@prefix vivo: . @prefix edm: . @prefix ns0: . @prefix dcterms: . @prefix skos: . vivo:departmentOrSchool "Forestry, Faculty of"@en ; edm:dataProvider "DSpace"@en ; ns0:degreeCampus "UBCV"@en ; dcterms:creator "Middleton, Collin Thomas"@en ; dcterms:issued "2016-07-04T15:44:21Z"@*, "2016"@en ; vivo:relatedDegree "Master of Science - MSc"@en ; ns0:degreeGrantor "University of British Columbia"@en ; dcterms:description "Upriver migrating adult Pacific salmon home to natal sites following natal water cues while also undergoing a suite of physiological changes to prepare for spawning. Migrants can encounter myriad environmental conditions that are physiologically and energetically challenging throughout these journeys. Many freshwater migration corridors have also been converted into hydroelectric systems (hydrosystems) that can change the composition of flows such that the availability and concentration of natal water cues can vary substantially. How such flow composition changes affect migration behaviour has rarely been examined, while the cumulative effects of environmental and physiological factors on the fate of migrating adult salmon in regulated rivers are not well understood. Using the Seton hydrosystem in British Columbia and two populations of sockeye salmon as a model, I conducted a field radio-tagging study that examined how physiology, temperature, and natal water concentration affected the behaviour and fate of adult salmon migrating through a regulated river while enroute to natal spawning sites. Most tagged migrants (89%) delayed in the outlet of the Seton powerhouse that discharges strong concentrations of natal water, and subsequently wandered in the mainstem Fraser River before continuing their upstream migration into the Seton River, where natal water cues can also vary. I found few associations between physiological stress and reproductive hormone levels with powerhouse delays and wandering, although fish with higher energy content were generally slower migrating through the whole hydrosystem. Higher temperatures and elevated natal water concentrations were associated with shorter delays at the powerhouse and less wandering, but only among late-run migrants. I found little evidence that the cumulative effects of physiology and environmental conditions during hydrosystem passage were related to survival to natal sites, suggesting that other factors aside from those encountered during hydrosystem passage (e.g. environmental factors prior to reaching the hydrosystem) may have played a role in influencing survival post dam-passage. My thesis provides the first detailed account of how varying natal water concentrations affects the homing behaviour of wild migrating adult salmon and how the cumulative effects of physiology and environmental conditions experienced during passage through a regulated migration corridor influences survival to natal sites."@en ; edm:aggregatedCHO "https://circle.library.ubc.ca/rest/handle/2429/58376?expand=metadata"@en ; skos:note """THE CUMULATIVE EFFECTS OF PHYSIOLOGY, TEMPERATURE, AND NATAL WATER CUES ON THE MIGRATION BEHAVIOUR AND SURVIVAL OF ADULT SOCKEYE SALMON DURING PASSAGE THROUGH THE SETON RIVER HYDROELECTRIC SYSTEM, BRITISH COLUMBIA by COLLIN THOMAS MIDDLETON B.Sc. (Honours), The University of British Columbia, 2012 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Forestry) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) June 2016 © Collin Thomas Middleton, 2016 ii Abstract Upriver migrating adult Pacific salmon home to natal sites following natal water cues while also undergoing a suite of physiological changes to prepare for spawning. Migrants can encounter myriad environmental conditions that are physiologically and energetically challenging throughout these journeys. Many freshwater migration corridors have also been converted into hydroelectric systems (hydrosystems) that can change the composition of flows such that the availability and concentration of natal water cues can vary substantially. How such flow composition changes affect migration behaviour has rarely been examined, while the cumulative effects of environmental and physiological factors on the fate of migrating adult salmon in regulated rivers are not well understood. Using the Seton hydrosystem in British Columbia and two populations of sockeye salmon as a model, I conducted a field radio-tagging study that examined how physiology, temperature, and natal water concentration affected the behaviour and fate of adult salmon migrating through a regulated river while enroute to natal spawning sites. Most tagged migrants (89%) delayed in the outlet of the Seton powerhouse that discharges strong concentrations of natal water, and subsequently wandered in the mainstem Fraser River before continuing their upstream migration into the Seton River, where natal water cues can also vary. I found few associations between physiological stress and reproductive hormone levels with powerhouse delays and wandering, although fish with higher energy content were generally slower migrating through the whole hydrosystem. Higher temperatures and elevated natal water concentrations were associated with shorter delays at the powerhouse and less wandering, but only among late-run migrants. I found little evidence that the cumulative effects of physiology and environmental conditions during hydrosystem passage were related to survival to natal sites, suggesting that other factors aside from those encountered during hydrosystem passage (e.g. environmental factors prior to reaching the hydrosystem) may have played a role in influencing survival post dam-passage. My thesis provides the first detailed account of how varying natal water concentrations affects the homing behaviour of wild migrating adult salmon and how the cumulative effects of physiology and environmental conditions experienced during passage through a regulated migration corridor influences survival to natal sites. iii Preface This research was carried out as one component of a large monitoring / research program commissioned by BC Hydro and the St’át’imc First Nation as part of the Bridge River Water Use Plan entitled: Effectiveness of Cayoosh flow dilution, dam operation, and fishway passage on delay and survival of upstream migration of salmon in the Seton-Anderson watershed. I held primary responsibility for study designs, collection and analysis of data, and the preparation of manuscripts for submission. Throughout this process I received considerable logistical support from my colleagues and St’át’imc eco-resources technicians, and guidance from my supervisor Dr. Scott Hinch and supervisory committee member Dr. Eduardo Martins. Collaborators on individual projects who were instrumental in development, fieldwork, data analyses, or manuscript preparation are listed as coauthors on manuscripts that will be submitted for publication. All study procedures were conducted in accordance with the guidelines of the Canadian Council of Animal Care administered by the University of British Columbia (A11-0125). iv Chapter 2: Effects of natal water concentration and temperature on migration behaviours of up-river migrating adult sockeye salmon Authors: Collin T Middleton, Scott G Hinch, Eduardo G Martins, Douglas C Braun, Nicholas J Burnett, Vanessa Minke-Martin, and Matthew T Casselman Acceptance: To be submitted July 2016 Comments: This study was conducted and written by CTM with assistance from EGM, DCB, NJB, VMM, and MTC, under the supervision and guidance of SGH who helped to conceptualize the study and with preparation of the manuscript Chapter 3: Migration behaviour and the cumulative effects of hydrosystem passage in relation to natal water cues, temperature, and individual characteristics of wild up-river migrating sockeye salmon Authors: Collin T Middleton, Scott G Hinch, Eduardo G Martins, Arthur L. Bass, Douglas C Braun, Nicholas J Burnett, Vanessa Minke-Martin, and Matthew T Casselman Acceptance: To be submitted July 2016 Comments: This study was conducted and written by CTM with assistance from EGM, ALB, DCB, NJB, VMM, and MTC, under the supervision and guidance of SGH who helped to conceptualize the study and with preparation of the manuscript. v Table of Contents Abstract ................................................................................................................................................... ii Preface ....................................................................................................................................................iii Table of Contents ................................................................................................................................. v List of Tables ...................................................................................................................................... vii List of Figures ....................................................................................................................................... x Acknowledgments ........................................................................................................................... xvi Chapter 1: Introduction ...................................................................................................................... 1 1.1 General introduction and background ................................................................................ 1 1.2 Study system and species ....................................................................................................... 6 1.3 Thesis overview and objectives ........................................................................................... 7 Chapter 2: Effects of natal water concentration and temperature on migration behaviours of up-river migrating adult sockeye salmon ........................................................... 9 2.1 Introduction ................................................................................................................................ 9 2.2 Methods ................................................................................................................................... 12 2.3 Results ...................................................................................................................................... 20 2.4 Discussion ............................................................................................................................... 24 2.5 Chapter 2 Tables .................................................................................................................... 31 2.6 Chapter 2 Figures .................................................................................................................. 39 Chapter 3: Migration behaviour and the cumulative effects of hydrosystem passage in relation to natal water cues, temperature, and individual characteristics of wild up-river migrating sockeye salmon .............................................................................................................. 44 3.1 Introduction ............................................................................................................................. 44 3.2 Methods ................................................................................................................................... 47 3.3 Results ...................................................................................................................................... 57 3.4 Discussion ............................................................................................................................... 62 3.5 Chapter 3 Tables .................................................................................................................... 72 3.6 Chapter 3 Figures .................................................................................................................. 79 Chapter 4: Summary and conclusions ......................................................................................... 85 vi References ........................................................................................................................................... 91 Appendix 1 ........................................................................................................................................ 104 Appendix 2 ........................................................................................................................................ 128 vii List of Tables Table 2.1 – AICc model selection statistics for generalized linear models and linear models predicting (1) the number of forays made into the Seton powerhouse, (2) the amount of wandering in the Fraser River between the release site and the Seton-Fraser confluence, and (3) the total amount of migration delay incurred by individuals in the Seton powerhouse tailrace (3) for Gates Creek and Portage Creek sockeye salmon. Abbreviations used for model variables include: (FRT) Fraser River temperature, (PHT) Seton powerhouse temperature, (Tdiff) temperature differential between the Fraser River and Seton powerhouse, (NW) natal water concentration, and (TD) tagging date. In the model 3 for the Portage Creek stock, TD is shown with FRT in parentheses to indicate that tagging date was substituted for Fraser River temperature. Note, ΔAICc is the difference in AICc values between model i and the top model in the candidate set. Models are ranked from lowest to highest ΔAICc, and by wi – the probability that a given model is the best in the 95% confidence set. Adj R2 is an estimate of the proportion of variance explained by each model, adjusted by the number of explanatory variables; this is only shown for linear models as generalized linear model fits were evaluated by Chi-square tests (see methods; Chapter 2)………. 31 Table 2.2 – Mean ± SD and range of fork length (cm), glucose (mmol L-1), lactate (mmol L-1), and testosterone (ng ml-1) for female (♀) and male (♂) Gates Creek and Portage Creek sockeye salmon by year. Note, preliminary analyses of physiological variables from the three datasets used in behavioural models 1, 2, and 3 yielded similar values; this table reports the data for stress and maturity indices used in model 3 as an example. Values for fork length, glucose, and lactate are pooled because preliminary analyses indicated there were not differences between sexes and populations. * Indicates significant difference at P < 0.05 in plasma testosterone between sexes and populations as assessed by two-way ANOVA…………………………………………………………………… 38 viii Table 3.1 – Summary statistics for reach-specific and total lower Seton hydrosystem migration times in hours for Gates Creek and Portage Creek sockeye salmon tracked in this study. Migration times are calculated based only on individuals that achieved the event of interest (i.e. entered the Seton River, arrived at the Seton Dam, passed the Seton Dam) and are presented by sex and fate; successful individuals were fish that eventually reached natal spawning sites; unsuccessful individuals were those not observed to reach natal sites……………………….. 72 Table 3.2 – AICC model selection statistics for models with Δ AICC < 2 for Gates Creek and Portage Creek sockeye salmon time-to-event proportional hazards regression predicting hydrosystem migration times from (1) release to Seton River entry, (2) Seton River entry to arrival at the Seton Dam, and (3) passage time of the Seton Dam. Models with ΔAICC < 2 are also given for generalized linear models for both populations predicting (4) survival to natal spawning sites following passage of the Seton Dam. Abbreviations used for model covariates include: (GSE) gross somatic energy, (FRT) Fraser River temperature, (SRT) Seton River temperature, (NW) natal water concentration, (DTT) Seton Dam tailrace temperature, (Ddis) Seton Dam discharge, (Flow) flow scenario, (TMT) total migration time, and (DD) degree days. ΔAICC is the difference in AICC values between model i and the top model in the candidate set. Models are ranked from lowest to highest ΔAICC, and by wi – the probability that a given model is the best in the 95% confidence set…………………………………………………………………… 76 Table A1.1 – Location of radio receivers and combined percent time in operation during the 2013 and 2014 study years………………………………………………… 104 Table A1.2 – Fate of all female (♀) and male (♂) Gates Creek and Portage Creek sockeye salmon used in models predicting (1) the number of forays into the Seton powerhouse, (2) wandering in the Fraser River between the release site and Seton-Fraser confluence, and (3) the total amount of migration delay incurred by individuals in the Seton powerhouse tailrace. Fates of individuals are given as a percentage of the total number of fish used in each model and by sample size…………………………………………………………………………….. 105 ix Table A1.3 – AICc model selection statistics for the 95% confidence set of models for generalized linear models predicting (1) the number of forays into the Seton Powerhouse, (2) the amount of wandering in the Fraser River between the release site and the Seton-Fraser confluence, and linear models predicting (3) the total amount of migration delay incurred by individuals in the Seton Powerhouse tailrace for Gates Creek and Portage Creek sockeye salmon. Abbreviations used for model predictor variables include: (Gluc) Glucose, (Lact) Lactate, (Test) Testosterone, (FRT) Fraser River Temperature, (PHT) Powerhouse Temperature, (Tdiff) Temperature differential between the Fraser River and Seton Powerhouse tailrace, (NW) Natal Water concentration, and (TD) Tagging Date. Generalized linear model fits were evaluated by Chi-square tests (see methods); Adj R2 is an estimate of the proportion of variance explained by each model, adjusted by the number of explanatory variables and is only shown for linear models of powerhouse delay. ΔAICc is the difference in AICc values between model i and the top model in the candidate set. Models are ranked from lowest to highest ΔAICc, and by wi – the probability that a given model is the best in the 95% confidence set………………………………………………………………….. 106 Table A2.1 - Model-averaged results for sensitivity analyses testing ‘original’, ‘lower odds’, ‘upper odds’, and ‘event-only’ models to assess the effects of informative censoring on time-to-event regression analysis. Model averaged odds ratios with 95% confidence intervals that do not include zero are shown in bold………… 132 x List of Figures Figure 2.1 – Study area, located in the Seton-Anderson watershed in southwestern British Columbia, Canada (inset). Sockeye salmon (Oncorhynchus nerka) capture, release, and radio-telemetry sites, along with temperature and conductivity monitoring locations are indicated by the legend and map. The location of the Seton powerhouse tailrace and the visible extent of the powerhouse discharge plume of natal Seton Lake water is shown extending into the Fraser River, as is the extent of the Seton River plume that fluctuates in its concentration of natal water given relative contributions of Cayoosh Creek discharge……………….. 39 Figure 2.2 – Mean daily Fraser River (solid lines) and Seton powerhouse (dashed lines) temperatures (panel a), and natal water concentration of the Seton River and its plume (panel b). Measurements for the 2013 study are given by grey lines and for the 2014 study by black lines. Shaded grey boxes represent the periods in which fish from each population were tagged. Solid red lines represent the current natal water concentration targets of 80% and 90% during the Gates Creek and Portage Creek sockeye salmon migrations, respectively, and are approximately proportional to the migration timing of each population through the hydrosystem…………………………………………………………………….. 40 Figure 2.3 – Panels a & b – Histograms of the number of forays female (grey bars) and male (black bars) Gates Creek (a) and Portage Creek (b) sockeye salmon made into the Seton powerhouse as a proportion of all the individuals included in models predicting this behaviour. In panel b, 1 female that made 24 forays was removed from the histogram for clarity. Panels c & d – Model-averaged standardized coefficient estimates for models predicting the number of forays Gates Creek (c) and Portage Creek (d) sockeye salmon made into the Seton powerhouse. Coefficient estimates with 95% confidence intervals that do not cross zero are highlighted by solid black circles. Vertical dashed lines indicates the coefficient value of zero. Abbreviations for predictor variables include: (FRT) Fraser River temperature, (PHT) Seton powerhouse temperature, (Tdiff) temperature differential between the Fraser River and Seton powerhouse, and xi (NW) natal water concentration. FRT is shown with TD in parentheses to indicate that tagging date was substituted for Fraser River temperature in the Portage Creek foray model (panel d only). Note the differences in x-axes scales between panels…………………………………………………………………………… 41 Figure 2.4 – Panels a & b – Histograms of wandering (the number of back-and-forth movements) in the Fraser River between the release site and the Seton-Fraser confluence for female (grey bars) and male (black bars) Gates Creek (a) and Portage Creek (b) sockeye salmon shown as a proportion of all individuals included in the models predicting this behaviour for each population. In panel a, 1 female that wandered 9 times was removed from the histogram for clarity. Panels c & d – Model-averaged standardized coefficient estimates for models predicting wandering for Gates Creek (c) and Portage Creek (d) sockeye salmon. Coefficient estimates with 95% confidence intervals that do not include zero are highlighted by solid black circles. Vertical dashed line indicates the coefficient value of zero. Abbreviations for predictor variables include: (FRT) Fraser River temperature, (PHT) Seton powerhouse temperature, (Tdiff) temperature differential between the Fraser River and Seton powerhouse, and (NW) natal water concentration. Note the differences in x-axes scales between panels………………………….. 42 Figure 2.5 – Panel a – Beanplots of total migration delay in the Seton powerhouse tailrace for female (black beans) and male (grey beans) Gates Creek and Portage Creek sockeye salmon used in models predicting this behaviour. Shaded polygons represent the distribution of individual delay times (small horizontal lines) and bold horizontal lines represent means. Panel b & c – Model-averaged standardized coefficient estimates from models predicting the total amount of migration delay incurred by Gates Creek (b) and Portage Creek (c) sockeye salmon in the Seton powerhouse tailrace. Coefficient estimates with 95% confidence intervals that do not include zero are highlighted by solid black circles. Vertical dashed line indicates the coefficient value of zero. Abbreviations for model variables include: (FRT) Fraser River temperature, (PHT) Seton powerhouse temperature, (Tdiff) temperature differential between the Fraser xii River and Seton powerhouse, and (NW) natal water concentration. FRT is shown with TD in parentheses to indicate that tagging date was substituted for Fraser River Temperature in the Portage Creek delay model (panel c only). Note the differences in x-axes scales between panels……………………………………. 43 Figure 3.1 – The Seton-Anderson watershed and locations of Gates Creek and Portage Creek sockeye salmon (O. nerka) spawning grounds (large inset), and the Seton hydrosystem (red dashed rectangle in large inset and main map) in southwestern British Columbia, Canada (small inset). The three study reaches of the hydrosystem are indicated by grey shaded polygons on the main map. Fish capture / release sites, radio-telemetry sites, temperature monitoring sites, and dam locations are indicated by the legend and map. The location of the Seton powerhouse tailrace and extent of its discharge plume of natal Seton Lake water is shown extending into the Fraser River, as is the extent of the Seton River plume that fluctuates in its concentration of natal water………………………………. 79 Figure 3.2 – Mean daily temperature (a), natal water concentration of the lower Seton River (b), and discharge from the Seton Dam (c) during the 2013 (grey lines) and 2014 (black lines) Gates Creek (~ 1 August – 5 September) and Portage Creek (~ 15 September – 31 October) sockeye salmon migrations through the lower Seton hydrosystem. Temperatures in panel ‘a’ are given for the Fraser River at Texas Creek (solid lines), the Seton River (dashed lines), and the Seton Dam tailrace (dotted lines)……………………………………………………………………. 80 Figure 3.3 - Panel a & b – Passage time curves through reach 1 of the Seton hydrosystem from release in the Fraser River to Seton River entry for Gates Creek (black lines) and Portage Creek (grey lines) female (a) and male (b) sockeye salmon by fate. Solid lines indicate eventually successful migrants (survived to natal spawning grounds); broken lines indicate unsuccessful migrants (did not survive to natal spawning grounds); open circles indicate censored individuals. Panel c & d – Model-averaged estimates of odds ratios for standardized covariates in time-to-event proportional hazards regression for (c) Gates Creek and (d) Portage Creek passage times through the first reach of the Seton xiii hydrosystem. Error bars represent 95% confidence intervals for odds ratios; closed black circles highlight odds ratios with confidence intervals that do not include zero. Vertical dashed lines indicate the coefficient value of zero. Abbreviations for covariates are: (GSE) gross somatic energy; (FRT) Fraser River temperature; (SRT) Seton River temperature; (NW) natal water concentration. Precise values for odds ratios and 95% CIs are given by covariate in Appendix 2…………………………………………………………………………………. 81 Figure 3.4 – Panel a & b – Passage time curves through reach 2 of the Seton hydrosystem from Seton River entry to arrival at the Seton Dam for Gates Creek (black lines) and Portage Creek (grey lines) female (a) and male (b) sockeye salmon by fate. Solid lines indicate eventually successful migrants (survived to natal spawning grounds); broken lines indicate unsuccessful migrants (did not survive to natal spawning grounds); open circles indicate censored individuals. Panel c & d – Model-averaged estimates of odds ratios for standardized covariates in time-to-event proportional hazards regression for Gates Creek (c) and Portage Creek (d) passage times through the second reach of the Seton hydrosystem. Error bars represent 95% confidence intervals for odds ratios; closed black circles highlight odds ratios with confidence intervals that do not include zero. Vertical dashed lines indicate the coefficient value of zero. Abbreviations for coefficients are: (GSE) gross somatic energy; (SRT) Seton River temperature; (NW) natal water concentration. Precise values for odds ratios and 95% CIs are given by covariate in Appendix 2…………………………… 82 Figure 3.5 – Panel a & b – Seton Dam passage time curves for Gates Creek (black lines) and Portage Creek (grey lines) female (a) and male (b) sockeye salmon by fate. Solid lines indicate eventually successful migrants (survived to natal spawning grounds); broken lines indicate unsuccessful migrants (did not survive to natal spawning grounds); open circles indicate censored individuals. Panel c & d – Model-averaged estimates of odds ratios for standardized covariates in time-to-event proportional hazards regression for Seton Dam passage time of Gates Creek (c) and Portage Creek (d) sockeye salmon. Error bars represent 95% confidence xiv intervals for odds ratios; closed black circles highlight odds ratios with confidence intervals that do not include zero. Vertical dashed lines indicate the coefficient value of zero. Abbreviations for coefficients are: (GSE) gross somatic energy; (Dtemp) Seton Dam tailrace temperature; (Ddis) Seton Dam discharge; (Flow) binary covariate for experimental flow scenario change (Gates migration only). Precise values for odds ratios and 95% CIs are given by covariate in Appendix 2……………………………………………………………………………….… 83 Figure 3.6 – Panel a – Beanplot of cumulative hydrosystem migration times from release in the Fraser River to upstream of the Seton Dam for Gates Creek and Portage Creek male and female sockeye salmon that were ultimately successful (survived to natal spawning grounds; black beans) and unsuccessful (did not survive to natal spawning grounds; grey beans) following dam passage. Beans represent distribution of individual passage time times (small horizontal lines); large black horizontal lines represent means. Panel b & c – Model-averaged standardized coefficient estimates from generalized linear models predicting survival to natal spawning sites for all Gates Creek (b) and Portage Creek (c) sockeye salmon that passed the Seton Dam. Error bars represent 95% confidence intervals for coefficient estimates; closed black circles highlight coefficients with confidence intervals that do not include zero. Vertical dashed lines indicate the coefficient value of zero. Abbreviations for coefficients are: (GSE) gross somatic energy; (TD) tagging date as an index of run timing; (DD) degree-day accumulation from release in the Fraser River to Seton Dam passage; (TMT) total migration time from release to dam passage. (Flow) binary covariate for experimental flow scenario change experienced at the time of dam passage (Gates migration only). Precise values for coefficients and 95% CIs are given by covariate in Appendix 2…………………………………………………………………………………. 84 Figure A1.1 – Water conductivity measured at sites in Cayoosh Creek (CC; n = 127), the upper Seton River below the Seton Dam (USR; n = 127), the Lower Seton River below its confluence with Cayoosh Creek (LSR; n = 125), and in the Seton powerhouse tailrace (PhTr; n = 127) used to distinguish between the olfactory xv signatures of each water source. Data from the 2013 and 2014 study years were pooled. Solid lines inside boxes represent median values; upper and lower whiskers represent sample minimum and maximums. Lowercase letters indicate significant differences in conductivity between sites determined by a Kruskal-Wallis test at α < 0.05………………………………………………………… 130 xvi Acknowledgments To begin, I’d like to thank my supervisor Scott Hinch for giving me the opportunity to pursue something I really do love. Remember way back in the 2nd year of my undergrad when I asked you for a job and you rolled your eyes and told me to come back in a few years? Well, I never thought I’d be in the position I am today, and your continued support and guidance have made this all possible. And to Natalie Sopinka – if it weren’t for you believing that I was the best undergrad to pick and weight dead eggs in the wetlab, and then write a thesis about it, this entire journey likely would have never begun. So to Scott and Natalie, I am forever grateful for you two believing that a guy with a normal GPA and a fondness for salmonids could achieve what I have today, thank you! To my mom and dad and brother. Thank you for endless support, encouragement, positivity, and for believing in me throughout the course of my (many) years in school. If it weren’t for you guys I would not be the man (or scientist? weird.) I am today. And Cassandra – I’m not sure I would have made it through this process without your seemingly never-ending love, support, and patience. I can’t imagine having done it without you, thank you. Eduardo Martins and Doug Braun, thank you for your quantitative wizardry and for teaching me that R is awesome and that stats are actually super cool and exciting! As a guy who didn’t even finish high school math, I would have never believed I would write these words. Also thank you to Kathy Martin and David Patterson for their input as committee members and for helping steer the direction of my research by providing valuable feedback on my work. Of course, none of this would have been possible without the help of my friends (I guess some would call you colleagues or lab mates, but you guys are my friends first). In no particular order, I owe huge thank you’s to Vanessa Minke-Martin, Nolan Bett, Nich Burnett, Matt Casselman, Carson White, Matt Drenner, Art Bass, Nathan Furey, Katrina Cook, Steve Healy, the entire DFO Environmental Watch team, and Andrew Lotto for helping with field work, analyses, lab work, being good housemates, and all-round good company. xvii I’d also like to thank all members of the St’át’imc First Nation and BC Hydro in Lillooet, and the N’Quatqua First Nation in D’Arcy. Your willingness and enthusiasm for my research and the entire BRGMON-14 project is greatly appreciated and a strong showing of your commitment to the salmon of the Seton-Anderson watershed. In particular I would like to thank Wesley Payne, Bonnie Dunn, Roxx Ledoux, Jessica Hopkins, the O’Donahey’s, and Chris Fletcher for your dedicated support in the field and for making me feel welcome in your respective communities. Financial support for my research came from a NSERC CGS-M scholarship, MITACS, St’át’imc Government Services, and BC Hydro. My apologies for the long-windedness of these acknowledgements, but anyone who has ever reviewed any of my writing will tell you, this is pretty much par for the course. 