Modeling the Occurrence of Hybridization between ...

ASSESSING THE EXTENT OF HYBRIDIZATION BETWEEN WESTSLOPE CUTTHROAT TROUT AND INTRODUCED RAINBOW TROUT IN THE UPPER KOOTENAY RIVER, BRITISH COLUMBIA by Stephen N. Bennett A dissertation submitted in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY in Fisheries Biology Approved: Dr. Jeff Kershner Major Professor

Dr. Brett Roper Committee Member

Dr. Phaedra Budy Committee Member

Dr. Karen Beard Committee Member

Dr. Mark Miller Committee Member

Dr. Byron Burnham Dean of Graduate Studies UTAH STATE UNIVERSITY Logan, Utah 2007

ii

Copyright © Stephen N. Bennett 2007 All Rights Reserved

iii ABSTRACT Assessing the Current and Potential Extent of Hybridization Between Westslope Cutthroat Trout and Introduced Rainbow Trout in the Upper Kootenay River, British Columbia by Stephen N. Bennett, Doctor of Philosophy Utah State University, 2007 Major Professor: Dr. Jeff Kershner Department: Watershed Sciences The distribution and genetic integrity of all inland cutthroat trout (Oncorhynchus clarki ssp.) have been negatively affected by the introduction of non-native salmonid species throughout their range. Westslope cutthroat trout (WCT, O. c. lewisi) have the largest historic distribution of all the inland cutthroat trout. In the USA the range of WCT has contracted and many remaining populations have hybridized with introduced salmonids, especially rainbow trout (RBT, O. mykiss). In this study I evaluate the genetic integrity of WCT in the Upper Kootenay River, British Columbia. We used diagnostic genetic markers and the program NEWHYBRIDS to determine the genotypes of 2,670 fish at 45 sites between 1999 and 2006. A broad hybrid zone stretching over 200 km was observed with WCT backcross individuals as the dominant hybrid type present (6%). Population genetic analysis revealed that much of the hybridization is

iv relatively recent (2-3 generations) and likely results from stocking RBT in Koocanusa Reservoir. I assessed the ability of a variety of environmental and propagule pressure related variables to predict the levels of introgression using GIS and logistic regression analysis and found that a relative measure of propagule pressure was the best predictor of introgression. Long-term monitoring revealed that most low elevation sites were close to becoming complete hybrid swarms, and introgression at mid elevations < 80 km from the Koocanusa Reservoir was increasing as a result of hybrids straying from the low elevation sites. We recommend that the sources of RBT be more thoroughly identified, remaining pure WCT populations be confirmed, more inventory work be conducted to determine which streams are non-fish bearing, a ban on all fertile RBT stocking be implemented, and more studies be undertaken to evaluate the life history characteristics of the native WCT. (201 pages)

v DEDICATION I dedicate this dissertation to my daughters and wife. Nicola and Mackenzie grew into strong and confident adolescents while I toiled on this project, and I am proud of their many accomplishments since we moved to Utah. I hope that they will be able to find as much joy at work and play as I have. And to Doris, it is “inconceivable” to think of life without you.

vi ACKNOWLEDGMENTS I would like to thank Dr. Reese Halter and the Global Forest Foundation for supporting this project from the beginning and having the commitment to continue long-term support, especially when there was little interest from other agencies. This project has been funded in part by Global Forest since its inception in 1999 (GF-18-2002-137). I would also like to thank the USDA Forest Service, Fish and Aquatic Ecology Unit in Logan, Utah, and my major advisor, Dr. Jeff Kershner, for providing financial support and mentoring during my studies. It was with a healthy dose of naiveté that my family moved to Utah and there were many obstacles along the way that could have derailed my studies, but Jeff always found ways to make this project a success. I also want to thank Emily Rubidge, who graciously provided me with all the genetic data collected prior to 2002. Emily also endured many questions and requests for help as I tried to learn basic genetic analysis techniques and build a comprehensive database. And to Dr. Paul Wolf a special thank you for generously providing me with lab space, guidance, and support while I attempted to learn the black magic art of PCR! Dr. Steve Larson also graciously allowed me to use lab space in the USDA, Forage and Range Research Lab to analyze my data with GeneMapper. These projects invariably require the development of multiple skill sets and I would have been unable to do such without the dedicated help of many people. Brett Roper became my de facto advisor when Jeff moved to Bozeman, MT, and

vii I harassed Brett continually on all matters statistical – if I turned a corner in my ability to understand statistical issues, Brett is solely responsible. John Olson practically taught me how to use ARC GIS and Ryan Hill also helped me writing ARCINFO code and debugging my models. David Mayhood provided valuable information on the status of westslope cutthroat in Alberta and was always willing to discuss broader conservation issues. John Bell, Mike Hensler, Mickey MacDonald, and Bill Westover provided information on stocking records and valuable insight in management activities within the Upper Kootenay River. The Mirkwood crew of Peter Corbett and John Addison (Coach) were camping and work mates extraordinaire. Any crew that cooks a leg of lamb in the field is alright by me! Peter was also my constant sounding board for ideas and has been, since I joined Mirkwood Ecological in 1994. Coach produced all the maps for this study and was always a treat to work with. Finally, thanks to Doris, Nicola, and Mackenzie Bennett for putting up with a sometimes grumpy guy and partaking in this cross border adventure. Having this opportunity to go back to school and live and play in the Great Basin has been an experience I will always cherish. Stephen N. Bennett

viii CONTENTS Page ABSTRACT ..........................................................................................................iii ACKNOWLEDGMENTS .......................................................................................vi LIST OF TABLES ................................................................................................xii LIST OF FIGURES ..............................................................................................xv 1. RESEARCH GOALS AND BACKGROUND INFORMATION ........................... 1 Introduction.................................................................................................... 2 Westslope Cutthroat Trout Evolution and Distribution ................................... 4 Westslope Cutthroat Trout Status and Threats ............................................. 6 Study Area and Background.......................................................................... 9 Invasion Biology .......................................................................................... 11 Impact of Introduced Species .............................................................. 11 Geography of Invasion ........................................................................ 13 Hybridization................................................................................................ 19 Fitness of Hybrids................................................................................ 20 Hybrid Zone Models and Structure ...................................................... 21 Influences of Hybridization on Invasiveness ........................................ 22 Literature Cited............................................................................................ 24 2. ASSESSING THE GENETIC INTEGRITY OF A WESTSLOPE CUTTHROAT TROUT STRONGHOLD AT THE NORTHERN PERIPHERY OF ITS RANGE ......................................................................... 35 Abstract ....................................................................................................... 35 Introduction.................................................................................................. 36 Study Area................................................................................................... 41 Methods....................................................................................................... 42 Fish Capture and Tissue Collection..................................................... 44 DNA Extraction and Amplification........................................................ 45 Classification of Hybrids ...................................................................... 46 Distribution of Hybrids and Hybrid Zone Structure............................... 49

ix Results ........................................................................................................ 50 Classification of Hybrids ...................................................................... 50 Distribution of Genotypes and Hybrid Zone Structure ......................... 51 Introgression and Distribution of RBT Alleles ...................................... 52 Population Genetic Analysis ................................................................ 53 Discussion ................................................................................................... 54 Distribution and Abundance of Hybrids ............................................... 54 Genotypes and Hybrid Zone Structure ................................................ 55 Sample Bias ........................................................................................ 57 Hybrid Movement ................................................................................ 57 Level of Introgression and Age of the Hybrid Events........................... 59 Management Implications.................................................................... 61 Conclusion................................................................................................... 62 Literature Cited............................................................................................ 63 3. INFLUENCE OF PROPAGULE PRESSURE AND STREAM CHARACTERISTICS ON INTROGRESSION BETWEEN NATIVE WESTSLOPE CUTTHROAT TROUT AND INTRODUCED RAINBOW TROUT IN BRITISH COLUMBIA..................................................................... 82 Abstract ....................................................................................................... 82 Introduction.................................................................................................. 83 Methods....................................................................................................... 88 Sample Site Selection and Survey Design .......................................... 88 Fish Capture ........................................................................................ 89 DNA Analysis....................................................................................... 90 Quantifying Introgression..................................................................... 90 Power to Detect Hybridization ............................................................. 91 Analysis of Model Parameters ............................................................. 92 Model Development and Assessment ................................................. 99 Model Validation ................................................................................ 102 Results ...................................................................................................... 102 Distribution of RBT Alleles ................................................................. 102 Propagule Pressure........................................................................... 103 Model Results.................................................................................... 104 Model Validation ................................................................................ 105 Discussion ................................................................................................. 106

x Propagule Pressure........................................................................... 106 Environmental Factors....................................................................... 107 Model Limitations............................................................................... 109 Conclusion................................................................................................. 112 Literature Cited.......................................................................................... 113 4. LEVELS OF INTROGRESSION IN WESTSLOPE CUTHTROAT POPULATIONS EIGHT YEARS AFTER CHANGES TO RAINBOW TROUT STOCKING PROGRAMS IN SOUTHEASTERN BRITISH COLUMBIA.................................................................................................... 130 Abstract ..................................................................................................... 130 Introduction................................................................................................ 131 Study Area......................................................................................... 134 Background ....................................................................................... 135 Methods..................................................................................................... 137 Monitoring sites ................................................................................. 137 DNA Extraction and Amplification...................................................... 139 Classifying Hybrids ............................................................................ 139 Power to Detect Introgression ........................................................... 141 Data Analysis..................................................................................... 142 Influence of Discharge....................................................................... 143 Results ...................................................................................................... 144 Distribution of Hybrids and Genotypes .............................................. 144 Trends ............................................................................................... 145 Influence of Discharge....................................................................... 147 Discussion ................................................................................................. 147 Closed Sites ...................................................................................... 149 Open Sites......................................................................................... 151 Effect of Discharge ............................................................................ 155 Conclusion................................................................................................. 155 Literature Cited.......................................................................................... 155 5. CONCLUSION.............................................................................................. 174

xi APPENDIX ....................................................................................................... 177 VITA ................................................................................................................. 179

xii LIST OF TABLES Table

Page

1. Results of genetic analysis in 1986 and 1999 on selected streams throughout the Upper Kootenay River. Pure refers to the presence of pure WCT only, hybrids refers to the presence of WCT x RBT hybrids. .................................. 9 2.

Summary of sample site characteristics, distance from the Koocanusa Reservoir, and the number of fish captured from 1999-2006. Only fish that were successfully diagnosed at > 2 diagnostic nuclear loci are included in the sample sizes.......................................................................................... 71

3.

PCR laboratory protocols for multiplexed loci used to identify WCT, RBT, and hybrids in the Upper Kootenay River, BC. All loci are co-dominant and diagnostic for WCT and RBT (Ostberg and Rodriguez 2004). PCR mix used Qiagen Multiplex mix (commercial) with 1 unit Hot Start DNA Polymerase, 3 mM MgCl2, at pH 8.7. ................................................................................. 72

4.

Genotype frequency classes and expected proportions of loci with 0, 1, and 2 genes originating from WCT hybridizing with RBT after two generations (Anderson and Thompson 2002)................................................................. 73

5.

The pooled genotypes of all fish captured by site between 1999-2006 as determined by the NEWHYBRIDS model (Anderson and Thompson 2002). Closed sites are arranged in ascending order from lowest to highest elevation and open sites are arranged in ascending order from closest to furthest from the Koocanusa Reservoir. All samples were analyzed separately with NEWHYBRIDS and then results were pooled across sites that had multiple samples. Site = site number where the sample was collected (see Figure 2 for locations), N = total number of fish captured per site, Genotype = six genotype classes as determined by NEWHYBRIDS (see Table 4 for definitions), %RBT = the percent of RBT alleles by sample site, and Pwr = power to detect 1% RBT alleles at a sample site based on our sample size and the number of markers used....................................... 74

6.

Average inbreeding coefficient (FIS) values (a) and p values (b) for each site and sample that significantly deviated from Hardy-Weinberg expectations (Ave.) and for each of the seven loci examined separately (listed by loci). . 76

7.

P values for exact tests of linkage equilibrium for all pairwise combinations of seven nuclear loci using GDS (Lewis and Zaykin 2001) for samples collected between 2001 and 2006. Significant linkage disequilibrium for each pairwise comparison is assumed for p values < 0.0001 based on a Bonferroni

xiii adjusted p value (i.e., 0.05/462). Only sites with > 20 fish captured are included in the analysis. A tally of the number of pairwise tests that were significant at p = 0.05 are also included in recognition of fact that Bonferroni corrections maintain a low chance of making a Type I error that results in an increase chance of Type II errors (Moran 2003; Verhoeven et al. 2005). Asterisks indicate streams that deviated from Hardy-Weinberg expectations. ............................................................................................... 77 8.

Summary statistics for independent variables measured at each sample site (n=45). The variables were grouped into two categories: physical site characteristics and characteristics related to the potential propagule pressure. ................................................................................................... 121

9. Summary of model selection statistics for evaluating the level of introgression between WCT and RBT in the Upper Kootenay River, BC........................ 122 10. Model results for the best multinomial logistic regression model for predicting introgression between WCT and RBT in the Upper Kootenay River, BC. . 124 11. The sign of the coefficients of all variables that had significant correlations with the level of introgression between WCT and RBT (p < 0.005) and the predicted response of the variable. ........................................................... 124 12. Genotype frequency classes and expected proportions of loci with 0, 1, and 2 genes originating from WCT hybridizing with RBT after two generations (Anderson and Thompson 2002)............................................................... 163 13. Summary of number of presumed pure WCT and hybrids (i.e., fish with > 1 RBT allele) by sample site and year in the Upper Kootenay River, BC. Chisquare test statistics refer to test results over all years with in a sample site. Sites with < 1 observed WCT or hybrid were excluded from tests and Yates correction factor used for all tests with observed values < 5. Stream = stream name, Type = whether the stream can be accessed by fish from the Koocanusa Reservoir (open), or is inaccessible due to migration barriers (closed), Elev. = elevation of the site, Dist. = stream distance to the Koocanusa Reservoir from the sample site, No. WCT = number of WCT; No. Hyb = number of hybrids (i.e., fish with > 1 RBT allele), Prop WCT = proportion WCT, Prop Hyb = proportion hybrids, Pwr = the power to detect 1% RBT alleles, X2 = Chi-square statistic, df = degrees of freedom, P = p value for the X2 test. Test statistics are for tests between years within sample sites. All test statistics are listed on the first line of the site. Test between closed and open sites listed on the subtotal line for closed sites. Significant differences are denoted by lower case letters. .......................................... 164

xiv 14. Proportion of each genotype frequency class by sample site and year in the Upper Kootenay River, BC: 1999-2006. Sites arranged closed sites (above migration barriers) and open sites (below migration barriers). Closed sites arranged by elevation from lowest to highest and open sites arranged by distance to the Koocanusa Reservoir from nearest to furthest (see Table 13). Genotype classifications determined by NEWHYBRIDS model (Anderson and Thompson 2002). ............................................................................... 166

xv LIST OF FIGURES Figure

Page

1. Distribution of westslope cutthroat (dark shade) and coastal cutthroat (gray shade) (in Costello and Rubidge 2004, adapted from Behnke 2002)............. 5 2. Study site and genetic sample site locations collected in the Upper Kootenay River: 1999-2006. See table 2 for a list of the sample site names. Pie charts represent the percent of WCT (white) and RBT (black) alleles detected at each site. For sites that were sampled in more than one year the charts represent the average. Black bars represent hydro electric dams that are barriers to upstream fish migration. Gray x’s are fish migration barriers > 2 m tall and gray dots represent RBT stocking locations. ................................... 78 3. Examples of three genotype distributions identified at sites where hybridization between WCT and RBT has occurred: a) pure WCT, upper Bull River, b) bimodal distribution, lower Gold Creek, and c) unimodal left skewed WCT, Perry Creek. See Table 4 for descriptions of the different genotype categories and Table 5 for the pooled proportion of genotypes at all sites. Bimodal distribution (Gold Creek) is significantly different than unimodal left skewed distribution (Perry Creek, p < 0.000001) based on Fisher’s exact test using RxC (Miller 1997)................................................................................ 80 4. Spearman’s Rank Correlation of the distance to the Koocanusa Reservoir (black triangles) and elevation (gray squares) versus the pooled average percent of RBT alleles for “open” sites only (n = 27, elevation R2 = 0.39, p = 0.0003, distance R2 = 0.49, p < 0.0001)....................................................... 81 5. Study area and location of hybridization monitoring sites (n = 45) in the Upper Kootenay River, BC. Sample sites are 1) lower Alexander Ck 2) Upper Alexander Ck, 3) lower Alki Ck, 4) lower Bloom Ck, 5) lower Bull R, 6) upper Bull R, 7) lower Caven Ck, 8) lower Coal Ck, 9) lower Wild Horse R, 10) lower East White R, 11) lower Elk R, 12) mid Elk R, 13) upper Elk R, 14) mid Findlay Ck, 15) upper Fording R, 16) lower Forsyth Ck, 17) lower Gold Ck, 18) upper Gold Ck, 19) lower Grave Ck, 20) upper Kootenay R, 21) mid Wigwam R, 22) lower Lodgepole Ck, 23) upper Lodgepole Ck, 24) lower Lussier R, 25) upper Lussier R, 26) mid Mather Ck, 27) lower Meachen Ck, 28) mid Meachen Ck, 29) lower Michel Ck, 30) mid Michel Ck, 31) upper Michel Ck, 32) lower Morrissey Ck, 33) mid North White R, 34) upper North White R, 35) Wheeler Ck, 36) lower Perry Ck, 37) lower Sand Ck, 38) mid Sand Ck, 39) lower Simpson R, 40) lower Skookumchuck Ck, 41) mid Skookumchuck Ck, 42) lower St. Mary R, 43) upper St. Mary R, 44) upper Summit Ck, 45) lower Tepee Ck. Pie charts = % WCT alleles (clear) and %

xvi RBT alleles (black), gray dots = stocking sites with outlets, x = fish migration barrier, and black bars = hydro electric dams (also migration barriers)...... 125 6. The percent of fish movement from a stocking site based on the assumption that 61% of all fish will not stray any further than 10 km. The % straying is a function of a constant intercept and decay rate and the distance from the ( −0.05*dis tan ce ( km )) source as: % Straying = 100 * e ................................................. 127 7. The frequency distribution of the % RBT alleles by site in the Upper Kootenay River watershed: 1999-2006 (n=45)........................................................... 127 8. The number of stocking sites and RBT stocked in lakes with outlets by elevation within the Upper Kootenay River, BC from 1915-2006. .............. 128 9. The predicted probability of the level of introgression based on the GIS derived propagule pressure measure (GISPP). The zero line is the predicted probability of a site having zero introgression, the low line is the predicted probability of a site having < 10% introgression, and the high line is the predicted probability of a site having > 10% introgression at the different levels of propagule pressure (i.e., potential number of RBT). .................... 129 10. Study area and location of hybridization monitoring sites in the Upper Kootenay River, BC. Site numbers are 1) Michel Creek, 2) Grave Creek, 3) Alexander Creek, 4) Gold Creek, 5) Skookumchuck Creek, 6) Perry Creek, 7) Bloom Creek, 8) Tepee Creek, 9) Kootenay River, 10) Simpson River, 11) Elk River, 12) Bull River, and 13) St. Mary River. Small gray dots = RBT stocking sites, gray x’s = fish migration barriers > 2 m, and black bars = hydroelectric dams that are also fish migration barriers. ............................ 168 11. Proportion of genotypes based on a NEWHYBIRDS model classification (Anderson and Thompson 2002) for closed (black bars) vs. open (gray bars) sites in the Upper Kootenay River: 1999-2006. Genotypes are listed as a continuum from genotypes with the most WCT alleles (WCT_bx) on the right to genotypes with the least WCT alleles on the right (RBT). See Table 12 for definitions of the genotype classes. ........................................................... 170 12. The predicted and observed proportion of hybrids at Perry Creek. Predicted observations are based on a logit model developed from the observed proportion of hybrids sampled from 1997-2006. The proportion of hybrids in Perry Creek is expected to increase by 19.5% from 2006 to 2007 (N = 460, Odds Ratio = 1.195, 95% Wald CL = 1.088-1.314, p < 0.0002). ................ 171 13. The change in the genotype frequencies among years for the three longterm monitoring sites: a) Perry Creek, b) Gold Creek, and c) Michel Creek. Genotype classes were combined for presentation purposes, clear bars =

xvii WCT_bx, black bars = F1/F2, and Gray bars = RBT_types. Pure WCT types are not included in the graphs. ................................................................... 172 14. The change in the genotype frequencies between years for two of the revisit sites: a) Skookumchuck Creek and b) Alexander Creek. Genotype classes were combined for presentation purposes, clear bars = WCT_bx, black bars = F1/F2, and Gray bars = RBT_types. Pure WCT types are not included in the graphs. ................................................................................................. 173

CHAPTER 1 RESEARCH GOALS AND BACKGROUND INFORMATION The goal of this research is to better understand current and future threats to the genetic integrity of native westslope cutthroat trout (WCT, Oncorhynchus lewisi) posed by hybridization with introduced rainbow trout (RBT, O. mykiss) in the Upper Kootenay River, British Columbia (BC). Hybridization refers to the mating between individuals from two genetically distinct entities that may or may not produce viable offspring (Taylor 2004). Introgressive hybridization (introgression) is defined as hybridization that produces viable offspring and results in the movement of alleles from one genetically distinct entity to another. This research focuses on determining landscape level distributions of WCT, RBT, and their hybrids using genetic analysis of fish tissue samples collected throughout the watershed and GIS modeling. A model is developed and tested to determine if readily available abiotic variables can be used to predict the level of introgression. This chapter reviews the distribution and status of WCT, the management history of the study area, the theory of invasion biology, and the causes and consequences of hybridization. The remaining research components of this dissertation are presented in four chapters that will be submitted for individual publication and a concluding chapter that summarizes all of my findings.

2 Introduction Extinctions of species caused by human are happening at an alarming rate and evolutionary processes that create diversity are being disrupted (Myers and Knoll 2001). Freshwater species are disproportionately more threatened than terrestrial species in much of the United States of America (USA) (Richter et al. 1997; Loftus and Flather 2000; Warren et al. 2000). For example, fewer than 20% of terrestrial vertebrates in the USA are at risk of extinction, whereas almost 40% of amphibians and fish, and 67% of unionid mussels are at risk of extinction (Richter et al. 1997). Habitat destruction and competition from introduced species are consistently recognized as the two leading causes of species extinctions in both terrestrial and aquatic ecosystems (Miller et al. 1989; Warren and Burr 1994; Richter et al. 1997; Wilcove et al. 1998). The study of “invasion biology” has increased rapidly in response to the impacts of species introductions worldwide (Carey et al. 1996; Vermeij 1996; Reichard and White 2003). Many species introductions are a direct result of dramatic growth in transoceanic and transcontinental travel and trade in the last century (Elton 1958). It is estimated that 50,000 species have been introduced intentionally or unintentionally to the USA (Pimentel et al. 2000), and over 4,000 introduced vascular plants are now naturalized in the USA accounting for about 20% of all vascular plants (Davis 2003). Hawaii has over 4,600 introduced plant species, which is three times the estimated number of native species (Soule 1990), and in the western USA and Canada many watersheds now have more

3 introduced aquatic species than native species (Crossman 1991; Wydoski and Whitney 2003). Salmonid fishes have been introduced more often than almost any group of fishes because of their popularity as sport fish and the ease with which they can be reared and transported. In particular, the RBT, which were native to western North America and eastern Siberia (Behnke 1992), have been introduced more than any other fish species, and are now found on all continents except Antarctica (Welcomme 1992). Salmonid species are prone to hybridization because of behaviors that fail to fully isolate species during reproduction and competition for limited spawning habitat (Behnke 1992; Allendorf and Waples 1996). Therefore, it is not unexpected that hybridization between native and introduced salmonids is now common because of the widespread stocking of salmonids outside their native range. In the western USA and Canada, rainbow trout (RBT, Oncorhynchus mykiss) readily hybridize with native inland cutthroat trout (O. clarki subsp.) and as a partial result, two sub-species of inland cutthroat trout are now extinct, and the remaining sub-species have all suffered range contractions (Young 1995; Duff 1996; Behnke 2002). Introgressive hybridization is now considered the largest threat to the persistence of most sub-species of inland cutthroat trout (Allendorf and Leary 1988; Leary et al. 1995; Behnke 2002).

4 Westslope Cutthroat Trout Evolution and Distribution

The ancestors of the salmonid family are believed to have evolved about 100 million years ago (MYA) with the doubling of their chromosome number (tetraploidization) that separated them from most other fishes (Behnke 2002). Cutthroat and RBT are closely related salmonid species that likely diverged from each other between 3.5 and 8.5 MYA (McKay et al. 1996). Both cutthroat and RBT species have a high degree of within species variation as noted by the numerous recognized sub-species of each species (Behnke 2002). Deciphering the taxonomy of these two species has been complicated because of the extreme genetic, morphological, life history, and ecological variations within the species (Allendorf and Waples 1996). Furthermore, phylogenetic reconstruction has been complicated by one or more apparent hybridization event(s) (reticulate evolution) after the species initially diverged (Allendorf and Waples 1996; Behnke 2002; Brown et al. 2004). The westslope cutthroat trout (WCT) is one of 13 sub-species of interior cutthroat trout native to western North America (Behnke 1992). WCT have the most geographically widespread range of all interior sub-species of cutthroat trout and are found in Alberta, BC, Washington, Idaho, Montana, Oregon, and Wyoming encompassing most of the upper Missouri Basin and headwaters of the Columbia River (Figure 1, Behnke 1992). The distribution of WCT appears to have been determined approximately 70,000 years ago by the formation of barrier falls on many of the large tributaries to the Columbia River including the

5 Kootenay, Clark Fork, Pend ‘Oreille, and Spokane Rivers. It is hypothesized that WCT were able to colonize above barrier falls because water levels were higher during the glacial retreat, or possibly because barriers formed following glacial retreat as the land mass rebounded (Behnke 1992). It is also speculated that WCT were isolated above these barrier falls and survived in Montana, Idaho, and Washington during the most recent glacial period. RBT appear to have been restricted to the lower Columbia River during this period allowing WCT to colonize large inland portions of North America in isolation from RBT (Behnke 1992).

Figure 1. Distribution of westslope cutthroat (dark shade) and coastal cutthroat (gray shade) (in Costello and Rubidge 2004, adapted from Behnke 2002).

6 WCT evolved with relatively few other fish species over most of their range and may have been shaped more by environmental conditions than by interspecific interactions (Griffith 1988). In the Upper Kootenay River, WCT evolved primarily with bull trout (Salvelinus confluentus) and Rocky Mountain whitefish (Prosopium williamsoni) (McPhail and Carveth 1992). WCT are generalist predators likely due to the cold and often turbid waters of the Upper Kootenay River and its tributaries, and the lack of other trout species (Griffith 1988, McPhail & Carveth 1992). Where WCT have evolved with RBT the two species are segregated along elevational gradients, with WCT occupying higher elevation sites (Hanson 1977; Platts 1979). Paul and Post (2001) found that RBT tend to redistribute to lower elevation sites when they are initially introduced into headwater lakes above populations of WCT. When RBT are introduced to areas with allopatric WCT populations, they often hybridize, indicating that WCT lack innate isolating mechanisms that allow them to co-exist with RBT (Behnke 1992). Westslope Cutthroat Trout Status and Threats There have been numerous recent efforts to determine the status of WCT, especially in the USA (Liknes and Graham 1988; McIntyre and Rieman 1995; Young 1995; Van Eimeren 1996; Shepard 1997; Shepard 2003). There is a general consensus that the distribution and abundance of WCT has decreased from historic levels over much of their range due mainly to habitat destruction, over fishing, and competition and hybridization with non-native salmonids. WCT are particularly susceptible to these stressors because they are often found in

7 small headwater streams, are stream resident, easily angled, intolerant of high water temperatures, repeat spawners, and depend on riparian and instream cover, natural flow and stream hydrological features, and well-oxygenated gravel substrate for spawning (Haas 1998). Despite these current threats to WCT, genetic introgression with RBT is thought to be the biggest threat to the future persistence of WCT throughout their range (Allendorf and Leary 1988; Mayhood 2000; Rubidge et al. 2001). In this context, complete introgression is defined as the amount of hybridization that would have to occur before the native genotype no longer exists (Leary et al. 1995). Once complete introgression has occurred the population is often referred to as a “hybrid swarm.” Detailed status reviews of WCT have been hampered by a lack of current population and genetic data, and there is considerable uncertainty about the status of numerous populations. Overall in the USA it is estimated that there has been a 40% reduction in the range of WCT, and between 65 and 87% of the occupied habitat does not contain genetically pure WCT (Shepard et al. 2005). In local areas WCT populations have been severely depressed, such as in parts of Montana where WCT are thought to occupy less than 28% of their historic range and less than 3% of their current range is occupied by genetically pure populations (Liknes and Graham 1988). In the Missouri River Basin 90% of the 144 populations known to have at least 90% genetic purity are at “high” to “very high” risk of becoming extinct (Shepard 1997).

8 The situation is similar in parts of the Canadian range (Mayhood 2000; Costello and Rubidge 2004; Costello 2006). The number of genetically pure populations of WCT has declined in Alberta by as much as 95%, and most of the remaining populations exist in isolated headwater tributaries (Mayhood 2000). Non-native salmonids such as RBT, brown trout (Salmo trutta), eastern brook trout (Salvelinus fontinalis), Yellowstone cutthroat trout (O. c. bouvieri), and golden trout (O. mykiss aguabonita) have been stocked throughout Alberta and now dominate many historic WCT streams (Mayhood 2000). In BC, the distribution of WCT is similar to its historic distribution, but the status of many populations are unknown (Costello and Rubidge 2004). The most recent genetic surveys in BC suggest that the genetic integrity of WCT populations is declining (Rubidge et al. 2001; Rubidge and Taylor 2004). A genetic assessment of the Upper Kootenay River in 1986 found one tributary stream out of the seven surveyed had evidence of WCT x RBT hybrids (Leary et al. 1987). A repeat sampling of the same streams in 1999 found four streams with hybrids present (Table 1, Rubidge et al. 2001). Further sampling in 2000 found hybrids present in other streams throughout the watershed with some populations having hybrids in excess of 35% of the samples collected (Rubidge 2003). The results of follow-up studies (1999 and 2000) and the original study both had about a 90% chance of detecting the presence of 1% RBT alleles and thus represent a conservative estimate of the level of hybridization (Rubidge 2003). Hybrids were found across all four life history stages with the highest

9 percentage found in the younger age groups suggesting that the number of hybrids was increasing (Rubidge et al. 2001).

Table 1. Results of genetic analysis in 1986 and 1999 on selected streams throughout the Upper Kootenay River. Pure refers to the presence of pure WCT only, hybrids refers to the presence of WCT x RBT hybrids. Genetic Survey Results Sample Location Upper Kootenay River North Fork White River Skookumchuck River Bull River Upper Elk River Lower Elk River Wigwam River Gold Creek Upper St Mary River Lower St Mary River 1 2

19861 Not sampled Hybrids Pure Pure Pure Not sampled Pure Pure Pure Not sampled

19992 Hybrids Hybrids Hybrids Pure Pure Hybrids Hybrids Hybrids Pure Hybrids

Leary et al. 1987 Rubidge et al. 2001 Study Area and Background The study area encompasses the Canadian portion of the Kootenay River

from its headwaters in Kootenay National Park down to the Canada/United States (U.S.) border near Newgate, BC. The Kootenay River is a large order 7th tributary (Strahler 1957) to the Columbia River in southeastern BC with a mean annual discharge of 295.6 m3. This portion of the Kootenay River is in the East Kootenay Region of BC and is bounded by the Rocky Mountains to the east and

10 the Purcell Mountains to the west. The study area is approximately 250 km long and the drainage area is approximately 18,500 km2. Like most watersheds throughout western North America, non-native fish have been stocked extensively throughout the Kootenay drainage, particularly in low elevation lakes (MWLAP 2006). The first records of RBT stocking in the Upper Kootenay River begin in 1915 and almost 20 million RBT have been stocked since then at 114 sites in over 2,500 individual stocking events (MWLAP 2006). The Libby dam was completed in 1972 on the Kootenay River at Libby, Montana, at a bedrock chute which probably isolated WCT above the falls from RBT in the Lower Kootenay River (Behnke 1992). The dam created a 170 km long reservoir that spans the Canada/USA border (Whatley 1972). For several years both the USA and Canada attempted to establish WCT in the reservoir with little success (B. Westover, Ministry of Water, Land, and Air Protection, personal communication). Between 1986 and 1998 RBT were stocked in the Koocanusa Reservoir on both the Canada and U.S. side of the border (MFWP 2001; MWLAP 2006). An average of 5000 yearling and adult RBT/year were stocked in Kikomun Creek, a tributary to the Koocanusa Reservoir on the Canadian side, and ~ 41,900 RBT/year were stocked directly into Koocanusa Reservoir on the U.S. side during this time. Stocking RBT in lakes and reservoirs with potential outlets in the Upper Kootenay River was stopped in 1999 and replaced with WCT stocking or stocking of triploid (presumed sterile) RBT in all but Premier Lake (M. MacDonald, Go Fish BC, personal communication).

