Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Bull trout (Salvelinus confluentus) spawning migrations in the Duncan River : insights from telemetry… O'Brien, David Sean 2001

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

Item Metadata

Download

Media
831-ubc_2001-0258.pdf [ 6.36MB ]
Metadata
JSON: 831-1.0089986.json
JSON-LD: 831-1.0089986-ld.json
RDF/XML (Pretty): 831-1.0089986-rdf.xml
RDF/JSON: 831-1.0089986-rdf.json
Turtle: 831-1.0089986-turtle.txt
N-Triples: 831-1.0089986-rdf-ntriples.txt
Original Record: 831-1.0089986-source.json
Full Text
831-1.0089986-fulltext.txt
Citation
831-1.0089986.ris

Full Text

J B U L L TROUT (SALVELINUS CONFLUENTUS) SPAWNING MIGRATIONS IN THE D U N C A N RIVER: INSIGHTS F R O M T E L E M E T R Y A N D D N A . David Sean O'Brien BSc, University of British Columbia, 1992 A THESIS SUBMITTED IN PARTIAL F U L F I L M E N T OF T H E REQUIREMENTS F O R T H E D E G R E E O F M A S T E R O F S C I E N C E in T H E F A C U L T Y O F G R A D U A T E S T U D I E S ( D E P A R T M E N T OF Z O O L O G Y ) We accept this thesis as conforming to the required standard University of British Columbia April 2001 © David Sean O'Brien, 2001 In p r e s e n t i n g t h i s t h e s i s i n p a r t i a l f u l f i l m e n t of the requirements f o r an advanced degree at the U n i v e r s i t y of B r i t i s h Columbia, I agree that the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r reference and study. I f u r t h e r agree that permission f o r extensive copying of t h i s t h e s i s f o r s c h o l a r l y purposes may be granted by the head of my department or by h i s or her r e p r e s e n t a t i v e s . I t i s understood that copying or p u b l i c a t i o n of t h i s t h e s i s f o r f i n a n c i a l gain s h a l l not be allowed without my w r i t t e n permission. Department of The U n i v e r s i t y of B r i t i s h Columbia Vancouver, Canada Date /jf)yii 23; Zoo! Abst rac t R a d i o telemetry and microsatell i te D N A analyses were used to descr ibe s p a w n i n g migrat ions and the spatial scale o f genetic differentiat ion a m o n g popula t ions o f b u l l trout (Salvelinus confluentus) in the D u n c a n R i v e r system, southeastern B r i t i s h C o l u m b i a . O v e r two years, 66 radio- tagged b u l l trout were tracked to destinations in the u p p e r D u n c a n R i v e r . T h e distr ibution o f b u l l trout b y destination in the upper D u n c a n R i v e r and m i g r a t i o n t i m i n g a m o n g al l destinations d i d not v a r y between the two study years. O n e third o f b u l l trout t racked for two s p a w n i n g migrat ions switched s p a w n i n g stream f r o m one year to the next. T h e r e was also a trend for increasing size o f radio-tagged b u l l trout with s p a w n i n g stream size. T h i s suggests that b u l l trout in the upper D u n c a n R i v e r do not have f idel i ty at the scale o f i n d i v i d u a l s p a w n i n g stream. M i c r o s a t e l l i t e D N A analysis o f j u v e n i l e b u l l trout f r o m two s p a w n i n g streams in the upper D u n c a n R i v e r indicate that there is little between stream variat ion. E x a c t tests o f p o p u l a t i o n different iat ion over allele frequencies o f six p o l y m o r p h i c microsatel l i te loc i suggest no s ignif icant genetic differentiat ion between the t w o s p a w n i n g streams. A n analysis o f m o l e c u l a r var iance ( A M O V A ) parti t ioned a s m a l l (< 1%) but s igni f i cant a m o u n t o f genetic variance to the between s p a w n i n g stream variance parti t ion. T h e estimate o f between s p a w n i n g stream F S T was 0.01. T h e s e estimates o f l o w genetic var ia t ion between s p a w n i n g streams, espec ia l ly w h e n c o m b i n e d w i t h direct ev idence f r o m telemetry, suggest that b u l l trout in the upper D u n c a n R i v e r do not h o m e at the scale o f i n d i v i d u a l s p a w n i n g streams. Table of Contents Abstract u Table of Contents • »' Table of Tables v Table of Figures v" Acknowledgements >x C H A P T E R 1 - G E N E R A L I N T R O D U C T I O N 1 C H A P T E R 2 - B U L L T R O U T (SALVELINUS CONFL UENTUS) R A D I O T E L E M E T R Y IN T H E D U N C A N R I V E R : M I G R A T O R Y T I M I N G A N D INSIGHT I N T O T H E M E C H A N I S M O F S P A W N I N G S T R E A M S E L E C T I O N 6 2.1 INTRODUCTION 6 Study Site 8 2.2 METHODS 11 Capture and tagging U Tracking Data analysis ^ 2.3 RESULTS 1 9 Capture and tagging 19 Destinations in the upper Duncan River 20 Migratory timing comparisons 23 Multi-year migrations 28 Houston Creek fish trap 29 2.4 DISCUSSION 30 Bull trout migrations in the upper Duncan River 31 Spawning stream switching '. 33 Houston Creek fish trap 34 Conservation implications 34 2.5 APPENDIX 3 6 C H A P T E R 3 - A N E V A L U A T I O N O F G E N E T I C D I F F E R E N T I A T I O N B E T W E E N T W O B U L L T R O U T (SALVELINUS CONFL UENTUS) S P A W N I N G S T R E A M S IN T H E U P P E R D U N C A N R I V E R USING M I C R O S A T E L L I T E S 38 3.1 INTRODUCTION 38 3.2 METHODS 4 2 Tissue collection 42 iii Microsatellite amplification 47 Data analysis 47 3.3 R E S U L T S 49 Microsatellite loci 49 Genetic population structure 53 3.4 DISCUSSION 55 Number of migrants 59 Comparison to previous bull trout studies 61 Conservation implications 63 3.5 A P P E N D I X 64 C H A P T E R 4 - G E N E R A L C O N C L U S I O N S 71 Migratory timing and spawning locations in the upper Duncan River 71 Spawning stream selection 71 Population structure and gene/low 72 Comparison between methods 75 Conservation and management implications 74 L I T E R A T U R E C I T E D 76 iv Table of Tables Table 2-1. Characteristics of radio tags employed in the study 11 Table 2-2. Migratory destinations and A N O V A groups of radio-tagged bull trout in the upper Duncan River. Destination 3, mainstem upper Duncan River, is broken into five sub-categories (3.a. - 3.e., italics) which are separated by tributary confluences (Fig. 2-6) 21 Table 2-3. Two factor A N O V A comparisons of migration data for radio-tagged bull trout by analysis group (see Table 2.2) and year of the study 24 Table 2-4. The catchment area of tributaries targeted by radio-tagged bull trout in 1995 and 1996 26 Table 2-5. Overwinter out-migration dates of individual radio-tagged bull trout through the discharge structure at Duncan Dam by year of migration to the upper Duncan River 28 Table 2-6. Destinations of radio-tagged bull trout migrating to the upper Duncan River in two consecutive years 29 Table 2-7. The biological data collected from radio-tagged bull trout migrating to destinations in the upper Duncan River. 'Ch. ' is tag channel, 'Floy' is the number of the external tag, ' F L ' is fork length in mm, and mass is measured in kg 36 Table 2-8. Summary of migration statistics and destinations of all radio-tagged bull trout migrating to destinations in the upper Duncan River (used in 2 way A N O V A ) . Tn' is in-migration, 'Res' is residence, 'Out' is out-migration, 'Dest' is destination (see Table 2-2) rate units are km*day"1, and time units are days 37 Table 3-1. Allele size (bp) and frequency (freq.) over all bull trout samples for nine microsatellite loci 51 Table 3-2. Sample size (n), number of alleles amplified (A), and observed (H (obs)) and expected (H (exp)) heterozygosity values by stream, age class group, and locus, as well as estimated probability values (P) of H W E tests 52 Table 3-3. Estimated probability values of exact tests of population differentiation from allele frequencies across loci between all age class groups (0+ - 2+) from Houston Creek (H) and Westfall River (W) 53 Table 3-4. Estimates of FST and RST within and across loci for pooled age class samples by stream. Confidence intervals (C.I.) for across loci estimates are included.... 54 Table 3-5. Estimated probability values of exact tests of population differentiation, based on allele frequency differences (Infinite Alleles Model), between Houston Creek and Westfall River samples (pooled across age classes) within loci, and across all loci (bold). 'SE' is the standard error of the estimate 54 v Table 3-6. Analysis of Molecular Variance ( A M O V A ) partitioning for both pooled and separate age classes by stream. Estimated probability values associated with variance partitions are included. 'SE' is the standard error of the estimate 55 Table 3-7. Loci specific PRC protocols. Reaction step abbreviations are: D - denature, A - anneal, E - extend. Time notation is m:ss. The reactions of starred loci included DMSO (see text) 64 v i Table of Figures Figure 2-1. Map of the Duncan River system and Kootenay Lake. Inset map locates the study area within British Columbia 10 Figure 2-2. Dates of tracking flights in the Duncan River system 14 Figure 2-3. Kootenay Lake and the Duncan River with kilometer designations used throughout the study. The location of data logging telemetry fixed stations (circles) and the Houston Creek trap (square) are indicated 16 Figure 2-4. Example of a bull trout migration pattern based on radio telemetry location fixes (connected squares) showing the function of the residency mark in defining the boundaries between the in-migration, residence and out-migration phases. 18 Figure 2-5. The fork length (FL) frequency histograms of all bull trout captured at Duncan Dam in 1995 and 1996 ( 'DD' - outline and right axis, n = 533) and of all radio-tagged bull trout migrating to destinations in the upper Duncan River ('RT' - gray bars and left axis, n=66) 20 Figure 2-6. Map of the upper Duncan River system with kilometer designations upstream of Duncan Dam 22 Figure 2-7. Data used in the calculation of bull trout fork length by destination and year two factor A N O V A . Open circles and solid squares are 1995 and 1996 data, respectively. Error bars are ± 1 SE. Destinations ( A N O V A groups - Table 2.2) are approximately (see text) ordered by stream size from largest to smallest (left to right). Sample sizes are indicated next to data points 24 Figure 2-8. The regression of catchment size to instream flow estimate of five tributaries of the upper Duncan River. The regression (y = 0.029x - 0.636; solid line) is significant (P < 0.001, r 2 = 0.997) 26 Figure 2-9. Fork length of radio-tagged bull trout (n = 49) by catchment area of destination tributaries in the upper Duncan River. The regression (y = 0.301x + 669.2; solid line) is significant (P = 0.03, r 2 = 0.095). A linear regression excluding the largest fish from this figure (open circle) is not significant (P = 0.08, see text) 27 Figure 2-10. The fork length (FL) frequency histogram of bull trout captured at the Houston Creek trap in 1996 (gray bars, n = 47) 30 Figure 3-1. Juvenile bull trout tissue sampling locations (n = 14) in Westfall River 44 Figure 3-2. Juvenile bull trout tissue sampling locations (n = 13) in Houston Creek 45 Figure 3-3. Fork length frequency of juvenile bull trout captured in the Westfall River, 15 July - 12 August 1997 (n = 167) 46 Figure 3-4. Fork length frequency of juvenile bull trout captured in the Houston Creek, 20 - 22 August 1997 (n = 104) 46 vii Figure 3-5. A composite autoradiograph of the six polymorphic microsatellite loci scored in bull trout from Houston Creek and the Westfall River. Sample lanes are indicated with an 'S'. Allele sizes (bp) are indicated (arrows) 50 Figure 3-6. The allele frequencies of microsatellite locus Ogo 2 from sampled bull trout by age class (0+, 1+, and 2+ top to bottom) and stream (Houston Creek - right, Westfall River - left) 65 Figure 3-7. The allele frequencies of microsatellite locus Omy 11 from sampled bull trout by age class (0+, 1+, and 2+ top to bottom) and stream (Houston Creek - right, Westfall River - left) 66 Figure 3-8. The allele frequencies of microsatellite locus Sco 19 from sampled bull trout by age class (0+, 1+, and 2+ top to bottom) and stream (Houston Creek - right, Westfall River - left) 67 Figure 3-9. The allele frequencies of microsatellite locus Sco 23 from sampled bull trout by age class (0+, 1+, and 2+ top to bottom) and stream (Houston Creek - right, Westfall River-left) 68 Figure 3-10. The allele frequencies of microsatellite locus Ssa 197 from sampled bull trout by age class (0+, 1+, and 2+ top to bottom) and stream (Houston Creek -right, Westfall River - left) 69 Figure 3-11. The allele frequencies of microsatellite locus Ssa 456 from sampled bull trout by age class (0+, 1+, and 2+ top to bottom) and stream (Houston Creek -right, Westfall River - left) 70 viii Acknowledgements This thesis would not have been possible without the contribution of many people, and their assistance is gratefully acknowledged. My supervisor, Dr. Don McPhail, and supervisory committee members, Dr. Eric Taylor and Dr. Robin Liley, provided ideas, encouragement and assistance all along the way. Special thanks to my committee for much needed conceptual and editorial comment during the crunch. Thanks for the love, encouragement and support provided by my parents during years of post secondary study. Ric Olmsted (BCHydro), Colin Spence (MELP), John Bell (MELP), Harald Manson (CBFWCP), Steve Arndt (CBFWCP) and Maureen DeHaan (CBFWCP) provided a wealth of experience, ideas and encouragement that was vital to the success the field portion of the research. Rayanne McKay and Mark Phillpotts provided excellent assistance in the field. Murray Pearson, Peter Mylechreest, Gordon Haas, James Baxter, Ernest Keeley, Jason Ladell, Ian Kusabs, Steve Latham, Mike Stamford, Janelle Curtis, the unnamed Tay-Mac's, Abraham bij de Vaate, Lynda Ritchie, Gary Birch, Dean den Biesen and Mark Hamilton gave technical help, advice, and support. The Zoology Computing Unit, and particularly Alistair Blachford, assisted with digitizing maps and general analysis help. Thanks to Ken Ashley for first getting me involved in this study. The staff at Duncan Dam, especially Len Wiens, Ray Wheldon and the late Dan Burt went out of their way on many occasions to assist my work. I would like to acknowledge 'Dutchy' Wageningen for having the foresight to let bull trout through the dam in the first place. The Columbia Basin Fish & Wildlife Compensation Program provided project funding and field equipment. NSERC, through grants to D. J. McPhail and E. B. Taylor, supported a portion of operating costs in the lab. BCHydro provided safety equipment and training for Duncan Dam work, and a jet boat used to access the Duncan River. The Ministry of Environment, Lands and Parks provided staff housing, fence material, and fixed receiver stations. Chapter 1 - General Introduction The way species are subdivided into reproductive groups, or populations, is critical for conservation and management (Vogler and DeSalle 1994; Bohonak 1999). Different populations within a species often reflect different current and recent historical evolutionary events. Preserving this within species genetic diversity is thought ensure long-term species viability (Moritz 1994; Avise 1995). To preserve, or manage, this diversity requires an understanding of the geographic scale over which populations occur. Divergence among populations is a function of the strength of local selection and the amount of gene flow among populations. Important determinants of the amount of gene flow between two populations include dispersal ability and the presence and location of barriers to dispersal (Waples 1998; Bohonak 1999). In the absence of barriers to dispersal, the magnitude of migration and gene flow between otherwise separate populations is partly a function of geographic proximity. Adjacent populations exchange migrants more frequently than populations separated by greater distances. This phenomenon is called "isolation by distance" (Wright 1943; Slatkin 1993). In organisms that migrate between feeding and breeding locations, the level of population divergence is often a function of degree of fidelity to discrete reproductive sites. If individuals born in different places home (i.e. return) to specific reproductive sites, this could influence interpopulation divergence. Homing decreases gene flow among reproductive sites and, potentially, can re-enforce geographic isolation among populations (Taylor 1991). Without gene flow, populations in different environments will adapt to local conditions, and diverge (Barton and Hewitt 1989). Even in the absence of strong selection, isolated populations will diverge due to genetic drift (Wright 1931); however, the migration of even a small number of reproductive individuals between populations will prevent the fixation of local genotypes (Wright 1931; Haiti and Clark 1989; Mills and Allendorf 1996). From a management or conservation perspective, differentiating among divergent populations is, therefore, a species specific task. The processes leading to interpopulation divergence are well studied in salmonid fishes. 