1 Chapter 1: Introduction 1.1 General introduction and background Migrations are characterized by the predictable and directed movements of organisms over often long distances, with sensory responses to external and internal cues directing shifts between habitat types for the purposes of foraging, reproduction, and/or survival (Dingle 2014). Although migrations occur across taxa, one of the most well-known and well-studied examples of this phenomenon are those of anadromous fishes – particularly semelparous Pacific salmon (Oncorhynchus spp.). Pacific salmon are hatched in freshwater streams or lakes, then migrate to the ocean where they feed and grow in more productive marine waters (Gross et al. 1988). After ~ 1 to 3 years in the ocean, increasing levels of reproductive hormones initiate return migrations and adult salmon utilize a series of sensory cues to ‘home’ toward natal freshwater streams where they will spawn and eventually die (Hinch et al. 2006). Such mass migrations to natal freshwater sites help increase the chances that migrants will find mates and habitat suitable for spawning and egg survival (Quinn 2005), and also ensures the evolution of unique, locally adapted populations (Keefer and Caudill 2014). The implications of the semelparous Pacific salmon life history are such that an entire lifetime of reproductive success depends on the successful completion of these migrations. Given that some adult Pacific salmon migrate thousands of kilometers from the ocean to natal freshwater sites (Dittman and Quinn 1996), these journeys are inherently complex, energetically costly, and often risky (Dingle 2014). Pacific salmon are perhaps most famous for the remarkable fidelity and precision of adult migrations that characterize certain species (e.g. sockeye salmon O. nerka), which is made possible by the detection and perception of sensory cues that guide homing (Dittman and Quinn 1996). Beginning at sea, adults navigate toward coastal waters by following a combination of celestial cues, polarized light, magnetic fields, and a bi-coordinate map-compass system (1964; Quinn and Groot 1984; Hansen and Jonsson 1994; Ueda 2011; Putman et al. 2014). In the near shore environment, navigation may be guided by many of these same mechanisms, although visual and olfactory cues, currents, salinity, temperature, and riverine inputs are likely more important to homing directly 2 prior to entry into freshwater (Ueda 2011; Doving and Stabell 2003; Drenner et al. 2015). Upon entering freshwater, adult salmon rely primarily on olfaction to detect directional cues and guide homing (Hasler and Wisby 1951). As juveniles, salmon imprint on the unique chemical signatures or ‘odour bouquets’ of their natal streams; returning as adults, migrants detect varying concentrations of these cues and use them as freshwater ‘signposts’ to guide homing back to natal sites (Hasler and Scholz 1983). In large migration corridors, the availability of homestream olfactory cues can vary substantially with proximity to natal tributaries; to reconcile this issue, adult salmon likely also follow pheromones from co-migrating conspecifics up to a point where natal cues are strong enough to provide precise directional cues that lead to natal tributaries (Nordeng 1977; Bett and Hinch 2015). Olfactory-guided homing is of course not the only component of adult salmon migrations that dictate success, as individual fish traits and other environmental features also play a large role. Indeed, there are many physiological and energetic demands associated with migrating such long distances, which in large part determine the behaviour and fate of migrants before reaching natal spawning grounds (Quinn 2005; Hinch et al. 2006). In addition, adults also cease feeding prior to entering freshwater, thus all upriver migration, gonad development, and spawning is fueled solely by energy reserves acquired while at sea (Brett 1995). Upon entry into freshwater, adult salmon undergo a suite of physiological changes that are characterized by increases in stress and reproductive hormones to prepare for spawning (Quinn 2005; Hinch et al. 2006). Many fish also gain significant amounts of elevation (Eliason et al. 2011) and traverse hydraulically challenging flows during these migrations (Rand et al. 2006). A number of recent studies have used adult sockeye salmon as a model from which to examine how these physiological and energetic demands underlie migration behaviour and survival (reviewed in Hinch et al. 2006). For instance, sockeye salmon with more advanced reproductive status and lower somatic energy reserves are known to migrate faster upriver than relatively less mature fish with more energy (Crossin et al. 2008; Hanson et al. 2008). Fish that die enroute to spawning grounds often exhibit higher levels of stress indices (e.g. plasma cortisol, glucose, and lactate; Cooke et al. 2006), lower somatic energy reserves, higher levels of reproductive hormones, and have been shown to migrate 3 faster than fish that successfully reach natal sites (Young et al. 2006). Traversing hydraulically challenging flows can also increase stress, deplete energy reserves (Hinch and Rand 1998; Standen et al. 2002), and require migrants to utilize non-optimal swimming strategies that have latent effects on survival and spawning success (Burnett et al. 2014). With respect to homing, however, stressed and maturing adult salmon may also exhibit heightened olfactory sensitivity and enhanced long-term memory recall of imprinted odours (Nevitt et al. 1994; Carruth et al. 2002; Ueda 2011), though it remains to be examined how different stress and energy levels affect responses to natal water cues and in turn the homing behaviour of wild adult salmon. Although a number of environmental factors can influence the migration biology of fish (e.g. flow, turbidity, dissolved oxygen, temperature, natal water), water temperature has been coined the ‘ecological master factor’ (Brett 1971), primarily because it regulates so many aspects of migration. For Pacific salmon, this includes migration rates (Hanson et al. 2008; Goniea et al. 2011), physiological condition (Hinch et al. 2006; Donaldson et al. 2010), energy use (Burnett et al. 2013), and survival to natal sites (Martins et al. 2011; Hinch et al. 2012). Adult Columbia and Fraser River sockeye salmon for example, have been known to display faster migration rates at the onset of stressful temperatures, but in turn exhibit greater enroute mortality (Young et al. 2006, Keefer et al. 2008b). High temperatures can also increase maturation (reviewed in Hinch et al. 2006), deplete energy reserves (Hinch and Rand 1998; Burnett et al. 2013), increase susceptibility to freshwater pathogens and diseases (Wagner et al. 2005; Miller et al. 2011), and reduce aerobic scope (Eliason and Farrell 2015). Any or all of which are known to affect behaviour and reduce survival (Hinch et al. 2012). Not surprisingly, Pacific salmon have evolved to seek cool water refuges (e.g. hypolimnion of lakes, tributary outlets; Newell and Quinn 2005; Goniea et al. 2011) to reduce these thermal related issues and increase chances of migration survival (Mathes et al. 2010). The use of thermal refuges also appears to be particularly important for reproductively advanced females with lower energy reserves, who likely exploit cooler temperatures to slow rates of reproductive development (Roscoe et al. 2010a). However, delaying too long in thermal refuges has also been shown to reduce the probability of surviving to natal sites (Keefer et al. 2009), and thus there are trade-offs associated with this behaviour. Clearly, 4 a great deal of research has demonstrated how physiology, energy, and temperature all shape migration behaviour and survival of adult Pacific salmon; however, the relative role of natal water cues in this equation remains largely unexplored. In a fairly short period of time on the Pacific salmon evolutionary scale, human activities have added additional migration challenges that have resulted in regional declines and local extinctions (Noakes et al. 2000, Tear et al. 2005, Young 2011). For example, whether in marine, coastal, or riverine environments, migrants are likely to encounter fisheries at multiple scales (e.g. commercial, recreational, subsistence); release or escape from which can cause injuries and stress that ultimately result in slowed migrations and enroute mortality (Donaldson et al. 2011; Robinson et al. 2015). Human induced climate change has also at least in part caused many freshwater migration corridors to warm in recent years (Patterson et al. 2007, Isaak et al. 2012), with many adult salmon now routinely encountering temperatures well outside of thermal optima that can result in mortality (e.g. Eliason et al. 2011). Many climate models are also predicting future increases in regional warming throughout these migration corridors (Ferrari et al. 2007; Ficklin et al. 2014), thus adult salmon that migrate at precise times across years are likely to be exposed to increasingly stressful temperatures that could result in increased behavioural changes and mortality (Mantua et al. 2010; Hinch and Martins 2011; Hinch et al. 2012). The single greatest change to freshwater migration corridors worldwide, however, has been through the impoundment or diversion of waterways for irrigation, flood control, water storage, and hydroelectric generation (hydrosystems) (Nilsson et al. 2005; Vörösmarty et al. 2010). For adult salmon, hydrosystem passage can have individual to population level effects on migration behaviour and survival (reviewed in Waples et al. 2008; Thorstad et al. 2008). Arguably the most notable effects of hydrosystems on adult salmon migrations is that dam passage can require significant energy usage, resulting in extensive migration delays and latent effects on survival and spawning success (Gowans et al. 1999, 2003; Keefer et al. 2004b; Pon et al. 2009a; Burnett et al. 2013, 2014). However, hydrosystems are also known to change the hydrology of migration corridors (Young et al. 2011), and one of the most often overlooked aspects of hydrosystem 5 passage is how changes to flow composition can affect migration behaviour and in turn survival. For example, impoundment has increased river channel cross sections and led to odour diffusion from natal tributaries in the Columbia River hydrosystem, and many adult Chinook salmon (O. tshawytscha) now exhibit behaviours reflective of searching for natal homestream cues and orientation challenges, leading to delayed migration and potentially increased energy use (Bugert et al. 1997; Keefer et al. 2008a). In other regulated migration corridors, diversions of natal water sources from turbine outlets can re-distribute directional cues, causing significant migration delays (e.g. Fretwell 1989). For example, adult Atlantic salmon (Salmo salar) passing hydrosystems in Sweden and Norway display signs of confusion and are known to ‘yo-yo’ or wander between dams and turbine outlets (Lundqvist et al. 2008), where discharge of natal water at higher velocities than mainstem flows can also result in migration delays and upstream passage failures (Thorstad et al. 2003b, 2008). Indeed, many lab-based experiments have demonstrated that even small changes to the distribution and concentration of natal water cues can result in behavioural responses in adult salmon (reviewed in Bett and Hinch 2015); however, considerably less attention has been given to how this could affect the behaviour or survival of adults homing through corridors where natal water cues may indeed change. The literature on adult salmon homing through regulated rivers is extensive (see reviews in Waples et al. 2008; Thorstad et al. 2008; Roscoe and Hinch 2010; Keefer and Caudill 2014), but most previous studies have focused only on how environmental conditions (i.e. temperature and discharge) and the physiological state of migrants affect isolated aspects of passage (e.g. dam passage, reservoir migration rates) and in turn affect behaviour and survival (e.g. Gowans et al. 1999; Keefer et al. 2004b; Roscoe et al. 2010a, 2010b; Burnett et al. 2013). In the few studies that have examined passage through entire hydrosystem corridors, researchers have found evidence that slowed migration past multiple dams and reaches is associated with a reduction in survival (e.g. Naughton et al. 2005; Caudill et al. 2007). However, it is not known how the interactions between individual fish traits (i.e. physiological and energetic state) and environmental conditions (i.e. temperature, discharge, natal water concentration) influences behaviour in 6 hydrosystem corridors, or whether the cumulative effects of hydrosystem passage ultimately affect survival. 1.2 Study system and species The Seton River, located approximately 350 km upriver from the Pacific Ocean in the Seton-Anderson watershed in Southwestern British Columbia, Canada, is one of the only tributaries of the Fraser River that is regulated for hydroelectric power production (hereafter Seton hydrosystem). The Seton hydrosystem is home to 5 species of anadromous Pacific salmon, including 2 populations of sockeye. Gates Creek sockeye salmon are considered a ‘early summer run’ population of Fraser River sockeye that migrate through the hydrosystem beginning in late July and continuing through early September, while Portage Creek sockeye are a ‘late run’ population that migrates through the system beginning in mid September, continuing through early November. Both populations must pass through the lower Seton hydrosystem to reach natal spawning tributaries in the upper reaches of the watershed in the creeks of their namesakes. The Seton hydrosystem is a complex network of reservoirs, inter-basin diversions, and power generating stations operated by BC Hydro (see Casselman et al. 2015; Figure 2.1 & 3.1). Migrating adult sockeye salmon may first encounter the hydrosystem downstream from the Seton-Fraser confluence on the western bank of the Fraser River at the Seton Generating Station (hereafter ‘powerhouse’). This facility discharges natal Seton Lake water diverted by the Seton Dam from an underwater turbine outlet into a tailrace directly adjacent to the Fraser River. There is no fish passage provided at this facility and fish are prevented from entering the underwater outlet by steel grates. The Seton-Fraser confluence upstream 1.5 km from the powerhouse is the only migration corridor for migrants returning to the hydrosystem. The 5 km long Seton River is regulated by the Seton Dam and made passable to migratory fish by a vertical slot fishway. The only tributary of the Seton River is Cayoosh Creek, which drains a separate watershed. At times of increased discharge from Cayoosh Creek, its flows can ‘dilute the olfactory signature’ of the Seton River (Fretwell 1989) and thereby reduce the concentration of natal homestream cues in the lower Seton River and its plume. 7 Shortly after construction of the hydrosystem in 1956, large congregations of adult sockeye were reported milling in the powerhouse tailrace (Andrew and Geen 1958). In the late 1970’s and early 1980’s, studies conducted by the International Pacific Salmon Fisheries Commission (IPSFC) confirmed that the upstream migrations of both Gates Creek and Portage Creek sockeye could be delayed at the powerhouse. Furthermore, they demonstrated that migrants often had difficulty navigating beyond this point because the variable concentrations of natal water cues emanating from the Seton River were relatively weak compared to the pure natal Seton Lake water discharged in the powerhouse tailrace (summarized in Fretwell 1989). Results from behavioural choice experiments and limited radio-telemetry studies suggested that minimizing delay and encouraging upstream migration past the powerhouse and into the Seton River was dependent on maintaining natal water concentrations in the Seton River at or above 80% and 90% during the Gates Creek and Portage Creek migrations, respectively (Fretwell 1989). BC Hydro currently implements operational targets to maintain these natal water concentrations, although large fluctuations can still occur due to dam operational changes or precipitation events, and migration delays at the powerhouse and wandering behaviour in the Fraser River between the powerhouse and Seton-Fraser confluence are still observed in both populations (Roscoe and Hinch 2008). A large amount of research has emerged from the Seton hydrosystem in the past decade. Although a great deal has been learned from these studies, they have largely focused on the physiological and energetic correlates of Seton Dam passage, and post- passage behaviour and survival (Pon et al. 2009a, 2009b; Roscoe et al. 2010a, 2010b; Burnett et al. 2013, 2014). To date, no studies have explored how the unique environmental conditions in this system in combination with the physiological and energetic state of individuals interact to shape behaviour and passage through the hydrosystem. How passage behaviour ultimately affects survival to natal spawning sites has also been a lingering question since fish were first observed delaying at the powerhouse (Andrew and Geen 1958; Fretwell 1989). 1.3 Thesis overview and objectives In this thesis, I used the Seton hydrosystem and its two populations of sockeye salmon as a model from which to examine the relative roles that physiology, energy, and 8 environmental conditions can have on the behaviour and fate of sockeye salmon during the final stages of their spawning migration through a natal watershed, and secondly, the impacts hydrosystems can have on these migrations. In Chapter 2, I report the findings from two field studies that used biotelemetry to quantify the total amount of migration delay incurred by sockeye salmon in the Seton powerhouse tailrace, and assess the effectiveness of the current natal water targets in mitigating such delays. This chapter also evaluates the factors that contribute to initial attraction to the powerhouse and subsequent wandering behaviour of sockeye salmon in a reach of the migration corridor where the concentration and availability of natal water cues can vary on an hourly basis. Chapter 3 takes a broader look at migration and examines the relative roles that energy levels, natal water cues, temperature, and dam discharge have on sockeye salmon migration behaviour while passing through a regulated migration corridor, and ultimately assesses the cumulative effects of this passage experience on survival to natal sites. In Chapter 4 – my conclusion – I synthesize the findings from these studies, discuss some of the research limitations and implications of my results to local fisheries management, and suggest possible directions for future investigations. 9 Chapter 2: Effects of natal water concentration and temperature on migration behaviours of up-river migrating adult sockeye salmon 2.1 Introduction The behaviours of anadromous fishes during reproductive migrations are some of the most complex of any group of organisms (Dingle and Drake 2007), in large part because of the variable abiotic conditions encountered throughout these journeys and the inherent physiological changes associated with reproductive development (Alerstam et al. 2003; Donaldson et al. 2011). Pacific salmon (Oncorhynchus spp.) epitomize this complexity. Up-river migrating adults use olfactory cues to locate natal areas (Keefer and Caudill 2014) and do so while maturing and coping with environmental features that can lead to physiological stress and increased energy usage (e.g. high discharge or high temperature; Hinch et al. 2006). Because adults cease feeding prior to starting freshwater spawning migrations, they must complete migration, gonad development, and spawn using finite energy reserves (Brett 1971). Thus any environmental features that may slow or delay migrations (e.g. Bugert et al. 1997; Keefer et al. 2008a) could affect migration or spawning success (Burnett et al. 2016). Many freshwater corridors utilized by migratory salmon have been altered by impoundments or diversions, often necessitating the use of passage facilities (see reviews in Waples et al. 2008; Thorstad et al. 2008). While numerous studies have examined how changes in the amount of flow can affect migratory salmon (e.g. Quinn et al. 1997; Keefer et al. 2004b; Murchie et al. 2008), there have been few direct investigations of how changes in the specific composition of flows affect migrations. For example, multiple impoundments of the Columbia River, USA have increased river channel cross-sections and led to substantial odour diffusion from natal tributaries (Quinn 2005; Keefer and Caudill 2014). The wandering behaviour many adult salmon now exhibit in this system could be reflective of orientation challenges and active searching for natal water cues, though this is largely speculation (Keefer et al. 2008a). Adult salmon are known to be attracted to and delay their migrations in power station outlets which discharge natal water along migration routes (Andrew and Geen 1958; Thorstad et al. 2003b, 2008). At these facilities, numerous bouts of wandering in and out of mainstem flows can result in 10 substantial migration delays or failure to continue migrations upstream to pass dams (Fretwell 1989; Lundqvist et al. 2008). Indeed, lab-based experiments have revealed that migration responses in adult salmonids can be triggered by even slight changes in natal water concentrations (reviewed in Bett and Hinch 2015), yet comparatively little empirical investigation has examined how such changes to natal water along migration routes can influence behaviour in wild migrating adult salmon. In recent years, freshwater migration corridors have been warming (e.g. Patterson et al. 2007; Isaak et al. 2012) and many Pacific salmon now frequently encounter temperatures that can exceed population-specific thermal limits (Eliason et al. 2011). For adult sockeye salmon (O. nerka), exposure to high temperatures accelerates maturation (Hinch et al. 2006), depletes energy reserves (Hinch and Rand 1998; Burnett et al. 2014), increases physiological stress and incidence of disease development (Young et al. 2006; Miller et al. 2011), and reduces aerobic scope (Eliason and Farrell 2015); any or all of which can lead to enroute mortality (e.g. Keefer et al. 2009; Hinch et al. 2012). Pacific salmon will seek cool-water refuges (e.g. hypolimnion of lakes, tributary outlets; Newell and Quinn 2005; Goniea et al. 2011) to reduce these thermal related issues and increase chances of migration survival (Mathes et al. 2010). However, the use of thermal refuges could limit exposure to natal water cues and potentially increase migration delays. Delaying too long in thermal refuges has also been shown to reduce probability of surviving to natal sites by migrating adult salmon (e.g. Columbia River steelhead O. mykiss; Keefer et al. 2009). The physiological state of salmon can have strong effects on migration behaviours (Hinch et al. 2006). For instance, in ocean migrating adult Fraser River sockeye salmon, elevated levels of physiological stress (as reflected by plasma stress metabolites) is associated with delayed river entry and slowed migration rates (Cooke et al. 2006; Crossin et al. 2007). In contrast, advanced maturation (as reflected by high levels of plasma testosterone) is correlated with earlier river entry and fast migration rates (Sato et al. 1997; Crossin et al. 2009). The important role of physiology is further emphasized in how sexes can differ in migration behaviours. Females devote more energy to gonad development (Crossin et al. 2008) and typically have higher levels of reproductive 11 hormones than males throughout freshwater migrations (Truscott et al. 1986). In the Fraser River, female sockeye are energetically more efficient swimmers (Hinch and Rand 2000), but are often less successful in passing hydraulically challenging areas (Hinch and Bratty 2000; Burnett et al. 2013) compared to males. How an individual’s physiology or sex affects migration behaviour in regulated systems where natal water concentrations and flow composition can change during migration has not been examined. For this chapter, I used non-invasive biopsy and radio telemetry to track wild adult sockeye salmon from two populations as they migrated during the summer and fall from a large mainstem system (Fraser River) into a small tributary with regulated flows (Seton River) enroute to spawning areas. Immediately prior to entering the tributary, migrants pass by a powerhouse tailrace which may be attractive because it discharges water containing strong natal cues and is sometimes cooler than the mainstem offering a potential thermal refuge. Natal cue composition in the tributary can vary daily resulting from the operation of a diversion dam – thus causing the availability and strength of natal cues entering the mainstem just upstream of the powerhouse to also vary. This locale provides an ideal system in which to examine the relative roles that flow composition (e.g. natal water component), water temperature, and thermal refugia can have on individual adult salmon’s behaviour as they migrate towards spawning grounds. I predicted that, if variation in natal water concentration does indeed affect migration behaviour, adult sockeye salmon would spend more time in the powerhouse tailrace when upstream natal water composition in the mainstem Fraser River was low. Furthermore, if higher concentrations of natal water in the Seton River and its plume are meant to encourage upstream migration, I predicted that fish would display behaviours indicative of ‘migration confusion’ such as enhanced wandering when natal water concentration was reduced, but that migrants would make more directed migrations towards their tributary when natal water composition was higher. I further hypothesized that if mainstem Fraser River temperatures were extremely warm or approaching physiologically critical levels, fish would spend more time delaying in the powerhouse tailrace, potentially using this area as a thermal refuge. Lastly, given the physiological condition and sex of migrants are known to differentially affect migration behaviour, I 12 predicted that sockeye salmon who were initially more mature would migrate faster and more directly towards the natal tributary, whereas those which were more physiologically stressed would delay their migration, likely in the powerhouse tailrace. I also expected that females would exhibit less wandering and less delay at the powerhouse given their more limited energy budget. 2.2 Methods 2.2.1 Study system The Seton hydrosystem is a complex network of reservoirs, inter-basin diversions, and power generating stations operated by BC Hydro (see Casselman et al. 2015). Adult sockeye salmon first encounter the Seton Generating Station (powerhouse) ~ 340 km from the Pacific Ocean on the western bank of the Fraser River near the confluence of the Seton and Fraser Rivers (Figure 2.1). At the powerhouse, natal Seton Lake water diverted by the Seton Dam is discharged from an underwater outlet into a 5-meter deep tailrace directly adjacent to the Fraser River (Figure 2.1). There is no fish passage at this facility and individuals are prevented from entering the underwater outlet by steel grates. During adult sockeye salmon migrations, mean discharge from the powerhouse is ~ 84.5 m3s-1, or ~ 5% of the mean 1492.5 m3s-1 of the adjacent mainstem Fraser River. Discharge from the powerhouse creates a plume of natal Seton Lake water that visibly extends ~ 800 m downstream into the Fraser River (Figure 2.1). Upstream 1.5 km from the powerhouse and ~ 3 km past the extent of its discharge plume is the Seton-Fraser confluence (Figure 2.1). The 5 km long Seton River (22.7 m3s-1mean annual discharge) is regulated by the Seton Dam and made passable to migratory fish by a 107 m long, 32-pool vertical slot fishway. Cayoosh Creek (13.9 m3s-1 mean annual discharge) is the only tributary of the Seton River; it drains a separate watershed (Figure 2.1). At times of increased discharge from Cayoosh Creek, which occurs during operational changes at the Seton Dam or at Walden North (the small diversion dam on Cayoosh Creek), or during large precipitation events, flows from this tributary can ‘dilute the olfactory signature’ of the Seton River (Fretwell 1989) and thereby reduce the concentration of natal homestream cues in the lower Seton River and its plume. It is the combination of natal water discharged at the powerhouse and reduction in natal water concentration in the Seton River that creates the olfactory conditions of interest in this study (Figure 2.1). 13 Gates Creek (hereafter GC) and Portage Creek (hereafter PC) sockeye salmon are the two populations of Fraser River sockeye salmon that spawn and rear in this system, representing distinct ‘early-summer’ and ‘late-run’ populations, respectively. The GC population typically begins migrating into the hydrosystem in late July, travelling ~ 55 km past the Seton Dam through Seton and Anderson Lakes to reach spawning sites at Gates Creek. Whereas the PC population begins migrating into the hydrosystem at the end of September, travelling ~ 20 km past the Seton Dam through Seton Lake to reach spawning sites in Portage Creek. Studies conducted throughout the 1970’s by the International Pacific Salmon Fisheries Commission (IPSFC) indicated that the upstream migration of both populations could be delayed at the powerhouse, and that migrants often had difficulty navigating beyond this point (summarized in Fretwell 1989). Results of behavioural choice (Y-maze) experiments and limited radio-telemetry studies suggested that minimizing delay and encouraging upstream migration past the powerhouse was dependent on maintaining natal water levels in the Seton River and its plume at or above 80% and 90% during the GC and PC migrations, respectively (Fretwell 1989). BC Hydro currently operates the hydrosystem with measures in place to maintain these natal water targets, though large fluctuations still occur, and migration delays at the powerhouse followed by wandering in the Fraser River are still observed in both populations (Roscoe and Hinch 2008). 2.2.2 Fish capture and tagging It was not possible to capture Seton sockeye in the Fraser River mainstem. Therefore, all fish were captured using a full-spanning fence and trap at a site 200 m downstream from the Seton Dam in the Seton River (Figure 2.1). In total, I radio-tagged and monitored the movements of 517 sockeye salmon through the summer and fall of 2013 and 2014. Specifically, 138 GC sockeye salmon were tagged between 5 August and 2 September 2013, and 166 were tagged between 5 August and 7 September 2014. In 2013, PC sockeye salmon co-migrated with ~ 0.8 million pink salmon (Oncorhynchus gorbuscha) returning to the Seton River. This made targeting PC sockeye salmon with the fence and trap extremely difficult – to the extent that only 24 individuals were tagged from 3 – 9 October 2013. However no pink salmon were present in 2014, enabling 189 PC sockeye to be tagged from 27 September to 9 October. 14 Radio transmitters [(43 mm length × 16 mm diameter; 15.2 g in air); Sigma Eight, Newmarket, Ontario, Canada] were gastrically inserted into individual fish. Tags had a burst-rate of 3 s and a unique digital identification code. Half-duplex passive integrated transponder (PIT) tags (32 mm X 3.65 mm; Texas Instruments, Dallas) were also implanted into individual fish; individually labeled spaghetti tags (Floy Manufacturing, Seattle) were attached externally through the dorsum of each fish. A tissue sample from the adipose fin and a 1.5 ml blood sample from the caudal vasculature were taken for population identification and physiological assays. Average time to tag and biopsy sample each fish was ~ four minutes (± 1.3 min; SD). Fish were not anesthetized to minimize handling and related stress (Cooke et al. 2005). My handling and intra-gastric tagging methods followed procedures used in many adult salmon tracking studies and are known not to affect behaviour or cause mortality (e.g. Ramstad and Woody 2003; Cooke et al. 2005). All procedures followed the methods described in greater detail in (Pon et al. 2009a; Roscoe et al. 2010b; Burnett et al. 2013, 2014), and were conducted in accordance with the guidelines of the Canadian Council of Animal Care administered by the University of British Columbia (A11-0125). I selected fish for tagging based on visual inspection of injury, external pathogens and somatic lipid content (measured by a hand-held microwave energy meter – Fatmeter model 692; Distell Inc., West Lothian, Scotland, UK; see Crossin and Hinch 2005). Severely injured fish (e.g. lacerations exposing the coelomic cavity or skeletal system) or those with extensive fungus cover on body or gills were not tagged, nor were those with high somatic lipid concentrations. An earlier study which compared DNA population identification to Fatmeter output from fish captured at the fish fence revealed that these high energy fish were strays which had entered the Seton River but whose spawning grounds were several 100 km further along the Fraser River (Casselman et al. 2015). Sex was assigned by analysis of 17-β estradiol and testosterone concentrations in blood plasma. Plasma concentrations of glucose and lactate were used as indices of stress (Cooke et al. 2008), and testosterone concentrations were used to infer maturity level (Crossin et al. 2007). All blood plasma analyses followed the methods described in (Roscoe et al. 2010b). 15 In order to examine how migrants behaved as they first encountered the powerhouse outflows and Seton River plume, tagged fish were transported by truck downstream of the powerhouse for release into the Fraser River. Groups of ten to twelve individuals (approximate 50:50 sex ratio) were moved via a 1000 L insulated and oxygenated transport tank, and released within 30 min (± 10 min; SD) of tagging, 1.5 km downstream from the powerhouse on the western bank of the Fraser River (Figure 2.1). The use of passive capture methods followed by transport have been used in other studies with little effect on behaviour or mortality (Keefer et al. 2004a, 2009; Hubert et al. 2012). Moreover, methods of upstream capture and downstream release are common in studies of salmon migrating through regulated rivers (Thorstad et al. 2003b; Naughton et al. 2005; Caudill et al. 2007), and there is little evidence suggesting adult salmon have the ability to learn migration routes or consequently behave differently from non-study fish (Hansen and Jonsson 1994; Thorstad et al. 2003b). 2.2.3 Telemetry monitoring and quantifying behaviour I used a combination of fixed SRX (Lotek Wireless, Inc.) and Orion (Sigma Eight, Inc.) radio-receivers fitted with aerial Yagi antennas at different locations in the hydrosystem to monitor fish movement (Figure 2.1). The powerhouse receiver was configured to only detect tags only in the tailrace and not in the adjacent Fraser River. The receiver at the Seton-Fraser confluence was configured to only detect tags within a ~ 100 m radius of the Seton River mouth. A receiver at the Fraser River release site served as a ‘gate’ for detecting individuals exhibiting fallback in the mainstem Fraser River, whereas receivers upstream in the Seton River confirmed Seton River entry or arrival at the dam by tagged fish (Figure 2.1). All receivers operated > 85% of the time, providing sufficient ability to monitor fish movements (Appendix 1). Raw telemetry data were filtered to remove false detections (e.g. likely false detections of study tag ID’s, non-study tag ID’s, detections recorded before release; Beeman & Perry 2012) and used to generate migration histories. To ensure quality control of the data, plots of individual migration histories were reviewed for each fish to exclude spatio-temporally implausible records and to quantify behaviour and fate. I quantified migration behaviour by (1) establishing the number of forays individuals made into the powerhouse tailrace as a measure of attraction to this facility, (2) determining the 16 number of back-and-forth movements in the Fraser River between the release site, the powerhouse, and the Seton-Fraser confluence as an index of migration confusion (hereafter ‘wandering’), and (3) summing the duration of each foray into the powerhouse tailrace as an estimate of the total amount of migration delay incurred by individuals at this facility. To ensure the number of forays and thus delay at the powerhouse were representative of true behaviour I calculated each metric based on detections that were part of ‘residence events’. Residence events for a given fish were defined as a series of consecutive detections at a given receiver separated by no more than 30 minutes. The duration of a residence event could continue indefinitely until being terminated when (1) the time elapsed between any two consecutive detections was > 30 minutes, or when (2) a detection occurred at another receiver site. Visual examinations of migration histories were also used to count the number of back-and-forth movements (i.e. wandering) in the Fraser River. I classified every change in direction prior to final assignment of fate as a single wandering event. Fish with wandering values of zero exhibited only direct movements in the Fraser River, while subsequently increasing numbers indicated the number of changes in direction prior to final assignment of fate. Fate was assigned to individuals as having (a) successfully entered the Seton River if their last known detection occurred at any receiver upstream of the Seton-Fraser confluence into the Seton-Anderson watershed, (b) passed the Seton River and migrated further upstream in the Fraser River if their last detection occurred at the Seton-Fraser confluence, (c) migrated downstream in the mainstem Fraser River if their last detection occurred at the release site receiver, or (d) be of unclassified fate if detections did not conform any of the aforementioned patterns. Of the 304 GC sockeye salmon originally released for this study, 90% (274 of 304) successfully entered the Seton River, 4% (12 of 304) overshot the Seton River, 3% (9 of 304) migrated back downstream in the Fraser, and 3% (9 of 304) had unclassified fates. Of the 213 PC fish originally released, 80% (170 of 213) successfully entered the Seton River, 6% (13 of 213) overshot the Seton River, 13% (28 of 213) returned downstream, and 1% (2 of 213) had unclassified fates. Fish that fell back downstream after release and did not re-ascend may have suffered from some form of experimenter-induced stress. However, given that this was a relatively small component of my fish, and 93% and 86% of GC and PC fish, 17 respectively, reached the Seton River suggests that my handling and transport approaches probably had minimal effects on fish behaviour and short-term mortality overall. 