11 Invasion Biology Impact of Introduced Species Kohler & Courtenay (2003) describe five negative impacts of introduced species: habitat alteration, introduction of diseases, trophic alterations (predation), spatial alterations (competition), and gene pool deterioration (hybridization). Grass carp (Ctenopharyngoden idella) are often cited as an example of an introduced species that caused habitat alteration (Bain 1993). Carp disturb large amount of sediments while foraging on aquatic vegetation and increase water turbidity, which in turn reduces aquatic photosynthesis. Introduced zebra mussels (Dreissena polymorpha) have had the opposite effect in the Great Lakes where they have become established in such large numbers that their filtering ability has increased water clarity beyond natural levels and reduced phytoplankton availability for other native species (Johnson and Padilla 1996). Introduced species have also been the vectors for the spread of disease which they themselves are relatively immune. For example, introduced brown trout (Salmo trutta) are thought to have been the hosts that spread whirling disease in North America, which is now threatening many native trout species (Markiw 1989; Allendorf et al. 2001b). Predation by introduced species has caused severe reductions and extinctions of native species as was the case in Lake Victoria, Africa, where the Nile perch (Lates niloticus) was introduced, causing serious declines in many of the native cichlid species (Kaufman 1992). However, with the recent over fishing

12 of Nile perch, many cichlid species are recovering and some species that were thought to be extinct are now being rediscovered (Balirwa et al. 2003). Predation by introduced brown trout has also negatively impacted native species including many galaxids in New Zealand (Crowl et al. 1992; Townsend 1996, 2003). Intraspecific competition for suitable feeding and hiding cover is intense in salmonids, and larger fish often exclude smaller fish from optimal habitat (Chapman 1966). Interspecific competition between introduced and native salmonids can also be intense. Introduced eastern brook trout have been shown to decrease numbers of bull trout (Salvelinus confluentus) and negatively affect their foraging rates and use of cover (Nakano et al. 1998). Cutthroat trout are generally less aggressive than other trout and char species (Griffith 1988). In experimental stream reaches cutthroat fry were unable to displace RBT fry, but the RBT were able to displace the cutthroat (Hanson 1977). RBT have also been able to displace native eastern brook trout (Moore and Ridley 1984; Larson et al. 1995) and eastern brook trout have displaced inland cutthroat (Gowan and Fausch 1996; Peterson and Fausch 2003). Cutthroat are thought to be particularly susceptible to predation and competition because they are slow to mature, emerge from the gravel later than other salmonids, are less aggressive, lack morphological specialization, and are often smaller than other salmonids (Griffith 1988).

13 Geography of Invasion The initial consequences and potential impacts of introduced species are fairly well understood, but the mechanisms of invasion are not (Krueger and May 1991). In most cases the introduction of a species to a new environment is rarely documented from the initial release to the subsequent establishment, and few controlled experiments have been conducted to determine factors that affect a species success in invading a new area (Kolar and Lodge 2001; Peterson 2002). Recent efforts at developing invasion biology theory have relied primarily on describing the attributes of invasive species and invaded communities (Peterson 2002). There are typically three stages of species invasions that have been described (Moller 1996; Vermeij 1996; Kolar and Lodge 2001). First the species is transported to a new habitat (naturally or human aided), then the species establishes or integrates itself in the new habitat, and finally the species may become well established and possibly spread to new areas. Predicting which species are likely to complete these stages has proved very difficult (Kolar and Lodge 2001). Invasions have been documented in practically all ecosystem types by every level of organism (Franklin et al. 1971; Butler 1986; Carey 1996; Carlton 1996b; Johnson and Padilla 1996; Dark et al. 1998; Allendorf and Lundquist 2003; Grigorovich et al. 2003; Bohlen et al. 2004). However, despite enormous numbers of introductions throughout the world, there are not always dire consequences (at least in the short-term), and many introduced species fail to develop self sustaining populations (Elton 1958). Williamson (1996) outlines the

14 evidence for the “”tens rule” which suggests that approximately 10% of species introductions are successful and only 10% of the successful introductions become “pests.” There is some evidence that island habitats are more susceptible to invasion than the mainland for a combination of the following reasons: island ecosystems are newer and have experienced more extinctions in the past, are more simple systems (less species), and are invaded by species that are dramatically different than native species (Vermeij 1996). Marine environments have been particularly susceptible to invasions by species introduced via ballast discharges from commercial tanker ship traffic (Carlton 1996b; Grigorovich et al. 2003; Niimi and Reid 2003), although introductions as a result of the pet trade are also a growing concern (Semmens et al. 2004). Some invasion biologists believe that the attributes of the native community are key to understanding its invasion potential, whereas others suggest that the characteristics of the invader will ultimately decide the success of an invasion (Simberloff and Boecklen 1991). Carlton (1996b) noted several invasion scenarios that appear to be common to successful invasions which include changes in the donor/recipient habitat that can create an invasion “window,” the creation of new donor regions, stochastic inoculation events, and/or the dispersal vector changes (altering the inoculation rate). In the background of all the invasion biology research is also the basic problem of a lack of accurate species distribution data including historic occurrence records. Cryptogenic species (i.e., those that are neither clearly native or non-native) may

15 be more common than currently recognized which could further complicate the study of invasive species (Carlton 1996a). As with many emerging ecological theories, there is no single theory that adequately describes the attributes of all invasive species or communities susceptible to invasion. However, Moyle and Light (1996a) used an extensive review of invasion case histories and community assembly theory to develop 12 empirically derived rules of aquatic invasions as follows: 1) most introduced species fail to become established, 2) most successful invaders are integrated without major negative effects, 3) all aquatic systems are invasible (invasibility is NOT related to diversity), 4) major community effects of invasions are most often observed where the number of species is low, 5) less disturbed systems are more likely to be invaded by piscivores, omnivores, and detritivores, 6) piscivores are more likely to disrupt systems (cascading effects), 7) in moderately disturbed systems any species with the right physiological and morphological conditions can become established, 8) persistence of the invader will depend on this match being maintained, 9) invaders are more likely to be successful when natives have been perturbed, 10) long-term integration is more likely in permanently disturbed systems, 11) invasibility of natural aquatic systems is related to interactions between environmental variability, predictability, and severity, and 12) invaders are most likely to extirpate natives in systems with extremely high or low variability. Many of these assertions have also been made for terrestrial invasions (Elton 1958; Moller 1996; Chapman et al. 2004; Rose and Hermanutz 2004).

16 Salmonid invasions in the western USA, BC, and Alberta typically have resulted from extensive stocking of non-native RBT, eastern brook trout, brown trout, and non-native cutthroat sub-species throughout the 1900’s. In many cases these non-natives have established self-sustaining naturalized populations (Crossman 1991; Behnke 1992), and have been implicated in native cutthroat declines (Young 1995; Duff 1996; Shepard 2003). Many salmonid streams are thought to be very susceptible to invasion by non-natives because they are often invaded by other salmonids with similar niche requirements, they have low primary productivity, and many have suffered habitat degradation (which may reduce native populations) (Moyle 1986; Krueger and May 1991). Invading fish species have been shown to spread both upstream and downstream from introduction sites (Adams et al. 2001, Hitt 2002). RBT stocked in the Koocanusa Reservoir are suspected of migrating upstream and hybridizing with native WCT throughout the Kootenay River Watershed (Rubidge et al. 2001). Hitt (2002) noted the same phenomenon in the Flathead River Watershed, where RBT spread upstream from Flathead Lake and hybridized with native WCT. Upstream invasion is limited by physical barriers such as waterfalls and gradient barriers whereas downstream invasion is not. Downstream invasions have been documented where formerly fishless high elevation lakes are stocked and fish escaped downstream (Adams et al. 2001). Downstream invasions of this nature pose a disproportionately large threat to native fish because the invading species is not restricted by physical barriers. Fish have been documented moving downstream over gradients > 80% and waterfalls of 18 m and invading

17 areas as far as 22 km downstream (Adams et al. 2001). Fish moving downstream in an exploratory nature also become de facto dispersers because they cannot return upstream to the lake if barriers are present. Behavioral interactions in salmonids are often size dependant, with larger fish dominating smaller fish (Chapman 1966). Non-native fish that invade from headwater lakes can have a competitive advantage over native stream dwelling fish because fish typically have faster growth rates in lakes. Recent criticism of invasion biology research, especially as it relates to salmonid invasions, has pointed out that very few studies have focused on determining the population level mechanisms that are important for invasions to be successful (Peterson and Fausch 2003). Peterson and Fausch (2003) also point out that for a native species to suffer declines during the invasion of a nonnative species, one or more of the following biotic interactions must occur that negatively impacts the native species: a declining reproductive rate, declining survival rate, net emigration, or introduction of disease by the invader. Current knowledge of how invading species out-compete native species is lacking because studies have not focused on identifying the specific mechanisms and age classes where these effects are happening. However, these factors may be much less important in cases where invading species can hybridize with the native species. In cases where the invader and native species can hybridize, native salmonid populations can become introgressed with non-native populations simply by behavior that allows the interbreeding of the two groups. For example, salmonids are broadcast spawners, and as such there is

18 opportunity for smaller (“less fit”) males to “sneak” in and fertilize larger females’ eggs during spawning. This has been documented for several trout and char populations (e.g. Kitano et al. 1994; Baxter et al. 1997; Ostberg et al. 2004). Different species of salmonids also tend to congregate in specific habitats during spawning. It is therefore possible for non-native individuals to breed with native fish, even when they normally occupy different habitats during the rest of the year. For example, cutthroat trout that live in upper elevations may spawn in downstream areas where non-native RBT reside. All individuals in the cutthroat population will contain some rainbow trout alleles (i.e., become introgressed) if a certain level of interbreeding occurs. Therefore, factors that influence the likelihood of two species hybridizing will likely control the rate of these types of invasions. The following factors have been identified as potential controls on the rate of hybridization: presence/absence of migration barriers (Thompson and Rahel 1998; Hilderbrand and Kershner 2000; Adams et al. 2001; Novinger and Rahel 2003), flow regime (Strange et al. 1992; Fausch et al. 2001), elevation, gradient, stream width, water temperature (Larson and Moore 1985; Fausch 1989; Paul and Post 2001; Weigel et al. 2003), distance to source (i.e., non-native stocking locations) (Rubidge et al. 2001, Hitt 2002), propagule pressure (Kolar and Lodge 2001; Marchetti et al. 2004; Colautti 2005), and amount of disturbed habitat (Moyle and Light 1996a, 1996b). These factors can be grouped into two broad categories: abiotic (migration barriers, flood disturbance regime, elevation, gradient, stream width,

19 temperature) and management factors (stocking location, propagule pressure, habitat disturbance). The abiotic factors are factors that have the potential to limit the ability of non-native fish from migrating to or surviving in particular habitats. The management factors are factors that relate to the amount and source of the non-natives (i.e., how many have been stocked and where) and to the condition of the habitat (has it been altered from past conditions). Clearly the presence of migration barriers is a primary factor determining both the historic distribution of fish species in stream systems (Behnke 1992, McPhail & Carveth 1992) and to a lesser degree the extent of non-native fish species (Kruse et al. 1997; Novinger and Rahel 2003). The presence/absence of a source of non-native species (stocking location) is also crucial in understanding the potential distribution of non-natives. Beyond these two factors, the other highlighted factors likely all act in various ways to regulate the intensity and potential success of non-native species ability to hybridize with native species. Hybridization Hybridization is recognized as a natural phenomenon that can play an important role in plant and animal evolution by creating recombinant species and evolutionary novelties (Jiggins and Mallet 2000; Allendorf et al. 2001a; Barton 2001). For example, the origin of tetraploidy in salmonids is suspected to have resulted from a hybrid event (Allendorf and Waples 1996), and there is evidence of past hybridization events between bull trout and dolly varden (Salvelinus malma) (Baxter et al. 1997), and cutthroat and RBT (Leary et al. 1987; Brown et

20 al. 2004). Even the entire vertebrate taxa may have had polyploid hybrids as an important part of their evolution (Lynch and Conery 2000). There are numerous examples of how environmental and ecological factors influence the success or failure of hybrids between closely related species (Grant and Grant 1993; Schluter 1993). Fitness of Hybrids Despite the potential importance of hybrid events in speciation, there appears to be a general consensus that hybrids are more often less fit than their parental species (Arnold and Hodges 1995). Theoretically, mating between two different species should result in no offspring being produced, offspring being produced with significantly reduced fitness due to genetic incompatibilities (Hawkins and Foote 1998), or outbreeding depression (Leary et al. 1985; Ledig 1986; Krueger and May 1991; Allendorf and Waples 1996; Utter 2003). Indeed, there is a considerable amount of empirical genetic evidence that supports these predictions. Hybrids between horses and donkeys are sterile (mules), many fish hybrids are sterile (Hubbs 1955; Kanda et al. 2002a), some flower hybrids produce fewer seeds of the parental species (Arnold and Hodges 1995), and there can be decreased developmental stability in rainbow trout (RBT) and cutthroat trout hybrids (Leary et al. 1985). However, despite the large amount of empirical evidence of reduced hybrid fitness, there are examples of hybridization between salmonid species resulting in introgression and the production of hybrid swarms (Allendorf and Waples 1996; Rubidge et al. 2001; Docker et al. 2003;

21 Hitt et al. 2003; Weigel et al. 2003). This implies that some hybrids are fit enough to survive and breed. Arnold and Hodges (1995) conclude that hybrids are not uniformly unfit, but rather range in fitness depending on environmental conditions present. Hybrid Zone Models and Structure Taylor (2004) reviewed the literature on hybrid zone models and found that there were four common models of hybrid zone structure that were particularly relevant to salmonid species: tension zone, bounded hybrid superiority zone, mosaic hybrid, and evolutionary novelty zone. The tension zone model is an environment-independent model that assumes that there is selection against hybrids and that hybrid zones are maintained by dispersal of parental offspring into the hybrid zone. The bounded hybrid superiority and mosaic zone models are both environment-dependent models that predict that hybrids have intermediate fitness and will have superior fitness in intermediate habitats. The evolutionary novelty model is a mixture of environment-independent and dependent models that allows for hybrids to demonstrate higher and lower fitness than parentals, and ultimately form new species. In instances where hybridization and introgression are common (i.e., salmonid populations) the tension zone model is not appropriate, because a fundamental assumption of the tension zone model is that hybrids are less fit. There is evidence that many salmonid hybrid zones conform to the environment-dependent zone models, especially the mosaic and evolutionary novelty models (Taylor 2004).

22 Many allopatric cutthroat populations have been threatened by introgression when non-native rainbow trout have been introduced (Young 1995; Duff 1996). There is also evidence that natural hybrid zones formed between sympatric populations of cutthroat and rainbow trout can be altered when nonnative rainbow trout are introduced, with the result that the hybrid zone increases (Docker et al. 2003). This may be a result of what is described as “propagule pressure” which has emerged as an important factor in invasion biology (Kolar and Lodge 2001). Changes in the hybrid zone structure can lead to outcomes that range from maintenance of both parental species to complete introgression and loss of locally adapted gene complexes in the native species (Leary et al. 1995). Influences of Hybridization on Invasiveness Allendorf and Lundquist (2003) asked why invading species out-compete native species that are theoretically better suited to local environmental conditions. They suggest that there are five possible reasons why invaders outcompete native species: 1) they are better competitors, 2) their enemies are absent, 3) the native species’ superior local adaptations are only expressed under extreme environmental conditions, 4) human impacts have reduced the abundance of natives prior to the arrival of the invader, and 5) the invader displays phenotypic plasticity. Of these possible advantages that RBT and their hybrids may have, it seems that the most important in my project are 1, 3, and 5.

23 RBT have been shown to be more aggressive than cutthroat and they can kill or displace them from optimal habitat in laboratory and natural settings (Griffith 1988). Therefore, if cutthroat and rainbow trout are producing viable hybrids, then it is possible that the hybrids may be able to out-compete pure cutthroat for feeding and spawning habitat. Also, cutthroat may have local adaptations that only play a significant role under extreme environmental conditions. They may be able to out-compete rainbow in years with extremely cold temperatures, when there are successive high runoff years, or at higher elevations where stream conditions are more extreme. The recent “warming trend” where snow packs and runoff discharges have been below historic averages may be allowing RBT and their hybrids to out-compete cutthroat. Finally, because many salmonids have a high degree of plasticity in life history characteristics, RBT and their hybrids may be able to compete with native cutthroat by shifting their life histories to adapt to conditions that are similar to those faced by native cutthroat’s (i.e., timing of spawning, migrating strategies, etc.). In summary, hybrids of numerous taxa have been shown to have varying levels of fitness which is ultimately controlled by genomic compatibility. However, when two species have compatible gametes it appears that environmental factors will play a large role in the relative fitness of hybrids. The increase in anthropogenic induced hybridization by transplanting species outside their home range, the human impacts on habitat (dams, water diversion, fragmentation, etc.), and environmental conditions (global warming, contamination, etc.) all

24 contribute to the potential that hybrids may out-compete parentals, leading to continued loss of native gene complexes. Literature Cited Adams, S., C. A. Frissell, and B. E. Rieman. 2001. Geography of invasion in mountain streams: consequences of headwater lake fish introductions. Ecosystems 2001:296-307. Allendorf, F. W., and R. F. Leary. 1988. Conservation and distribution of genetic variation in polytypic species, the cutthroat trout. Conservation Biology 2(2):170-184. Allendorf, F. W., R. F. Leary, P. Spruell, and J. K. Wenburg. 2001a. The problems with hybrids: setting conservation guidelines. Trends in Ecology and Evolution 16(11):613-622. Allendorf, F. W., and L. L. Lundquist. 2003. Introduction: population biology, evolution, and control of invasive species. Conservation Biology 17(1):2430. Allendorf, F. W., P. Spruell, and F. M. Utter. 2001b. Whirling disease and wild trout: Darwinian fisheries management. Fisheries 26(5):27-29. Allendorf, F. W., and R. S. Waples. 1996. Conservation and genetics of salmonid fishes. In J.C. Avise and J.L Hamrick (eds). Conservation genetics: case histories from nature. Chapman and Hall, New York, NY:238-280. Arnold, M. L., and S. A. Hodges. 1995. Are natural hybrids fit or unfit relative to their parents? Trends in Ecology & Evolution 10:67-71. Bain, M. B. 1993. Assessing Impacts of Introduced Aquatic Species - Grass Carp in Large Systems. Environmental Management 17(2):211-224. Balirwa, J. S., C. A. Chapman, L. J. Chapman, I. G. Cowx, K. Geheb, L. Kaufman, R. H. Lowe-McConnell, O. Seehausen, J. H. Wanink, R. L. Welcomme, and F. Witte. 2003. Biodiversity and fishery sustainability in the Lake Victoria Basin: An unexpected marriage? Bioscience 53(8):703715. Barton, N. H. 2001. The role of hybridization in evolution. Molecular Ecology 10(3):551-568.

25 Baxter, J. S., E. B. Taylor, R. H. Devlin, J. Hagen, and J. D. McPhail. 1997. Evidence for natural hybridization between Dolly Varden ( Salvelinus malma) and bull trout ( Salvelinus confluentus) in a northcentral British Columbia watershed. Canadian Journal of Fisheries and Aquatic Sciences 54:421–429. Behnke, R. J. 1992. Native trout of western North America. American Fisheries Society, Monograph 6. Behnke, R. J. 2002. Trout and salmon of North America. Simon and Schuster Inc., New York. Bohlen, P. J., P. M. Groffman, T. J. Fahey, M. C. Fisk, E. Suarez, D. M. Pelletier, and R. T. Fahey. 2004. Ecosystem consequences of exotic earthworm invasion of north temperate forests. Ecosystems 7(1):1-12. Brown, K. H., S. J. Patton, K. E. Martin, K. M. Nichols, R. Armstrong, and G. H. Thorgaard. 2004. Genetic analysis of interior Pacific Northwest Oncorhynchus mykiss reveals apparent ancient hybridization with westslope cutthroat trout. Transactions of American Fisheries Society 133:1078-1088. Butler, D. R. 1986. Conifer invasion of subalpine meadows, Central Lemhi Mountains, Idaho. Northwest Science 60(3):166-173. Carey, J. R. 1996. The incipient Mediterranean fruit fly population in California: implications for invasions biology. Ecology 77(6):1690-1697. Carey, J. R., P. B. Moyle, M. Rejmanek, and G. J. Vermeij. 1996. Special issue, invasion biology: preface. Biological Conservation 78:1-2. Carlton, J. T. 1996a. Biological invasions and cryptogenic species. Ecology 77(6):1653-1655. Carlton, J. T. 1996b. Pattern, process, and prediction in marine invasion ecology. Biological Conservation 78:97-106. Chapman, D. W. 1966. Food and space as regulators of salmonid populations in streams. American Naturalist 100:345-357. Chapman, R. N., D. M. Engle, R. E. Masters, and D. M. Leslie. 2004. Tree invasion constrains the influence of herbaceous structure in grassland bird habitats. Ecoscience 11(1):55-63.

26 Colautti, R. I. 2005. Are characteristics of introduced salmonid fishes biased by propagule pressure? Canadian Journal of Fisheries and Aquatic Sciences 62(4):950-959. Costello, A. B. 2006. Status of the westslope cutthroat trout (Oncorhynchus clarkii lewisii) in Alberta. Prepared for Alberta Sustainable Resource Development (SRD) and Alberta Conservation Association (ACA). Alberta Wildlife Status Report No. 61:47 p. Costello, A. B., and E. Rubidge. 2004. DRAFT COSEWIC status report on cutthroat trout (Oncorhynchus clarki subspp.). Prepared for the Committee on the Status of Endangered Wildlife in Canada. Crossman, E. J. 1991. Introduced freshwater fishes: a review of the North American perspective with emphasis on Canada. Canadian Journal of Fisheries and Aquatic Sciences 48(Supplement 1):46-57. Crowl, T. A., C. R. Townsend, and A. R. McIntosh. 1992. The impact of introduced brown and rainbow trout on native fish: the case of Australasia. Reviews in Fish Biology and Fisheries 2:217-241. Dark, S., R. J. Gutierrez, and G. I. Gould. 1998. The barred owl (Strix varia) invasion in California. The Auk 115(1):50-56. Davis, M. A. 2003. Biotic globalization: does competition from introduced species threaten biodiversity. Bioscience 53(5):481-489. Docker, M. F., A. Dale, and D. D. Heath. 2003. Erosion of interspecific reproductive barriers resulting from hatchery supplementation of rainbow trout sympatric with cutthroat trout. Molecular Ecology 12(12):3515-3521. Duff, D. A. 1996. Conservation assessment for inland cutthroat trout: distribution, status and habitat management implications. USDA For. Ser. Intermountain Region, Ogden, Utah. Elton, C. S. 1958. The ecology of invasions by animals and plants. Chapman and Hall, London. Fausch, K. D. 1989. Do gradient and temperature affect distributions of, and interactions between, brook charr (Salvelinus fontinalis) and other resident salmonids in streams? Physiology and Ecology Japan Special Volume 1:303-322.

27 Fausch, K. D., Y. Taniguchi, S. Nakano, G. D. Grossman, and C. R. Townsend. 2001. Flood disturbance regimes influence rainbow trout invasion success among five Holartic regions. Ecological Applications 11:1438-1455. Franklin, J. F., W. H. Moir, G. W. Douglas, and C. Wiberg. 1971. Invasion of subalpine meadows by trees in the Cascade Range, Washington and Oregon. Arctic and Alpine Research 3:215-224. Gowan, C., and K. D. Fausch. 1996. Long-term demographic responses of trout populations to habitat manipulation in six Colorado streams. Ecological Applications 6(3):931-946. Grant, R. B., and P. R. Grant. 1993. Evolution of Darwin's finches caused by a rare climatic event. Proceedings of Royal Society of London, Series B 251(1331):111-117. Griffith, J. S. 1988. Review of competition between cutthroat trout and other salmonids. American Fisheries Society Symposium 4:134-140. Grigorovich, I. A., R. I. Colautti, E. L. Mills, K. Holeck, A. G. Ballert, and H. J. Maclsaac. 2003. Ballast-mediated animal introductions in the Laurentian Great Lakes: retrospective and prospective analyses. Canadian Journal of Fisheries and Aquatic Sciences 60(6):740-756. Haas, G. 1998. Indigenous fish species potentially at risk in BC, with recommendations and prioritizations for conservation, forestry/resource use, inventory, and research. Ministry of Fisheries, Research and Development Branch Fisheries Management Report No. 105. Hanson, D. L. 1977. Habitat selection and spatial interaction in allopatric and sympatric populations of cutthroat and steelhead trout. Doctoral dissertation. University of Idaho, Moscow. Hawkins, D. K., and C. J. Foote. 1998. Early survival and development of coastal cutthroat trout (Oncorhynchus clarki clarki), steelhead (Oncorhynchus mykiss), and reciprocal hybrids. Canadian Journal of Fisheries and Aquatic Sciences 55:2097-2104. Hilderbrand, R. H., and J. L. Kershner. 2000. Conserving inland cutthroat trout in small streams: how much stream is enough? North American Journal of Fisheries Management 20:513-520. Hitt, N. P. 2002. Hybridization between westslope cutthroat trout (Oncorhynchus clarki lewisi) and rainbow trout (O. mykiss): distribution and limiting

28 factors. Master Thesis, Biological Sciences, University of Montana, Missoula, MT. Hitt, N. P., C. A. Frissell, C. C. Muhlfeld, and F. W. Allendorf. 2003. Spread of hybridization between native westslope cutthroat trout, Oncorhynchus clarki lewisi, and nonnative rainbow trout, Oncorhynchus mykiss. Canadian Journal of Fisheries and Aquatic Sciences 60:1440-1451. Hubbs, C. L. 1955. Hybridization between fish species in nature. Systematic Zoology 4(1):1-20. Jiggins, C. D., and J. Mallet. 2000. Bimodal hybrid zones and speciation. Trends in Ecology and Evolution 15(6):251-255. Johnson, L. E., and D. K. Padilla. 1996. Geographic spread of exotic species: ecological lessons and opportunities from the invasion of the zebra mussel (Dreissena polymorpha). Biol. Cons. 78:23-33. Kanda, N., R. F. Leary, and F. W. Allendorf. 2002. Evidence of introgressive hybridization between bull trout and brook trout. Transactions of the American Fisheries Society 131(4):772-782. Kaufman, L. 1992. Catastrophic change in species-rich freshwater ecosystems: lessons from Lake Victoria. Bioscience 42:846-858. Kitano, S., K. Maekawa, S. Nakano, and K. D. Fausch. 1994. Spawning behavior of bull trout in the Upper Flathead Drainage, Montana, with special reference to hybridization with brook trout. Transactions of the American Fisheries Society 123(6):988-992. Kohler, C. C., and W. R. J. Courtenay. 2003. American Fisheries Society Position on Introductions of Aquatic Species. web address http://www.afsifs.vt.edu/afspos.html. Kolar, C. S., and D. M. Lodge. 2001. Progress in invasion biology: predicting invaders. Trends in Ecology & Evolution 16:199-204. Krueger, C. C., and B. May. 1991. Ecological and genetic effects of salmonid introductions in North America. Canadian Journal of Fisheries and Aquatic Sciences 48 (Supplement 1):66-77. Kruse, C. G., W. A. Hubert, and F. J. Rahel. 1997. Geomorphic influences on the distribution of Yellowstone cutthroat trout in the Absaroka Mountains, Wyoming. Transactions of American Fisheries Society 126:418-427.

29 Larson, G. L., and S. E. Moore. 1985. Encroachment of exotic rainbow trout into stream populations of native brook trout in the southern Appalachian mountains. Transactions of the American Fisheries Society 114:195-203. Larson, G. L., S. E. Moore, and B. Carter. 1995. Ebb and flow of encroachment by nonnative rainbow trout in a small stream in the Southern Appalachian Mountains. Transactions of the American Fisheries Society 124:613-622. Leary, R., F. Allendorf, and G. Sage. 1995. Hybridization and introgression between introduced and native fish. Pages 91-101 in H.L. Schramm, Jr., and R.G. Piper, editors. Uses and effects of cultured fishes in aquatic ecosystems. American Fisheries Society, Symposium 15, Bethesda, Maryland. Leary, R., F. W. Allendorf, and K. L. Knudsen. 1987. Genetic divergence among populations of westslope cutthroat trout in the upper Kootenay River drainage, BC. Dept of Zoology University of Montana Population Genetics Laboratory Report 87/1. Leary, R. F., F. W. Allendorf, and K. L. Knudsen. 1985. Developmental instability and high meristic counts in interspecific hybrids of salmonid fishes. Evolution 39:1318-1326. Ledig, F. T. 1986. Heterozygosity, heterosis, and fitness in outbreeding plants. Pages 77-104 in M. E. Soule, editor. Conservation Biology: The Science of Scarcity and Diversity. Sinauer Associates, Inc., Sunderland, Massachusetts. Liknes, G., and P. Graham. 1988. Westslope cutthroat trout in Montana: Life history, status, and management. American Fisheries Society Symposium 4:53-60. Loftus, A., and C. Flather. 2000. Fish and other aquatic resource trends in the USA: A technical document supporting the 2000 USDA Forest Service RPA assessment. Gen. Tech. Rep. RMRS-GTR-53. USDA, Forest Service, Fort Collins, Colorado. Lynch, M., and J. S. Conery. 2000. The evolutionary fate and consequences of duplicate genes. Science 290:1151-1155. Marchetti, M. P., P. B. Moyle, and R. Levine. 2004. Alien fishes in California watersheds: characteristics of successful and failed invaders. Ecological Applications 14(2):587-596.

30 Markiw, M. E. 1989. Portals of Entry for Salmonid Whirling Disease in RainbowTrout. Diseases of Aquatic Organisms 6(1):7-10. Mayhood, D. 2000. Provisional evaluation of the status of westslope cutthroat trout in Canada. At risk proceedings of a conference on the biology and management of species and habitats at risk. Feb. 15-19, 1999, Kamloops, British Columbia. McIntyre, J. D., and B. E. Rieman. 1995. Westslope cutthroat trout In M.K. Young (Tech. Ed.) Conservation assessment for inland cutthroat trout. USDA For. Ser. Gen. Tech. Rep. RM-GTR-256. McKay, S. J., R. H. Devlin, and M. J. Smith. 1996. Phylogeny of Pacific salmon and trout based on growth hormone type-2 and mitochondrial NADH dehydrogenase subunit 3 DNA sequences. Canadian Journal of Fisheries and Aquatic Sciences 53:1165-1176. McPhail, J. D., and R. Carveth. 1992. A foundation for conservation: the nature and origin of the freshwater fish fauna of British Columbia. Fish Museum, Dept. Zoology, University of British Columbia, Vancouver, BC. MFWP. 2001. Stocking records for Koocanusa Reservoir. Montana Department of Fish, Wildlife, and Parks, Libby, Montana. Miller, R. R., J. D. Williams, and J. E. Williams. 1989. Extinctions of North American fishes during the past century. Fisheries 14(6):22-38. Moller, H. 1996. Lessons for invasion theory from social insects. Biological Conservation 78:125-142. Moore, S. E., and B. Ridley. 1984. A summary of changing standing crops of native brook trout in response to removal of sympatric rainbow trout in Great Smoky Mountains National Park. Journal of the Tennessee Academy of Science 59(4):76-77. Moyle, P. B. 1986. Fish introductions into North America: patterns and ecological impact. Pages 27-43 in Mooney, H.A. and J.A. Drake (eds.) Ecology of biological invasions in North America and Hawaii. Springer-Verlag, New York, NY. Moyle, P. B., and T. Light. 1996a. Biological invasions of fresh water: empirical rules and assembly theory. Biol. Cons. 78:149-161. Moyle, P. B., and T. Light. 1996b. Fish invasions in California: do abiotic factors determine success. Ecology 77(6):1666-1670.

31 MWLAP. 2006. Regional fish stocking records: Kootenay Region, 2006. Ministry of Water, Land and Air Protection, Victoria, British Columbia. Myers, N., and A. H. Knoll. 2001. The biotic crisis and the future of evolution. Proceedings of National Academy of Sciences 98:5389–5392. Nakano, S., S. Kitano, K. Nakai, and K. D. Fausch. 1998. Competitive interactions for foraging microhabitat among introduced brook char, Salvelinus fontinalis, and native bull char, S. confluentus, and westslope cutthroat trout, Oncorhynchus clarki lewisi, in a Montana stream. Environmental Biology of Fishes 52:345-355. Niimi, A. J., and D. M. Reid. 2003. Low salinity residual ballast discharge and exotic species introductions to the North American Great Lakes. Marine Pollution Bulletin 46(10):1334-1340. Novinger, D. C., and F. J. Rahel. 2003. Isolation management with artificial barriers as a conservation strategy for cutthroat trout in headwater streams. Conservation Biology 17(3):772-781. Ostberg, C., S. Slatton, and R. Rodriguez. 2004. Spatial partitioning and asymmetric hybridization among sympatric coastal steelhead trout (Oncorhynchus mykiss irideus), coastal cutthroat trout (O. clarki clarki) and interspecific hybrids. Molecular Ecology 13:2773–2788. Paul, A. J., and J. R. Post. 2001. Spatial distribution of native and nonnative salmonids in streams of the Eastern Slopes of the Canadian Rocky Mountains. Transactions of American Fisheries Society 130:417-430. Peterson, D. P. 2002. Population ecology of an invasion: demography, dispersal, and effects of nonnative brook trout on native cutthroat trout. PhD dissertation, Department of Fishery and Wildlife Biology, Colorado State University, Fort Collins, Colorado. Peterson, D. P., and K. D. Fausch. 2003. Testing population-level mechanisms of invasion by a mobile vertebrate: a simple conceptual framework for salmonids in streams. Biological Invasions 5:239-259. Pimentel, D., L. Lach, R. Zuniga, and D. Morrison. 2000. Environmental and economic costs of non-indigenous species in the United States. Bioscience 50:53-65. Platts, W. S. 1979. Relationships among stream order, fish populations, and aquatic geomorphology in an Idaho river drainage. Fisheries 4:5-9.