1 Generally, salmonid fishes tend to home (i.e. return to their natal streams), albeit with varying degrees of fidelity (Scheer 1939; Banks 1969; Nordeng 1977; Quinn 1993; McElhany et al. 2000). In these fish, homing typically leads to geographic isolation and local adaptation, often within individual spawning streams (Scheer 1939; Quinn and Dittman 1990; Taylor 1991). Homing probably is especially important for semelparous Pacific salmon (Oncorhynchus spp.): it increases the probability that the one chance for reproduction is successful. Banks (1969) commented that iteroparous Atlantic salmon (Salmo salar) may home less accurately than semelparous Pacific salmon, since Salmo has scarcely speciated relative to Oncorhynchus. Management and conservation of salmonid fishes focuses, or attempts to focus, at the level of genetically distinct units (Allendorf and Phelps 1981; Taylor 1991; Utter and Ryman 1993; Allendorf et al. 1997). Critically important to this management objective is knowledge regarding where genetically distinct groups occur and, in the case of salmonid fishes, this often corresponds to specific spawning streams (Ryman 1983; Taylor 1991). Bull trout (Salvelinus confluentus), is an iteroparous salmonid native to northwestern North America (Cavender 1978; Haas and McPhail 1991). Historically, bull trout ranged over 25° of latitude from the McCloud River in northern California to the Mackenzie River in the North West Territories and between the Pacific Ocean and the eastern slopes of the Rocky Mountains (Haas and McPhail 1991; McPhail and Baxter 1996). Because of population declines in the U.S., bull trout in that country are listed as threatened under the Endangered Species Act (Barry 1999). Usually, these declines are attributed to habitat degradation, habitat fragmentation, and exotic species introductions (Rieman and Mclntyre 1993; Watson and Hillman 1997). Bull trout are long lived (potentially > 20 years) that occur as one of four life history types: resident, fluvial, lacustrine, or anadromous (Fraley and Shepard 1989; Goetz 1989; McPhail and Baxter 1996). Bull trout spawn in running freshwater in the fall. Like other Salvelinus species, they are adapted to cold water, and their distribution suggests strict temperature requirements (Rieman and Mclntyre 1993; Parkinson and Haas 1996). In all but the resident life history form, bull trout migrate between geographically separate spawning and feeding areas. Eggs incubate over winter and hatch early in spring. Juvenile bull 2 trout remain in the natal stream for as long as three summers before moving to either larger streams, lakes or nearshore marine areas (McPhail and Murray 1979; Goetz 1989; Haas and McPhail 1991; McPhail and Baxter 1996). Age of first maturity may be four years, although more commonly bull trout do not mature until age five to seven (Allen 1980; Fraley and Shepard 1989; Goetz 1989). Bull trout spawning migrations are not well studied. In a review of migratory timing information, McPhail and Baxter (1996) found that there was no appreciable difference in spawning time across 20° of latitude. Generally, bull trout spawning migrations begin in the spring or early summer, with fish arriving at a spawning destination in early to late August. Spawning occurs from early September to mid October, with fish leaving spawning areas shortly after spawning is completed (McPhail and Baxter 1996). Several studies have addressed migratory timing with radio telemetry (McLeod and Clayton 1994; Schill et al. 1994; Swanberg 1997; O'Brien and Chamberlain 2000). These telemetry studies determine timing of migrations more specifically for particular regions, and help to define seasonal habitat use by bull trout for conservation and management strategies. Telemetry also can help identify limits and barriers to dispersal. Genetic studies suggest that bull trout occur as two major phylogenetic groups: a coastal and interior group (Spruell and Allendorf 1997; Taylor et al. 1999). This broad scale structure is attributed to recent historical isolation within two glacial refugia south of the Cordilleran ice sheet (Taylor et al. 1999). Genetic studies also generally find low genetic diversity in restricted geographic areas, such as within major watersheds of large river systems like the Columbia and Fraser, with relatively high differentiation between such regions (Leary et al. 1993; Spruell and Allendorf 1997; Williams et al. 1997; Costello and Taylor unpubl. MS). Most evidence suggests that bull trout home accurately to natal streams for reproduction (as reviewed by Goetz 1989), although some studies have suggested that homing may not be accurate at the scale of natal stream (McPhail 1979; Oliver 1979; see also references in Goetz 1989). My thesis focuses on the application of radio telemetry and molecular genetic techniques to an investigation of the relationships between bull trout spawning 3 migrations, patterns of movement among potential spawning locations, and the spatial scale of population structure. I have organized the thesis into two primary chapters around the telemetry and molecular genetic methods I used. Each primary chapter is written as a stand-alone manuscript; although, because of repeated citations in each chapter, only one list of cited literature is presented. This organization, although convenient for later publication of results, requires repetition of introductory and discussion material in each chapter. The chapters are structured as follows: Chapter 2. Bull trout (Salvelinus confluentus) radio telemetry in the Duncan River: migratory timing and insight into the mechanism of spawning stream selection. This chapter presents results of a three year (1995-97) radio telemetry study of bull trout migrations in the Duncan River, a tributary to Kootenay Lake in southeastern British Columbia. I describe the migratory timing of bull trout moving from Kootenay Lake to destinations in the upper Duncan River. I compare the number of bull trout migrating to specific destinations over two years, and test whether the size of radio-tagged bull trout is related to the size of the tributaries to which the fish migrated. I estimate of the total number of bull trout in the upper Duncan River in one year on the basis of migrants trapped leaving a spawning stream. I also describe the migration of six bull trout to the upper Duncan River in two consecutive years, and changes in their migratory destinations between years. I use my estimate of total numbers from the trap, and the proportion of fish changing destination to provide a rough estimate of the magnitude of switching among spawning streams. Chapter 3. A n evaluation of genetic differentiation between two bull trout (Salvelinus confluentus) spawning streams in the upper Duncan River using microsatellites. I present microsatellite allele frequency data from juvenile bull trout collected in two spawning tributaries to the upper Duncan River. From these data, I calculate measures of population differentiation and examine population divergence among these 4 spawning streams. Under the assumptions of the island model of population structure (Wright 1943), I estimate the number of migrants between the two spawning streams. I compare this estimate of migration qualitatively with the telemetry results presented in Chapter 2. I discuss my results in relation to other genetic studies of bull trout at similar geographic scales. Chapter 4. General Conclusions In this chapter, I summarize the results presented in Chapters 2 and 3 as they relate to bull trout in the upper Duncan River. I compare the results of the telemetry and genetic methods used. I also discuss the implications of my results in relation to the management and conservation of bull trout in general. 5 Chapter 2 - Bull trout {Salvelinus confluentus) radio telemetry in the Duncan River: migratory timing and insight into the mechanism of spawning stream selection 2.1 Introduction Bull trout (Salvelinus confluentus) is an iteroparous salmonid native to northwestern North America (Cavender 1978; Haas and McPhail 1991). Prior to the taxonomic work of Cavender (1978) and Haas and McPhail (1991), bull trout were considered Dolly Varden (Salvelinus malma). Historically, bull trout ranged over 25° of latitude from the McCloud River in northern California to the Mackenzie River in the North West Territories and between the Pacific Ocean and the eastern slopes of the Rocky Mountains (McPhail and Baxter 1996). Because of population declines in the U.S., bull trout are listed as threatened under the Endangered Species Act (Barry 1999). Usually, these declines are attributed to habitat degradation, habitat fragmentation and exotic species introductions (Rieman and Mclntyre 1993; Watson and Hillman 1997). In British Columbia (BC), bull trout are 'blue listed', meaning that they are a species of special conservation concern. Bull trout are the only fish species identified by the BC Forest Practices Code as requiring special forest harvesting related management practices (Haas 1998). Bull trout are long lived (potentially > 20 years; McPhail and Baxter 1996), spawn in running freshwater, and occur as one of four life history types: resident, fluvial, lacustrine or anadromous (Fraley and Shepard 1989; Goetz 1989; McPhail and Baxter 1996). Spawning occurs in the fall. Eggs incubate over winter and hatch in spring. Juvenile bull trout remain in the natal stream for as long as three summers before moving to larger streams, lakes or nearshore marine areas (McPhail and Murray 1979; Goetz 1989; Haas and McPhail 1991; McPhail and Baxter 1996). Age of first maturity may be four years, although more commonly bull trout do not mature until age five to seven (Allen 1980; Fraley and Shepard 1989; Goetz 1989). Spawning frequency is uncertain, although it is unlikely that spawning occurs every year after first maturity (Allen 1980; Goetz 1989; McPhail and Baxter 1996). With the exception of the resident life history, bull trout 6 migrate between geographically separate spawning and feeding areas. Bull trout migrations are often in excess of 200 km (Clayton and Mcleod 1993; Schill et al. 1994; Swanberg 1997; O'Brien and Chamberlain 2000), and in rare instances, where movement is not blocked, can exceed 500 km (Burrows et al. in press). Until recently, little quantitative migration data were available. Much of the literature on bull trout migratory behaviour consists of provincial and state fisheries agency reports. Two reviews of this literature (Goetz 1989; McPhail and Baxter 1996) have summarized data from throughout the range of bull trout. In general, it is known that bull trout migrate throughout the summer, arriving at spawning locations from mid August to early October. Spawning activity has been observed as early August (Metolius River tributaries - Ratliff 1987 as cited by Goetz 1989) and as late as October (MacKenzie Creek - McPhail and Murray 1979). In their review, McPhail and Baxter (1996) found no suggestion of differences in the timing of spawning across 20° latitude. It appears that the timing of spawning, and presumably the timing of spawning migration, varies little over the range of bull trout. Swanberg (1996; 1997) conducted a radio-telemetry project in the Blackfoot River in Montana. Bull trout were collected from locations in the Blackfoot River, radio-tagged, and tracked to spawning locations in tributary streams. Spawning migrations began as early as June, with bull trout arriving at spawning locations in early July. Spawning did not take place until late September. Swanberg (1997) documented strong fidelity to feeding locations in the mainstem Blackfoot River. After round trip spawning migrations of up to 150 km, 86 % of bull trout returned to within 20 m of their original point of capture in the mainstem river (Swanberg 1997). Schill et al. (1994) examined bull trout spawning migrations with radio telemetry in the Rapid River, a tributary of the Salmon River in Idaho. Movement into the Rapid River began as early as April 30, and bull trout generally arrived at spawning locations in early August. Spawning occurred in early September. Schill et al. (1994) documented high post spawning mortality in the Rapid River: only 31% of the tagged bull trout left the Rapid River and returned to feeding areas in the Salmon River. It is usually assumed that bull trout home to their natal streams (Goetz 1989); however, despite suggestions that this may not be the case (McPhail and Murray 1979), 7 the mechanism of spawning stream selection by bull trout has not been examined. McPhail and Murray (1979), based on observations from two years of data from a trap on a small bull stream in British Columbia, suggested that as bull trout age and grow, they switch to larger spawning streams. They based this suggestion on the observation of one fish returning to McKenzie Creek in both study years, only to leave the stream without spawning in the second year. The year it left without spawning, this fish was the largest bull trout to have entered the creek. McPhail and Murray (1979) also found that the mean size of bull trout spawning in McKenzie Creek was smaller than the mean size of bull trout captured by anglers from Upper Arrow Lake, into which the creek flows. In addition, they observed that larger streams flowing into Upper Arrow Lake contained larger spawning bull trout. I used radio telemetry to examine the migratory behaviour of bull trout in the Duncan River, BC. I wanted to determine the extent and timing of spawning migrations of bull trout to destinations in the upper Duncan River. In addition, I wanted to identify bull trout spawning locations. I also tested whether there was a positive relationship between size of destination stream and size of fish as suggested by McPhail and Murray (1979). My study was conducted within a larger project designed to identify bull trout spawning locations in the upper Duncan River and examine the effect of a water storage dam on the Duncan River on bull trout migrations. The questions I addressed were • What is the timing of bull trout migrations in the Duncan River? • What streams in the upper Duncan River are used by bull trout for spawning? • Is there a relationship between size of bull trout and the size of their chosen spawning stream, as suggested by McPhail and Murray (1979)? Study Site The Duncan River, located in southeastern British Columbia, drains from the Selkirk and Purcell mountains into Kootenay Lake and, ultimately, into the Columbia River (Fig. 2-1). On 31 July 1967, Duncan River was impounded by the closure of Duncan Dam, located approximately 10 kilometers upstream of the river's confluence with Kootenay Lake. Duncan Dam is 40 m high and creates a 72 km reservoir. The dam provides water storage (1.73 billion m 3), but has no power generation facilities 8 (Anonymous 1986). Upstream of the Duncan Reservoir, the upper Duncan River drains an area of nearly 1300 km . The migration route of bull trout from feeding areas in Kootenay Lake to spawning locations in the upper Duncan River requires passage through Duncan Dam. Studies prior to 1967 predicted that the migration route of bull trout would be blocked by impoundment (Peterson and Withler 1965). In 1968, the first senior operator at the dam, 'Dutchy' Wageningen, altered spring and summer discharge at Duncan Dam to allow bull trout access to the upper river. Currently, this 'fish transfer' procedure continues under the supervision of BCHydro, the BC Ministry of Environment, Lands and Parks - Fisheries Section (MELP), and Len Wiens - the present senior operator. Tag recapture information prior to this study suggested that bull trout are able to ascend the dam (Fleck 1977; J. Bell, M E L P , pers. comm.), but in 1994, aerial surveys of the upper Duncan River system failed to locate evidence of spawning (C. Spence, M E L P , pers. comm.). Tagging studies of bull trout from the Duncan River provide evidence of migrations throughout the larger Kootenay River system, with tag returns ranging from Trout Lake in the north to the confluence of the Kootenay and Moyie rivers in the south (Fleck 1977; Fig. 2-1). 9 Figure 2-1. Map of the Duncan River system and Kootenay Lake. Inset map locates the study within British Columbia. 2.2 Methods Capture and tagging During this study, I used several different sized radio tags, which varied in transmission duration and frequency (Table 1). A l l radio tags were digitally encoded. This allowed reception equipment to distinguish individual signals. Lotek Engineering, Newmarket, Ontario manufactured the radio tags and reception equipment. Table 2-1. Characteristics of radio tags employed in the study. Channel Frequency Number Duration Diameter Length Antenna Mass (g) (MHz) deployed (months) (mm) (mm) length (mm) 5 149.400 31 12 11 50 400 8.2 15 148.400 16 6 11 45 370 7.1 20 149.700 15 24 15 50 400 15.5 22 149.740 16 12 11 50 400 8.2 31 150.120 18 12 11 50 400 8.2 I obtained bull trout for radio tagging from four locations in the Duncan River system: the discharge structure at Duncan Dam, the tailrace channel immediately downstream of the dam, the lower Duncan River (between the dam and Kootenay Lake), and the upper Duncan River. Bull trout were captured by angling with spoons at locations other than the discharge structure at Duncan Dam. Those captured at the dam were fish that had actively moved into the flip bucket at the downstream end of the discharge tunnel. Presumably they were attempting to move through the dam. By stopping flow through the tunnel, and lowering water levels in the flip bucket with a pump, I was able to access bull trout gathering there. BCHydro staff at Duncan Dam facilitated my access to bull trout in the flip bucket by manipulating discharge, controlling pumps and assisting with fish capture. A l l bull trout present in the drained flip bucket were tagged with external individually numbered 'T-anchor' style tags (Floy Tag Inc. Seattle, Washington) and measured (fork length) with a flexible tape. BCHydro, Ministry of Environment and Columbia Basin Compensation Program staff and volunteers assisted with capture and tagging of bull trout in the flip bucket. I selected bull trout from the flip bucket captures for radio tagging. 11 I did not select bull trout randomly for transmitter application. The weight (in water) of surgically implanted tags should not exceed 2% of the weight (in air) of the fish destined to carry it, and ideally this ratio should not exceed 1% (Winter 1983; Marty and Summerfelt 1986; McLeod and Clayton 1993). By selecting bull trout that exceeded 500 mm fork length (FL), in all cases I kept the weight ratio below 1%. I attempted to radio tag only male bull trout; however, sex determination before surgery was difficult, because few secondary sexual characteristics were evident. I targeted males to limit effects of radio tagging on reproduction. When possible during surgery, I determined sex by internal examination of the gonads. I anaesthetized selected fish in a cooler containing 30 L of water with tricane methane sulfonate (MS222) at 100 ppm. Whenever possible, I used the anesthetic bath on two or three fish that did not undergo surgery before anaesthetizing a surgical candidate. This was done to lower the stress of reduced pH brought about by the addition of MS222 (Bidgood 1980). I anaesthetized bull trout to stage 4 for surgery. Stage 4 anesthesia is characterized by loss of both equilibrium and reactivity to external stimuli as well as a slow but regular opercular rate (McKinley et al. 1992). Before surgery, I measured mass using a spring scale with the fish supported in either a collapsible fish tube or weighing bag of known mass. Subsequent to measuring, a radio tag was inserted into the abdominal cavity (as described by McLeod and Clayton 1993). Briefly, I placed the anaesthetized fish on its back in a wet neoprene lined wooden 'V'-trough and covered it, except for the region of incision, with moist towels. In 1995,1 made a 2.5 cm long incision into the abdominal cavity immediately anterior to the pelvic girdle and approximately 1 cm off, and parallel to, the mid-ventral line. In 1996, the incision was made directly on the mid-ventral line. This was done to reduce damage to the ventral musculature. I then inserted a stainless steel needle (16 gauge, 5 cm long) through the body wall from posterior to and above the pelvic girdle into the incision. I threaded the whip antenna of the radio tag through the needle, removed the needle, and inserted the tag into the abdominal cavity. I closed the incision with three or four interrupted sutures and applied betadine to the closed incision and antenna exit wound. To recover the fish after surgery, they were held directly in a stream of flowing water. When the fish was capable of maintaining equilibrium (stage 2 anesthesia - McKinley et al. 12 1992), it was placed in a collapsible fish tube and allowed to recover for 30 - 180 min before release. Prior to 28 June 1995, all radio-tagged bull trout captured at Duncan Dam (22 fish) were released in the flip bucket at Duncan Dam. I anticipated that the majority of these fish would be tracked as they ascended the discharge tunnel; however, I did not track any of these initial releases in Duncan Reservoir or the upper Duncan River. After 28 June in 1995,1 released all bull trout radio tagged at Duncan Dam directly into Duncan Reservoir (25 fish). To ensure a large sample size of migrants to the upper Duncan River in 1996,1 released 34 bull trout radio tagged from flip bucket captures into the reservoir. Six bull trout radio tagged at the dam in 1996 were released back into the flip bucket. A l l bull trout that were radio tagged away from the dam were released at the point of capture. In 1996,1 recaptured six bull trout carrying radio tags from 1995. As these tags should have stopped functioning before the end of the migration period in 1996 I attempted to surgically remove old tags and replace them with new ones. In two of the replacement attempts, I could not remove the old tag and so clipped its antenna close to the body to minimize drag before inserting a new tag. In the other replacements I removed the old tag without incident. I recaptured two radio-tagged bull trout in the flip bucket in 1997, and removed (but did not replace) one of the tags. In addition to radio tagging in 1996,1 operated a fish trap on Houston Creek at the confluence of Houston Creek and the upper Duncan River. The trap was operational for a total of 91 h from 24 September to 3 October 1996. The trap was constructed with 2.5 cm 2 plastic coated wire mesh supported by iron fence posts pounded into the stream bottom. The trap box was a 2.5 * 1.5* 1.0 m wood framed box enclosed by 1.0 cm plastic mesh, with a firnnel entrance on the upstream side. I estimated the number of bull trout present during the spawning period in Houston Creek from the number of radio-tagged and untagged fish captured in the trap and the known number of radio-tagged bull trout present in the creek using the Peterson estimate (as described by Greenwood 1996). In an attempt to roughly quantify the number of bull trout migrating into the upper Duncan River in 1996,1 assumed that the proportion of radio-tagged to untagged bull trout was the same in all other location within the upper Duncan River. 