HEAT TRANSFER, THERMAL PREFERENCE, AND BEHAVIOURAL THERMOREGULATION OF ADULT FEMALE BABINE LAKE SOCKEYE SALMON by Avery Dextrase B.Sc. (Hons), Dalhousie University, 2020 THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN NATURAL RESOURCES AND ENVIRONMENTAL SCIENCE - BIOLOGY UNIVERSITY OF NORTHERN BRITISH COLUMBIA May 2024 © Avery Dextrase, 2024 ii Abstract During their upriver migration to spawning ground, many adult Sockeye Salmon (Oncorhynchus nerka) pass through lakes that provide wide ranges of water temperatures and opportunities to behaviourally thermoregulate. There is still uncertainty regarding if and how effectively Sockeye Salmon behaviourally thermoregulate in lakes, what factors might drive the thermoregulation, and how thermoregulatory behaviour impacts fitness. The purpose of this thesis was to study the thermal ecology of adult Sockeye Salmon by estimating their thermal preference and how effectively they behaviourally thermoregulated while migrating through Babine Lake where up to approximately 90% of Sockeye Salmon returning to the Skeena River spawn. Additionally, I investigated what parameters impact the thermal preferences and thermoregulatory behaviour of the salmon, and if more effective behavioural thermoregulators had greater spawning success. External attached temperature logger tags monitored the thermal experience of Sockeye Salmon in Babine Lake, and shuttle box tests were conducted to estimate the thermal preference ranges of the salmon as well as to estimate the coefficient of heat exchange (݇) from their body surface to body core. The ݇ coefficient was used in conjunction with the external temperature logger data to estimate body temperatures of the salmon in Babine Lake, and the body temperatures, thermal preference estimates, and lake water temperature data were incorporated to estimate how effectively the tagged Sockeye Salmon behaviourally thermoregulated while passing through Babine Lake. Chapter One of this thesis focuses on estimating ݇ for adult Sockeye Salmon and applying these estimates to predict the body temperatures of the externally tagged Sockeye Salmon in Babine Lake. Chapter Two focuses more so on behaviour, including estimates of the thermal preference and effectiveness of behavioural thermoregulation of Babine Lake Sockeye Salmon. The median estimated ݇ of the iii salmon was 0.08 min-1 and was slightly lower for fish with higher condition factor (Fulton’s), when the body was cooling, and when water temperatures were closer to the body temperature. Metabolic heat production (ܶ݉) appeared to have a small but notable impact on body temperature. The median model-averaged fitted thermal preference range (interquartile) of the shuttle box fish was 11.84°C to 12.32°C, and the median predicted thermal preference range of the radio tagged fish was 13.39°C to 13.89°C. The salmon appeared to behaviourally thermoregulate quite effectively in Babine Lake, although there was a lot of individual variation. There was high uncertainty for relationships between the thermal preference of the salmon or the effectiveness of their thermoregulatory behaviour in Babine Lake and their relative infectious burden, mass, and gross somatic energy density, or the time of day (day/night). Salmon with higher heat stress scores tended to prefer warmer temperatures, presumably due to an association between acclimation temperatures and temperature selection. Most of the salmon had high spawning success, with more effective thermoregulators tending to have had greater spawning success. Babine Lake provides important thermal habitat to adult Sockeye Salmon, and this habitat and effective behavioural thermoregulation could become imperative as North American rivers and lakes continue to warm with climate change. Table of Contents Abstract...............................................................................................................................ii Table of Contents...............................................................................................................iv List of Tables.....................................................................................................................vii List of Figures...................................................................................................................viii Acknowledgements...........................................................................................................xii Dedication........................................................................................................................xiii CHAPTER ONE: Heat Exchange Between Sockeye Salmon and the Ambient Environment.........................................................................................................................1 Introduction..............................................................................................................1 Methods..................................................................................................................10 Fish Capture and Data Collection.................................................................10 Data Analysis................................................................................................12 Results....................................................................................................................18 Model Selection and Averaging....................................................................18 Model Averaged Estimates...........................................................................19 Visualizing Body Temperature Rate of Change...........................................22 Sockeye Salmon Body and Ambient Water Temperatures in Babine Lake.23 Discussion..............................................................................................................28 