32 Reichard, S. H., and P. S. White. 2003. Invasion biology: an emerging field of study. Annals of the Missouri Botanical Garden 90(1):64-66. Richter, B. D., D. P. Braun, M. A. Mendelson, and L. L. Master. 1997. Threats to imperiled freshwater fauna. Conservation Biology 11(5):1081-1093. Rose, M., and L. Hermanutz. 2004. Are boreal ecosystems susceptible to alien plant invasion? Evidence from protected areas. Oecologia 139(3):467477. Rubidge, E. 2003. Molecular analysis of hybridization between native westslope cutthroat trout (Oncorhynchus clarki lewisi) and introduced rainbow trout (O. mykiss) in southeastern British Columbia. Master's thesis, Department of Zoology, University of British Columbia, Vancouver. Rubidge, E., P. Corbett, and E. B. Taylor. 2001. A molecular analysis of hybridization between native westslope cutthroat trout and introduced rainbow trout in southeastern British Columbia, Canada. Journal of Fish Biology 59:42-54. Rubidge, E. M., and E. B. Taylor. 2004. Hybrid zone structure and the potential role of selection in hybridizing populations of native westslope cutthroat trout (Oncorhynchus clarki lewisi) and introduced rainbow trout (Omykiss). Molecular Ecology 13(12):3735-3749. Schluter, D. 1993. Ecological speciation in postglacial fishes. Philosophical Transactions of the Royal Society of London, Series B: Biological Sciences 352:807-814. Semmens, B. X., E. R. Buhle, A. K. Salomon, and C. V. Pattengill-Semmens. 2004. A hotspot of non-native marine fishes: evidence for the aquarium trade as an invasion pathway. Marine Ecology-Progress Series 266:239244. Shepard, B. 1997. Status and risk of extinction for westslope cutthroat trout in the upper Missouri River basin, Montana. North American Journal of Fisheries Management 17:1158-1172. Shepard, B. B. 2003. Status of westslope cutthroat trout (Oncorhynchus clarki lewisi) in the United States: 2002. Westslope cutthroat interagency conservation team, Montana Fish, Wildlife, and Parks and Montana Cooperative Fishery Research Unit, Bozeman, Montana.

33 Shepard, B. B., B. E. May, and W. Urie. 2005. Status and conservation of westslope cutthroat trout within the western United States. North American Journal of Fisheries Management 25(4):1426-1440. Simberloff, D., and W. Boecklen. 1991. Patterns of extinction in the introduced Hawaiian avifauna: a reexamination of the role of competition. American Naturalist 138:300-327. Soule, M. E. 1990. The onslaught of alien species, and other challenges in the coming decades. Conservation Biology 4(3):233-239. Strahler, A. N. 1957. Quantitative analysis of watershed geomorphology. Transactions of American Geophysical Union 38:913-920. Strange, E. M., P. B. Moyle, and T. C. Foin. 1992. Interactions between stochastic and deterministic processes in stream fish community assembly. Environmental Biology of Fishes 36:1-15. Taylor, E. B. 2004. Evolution in mixed company: inferences from studies of natural hybridization in Salmonidae. Pages 232-263 In A.P. Hendry and S. Stearns (eds) Evolution illuminated: salmon and their relatives. Oxford Press, Oxford, United Kingdom. Thompson, P., and F. Rahel. 1998. Evaluation of artificial barriers in small rocky mountain streams for preventing the upstream movement of brook trout. North American Journal of Fisheries Management 18:206-210. Townsend, C. R. 1996. Invasion biology and ecological impacts of brown trout Salmo trutta in New Zealand. Biological Conservation 78:13-22. Townsend, C. R. 2003. Individual, population, community, and ecosystem consequences of a fish invader in New Zealand streams. Conservation Biology 17(1):38-47. Utter, F. 2003. Genetic impacts of fish introductions. In E.M. Hallerman (ed) Population genetics: principles and applications for fisheries scientists American Fisheries Society, Bethesda, Maryland. Van Eimeren, P. 1996. Westslope cutthroat trout In D.A. Duff (Tech. Ed.) Conservation assessment for inland cutthroat trout: distribution, status and habitat implications. USDA, Forest Service, Intermountain Region, Ogden, Utah. Vermeij, G. J. 1996. An agenda for invasion biology. Biological Conservation 78:3-9.

34 Warren, M. L., and B. M. Burr. 1994. Status of freshwater fishes of the United States: overview of an imperiled fauna. Fisheries 19(1):6-18. Warren, M. L. J., B. M. Burr, S. J. Walsh, H. L. J. Bart, R. C. Cashner, D. A. Etnier, B. J. Freeman, B. R. Kuhajda, R. L. Mayden, H. W. Robison, S. T. Ross, and W. C. Starnes. 2000. Diversity, distribution, and conservation status of the native freshwater fishes of the Southern United States. Fisheries 25(10):7-31. Weigel, D. E., J. T. Peterson, and P. Spruell. 2003. Introgressive hybridization between native cutthroat trout and introduced rainbow trout. Ecological Applications 13(1):38-50. Welcomme, R. L. 1992. A history of international introductions of inland aquatic species. ICES marine science symposia 194:3-14. Whatley, M. 1972. Effects on fish in the Kootenay River of construction of Libby Dam. Fisheries Management Report No. 65. Fish Habitat Protection Section, Fish and Wildlife Branch, Dept. of Recreation and Conservation, Victoria, BC. Wilcove, D. S., D. Rothstein, J. Dubow, A. Phillips, and E. Losos. 1998. Quantifying threats to imperiled species in the United States. Bioscience 48(8):607-615. Williamson, M. 1996. Biological invasions. Chapman and Hall, London. Wydoski, R. S., and R. R. Whitney. 2003. Inland fishes of Washington. American Fisheries Society, Second Edition, Bethesda, Maryland. Young, M. 1995. Conservation assessment for inland cutthroat trout. USDA For. Ser., RM-GTR-256, Fort Collins, Colorado.

35 CHAPTER 2 ASSESSING THE GENETIC INTEGRITY OF A WESTSLOPE CUTTHROAT TROUT STRONGHOLD AT THE NORTHERN PERIPHERY OF ITS RANGE Abstract Many westslope cutthroat trout (WCT) populations in the United States of America (USA) are threatened by hybridization with non-native trout. However, less is known about the genetic status of WCT populations in Canadian portion of their range. The Upper Kootenay River in British Columbia (BC) was considered a stronghold for WCT, but a preliminary study conducted from 1999 to 2001 found evidence of recent hybridization likely caused by stocking introduced rainbow trout (RBT, Oncorhynchus mykiss) in the Koocanusa Reservoir in the late 1980s. We assessed the distribution of native WCT, RBT, and their hybrids in the Upper Kootenay River using seven co-dominant, simple sequence repeat markers, and the program NEWHYBRIDS Version 1.1 beta. We determined the genotypes of 2,670 fish collected from 31 different streams between 1999 and 2006. Most fish were classified as pure WCT (85%), whereas pure RBT, RBT backcross and F1 individuals were rare (combined 2.8%). However, a broad hybrid zone stretching over 200 km was observed with WCT backcross individuals as the dominant hybrid type present (6%). Sites with no fish migration barriers between them and the Koocanusa Reservoir (open sites) had significantly more hybrid genotypes than sites isolated above migration barriers

(closed sites; X2 = 60.15, df = 5, p < 0.00001). Open sites < 90 km from the

36

reservoir had more RBT, RBT backcross, and F1 types than sites further from the reservoir, and hybrids appear to be the main vector for the dispersal of RBT alleles to upstream sites. The hybrid zone appears to be limited by environmentally dependent factors (ecotone model). Management efforts should focus on identifying and limiting the sources of RBT in the watershed, identifying the remaining pure WCT populations, and learning more about the movement of WCT within the hybrid zone. Introduction Human-caused species extinctions are happening at an alarming rate, and evolutionary processes that create diversity are being disrupted (Myers and Knoll 2001). Habitat destruction and competition from introduced species are often the leading causes of species extinctions (Miller et al. 1989; Warren and Burr 1994; Richter et al. 1997; Wilcove et al. 1998), and freshwater species are disproportionately more threatened than terrestrial species in the USA (Warren and Burr 1994; Richter et al. 1997; Loftus and Flather 2000). About 50,000 species have been introduced intentionally or unintentionally to the USA (Pimentel et al. 2000), and in western North America many watersheds now have more introduced aquatic species than native species (Crossman 1991; Wydoski and Whitney 2003). Kohler and Courtenay (2003) describe five negative impacts of introduced species: habitat alteration, introduction of diseases, trophic alterations (predation), spatial alterations (competition), and gene pool

37 deterioration (hybridization). The increased availability of inexpensive molecular techniques has allowed investigators to study the negative effects of hybridization, especially the potential of hybridization to cause species extinctions (Rhymer and Simberloff 1996). Salmonid fishes have been introduced more often than almost any group of fishes because of their popularity as sport fish and the ease with which they can be reared and transported (Crossman 1991; Welcomme 1992). In particular, RBT, which are native to western North America and eastern Siberia (Behnke 1992), have been introduced more than any other fish species, and are now found on all continents except Antarctica (Welcomme 1992). Salmonids are prone to hybridization because of behaviors that fail to fully isolate species during reproduction and competition for limited spawning habitat (Behnke 1992; Allendorf and Waples 1996). Therefore, it is not unexpected that natural- and human-induced hybridization between native and introduced salmonids is common (e.g. Jansson and Ost 1997; Kanda et al. 2002a; Brown et al. 2004). There is evidence that species of Oncorhynchus hybridize less frequently than other genera in the salmonid family, and it is suspected that complex life histories that include anadromy may select against hybrids (Utter 2000). However, many non-anadromous species of Oncorhynchus, including the 11 extant subspecies of inland cutthroat trout (O. clarki subsp.), appear to be especially susceptible to hybridization with introduced RBT, and hybridization is often considered one of the greatest threats to the persistence of most sub-

38 species of inland cutthroat trout (Allendorf and Leary 1988; Leary et al. 1995; Behnke 2002). Introgression, by which the alleles of one distinct population move into another distinct population, can occur when hybrids are viable and fertile (Arnold 1997). Overlapping zones of two distinct populations that are hybridizing are referred to as hybrid zones and the structure of these zones in time and space can indicate what forces are ultimately controlling hybridization and introgression (Taylor 2004). There is evidence that many salmonid hybrid zones conform to the environment-dependent zone models, whereby the extent of a hybrid zone is limited by environmental factors (Arnold 1997; Taylor 2004; Ostberg and Rodriguez 2006). In this study we assess a hybrid zone between native westslope cutthroat trout (WCT, O. c. lewisi) and introduced RBT in the Upper Kootenay River, BC. The westslope cutthroat trout is a subspecies of inland cutthroat trout native to western North America (Behnke 1992). WCT have the most geographically widespread range of all interior subspecies of cutthroat trout and are found in Alberta, BC, Idaho, Montana, Oregon, Washington, and Wyoming encompassing most of the upper Missouri Basin and headwaters of the Columbia River (Behnke 1992). The distribution of WCT appears to have been determined approximately 70,000 years ago by the formation of barrier falls on many of the large tributaries to the Columbia River including the Kootenay, Clark Fork, Pend ‘Oreille, and Spokane Rivers. It is hypothesized that WCT were able to colonize above barrier falls because water levels were higher during the glacial retreat, or

39 possibly because barriers formed following glacial retreat as the land mass rebounded (Behnke 1992). Westslope cutthroat trout may have been isolated above these barrier falls and survived in Montana, Idaho, and Washington during the most recent glacial period (Allendorf et al. 1980; Behnke 1992). Rainbow trout appear to have been restricted to the lower Columbia River during this period allowing WCT to colonize large inland portions of North America in isolation from RBT (Behnke 1992). There are some locations within the historic range of WCT where they are naturally sympatric with RBT. In these situations, the two species are segregated along elevational gradients, with WCT occupying higher elevation sites (Platts 1979). This same pattern is found in sympatric populations of anadromous RBT (i.e., steelhead) and coastal cutthroat trout (O. c. clarki) (Hanson 1977; Ostberg et al. 2004). Paul and Post (2001) found that RBT tend to redistribute to lower elevation sites when they are introduced into headwater lakes above populations of WCT. When RBT are introduced to areas with allopatric WCT populations, they often hybridize, indicating that these WCT lack innate isolating mechanisms that allow them to co-exist with RBT (Behnke 1992). In the USA, the range of WCT has decreased by about 40% and between 65 and 87% of the occupied habitat does not contain genetically pure WCT (Shepard et al. 2005). The situation appears to be similar in parts of the Canadian range. The number of genetically pure populations of WCT has declined in Alberta by as much as 95% from their former range, and most of the remaining populations exist in isolated headwater tributaries (Mayhood 2000).

40 The Committee on the Status of Endangered Wildlife in Canada (COSEWIC) lists WCT as threatened in Alberta and a recent WCT status report commissioned by the Alberta provincial government estimates that less than 7,000 genetically pure adult fish remain in the province (Costello 2006). The Upper Kootenay River contains the largest continuous range of WCT in BC, and WCT populations may be less affected by hybridization in this area. A cursory genetic assessment in 1986 found that only one tributary stream out of the seven surveyed had evidence of WCT x RBT hybrids (Leary et al. 1987). The stream with hybrids was the North White River which is directly upstream of Whiteswan Lake where over 1.5 million RBT have been stocked since 1931 (MWLAP 2006). However, a repeat sampling of the same seven streams in 1999 found four streams with hybrids present indicating that the genetic integrity of WCT populations has declined (Rubidge et al. 2001). Further sampling in 2000 and 2001 found hybrids present in other streams with some sites having > 35% hybrids (Rubidge and Taylor 2004, 2005). Rubidge and Taylor (2004; 2005) concluded that the increase in hybridization is relatively recent, and that a RBT stocking program in Koocanusa Reservoir, which began in 1988, is likely the primary cause for the increase. In BC, WCT are listed as “species of special concern” by the federal government (Costello and Rubidge 2004) and as “threatened” by the provincial government (Bennett 2002). Our objectives were to determine the distribution and abundance of WCT, RBT, and their hybrids, and measure the level of introgression between native WCT and introduced RBT in the Upper Kootenay River. We classify fish into six

41 genotype frequency classes and compute the probability a fish belongs to each class using NEWHYBRIDS (Anderson and Thompson 2002). This work is an expansion of the previous work of Rubidge (2003) and is important because it appears that the BC population of WCT may represent a stronghold in the northern periphery of the species range. Study Area The study area encompasses the Canadian portion of the Kootenay River from its headwaters in Kootenay National Park down to the Canada/U.S. border near Newgate, BC (Figure 2). This portion of the Kootenay River is in the East Kootenay Region of BC and is bounded by the Rocky Mountains to the east and the Purcell Mountains to the west. The study area is approximately 250 km long and the drainage area is approximately 18,500 km2. The Kootenay River is the largest tributary to the Columbia River in Canada and is a 7th order stream at the Canada/U.S. border (Strahler 1957). The headwater region of the study area is dominated by large mountainous areas with numerous glaciers and snowfields. Like most watersheds throughout western North America, non-native fish have been stocked extensively throughout the Upper Kootenay River, particularly in low elevation lakes (MWLAP 2006). The first records of RBT stocking in the Upper Kootenay River begin in 1915, and almost 20 million RBT have been stocked since then at 114 sites in over 2,500 individual stocking events (MWLAP 2006).

42 The Koocanusa Reservoir is a 170 km long reservoir on the Kootenay River that was formed after the completion of the Libby dam in 1972 at Libby, Montana (Whatley 1972). The dam was located at a natural barrier which was suspected of isolating WCT upstream from RBT in the Lower Kootenay River (Allendorf et al. 1980; Behnke 1992). Fisheries agencies in both Canada and the USA were unsuccessful in attempts to establish WCT in the reservoir (B. Westover, personal communications, 2003). A policy was then developed to establish Gerrard strain RBT in the reservoir, and between 1986 and 1998 Gerrard strain RBT were stocked in tributaries to the reservoir and in the reservoir itself (MFWP 2001; MWLAP 2006). Growing conservation concerns prompted stocking of RBT to be replaced with WCT stocking, or triploid RBT stocking, in lakes and reservoirs with potential outlets in the Upper Kootenay River starting in 1999. Methods This project builds on previous sampling that occurred from 1999-2001 (Rubidge 2003). The raw data collected during the previous sampling was combined with samples collected from 2002-2006 to form one comprehensive database on which most analyses were conducted. All sample sites were selected in tributary streams (3rd to 6th order streams) to the Upper Kootenay River in a systematic fashion throughout the watershed at varying distances from the Koocanusa Reservoir, and at a range of elevations within streams. Each stream was divided into three reaches of equal length (i.e., lower, mid, upper)

43 and a representative sample site was selected in a reach based on access and location. We chose this basic design because the previous study had implicated Koocanusa Reservoir as the cause of increased hybridization and other similar studies suggested that elevation played a significant role in the distribution of WCT, RBT, and their hybrids (Paul and Post 2001; Rubidge et al. 2001; Weigel et al. 2003; Rubidge and Taylor 2005). Several sample sites were located above known fish migration barriers where presumed pure populations of WCT existed. All sites were classified as either open or closed to fish migration. Sites were considered open if introduced RBT could potentially move from the Koocanusa Reservoir to a site (Figure 2). Streams were considered closed if they were isolated from the Koocanusa Reservoir by a fish migration barrier. Fish migration barriers were identified with the use of an existing BC Ministry of Water, Land, and Air Protection (MWLAP) database of fish obstructions compiled from extensive fish and habitat inventories (http://www.env.gov.bc.ca/fish/fiss/index.html). This database is the result of extensive efforts by the BC government to synthesize existing fisheries information into a comprehensive database. Also, since the early 1990s the government has worked to develop detailed inventory standards and quality assurance protocols for all fish and wildlife inventories (http://ilmbwww.gov.bc.ca/risc/). The barrier database is likely accurate for the larger streams we worked in, but we suspect that barriers on smaller streams (order 1 and 2) are not well represented because these streams are underrepresented in the database. We classified obstructions in the database as

44 complete fish migration barriers if they met the following criteria: 1) if the obstruction was labeled as the upstream limit or fish migration (usually as a result of a field assessment and fish sampling above and below the feature), and 2) any obstruction > 2 m tall that was described as either bedrock, cascade, chute, waterfall, or human constructed. We chose these criteria to classify barriers based on the demonstrated jumping ability of inland trout (Thompson and Rahel 1998; Novinger and Rahel 2003; Kondratieff and Myrick 2006). We considered all beaver dam, canyon, culvert, and debris obstructions in the database as “porous” or “ephemeral” obstructions, and unlikely to prevent upstream fish migration based on studies of trout movement over barriers (Thompson and Rahel 1998; WFL 2003; Coffman 2005). Fish Capture and Tissue Collection We sampled fish from 31 streams and 45 sites (Table 2). Tissue samples were collected and analyzed from 2,670 fish that were captured between 1999 and 2006. Sixty-eight separate samples (number of fish ranging from 6 to 299) were collected from revisits, long-term monitoring at some sites, and an extensive cohort analysis in 2000 (Rubidge and Taylor 2004). Sample sites ranged from 1 to 241 km from the Koocanusa Reservoir and from 753 to 1,566 m elevation. All fish were caught and tissue collected as per Rubidge et al. (2001). At each sample site we attempted to capture 30 fish. The majority of the fish (85%) were captured by angling, followed by dip nets (12%), electroshocking (2.5%),

45 and minnow traps (0.5%). Multiple age classes were sampled at each site and sample reaches ranged from 1-3 km long, theoretically reducing the likelihood of sampling siblings (Weigel et al. 2003). We measured each fish captured to the nearest 0.5 cm and used a length frequency analysis to estimate the age of each fish. For all fingerling, juvenile, and adult fish, we clipped a small piece of the lower caudal fin and placed it in 1.5 mL of 95% ethanol. Tissue samples were collected from fish in the order they were caught as we moved upstream. Fry were collected whole. Samples were stored at room temperature until the DNA extraction process was completed. All sampling occurred during summer low flow conditions, typically from mid-July to early September. DNA Extraction and Amplification Tissue samples collected from 1999 to 2001 were analyzed by Rubidge et al. (2001) and Rubidge and Taylor (2004). For tissue samples collected from 2002-2006 we used a standard DNA salt-extraction technique (Aljanabi and Martinez 1997) and then the DNA was diluted to 5 ng/µl following quantification with fluorometry. The Nevada Genomics Center (http://www.ag.unr.edu/genomics) performed all PCR and fragment sizing on new samples using an Applied Biosystems Prism 3730 DNA Analyzer (Foster City, CA, U.S.A). We used seven fluorescently labeled simple sequence repeat (SSR) co-dominant, markers isolated from several species of inland cutthroat and rainbow trout (Ostberg and Rodriguez 2004). The markers were diagnostic and produced species specific DNA fragment sizes for either WCT or RBT, and

46 hybrids produced both fragments. Polymerase chain reaction (PCR) was performed in 15 µl reactions using 20ng DNA and reagent concentrations listed in Table 3. The thermal cycling protocol for each multiplexed reaction was an initial heating to 95º C for 15 min followed by 34 cycles of 95º C (30 sec), 57º C (1.5 min) and 72º C (30 sec) and finally 30 min at 62º C. Individuals were genotyped automatically using auto bin features in Genemapper v3.0 (Applied Biosystems) and then each individual was screened manually at each locus to ensure correct allele assignments. We only counted alleles if the peaks met all of the following criteria: peak height > 100, peak height > two times any background noise, only peaks within + two base pairs of the expected allele sizes for each particular locus, and peak height ratio of each allele was > 0.4. We used a test group of 32 fish to assess whether we could compare our data generated with seven diagnostic markers to the four marker data used by Rubidge (2003). This comparison indicated that both marker sets classified fish the same (matched pair two tailed t-test N = 32, df = 31, t value = 0.75, p = 0.458). Classification of Hybrids We used the Bayesian model NEWHYBRIDS developed by Anderson and Thompson (2002) to classify fish into genotype classes. The model uses Markov chain Monte Carlo (MCMC) sampling to compute the posterior probability that individuals in a sample belong to parental and hybrid categories. We specified six genotype frequency classes based on the expected genotype frequencies arising from two generations of interbreeding between WCT and RBT populations (Table

47 4, Anderson and Thompson 2002). We specified prior allele frequencies for the model because the loci we used were presumed diagnostic for WCT and RBT (Ostberg and Rodriguez 2004). Each species specific allele was assigned a frequency of 10,000 for the diagnostic allele and 0 for non-diagnostic allele (i.e., the allele presumed to be absent from either WCT or RBT). We pooled the samples collected during different years at the same site because the proportion of RBT alleles did not change between years at most sites (Chapter 4). We ran the model for each site separately because the model uses the allele frequencies of individuals in the sample to assign posterior probabilities and treats all individuals simultaneously. We ran 25,000 sweeps of the MCMC for each sample and compared the results from several independent runs on test samples to see if the results were consistent. Individuals that were not classified into a particular genotype frequency class with a > 0.9 posterior probability were classified as unknown hybrid types (i.e., all fish that were classified with posterior probabilities < 0.9 had at least one RBT allele). A fish classified as a pure WCT with a 0.90 posterior probability has a 90% chance of being a pure WCT provided all the assumptions of the model are met. The model assumes that populations were in Hardy-Weinberg and linkage equilibrium prior to the populations mixing, samples are independent, and the number of generations of hybridization is < 2 generations (Anderson and Thompson 2002). We could not test if the populations were in Hardy Weinberg or linkage equilibrium prior to mixing because the populations have been hybridizing since at least 1999. If Hardy Weinberg and linkage disequilibrium are present the

48 model assumes they are entirely the result of admixture of the two populations and it uses this information to determine expected genotype frequencies. If these assumptions are violated, an overestimation of the confidence in hybrid classification can occur. In particular, it is not possible to precisely resolve the genotype frequency of individuals when there has been > 3 generations of inbreeding between two populations because the expected proportions of multilocus genotypes are identical for F2 and F3 generations (Anderson and Thompson 2002). If there has been more than two generations of inbreeding this will result in individuals being assigned to either WCT_bx or F2 genotypes that are likely later generation backcrosses or a mixture of backcrosses and pure types (E. Anderson, Fisheries Ecology Division, NOAA, personal communication). The NEWHYBRIDS classification of genotypes is similar to manual classification of hybrid genotypes used by Ostberg and Rodriguez (2006) and Rubidge and Taylor (2004). However, NEWHYBRIDS can be an improvement over manual method because the posterior probabilities that the model generates provide a measure of confidence in the assignment of an individual to each genotype category. Allele frequencies should conform to Hardy-Weinberg expectations if one randomly mixing population is present at a site, and recent hybridization between populations should result in significant linkage disequilibrium and marked deficiencies in heterozygotes due to either assortive mating or selection against hybrids (Campton 1987). We tested whether our samples conformed to HardyWeinberg expectations using the exact test in Genetic Data Analysis (GDA)

49 Version 1.0 (Lewis and Zaykin 2001). We tested the assumption that there was no association between loci by conducting pairwise comparisons between loci with GDA. The resulting test statistics were adjusted for multiple tests using Bonferroni corrections (Holm 1979). Bonferroni corrections maintain a lower chance of making a type I error at the direct cost of making more type II errors (Verhoeven et al. 2005). Distribution of Hybrids and Hybrid Zone Structure We calculated the proportion of individuals by genotype at each site to characterize the distribution of genotypes. The distribution of genotypes at a site can indicate the selective processes acting in the hybrid zone (Jiggins and Mallet 2000). Distributions are described as unimodal skewed towards a parental type, flat (uniform distribution of genotypes), or bimodal (both parental types common, Jiggins and Mallet 2000). We tested for differences between genotype distributions using the program RxC developed by Miller (1997) that uses Fisher's Exact test on any sized contingency table through the use of the Metropolis algorithm (Raymond 1995). We tested the hypothesis that the Koocanusa Reservoir was the main source of RBT by comparing the distribution of genotypes between open and closed sites using a Chi-square test. We assumed that if the Koocanusa Reservoir was responsible for the increase in hybridization, we would find more F1, F2, RBT_bx and pure RBT near the reservoir. We also calculated the % RBT alleles at each site as described in Rubidge and Taylor (2004). We tested the

50 relationship between the % RBT alleles per site and both elevation and distance from the Koocanusa Reservoir using the Spearman Rank Correlation. We also calculated our power to detect RBT alleles at a site. The power of the experimental design to detect the presence of RBT alleles is equal to β or 1 – α (Kanda et al. 2002b) where α:

α = (1-q)2nx

(Equation 1)

q = the desired frequency of RBT alleles you wish to detect, n = the number of fish sampled, and x = the number of diagnostic markers. For seven markers and a sample size of 30 fish the probability of detecting 1% RBT alleles is 98.5% (i.e., 1 – α where α = 0.015). The probability of detecting < 1% RBT alleles drops off rapidly regardless of the number of markers used and makes detection of < 1% RBT alleles expensive (Boecklen and Howard 1997). Results Classification of Hybrids The Bayesian model NEWHYBRIDS classified 2,569 (96.2%) fish into a single genotype frequency class with a > 0.9 posterior probability of belonging to a particular genotype class. Fish with a posterior probability of < 0.9 were classified as unknown hybrids because they all had at least one RBT allele. The majority of fish were classified as being pure WCT (84.8%) followed by WCT_bx (6.5%) and unknown (3.8%, Table 5). Very few fish were classified as F2 (2.4%),

51 pure RBT (1.9%), RBT_bx (0.5%), and F1 (0.3%). Of the 31 sites that had hybrid genotypes present, the majority of sites (71%) had only later generation (WCT_bx and F2) hybrids. Distribution of Genotypes and Hybrid Zone Structure There were 14 (31%) sites that were putatively pure WCT, 10 of which were above migration barriers (Table 5, Figure 5). However, only four putatively pure WCT sites were isolated by migration barriers from known sources of RBT (Bull River, Fording River, Meachen Creek, and Sand Creek). There were significantly more pure WCT and significantly less hybrid genotypes at the closed sites compared to the open sites (X2 = 60.15, df = 5, p < 0.00001). Three types of genotype distributions were apparent: pure WCT, unimodal left skewed to WCT, and partially bimodal (Figure 3). All of the bimodal genotype distributions were at open sites < 30 km from the reservoir (Table 5). The bimodal (Gold Creek) and unimodal left skewed to WCT (Perry Creek) distributions were significantly different from each other based on Fisher’s exact test performed using RxC (p < 0.00001, Miller 1997). Open sites with unimodal genotype distributions that were left skewed to WCT generally progressed from a high proportion of WCT_bx and F2 individuals at moderate distances form the reservoir (i.e., 30-80 km) to being predominantly pure WCT and WCT_bx types > 90 km from the reservoir (Table 5). However, pure RBT were detected up to 87 km from the reservoir (e.g., Skookumchuck

52 Creek) and F1 genotypes were detected up to 116 km from the reservoir (e.g., East White River). The majority of closed sites (53%) had putatively purer populations (i.e., no hybrid genotypes detected). Most of the closed sites (74%) that we sampled were located in the Elk River watershed above a fish migration barrier (Figure 2). There is over 115 km of mainstem habitat in the Elk River isolated from the Koocanusa Reservoir above the barrier. No pure RBT and only one RBT_bx were detected at closed sites in the Elk River (e.g., lower Alexander Creek). WCT_bx and F2 genotypes were the most common hybrid types in the Elk River above the dam, and most hybrids were found at lower Alexander Creek which had 22% hybrid genotypes. Introgression and Distribution of RBT Alleles Three broad levels of introgression were observed at sites: 0%, < 10%, and > 10% RBT alleles. The % RBT alleles were negatively correlated with the distance sites were from the Koocanusa Reservoir and site elevation (n = 45, distance R2 = -0.21 p = 0.0012 and elevation R2 = -0.21 p = 0.0011). When only open sites were used in the analysis, the correlation between the % RBT alleles and distance from the reservoir strengthened to 0.49 and the correlation between the % RBT and elevation increased to 0.39 (Figure 4). The trend for lower elevation sites to have more hybridization was consistent in 12 of the 15 streams where we sampled downstream and upstream locations within the same stream.

53 The only streams where we found more hybrids in upstream sites were Alexander Creek, Michel Creek, and Wigwam River (Table 4, Figure 2). Population Genetic Analysis Tests for deviations from Hardy-Weinberg equilibrium and the presence of linkage disequilibrium for samples collected from 1999-2001 were reported in Rubidge and Taylor (2004). They found four sites that deviated from HardyWeinberg expectations and seven sites that had significant linkage disequilibrium. Four streams (five total samples) collected from 2002-2006 had deviated significantly from Hardy-Weinberg expectations (Table 6). These sites were in the lower reaches of the Elk River, Gold Creek, Sand Creek, and Skookumchuck Creek. There was also evidence of linkage disequilibrium in all of the sites that deviated from Hardy-Weinberg expectations (Table 7). All sites that did not conform to Hardy-Weinberg expectations displayed significant linkage disequilibrium at every pairwise loci comparison (P < 0.005). All the sites that had deviated from Hardy Weinberg expectations and had significant linkage disequilibrium had genotype frequency distributions that were bimodal between WCT and RBT (Table 5). Numerous other populations displayed some linkage disequilibrium but were not significant when Bonferroni corrections were applied (Table 7). The number of differences between pairwise loci comparisons varied widely within populations from only two (e.g., Alexander Creek ALC, 2003) to 20 (lower East White River, 2004) out of a possible 21 pairwise comparisons. Sites with

54 significant linkage disequilibrium, but that conformed to Hardy-Weinberg expectations were closed sites clustered near Summit Lake, open sites at mid elevations, or open sites > 50 km from Koocanusa Reservoir (Table 5). Discussion Distribution and Abundance of Hybrids The expansion of surveys to new streams, upper elevations, sites above known fish migration barriers, and sites near Koocanusa Reservoir indicates that hybridization between WCT and RBT has occurred throughout much of the Upper Kootenay River. Our study found an extensive hybrid zone reaching from the Canada/U.S. border for over 200 km upstream to near the headwaters of the Upper Kootenay River. We detected hybrid genotypes at almost 70% of all sample sites (31/45). The extent of hybridization between WCT and RBT in the Upper Kootenay River is similar to that of other areas within the native range of WCT. For example, WCT x RBT hybrids were found at > 55% of the sites surveyed in the Flathead River drainage, Montana (Hitt et al. 2003; Boyer 2006), > 64% of the sites in the Middle Fork Salmon River and Clearwater River drainages, Idaho (Campbell and Cegelski 2003; Weigel et al. 2003), and 83% of the sites in Stehekin River drainage, Washington (Ostberg and Rodriguez 2006). The distribution of hybrids between RBT and other inland trout species is also extensive. For example, Kruse et al. (2000) found genetically pure Yellowstone cutthroat trout (O. c. bouvieri) in only 26% of the 104 streams surveyed in

55 Wyoming, and hybrids between RBT and Apache trout (Oncorhynchus apache) were found at 35% of the sites surveyed in Arizona (Carmichael et al. 1993). Genotypes and Hybrid Zone Structure The majority of fish in the upper Kootenay River system were classified as pure WCT using NEWHYBRIDS, and most sites with hybrids had genotype distributions that were skewed heavily towards pure WCT. Sites with F1 and RBT genotypes all tended to be close to the Koocanusa Reservoir. Rubidge and Taylor (2004) found a similar distribution of genotypes in the Upper Kootenay River by manually classifying each individual using the presence or absence of WCT and RBT alleles at each locus. Ostberg and Rodriguez (2006) also used the manual multilocus approach to determine genotypes in a hybrid zone between WCT and RBT in Washington, and observed that F1 and RBT genotypes were more common near known stocking locations of RBT. Manually classifying each individual based on homozygous or heterozygous alleles does not consider the other individuals captured at a site, whereas NEWHYBRIDS gives weight to the proportion of individuals from each of the different genotypes (e.g., the manual approach would classify an individual that was heterozygous at all loci as a F1, but NEWHYBRIDS would weight the probability of the individual being an F1 based on the presence of other F1s in the sample). The prevalence of pure WCT genotypes and WCT skewed genotype distributions in our study, at sites upstream of the suspected RBT source, could have been the result of a past hybridization event that is slowly being diluted by

56 backcrossing of hybrids with pure WCT individuals. However, we agree with Rubidge and Taylor (2004) that there was likely not widespread hybridization prior to the construction of the Koocanusa Reservoir, because it would have probably been detected by Leary et al. (1987). A more likely scenario is that the distribution of hybrids we found was the result of sites near the Koocanusa Reservoir being a source of RBT alleles that then spread to upper elevation sites. Our estimates of the proportion of genotypes are likely biased because hybridization, although recent, has likely been occurring for > 3 generations (since at least 1999 and possibly as early as 1988 or earlier). The average rate of maturity for WCT is 4 years (range 2-6 years, Behnke 1992) which suggests that at least 4-5 generations of inbreeding may have occurred. This violates the assumption of NEWHYBRIDS that only two generations of inbreeding have occurred. However, because it is apparent that hybridization is not complete (i.e., hybrid swarms have not developed), NEWHYBRIDS is still able to differentiate between most genotype classes. It is likely however, that we have overestimated the number of WCT_bx and F2 individuals in the populations. These individuals may belong to later generation backcrosses (F3 and greater) and backcrosses between > F2s and other pure and hybrid types. This violation of the model assumptions will also cause an overestimation of the proportion of pure WCT and therefore, our estimates of hybridization levels are an underestimate of the true amount present.