13 I also conducted stream walks (late September) in 1995 and 1996 through the portions of Houston Creek, Westfall River and the mainstem upper Duncan River looking for evidence of spawning. These searches focused on areas where radio-tagged bull trout were located. Tracking I used a Lotek SRX-400A version 4.01/W5 receiver to track radio-tagged bull trout by truck, boat and aircraft. While tracking on the ground or by boat, I used either a hand held three element antenna or a short range whip antenna. A Cessna 172 aircraft was used for the majority of aerial tracking, but I tracked from a Eurocopter AS 350B helicopter on two occasions and a Cessna 337 aircraft once. Antenna setup differed by aircraft: the Cessna 172 had a single forward pointing four element antenna attached to the port side wing strut, the AS 350B had a single three element antenna attached to the nose of the aircraft, and the Cessna 337 had a downward-pointing two element antenna attached to each wing strut. In the Cessna 337, a switch-box allowed the receiver to 'listen' to both antennae simultaneously or either individually during tracking to help resolve the direction of receptions. When tracking from aircraft, I recorded location of receptions as both written descriptions and coordinates from a GPS. Generally, I maintained a bi-weekly schedule of tracking flights during the in-migration and spawning phase in 1995 and 1996 (Fig. 2-2). As the first radio-tagged fish began to leave their spawning sites, flight frequency was increased. In 1997 and 1998,1 flew once in the fall to identify number and location of spawning bull trout. 1998 -, ._ 199? co CD 1996 1995 -I • • # • • • • • • • • • • • # • • • • -1 1 1 1 1 1 1—r 1 1 r 1 1 1 1 1 1 1 1 1 H 30-Apr 30-May 29-Jun 29-Jul 28-Aug 27-Sep 27-Oct 26-Nov 26-Dec Date Figure 2-2. Dates of tracking flights in the Duncan River system. 14 I also used two Lotek SRX-400A version 4.01/W17T receivers as data logging fixed stations during 1996. Fixed station sites consisted of a battery-powered (1 deep-cycle 12V battery) receiver connected to two antennae oriented upstream and downstream to resolve direction of travel. Fixed station sites were located at Duncan Dam and 42.5 km upstream of the dam on the upper Duncan River (Fig. 2.3). I programmed fixed receivers to continuously scan each radio tag frequency in sequence for a five-second interval on each antenna, and record any reception in memory. With this scan cycle, each frequency was scanned for a minimum of 12 min each hour. I downloaded reception information to a portable computer when the receiver battery was replaced (bi-monthly). Data analysis I converted raw aerial, ground, and remote reception data to distance upstream (+) or downstream (-) of the Duncan Dam in km. Multiple receptions of an individual bull trout in one day were reduced, by averaging the locations, to a single daily reception. Figure 2-3 illustrates the kilometer designations used throughout the study. To begin analysis of the telemetry data, I characterized the destination of individual radio-tagged bull trout by the tributary or the section of the mainstem upper Duncan River to which they migrated. I then plotted the reception locations for each tagged individual as date vs. distance from Duncan Dam. Only the movements of those bull trout migrating into the upper Duncan River are presented. Unless otherwise noted, all statistics are presented with the 95% confidence intervals. 15 Figure 2-3. Kootenay Lake and the Duncan River with kilometer designations used throughout the study. The location of data logging telemetry fixed stations (circles) and the Houston Creek trap (square) are indicated. 16 I first compared the distribution of bull trout by destination in the upper Duncan River between 1995 and 1996 using a Fisher exact test. I then divided the movements of bull trout migrating into the upper Duncan River into three phases: the in-migration, spawning site residence, and out-migration. I considered the in-migration phase underway when a radio-tagged individual was released after tagging or continuous upstream movement began, and it continued until upstream movement ended. The spawning site residence phase continued from the end of the in-migration phase until a continuous downstream movement began. I considered the out-migration phase complete when a tag was first detected in the Duncan Reservoir (1995) or when it had passed below the fixed station receiver located near the inlet of the upper Duncan River to the Duncan Reservoir (1996). To simplify these data, for each destination category I defined a separate arbitrary distance from Duncan Dam to distinguish between the three migration phases. The residency mark was positioned to include all similar migrants to a particular destination and was based on cessation of upstream movement by those migrants. For example, when a radio-tagged bull trout migrating to Houston Creek (a tributary of the upper Duncan River) crossed a mark 100 km upstream of Duncan Dam, I considered the in-migration phase to have ended and the spawning site residency phase to have begun. Similarly, when the 100 km residency mark was crossed in the downstream direction by bull trout leaving Houston Creek, I considered the spawning site residence phase complete and the out-migration phase to have started (Fig. 2-4). Most often, I did not locate radio-tagged bull trout as they crossed the residency mark; I inferred the date from the closest receptions immediately upstream and downstream. 17 By dividing movement patterns into these three phases (Fig. 2-4), I was able to calculate 9 migration statistics: • in-migration start date / time (days) / rate (knrday"1) • residency start date / time (days), • out-migration start date / time (days) / rate (km»day"') / end date. 2-Jul 22-Jul 11-Aug 31-Aug 20-Sep 10-Oct 30-Oct Date Figure 2-4. Example of a bull trout migration pattern based on radio telemetry location fixes (connected squares) showing the function of the residency mark in defining the boundaries between the in-migration, residence and out-migration phases. I compared the nine migration statistics, along with FL and mass, over year of migration and destination category using two-factor analysis of variance (ANOVA). This study was not designed to fit the A N O V A model and, as such, I used it only as an exploratory tool. The first factor in the A N O V A analysis was the year of migration. For bull trout with two years of migration data in the upper Duncan River, I considered each migration year as independent in terms of the A N O V A analysis. The second factor in the analysis was destination. Because of small sample sizes to some destinations, I grouped low-use destinations into more broad categories for analysis. Where no differences were found by the two-factor A N O V A analysis, I pooled migration statistics. 18 I compared, using t-tests, the migration statistics of bull trout migrating through the discharge structure to those that were moved above, to determine if moving bull trout above Duncan Dam changed migration timing. In order to determine the relative sizes of spawning tributaries in the upper Duncan River system, I estimated the catchment area (km ) of each targeted tributary from 1:50,000 topographic maps. I regressed instream flow estimates of five tributaries calculated from portable flow-meter transects (Anonymous 1990) against catchment area to evaluate the usefulness of catchment area as a predictor of flows in the upper Duncan River system. The five instream flow measurements were made over a two-day period in October 1996 (stable low flows). I then regressed the mean FL of radio-tagged bull trout against catchment area of their destination tributaries. 2.3 Results Capture and tagging Ninety four bull trout were tagged over the duration of the project. Two bull trout died within one hour of the surgical procedure (prior to release). Anaesthesia time averaged 360 ± 28 s (n = 92). Surgery time, measured as the total time from removal of the fish from the anaesthetic to the start of recovery, averaged 628 ± 3 1 s (n = 95). Post surgery recovery, from removal of the radio-tagged bull trout from the surgical trough until equilibrium and reactivity was regained, averaged 351 ± 35 s (n = 94). The biological data collected from radio-tagged bull trout that migrated to destinations in the upper Duncan River are appended (n = 66). I found no significant difference in fork length and weight distributions of radio-tagged bull trout across years (t-tests, P = 0.882 and 0.468, respectively) and pooled these data. Fork length (Fig. 2-5) and mass of only the radio-tagged bull trout migrating to the upper Duncan River averaged 703 ± 14 mm and 3575 ± 240 g (n = 66). There was no significant difference in mean FL between all bull trout captures from the flip bucket in 1995 and 1996 (two-tailed t-test, P = 0.558) and again I pooled the data. In 1995 and 1996, 533 bull trout (mean FL = 652.8 ± 5.0 mm) were captured and measured in the flip bucket at Duncan Dam. The average FL of radio-tagged bull trout that migrated to the 19 upper Duncan River was significantly larger than average FL of bull trout captured at Duncan Dam that were not radio-tagged (one-tailed t-test, P < 0.001; Figure 2-5). -r 80 - 70 -• 60 a -• 50 g" - 4 0 J 4so 3 D 20 0 & 410 =• - i — i — i — r 8 8 8 R S 8 8 8 8 S 8 8 8 8 § 8 8 8 R S 8 8 8 R S Fork length (mm) Figure 2-5. The fork length (FL) frequency histograms of all bull trout captured at Duncan Dam in 1995 and 1996 ( 'DD' - outline and right axis, n = 533) and of all radio-tagged bull trout migrating to destinations in the upper Duncan River ( 'RT' - gray bars and left axis, n=66). Destinations in the upper Duncan River In total, I tracked 66 radio-tagged bull trout to destinations in the upper Duncan River over 1995 and 1996 (Table 2-2). The majority of these bull trout were physically moved from below Duncan Dam and released in the reservoir. On average, each of the fish migrating to the upper Duncan River was located 28.5 ± 7.5 times during the study. Only six radio-tagged bull trout migrated through the discharge structure at Duncan Dam without interference. Figure 2-6 illustrates the upper Duncan River and location of major tributaries. The distribution of bull trout by migratory destination (seven categories; Table 2-2) was not significantly different between 1995 and 1996 (Fisher exact test, P = 0.91). Destinations with small numbers of radio-tagged bull trout migrating to them were grouped, giving a total of four destination categories for A N O V A analysis (Table 2-2). 20 Table 2-2. Migratory destinations and A N O V A groups of radio-tagged bull trout in the upper Duncan River. Destination 3, mainstem upper Duncan River, is broken into five sub-categories (3.a. - 3.e., italics) which are separated by tributary confluences (Fig. 2-6). Migratory Destination 1995 1996 Analytical Migrants % of Total Migrants % of Total Group 1.) Houston Cr. 8 30.8 13 32.5 l 2.) Westfall R. 6 23.1 10 25 2 3.) mainstem upper Duncan River 9 34.6 8 20 3 3.a.) above Houston Cr. 2 7.7 3 7.5 3 3.b.) Hatteras Cr. to Houston Cr. 4 15.4 2 5 3 3.c.) Laidlaw Cr. to Hatteras Cr. 1 3.8 I 2.5 3 3.d.) Westfall R. to Laidlaw Cr. 2 7.7 I 2.5 3 3.e.) Giegerich Cr. to Westfall R. 0 0 1 2.5 3 4.) Giegerich Cr. 2 7.7 5 12.5 4 5.) Stevens Cr. 1 3.8 2 5.0 4 6.) Hatteras Cr. 0 0 1 2.5 4 7.) Marsh Adams Cr. 0 0 1 2.5 4 T O T A L S 26 100 40 100 21 Figure 2-6. Map of the upper Duncan River system with kilometer designations upstream of Duncan Dam. 22 Migratory timing comparisons Radio-tagged bull trout migrating to destinations in the upper Duncan River system showed little variation in timing of migration. The two factor A N O V A comparisons of migration rate and timing did not vary by destination, between years, or year by destination interaction. The A N O V A results are summarized in Table 2-3. There was a suggested relationship to destination for FL, mass and out-migration time, but not between years or in terms of the interaction between year and destination. As out-migration time was not significantly correlated (Pearsons correlation coefficient, r = 0.236, P = 0.09) with out-migration distance I pooled the out-migration time data. I plotted fork length data for each A N O V A group (Fig. 2-7) to show the relationship between length and destination suggested by the two factor A N O V A result. In Figure 2-7,1 suggest the mainstem upper Duncan River destinations represent the physically smallest destination category: this is not technically correct, and reflects an assumption that bull trout migrating to these destinations did not spawn. Only the fork length of bull trout migrating to Westfall River and destinations in the mainstem upper Duncan River were significantly different using a Tukey test (P = 0.01) in a pooled by destination comparison. Two of the migration statistics used in this analysis are confounded. Migration rate statistics, in-migration and out-migration rate, are calculated as distance per day. The destination factor of the A N O V A model is also related to distance. As a result, the migration rate calculations are confounded due to the relationship between the distance inherent in both rate and the destination factor of the A N O V A . The A N O V A calculations involving rate are included because they do give insights into influence of the year of migration on migration rates. A table presenting all migration data used in the A N O V A analysis, by radio-tag number and destination category, is appended. 23 Table 2-3. Two factor ANOVA comparisons of migration data for radio-tagged bull trout by analysis group (see Table 2.2) and year of the study. Migration Statistic P-value P-value P - value Pooled Destination Year Interaction Mean ± 95 % C.I. (n) (DesfYear) In-migration Start Date 0.280 0.357 0.833 8 July ±4.4 days (64) In-migration Rate ** 0.010** 0.332 0.825** 1.80 ± 0.17 km/d (64) In-migration Time 0.247 0.754 0.981 51.4 ±4.2 days (64) Residency Start Date 0.630 0.358 0.482 29 Aug. ± 2.9 days (64) (In-migration End Date) Residency Time 0.489 0.650 0.533 . 31.7 ±3.4 days (53) Out-migration Start Date ' 0.084 0.154 0.733 28 Sept. ± 2.0 days (53) (Residency End Date) Out-migration Rate ** 0.350** 0.102 0.711** 9.93 ± 2.78 km/d (52) Out-migration Time 0.017* 0.655 0.561 8.99 ± 2.69 days (52) Out-migration End Date 0.849 0.574 0.619 8 Oct. ± 2.68 days (52) Fork Length (mm) 0.043 0.351 0.360 suggested differences (64) Mass (g) 0.019 0.471 0.257 suggested differences (64) * out-migration time was not correlated with distance (see text) and pooled ** comparison is confounded by similarity between factor and data variable (see text). 800 T 775 + 750 4-725 I 700 675 650 3 * 8 * + 13 10 Mainstem upper Duncan River Other tributaries i Houston Creek Westfall River Destination Figure 2-7. Data used in the calculation of bull trout fork length by destination and year two factor ANOVA. Open circles and solid squares are 1995 and 1996 data, respectively. Error bars are ± 1 SE. Destinations (ANOVA groups - Table 2.2) are approximately (see text) ordered by stream size from smallest to largest (left to right). Sample sizes are indicated next to data points. 24 The migration statistics of radio-tagged bull trout moving through the discharge structure at Duncan Dam were not different from bull trout transported and released upstream (two sample t-tests; minimum P = 0.092 for all migration statistics except in-migration" and out-migration rate). I confirmed spawning activity within the residency area of radio-tagged bull trout in Houston Creek and the Westfall River. Locating bull trout redds was easy in these two tributaries: redds were located at all locations where radio-tagged bull trout were located during the last week of the spawning site residency period. I was unable to locate evidence of spawning in sections of the mainstem upper Duncan River upstream of Hatteras Creek (Fig. 2-6) where radio-tagged bull trout were located, despite thorough searches. I did not search for evidence of spawning in the mainstem upper Duncan River below Hatteras Creek; however, the mainstem river downstream of Hatteras Creek (Fig. 2-6) does not contain habitat usually associated with bull trout spawning (Goetz 1984; McPhail and Baxter 1996). I assume that the bull trout migrating to destinations in the upper Duncan River were not migrating to spawn based on similar bull trout non-spawning migrations observed by Swanberg (1997) and O'Brien and Chamberlain (2000). In total, 49 radio-tagged bull trout migrated to presumed spawning tributaries of the upper Duncan River during the study. Catchment area was a useful predictor of stream flow in the upper Duncan River system. A linear regression of stream flow estimates on catchment area was significant (P < 0.001; Fig. 2-8). Although the sample size is small, the catchment size of different tributaries explained > 99% of the variation in estimated stream flows (y - 0.029x - 0.636, r 2 = 0.997). Because I was unable to estimate flows from all streams targeted by radio-tagged bull trout, I used catchment area as a surrogate (Table 2.4). I plotted the fork length of all radio-tagged bull trout (1995 and 1996 combined) against catchment area of the destination tributary in Figure 2-9. A linear regression of fork length on catchment area was significant (P = 0.030), although little of the variation in length was explained by increasing catchment area (y = 0.301x + 669.2 r 2= 0.095; Fig.2-9). The relationship between fork length and catchment area is not robust: when the largest individual (875 mm FL from Westfall River - 230 km 2 catchment; Fig. 2-9) is removed, the regression is not significant (y = 0.235 + 675.5, r2 = 0.065, P = 0.077). 