0.1 0.077 0 0265 0.1 0.06 0 0 0.008 0.021 0 0.045 0 0.115 0 0.063269 0.1 0.02 0 0 0.008 0.016 0 0 0 0.038 0 0273 0 0.02 0 0 0.008 0.011 0 0.045 0 0 0.063 0277 0 0.02 0 0 0.023 0.021 0 0 0 0 0.031 0281 0 0 0 0 0.016 0.021 0 0 0 0 0 0285 0 0 0 0 0 0.011 0 0 0 0 0 0289 0 0 0 0 0.031 0.016 0.005 0 0.1 0 0 0293 0 0.02 0 0 0.016 0 0.011 0 0 0.038 0 0297 0.1 0 0 0 0 0 0 0 0 0 0 0301 0 0 0 0 0 0 0.011 0 0 0 0.031 0309 0 0 0 0 0.016 0.011 0 0 0 0 0 0313 0 0 0 0 0.016 0 0 0 0 0 0 0317 0 0 0 0 0.008 0 0 0 0 0 0 0.063321 0 0 0 0 0.023 0.016 0 0 0 0 0 0325 0 0 0 0 0 0.005 0 0 0 0 0 0329 0 0 0 0 0 0.005 0 0 0 0 0 0333 0 0 0 0 0.008 0.011 0.038 0 0 0 0 0337 0 0 0 0 0.023 0 0.005 0 0 0 0 0341 0 0 0 0 0.008 0.005 0 0 0 0 0 0345 0 0 0 0 0.039 0 0 0 0.1 0 0 0349 0 0 0.023 0 0.031 0 0 0 0 0.038 0 0353 0 0 0 0 0 0 0 0 0 0.038 0 0357 0 0 0.023 0 0.031 0.005 0 0 0 0 0 0361 0 0 0.023 0 0.016 0 0 0 0 0 0 0365 0 0 0 0 0 0.016 0 0 0 0 0 0369 0 0 0.023 0 0.016 0.005 0 0 0 0 0 0373 0 0 0 0 0.016 0 0 0 0 0 0 0381 0 0 0 0 0.023 0.005 0.005 0 0 0 0 0385 0 0 0 0 0.016 0 0.016 0 0 0 0 0389 0 0 0 0 0 0.005 0.011 0 0 0 0 0393 0 0 0 0 0.008 0 0.011 0 0 0 0 0397 0 0 0 0 0.016 0 0.005 0 0 0 0 0401 0 0 0 0 0 0.005 0 0 0 0 0 0409 0 0 0 0 0 0 0.005 0 0 0 0 0417 0 0 0 0 0 0.011 0 0 0 0 0 0421 0 0 0 0 0.016 0 0 0 0 0 0 0429 0 0 0 0 0 0.005 0 0 0 0 0 0437 0 0 0 0 0.008 0.005 0 0 0 0 0 0Rhca23 N 4 25 22 11 63 96 86 10 6 13 16 8Continued on next page120AppendixB.Chapter3Table B.4 – continued from previous pageLocationLocus Allele PorterCreek,WASatsopRiver,WABertrandCreek,BCBrunetteRiver,BCCoquit-lamRiver,BCAlou-etteRiver,BCKanakaCreek,BCNorrishCreek,BCFraserRiver,BCCoqui-hallaRiver,BCBeaverCreek,BCBeaverRiver,BC204 0 0 0 0 0 0 0 0 0 0.038 0 0212 0 0 0 0 0 0 0 0 0 0 0 0.188214 0 0 0 0 0 0.01 0 0.15 0.083 0.308 0 0216 0 0 0 0 0.008 0.01 0 0 0 0.038 0 0218 0 0 0 0 0 0 0 0 0 0 0 0.375234 0 0 0 0 0 0 0 0 0 0.038 0.75 0236 0 0 0 0 0 0 0.006 0 0 0.154 0.156 0238 0 0.04 0 0 0 0 0.006 0.1 0.167 0.038 0 0240 0 0.02 0 0 0 0 0 0 0 0 0 0244 0 0 0 0 0 0 0 0 0 0.038 0 0246 0 0.04 0 0 0 0 0 0 0 0 0 0248 0 0 0 0 0 0.031 0 0 0.25 0.154 0.094 0250 0 0 0 0 0 0.005 0 0 0 0 0 0254 0 0 0 0 0.056 0.036 0.012 0 0 0 0 0.25256 0.25 0.16 0 0 0.04 0.01 0.023 0 0 0 0 0.125258 0.25 0.04 0.114 0 0.095 0.13 0.25 0.45 0.083 0 0 0.063260 0.25 0.22 0.386 1 0.524 0.328 0.506 0.1 0.25 0.077 0 0262 0 0.08 0.136 0 0.008 0.026 0 0 0 0 0 0264 0.25 0.14 0 0 0.048 0.125 0.023 0 0.167 0.115 0 0266 0 0.04 0 0 0 0.005 0.006 0 0 0 0 0268 0 0.12 0.364 0 0.206 0.281 0.157 0.05 0 0 0 0270 0 0.08 0 0 0 0 0 0 0 0 0 0272 0 0 0 0 0.016 0 0 0 0 0 0 0274 0 0.02 0 0 0 0 0 0 0 0 0 0278 0 0 0 0 0 0 0.006 0 0 0 0 0288 0 0 0 0 0 0 0 0.15 0 0 0 0300 0 0 0 0 0 0 0.006 0 0 0 0 0121AppendixB.Chapter3Table B.5: Pairwise FST values for all sampling locations. All are significant at P < 0.05.PorterCr.W.ForkSatsopR.BertrandCr.BrunetteR.Coquit-lamR.Alou-etteR.KanakaCr.NorrishCr.FraserR.Coqui-hallaR.BeaverCr., BCBeaverR., ONPorter Cr. — 0.0227 0.1005 0.3463 0.0896 0.0789 0.0832 0.1134 0.0682 0.0883 0.1704 0.1121W.Fork Satsop R. 0.0227 — 0.0799 0.2721 0.0797 0.0896 0.0797 0.128 0.0926 0.1039 