57 Sample Bias We did not test for differences in capture rates between WCT and RBT which would have required a double sampling technique to estimate the effectiveness of angling to capture the two species (Eberhardt and Simmons 1987). If RBT were easier to capture it likely that their hybrids may also have been more easily captured compared to pure WCT based on hybrids having intermediate characteristics of the parental types (Arnold 1997). We conducted some snorkel surveys on Gold Creek in 2003 and observed 4.9% RBT (5/103) in a 600 m reach (S. Bennett unpublished data). This conforms very closely to the 5.5% RBT detected using genetic analysis at this site (Table 5). We also snorkel surveyed Michel Creek in the mid and upper reach where we collected genetic samples and did not observe any RBT and none were detected via genetic analysis (Table 5). These results suggest that our samples collected via angling were likely representative of the populations at these sites. Hybrid Movement It appears that pure RBT are not spreading RBT alleles to sites higher in the watershed, because after eight years of sampling we have not observed any pure RBT > 90 km from the reservoir. Hitt et al. (2003) and Rubidge and Taylor (2004) have speculated that the lack of pure RBT and F1 individuals at higher elevations indicates that hybrids are likely spreading RBT alleles upstream. Evidence to support this speculation was found in the Flathead drainage in Montana by a recent study (Boyer 2006). Using a Bayesian admixture model,

58 Boyer (2006) determined that the source of RBT alleles in upstream populations was a low elevation hybrid swarm (Abott Creek). Boyer (2006) also showed that two models of dispersal were likely represented: stepping-stone (Kimura and Weiss 1964) and continent-island (Wright 1931) dispersal. In the stepping-stone model, RBT alleles spread in an upstream direction to neighboring sites, the level of introgression decreases with distance from the source, and F1 and RBT types are confined to sites near the source. The island-continent model assumes equal probability of dispersal from a source to all subpopulations, and F1 individuals move from the source to non-adjacent sites via long distance dispersal (Boyer 2006). Both of these models may present an accurate representation of hybridization in the upper Kootenay watershed. There is a general trend for the % RBT alleles and the proportion of hybrid genotypes to decrease as sites become further from Koocanusa Reservoir (i.e., suggesting stepping-stone model). There are also some situations where F1 and RBT types were found at a site with no other F1 or RBT types present at adjacent sites, such as at lower Skookumchuck Creek and upper Alexander Creek (i.e., continent-island model).This suggests that RBT alleles may be moving throughout the watershed via a variety of dispersal processes that may be difficult to predict. Despite most RBT being restricted to lower elevation areas, the movement of RBT alleles upstream via F1 and backcross hybrids suggests that putatively pure WCT sites below migration barriers, and sites with currently low levels of RBT alleles are vulnerable. The potential therefore exists for these sites to develop into hybrids swarms in the future if stream conditions for either hybrids or

59 pure RBT are favorable. There is little evidence that introgression of RBT alleles will not continue in the absence of physical barriers. Physical migration barriers have been the only consistent protection for pure populations in this and other studies (Hitt et al. 2003; Van Houdt et al. 2005; Ostberg and Rodriguez 2006). There is also strong evidence that hybrids between WCT and RBT are not significantly less fit than parental types at our study area (Rubidge and Taylor 2004). However, even when F1 hybrids are less fit than parental types, successful backcrossing between F1s and parental types can lead to the spread of introgression, and eventually hybrid swarms will form (Epifanio and Philipp 2001; Wolf et al. 2001; Ostberg and Rodriguez 2006). Level of Introgression and Age of the Hybrid Events Sites with > 20% introgression were relatively rare and exclusively confined to low elevation sites near a known source of RBT (i.e., Koocanusa Reservoir). This mirrored the distribution of the % RBT alleles found in the previous study in the Upper Kootenay River (Rubidge 2003). However, sites within the USA tend to have higher levels of hybridization. For example, the percent of sites with > 30% RBT alleles in our study was 9% compared to 16% in Montana (Boyer 2006), 33% in Washington (Ostberg and Rodriguez 2006), and 51% in Idaho (Weigel et al. 2003). This could be related to the length of time since the successful introduction of RBT (i.e., the time since RBT were introduced, survived, and reproduced) among studies. Unfortunately, when RBT establish and start to reproduce is not typically known. Rubidge and Taylor

60 (2004) attributed significant heterozygote deficiencies and linkage disequilibrium at low elevation sites as evidence of recent hybridization in the Upper Kootenay (see also Forbes and Allendorf 1991). When a site becomes a random mating hybrid swarm there should be no associations among parental alleles (Jiggins and Mallet 2000). The absence of linkage disequilibrium at many of the upper elevation sites is, however, not necessarily evidence that the hybridization at these sites is not recent because linkage disequilibrium is harder to detect, or absent, when RBT alleles that are spread by backcross and F2 individuals (Rubidge and Taylor 2004) or when RBT alleles are rare compared to WCT alleles as was the case in many of our populations. Also, we used a Bonferroni correction on the multiple pairwise comparisons for linkage disequilibrium and Moran (2003) and Garcia (2004) warn that only very strong effects will be significant when using Bonferroni corrections. Long-term monitoring of one low elevation site near the Koocanusa Reservoir (lower Gold Creek) indicates that the level of introgression is not changing in low elevation sites but may be increasing at mid-elevation sites within 90 km of the Koocanusa Reservoir (Chapter 4). This suggests that there are opposing forces maintaining this situation. Rubidge and Taylor (2004) suggested that a continual influx of RBT from the Koocanusa Reservoir to the lower Gold Creek site may be maintaining the high level of introgression. We suspect that it is also the continual addition of pure WCT from upstream areas that may be keeping these sites from turning into complete hybrid swarms. Ostberg and Rodriguez (2006) described a similar situation where pure RBT

61 above a migration barrier may have been moving downstream and hybridizing with native WCT and maintaining a hybrid swarm. It is likely that some pure WCT wash over barriers (such as the Bull River and Elk River hydro dams) and maintain the bimodal distribution of WCT and RBT with a mixture of hybrid types we observed. The apparent increase of introgression upstream of these bimodal, introgressed populations indicates they may be acting as a source of RBT alleles that are spreading upstream. The pattern of introgression we found (i.e., few hybridized high elevation sites) also suggests that RBT stocked in high elevation lakes prior to the formation of the Koocanusa Reservoir were not able to escape the lakes, unable to spawn at high elevations, or moved down stream until they found suitable conditions. Paul and Post (2001) observed that RBT stocked in high elevation lakes only established populations in low elevations stream reaches in Alberta. However, we did find at least one exception to this trend in lower Alexander Creek (Figure 2). Here the highest level of introgression was found upstream (near Summit Lake the presumed source of RBT). This indicates that there may be some suitable habitat for RBT at high elevation sites. Management Implications If hybrid swarms have developed (i.e., no individuals in the population are pure WCT) there are no management options for restoring pure WCT populations besides complete removal of the fish and replacement with pure individuals from another source (Leary et al. 1995). In order to choose the correct

62 management action, techniques are required to correctly identify individuals in a hybrid zone into classes (e.g., pure WCT or RBT, F1, F2, backcross, etc.). However, when two species have been interbreeding for several generations there are high misclassification rates associated with distinguishing between backcross and pure individuals (Boecklen and Howard 1997). For example, if we assume that hybridization has been occurring for at least four generations in the Upper Kootenay River, the probability of misclassifying a third generation backcross as a pure WCT is 0.6 when using a sample size of 30 and 7 diagnostic markers (Boecklen and Howard 1997). This is a conservative estimate of the misclassification rate because Boecklen and Howard (1997) assumed that backcrossed individuals would only cross with parental types and not other backcrosses. This means that at sites where we found 80% pure WCT, the true number of pure WCT is at best 48% (i.e., 0.6 misclassification rate x 80% observed pure WCT). Conclusion A broad hybrid zone has developed in the Upper Kootenay River that appears to be primarily a result of recent RBT stocking in the Koocanusa Reservoir. Sites near the reservoir have high levels of introgression and it may be only the addition of pure WCT from other sites (upstream) that are preventing complete hybrid swarms from developing. Open sites further from the reservoir are likely to continue to have increased levels of introgression and may become hybrid swarms due to a combination of stepping-stone movements from areas of

63 high introgression to adjacent sites, and long-distance migration of F1 and RBT from areas of high introgression to non-adjacent sites. It appears that the genetic integrity of the supposed strong hold of WCT at the northern periphery has been diminished, but the spread of hybridization is relatively recent and not complete. Allendorf et al. (2001a) suggest that an emphasis be placed on identifying and protecting the remaining pure populations in areas with “widespread introgression” between native and introduced species. This study has identified 14 putatively pure populations of WCT, although only 4 are protected from all sources of RBT. Other management priorities should be the further identification and elimination of the current sources of RBT, and studies to better understand the population dynamics of WCT and how the movement of pure WCT may influence levels of introgression. Literature Cited Aljanabi, S. M., and I. Martinez. 1997. Universal and rapid salt-extraction of high quality genomic DNA for PCR-based techniques. Nucleic Acids Research 25(22):4692-4693. Allendorf, F. W., D. M. Espeland, D. T. Scow, and S. Phelps. 1980. Coexistence of native and introduced rainbow trout in the Kootenai River drainage. Proceedings of the Montana Academy of Sciences 39:28–36. Allendorf, F. W., and R. F. Leary. 1988. Conservation and distribution of genetic variation in polytypic species, the cutthroat trout. Conservation Biology 2(2):170-184. Allendorf, F. W., R. F. Leary, P. Spruell, and J. K. Wenburg. 2001. The problems with hybrids: setting conservation guidelines. Trends in Ecology and Evolution 16(11):613-622.

64 Allendorf, F. W., and R. S. Waples. 1996. Conservation and genetics of salmonid fishes. In J.C. Avise and J.L Hamrick (eds). Conservation genetics: case histories from nature. Chapman and Hall, New York, NY:238-280. Anderson, E. C., and E. A. Thompson. 2002. A model-based method for identifying species hybrids using multilocus genetic data. Genetics 160:1217-1229. Arnold, M. L. 1997. Natural hybridization and evolution. Oxford University Press, Oxford, UK. Behnke, R. J. 1992. Native trout of western North America. American Fisheries Society, Monograph 6. Behnke, R. J. 2002. Trout and salmon of North America. Simon and Schuster Inc., New York. Bennett, S. N. 2002. Westslope Cutthroat Trout: Identified Wildlife Species Accounts. Prepared for Identified Wildlife Management Strategy, Water, Land, and Air Protection, Victoria, British Columbia. Boecklen, W. J., and D. J. Howard. 1997. Genetic analysis of hybrid zones: number of markers and power of resolution. Ecology 78(8):2611-2616. Boyer, M. C. 2006. Rainbow trout invasion and the spread of hybridization with native westslope cutthroat trout. Master's Thesis, University of Montana, Missoula. Brown, K. H., S. J. Patton, K. E. Martin, K. M. Nichols, R. Armstrong, and G. H. Thorgaard. 2004. Genetic analysis of interior Pacific Northwest Oncorhynchus mykiss reveals apparent ancient hybridization with westslope cutthroat trout. Transactions of American Fisheries Society 133:1078-1088. Campbell, M., and C. Cegelski. 2003. An assessment of hybridization and introgression between westslope cutthroat trout and both native and introduced trout in the Middle Fork Salmon River, Idaho: conservation and management implications. IDFG Report Number 03-49. Boise, Idaho. Campton, D. E. 1987. Natural hybridization and introgression in fishes. Chapter 7 In Ryman, N. and Utter, F. (eds.) Population genetics and fisheries management, University of Washington Press, Seattle.

65 Carmichael, G. J., J. N. Hanson, M. E. Schmidt, and D. C. Morizot. 1993. Introgression among Apache, cutthroat, and rainbow trout in Arizona. Transactions of American Fisheries Society 122:121-130. Coffman, J. S. 2005. Evaluation of a predictive model for upstream fish passage through culverts. Masters Thesis, Department of Biology, James Madison, University. Costello, A. B. 2006. Status of the westslope cutthroat trout (Oncorhynchus clarkii lewisii) in Alberta. Prepared for Alberta Sustainable Resource Development (SRD) and Alberta Conservation Association (ACA). Alberta Wildlife Status Report No. 61:47 p. Costello, A. B., and E. Rubidge. 2004. DRAFT COSEWIC status report on cutthroat trout (Oncorhynchus clarki subspp.). Prepared for the Committee on the Status of Endangered Wildlife in Canada. Crossman, E. J. 1991. Introduced freshwater fishes: a review of the North American perspective with emphasis on Canada. Canadian Journal of Fisheries and Aquatic Sciences 48(Supplement 1):46-57. Eberhardt, L. L., and M. A. Simmons. 1987. Calibrating population indices by double sampling. Journal of Wildlife Management 51(3):665-675. Epifanio, J. M., and D. P. Philipp. 2001. Simulating extinction of parental lineages from introgressive hybridization: the effects of fitness, initial proportions of parental taxa, and mate choice. Reviews in Fish Biology and Fisheries 10(3):339-354. Forbes, S. H., and F. W. Allendorf. 1991. Associations between mitochondrial and nuclear genotypes in cutthroat trout hybrid swarms. Evolution 45(6):1332-1349. Garcia, L. V. 2004. Escaping the Bonferroni iron claw in ecological studies. Oikos 105(3):657-663. Hanson, D. L. 1977. Habitat selection and spatial interaction in allopatric and sympatric populations of cutthroat and steelhead trout. Doctoral dissertation. University of Idaho, Moscow. Hitt, N. P., C. A. Frissell, C. C. Muhlfeld, and F. W. Allendorf. 2003. Spread of hybridization between native westslope cutthroat trout, Oncorhynchus clarki lewisi, and nonnative rainbow trout, Oncorhynchus mykiss. Canadian Journal of Fisheries and Aquatic Sciences 60:1440-1451.

66 Holm, S. 1979. A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics 6:65-70. Jansson, H., and T. Ost. 1997. Hybridization between Atlantic salmon (Salmo salar) and brown trout (S-trutta) in a restored section of the River Dalalven, Sweden. Canadian Journal of Fisheries and Aquatic Sciences 54(9):2033-2039. Jiggins, C. D., and J. Mallet. 2000. Bimodal hybrid zones and speciation. Trends in Ecology and Evolution 15(6):251-255. Kanda, N., R. F. Leary, and F. W. Allendorf. 2002a. Evidence of introgressive hybridization between bull trout and brook trout. Transactions of the American Fisheries Society 131(4):772-782. Kanda, N., R. F. Leary, P. Spruell, and F. W. Allendorf. 2002b. Molecular genetic markers identifying hybridization between the Colorado River-greenback cutthroat trout complex and Yellowstone cutthroat trout or rainbow trout. Transactions of the American Fisheries Society 131(2):312-319. Kimura, M., and G. H. Weiss. 1964. The stepping stone model of population structure and the decrease of genetic correlation with distance. Genetics 49:561-576. Kohler, C. C., and W. R. J. Courtenay. 2003. American Fisheries Society Position on Introductions of Aquatic Species. web address http://www.afsifs.vt.edu/afspos.html. Kondratieff, M. C., and C. A. Myrick. 2006. How high can brook trout jump? A laboratory evaluation of brook trout jumping performance. Transactions of the American Fisheries Society 135(2):361-370. Kruse, C. G., W. A. Hubert, and F. J. Rahel. 2000. Status of Yellowstone cutthroat trout in Wyoming waters. North American Journal of Fisheries Management 20:693-705. Leary, R., F. Allendorf, and G. Sage. 1995. Hybridization and introgression between introduced and native fish. Pages 91-101 in H.L. Schramm, Jr., and R.G. Piper, editors. Uses and effects of cultured fishes in aquatic ecosystems. American Fisheries Society, Symposium 15, Bethesda, Maryland. Lewis, P. O., and D. Zaykin. 2001. Genetic Data Analysis: Computer program for the analysis of allelic data. Version 1.0 (d16c). Free program distributed

67 by the authors over the internet from http://lewis.eeb.uconn.edu/lewishome/software.html. Loftus, A., and C. Flather. 2000. Fish and other aquatic resource trends in the USA: A technical document supporting the 2000 USDA Forest Service RPA assessment. Gen. Tech. Rep. RMRS-GTR-53. USDA, Forest Service, Fort Collins, Colorado. Mayhood, D. 2000. Provisional evaluation of the status of westslope cutthroat trout in Canada. At risk proceedings of a conference on the biology and management of species and habitats at risk. Feb. 15-19, 1999, Kamloops, British Columbia. MFWP. 2001. Stocking records for Koocanusa Reservoir. Montana Department of Fish, Wildlife, and Parks, Libby, Montana. Miller, M. 1997. RxC program. Available at: http://www.marksgeneticsoftware.net. Miller, R. R., J. D. Williams, and J. E. Williams. 1989. Extinctions of North American fishes during the past century. Fisheries 14(6):22-38. Moran, M. D. 2003. Arguments for rejecting the sequential Bonferroni in ecological studies. Oikos 100(2):403-405. MWLAP. 2006. Regional fish stocking records: Kootenay Region, 2006. Ministry of Water, Land and Air Protection, Victoria, British Columbia. Myers, N., and A. H. Knoll. 2001. The biotic crisis and the future of evolution. Proceedings of National Academy of Sciences 98:5389–5392. Novinger, D. C., and F. J. Rahel. 2003. Isolation management with artificial barriers as a conservation strategy for cutthroat trout in headwater streams. Conservation Biology 17(3):772-781. Ostberg, C., S. Slatton, and R. Rodriguez. 2004. Spatial partitioning and asymmetric hybridization among sympatric coastal steelhead trout (Oncorhynchus mykiss irideus), coastal cutthroat trout (O. clarki clarki) and interspecific hybrids. Molecular Ecology 13:2773–2788. Ostberg, C. O., and R. J. Rodriguez. 2004. Bi-parentally inherited speciesspecific markers identify hybridization between rainbow trout and cutthroat trout subspecies. Molecular Ecology Notes 4(1):26-29. Ostberg, C. O., and R. J. Rodriguez. 2006. Hybridization and cytonuclear associations among native westslope cutthroat trout, introduced rainbow

68 trout, and their hybrids within the Stehekin River drainage, North Cascades National Park. Transactions of the American Fisheries Society 135(4):924-942. Paul, A. J., and J. R. Post. 2001. Spatial distribution of native and nonnative salmonids in streams of the Eastern Slopes of the Canadian Rocky Mountains. Transactions of American Fisheries Society 130:417-430. Pimentel, D., L. Lach, R. Zuniga, and D. Morrison. 2000. Environmental and economic costs of non-indigenous species in the United States. Bioscience 50:53-65. Platts, W. S. 1979. Relationships among stream order, fish populations, and aquatic geomorphology in an Idaho river drainage. Fisheries 4:5-9. Rhymer, J. M., and D. Simberloff 1996. Extinction by hybridization and introgression. Annual Review of Ecology and Systematics 27:83-109. Richter, B. D., D. P. Braun, M. A. Mendelson, and L. L. Master. 1997. Threats to imperiled freshwater fauna. Conservation Biology 11(5):1081-1093. Rubidge, E. 2003. Molecular analysis of hybridization between native westslope cutthroat trout (Oncorhynchus clarki lewisi) and introduced rainbow trout (O. mykiss) in southeastern British Columbia. Master's thesis, Department of Zoology, University of British Columbia, Vancouver. Rubidge, E., P. Corbett, and E. B. Taylor. 2001. A molecular analysis of hybridization between native westslope cutthroat trout and introduced rainbow trout in southeastern British Columbia, Canada. Journal of Fish Biology 59:42-54. Rubidge, E. M., and E. B. Taylor. 2004. Hybrid zone structure and the potential role of selection in hybridizing populations of native westslope cutthroat trout (Oncorhynchus clarki lewisi) and introduced rainbow trout (Omykiss). Molecular Ecology 13(12):3735-3749. Rubidge, E. M., and E. B. Taylor. 2005. An analysis of spatial and environmental factors influencing hybridization between native westslope cutthroat trout (Oncorhynchus clarki lewisi) and introduced rainbow trout (O. mykiss) in the upper Kootenay River drainage, British Columbia. Conservation Genetics 6(3):369-384. Shepard, B. B., B. E. May, and W. Urie. 2005. Status and conservation of westslope cutthroat trout within the western United States. North American Journal of Fisheries Management 25(4):1426-1440.

69 Strahler, A. N. 1957. Quantitative analysis of watershed geomorphology. Transactions of American Geophysical Union 38:913-920. Taylor, E. B. 2004. Evolution in mixed company: inferences from studies of natural hybridization in Salmonidae. Pages 232-263 In A.P. Hendry and S. Stearns (eds) Evolution illuminated: salmon and their relatives. Oxford Press, Oxford, United Kingdom. Thompson, P., and F. Rahel. 1998. Evaluation of artificial barriers in small rocky mountain streams for preventing the upstream movement of brook trout. North American Journal of Fisheries Management 18:206-210. Utter, F. M. 2000. Patterns of subspecific anthropogenic introgression in two salmonid genera. Reviews in Fish Biology and Fisheries 3:1-15. Van Houdt, J. K. J., J. Pinceel, M. C. Flamand, M. Briquet, E. Dupont, F. A. M. Volckaert, and P. V. Baret. 2005. Migration barriers protect indigenous brown trout (Salmo trutta) populations from introgression with stocked hatchery fish. Conservation Genetics 6(2):175-191. Verhoeven, K. J. F., K. L. Simonsen, and L. M. McIntyre. 2005. Implementing false discovery rate control: increasing your power. Oikos 108(3):643-647. Warren, M. L., and B. M. Burr. 1994. Status of freshwater fishes of the United States: overview of an imperiled fauna. Fisheries 19(1):6-18. Weigel, D. E., J. T. Peterson, and P. Spruell. 2003. Introgressive hybridization between native cutthroat trout and introduced rainbow trout. Ecological Applications 13(1):38-50. Welcomme, R. L. 1992. A history of international introductions of inland aquatic species. ICES marine science symposia 194:3-14. WFL. 2003. Elk River westslope cutthroat trout radio telemetry study: 2000-2002. Prepared by Westslope Fisheries Ltd. Prepared for Columbia-Kootenay Fisheries Renewal Partnership, Cranbrook, British Columbia. Whatley, M. 1972. Effects on fish in the Kootenay River of construction of Libby Dam. Fisheries Management Report No. 65. Fish Habitat Protection Section, Fish and Wildlife Branch, Dept. of Recreation and Conservation, Victoria, BC.

70 Wilcove, D. S., D. Rothstein, J. Dubow, A. Phillips, and E. Losos. 1998. Quantifying threats to imperiled species in the United States. Bioscience 48(8):607-615. Wolf, D. E., N. Takebayashi, and L. H. Rieseberg. 2001. Predicting the risk of extinction through hybridization. Conservation Biology 15(4):1039-1053. Wright, S. 1931. Evolution in Mendelian populations. Genetics 16:97–159. Wydoski, R. S., and R. R. Whitney. 2003. Inland fishes of Washington. American Fisheries Society, Second Edition, Bethesda, Maryland.

71 Table 2. Summary of sample site characteristics, distance from the Koocanusa Reservoir, and the number of fish captured from 1999-2006. Only fish that were successfully diagnosed at > 2 diagnostic nuclear loci are included in the sample sizes.

StreamName Alexander Creek Alexander Creek Alki Creek Bloom Creek Bull River Bull River Caven Creek Coal Creek East White River Elk River Elk River Elk River Findlay Creek Fording River Forsyth Creek Gold Creek Gold Creek Grave Creek Kootenay River Lodgepole Creek Lodgepole Creek Lussier River Lussier River Mather Creek Meachen Creek Meachen Creek Michel Creek Michel Creek Michel Creek Morrissey Creek North White River North White River Perry Creek Sand Creek Sand Creek Simpson River Skookumchuck Creek Skookumchuck Creek St. Mary River St. Mary River Summit Creek Tepee Creek Wheeler Creek Wigwam River Wild Horse River Mean Total

Width Site Closed a b No. /Open Reach (m)

Elev. c (m)

No. Fish Captured Dist d (km) 1999 2000 2001 2002 2003 2004 2005 2006 Total

6.7 4.3 2.5 7.3 80 28.7 4.3 11 11 115 58 26 35 12 10 17.7 4.9 10.3 14 27.5 7.4 35 13 7.8 17.7 16.7 33 22 9 7 40 23.3 8.5 22 15 35 31 18.1 73.5 26 1.5 7 2.5 34 17.5

1280 1455 984 1020 753 1117 1073 1123 1266 777 958 1434 996 1566 1402 835 1060 1187 1169 1005 1301 809 1528 894 990 1062 1170 1282 1435 1016 1295 1449 914 756 1204 1240 923 1031 769 1029 1316 1167 1369 1152 831

99.4 112.7 82.9 26.2 1.5 39.3 26.4 55.6 116.1 5.2 37.9 155.2 111.0 143.3 150.0 6.4 34.1 98.7 226.0 23.6 38.0 78.8 136.0 53.1 86.3 88.0 91.2 103.9 117.0 33.1 119.3 136.2 57.7 1.0 29.0 241.1 86.8 98.9 30.7 103.6 104.6 41.3 107.4 40.0 29.1

22.44 1119.82

80.1

1 2 3 4 6 5 7 8 10 12 13 11 14 15 16 17 18 19 20 23 22 24 25 26 28 27 29 30 31 32 33 34 36 38 37 39 40 41 42 43 44 45 35 21 9

Closed Low Closed Upper Open Low Open Low Open Low Closed Mid Open Low Closed Low Open Low Open Low Closed Mid Closed Upper Closed Mid Closed Upper Closed Low Open Low Open Upper Closed Low Open Upper Open Low Closed Upper Open Low Open Upper Open Mid Open Low Closed Mid Closed Low Closed Mid Closed Upper Closed Low Open Mid Open Upper Open Low Open Low Closed Mid Open Low Open Low Open Mid Open Low Open Upper Closed Upper Open Low Closed Low Open Mid Open Low

30

20 27

67 11

29 29

36

23 19 40 30 28 20

38

29 32 34 23

36

30 30

31 13 6

14

30 7

40

30 28 30 29 28 41 24 30

30

99

31

29 28 25

31

30

30 33 31

214

29 40 33 31

31 56 31 12

30

9 25

195 299 27 30

30 30

34 45 356 1249

158

195

289

272

61

50 27 67 40 29 59 19 40 30 28 20 67 32 34 23 127 30 20 60 30 28 30 29 28 41 24 29 149 25 30 33 31 461 31 12 38 65 33 195 330 27 60 30 34 45

90 2670

a

72 Closed refers to sites isolated from the Koocanusa Reservoir by fish migration

barriers, b Stream width, c Elevation, and d distance from the Koocanusa Reservoir.

Table 3. PCR laboratory protocols for multiplexed loci used to identify WCT, RBT, and hybrids in the Upper Kootenay River, BC. All loci are co-dominant and diagnostic for WCT and RBT (Ostberg and Rodriguez 2004). PCR mix used Qiagen Multiplex mix (commercial) with 1 unit Hot Start DNA Polymerase, 3 mM MgCl2, at pH 8.7. Locus

Primer uM

OCC34

0.2

OCC35

0.1

OCC36

0.4

OCC37

0.2

OCC38

0.2

OCC42

0.2

OM55

0.2

Primer Sequence CCA TAT ACT GTA GTT CAT AGTC AAG AAT GTT GAG GCT AAA ACG GGG TTA AGG TAA GGA TCA GTT GCA TTT TCC TGG GCT GTT AT ACC GTC TGG TGC GCA ACA TT CCG GTG TAT GGG AGC ATT TGA CAT GAT CTG CTC AAC AGG AT AGG TCT CTG GGC CAA TCT AT GCT ATC CTA ATG GTG TGC TT TAA GTG CCC TGT CAA CAA GA GAC GTA CAT GGT ATA TAA CGT CCC CAA AAC ACG TGT TAA TAT AAC CTG CCG ACT TCA ACA AT CAC GCT CAA AAC ACT AGC TT

73 Table 4. Genotype frequency classes and expected proportions of loci with 0, 1, and 2 genes originating from WCT hybridizing with RBT after two generations (Anderson and Thompson 2002).

GFCb WCT RBT F1 F2 WCT_bx RBT_bx a

Expected portion of loci with WCT and 11 12 21 22 1.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 0.00 0.50 0.50 0.00 0.25 0.25 0.25 0.25 0.50 0.25 0.25 0.00 0.00 0.25 0.25 0.50

1 = WCT allele; 2 = RBT allele; 11 = homozygous for WCT, 12 or 21 =

heterozygous, and 22 = homozygous for RBT. b

GFC = genotype frequency class name; WCT = pure WCT; RBT = pure RBT,

F1 = hybrid produced from breeding of pure WCT and pure RBT; F2 = hybrid produced from breeding of two F1 hybrids; WCT_bx = backcrossed WCT produced from breeding of pure WCT and F2 hybrid; RBT_bx = backcross RBT produced from breeding a pure RBT with a F2 hybrid.

74 Table 5. The pooled genotypes of all fish captured by site between 1999-2006 as determined by the NEWHYBRIDS model (Anderson and Thompson 2002). Closed sites are arranged in ascending order from lowest to highest elevation and open sites are arranged in ascending order from closest to furthest from the Koocanusa Reservoir. All samples were analyzed separately with NEWHYBRIDS and then results were pooled across sites that had multiple samples. Site = site number where the sample was collected (see Figure 2 for locations), N = total number of fish captured per site, Genotype = six genotype classes as determined by NEWHYBRIDS (see Table 4 for definitions), %RBT = the percent of RBT alleles by sample site, and Pwr = power to detect 1% RBT alleles at a sample site based on our sample size and the number of markers used.