25 0 T 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 I I I 0 50 100 150 200 250 Catchment area (km2) Figure 2-8. The regression of catchment size to instream flow estimate of five tributaries of the upper Duncan River. The regression (y = 0.029x - 0.636; solid line) is significant (P < 0.001, r 2= 0.997). Table 2-4. The catchment area of tributaries targeted by radio-tagged bull trout in 1995 and 1996. Stream Catchment area (km2) Westfall River 230 Giegerich Creek 113 Houston Creek 107 Hatteras Creek 65 Stevens Creek 51 Marsh Adams Creek 46 26 E E £Z 0> o L L 9 0 0 8 5 0 8 0 0 750 700 6 5 0 600 5 5 0 H 5 0 0 - i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i i 0 50 100 150 2 0 0 2 5 0 Catchment area (km ) Figure 2-9. Fork length of radio-tagged bull trout (n = 49) by catchment area of destination tributaries in the upper Duncan River. The regression (y = 0.301x + 669.2; solid line) is significant (P = 0.03, r 2= 0.095). A linear regression excluding the largest fish from this figure (open circle) is not significant (P = 0.08, see text). For analysis of migration statistics, I considered the migration of bull trout to the upper Duncan River over when they returned to Duncan Reservoir. From subsequent locations, via both angler recaptures and telemetry fixes, most were tracked back to Kootenay Lake after reaching Duncan Reservoir. Over the winter of 1995/96, 82.6 % (19 of 23) radio-tagged bull trout left the reservoir through the discharge structure. I excluded three bull trout confirmed killed by anglers after the 1995 migration from this calculation. Over the winter of 1996/97, 74.3 % (26 of 35) of radio-tagged bull trout were confirmed to have left the reservoir; excluding three expired transmitters and two presumed mortalities. I rarely located radio-tagged bull trout in lakes, because of radio signal attenuation with depth in water (Winter 1983). I did however track bull trout throughout Kootenay Lake, and also located one radio-tagged bull trout at the confluence of the Moyie and Kootenay rivers in Idaho (Fig. 2-1) the summer after it had been tracked in the upper Duncan River. Radio telemetry confirmed all of the suggested movement patterns of bull trout in the 27 Kootenay River collected previously using conventional individually numbered Floy-tags (Fleck 1977). I also determined the date bull trout moved downstream through Duncan Dam. In 1995,1 only estimated the date based on first location downstream, but over the winter of 1996/97,1 placed a remote data-logging receiver at the base of the dam (Table 2-5) to record actual dates of downstream passage. Over the winter of 1996/97,1 tracked 13 bull trout moving downstream through Duncan Dam. On average, bull trout left Duncan Reservoir on 23 January 1997 (± 30 d) after summer and fall migrations to the upper Duncan River in 1996. Table 2-5. Overwinter out-migration dates of individual radio-tagged bull trout through the discharge structure at Duncan Dam by year of migration to the upper Duncan River. 1995 migration 1996 migration 16-Oct-95* 8-Oct-96* 16-Oct-95* 20-Oct-96* 16-Oct-95* 26-Dec-96 31-Oct-95* 25-Jan-97 25-Nov-95* l-Feb-97 13-Aug-96* 7-Feb-97 10-Feb-97 13-Feb-97 16-Feb-97 19-Feb-97 24-Feb-97 10-Mar-97 ll-Mar-97 •approximate only - first reception downstream of the dam using manual receiver Multi-year migrations Six radio-tagged bull trout were tracked migrating to the upper Duncan River in two consecutive spawning years. Two of the six entered different tributary streams in the second migration year (Table 2-6). If I consider only those bull trout migrating to presumed spawning locations in 1995 (17 fish), two (11.8%) returned in 1996, and both entered a different tributary. In both cases, the bull trout switched to a tributary with a smaller catchment area than their first tributary destination. Because many radio tags expired before the spawning migration in 1997,1 cannot estimate the number migrating in both 1996 and 1997. 28 Table 2-6. Destinations of radio-tagged bull trout migrating to the upper Duncan River in two consecutive years. Radio tag F.L. (mm) Sex Destination Distance (km) number at tagging year 1 year 2 between destinations 22 675 F Westfall R. Stevens Cr. 11.7 30 655 F Mainstem Mainstem 11.5 35 755 M Stevens Cr. Marsh Adams Cr. 18.7 39 655 M Mainstem Mainstem 12.5 40 690 F Mainstem Mainstem 3.5 64 685 M Houston Cr. Houston Cr. 4.0 Houston Creek fish trap I captured 47 bull trout emigrating from Houston Creek in a fish trap. Of these out-migrants, three were radio-tagged (6.4 %) and one was marked with a Floy-tag at Duncan Dam. The mean fork length of trapped out-migrants was 564 ± 27 mm (n = 45, Figure 2-10). The mean fork length of fish captured at the trap was significantly smaller than that of fish captured at Duncan Dam that were not radio tagged (pooled 1995 and 1996 dam captures; 1 tailed t-test, P < 0.001). Twenty one were female, 23 were male, a single bull trout (FL 411 mm) was of indeterminate sex, and two escaped before examination. A l l but the smallest bull trout captured at Houston Creek showed clear evidence of spawning. Females were gaunt and all males had recent scarring on the caudal fin and peduncle, suggesting spawning (McPhail and Baxter 1996). A l l males and females expressed milt or eggs when anaesthetized and handled. The mark recapture estimate of the total number of bull trout in Houston Creek was 167. Assuming the proportion of radio-tagged to untagged fish was the same at all destinations in the upper Duncan River, approximately 500 bull trout migrated into the system in 1996 (Houston Creek was 32.5% of the total number of migrants in 1996 - Table 2-2). Based on the proportion of fish migrating in consecutive years, I estimate that roughly 40 to 50 bull trout switch spawning stream from year to year in the upper Duncan River. 29 8 j 7 •• 6 • Fork Length (mm) Figure 2-10. The fork length (FL) frequency histogram of bull trout captured at the Houston Creek trap in 1996 (gray bars, n = 47). 2.4 Discussion The mean fork length of bull trout receiving radio tags (703 ± 14 mm) was significantly larger than the mean size of fish captured at Duncan Dam. This discrepancy reflects my selection of larger bull trout for tagging due to size constraints associated with the implantation of tags. In this study I have assumed that radio tagging does not change the migratory behaviour of bull trout. Except for a short period immediately following tagging, many studies (Stasko and Pincock 1977; Winter 1983; Marty and Summerfelt 1986; Swanberg 1996; Thorstad et al. 2000) suggest that if the transmitter is small relative to the size of the fish tagged, the effect of transmitter implantation on fish behaviour and movement can safely be ignored. By 'small', I follow the definition of Marty and Summerfelt (1986): the mass (in water) of surgically implanted tags should not exceed 2% of the mass (in air) of the fish destined to carry it and, ideally, this ratio should not exceed 1%. I kept the tag to fish weight ratio below 1% in all cases. In this study, I did not explicitly test the assumption that radio tagging did not effect behaviour, but several of my observations corroborate the findings of others. I found that bull trout carrying transmitters often migrated to the extent of available habitat. I also located untagged bull trout in proximity to radio-tagged fish at all locations I examined, including a tagged bull trout male 30 spawning with an untagged female in Westfall River. This suggests that their behaviour was similar. Untagged bull trout were also captured during out-migration from Houston Creek at the same time as radio-tagged individuals. Bull trout migrations in the upper Duncan River Radio telemetry proved useful in obtaining data on bull trout spawning migrations in the upper Duncan River. The pattern of bull trout migration typically consisted of in-migration to a destination in the upper Duncan River, residence within a short distance of that destination, and a rapid out-migration relative to the time of in-migration and residence. Comparisons of the 1995 and 1996 migration statistics show the two years to be very similar (Table 2-3). Time of in-migration, residence, and out-migration do not differ between destinations in the upper Duncan River or the year of migration (Table 2-3). The average bull trout migrating to a destination in the upper Duncan River began moving upstream of the dam on 8 July. The average bull trout arrived at its destination on 29 August (51 day in-migration) and remained there for 32 days until 28 September. In-migration rate averaged just under 2 km*day''. Moving more than 9 km* day"1, the average out-migrant returned to Duncan Reservoir on 8 October. Bull trout leave Duncan Reservoir, and return to Kootenay Lake, throughout the winter (Table 2-5). The timing of spawning migrations in the Duncan River fall within the range of bull trout migratory timing as reviewed by Goetz (1984) and McPhail and Baxter (1996). The impact of releasing bull trout captured and tagged downstream of Duncan Dam in the reservoir upstream on migratory timing and behaviour is a concern. The migratory timing of bull trout moved over Duncan Dam did not differ from those that moved through it (P > 0.092), suggesting that the release of bull trout upstream of the dam did not alter migration behaviour. I only tagged fish after they actively moved into the discharge structure at Duncan Dam. Movement into the discharge structure suggests that bull trout were attempting to move through. O'Brien and Chamberlain (2000) compared the timing of bull trout spawning migrations in the Adams River with my results. The dates radio-tagged bull trout arrived and left destination areas in Adams River (no dam) were the same as Duncan River bull trout. This suggests that the impact of the dam on the 31 timing of spawning migrations is minimal for those fish successfully passing upstream through the dam. In total, six tributaries and portions of the mainstem upper Duncan River were the destinations of migrating bull trout. The two largest tributaries, Westfall River and Houston Creek, account for over half of the tagged migrants (Table 2-2). The distribution of bull trout among destinations (Table 2-2) did not vary significantly between 1995 and 1996 (Fisher exact test, P = 0.91). I evaluated whether this distribution was related to stream size, using catchment area as a surrogate (Fig. 2-8), in terms of the size (Fig. 2-9) of bull trout migrants as suggested by McPhail and Murray (1979). For this test, I used only bull trout migrating to tributary destinations and excluded those migrating to destinations in the mainstem river (see below). There was a slight positive relationship between the size of radio-tagged bull trout and the size of the tributary they migrated too (linear regression, P = 0.03), but this relationship was not robust. Perhaps this relationship would be better resolved if smaller bull trout had been radio-tagged. McPhail and Murray (1979) suggested that bull trout may be physically restricted from smaller spawning streams as they grow. The mean sizes of spawning bull trout McPhail and Murray (1979) captured were 447 and 438 mm standard length and 1260 and 1097 g for males and females, respectively. These were probably first time spawners. Duncan Dam bull trout captures were on average more than 200 mm larger than the bull trout McPhail and Murray (1979) captured in their study. The largest bull trout I captured at Duncan Dam was 920 mm and 9500 g. This amount of variation in size supports McPhail and Murray's (1979) suggestion: as bull trout grow, spawning habitat preferences probably change. If bull trout switch spawning stream based on size, as the regression of bull trout size on tributary catchment area suggests, it explains a reduction in fidelity to natal stream. It suggests that bull trout may not home accurately at the scale of individual spawning streams. It is also possible that the apparent relationship between bull trout size and spawning tributary catchment area could reflect the effect of selection in streams of differing size for appropriately sized bull trout. I assumed that bull trout migrating to mainstem destinations in the upper Duncan River did not spawn. I based this assumption on both negative spawning survey results and a general lack of suitable spawning habitat at mainstem destinations compared to 32 literature descriptions (Baxter and McPhail 1996; McPhail and Baxter 1996). Two other radio telemetry studies have suggested that some bull trout that participate in the spawning migration do not spawn. Swanberg (1996; 1997) found that as many as 70 % of bull trout radio-tagged in the Blackfoot River in Idaho did not migrate all the way to spawning areas, and did not spawn. Swanberg (1996; 1997) found that these fish were generally smaller, began migrations later, and left tributaries earlier than spawning fish. Similar results were obtained in the Adams River, B C (O'Brien and Chamberlain 2000). Over two years, approximately 40 % of the radio-tagged bull trout migrating into the Adams River did not leave the mainstem river. In Adams River, bull trout remaining in the mainstem were smaller than those moving into spawning tributaries, and they emigrated from the river prior to the emigration of individuals that spawned in tributaries. Bull trout migrating to mainstem upper Duncan River destinations were generally smaller than other migrants (Fig. 2-7), although all were large enough to be mature based on literature accounts (Goetz 1989; McPhail and Baxter 1996). Unfortunately, I was unable to collect bull trout at mainstem destinations and examine them for their level of maturity. Swanberg (1996; 1997) suggested that non-spawning migrants were moving to avoid seasonal high temperatures in the mainstem Blackfoot River. Temperatures in the mainstem Blackfoot River can exceed 20 °C in July. Swanberg (1997) specifically dismissed the hypothesis of foraging for these non-spawning migrants, as prey species were rare in tributaries of the Blackfoot R. Seasonally high temperatures may occasionally exceed 20 °C in the surface waters of Kootenay lake or the Duncan Reservoir, but such temperatures are easily avoidable with depth. In the upper Duncan River, I did not sample mainstem locations enough to estimate abundances of bull trout prey fish species; although, in all sampling of spawning tributaries (Chapter 3), I encountered only bull trout. Non-spawning migrations in the upper Duncan River do not seem related to temperature or foraging. Spawning stream switching Two of six bull trout making repeat migrations switched destination tributary (Table 2-5) in the second migration year. Of the six bull trout tracked in consecutive years, the sex ratio was equal and one male and one female migrated to a different presumed spawning tributary (Table 2-5). Contrary to the hypothesis of McPhail and 33 Murray (1979), bull trout switching tributaries changed to tributaries with smaller catchment areas. The observed destination switching suggests that there may be gene flow among spawning streams. I was not able to confirm if stream-switching fish actually spawned in each stream. I was able to directly confirm spawning activity only in Houston Creek and Westfall River because of limited access in other streams. With results like those of Swanberg (1997) and suggested by O'Brien and Chamberlain (2000), where some bull trout migrated without spawning, it is important to determine if the observation of spawning stream switching results in gene flow (see Chapter 3). Houston Creek fish trap The bull trout captured during out-migration from Houston Creek were on average much smaller than those captured at Duncan Dam (Fig. 2-10). This suggests that not all bull trout spawning in Houston Creek had migrated from Kootenay Lake. The majority of smaller, and probably younger, bull trout trapped in Houston Creek were bull trout likely migrating from Duncan Reservoir. I used bull trout captured during out-migration from Houston Creek to estimate the number of bull trout present in Houston Creek for spawning in 1996. Using this estimate (167 fish) I roughly extrapolated the number present in the upper Duncan River in 1996 (-500 bull trout). I assumed the number present in 1995 was similar to 1996, and then estimated numbers of untagged bull trout that change destination tributary each year from the number of tagged bull trout that did so over 1995 and 1996 (approximately 40 to 50 fish). Clearly this extrapolated estimate is imprecise; nevertheless, it is valuable in that it provides me a comparison with estimates of migration from genetic analysis presented in Chapter 3. Conservation implications Without confirmation of spawning by bull trout at most locations in the upper Duncan River, the conservation implications of these telemetry results are difficult to interpret. The potentially most important result from a conservation, or management, perspective is the apparent switching between spawning sites by bull trout in consecutive 34 years. This result suggests that individual spawning streams do not represent genetically divergent populations in the upper Duncan River. The apparent non-spawning movements of more than 25 % of the radio-tagged bull trout into the upper Duncan River is also important from a conservation perspective. The knowledge that many more bull trout than would be enumerated at spawning sites have migrated nearly the same distance as spawning fish changes suggests habitat protection may be necessary beyond only the area of specific spawning sites. It is important to know that these non-spawning bull trout move at similar times to fish that spawn. Protection measures, such as seasonal angling closures along migration routes, will benefit both spawning and non-spawning migrants. 35 2.5 Appendix Table 2-7. The biological data collected from radio-tagged bull trout migrating to destinations in the upper Duncan River. 'Ch.' is tag channel, 'Floy' is the number of the external tag, 'FL' is fork len lag Code Date Tagged Hoy T L - Mass •SeT 4 5 104 13-Jun-95 577W 640 2.80 M 4 20 102 3-Jul-96 577W 715 3.30 M 10 5 111 14-Jun-95 637W 730 4.10 M 10 20 101 14-Jun-96 636Y 770 4.80 M 20 22 123 27-Jun-95 748W 630 2.75 M 20 20 111 14-Aug-96 748W 720 4.20 M 21 22 124 27-Jun-95 749W 710 3.30 M 22 22 125 28-Jun-95 763W 675 2.80 F 23 22 126 28-Jun-95 801W 810 5.45 F 24 22 127 28-Jun-95 802W 655 2.70 M 25 22 128 29-Jun-95 803W 745 4.35 F 26 22 129 29-Jun-95 761W 705 3.35 M 27 22 131 29-Jun-95 804W 670 3.20 F 28 22 133 29-Jun-95 812W 670 3.05 M 29 5 100 5-Jul-95 810W 780 4.80 M 30 22 134 5-Jul-95 829W 655 2.90 F 31 22 135 5-Jul-95 833W 640 2.90 M 32 22 136 5-Jul-95 520W 645 2.80 M 33 22 137 5-Jul-95 775W 685 3.40 M 34 31 138 13-Jul-95 518Y 850 5.90 M 35 31 141 18-Jul-95 520Y 725 3.50 M 35 5 134 14-Aug-96 520Y 755 4.00 M 36 31 142 18-Jul-95 524Y 875 7.60 M 37 31 143 18-Jul-95 525Y 650 2.80 F 38 31 140 19-Jul-95 533W 665 3.10 M 39 31 144 19-Jul-95 890W 655 2.90 M 39 20 112 14-Aug-96 822Y 695 3.40 M 40 31 145 19-Jul-95 554Y 690 3.20 F 41 31 147 19-Jul-95 557Y 665 2.80 F 42 31 148 19-Jul-95 691W 725 4.00 M 43 31 146 20-Jul-95 575Y 715 3.80 M 44 31 149 20-Jul-95 689W 700 3.40 F 45 31 150 20-Jul-95 675W 685 3.10 F 46 31 151 20-Jul-95 619W 680 2.80 F 49 15 2 3-Sep-95 712W 810 5.00 F 51 31 153 6-Jun-96 848Y 780 4.40 M 53 5 121 14-Jun-96 642 Y 650 2.80 M 54 5 122 14-Jun-96 643Y 685 3.10 M 55 5 123 14-Jun-96 638Y 695 3.80 M 56 5 124 14-Jun-96 641Y 720 4.00 M 57 31 154 14-Jun-96 637Y 665 3.00 M 58 5 125 18-Jun-96 846Y 685 3.60 M 59 5 126 18-Jun-96 664W 695 3.50 F 60 5 127 18-Jun-96 557W 685 3.00 M 61 5 128 3-Jul-96 815Y 670 2.70 M 62 5 129 3-Jul-96 879W 610 1.80 F 64 20 103 3-Jul-96 812Y 685 3.50 M 65 20 104 3-Jul-96 814Y 765 5.10 M 66 20 105 3-Jul-96 915Y 685 3.20 M 67 15 3 10-Jul-96 817Y 690 3.00 M 68 15 4 10-Jul-96 907Y 595 2.10 M 69 15 5 10-Jul-96 604W 660 2.40 M 71 20 107 10-Jul-96 821Y 755 3.50 M 72 20 108 10-Jul-96 819Y 800 4.60 M 73 20 109 10-Jul-96 816Y 770 4.40 M 74 20 110 10-Jul-96 818Y 755 3.40 M 75 15 6 15-Jul-96 8019Y 745 4.40 M 76 15 7 15-Jul-96 922Y 625 2.80 M 77 15 9 15-Jul-96 8017Y 695 3.50 F 78 15 10 15-Jul-96 828W 700 3.30 M 79 15 11 15-Jul-96 8036Y 675 2.90 M 80 15 12 • 15-Jul-96 8007Y 600 2.50 M 81 20 114 15-Jul-96 923Y 760 4.90 M 82 20 115 15-Jul-96 611Y 780 5.40 M 83 5 131 7-Aug-96 672W 680 3.50 M 87 5 135 27-Aug-96 8099Y 635 3.60 M 36 Table 2-8. Summary of migration statistics and destinations of all radio-tagged bull trout migrating to destinations in the upper Duncan River (used in 2 way ANOVA). 'In' is in-migration, 'Res' is residence, 'Out' is out-migration, 'Dest' is destination (see Table 2-Start in In time In rate ln/res Res time Res/out out rate out time End out Out time Dest 26 17-Jun 6! 