2.2.4 Environmental data collection and study conditions I used hourly measurements from the Water Survey of Canada (available from http://wateroffice.ec.gc.ca) to monitor discharge in Cayoosh Creek (Station No. 08ME002) and the upper Seton River below the Seton Dam (Station No. 08ME003). Hourly discharge data from the powerhouse were provided by BC Hydro. I calculated hourly lower Seton River discharge as the sum of upper Seton River discharge and Cayoosh Creek discharge plus a constant 2.1 m3s-1 of lake water diverted to an artificial spawning channel and returned into the lower Seton River before the Seton-Fraser confluence. I then calculated the concentration of natal water in the lower Seton River and its plume using discharges in the following equation: 𝑁𝑎𝑡𝑎𝑙 𝑤𝑎𝑡𝑒𝑟 (%)= 100 − (100 × (𝐶𝑎𝑦𝑜𝑜𝑠ℎ 𝐶𝑟𝑒𝑒𝑘𝑈𝑝𝑝𝑒𝑟 𝑆𝑒𝑡𝑜𝑛 𝑅𝑖𝑣𝑒𝑟 + 𝐶𝑎𝑦𝑜𝑜𝑠ℎ 𝐶𝑟𝑒𝑒𝑘 + 𝑆𝑝𝑎𝑤𝑛𝑖𝑛𝑔 𝐶ℎ𝑎𝑛𝑛𝑒𝑙)) I confirmed differences in the water chemistry between natal Seton River and non-natal Cayoosh Creek water, which could reflect differences in olfactory cues, by collecting daily conductivity measurements from Cayoosh Creek, the upper and lower Seton River, and the powerhouse using a hand held YSI Pro30 conductivity meter (YSI Inc., Yellow Springs, OH, USA) (Appendix 1). Conductivity has previously been used to distinguish between water sources utilized by other anadromous fishes that use olfaction to guide homing (e.g. Leduc et al. 2010, Vrieze et al. 2011). Hourly water temperatures were measured and collected using TidbiT v2 data loggers (Onset HOBO data loggers, Bourne, MA) at locations in the Fraser River, the powerhouse, and the lower Seton River (Figure 2.1). 2.2.5 Data analysis Statistical models I used generalized linear (GLM) and linear models (LM) to relate migration behaviour to characteristics of individual migrants and the environmental conditions they encountered in the lower reaches of the Seton hydrosystem. My analyses included three 18 specific models to examine: (1) the number of forays made into the powerhouse (Poisson or Negative Binomial GLM), (2) the number of wandering events between the powerhouse and Seton-Fraser confluence (Poisson or Negative Binomial GLM), and (3) the total amount of migration delay incurred at the powerhouse (LM). I conducted all analyses on the GC and PC populations independently because of the different run-timing and environmental conditions experienced by the two populations. Model construction All global models included explanatory variables for sex, stress (lactate and glucose), and maturity levels (testosterone) at the time of tagging. Variables summarizing temperatures encountered in the Fraser River and in the powerhouse tailrace were included in each model, as was a summary measure of natal water concentration in the Seton River plume. Neither Fraser River nor Seton River discharge were included as predictors in the models because GC and PC sockeye migrate upriver when flows are lowest in the migration season and at levels unlikely to affect behavior (Rand et al. 2006; Macdonald et al. 2007). Moreover, temperature and discharge were highly correlated (r = 0.89). Powerhouse discharge was also not included in any analyses because all fish included in the models experienced consistent discharge of ~ 87.3 m3s-1 in the powerhouse tailrace. All calculations of environmental explanatory variables were based on residence events and dependent on each model response. I calculated all variables as means to ensure the best representation of overall conditions experienced, or means weighted by residence event duration to emphasize the effect of conditions experienced over longer time. In model 1, I used measurements that occurred 30 seconds prior to the first detection of every foray into the powerhouse to best represent what conditions were like the moment before an individual moved into the powerhouse tailrace. In model 2, I weighted the means of conditions based on the duration of residence events that occurred while individuals changed migration direction in the Fraser River between the release site, the powerhouse, and the Seton-Fraser confluence. In model 3, I again weighted the means of conditions by the duration of each foray into the powerhouse, but then took an overall average of these conditions to use as a final explanatory variable in the model. In each of the models, I included explanatory variables that were measured on different scales (e.g. 19 C, %). To facilitate comparisons of the relative effect size of each explanatory variable, I standardized the data by centering (subtracting the mean) and dividing by two standard deviations (Gelman 2008; Schielzeth 2010). Prior to any analyses, I applied the data exploration protocols described in Zuur et al. (2010) to each model separately. Examinations of Cleveland dotplots were used to identify outliers and variables with inordinate values to the majority of observations were removed (1 – 4 observations per model). Pearson correlation coefficients > 0.7 and variance inflation factors > 3 were used as thresholds to identify collinear variables (Zuur et al. 2010). In models 1 and 3 for PC sockeye salmon, the Fraser River temperature variable was highly collinear with the natal water concentration variable. To account for this, I substituted tagging date for Fraser River temperature because the two were related (r = -0.68), but tagging date was not collinear with the natal water concentration variable (VIF < 3). I constructed all models based on a sample size of at least ten observations per explanatory variable. Models 1and 2 were initially fit with a Poisson GLM, then assessed for over-dispersion by summing the square of the Pearson residuals and dividing by the degrees of freedom (McCullagh and Nelder 1989). Models with a dispersion parameter greater than 1 were deemed overdispersed and re-fit with a negative binomial GLM (O'Hara and Kotze 2010). All subsequent GLMs fitted the data adequately as assessed by a Chi-square test (P > 0.05; Smyth 2003). In model 3, I log-transformed the response variables to satisfy assumptions of linearity and homogeneity of residual variance (both assessed visually); linear model fits were evaluated using adjusted R2. Model selection and multimodel averaging In each analysis, I fit models with all possible subsets of explanatory variables to the data. All models were then ranked using the bias-corrected Akaike Information Criterion (AICC) and compared using their AICC weights (wi), which describes the probability of a given model in a candidate set as being the most parsimonious, given the data (Burnham and Anderson 2003). Uncertainty in model selection was accounted for by calculating model-averaged estimates of the coefficients using the ‘natural average’ method (Grueber et al. 2011). Only models included in the 95% confidence set of models 20 were used for model averaging (Burnham and Anderson 2003). I present model selection statistics for all models with ΔAIC values < 2 in Table 2.1. All statistical analyses were conducted in R using the MuMIn package (R Development Core Team, 2008; version 3.1.2). 2.3 Results 2.3.1 Environmental conditions The mean Fraser River temperature experienced by GC sockeye salmon in this study was 18.3C (±1.3C SD), though temperatures varied considerably over the entire 2013 and 2014 migrations, with some periods reaching near record summer highs (~ 22C; Figure 2.2a). On average, GC fish encountered temperatures in the powerhouse of 17.9C (±1.1C SD); there were many opportunities for migrants to seek thermal refuge in this area as temperatures were often 0.5 – 4.0C cooler than the adjacent Fraser River; though temperatures in this area could also be warmer than the Fraser River as well (Figure 2.2a). For more than 83% of both study years, natal water concentration in the Seton River plume remained above the recommended 80% target for the GC population (Figure 2.2b), with tagged fish on average experiencing concentrations of 90.3% (±4.6% SD). PC sockeye salmon are a late-run population that migrates during the fall months, thus they encountered Fraser River temperatures that were substantially cooler than during the GC migration (Figure 2.2a). For example, PC fish tracked in this study experienced mean Fraser River temperatures of 11.8C (±0.7C SD), while they experienced 14.6C (±0.5C SD) on average at the powerhouse (Figure 2.2a). Natal water concentration in the Seton River plume remained above the 90% target for the PC population for over 87% of both 2013 and 2014 migrations, while the mean concentration experienced by PC fish in this study was 91.4% (±1.2% SD) (Figure 2.2b). 2.3.2 Stress and reproductive hormone levels I compared mean levels of plasma glucose, lactate, and testosterone between sexes, populations, and years. There were no differences between years within each of the parameters measured for each population (Table 2.2). Glucose levels were very similar among males and females, both within and between the populations (Table 2.2). However, 21 lactate was ~1.4 times higher in females compared to males in each population, and ~1.3 times higher among all GC fish relative to PC fish (Table 2.2). Testosterone was considerably higher in females compared to males for both populations [(5.6 times higher in GC population) (4.9 times higher in PC population)], whereas PC sockeye salmon had 2.2 times higher levels of this hormone than GC sockeye salmon (Table 2.2). 2.3.3 Powerhouse attraction and forays Of all tagged GC and PC sockeye salmon initially released into the Fraser River, 87% (265 of 304) and 91% (193 of 213), from each population respectively, made at least one foray into the powerhouse. However, because only 255 GC and 90 PC sockeye salmon had complete stress and maturity profiles, only these fish were included in models predicting this behaviour (fates given by sex and population in Appendix 1). Fifty six percent (142 of 255) of these GC fish made only one foray into the powerhouse, while the remainder made up to 8 forays (Figure 2.3 a). In contrast, 43% of PC fish (39 of 90) made only one foray into the powerhouse, while the remainder made up to 17, with one PC female making 24 forays (Figure 2.3 b). PC sockeye salmon (3.6 forays ± 4.0 SD) made nearly twice as many forays into the powerhouse as GC fish overall (1.9 forays ± 1.4 SD) (Figure 2.3 a & b). AIC model selection results indicated a large amount of uncertainty in foray models for GC fish, whereas there was less uncertainty in models for PC fish (Appendix 1). Nevertheless, there are trends among explanatory variables that are consistent between populations and important to highlight. For instance, sex was absent from the majority of models in the 95% confidence sets for both populations because males and females made similar numbers of forays (Table 2.1; Figure 2.3 a & b). Model-averaged estimates for the effect of glucose in both populations were negative and ~ 2× larger than those of lactate, suggesting individuals with elevated levels of glucose may have made fewer forays into the powerhouse (Figure 2.3 c & d). Testosterone effects were positive for both populations, implying that increasingly mature fish may have made more forays into this facility (Figure 2.3 c & d). However, for both glucose and testosterone effects in both populations, the 95% confidence intervals (CIs) included zero, thus confidence in these predictions was low (Figure 2.3 c & d). 22 Natal water concentration and powerhouse temperature had the largest effects on forays among PC fish and were included in the majority of models from the 95% set for this population (Table 2.1). Estimates for these effects were negative and had CIs that did not include zero, indicating that PC fish that encountered higher natal water concentrations or elevated powerhouse temperatures made fewer forays into this facility (Figure 2.3 d). Foray behaviour among GC sockeye salmon was largely unaffected by natal water concentration or powerhouse temperatures as these effects were absent from most of the models in the 95% confidence set for this population (Table 2.1; Figure 2.3 c). There was little indication that foray behaviour in fish from either population was affected by the temperature differential between the powerhouse and Fraser River (Figure 2.3 c & d). 2.3.4 Wandering Models predicting wandering behaviour included 235 GC and 78 PC sockeye salmon (fates given by sex and population in Appendix 1). Eighty three percent (195 of 235) and 73% (57 of 78) of GC and PC sockeye salmon, respectively, migrated upstream in the Fraser River without exhibiting any back-and-forth movements (Figure 2.4 a & b). Males and females within each population displayed similar amounts of wandering. There was more variability in this behaviour among PC fish as 27% of PC migrants wandered 1 to 5 times, while 17% of the GC migrants wandered back-and-forth 1 to 4 times (Figure 2.4 a & b). There was a fair amount of uncertainty in the model selection results (Appendix 1); however, like the foray models, there are trends among predictors that are important to highlight. Lactate was included in the majority of the models from the 95% confidence set for both populations, with negative effects that were more than twice that of glucose, suggesting that wandering may have been reduced in fish with higher levels of this metabolic stress variable (Table 2.1; Figure 2.4 c & d). In each case, CIs for these effect estimates of lactate only narrowly included zero (Figure 2.4 c & d). The negative effect of sex in the GC model was driven primarily by 1 female that wandered 9 times (Figure 2.4 a). 23 Fraser River temperature had little effect on wandering among GC fish but was included in the majority of the models in the 95% confidence set for the PC population suggesting this behaviour was likely reduced in PC fish that encountered warmer Fraser River temperatures; again, CI for this estimate only narrowly included zero (Table 2.1; Figure 2.4 c & d). The effects of temperatures at the powerhouse had similar size but opposite effects on both stocks, implying warmer temperatures at this facility may have been associated with increased wandering among GC sockeye salmon and less among PC sockeye salmon (Figure 2.4 c & d). There was more support for the effect of powerhouse temperature among GC fish models (Table 2.1). There was no evidence that the temperature differential between the powerhouse tailrace and the Fraser River influenced wandering in either population (Table 2.1; Figure 2.4 c & d). Natal water was included in the majority of the models from the 95% confidence set for PC fish (Table 2.1). This variable had the largest effect of all predictors on wandering among PC fish and had CI that did not include zero (Figure. 2.4 d). These results suggested that wandering declined among PC sockeye salmon that encountered higher concentrations of natal water in the Seton River plume. Wandering among GC fish in contrast was largely unaffected by changes in natal water (Figure 2.4 c). 2.3.5 Powerhouse delay Models predicting the total amount of migration delay incurred by individuals in the powerhouse tailrace included 256 GC and 90 PC sockeye salmon (fates given by sex and population in Appendix 1). Mean delay times of GC females (5.3 hours ± 6.4 SD) were only slightly higher than GC males (3.4 hours ± 3.9 SD), and only slightly lower among PC females (19.5 hours ± 18.5 SD) compared to PC males (23.4 hours ± 27.1 SD), and there was no effect of sex on delay in models for either population (Figure 2.5 c & d). Overall, PC sockeye salmon (21.1 hours ± 22.3 SD) delayed ~ 5✕ longer than GC fish (4.5 hours ± 5.6 SD) (Table 2.1; Figure 2.5 a). Glucose was included in the majority of models from the 95% confidence set, had the largest effect on delay among GC fish, and its CI did not include zero (Table 2.1; Figure 2.5 b). This negative effect of glucose was 9✕ greater than that of lactate and indicated that GC fish with elevated levels of glucose incurred less delay in the 24 powerhouse (Figure 2.5 b); neither glucose nor lactate had any effect on delay among PC fish (Figure 2.5 c). There was little support for testosterone effects on delay in either population (Figure 2.5 c & d). Natal water concentration was included in the majority of models from the 95% confidence set for the PC population, with a negative effect that was the largest of all predictors. This finding indicates that delay at the powerhouse decreased when PC fish encountered higher concentrations of natal water emanating from the Seton River plume (Table 2.1; Figure 2.5 c). Effects of temperatures in the powerhouse itself were also well-supported in models for the PC population, with a strong negative effect of this parameter indicating that PC fish delayed less during periods of elevated temperatures at this facility (Table 2.1; Figure 2.5 c). Although a similar trend in the effects of powerhouse temperature was also observed for GC fish, CIs for this effect included zero (Figure 2.5 b). There was no indication that the temperature differential between the powerhouse tailrace and the Fraser River affected delay in either population (Figure 2.5 b & c). 2.4 Discussion All migrants in this study entered the powerhouse tailrace as they migrated up the Fraser River enroute to their natal (Seton River) tributary and spent varying amounts of time in this area before resuming their migration. Adult Atlantic salmon migrating up rivers in Norway and Sweden have been found to be attracted to turbine outlets making several movements in and out of these tailrace areas (Thorstad et al. 2003a; Lundqvist et al. 2008). In these studies, tailrace discharges were usually far greater than bypass flows thus discharge was the primary cause of tailrace attraction. However, in my study, spill from the powerhouse was only ~5% of the mainstem Fraser River suggesting that initial attraction to the powerhouse occurs either because of the strong olfactory cues coming directly from natal Seton Lake which is discharged at the powerhouse, and/or because the powerhouse provides some degree of thermal refuge during the GC migration. One of the most striking findings in this study was the population-specific differences I observed in how individual migrants behaviourally responded to environmental factors they encountered in this short segment of their shared migration corridor. PC sockeye spent 4 – 5 times longer than GC sockeye in the powerhouse 25 tailrace (mean ~ 25 hours vs. 5 hours, respectively). Late run populations of Fraser River sockeye (e.g. PC fish) are known to migrate slower up the Fraser River than summer runs (e.g. GC fish) (English et al. 2005), and unlike summer runs, late run fish usually mill in natal lakes prior to entering spawning grounds (Hinch et al. 2012), so this extended delay in the tailrace is not unexpected especially given PC fish are only ~ 8 km from their natal lake and spawning grounds. As I had anticipated, changes in the concentration of natal water had a strong effect on migration behaviour, but only for the PC population. Specifically, lower natal water concentrations emanating from the Seton River into the Fraser River were associated with more forays into the powerhouse tailrace, more overall time in the tailrace, and more wandering in the Fraser River. It thus appeared that PC migrants became confused when natal water concentrations were reduced and were less certain of their intended trajectories. Given the narrower natal water concentration range experienced by PC migrants during the study (~ 85 – 94% natal water) compared to that of GC migrants (~ 60 – 95% natal water) it is possible that PC migrants are more sensitive to subtle changes in olfactory cues. One explanation for this is that PC fish were more mature than GC fish (based on higher testosterone concentrations) and more mature fish can exhibit heightened discriminatory capability of homestream odours, and thus may be more sensitive to even slight changes in natal water concentration (Nevitt and Dittman 1998; Ueda 2011). Further, PC fish also encounter relatively benign temperature conditions in the Fraser River (discussed further below), thus the primary challenge this population faces involves olfactory ‘obstacles’ at this stage in their migration, possibly leading to an acute sensitivity to small changes in this abiotic factor. High migration temperatures are stressful for adult salmon and upriver migrants will alter their migration times to avoid peak temperatures (e.g. Columbia River sockeye & steelhead; Quinn and Adams 1996; Robards and Quinn 2011), and, seek cool water refugia in tributaries & lakes when they can (Columbia & Fraser River sockeye & Chinook; Hodgson and Quinn 2002, Goniea et al. 2011). Thermal data loggers implanted in upriver migrating Fraser River sockeye salmon have revealed that the Fraser River mainstem provides very limited areas of thermal refuge during migration (Donaldson et al. 2009) so I had hypothesized that the powerhouse tailrace could serve as a thermal refuge for migrants if mainstem temperatures were high. Indeed many GC fish 26 experienced Fraser River temperatures that were well above optimal (~ 17C; Lee et al. 2003) and near their measured thermal limit (~ 21C; Eliason et al. 2011) where aerobic scope collapses and fish must rely on anaerobic metabolism risking acidosis and cardiorespiratory failure (Farrell et al. 2008). However, I found no evidence for GC fish that the number of forays into, or delay in, the powerhouse tailrace increased when tailrace water was cooler or when the temperature differential between the tailrace and the Fraser River was large. There were in fact many periods when tailrace temperatures were very similar to, or even greater than, Fraser River temperatures, thus opportunities for the powerhouse tailrace to act as a thermal refuge were limited. The fact that GC fish spent relatively little time in the powerhouse tailrace, and instead migrated quickly into the Seton River suggests that if they were in need of some form of thermal refuge, they would find this in the Seton River which was generally 1-3C cooler than the Fraser River during their migratory period (Casselman et al. 2015), and/or in the hypolimnion of Seton Lake, which GC sockeye are known to occupy enroute to spawning grounds (Roscoe et al. 2010a) and is only ~ 6 km from the powerhouse tailrace. During the migration of PC fish, there was likely no need for individuals to seek thermal refuge in the tailrace as the Fraser River was always 2 – 3C cooler, yet temperature still seemed to play a role in behaviour in that warmer tailrace temperatures were associated with shorter delays and less forays into the tailrace of this facility. Caudill et al. (2013) found that adult Chinook salmon and steelhead would turn away from high temperatures in fish ladders and remain in cooler dam tailraces in the Snake River, Washington. In a similar manner, PC fish may choose to avoid the powerhouse tailrace and spend more time in the cooler Fraser River where temperatures are closer to optimum for late-run sockeye (Lee et al. 2003; Farrell et al. 2008; Eliason et al. 2011). As discussed earlier, high concentrations of natal water in the Fraser were associated with less delay and fewer forays at the tailrace but the interactive roles of olfaction and thermal biology in altering behaviour of migrating PC fish could not be examined in these models because temperature and natal water variables were highly collinear, so this specific issue remains unresolved. 27 I pooled data from the 2013 and 2014 studies so that I had adequate sample sizes to examine how behaviours were influenced by a large number of environmental variables simultaneously. One advantage of this approach is that it enabled a broader range of water temperatures and natal flow composition to be examined. If fish condition varied between years, this could have compromised my interpretations. However, the initial size (length) and physiological state (plasma glucose, lactate, testosterone) of migrants at their time of release did not differ within a population between years suggesting that my results were not driven by year-specific fish condition issues. I had expected that initial physiological state would play a role in affecting population specific behaviours and had predicted that sockeye which were initially more mature would migrate more directly toward their natal tributary while those which were more physiologically stressed would delay their migration, likely in the powerhouse tailrace where they might be able to recover from stressful conditions away from the mainstem. However I found few associations, within a population, that indices of physiological stress or maturation influenced behaviour. The only exception was GC fish, which, contrary to my expectation, spent less time delaying in the powerhouse tailrace when their plasma glucose concentrations were relatively high. Elevated glucose can be indicative of recovery from swimming fatigue and excessive handling or confinement (Nielsen et al. 1994; Farrell et al. 1998; Kubokawa et al. 1999). It is possible that because GC fish had migrated ~ 340 km up river during peak summer temperatures that they were experiencing some level of modest physiological stress. Plasma glucose levels at time of capture were well above baseline, though similar to levels found previously in GC fish captured at this local in 2007 (Roscoe et al. 2010b), but not extremely high relative to values recorded for summer run sockeye captured in the lower Fraser River (Donaldson et al 2010). Because I found there was only limited opportunity for thermal refuge in the powerhouse tailrace for GC fish, the most stressed individuals may have chosen to move more quickly through the study area to the Seton River where flow rates and temperatures were lower and better for physiological recovery. Indeed, it has been shown that sockeye salmon in the Columbia River migrate more rapidly toward natal tributaries as mainstem temperatures increase (Naughton et al. 2005; Keefer et al. 2008b). 28 I had hypothesized that given their more limited energy budgets for swimming activities, females would exhibit less wandering and less delay at the powerhouse compared to males. Testosterone levels were substantially higher among females from both populations, which is typical of sockeye salmon throughout their freshwater migration (Truscott et al. 1986; Crossin et al. 2007; Donaldson et al. 2010). Testosterone also increases steadily in both sexes with decreasing distance to natal sites as oocytes and testes prepare for final maturation (Truscott et al. 1986; Hinch et al. 2006), perhaps explaining why PC fish had nearly twice as high plasma testosterone as GC fish given the much shorter distance remaining to PC spawning grounds. Despite being more mature, females from both populations did not behave differently than males. This was unexpected given that recent physiological telemetry investigations involving GC sockeye salmon at the nearby Seton Dam found that females swam differently than males, exhibiting much more anaerobic effort (more burst swimming) compared to males when passing through the Seton Dam tailrace and through the adjoining fishway (Burnett et al. 2014). Towards the end of freshwater migrations, female sockeye often suffer higher levels of mortality than males (Hodgson and Quinn 2002; Keefer et al. 2004a, 2004c, 2008a, 2008b; Hanson et al. 2008; Donaldson et al. 2010; Martins et al. 2012a) possibly caused by exhaustion of energy reserves or cardiorespiratory collapse since they have a smaller ventricular mass (Sandblom et al. 2009) and thus may have to rely more heavily on anaerobic burst swimming to negotiate fast flows (Burnett et al. 2014). The fact that I did not observe behavioural differences between sexes is likely because this short study reach did not offer significant enough energetic challenges or obstacles (e.g. high discharge, temperatures frequently above optimal) necessitating females to either attempt to conserve energy by slowing or delaying migration, or, requiring large levels of burst swimming. 2.4.1 Management implications It has long been recognized that the environmental conditions created by the Seton powerhouse and fluctuating natal water concentrations from the Seton River could cause significant delays for upriver migrating sockeye salmon which could have serious consequences for successful arrival at the Seton Dam and passage through its fishway (Fretwell 1989). Water preference experiments (‘Y-maze’ behaviour studies) conducted 29 in the late 1970s and early 1980s demonstrated that adult sockeye were capable of discriminating between Seton River water and Cayoosh Creek water, and that olfaction was the primary sensory mechanism used by fish to select water source differences for homing through the hydrosystem (Fretwell 1989). The threshold of sensitivity to the proportion of Cayoosh Creek water mixed with Seton River water differed between GC fish (20% Cayoosh Creek) and PC fish (10% Cayoosh Creek). These studies provided in-situ experimental evidence suggesting that the powerhouse tailrace delay issue could be mitigated by controlling Seton River discharge to maintain concentrations equal to or greater than the threshold of sensitivity exhibited by the two populations (Fretwell 1989). Based on these results, BC Hydro built a diversion dam on Cayoosh Creek so they could attempt to maintain the Cayoosh Creek component of the Seton River discharge to 20% and 10% during the upstream migration period for GC and PC sockeye runs, respectively. Recent Y-maze experiments with both populations have re-confirmed these olfactory sensitivity thresholds (Casselman et al. 2015). In the present study, natal water concentrations generally exceeded threshold targets except for a one-week period during the 2013 GC migration (Figure 2.2). Despite a large range of natal water concentrations experienced by GC sockeye, I could not find any effects of this factor on delay or other behaviours and these fish seemed to have little difficulty rapidly homing into the Seton River. On the other hand, even though threshold natal water concentrations were exceeded for PC migrants, relatively lower natal water concentrations enhanced tailrace delay and slowed migration. Given the apparent olfactory sensitivity displayed by PC fish, the 10% threshold may need to be re-visited by managers. Fraser River water temperatures can have an incremental effect on sockeye migration success, and affect the condition of fish when they arrive at the Seton system. Peak summer water temperatures in the Fraser River have increased > 2C in the past 60 years, with recent years exhibiting record high levels and all climate models indicating even warmer peak temperatures in near future years (Patterson et al. 2007; Ferrari et al. 2007). GC fish now typically encounter Fraser River temperatures exceeding 18 – 19 C, temperatures that are extremely stressful, with prolonged exposure causing migration mortality (Eliason et al. 2011). Even though GC fish did not delay for long periods in the powerhouse tailrace enroute to the Seton River, the fact that they are increasingly 30 encountering thermally stressful migrations prior to reaching the Seton hydrosystem means that any small delay caused by the powerhouse or by varying olfactory cues from the Seton River could be devastating if delays expose fish to additional thermal stress. Thus managers should continue to strive to achieve or exceed the natal water targets established for this population. 31 2.5 Chapter 2 Tables Table 2.1 – AICc model selection statistics for generalized linear models and linear models predicting (1) the number of forays made into the Seton powerhouse, (2) the amount of wandering in the Fraser River between the release site and the Seton-Fraser confluence, and (3) the total amount of migration delay incurred by individuals in the Seton powerhouse tailrace (3) for Gates Creek and Portage Creek sockeye salmon. Abbreviations used for model variables include: (FRT) Fraser River temperature, (PHT) Seton powerhouse temperature, (Tdiff) temperature differential between the Fraser River and Seton powerhouse, (NW) natal water concentration, and (TD) tagging date. In the model 3 for the Portage Creek stock, TD is shown with FRT in parentheses to indicate that tagging date was substituted for Fraser River temperature. Note, ΔAICc is the difference in AICc values between model i and the top model in the candidate set. Models are ranked from lowest to highest ΔAICc, and by wi – the probability that a given model is the best in the 95% confidence set. Adj R2 is an estimate of the proportion of variance explained by each model, adjusted by the number of explanatory variables; this is only shown for linear models as generalized linear model fits were evaluated by Chi-square tests (see methods; Chapter 2). 