Condition factor, Body Size, and ݇...............................................................29 Heating, Cooling, and ݇................................................................................30 Absolute Temperature Difference and ݇......................................................32 Effect of Tm on Body Temperature..............................................................33 Predicted Body and Observed Ambient Temperatures in Babine Lake.......34 Additional Study Assumptions and Limitations...........................................36 Applications and Potential for Future Research...........................................39 References..............................................................................................................40 v CHAPTER TWO: Thermal Preference and Behavioural Thermoregulation of Sockeye Salmon...............................................................................................................................48 Introduction............................................................................................................48 Methods..................................................................................................................57 Study Area....................................................................................................57 Fish Capture, Sampling, and Tagging...........................................................61 Thermal Preference Tests.............................................................................63 Monitoring Water Temperatures in Babine Lake.........................................66 Radio Telemetry and Fish Recovery.............................................................66 Gill Clip Processing......................................................................................69 Thermal Preference Analysis........................................................................71 Effectiveness of Behavioural Thermoregulation Estimations......................75 Relationship Between ‫ ܧ‬and Covariates.......................................................81 E and Spawning Success...............................................................................83 Results....................................................................................................................84 Measured Body Parameters..........................................................................84 Thermal Preference Results..........................................................................85 Days and Depths Experienced in Babine Lake.............................................91 Temperatures in Babine Lake and Quality of Thermal Habitat (݀݁)............92 Accuracy of Behavioural Thermoregulation (ܾ݀)........................................94 Effectiveness of Behavioural Thermoregulation (‫)ܧ‬....................................95 Effectiveness of Behavioural Thermoregulation Models.............................96 Spawning Success and Effectiveness of Behavioural Thermoregulation.....99 Discussion............................................................................................................101 Thermal Preference Results........................................................................101 Thermoregulation in Babine Lake..............................................................109 ‫ ܧ‬and Measured Body Parameters..............................................................112 ‫ ܧ‬and Spawning Success............................................................................115 vi Conclusion...........................................................................................................116 Additional Questions..................................................................................116 Research Implications.................................................................................117 References............................................................................................................119 Appendix..........................................................................................................................133 vii List of Tables Table 1.1. AIC scores and related statistics for candidate models for the heat coefficient, ݇. P is the number of parameters in the model, AIC is the AIC score, ΔAIC is the difference between the model AIC score and the top model AIC score, lik is the model likelihood, wAIC is the model AIC weight, and cwAIC is the cumulative AIC weight. In the fixed effects column, ܽ‫ ݀ݐ‬is the absolute temperature difference variable, ℎ݁ܽ‫ ݐ‬is the heating/cooling variable, and ܾܿ is condition factor.................................................................................................................................19 Table 2.1. The estimated area, water volume, and proportion of the total Babine Lake water volume (up to 55m deep) found in each five-meter depth interval..........................79 Table 2.2. The candidate models that had stacking weights greater than 0 (15 models), the difference in expected log predictive densities between these models, the model with the highest expected log predictive density (elpd Difference, model 5), and the standard error of these differences (Difference se)...................................................................................87 viii List of Figures Figure 1.1. A female Sockeye Salmon externally and internally tagged with temperature data loggers before beginning a thermal preference test in the shuttle box...................11 Figure 1.2. Model averaged estimates (with 95% confidence intervals) of the intercept and fixed effects for the heat coefficient ݇ (min-1)...........................................................20 