1.64 17-Aug 51 7-Oct 8.08 12 19-Oct 12 1 27 29-Jun 67 1.49 4-Sep 8 12-Sep 6.94 9 21-Sep 9 1 28 29-Jun 52 1.92 20-Aug 34 23-Sep 5.00 20 13-Oct 20 1 29 5-Jul 51 1.96 25-Aug 37 1-Oct 5.11 19 20-Oct 19 1 31 4-Jul 55 1.82 28-Aug 34 1-Oct 6.47 15 16-Oct 15 1 34 13-Jul 55 1.82 6-Sep 28 4-Oct 24.00 2 6-Oct 2 1 38 18-Jul 42 2.38 29-Aug 36 4-Oct 26.25 2 6-Oct 2 1 43 20-Jul 36 2.78 25-Aug 37 1-Oct 11.50 5 6-Oct 5 1 21 14-Jun 61 1.64 14-Aug 44 27-Sep 5.35 10 7-Oct 10 1 55 14-Jun 61 1.64 14-Aug 47 30-Sep 7.64 7 7-Oct 7 1 56 15-Jul 30 3.33 14-Aug 57 10-Oct 53.50 1 11-Oct 1 1 59 15-Jul 38 2.63 22-Aug 36 27-Sep 4.86 11 8-Oct 11 1 62 15-Jul 38 2.50 22-Aug 43 4-Oct 12.13 4 8-Oct 4 1 64 10-Jul 45 2.22 24-Aug 32 25-Sep 3.82 14 9-Oct 14 1 67 10-Jul 47 2.13 26-Aug 32 27-Sep 4.46 12 9-Oct 12 1 68 10-Jul 47 2.13 26-Aug 1 71 10-Jul 55 1.82 3-Sep 23 26-Sep 4.46 12 8-Oct 12 1 73 3-Jul 62 1.61 3-Sep 28 1-Oct 13.38 4 5-Oct 4 1 75 15-Jul 50 2.00 3-Sep 1 76 3-Jul 71 1.41 12-Sep 13 25-Sep 5.94 9 4-Oct 9 1 80 18-Jun 87 1.15 13-Sep 1 81 7-Jun 102 0.98 17-Sep 1 22 28-Jun 67 1.12 3-Sep 24 27-Sep 19.00 2 29-Sep 2 2 23 28-Jun 71 1.06 7-Sep 23 30-Sep 12.50 4 4-Oct. 4 2 25 29-Jun 47 1.60 15-Aug 53 7-Oct 11.17 6 13-Oct 6 2 33 5-Jul 30 2.50 4-Aug 54 27-Sep 6.82 11 8-Oct 11 2 36 18-Jul 57 1.21 13-Sep 30 13-Oct 8.83 3 16-Oct 3 2 49 3-Sep 27 0.46 30-Sep 8 8-Oct 14.40 5 13-Oct 5 2 10 14-Jun 70 1.07 23-Aug 36 28-Sep 3.17 9 7-Oct 9 2 20 14-Aug 32 2.34 15-Sep 31 16-Oct 28.50 1 17-Oct 1 2 57 14-Jun 74 1.01 27-Aug 2 61 3-Jul 44 1.70 16-Aug 42 27-Sep 3.17 9 6-Oct 9 2 65 3-Jul 63 1.19 4-Sep 25 29-Sep 3.17 9 8-Oct 9 2 72 10-Jul 46 1.63 25-Aug 30 24-Sep 2.38 12 6-Oct 12 2 74 10-Jul 62 1.11 10-Sep 29 9-Oct 11.25 2 11-Oct 2 2 77 15-Jul 31 2.42 15-Aug 2 78 15-Jul 59 1.27 12-Sep 2 79 15-Jul 36 2.08 20-Aug 36 25-Sep 7.13 4 29-Sep 4 2 24 27-Jun 68 1.25 3-Sep 32 5-Oct 0.74 51 25-Nov 51 3 30 2-Jul 56 1.52 27-Aug 31 27-Sep 4.21 19 16-Oct 19 3 37 18-Jul 44 1.70 31-Aug 3 39 19-Jul 47 2.02 4-Sep 26 30-Sep 18.25 4 4-Oct 4 3 40 19-Jul 52 1.63 9-Sep 21 30-Sep 1.53 19 19-Oct 19 3 41 19-Jul 31 2.42 19-Aug 46 4-Oct 32.50 1 5-Oct 1 3 42 19-Jul 32 2.66 20-Aug 45 4-Oct 3.64 22 26-Oct 22 3 44 20-Jul 33 2.88 22-Aug 32 23-Sep 4.77 11 4-Oct 11 3 46 20-Jul 31 2.58 20-Aug 20 9-Sep 6.25 6 15-Sep 6 3 30 7-Aug 37 2.62 13-Sep 3 39 14-Aug 30 2.83 13-Sep 13 26-Sep 3.50 11 7-Oct 11 3 40 14-Aug 27 3.15 10-Sep 16 26-Sep 1.93 20 16-Oct 20 3 58 18-Jun 58 1.67 . . 15-Aug 42 26-Sep 4.21 12 8-Oct 12 3 60 18-Jun 60 1.00 17-Aug 39 25-Sep 1.93 7 2-Oct 7 3 66 3-Jul 52 1.63 24-Aug 3 69 10-Jul 44 2.20 23-Aug 15 7-Sep 2.20 23 30-Sep 23 3 32 5-Jul 60 1.00 3-Sep 31 4-Oct 17.50 1 5-Oct 1 4 35 18-Jul 62 1.13 18-Sep 11 29-Sep 5.50 5 4-Oct 5 4 45 20-Jul 47 1.28 5-Sep 25 30-Sep 9.50 4 4-Oct 4 4 4 3-Jul 34 1.76 6-Aug 53 28-Sep 1.69 8 6-Oct 8 4 22 12-Jul 35 1.86 16-Aug 40 25-Sep 2.06 9 4-Oct 9 4 35 14-Aug 20 3.90 3-Sep 23 26-Sep 2.63 12 8-Oct 12 4 51 6-Jun 90 0.67 4-Sep 20 24-Sep 0.90 15 9-Oct 15 4 53 14-Jun 86 0.70 8-Sep 13 21-Sep 16-Oct 4 54 14-Jun 70 0.86 23-Aug 47 9-Oct 1.93 7 7 4 37 Chapter 3 - An evaluation of genetic differentiation between two bull trout (Salvelinus confluentus) spawning streams in the upper Duncan River using microsatellites. 3.1 Introduction Most species are structured into semi-isolated reproductive populations. Such populations tend to diverge from other populations through selection or genetic drift (Slatkin 1987). Preserving this genetic diversity is considered paramount for conservation (Moritz 1994; Avise 1995). Understanding the spatial scale over which such genetic population structure occurs is, therefore, an important conservation priority. The geographic scale over which species are divided into populations is a function gene flow. In general, a small amount of gene flow between otherwise separate populations will counter genetic divergence due to selection or drift (Wright 1931; Hartl and Clark 1989; Mills and Allendorf 1996). The amount of gene flow is related to barriers to dispersal, a species' inherent dispersal ability and reproductive site fidelity (Waples 1998; Bohonak 1999). The distance animals can move obviously affects population structure. For populations in close proximity, relative to dispersal ability, there should be more gene flow than between distant populations. This is 'isolation by distance' (Wright 1943; Slatkin 1993). Barriers to migration and gene flow enhance isolation among populations, resulting in a finer spatial scale over which species are structured into genetically distinct populations. In marine organisms, where barriers to gene flow are slight (e.g. Palumbi et al. 1997), the amount of genetic differentiation between populations is generally small (Ward et al. 1994; Waples 1998). In contrast, where physical barriers to movement are more common, such as among populations of freshwater fish, genetic differentiation among populations is stronger (Ward et al. 1994). The degree to which animals show fidelity to natal sites also has important implications for the level of gene flow and resulting genetic population structure. Reproductive site fidelity, or homing, is an adaptive trait, because it ensures that 38 reproduction occurs where conditions suitable for survival, growth and reproduction occurred in the past (Lindsey et al. 1959; Northcote 1997). Thus, the spatial scale of genetic divergence among populations should be finer in animals that home to their exact birth site, than, for example, animals that home to the general natal area. The processes that affect population structure, and divergence among populations specifically, have been well studied in salmonid fishes. In salmonid fishes, particularly Pacific salmon, there is evidence that fidelity to natal areas, or homing, results in fine scale population divergence (Scheer 1939; Banks 1969; Mclsaac and Quinn 1988; Quinn and Dittman 1990; Taylor 1991). Distinct populations are detectable, through morphological, behavioural or genetic differences, in salmonids from streams in close geographic proximity (Quinn et al. 1987; Taylor 1991; Angers et al. 1995), within the same stream (Taft and Shapovalov 1938; Tallman and Healey 1994; Costello and Taylor unpubl. MS), or even within the same stream reach (Hansen et al. 1997; Carlsson et al. 1999). Homing may be especially important for semelparous Pacific salmon, in that it increases the chance that their only reproduction is successful. Clearly homing is not perfect in Pacific salmon, since much of their current range was recolonised following deglaciation (Quinn and Dittman 1990). Rates of straying among natal streams can be high quite high (>50%) in some species of Pacific salmon (Shaklee and Varnavskaya 1994; McElhany et al. 2000), especially in fish of hatchery origin (Quinn 1993). As suggested previously, less accurate homing will tend to increase the geographic area of populations. The bull trout is an iteroparous salmonid native to north western North America (Haas and McPhail 1991). Because of population declines in the southern portion of its range (Barry 1999), bull trout are the subject of conservation concern. These declines are usually attributed to habitat loss, habitat fragmentation and exotic species introductions (Goetz 1984; Rieman and Mclntyre 1993; Watson and Hillman 1997). To help define the spatial scale at which conservation efforts should be focused (Vogler and DeSalle 1994), it is important to define the geography of distinct bull trout populations. Unfortunately, the strength of natal stream fidelity is not known in bull trout (McPhail and Murray 1979; Goetz 1984; McPhail and Baxter 1996). 39 The spatial scale of genetic population structure in bull trout has been studied at three different levels: a species wide level (Spruell and Allendorf 1997; Taylor et al. 1999), large regional levels (Leary et al. 1993; Spruell and Allendorf 1997; Williams et al. 1997; Taylor et al. 1999; Latham and Taylor in press; Costello and Taylor unpubl. MS), and smaller local or small watershed levels (Spruell et al. 1999; Costello and Tayor unpubl. MS). Over the range of the species, both mitochondrial D N A (mtDNA) and microsatellite studies (Spruell and Allendorf 1997; Taylor et al. 1999) suggest that bull trout are broadly divided into a coastal and an interior group. Subdivision on this broad geographic scale is attributed to historical isolation in two glacial refugia south of the Cordilleran ice sheet (Taylor et al. 1999). At a regional level, below that of large drainages like the Columbia or Fraser rivers, bull trout are structured into clusters of similar groups within regions that are differentiated from other such groups (Leary et al. 1993; Spruell and Allendorf 1997; Williams et al. 1997; Taylor et al. 1999; Latham and Taylor in press; Costello and Taylor unpubl. MS). At these two relatively broad scales, bull trout generally display low levels of genetic variability. This low variability is attributed to historical demographic processes such as founder effects resulting from postglacial dispersal (Leary et al. 1993) and more recent bottlenecks (Spruell and Allendorf 1997; Taylor et al. 1999; Costello and Taylor unpubl. MS). Two earlier studies characterized genetic population structure at the level of individual watersheds. Spruell et al. (1999) examined Lightning Creek, a tributary to the Clark Fork River in Idaho, that drains an area of 220 km 2 . Costello and Taylor (unpubl. MS) examined population structure within and between Peace and Kootenay river tributaries in British Columbia and Red Deer and Oldman rivers in Alberta. The spatial scales that Costello and Taylor (unpubl. MS) examined within river systems ranged from roughly 300 to 3000 km . In all but the case of one Kootenay River tributary (Wigwam River) both studies found significant subpopulation variation within watersheds (Spruell et al. 1999; Costello and Taylor unpubl. MS). In Chapter 2,1 presented telemetry evidence that over a two year period some bull trout in the upper Duncan River switched spawning streams. This suggests that bull trout may not home to specific streams in the upper Duncan River system. The upper 40 Duncan River drains an area of nearly 1300 km . As discussed in Chapter 2, the observation that some bull trout change streams does not necessarily mean successful reproduction at both locations. I did not confirm spawning at all locations. Also, because the natal stream(s) of the radio-tagged bull trout are unknown, I cannot use the data in Chapter 2 to make direct inferences about stream fidelity, other than noting that fish switched migration destinations over two years. If, over their lives, bull trout switch spawning streams this should tend to increase the geographic area over which bull trout remain a single, genetically homogeneous population. I used microsatellites to compare two spawning tributaries within the upper Duncan River. The purpose was to test the prediction (based on the telemetry data) that bull trout in the upper Duncan system are a single genetic population. Microsatellites are selectively neutral nuclear loci that show high levels of allelic variation and, as such, they are useful for differentiating closely related species, populations, or even individuals (Jarne and Lagoda 1996). I collected microsatellite allele frequency data from Houston Creek and the Westfall River. These streams were chosen because of their relatively high use by radio-tagged bull trout (Chapter 2) and the observation that bull trout spawned in both streams. The mouths of Houston Creek and Westfall River are separated by approximately 30 km, and their primary spawning areas (identified from telemetry) are separated by approximately 60 km. I used the allele frequency data to estimate the level of genetic differentiation between Houston Creek and Westfall River to address the following questions: • Do bull trout in different spawning streams in the upper Duncan River belong to genetically distinct populations? • How does spawning stream switching (as estimated from telemetry) compare to estimates of gene flow derived from microsatellite data? Based on the suggestion of high gene flow from my telemetry results (Chapter 2), and despite the results of genetic studies at similar spatial scales, I predict that bull trout in Houston Creek and Westfall River represent a single, genetically homogeneous population. 41 3.2 Methods T i s s u e co l lect ion I captured juvenile bull trout by backpack electrofishing at sites in Houston Creek and the Westfall River. The relatively wide spacing of sample locations in Westfall R. (Fig. 3-1) compared to those in Houston Cr. (Fig. 3-2) is a function of access in the two systems. A road extends along the length of Westfall R. and allows relatively easy access. We accessed all but one site in Houston Cr. by foot from the confluence of the creek and the upper Duncan River mainstem. One site in Houston Cr. was accessed by helicopter. At each site, a GPS coordinate was recorded, and a single electrofishing pass made through suitable juvenile habitat. I focused on side channels, woody debris and channel margins (Baxter and McPhail 1996). Electrofishing continued at each site until the target numbers of individuals were captured (see below). The type of tissue collected varied as a function of fish size. Tissue samples of young-of-the-year bull trout consisted of the entire animal. For juveniles > 55 mm, the adipose fin was removed. In all cases, the tissue was placed in an individually numbered snap top centrifuge tube containing 95% ethyl alcohol and refrigerated until processed in the laboratory. No mature bull trout were encountered during this sampling. I separated tissue samples from each stream into three age classes based on fork length (FL) frequency, and in this way grouped samples for further analysis. In Westfall R., bull trout < 54 mm were considered 0+, 1+ from 54 to 93 mm, and 2+ from 105 to 155 mm (Fig. 3-3). In Houston Cr., I considered fish < 60 mm 0+, from 60 to 100 mm 1+, and those > 110 mm 2+ (Fig. 3-4). By setting age breaks at these lengths, I avoided including samples that were difficult to clearly place in a specific age category. I attempted to reach a sample size of 20 individuals from each age class in each stream, but 2+ bull trout were rare in my samples from Houston Cr. Most of the tissue sampling occurred in 1997; however, I used four 1+ samples collected from Houston Cr. in 1996 to bolster the sample size in the 2+ age class from 1997. Allendorf and Phelps (1981) cautioned specifically against attempting to estimate population structure from juveniles, because allele frequencies estimated from juveniles may not reflect the allele frequencies of adults. In a study evaluating the 42 problem of relatedness when sampling juveniles for molecular analysis, Hansen et al. (1997) found that young-of-the-year brown trout (Salmo truttd) captured within short stream sections came from a small number of full-sibling families. This biased estimates of allele frequencies. To reduce the likelihood of sampling families, I used no more than three young-of-the-year bull trout from each sampling site, and maintained a minimum of 250 m between all sample sites. Although Hansen et al. (1997) found that the problem of sampling families was not apparent in age classes older than young-of-the-year, I also used no more than three fish of older age classes from each sample site. 43 Figure 3-1. Juvenile bull trout tissue sampling locations (n = 14) in Westfall River. 44 F i g u r e 3-2. Juveni le b u l l trout tissue s a m p l i n g locations (n = 13) in H o u s t o n C r e e k . 45 18 -r Z3 o .Q E Fork Length (mm) Figure 3-3. Fork length frequency of juvenile bull trout captured in the Westfall River, 15 July 12 August 1997 (n= 167). 10 -r U o 1 <D .Q E Fork Length (mm) Figure 3-4. Fork length frequency of juvenile bull trout captured in the Houston Creek, 20 - 22 August 1997 (n= 104). 46 Microsatellite amplification Total D N A was extracted as described by Taggart et al. 1992 and Taylor et al. 1996. Briefly, I suspended tissue in 2 m M EDTA and digested with Pronase (24 h at 37 °C with rotation) and RNAase (1 h at 37 °C). The D N A was extracted using standard phenol:chloroform protocols followed by ethanol precipitation. I resuspended the extracted D N A in TE buffer and quantified the concentration by spectrophotometry. I stored extracted D N A samples at -20 °C. From extracted D N A samples, nine microsatellite loci were amplified separately with the polymerase chain reaction (PCR). The microsatellite loci I used came from two sources: seven were isolated from other species (Ogo 2 - Olsen et al. 1998; Omy 11 -Morris et al. 1996; Ots 101 - Nelson and Beacham 1999; Ssa 197 - O'Reilly et al. 1996; Ssa 311 and Ssa 456 - Slettan et al. 1995; Sfo 18 - Angers et al. 1995) and two were isolated from a bull trout genomic library (Sco 19 and 23; E.B. Taylor, unpubl. data). A l l of the loci isolated from other species are polymorphic in populations of bull trout in British Columbia and Montana (Spruell et al. 1999; Costello and Taylor unpubl. MS). One primer from each primer pair was P 3 2 end-labeled, using a Polynucleotide Kinase kit (New England Biolabs). Locus specific PCR protocols are appended. A l l PCR reactions were run on a M J Research PTC-100 thermocycler. A l l samples were amplified in 10 uL reaction volumes containing: 100 ng template D N A and 0.5 U Taq D N A polymerase (Gibco/BRL) in 10 pi of 20mM Tris-HCl (pH 8.4), 50 m M KC1, 1.5mM M g C l 2 , 800 p M dNTPs, 0.5 pmol of [v^PJ-dATP labeled forward primer, 2.5 pmol unlabelled forward primer, and 6 pmol of reverse primer in dEI^O. Four of the loci (Ogo 2, Ots 101, Sco 19, SKI 311) also included 10% dimethyl sulfoxide (DMSO) in the reaction volume (Winship 1989). Amplification products were size fractionated on 6% denaturing polyacrylamide gels and autoradiographed. Alleles sizes were determined using a size standard ( M l 3 sequence) run in two or more lanes of each gel. Data analysis Allele frequencies, heterozygosity (both observed and expected), and tests for departures from Hardy-Weinburg Equilibrium (HWE) were calculated using TFPGA 1.3 47 (Miller 1997) for all loci scored by age class and stream. I excluded monomorphic loci from subsequent analyses. I used GENEPOP 3.2 (Raymond and Rousset 1995) to test for genotypic linkage disequilibrium among loci and to calculate across loci pairwise exact tests of genetic differentiation based on allele frequencies among age classes within streams. In all simultaneous statistical tests, I used the sequential Bonferroni correction (Rice 1989) to adjust alpha values. To estimate the amount of genetic differentiation between Houston Cr. and Westfall R., I used TFPGA and RST C A L C 2.2 (Goodman 1997) to calculate locus specific and across loci estimates of FST and RST, respectively, for samples pooled by stream. Allele data were entered as the number of repeat units (with the smallest allele size being one repeat unit) for calculation of RST estimates using RST C A L C . I also calculated locus specific and across loci exact tests of genetic differentiation based on allele frequencies for samples pooled by stream (TFPGA). As an alternate test of differentiation between Houston Cr. and Westfall R., I calculated hierarchical structuring of genetic variance using analysis of molecular variance ( A M O V A ; Excoffier et al. 1992) with ARLEQULN 2.0 (Schneider et al. 2000). I used only a FST based genetic variance matrix for A M O V A . In addition, I calculated A M O V A for both age structured samples by stream, and for samples pooled by stream. I calculated the number of migrants from the pooled by stream across loci estimate of FST (FST « l/(4N em + 1) - Wright 1943; Whitlock and McCauley 1999) to compare qualitatively with the magnitude of movement (migration) estimated from telemetry (Chapter 2). I used bootstrapped 95% confidence intervals of the FST estimate to calculate approximate confidence intervals of the migration estimate. 