32 Response variable and model log Lik AICc Δ AICc wi Adj R2 A) Powerhouse forays Gates (N=255) Glucose -421.42 848.94 0.00 0.03 - Glucose + PHT -420.67 849.50 0.56 0.02 - Glucose +Tdiff -420.68 849.52 0.57 0.02 - Intercept -422.82 849.69 0.74 0.02 - Tdiff -421.80 849.70 0.75 0.02 - Testosterone + PHT -420.84 849.84 0.89 0.02 - Testosterone -421.90 849.90 0.96 0.02 - Glucose + testosterone -420.87 849.91 0.97 0.02 - Glucose + testosterone + PHT -419.84 849.91 0.97 0.02 - Glucose + NW -420.89 849.94 1.00 0.02 - PHT -422.12 850.33 1.39 0.01 - Testosterone + Tdiff -421.18 850.52 1.57 0.01 - Glucose + Tdiff +NW -420.22 850.68 1.73 0.01 - Glucose + Testosterone + Tdiff -420.30 850.85 1.91 0.01 - Glucose + FRT + Tdiff -420.35 850.94 1.99 0.01 - Glucose + FRT + Tdiff -420.35 850.94 1.99 0.01 - Glucose + FRT + PHT -420.35 850.94 1.99 0.01 - Glucose + PHT + Tdiff -420.35 850.94 1.99 0.01 - 33 Response variable and model log Lik AICc Δ AICc wi Adj R2 A) Powerhouse forays Portage (N=90) TD + PHT + Tdiff + NW -204.36 421.73 0.00 0.06 - PHT + NW -206.68 421.84 0.11 0.06 - PHT + Tdiff + NW -205.56 421.84 0.11 0.06 - Testosterone + PHT + NW -205.65 422.02 0.29 0.06 - Sex + PHT + NW -206.00 422.72 0.99 0.04 - Glucose + PHT + NW -206.02 422.75 1.02 0.04 - Testosterone + TD + PHT + Tdiff + NW -203.78 422.93 1.20 0.04 - Testosterone + PHT + Tdiff + NW -204.99 422.99 1.26 0.03 - Glucose + PHT + Tdiff + NW -205.14 423.29 1.56 0.03 - Sex + PHT + Tdiff + NW -205.20 423.40 1.67 0.03 - Glucose + TD + PHT + Tdiff + NW -204.08 423.52 1.79 0.03 - Glucose + Testosterone + PHT + NW -205.27 423.55 1.82 0.03 - Sex + TD + PHT + Tdiff + NW -204.17 423.70 1.97 0.02 - 34 Response variable and model log Lik AICc Δ AICc wi Adj R2 B) Wandering Gates (N=235) Sex + Lactate + PHT -142.09 294.44 0.00 0.05 - Sex + Lactate + Testosterone + PHT -141.95 296.27 1.83 0.02 - Lactate + PHT -144.06 296.29 1.84 0.02 - Sex + Lactate + FRT + PHT -142.01 296.40 1.95 0.02 - Sex + Lactate + PHT + Tdiff -142.01 296.40 1.95 0.02 - Sex + Lactate + FRT + Tdiff -142.01 296.40 1.95 0.02 - Sex + Lactate + FRT + Tdiff -142.01 296.40 1.95 0.02 - 35 Response variable and model log Lik AICc Δ AICc wi Adj R2 B) Wandering Portage (N=78) Sex + Lactate + FRT + NW -73.44 160.07 0.00 0.03 - Lactate + FRT + NW -74.72 160.28 0.21 0.03 - FRT + NW -75.96 160.47 0.40 0.03 - Lactate + Testosterone + FRT + NW -73.94 161.07 1.00 0.02 - Lactate + FRT -76.34 161.22 1.15 0.02 - Lactate -77.52 161.36 1.29 0.02 - Sex + Lactate + FRT + Tdiff + NW -72.91 161.42 1.35 0.02 - Sex + Lactate + FRT + Tdiff + NW -72.91 161.42 1.35 0.02 - Sex + Lactate + FRT + PHT + NW -72.91 161.42 1.35 0.02 - Sex + Lactate + PHT + Tdiff + NW -72.91 161.42 1.35 0.02 - Lactate + FRT + Tdiff + NW -74.31 161.81 1.74 0.01 - Lactate + FRT + Tdiff + NW -74.31 161.81 1.74 0.01 - Lactate + FRT + PHT + NW -74.31 161.81 1.74 0.01 - Lactate + PHT + Tdiff + NW -74.31 161.81 1.74 0.01 - Sex + Lactate + PHT + NW -74.32 161.83 1.77 0.01 - Sex + Lactate -76.65 161.85 1.78 0.01 - Sex + FRT + NW -75.52 161.86 1.80 0.01 - FRT + Tdiff + NW -75.59 162.02 1.95 0.01 - FRT + Tdiff + NW -75.59 162.02 1.95 0.01 - FRT + PHT + NW -75.59 162.02 1.95 0.01 - 36 Response variable and model log Lik AICc Δ AICc wi Adj R2 C) Powerhouse delay Gates (N=256) Glucose + Tdiff + NW -481.06 972.35 0.00 0.03 0.05 Glucose + Testosterone + Tdiff + NW -480.02 972.38 0.03 0.03 0.06 Glucose + Tdiff -482.40 972.97 0.61 0.02 0.04 Glucose + NW -482.49 973.13 0.78 0.02 0.04 Glucose + PHT -482.49 973.14 0.79 0.02 0.04 Glucose + PHT + NW -481.68 973.60 1.24 0.02 0.04 Glucose + Testosterone + Tdiff -481.69 973.62 1.27 0.02 0.04 Glucose + Testosterone + NW -481.77 973.79 1.44 0.02 0.04 Glucose + FRT + Tdiff -481.79 973.82 1.46 0.02 0.04 Glucose + FRT + PHT -481.79 973.82 1.46 0.02 0.04 Glucose + PHT + Tdiff -481.79 973.82 1.46 0.02 0.04 Glucose + FRT + Tdiff -481.79 973.82 1.46 0.02 0.04 Glucose + FRT + Tdiff + NW -480.86 974.05 1.70 0.01 0.05 Glucose + FRT + PHT + NW -480.86 974.05 1.70 0.01 0.05 Glucose + PHT + Tdiff + NW -480.86 974.05 1.70 0.01 0.05 Glucose + FRT + Tdiff + NW -480.86 974.05 1.70 0.01 0.05 Glucose + Lactate + Tdiff + NW -480.87 974.07 1.72 0.01 0.05 Glucose -484.01 974.12 1.77 0.01 0.03 Glucose + Tdiff + NW -480.95 974.24 1.88 0.01 0.05 Glucose + Lactate + NW -482.02 974.28 1.92 0.01 0.04 37 Response variable and model log Lik AICc Δ AICc wi Adj R2 C) Powerhouse delay Portage (N=90) PHT + NW -163.15 334.76 0.00 0.07 0.23 PHT + NW + Glucose -162.08 334.88 0.12 0.06 0.24 PHT + NW + TD + Tdiff -161.20 335.41 0.65 0.05 0.26 PHT + NW + TD -162.56 335.83 1.07 0.04 0.24 PHT + NW + Glucose + TD -161.53 336.07 1.30 0.04 0.25 PHT + NW + Glucose + TD + Tdiff -160.46 336.29 1.53 0.03 0.27 PHT + NW + Glucose + Testosterone -161.83 336.67 1.90 0.03 0.25 PHT + NW + Glucose + Lactate -161.87 336.75 1.98 0.03 0.25 38 Table 2.2 – Mean ± SD and range of fork length (cm), glucose (mmol L-1), lactate (mmol L-1), and testosterone (ng ml-1) for female (♀) and male (♂) Gates Creek and Portage Creek sockeye salmon by year. Note, preliminary analyses of physiological variables from the three datasets used in behavioural models 1, 2, and 3 yielded similar values; this table reports the data for stress and maturity indices used in model 3 as an example. Values for fork length, glucose, and lactate are pooled because preliminary analyses indicated there were not differences between sexes and populations. * Indicates significant difference at P < 0.05 in plasma testosterone between sexes and populations as assessed by two-way ANOVA. Gates Creek Portage Creek 2013 (n = 111) 2014 (n = 145) 2013 (n = 19) 2014 (n = 71) Fork length (cm) 58.38 ± 2.9 (49 – 67.5) 60.1 ± 4.0 (51 – 69.0) 57.3 ± 1.8 (54.5 – 61.0) 59.7 ± 2.6 (50.5 – 67.0) Glucose (mmol L-1) 5.5 ± 1.4 (2.9 – 13.7) 4.9 ± 1.2 (2.5 – 10.1) 4.0 ± 1.0 (2.2 – 7.0) 4.8 ± 0.9 (2.6 – 8.0) Lactate (mmol L-1) 4.2 ± 3.4 (0.8 – 16.1) 4.5 ± 2.4 (0.8 – 11.0) 4.0 ± 2.1 (0.9 – 8.4) 3.4 ± 1.4 (1.1 – 8.5) Testosterone (ng ml-1) ♀* 82.0 ± 97.4 (1.1 – 398.1) (n = 63) 53.32 ± 56.5 (2.1 – 385.6) (n = 83) 148.8 ± 108.9 (18.3 – 365.2) (n = 13) 138.4 ± 73.3 (13.3 – 306.8) (n = 41) ♂* 16.7 ± 17.7 (0.9 – 83.6) (n = 48) 8.0 ± 6.63 (1.2 – 31.1) (n = 62) 88.3 ± 66.7 (33.5 – 216.2) (n = 6) 16.57 ± 10.0 (5.7 – 42.9) (n = 30) 39 2.6 Chapter 2 Figures Figure 2.1 – Study area, located in the Seton-Anderson watershed in southwestern British Columbia, Canada (inset). Sockeye salmon (Oncorhynchus nerka) capture, release, and radio-telemetry sites, along with temperature and conductivity monitoring locations are indicated by the legend and map. The location of the Seton powerhouse tailrace and the visible extent of the powerhouse discharge plume of natal Seton Lake water is shown extending into the Fraser River, as is the extent of the Seton River plume that fluctuates in its concentration of natal water given relative contributions of Cayoosh Creek discharge. 40 Temperature (°C)6810121416182022 (a)60708090100Natal Water (%)Aug 01 Aug 08 Aug 15 Aug 22 Aug 29 Sep 05 Sep 12 Sep 19 Sep 26 Oct 03 Oct 10 Oct 17 Oct 24 Oct 31(b)GC tagging PC taggingDateFigure 2.2 - Mean daily Fraser River (solid lines) and Seton powerhouse (dashed lines) temperatures (panel a), and natal water concentration of the Seton River and its plume (panel b); grey lines indicate 2013 measurements, black lines indicate 2014 measurements. Shaded grey boxes represent the periods in which fish from each population were tagged. Dashed red lines represent the current natal water concentration targets of 80% and 90% during the Gates Creek and Portage Creek sockeye salmon migrations, respectively, and are approximately proportional to the migration timing of each population through the hydrosystem. 41 1 3 5 7 9 11 13 15 170.00.10.20.30.40.50.6Proportion(a)1 3 5 7 9 11 13 15 17(b)−0.16 0.04 0.19NWTdiffPHTFRT (TD)TestosteroneLactateGlucoseSex (c)●●−1.1 −0.7 −0.3 0.1(d)Powerhouse foraysStandardized coefficientsFigure 2.3 – Panels a & b – Histograms of the number of forays female (grey bars) and male (black bars) Gates Creek (a) and Portage Creek (b) sockeye salmon made into the Seton powerhouse as a proportion of all the individuals included in models predicting this behaviour. In panel b, 1 female that made 24 forays was removed from the histogram for clarity. Panels c & d – Model-averaged standardized coefficient estimates for models predicting the number of forays Gates Creek (c) and Portage Creek (d) sockeye salmon made into the Seton powerhouse. Coefficient estimates with 95% confidence intervals that do not cross zero are highlighted by solid black circles. Vertical dashed line indicates the coefficient value of zero. Abbreviations for predictor variables include: (FRT) Fraser River temperature, (PHT) Seton powerhouse temperature, (Tdiff) temperature differential between the Fraser River and Seton powerhouse, and (NW) natal water concentration. FRT is shown with TD in parentheses to indicate that tagging date was substituted for Fraser River temperature in the Portage Creek foray model (panel d only). Note the differences in x-axes scales between panels. 42 0 1 2 3 4 50.00.20.40.60.8Proportion(a)0 1 2 3 4 5(b)−0.8 −0.2 0.4 1.0NWTdiffPHTFRTTestosteroneLactateGlucoseSex (c)Wandering●−2.2 −1.0 0.2(d)Wandering eventsStandardized coefficientsFigure 2.4 – Panels a & b – Histograms of wandering (the number of back-and-forth movements) in the Fraser River between the release site and the Seton-Fraser confluence for female (grey bars) and male (black bars) Gates Creek (a) and Portage Creek (b) sockeye salmon shown as a proportion of all individuals included in the models predicting this behaviour for each population. In panel a, 1 female that wandered 9 times was removed from the histogram for clarity. Panels c & d – Model-averaged standardized coefficient estimates for models predicting wandering for Gates Creek (c) and Portage Creek (d) sockeye salmon. Coefficient estimates with 95% confidence intervals that do not include zero are highlighted by solid black circles. Vertical dashed line indicates the coefficient value of zero. Abbreviations for predictor variables include: (FRT) Fraser River temperature, (PHT) Seton powerhouse temperature, (Tdiff) temperature differential between the Fraser River and Seton powerhouse, and (NW) natal water concentration. Note the differences in x-axes scales between panels. 43 51525354555657585Gates Creek Portage CreekPowerhouse delay (hrs) (a)●−0.6 0.0 0.3 0.6NWTdiffPHTFRT (TD)TestosteroneLactateGlucoseSex (b)●●−3.0 −2.0 −1.0 0.0 1.0(c)Standardized coefficientsFigure 2.5 – Panel a – Beanplots of total migration delay in the Seton powerhouse tailrace for female (black beans) and male (grey beans) Gates Creek and Portage Creek sockeye salmon used in models predicting this behaviour. Shaded polygons represent the distribution of individual delay times (small horizontal lines) and bold horizontal lines represent means. Panel b & c – Model-averaged standardized coefficient estimates from models predicting the total amount of migration delay incurred by Gates Creek (b) and Portage Creek (c) sockeye salmon in the Seton powerhouse tailrace. Coefficient estimates with 95% confidence intervals that do not include zero are highlighted by solid black circles. Vertical dashed line indicates the coefficient value of zero. Abbreviations for model variables include: (FRT) Fraser River temperature, (PHT) Seton powerhouse temperature, (Tdiff) temperature differential between the Fraser River and Seton powerhouse, and (NW) natal water concentration. FRT is shown with TD in parentheses to indicate that tagging date was substituted for Fraser River Temperature in the Portage Creek delay model (panel c only). Note the differences in x-axes scales between panels. 44 Chapter 3: Migration behaviour and the cumulative effects of hydrosystem passage in relation to natal water cues, temperature, and individual characteristics of wild up-river migrating sockeye salmon 3.1 Introduction Reproductive migrations of anadromous fishes are some of the most complex and fascinating phenomenon in nature (Dingle and Drake 2007). Migrations can be particularly arduous for semelparous adult Pacific salmon (Oncorhynchus spp.), as successful spawning requires that individuals migrate upriver often long distances to natal areas following imprinted olfactory cues (reviewed in Bett and Hinch 2015) while coping with environmental conditions (e.g. high temperatures and flows) that can impart physiological and energetic hardships (Hinch et al. 2006; Burnett et al. 2014). Adults also cease feeding prior to entering freshwater, therefore, migration, gonad development, and spawning is carried out solely on energy reserves acquired while at sea (Brett 1971). Thus any encounters with environmental features [e.g. high temperatures (Martins et al. 2011), fisheries (Donaldson et al. 2011), dams (Keefer et al. 2004b)] that could hinder or delay migrations could result in enroute mortality or affect spawning success (Burnett et al. 2016). Many freshwater corridors utilized by migratory salmon have been impounded or diverted to create hydroelectric systems (hydrosystems) (see reviews in Waples et al. 2008; Thorstad et al. 2008). Hydrosystems often require that fish pass turbine outlets and ascend multiple dams in order to reach natal spawning sites. Migratory delays incurred at isolated turbine outlet flows (e.g. Thorstad et al. 2003a) or in dam tailraces (e.g. Burnett et al. 2013) can lead to failure passing dams or to post-passage mortality. Caudill et al. (2007) found that adult Chinook salmon (O. Tshawytscha) and steelhead (O. mykiss) that do not survive to natal spawning tributaries are often slower at passing dams and reservoirs throughout the Columbia River hydrosystem compared to those that do survive. Even single dam passage events can have delayed carry-over effects, and negatively impact spawning success (Burnett et al. 2016). Metabolic and cardiac stress, and energy exhaustion are associated with migration through hydraulically challenging environments, and have been proposed as physiological mechanisms associated with immediate or 45 delayed mortality arising from dam passage (Burnett et al. 2013; 2014). However, migratory cues are also often altered in hydrosystems, and varying natal water concentrations can lead to delayed migrations. For example, Chinook salmon exhibited searching behaviours in reaches of the Columbia River hydrosystem where natal water was highly-diffuse (Keefer et al. 2008a), and Fraser River sockeye salmon (O. nerka) delayed migration in a power station outlet and wandered more when homestream cues were less available (Chapter 2). However, there has been comparatively little empirical research into how encounters with changing flow composition (e.g. olfactory cue changes), discharge, and/or varying temperatures are associated with migration behaviour and successful passage through an entire hydrosystem (Naughton et al. 2005). The physiological condition and sex of individuals also plays an important role in the behaviour and passage success of upriver migrating adult salmon (Hinch et al. 2006). In upriver migrating Fraser River sockeye salmon, reduced energy reserves are associated with faster migration rates (Hanson et al. 2008) yet increased difficulty in passing through challenging flows (Hinch and Rand 1998), which have both been linked to reduced survival (Young et al. 2006; Burnett et al. 2014). Female sockeye salmon devote significantly more energy to gonad development (Crossin et al. 2008), exhibit slower migration rates (Hanson et al. 2008), but have less success passing hydraulically challenging areas (Hinch and Bratty 2000; Burnett et al. 2013) and often have lower survival than males (Martins et al. 2012a). The Fraser River, like many migration corridors the world over, has also been warming in recent years (Patterson et al. 2007; Fenkes et al. 2015), and migrants now routinely encounter temperatures that exceed population-specific thermal optimum (Eliason et al. 2011; Eliason and Farrell 2015). For adult sockeye salmon, exposure to high temperatures can deplete energy reserves (Hinch et al. 2006), increase physiological stress (Young et al. 2006), and increase disease development (Wagner et al. 2005; Miller et al. 2011), any or all of which can lead to changes in migratory behaviour and increased en route mortality (Hinch et al. 2012). How physiological condition and sex influences behaviour and passage success through a regulated river with a dynamic and changing landscape of natal water cues as well as temperatures and discharge has not yet been examined. Nor have the cumulative effects of sex, physiological condition, and the environmental conditions encountered during 46 hydrosystem passage ever been explored in relation to post-hydrosystem passage survival to natal spawning grounds. In this study, I used non-invasive biopsy and radio-tracking to examine the upstream migration behaviour and post-hydrosystem passage survival of adult sockeye salmon as they passed from a large mainstem river (Fraser River) into a small tributary with regulated flows (Seton River) before ascending a dam while enroute to spawning grounds. The concentration of natal water cues in the Seton River can vary daily as a result of operations at two diversion dams, and in turn cause the availability and strength of natal water cues entering the mainstem Fraser River to also vary. I first assessed the relative roles that fish traits (sex, energy content, ability to reach spawning grounds), flow composition (amount of natal water), water temperature, and dam operations (flow patterns and discharge) can have on migration behaviour (measured by passage time) during passage through three specific reaches of the Seton hydrosystem. I then examined how the cumulative effects of environmental conditions encountered during hydrosystem passage and individual fish traits affected post-hydrosystem passage survival to natal spawning sites. As ultimately unsuccessful migrants, females, and fish with greater energy reserves are all known to migrate more slowly during adult return migrations, I predicted that adult sockeye which were destined not to survive to natal sites would be slower passing through each reach of the hydrosystem, and that females and migrants with higher energy reserves would be the slowest. Given reduced and diffuse concentrations of natal water can lead to extensive migration delays and increased wandering in mainstem rivers, I expected that when natal water concentration in the tributary was reduced, migrants would be slower passing from the mainstem into the tributary, and while migrating up to the dam. I further predicated that because of the increased metabolic and cardiac stress associated with elevated water temperature, passage times through the hydrosystem would be decreased in migrants that encountered warmer temperatures. Lastly, because female migrants and those with lower energy reserves also often exhibit higher en route mortality, I predicted that females and fish with lower energy would be more likely succumb to the cumulative effects of hydrosystem passage and exhibit higher post-hydrosystem passage mortality, as would 47 those who were slower passing the entire hydrosystem given the energetic costs and increased temperature exposure associated with prolonged migrations. 3.2 Methods 3.2.1 Study system The Seton River hydrosystem, located in the Seton-Anderson watershed in southwestern British Columbia provides a unique locale for examining the cumulative influences of flow composition, temperature and discharge on migrating adult Pacific salmon. The Seton hydrosystem is one of the few regulated tributaries of the Fraser River – the largest producer of wild salmon in Canada – and consists of a complex network of reservoirs, inter-basin diversions, and power generating stations operated by BC Hydro (Casselman 2015). Adult sockeye salmon first encounter the hydrosystem at the impassable Seton powerhouse ~ 340 km from the Pacific Ocean on the western bank of the Fraser River near the confluence of the Seton and Fraser Rivers (Figure 3.1). Adult sockeye salmon are known to delay their migrations for up to 8 days at this location because of the discharge of natal Seton Lake water in the tailrace of this facility (see Fretwell 1989; Chapter 2). Upstream 1.5 km from the powerhouse is the Seton-Fraser confluence. The 5 km long Seton River (22.7 m3s-1 mean annual discharge) is regulated by the Seton Dam and is the only migration corridor for adult salmon returning to the Seton-Anderson watershed. Cayoosh Creek (13.9 m3s-1 mean annual discharge) is the only tributary of the Seton River; it drains a separate watershed (Figure 3.1). At times of increased discharge from Cayoosh Creek, which occurs during operational changes at the Seton Dam, Walden North Dam, or during large precipitation events, flows from this tributary can ‘dilute the olfactory signature’ of the Seton River (Fretwell 1989) and thereby reduce the concentration of natal water cues in the lower Seton River and its plume relative to the pure natal lake water discharged from the powerhouse. To navigate through the lower hydrosystem and reach the Seton Dam, migrating adult salmon must utilize relatively low concentrations of natal water in the lower Seton River and its plume to guide homing past the powerhouse and enter the Seton River before arriving at the Seton Dam (Figure 3.1; Chapter 2). Upon arriving at the Seton Dam, fish must then negotiate high volumes of turbulent flows in the dam tailrace and find the entrance of a 32-pool, 107 m long vertical slot fishway in order to successfully pass the dam and 48 subsequently travel through one to two natal lakes to reach spawning sites (Figure 3.1). Passage through the Seton Dam tailrace and fishway can require the use of anaerobic swimming, which can lead to post-passage (i.e. latent) mortality if anaerobic swimming levels are high enough, which can reduce spawning success in migrants that are able to reach spawning grounds (Burnett et al. 2014; 2016). To date, passage through this hydrosystem has only been studied at discrete locales and reaches, while a whole-system approach has yet to be explored. Coho salmon (O. kisutch), Chinook salmon (O. Tshawytscha), pink salmon (O. gorbuscha), and steelhead (O. mykiss) all return the Seton-Anderson watershed. This chapter of my thesis, however, focused only on the distinct ‘summer’ Gates Creek (hereinafter GC) and ‘late-run’ Portage Creek (hereinafter PC) populations of Fraser River sockeye salmon that spawn and rear in the upper reaches of the watershed. The GC population typically begins migrating into the hydrosystem in late July with ~ 30,000 adults annually passing the Seton Dam (Casselman et al. 2015) that must travel an additional ~ 55 km through Seton and Anderson Lakes to reach spawning sites at Gates Creek. The PC population typically begins migrating into the system at the end of September, with ~ 35,000 adults annually that pass the dam (Casselman et al. 2015) and travel an additional ~ 20 km through Seton Lake to reach spawning sites in Portage Creek. Studies conducted throughout the 1970’s and early 1980’s by the International Pacific Salmon Fisheries Commission (IPSFC) indicated that delays at the powerhouse and interrupted migrations into and through the lower Seton River could be mitigated by maintaining natal water concentrations in the lower Seton River at or above 80% and 90% during the GC and PC migrations, respectively (Fretwell 1989; Chapter 2). BC Hydro currently operates the hydrosystem with measures in place to maintain these natal water concentrations. However, large fluctuations still occur, and migration delays at the powerhouse followed by wandering in the Fraser River between the powerhouse and the Seton-Fraser confluence are still observed in both populations (Roscoe and Hinch 2008). 3.2.2 Fish capture and tagging All fish were captured for radio-tagging using a full-spanning fence and trap at a site in the Seton River 200 m downstream from the Seton Dam (Figure 3.1). I targeted 49 only GC and PC fish for tagging using readings of high gross somatic energy (GSE) concentrations (measured by a hand-held microwave energy meter – Fatmeter model 692; Distell Inc., West Lothian, Scotland, UK; see Crossin and Hinch 2005) to identify and remove non-target fish. An earlier study which compared DNA population identification to Fatmeter output from fish captured at this fence revealed that high energy fish were strays that had entered the Seton River but whose spawning grounds were several 100 km further up the Fraser River (Casselman et al. 2015). GSE was also recorded and used as an index of energy content for each fish (Crossin et al. 2004; Crossin and Hinch 2005). I used secondary sexual characteristics (e.g. exaggerated kype and dorsal hump of males; rounder snout, wider ventral profile, and engorged vent of females) to maintain approximate 50:50 sex ratios while tagging and later confirmed sex by analysis of 17-β estradiol and testosterone concentrations in blood plasma. Fish with severe injuries (e.g. lacerations exposing the coelomic cavity or skeletal system) or those showing signs of disease (e.g. extensive fungus cover on body or gills) were not selected for tagging. Radio transmitters [(43 mm length × 16 mm diameter; 15.2 g in air); Sigma Eight, Newmarket, Ontario, Canada] transmitting a unique digital identification code every 3 s were gastrically inserted into individual fish. Half-duplex passive integrated transponder (PIT) tags (32 mm X 3.65 mm; Texas Instruments, Dallas) were also implanted into the dorsal musculature, with individually labeled spaghetti tags (Floy Manufacturing, Seattle) attached externally through the dorsum of each fish. A tissue sample from the adipose fin and a 1.5 ml blood sample from the caudal vasculature were taken for population identification and physiological assays, respectively. Average tagging and biopsy time was approximately four minutes (± 1.3 min; SD), and fish were not sedated to eliminate the consumption of anesthetics in traditional First Nations fisheries further upstream. All procedures followed the methods described in greater detail in (Roscoe et al. 2010b, Burnett et al. 2014) and were conducted in accordance with the guidelines of the Canadian Council of Animal Care administered by the University of British Columbia (A11-0125). In total, I radio-tagged and monitored the movements of 500 sockeye salmon through the summer and fall of 2013 and 2014. Specifically, 117 GC sockeye salmon 50 were tagged between 5 August and 2 September in 2013, and 166 were tagged between 5 August and 7 September 2014. In 2013, PC sockeye salmon co-migrated with approximately 0.8 million pink salmon returning to the Seton River. This made targeting sockeye salmon with the fence and trap particularly onerous – to the extent that only 24 sockeye were tagged from 3 – 9 October 2013. However, no pink salmon were present in the Seton River in 2014, enabling 193 PC sockeye to be tagged from 27 September to 9 October. Results from analysis of 17-β estradiol and testosterone concentrations in blood plasma confirmed that 55% (157 of 283) of tagged GC fish were females and 45% (126 of 238) were males, while 51% (110 of 217) of tagged PC fish were females and 49% (107 of 217) were males. All fish collection occurred from approximately 0900 – 1700 daily, with the fence remaining open for approximately 16 hours each day to allow for unimpeded migration. In order to examine migrants as they passed through the entire lower hydrosystem, tagged fish were transported by truck downstream of the powerhouse for release into the Fraser River. Groups of ten to twelve individuals (approximate 50:50 sex ratio) were moved via a 1000 L insulated and oxygenated transport tank, and released within 30 min (± 10 min; SD) of tagging, 1.5 km downstream from the powerhouse on the western bank of the Fraser River (Figure 3.1). It was not possible to capture Seton sockeye in the Fraser River mainstem. There are limited recreational and traditional First Nations fisheries in the Fraser River near the study area and near terminal spawning sites at GC and PC. Any fish that were reported caught in these fisheries were excluded from this study, a priori. 3.2.3 Telemetry monitoring and fate assignment I utilized combination of fixed SRX 400 (Lotek Wireless, Inc., Newmarket, Ontario, Canada) and Orion (Sigma Eight, Inc., Newmarket, Ontario, Canada) radio-receivers fitted with aerial Yagi antennas as the primary means of assessing the movements and survival of migrants (Figure 3.1). Pass-through PIT antennas at the entrance and exit of the Seton Dam fishway and at Gates Creek provided additional backup for fixed radio sites and established the exact time of dam passage (see Burnett et al. 2014 for more details regarding PIT antennas). All telemetry receivers operated > 85% of the time, providing sufficient ability to monitor fish movements and survival (Appendix 1). Mobile radio tracking on foot and by truck was conducted throughout the 51 hydrosystem approximately every 2 weeks from late August to late November to establish the location of fish in areas not covered by fixed receiver sites. Raw telemetry data were filtered to remove false detections (e.g. non-study tag ID’s, detections recorded before release, spatio-temporally implausible records) and generate migration histories, which were then individually reviewed and used to calculate hydrosystem passage times and classify the fate of each fish. 3.2.4 Survival estimates and passage time calculations Ultimately successful migrants included sockeye salmon with final telemetry records at natal spawning sites in either GC or PC; unsuccessful migrants included fish with final records outside of natal sites. I also used the telemetry data to estimate success and passage times through three specific reaches of the lower hydrosystem. Fish were classified as having successfully entered the Seton River if they were detected at any two receivers above the Seton-Fraser confluence. I calculated passage time through this reach – from release to Seton River entry (hereinafter ‘reach 1’) – from the date and time of release (+30 min for regaining orientation following transport) to the last detection at the Seton-Fraser confluence. Some fish wander back-and-forth in the Fraser River between the release site and the Seton-Fraser confluence due to conflicting olfactory cues and temperatures at the Seton powerhouse (Chapter 2), and this estimate of passage time into the Seton River accounted for any additional time that may have accumulated during wandering bouts. Of the fish that successfully entered the Seton River, those that were detected by receivers in the Seton Dam tailrace were deemed to have arrived at the Seton Dam. Passage times from Seton River entry to arrival at the Seton Dam (hereinafter ‘reach 2’) were calculated from river entry to the first detection in the Seton Dam tailrace. Of the fish that survived to the Seton Dam, those with detections at the fishway exit were deemed to have successfully passed the dam. Dam passage times (hereinafter ‘reach 3’) were calculated from the first detection in the Seton Dam tailrace to the last detection at the fishway entrance PIT antenna that corresponded to successful dam passage. Some fish migrate up to the Seton Dam tailrace or enter the fishway but return downriver for some time before returning to the dam and successfully passing, a behaviour know to be reflective of searching for the fishway entrance or unfavorable passage conditions (Burnett et al. 2013, 2014), and dam passage times would have reflected any additional 52 delays that may have occurred because of this behaviour. I am particularly confident in the accuracy of these reach-specific passages times because all automated time calculations were confirmed by my detailed visual review of migration histories for each individual fish. 3.2.5 Environmental data I used hourly measurements from the Water Survey of Canada (available from http://wateroffice.ec.gc.ca) to monitor discharge in Cayoosh Creek (Station No. 08ME002) and the upper Seton River below the Seton Dam (Station No. 08ME003). Because there are no hourly measurements available for discharge of the lower Seton River below its confluence with Cayoosh Creek, I calculated this parameter as the sum of upper Seton River discharge and Cayoosh Creek discharge, plus a constant 2.1 m3s-1 of Seton Lake water diverted to an artificial spawning channel that is returned into the lower Seton River before the Seton-Fraser confluence. The concentration of natal water in the lower Seton River and its plume was then calculated using discharges in the following equation: 𝑁𝑎𝑡𝑎𝑙 𝑤𝑎𝑡𝑒𝑟 (%)= 100 − (100 × (𝐶𝑎𝑦𝑜𝑜𝑠ℎ 𝐶𝑟𝑒𝑒𝑘𝑈𝑝𝑝𝑒𝑟 𝑆𝑒𝑡𝑜𝑛 𝑅𝑖𝑣𝑒𝑟 + 𝐶𝑎𝑦𝑜𝑜𝑠ℎ 𝐶𝑟𝑒𝑒𝑘 + 𝑆𝑝𝑎𝑤𝑛𝑖𝑛𝑔 𝐶ℎ𝑎𝑛𝑛𝑒𝑙)) To confirm differences in the water chemistry between natal Seton River and non-natal Cayoosh Creek water, which could reflect differences in olfactory cues, I collected daily conductivity measurements from Cayoosh Creek, the upper and lower Seton River, and the powerhouse using a hand held YSI Pro30 conductivity meter (YSI Inc., Yellow Springs, OH, USA) (Appendix 1). Conductivity has previously been used to distinguish between water sources utilized by other anadromous fishes that use olfaction to guide homing (e.g. Leduc et al. 2010; Vrieze et al. 2011). Hourly water temperatures were recorded at locations in the Fraser River, the Seton River, Cayoosh Creek, and the Seton Dam using TidbiT v2 data loggers (Onset HOBO data loggers, Bourn, MA) (Fig. 1). 