Percent by Genotype Stream Name/ Type

WCT_ bx

F2

F1

RBT_ bx

RBT

UNK

% RBT

Site

N

WCT

Pwr

Elk

12

20

100.0

0.0

79

Findlay

14

32

100.0

0.0

92

Morrissey

32

30

96.7

Meachen

28

24

100.0

Bull

6

59

100.0

Coal

8

40

90.0

Michel

29

29

93.1

Grave

19

20

95.0

Sand

38

12

100.0

Closed

3.3

10.0 6.9 5.0

Alexander

1

50

78.0

8.0

Michel

30

149

91.3

4.0

Lodgepole

23

28

100.0

Summit

44

27

92.6

Wheeler

35

30

70.0

13.3

8.0

1.3

90

0.0

85

0.0

100

1.3

96

0.5

98

0.7

94

0.0

82

4.0

11.1

96

4.7

1.7

100

0.0

98

3.7

3.7

2.1

98

6.7

10.0

8.0

99

Forsyth

16

23

100.0

0.0

96

Elk

13

67

100.0

0.0

100

Michel

31

25

100.0

0.0

97

75 Table 5. Continued.

Percent by Genotype Stream Name/ Type

Site

N

WCT

Alexander

2

27

92.6

Fording

15

34

100.0

726

93.8

subtotal/mean Closed

WCT_ bx

F2

F1

RBT_ bx

RBT

3.7 2.6

1.1

0.1

0.1

0.0

UNK

% RBT

Pwr

3.7

3.6

97

2.2

0.0

92

1.6

94

Open Sand

37

31

61.3

Bull

5

29

0.0

Elk

11

28

60.7

10.7

Gold

17

127

69.3

2.4

Lodgepole

22

30

33.3

Bloom

4

40

80.0

3.4

7.5

7

19

63.2

Wild Horse

9

45

62.2

37.8

St. Mary

42

195

80.0

13.8

Gold

18

30

86.7

Wigwam

21

34

91.2

45

60

90.0

Mather

26

28

53.6

Perry

36

461

73.8

Lussier

24

30

83.3

Alki

3

67

Meachen

27

41

Skookumchuck

40

65

86.2

Skookumchuck

41

33

90.9

St. Mary

43

330

100.0

East White

10

30

83.3

North White

33

33

87.9

Lussier

25

29

96.6

North White

34

31

100.0

1.6

22.6

9.7

30.6

99

82.8

0.0

97.0

90

25.0

3.6

29.8

98 100

4.7

20.0

Caven

Tepee

5.5

6.5 13.8

7.5

5.5

11.0

22.4

10.0

36.7

37.4

91

2.5

5.9

99

2.5

15.8

6.7

8.6

78

0.0

7.5

97

3.6

2.6

5.0

100

3.3

10.0

2.5

90

2.9

5.9

1.5

94

1.7

1.7

1.1

100

14.3

32.1

11.8

88

3.9

4.1

6.0

98

3.3

13.3

6.8

90

100.0

0.0

100

100.0

0.0

96

1.5

7.2

97

9.1

3.5

92

0.0

100

Kootenay

20

60

90.0

Simpson

39

38

100.0

18.2

21.1

9.2

3.1

6.7

10.0 6.1

8.3

6.0

99

6.1

4.0

92

3.4

1.2

98

0.0

99

1.7

2.1

100

0.2

100

subtotal/mean Open

1944

81.4

7.9

2.8

0.3

0.6

2.6

4.4

11.5

96

Total

2670

84.8

6.5

2.4

0.3

0.5

1.9

3.8

-

95

76 Table 6. Average inbreeding coefficient (FIS) values (a) and p values (b) for each site and sample that significantly deviated from Hardy-Weinberg expectations (Ave.) and for each of the seven loci examined separately (listed by loci).

a) Location Elk River Gold Creek Sand Creek Skookumchuck Creek

Site 1 1 1 1 1

Year 2004 2003 2006 2004 2003

Occ34 0.857 0.529 0.776 0.926 0.634

Occ35 0.740 0.529 0.669 0.926 0.517

Occ36 0.796 0.641 0.501 0.926 0.634

Site 1 1 1 1 1

Year 2004 2003 2006 2004 2003

Occ34 0.00000 0.02969 0.00000 0.00000 0.02188

Occ35 0.00000 0.03281 0.00031 0.00000 0.04125

Occ36 0.00000 0.01063 0.01031 0.00000 0.01688

FIS Occ37 0.796 0.641 0.691 0.761 0.517

Occ38 0.796 0.641 0.669 0.764 0.634

Occ42 0.857 0.641 0.601 0.848 0.517

p values Occ37 Occ38 0.00000 0.00000 0.01250 0.01031 0.00000 0.00031 0.00000 0.00000 0.05188 0.01844

Occ42 0.00000 0.01063 0.00125 0.00000 0.03719

Om55 0.857 0.641 0.770 0.857 0.634

Ave. 0.813 0.606 0.672 0.860 0.580

b) Location Elk River Gold Creek Sand Creek Skookumchuck Creek

Om55 Ave. 0.00000 0.00906 0.00000 0.00000 0.01969

77 Table 7. P values for exact tests of linkage equilibrium for all pairwise combinations of seven nuclear loci using GDS (Lewis and Zaykin 2001) for samples collected between 2001 and 2006. Significant linkage disequilibrium for each pairwise comparison is assumed for p values < 0.0001 based on a Bonferroni adjusted p value (i.e., 0.05/462). Only sites with > 20 fish captured are included in the analysis. A tally of the number of pairwise tests that were significant at p = 0.05 are also included in recognition of fact that Bonferroni corrections maintain a low chance of making a Type I error that results in an increase chance of Type II errors (Moran 2003; Verhoeven et al. 2005). Asterisks indicate streams that deviated from Hardy-Weinberg expectations. Strm Code ALC ALC BLC ELR* EWR GOC* GOC* GRC KOR LUR MIC MIC MIC MIC PEC PEC PEC SAC* SKC* SUC* WHC

Reach Lower Upper Lower Lower Lower Lower Lower Lower Upper Upper Mid Mid Mid Mid Mid Mid Mid Lower Lower Lower Lower

Year 2003 2004 2003 2004 2004 2003 2006 2002 2003 2004 2002 2003 2005 2006 2002 2004 2006 2004 2003 2002 2002

Occ34/ Occ34/ Occ34/ Occ34/ Occ34/ Occ34/ Occ35/ Occ35/ Occ35/ Occ35/ Occ35/ Occ36/ Occ36/ Occ36/ Occ36/ Occ37/ Occ37/ Occ37/ Occ38/ Occ38/ Occ42/ P Occ35 Occ36 Occ37 Occ38 Occ42 Om55 Occ36 Occ37 Occ38 Occ42 Om55 Occ37 Occ38 Occ42 Om55 Occ38 Occ42 Om55 Occ42 Om55 Om55 10 pairs

78 Figure 2. Study site and genetic sample site locations collected in the Upper Kootenay River: 1999-2006. See table 2 for a list of the sample site names. Pie charts represent the percent of WCT (white) and RBT (black) alleles detected at each site. For sites that were sampled in more than one year the charts represent the average. Black bars represent hydro electric dams that are barriers to upstream fish migration. Gray x’s are fish migration barriers > 2 m tall and gray dots represent RBT stocking locations.

79

80 a) 1 .0 0 0 .7 5 0 .5 0 0 .2 5 0 .0 0 W CT

W CT_bx

F2

F1

RBT_bx

RBT

F1

RBT_bx

RBT

F1

RBT_bx

RBT

b) 1 .0 0 0 .7 5 0 .5 0 0 .2 5 0 .0 0 W CT

W CT_bx

F2 c)

1 .0 0 0 .7 5 0 .5 0 0 .2 5 0 .0 0 W CT

W CT_bx

F2

Figure 3. Examples of three genotype distributions identified at sites where hybridization between WCT and RBT has occurred: a) pure WCT, upper Bull River, b) bimodal distribution, lower Gold Creek, and c) unimodal left skewed WCT, Perry Creek. See Table 4 for descriptions of the different genotype categories and Table 5 for the pooled proportion of genotypes at all sites. Bimodal distribution (Gold Creek) is significantly different than unimodal left skewed distribution (Perry Creek, p < 0.000001) based on Fisher’s exact test using RxC (Miller 1997).

Pooled Average % RBT Alleles

81

50

25

0 0

10

20

30

40

50

Rank Distance and Elevation

Figure 4. Spearman’s Rank Correlation of the distance to the Koocanusa Reservoir (black triangles) and elevation (gray squares) versus the pooled average percent of RBT alleles for “open” sites only (n = 27, elevation R2 = 0.39, p = 0.0003, distance R2 = 0.49, p < 0.0001).

82 CHAPTER 3 INFLUENCE OF PROPAGULE PRESSURE AND STREAM CHARACTERISTICS ON INTROGRESSION BETWEEN NATIVE WESTSLOPE CUTTHROAT TROUT AND INTRODUCED RAINBOW TROUT IN BRITISH COLUMBIA 1

Abstract Hybridization between introduced and native salmonids threatens the continued persistence of many inland cutthroat trout species. Invasion theory suggests that introduced species are more likely to establish and spread as the total number of individuals, and the number of times they are introduced (i.e., propagule pressure) increases. We tested the use of propagule pressure and a set of stream characteristics to predict the level of introgression between native westslope cutthroat trout and introduced rainbow trout in the Upper Kootenay River, British Columbia. We used GIS to model the potential propagule pressure of introduced rainbow trout using stocking records and fish migration barrier data. We determined the percent rainbow trout alleles present at 45 sites in 31 different streams using between four and seven co-dominant, diagnostic nuclear markers. The propagule pressure variable we derived using GIS best explained the level of introgression (n = 45, r2 = 0.58, p = 0.0002). This model was also four times more likely to explain introgression than the next best model that included a parameter for the number of stocking sites within 10 km, distance to the 1

Coauthored by Stephen Bennett and John Olson.

83 Koocanusa Reservoir, and a dummy variable for whether a site was above or below a migration barrier. This study used a novel way to describe propagule pressure and found that propagule pressure was able to predict introgression in a human-induced hybrid zone. These results also suggest that introgression is likely to increase throughout the watershed unless sources of rainbow trout are removed. Introduction The introduction of a species to a new environment is often poorly documented, and efforts at developing invasion biology theory have relied primarily on describing the attributes of invasive species and invaded communities (Kolar and Lodge 2001; Peterson 2002). There are typically three general stages of species invasions that have been described (Moller 1996; Vermeij 1996; Kolar and Lodge 2001). First the species is transported to a new habitat (naturally or human aided), then the species establishes itself in the new habitat, and finally the species may integrate into the new environment, become well established, and possibly spread to new areas. Predicting which species are likely to complete these stages is difficult (Kolar and Lodge 2001). Peterson and Fausch (2003) point out that for a native species to suffer declines during the invasion of a non-native species, one or more of the following biotic interactions must occur: a declining reproductive rate, declining survival rate, net emigration, or introduction of disease by the invader. However, when native and non-native species hybridize other factors may threaten the native

84 species. Hybridization refers to the mating between individuals from two genetically distinct entities that may or may not produce viable offspring (Taylor 2004). Introgressive hybridization (hereafter introgression) is defined as hybridization that produces viable offspring and results in the movement of alleles from one genetically distinct entity to another. A population is considered a hybrid swarm when all individuals in a population have alleles from both parental populations and no individuals are pure parental types (Leary et al. 1995). Hybridization is recognized as a natural phenomenon that can play an important role in plant and animal evolution by creating recombinant species and evolutionary novelties (Jiggins and Mallet 2000; Allendorf et al. 2001a; Barton 2001). However, human-induced hybridization can result in wasted reproductive effort for the native species (Kanda et al. 2002a), loss of co-adapted gene complexes (Allendorf and Waples 1996; Gilk et al. 2004), and potentially extinction (Rhymer and Simberloff 1996). Salmonid species are prone to hybridization because many species evolved in allopatry and lack behavioral mechanisms to isolate them during reproduction (Behnke 1992; Allendorf and Waples 1996). Therefore, it is not unexpected that hybridization between native and introduced salmonids is now common because of the widespread stocking of salmonids outside their native range (Welcomme 1992). In the western North America, introduced rainbow trout (RBT - Oncorhynchus mykiss) readily hybridize with native inland cutthroat trout (O. clarki subsp.). Introgression is often considered one of the greatest threats to

85 the persistence of most sub-species of inland cutthroat trout (Allendorf and Leary 1988; Leary et al. 1995; Behnke 2002). Fisheries managers have built numerous models to predict the presence of fish species using stream habitat variables in an effort to minimize the amount of field sampling required to effectively manage populations (Binns and Eiserman 1979; Wesche and Hubert 1989; Bozek and Rahel 1991; Kruse et al. 1997). However, there are comparatively few examples of models for predicting the presence of hybrids between native and introduced trout (but see Kruse 1998; Weigel et al. 2003; Rubidge and Taylor 2005). Despite the limited number of predictive cutthroat hybrid presence/absence models, there are several factors that are consistently cited in species invasion studies that may be associated with the level of introgression between native and non-native salmonids: migration barriers (Adams et al. 2001; Rubidge and Taylor 2005; Ostberg and Rodriguez 2006), flow regime (De Rito 2004), water temperature (De Rito 2004; Ostberg and Rodriguez 2006), gradient (Kruse et al. 1997; Adams et al. 2001), elevation (Kruse et al. 1997; Paul and Post 2001; Weigel et al. 2003), distance to non-native sources, and propagule pressure (Lockwood et al. 2005; Rubidge and Taylor 2005; Wonham et al. 2005a; Duggan et al. 2006; Lambrinos 2006). Many of the above studies have found conflicting results. Some studies have found environmental factors control the spread of hybridization (Weigel et al. 2003, Ostberg et al. 2006), whereas others studies have found that distance to nearest hybridized site predict hybridization presence (Hitt et al. 2003, Rubidge and Taylor 2005). However, many studies have either failed to measure

86 propagule pressure, or have not accounted for propagule pressure well enough to accurately assess its importance. We use the definition of propagule pressure as a combination of the total number of individuals released, and the number of times and places a species was introduced (Carlton 1996b). There is growing evidence that that propagule pressure may be a significant predictor of successful invasions (Drake et al. 2005; Lockwood et al. 2005; Travis et al. 2005; Von Holle and Simberloff 2005; Duggan et al. 2006). It is also evident that when propagule pressure is not accounted for it may confound efforts to model the success of introduced species (Colautti 2005). Propagule pressure is hard to quantify in accidental species introductions, such as ballastwater transfers (Wonham et al. 2005b). However, where government agencies or acclimatization societies have deliberately introduced species, propagule pressure has often been found to be the best predictor of establishment success of birds (Cassey et al. 2004) and mammals (Forsyth and Duncan 2001). Salmonid fish introductions provide a unique opportunity to test the role of propagule pressure in the establishment and spread of introduced species and hybridization, because salmonid introductions are relatively well documented in terms of both the timing, location, and number of propagules (i.e., total number of fish) introduced (Colautti 2005). Our first objective was to develop a novel way to estimate the potential propagule pressure of non-native RBT in the Upper Kootenay River in southeastern British Columbia (BC). Our second objective was to test whether this measure of propagule pressure, and a set of biotic and abiotic variables,

87 could predict the level of introgression between native westslope cutthroat trout (O. c. lewisi) and non-native rainbow trout. This project builds on a preliminary assessment of the genetic status of WCT populations in the Upper Kootenay River (Rubidge et al. 2001). The study area encompasses the Canadian portion of the Kootenay River from its headwaters in Kootenay National Park down to the Canada/U.S. border near Newgate, BC (Figure 5). The Kootenay River is a large order 7th tributary (Strahler 1957) to the Columbia River in southeastern BC with a mean annual discharge of 295.6 m3. This portion of the Kootenay River is in the East Kootenay Region of BC and is bounded by the Rocky Mountains to the east and the Purcell Mountains to the west. The study area is approximately 250 km long and the drainage area is approximately 18,500 km2. Like most watersheds throughout western North America, non-native fish have been stocked extensively throughout the Kootenay drainage, particularly in low elevation lakes (MWLAP 2006). Stocking records are available for RBT in the Upper Kootenay River from 1915 to present. Almost 20 million RBT have been stocked at approximately 114 sites in over 2,500 individual stocking events (MWLAP 2006). The Koocanusa Reservoir is an 170 km long reservoir on the Kootenay River that was formed after the completion of the Libby dam in 1972 at Libby, Montana (Whatley 1972). The dam was located at a bedrock chute which is suspected of historically isolating WCT upstream from RBT in the Lower Kootenay River (Behnke 1992). An attempt to establish WCT in the reservoir by

88 both the USA and Canada failed (B. Westover, Ministry of Water, Land, and Air Protection, personal communication). A policy was then developed to establish Gerrard strain RBT in the reservoir, and between 1986 and 1998 they were stocked in tributaries to the reservoir and in the reservoir itself (MFWP 2001; MWLAP 2006). Stocking RBT in lakes and reservoirs with potential outlets in the Upper Kootenay River was stopped in 1999 and replaced with WCT stocking or stocking of triploid RBT in all but Premier Lake (M. MacDonald, Go Fish BC, personal communication). Hybridization between WCT and RBT was detected as early as 1986 in the Upper Kootenay River (Leary et al. 1987). However, hybridization was only confirmed at one tributary (White River) out of the seven tributaries sampled. A follow-up study in 1999 found hybridization at four of the same seven tributaries sampled in 1986 (Rubidge et al. 2001). The increase in hybridization was attributed to the initiation of a RBT stocking program in the Koocanusa Reservoir (Rubidge et al. 2001). Methods Sample Site Selection and Survey Design This project builds on previous sampling that occurred from 1999 to 2001 (Rubidge 2003). The raw data collected during the previous sampling was combined with samples collected from 2002-2006 to form one comprehensive database on which most analyses were conducted. All sample sites were selected in tributary streams (3-6th order streams) to the Upper Kootenay River in

89 a systematic fashion throughout the watershed at varying distances from the Koocanusa Reservoir, and at a range of elevations within streams. Streams were divided into three reaches of equal length (i.e., lower, mid, and upper elevation), and a representative sample site was selected in a reach based on access and location (i.e., we wanted to collect from low, mid and upper reaches throughout the watershed). We chose this basic design because the previous study had implicated Koocanusa Reservoir as the cause of increased hybridization and other similar studies suggested elevation played a significant role in the distribution of WCT, RBT, and their hybrids (Paul and Post 2001, Rubidge et al. 2001, Weigel et al. 2003, Rubidge and Taylor 2005). Several sample sites were located above known fish migration barriers where presumed pure populations of WCT existed. All sites were classified as either open or closed to fish migration. Sites were considered open if introduced RBT could potentially move from the Koocanusa Reservoir to a site (Figure 5). Streams were considered closed if they were isolated from the Koocanusa Reservoir by a fish migration barrier. Fish Capture Tissue samples were collected and analyzed from 2,670 fish captured in 31 streams at 45 sample sites from 1999 to 2006. All fish were caught and tissues samples collected as per Rubidge et al. (2001). At each sample site we attempted to capture 30 fish. The majority of the fish (85%) were captured by angling, followed by dip nets (12%), electroshocking (2.5%), and minnow traps (0.5%). Multiple age classes were sampled at each site and sample reaches

90 ranged from 1 to 3 km long, to reduce the likelihood of sampling siblings (Weigel et al. 2003). We measured each fish to the nearest 0.5 cm and used a length frequency analysis to estimate the age of each fish (Johnson and Anderson 1974). For all fingerling, juvenile, and adult fish, we clipped a small piece of the lower caudal fin placed it in 1.5 mL of 95% ethanol. Fry were collected whole. All sampling occurred during summer low flow conditions, typically from mid-July to early September. DNA Analysis Descriptions of the laboratory methods can be found in Rubidge and Taylor (2004, 2005) for samples collected from 1999 to 2001, and in chapter 2 for samples collected from 2002-2006. Briefly, we used diagnostic, co-dominant, nuclear loci to differentiate WCT, RBT and their hybrids. Four loci were used for data collected from 1999 to 2001, and seven loci for data collected from 2002 to 2006. Loci were either restriction lengths polymorphisms (RFLPs) or simple sequence repeats (SSR) developed by Baker et al. (2002) and Ostberg and Rodriguez (2002, 2004). Quantifying Introgression We determined the % RBT alleles at each sample site as a measure of introgression (Weigel et al. 2003). We calculated the % RBT alleles as follows:

% RBT admixture = (RBT alleles / 2LN) * 100 (Equation 1)

91 where RBT alleles = the total number of RBT alleles detected at a site, L = the number of loci used, and N = the total number of fish captured at a site. Some sites were visited more than once as part of another study to assess whether the % RBT changed between years (Chapter 4). For all preceding analyses we pooled all samples collected across multiple years. The data were pooled because the %RBT alleles at each site stayed within the same introgression category (see Model Development and Assessment below) over the sample period. Power to Detect Hybridization The power of the experimental design to detect the presence of RBT alleles is equal to β or 1 – α (Kanda et al. 2002b). The following equation was used to calculate α:

α = (1-q)2nx

(Equation 2)

where q = the desired frequency of RBT alleles you wish to detect, n = the number of fish sampled, and x = the number of diagnostic markers. The combination of four markers and a sample size of 30 fish per site equates to a 91% probability of detecting 1% RBT alleles (i.e., 1 – α where α = 0.08963). For seven markers (n = 30) the probability of detecting 1% RBT alleles is 98.5%. When two species have been interbreeding for several generations there are high misclassification rates associated with distinguishing between backcross

92 and pure individuals (Boecklen and Howard 1997). For example, with four markers the probability of misclassifying first generation hybrids as pure WCT is 0.07, and with seven markers the probability is 0.01. The misclassification rate increases rapidly when classifying later generation backcrosses (i.e., BC-2) to 0.25 and 0.15 when using four markers and seven markers respectively. This means that our sampling results were probably an underestimate the true number of RBT alleles present. Analysis of Model Parameters We selected a set of nine independent variables with Pearson correlation coefficients (r2) < 0.5 for inclusion in our models (Table 8). We describe our rationale for using these variables and how we measured them below.

Mean May Water Temperature Competition between salmonids can be affected by water temperature (Fausch 1988; Griffith 1988), and water temperature is an important factor in the timing and location of spawning cutthroat and RBT (Henderson et al. 2000; De Rito 2004). We deployed temperature data loggers (Hobo and MadgeTech T1000 types) at 35 of the fish sample sites from August 2003 to August 2005 and used existing temperature data from another 21 sites to calculate the mean annual May water temperature. May temperature (MayTemp) was used in model building because RBT spawning tends to peak during this time (Ennis 1995; De Rito 2004). We predicted that higher levels of introgression would be positively associated with streams with higher mean MayTemp.

93 Biogeoclimatic Zone Broad generalizations about species richness and presence/absence can be made with information about the general biogeographic setting of an area (Vinson and Hawkins 1998; Hawkins et al. 2003). Ecosystems in BC have been classified using a system known as Biogeoclimatic Zone classification (BEC, Pojar et al. 1987). The BEC system groups ecosystems based on vegetation, soils, topography, and climate factors. Our sample sites fell within five zones ranging from a warm dry ponderosa pine (Pinus ponderosa) dominated zone to a cold and moist engelmann spruce (Picea engelmannii) subalpine fir (Abies lasiocarpa) dominated zone. Our sample size was too small to use all five zones in our models so we grouped the zones into two categories: warm and dry (Ponderosa Pine and Interior Douglas Fir Zones) and cool and moist (Interior Cedar Montane Spruce, Montane Spruce, and Englemann Spruce/Subalpine Fir Zones). We used a dummy variable to represent the BEC data and each sample site was assigned a “1” if it was in a warm dry zone or “0” otherwise. We assumed that these zones would reflect a variety of abiotic factors at the sites (e.g., precipitation, air temperature, water chemistry, etc.) and that sample sites within warm dry BEC zones would have higher levels of introgression because streams in these zones would have more optimal conditions for RBT (Paul and Post 2001).

94 Elevation Elevation has been found to be a significant predictor of the presence/absence of WCT x RBT hybrids (Weigel et al. 2003). The elevation was determined for the starting point of each sample site using 1:50,000 map layers and ARC GIS® (version 9.2). We predicted that elevation would be negatively associated with higher levels of introgression.

Flow Regime RBT typically spawn during the ascending limb of the hydrograph and WCT, and other cutthroat trout on the descending limb (Marotz and Farley 1986; Henderson et al. 2000; Allan 2001; De Rito 2004). This may be an adaptation of inland cutthroat trout to avoid spawning during periods of high stream bed mobility that typically occur during peak flows in the Intermountain West (Moller and Van Kirk 2003). We used regional hydrometric stations to estimate the date of the mean peak discharge for streams where genetic samples were collected. We characterized the flow regime of each site based on the mean number of days from January 1 to the peak discharge during the spawning season (beginning of April to the end of July). If a stream did not have a gauging station, we used the nearest station that was most representative of the watershed (i.e., watershed area, orientation, elevation range, etc.). We predicted that introgression would be negatively related to later peak flow dates.

95 Stream Width Stream width has been found to influence the presence/absence of native WCT and non-native RBT (Weigel et al. 2003), and native cutthroat trout (Kruse et al. 1997) with smaller streams having a higher abundance of pure cutthroat. Stream width typically increases in a downstream direction and is therefore negatively correlated with elevation; however, there can be a large amount of variability in stream width within a watershed (Bozek and Hubert 1992; Kruse et al. 1997). Our sample reaches were relatively long (1-3 km), therefore we used BC Ministry of Water, Land, and Air Protection (MWLAP) existing stream inventory data to assign a mean channel width (m) to each sample site (http://srmapps.gov.bc.ca/apps/fidq/). We predicted that levels of introgression would be higher with increasing stream width.

Migration Barriers Migration barriers may play a critical role in the control of introduced fishes and help limit hybridization between native cutthroat trout and RBT (Thompson and Rahel 1998; Adams et al. 2001; Novinger and Rahel 2003). The BC MWLAP maintains a database of stream obstructions that includes information on the location, height, length, type, and potential to prevent upstream fish migration. This database is the result of extensive efforts by the BC government to synthesize existing fisheries information into a comprehensive database. Also, since the early 1990s the government has worked to develop detailed inventory standards and quality assurance protocols for all fish and wildlife inventories

96 (http://ilmbwww.gov.bc.ca/risc/). We reviewed 1,520 obstructions in the database to determine if they were migration barriers (hereafter barriers). We classified an obstruction as a barrier if they met the following criteria: 1) if the obstruction was previously designated as the upstream limit or fish migration, and 2) any obstruction > 2 m tall that was described as either bedrock, cascade, chute, waterfall, or man made dam. All other obstructions were excluded from any further analysis. We chose these criteria for designating barriers because we felt it would be a conservative approach based on the demonstrated jumping ability of non-anadromous salmonid species which is typically < 2 m (Kondratieff and Myrick 2006). All sample sites were classified as either “open” or “closed” based on whether they were upstream of a barrier or not. Streams were considered open if introduced RBT could potentially move from the Koocanusa Reservoir into the streams (e.g., Gold Creek; Figure 5). Streams were considered closed if they were isolated from the Koocanusa Reservoir by a barrier (e.g., upper Bull River; Figure 5). Sites with other potential sources of RBT upstream (i.e., other stocked lakes) were considered closed if they were isolated from the Koocanusa Reservoir. We coded each site with a dummy barrier variable “0” for closed sites and “1” for open sites. We predicted that open sites and sites below barriers would have higher levels of introgression.

97 Propagule Pressure Invasions of introduced species are often more successful with increased propagule pressure (Williamson 1996; Kolar and Lodge 2001), and recent studies have shown that successful salmonid invaders are stocked more often and in higher numbers (Colautti 2005; Lockwood et al. 2005). We used existing stocking records to determine how many sites had been stocked and how many RBT had been introduced (MWLAP 2006). We then interviewed regional fisheries managers and hatchery staff, and reviewed 1:50,000 topographic maps to determine sites from which RBT could potentially escape. We assumed that lakes without outlets could not contribute to propagule pressure and excluded them from further analysis. For each sample site we calculated four measures of propagule pressure: the distance to the closest stocking site (ClstStkSt), the number of stocking sites within 10 km (StkSt10), the number of stocking events (i.e., the total number of times RBT were stocked at sites) within 10 km (StkEvt10), and the total number of RBT that were stocked within 10 km. These variables were strongly correlated with one another (r2 > 0.5) so we only retained StkSt10 for model building purposes. We also used GIS to derive a composite measure of propagule pressure within the entire watershed. We used the existing stocking records, barrier data, a 1:50,000 stream layer, 90 m resolution digital elevation model data (DEM), and the cost weighted feature in ARC GIS® (version 9.2) to derive this measure. First, a distance raster was created for each stocking site using the cost weighted feature in ARC GIS. This raster provided the stream distance from the stocking

98 site to all other points in the watershed. The fish migration barriers were then incorporated into each distance raster such that any barrier upstream of a stocking site would prevent fish movement beyond that point (i.e., the cost of moving beyond the barrier was essentially infinite), but barriers downstream of a stocking site would not. We then developed a fish movement model based on published data on the distances stocked RBT stray from the original site they were stocked (Figure 6, Cresswell 1981). The movement model restricted the straying distance of a percentage of all the RBT stocked at a particular site (i.e., 60% of all fish stocked at a site would not move beyond 10 km). We assumed that RBT stocked in streams could move downstream unrestricted and upstream as far as the first known fish migration barrier (Adams et al. 2001). Lakes with known inlet and/or outlet streams were classified similarly to stream stocking sites, where fish will be assumed to move downstream unrestricted and upstream as far as the next migration barrier. Next, a raster layer was created to represent the potential amount of propagule pressure associated with each stocking site using following formula:

PP = NoFish * e − x ( DistGrid )

(Equation 3)

where PP = potential propagule pressure for a single stocking site, NoFish = the total number of fish stocked at a site (all years combined), e = 2.718282, x = a constant decay rate for straying fish, and DistGrid = the distance to each stocking site. This equation produced a raster where each pixel represented the number

99 of RBT that could have potentially moved from a stocking site to each pixel within the raster. For example, if we used a decay rate (x) of -0.05 and 10,000 RBT were stocked into a lake, this model would predict that the total number of RBT that could potentially move 10 km upstream or downstream (assuming no barriers) would be 6,065. We then summed the propagule raster layers for all the stocking sites to come up with one composite propagule pressure layer. This final composite layer provided us with an estimate of the maximum potential number of RBT that could have moved to any point in the stream grid from all the stocking sites. We allowed fish to move in both directions simultaneously (i.e., the total number of RBT in the propagule layer was > the total number of fish stocked). This layer is a measure of the relative contribution of stocking sites to the overall propagule pressure in the Upper Kootenay River. We predicted that the level of introgression would be positively associated with high propagule pressure. Model Development and Assessment We used multinomial logistic regression to assess the influence of physical and biological site characteristics on the level of introgression between WCT and RBT. We defined three levels of introgression: no RBT alleles present (zero), < 10% RBT alleles present (low), and > 10% RBT alleles present (high) based on the frequency distribution of RBT alleles by sample site (Figure 7). We estimated a model predicting the lowest ordered value of the response value

100 (zero); therefore, all models predict the probability of no RBT alleles being present. We used the cumulative logit model in SAS (version 9.1) to asses the relationship between the response categories and our independent variables because the cumulative logit model assumes the response categories are ordered. In our case the response categories are naturally ordered from no RBT alleles present to a high proportion of RBT alleles present. The advantages of using an ordered logit model as compared to an unordered logit model are that the coefficients are easier to interpret and the hypothesis tests are more powerful (Allison 1999). We tested for multicollinearity between all independent variables using a Pearson correlation matrix. Only uncorrelated independent variables (r2 values one unit increase in the dependent variable), and as such their predicted effect on introgression would be an exponential function. Therefore, we used natural log transformation to linearize the effect of the GIS derived

101 propagule pressure and the distance to the Koocanusa Reservoir on the dependent variable. All other variables were not transformed. We tested the following 12 candidate models:

•

Biogeoclimatic zone (BEC)

•

Distance to Koocanusa

•

Elevation

•

Elevation and width

•

Flow regime and water temperature

•

GIS propagule pressure

•

Migration barriers

•

Proximity to stocking sites

•

Global Physical (elevation, width, flow regime, temperature, BEC)

•

Global Propagule (distance to Koocanusa Reservoir, number of stocking events within 10 km, GIS derived propagule pressure)

•

Global Propagule without GIS derived propagule pressure

•

Global model (all variables)

The relative plausibility of candidate models was assessed using the information-theoretic approach (Burnham and Anderson 1998). Our sample size was relatively small (n = 45), so we used AICc, a second order variant of AIC for small sample sizes, to assess our candidate models (Burnham and Anderson 1998). The model with the lowest AICc score was considered the best fitting model; however, the relative plausibility of each model was assessed by

102 calculating normalized relative likelihood or Akaike weights (wi). A given wi can be interpreted as the weight of evidence that a given model i is the best model given the data and the set of candidate models assessed. We tested for overdispersion of the data and non-independence between sites by approximating the variance inflation factor c. We estimated c by dividing the goodness of fit X2 value for the global model by the degrees of freedom of the global model (Burnham and Anderson 1998). Model Validation We estimated the performance of the best model (as determined by the lowest AICc score) by using a 10-fold cross-validation technique where one observation was left out and the logistic regression was calculated on the remaining n-1 observations (Efron 1983; Olden et al. 2002). We then assessed the accuracy of the best candidate model by determining the overall classification error rates. The classification error rate is the proportion of observations that were incorrectly classified (Peterson and Dunham 2001). Results Distribution of RBT Alleles Introgression (i.e., presence of RBT alleles) was found at 31/45 (68.9%) of the sites and was generally greater at lower elevations and closer to the Koocanusa Reservoir (Figure 5). Of the 14 sites where no RBT alleles were detected (i.e., suspected pure WCT populations), ten were above known fish

103 migration barriers, but only four were completely isolated upstream of all documented RBT sources. Open sites had a higher average % RBT alleles compared to closed sites (Open sites mean = 11.3, range 0-97, STD = 19.9, n = 27; Closed sites mean = 1.5, range = 0-9.3, STD = 2.8, n = 18; t-test = -2.07, p = 0.022). The majority of sites had < 10% RBT alleles (Figure 7). Propagule Pressure We found records of 87 RBT stocking sites dating back to 1915 that could be accurately located (i.e., able to determine GPS coordinates). Of the 87 sites that we could accurately locate, 52 (59.8%) sites were determined to have outlets (i.e., fish could presumably escape the lake and enter the watershed). The total number of RBT stocked in lakes with outlets was 8,283,793 RBT resulting from 1,165 individual stocking events. The majority of RBT stocked were fry/eyed egg (60%) and one year olds (28%). Very few (0.4%) age > 2 RBT were stocked and no ages were recorded for 12% of the stocked RBT. The median number of RBT stocked per event was 3,500 (mean = 7,110, SD = 11,896, range 2 to 166,698). The majority of stocking sites (70%) and RBT stocked (90%) were below 1200 m (Figure 8). The highest levels of GIS-derived propagule pressure were found in two distinct areas: clustered around and directly upstream of the Koocanusa Reservoir, and around Whiteswan Lake. Potential propagule pressure near the Koocanusa Reservoir and Whiteswan Lake was > 1,000,000 RBT. Most other areas had relatively low GISPP levels (< 100,000). The main areas that had

104 moderate GISPP levels (100,000 to 500,000) were around St. Mary Lake and Summit Lakes. Model Results The best model for predicting the absence of introgression (i.e., zero % RBT alleles) was the model that contained only the GIS derived propagule pressure variable (GISPP; Table 9). The Akaike weight (wi) score (0.798) and model r2 = 0.58 indicates good model fit and relatively high predictive power. The score test for the proportional odds assumption indicated that the model assumptions were not violated (X2 = 1.23, df = 1, p = 0.267). The score tests whether aggregating the three introgression groups into two groups provides similar results and non-significant results suggest that the grouping did not influence the outcome. The data did not appear to be overdispersed based on our approximation of the variance inflation factor (c = 1.13) and samples sites were assumed to be independent. The GISPP model was four times more likely to explain introgression than the next best model that included number of stocking sites within 10 km, distance to the Koocanusa Reservoir, and a dummy variable coded for whether a site was above or below a migration barrier. The negative parameter estimate indicates that an increase in the number of RBT stocked will result in a decreased probability of zero % RBT alleles at a site (i.e., an increased probability of introgression; Table 10). The odds ratio estimate for this model was 0.257 (95% confidence limit range 0.125-0.53). The odds ratio estimate can be interpreted as a 74% decrease in the predicted odds

105 of being in the zero category for every one unit increase in GISPP (i.e., [1 – 0.257] * 100 = 74.3%). A unit increase in this model is on the natural log scale. For ease of interpretation we converted the GISPP measure back to the potential number of RBT and plotted the predicted probability of a site having each of the three levels of introgression (Figure 9). The predicted probability of a site having zero introgression was > 0.95 when the GISPP level was < 3,000 RBT. When the level of GISPP was > 750,000 RBT the predicted probability the site would have a high level of introgression (i.e., > 10%) was > 0.5. We reviewed the sign of the coefficient for each model variable that had a significant p value to determine if the variables influence on the level of introgression equaled our predictions (Table 11). Elevation and distance to the Koocanusa both had positive measured responses indicating the odds of being in the zero category increased with each unit increase in both these variables (i.e., sites with higher elevations and further away from the Koocanusa had lower introgression). The BEC and barrier categories both had negative measured responses indicating that warm, dry sites that were open (below barriers) had lower odds of being in the zero introgression category. These four responses all fit with our original hypotheses. Model Validation The overall error rate of the best model was 0.37 based on ten-fold cross validation. The most common classification error was predicting zero introgression at a site when it was measured as a low introgression site. The

106 model did not predict any zero introgression sites that were measured as high introgression. Discussion Propagule Pressure Our study demonstrated that propagule pressure was a strong predictor of the level of introgression between native WCT and introduced RBT despite incomplete stocking records . Four of the top five models contained only propagule related variables (Table 9). The best model used a GIS-derived measure of how many stocked RBT could potentially move to a site from all stocking sources. The role of propagule pressure in species invasions is intuitively appealing and well supported by ecological theory (reviewed in Lockwood et al. 2005). The larger each population of invading propagules is, the more often they are introduced, and the more widely distributed they are, the less susceptible they will be to extinction due to stochastic environmental and genetic effects (e.g., catastrophic disturbance, reduced genetic diversity, and inbreeding). Therefore, increased propagule pressure can lead to an increased probability that an introduced species will be successful at all stages of invasion (Marchetti et al. 2004; Lockwood et al. 2005; Von Holle and Simberloff 2005). Propagule pressure is also a more parsimonious explanation for a species invasion success than using a combination of species characteristics and abiotic factors, and when propagule pressure is not accounted for it can confound more complex models (Colautti 2005).