48 3.3 Results Microsatellite loci Of the nine loci I amplified, four were polymorphic (95% criteria; Hartl and Clark 1989), two had rare second alleles (polymorphic at the 99% criteria), and three were monomorphic (Table 3-1). I will use the 99% criteria as my definition of polymorphic, meaning that the most common allele is at a frequency < 0.99 (Hartl and Clark 1989). A composite autoradiograph of the six polymorphic loci is presented in Figure 3-5. For the polymorphic loci, the mean number of alleles is 3.7, but over all nine scored loci, the mean is 1.6. Westfall River samples had two private alleles (Ogo 2 -148bp and Ssa 456 - 159bp) that occurred at frequencies of 0.008 and 0.017, respectively. Although I found all loci pairs in linkage equilibrium across stream level sample groups (P > 0.164), within age class samples P = 0.027 for the Ssa 197 - Sco 19 pair comparison in the Houston Creek 0+ sample group. After sequential Bonferroni correction (Rice 1989; 15 simultaneous tests, a = 0.003), this P value is not significant. Sample sizes and number of alleles by specific stream, age class group, and locus are presented in Table 3-2. Observed and expected heterozygosities are presented in Table 3-2. Estimates of P values for tests of H W E indicated no significant departures from equilibrium allele frequencies across all age sample groupings (P > 0.056; Table 3-2). The mean expected heterozygosity over all nine scored loci by age class group is 0.220 (min. 0.00 - max. 0.770). Over the six polymorphic loci only, the mean expected heterozygosity is 0.330 (min. 0.00 - max. 0.770; Table 3-2). Allele frequency plots for the six polymorphic loci by age class group and stream are appended. . 49 s s s s s s s s s s s s s s s s -•-170 -156 -154 '152 -146 202 200 •194 •188 •174 f 3> 1 Ogo 2 [fit Omy 77 > . ft * o s s s s s s s s s s s s s s Sco 19 s s s s s s s s s s s I L J r i ^159 •157 » * 1 $ Ssa 197 119 I 2 Sco 23 Ssa 456 Figure 3-5. A composite autoradiograph of the six polymorphic microsatellite loci scored in bull trout from Houston Creek and the Westfall River. Sample lanes are indicated with an 'S' . Allele sizes (bp) are indicated (arrows). 50 Table 3-1. Allele size (bp) and frequency (freq.) over all bull trout samples for nine microsatellite loci. Locus 1 2 3 Allele 4 5 6 7 Ogo 2 bp 146 148 152 154 156 170 freq. 0.315 0.005 0.095 0.257 0.293 0.036 Omy 11 bp 279 283 287 freq. 0.086 0.054 0.860 Ots 101 bp 100 freq. 1.000 Sco 19 bp 174 188 194 200 202 204 206 freq. 0.068 0.064 0.109 0.682 0.050 0.018 0.009 Sco 23 bp 193 195 freq. 0.559 0.441 Sfo 18 bp 150 freq. 1.000 Ssa 197 bp 119 123 freq. 0.050 0.950 Ssa 311 bp 120 freq. 1.0000 Ssa 456 bp 157 159 freq. 0.991 0.009 51 Table 3-2. Sample size (n), number of alleles amplified (A), and observed (H (obs)) and expected (H (exp)) heterozygosity values by stream, age class group, and locus, as well as estimated probability values (P) of HWE tests. Locus Westfall Houston 0+ 1+ 2+ 0+ 1+ 2+ Ogo 2 n 20 20 20 20 20 11 A 5 4 6 5 4 4 H (obs) 0.900 0.800 0.750 0.700 0.700 0.636 H (exp) 0.734 0.696 0.770 0.744 0.671 0.665 P 0.913 0.730 0.721 0.271 0.423 0.105 Omy 11 n 20 20 20 20 20 11 A 3 3 3 3 3 2 H (obs) 0.200 0.350 0.150 0.250 0.350 0.273 H (exp) 0.184 0.301 0.141 0.304 0.304 0.236 P 1.000 1.000 1.000 0.056 1.000 1.000 Ots 101 n 20 20 20 20 20 10 A 1 1 1 1 1 1 H (obs) - - - - - -H (exp) - - - - - -P - - - - - -Sco 19 n 20 20 20 20 20 10 A 6 6 6 5 4 2 H (obs) 0.550 0.550 0.500 0.600 0.350 0.200 H (exp) 0.596 0.600 0.486 0.616 0.306 0.185 P 0.266 0.206 0.733 0.096 1.000 1.000 Sco 23 n 20 20 20 20 20 11 A 2 2 2 2 2 2 H (obs) 0.450 0.600 0.450 0.500 0.600 0.546 H (exp) 0.469 0.495 0.399 0.420 0.455 0.496 P 1.000 0.659 1.000 0.618 0.326 1.000 Sfo 18 n 20 20 20 20 20 10 A 1 1 1 1 1 1 H (obs) - - - - - -H (exp) - - - - - -P - - - - - -Ssa 197 n 20 20 20 20 20 10 A 2 1 2 2 2 1 H (obs) 0.200 - 0.150 0.150 0.050 -H (exp) 0.180 - 0.139 0.139 0.049 -P 1.000 - 1.000 1.000 1.000 -Ssa 311 n 20 20 20 20 20 11 A 1 1 1 1 1 1 H (obs) - - - - - -H (exp) - - - - - -P - - - - - -Ssa 456 n 20 20 20 20 20 11 A 1 2 2 1 1 1 H (obs) - 0.050 0.050 - - -H (exp) - 0.049 0.049 - - -P - 1.000 1.000 - - -52 Genetic population structure I found little evidence of genetic divergence between Houston Creek and Westfall River bull trout with the microsatellite loci I amplified. After sequential Bonferroni correction, no pairwise age class comparison had significantly differing allele frequencies (P > 0.011, 15 simultaneous tests, a = 0.003 - Table 3-3), so I pooled samples by stream for further analysis. Locus-specific estimates of FST and RST (Table 3-4) again suggest little genetic differentiation between streams. The across loci estimates of FST and RST are 0.010 and 0.006 respectively. The bootstrapped 95% confidence intervals of both across loci FST and RST estimates include zero (Table 3-4). The .Sco 23 specific estimates are higher than the other loci (Table 3-4) but generally there is concordance among the polymorphic loci I scored. The number of effective migrants (Nem) calculated from the across loci estimate of FST is 24.8. The approximate 95% confidence intervals are 7.2< N em< °°. The between stream exact test of population differentiation across all loci is not significant (P = 0.284; Table 3-5). The probability value estimated for the Sco 23 specific exact test, P = 0.011, is not significant after Bonferroni correction (6 simultaneous tests, a = 0.008; Table 3-5). Table 3-3. Estimated probability values of exact tests of population differentiation from allele frequencies across loci between all age class groups (0+ - 2+) from Houston Creek (H) and Westfall River (W). Age group H0+ H1+ H2+ W0+ W1+ H1+ 0.011 H2+ 0.143 0.311 W0+ 0.910 0.147 0.403 W1+ 0.475 0.387 0.716 0.566 W2+ 0.050 0.015 0.238 0.391 0.171 53 Table 3-4. Estimates of F S T and R S T within and across loci for pooled age class samples by stream. Confidence intervals (C.I.) for across loci estimates are included. Locus Fst Rst Ogo 2 -0.006 -0.009 Omy 11 -0.001 -0.002 Sco 19 0.007 -0.002 Sco 23 0.045 0.044 Ssa 197 -0.005 -0.005 Ssa 456 0.007 0.008 Across loci estimate 0.010 0.006 Across loci 95% C.I. 0.034>Fst>-0.005 0.168>Rst>-0.001 Table 3-5. Estimated probability values of exact tests of population differentiation, based on allele frequency differences (Infinite Alleles Model), between Houston Creek and Westfall River samples (pooled across age classes) within loci, and across all loci (bold). 'SE' is the standard error of the estimate. Locus P value SE Ogo 2 0.694 0.016 Omy 11 0.547 0.009 Sco 19 0.480 0.015 Sco 23 0.011 0.002 Ssa 197 0.762 0.002 Ssa 456 0.503 0.003 Across loci estimate 0.284 Partitioning of genetic variation ( A M O V A ) for both pooled and separate age classes within streams are presented in Table 3-6. I present the A M O V A for separate age classes, despite the lack of pairwise differences between them, to compare the proportion of genetic variance partitioned among age classes within streams and between streams. Pooling age classes resulted in a 0.7% increase in genetic variance partitioned to the between stream partition (0.2 to 0.9%, Table 3-6 and Fig. 3-6). The between stream variance partition of the A M O V A is significant when age classes are pooled by stream (P = 0.036) but not when age classes are separate (P = 0.416; Table 3-6). A l l other variance partitions in the A M O V A are significant (Table 3-6). 54 Table 3-6. Analysis of Molecular Variance (AMOVA) partitioning for both pooled and separate age classes by stream. Estimated probability values associated with variance partitions are included. 'SE' is the standard error of the estimate. AMOVA Partitions pooled over age classes by stream Within streams Between streams P value estimate (SE) 99.1% 0.9% 0.036 (0.006) separate age classes Among age classes within streams P value estimate (SE) Within age class P value estimate (SE) Between streams P value estimate (SE) 97.6% 0.001 (0.001) 2.2% 0.002 (0.001) 0.2% 0.416 (0.014) 3.4 Discussion I was unable to collect many 2+ bull trout from Houston Creek. The apparent lack of 2+ bull trout in Houston Cr. probably reflects the smaller portion of the stream sampled (Fig. 3-1 vs. Fig. 3-2) and lower total catch, because the relative frequency of different age classes between Houston Cr. and Westfall R. appears similar (Figs. 3-3 and 3-4). The apparent size differences between the same age classes from the two streams (Figs. 3-3 and 3-4) probably represent differences in sampling date (> 30 d in most cases) rather than stream specific growth rates. More than half of the microsatellite loci amplified (Table 3-1) were monomorphic by the 95% criterion. I used these loci specifically because they were polymorphic in other populations of bull trout (Spruell et al. 1999; Costello and Taylor unpublished manuscript). There were no significant departures from HWE. Heterozygosities are low when averaged over loci (mean across loci = 0.220; Table 3-2), but this appears to be normal in bull trout (Spruell and Allendorf 1997), and as mentioned previously, the general finding of low genetic variability in bull trout has been attributed to historical demographic processes (Leary et al. 1993; Spruell and Allendorf 1997; Taylor et al. 1999; Costello and Taylor unpubl. MS). Although not significant, the estimated P values of pairwise exact tests between Houston Cr. 0+ and Houston Cr. 1+ (H0+ vs. H1+), H0+ and Westfall R. 2+ (H0+ vs. 55 W2+) and H1+ vs. W2+ samples (Table 3-3) suggest some allele frequency differences. In the case of the H0+ vs. H1+ comparison, the loci driving the differences were Sco 19 and Ogo 2. These two loci have the most allelic diversity in my samples (Sco 19 - seven alleles, Ogo 2 - six alleles; Table 3-1), and the difference suggested by the low estimated P value (Table 3-3) probably result from my within age group sample sizes. My sample sizes are certainly of issue at the individual age class level, particularly in the 2+ age class from Houston Cr. (Table 3-2). For the other two comparisons with low estimated P values (H0+ vs. W2+ and H1+ vs. W2+; Table 3-3), the locus driving the result was Sco 23 (two alleles; Table 3-1), suggesting that my sample sizes were not as important. After Bonferroni correction (Rice 1989), none of the pairwise age class exact tests were significantly different (Table 3-3). In calculated estimates of genetic differentiation, FST is consistently higher than RST (Table 3-4). The microsatellite allele sizes identified in Houston Cr. and Westfall R. bull trout are consistent with those identified by other bull trout studies with the same loci (Spruell et al. 1999; Taylor and Costello 1999; Latham and Taylor in press; Costello and Taylor unpublished manuscript). Because the same sized alleles exist at the same loci elsewhere in the range of bull trout, this suggests that they did not arise via mutation in Houston Cr. and Westfall R. If microsatellite alleles did not arise by mutation, and instead arrived when the area was recolonized, the estimate of RST is not as good an indicator of genetic differentiation as F S T- Taylor and McPhail 2000 argue that demographic processes (random genetic drift, founder effects and bottlenecks) likely outweigh the effect of mutation on genetic differentiation in threespine stickleback (Gasterosteus aculeatus) postglacially recolonizing BC lakes from the sea. With evidence for the importance of the same demographic processes in bull trout (Leary et al. 1993; Spruell and Allendorf 1997; Taylor et al. 1999; Costello and Taylor unpubl. MS), mutation likely plays a minor role in the microsatellite differentiation I found. There is little genetic differentiation between juvenile bull trout sampled from Houston Cr. and Westfall R. The Sco 23 locus (Table 3-4) appears to drive the only differences I found between the two streams. Besides Sco 23, there is good concordance across the loci I scored: they all indicate no differentiation between the streams (Table 3-56 4). These results are reflected in the non-significant exact test of differentiation based on allele frequencies across loci (P = 0.284; Table 3-5). The amount of genetic differentiation I found (Table 3-4) is very small compared to estimates from other fish species (Ward et al. 1994; Waples 1998), particularly other freshwater fish. My estimate of FST (0.01) between Houston Cr. and Westfall R . is six times less than the mean for anadromous fishes (mean marine FST estimate = 0.06), and more than an order of magnitude lower than the mean estimate for freshwater fishes (mean freshwater FST estimate = 0.222; as reviewed by Ward et al. 1994). Because of fewer strong barriers to dispersal, marine organisms generally have lower levels of genetic differentiation across their distributions (e.g. Palumbi et al. 1997; reviewed by Waples 1998). The low mean estimate of FST from the studies reviewed by Ward et al. (1994) may be because many were surveys of allozyme differentiation, which usually have lower resolution of genetic variation than microsatellites (Ferguson et al. 1995; Jarne and Lagoda 1996). Most of the loci I used had low allelic diversity (only two loci had > 4 alleles; Table 3-1) compared to many microsatellite systems (Hedrick 1999; Nelson and Beacham 1999); however, I still found diversity, just very little differentiation between Houston Cr. and Westfall R . Another way to evaluate my result is to compare it with the amount of differentiation expected between isolated populations diverging via genetic drift since founding from a common source. Imagine that Houston Creek and Westfall River represent two such populations. Assuming discreet generations of five years, no gene flow between populations of constant effective population size (N e = 300), and that colonization occurred 10,000 years ago, FST = 0.964 based on genetic drift alone (Nei and Chakravarti 1977 as cited in Waples 1998). I also used A M O V A (Table 3-6) to quantify genetic divergence between Houston Cr. and Westfall R bull trout. Because of the previous discussion of the applicability of RST based measures, I only calculated the A M O V A with FST based distances (Excoffier et al. 1992). With pooled age classes by stream, a significant portion (0.9%, P = 0.036; Table 3-6) of genetic variance is partitioned to differences between streams. Although sample sizes are small, I presented the A M O V A analysis with 57 separate age classes (Table 3-6), specifically to compare the amount of genetic variation attributed to the between stream partition relative to the among age classes within streams partition (Fig. 3-6). Just over 2 % (P = 0.002) of the total variation is attributed to the among age class partition in the analysis (Table 3-6). The pairwise among age classes (Table 3-1) and between stream exact test (Table 3-5) are both contradicted by these A M O V A results. This may reflect a difference in power between the two methods. Despite this apparent discrepancy, the amount of genetic differentiation between Houston Cr. and Westfall R. is still small. Less than one percent of genetic variance is explained by between stream differences. This A N O V A result, along with FST estimates and exact tests of population differentiation, suggest that the spatial scale of population divergence in upper Duncan River bull trout is larger than the spawning stream. As discussed in the methods section, a problem with this general result is the use of allele frequency data collected from juveniles. I replicated my samples over age classes (time), took precautions against sampling families within a year, and used multiple loci in an attempt to alleviate (as suggested by Hansen et al. 1997) the biases introduced by the Allendorf-Phelps effect (Allendorf and Phelps 1981; Waples 1998). As detailed by Waples (1998), the net bias introduced by the sampling of juveniles instead of the adult population would be to inflate estimates of population subdivision. My data, with low differentiation between age classes and in pooled between stream comparisons, suggest that the Allendorf-Phelps effect has not been important in my analysis. Sample sizes are also a concern in the interpretation of the general lack of between stream differences. The 0 estimator of FST I used (Weir and Cockerham 1984) adjusts population specific estimates by sample size; however, in one age class group (Houston Cr. 2+) my mean sample size over loci was 10.6 individuals (Table 3-2). When two of four polymorphic loci used have more than five alleles, this sample size is not sufficient. My pooled age class between stream analyses are based on mean sample sizes over six loci of 50.6 individuals from Houston Cr. and 60 individuals from Westfall R. Utter et al. 1992, describe a similar result when comparing chinook salmon (Oncorhynchus tshawytscha) populations in the Snake and Klamath rivers. Originally, attempts to differentiate these two groups of chinook were not successful using allozyme 58 markers (Utter et al. 1992). Only later, when more variable allozyme loci were used, did the two populations become distinguishable (Utter et al. 1992). With additional loci, perhaps more differentiation between Houston Cr. and Westfall R. bull trout would be apparent; although, in terms of the pooled age class test, my sample size (n > 50 per locus and stream) and number of loci (six polymorphic by the 99% criteria) are greater than many similar studies (e.g. Spruell et al. 1999; Latham and Taylor in press; Costello and Taylor unpubl. MS). Finally, a potential problem with attempting to define population structure (or lack thereof) using highly variable markers like microsatellites was raised by Hedrick (1999). In this study, with little evidence of population differences, this lack of differentiation may only reflect these specific microsatellite markers. The problem is that other loci, potentially loci defining adaptively significant traits, may give rise to biologically meaningful differences (Hedrick 1999). I tried to minimize this possibility by amplifying nine loci that were known to be polymorphic in bull trout. Averaging over more independent loci would certainly increase the accuracy of estimates of genetic differentiation (Waples and Teel 1990). Although I did not measure many adaptive traits, it is clear that migratory timing between the two streams does not differ (Chapter 2). The relative proportions of different juvenile age classes present during sampling for microsatellite analysis (Figs. 3-3 and 3-4) suggest that juvenile rearing times are similar. General life history, with migrations to and from Kootenay Lake, appears the same in bull trout from both streams. There was a significant, but not robust, trend for larger radio-tagged bull trout in larger tributary streams, but this may also reflect a lack of precise homing (Chapter 2). Number of migrants As described by Whitlock and McCauley (1999), estimation of the number of migrants (Nem) from FST is not accurate. The assumptions of the "Fantasy Island" model (Wright's island model of population subdivision - Wright 1943; Whitlock and McCauley 1999), which should be met for calculation of number of migrants from FST, have been violated in my calculations. Specifically, Wright's island model assumes: that the number of populations is infinite; population sizes are equal and do not fluctuate; 59 breeding is random; generations are discreet; each population contributes the same number of migrants, and the probability of migration to one population is equal to all others; the migration rate is small; and that there is no selection or mutation. Telemetry data showing variable number of tagged bull trout spawning in various spawning stream 'populations' suggest that population size is different among streams. Telemetry also suggests that the migration rate may be quite large. I also only sampled two of an unknown number of possible