0.1549 0.147Bertrand Cr. 0.1005 0.0799 — 0.2292 0.0498 0.0493 0.0409 0.1144 0.0887 0.1097 0.1664 0.1571Brunette R. 0.3463 0.2721 0.2292 — 0.1697 0.1894 0.2216 0.3051 0.312 0.266 0.3343 0.3432Coquitlam R. 0.0896 0.0797 0.0498 0.1697 — 0.0235 0.0491 0.1015 0.0552 0.0619 0.1394 0.1209Alouette R. 0.0789 0.0896 0.0493 0.1894 0.0235 — 0.0504 0.0857 0.0315 0.0548 0.1474 0.1255Kanaka Cr. 0.0832 0.0797 0.0409 0.2216 0.0491 0.0504 — 0.1161 0.0805 0.1152 0.1664 0.1598Norrish Cr. 0.1134 0.128 0.1144 0.3051 0.1015 0.0857 0.1161 — 0.0808 0.0859 0.1833 0.1657Fraser R. 0.0682 0.0926 0.0887 0.312 0.0552 0.0315 0.0805 0.0808 — 0.0338 0.126 0.1199Coquihalla R. 0.0883 0.1039 0.1097 0.266 0.0619 0.0548 0.1152 0.0859 0.0338 — 0.1192 0.1205Beaver Cr., BC 0.1704 0.1549 0.1664 0.3343 0.1394 0.1474 0.1664 0.1833 0.126 0.1192 — 0.1848Beaver Cr., ON 0.1121 0.147 0.1571 0.3432 0.1209 0.1255 0.1598 0.1657 0.1199 0.1205 0.1848 —122AppendixB.Chapter3Table B.6: Pairwise FST for all sampling locations, with each sympatric population broken down into subsetsfor those dace with LND mtDNA and those with NSD mtDNA. All are significant at P < 0.05, except for thosemarked in gray.PorterCreek(NSD)SatsopRiver(NSD)BertrandCreek(NSD)BrunetteRiver(NSD)Coq.RiverLNDCoq.RiverNSDAl.RiverLNDAl.RiverNSDKan.CreekLNDKan.CreekNSDNor.Creek(LND)FraserRiver(LND)Coqui-hallaRiver(LND)BeaverCreek,BC(LND)BeaverRiver,ON(LND)Porter Cr. — 0.0227 0.1005 0.3463 0.0853 0.0906 0.0755 0.0728 0.0747 0.0833 0.1134 0.0682 0.0883 0.1704 0.1121Satsop R. 0.0227 — 0.0799 0.2721 0.0784 0.0867 0.0868 0.084 0.0792 0.0715 0.128 0.0926 0.1039 0.1549 0.147Bertrand Cr. 0.1005 0.0799 — 0.2292 0.0475 0.0554 0.0509 0.0555 0.0375 0.0379 0.1144 0.0887 0.1097 0.1664 0.1571Brunette R. 0.3463 0.2721 0.2292 — 0.184 0.2059 0.1874 0.2136 0.264 0.2237 0.3051 0.312 0.266 0.3343 0.3432Coq. R. LND 0.0853 0.0784 0.0475 0.184 — −0.0005 0.0153 0.0234 0.039 0.0473 0.0972 0.0527 0.0596 0.1465 0.1118Coq. R. NSD 0.0906 0.0867 0.0554 0.2059 −0.0005 — 0.0208 0.039 0.045 0.0439 0.1084 0.0509 0.0587 0.1512 0.1258Al. R. LND 0.0755 0.0868 0.0509 0.1874 0.0153 0.0208 — 0.0027 0.0374 0.0454 0.0874 0.0262 0.0484 0.1411 0.1208Al. R. NSD 0.0728 0.084 0.0555 0.2136 0.0234 0.039 0.0027 — 0.0403 0.0531 0.0764 0.0363 0.0476 0.1422 0.1219Kan. Cr. LND 0.0747 0.0792 0.0375 0.264 0.039 0.045 0.0374 0.0403 — 0.0008 0.0943 0.0565 0.0929 0.1492 0.1354Kan. Cr. NSD 0.0833 0.0715 0.0379 0.2237 0.0473 0.0439 0.0454 0.0531 0.0008 — 0.115 0.0736 0.1098 0.162 0.1597Norrish Cr. 0.1134 0.128 0.1144 0.3051 0.0972 0.1084 0.0874 0.0764 0.0943 0.115 — 0.0808 0.0859 0.1833 0.1657Fraser R. 0.0682 0.0926 0.0887 0.312 0.0527 0.0509 0.0262 0.0363 0.0565 0.0736 0.0808 — 0.0338 0.126 0.1199Coquihalla R. 0.0883 0.1039 0.1097 0.266 0.0596 0.0587 0.0484 0.0476 0.0929 0.1098 0.0859 0.0338 — 0.1192 0.1205Beaver Cr., BC 0.1704 0.1549 0.1664 0.3343 0.1465 0.1512 0.1411 0.1422 0.1492 0.162 0.1833 0.126 0.1192 — 0.1848Beaver R., ON 0.1121 0.147 0.1571 0.3432 0.1118 0.1258 0.1208 0.1219 0.1354 0.1597 0.1657 0.1199 0.1205 0.1848 —123

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0167029/manifest

Comment

Related Items