3.2.6 Statistical analysis In Chapter 2, I found limited evidence for stress indices (i.e. plasma glucose and lactate) or maturity levels (i.e. plasma testosterone) affecting attraction to the powerhouse 53 or behaviour prior to entering the Seton River. These physiological indices also had very limited effects as covariates in preliminary models for this chapter. As such, I removed stress and maturity effects from the analyses described below, and subsequently examined the relationships between hydrosystem passage times and fate with environmental experience and individual energy content (GSE). Hydrosystem passage At the finest reach scale analysis of behaviour, I used time-to-event proportional hazards regression (hereinafter TTE) (Castro-Santo and Haro 2003; Kleinbaum et al. 2012) to investigate how fish traits (e.g. sex, GSE, fate) and the environmental conditions (e.g. temperature, natal water, dam discharge) encountered during passage were related to passage times through the three specific reaches of the hydrosystem described previously. TTE analyses model the time until an event occurs as a ‘hazard rate’ which describes the instantaneous probability of an event occurring (river entry, dam arrival/passage) for an individual at time t given (i) the event had not occurred prior to the start of t, and (ii) a set of covariate predictors describing environmental conditions and fish traits that may be constant (e.g. sex, GSE, fate) or time-varying (e.g. natal water concentration, river temperature, dam discharge) (Kleinbaum et al. 2012). Hazard rates are expressed as odds ratios comparing the probability of an event occurring within a given time interval for individuals belonging to different groups (e.g. males/females; successful/unsuccessful) or for a 1-unit increase in a continuous predictor variable; increasing hazards correspond to more rapid passage. TTE models are particularly well suited to analyzing hydrosystem passage times (e.g. Zabel et al. 2014) because individuals with incomplete telemetry records can be explicitly included in the modeling until their time of censoring (described below), and because covariates used to describe an individual’s passage experience can vary throughout time. Prior to each analysis, I developed candidate models that would best account for migration experience in each of the three hydrosystem reaches. Each model consisted of fixed covariates for the ultimate fate, sex, and GSE of migrants, as well as time-varying covariates calculated at hourly intervals to describe variation in river and dam tailrace environment. In models for reaches 1 and 2, river environment was described by temperature because of its effect on salmon behaviour and physiology (Hinch et al. 2006), 54 and natal water concentration in the lower Seton River because the behaviour of sockeye salmon in the lower reaches of the hydrosystem can vary with fluctuations in natal water cues (Fretwell 1989; Chapter 2). Because the time-varying covariates in the models for reach 2 were calculated upon arrival at the Seton Dam, above where natal water cues and temperatures are more variable given the influence of Cayoosh Creek, I incorporated a lag of 7 hours (mean passage from Seton River entry to the Seton-Cayoosh confluence) into each calculation to account for temperatures and natal water concentrations experienced below the Seton-Cayoosh confluence. Preliminary models of passage times from the Seton-Cayoosh confluence to the Seton Dam (where natal water cues do not vary) yielded similar results; therefore, I concluded models with a 7-hour lag were an appropriate account of passage times in relation to the conditions experienced from Seton River entry to arrival at the dam. Neither Fraser River nor Seton River discharge were included in the models for reaches 1 or 2 because GC and PC sockeye salmon migrate upriver when flows are at their yearly minimum and at levels unlikely to affect migration rates or behaviour (Rand et al. 2006; Macdonald et al. 2007). In the model for reach 3, the Seton Dam tailrace environment was described by time-varying covariates for temperature and discharge because GC sockeye salmon are known to require more anaerobic swimming effort while passing the Seton Dam, and exhibit less passage success during periods of higher discharge and temperature (Burnett et al. 2013, 2014). In addition, the model for reach 3 for GC fish also included a binary time-varying covariate to account for a change in attraction flows 10 m away from the fishway entrance during a management experiment conducted from 9 to 18 August 2014 (see Burnett et al. 2016). Fish were censored if they (i) failed to successfully enter the Seton River, (ii) entered the Seton River but failed to reach the Seton Dam, or (iii) reached the Seton Dam but failed to pass. I used the last available record from receivers throughout the lower hydrosystem in combination with visual review of individual migration histories to assign censoring times. Although TTE models are non-parametric with respect to the baseline hazard, there are additional assumptions that must be satisfied before models results can be safely interpreted (Kleinbaum et al. 2012). First is the assumption that censoring is not informative with respect to different groups. My models violated this assumption because 55 fish that ultimately did not survive were also censored more frequently than others. However, in a similar study of salmon passage through the Columbia River hydrosystem, this same assumption was violated and determined not to be of concern, as assessed by a series of sensitivity analyses (Caudill et al. 2007). Following the methods described in Caudill et al. (2007) [see Allison (2010) for more detail], I assessed the effects of such ‘informative censoring’ by assigning censoring times using three different criteria and comparing the estimated hazards with those of the original models of interest. If the odds ratios from these alternative models did not change markedly compared to the original models, I considered the overall effect of this violation on model conclusions to be minor (Caudill et al. 2007; Allison 2010). Briefly, the 3 alternative criteria I used in the sensitivity analyses were to (1) assume that all fish achieved each ‘event’ (i.e. river entry/dam arrival/dam passage) by assigning censoring times as event times, (2) assume that all censored fish would have remained ‘at-risk’ until the end of observation by assigning censoring times equal to the longest observed event-time, and (3) by only including fish that achieved each ‘event’ to directly examine how the ultimate fate of these individuals affected passage times. Results from this sensitivity analysis suggested violation of the assumption that censoring is not informative did not seriously bias model outcomes or provide false evidence of passage time relationships with covariates (for a detailed description of sensitivity analyses and results see Appendix 2). A second critical assumption of TTE models is that the hazards between groups remain proportional over time (Kleinbaum et al. 2012). Thus I included interactions for sex × time and fate × time in all models that both statistically tested and controlled for any violation of this assumption (Fox and Weisberg 2011). Additional assumptions of non-influential data and linearity in the relationship between the log hazard and covariates were also assessed visually by plotting residuals (Fox and Weisberg 2011). Survival to natal spawning grounds At a broader whole-system scale, I examined how the cumulative effects of passage through the entire Seton hydrosystem affected post-dam passage survival to natal spawning sites. Specifically, I used generalized linear models (GLMs) to assess how survival from fishway exit to natal sites was related to total hydrosystem passage time and degree-day accumulation from release to dam passage, as well as the sex, energetic 56 state, and capture/release date of all fish that successfully passed the Seton Dam. Consistent with the TTE model of reach 3 for GC fish, I also assessed how experience of an experimental flow change in the Seton Dam tailrace immediately prior to passage may have affected the survival of fish from this population. Variable selection, model selection, and multimodel averaging Prior to any analyses, I examined all covariates included in each of the models for collinearity and the presence of outliers following the data exploration protocols described in Zuur et al. (2010). Tagging/release date (TD) was highly collinear with temperature in the reach 1 and 3 models for both populations (reach 1: r = -0.8; VIF > 3; reach 3: r = -0.7; VIF > 3), as well as with maximum dam discharge in the survival GLM for GC fish (r = -0.8; VIF > 3). Total migration time was highly correlated with accumulated thermal units (ATUs) and maximum temperature exposure (r = 1.0; VIF > 3) in the survival GLM for both populations, but not with degree days (DDs) (r < 0.4; VIF < 1. I opted to include covariates for temperature and discharge rather than TD in the TTE models, as these effects were more biologically relevant to addressing the objectives of this study. Whereas in the survival GLMs, I included TD rather than discharge and temperature because TD could be used as a proxy to assess the effects of maximum temperature and discharge on survival, but also account for the run timing of individuals. Similarly, total migration time from release to dam passage and DDs, rather than ATUs or maximum temperature exposure were included in survival GLMs because both maximum temperature exposure and ATUs would inherently increase in fish with longer cumulative passage times, while DD provided an appropriate measure of cumulative temperature experience during hydrosystem passage. Moreover, inclusion of total passage time in this final GLM also allowed me to test whether the probability of surviving to natal sites was reduced in slower migrants (e.g. Naughton et al. 2005; Caudill et al. 2007). I conducted all analyses on the GC and PC populations independently, and pooled the data from the 2013 and 2014 study years because I was primarily interested in the effects of fate, fish traits, and environmental conditions on passage times rather than annual variation in behaviour. Plus any differences the physiological condition of migrants was not different between years (Chapter 2), thus any differences in behaviour would have likely mainly been driven by temperature, which 57 was included in all analyses. No observations for any of the covariates included in the models were considered outliers or removed. For each analysis, I fit models with all possible subsets of covariates to the data, and then ranked these models using bias-corrected Akaike Information Criteria (AICC), which compares each model by its relative support for the data using AICC weights (wi) which describe the probability of a model in a candidate set as the most parsimonious (Burnham and Anderson 2003). Sex × time and fate × time interactions were fixed in all TTE models so assumptions were not violated. Uncertainty in model-selection was accounted for by calculating model-averaged estimates of odds ratios and unconditional confidence intervals (CIs) from a 95% confidence set of models using the ‘zero’ method (Burnham and Anderson 2003, Grueber et al. 2011). All continuous model covariates were standardized by centering and dividing by two standard deviations (Gelman 2008) to allow for the comparison of the relative effect size of each variable and the direct interpretation of the main effects of variables involved in interactions (Schielzeth 2010). Covariates were considered to have predictive power when 95% CIs for estimates of odds ratios did not include zero. AICC model selection statistics for all models with ΔAIC < 2 are given by response and population in Table 3.2. All statistical analyses were conducted in R (R Development Core Team 2008). 3.3 Results 3.3.1 Environmental conditions The mean Fraser River temperature experienced by GC sockeye salmon tracked throughout the two study periods was 18.3C (± 1.2C SD), though some fish encountered temperatures as high as 21.2C, with considerable variation in temperature over the duration of the 2013 and 2014 migrations (Figure 3.2 a). Seton River temperatures experienced by GC fish through the duration of the study were on average cooler than the Fraser River at 17.6C (± 1.0C SD), though some fish could have encountered highs of ~ 22C during periods of the 2013 and 2014 migrations (Figure 3.2 a). The mean concentration of natal water GC fish encountered while migrating in the Fraser River and lower Seton River was 92.8% (± 3.4% SD), with more variability throughout the 2013 migration compared to 2014 (Figure 3.2 b). The mean temperature 58 experienced by GC fish in the Seton Dam tailrace was 17.8C (± 1.1C SD), though again, there was a considerable amount of variability in temperatures over the 2013 and 2014 migrations, with highs nearing ~ 24C in the dam tailrace in 2013 (Figure 3.2a). Mean discharge experienced by GC fish while in the Seton Dam tailrace prior to passing was 26.3 m3s-1 (± 2.3 m3s-1 SD), with some fish experiencing 33.8 m3s-1 during the management experiment conducted in 2014 (Burnett et al. 2016), as described previously (Figure 3.2 c). In contrast, late-run PC sockeye salmon migrate during the fall months and thus encounter conditions that are markedly different from summer-run GC conspecifics. For example, the mean Fraser River temperature experienced by tracked PC fish was 11.8C (± 1.0C SD), with considerably less variability in temperatures over the duration of the 2013 and 2014 migrations (Figure 3.2 a). Because of the influence of warm epilimnetic Seton Lake water discharged from the Seton Dam, the mean temperature experience of tagged PC fish in the Seton River was 14.3C (± 0.5C SD), approximately 2.5C warmer on average than the Fraser River. Similarly, mean temperature in the Seton Dam tailrace experienced by PC fish prior to passing the dam was 14.5C (± 0.8C), with less variability over the two years of migration (Figure 3.2 a). PC fish also experienced a mean discharge of 17.2 m3s-1 from the Seton Dam that was considerably less than what GC fish experienced, which also continued to decline over both PC migration years (Figure 3.2 c). Mean natal water concentration experienced by PC fish in the Fraser River and lower Seton River was only slightly higher than the target level at 91% (± 1.7% SD) (Figure 3.2 b). 3.3.2 Hydrosystem passage Seton River entry Nearly all of GC (92%; 258 of 283) and PC (87%; 188 of 217) sockeye salmon tracked for this study passed into the Seton River following release. Median passage times through reach 1 were 18.8 hours and 73 hours for the GC and PC populations, respectively (Figure 3.3 a & b), though 3% of GC fish (8 of 258) took more than 60 hours to pass, while 8.5% of PC fish (16 of 188) took over 200 hours. Summer run GC fish 59 were also 125% faster than PC fish on average while transiting this 3.5 km reach (Table 3.1). Results from the AIC model selection and multi-model averaging revealed that sex had the strongest influence on passage times through reach 1 for both populations (Table 3.2), as mean passage times were 36% and 32% faster among males compared to females from the GC and PC, respectively (Table 3.1). Odds ratios for sex indicated that GC and PC males were 4× (e1.4) and 2× (e0.86), respectively, more likely than females to enter the Seton River at any given hour after release (Figure 3.3 c & d). The fate of individuals was also a strong predictor of passage times through reach 1, as successful GC and PC fish (those that survived to natal spawning sites) were, respectively, 21% and 33% faster and 2× (GC: e0.81; PC: e0.66) more likely than unsuccessful fish (those that did not survive to natal sites) to enter the Seton River at any given hour post-release (Figure 3.3 c & d). PC fish with higher somatic energy reserves were slower passing through reach 1, as the odds ratio for GSE was less than zero, indicating a 30% decrease (e-0.33) in the instantaneous probability of river entry per 1.5 unit increase in energy in fish from this population (Figure 3.3 d). A similar trend was true for GC fish, though the estimated odds ratio for GSE in this population had CIs that included zero (Figure 3.3 c). Neither temperature nor natal water cues appeared to be related to passage times through this first reach of the hydrosystem for either population (Table 3.2; Figure 3.3 c & d). There was also no evidence for any interactions between time with sex or fate (Table 3.2; Fig. 3c,d). Seton River entry to Seton Dam The odds of arriving at the Seton Dam following Seton River entry were high, as 98% (253 of 258) and 97% (183 of 188) of GC and PC sockeye salmon, respectively, that entered the Seton River were also detected in the Seton Dam tailrace. The differences in passage times that characterized the two populations through reach 1 largely disappeared in reach 2, as both populations had mean and median passage times through reach 2 of 15 hours and 11 hours, respectively (Table 3.2; Figure 3.4 a & b). However, 15% (39 of 258) of GC fish and 18% (34 of 188) of PC fish took over 20 hours to swim the 5 km Seton River, while one PC individual took > 200 hours (Figure 3.4 a & b). 60 Mean passage times through the Seton River were 17% and 62% faster for ultimately successful GC and PC migrants relative to unsuccessful fish (Table 3.1). Odds ratios for fate were the largest in models for both populations, indicating that eventually successful GC and PC fish were 3.4× (e1.21) and 2.6× (e0.94), respectively, more likely than unsuccessful fish to arrive at the dam at any given hour since entering the Seton (Figure 3.4 c & d). While there was little difference in mean passage times between sexes within populations through reach 2 (Table 3.1), the odds ratio for sex in the GC model indicated males were twice (e0.7) as likely as females to arrive at the Seton Dam at any given hour since entering the Seton (Figure 3.4 c). There was also support from the models for interactions between time with sex and fate affecting passage times through reach 2 (Table 3.2; Figure 3.4 c & d); however, the odds ratios for these interactions in both cases were very close to zero (e-0.06 and e-0.05) (Figure 3.4 c & d). Consistent with the models from reach 1, fish with higher energy reserves (GSE) also passed more slowly through reach 2 (Figure 3.4 c & d). Though in this case there was only confidence in this effect for GC fish, with the odds ratio indicating a 40% (e-0.48) decrease in the instantaneous probability of arrival at the Seton Dam for every 1.5 unit increase in GSE in GC sockeye (Figure 3.4 c). There was also a similar trend for GSE among PC fish, but CIs for the odds ratio included zero (Table 3.2; Figure 3.4 d). The temperature of the Seton River had no effect on passage times through reach 2 for either population, nor did the concentration of natal water cues affect PC fish passage through this reach (Table 3.2; Figure 3.4 c & d). Among GC fish, however, the odds ratio for natal water was negative, indicating arrival hazard decreased following increases in natal water concentration – implying that GC fish passed more slowly through this reach when natal water concentrations were elevated (Figure 3.4 c). Seton Dam passage Most sockeye salmon that arrived at the Seton Dam also passed upstream beyond the dam, as 91% (229 of 253) and 87% (160 of 183) of GC and PC fish, respectively, that were detected in the Seton Dam tailrace were also detected exiting the fishway into the dam forebay. Mean dam passage times of PC fish were nearly twice as long as they were for GC fish, while 9% (24 of 253) of GC and 24% (44 of 183) of PC fish took over 20 61 hours to pass, with one ultimately successful GC male passing the dam in 96 hours (Table 3.1; Figure 3.5 a & b). GC and PC sockeye that survived to natal spawning sites were 17% and 7% faster passing the Seton Dam, respectively, than fish that passed the dam but failed to survive to natal sites (Table 3.2). Fate also had the largest effect on Seton Dam passage times, with odds ratios indicating that GC and PC fish that survived to natal spawning sites were 1.8× (e0.62) and 1.6× (e0.50), respectively, more likely than fish that passed the dam but did not survive to natal sites to pass the dam at any given hour since arriving in the dam tailrace (Figure 3.5 a, b, c & d). There was no effect of sex on passage times for either population, because males and females passed the dam in similar amounts of time (Table 3.2); there was also no relationship between energy levels and passage for either population (Table 3.2; Figure 3.5 a, b, c & d). Discharge in the Seton Dam tailrace was not related to passage time in either population, nor was there any evidence for an effect of the experimental flow scenario on the dam passage times for GC fish (Table 3.2; Figure 3.5 c & d). Passage times among PC fish were largely unaffected by temperatures in the dam tailrace; however, there was a strong effect of tailrace temperature on passage times for GC fish (Figure 3.5 c & d). The odds ratio for temperature in this case indicated that the instantaneous probability of GC fish passing the dam increased 1.6× (e0.45) for every 2C increase in tailrace temperature above the mean (Figure 3.5 c), implying that GC fish passed the dam more rapidly following increases in tailrace temperature. 3.3.3 Cumulative effects of passage and survival to natal sites Mean cumulative migration times through the entire hydrosystem, from release in the Fraser River to exiting the top pool of the Seton Dam fishway, were 94% faster among GC fish compared to PC fish, while males from both populations were 17% and 25% faster, respectively, than females (Table 3.1; Fig 3.6 a). GC and PC fish that ultimately survived to natal sites were also 17% and 25%, respectively, faster on average when compared to ultimately unsuccessful fish (Table 3.1; Figure 3.6 a). 62 Of all tracked fish that passed through the lower hydrosystem to successfully ascend the Seton Dam, 68% (156 of 229) and 74% (119 of 160) survived to natal spawning sites at GC and PC, respectively. Model averaged coefficient estimates describing how the cumulative effects of hydrosystem passage and fish traits affected survival indicated that male GC fish were more than twice as likely (e0.8) as females to survive to natal sites, while there was no effect of sex on post-passage survival for PC fish (Table 3.2; Figure 3.6 b & c). As such, GC post-passage survival to natal sites was 60% (77 of 128) for females and 78% (79 of 101) for males, while for PC females it was 71% (54 of 76) and 80% (66 of 83) for PC males (Table 3.1). Tagging date, degree-day accumulation, and GSE all had no effect on the probability of post passage survival in fish from both populations that passed the dam (Table 3.2; Figure 3.6 b & c). There was some indication from the models that survival probability was decreased in fish with greater cumulative migration times, although confidence in these effects was low because 95% CIs included zero (Table 3.2; Figure 3.6 b & c). There was also considerable uncertainty in the estimated effect of the experimental flow scenario on the survival of GC fish, as only 16% (31 of 198) of GC migrants that passed the dam encountered these alternate conditions (Table 3.2; Figure 3.6 b). 3.4 Discussion Approximately 65,000 GC and PC sockeye salmon pass through the Seton hydrosystem annually (Casselman et al. 2015); however, this is the first study to explore the factors that drive behaviour during passage through multiple reaches of the hydrosystem where unique environmental characteristics could have differing effects on behaviour, and on the cumulative effects of post-passage survival to natal spawning sites. While the overall post-release survival of GC and PC migrants to their respective natal spawning sites was only slightly greater than half for both populations (GC = 55%; PC = 55%), passage through the hydrosystem was high (GC = 80%; PC = 74%) and consistent with past estimates of dam passage for GC sockeye in the system [e.g. 80% (Roscoe et al. 2010b) & 62% (Burnett et al. 2013)]. However, over one quarter of migrants (GC = 32%; PC = 26%) failed to reach natal spawning sites after passing the dam, which was again consistent with post-dam passage mortality estimates from recent studies of the GC population [e.g. 20% (Roscoe et al. 2010a) & 33% (Burnett et al. 2014)]. Presumably 63 many of these migrants died in Seton or Anderson Lakes after succumbing to rigours of their past migration experience (e.g. Burnett et al. 2013, 2014); however, manual tracking surveys conducted during a follow up study in 2015 also revealed a portion of mortality (~7%) was due to unreported fisheries removals upstream of the Seton Dam (Casselman et al. 2016). One of the most apparent findings from this study were the population-specific differences in passage behaviour throughout the entire hydrosystem, specifically, PC sockeye salmon took nearly 3 times as long as GC sockeye to pass from below the powerhouse to above the dam (mean = ~ 112 hrs vs. ~ 41 hrs, respectively), with most delay occurring in the Fraser River prior to entering the Seton, and in the Seton Dam tailrace prior to passage. This was not unexpected however, as late run Fraser River sockeye (e.g. PC fish) are known to be slower migrants than summer runs (e.g. GC fish) (English et al. 2005), and in Chapter 2 I found that PC fish delay longer in the Seton powerhouse tailrace and wander more in the Fraser River than GC fish. The extended delays passing the Seton Dam were also not unexpected given late runs usually mill in natal lakes prior to entering spawning grounds (Mathes et al. 2010), and PC fish pass the dam directly into Seton Lake (their natal lake), which is only ~20 km from spawning grounds. The most striking finding from this study was the strong, consistent relationships between passage times and the ultimate fate of individuals from both populations. That is, fish that ultimately survived to natal sites were consistently faster passing through each reach of the hydrosystem relative to those that did not survive. I had anticipated this trend given this relationship has been observed in upriver migrating salmon passing the multiple dams and reservoirs of the highly-impacted Columbia River hydrosystem (e.g. Chinook salmon and steelhead; Caudill et al. 2007). However, these findings also raise additional questions about whether this fate – time relationship is a result of impoundment, or whether it would exist in more natural systems as an artifact of previous upriver migration experience (Caudill et al. 2007), and also whether the cumulative effects of prolonged migration through the Seton hydrosystem affect survival to natal sites (e.g. Naughton et al. 2005). 64 In many regulated systems, the high en route loses of migrating adult salmonids are thought to be a result of temperature and hydraulic changes to migration corridors that increase the metabolic costs of migration (see reviews in Waples et al. 2008; Thorstad et al. 2008). Passage through the Seton hydrosystems is certainly not without its difficulties that may have contributed to the fate – time relationship observed in this study [e.g. powerhouse delays (Fretwell 1989; Chapter 2), dam passage (Roscoe et al. 2010b; Burnett et al. 2013)]; however, GC and PC migrants also experience ~ 340 km of additional upriver migration in the un-impounded mainstem Fraser River prior to entering the hydrosystem where they can be exposed to a number of different factors that could have also influenced this relationship. Indeed, encounters in the mainstem Fraser with high temperatures (e.g. Patterson et al. 2007; Donaldson et al. 2009), hydraulically challenging rapids (e.g. Hinch and Rand 1998; Hinch and Bratty 2000), fisheries e.g. (Nguyen et al. 2013; Raby et al. 2015; Robinson et al. 2015), or diseases and pathogens (e.g. Wagner et al. 2005; Miller et al. 2011) could all increase physiological stress, deplete energy reserves, prolong migration, and reduce survival odds prior to entering the hydrosystem (Hinch et al. 2006). My observations that ultimately unsuccessful fish were slower migrating up the un-impounded mainstem Fraser River and Seton River where discharge levels were unlikely to affect behaviour (Macdonald et al. 2007) suggest that the slowed and eventually unsuccessful migrations of some fish during passage through regulated systems may at least in part be unrelated to impoundment and partially predetermined by the rigours of long-distance migrations (Caudill et al. 2007). Relatively slow migrations through the Columbia River hydrosystem have been linked to sockeye salmon failing to reach natal tributaries and are thought to be a result of individual fish characteristics and the cumulative effects of migration (Naughton et al. 2005). Thus I had expected that an individual’s passage experience through the Seton hydrosystem as well as sex and energy content would play a role in survival to natal sites. I had predicted that slower migrants and those exposed to higher degree-days of water temperature would not survive as well relative to faster fish, nor would females survive as well as males, while fish with higher energy reserves would exhibit greater survival. In general, I found survival to natal sites was indeed lower in migrants that were slower transiting the hydrosystem. However, I found little indication, within a population, that 65 the cumulative effects of hydrosystem migration experience or fish traits influenced post-passage survival, with the exception of GC fish, for which males were more likely to survive than females. I had expected this reduced survival of GC females, as summer-run female Fraser River sockeye often exhibit greater en route mortality than males (Martins et al. 2012a), and past studies have documented poorer survival to spawning grounds of GC females (Roscoe et al. 2010b; Burnett et al. 2014, 2016). Similarly, I had expected equal survival among PC sexes, as this has been observed in past studies of late-run adult Fraser River sockeye (e.g. Martins et al. 2012b), likely because of the significantly cooler temperatures and discharge conditions these fish encounter throughout their upriver migration, which could greatly reduce the metabolic and physiological costs of migration and increase survival (Clabough et al. 2006; Hinch et al. 2006). That I found no effects of cumulative hydrosystem migration experience or fish traits on post-dam passage survival – aside from sex – was contrary to what I had expected. However there are many other aspects of an individual’s migration experience and physiological condition that I did not measure that could predispose some migrants to post-dam passage mortality regardless of their behaviour during hydrosystem passage. For instance, physiological impairments arising from escape or release from lower Fraser River fisheries, and pathogen and disease development brought on by injuries and warming rivers are two examples of recent emerging issues known to have latent effects on upriver migrating Fraser River salmon (Donaldson et al. 2011; Miller et al. 2014; Raby et al. 2015). Apart from affecting survival, the cumulative stresses of negotiating the hydrosystem and passing the Seton Dam may impart additional latent carry-over effects that affect post-passage behaviour in natal lakes such as seeking cooler temperatures to cope with stress (e.g. Roscoe et al. 2010a; Casselman et al. 2016), or reduce spawning success (Burnett at al. 2016). I had hypothesized that given their more limited energy budgets for swimming activities (Hinch and Rand 1998; Standen et al. 2002), slower migration rates in the Fraser River (Hanson et al. 2008), and longer delays in the Seton Dam tailrace compared to males (Roscoe et al. 2010b; Burnett et al. 2013), females would be slower passing through the hydrosystem. This was indeed the case, although not through every reach. 