populations that have overlapping generations. Despite these breaches of the assumptions of the model, I decided that the estimation of N e m was justified to allow at least a qualitative comparison of movement rates between spawning streams in the upper Duncan R. based on genetic and telemetry data. The problem of estimation of N e m from FST is particularly acute at small values of FST, as found in this study. When I used the bootstrap estimates of the 95% confidence interval to calculate approximate confidence intervals for my estimate of N e m , the range (from 7.2 to an infinite number of individuals) highlights this problem. N e m estimates are very sensitive to small changes in FST when FST < 0.01 (Waples 1989; Whitlock and McCauley 1999). On a purely qualitative level, my estimates of 'migration' between spawning streams based on telemetry and genetic data are similar. As reported in Chapter 2, an approximate estimate of the total number of bull trout changing spawning destination in the upper Duncan R. over two consecutive years is 40 to 50 individuals. The FST based estimate of 24.8 effective migrant individuals per generation equals approximately 5 effective migrants per year, between Houston Cr. and Westfall R. specifically, with a five year generation time. This qualitative agreement suggests two important things: first, that my violation of assumptions of the calculation of number of migrants from FST was not completely unrealistic; and second, that the telemetry spawning stream switching result was not spurious, and likely reflects a high level of current and historical gene flow between, at least, Houston Cr. and Westfall R. The impact of gene flow is dependent on population size (Hartl and Clark 1989). To best evaluate the effects of gene flow on population structure, I need an estimate of migration rate (m). The effective population size, N e , is the idealized number of breeding 60 individuals giving rise to observed levels of random genetic drift (Haiti and Clark 1989). In fishes, the proportion of census size that N e reflects is generally less than 20% (Frankham 1995; Miller and Kapuscinski 1997). If I assume that the number of spawning individuals in both Houston Cr. and Westfall R. average approximately 100 to 150 individuals (from Houston Cr. trap data - Chapter 2), N e may be as small as 20 to 30 individuals in each stream. With N e = 25, m = 1, suggesting that nearly all bull trout move between Houston Cr. and Westfall R. for spawning. With N e equal to estimates of spawning census size ( N e =150), migration rate is approximately 15 to 25 % per generation. In the most conservative case, using the lower 95% C.I. of the N e m estimate (7.8 individuals) and census size as N e (=150), migration rate is still > 5 %. A migration rate of 5% would tend to genetically homogenize otherwise separate populations (Haiti and Clark 1989; Mills and Allendorf 1996). Comparison to previous bull trout studies My results differ from those of other fine-scale bull trout genetic studies (Spruell et al. 1999; Costello and Taylor unpublished manuscript). Spruell et al. (1999) estimated FST at 0.063 between samples of bull trout from five tributary populations within the Lightning Cr. drainage, and found significant differentiation between all population pairs. Lightning Cr. is tributary to the Clark Fork River, Idaho, and drains approximately 220 km 2 into Lake Pend Oreille (Spruell et al. 1999), thus, it is less than 20 % the size of upper Duncan River. Costello and Taylor (unpubl. MS) estimated FST values ranging from 0.01 to 0.31, 0.01 to 0.68, and 0.08 to 0.93 across bull trout populations within watersheds tributary to the Peace, Kootenay, and Red Deer and Oldman rivers, respectively. Sample sites in Costello and Taylor's study (unpubl. MS) were separated by as much as 200 km in each system. Generally, where Costello and Taylor (unpubl. MS) estimated small, but significant, pairwise FST values, sample sites were within the same watershed or even the same stream. Only in one small watershed did Costello and Taylor (unpublished MS) find no significant genetic differentiation among samples (see below). One difference between my study and the other studies is that Houston Cr. and Westfall R. seem to support only migratory bull trout (Chapter 2). Costello and Taylor (unpublished manuscript) found that if they removed samples collected from above 61 barriers to migration (populations geographically isolated from migrant populations) from their analysis, it reduced within watershed estimates of FST by approximately one half. Spruell et al. (1999) give no specifics detailing which, i f any, of the populations they sampled were isolated. Perhaps sampling isolated populations inflated their estimates of genetic differentiation. Latham and Taylor (in press), examined population structuring among bull trout spawning streams tributary to Arrow Lake in the Columbia River, adjacent to Duncan River. Maximum distance between sampled streams in Arrow Lake and the Columbia was approximately 200 km (Latham and Taylor in press). They found evidence of genetic divergence among these streams, specifically a north - south shift in both microsatellite allele frequencies and mtDNA haplotype patterns. Latham and Taylor (in press) estimated a FST of 0.11 for microsatellites across their sampling area, excluding above barrier populations. They interpreted their results as strong evidence for either accurate homing or low reproductive success in straying fish. Latham and Taylor's (in press) general result, like that of Spruell et al. 1999 and most areas examined by Costello and Taylor (unpubl. MS) stand in sharp contrast to my results. My findings are consistent with genetic results from one other watershed. Costello and Taylor (unpubl. MS) found insignificant pairwise FST among three sample sites within the Wigwam River, a tributary of the Elk River that drains nearly 800 km . The Elk River flows into Kokanoosa Reservoir, on the Kootenay River upstream of Kootenay Lake. Like the upper Duncan River, a lacustrine population of bull trout (from Kokanoosa Reservoir) spawns in Wigwam River; although, prior to the construction of Libby Dam (1972), this would have been a fluvial population. There are several possible reasons why my findings from the upper Duncan River differ from most other bull trout studies. One possible reason, as described previously, is that my sites were not isolated from each other. Telemetry results (Chapter 2), and the apparent absence of large mature fish prior to arrival of migrating adults, suggest that Houston Cr. and Westfall R. do not contain resident adult bull trout. Another possibility is that the spatial scale of my comparison is too small; however, one purpose of my study was to determine if more than one population spawned in the upper Duncan 62 River. Although the streams I sampled for this comparison are geographically close together, distances are apparently more than enough for significant population divergence in bull trout (Spruell et al. 1999; Costello and Taylor unpubl. MS), and certainly other salmonids (Quinn et al. 1987; Taylor 1991; Tallman and Healey 1994; Angers et al. 1995; Carlsson et al. 1999). Perhaps Duncan Dam has had an impact on genetic population structure within the upper Duncan R., by restricting access from Kootenay Lake. In 1997, the dam had been present for 30 years. At most, this equates to six bull trout generations and the impact of gene flow and other demographic processes on allele frequency differences among populations are usually integrated over longer time scales than this (Whitlock 1992). If bull trout from separate spawning streams in the upper Duncan River represented genetically distinct populations prior to 1967 and the effect of the dam was to make all of these populations breed randomly with one another, six generations would be enough to homogenize any genetic differentiation that had existed. This scenario seems unlikely, given evidence for movement through the dam in both directions by radio-tagged bull trout (Chapter 2). Conservation implications The genetic comparison between Houston Cr. and Westfall R. bull trout suggests that the switching behaviour suggested in Chapter 2 produces gene flow among upper Duncan tributaries. Although the microsatellite comparison includes only Houston Creek and Westfall River, the telemetry results indicate that this result applies across the streams entered by radio-tagged bull trout in the upper Duncan River. Thus, the population boundary of bull trout in the upper Duncan River apparently includes all spawning tributaries. So, at what spatial scale are bull trout migrating to the upper Duncan River divided into genetically distinct populations? Without comparisons outside the upper Duncan River, it is only possible to speculate. Based on other studies of genetic differentiation, the next available watershed unit above individual spawning streams is the whole upper Duncan River. Perhaps, a comparison of migrant bull trout from Kootenay Lake entering the Duncan and Lardeau rivers might reveal genetic divergence between these major Kootenay Lake inlet rivers. My results suggest that from 63 the perspective of conservation and management, the bull trout that migrate from Kootenay Lake to the upper Duncan River should be considered part of the same, essentially panmictic, population. 3.5 Appendix Table 3-7. Loci specific PRC protocols. Reaction step abbreviations are: D - denature, A -anneal, E - extend. Time notation is m:ss. The reactions of starred loci included DMSO (see text). PCR Profile Locus Step 1 D A E cycles Step 2 D A E cycles Step 3 D A E cycles Ogo 2 * temp. 95 1 94 56 72 5 93 55 72 25 time 3:00 1:00 1:00 1:00 1:00 1:00 1:00 Omy 11 temp. 95 56 72 1 94 56 72 5 92 54 72 25 time 3:00 2:00 1:00 1:00 1:00 1:00 1:00 0:30 0:30 Ots 101 * temp. 95 1 94 56 72 5 93 55 72 25 time 3:00 1:00 1:00 1:00 1:00 1:00 1:00 Sco 19* temp. 95 52 72 1 94 52 72 5 92 50 72 25 time 3:00 2:00 3:00 1:00 1:00 1:00 1:00 1:00 1:00 Sco 23 temp. 95 61 72 1 94 61 72 5 92 60 72 25 time 2:00 1:00 1:00 1:00 1:00 1:00 0:45 0:45 0:30 Sfo 18 temp. 94 65 72 15 94 65 72 20 time 1:00 0:35 0:10 0:35 0:35 0:10 Ssa 197 temp. 95 58 72 1 94 58 72 5 94 57 72 25 time 2:00 1:00 0:30 1:00 1:00 0:30 0:45 0:45 0:30 Ssa 311 * temp. 95 1 94 56 72 5 93 55 72 25 time 3:00 1:00 1:00 1:00 1:00 1:00 1:00 Ssa 456 temp. 95 1 94 56 72 5 93 55 72 25 time 3:00 1:00 1:00 1:00 1:00 1:00 1:00 64 0.75 0.5 0.25 0.75 0.5 0.25 H 0.75 H 0.5 0.25 H I* It. 1—• 1 •—|—•—'—i—'—•—I—' i " I 1 I I I r 146 148 152 154 156 170 146 148 152 154 156 170 Figure 3-6. The allele frequencies of microsatellite locus Ogo 2 from sampled bull trout by age class (0+, 1+, and 2+ top to bottom) and stream (Houston Creek - right, Westfall River - left). 65 Figure 3-7. The allele frequencies of microsatellite locus Omy 11 from sampled bull trout by age class (0+, 1+, and 2+ top to bottom) and stream (Houston Creek - right, Westfall River - left). 66 0.75 -0.5 -0.25 -B •j r n • — i i l l r~I i B WM 1 1 1 1 l l 174 188 194 200 202 204 206 174 188 194 200 202 204 20fi Figure 3-8. The allele frequencies of microsatellite locus Sco 19 from sampled bull trout by age class (0+, 1+, and 2+ top to bottom) and stream (Houston Creek - right, Westfall River - left). 67 0.75 0.5 -0.25 0.75 0.5 0.25 0.75 0.5 0.25 " 0 --!f.r 'fa f§p|||||l IMRS 193 Figure 3-9. The allele frequencies of microsatellite locus Sco 23 from sampled bull trout by age class (0+, 1+, and 2+ top to bottom) and stream (Houston Creek - right, Westfall River - left). 68 0.75 -0.5 0.25 0.75 0.5 0.25 0.75 0.5 0.25 »*• — I J JL • i i .'i«r«- 1 I 1 1 9 1 2 3 1 1 9 1 2 3 Figure 3-10. The allele frequencies of microsatellite locus Ssa 197 from sampled bull trout by age class (0+, 1+, and 2+ top to bottom) and stream (Houston Creek - right, Westfall River - left). 69 1 -I 0.75 -0.25 -0.75 H 0.5 -0.25 H liiijBSiSi! iillffillllPHllil SiiSISiiilliit ••1 •HHHH INHHH1 S^ ^^ H|i|i||S8Hp HJHHIJK HHHI * J p - -*S,* ,n.i r -.- , sM* • -^^ WSIiSHiiiHiiili B-isiiiiiii*': ^^^^^^^^^^^ IKIflHHp .r,i.. . o -'—1 . 1 1 1 — i 1 ' ' 157 159 157 159 ;ure 3-11. The allele frequencies of microsatellite locus Ssa 456 from sampled bull trout by age class (0+, 1+, and 2+ top to bottom) and stream (Houston Creek - right, Westfall River-left). Chapter 4 - General conclusions Migratory timing and spawning locations in the upper Duncan River I wanted to determine the timing of bull trout migrations in the Duncan River. In addition, I wanted to identify which streams bull trout use for spawning in the upper Duncan River. Using radio telemetry, I recorded the migration of bull trout from Kootenay Lake to spawning locations in the upper Duncan River, where I identified six putative bull trout spawning streams. Over two years, the proportion of bull trout migrating to these streams and destinations in the mainstem upper Duncan River did not change. The timing of migrations did not vary over the two years of detailed telemetry. These predictable migrations allow conservation and management measures, such as angling restrictions, land use decisions, or flow releases from dams, to focus on specific locations and times that will best benefit the migrating fish. Apparently, bull trout migrating to destinations in the mainstem upper Duncan River do not spawn. Swanberg (1997) suggested that similar findings for bull trout in the Blackfoot River were related to avoidance of high summer water temperatures. The high summer temperatures in Kootenay Lake and Duncan River probably are not warm enough to explain these non-spawning migrations. The non-spawning movements were not related to feeding, as prey fish were scarce in the upper Duncan River during the summer and fall sampling periods. This result suggests that adult bull trout use more river habitat in the summer than suggested from spawning area use alone. Perhaps to best protect bull trout, habitat protection measures should focus on the upper reaches of spawning systems, not just specific spawning locations. Spawning stream selection I examined the hypothesis that bull trout select spawning stream based on size, as suggested by McPhail and Murray (1979). The size of radio-tagged bull trout, both in length and mass, differed by destination in the upper Duncan River. There was a significant, but not robust, relationship between fish length and catchment area of the tributary they entered. This result suggests support for the suggestion that bull trout 71 switch to larger spawning streams as they age and grow (McPhail and Murray 1979). The result may also mean that there is a general lack of accurate homing at the level of spawning stream within the upper Duncan River. If bull trout spawning habitat requirements change as they grow, then perhaps the lacustrine life history of Kootenay Lake bull trout, and the large sizes they attain, result in reduced homing accuracy. Of six bull trout tracked migrating into the upper Duncan River in consecutive years, four changed their destination in the upper Duncan River by 10 km from the previous year. Two of the six migrated to putative spawning tributaries in both tracking years, and both switched to a stream with a smaller catchment area during the second year. This result is contradictory to McPhail and Murray's (1979) hypothesis. The stream switching behaviour of these fish suggests that bull trout do not home to specific spawning streams in the upper Duncan River. Without confirmation of spawning at both locations, switching tributaries does not necessarily reflect gene flow between spawning locations. I used fish trap captures and known numbers of radio-tagged fish to extrapolate a rough number of spawning-stream-switchers. This extrapolation is very crude, but represents my attempt to qualitatively compare the results of telemetry with estimates of the number of migrants inferred from microsatellite allele frequency data. Population structure and gene flow I evaluated the spatial scale of population structure in the upper Duncan River. I asked if bull trout in different spawning streams act as genetically distinct populations. I was also interested in qualitative comparisons between the amount of movement between spawning streams and estimates of gene flow from molecular data. I collected microsatellite data from juvenile bull trout sampled in Houston Creek and Westfall River (two confirmed spawning tributaries in the upper Duncan River). A Fisher exact test of differentiation between the two tributaries using six polymorphic loci was not significant. An A M O V A , however, suggests a small (< 1%) but significant component of genetic variation is attributable to the between stream variance component. This suggests that bull trout in the upper Duncan River, as suggested by telemetry results, but contrary to 72 most other bull trout genetic studies done at similar spatial scales, are not divided into separate populations in different spawning tributaries. I calculated the number of effective migrants from estimates of genetic differentiation between spawning streams. This estimate is similar to the extrapolation of the number of spawning-stream-switchers using telemetry. A conservative estimate of migration rate is > 5 %. The difference between this result and other results from studies of bull trout and other fish species is of concern. The spatial scale I examined is small; however, significant differentiation has been found at smaller geographic scales (Spruell et al. 1999; Costello and Taylor unpubl. MS). Another possibility is that some of the differentiation found in other studies was related to sampling resident bull trout populations. In some cases there is no clear indication of the status of populations sampled (Spruell et al. 1999), while in others (Latham and Taylor in press; Costello and Taylor unpubl. MS), above barrier populations were removed and analysed separately. My results are reasonable considering that pooled sample size is > 50 for six polymorphic microsatellite loci in each spawning stream. It is possible that uncovered differentiation, and perhaps adaptive differences, exist between the two streams I compared (Utter et al. 1992;Hedrick 1999). Results similar to this study were found in the Wigwam River (Costello and Taylor, unpubl. MS). Wigwam River bull trout have a similar life history to the bull trout migrating into the upper Duncan River (Oliver 1979), and sample sites are from areas below barriers to migration (Costello and Taylor, unpubl. MS). Perhaps the similar life history, and growth to large size, plays an important role in the apparent lack of population divergence within the Wigwam and upper Duncan rivers. Comparison between methods Direct evidence of bull trout switching spawning stream from telemetry was reflected in estimates of genetic differentiation between Houston Creek and Westfall River. This result is not surprising. Often, a criticism of direct monitoring of dispersal or movement for making inferences about gene flow is that rare but biologically important movements are easily missed (Whitlock and McCauley 1999). In this study, by 73 restricting focus to the upper Duncan River, I was able to document a relatively common movement that indicated potential for high levels of gene flow. The microsatellite comparison presented in chapter 3 was certainly justified, as the telemetry data only suggested gene flow. On a qualitative level, the estimate of spawning-stream-switchers from telemetry was similar to the equivalent estimate, the number of effective migrants, from microsatellite allele frequencies. This suggests that all of the bull trout that switch spawning tributaries in the upper Duncan River successfully spawn in both locations. Also, this result suggests that the simple relationship between FST and N e m I used was robust to violations of the assumptions of the Wright's (1943) island model of population structure. Conservation and management implications The movement data I have collected allows conservation and management activities in the Duncan River to focus specifically on migrating fish. Habitat protection, angling regulations, and dam operation can be seasonally modified to better protect and assist bull trout movement and reproduction. As suggested above, the migration of non-spawning bull trout to areas downstream of spawning locations also provides information relevant to conservation and management. It seems that more habitat than would be predicted based on location of spawning sites is important for migrating bull trout. Perhaps habitat protection efforts should include more the upper reaches of bull trout spawning systems, not just spawning locations. The bull trout is an iteroparous salmonid, and as such accurate homing may provide less selective advantage for spawning than in semelparous species like Pacific salmon (Banks 1969). A missed opportunity for reproduction in bull trout may not greatly reduce fitness. In addition, bull trout appear specialized for cold water habitats (Rieman and Mclntyre 1993; Parkinson and Haas 1996). Perhaps the ephemeral nature of cold water habitats (often associated with mountainous terrain and glaciation) contributes to a lower selective advantage for homing to the natal stream. Indeed, some radio-tagged bull trout in the upper Duncan River migrated to the upper extent of available habitat. This suggests the species may have exploratory behaviour that is adaptive in ephemeral habitats. 