Figure 1.3. Model-averaged estimates of the heat coefficient, ݇, for a fish with the minimum (0.78), average (0.96), and maximum (1.15) condition factors measured for the shuttle box test fish. Figure A represents ݇ estimated with an absolute temperature difference of 0°C (the minimum observed absolute temperature difference in shuttle box tests), B represents ݇ estimated with an absolute temperature difference of 0.37°C (the average observed temperature difference in the shuttle box tests), and C represents ݇ estimated with an absolute temperature difference of 2.47°C (the maximum observed temperature difference in the shuttle box tests).................................................................21 Figure 1.4. Recorded vs. model-averaged predicted body temperatures for the fish during their shuttle box tests. The orange line represents equal predicted and recorded temperatures.......................................................................................................................22 Figure 1.5. Model-averaged predictions of body temperature over time of a Sockeye Salmon with a condition factor of 0.96. Plot A shows how the body temperature of a fish would warm after beginning at 8°C, 12°C, and 16°C and being placed in 20°C water. Plot B shows how the body temperature of a fish would cool after beginning at 20°C, 16°C, and 12°C and being placed in 8°C water...........................................................................23 Figure 1.6. The relationship between the square-root transformed mean rate of water temperature change and square-root transformed mean absolute difference between the body temperatures of radio-tagged Sockeye Salmon in Babine Lake (model-averaged predictions) and the water temperatures they encountered. As mentioned in the methods section, these values were calculated for each recording (taken every 1.5 minutes) then averaged for the corresponding individual and hour. The orange line represents the linear relationship modelled between the variables when assuming zero residual error and a random deviation from the intercept of zero. The equation of the line has |ܶܽ − ܾܶ| representing the square rooted average absolute difference in body and water temperatures of an individual salmon, and ‫ݔ‬௪௧௖ representing the average rate of change in water temperature experienced by an individual during an hour in Babine Lake. The R2 value represents both the marginal and conditional R2......................................................25 Figure 1.7. Twenty-four hour snapshot of water temperatures (recorded every 1.5 minutes) experienced by radio-tagged Sockeye Salmon ID 11 and 63 and their corresponding predicted body temperatures that occurred while passing through Babine Lake on their way to spawning grounds............................................................................27 ix Figure 2.1. Map of the Skeena River, Babine River, Nilkitkwa Lake, and Babine Lake. The small blue portion of river between Nilkitkwa Lake and Babine Lake is known as Rainbow Alley. The river and lake shapefiles were retrieved from the BC Freshwater Atlas webpage in March 2023. The base maps are from ESRI (primary map) and OpenStreetMap (inset)......................................................................................................58 Figure 2.2. Map of the study area, including Babine River, Nilkitkwa Lake, Babine Lake, Morrison Creek, Fulton River, and Pinkut Creek. The small light blue portion of river between Nilkitkwa Lake and Babine Lake is known as Rainbow Alley. The tagging location shown on the map is also the location of the Babine River counting fence and where the thermal preference tests were conducted (See the “thermal preference tests” section below for details). The locations of the receiver station towers and the approximate locations of the temperature logger strings in 2021 are also shown (See the “Monitoring Water Temperatures in Babine Lake” and “Radio Telemetry and Fish Recover” sections below for more details)........................................................................60 Figure 2.3. Babine Lake at the mouth of Pinkut Creek in September 2021. The red shapes in the water are adult Sockeye Salmon..............................................................................61 Figure 2.4. A radio-tagged Sockeye Salmon in the Babine River after tagging on September 1st, 2021............................................................................................................62 Figure 2.5. Simplified schematic of the shuttle box. The camera tracks the location of this fish, informing the temperature control system if the fish is in the cool (blue) or warm (red) side of the shuttle box. The temperature control system warms both sides of the shuttle box if the fish is in the warm side (panel A) and cools both sides if the fish is in the cool side (panel B).......................................................................................................65 Figure 2.6. Volunteer Spencer Smith using the mobile radio tracking unit (black box, wire, and antenna visible) to locate a radio-tagged fish in Fulton River in September 2021..................................................................................................................................68 Figure 2.7. Boxplots of the measured masses (g), gross somatic energies (GSE, MJ • kg), relative infectious burdens (RIB), and levels of heat stress of the shuttle box thermal preference tested fish and the recovered radio-tagged fish. The data from the salmon whose shuttle box tests were not included in the thermal preference models (test 12 and 13) are not included. The boxplots include the variable median (horizontal bar), interquartile range (Q25 – Q75, lower and upper