107 In the Upper Kootenay River, the stocking of RBT in lakes with outlets peaked in the 1950s, and has averaged about 900,000 a decade since then (MWLAP 2006). If propagule pressure alone was responsible for the introgression levels we observed, Leary et al. (1987) should have found more widespread introgression in the late 1980s. However, Rubidge and Taylor (2005) found evidence that the introgression increased as a result of stocking RBT in Koocanusa Reservoir, which did not begin until 1988. A possible explanation for these findings is that environmental factors prevented most RBT from successfully spawning prior to 1988. Changes in environmental factors that benefit RBT have been suggested as a mechanism for recent increases in hybridization between RBT and Yellowstone cutthroat trout (Henderson et al. 2000; Van Kirk and Jenkins 2005). If environmental factors do limit introgression it does not necessarily weaken the role of propagule pressure. Instead this stresses how continued propagule pressure can lead to situations where species can invade when a “window of opportunity” arises. Presumably the greater the propagule pressure the more likely the presence of invaders (i.e., RBT) will coincide with optimal environmental conditions. Environmental Factors Despite the growing support for the role of propagule pressure as a key factor in introduced species invasions, we recognize that invasions are inherently multivariate in nature (Marchetti et al. 2004). A global model that included a wide variety of environmental and species characteristics was the best predictor of the

108 establishment of introduced aquatic species in watersheds throughout California (Marchetti et al. 2004), and the only model in the top five models that did not included non-propagule pressure related variables in our study was the global model (Table 9). This lends support to the contention that, given the right opportunity, any species can be a successful invader (Moyle and Light 1996a). Several of the other environmental variables we tested had significant correlations with the level of introgression which further support the contention that environmental factors can control the establishment and spread of introduced species as well as the level of introgression between hybridizing species. For example, migration barriers limit the movement of non-native salmonids and protect native populations upstream (Harig et al. 2000; Novinger and Rahel 2003; Van Houdt et al. 2005), and we found most of the pure WCT populations above barriers. Elevation is also negatively correlated with the level of introgression (Paul and Post 2001; Weigel et al. 2003), and we found a decrease in the level of introgression with increasing elevation. This suggests that RBT may have biological limitations that prevent them from occupying high elevation areas. Tests of thermal tolerances of RBT and WCT tend to support this theory. For example, RBT have been found to have higher upper tolerances for water temperature (24 °C) compared to WCT (19 °C), and RBT appear to grow over a wider range of water temperatures (Bear et al. 2007). The differences in temperature tolerances between RBT and WCT may partly explain why RBT are often restricted to lower elevation sites (Paul and Post 2001; Weigel et al. 2003; Rubidge et al. 2001).

109 Of equal concern is how environmental factors affect the fitness of the hybrids between native and introduced species (Leary et al. 1995). It is clear that hybrids between WCT and RBT are fertile and their fitness appears to be equal to that of the parentals (Rubidge and Taylor 2004). There is strong evidence that hybrids are spreading RBT alleles in the Upper Kootenay Watershed (Rubidge et al. 2001) and in other parts of the WCT range (Hitt et al. 2003; Boyer 2006). The implications of these findings are that environmental conditions that restrict RBT to the lower elevation streams are not equally restricting hybrids. This may be due to hybrids having physiological tolerances that are intermediate between the parentals (Arnold 1997; Seiler and Keeley 2007). Presumably WCT_bx individuals and later generation backcrosses would have similar physiological tolerances to pure WCT which could lead to hybrids occurring throughout the watershed if not restricted by migration barriers. Model Limitations Our estimate of RBT propagule pressure is a relative measure because it does not model the contribution of individual stocked RBT. Instead, our model predicts the maximum number of RBT that could move to any site within the Upper Kooteney River from stocking locations assumed to have outlets. This approach allows more RBT to occur in the watershed than were originally stocked because fish can move simultaneously upstream and downstream from a single stocking site. Therefore, the relative propagule pressure we measured is higher than what would have occurred at any one site. Thus, when interpreting

110 the model results, the actual number of RBT required for a site to have a > 0.5 probability of having > 10% introgression is less than the 750,000 we found (Figure 9). It is possible that other sources of RBT were not accounted for in our estimate of propagule pressure. Naturalized populations of RBT and straying of RBT and hybrids from introgressed populations could be contributing to propagule pressure in the watershed. In the Flathead River there is evidence that introgressed populations can spread RBT alleles to other sites via hybrid straying (Hitt et al. 2003; Boyer 2006). Boyer (2006) also found strong evidence in support of a stepping stone model (Kimura and Weiss 1964) and the continentisland model (Wright 1931). When introgression spread by hybrids to adjacent sites it fit the stepping stone model, and when introgression spread by longdistance (> 50 km) dispersal of hybrids it fir the continent-island model was more appropriate. This indicates that not only are original stocking sites a source of propagule pressure in the watershed, but new “source” pools can be created as hybrid individuals disperse from introgressed populations and the process continues upstream or downstream. We relied on government stocking records and fish habitat inventory databases to provide data on the number of RBT stocked and the locations of migration barriers within the watershed. These sources of data are continually being updated and the BC government is still in the process of standardizing data collection methods and synthesizing historic data and new data into a comprehensive database. We suspect there are some historic stocking sites that

111 were not documented in the databases we used and we identified several sites that had been stocked but the location was not recorded. As an example, Rubidge and Taylor (2005) determined that almost 3 million RBT were stocked in the Upper Kootenay River between 1915 and 1998. However, we determined that almost 20 million RBT were stocked between 1915 and 2006 (not including triploids) using the same data source (i.e., BC MWLAP).This discrepancy was likely due to the intensive data synthesis efforts that were occurring during the late 1990s as numerous databases were reviewed and combined by the BC Resource Inventory Standards Committee ( http://ilmbwww.gov.bc.ca/risc/ ). Also complicating our analysis was our reliance on expert opinion to determine if stocking sites had outlets. We know of at least one site (e.g., Summit Lake) that was wrongly assumed to have no outlet. Upon field inspection we found a defined channel indicating that the lake frequently had an outlet and we captured fry in the outlet. An inspection of all stocking sites is required to confirm their outlet status and allow for better assessment of each stocking sites potential role in contributing to the overall RBT propagule pressure. The migration barrier data base is likely accurate for larger streams where the majority of our sampling occurred (i.e., > 3rd order). These larger streams have been inventoried numerous times and it is unlikely that any mainstem barriers have not been identified. However, smaller tributaries are underrepresented in the database and there are likely numerous migration barriers that have not been identified. The effect of this barrier data structure is that propagule

112 pressure in the smaller streams will be overestimated because RBT will not be able to access these streams due to migration barriers. Conclusion Our estimate of relative propagule pressure demonstrates the importance of this measure in salmonid invasion events. This initial attempt provides further evidence of the general importance of propagule pressure and in the difficulty of quantifying it accurately, even when relatively precise records of introductions have been kept. We agree with those calling for more focus to be placed on assessing the role of propagule pressure for these reasons (Lockwood et al. 2005; Verling et al. 2005). Despite the importance of propagule pressure, environmental factors do appear to limit the extent of RBT to lower elevation streams in many parts of the WCT range. However, hybrids appear not to be restricted by environmental factors and growing evidence suggests that hybrids are the main vector for spreading introgression. Our model results suggest that introgression will spread throughout the watershed unless prevented by migration barriers. The recent cessation of RBT stocked in the Upper Kootenay River is certainly a good first step in reducing the threats to the native WCT, but it would be naïve to think that this alone will prevent the further introgression between RBT and WCT. Monitoring of introgression levels and continued efforts to determine and eliminate sources of RBT will be essential for conservation of the

113 remaining pure WCT populations, and potential restoration of those with currently low levels of introgression. Literature Cited Adams, S., C. A. Frissell, and B. E. Rieman. 2001. Geography of invasion in mountain streams: consequences of headwater lake fish introductions. Ecosystems 2001:296-307. Allan, J. 2001. Fisheries investigations in Line Creek. Prepared for Crows Nest Resources Ltd. Pices Environmental Consulting Services Ltd, Red Deer, Alberta. Allendorf, F. W., and R. F. Leary. 1988. Conservation and distribution of genetic variation in polytypic species, the cutthroat trout. Conservation Biology 2:170-184. Allendorf, F. W., R. F. Leary, P. Spruell, and J. K. Wenburg. 2001. The problems with hybrids: setting conservation guidelines. Trends in Ecology and Evolution 16:613-622. Allendorf, F. W., and R. S. Waples. 1996. Conservation and genetics of salmonid fishes. In J.C. Avise and J.L Hamrick (eds). Conservation genetics: case histories from nature. Chapman and Hall, New York, NY:238-280. Allison, P. D. 1999. Logistic regression using the SAS system: theory and application. SAS Institute Inc, Cary, NC. Arnold, M. L. 1997. Natural hybridization and evolution. Oxford University Press, Oxford, UK. Baker, J., P. Bentzen, and P. Moran. 2002. Molecular markers distinguish coastal cutthroat trout from coastal rainbow trout/steelhead and their hybrids. Transactions of the American Fisheries Society 131:404–417. Barton, N. H. 2001. The role of hybridization in evolution. Molecular Ecology 10:551-568. Bear, E. A., T. E. McMahon, and A. V. Zale. 2007. Comparative thermal requirements of westslope cutthroat trout and rainbow trout: implications for species interactions and development of thermal protection standards. Transactions of the American Fisheries Society 136:1113–1121.

114 Behnke, R. J. 1992. Native trout of western North America. American Fisheries Society, Monograph 6. Behnke, R. J. 2002. Trout and salmon of North America. Simon and Schuster Inc., New York. Binns, N. A., and F. M. Eiserman. 1979. Quantification of fluvial trout habitat in Wyoming. Transactions of American Fisheries Society 108:215-228. Boecklen, W. J., and D. J. Howard. 1997. Genetic analysis of hybrid zones: number of markers and power of resolution. Ecology 78:2611-2616. Bozek, M. A., and W. A. Hubert. 1992. Segregation of resident trout in streams as predicted by three habitat dimensions. Canadian Journal of Fisheries and Aquatic Sciences 70:886-890. Bozek, M. A., and F. J. Rahel. 1991. Assessing habitat requirements of young Colorado River cutthroat by use of macrohabitat and microhabitat analyses. Transactions of American Fisheries Society 120:571-581. Burnham, K. P., and D. R. Anderson. 1998. Model selection and inference: a practical information-theoretic approach. Springer-Verlag, New York. Carlton, J. T. 1996. Pattern, process, and prediction in marine invasion ecology. Biological Conservation 78:97-106. Cassey, P., T. M. Blackburn, S. Sol, R. P. Duncan, and J. L. Lockwood. 2004. Global patterns of introduction effort and establishment success in birds. Proceedings of the Royal Society of London Series B-Biological Sciences 271:S405-S408. Colautti, R. I. 2005. Are characteristics of introduced salmonid fishes biased by propagule pressure? Canadian Journal of Fisheries and Aquatic Sciences 62:950-959. Cresswell, R. C. 1981. Post-stocking movements and recapture of hatcheryreared trout released into flowing water-a review. Journal of Fish Biology 18:429-442. De Rito, J. N. J. 2004. Assessment of reproductive isolation between Yellowstone cutthroat trout and rainbow trout in the Yellowstone River, Montana. Master's thesis, Fish and Wildlife Management, Montana State University, Bozeman.

115 Drake, J. M., P. Baggenstos, and D. M. Lodge. 2005. Propagule pressure and persistence in experimental populations. Biology Letters 1:480-483. Duggan, I. C., C. Rixon, and H. J. MacIsaac. 2006. Popularity and propagule pressure: determinants of introduction and establishment of aquarium fish. Biological Invasions 8:377–382. Efron, B. 1983. Estimating the error rate of a prediction rule: improvement on cross-validation. Journal of the American Statistical Association 78:316331. Ennis, G. 1995. Literature Review of the life history requirements and mitigation/compensation strategies for selected fish species in the Peace, Liard, and Columbia River drainages of BC. DFO. Fausch, K. D. 1988. Tests of competition between native and introduced salmonids in streams: what have we learned? Canadian Journal of Fisheries and Aquatic Sciences 45:2238-2246. Forsyth, D. M., and R. P. Duncan. 2001. Propagule size and the relative success of exotic ungulate and bird introductions to New Zealand. American Naturalist 157:583-595. Gilk, S. E., I. A. Wang, C. L. Hoover, W. W. Smoker, S. G. Taylor, A. K. Gray, and A. J. Gharrett. 2004. Outbreeding depression in hybrids between spatially separated pink salmon, Oncorhynchus gorbuscha, populations: marine survival, homing ability, and variability in family size. Environmental Biology of Fishes 69:287-297. Griffith, J. S. 1988. Review of competition between cutthroat trout and other salmonids. American Fisheries Society Symposium 4:134-140. Harig, A. L., K. D. Fausch, and M. K. Young. 2000. Factors influencing success of greenback cutthroat trout translocations. North American Journal of Fisheries Management 20:994-1004. Hawkins, B. A., R. Field, H. V. Cornell, D. J. Currie, J. F. Guegan, D. M. Kaufman, J. T. Kerr, G. G. Mittelbach, T. Oberdorff, E. M. O'Brien, E. E. Porter, and J. R. G. Turner. 2003. Energy, water, and broad-scale geographic patterns of species richness. Ecology 84:3105-3117. Henderson, R., J. L. Kershner, and C. A. Toline. 2000. Timing and location of spawning by nonnative wild rainbow trout and native cutthroat trout in the South Fork Snake River, Idaho, with implications for hybridization. North American Journal of Fisheries Management 20:584-596.

116 Jiggins, C. D., and J. Mallet. 2000. Bimodal hybrid zones and speciation. Trends in Ecology and Evolution 15:251-255. Johnson, D. L., and R. O. Anderson. 1974. Evaluation of a 12-inch length limit on largemouth bass in Philips Lake, 1966-1973. Pages 106-116 in J. L. Funk, editor. Symposium on largemouth bass overharvest and management in small impoundments. Special Publication 3, North Central Division, American Fisheries Society, Bethesda, Maryland. Kanda, N., R. F. Leary, and F. W. Allendorf. 2002a. Evidence of introgressive hybridization between bull trout and brook trout. Transactions of the American Fisheries Society 131:772-782. Kanda, N., R. F. Leary, P. Spruell, and F. W. Allendorf. 2002b. Molecular genetic markers identifying hybridization between the Colorado River-greenback cutthroat trout complex and Yellowstone cutthroat trout or rainbow trout. Transactions of the American Fisheries Society 131:312-319. Kimura, M., and G. H. Weiss. 1964. The stepping stone model of population structure and the decrease of genetic correlation with distance. Genetics 49:561-576. Kolar, C. S., and D. M. Lodge. 2001. Progress in invasion biology: predicting invaders. Trends in Ecology & Evolution 16:199-204. Kondratieff, M. C., and C. A. Myrick. 2006. How high can brook trout jump? A laboratory evaluation of brook trout jumping performance. Transactions of the American Fisheries Society 135:361-370. Kruse, C. G. 1998. Influence of non-native trout and geomorphology on the distributions of indigenous trout in the Yellowstone River drainage of Wyoming. PhD dissertation, Department of Zoology and Physiology, University of Wyoming, Laramie. Kruse, C. G., W. A. Hubert, and F. J. Rahel. 1997. Geomorphic influences on the distribution of Yellowstone cutthroat trout in the Absaroka Mountains, Wyoming. Transactions of American Fisheries Society 126:418-427. Lambrinos, J. G. 2006. Spatially variable propagule pressure and herbivory influence invasion of chaparral shrubland by an exotic grass. 147:327334. Leary, R., F. Allendorf, and G. Sage. 1995. Hybridization and introgression between introduced and native fish. Pages 91-101 in H.L. Schramm, Jr.,

117 and R.G. Piper, editors. Uses and effects of cultured fishes in aquatic ecosystems. American Fisheries Society, Symposium 15, Bethesda, Maryland. Leary, R., F. W. Allendorf, and K. L. Knudsen. 1987. Genetic divergence among populations of westslope cutthroat trout in the upper Kootenay River drainage, BC. Dept of Zoology University of Montana Population Genetics Laboratory Report 87/1. Lockwood, J. L., P. Cassey, and T. Blackburn. 2005. The role of propagule pressure in explaining species invasions. Trends in Ecology and Evolution 20:223-228. Marchetti, M. P., P. B. Moyle, and R. Levine. 2004. Alien fishes in California watersheds: characteristics of successful and failed invaders. Ecological Applications 14:587-596. Marotz, B., and J. Farley. 1986. Instream flow needed for successful migration and rearing of rainbow and westslope cutthroat trout in selected tributaries of the Kootenai River. US Dept. Energy, Bonneville Power Admin., Portland, Oregon. MFWP. 2001. Stocking records for Koocanusa Reservoir. Montana Department of Fish, Wildlife, and Parks, Libby, Montana. Moller, H. 1996. Lessons for invasion theory from social insects. Biological Conservation 78:125-142. Moyle, P. B., and T. Light. 1996. Biological invasions of fresh water: empirical rules and assembly theory. Biol. Cons. 78:149-161. MWLAP. 2006. Regional fish stocking records: Kootenay Region, 2006. Ministry of Water, Land and Air Protection, Victoria, British Columbia. Novinger, D. C., and F. J. Rahel. 2003. Isolation management with artificial barriers as a conservation strategy for cutthroat trout in headwater streams. Conservation Biology 17:772-781. Olden, J. D., D. A. Jackson, and P. R. Peres-Neto. 2002. Predictive models of fish species distributions: a note on proper validation and chance predictions. Transactions of the American Fisheries Society 131:329–336. Ostberg, C. O., and R. J. Rodriguez. 2002. Novel molecular markers differentiate Oncorhynchus mykiss (rainbow trout and steelhead) and the O. clarki (cutthroat trout) subspecies. Molecular Ecology 2:197-202.

118 Ostberg, C. O., and R. J. Rodriguez. 2004. Bi-parentally inherited speciesspecific markers identify hybridization between rainbow trout and cutthroat trout subspecies. Molecular Ecology Notes 4:26-29. Ostberg, C. O., and R. J. Rodriguez. 2006. Hybridization and cytonuclear associations among native westslope cutthroat trout, introduced rainbow trout, and their hybrids within the Stehekin River drainage, North Cascades National Park. Transactions of the American Fisheries Society 135:924-942. Paul, A. J., and J. R. Post. 2001. Spatial distribution of native and nonnative salmonids in streams of the Eastern Slopes of the Canadian Rocky Mountains. Transactions of American Fisheries Society 130:417-430. Peterson, D. P. 2002. Population ecology of an invasion: demography, dispersal, and effects of nonnative brook trout on native cutthroat trout. PhD dissertation, Department of Fishery and Wildlife Biology, Colorado State University, Fort Collins, Colorado. Peterson, J. T., and J. B. Dunham. 2001. Combining inferences from models of capture efficiency, detectability, and suitable habitat to classify landscapes for conservation of threatened bull trout. Conservation Biology 17:10701077. Pojar, J., K. Klinka, and D. V. Meidinger. 1987. Biogeoclimatic ecosystem classification in British Columbia. Forest Ecology and Management 22:119-154. . Rhymer, J. M., and D. Simberloff 1996. Extinction by hybridization and introgression. Annual Review of Ecology and Systematics 27:83-109. Rubidge, E., P. Corbett, and E. B. Taylor. 2001. A molecular analysis of hybridization between native westslope cutthroat trout and introduced rainbow trout in southeastern British Columbia, Canada. Journal of Fish Biology 59:42-54. Rubidge, E. M., and E. B. Taylor. 2005. An analysis of spatial and environmental factors influencing hybridization between native westslope cutthroat trout (Oncorhynchus clarki lewisi) and introduced rainbow trout (O. mykiss) in the upper Kootenay River drainage, British Columbia. Conservation Genetics 6:369-384. Strahler, A. N. 1957. Quantitative analysis of watershed geomorphology. Transactions of American Geophysical Union 38:913-920.

119 Taylor, E. B. 2004. Evolution in mixed company: inferences from studies of natural hybridization in Salmonidae. Pages 232-263 In A.P. Hendry and S. Stearns (eds) Evolution illuminated: salmon and their relatives. Oxford Press, Oxford, United Kingdom. Thompson, P., and F. Rahel. 1998. Evaluation of artificial barriers in small rocky mountain streams for preventing the upstream movement of brook trout. North American Journal of Fisheries Management 18:206-210. Travis, J. M., M. Hammershoj, and C. Stephenson. 2005. Adaptation and propagule pressure determine invasion dynamics: insights from a spatially explicit model for sexually reproducing species. Evolutionary Ecology Research 7:37-51. Van Houdt, J. K. J., J. Pinceel, M. C. Flamand, M. Briquet, E. Dupont, F. A. M. Volckaert, and P. V. Baret. 2005. Migration barriers protect indigenous brown trout (Salmo trutta) populations from introgression with stocked hatchery fish. Conservation Genetics 6:175-191. Van Kirk, R., and A. Jenkins. 2005. Hydrologic alteration in the Upper Teton watershed and its implications for cutthroat trout restoration. Project completion report for Friends of the Teton River, Driggs, Idaho. Verling, E., G. M. Ruiz, L. D. Smith, B. Galil, A. W. Miller, and K. R. Murphy. 2005. Supply-side invasion ecology: characterizing propagule pressure in coastal ecosystems. 272:1249-1256. Vermeij, G. J. 1996. An agenda for invasion biology. Biological Conservation 78:3-9. Vinson, M. R., and C. P. Hawkins. 1998. Biodiversity of stream insects: variation at local, basin, and regional scales. Annual Review of Entomology 43:271293. Von Holle, B., and D. Simberloff. 2005. Ecological resistance to biological invasion overwhelmed by propagule pressure. Ecology 86:3212-3218. Weigel, D. E., J. T. Peterson, and P. Spruell. 2003. Introgressive hybridization between native cutthroat trout and introduced rainbow trout. Ecological Applications 13:38-50. Welcomme, R. L. 1992. A history of international introductions of inland aquatic species. ICES marine science symposia 194:3-14.

120 Wesche, T. A., and W. A. Hubert. 1989. Relation of geomorphology to stream habitat and trout standing stock in small Rocky Mountain streams. In F. Richardson and R.H. Hamre (eds) Wild Trout IV, Proceedings of the symposium, Yellowstone National Park, September 18-19, 1989:170-171. Whatley, M. 1972. Effects on fish in the Kootenay River of construction of Libby Dam. Fisheries Management Report No. 65. Fish Habitat Protection Section, Fish and Wildlife Branch, Dept. of Recreation and Conservation, Victoria, BC. Williamson, M. 1996. Biological invasions. Chapman and Hall, London. Wonham, M. J., M. A. Lewis, and H. J. MacIsaac. 2005a. Minimizing invasion risk by reducing propagule pressure: a model for ballast-water exchange. 3:473-478. Wonham, M. J., M. A. Lewis, and H. J. MacIsaac. 2005b. Minimizing invasion risk by reducing propagule pressure: a model for ballastwater exchange. Frontiers in Ecology and the Environment 3:473-478. Wright, S. 1931. Evolution in Mendelian populations. Genetics 16:97–159.

121 Table 8. Summary statistics for independent variables measured at each sample site (n=45). The variables were grouped into two categories: physical site characteristics and characteristics related to the potential propagule pressure.

Variable Description/Group Physical Mean May water temperature (°C) Biogeoclimatic Zone* Elevation (m) Average day of peak flow Stream width (m)

Abbreviation

MayTemp BEC Elev QAveMaxD Width

Mean 5.35 1,120 152 22.4

Potential Propagule Pressure 270,477 GIS Propagule Pressure (no. fish) GISPP 0.60 Above or below barrier** Barrier 80.1 Distance to Koocanusa Reservoir (km) DistRes 0.4 No. of stocking sites within 10 km StkSt10 * Dummy variables coded 0 for not warm/dry and 1for warm/dry ** Dummy variable coded 0 closed sites and 1 for open sites.

Min

Max

SD

3.04 753 134 1.5

8.50 1,566 168 115.0

1.15 222 10 22.3

1.0 0

1,572,804 1.00 241.1 3

340,856 0.50 54.9 2

122 Table 9. Summary of model selection statistics for evaluating the level of introgression between WCT and RBT in the Upper Kootenay River, BC. Model Selection Statistics*

Model Name

Independent Variables

K

AIC

AICC value

GIS Propagules

GISPP

3

63.75

64.33

0.00

1.000 0.798

Global Propagule

Barrier,DstRes,GISPP,StkSt10

6

64.89

67.10

2.77

0.250 0.200

794

25.02

0.64

0.5385

Global

All variables

11

69.69

77.69

13.35

0.001 0.001

1297

0.13

0.70

0.334

Global Propagule without GISPP DstRes,StkSt10,Barrier

5

77.13

78.67

14.34

0.001 0.001

3356

0.08

0.45

0.2762

Distance to Koocanusa

DstRes

3

79.98

80.57

16.24

0.000 0.000

5843

0.03

0.33

0.1151

Biogeoclimatic Zone

BEC

3

81.09

81.68

17.35

0.000 0.000

65153

0.02

0.31

0.8268

Open or Closed

Barrier

3

85.92

86.50

22.17

0.000 0.000

82002

0.00

0.21

0.7405

Elevation

Elev

3

86.38

86.96

22.63

0.000 0.000 >100000

0.00

0.20

0.1405

Global Physical

BEC,Elev,Flow,Temp,Width

7

85.03

88.06

23.72

0.000 0.000 >100000

0.00

0.38

0.4906

Elevation and Width

Elev,Width

4

88.31

89.31

24.97

0.000 0.000 >100000

0.00

0.20

0.2065

Water temp and flow

Flow,Temp

4

92.82

93.82

29.49

0.000 0.000 >100000

0.00

0.10

0.5296

Proximity to Stocking Sites

StkSt10

3

93.24

93.83

29.50 sum

0.000 0.000 >100000 1.253

0.00

0.04

0.2964

Δ AIC Rel. Like

wi

Evid. %max Ratio wi

2

r

Prop Odds

4 100.00

0.58

0.2672

* AIC - Akaike's Information Criteria; AICC - AIC corrected for small sample size; Δ AIC - difference between model with lowest AIC and every other model; K - number of parameters (including two intercepts); Rel. Like. - exp (-0.5*ΔAICc),

123 relative likelihood of the model, given the data; Evid. Ratio - evidence ratio, ratio of the wI for a given model vs. wi for best model; r2 - maximum rescaled r2, adjusts generalized r2 due to upper bound being < 1; Prop Odds - p value for the scored test for the proportional odds assumption test that the grouping of the response variable did not influence the results.

124 Table 10. Model results for the best multinomial logistic regression model for predicting introgression between WCT and RBT in the Upper Kootenay River, BC. 95% CI for odds ratio Odds Model Estimated Standard 2 Parameter Coefficient Error Wald X P value Ratio Intercept zero 14.611 4.297 11.559 0.001 Intercept low 18.513 4.706 15.477 10% introgression at the different levels of propagule pressure (i.e., potential number of RBT).

130 CHAPTER 4 LEVELS OF INTROGRESSION IN WESTSLOPE CUTHTROAT POPULATIONS EIGHT YEARS AFTER CHANGES TO RAINBOW TROUT STOCKING PROGRAMS IN SOUTHEASTERN BRITISH COLUMBIA Abstract Westslope cutthroat trout (WCT) have suffered declines throughout most of their historic range in part due to hybridization with rainbow trout (RBT). Introgressive hybridization (introgression) between native WCT and introduced RBT in the Upper Kootenay River, British Columbia appears to be recent and is attributed to RBT stocking in Koocanusa Reservoir. In 1998, RBT stocking was stopped or replaced with stocking of triploid RBT in the watershed to prevent further introgression. The goal of this research was to determine the effect of the change in stocking practices on the level of introgression, and the structure of the hybrid zone between WCT and RBT. We monitored 13 sites from 1999 to 2006 using co-dominant, diagnostic nuclear markers. We classified fish as pure WCT if no RBT alleles were detected and hybrids if they had > 1 RBT allele present. We used a Bayesian hybrid model, NEWHYBIRDS, to classify fish into genotypes to assess the structure of the hybrid zone. Sample sites were categorized as open if no fish migration barriers existed between the site and the reservoir, and closed if migration barriers were present between the site and the reservoir. Open sites consistently had more hybrids (16.3%) than closed sites (6.8%; X2 = 21.15, df = 1, p < 0.0001), and no pure RBT were detected at closed sites. Later generation

131 WCT backcross individuals were the most common genotype at closed sites and open sites far from the reservoir (i.e., > 100 km), with a non-significant trend towards decreasing introgression. At open sites near the reservoir introgression levels stayed relatively high (20-30%) and pure RBT and RBT backcross individuals were common. Introgression significantly increased at sites intermediate distances upstream from the reservoir (i.e., 60-100 km). We also assessed the influence of the maximum and minimum discharge ratio (Max/Min Q) on the ratio of WCT to hybrids at three sites with at least 4 years of monitoring data. We found no significant trends between Max/Min Q and WCT/Hybrid (p > 0.21) and suspect that large Max/Min Q (72.5, N = 18, SD = 21.83, range 35.19116.15) may be partly responsible for low levels of introgression in the Upper Kootenay River compared to other parts WCT range. The new stocking program does not appear to have reduced introgression in Upper Kootenay River, at least in the short term (i.e., < decade), and more active management strategies will be required to prevent further introgression and loss of unique WCT populations. Introduction Non-native fish introductions have resulted in many watersheds in the western USA and Canada having more introduced aquatic species than native species (Crossman 1991; Wydoski and Whitney 2003). Salmonid fishes have been introduced more often than almost any group of fishes because of their popularity as sport fish and the ease with which they can be reared and transported. In particular, rainbow trout (RBT), which are native to western North

132 America and eastern Siberia (Behnke 1992), have been introduced more than any other fish species, and are now found on all continents except Antarctica (Welcomme 1992). Salmonid species are prone to hybridization because of behaviors that fail to fully isolate species during reproduction and competition for limited spawning habitat (Behnke 1992; Allendorf and Waples 1996). Therefore, it is not unexpected that hybridization between native and introduced salmonids is now common because of the widespread stocking of salmonids outside their native range. For example, in the western USA and Canada, RBT readily hybridize with native inland cutthroat trout (O. clarki subsp.) which has resulted in introgression (Allendorf and Leary 1988; Leary et al. 1995). Introgression is the “repeated backcrossing of hybrid descendants with a parental line … , resulting in the incorporation of genes from one gene pool into another” (Hallerman 2003). When introgression progresses to where no individuals in a population are derived from the parental species the population is considered a hybrid swarm (Leary et al. 1995). Introgression between non-native RBT and inland cutthroat trout has caused range contractions and extinctions of local populations of all the extant sub-species of inland cutthroat trout (Young 1995; Duff 1996; Behnke 2002), and introgressive hybridization is now considered the largest threat to the persistence of most cutthroat sub-species (Allendorf and Leary 1988; Leary et al. 1995; Behnke 2002). The management of non-native salmonids within native cutthroat trout habitat has begun to change in an effort to slow or reverse the adverse affects

133 from hybridization and competition. Several management techniques are currently being practiced, including reduction of RBT and other non-native stocking programs and/or stocking of native fish (Leary et al. 1995), stocking of triploid (sterile) fish (Kozfkay et al. 2006), isolating native headwater trout populations with construction of migration barriers (Thompson and Rahel 1998), and non-native removal (Moore and Ridley 1984; Kulp and Moore 2000; Ruzycki et al. 2003). Many of these techniques have been used in combination. Complete removal of non-native fish from relatively small streams is difficult and costly (Shepard 2002), and construction of migration barriers has been recognized as only a temporary fix because of the problems associated with effectively isolating small populations and restricting gene flow (Hilderbrand and Kershner 2000; Novinger and Rahel 2003). In this paper we present eight years of monitoring data on introgression between non-native RBT and native westslope cutthroat trout (WCT) after RBT stocking was greatly reduced, or replaced with triploid stocking programs in the Upper Kootenay River, British Columbia (BC). Our specific objectives were to determine 1) the distribution of WCT x RBT genotypes at monitoring sites within the Upper Kootenay River, and 2) if the reduction of RBT stocking and implementation of triploid RBT stocking within the Upper Kootenay watershed have reduced introgression between WCT and RBT. If these management changes were successful, one possible response could be a decrease in introgression from 1999 to 2006. A decrease in introgression may indicate that the levels of hybridization between WCT and RBT were maintained by RBT

134 stocking prior to the change stocking practices. If the level of introgression has not changed, or if introgression has continued to increase, it could indicate RBT have established naturalized populations, and that other management strategies may be required to reduce introgression levels. We also tested if there was a relationship between levels of introgression and a measure of stream maximum/minimum discharge. Lower maximum spawning flow /minimum winter flow discharge ratios can increase RBT spawning success and subsequently lead to more hybridization between RBT and native cutthroat trout (Moller and Van Kirk 2003). Study Area The study area encompasses the Canadian portion of the Kootenay River from its headwaters in Kootenay National Park down to the Canada/U.S. border near Newgate, BC (Figure 10). The Kootenay River is a large order 7th tributary (Strahler 1957) to the Columbia River in southeastern BC with a mean annual discharge of 295.6 m3. This portion of the Kootenay River is in the East Kootenay Region of BC and is bounded by the Rocky Mountains to the east and the Purcell Mountains to the west. The study area is approximately 250 km long and the drainage area is approximately 18,500 km2. Like most watersheds throughout western North America, non-native fish have been stocked extensively in lakes throughout the watershed (MWLAP 2006). Since 1915, almost 20 million RBT have been stocked into the Upper Kootenay River in 2,531 separate stocking events (MWLAP 2006).