74 Comparison of the result from Houston Creek and Westfall River to other bull trout genetic studies suggest that barriers to migration play an important role in genetic divergence in bull trout. For conservation purposes, when surveying bull trout genetic variation, care must be taken to identify populations isolated by migration barriers. It is important to note that barriers to migration are some times obvious, such as waterfalls or mountain ranges; however, they are not always readily apparent (Waples 1998; Bohonak 1999). My results suggest that where bull trout have open access to habitat, the spatial scale of populations may be larger than the typical unit of conservation and management -individual spawning streams - in salmonids (Taylor 1991). 75 Literature cited Allen, J. H . 1980. Life history notes on the Dolly Varden char (Salvelinus malma) in the Upper Clearwater River, Alberta. Red Deer, Alberta, Alberta Energy and Natural Resources, Fish and Wildlife Division. Allendorf, F. W., D. Bayles, D. Bottom, K. P. Currens, C. A . Frissell, D. Hankin, J. A . Lichatowich, W. Nehlsen, P. C. Trotter, and T. H . Williams. 1997. Prioritizing Pacific salmon stocks for conservation. Cons. Biol. 11: 140-152. Allendorf, F. W., and S. R. Phelps. 1981. Use of allelic frequencies to describe population structure. Can. J. Fish. Aquat. Sci. 38: 1507-1514. Angers, B., L. Bernatchez, A . Anger, and L. Desgroseillers. 1995. Specific microsatellite loci for brook char reveal strong population subdivision on a microgeographic scale. J. Fish Bio. 45: 177-185. Anonymous. 1986. Annual Report to the Governments of the United States and Canada. Washington, DC and Ottawa, O N , Columbia River Treaty Permanent Engineering Board. Anonymous. 1990. Flo-Mate Model 2000 Portable Water Flowmeter Instruction Manual. Frederick, Maryland, Marsh-McBirney Inc. Avise, J. C. 1995. Mitochondrial D N A polymorphism and a connection between genetics and demography of relevance to conservation. Cons. Biol. 9: 686-690. Banks, J. W. 1969. A review of the literature on the upstream migration of adult salmonids. J. Fish Bio. 1: 85-136. Barry, D. 1999. Endangered and threatened wildlife and plants; determination of threatened status for bull trout in the conterminous United States; final rule. U.S. Federal Register 64: 58910-58933. Barton, N . H . , and G. M . Hewitt. 1989. Adaptation, speciation and hybrid zones. Nature 341:497-503. Baxter, J. S., and J. D. McPhail. 1996. Bull trout spawning and rearing habitat requirements: summary of the literature. Vancouver, British Columbia, B.C. Ministry of Environment, Lands and Parks, Fisheries Branch. Bidgood, B. F. 1980. Field surgical procedure for implantation of radio tags in fish. Edmonton, Alberta, Alberta Department of Energy and Natural Resources - Fish and Wildlife Division. Bohonak, A . J. 1999. Dispersal, gene flow, and population structure. Quarterly Rev. Biol. 74: 21-45. Burrows, J., T. Euchner, and N . Baccante. in press. Bull trout movement patterns: Halfway River and Peace River Progress. Bull Trout II Conference Proceedings, Banff, A B , Trout Unlimited Canada. 76 Carlsson, J., K . H . Olsen, J. Nilsson, 0 . 0verli, and O. B. Stabell. 1999. Microsatellites reveal fine-scale genetic structure in stream-living brown trout. J. Fish Bio. 55: 1290-1303. Cavender, T. M . 1978. Taxonomy and distribution of the bull trout Salvelinus confluentus (Suckley), from the American northwest. Calif. Fish Game 64: 139-174. Clayton, T., and C. Mcleod. 1993. Seasonal movements of radio tagged fish in the upper Athabasca River summer 1992 to late winter 1993. Edmonton, R. L. & L. Environmental Services Ltd. Costello, A . B., and E. B. Taylor, unpublished manuscript. The role of historical and ecological factors in structuring genetic variation within species: an examination using genetic structure of bull trout, Salvelinus confluentus (Pisces: Salmonidae). Excoffier, L. , P. E. Smouse, and J. M . Quattro. 1992. Analysis of molecular variance inferred from metric distances among D N A haplotypes: application to human mitochondrial D N A restriction data. Genetics 131: 479-491. Ferguson, A . , J. B. Taggart, P. A . Prodohl, O. McMeel, C. Thompson, C. Stone, P. McGinnity, and R. A . Hynes. 1995. The application of molecular markers to the study and conservation offish populations, with special reference to Salmo. J. Fish Bio. 47: 103-126. Fleck, J. L. 1977. Summary of Dolly Varden tagging program Meadow Creek 1970-1976. Nelson, British Columbia Fish and Wildlife Branch. Fraley, J. J., and B. B. Shepard. 1989. Life history, ecology and population status of migratory bull trout (Salvelinus confluentus) in the Flathead Lake and river system, Montana. Northwest Science 63: 133-143. Frankham, R. 1995. Inbreeding and extinction: a threshold effect. Cons. Biol. 9: 792-799. Goetz, F. 1989. Biology of the bull trout Salvelinus confluentus a literature review. Eugene, Oregon, Willamette National Forest. Goodman, S. J. 1997. RST C A L C : a collection of computer programs for calculating estimates of genetic differentiation from microsatellite data and determining their significance. Mol . Ecol. 6: 881-885. Greenwood, J. J. D. , Ed. 1996. Ecological Census Techniques - a handbook. New York, Cambridge University Press. Haas, G. R. 1998. Indigenous fish species potentially at risk in B.C. with recommendations and prioritizations for conservation, forestry/resource use, inventory and research. Vancouver, B.C. Ministry of Environment, Lands & Parks. Haas, G. R., and J. D. McPhail. 1991. Systematics and distributions of Dolly Varden (Salvelinus malma) and bull trout (Salvelinus confluentus) in North America. Can. J. Fish. Aquat. Sci. 48: 2191-2211. 77 Hansen, M . M . , E. E. Nielsen, and K. -L . D. Mensberg. 1997. The problem of sampling families rather than populations: relatedness among individuals in samples of juvenile brown trout Salmo trutta L. Mol . Ecol. 6: 469-474. Hartl, D. L. , and A . G. Clark. 1989. Principles of Population Genetics. Sunderland, Mass., Sinauer Associates, Inc. Hedrick, P. W. 1999. Perspective: highly variable loci and their interpretation in evolution and conservation. Evolution 53: 313-318. Jarne, P., and P. J. L. Lagoda. 1996. Microsatellites, from molecules to populations and back. TREE 11: 424-429. Latham, S. J., and E. B. Taylor, in press. Hatchery-mediated reduction of genetic differentiation among bull trout populations: assessment of mechanisms and consequences using neutral genetic molecular markers. Bull trout II Conference Proceedings, Banff, A B , Trout Unlimited Canada. Leary, R. F., F. W. Allendorf, and S. H . Forbes. 1993. Conservation genetics of bull trout in the Columbia and Klamath river drainages. Cons. Biol. 7: 856-865. Lindsey, C. C , T. G. Northcote, and G. F. Hartman. 1959. Homing of rainbow trout to inlet and outlet spawing streams at Loon Lake, British Columbia. J. Fish. Res. Board Can. 16: 695-719. Marty, G. D., and R. C. Summerfelt. 1986. Pathways and mechanisms for expulsion of surgically implanted dummy transmitters from channel catfish. Trans. A m . Fish. Soc. 115: 577-589. McElhany, P., M . Ruckelshaus, M . J. Ford, T. Wainwright, and E. Bjorkstedt. 2000. Viable salmonid populations and the recovery of evolutionary significant units, U.S. Dept. Commer. (available on line) Mclsaac, D. O., and T. P. Quinn. 1988. Evidence for a hereditary component in homing behavior of chinook salmon (Oncorhynchus tshawytscha). Can. J. Fish. Aquat. Sci. 45: 2201-2205. McKinley, R. S., G. Power, and H . E. Kowalyk. 1992. Transmitter attachment/implant -laboratory manual. Waterloo, Ont, University of Waterloo, Biology Department. McLeod, C , and T. Clayton. 1993. Fish Radio Telemetry Demonstration Project, Upper Athabasca River May to August, 1992. Edmonton, Alberta, R .L .&L. Environmental Services Ltd. McLeod, T., and C. Clayton. 1994. Seasonal Movements of Radio Tagged Fish, Upper Athabasca River August, 1992 to March, 1993. Edmonton, Alberta, R .L .&L. Environmental Services Ltd. McPhail, J. D., and J. S. Baxter. 1996. A review of bull trout (Salvelinus confluentus) life-history and habitiat use in relation to compensation and improvement opportunities. Vancouver, British Columbia, Province of British Columbia Ministry of Environment, Lands and Parks, Fisheries Branch. 78 McPhail, J. D., and C. B. Murray. 1979. The early life-history and ecology of Dolly Varden (Salvelinus malma) in the Upper Arrow Lakes. Vancouver, B.C., Dept. of Zoology and Institute of Animal Resource Ecology, U.B.C. Miller, L. M . , and A . R. Kapuscinski. 1997. Historical analysis of genetic variation reveals low effective population size in a northern pike (Esox lucius) population. Genetics 147: 1249-1258. Miller, M . P. 1997. Tools for population genetic analyses (TFPGA) 1.3: a Windows program for the analysis of allozyme and molecular population genetic data., Computer software distributed by author. Mills , L. S., and F. W. Allendorf. 1996. The one-migrant-per-generation rule in conservation and management. Cons. Biol. 10: 1509-1518. Moritz, C. 1994. Defining 'Evolutionarily Significant Units' for conservation. TREE 9: 373-375. Morris, D. B., K. Richard, and J. M . Wright. 1996. Microsatellites from rainbow trout (Oncorhynchus mykiss) and their use for genetic study of salmonids. Can. J. Fish. Aquat. Sci. 53: 120-126. Nei, M . , and A . Chakravarti. 1977. Drift variances of FST and GST statistics obtained from a finite number of isolated populations. Theor. Pop. Biol. 11:307-325. Nelson, R. J., and T. D. Beacham. 1999. Isolation and cross species amplification of microsatellite loci useful for the study of Pacific salmon. Animal Genetics 30: 228-229. Nordeng, H . 1977. A pheromone hypothesis for the homeward migration of anadromous salmonids. OIKOS 28: 155 - 159. Northcote, T. G. 1997. Potadromy in salmonidae - living and moving in the fast lane. N . A m . J. Fish. Man. 17: 1029-1045. O'Brien, D. S., and M . W. Chamberlain. 2000. 1999 Adams Bull Trout Telemetry Project Progress Report. Kamloops BC, Ministry of the Environment, Lands & Parks. Oliver, G. G. 1979. A final report on the present fisheries use of the Wigwam River with an emphasis on the migration, life history and spawning behaviour of Dolly Varden char (Salvelinus malma) (Walbaum). Nelson, B.C., B.C. Fish and Wildlife Branch, Kootenay Region. Olsen, J. B., P. Bentzen, and J. E. Seeb. 1998. Characterization of seven microsatellite loci derived from pink salmon. Mol . Ecol. 7: 1083-1090. O'Reilly, P. T., L. Hamilton, S. McConnell, and J. M . Wright. 1996. Rapid analysis of genetic variation in Atlantic salmon (Salmo salar) by PCR multiplexing of dinucleotide and tetranucleotide microsatellites. Can. J. Fish. Aquat. Sci. 53: 2292-2298. 79 Palumbi, S. R., G. Grabowsky, T. Duda, L. Geyer, andN. Tachino. 1997. Speciation and population genetic structure in tropical pacific sea urchins. Evolution 51: 1506-1517. Parkinson, E., and G. R. Haas. 1996. The role of macrohabitat variables and temperature in defining the range of bull trout. Vancouver, B.C., B C Fisheries Branch -Research and Development Section. Peterson, G. R., and I. L. Withler. 1965. Effects on fish and game species of development of Duncan Dam for hydro-electric purposes, British Columbia Fish and Wildlife Branch. Quinn, T. P. 1993. A review of homing and straying of wild and hatchery-produced salmon. Fisheries Res. 18: 29-44. Quinn, T. P., and A. H . Dittman. 1990. Pacific salmon migrations and homing: mechanisms and adaptive significance. TREE 5: 174-177. Quinn, T. P., C. C. Wood, L. Margolis, B. E. Riddell, and H . K . D. 1987. Homing in wild sockeye salmon (Oncorhynchus nerka) populations as inferred from differences in parasite prevalance and allozyme allele frequencies. Can. J. Fish. Aquat. Sci. 44: 1963-1971. Raymond, M . , and F. Rousset. 1995. GENEPOP (1.2): Population genetics software for exact tests and ecumenism. J. Heredity 86: 248 - 249. Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43: 223-225. Rieman, B. E., and J. D. Mclntyre. 1993. Demographic and habitat requirements for conservation of bull trout. Ogden, Utah, United States Department of Agriculture: Forest Service, Intermountain Research Station. Ryman, N . 1983. Patterns of distribution of biochemical genetic variation in salmonids: differences between species. Aquaculture 33: 1-21. Scheer, B. T. 1939. Homing instincts in salmon. Quarterly Rev. Biol. 14: 408-420. Schill, D., R. Thurow, and P. Kline. 1994. Seasonal movement and spawning mortality of fluvial bull trout in Rapid River, Idaho, Idaho Department of Fish & Game. Schneider, S., D. Roessli, and L. Excoffier. 2000. A R L E Q U I N version 2: a software for population genetic analysis., Genetics and Biometry Laboratory, University of Geneva, Switzerland. Shaklee, J. B., and N . V . Varnavskaya. 1994. Electrophorectic characterization of odd-year pink salmon (Oncorhynchus gorbuscha) populations from the Pacific coast of Russia, and comparisons with selected North American populations. Can. J. Fish. Aquat. Sci. 51: 158-171. Slatkin, M . 1987. Gene flow and the geographic structure of natural populations. Science 236:787-792. Slatkin, M . 1993. Isolation by distance in equilibrium and non-equilibrium populations. Evolution 47: 264-279. 80 Slettan, A . , I. Olsaker, and O. Lie. 1995. Atlantic salmon, Salmo salar, microsatellites at the SSOSL25, SSOSL85, SSOSL311, SSOSL417 loci. Animal Genetics 26: 281-282. Spruell, P., and F. W. Allendorf. 1997. Nuclear D N A analysis of Oregon bull trout final report. Missoula, MT, University of Montana. Spruell, P., B. E. Reiman, K. L. Knudsen, F. M . Utter, and F. W. Allendorf. 1999. Genetic population structure within streams: microsatellite analysis of bull trout populations. Ecol. Freshwater Fish 8: 114-121. Stasko, A . B., and D. G. Pincock. 1977. Review of underwater biotelemetry, with emphasis on ultrasonic techniques. J. Fish. Res. Board Can. 34: 1261-1285. Swanberg, T. 1996. The movement and habitat use of fluvial bull trout in the Upper Clark Fork River drainage. MSc. Thesis, Department of Biological Sciences, University of Montana. Swanberg, T. R. 1997. Movements of and habitat use by fluvial bull trout in the Blackfoot River, Montana. Trans. Am. Fish. Soc. 126: 735-746. Taft, A . C , and L. Shapovalov. 1938. Homing instinct and straying among steelhead trout (Salmo gairdnerii) and silver salmon (Oncorhynchus kisutch). Calif. Fish Game 24: 118-125. Taggart, J. B., P. A . Hynes, P. A . Prodohl, and A. Ferguson. 1992. A simplified protocol for routine total D N A isolation from salmonid fishes. J. Fish Bio. 40: 963-965. Tallman, R. F., and M . C. Healey. 1994. Homing, straying, and gene flow among seasonally separated populations of chum salmon (Oncorhynchus keta). Can. J. Fish. Aquat. Sci. 51: 577-588. Taylor, E. B. 1991. A review of local adaptation in Salmonidae, with particular reference to Pacific and Atlantic salmon. Aquaculture 98: 185-207. Taylor, E. B., C. J. Foote, and C- C. Wood. 1996. Molecular genetic evidence for parallel life-history evolution within a Pacific salmon (sockeye salmon and kokanee, Oncorhynchus nerka). Evolution 50: 401-416. Taylor, E. B., and J. D. McPhail. 2000. Historical contingency and ecological determinism interact to prime speciation in sticklebacks, Gasterosteus. Pro. R. Soc. Lond. B267: 1-10. Taylor, E. B., S. Pollard, and D. Louie. 1999. Mitochondrial D N A variation in bull trout (Salvelinus confluentus) from northwestern North America: implications for zoogeography and conservation. Mol . Ecol. 8: 1155-1170. Thorstad, E. B., F. Okland, and B. Finstad. 2000. Effects of telemetry transmitters on the swimming performace of adult Atlantic salmon. J. Fish Bio. 57: 531 - 535. Utter, F., and N . Ryman. 1993. Genetic markers and mixed stock fisheries. Fisheries 18: 11-20. 81 Utter, F. M . , R. S. Waples, and D. J. Teel. 1992. Genetic isolation of previously indistinguishable chinook salmon populations of the Snake and Klamath Rivers: limitations of negative data. Fish. Bull. 90: 770-777. Vogler, A . P., and R. DeSalle. 1994. Diagnosing units of conservation and management. Cons. Biol. 8: 354-363. Waples, R. S. 1991. Pacific salmon, Oncorhynchus spp., and the definition of "species" under the Endangered Species Act. Mar. Fish. Rev. 53: 11-22. Waples, R. S. 1998. Separating the wheat from the chaff: patterns of genetic differentiation in high gene flow species. J. Heredity 89: 438-450. Waples, R. S., and D. J. Teel. 1990. Conservation genetics of pacific salmon 1. Temporal changes in allele frequency. Cons. Biol. 4: 144-156. Ward, R. D., M . Woodwark, and D. O. F. Skibinski. 1994. A comparison of genetic diversity levels in marine, freshwater, and anadromous fishes. J. Fish Biol. 44: 213-232. Watson, G., and T. W. Hillman. 1997. Factors affecting the distribution and abundance of bull trout: an investigation at hierarchical scales. N . A m . J. Fish. Man. 17: 237-252. Weir, B. S., and C. C. Cockerham. 1984. Estimating F-statistics for the analysis of population structure. Evolution 38: 1358-1370. Whitlock, M . C. 1992. Temporal fluctuations in demographic parameters and the genetic variance among populations. Evolution 46: 608-615. Whitlock, M . C , and D. E. McCauley. 1999. Indirect measures of gene flow and migration: Fst * l/(4Nm +1). Heredity 82: 117-125. Williams, R. N . , R. P. Evans, and D. K . Shiozawa. 1997. Mitochondrial D N A diversity in bull trout from the Columbia Basin. Friends of the bull trout, Calgary, Alberta, Bull Trout Task Force - Trout Unlimited Canada: 283-297. Winship, P. R. 1989. A n improved method for directly sequencing PCR amplified material using dimethyl sulphoxide. Nucleic Acids Res. 17: 1266. Winter, J. D. 1983. Underwater Biotelemetry. in: Fisheries Techniques. Eds. L. A . Nielsen and D. L. Johnson. Bathesda, Maryland, American Fisheries Society: 371-395. Wright, S. 1931. Evolution in Mendelian populations. Genetics 16: 97-159. Wright, S. 1943. Isolation by distance. Genetics 28: 114-138. 82 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0089986/manifest

Comment

Related Items