bounds of the coloured boxes), whiskers (lowest and highest values outside of the interquartile range and within 1.5 x the interquartile range, vertical lines), and outliers (values extending past the interquartile range by more than ± 1.5 x the interquartile range, circles)..............................................85 1 x Figure 2.8. The mean effect size of the fixed parameters included in 15 thermal preference models with non-zero stacking weights. The effect sizes on both the lower (Q25) and upper (Q75) bounds of the thermal preference ranges are included. Horizontal orange and blue lines represent the 95% credible intervals of the fixed effect estimates. Vertical gray lines correspond to an effect size of zero.....................................................89 Figure 2.9. Violin plots of the predicted thermal preference ranges (interquartile range – Q25, Q75) of the 23 recovered radio-tagged female Sockeye Salmon during the civil daytime and nighttime. The boxplots include the parameter median (horizontal bar), interquartile range (Q25 – Q75, lower and upper bounds of the coloured boxes), whiskers (lowest and highest values outside of the interquartile range and within 1.5 x the interquartile range, vertical lines), and outliers (values extending past the interquartile range by more than ± 1.5 x the interquartile range, circles). The solid colours represent the densities of thermal preference estimates....................................................................90 Figure 2.10. The date and time each recovered radio-tagged fish was released after tagging and the estimated time they entered and exited Babine Lake in 2021. The bars represent the duration each fish was assumed to be in Babine Lake.................................92 Figure 2.11. The predicted hourly temperatures from 0 to 60 meters in Babine Lake from August 27th, 2021, to September 27th, 2021 (the time period recovered radio-tagged sockeye salmon were believed to be in Babine Lake). The temperatures were predicted with a GAMM model and eight stations of Fisheries and Oceans Canada temperature logger data..........................................................................................................................94 Figure 2.12. Violin plots of the estimated hourly thermal habitat quality (݀݁, °C), accuracy of thermoregulation (ܾ݀, °C), and effectiveness of behavioural thermoregulation (‫ )ܧ‬of the recovered radio-tagged female Sockeye Salmon while they were predicted to be in Babine Lake in August and September 2021. The boxplots include the median (horizontal bar), interquartile range (Q25 – Q75, lower and upper bounds of the coloured boxes), whiskers (lowest and highest values outside of the interquartile range and within 1.5 x the interquartile range, vertical lines), and outliers (values extending past the interquartile range by more than ± 1.5 x the interquartile range, circles). The solid colours represent the densities of estimates....................................................................................96 Figure 2.13 Model-averaged estimates and associated 95% confidence interval for the covariates present in the models included in the 95% confidence set for the best model of behavioural thermoregulation effectiveness (‫)ܧ‬. The expected ‫ ܧ‬for the daytime is represented by the intercept, which had a model average estimate of 1.40 (95% CI: 0.65 – 2.15) ..................................................................................................................................98 xi Figure 2.14 Model-averaged predictions of the effectiveness of behavioural thermoregulation for a Sockeye Salmon with the following covariate values, based on the ranges of covariates measured in the radio-tagged and thermal preference test fish: A. Sockeye Salmon with the minimum (1000g) and maximum measured masses (2300g); B. Sockeye Salmon with the minimum (4.60MJ•kg-1) and maximum measured GSE (6.69MJ•kg-1); C. Sockeye Salmon with the minimum (0.06) and maximum (0.95) heat stress value; and D. Sockeye Salmon with the minimum (0.0) and maximum estimated RIB (1.93). Predictions were made for the two values of each variable for day and night, while keeping all other variables at their average value in the data set.............................99 Figure 2.15. The approximate percentage of eggs retained after spawning and the median hourly effectiveness of behavioural thermoregulation scores of 15 of the recovered radiotagged Sockeye Salmon. Eight recovered fish are excluded from this plot due to only their tag being recovered (with no body) or because their body cavity was not intact upon recovery. A darker orange circle indicates overlapping points........................................100 xii Acknowledgements I could not have made it here without the enormous amount of support I received from so many people! I would like to start by thanking Donna MacIntyre and the Lake Babine Nation fisheries department for collaborating with us on the project and providing their expertise and support. I am very grateful for the opportunity to have worked with you and thankful to the Lake Babine Nation communities whose waters and land this work was completed on. I would also like to thank our collaborators from Fisheries and Oceans Canada. Thank you to Dr. Daniel Selbie, Dr. Svein Vagle, and Lucas Pon for helping us understand the limnology of Babine Lake and providing us with Babine Lake temperature data. Thank you to Dr. Kristina Miller-Saunders, Tobi Ming, and the Pacific Biological Station team for processing the salmon gill clips and helping us analyse and understand the results. Thank you to Mitchell Harborne and the teams at the Fulton River and Pinkut Creek spawning channels for your support, and thank you David Patterson and Kendra