135 Background Hybridization between WCT and RBT was first detected in 1986 in the Upper Kootenay River (Leary et al. 1987). However, hybridization was only confirmed at one tributary (White River) out of the seven tributaries sampled. A follow-up study in 1999 found hybridization in four of the same seven tributaries sampled in 1986 (Rubidge et al. 2001). The spread of hybridization was attributed to the initiation of a RBT stocking program in the Koocanusa Reservoir beginning in 1988. Koocanusa Reservoir was created by the construction of Libby dam in 1972 on the Kootenay River at Libby, Montana. The dam was created at a large bedrock chute which is believed to have isolated WCT upstream from RBT in the Lower Kootenay River for several thousand years (Behnke 1992). The dam created a 170 km long reservoir that spans the Canada/U.S. border (Whatley 1972). Attempts to establish WCT in the reservoir by both Canada and the USA failed (B. Westover, MWLAP, personal communication). A policy was then developed to establish Gerrard strain RBT in the reservoir, and between 1986 and 1998 RBT were stocked in tributaries to the reservoir and in the reservoir itself (MFWP 2006; MWLAP 2006). Due to conservation concerns, stocking RBT in lakes and reservoirs with potential outlets in the Upper Kootenay River was stopped in 1999 and replaced with WCT stocking or stocking of triploid RBT throughout the watershed (M. MacDonald, Go Fish BC, personal communication).

136 We established three long-term monitoring sites where genetic samples were collected starting in 1999 to investigate the effect of these management changes on the level of introgression. The three sites were Gold Creek, Perry Creek, and Michel Creek (Figure 10). Gold Creek flows directly into the Koocanusa Reservoir and in 1999 and 2000 genetic surveys found 33% hybrids and 20% RBT alleles at the Gold Creek monitoring site (Rubidge et al. 2001; Rubidge and Taylor 2005). Perry Creek is a known WCT spawning tributary to the lower St. Mary River. The St. Mary River is a large, productive tributary to the Kootenay River and enters the Kootenay River approximately 30 km (river km) upstream of the Koocanusa Reservoir. Initial genetic surveys found approximately 20% hybrids and 5% RBT alleles at the Perry Creek monitoring site (Rubidge et al. 2001; Rubidge and Taylor 2005). Perry Creek and Gold Creek will be referred to hereafter as “open sites” because there are no barriers to upstream movement of RBT between these sites and the Koocanusa Reservoir. Michel Creek is a tributary to the mid-section of the Elk River which in turn flows into Koocanusa Reservoir just north of the Canada/U.S. border. Unlike Gold or Perry Creeks, Michel Creek and most of the Elk River are isolated from the remainder of the Kootenay River by a large hydroelectric dam at Elko, approximately 15 km upstream of the confluence of the Elk and Kootenay Rivers (Figure 10). For this reason we will refer to Michel Creek as a “closed” system. Michel Creek is just 10 km downstream of Summit Lake which was stocked with between 3,000-50,000 RBT a year for 20 years between 1939 and 1995

137 (MWLAP 2006). RBT are suspected of escaping from Summit Lake during high runoff years when the lake is connected to Summit Creek, an ephemeral outlet creek that is a tributary to Michel Creek. In 2001 approximately 6% hybrids and 0.5% RBT alleles were present in the Michel Creek population (Rubidge et al. 2001; Rubidge and Taylor 2005). Methods Monitoring sites We selected 13 monitoring sites to represent the variety of streams within the Upper Kootenay River based on preliminary genetic assessment (Rubidge et al. 2001). The monitoring sites were divided into three broad categories: 1) open sites with high initial levels of introgression (e.g., Perry and Gold Creeks), closed sites with low-moderate levels of introgression (e.g., Michel and Grave Creeks), and sites with no apparent introgression (e.g., putatively pure; Upper Bull and St Mary Rivers). Gold Creek, Michel Creek, and Perry Creek were sampled > 4 years from 1999 to 2006. All other sites were sampled 2-3 years from 1999 to 2006. Each year we attempted to capture a minimum of 30 fish at each site. The length of stream required to capture this many fish varied but typically averaged 1-3 km. All age classes except fry (i.e., fingerling, juvenile, and adult) were sampled using angling. A small piece of the lower caudal fin was clipped from each fish captured and stored in 1.5 mL of 95% ethanol for genetic analysis. Tissue samples were

138 collected from fish in the order they were caught. All field sampling occurred during summer low flow conditions, typically from mid July to mid August. We did not separate fish into age classes for most of the analyses. Our intent was to assess the trend in introgression at the population level not within individual age classes. We took this approach because our sample sizes were not generally large enough to allow separation by age class within a site. Also, we did not have scale samples from each site and could not rely on length frequency analysis to determine fish age across such a large scale (i.e., large elevation differences between sites and watershed > 250 km long). The only exception to the above sampling scheme was at Perry Creek where we restricted our sampling to fry and fingerlings because this site had been sampled this way during previous studies (Rubidge et al. 2001, Rubidge and Taylor 2004). We sampled fry and fingerlings in Perry Creek in order to assess introgression in recent cohorts. Fry were assumed to represent fish that were born that year (i.e., age 0) and fingerling were assumed to represent fish born in the previous year (i.e., age 1). In most years > 30 fry were captured within the same 1 km section of stream along the stream margin and in side channels using a dip net. In years when we were unable to collect fry due to budget constraints or sample conditions, fingerlings were collected the following year along with fry of the year. The fingerlings were then used to infer the genetic make up of the previous year’s spawning event. We collected samples representative of annual spawning events in Perry Creek from 1999 to 2006 except for 2005. Fry were considered to be any fish < 60 mm and fingerlings were considered to be fish >

139 60 and < 130 based on a length frequency analysis from a previous study (Rubidge et al. 2001) and an extensive length at age study from a nearby watershed (Fraley and Shepard 2005). DNA Extraction and Amplification Descriptions of the laboratory methods can be found in Rubidge and Taylor (2004, 2005) for samples collected from 1999-2001, and in chapter 2 for samples collected from 2002-2006. Briefly, we used co-dominant, diagnostic, nuclear loci to differentiate WCT, RBT and their hybrids. Four loci were used for fish collected from 1999 to 2001, and seven loci for fish collected from 2002 to 2006. Loci were either restriction lengths polymorphisms (RFLPs) or simple sequence repeats (SSR) developed by Baker et al. (2002) and Ostberg and Rodriguez (2002; 2004). We used a test group of 32 fish with both sets of markers to determine whether we could compare the results of the original analysis using four markers with the new seven SSR markers. There was no significant difference in how each marker set classified the test group (matched pair two-tailed t-test, N = 32, df = 31, t value = 0.75, p = 0.458), and 30 of the 32 fish analyzed were classified as the same genotype. Classifying Hybrids We used two different techniques to classify fish. The first technique was based on the presence or absence of RBT alleles. This technique is the same technique used in a previous study (Rubidge et al. 2001). If the fish was homozygous at all loci for WCT alleles it was classified as a pure WCT. If the fish

140 was heterozygous at one or more loci it was classified as a hybrid. We will refer to this classification method hereafter as the “presence/absence” method. The second technique we used to classify fish was the Bayesian model, NEWHYBRIDS (Anderson and Thompson 2002). The advantages of NEWHYBRIDS are that it is based on a genetic model, does not require the user to know the allele frequencies of the parental species, does not require the use of diagnostic loci, incorporates uncertainty, and it provides the posterior probability that an individual belongs to distinct hybrid genotypes (Anderson and Thompson 2002). The model assumes that populations are in Hardy-Weinberg equilibrium, loci are unlinked, samples are independent, and the number of generations of hybridization is < 2-3 generations (Anderson and Thompson 2002). If these assumptions are violated it can cause an overestimation of the confidence in hybrid classification. We will refer to this classification method hereafter as the “genotype” method. It is not possible to precisely resolve the genotype frequency of individuals when there has been more > 3 generations of inbreeding between two populations because the expected proportions of multi-locus genotypes are identical for F2 and F3 generations (Anderson and Thompson 2002). If there has been more than two generations of inbreeding this will result in individuals being assigned to either WCT_bx or F2 genotypes that are likely later generation backcrosses or a mixture of backcrosses and pure types (E. Anderson, Fisheries Ecology Division, NOAA, personal communication). The NEWHYBRIDS classification of genotypes is similar to the manual classification of hybrid

141 genotypes used by Ostberg and Rodriguez (2006) and Rubidge and Taylor (2004). However, NEWHYBRIDS can be an improvement over the manual method because the posterior probabilities that the model generates provide a measure of confidence in the assignment of an individual to each genotype category. For more discussion on NEWHYBRIDS see Chapter 2. For all analyses, we used the six default genotype frequency classes to assign each fish a genotype (Table 12). Once the posterior probabilities were calculated, each fish was assigned a genotype based on which ever genotype frequency class had the highest posterior probability. The majority (95%) of all fish classified with the Bayesian model were assigned a genotype frequency class with a > 0.95 posterior probability. A posterior probability of 0.95 assigned to the “WCT pure” genotype is equivalent to the fish having a 95% probability of being a pure WCT provided all the assumptions of the model are met. Power to Detect Introgression The power to detect the presence of RBT alleles is equal to β or 1 – α (Kanda et al. 2002b). The following equation was used to calculate α:

α = (1-q)2nx

(Equation 1)

where q = the desired frequency of RBT alleles you wish to detect, n = the number of fish sampled, and x = the number of diagnostic markers. Therefore, the combination of four markers and a sample size of 30 fish per site equates to

142 a 0.91 probability of detecting 1% RBT alleles (i.e., 1 – α where α = 0.08963). For seven markers the probability of detecting 1% RBT alleles is 0.985. Data Analysis We used the chi-square test (X2) to test for differences in the proportion of hybrids (presence/absence classification method) between open and closed sites. We also used the X2 to determine if the proportion of hybrids changed over time at each of the sample sites. For the Perry Creek site (most years of monitoring), we also used logistic regression to predict the potential change in the proportion of hybrids over time. The dependent variable in the analysis was the proportion of hybrids found at each site and the independent variable was year. Logistic regression was used for this analysis because it is more appropriate than linear regression when the dependent variable is dichotomous (Allison 1999).The genotype classification method was used to compare the distribution of genotypes (i.e., hybrid zone structure) between sites and years. A X2 analysis could not be used to determine if the distribution of genotypes varied by year because many of the genotype categories had expected values < 1 (Zar 1984). Therefore, we used the genotype distributions to visually assess trends in hybrid zone structure at sites. All statistical analyses were conducted using SAS© version 9.1 and statistical significance was set at α < 0.05 unless otherwise stated.

143 Influence of Discharge We used a technique developed by Moller and Van Kirk (2003) to assess if a relationship was present between the proportion of hybrids and average maximum spawning and minimum low flow ratio (max/min flow). Moller and Van Kirk (2003) found a significant positive relationship between the recruitment of RBT and low max/min flow ratios. Hydrographs with low (i.e., < 10) max/min flow also had low, flat, long-duration peaks which tended to be more favorable for RBT spawning. No RBT successfully spawned when max/min flow ratios were > 14. We hypothesized that low max/min flows would be correlated with increased introgression in our study area because favorable flow conditions for RBT spawning should increase the potential of hybridizing with WCT. WCT/Hybrid ratio was calculated using the estimated ratio of WCT/Hybrids at the time each fish was a fry (i.e., age 0). The date each fish was at the fry stage was back-calculated using the date captured and the age of the fish at the time of capture based on scale analyses, length frequency histograms, and comparison of length at age data from a nearby watershed (Fraley and Shepard 2005). The max/min flow ratio was calculated using discharge data from hydrologic gauging stations near the sample sites. The average maximum flow was calculated from April to July to coincide with the peak spawning period of RBT and WCT (Henderson et al. 2000; WFL 2003; De Rito 2004; WFL 2004). The minimum flow was calculated over the winter low flow period from October to March. Regression analysis was then used to test for a linear relationship between the WCT/Hybrid and max/min flow ratios.

144 Results Distribution of Hybrids and Genotypes We revisited 13 sites between 1999 and 2006 and collected and genetically analyzed 1,562 fish (Table 13). The elevation of the monitoring sites ranged from 835-1434 m and from 6-241 km upstream of Koocanusa Reservoir (Table 13). Overall 1,342 (85.9%) of all fish were classified as pure WCT (i.e., fish with no RBT alleles) (Table 13). More hybrids were captured at open sites (16.3%) compared to closed sites (6.8%; X2 = 21.5, df = 1, p < 0.0001). The percent of hybrids at a site ranged from 0% (e.g., Elk River and Bull River) to 50% (Perry Creek). In general, the proportion of hybrids was higher at lower elevations and at sites closer to the Koocanusa Reservoir (Table 13). All fish that were classified as pure WCT using the simple hybrid classification approach were also classified as pure WCT with the Bayesian classification model because we used markers that were assumed to be 100% diagnostic (i.e., > 1 RBT allele present = some type of hybrid; Ostberg and Rodriguez 2002, 2004). The most common hybrid genotypes were WCT backcrosses (WCT_bx) and F2s at both closed and open sites (Figure 11). First generation (F1), RBT backcrosses (RBT_bx), and presumed pure RBT (i.e., no WCT alleles) classes combined made up 0.05% of the fish captured at closed sites and 2.1% of the fish captured at open sites. No pure RBT were captured at any closed sites. Open sites consistently had a greater proportion of each hybrid genotype class than closed sites (Figure 11).

145 Trends There was an increase in the number of hybrids at Perry Creek (Table 13; N = 460, X2 = 21.03, df = 1, p = 0.007). The logistic regression analysis of these data suggested that the proportion of hybrids at Perry Creek will increase from 2006 to 2007 by 19.5% (N = 460, Odds Ratio 1.195, 95% Wald CL = 1.0881.314, p < 0.0002). We plotted the proportion of hybrids at Perry Creek from 1985 to 2030 using the logit model developed with our long-term monitoring data (Figure 12). The model predicts that the proportion of hybrids was 5.3% in 1990 and will be 83% by 2015. At Gold Creek and Michel Creek there was no change in the proportion of hybrids over the sampling period (Table 13; p > 0.41). We did not detect any RBT alleles at the three putatively pure WCT sites upon re-sampling (Table 13). The power to detect 1% RBT alleles ranged from 94.1% to 100% (see equation 1 for details). At the remaining closed sites we revisited, there was a decreasing trend in the proportion of hybrids, but this trend was only significant at Alexander Creek (Table 13; p < 0.043). At the open sites there was no change in the proportion of hybrids at Bloom Creek, Kootenay River, Tepee Creek, and Simpson River (Table 13). At Skookumchuck Creek the proportion of hybrids increased (Table 13; p = 0.009). There was a pattern in the changes in the proportion of hybrids between the first and last sample period for each site (Table 13). At the open sites there were two apparent trends. At low elevation, open sites close to the reservoir (e.g., Gold, Skookumchuck, and Perry Creeks) there was an increase in the proportion of hybrids detected between the first and last sampling periods.

146 However, at higher elevation open sites, further from the reservoir (e.g., Bloom, Tepee, Kootenay, and Simpson), there was a decrease in the proportion of hybrids detected between1999 and 2003 (Table 13). The trends in the change of the proportion of hybrid genotypes for the Bayesian model analysis were similar to the changes in the simple hybrid classification analysis above (Table 14). We grouped the genotypes into the following classes to simplify the presentation of the data (Figure 13): WCT_bx (only WCT_bx types), F1/F2 (F1 and F2 types), and RBT_types (RBT_bx and pure RBT). At Perry Creek the number of WCT_bx and F2 individuals tended to increase over time, and only WCT_bx and F2 hybrids were detected at Perry Creek except for one RBT in 1999. At both Michel and Gold Creek there was no apparent trend in the distribution of genotypes over time (Figure 13). Gold Creek was the only site that consistently had RBT_bx and RBT present during every year it was monitored. Among the remaining revisit sites, only Skookumchuck Creek appeared to have increasing proportions of F1/F2 and RBT_types (Figure 14; Table 14). Skookumchuck Creek was the only site where RBT were detected in the repeat sampling, but not the initial sampling. There was a consistent decrease in the proportion of hybrid genotypes at the Kootenay River, Tepee Creek, and the Simpson River Bloom Creek sites (Table 14). The proportion of F1/F2 and RBT_types appeared to decrease at Alexander Creek (Figure 14), and the sample size was too small at Grave Creek to detect a trend.

147 Influence of Discharge The influence of maximum and minimum discharge ratios (Max/Min Q) on the ratio of WCT to hybrids (any fish with > 1 RBT allele) was evaluated for Gold Creek, Michel Creek, and Perry Creek. The average minimum winter low flow discharge was 0.85 m3/sec (N = 18, SD = 0.19, range 0.52-1.04), the average maximum spawning period discharge was 58.51 m3/sec (N = 18, SD = 13, range 34.3-72.1), and the average max/min ratio was 72.5 (N = 18, SD = 21.83, range 35.19-116.15). There were weak positive associations, but non-significant trends between WCT/Hybrid and Max/Min Q at all three long-term monitoring sites (Gold Creek – R2 = 0.184 , df = 9, F = 1.802, p = 0.216; Michel Creek – R2 = 0.028 , df = 7, F = 0.176, p = 0.690; Perry Creek – R2 = 0.042 , df = 8, F = 0.308, p = 0.597). Discussion We expected the proportion of hybrids and pure RBT to decrease from 1999 to 2006 with the reduction of RBT stocking throughout the Upper Kootenay River if stocking was a major factor contributing to the observed increase in the level and distribution of hybrids. This expectation assumes that RBT were being sustained by continued stocking, and that WCT and RBT hybrids were only produced by recently introduced RBT. Stocking of RBT is equivalent to sustained propagule pressure which is often cited in the invasion biology literature as a controlling factor in the rate of establishment and spread of invasive species (Colautti 2005; Lockwood et al. 2005; Duggan et al. 2006; Martinez-Ghersa and

148 Ghersa 2006). Increased propagule pressure has been shown to alter natural hybrid zone dynamics between coastal cutthroat (O .c. clarki) and steelhead (Docker et al. 2003). Where hatchery RBT were introduced 51% of the population were hybrids compared to only 10% where the two species cooccurred without RBT stocking. We found levels of introgression at our study area comparable to when our monitoring first started in 1999. We only failed to detect hybrids at two sites where we initially found them (Grave Creek and Simpson River). Both of these sites had very low initial levels of introgression (one individual WCT_bx per site) and unfortunately our follow-up sample sizes were small and had low power (< 0.72), which could have contributed to the failure to detect hybrids (Table 13). Grave Creek is also the outlet to Grave Lake which was stocked with almost a million RBT from 1923 to 1998. Since only one WCT_bx has been captured at Grave Creek, it suggests that RBT can not escape from Grave Lake, and the hybrid was likely a stray fish produced near Summit Creek or lower Alexander Creek. We suspect that the only Simpson River hybrid found was also a stray hybrid from the nearest source, Marion Lake which is approximately 45 km upstream of the Simpson River sample site (Figure 10). The hybrid found at the Simpson River site could also have come from the Koocanusa Reservoir or another tributary as predicted by the stepping stone model (Kimura and Weiss 1964) and observed in the Flathead watershed (Boyer 2006).

149 Closed Sites Introgression at the closed sites remained relatively constant over the monitoring period. The four closed sites within the Elk River system (Alexander Creek, Grave Creek, Michel Creek, and the Upper Elk) were presumed to have been exposed to “pulses” of RBT that occurred in the mid- to late-1990s when Summit Lake flooded and presumably released stocked RBT into Summit Creek, a tributary to Alexander Creek. The fact that no RBT and very few RBT_bx or F1 individuals have been detected at these sites in the Elk River indicate that the pulses of RBT were likely not sustained long enough to establish RBT or that RBT are at extremely low levels in the system. There is also evidence that RBT did not spread far from this potential source, as very few hybrids have ever been found further downstream in the lower Elk River mainstem, its tributaries, or further upstream in Michel Creek (Rubidge et al. 2001). Introgression at closed sites decreases the further away from Summit Lake the samples were taken. This pattern has been reported in several other investigations of hybridization between WCT and RBT, and it indicates that propagule pressure may play an important role in the spread of hybridization (Chapter 3). No hybrids were detected at the Upper Elk River site (~ 80 km stream distance from Summit Creek) despite the lack of migration barriers between the sites. This is relatively unusual as most pure WCT sites identified within the Upper Kootenay watershed and other sites in the historic WCT range have been located above migration barriers (Young et al. 2004; Rubidge and Taylor 2005; Ostberg and Rodriguez 2006). Possible explanations for the lack of

150 hybrids at the Upper Elk River site are that our sampling was insufficient to detect low levels of hybrids, the propagule pressure was not sufficient to establish hybrids 80 km from the source, or environmental factors prohibited the establishment of RBT (Weigel et al. 2003). We are confident that our sample size was large enough to detect very low levels of RBT introgression in all sites except Grave Creek and Simpson Creeks. For example, we captured a total of 90 fish in the Upper Elk (from 1999, 2002, 2005 combined) and had an average power of 0.964 to detect as little as 1.0% RBT alleles (Kanda et al. 2002a). The establishment of RBT after sporadic releases from Summit Lake may have been hampered by a variety of factors including cold water temperatures (Bear et al. 2007). Rainbow trout, both sympatric with WCT and introduced, have consistently shown a preference for lower elevation stream reaches (Paul and Post 2001; Hitt et al. 2003; Weigel et al. 2003; Rubidge and Taylor 2005; Boyer 2006). The elevation of the Elk River site coincides with colder water temperatures (WFL 2003) which may favor native WCT over RBT. This same scenario was speculated for the observed lower levels of introgression (30% introgression (Ostberg and Rodriguez 2006). At Michel Creek introgression stayed at a relatively low but constant level of approximately 10% (Table 13, Figure 13). It was encouraging that this site showed no evidence of an increase in introgression over the monitoring period. This result strengthens our conclusion that the “pulse” of RBT this area likely

151 received was insufficient to establish RBT in the Upper Elk and its tributaries. However, it also suggests that it may be many more generations before hybrid levels fall below detectable levels assuming no more RBT enter the system. Open Sites Changes in levels of introgression at open sites appear to be heavily influenced by the location of the site in relation to the Koocanusa Reservoir and/or reflect the population dynamics of the stream. At Perry Creek, where we have caught and genetically identified 460 fish that represent progeny from all years from 1997 to 2006 (except 2005), all but one fish were pure WCT, WCT_bx, or F2 individuals. Therefore, it is likely that hybrids and not pure RBT are spawning in Perry Creek. Several other studies have also suggested that hybrids are the predominant disperser of RBT alleles in the northern periphery of the WCT range (Hitt et al. 2003, Rubidge and Taylor 2005, Boyer 2006, Ostberg and Rodriguez 2006). Straying rates of WCT are presumed to be relatively low because of the large observed genetic divergence based on microsatellite and allozyme analyses between many adjacent populations (i.e., neighboring tributaries) in our study area (Leary et al. 1987, Taylor et al. 2004) and other WCT areas (Allendorf and Leary 1988, Boyer 2006). The observed increase in the proportion of WCT_bx and F2 hybrids at Perry Creek, despite the reduction of RBT stocking, suggests that earlier generation hybrids (F1) and RBT_bx fish are being produced outside of Perry Creek and then straying to Perry Creek to spawn.

152 Increases in straying rates of hybrids have been documented in the Flathead watershed (Hitt et al. 2003; Boyer 2006) and have been suspected in the spread of hybridization in our study area (Rubidge 2004, 2005) and others (Campbell et al. 2002; Ostberg and Rodriguez 2006). This type of hybridization would likely be very hard to detect without genetic analysis because of the low levels of RBT introgression (i.e., fish have phenotypic characteristics of pure WCT) and could lead to situations where managers assume populations are pure WCT. There is currently considerable controversy regarding the appropriate management actions that should be enacted in these situations (Allendorf et al. 2005; Campton and Kaeding 2005). The value of populations with low levels of introgression can not be decided on biological criteria alone, and a policy will likely need to be developed that weighs a variety of factors including regional and global status, economic values, and genetic criteria. The rate of increase in the proportion of hybrids at Perry Creek is alarming because it suggests that all fish spawned at this location in Perry Creek could be hybrids within 2 to 3 generations, despite the cessation of RBT stocking since 1998. Straying of these hybrids could then possibly threaten the putatively pure WCT populations upstream of Perry Creek in the Upper St. Mary River. No barriers to migration exist between these sites, and it is unclear why the Upper St. Mary River has remained free of RBT and hybrids despite being relatively close to Koocanusa Reservoir and at a relatively low elevation. At Gold Creek, the other open, long-term monitoring site, there was no trend in the levels of introgression detected over the monitoring period. Gold

153 Creek had the highest initial level of introgression of any of our monitoring sites (31%). We suspect that this site may have reached some sort of dynamic equilibrium due to the apparent population structure in Gold Creek. Unlike most of the other streams we sampled (excluding Perry Creek), Gold Creek tended to have more fingerling and juvenile fish in the lower reach, whereas the other streams were dominated by adult fish. We speculate that this population structure reflects the biology of WCT and introduced RBT in Gold Creek. We think that the lower reaches of Gold Creek are predominately used for spawning by fluvial adults and adfluvial fish from the Koocanusa Reservoir. Adjacent tributaries and the upper reaches have a resident WCT population. We may have been sampling a mixture of the progeny of the adfluvial and fluvial spawners, and migrants from the upstream WCT population. This could explain why introgression has not increased at Gold Creek. A similar scenario has been observed in Washington where pure RBT and WCT from above migration barriers were suspected of moving downstream into a hybrid zone of the two species and maintaining relatively constant levels of hybridization despite RBT having been introduced into the native WCT habitat for > 50 years (Ostberg and Rodriguez 2006). This hypothesis is strengthened by the data from Bloom Creek and Tepee Creek which are both tributaries to Gold Creek. Both of these tributaries had lower levels of introgression with the highest elevation site (Tepee Creek) having the lowest introgression. Both of these sites could be sources of pure WCT to the lower Gold Creek site. Although Bloom and Tepee Creek are at higher elevations than Gold Creek, they are still at lower elevations than many

154 other streams with higher levels of introgression and they are relatively close to the reservoir. However, they are smaller than Gold Creek ( 14 prevent successful RBT spawning. This may be one explanation of why our streams do not have higher levels of introgression despite almost 100 years of RBT stocking. The hydrologic conditions for optimal RBT reproduction may be severely limited and favor the late spawning strategy of WCT (on the downward slope of the hydrograph) versus RBT that tend to spawn earlier (De Rito 2004; Henderson et al. 2000). Conclusion When Wolf et al. (2001) modeled plant hybridization pre-zygotic barriers were more important than post-zygotic barriers in predicting extinction. They also

156 found that stability in a hybrid zone could only be maintained when habitat segregation was included in the model. Unfortunately, when RBT are introduced into native WCT habitat, there appears to be few pre-zygotic barriers as evidenced by the extraordinary large number of native WCT populations that now have high levels of introgression with RBT (Mayhood 2000; Rubidge et al. 2001; Weigel et al. 2003; Shepard et al. 2005). The reduction of RBT stocking in the Upper Kootenay River likely came too late to prevent further spread of RBT alleles throughout the watershed. However, in closed sites that were apparently only exposed to infrequent pulses of RBT, it does not appear that introgression is increasing. These sites may return to undetectable levels of introgression over the next several generations. In the open sites however, it appears that heavily hybridized populations near the Koocanusa Reservoir have become the new sources of RBT alleles that are being spread primarily via hybrid straying. Some sites appear to have increased introgression over relatively short time period (2-3 generations of monitoring), and more hybrid swarms could develop if immediate management action is not taken. Management options are limited with the removal of RBT being the most commonly proposed action (Moore and Ridley 1984; Thompson and Rahel 1996; Quist and Hubert 2004). Prior to any management action a full assessment of the sources of RBT in the Kootenay watershed should be conducted to prioritize streams for treatment.

157 Literature Cited Allendorf, F. W., R. E. Leary, N. P. Hitt, K. L. Knudsen, M. C. Boyer, and P. Spruell. 2005. Cutthroat trout hybridization and the US Endangered Species Act: One species, two policies. Conservation Biology 19(4):13261328. Allendorf, F. W., and R. F. Leary. 1988. Conservation and distribution of genetic variation in polytypic species, the cutthroat trout. Conservation Biology 2(2):170-184. Allendorf, F. W., and R. S. Waples. 1996. Conservation and genetics of salmonid fishes. In J.C. Avise and J.L Hamrick (eds). Conservation genetics: case histories from nature. Chapman and Hall, New York, NY:238-280. Allison, P. D. 1999. Logistic regression using the SAS system: theory and application. SAS Institute Inc, Cary, NC. Anderson, E. C., and E. A. Thompson. 2002. A model-based method for identifying species hybrids using multilocus genetic data. Genetics 160:1217-1229. Baker, J., P. Bentzen, and P. Moran. 2002. Molecular markers distinguish coastal cutthroat trout from coastal rainbow trout/steelhead and their hybrids. Transactions of the American Fisheries Society 131:404–417. Bear, E. A., T. E. McMahon, and A. V. Zale. 2007. Comparative thermal requirements of westslope cutthroat trout and rainbow trout: implications for species interactions and development of thermal protection standards. Transactions of the American Fisheries Society 136:1113–1121. Behnke, R. J. 1992. Native trout of western North America. American Fisheries Society, Monograph 6. Behnke, R. J. 2002. Trout and salmon of North America. Simon and Schuster Inc., New York. Boyer, M. C. 2006. Rainbow trout invasion and the spread of hybridization with native westslope cutthroat trout. Master's thesis, University of Montana, Missoula, Montana. Campbell, M. R., J. Dillon, and M. S. Powell. 2002. Hybridization and introgression in a managed, native population of Yellowstone cutthroat

158 trout: genetic detection and management implications. Transactions of the American Fisheries Society 131(3):364-375. Campton, D. E., and L. R. Kaeding. 2005. Westslope cutthroat trout, hybridization, and the US Endangered Species Act. Conservation Biology 19(4):1323-1325. Colautti, R. I. 2005. Are characteristics of introduced salmonid fishes biased by propagule pressure? Canadian Journal of Fisheries and Aquatic Sciences 62(4):950-959. Crossman, E. J. 1991. Introduced freshwater fishes: a review of the North American perspective with emphasis on Canada. Canadian Journal of Fisheries and Aquatic Sciences 48(Supplement 1):46-57. De Rito, J. N. J. 2004. Assessment of reproductive isolation between Yellowstone cutthroat trout and rainbow trout in the Yellowstone River, Montana. Master's thesis, Fish and Wildlife Management, Montana State University, Bozeman. Docker, M. F., A. Dale, and D. D. Heath. 2003. Erosion of interspecific reproductive barriers resulting from hatchery supplementation of rainbow trout sympatric with cutthroat trout. Molecular Ecology 12(12):3515-3521. Duff, D. A. 1996. Conservation assessment for inland cutthroat trout: distribution, status and habitat management implications. USDA For. Ser. Intermountain Region, Ogden, Utah. Duggan, I. C., C. Rixon, and H. J. MacIsaac. 2006. Popularity and propagule pressure: determinants of introduction and establishment of aquarium fish. Biological Invasions 8:377–382. Fraley, J., and B. B. Shepard. 2005. Age, growth, and movements of westslope cutthroat trout, Oncorhynchus clarki lewisi, inhabiting the headwaters of a wilderness river. Northwest Science 79(1):12-21. Hallerman, E. M. 2003. Population genetics: principles and applications for fisheries scientists. American Fisheries Society, Bethesda, Maryland. Henderson, R., J. L. Kershner, and C. A. Toline. 2000. Timing and location of spawning by nonnative wild rainbow trout and native cutthroat trout in the South Fork Snake River, Idaho, with implications for hybridization. North American Journal of Fisheries Management 20:584-596.