66 Females from both populations were markedly slower than males passing through the first reach of the hydrosystem. GC females continued with this pattern following their entry into the Seton River, however, PC females sped up and passed through this second reach at the same speed as PC males. Which, though unexpected, makes sense, given PC fish are ~ 25 km closer to natal spawning sites and more mature than GC fish at this stage in their migration (Chapter 2), and female sockeye are known to migrate faster with decreasing proximity to natal sites and increases in reproductive hormones (Sato et al. 1997). Sex had no effect on Seton Dam passage times because equal proportions of males and females within each population passed the dam, and did so in similar amounts of time. In past studies from this system where differences in passage delay and success were observed between sexes, all fish were captured from the top pool of the fishway (Roscoe et al. 2010a, 2010b; Burnett et al. 2013). Burnett et al. (2014) recently found that female GC sockeye salmon swim with much more anaerobic effort than males when passing the Seton Dam, which likely had confounding effects on dam passage estimates in these past studies in that fishway capture probably doubled the effort required by females to navigate the turbulent flows of the tailrace and resulted in longer delays and lower passage success. Indeed, after moving the capture location downstream of the dam and using ‘fishway-naïve’ fish caught from the same fence in the same year as the present study, Burnett et al. (2014) found no effect of sex on dam passage delay or success. I had predicted that fish with higher energy reserves would migrate slower (e.g. Young et al. 2006; Hanson et al. 2008), which was indeed the case, although only through the first two reaches of the hydrosystem. Throughout their upriver migrations, sockeye salmon pass through river segments that differ greatly in their hydraulic characteristics, and spend the most amount of energy in areas of complex and challenging flows (Hinch and Rand 1998; Standen et al. 2002; Burnett et al. 2014). Fish with lower energy in this study likely migrated faster while in the Fraser River and through the Seton River in an effort to reduce transit time and maximize time on spawning grounds (Hanson et al. 2008); indeed discharge in these two reaches was relatively low and would have made this possible. However, these faster-migrating fish also potentially depleted additional energy reserves while doing so, and upon encountering the Seton Dam had 67 reduced their capacity to meet the energetic requirements of dam passage (e.g. Burnett et. al 2014), resulting in passage times equal to those of migrants with higher energy content. Although degree-day accumulation during passage through the hydrosystem was unrelated to post-passage survival to natal sites, the acute stress of high migration temperatures can cause adult upriver migrants to alter their behaviour to avoid peak temperatures (e.g. Columbia River sockeye and steelhead; Keefer et al. 2008 & 2009). Past studies of adult Fraser River sockeye have revealed that fish migrate faster in upstream river segments above Hell’s Gate when temperatures are high (Hanson et al. 2008), thus I had hypothesized that GC migrants encountering Fraser River and Seton River temperatures well above optimal (~17C; Lee et al. 2003) would travel faster. While there was little confidence in high temperatures affecting passage times through the first two reaches of the hydrosystem, there was a strong effect of temperature on Seton Dam passage times, indicating elevated tailrace temperatures led to more rapid dam passage. Interestingly, this change in temperature also corresponded to the onset of temperatures near the critical thermal limit (TCRIT) for the GC population (~21C; Eliason et al. 2011). Above TCRIT, aerobic scope collapses and fish rely on anaerobic metabolism risking acidosis and cardiorespiratory failure (Farrell et al. 2008). This is concerning as faster passage during stressful temperatures would likely exacerbate the effects of rapid passage, which are known to require excessive burst swimming that often results in increased mortality, irrespective of temperature (Burnett et al. 2014). That PC were largely unaffected by temperatures throughout the hydrosystem is in line with what I had expected given Fraser and Seton River temperatures were significantly cooler than during the GC run, and much closer to optimal temperatures for late run Fraser River sockeye (Eliason et al. 2011). One of the most unique aspects of this study was my ability to examine how variable concentrations of natal water cues affected passage through the lower hydrosystem. TTE analyses revealed fish from both populations encountered natal water concentrations that ranged from ~ 73 – 95% during passage through the first two reaches of the hydrosystem. In Chapter 2, I described how wandering and delay of GC sockeye in the first reach of the hydrosystem was largely unaffected by variable natal water cues 68 emanating from the Seton River plume, while on the other hand PC sockeye seemed more ‘confused’ and less certain of their intended trajectories when natal water concentrations were reduced. I had thus expected that given such variable proportions of natal water, PC fish would exhibit slowed passage through the whole hydrosystem during periods of lower natal water concentrations, yet I found no evidence for this. One possible explanation for this is that PC fish are nearing full maturity at this stage in their migration (Chapter 2), and more mature fish can exhibit heightened discriminatory capability of homestream odours (Carruth et al. 2002; Ueda 2011). Which combined with the increased motivation to complete migration that is related to elevated reproductive hormones (e.g. Sato et al. 1997), may have been responsible for PC sockeye not reducing the speed of their migrations, even when encountering such variable natal water concentrations. In Chapter 2, I discussed how the heightened olfactory sensitivity of PC fish might lead to extensive delays in the Seton powerhouse and wandering in the Fraser River that could have consequences further along the migration route. However, the present results also suggest this sensitivity may be advantageous in that once PC migrants enter the Seton River they could be less likely to delay their migrations any further, even when natal water cues are relatively low. In contrast to PC fish, I observed that GC fish migrated more slowly up the Seton River when natal water concentration in the Seton River was high, which given the results from Chapter 2 for this population, was contrary to what I had predicted. However, GC fish were also less mature than PC fish (Chapter 2), and passed through this reach more slowly if they had higher energy content. In this case, encounters with higher natal water concentrations along with greater energy reserves likely sent strong signals to GC fish to slow down migrations and encourage enroute gonad development rather than arriving on spawning grounds prematurely, as spawning dates are generally fixed and immature fish would be unable to spawn should they arrive early (Hinch et al. 2006). Roscoe et al. (2010a) found that less mature GC migrants with higher energy reserves tended to hold for longer durations in the natal lakes above the Seton Dam where they were presumably maturing before continuing on to spawning grounds at Gates Creek. My present results also suggest this relationship exists in GC migrants prior to passing the Seton Dam as well. 69 Apart from describing passage through the lower Seton hydrosystem, another advantage of my detailed review of the telemetry data was that it allowed me to make additional inferences about the final fate of fish not observed to have passed the dam. Rather then entering the Seton River after release, I observed that 11% (54 of 500) of migrants either overshot their natal tributary (18 of 54), or migrated back downstream in the Fraser River (36 of 54). A portion of these fish (4%; 20 of 500) may have incurred some degree of experimenter-induced stress that led to eventual post-release mortality in the mainstem Fraser (confirmed by manual radio-tracking surveys). However, after accounting for these losses, the proportion of migrants that did not re-ascend the Seton hydrosystem was similar to rates of straying observed in other populations of sockeye salmon from the Fraser and Columbia Rivers (~ 5%; reviewed in Keefer and Caudill 2014). All tagged fish in my study were confirmed to be either GC or PC migrants via screening with a Fatmeter or DNA analysis (see Casselman et al. 2015), thus the notion that some of these migrants may have indeed strayed after release is valid. This was further supported by observations of my tagged GC and PC fish in telemetry records from concurrent studies in downstream Fraser River tributaries (e.g. the Thompson River) and spaghetti tagged fish observed in tributaries further upstream in the Fraser River (e.g. the Bridge River; Casselman et al. 2015). Another 11% of my study fish (47 of 436) were detected in the Seton Dam tailrace but failed to pass, which is consistent with estimates of passage failure for adult salmon at dams in the Columbia River and many European hydrosystems (Gowans et al. 2003; Keefer and Joosten 2008; Thorstad et al. 2008). It is often difficult, however, to discern from telemetry data whether fish actually die in dam tailraces, succumbing to the energetic and physiological costs of traversing hydraulically challenging tailrace flows (e.g. Burnett et al. 2013, 2014), or whether they simply exhibit fallback, never to re-ascend (e.g. Boggs et al. 2004). Roscoe et al. (2009) postulated that GC sockeye salmon that fail to pass the Seton Dam might seek alternate passage routes and fallback from the dam rather than die in the tailrace. However, an unexpected advantage of the fence/trap below the Seton Dam was its ability to collect fish carcasses that washed downstream from the dam. Over the duration of the study, I counted ~ 650 dead sockeye salmon that had collected on the upstream side of the fence (including 8 radio-tagged fish), lending 70 support to the telemetry data that indicate a certain proportion of fish from the GC and PC populations do indeed die in the dam tailrace rather than seek alternate passage (Casselman et al. 2015). 3.4.1 Management implications It has long been recognized that the fluctuating natal water cues in the lower Seton hydrosystem can be a significant migration obstacle to upriver migrating sockeye, leading to long delays in the Seton powerhouse and difficulty entering and ascending the Seton River (Fretwell 1989). Results from water preference studies using ‘Y-maze’ behavioural choice experiments conducted in the late 1970’s and re-visited again in 2013 and 2014 provided in-situ experimental evidence that delays in the lower hydrosystem could be mitigated by controlling Seton Dam discharge so that the relative contribution of Cayoosh Creek water to the lower Seton River was no more than 80% and 90% during the GC and PC migrations, respectively (Fretwell 1989; Casselman et al. 2015). While these previous studies (e.g. Y-mazes, Fretwell 1989; Casselman et al. 2015) and my observations from Chapter 2 examining responses to natal water in the powerhouse and Fraser River alone provide valuable insight into some of the mechanisms underlying behaviour and delay of GC and PC migrants in this system, these results are only part of the equation when it comes to examining how fish utilize natal water during passage through the whole hydrosystem. That I observed migrants in the present study from both populations encountered natal water concentrations that fell above and below the 80% and 90% targets with no overall effect on passage through the hydrosystem suggests the current natal water targets are acceptable for mitigating delay and encouraging passage, but also that managers should strive to maintain the highest natal water levels possible during sockeye migrations through the hydrosystem, especially during the PC migration (Chapter 2). However, these results may also be an impetus for managers to explore a whole system experimental approach to natal water manipulation in order to understand how this factor may have differing effects at different reaches and at different stages of maturation for the two populations of sockeye that pass through the hydrosystem. Fraser and Seton River temperatures can have a confounding effect on sockeye salmon migration success, affecting the condition and in turn behaviour of migrants as they pass through the hydrosystem. In the past 60 years, average summer river 71 temperatures in the Fraser River watershed have increased > 2C (Patterson et al. 2007), with many fish now typically encountering temperatures that meet or exceed population-specific thermal limits, which can be extremely stressful with prolonged exposure, and can lead to en route mortality (Eliason et al. 2011). There was also some indication from the models that migrants from both populations took longer to enter the Seton River following release during periods of elevated Fraser River temperatures, which is concerning given migration delays at high temperatures can be fatal. Herein is more incentive for managers to strive for maintaining the highest concentrations of natal water possible so that migrants are able to uninterruptedly enter the Seton River where they may encounter somewhat cooler temperatures than the Fraser (Chapter 2). Even though temperature was unrelated to passage through the Seton River itself, observations that temperatures reached near lethal limits for sockeye salmon (~24C; Servizi and Jensen 1977) in the Seton Dam tailrace, and that GC fish passed the dam more rapidly at the onset of stressful temperatures is cause for concern given the implications of increased mortality following rapid passage during such temperatures (e.g. Burnett et al. 2014). Managers may be able to mitigate some of the migration stress adult sockeye face during dam passage at peak temperatures by altering operations of the powerhouse so that cooler water from Seton Lake flows more directly into the Seton Dam fishway (Casselman et al. 2015), and/or by altering the hydraulic conditions in the Seton Dam tailrace through changes in water release location which has been shown to reduce metabolic stress and enhance migration survival to spawning grounds in GC sockeye salmon (e.g. Burnett et al. 2016). These sorts of actions may be particularly imperative in the coming years, as climate models predict future regional warming and higher peak summer temperatures throughout the Fraser River watershed (Ferrari et al. 2007; Patterson et al. 2007; Martins et al. 2011). 72 3.5 Chapter 3 Tables Table 3.1 – Summary statistics for reach-specific and total lower Seton hydrosystem migration times in hours for Gates Creek and Portage Creek sockeye salmon tracked in this study. Migration times are calculated based only on individuals that achieved the event of interest (i.e. entered the Seton River, arrived at the Seton Dam, passed the Seton Dam) and are presented by sex and fate; successful individuals were fish that eventually reached natal spawning sites; unsuccessful individuals were those not observed to reach natal sites. Mean Median Range SD SE Time from Fraser River release to Seton River entry (hours) Gates Creek Females N = 142 26.37 20.81 3.96 – 97.71 15.36 1.28 Males N = 116 18.26 15.78 2.72 – 103.15 13.85 1.28 Successful N = 156 20.74 17.86 2.94 – 103.13 15.24 1.22 Unsuccessful N = 102 25.62 21.82 2.68 – 78.14 14.69 1.46 Portage Creek Females N = 88 112.65 91.10 15.84 – 468.21 83.57 8.91 Males N = 100 81.43 62.73 5.23 – 268.92 64.65 6.46 Successful N = 119 83.73 65.35 5.85 – 305.09 60.30 5.53 Unsuccessful N = 69 117.29 95.65 5.23 – 468.17 93.04 11.20 73 Mean Median Range SD SE Time from Seton River entry to Seton Dam arrival (hours) Gates Creek Females N = 139 13.96 11.19 2.46 – 46.05 8.53 0.72 Males N = 114 13.70 10.31 4.93 – 82.56 11.33 1.06 Successful N = 156 12.94 10.24 4.54 – 82.56 9.92 0.79 Unsuccessful N = 97 15.30 12.80 2.46 – 52.51 9.67 0.98 Portage Creek Females N = 88 14.76 9.73 3.24 – 62.52 12.27 1.31 Males N = 97 16.57 11.29 2.94 – 201.78 22.44 2.28 Successful N = 121 12.01 9.75 4.57 – 61.35 7.14 0.65 Unsuccessful N = 64 22.70 13.74 2.94 – 201.78 28.37 3.55 74 Mean Median Range SD SE Seton Dam passage time (hours) Gates Creek Females N = 128 7.76 3.02 1.07 – 59.97 10.01 0.88 Males N = 101 7.97 2.75 1.02 – 96.99 13.55 1.35 Successful N = 156 7.42 2.83 1.02 – 96.99 12.05 0.97 Unsuccessful N = 73 8.79 3.49 1.04 – 55.49 10.86 1.27 Portage Creek Females N = 77 15.44 10.03 1.12 – 74.19 16.33 1.86 Males N = 83 15.73 11.24 1.20 – 74.77 17.52 1.92 Successful N = 119 15.29 10.18 1.12 – 74.77 16.62 1.52 Unsuccessful N = 41 16.45 8.63 1.20 – 65.66 17.90 2.80 75 Mean Median Range SD SE Total migration time from release to Seton Dam passage (hours) Gates Creek Females N = 128 46.27 43.38 15.65 – 125.68 19.01 1.68 Males N = 101 39.00 33.68 12.32 – 121.54 22.15 2.20 Successful N = 156 40.59 36.80 12.32 – 121.54 20.82 1.67 Unsuccessful N = 73 48.35 44.44 20.30 – 125.68 19.64 2.30 Portage Creek Females N = 77 135.14 123.47 25.56 – 449.83 75.56 8.67 Males N = 83 105.21 87.52 17.66 – 358.35 72.33 7.94 Successful N = 119 111.69 98.09 17.72 – 358.35 66.51 6.07 Unsuccessful N = 41 143.55 128.53 17.66 – 449.83 94.10 15.07 76 Table 3.2 – AICC model selection statistics for models with Δ AICC < 2 for Gates Creek and Portage Creek sockeye salmon time-to-event proportional hazards regression predicting hydrosystem migration times from (1) release to Seton River entry, (2) Seton River entry to arrival at the Seton Dam, and (3) passage time of the Seton Dam. Models with ΔAICC < 2 are also given for generalized linear models for both populations predicting (4) survival to natal spawning sites following passage of the Seton Dam. Abbreviations used for model covariates include: (GSE) gross somatic energy, (FRT) Fraser River temperature, (SRT) Seton River temperature, (NW) natal water concentration, (DTT) Seton Dam tailrace temperature, (Ddis) Seton Dam discharge, (Flow) flow scenario, (TMT) total migration time, and (DD) degree days. ΔAICC is the difference in AICC values between model i and the top model in the candidate set. Models are ranked from lowest to highest ΔAICC, and by wi – the probability that a given model is the best in the 95% confidence set. Model log Lik AICC Δ AICC wi A) Time to Seton River entry Gates N = 283 Censored = 25 Sex + Fate + GSE + FRT + SRT + (Sex × Time) + (Fate × Time) -1088.8 2191.7 0.00 0.25 Sex + Fate + GSE + SRT + (Sex × Time) + (Fate × Time) -1090.6 2193.2 1.56 0.12 Sex + Fate + GSE + FRT + SRT + NW + (Sex × Time) + (Fate × Time) -1088.7 2193.4 1.76 0.11 Sex + Fate + FRT + SRT -1090.7 2193.5 1.86 0.09 Portage N = 217 Censored = 29 Sex + Fate + GSE + FRT + SRT + (Sex × Time) + (Fate × Time) -811.2 1636.3 0.00 0.26 Sex + Fate + GSE + FRT + (Sex × Time) + (Fate × Time) -812.3 1636.5 0.22 0.23 Sex + Fate + GSE + FRT + SRT + NW + (Sex × Time) + (Fate × Time) -810.8 1637.5 1.18 0.14 Sex + Fate + GSE + FRT + NW + (Sex × Time) + (Fate × Time) -812.0 1638.0 1.69 0.11 77 Model log Lik AICc Δ AICc wi B) Time from Seton River entry to Seton Dam Gates N = 258 Censored = 5 Sex + Fate + GSE + NW + (Sex × Time) + (Fate × Time) -1095.1 2202.3 0.00 0.65 Sex + Fate + GSE + NW + SRT + (Sex × Time) + (Fate × Time) -1094.8 2203.7 1.43 0.32 Portage N = 188 Censored = 5 Sex + Fate + GSE + (Sex × Time) + (Fate × Time) -779.1 1568.1 0.00 0.32 Sex + Fate + (Sex × Time) + (Fate × Time) -780.6 1569.3 1.13 0.18 Sex + Fate + GSE + SRT + (Sex × Time) + (Fate × Time) -788.8 1569.7 1.60 0.15 Sex + Fate + GSE + NW + (Sex × Time) + (Fate × Time) -788.9 1570.0 1.85 0.13 C) Seton Dam passage time Gates N = 253 Censored = 24 Sex + Fate + DTT + (Sex × Time) + (Fate × Time) -1014.8 2039.7 0.00 0.29 Sex + Fate + DTT + Ddis + (Sex × Time) + (Fate × Time) -1014.3 2040.7 0.96 0.18 Sex + Fate + DTT + Flow + (Sex × Time) + (Fate × Time) -1014.7 2041.3 1.63 0.13 Sex + Fate + GSE + DTT + (Sex × Time) + (Fate × Time) -1014.7 2041.4 1.66 0.13 Portage N = 183 Censored = 23 Sex + Fate + DTT + Ddis + (Sex × Time) + (Fate × Time) -695.3 1402.5 0.00 0.28 Sex + Fate + Ddis + (Sex × Time) + (Fate × Time) -696.7 1403.5 0.98 0.17 Sex + Fate + GSE + DTT + Ddis + (Sex × Time) + (Fate × Time) -694.8 1403.7 1.15 0.16 Sex + Fate + (Sex × Time) + (Fate × Time) -697.9 1404.0 1.46 0.13 78 Model log Lik AICc Δ AICc wi D) Survival to spawning grounds Gates N = 229 Successful = 156 Unsuccessful = 73 Sex + TMT -133.83 273.8 0.00 0.11 Sex + TMT + Flow -133.01 274.2 0.42 0.09 Sex + TMT + Flow + TD -132.09 274.5 0.68 0.08 Sex + Flow + TD + DD -132.21 274.7 0.91 0.07 Sex + TMT + Flow + TD + DD -131.39 275.2 1.39 0.05 Sex + TMT + TD -133.71 275.6 1.83 0.04 Sex + TMT + DD -133.79 275.8 1.98 0.04 Sex + TMT + GSE -133.79 275.8 1.98 0.04 Portage N = 160 Successful = 119 Unsuccessful = 41 TMT -86.09 176.3 0.00 0.16 TMT + DD -85.33 176.8 0.55 0.12 TMT + Sex -85.73 177.6 1.35 0.08 TMT + TD -85.90 178.0 1.71 0.07 79 3.6 Chapter 3 Figures Figure 3.1 – The Seton-Anderson watershed and locations of Gates Creek and Portage Creek sockeye salmon (O. nerka) spawning grounds (large inset), and the Seton hydrosystem (red dashed rectangle in large inset and main map) in southwestern British Columbia, Canada (small inset). The three study reaches of the hydrosystem are indicated by grey shaded polygons on the main map. Fish capture / release sites, radio-telemetry sites, temperature monitoring sites, and dam locations are indicated by the legend and map. The location of the Seton powerhouse tailrace and extent of its discharge plume of natal Seton Lake water is shown extending into the Fraser River, as is the extent of the Seton River plume that fluctuates in its concentration of natal water. ∧ ∧ ∧ ∧ Powerhouse Fraser River Creek River Upper Lower River Natal ¤ Water Diversion Gates Creek Portage Creek Lake British Columbia 200 km Seton Dam & fishway Radio receiver Release loca on Seton River plume Powerhouse plume Water flow direc on ¤ Capture loca on Walden North dam 500 m NTemperature / Conduc vity 3 km Reach 2 Reach 1 Reach 3 80 Temperature (°C)681012141618202224 (a)Natal Water (%)60708090100(b)Dam discharge (m3s−1)Aug 01 Aug 08 Aug 15 Aug 22 Aug 29 Sep 05 Sep 12 Sep 19 Sep 26 Oct 03 Oct 10 Oct 17 Oct 24 Oct 31152025303540(c)DateFigure 3.2 – Mean daily temperature (a), natal water concentration of the lower Seton River (b), and discharge from the Seton Dam (c) during the 2013 (grey lines) and 2014 (black lines) Gates Creek (~ 1 August – 5 September) and Portage Creek (~ 15 September – 31 October) sockeye salmon migrations through the lower Seton hydrosystem. Temperatures in panel ‘a’ are given for the Fraser River at Texas Creek (solid lines), the Seton River (dashed lines), and the Seton Dam tailrace (dotted lines). 81 00.250.50.751Females(a)Proportion entered Seton River0 50 100 150 200 250 300 350 400 450 50000.250.50.751Time from Fraser River release (hours)Males(b)●●−0.4 0.0 0.4 0.8 1.2NWSRTFRTGSEFate x TimeFateSex x TimeSex (c) ●●●−0.5 −0.1 0.3 0.7(d)Standardized coefficientsFigure 3.3 - Panel a & b – Passage time curves through reach 1 of the Seton hydrosystem from release in the Fraser River to Seton River entry for Gates Creek (black lines) and Portage Creek (grey lines) female (a) and male (b) sockeye salmon by fate. Solid lines indicate eventually successful migrants (survived to natal spawning grounds); broken lines indicate unsuccessful migrants (did not survive to natal spawning grounds); open circles indicate censored individuals. Panel c & d – Model-averaged estimates of odds ratios for standardized covariates in time-to-event proportional hazards regression for (c) Gates Creek and (d) Portage Creek passage times through the first reach of the Seton hydrosystem. Error bars represent 95% confidence intervals for odds ratios; closed black circles highlight odds ratios with confidence intervals that do not include zero. Vertical dashed lines indicate the coefficient value of zero. Abbreviations for covariates are: (GSE) gross somatic energy; (FRT) Fraser River temperature; (SRT) Seton River temperature; (NW) natal water concentration. Precise values for odds ratios and 95% CIs are given by covariate in Appendix 2. 82 00.250.50.751Proportion arrived at Seton DamFemales(a)0 10 20 30 40 50 60 70 80 9000.250.50.751Time from Seton River entry (hours)Males(b)●●●●●●−0.5 0.1 0.4 0.7 1.0 1.3NWSRTGSEFate x TimeFateSex x TimeSex (c)●−0.2 0.1 0.4 0.7 1.0(d)Standardized coefficientsFigure 3.4 – Panel a & b – Passage time curves through reach 2 of the Seton hydrosystem from Seton River entry to arrival at the Seton Dam for Gates Creek (black lines) and Portage Creek (grey lines) female (a) and male (b) sockeye salmon by fate. Solid lines indicate eventually successful migrants (survived to natal spawning grounds); broken lines indicate unsuccessful migrants (did not survive to natal spawning grounds); open circles indicate censored individuals. Panel c & d – Model-averaged estimates of odds ratios for standardized covariates in time-to-event proportional hazards regression for Gates Creek (c) and Portage Creek (d) passage times through the second reach of the Seton hydrosystem. Error bars represent 95% confidence intervals for odds ratios; closed black circles highlight odds ratios with confidence intervals that do not include zero. Vertical dashed lines indicate the coefficient value of zero. Abbreviations for coefficients are: (GSE) gross somatic energy; (SRT) Seton River temperature; (NW) natal water concentration. Precise values for odds ratios and 95% CIs are given by covariate in Appendix 2. 83 00.250.50.751Proportion passed Seton DamFemales(a)0 10 20 30 40 50 60 70 80 90 10000.250.50.751Time from Seton Dam arr ival (hours)Males(b)●●−0.05 0.15 0.35 0.55FlowDdisDtempGSEFate x TimeFateSex x TimeSex (b)−0.2 0.0 0.2 0.4 0.6(c)Standardized coefficientsFigure 3.5 – Panel a & b – Seton Dam passage time curves for Gates Creek (black lines) and Portage Creek (grey lines) female (a) and male (b) sockeye salmon by fate. Solid lines indicate eventually successful migrants (survived to natal spawning grounds); broken lines indicate unsuccessful migrants (did not survive to natal spawning grounds); open circles indicate censored individuals. Panel c & d – Model-averaged estimates of odds ratios for standardized covariates in time-to-event proportional hazards regression for Seton Dam passage time of Gates Creek (c) and Portage Creek (d) sockeye salmon. Error bars represent 95% confidence intervals for odds ratios; closed black circles highlight odds ratios with confidence intervals that do not include zero. Vertical dashed lines indicate the coefficient value of zero. Abbreviations for coefficients are: (GSE) gross somatic energy; (Dtemp) Seton Dam tailrace temperature; (Ddis) Seton Dam discharge; (Flow) binary covariate for experimental flow scenario change (Gates migration only). Precise values for odds ratios and 95% CIs are given by covariate in Appendix 2. 84 050100150200250300350400450GC Females GC Males PC Females PC MalesTotal migration time (hours) (a)●−0.5 0.0 0.4 0.8FlowTMTDDTDGSESex (b)−0.8 −0.4 0.0(c)Standardized coefficientsFigure 3.6 – Panel a – Beanplot of cumulative hydrosystem migration times from release in the Fraser River to upstream of the Seton Dam for Gates Creek and Portage Creek male and female sockeye salmon that were ultimately successful (survived to natal spawning grounds; black beans) and unsuccessful (did not survive to natal spawning grounds; grey beans) following dam passage. Beans represent distribution of individual passage time times (small horizontal lines); large black horizontal lines represent means. Panel b & c – Model-averaged standardized coefficient estimates from generalized linear models predicting survival to natal spawning sites for all Gates Creek (b) and Portage Creek (c) sockeye salmon that passed the Seton Dam. Error bars represent 95% confidence intervals for coefficient estimates; closed black circles highlight coefficients with confidence intervals that do not include zero. Vertical dashed lines indicate the coefficient value of zero. Abbreviations for coefficients are: (GSE) gross somatic energy; (TD) tagging date as an index of run timing; (DD) degree-day accumulation from release in the Fraser River to Seton Dam passage; (TMT) total migration time from release to dam passage. (Flow) binary covariate for experimental flow scenario change experienced at the time of dam passage (Gates migration only). Precise values for coefficients and 95% CIs are given by covariate in Appendix 2. 85 Chapter 4: Summary and conclusions Sockeye salmon are famous for their long distance reproductive migrations that are carried out with remarkable fidelity and precision from the high seas to natal freshwater spawning sites. My thesis research examined the migration behaviour and success of these fish during the final stages of this migration as adults passed from a large mainstem river into a natal tributary affected by hydroelectric development while enroute to spawning grounds. In this hydrosystem, discharge from a powerhouse and mixing of water sources can cause large fluctuations in the availability and concentration of natal water cues that adults use to guide homing. This unique set of conditions also allowed me to explore how varying concentrations of natal water can affect the behaviour of adult salmon near the final stages of their migration. In Chapter 2, I used non-invasive biopsy and radio telemetry to measure physiological condition and to monitor the total amount of migration delay incurred by individuals in the tailrace of a powerhouse and the subsequent wandering behaviour of fish between this facility and the entrance to their natal migration corridor. I used linear- and generalized-linear models combined with model selection techniques to investigate whether these behaviours were related to natal water cues and temperature, as well as the stress and reproductive status of individuals. In Chapter 3, I examined migration behaviour during passage through the entire hydrosystem, from downstream of the powerhouse to upstream of the dam, and then assessed post-passage survival to natal sites. I again applied model selection techniques but in this chapter used proportional hazards regression to describe in detail how natal water cues, temperature, energy status, and the ultimate fate of individuals was related to passage times through three specific reaches of the hydrosystem. I then summarized the migration experience of all fish that successfully passed through the hydrosystem, and used generalized linear models to examine how the cumulative effects of this hydrosystem migration experience affected post-passage survival to natal sites. Below I discuss my findings and how they relate to the ecology of homing anadromous salmon, some of the shortcomings of my research, and how this information can be used to inform local fisheries managers about the conservation of sockeye salmon in the Seton-Anderson watershed. 86 Consistent with past observations and studies in this system (Andrew and Geen 1958, Fretwell 1989), I found that nearly all tagged sockeye-salmon (GC = 87%, PC = 91%; Chapter 2) released over the two years of my study delayed in the powerhouse and wandered in the Fraser River before eventually passing into the Seton River – which nearly all fish did (GC = 92%, PC = 87%; Chapter 3). I also found that over three quarters of all tagged migrants were eventually able to re-ascend the hydrosystem and successfully pass the Seton Dam (Chapter 3). PC fish delayed roughly five times longer than GC fish in the powerhouse (Chapter 2), and took nearly fives times as long to pass through the whole Seton hydrosystem following release (Chapter 3). Though striking, these differences were not unexpected, as milling/delay and generally slower migrations are an important part of the life history of late run Fraser River sockeye salmon (e.g. PC fish) (English et al. 2005). Despite being more mature and having higher levels of stress indices, females from both populations did not exhibit more powerhouse delay or wandering behaviour than males. In fact, I found few associations, within a population, that physiological stress or maturation influenced these behaviours. This was likely because delaying in the powerhouse and wandering in the Fraser River did not offer significant enough energetic challenges or obstacles (e.g. high discharge, temperatures) necessitating females or more stressed fish attempt to conserve energy by delaying or wandering less. Indeed, when examining behaviour at a broader scale during passage through the whole hydrosystem, females and those with higher energy content were often markedly slower compared to males and fish with lower energy, which is consistent with other reach-scale observations of migration rates for Fraser River sockeye salmon (Young et al. 2006; Hanson et al. 2008). GC females were also far less likely than males to survive to natal spawning sites following passage of the Seton Dam, which is consistent with previous observations of GC sockeye salmon mortality in this system (Roscoe et al. 2010b; Burnett et al. 2013), and likely a result of the physiological and energetic demands of passing the Seton Dam (Burnett et al. 2014). It is concerning that females consistently incur greater losses than males after passing the dam given that the number of progeny in subsequent years 87 ultimately depends on the number of females that survive to spawn (Quinn 2005). With this in mind, I recommend that managers continue to explore options to mitigate the stresses associated with passing the Seton Dam, as it has recently been shown that changes to the hydraulic conditions in the dam tailrace can improve female survival to spawning grounds (Burnett et al. 2016). I would further argue that my results indicating GC fish passed the dam more rapidly at the onset of stressful temperatures (Chapter 3) add to the necessity and urgency of exploring such alternative passage options, as rapid dam passage during high temperatures would only exacerbate the physiological and energetic demands of passage (e.g. Burnett et al. 2014) and result in further mortality, particularly among females and as temperatures in the Fraser River watershed continue to warm (Ferrari et al. 2007). The majority of tagged PC and GC fish incurred delays of varying degrees at the powerhouse, in the Fraser River, and at the Seton Dam prior to ascending the fishway. Many of these fish also experienced extremely high temperatures and relatively low concentrations of natal water, and exhibited varying levels of energy, stress, and maturation. While some of these factors contributed to slowed migrations in reaches leading up to the Seton Dam, I had also expected that any or all of these features would have contributed to patterns of migration mortality following dam passage, e.g. there would either be carry-over or latent effects of hydrosystem passage (Burnett et al. 2016). Aside from the sex of GC fish (females did not survive as well as males) and that slower migrating fish generally did not survive as well (Chapter 3), I found little evidence to suggest that a fish’s post-dam passage fate was related to any of the pre-dam passage variables I examined. One explanation is that passage for many individuals (especially those in 2014) was through relatively benign flow and temperature conditions. Had temperatures been at or near lethal levels in the Fraser or Seton Rivers, as has been observed in other recent years, or if flows in the Seton River were significantly higher (as has been observed in 2010; Roscoe 2009), I may have observed stronger latent effects of hydro-system passage. Longer passage times through the Columbia River hydrosystem have been associated with reduced survival to natal sites in sockeye salmon (Naughton et al. 2005); however, I suspect the mechanisms underlying these results are very different. For instance, Columbia River sockeye salmon pass up to nine dams in order to reach 88 spawning grounds (Naughton et al. 2005), so energetic and physiological demands would have almost certainly be more pronounced in this system. Whereas sockeye salmon in my study migrated ~ 340 km upstream in the un-impounded Fraser River prior to entering the Seton hydrosystem where they likely encountered a suite of stressors that could result in slowed hydrosystem passage and post-passage failure reaching natal sites in some years – factors that I did not examine in my study. For example, physiological impairment arising from escape/release from the gauntlet of lower Fraser River fisheries, or pathogen and disease development brought on by injuries and warming rivers can cause slowed and ultimately failed migrations (Donaldson et al. 2011; Miller et al. 2014; Raby et al. 2015). Indeed, I observed a large portion of GC and PC migrants with obvious physical signs of fisheries encounters and disease (e.g. hook wounds, gillnet marks, body/gill fungus) while tagging fish for my study. In 2015, 30-50% of migrants arriving at the capture fence on some days had visible signs of previous net entanglement (Casselman et al. 2016). I opted not to tag visibly injured or diseased fish, but had I done so, I would expect that those fish would have had enhanced difficulty passing through the hydrosystem. In 2015 our group observed 100s of dead adult sockeye washed up on the ‘upstream’ side of the capture fence; these were migrants that had been allowed to pass through the fence without tagging because they had visible signs of injury – and these fish were not able to ascend the dam (Casselman et at. 2016). Also I cannot conclude from my analyses that a fish’s arrival at natal sites resulted in its successful spawning. Burnett et al. (2016) recently found that apart from affecting survival, passing the Seton Dam can have carry-over effects that affect egg deposition in GC females on spawning grounds. Thus, I suggest that future studies that examine the cumulative effects of passage through the Seton hydrosystem, or any other system, account for the health of individual migrants as well as assess the spawning success of fish that arrive on spawning grounds (e.g. Burnett et al. 2016). Arguably the most unique aspect of this thesis was the examination of how varying concentrations of natal water cues affected the homing behaviour of GC and PC sockeye salmon. Despite a large range of natal water experienced by both populations (~ 75 – 95%), GC and PC fish elicited contrasting behavioural responses on the two different scales at which I examined behaviour. On the finer scale of powerhouse delay 89 and wandering in the Fraser River, natal water concentrations had very little effect on GC fish behaviour. While on the other hand, even slight decreases in this parameter were associated with PC fish increasing wandering and delay in the powerhouse. I had expected these population-specific behavioural responses to natal water to scale up and persist when examining passage through the first two reaches of the hydrosystem; although on the contrary, passage of PC fish was largely unaffected by varying natal water concentrations, while relatively higher concentrations of natal water caused GC fish to slow down when migrating up the Seton River. I suspect these results may be explained by the differences in maturity and energy levels between fish from the two populations. The less mature GC fish with more energy probably slowed their migrations when encountering strong natal water cues in the Seton River in an effort to time the ripening of their gonads with arrival on spawning grounds, similar to how less mature GC fish with higher energy content have been observed to hold longer in the two natal lakes above the Seton Dam before carrying on to spawning grounds (Roscoe et al. 2010a). PC fish were far more mature than GC fish, which has been associated with increased olfactory sensitivity (Nevitt and Dittman 1998, Carruth et al. 2002), and may explain why even small changes in natal water led these fish to increase powerhouse delay and Fraser River wandering. However, this increased olfactory sensitivity may also prove advantageous to fish homing through this system in that once PC migrants pass the powerhouse and enter the Seton River, they may be less likely to alter their behaviour, and continue migrating up the Seton even if natal water levels fluctuate. While a number of lab based experiments have demonstrated how even slight changes to natal water cues can elicit behavioural responses in adult salmon (reviewed in Bett and Hinch 2015), in this thesis I have provided some of the first detailed descriptions of how such changes to natal water concentrations can affect behaviour in wild, homing adult salmon. With respect to the efficacy and maintenance of the 80% and 90% natal water targets during the GC and PC migrations, respectively, my observations that nearly all tagged fish experienced concentrations above and below these targets yet still delayed in the powerhouse, wandered in the Fraser River, and eventually entered the Seton River to migrate up to the dam, suggest these targets are acceptable, and that managers continue to strive to maintain the highest levels possible, especially during the PC migration. In 90 addition, natal water conditions over the two years of my field studies were far more variable in 2013 than in 2014, thus some of the differences in behaviour may have also been confounded by annual variability in fish condition and other environmental factors unique to specific years that I did not measure. Though the physiological condition of tagged fish did not differ between years (Chapter 2), large reductions in natal water in 2013 coincided with temperatures above 22C in the Fraser and Seton Rivers, which is well above thermal optima for the GC population (Lee et al. 2003) and known to affect swimming performance, behaviour, and survival (Eliason et al. 2011). Thus my study would have benefitted from a more controlled experimental approach to examining the effects of natal water, which could be done by manipulating natal water concentrations in situ during migrations when other confounding factors such as temperature are not likely to be an issue. Indeed, such an experiment was conducted in August 2015, with operations at Walden North dam intentionally altered so as to decrease natal water concentrations in the Seton River below target thresholds (below 80%) for one week during the GC migration, followed by an increase in natal water levels above target thresholds for another week (Casselman et al. 2016). Preliminary results suggest that salmon that experienced below target natal water concentrations exhibited increased wandering behaviour in the Fraser River and decreased survival to Seton Dam as compared with fish that experienced above-target concentrations. Survival to Seton Dam for fish released during the week of decreased natal water concentrations was up to 25% lower than fish released in the following week. However, there was no difference in the time spent in the powerhouse tailrace or in the time to reach Seton Dam. My results, taken together with those of this recent experiment (and those of recent Y-maze behavioural experiments Casselman et al. 2016) illustrate how subtle changes in olfactory cues can influence migratory behaviour, and how important it is for managers to ensure that current threshold natal water targets applied to the Seton hydrosystem are maintained or increased during the PC migration and monitored into the future. 91 References Alerstam, T., Hedenström, A., and Åkesson, S. 2003. Long‐ distance migration: evolution and determinants. Oikos 103:247–260. Allison, P.D. 2010. Survival analysis using SAS. SAS Institute. Andrew, F.J., and Geen, G.H. 1958. Sockeye and pink salmon investigations at the Seton Creek hydroelectric installation. International Pacifc Salmon Fisheries Commission. New Westminster, B.C., Canada. Beeman, J.W., and Perry R.W. 2012, Bias from false-positive detections and strategies for their removal in studies using telemetry. Pp 505 – 518 in Telemetry Techniques: A User Guide for Fisheries Research. Adams, N S, Beeman, J W & Eiler, J H (eds). American Fisheries Society, Bethesda, MD. Bett, N.N., and Hinch, S.G. 2015. Olfactory navigation during spawning migrations: a review and introduction of the Hierarchical Navigation Hypothesis. Biological Reviews. doi:10.1111/brv.12191. Boggs, C.T., Keefer, M.L., Peery, C.A., Bjornn, T.C., and Stuehrenberg, L.C. 2004. Fallback, reascension, and adjusted fishway escapement estimates for adult Chinook salmon and steelhead at Columbia and Snake River Dams. Transactions of the American Fisheries Society 133:932–949. Brett, J.R. 1971. Energetic responses of salmon to temperature - study of some thermal relations in physiology and freshwater ecology of sockeye salmon (Oncorhynchus nerka). American Zoologist 11: 99–113. Bugert, R.M., and Mendel, G.W. 1997. Adult returns of subyearling and yearling fall Chinook salmon released from a Snake River hatchery or transported downstream. North American Journal of Fisheries Management 17:638–651. Burnett, N.J., Hinch, S.G., Bett, N.N., Braun, D.C., Casselman, M.T., Cooke, S.J., Gelchu, A., Lingard, S., Middleton, C.T., Minke-Martin, V., and White, C.F.H. 2016. Reducing carryover effects on the migration and spawning success of sockeye salmon through a management experiment of dam flows. River Research and Applications 00:000-000. Burnett, N.J., Hinch, S.G., Braun, D.C., Casselman, M.T., Middleton, C.T., Wilson, S.M., and Cooke, S.J. 2014. Burst swimming in areas of high flow: delayed consequences of anaerobiosis in wild adult sockeye salmon. Physiological and Biochemical Zoology 87: 587–598. 92 Burnett, N.J., Hinch, S.G., Donaldson, M.R., Furey, N.B., Patterson, D.A., Roscoe, D.W., and Cooke, S.J. 2013. Alterations to dam-spill discharge influence sex-specific activity, behaviour and passage success of migrating adult sockeye salmon. Ecohydrology 7. Burnham, K.P., and Anderson, D.R. 2003. Model selection and multimodel inference. Springer Science & Business Media, New York, NY. Carruth, L.L.L., Jones, R.E.R., and Norris, D.O.D. 2002. Cortisol and Pacific salmon: a new look at the role of stress hormones in olfaction and home-stream migration. Integrative and Comparative Biology 42:574–581. Casselman, M.T., Burnett, N.J., Bett, N.N., Middleton, C.T., Minke-Martin, V.M., Braun, D.C., McCubbing, D., and Hinch, S.G. 2015. BRGMON-14 Effectiveness of Cayoosh Creek flow dilution, dam operation, and fishway passage on delay and survival of upstream migration of salmon in the Seton-Anderson watershed. Annual Report - 2014. Report prepared for St’át’imc Government Services and BC Hydro. The University of British Columbia, Vancouver, BC. 67p. Casselman, M.T., Bett, N.N., Drenner, S.M., Middleton, C.T., Minke-Martin, V.M., and Hinch, S.G. 2016. BRGMON-14 Effectiveness of Cayoosh Creek flow dilution, dam operation, and fishway passage on delay and survival of upstream migration of salmon in the Seton-Anderson watershed. Annual Report - 2015. Report prepared for St’át’imc Government Services and BC Hydro. The University of British Columbia, Vancouver, BC. Castro-Santos, T., and Haro, A. 2003. Quantifying migratory delay: a new application of survival analysis methods. Canadian Journal of Fisheries and Aquatic Sciences 60: 986–996. Caudill, C.C., Daigle, W.R., Keefer, M.L., Boggs, C.T., Jepson, M.A., Burke, B.J., Zabel, R.W., Bjornn, T.C., and Peery, C.A. 2007. Slow dam passage in adult Columbia River salmonids associated with unsuccessful migration: delayed negative effects of passage obstacles or condition-dependent mortality? Canadian Journal of Fisheries and Aquatic Sciences 64:979–995. Caudill, C.C., Keefer, M.L., Clabough, T.S., Naughton, G.P., Burke, B.J., and Peery, C.A. 2013. Indirect effects of impoundment on migrating fish: temperature gradients in fish ladders slow dam passage by adult Chinook salmon and steelhead. PLoS ONE 8: doi:10.1371/journal.pone.0085586. Clabough, T.S., Caudill, C.C., Bjornn, T.C., and Stuehrenberg, L.C. 2006. Associations between adult salmon and steelhead body temperature during upstream migration and estimated environmental temperatures in lower granite reservoir during cold water releases from Dworshak Reservoir, 2001-2002. Idaho Cooperative Fish and Wildlife Research Unit, University of Idaho, Moscow, Idaho, USA. 93 Cooke, S.J., Crossin, G.T., Patterson, D.A., English, K.K., Hinch, S.G., Young, J.L., Alexander, R.F., Healey, M.C., Van Der Kraak, G., and Farrell, A.P. 2005. Coupling non-invasive physiological assessments with telemetry to understand inter-individual variation in behaviour and survivorship of sockeye salmon: development and validation of a technique. Journal of Fish Biology 67:1342–1358. Cooke, S.J., Hinch, S.G., Crossin, G.T., Patterson, D.A., English, K.K., Healey, M.C., Shrimpton, J.M., Van Der Kraak, G., and Farrell, A.P. 2006. Mechanistic basis of individual mortality in Pacific salmon during spawning migrations. Ecology 87: 1575–1586. Crossin, G.T., and Hinch, S.G. 2005. A nonlethal, rapid method for assessing the somatic energy content of migrating adult Pacific salmon. Transactions of the American Fisheries Society 134: 184–191. Crossin, G.T., Hinch, S.G., and Farrell, A.P. 2004. Somatic energy of sockeye salmon Oncorhynchus nerka at the onset of upriver migration: a comparison among ocean climate regimes. Fisheries Oceanography 13:345-349. Crossin, G.T., Hinch, S.G., Cooke, S.J., Cooperman, M.S., Patterson, D.A., Welch, D.W., Hanson, K.C., Olsson, I., English, K.K., and Farrell, A.P. 2009. Mechanisms influencing the timing and success of reproductive migration in a capital breeding semelparous fish species, the sockeye salmon. Physiological and Biochemical Zoology 82: 635–652. Crossin, G.T., Hinch, S.G., Cooke, S.J., Welch, D.W., Batten, S.D., Patterson, D.A., Van Der Kraak, G., Shrimpton, J.M., and Farrell, A.P. 2007. Behaviour and physiology of sockeye salmon homing through coastal waters to a natal river. Marine Biology 152: 905–918. Crossin, G.T., Hinch, S.G., Cooke, S.J., Welch, D.W., Patterson, D.A., Jones, S.R.M., Lotto, A.G., Leggatt, R.A., Mathes, M.T., Shrimpton, J.M., Van Der Kraak, G., and Farrell, A.P. 2008. Exposure to high temperature influences the behaviour, physiology, and survival of sockeye salmon during spawning migration. Canadian Journal of Zoology 86: 127–140. Dingle, H. 2014. Migration - the biology of life on the move. Oxford University Press, UK. Dingle, H., and Drake, V.A. 2007. What is migration? BioScience 57: 113. Dittman, A., and Quinn, T. 1996. Homing in Pacific salmon: mechanisms and ecological basis. Journal of Experimental Biology 199: 83-91. 94 Donaldson, M.R., Cooke, S.J., Patterson, D.A., Hinch, S.G., Robichaud, D., Hanson, K.C., Olsson, I., Crossin, G.T., English, K.K., and Farrell, A.P. 2009. Limited behavioural thermoregulation by adult upriver-migrating sockeye salmon (Oncorhynchus nerka) in the Lower Fraser River, British Columbia. Canadian Journal of Zoology 86: 480–490. Donaldson, M.R., Hinch, S.G., Patterson, D.A., Farrell, A.P., Shrimpton, J.M., Miller Saunders, K.M., Robichaud, D., Hills, J., Hruska, K.A., Hanson, K.C., English, K.K., Van Der Kraak, G., and Cooke, S.J. 2010. Physiological condition differentially affects the behavior and survival of two populations of sockeye salmon during their freshwater spawning migration. Physiological and Biochemical Zoology 83: 446–458. Donaldson, M.R., Hinch, S.G., Patterson, D.A., Hills, J., Thomas, J.O., Cooke, S.J., Raby, G.D., Thompson, L.A., Robichaud, D., English, K.K., and Farrell, A.P. 2011. The consequences of angling, beach seining, and confinement on the physiology, post-release behaviour and survival of adult sockeye salmon during upriver migration. Fisheries Research 108:133–141. Doving, K.B., and Stabell, O.B. 2003.Trails in open waters: sensory cues in salmon migration. In Sensory Processing in Aquatic Environments. Springer-Verlag, New York, Inc. Drenner, S.M., Hinch, S.G., Martins, E.G., Furey, N.B., Clark, T.D., Cooke, S.J., Patterson, D.A., Robichaud, D., Welch, D.W., Farrell, A.P., and Thomson, R.E. 2015. Environmental conditions and physiological state influence estuarine movements of homing sockeye salmon. Fisheries Oceanography 24: 307–324. Eliason, E.J., and Farrell, A.P. 2015. Oxygen uptake in Pacific salmon Oncorhynchus spp.: when ecology and physiology meet. Journal of Fish Biology: doi:10.1111/jfb.12790 Eliason, E.J., Clark, T.D., Hague, M.J., Hanson, L.M., Gallagher, Z.S., Jeffries, K.M., Gale, M.K., Patterson, D.A., Hinch, S.G., and Farrell, A.P. 2011. Differences in thermal tolerance among sockeye salmon populations. Science 332: 109–112. English, K.K., Koski, W.R., Sliwinski, C., Blakley, A., Cass, A., and Woodey, J.C. 2005. Migration timing and river survival of late-run Fraser River sockeye salmon estimated using radiotelemetry techniques. Transactions of the American Fisheries Society 134: 1342–1365. Farrell, A.P., Gamperl, A.K., and Birtwell, I.K. 1998. Prolonged swimming, recovery and repeat swimming performance of mature sockeye salmon Oncorhynchus nerka exposed to moderate hypoxia and pentachlorophenol. Journal of Experimental Biology 201: 2183–2193. 95 Farrell, A.P., Hinch, S.G., Cooke, S.J., Patterson, D.A., Crossin, G.T., Lapointe, M., and Mathes, M.T. 2008. Pacific salmon in hot water: applying aerobic scope models and biotelemetry to predict the success of spawning migrations. Physiological and Biochemical Zoology 81: 697–708. Fenkes, M., Shiels, H.A., Fitzpatrick, J.L., and Nudds, R.L. 2015. The potential impacts of migratory difficulty, including warmer waters and altered flow conditions, on the reproductive success of salmonid fishes. Comparative Biochemistry and Physiology, Part A: 1–11. Ferrari, M.R., Miller, J.R., and Russell, G.L. 2007. Modeling changes in summer temperature of the Fraser River during the next century. Journal of Hydrology 342: 336–346. Ficklin, D.L., Barnhart, B.L., Knouft, J.H., Stewart, I.T., Maurer, E.P., Letsinger, S.L., and Whittaker, G.W. 2014. Climate change and stream temperature projections in the Columbia River basin: habitat implications of spatial variation in hydrologic drivers. Hydrology and Earth Systems Science 18: 4897–4912. Fox, J.F., and Weisberg, S. 2011. Cox proportional-hazards regression for survival data in R: An appendix to An R companion to applied regression, Second Edition. Sage Publishing. Fretwell, M.R. 1989. Homing behaviour of adult sockeye salmon in reponse to a hydroelectric diversion of homestream waters at Seton Creek. International Pacific Salmon Fisheries Commission. New Westminster, BC, Canada. Gelman, A. 2008. Scaling regression inputs by dividing by two standard deviations. Statistics in Medicine 27: 2865–2873. Goniea, T.M., Keefer, M.L., Bjornn, T.C., Peery, C.A., Bennett, D.H., and Stuehrenberg, L.C. 2011. Behavioral thermoregulation and slowed migration by adult fall Chinook salmon in response to high Columbia River water temperatures. Transactions of the American Fisheries Society 135: 408–419. Gowans, A.R.D., Armstrong, J.D., and Priede, I.G. 1999. Movements of adult Atlantic salmon in relation to a hydroelectric dam and fish ladder. Journal of Fish Biology 54: 713–726. Gowans, A.R.D., Armstrong, J.D., Priede, I.G., and Mckelvey, S. 2003. Movements of Atlantic salmon migrating upstream through a fish‐ pass complex in Scotland. Ecology of Freshwater Fish 12: 177–189. Gross, M.R., Coleman, R.M., and McDowall, R.M. 1988. Aquatic productivity and the evolution of diadromous fish migration. Science 239: 1291–1293. 96 Grueber, C.E., Nakagawa, S., Laws, R.J., and Jamieson, I.G. 2011. Multimodel inference in ecology and evolution: challenges and solutions. Journal of Evolutionary Biology 24: 699–711. Hansen, L.P., and Jonsson, B. 1994. Homing of Atlantic salmon: effects of juvenile learning on transplanted post-spawners. Animal Behaviour 47: 220–222. Hanson, K.C., Cooke, S.J., Hinch, S.G., Crossin, G.T., Patterson, D.A., English, K.K., Donaldson, M.R., Shrimpton, J.M., Van Der Kraak, G., and Farrell, A.P. 2008. Individual variation in migration speed of upriver‐ migrating sockeye salmon in the Fraser River in relation to their physiological and energetic status at marine approach. Physiological and Biochemical Zoology 81: 255–268. Hasler, A.D., and Scholz, A.T. 1983. Olfactory imprinting and homing in salmon. Springer-Verlag. Hasler, A.D., and Wisby, W.J. 1951. Discrimination of stream odors by fishes and its relation to parent stream behavior. The American Naturalist 85: 223–238. Hinch, S.G., and Bratty, J. 2000. Effects of swim speed and activity pattern on success of adult sockeye salmon migration through an area of difficult passage. Transactions of the American Fisheries Society 129: 598–606. Hinch, S.G., and Martins, E.G. 2011. A review of potential climate change effects on survival of Fraser River sockeye salmon and an analysis of interannual trends in en route loss and pre-spawn mortality. The Cohen Commission of Inquiry into the Decline of Sockeye Salmon in the Fraser River. Vancouver, B.C., Canada. Hinch, S.G., and Rand, P.S. 1998. Swim speeds and energy use of upriver-migrating sockeye salmon (Oncorhynchus nerka): role of local environment and fish characteristics. Canadian Journal of Fisheries and Aquatic Sciences 55: 1821–1831. Hinch, S.G., and Rand, P.S. 2000. Optimal swimming speeds and forward-assisted propulsion: energy-conserving behaviours of upriver-migrating adult salmon. Canadian Journal of Fisheries and Aquatic Sciences 57: 2470–2478. Hinch, S.G., Cooke, S.J., Farrell, A.P., Miller, K.M., Lapointe, M., and Patterson, D.A. 2012. Dead fish swimming: a review of research on the early migration and high premature mortality in adult Fraser River sockeye salmon Oncorhynchus nerka. Journal of Fish Biology 81: 576–599. Hinch, S.G., Cooke, S.J., Healey, M.C., and Farrell, A.P. 2006. Behavioural physiology of fish migrations: salmon as a model approach. In Fish Physiology: 239–295. 97 Hodgson, S., and Quinn, T.P. 2002. The timing of adult sockeye salmon migration into fresh water: adaptations by populations to prevailing thermal regimes. Canadian Journal of Zoology 80: 542–555. Hubert, W.A., Pope, K.L., and Dettmers, J.M. 2012. Passive Capture Techniques. Nebraska Cooperative Fish and Wildlife Research Unit -- Staff Publications. Paper 111. Isaak, D.J., Wollrab, S., Horan, D., and Chandler, G. 2012. Climate change effects on stream and river temperatures across the northwest U.S. from 1980–2009 and implications for salmonid fishes. Climatic Change 113: 499–524. Keefer, M.L., and Caudill, C.C. 2014. Homing and straying by anadromous salmonids: a review of mechanisms and rates. Reviews in Fish Biology and Fisheries 24: 333-368. Keefer, M.L., and Joosten, D.C. 2008. Adult salmon and steelhead passage through fishways and transition pools at Bonneville Dam, 1997-2002. Report for project MPE-P-95-1 for U.S. Army Corps of Engineers and Benneville Power Administration. Portland, OR, USA. Keefer, M.L., Caudill, C.C., Peery, C.A., and Boggs, C.T. 2008a. Non-direct homing behaviours by adult Chinook salmon in a large, multi-stock river system. Journal of Fish Biology 72: 27–44. Keefer, M.L., Peery, C.A., and Heinrich, M.J. 2008b. Temperature-mediated en route migration mortality and travel rates of endangered Snake River sockeye salmon. Ecology of Freshwater Fish 17: 136–145. Keefer, M.L., Peery, C.A., and High, B. 2009. Behavioral thermoregulation and associated mortality trade-offs in migrating adult steelhead (Oncorhynchus mykiss): variability among sympatric populations. Canadian Journal of Fisheries and Aquatic Sciences 66: 1734–1747. Keefer, M.L., Peery, C.A., and Jepson, M.A. 2004a. Upstream migration rates of radio‐tagged adult Chinook salmon in riverine habitats of the Columbia River basin. Journal of Fish Biology 65: 1126-1141. Keefer, M.L., Peery, C.A., Bjornn, T.C., Jepson, M.A., and Stuehrenberg, L.C. 2004b. Hydrosystem, dam, and reservoir passage rates of adult Chinook salmon and steelhead in the Columbia and Snake Rivers. Transactions of the American Fisheries Society 133: 1413–1439. Keefer, M.L., Peery, C.A., Jepson, M.A., Tolotti, K.R., Bjornn, T.C., and Stuehrenberg, L.C. 2004c. Stock-specific migration timing of adult spring–summer Chinook salmon in the Columbia River basin. North American Journal of Fisheries Management 24: 1145–1162. 98 Kleinbaum, D.G., and Klein, M. 2012. Survival Analysis: a self-learning text. Third Edition. Springer Science & Business Media, New York. Kubokawa, K., Watanabe, T., Yoshioka, M., and Iwata, M. 1999. Effects of acute stress on plasma cortisol, sex steroid hormone and glucose levels in male and female sockeye salmon during the breeding season. Aquaculture 172: 335–349. Leduc, A.O.H.C., Roh, E., Macnaughton, C.J., Benz, F., Rosenfeld, J., and Brown, G.E. 2010. Ambient pH and the response to chemical alarm cues in juvenile Atlantic salmon: mechanisms of reduced behavioral responses. Transactions of the American Fisheries Society 139: 117–128. Lee, C.G., Farrell, A.P., Lotto, A., MacNutt, M.J., Hinch, S.G., and Healy, M.C. 2003. The effect of temperature on swimming performance and oxygen consumption in adult sockeye (Oncorhynchus nerka) and coho (O. kisutch) salmon stocks. Journal of Experimental Biology 206: 3239–3251. Lundqvist, H., Rivinoja, P., Leonardsson, K., and McKinnell, S. 2008. Upstream passage problems for wild Atlantic salmon (Salmo salar L.) in a regulated river and its effect on the population. Hydrobiologia 602: 111–127. Macdonald, J.S., Morrison, J., Patterson, D.A., and Heinonen, J. 2007. Examination of factors influencing Nechako River discharge, temperature, and aquatic habitats. Canadian Technical Report of Fisheries and Aquatic Sciences 2773. Mantua, N., Tohver, I., and Hamlet, A. 2010. Climate change impacts on streamflow extremes and summertime stream temperature and their possible consequences for freshwater salmon habitat in Washington State. Climatic Change 102: 187–223. Martins, E.G., Hinch, S.G., Patterson, D.A., Hague, M.J., Cooke, S.J., Miller, K.M., Robichaud, D., English, K.K., and Farrell, A.P. 2012a. High river temperature reduces survival of sockeye salmon (Oncorhynchus nerka) approaching spawning grounds and exacerbates female mortality. Canadian Journal of Fisheries and Aquatic Sciences 69: 330-342. Martins E.G., Hinch, S.G., Cooke, S.J., and Patterson, D.A. 2012b. Climate effects on growth, phenology, and survival of sockeye salmon (Oncorhynchus nerka): a synthesis of the current state of knowledge and future research directions. Reviews in Fish Biology and Fisheries 22: 887–914. Martins, E.G., Hinch, S.G., Patterson, D.A., Hague, M.J., Cooke, S.J., Miller, K.M., Lapointe, M.F., English, K.K., and Farrell, A.P. 2011. Effects of river temperature and climate warming on stock-specific survival of adult migrating Fraser River sockeye salmon (Oncorhynchus nerka). Global Change Biology 17: 99–114. 99 Mathes, M.T., Hinch, S.G., Cooke, S.J., Crossin, G.T., Patterson, D.A., Lotto, A.G., and Farrell, A.P. 2010. Effect of water temperature, timing, physiological condition, and lake thermal refugia on migrating adult Weaver Creek sockeye salmon (Oncorhynchus nerka). Canadian Journal of Fisheries and Aquatic Sciences 67:70-84. McCullagh, P., and Nelder, J.A. 1989. Generalised linear models. Chapman and Hall. London. Miller, K.M., Li, S., Kaukinen, K.H., Ginther, N., Hammill, E., Curtis, J.M.R., Patterson, D.A., Sierocinski, T., Donnison, L., Pavlidis, P., Hinch, S.G., Hruska, K.A., Cooke, S.J., English, K.K., and Farrell, A.P. 2011. Genomic signatures predict migration and spawning failure in wild Canadian salmon. Science 331: 214–217. Miller, K.M., Teffer, A., Tucker, S., Li, S., Schulze, A.D., Trudel, M., Juanes, F., Tabata, A., Kaukinen, K.H., Ginther, N.G., Ming, T.J., Cooke, S.J., Hipfner, J.M., Patterson, D.A., and Hinch, S.G. 2014. Infectious disease, shifting climates, and opportunistic predators: cumulative factors potentially impacting wild salmon declines. Evolutionary Applications 7: 812-855. Murchie, K.J., Hair, K.P.E., Pullen, C.E., Redpath, T.D., Stephens, H.R., and Cooke, S.J. 2008. Fish response to modified flow regimes in regulated rivers: research methods, effects and opportunities. River Research and Applications 24: 197–217. Naughton, G.P., Caudill, C.C., Keefer, M.L., Bjornn, T.C., Stuehrenberg, L.C., and Peery, C.A. 2005. Late-season mortality during migration of radio-tagged adult sockeye salmon (Oncorhynchus nerka) in the Columbia River. Canadian Journal of Fisheries and Aquatic Sciences 62: 30–47. Neave, F. 1964. Ocean migrations of Pacific salmon. Journal of the Fisheries Research Board of Canada 21: 1227-1244. Nevitt, G., and Dittman, A. 1998. A new model for olfactory imprinting in salmon. Integrative Biology Issues News and Reviews 6: 215-223. Nevitt, G.A., Dittman, A.H., Quinn, T.P., and W J Moody, J. 1994. Evidence for a peripheral olfactory memory in imprinted salmon. PNAS 91: 4288–4292. Newell, J.C., and Quinn, T.P. 2005. Behavioral thermoregulation by maturing adult sockeye salmon (Oncorhynchus nerka) in a stratified lake prior to spawning. Canadian Journal of Zoology 83: 1232–1239. Nguyen, V.M., Rudd, M.A., Hinch, S.G., and Cooke, S.J. 2013. Recreational anglers' attitudes, beliefs, and behaviors related to catch-and-release practices of Pacific salmon in British Columbia. Journal of Environmental Management 128: 852–865. 100 Nielsen, M.E., Boesgaard, L., Sweeting, R.M., McKeown, B.A., and Rosenkilde, P. 1994. Plasma levels of lactate, potassium, glucose, cortisol, growth hormone and triiodo- L-thyronine in rainbow trout (Oncorhynchus mykiss) during exercise at various levels for 24 h. Canadian Journal of Zoology 72: 1643–1647. Nilsson, C., Reidy, C.A., Dynesius, M., and Revenga, C. 2005. Fragmentation and flow regulation of the world's large river systems. Science 308: 405–408. Noakes, D.J., Beamish, R.J., and Kent, M.L. 2000. On the decline of Pacific salmon and speculative links to salmon farming in British Columbia. Aquaculture 183: 363–386. Nordeng, H. 1977. A pheromone hypothesis for homeward migration in anadromous salmonids. Oikos 28: 155. O'Hara, R.B., and Kotze, D.J. 2010. Do not log-transform count data. Methods in Ecology and Evolution 1: 118–122. Patterson, D.A., Macdonald, J.S., Skibo, K.M., Barnes, D.P., Guthrie, I., and Hills, J. 2007. Reconstructing the summer thermal history for the lower Fraser River, 1941 to 2006, and implications for adult sockeye salmon. Canadian Technical Report of Fisheries and Aquatic Sciences 2724. 43p. Pon, L.B., Hinch, S.G., Cooke, S.J., Patterson, D.A., and Farrell, A.P. 2009a. Physiological, energetic and behavioural correlates of successful fishway passage of adult sockeye salmon Oncorhynchus nerka in the Seton River, British Columbia. Journal of Fish Biology 74: 1323–1336. Pon, L.B., Hinch, S.G., Cooke, S.J., Patterson, D.A., and Farrell, A.P. 2009b. A comparison of the physiological condition, and fishway passage time and success of migrant adult sockeye salmon at Seton River Dam, British Columbia, under three operational water discharge rates. North American Journal of Fisheries Management 29: 1195–1205. Putman, N.F., Scanlan, M.M., Billman, E.J., O’Neil, J.P., Couture, R.B., Quinn, T.P., Lohmann, K.J., and Noakes, D.L.G. 2014. An inherited magnetic map guides ocean navigation in juvenile Pacific salmon. Current Biology 24: 446–450. Quinn, T.P. 2005. The behaviour and ecology of Pacific salmon and trout. University Press, Seattle, WA. 320p. Quinn, T.P., and Adams, D.J. 1996. Environmental changes affecting the migratory timing of American shad and sockeye salmon. Ecology 77: 1151–1162. Quinn, T.P., and Groot, C. 1984. Pacific salmon (Oncorhynchus) migrations: orientation versus random movement. Canadian Journal of Fisheries and Aquatic Science 41: 1319-1324. 101 Quinn, T.P., Hodgson, S., and Peven, C. 1997. Temperature, flow, and the migration of adult sockeye salmon (Oncorhynchus nerka) in the Columbia River. Canadian Journal of Fisheries and Aquatic Sciences 54: 1349-1360. Raby, G.D., Clark, T.D., Farrell, A.P., Patterson, D.A., Bett, N.N., Wilson, S.M., Willmore, W.G., Suski, C.D., Hinch, S.G., and Cooke, S.J. 2015. Facing the river gauntlet: understanding the effects of fisheries capture and water temperature on the physiology of coho salmon. PLoS ONE 10. Ramstad, K.M., and Woody, C.A. 2003. Radio tag retention and tag-related mortality among adult sockeye salmon. North American Journal of Fisheries Management 23: 978–982. Rand, P.S., Hinch, S.G., Morrison, J., Foreman, M.G.G., MacNutt, M.J., Macdonald, J.S., Healey, M.C., Farrell, A.P., and Higgs, D.A. 2006. Effects of river discharge, temperature, and future climates on energetics and mortality of adult migrating Fraser River sockeye salmon. Transactions of the American Fisheries Society 135: 655–667. Robards, M.D., and Quinn, T.P. 2011. The migratory timing of adult summer-run steelhead in the Columbia River over six decades of environmental change. Transactions of the American Fisheries Society 131: 523-536. Robinson, K.A., Hinch, S.G., Raby, G.D., Donaldson, M.R., Robichaud, D., Patterson, D.A., and Cooke, S.J. 2015. Influence of postcapture ventilation assistance on migration success of adult sockeye salmon following capture and release. Transactions of the American Fisheries Society 144: 693–704. Roscoe, D.W., and Hinch, S.G. 2008. Fishway passage, water diversion and warming temperatures: Factors limiting successful spawning migration of Seton-Anderson watershed sockeye salmon. Final Report for the Bridge Coastal Restoration Program, Project 07.BRG01. Roscoe, D.W. 2009. Hydro-system related mortality and in-lake behaviour of migrating adult sockeye salmon in the Seton-Anderson watershed, British Columbia. M.Sc. Thesis. The University of British Columbia, Vancouver, BC, Canada. Roscoe, D.W., and Hinch, S.G. 2010a. Effectiveness monitoring of fish passage facilities: historical trends, geographic patterns and future directions. Fish and Fisheries 11: 12–33. Roscoe, D.W., Hinch, S.G., Cooke, S.J., and Patterson, D.A. 2010b. Behaviour and thermal experience of adult sockeye salmon migrating through stratified lakes near spawning grounds: the roles of reproductive and energetic states. Ecology of Freshwater Fish 19: 51–62. 102 Roscoe, D.W., Hinch, S.G., Cooke, S.J., and Patterson, D.A. 2010c. Fishway passage and post-passage mortality of up-river migrating sockeye salmon in the Seton River, British Columbia. River Research and Applications 27: 693–705. Sandblom, E., Clark, T.D., Hinch, S.G., and Farrell, A.P. 2009. Sex-specific differences in cardiac control and hematology of sockeye salmon (Oncorhynchus nerka) approaching their spawning grounds. American Journal of Physiology - Regulatory, Integrative and Comparative Physiology 297: R1136–R1143. Sato, A., Ueda, H., Fukaya, M., Kaeriyama, M., Zohar, Y., Urano, A., and Yamauchi, K. 1997. Sexual differences in homing profiles and shortening of homing duration by gonadotropin-releasing hormone analog implantation in lacustrine sockeye salmon (Oncorhynchus nerka) in Lake Shikotsu. Zoological Science 14: 1009–1014. Schielzeth, H. 2010. Simple means to improve the interpretability of regression coefficients. Methods in Ecology and Evolution 1: 103–113. Servizi, J.A., and Jensen, J.O.T. 1977. Resistance of adult sockeye salmon [Oncorhynchus nerka] to acute thermal shock [Pacific Ocean]. International Pacific Salmon Fisheries Commission, New Westminster, British Columbia, Canada. Smyth, G.K. 2003. Pearson's goodness of fit statistic as a score test statistic. Lecture Notes-Monograph Series 40: 115–126. Institute of Mathematical Statistics. Standen, E.M., Hinch, S.G., Healey, M.C., and Farrell, A.P. 2002. Energetic costs of migration through the Fraser River Canyon, British Columbia, in adult pink (Oncorhynchus gorbuscha) and sockeye (Oncorhynchus nerka) salmon as assessed by EMG telemetry. Canadian Journal of Fisheries and Aquatic Sciences 59: 1809–1818. Tear, T.H., Kareiva, P., Angermeier, P.L., Comer, P., Czech, B., Kautz, R., Landon, L., Mehlman, D., Murphy, K., Ruckleshaus, M., Scott, J.M., and Wilhere, G. 2005. How much is enough? The recurrent problem of setting measurable objectives in conservation. BioScience 55: 835–849. Thorstad, E.B., Okland, F., Aarestrup, K., and Heggberget, T.G. 2008. Factors affecting the within-river spawning migration of Atlantic salmon, with emphasis on human impacts. Reviews in Fish Biology and Fisheries 18: 345–371. Thorstad, E.B., Okland, F., Johnsen, B.O., and Næsje, T.F. 2003a. Return migration of adult Atlantic salmon, Salmo salar, in relation to water diverted through a power station. Fisheries Management and Ecology 10: 13–22. Thorstad, E.B., Okland, F., Kroglund, F., and Jepsen, N. 2003b. Upstream migration of Atlantic salmon at a power station on the River Nidelva, Southern Norway. Fisheries Management and Ecology 10: 139–146. 103 Truscott, B., Idler, D.R., So, Y.P., and Walsh, J.M. 1986. Maturational steroids and gonadotropin in upstream migratory sockeye-salmon. General and Comparative Endocrinology 62: 99–110. Ueda, H. 2011. Physiological mechanism of homing migration in Pacific salmon from behavioral to molecular biological approaches. General and Comparative Endocrinology 170: 222–232. Vörösmarty, C.J., McIntyre, P.B., Gessner, M.O., and Dudgeon, D. 2010. Global threats to human water security and river biodiversity. Nature 467: 555-561. Vrieze, L.A., Bergstedt, R.A., and Sorensen, P.W. 2011. Olfactory-mediated stream-finding behavior of migratory adult sea lamprey (Petromyzon marinus). Canadian Journal of Fisheries and Aquatic Sciences 68: 523–533. Wagner, G.N., Hinch, S.G., Kuchel, L.J., Lotto, A., Jones, S.R., Patterson, D.A., Macdonald, J.S., Kraak, G.V.D., Shrimpton, M., English, K.K., Larsson, S., Cooke, S.J., Healey, M.C., and Farrell, A.P. 2005. Metabolic rates and swimming performance of adult Fraser River sockeye salmon (Oncorhynchus nerka) after a controlled infection with Parvicapsula minibicornis. Canadian Journal of Fisheries and Aquatic Sciences 62: 2124–2133. Waples, R.S., Zabel, R.W., Scheeuerell, M.D., and Sanderson, B.L. 2008. Evolutionary responses by native species to major anthropogenic changes to their ecosystems: Pacific salmon in the Columbia River hydropower system. Molecular Ecology 17: 84–96. Young, J.L., Hinch, S.G., Cooke, S.J., Crossin, G.T., Patterson, D.A., Farrell, A.P., Kraak, G.V.D., Lotto, A.G., Lister, A., Healey, M.C., and English, K.K. 2006. Physiological and energetic correlates of en route mortality for abnormally early migrating adult sockeye salmon (Oncorhynchus nerka) in the Thompson River, British Columbia. Canadian Journal of Fisheries and Aquatic Sciences 63: 1067–1077. Young, K.A. 2011. Managing the decline of Pacific salmon: metapopulation theory and artificial recolonization as ecological mitigation. Canadian Journal of Fisheries and Aquatic Sciences 56: 1700–1706. Young, P.S., Cech, J.J., Jr, and Thompson, L.C. 2011. Hydropower-related pulsed-flow impacts on stream fishes: a brief review, conceptual model, knowledge gaps, and research needs. Reviews in Fish Biology and Fisheries 21: 713–731. Zabel, R.W., Burke, B.J., Moser, M.L., and Caudill, C.C. 2014. Modeling temporal phenomena in variable environments with parametric models: An application to migrating salmon. Ecological Modeling 273: 23–30. Zuur, A.F., Ieno, E.N., and Elphick, C.S. 2010. A protocol for data exploration to avoid common statistical problems. Methods in Ecology and Evolution 1: 3–14. 