159 Hilderbrand, R. H., and J. L. Kershner. 2000. Conserving inland cutthroat trout in small streams: how much stream is enough? North American Journal of Fisheries Management 20:513-520. Hitt, N. P., C. A. Frissell, C. C. Muhlfeld, and F. W. Allendorf. 2003. Spread of hybridization between native westslope cutthroat trout, Oncorhynchus clarki lewisi, and nonnative rainbow trout, Oncorhynchus mykiss. Canadian Journal of Fisheries and Aquatic Sciences 60:1440-1451. Kanda, N., R. F. Leary, and F. W. Allendorf. 2002a. Evidence of introgressive hybridization between bull trout and brook trout. Transactions of the American Fisheries Society 131(4):772-782. Kanda, N., R. F. Leary, P. Spruell, and F. W. Allendorf. 2002b. Molecular genetic markers identifying hybridization between the Colorado River-greenback cutthroat trout complex and Yellowstone cutthroat trout or rainbow trout. Transactions of the American Fisheries Society 131(2):312-319. Kimura, M., and G. H. Weiss. 1964. The stepping stone model of population structure and the decrease of genetic correlation with distance. Genetics 49:561-576. Kozfkay, J. R., J. C. Dillon, and D. J. Schill. 2006. Routine use of sterile fish in salmonid sport fisheries: are we there yet? Fisheries 31(8):392-401. Kulp, M. A., and S. E. Moore. 2000. Multiple electroshocking removals for eliminating rainbow trout in a small Southern Appalachian stream. North American Journal of Fisheries Management 20:259-266. Leary, R., F. Allendorf, and G. Sage. 1995. Hybridization and introgression between introduced and native fish. Pages 91-101 in H.L. Schramm, Jr., and R.G. Piper, editors. Uses and effects of cultured fishes in aquatic ecosystems. American Fisheries Society, Symposium 15, Bethesda, Maryland. Leary, R., F. W. Allendorf, and K. L. Knudsen. 1987. Genetic divergence among populations of westslope cutthroat trout in the upper Kootenay River drainage, BC. Dept of Zoology University of Montana Population Genetics Laboratory Report 87/1. Lockwood, J. L., P. Cassey, and T. Blackburn. 2005. The role of propagule pressure in explaining species invasions. Trends in Ecology and Evolution 20(5):223-228.

160 Martinez-Ghersa, M. A., and C. M. Ghersa. 2006. The relationship of propagule pressure to invasion potential in plants. Euphytica 148(1-2):87-96. Mayhood, D. 2000. Provisional evaluation of the status of westslope cutthroat trout in Canada. At risk proceedings of a conference on the biology and management of species and habitats at risk. Feb. 15-19, 1999, Kamloops, British Columbia. MFWP. 2001. Stocking records for Koocanusa Reservoir. Montana Department of Fish, Wildlife, and Parks, Libby, Montana. Moore, S. E., and B. Ridley. 1984. A summary of changing standing crops of native brook trout in response to removal of sympatric rainbow trout in Great Smoky Mountains National Park. Journal of the Tennessee Academy of Science 59(4):76-77. MWLAP. 2006. Regional fish stocking records: Kootenay Region, 2006. Ministry of Water, Land and Air Protection, Victoria, British Columbia. Novinger, D. C., and F. J. Rahel. 2003. Isolation management with artificial barriers as a conservation strategy for cutthroat trout in headwater streams. Conservation Biology 17(3):772-781. Ostberg, C. O., and R. J. Rodriguez. 2002. Novel molecular markers differentiate Oncorhynchus mykiss (rainbow trout and steelhead) and the O. clarki (cutthroat trout) subspecies. Molecular Ecology 2:197-202. Ostberg, C. O., and R. J. Rodriguez. 2004. Bi-parentally inherited speciesspecific markers identify hybridization between rainbow trout and cutthroat trout subspecies. Molecular Ecology Notes 4(1):26-29. Ostberg, C. O., and R. J. Rodriguez. 2006. Hybridization and cytonuclear associations among native westslope cutthroat trout, introduced rainbow trout, and their hybrids within the Stehekin River drainage, North Cascades National Park. Transactions of the American Fisheries Society 135(4):924-942. Paul, A. J., and J. R. Post. 2001. Spatial distribution of native and nonnative salmonids in streams of the Eastern Slopes of the Canadian Rocky Mountains. Transactions of American Fisheries Society 130:417-430. Quist, M. C., and W. A. Hubert. 2004. Bioinvasive species and the preservation of cutthroat trout in the western United States: ecological, social, and economic issues. Environmental Science & Policy 7(4):303-313.

161 Rubidge, E., P. Corbett, and E. B. Taylor. 2001. A molecular analysis of hybridization between native westslope cutthroat trout and introduced rainbow trout in southeastern British Columbia, Canada. Journal of Fish Biology 59:42-54. Rubidge, E. M., and E. B. Taylor. 2004. Hybrid zone structure and the potential role of selection in hybridizing populations of native westslope cutthroat trout (Oncorhynchus clarki lewisi) and introduced rainbow trout (Omykiss). Molecular Ecology 13(12):3735-3749. Rubidge, E. M., and E. B. Taylor. 2005. An analysis of spatial and environmental factors influencing hybridization between native westslope cutthroat trout (Oncorhynchus clarki lewisi) and introduced rainbow trout (O. mykiss) in the upper Kootenay River drainage, British Columbia. Conservation Genetics 6(3):369-384. Ruzycki, J. R., D. A. Beauchamp, and D. L. Yule. 2003. Effects of introduced lake trout on native cutthroat trout in Yellowstone Lake. Ecological Applications 13(1):23-37. Shepard, B. B. 2002. A native westslope cutthroat trout population responds positively after brook trout removal and habitat restoration. Intermountain Journal of Sciences 8(3):193-214. Shepard, B. B., B. E. May, and W. Urie. 2005. Status and conservation of westslope cutthroat trout within the western United States. North American Journal of Fisheries Management 25(4):1426-1440. Strahler, A. N. 1957. Quantitative analysis of watershed geomorphology. Transactions of American Geophysical Union 38:913-920. Thompson, P., and F. Rahel. 1996. Evaluation of depletion-removal electrofishnig of brook trout in small Rocky Mountain streams. North American Journal of Fisheries Management 16:332-339. Thompson, P., and F. Rahel. 1998. Evaluation of artificial barriers in small rocky mountain streams for preventing the upstream movement of brook trout. North American Journal of Fisheries Management 18:206-210. Van Kirk, R. W., W. C. Schrader, and S. L. Moller. 2006. Effects of hydrologic regime and its alteration on nonnative trout invasion success in the Snake River, USA. Canadian Journal of Fisheries and Aquatic Sciences in review.

162 Weigel, D. E., J. T. Peterson, and P. Spruell. 2003. Introgressive hybridization between native cutthroat trout and introduced rainbow trout. Ecological Applications 13(1):38-50. Welcomme, R. L. 1992. A history of international introductions of inland aquatic species. ICES marine science symposia 194:3-14. WFL. 2003. Elk River westslope cutthroat trout radio telemetry study: 2000-2002. Prepared by Westslope Fisheries Ltd. Prepared for Columbia-Kootenay Fisheries Renewal Partnership, Cranbrook, BC. WFL. 2004. St. Mary River westslope cutthroat trout radio telemetry study: 20012004. Prepared by Westslope Fisheries Ltd. Prepared for ColumbiaKootenay Fisheries Renewal Partnership, Cranbrook, BC. Whatley, M. 1972. Effects on fish in the Kootenay River of construction of Libby Dam. Fisheries Management Report No. 65. Fish Habitat Protection Section, Fish and Wildlife Branch, Dept. of Recreation and Conservation, Victoria, BC. Wolf, D. E., N. Takebayashi, and L. H. Rieseberg. 2001. Predicting the risk of extinction through hybridization. Conservation Biology 15(4):1039-1053. Wydoski, R. S., and R. R. Whitney. 2003. Inland fishes of Washington. American Fisheries Society, Second Edition, Bethesda, Maryland. Young, M. 1995. Conservation assessment for inland cutthroat trout. USDA For. Ser., RM-GTR-256, Fort Collins, Colorado. Young, S. F., J. G. McLellan, and J. B. Shaklee. 2004. Genetic integrity and microgeographic population structure of westslope cutthroat trout, Oncorhynchus clarki lewisi, in the Pend Oreille Basin in Washington. Environmental Biology of Fishes 69(1-4):127-142. Zar, J. H. 1984. Biostatistical analysis: 2nd Edition. Simon and Schuster, Englewood Cliffs, New Jersey.

163 Table 12. Genotype frequency classes and expected proportions of loci with 0, 1, and 2 genes originating from WCT hybridizing with RBT after two generations (Anderson and Thompson 2002).

GFCb WCT RBT F1 F2 WCT_bx RBT_bx

a

Expected portion of loci with WCT and RBT allelesa 11 12 21 22 1.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 0.00 0.50 0.50 0.00 0.25 0.25 0.25 0.25 0.50 0.25 0.25 0.00 0.00 0.25 0.25 0.50

1 = WCT allele; 2 = RBT allele; 11 = homozygous for WCT, 12 or 21 =

heterozygous, and 22 = homozygous for RBT. b

GFC = genotype frequency class; WCT = pure WCT; RBT = pure RBT, F1 =

hybrid produced from breeding of pure WCT and pure RBT; F2 = hybrid produced from breeding of two F1 hybrids; WCT_bx = backcrossed WCT produced from breeding of pure WCT and F2 hybrid; RBT_bx = backcross RBT produced from breeding a pure RBT with a F2 hybrid.

164 Table 13. Summary of number of presumed pure WCT and hybrids (i.e., fish with > 1 RBT allele) by sample site and year in the Upper Kootenay River, BC. Chisquare test statistics refer to test results over all years with in a sample site. Sites with < 1 observed WCT or hybrid were excluded from tests and Yates correction factor used for all tests with observed values < 5. Stream = stream name, Type = whether the stream can be accessed by fish from the Koocanusa Reservoir (open), or is inaccessible due to migration barriers (closed), Elev. = elevation of the site, Dist. = stream distance to the Koocanusa Reservoir from the sample site, No. WCT = number of WCT; No. Hyb = number of hybrids (i.e., fish with > 1 RBT allele), Prop WCT = proportion WCT, Prop Hyb = proportion hybrids, Pwr = the power to detect 1% RBT alleles, X2 = Chi-square statistic, df = degrees of freedom, P = p value for the X2 test. Test statistics are for tests between years within sample sites. All test statistics are listed on the first line of the site. Test between closed and open sites listed on the subtotal line for closed sites. Significant differences are denoted by lower case letters. Hybrid classification Stream/ Type

Test statistics

Elev (m)

Dist (km)

Year

No. WCT

No. Hyb

Prop WCT

Prop Hyb

N

Pwr

Bull

1117

39

1999

36

0

1.00

0.00

36

95

2003

23

0

1.00

0.00

23

96

Grave

1187

99

2002

12

1

0.92

0.08

13

84

2005

7

0

1.00

0.00

7

63

2000

20

10

0.67

0.33

30

99

2003

19

1

0.95

0.05

20

94

2001

28

2

0.93

0.07

30

99

2

df

P

-

-

-

-

-

-

4.1

1

0.043

3.8

4

0.440

X

Closed

Alexander

Michel

1280

1282

99

104

z

165 Table 13. Continued. Hybrid classification Stream/ Type

Elk

Elev (m)

1434

Dist (km)

155

Test statistics

Year

No. WCT

No. Hyb

Prop WCT

Prop Hyb

N

Pwr

2002

25

5

0.83

0.17

30

99

2003

26

2

0.93

0.07

28

98

2005

28

3

0.90

0.10

31

99

2006

29

1

0.97

0.03

30

99

1999

38

0

1.00

0.00

38

95

2002

29

0

1.00

0.00

29

98 96

2005

23

0

1.00

0.00

23

-

343

25

0.93

0.07

368

1999

25

11

0.69

0.31

36

94

2000

19

11

0.63

0.37

30

99

2003

25

6

0.81

0.19

31

99

2006

19

11

0.63

0.37

30

99

2000

19

11

0.63

0.37

30

91

2003

25

4

0.86

0.14

29

98

2000

25

5

0.83

0.17

30

91

2003

29

1

0.97

0.03

30

99

1997

10

2

0.83

0.17

12

62

1998

18

2

0.90

0.10

20

80

1999

91

26

0.78

0.22

117

100

2000

109

39

0.74

0.26

148

100

2001

40

6

0.87

0.13

46

98

2002

22

9

0.71

0.29

31

99

2003

18

11

0.62

0.38

29

98

2004

17

10

0.63

0.37

27

98

2006

15

15

0.50

0.50

30

99

1999

38

2

0.95

0.05

40

96

2003

18

7

0.72

0.28

25

97

1999

31

0

1.00

0.00

31

92

2000

299

0

1.00

0.00

299

100

1999

12

2

0.86

0.14

14

65

2003

38

2

0.95

0.05

40

100

2002

28

1

0.97

0.03

29

98

2003

9

0

1.00

0.00

9

72

subtotal Open

-

999

195

0.84

0.16

1194

-

Total

-

1342

220

0.86

0.14

1562

-

subtotal Closed

2

df

P

-

-

-

21.1

1

0.000

2.9

3

0.411

3.0

1

0.086

1.7

1

0.197

21.0

8

0.007

x

6.8

1

0.009

w

-

-

-

0.3

1

0.583

-

-

-

X

y

Open Gold

Bloom

835

1020

6

26

Tepee

1167

41

Perry

914

58

Skookum

St. Mary Kootenay Simpson

923

1029 1169 1240

87

104 226 241

166 Table 14. Proportion of each genotype frequency class by sample site and year in the Upper Kootenay River, BC: 1999-2006. Sites arranged closed sites (above migration barriers) and open sites (below migration barriers). Closed sites arranged by elevation from lowest to highest and open sites arranged by distance to the Koocanusa Reservoir from nearest to furthest (see Table 13). Genotype classifications determined by NEWHYBRIDS model (Anderson and Thompson 2002).

Stream Name/ Type

Year

WCT

1999 2003 2002 2005 2000 2003 2001 2002 2003 2005 2006 1999 2002 2005 -

1.00 1.00 0.92 1.00 0.67 0.95 0.93 0.83 0.93 0.90 0.97 1.00 1.00 1.00 0.93

1999 2000 2003 2006 2000 2003 1997 1998 1999 2000

0.69 0.63 0.81 0.63 0.63 0.86 0.83 0.90 0.78 0.74

WCT_ bx

Genotype* F2 F1

RBT _bx

RBT

N

Closed Bull R Grave Cr Alexander Cr Michel Cr

Elk R

Total Closed

36 23 13 7 30 20 30 30 28 31 30 38 29 23 368

0.08 0.17 0.05 0.07 0.13 0.07 0.10 0.03

0.13

0.03

0.00 0.05

0.00 0.01

0.00 0.00

0.25 0.10

0.13

0.03

Open Gold Cr

Bloom Cr Perry Cr

0.07 0.10 0.10 0.10 0.08 0.05 0.19 0.20

0.07 0.27

0.03 0.07

0.13 0.00

0.06 0.07 0.06 0.10

0.03 0.08 0.05 0.03 0.07

0.01

36 30 31 30 30 29 12 20 117 148

167 Table 14. Continued.

Stream Name/ Type

Tepee Cr Skookumchuck Cr St. Mary R Kootenay R Simpson R Total Open Total *

Year

WCT

2001 2002 2003 2004 2006 2000 2003 1999 2003 1999 2000 1999 2003 2002 2003 -

0.87 0.71 0.62 0.63 0.50 0.83 0.97 0.95 0.72 1.00 1.00 0.86 0.95 0.97 1.00 0.84 0.86

WCT_ bx 0.11 0.29 0.38 0.37 0.40 0.13 0.03 0.05 0.16

Genotype* F2 F1

RBT _bx

RBT

0.02

0.10 0.03

0.00

0.00

0.04

0.00

0.08

0.14 0.05 0.03 0.11 0.09

0.04 0.03

0.00 0.00

0.01 0.01

0.01 0.01

N 46 31 29 27 30 30 30 40 25 31 299 14 40 29 9 1,193 1,561

See table 12 for definitions of the hybrid genotype classes. N = number of fish

captured per site and year.

168 Figure 10. Study area and location of hybridization monitoring sites in the Upper Kootenay River, BC. Site numbers are 1) Michel Creek, 2) Grave Creek, 3) Alexander Creek, 4) Gold Creek, 5) Skookumchuck Creek, 6) Perry Creek, 7) Bloom Creek, 8) Tepee Creek, 9) Kootenay River, 10) Simpson River, 11) Elk River, 12) Bull River, and 13) St. Mary River. Small gray dots = RBT stocking sites, gray x’s = fish migration barriers > 2 m, and black bars = hydroelectric dams that are also fish migration barriers.

169

170 0.12

Proportion

0.10 0.08 0.06 0.04 0.02 0.00 WCT_bx

F2

F1

RBT_bx

RBT

Genotype

Figure 11. Proportion of genotypes based on a NEWHYBIRDS model classification (Anderson and Thompson 2002) for closed (black bars) vs. open (gray bars) sites in the Upper Kootenay River: 1999-2006. Genotypes are listed as a continuum from genotypes with the most WCT alleles (WCT_bx) on the right to genotypes with the least WCT alleles on the right (RBT). See Table 12 for definitions of the genotype classes.

171

Proportion of Hybrids

1 0.75 Predicted 0.5

Observed

0.25 0 1980

1990

2000

2010

2020

2030

Year

Figure 12. The predicted and observed proportion of hybrids at Perry Creek. Predicted observations are based on a logit model developed from the observed proportion of hybrids sampled from 1997-2006. The proportion of hybrids in Perry Creek is expected to increase by 19.5% from 2006 to 2007 (N = 460, Odds Ratio = 1.195, 95% Wald CL = 1.088-1.314, p < 0.0002).

172 b)

Proportion

0.30

0.20

0.10

0.00 99

00

03

06

Year

c)

Proportion

0.15

0.10

0.05

0.00 2001

2002

2003

2005

2006

Year

a)

Proportion

0.40 0.30 0.20 0.10 0.00 97

98

99

00

01

02

03

04

06

Year

Figure 13. The change in the genotype frequencies among years for the three long-term monitoring sites: a) Perry Creek, b) Gold Creek, and c) Michel Creek. Genotype classes were combined for presentation purposes, clear bars =

173 WCT_bx, black bars = F1/F2, and Gray bars = RBT_types. Pure WCT types are not included in the graphs.

a) 0.20

Proportion

0.15

0.10

0.05

0.00 1999

2003

Year

b) 0.20

Proportion

0.15 0.10 0.05 0.00 2000

2003 Year

Figure 14. The change in the genotype frequencies between years for two of the revisit sites: a) Skookumchuck Creek and b) Alexander Creek. Genotype classes were combined for presentation purposes, clear bars = WCT_bx, black bars = F1/F2, and Gray bars = RBT_types. Pure WCT types are not included in the graphs.

174 CHAPTER 5 CONCLUSION In my dissertation, I expanded on a preliminary assessment of the genetic status of westslope cutthroat (WCT) in the Upper Kootenay River in southeastern British Columbia (BC). I determined the distribution of WCT, rainbow trout (RBT), and their hybrids, modeled the potential propagule pressure exerted by RBT stocking, assessed trends in the level of introgression, and estimated the current and potential genetic status of WCT throughout the watershed. I used a combined data set of 2,670 fish diagnosed at between four and seven codominant nuclear loci to assess the genetic status of individual fish and sites (i.e., populations). Westslope cutthroat populations have experienced significant declines across their native range, and in most cases, hybridization between WCT and introduced salmonids (particularly RBT) is the main cause of the decline. In general, pure WCT only persist in higher elevation streams and above fish migration barriers where RBT are introduced into WCT habitat. I found the same trend in the Upper Kootenay River. However, I also found evidence that increased propagule pressure could cause high levels of introgression in mid to high elevation areas (i.e., near Summit Lake and Whiteswan Lake). Propagule pressure was also the best predictor of the level of introgression out of the 12 models I tested.

175 I evaluated data that included a measure of introgression between WCT and RBT from 1997 to 2006. These data indicated that the level of introgression did not decrease at sites that had high levels of introgression at the beginning of the study (i.e., Gold Creek) despite the cessation of RBT stocking in 1998 throughout the watershed. Sites far from the Koocanusa Reservoir appeared to have similar levels of introgression over the study, but sites at intermediate distances (i.e., 60-80 km) from the reservoir had increases in the number of hybrids (i.e., Perry Creek and Skookumchuck Creek). I concluded that the level of introgression is likely to increase at sites that currently have low levels of introgression via hybrids produced in low elevation areas straying to upstream sites. The level of introgression between WCT and RBT in the Upper Kootenay River is lower than other areas within the WCT range; however, the hybrid zone I identified is still extensive and it may increase unless management actions can reduce sources of RBT. I recommend that the priorities for management be as follows:

•

Identify the remaining pure populations.

•

Continue a ban on stocking of RBT within the watershed.

•

Develop management strategies to protect the remaining pure populations from potential hybridization and other development activities.

•

Identify sources of RBT within the watershed. This should include an assessment of lakes that have been stocked in the past, and streams where RBT may have naturalized.

•

176 Determine how the movement of pure WCT within streams and between drainages affects the level of introgression at sites, especially where WCT move downstream over migration barriers from pure WCT populations into sites with varying levels of introgression.

•

Assess the effectiveness of RBT removal in trial areas that are potential sources of RBT alleles (i.e., tributaries to the Koocanusa Reservoir).

177

APPENDIX

178 Release Letter for John Olson as coauthor of Chapter 3. December 6, 2007 Stephen Bennett Box 579, Newton, Utah 84327 tel. (435) 563 5766 Dear John Olson I am in the process of preparing my dissertation in Watershed Sciences at Utah State University. I hope to complete in the Fall of 2007. I am requesting to submit your name as a coauthor for Chapter 3 of my dissertation titles “Influence of propagule pressure and stream characteristics on introgression between native westslope cutthroat trout and introduced rainbow trout in British Columbia”. You will be listed as the second coauthor along with Dr. Jeff Kershner. Please indicate your approval of this request by signing in the space provided. If you have any questions, please call me at the number above. Thank you

Stephen Bennett I hereby give permission to Stephen Bennett to use my name as a coauthor in Chapter 3 of his dissertation (see title above). Signed:

Date:

179 VITA STEPHEN N. BENNETT EDUCATION Fisheries Biology (Ph.D.), Aquatic, Watershed, and Earth Resources, Utah State University, Logan, Utah, USA –graduation Fall 2007 Resource and Environmental Management (M.R.M.), Simon Fraser University, Burnaby, British Columbia, Canada - Graduated August 1994 Wildlife Biology (B.Sc. Honors), University of Montana, Missoula, Montana, USA - Graduated June 1990 Renewable Resource Management (Diploma), Lethbridge Community College, Lethbridge, Alberta, Canada - Graduated May 1987 CAREER GOALS My career goals are to improve our management of natural resources, and help steer human societies towards more sustainable development of natural resources via conducting research on fisheries and wildlife ecology related issues, and training future resource scientists by teaching at the university level. EMPLOYMENT SUMMARY 2007

Post Doctoral Research, US Forest Service

Primary author and researcher responsible for developing a Fish Inventory and Monitoring Manual for all National Forest Lands. Project requires a detailed review of survey design theory, planning, and design, literature review of field techniques used for sampling fish in wadeable streams, and interviews with various fisheries scientists. Spring 2006

Adjunct Professor, Utah State University

Taught a two credit combined senior undergraduate and graduate level Watershed and Geographic Information Analysis course. Duties included developing and presenting a lecture program, coordinating guest lecturers and geographic information systems laboratory exercises, and evaluating student

180 performance. The course involves one two hour discussion/seminar period per week. Fall 2005

Adjunct Professor, Weber State University

Taught a four credit, senior undergraduate level Ichthyology course during the fall semester. Duties included developing and presenting a lecture, field study, and laboratory program. The course involved three 50 minute lectures and one four hour lab per week. I was responsible for the creation and grading of all assignments and tests. Spring 2002

FRWS 1200 Grader

Responsible for grading exams and homework assignments for 50 students as well as being the proctor for the midterm and final exams. 1994 – 2002

Biologist, Mirkwood Ecological Consultants

I worked as a project and field manager on a wide variety of fish and wildlife projects and environmental impact assessments throughout British Columbia. 1990 – 1994

Graduate Student at Simon Fraser University

Developed and conducted a two year field study on the bird use of wildlife trees and specialized in forest ecology and wildlife management. 1985 – 1990

Technician, Department of Fisheries and Oceans

Responsible for year round activities of a salmon hatchery, enforcing fishing regulations on local streams, and assessing new regulations on salmon stocks. STUDENT ACTIVITIES and AWARDS

• • • • •

Volunteer organizer at the Bonneville Chapter of the American Fisheries AGM in Park City, March 2006 Graduate Student Representative for the Aquatic, Watershed, and Earth Resources Department, 2003-2005 Volunteer organizer and Presenter at the American Fisheries Society Western Division meeting in Salt Lake City, March 2004 Volunteer and student representative at the Bear River Festival in Logan, May 2003 and 2004 Volunteer for Common Grounds Outdoor Adventures Organization kokanee salmon viewing, September 2003 and 2004

181 Active member of the student chapter of the American Fisheries Society, 2002 to present • Terri Lynn Steel Scholarship Award, College of Natural Resources, Utah State University • Science Council of B.C., Graduate Research Engineering and Technology Award (1992 to 1994)

•

REFERENCES Peter Corbett, Manager, Mirkwood Ecological Consultants Ltd. Winlaw, B.C. Telephone: (250) 226 7249 Dr. Jeffery Kershner 2, Center Director USGS Northern Rocky Mountain Science Center, Bozeman, MT Telephone: (406) 994-5304 Dr. Ken Lertzman 3, Associate Professor, School of Resource and Environmental Mgt, Simon Fraser University, Burnaby, B.C. Telephone: (778) 782-3069 Dr. Brett Roper, Aquatic Ecologist, Fish and Aquatic Ecology Unit, USDA Forest Service, Logan, Utah. Telephone: (435) 755 3566 PUBLISHED AND UNPUBLISHED REPORTS Bennett, S.N. and B.B. Roper. 2007. Fish inventory and monitoring technical guide for wadeable streams on National Forests (DRAFT). General Technical Report, USDA, Forest Service, Fish and Aquatic Ecology Unit, Logan, Utah. Bennett, S., P. Corbett, et al. 2004. Status of westslope cutthroat trout and the potential extent of hybridization with non-native rainbow trout in the Upper Kootenay River, BC. Paper presented at the American Fisheries Society Western Regional meeting, Salt Lake City, UT. Feb 29-March 4, 2004. Bennett, S. N. 2002. Westslope Cutthroat Trout: Identified Wildlife Species Accounts. Prepared for Identified Wildlife Management Strategy, Water, Land, and Air Protection, Victoria, BC.

2 3

Major advisor for PhD Major advisor for Master’s

182 Addison, J., Bennett, S. and P. Corbett. 2002. Reconnaissance (1:20,000) Fish and Fish Habitat Inventory (Phase IV-VI) of Tributaries to Revelstoke Reservoir. Prepared for Downie Creek Sawmills, Revelstoke, BC. Robertson, I., K. McIntosh, S. Bennett, S. Kesting, V. Palermo, P. Warburton, and P. Kneen. 2001. An examination of species at risk in areas with BC Hydro facilities. Prepared for BC Hydro, Vancouver, BC. Bennett, S. 2001. Raptor Survey in the Bernard Creek Drainage. Prepared for the Ministry of Forests, Kootenay Lake District, Nelson BC. Chytyk, P., Cooper, J. and S. Bennett. 2001. Northern goshawk inventory of TFL 14, Chetwynd, BC. Prepared for Canfor – Chetwynd Operations, Chetwynd, BC. Bennett, S. 2001. Songbird surveys and wildlife habitat assessment from Fort Shepherd to Castlegar, BC. Prepared for Larkspur Biological Consultants Ltd., Castlegar, BC. Bennett, S. and P. Corbett. 2000. Murphy Creek Fish and Fish Habitat Assessment for a proposed hydroelectric facility. Prepared for Columbia River Ranches Ltd. Bennett, S. and P. Corbett. 2000. Fisheries and wildlife assessment of the proposed 230 kV development from South Slocan to Fort Shepherd. Prepared for West Kootenay Power, Trail, BC. Bennett, S. N., P. Sherrington, et al. 2000). Habitat use and distribution of listed neotropical migrant songbirds of northeastern British Columbia. In L.M. Darling (ed) Proceedings of a conference on the biology and management of species and habitats at risk BC Environment, Lands, and Parks, Victoria, BC. Houde, I., S. Bennett and S. Clow. 2000. Owl surveys in the Arrow Forest District. Prepared for Innovative Forestry Practices Agreement, Nelson, BC. Bennett, S., P. Sherrington, P. Johnstone, and B. Harrison. 2000. Habitat use and distribution of listed neotropical migrant songbirds in northeastern British Columbia. In L. Darling (Ed.) At Risk: Proceedings of a conference on the biology and management of species and habitats at risk. Ministry of Environment, Lands and Parks, Victoria, BC.

183 Chytyk, P., Cooper, J. and S. Bennett. 2000. Northern goshawk inventory of TFL 14, Chetwynd, BC. Prepared for Canfor – Chetwynd Operations, Chetwynd, BC. Bennett, S. 2000. Raptor survey in the West Kokanee Creek Area. Prepared for the Ministry of Forests, Kootenay Lake District, Nelson BC. Bennett, S. and P. Corbett. 1999. Boreal Owl Surveys in the Engelmann Spruce Subalpine Fir Zone of the Arrow Forest District: 1995-1999. Prepared for Ministry of Forests, Arrow District Office, Castlegar, BC. Bennett, S., P. Sherrington, and W. Schaffer. 1998. Northern goshawk and diurnal raptor inventory of the Fort Nelson Forest District. Prepared for Ministry of Environment, Lands and Parks, Habitat Protection Branch, Fort St. John, British Columbia. Bennett, S. 1999. Fort Nelson Forest Bird Inventory, 1998: Smith and Dunedin Drainages. Prepared for Ministry of Environment, Lands and Parks, Fort St. John, BC. Bennett, S. 1999. Fisheries and Wildlife Habitat Assessment of the Trans Canada Highway Upgrade West of Revelstoke. Prepared for Ministry of Transportation and Highways, Victoria, BC. Bennett, S. and P. Corbett. 1998. TFL 14 Stream Crossing and Ford Assessments. Prepared for Crestbrook Forest Industries. Project #: 97WRP-FRBC-15 Bennett, S., P. Corbett, T. Elhers. 1997. Fisheries and Wildlife Habitat Assessment of the Trans Canada Highway Upgrade West of Revelstoke: Final Report. Prepared for the Ministry of Transportation and Highways, Highway Environment Branch, Victoria, British Columbia Robertson, I., S. Bennett, and N. Page 1997. Downton Reservoir Project: Wildlife Surveys – Spring 1997 and preliminary assessment of 1996 deep drawdown. Unpublished report for BC Hydro, Vancouver, BC. Bennett, S. and Corbett. 1996. Level 1 Fish Habitat Assessment of Caribou and McMurdo Creek: A Watershed Restoration Project. Prepared for Forest Renewal BC. Bennett, S. and P. Corbett. 1996. TFL 14 Owl, woodpecker, and wildlife tree inventory: 1996 progress report. Prepared for Ministry of Environment, Lands and Parks, Nelson, BC.

184 Bennett, S. and K. Enns. 1996. A bird inventory of the Boreal White and Black Spruce Biogeoclimatic Zone near the Big-Bend of the Liard River. Prepared for BC Environment, Ft St. John, BC. Addison, J. Elhers, T., and S. Bennett. 1995. TFL 55 stream inventory. Prepared for Evans Forest Products, Malakwa, BC. Robertson, I. G. Ryder, Bennett, S., D. Corbett, and P. Corbett. 1995. Keenleyside Powerplant Project: Environmental Assessment Study Land Resources. Prepared for BC Hydro, Vancouver, BC. Bennett, S. 1995. Gable Creek owl and woodpecker survey: final report. Prepared for Ministry of Environment, Lands and Parks, Penticton, BC. Gyug, L. and S. Bennett. 1995. Bird use of wildlife tree patches 25 years after clearcutting. Prepared for Ministry of Environment, Lands and Parks, Penticton, BC. Gyug, L. and S. N. Bennett 1995. Bird use of wildlife tree patches 25 years after clearcutting. BC Ministry of Environment, Penticton, BC. Bennett, S. 1994. Initiatives in wildlife tree management: an evaluation of the high-cut stumping technique. Masters Thesis, School of Resource and Environmental Management, Report 152, Simon Fraser University, Burnaby, BC. 75 p. Bennett, S. N. 1994. Revelstoke TSA stream inventory. Prepared for the Ministry of Forests, Revelstoke, BC. Bennett, S. and P. Johnstone. 1992. Bird Use of Seed Trees in the Saunier Creek Drainage. Prepared for the Ministry of Environment, Lands and Parks, Penticton, BC.