104 Appendix 1 Table A1.1 – Location of radio receivers and combined percent time in operation during the 2013 and 2014 study years. Receiver location % time in operation Release site 92.4 Seton Powerhouse 98.3 Seton-Fraser confluence 86.5 Lower Seton River 94.8 Seton-Cayoosh confluence 91.2 Seton Dam 95.7 Dam fishway entrance PIT 92.7 Dam fishway exit PIT 96.2 Portage Creek 96.6 Gates Creek 99.5 105 Table A1.2 – Fate of all female (♀) and male (♂) Gates Creek and Portage Creek sockeye salmon used in models predicting (1) the number of forays into the Seton powerhouse, (2) wandering in the Fraser River between the release site and Seton-Fraser confluence, and (3) the total amount of migration delay incurred by individuals in the Seton powerhouse tailrace. Fates of individuals are given as a percentage of the total number of fish used in each model and by sample size. Entered Seton River Overshot Seton River Downstream Fraser River Unclassified ♀ (n) ♂ (n) ♀ (n) ♂ (n) ♀ (n) ♂ (n) ♀ (n) ♂ (n) Model 1 (Forays) Gates (255) 52% (132) 41% (105) 1.6% (4) 1.6% (4) 2% (5) 0.4% (1) 1.6% (4) 0% (0) Portage (90) 41% (37) 28% (25) 4% (4) 3% (3) 14% (13) 7% (7) - 1% (1) Model 2 (Wandering) Gates (236) 56% (132) 42.5% (100) - - 1% (2) 0.5% (1) - - Portage (78) 47% (37) 35% (27) - 2% (2) 8% (6) 8% (6) - - Model 3 (Powerhouse delay) Gates (256) 52% (133) 41% (105) 1.5% (4) 1.5% (4) 2% (5) 0.5% (1) 1.5% (4) - Portage (90) 41% (37) 28% (25) 3% (3) 3% (3) 16% (14) 8% (7) - 1% (1) 106 Table A1.3 – AICc model selection statistics for the 95% confidence set of models for generalized linear models predicting (1) the number of forays into the Seton Powerhouse, (2) the amount of wandering in the Fraser River between the release site and the Seton-Fraser confluence, and linear models predicting (3) the total amount of migration delay incurred by individuals in the Seton Powerhouse tailrace for Gates Creek and Portage Creek sockeye salmon. Abbreviations used for model predictor variables include: (Gluc) Glucose, (Lact) Lactate, (Test) Testosterone, (FRT) Fraser River Temperature, (PHT) Powerhouse Temperature, (Tdiff) Temperature differential between the Fraser River and Seton Powerhouse tailrace, (NW) Natal Water concentration, and (TD) Tagging Date. Generalized linear model fits were evaluated by Chi-square tests (see methods); Adj R2 is an estimate of the proportion of variance explained by each model, adjusted by the number of explanatory variables and is only shown for linear models of powerhouse delay. ΔAICc is the difference in AICc values between model i and the top model in the candidate set. Models are ranked from lowest to highest ΔAICc, and by wi – the probability that a given model is the best in the 95% confidence set. Gates Creek sockeye salmon powerhouse forays Intercept Sex Gluc Lact Test FRT PHT Tdiff NW DF Log Likelihood AICc ΔAIC w 0.67 - -0.17 - - - - - - 3 -421.42 848.94 0.00 0.03 0.67 - -0.17 - - - 0.11 - - 4 -420.67 849.50 0.56 0.02 0.67 - -0.15 - - - - -0.12 - 4 -420.68 849.52 0.57 0.02 0.67 - - - - - - - - 2 -422.82 849.69 0.74 0.02 0.67 - - - - - - -0.14 - 3 -421.80 849.70 0.75 0.02 0.67 - - - 0.14 - 0.14 - - 4 -420.84 849.84 0.89 0.02 0.67 - - - 0.12 - - - - 3 -421.90 849.90 0.96 0.02 0.67 - -0.15 - 0.09 - - - - 4 -420.87 849.91 0.97 0.02 0.67 - -0.14 - 0.12 - 0.14 - - 5 -419.84 849.91 0.97 0.02 107 Gates Creek sockeye salmon powerhouse forays Intercept Sex Gluc Lact Test FRT PHT Tdiff NW DF Log Likelihood AICc ΔAIC w 0.67 - -0.18 - - - - - -0.10 4 -420.89 849.94 1.00 0.02 0.67 - - - - - 0.11 - - 3 -422.12 850.33 1.39 0.01 0.67 - - - 0.10 - - -0.12 - 4 -421.18 850.52 1.57 0.01 0.67 - -0.16 - - - - -0.11 -0.09 5 -420.22 850.68 1.73 0.01 0.67 - -0.14 - 0.08 - - -0.10 - 5 -420.30 850.85 1.91 0.01 0.67 - -0.16 - - 0.09 - -0.17 - 5 -420.35 850.94 1.99 0.01 0.67 - -0.16 - - 0.09 - -0.17 - 5 -420.35 850.94 1.99 0.01 0.67 - -0.16 - - -0.09 0.16 - - 5 -420.35 850.94 1.99 0.01 0.67 - -0.16 - - - 0.08 -0.08 - 5 -420.35 850.94 1.99 0.01 0.67 - -0.17 -0.02 - - - - - 4 -421.39 850.94 2.00 0.01 0.67 + -0.17 - - - - - - 4 -421.42 851.00 2.06 0.01 0.67 - -0.17 - - 0.00 - - - 4 -421.42 851.01 2.06 0.01 0.67 - - - - - - - -0.08 3 -422.48 851.06 2.11 0.01 0.67 - -0.18 - - - 0.09 - -0.06 5 -420.45 851.14 2.20 0.01 0.67 - - - - - - -0.13 -0.07 4 -421.50 851.15 2.21 0.01 0.67 - -0.16 - 0.08 - - - -0.08 5 -420.47 851.18 2.24 0.01 0.67 - - - - - 0.07 -0.11 - 4 -421.56 851.29 2.35 0.01 0.67 - - - - 0.08 - -0.18 - 4 -421.56 851.29 2.35 0.01 0.67 - - - - 0.08 - -0.18 - 4 -421.56 851.29 2.35 0.01 0.67 - - - - -0.12 0.17 - - 4 -421.56 851.29 2.35 0.01 108 Gates Creek sockeye salmon powerhouse forays Intercept Sex Gluc Lact Test FRT PHT Tdiff NW DF Log Likelihood AICc ΔAIC w 0.67 - - - 0.11 - - - -0.06 4 -421.67 851.49 2.55 0.01 0.67 - - - 0.13 -0.07 0.17 - - 5 -420.63 851.51 2.56 0.01 0.67 - - - 0.13 0.12 - -0.18 - 5 -420.63 851.51 2.56 0.01 0.67 - - - 0.13 0.12 - -0.18 - 5 -420.63 851.51 2.56 0.01 0.67 - - - 0.13 - 0.11 -0.07 - 5 -420.63 851.51 2.56 0.01 0.67 - -0.17 -0.03 - - 0.11 - - 5 -420.63 851.51 2.56 0.01 0.68 + -0.17 - - - 0.12 - - 5 -420.65 851.54 2.59 0.01 0.69 + - - - - - - - 3 -422.74 851.57 2.63 0.01 0.67 + -0.15 - - - - -0.12 - 5 -420.67 851.58 2.64 0.01 0.69 + - - - - - -0.14 - 4 -421.72 851.59 2.65 0.01 0.67 - -0.15 0.00 - - - -0.12 - 5 -420.68 851.60 2.65 0.01 0.67 - -0.15 -0.06 0.14 - 0.14 - - 6 -419.63 851.60 2.65 0.01 0.67 - - - - -0.02 - - - 3 -422.79 851.67 2.72 0.01 0.67 - - -0.05 0.16 - 0.14 - - 5 -420.71 851.67 2.72 0.01 0.67 - -0.15 -0.05 0.11 - - - - 5 -420.72 851.69 2.74 0.01 0.67 - - 0.03 - - - -0.14 - 4 -421.77 851.69 2.75 0.01 0.67 - - 0.00 - - - - - 3 -422.82 851.73 2.79 0.01 0.67 - -0.14 - 0.11 - 0.11 -0.05 - 6 -419.71 851.76 2.82 0.01 0.67 - -0.14 - 0.11 0.13 - -0.17 - 6 -419.71 851.76 2.82 0.01 0.67 - -0.14 - 0.11 0.13 - -0.17 - 6 -419.71 851.76 2.82 0.01 109 Gates Creek sockeye salmon powerhouse forays Intercept Sex Gluc Lact Test FRT PHT Tdiff NW DF Log Likelihood AICc ΔAIC w 0.67 - -0.14 - 0.11 -0.06 0.16 - - 6 -419.71 851.76 2.82 0.01 0.67 - - -0.04 0.13 - - - - 4 -421.82 851.80 2.86 0.01 0.67 - -0.15 - 0.11 - 0.12 - -0.04 6 -419.75 851.84 2.90 0.01 0.65 + -0.15 - 0.11 - - - - 5 -420.81 851.86 2.92 0.01 0.67 - - - 0.14 - 0.13 - -0.02 5 -420.82 851.88 2.94 0.01 0.67 - -0.15 - 0.10 0.03 - - - 5 -420.83 851.90 2.95 0.01 0.67 + - - 0.15 - 0.14 - - 5 -420.84 851.91 2.97 0.01 0.67 - -0.18 -0.03 - - - - -0.10 5 -420.84 851.92 2.97 0.01 0.65 + -0.15 - 0.13 - 0.14 - - 6 -419.79 851.93 2.98 0.01 0.67 - - - 0.13 0.02 - - - 4 -421.89 851.93 2.99 0.01 0.66 + - - 0.13 - - - - 4 -421.89 851.94 3.00 0.01 0.67 - -0.18 - - -0.03 - - -0.10 5 -420.86 851.95 3.01 0.01 0.68 + -0.17 - - - - - -0.10 5 -420.88 851.99 3.05 0.01 0.69 + - - - - 0.12 - - 4 -421.97 852.09 3.15 0.01 0.67 - - - 0.09 - - -0.12 -0.06 5 -420.94 852.13 3.19 0.01 0.67 - - - - - 0.10 - -0.05 4 -422.01 852.18 3.24 0.01 0.67 - -0.15 - 0.07 - - -0.10 -0.08 6 -419.93 852.20 3.25 0.01 110 Portage Creek sockeye salmon powerhouse forays Intercept Sex Gluc Lact Test TD PHT Tdiff NW DF Log Likelihood AICc ΔAIC w 1.21 - - - - -0.42 -1.16 -0.64 -0.93 6 -204.36 421.73 0.00 0.06 1.23 - - - - - -1.08 - -0.80 4 -206.68 421.84 0.11 0.06 1.22 - - - - - -1.04 -0.28 -0.82 5 -205.56 421.84 0.11 0.06 1.22 - - - 0.27 - -1.05 - -0.82 5 -205.65 422.02 0.29 0.06 1.31 + - - - - -1.10 - -0.85 5 -206.00 422.72 0.99 0.04 1.22 - -0.24 - - - -1.05 - -0.87 5 -206.02 422.75 1.02 0.04 1.20 - - - 0.20 -0.42 -1.15 -0.58 -0.94 7 -203.78 422.93 1.20 0.04 1.21 - - - 0.21 - -1.03 -0.23 -0.83 6 -204.99 422.99 1.26 0.03 1.21 - -0.19 - - - -1.02 -0.25 -0.87 6 -205.14 423.29 1.56 0.03 1.28 + - - - - -1.07 -0.25 -0.85 6 -205.20 423.40 1.67 0.03 1.20 - -0.16 - - -0.40 -1.14 -0.60 -0.96 7 -204.08 423.52 1.79 0.03 1.22 - -0.18 - 0.23 - -1.03 - -0.87 6 -205.27 423.55 1.82 0.03 1.25 + - - - -0.40 -1.17 -0.59 -0.95 7 -204.17 423.70 1.97 0.02 1.22 - - 0.10 - - -1.04 -0.29 -0.84 6 -205.43 423.87 2.14 0.02 1.23 - - 0.07 - - -1.08 - -0.82 5 -206.61 423.93 2.20 0.02 1.20 - - 0.07 - -0.41 -1.16 -0.64 -0.94 7 -204.29 423.94 2.21 0.02 1.23 - - - - 0.03 -1.07 - -0.80 5 -206.67 424.05 2.32 0.02 1.29 + -0.20 - - - -1.07 - -0.90 6 -205.55 424.12 2.39 0.02 1.25 + - - 0.22 - -1.07 - -0.84 6 -205.60 424.22 2.49 0.02 1.22 - - 0.02 0.26 - -1.05 - -0.83 6 -205.64 424.30 2.57 0.02 111 Portage Creek sockeye salmon powerhouse forays Intercept Sex Gluc Lact Test TD PHT Tdiff NW DF Log Likelihood AICc ΔAIC w 1.22 - - - 0.27 -0.02 -1.06 - -0.83 6 -205.64 424.30 2.57 0.02 1.22 - -0.27 0.12 - - -1.04 - -0.91 6 -205.80 424.62 2.89 0.02 1.21 - -0.16 - 0.18 - -1.01 -0.21 -0.87 7 -204.71 424.78 3.05 0.01 1.31 + - - - 0.02 -1.10 - -0.85 6 -206.00 425.01 3.28 0.01 1.20 - -0.12 - 0.19 -0.40 -1.13 -0.56 -0.96 8 -203.62 425.01 3.28 0.01 1.31 + - 0.00 - - -1.10 - -0.85 6 -206.00 425.02 3.29 0.01 1.22 - -0.24 - - 0.02 -1.04 - -0.86 6 -206.01 425.03 3.30 0.01 1.21 - -0.22 0.14 - - -1.02 -0.26 -0.91 7 -204.88 425.13 3.40 0.01 1.27 + -0.17 - - - -1.04 -0.23 -0.89 7 -204.88 425.13 3.40 0.01 1.21 - - 0.05 0.19 - -1.03 -0.23 -0.84 7 -204.95 425.26 3.53 0.01 1.24 + - - 0.17 - -1.04 -0.22 -0.84 7 -204.96 425.28 3.55 0.01 1.20 - - 0.03 0.20 -0.42 -1.15 -0.58 -0.94 8 -203.77 425.32 3.59 0.01 1.20 + - - 0.22 -0.42 -1.15 -0.59 -0.94 8 -203.78 425.34 3.61 0.01 1.20 - -0.18 0.10 - -0.38 -1.13 -0.59 -0.99 8 -203.92 425.62 3.89 0.01 1.24 + -0.14 - - -0.38 -1.15 -0.56 -0.97 8 -203.95 425.67 3.94 0.01 1.27 + - 0.04 - - -1.06 -0.25 -0.86 7 -205.17 425.71 3.98 0.01 1.22 - -0.20 0.07 0.22 - -1.03 - -0.89 7 -205.20 425.77 4.03 0.01 1.24 + -0.18 - 0.20 - -1.04 - -0.88 7 -205.24 425.84 4.11 0.01 1.22 - -0.18 - 0.24 -0.02 -1.04 - -0.88 7 -205.26 425.88 4.15 0.01 1.25 + - 0.03 - -0.40 -1.17 -0.60 -0.95 8 -204.15 426.09 4.36 0.01 112 Portage Creek sockeye salmon powerhouse forays Intercept Sex Gluc Lact Test TD PHT Tdiff NW DF Log Likelihood AICc ΔAIC w 1.23 - - 0.08 - 0.04 -1.06 - -0.81 6 -206.58 426.17 4.44 0.01 1.28 + -0.22 0.06 - - -1.07 - -0.91 7 -205.51 426.40 4.67 0.01 1.29 + -0.20 - - 0.01 -1.07 - -0.90 7 -205.55 426.47 4.74 0.01 1.25 + - - 0.22 -0.02 -1.07 - -0.84 7 -205.60 426.56 4.83 0.01 1.25 + - 0.01 0.22 - -1.07 - -0.84 7 -205.60 426.57 4.84 0.01 1.22 - - 0.02 0.27 -0.02 -1.06 - -0.83 7 -205.64 426.64 4.91 0.01 113 Gates Creek sockeye salmon wandering Intercept Sex Gluc Lact Test FRT PHT Tdiff NW DF Log Likelihood AICc ΔAIC w -1.15 + - -0.90 - - 0.77 - - 5 -142.09 294.44 0.00 0.05 -1.13 + - -0.86 -0.22 - 0.73 - - 6 -141.95 296.27 1.83 0.02 -1.41 - - -0.80 - - 0.71 - - 4 -144.06 296.29 1.84 0.02 -1.17 + - -0.86 - -0.18 0.87 - - 6 -142.01 296.40 1.95 0.02 -1.17 + - -0.86 - - 0.69 -0.16 - 6 -142.01 296.40 1.95 0.02 -1.17 + - -0.86 - 0.71 - -0.82 - 6 -142.01 296.40 1.95 0.02 -1.17 + - -0.86 - 0.71 - -0.82 - 6 -142.01 296.40 1.95 0.02 -1.15 + - -0.90 - - 0.75 - -0.06 6 -142.07 296.52 2.07 0.02 -1.15 + 0.04 -0.90 - - 0.77 - - 6 -142.08 296.53 2.09 0.02 -1.12 + - -0.75 - - - - - 4 -144.25 296.68 2.24 0.02 -1.19 + - -0.67 - - - -0.49 - 5 -143.43 297.12 2.67 0.01 -1.40 - - -0.59 - - - -0.60 - 4 -144.62 297.41 2.96 0.01 -1.42 - - -0.72 - 0.57 - -0.87 - 5 -143.70 297.67 3.22 0.01 -1.42 - - -0.72 - -0.38 0.93 - - 5 -143.70 297.67 3.22 0.01 -1.42 - - -0.72 - - 0.56 -0.35 - 5 -143.70 297.67 3.22 0.01 -1.42 - - -0.72 - 0.57 - -0.87 - 5 -143.70 297.67 3.22 0.01 -1.37 - - - - - - -0.67 - 3 -145.80 297.70 3.26 0.01 -1.15 + - -0.57 -0.52 - - -0.57 - 6 -142.73 297.83 3.39 0.01 -1.13 + - - - - 0.61 - - 4 -144.83 297.84 3.40 0.01 -1.37 - - -0.66 - - - - - 3 -145.88 297.86 3.42 0.01 114 Gates Creek sockeye salmon wandering Intercept Sex Gluc Lact Test FRT PHT Tdiff NW DF Log Likelihood AICc ΔAIC w -1.13 + - - -0.67 - - -0.68 - 5 -143.81 297.89 3.44 0.01 -1.09 + - -0.68 -0.39 - - - - 5 -143.84 297.95 3.51 0.01 -1.10 + - -0.86 - 0.33 - - - 5 -143.87 297.99 3.55 0.01 -1.15 + - -0.77 -0.31 -0.29 0.87 - - 7 -141.78 298.05 3.61 0.01 -1.15 + - -0.77 -0.31 0.61 - -0.82 - 7 -141.78 298.05 3.61 0.01 -1.15 + - -0.77 -0.31 - 0.60 -0.26 - 7 -141.78 298.05 3.61 0.01 -1.15 + - -0.77 -0.31 0.61 - -0.82 - 7 -141.78 298.05 3.61 0.01 -1.41 - -0.17 -0.81 - - 0.70 - - 5 -143.94 298.14 3.69 0.01 -1.18 + - - - - - -0.59 - 4 -145.00 298.18 3.73 0.01 -1.10 + - -0.75 - - - - -0.23 5 -144.02 298.30 3.86 0.01 -1.41 - - -0.80 - - 0.74 - 0.09 5 -144.02 298.31 3.87 0.01 -1.41 - - -0.81 0.06 - 0.72 - - 5 -144.05 298.36 3.91 0.01 -1.12 + - -0.85 -0.22 - 0.71 - -0.07 7 -141.93 298.36 3.92 0.01 -1.12 + 0.03 -0.86 -0.22 - 0.73 - - 7 -141.95 298.39 3.95 0.01 -1.35 - - - - - 0.58 - - 3 -146.17 298.45 4.00 0.01 -1.16 + - -0.86 - -0.19 0.85 - -0.08 7 -141.99 298.47 4.02 0.01 -1.16 + - -0.86 - - 0.67 -0.18 -0.08 7 -141.99 298.47 4.02 0.01 -1.16 + - -0.86 - 0.68 - -0.80 -0.08 7 -141.99 298.47 4.02 0.01 -1.16 + - -0.86 - 0.68 - -0.80 -0.08 7 -141.99 298.47 4.02 0.01 -1.16 + 0.07 -0.86 - -0.20 0.89 - - 7 -141.99 298.48 4.04 0.01 115 Gates Creek sockeye salmon wandering Intercept Sex Gluc Lact Test FRT PHT Tdiff NW DF Log Likelihood AICc ΔAIC w -1.16 + 0.07 -0.86 - - 0.69 -0.18 - 7 -141.99 298.48 4.04 0.01 -1.16 + 0.07 -0.86 - 0.71 - -0.83 - 7 -141.99 298.48 4.04 0.01 -1.16 + 0.07 -0.86 - 0.71 - -0.83 - 7 -141.99 298.48 4.04 0.01 -1.11 + - - - - - - - 3 -146.19 298.49 4.05 0.01 -1.14 + 0.03 -0.90 - - 0.76 - -0.06 7 -142.07 298.63 4.19 0.01 -1.38 - - - -0.42 - - -0.75 - 4 -145.28 298.74 4.29 0.01 -1.16 + - -0.68 - - - -0.48 -0.23 6 -143.19 298.75 4.31 0.01 -1.13 + -0.03 -0.75 - - - - - 5 -144.25 298.76 4.32 0.01 -1.32 - - - - - - - - 2 -147.38 298.82 4.37 0.01 -1.08 + - - -0.44 - 0.55 - - 5 -144.29 298.84 4.40 0.01 -1.37 - - - - -0.56 0.92 - - 4 -145.38 298.92 4.48 0.01 -1.37 - - - - 0.38 - -0.86 - 4 -145.38 298.92 4.48 0.01 -1.37 - - - - - 0.37 -0.51 - 4 -145.38 298.92 4.48 0.01 -1.37 - - - - 0.38 - -0.86 - 4 -145.38 298.92 4.48 0.01 -1.06 + - - -0.54 - - - - 4 -145.39 298.96 4.52 0.01 -1.17 + - - - 0.46 - -0.81 - 5 -144.38 299.02 4.58 0.01 -1.17 + - - - -0.43 0.87 - - 5 -144.38 299.02 4.58 0.01 -1.17 + - - - - 0.45 -0.39 - 5 -144.38 299.02 4.58 0.01 -1.17 + - - - 0.46 - -0.81 - 5 -144.38 299.02 4.58 0.01 -1.41 - - -0.52 -0.27 - - -0.65 - 5 -144.41 299.09 4.65 0.01 116 Portage Creek sockeye salmon wandering Intercept Sex Gluc Lact Test FRT PHT Tdiff NW DF Log Likelihood AICc ΔAIC w -0.44 + - -1.09 - -1.69 - - -2.03 6 -73.44 160.07 0.00 0.03 -0.73 - - -0.79 - -1.72 - - -1.80 5 -74.72 160.28 0.21 0.03 -0.63 - - - - -1.84 - - -2.24 4 -75.96 160.47 0.40 0.03 -0.77 - - -0.96 0.62 -1.68 - - -1.94 6 -73.94 161.07 1.00 0.02 -0.69 - - -0.98 - -0.81 - - - 4 -76.34 161.22 1.15 0.02 -0.61 - - -0.89 - - - - - 3 -77.52 161.36 1.29 0.02 -0.41 + - -1.09 - -2.75 - 1.02 -2.45 7 -72.91 161.42 1.35 0.02 -0.41 + - -1.09 - -2.75 - 1.02 -2.45 7 -72.91 161.42 1.35 0.02 -0.41 + - -1.09 - -1.34 -0.81 - -2.45 7 -72.91 161.42 1.35 0.02 -0.41 + - -1.09 - - -1.59 -0.97 -2.45 7 -72.91 161.42 1.35 0.02 -0.72 - - -0.79 - -2.73 - 0.92 -2.22 6 -74.31 161.81 1.74 0.01 117 Portage Creek sockeye salmon wandering Intercept Sex Gluc Lact Test FRT PHT Tdiff NW DF Log Likelihood AICc ΔAIC w -0.72 - - -0.79 - -2.73 - 0.92 -2.22 6 -74.31 161.81 1.74 0.01 -0.72 - - -0.79 - -1.45 -0.74 - -2.22 6 -74.31 161.81 1.74 0.01 -0.72 - - -0.79 - - -1.57 -1.05 -2.22 6 -74.31 161.81 1.74 0.01 -0.32 + - -1.19 - - -1.51 - -1.82 6 -74.32 161.83 1.77 0.01 -0.34 + - -1.18 - - - - - 4 -76.65 161.85 1.78 0.01 -0.43 + - - - -1.84 - - -2.46 5 -75.52 161.86 1.80 0.01 -0.63 - - - - -2.89 - 0.93 -2.74 5 -75.59 162.02 1.95 0.01 -0.63 - - - - -2.89 - 0.93 -2.74 5 -75.59 162.02 1.95 0.01 -0.63 - - - - -1.60 -0.74 - -2.74 5 -75.59 162.02 1.95 0.01 -0.63 - - - - - -1.66 -1.15 -2.74 5 -75.59 162.02 1.95 0.01 -0.66 - - -1.02 - - -0.71 - - 4 -76.75 162.04 1.98 0.01 118 Portage Creek sockeye salmon wandering Intercept Sex Gluc Lact Test FRT PHT Tdiff NW DF Log Likelihood AICc ΔAIC w -0.67 - - -0.90 - - - -0.64 - 4 -76.78 162.12 2.05 0.01 -0.46 + - -1.22 - -0.73 - - - 5 -75.65 162.14 2.07 0.01 -0.64 - - - 0.38 -1.82 - - -2.35 5 -75.67 162.17 2.11 0.01 -0.73 - -0.33 -0.79 - -1.70 - - -1.93 6 -74.54 162.26 2.19 0.01 -0.45 + -0.20 -1.08 - -1.68 - - -2.10 7 -73.37 162.34 2.28 0.01 -0.63 - -0.36 - - -1.83 - - -2.41 5 -75.77 162.38 2.31 0.01 -0.48 + - -1.09 0.14 -1.69 - - -2.03 7 -73.42 162.44 2.37 0.01 -0.49 - - - - - - - - 2 -79.15 162.45 2.39 0.01 -0.65 - - -1.07 0.56 - - - - 4 -76.95 162.45 2.39 0.01 -0.67 - - -0.87 - - -1.37 - -1.39 5 -75.82 162.48 2.41 0.01 -0.77 - - -0.97 0.66 -2.73 - 1.00 -2.36 7 -73.44 162.48 2.42 0.01 119 Portage Creek sockeye salmon wandering Intercept Sex Gluc Lact Test FRT PHT Tdiff NW DF Log Likelihood AICc ΔAIC w -0.77 - - -0.97 0.66 -2.73 - 1.00 -2.36 7 -73.44 162.48 2.42 0.01 -0.77 - - -0.97 0.66 -1.35 -0.80 - -2.36 7 -73.44 162.48 2.42 0.01 -0.77 - - -0.97 0.66 - -1.58 -0.97 -2.36 7 -73.44 162.48 2.42 0.01 -0.72 - - -1.12 0.47 -0.75 - - - 5 -75.92 162.67 2.61 0.01 -0.40 + - -1.29 - - -0.68 - - 5 -75.93 162.70 2.63 0.01 -0.73 - - -1.06 0.73 - -1.47 - -1.69 6 -74.84 162.87 2.80 0.01 -0.42 + - -1.16 - - - -0.57 - 5 -76.06 162.96 2.89 0.01 -0.55 - - - - -0.68 - - - 3 -78.36 163.05 2.98 0.01 -0.78 - -0.30 -0.97 0.60 -1.67 - - -2.06 7 -73.79 163.17 3.11 0.01 -0.55 - - - - - -1.39 - -1.85 4 -77.32 163.19 3.12 0.01 -0.62 - -0.34 -0.93 - - - - - 4 -77.34 163.23 3.16 0.01 -0.54 - - - - - - -0.65 - 3 -78.46 163.24 3.18 0.01 -0.69 - - -1.18 0.53 - -0.69 - - 5 -76.23 163.28 3.22 0.01 -0.42 + - - - -2.92 - 0.99 -2.97 6 -75.09 163.36 3.30 0.01 120 Portage Creek sockeye salmon wandering Intercept Sex Gluc Lact Test FRT PHT Tdiff NW DF Log Likelihood AICc ΔAIC w -0.42 + - - - - -1.69 -1.12 -2.97 6 -75.09 163.36 3.30 0.01 -0.69 - -0.13 -1.00 - -0.78 - - - 5 -76.31 163.45 3.39 0.01 -0.69 - - -1.00 - -0.71 -0.16 - - 5 -76.32 163.47 3.40 0.01 -0.69 - - -1.00 - - -0.56 -0.51 - 5 -76.32 163.47 3.40 0.01 -0.69 - - -1.00 - -0.98 - 0.19 - 5 -76.32 163.47 3.40 0.01 -0.69 - - -1.00 - -0.98 - 0.19 - 5 -76.32 163.47 3.40 0.01 -0.69 - - -1.06 0.49 - - -0.58 - 5 -76.34 163.51 3.44 0.01 -0.60 - - -0.86 - - - - -0.17 4 -77.49 163.53 3.47 0.01 -0.64 - - - 0.41 -2.90 - 0.99 -2.86 6 -75.25 163.68 3.61 0.01 -0.64 - - - 0.41 -2.90 - 0.99 -2.86 6 -75.25 163.68 3.61 0.01 -0.64 - - - 0.41 -1.54 -0.79 - -2.86 6 -75.25 163.68 3.61 0.01 -0.64 - - - 0.41 - -1.67 -1.11 -2.86 6 -75.25 163.68 3.61 0.01 121 Gates Creek sockeye salmon powerhouse delay Intercept Sex Gluc Lact Test FRT PHT Tdiff NW Adj R2 DF Log Likelihood AICc ΔAIC w 4.71 - -0.52 - - - - 0.34 0.33 0.05 5 -481.06 972.35 0.00 0.03 4.71 - -0.46 - 0.30 - - 0.38 0.37 0.06 6 -480.02 972.38 0.03 0.03 4.71 - -0.57 - - - - 0.36 - 0.04 4 -482.40 972.97 0.61 0.02 4.71 - -0.47 - - - - - 0.35 0.04 4 -482.49 973.13 0.78 0.02 4.71 - -0.51 - - - -0.35 - - 0.04 4 -482.49 973.14 0.79 0.02 4.71 - -0.47 - - - -0.27 - 0.27 0.04 5 -481.68 973.60 1.24 0.02 4.71 - -0.52 - 0.25 - - 0.40 - 0.04 5 -481.69 973.62 1.27 0.02 4.71 - -0.42 - 0.25 - - - 0.39 0.04 5 -481.77 973.79 1.44 0.02 4.71 - -0.54 - - -0.27 - 0.52 - 0.04 5 -481.79 973.82 1.46 0.02 4.71 - -0.54 - - 0.28 -0.49 - - 0.04 5 -481.79 973.82 1.46 0.02 4.71 - -0.54 - - - -0.24 0.26 - 0.04 5 -481.79 973.82 1.46 0.02 4.71 - -0.54 - - -0.27 - 0.52 - 0.04 5 -481.79 973.82 1.46 0.02 4.71 - -0.51 - - -0.16 - 0.44 0.29 0.05 6 -480.86 974.05 1.70 0.01 4.71 - -0.51 - - 0.30 -0.41 - 0.29 0.05 6 -480.86 974.05 1.70 0.01 4.71 - -0.51 - - - -0.15 0.28 0.29 0.05 6 -480.86 974.05 1.70 0.01 4.71 - -0.51 - - -0.16 - 0.44 0.29 0.05 6 -480.86 974.05 1.70 0.01 4.71 - -0.49 0.13 - - - 0.31 0.34 0.05 6 -480.87 974.07 1.72 0.01 4.71 - -0.52 - - - - - - 0.03 3 -484.01 974.12 1.77 0.01 4.75 + -0.50 - - - - 0.33 0.32 0.05 6 -480.95 974.24 1.88 0.01 4.71 - -0.44 0.20 - - - - 0.37 0.04 5 -482.02 974.28 1.92 0.01 4.71 - -0.45 0.05 0.28 - - 0.37 0.37 0.06 7 -480.00 974.45 2.10 0.01 4.71 - -0.46 - 0.28 -0.05 - 0.40 0.36 0.06 7 -480.01 974.47 2.12 0.01 122 Gates Creek sockeye salmon powerhouse delay Intercept Sex Gluc Lact Test FRT PHT Tdiff NW Adj R2 DF Log Likelihood AICc ΔAIC w 4.71 - -0.46 - 0.28 0.38 -0.38 - 0.36 0.06 7 -480.01 974.47 2.12 0.01 4.71 - -0.46 - 0.28 - -0.04 0.36 0.36 0.06 7 -480.01 974.47 2.12 0.01 4.71 - -0.46 - 0.28 -0.05 - 0.40 0.36 0.06 7 -480.01 974.47 2.12 0.01 4.69 + -0.46 - 0.31 - - 0.38 0.37 0.06 7 -480.01 974.47 2.12 0.01 4.71 - -0.48 0.17 - - -0.35 - - 0.04 5 -482.13 974.50 2.15 0.01 4.76 + -0.54 - - - - 0.35 - 0.04 5 -482.22 974.68 2.33 0.01 4.71 - -0.44 0.19 - - -0.26 - 0.29 0.05 6 -481.22 974.78 2.43 0.01 4.71 - -0.55 0.09 - - - 0.34 - 0.04 5 -482.30 974.84 2.49 0.01 4.71 - -0.48 - 0.13 - -0.33 - - 0.04 5 -482.30 974.84 2.49 0.01 4.71 - -0.43 - 0.19 - -0.22 - 0.31 0.05 6 -481.26 974.86 2.51 0.01 4.71 - -0.43 - 0.32 0.22 - - 0.44 0.05 6 -481.28 974.89 2.54 0.01 4.76 + -0.45 - - - - - 0.34 0.04 5 -482.33 974.90 2.55 0.01 4.71 - -0.49 - - 0.11 - - 0.38 0.04 5 -482.35 974.93 2.58 0.01 4.75 + -0.49 - - - -0.34 - - 0.04 5 -482.37 974.98 2.62 0.01 4.71 - -0.51 - 0.20 0.33 -0.48 - - 0.05 6 -481.35 975.05 2.69 0.01 4.71 - -0.51 - 0.20 -0.21 - 0.51 - 0.05 6 -481.35 975.05 2.69 0.01 4.71 - -0.51 - 0.20 - -0.19 0.31 - 0.05 6 -481.35 975.05 2.69 0.01 4.71 - -0.51 - 0.20 -0.21 - 0.51 - 0.05 6 -481.35 975.05 2.69 0.01 4.71 - - - 0.39 - - 0.33 0.44 0.04 5 -482.50 975.24 2.89 0.01 4.71 - -0.48 - 0.19 - - - - 0.03 4 -483.58 975.33 2.98 0.01 4.71 - -0.41 0.15 0.21 - - - 0.39 0.05 6 -481.53 975.40 3.05 0.01 4.71 - -0.50 0.17 - - - - - 0.03 4 -483.67 975.50 3.15 0.01 123 Gates Creek sockeye salmon powerhouse delay Intercept Sex Gluc Lact Test FRT PHT Tdiff NW Adj R2 DF Log Likelihood AICc ΔAIC w 4.75 + -0.46 - - - -0.26 - 0.26 0.04 6 -481.59 975.51 3.16 0.01 4.71 - -0.52 0.12 - -0.29 - 0.50 - 0.04 6 -481.62 975.57 3.22 0.01 4.71 - -0.52 0.12 - 0.24 -0.47 - - 0.04 6 -481.62 975.57 3.22 0.01 4.71 - -0.52 0.12 - - -0.26 0.23 - 0.04 6 -481.62 975.57 3.22 0.01 4.71 - -0.52 0.12 - -0.29 - 0.50 - 0.04 6 -481.62 975.57 3.22 0.01 4.77 + -0.49 - - - - - - 0.03 4 -483.76 975.67 3.32 0.01 4.75 + -0.52 - - -0.26 - 0.51 - 0.04 6 -481.67 975.68 3.33 0.01 4.75 + -0.52 - - 0.28 -0.47 - - 0.04 6 -481.67 975.68 3.33 0.01 4.75 + -0.52 - - - -0.23 0.26 - 0.04 6 -481.67 975.68 3.33 0.01 4.75 + -0.52 - - -0.26 - 0.51 - 0.04 6 -481.67 975.68 3.33 0.01 4.72 + -0.52 - 0.23 - - 0.39 - 0.04 6 -481.68 975.70 3.34 0.01 4.71 - -0.52 0.02 0.24 - - 0.39 - 0.04 6 -481.68 975.70 3.35 0.01 4.71 - -0.48 0.14 - -0.18 - 0.42 0.30 0.05 7 -480.63 975.71 3.36 0.01 4.71 - -0.48 0.14 - 0.26 -0.39 - 0.30 0.05 7 -480.63 975.71 3.36 0.01 4.71 - -0.48 0.14 - - -0.16 0.25 0.30 0.05 7 -480.63 975.71 3.36 0.01 4.71 - -0.48 0.14 - -0.18 - 0.42 0.30 0.05 7 -480.63 975.71 3.36 0.01 4.71 - - - 0.34 - - - 0.45 0.03 4 -483.85 975.86 3.51 0.01 4.72 + -0.42 - 0.24 - - - 0.38 0.04 6 -481.77 975.88 3.53 0.01 4.75 + -0.50 - - 0.30 -0.40 - 0.28 0.05 7 -480.77 976.00 3.65 0.01 4.75 + -0.50 - - -0.15 - 0.43 0.28 0.05 7 -480.77 976.00 3.65 0.01 4.75 + -0.50 - - - -0.14 0.28 0.28 0.05 7 -480.77 976.00 3.65 0.01 124 Portage Creek sockeye salmon powerhouse delay Intercept Sex Gluc Lact Test TD PHT Tdiff NW Adj R2 DF Log Likelihood AICc ΔAIC w 2.24 - - - - - -1.41 - -2.64 0.23 4 -163.15 334.76 0.00 0.07 2.24 - 0.46 - - - -1.44 - -2.76 0.24 5 -162.08 334.88 0.12 0.06 2.24 - - - - -1.12 -1.39 -0.94 -2.70 0.26 6 -161.20 335.41 0.65 0.05 2.24 - - - - -0.35 -1.42 - -2.59 0.24 5 -162.56 335.83 1.07 0.04 2.24 - 0.45 - - -0.34 -1.45 - -2.71 0.25 6 -161.53 336.07 1.30 0.04 2.24 - 0.38 - - -1.02 -1.42 -0.83 -2.79 0.27 7 -160.46 336.29 1.53 0.03 2.24 - 0.51 - -0.24 - -1.38 - -2.65 0.25 6 -161.83 336.67 1.90 0.03 2.25 - 0.51 -0.22 - - -1.43 - -2.83 0.25 6 -161.87 336.75 1.98 0.03 2.30 + - - - - -1.42 - -2.66 0.23 5 -163.04 336.80 2.04 0.02 2.24 - - - -0.14 - -1.38 - -2.57 0.23 5 -163.06 336.83 2.07 0.02 2.24 - - -0.10 - - -1.40 - -2.66 0.23 5 -163.10 336.92 2.15 0.02 2.24 - - - - - -1.41 -0.01 -2.64 0.23 5 -163.15 337.01 2.24 0.02 2.24 - 0.46 - - - -1.44 0.02 -2.75 0.24 6 -162.08 337.18 2.41 0.02 2.24 + 0.47 - - - -1.44 - -2.75 0.24 6 -162.08 337.18 2.41 0.02 2.23 - - - -0.20 -1.15 -1.34 -1.00 -2.61 0.26 7 -161.02 337.40 2.63 0.02 125 Portage Creek sockeye salmon powerhouse delay Intercept Sex Gluc Lact Test TD PHT Tdiff NW Adj R2 DF Log Likelihood AICc ΔAIC w 2.24 - - -0.11 - -0.35 -1.41 - -2.62 0.24 6 -162.51 338.03 3.26 0.01 2.24 - - - -0.10 -0.34 -1.40 - -2.55 0.24 6 -162.51 338.04 3.28 0.01 2.24 - 0.49 - -0.20 -0.31 -1.40 - -2.63 0.26 7 -161.35 338.07 3.30 0.01 2.22 - - - - - - - -1.43 0.18 3 -165.90 338.08 3.32 0.01 2.39 + - - -0.39 - -1.33 - -2.51 0.24 6 -162.63 338.27 3.51 0.01 2.24 - 0.57 -0.27 -0.29 - -1.35 - -2.72 0.25 7 -161.51 338.38 3.62 0.01 2.25 + 0.45 - - -0.34 -1.45 - -2.71 0.25 7 -161.53 338.42 3.66 0.01 2.22 - 0.43 - - - - - -1.51 0.19 4 -165.02 338.50 3.74 0.01 2.19 + 0.42 - - -1.05 -1.42 -0.88 -2.79 0.27 8 -160.40 338.59 3.82 0.01 2.24 - 0.41 -0.10 - -0.98 -1.41 -0.79 -2.82 0.27 8 -160.42 338.62 3.85 0.01 2.21 - - - - -1.14 - -0.98 -1.52 0.21 5 -163.99 338.70 3.94 0.01 2.32 + 0.46 - -0.37 - -1.35 - -2.61 0.25 7 -161.70 338.78 4.01 0.01 2.24 - - -0.12 -0.15 - -1.36 - -2.59 0.23 6 -162.99 339.00 4.24 0.01 2.24 - 0.51 - -0.24 - -1.38 -0.02 -2.65 0.25 7 -161.82 339.01 4.25 0.01 126 Portage Creek sockeye salmon powerhouse delay Intercept Sex Gluc Lact Test TD PHT Tdiff NW Adj R2 DF Log Likelihood AICc ΔAIC w 2.24 - - - -0.15 - -1.37 -0.04 -2.57 0.23 6 -163.05 339.12 4.35 0.01 2.24 - - -0.10 - - -1.40 0.01 -2.66 0.23 6 -163.10 339.21 4.45 0.01 2.22 - - - - -0.34 - - -1.38 0.19 4 -165.39 339.24 4.48 0.01 2.32 + - - -0.34 -1.09 -1.31 -0.93 -2.57 0.27 8 -160.86 339.50 4.73 0.01 2.24 + 0.47 - - - -1.44 0.02 -2.75 0.24 7 -162.08 339.53 4.76 0.01 2.39 + - - -0.34 -0.33 -1.35 - -2.49 0.24 7 -162.09 339.55 4.79 0.01 2.22 - 0.50 - -0.35 - - - -1.44 0.20 5 -164.48 339.68 4.92 0.01 2.22 - - - -0.25 - - - -1.36 0.18 4 -165.62 339.72 4.95 0.01 2.22 - 0.43 - - -0.32 - - -1.46 0.20 5 -164.53 339.78 5.01 0.01 2.23 - - 0.00 -0.20 -1.15 -1.34 -1.00 -2.61 0.26 8 -161.02 339.81 5.04 0.01 2.21 - 0.35 - - -1.05 - -0.88 -1.57 0.22 6 -163.41 339.82 5.06 0.01 2.24 - 0.56 -0.27 -0.25 -0.31 -1.37 - -2.69 0.26 8 -161.04 339.86 5.09 0.01 127 CC USR LSR PhTr80100120140160Conductivity (S/m)abcbFigure A1.1 – Water conductivity measured at sites in Cayoosh Creek (CC; n = 127), the upper Seton River below the Seton Dam (USR; n = 127), the Lower Seton River below its confluence with Cayoosh Creek (LSR; n = 125), and in the Seton powerhouse tailrace (PhTr; n = 127) used to distinguish between the olfactory signatures of each water source. Data from the 2013 and 2014 study years were pooled. Solid lines inside boxes represent median values; upper and lower whiskers represent sample minimum and maximums. Lowercase letters indicate significant differences in conductivity between sites determined by a Kruskal-Wallis test at α < 0.05. 128 Appendix 2 For each reach-specific TTE analysis, by definition, fish that were ultimately unsuccessful (i.e. those that did not survive to natal spawning grounds) were censored more frequently than those that were successful (i.e. did survive to natal spawning grounds) (see manuscript figures 3 – 6). This violates a primary assumption of TTE analysis and prompted me to examine the potential for this ‘informative censoring’ to bias results and confound my interpretation of model outcomes. Following the methods described in (Caudill et al. 2007; Allison 2010), I conducted a series of sensitivity analyses for each of the hydrosystem passage time models that differed in the way they handled censored individuals to assess the effects of informative censoring on model conclusions. First, I created a ‘lower’ odds model that changed all censoring times to event times and thus assumed all fish achieved the ‘event’ of interest. In the second ‘upper’ odds model, I assigned censoring times equal to the longest observed event time, thereby assuming that all censored fish would have remained able to achieve the event until the end of observation. In the third ‘event only’ model, I directly tested for later effects of fate on migration and dam passage times by excluding all censored individuals from each analysis a priori. AICC model selection and multi-model averaged results from these sensitivity analyses indicated that the informative censoring of ultimately unsuccessful fish had little effect on model outcomes. In every case where odds ratios for the fate or fate × time covariates in the original models (those with unaltered censoring observations) had 95% CIs that did not include zero, at least one or more of the same covariates in each of the sensitivity models had 95% CIs that also did not include zero (Table A2.1). In addition, estimates for odds ratios of all covariates in the alternative models did not change markedly in their sizes or directions relative to the original models (Table A2.1). Moreover, in every case where fate had strong predictive power in the original model, this was also true in the ‘event only’ model, lending further support to the observation that slower migration and dam passage times were at least in part related to the ultimate fate of individuals (Table A2.1). 129 Table A2.1 - Model-averaged results for sensitivity analyses testing ‘original’, ‘lower odds’, ‘upper odds’, and ‘event-only’ models to assess the effects of informative censoring on time-to-event regression analysis. Model averaged odds ratios with 95% confidence intervals that do not include zero are shown in bold. Model 1: Release to Seton River Entry Covariates Gates Creek SK Model Sex Sex x Time GSE FRT SRT NW Fate Fate x Time Original Odds ratio 1.40 -0.04 -0.22 -0.19 0.20 0.03 0.81 -0.01 95% CI 0.80, 1.99 -0.07, -0.01 -0.54, 0.11 -0.53, 0.15 -0.14, 0.53 -0.16, 0.21 0.25, 1.35 -0.03, 0.01 Lower Odds ratio 1.15 -0.03 -0.25 -0.03 0.17 0.02 0.22 -0.00 95% CI 0.64, 1.65 -0.05, -0.01 -0.55, 0.03 -0.21, 0.15 -0.14, 0.48 -0.14, 0.20 -0.24, 0.69 -0.02, 0.01 Upper Odds ratio 1.52 -0.05 -0.11 -0.46 0.09 0.06 0.49 0.01 95% CI 0.95, 2.08 -0.07, -0.02 -0.42, 0.18 -0.74, -0.18 -0.17, 0.36 -0.23, 0.36 -0.01, 1.00 -0.01, 0.02 Event-only Odds ratio 1.23 -0.03 -0.31 -0.06 0.25 0.04 0.79 -0.02 95% CI 0.66, 1.79 -0.06, -0.01 -0.60, -0.02 -0.31, 0.18 -0.09, 0.58 -0.17, 0.25 0.21, 1.28 -0.04, -0.00 130 Model 1: Release to Seton River Entry Covariates Portage Creek SK Model Sex Sex x Time GSE FRT SRT NW Fate Fate x Time Original Odds ratio 0.86 -0.00 -0.33 -0.47 -0.22 0.05 0.66 0.00 95% CI 0.28, 1.43 -0.01, 0.01 -0.62, -0.04 -1.06, 0.12 -0.70, 0.24 -0.20, 0.31 0.09, 1.24 -0.00, 0.01 Lower Odds ratio 0.48 -0.00 -0.34 -0.26 -0.29 0.06 0.32 0.00 95% CI -0.03, 0.99 -0.00, 0.00 -0.60, -0.08 -0.79, 0.27 -0.74, 0.16 -0.18, 0.31 -0.18, 0.84 -0.00, 0.00 Upper Odds ratio 1.52 -0.04 -0.12 -0.46 0.09 0.06 0.50 0.01 95% CI 0.95, 2.08 -0.07, -0.02 -0.42, 0.18 -0.74, -0.18 -0.17, 0.36 -0.23, 0.36 -0.01, 1.00 -0.00, 0.02 Event-only Odds ratio 1.23 -0.03 -0.31 -0.07 0.25 0.04 0.75 0.75 95% CI 0.66, 1.80 -0.05, -0.01 -0.60, -0.02 -0.32, 0.18 -0.09, 0.58 -0.17, 0.26 0.21, 1.28 0.21, 1.28 131 Model 2: Seton River entry to Seton Dam Covariates Model Sex Sex x Time GSE SRT NW Fate Fate x Time Gates Creek SK Original Odds ratio 0.70 -0.06 -0.48 -0.03 -0.43 1.21 -0.05 95% CI 0.63, 1.78 -0.10, -0.02 -0.70, -0.24 -0.21, 0.14 -0.70, -0.14 0.63, 1.78 -0.08, -0.01 Lower Odds ratio 0.70 -0.06 -0.45 -0.02 -0.39 1.12 -0.04 95% CI 0.15, 1.25 -0.10, -0.02 -0.67, -0.21 -0.17, 0.13 -0.68, -0.09 0.57, 1.67 -0.08, -0.01 Upper Odds ratio 0.26 -0.03 -0.50 -0.02 -0.43 0.88 -0.02 95% CI -0.22, 0.75 -0.05, 0.01 -0.73, -0.28 -0.19, 0.13 -0.68, -0.19 0.33, 1.43 -0.05, 0.02 Event-only Odds ratio 0.56 -0.05 -0.54 -0.02 -0.42 1.32 -0.05 95% CI -0.01, 1.13 -0.09, -0.01 -0.76, -0.32 -0.18, 0.13 -0.71, -0.13 0.74, 1.90 -0.09, -0.02 132 Model 2: Seton River entry to Seton Dam Covariates Portage Creek SK Model Sex Sex x Time GSE SRT NW Fate Fate x Time Original Odds ratio -0.04 -0.01 -0.19 0.02 -0.01 0.94 -0.01 95% CI -0.57, 0.49 -0.04, 0.02 -0.55, 0.17 -0.17, 0.22 -0.18, 0.15 0.31, 1.56 -0.05, 0.02 Lower Odds ratio -0.13 -0.01 -0.25 0.05 -0.02 0.87 -0.01 95% CI -0.65, 0.38 -0.03, 0.02 -0.62, 0.12 -0.20, 0.31 -0.19, 0.16 0.26, 1.48 -0.04, 0.02 Upper Odds ratio -0.50 0.01 -0.38 -0.00 -0.03 0.62 0.00 95% CI -1.15, 0.15 -0.01, 0.03 -0.83, 0.06 -0.24, 0.24 -0.29, 0.22 -0.13, 1.37 -0.04,0.04 Event-only Odds ratio -0.13 -0.01 -0.31 0.09 -0.02 0.81 -0.01 95% CI -0.66, 0.41 -0.03, 0.02 -0.69, 0.07 -0.24, 0.43 -0.21, 0.16 0.18, 1.44 -0.04, 0.03 133 Model 3: Seton Dam Passage Covariates Model Sex Sex x Time GSE Dtemp Ddis Fate Fate x Time Flow Gates Creek SK Original Odds ratio 0.01 -0.02 -0.04 0.45 0.05 0.62 0.01 0.02 95% CI -0.34, 0.34 -0.04, 0.01 -0.24, 0.17 0.16, 0.74 -0.16, 0.27 0.25, 0.99 -0.01, 0.04 -0.23, 0.27 Lower Odds ratio 0.04 -0.01 -0.06 0.39 0.02 0.51 -0.01 -0.00 95% CI -0.28, 0.37 -0.28, 0.36 -0.27, 0.17 0.11, 0.67 -0.14, 0.19 0.17, 0.86 -0.03, 0.02 -0.23, 0.23 Upper Odds ratio 0.03 -0.03 -0.04 0.49 0.07 0.52 0.04 0.03 95% CI -0.31, 0.37 -0.05, -0.00 -0.29, 0.20 0.19, 0.79 -0.15, 0.30 0.16, 0.89 0.02, 0.79 -0.25, 0.31 Event-only Odds ratio 0.16 -0.02 -0.05 0.48 0.02 0.46 -0.02 -0.02 95% CI -0.20, 0.52 -0.05, 0.01 -0.27, 0.16 0.18, 0.79 -0.15, 0.18 0.06, 0.86 -0.06, 0.01 -0.25, 0.22 134 Model 3: Seton Dam Passage Covariates Model Sex Sex x Time GSE Dtemp Ddis Fate Fate x Time Flow Portage Creek SK Original Odds ratio -0.13 -0.01 -0.03 0.21 0.23 0.50 0.02 - 95% CI -0.60, 0.32 -0.02, 0.01 -0.22, 0.15 -0.31, 0.73 -0.18, 0.65 -0.02, 1.00 -0.00, 0.04 - Lower Odds ratio -0.15 -0.01 -0.18 0.40 0.10 0.23 0.01 - 95% CI -0.46, 0.15 -0.01, 0.00 -0.44, 0.07 0.09, 0.72 -0.15, 0.36 -0.10, 0.55 0.00, 0.02 - Upper Odds ratio -0.20 -0.00 -0.02 0.28 0.35 0.50 0.03 - 95% CI -0.67, 0.25 -0.02, 0.01 -0.19, 0.14 -0.22, 0.78 -0.05, 0.75 -0.01, 1.01 0.01, 0.05 - Event-only Odds ratio -0.20 -0.00 -0.02 0.28 0.35 0.50 0.03 - 95% CI -0.68, 0.26 -0.02, 0.01 -0.19, 0.14 -0.22, 0.78 -0.05, 0.75 -0.02, 1.01 0.01, 0.05 - """@en ; edm:hasType "Thesis/Dissertation"@en ; vivo:dateIssued "2016-09"@en ; edm:isShownAt "10.14288/1.0305714"@en ; dcterms:language "eng"@en ; ns0:degreeDiscipline "Forestry"@en ; edm:provider "Vancouver : University of British Columbia Library"@en ; dcterms:publisher "University of British Columbia"@en ; dcterms:rights "Attribution-NonCommercial-NoDerivatives 4.0 International"@* ; ns0:rightsURI "http://creativecommons.org/licenses/by-nc-nd/4.0/"@* ; ns0:scholarLevel "Graduate"@en ; dcterms:title "The cumulative effects of physiology, temperature, and natal water cues on the migration behaviour and survival of adult sockeye salmon during passage through the Seton River hydroelectric system, British Columbia"@en ; dcterms:type "Text"@en ; ns0:identifierURI "http://hdl.handle.net/2429/58376"@en .