Robinson for assisting with the study design and logistics. I am so thankful for everyone that helped me in the field. Thank you Brandon West, Spencer Smith, Celeste Kieran, Paul Szirmay-Kalos, Emily Mason, Courtney Griffis, Jacob Newman, Justin Chow, Joe Bottoms, Sebastian Dalgarno, Dan Scurfield, Kristen Peck, Tyler Winther, Ryan Whitmore, Devon Smith, and Carly Walters. An additional huge thank you to my amazing parents, not only for their support, but also for coming all the way from Ontario to help me for a week of field work! I also want to acknowledge my amazing partner and the members of the Freshwater Fish Ecology Lab (many of whom helped with field work) for their encouragement and support. I am very grateful to NSERC and CFI for providing funding, and UNBC, the Ministry of Advanced Education, Skills and Training, Patrick Lloyd, and the Pacific Salmon Foundation for the bursaries and scholarships that helped me through my degree. Additionally, thank you to Quantum Industrial Solutions and Raven RSM for providing equipment to the project, and Coopdogg’s Fishing Lodge for their hospitality. Last, but certainly not least, I would like to thank my wonderful supervisor and committee members Dr. Eduardo Martins, David Patterson, and Dr. Mark Shrimpton. Thank you so much for all of the support, guidance, and learning opportunities over the last three years. xiii Dedication This thesis is dedicated to my dad, Al, who is not here to read the final copy but supported me every step of the way. Dad, you will always be an inspiration and truly "salmon" special. Love you. CHAPTER ONE Heat Exchange Between Sockeye Salmon and the Ambient Environment Introduction When it comes to the regulation of body temperature, animals are often grouped into two broad categories: endotherms and ectotherms (Reynolds, 1979). In general, endotherms use internal processes to regulate and stabilize their body temperature in the face of variable external temperatures (i.e., physiological thermoregulation), and their body heat is generated primarily by their metabolic output (Reynolds, 1979). Contrarily, the body temperature of ectotherms is primarily dictated by the surrounding environment as most ectothermic animals do not have internal mechanisms of body temperature regulation (Reynolds, 1979). Heat transfer between an animal and its surrounding environment occurs via conduction, convection, evapotranspiration, and radiation (Gaur et al., 2019). In terrestrial environments, sources of heat transfer between a body and the environment include infrared and solar radiation, conduction across the body surface, sweating, water evaporation while breathing, and excretion of feces and urine (although the last may be considered insignificant) (Brill et al., 1994; Reynolds, 1979). For aquatic, gill-breathing ectotherms (i.e., organisms, such as most fishes, that regulate body temperature by gaining or loosing heat from/to the environment), the avenues of heat transfer with their surrounding environment are less variable, with nearly all heat exchange occurring through conduction across the body and fins and through convection via the 2 cardiovascular system, primarily through perfusion of the gills with blood (Brill et al., 1994; Reynolds, 1979; Stevens & Sutterlin, 1976). A laboratory study on Sea Ravens (Hemitripterus americanus) found that the majority (70 to 90%) of metabolic body heat was lost through fins and the body wall, and only about 10 to 30% occurred through the gills (Stevens & Sutterlin, 1976). Similarly, Masser and Neill (1986) found that 20 to 40% of heat exchange between the bodies of Channel Catfish (Ictalurus punctatus) and Bluegills (Lepomis macrochirus) and ambient water occurred through their gills. The body temperature of most fishes approaches that of the water over time (when water temperature remains relatively constant), and remains equal to or within a few degrees of the water temperature after stabilization (Carey et al., 1971; Dean, 1976; Pépino et al., 2015). The temperature difference after stabilization may be due to a slight retention of internally, or metabolically, produced heat (Dean, 1976). For example, Stevens and Sutterlin (1976) found that approximately 10-20% of heat in the blood was retained after passing through the gills of Sea Ravens. Activity rates may also contribute to the difference between body and water temperature after body temperature stabilizes, as lower levels of activity reduce respiration rates resulting in less heat exchange occurring via the gills (Carey et al., 19771; Dean, 1976). Indeed, Dean (1976) found that when Rainbow Trout (Oncorhynchus mykiss) were actively swimming, deep white muscle tissue temperatures were equivalent to the water temperature but were 0.4°C – 1.1°C greater than the water temperature when resting (though both the resting and active temperatures of more surficial red muscle tissue and the gut remained very similar to the ambient water temperature). Exceptions to the trend of equilibrating body temperatures 3 are seen in members of the Thunnini group (tuna), Opah (Lampris guttatis) and Lamnidae family (mackerel sharks) whose body temperatures regularly remain above ambient water temperatures (Carey et al., 1971). Fishes in these groups possess effective countercurrent heat exchange systems containing “retia mirabilia”, specialized bundles of arteries and veins that allow for increased retention of blood heat (Graham & Dickson, 2001; Stevens et al., 1974). Another exception is the Scalloped Hammerhead Shark (Sphryna lewini) that closes its gill slits during deep dives to cool water to temporarily supress respiratory heat exchange and retain body heat (Royer et al., 2023). As most fishes are ectotherms, water temperature significantly impacts their physiology, behaviour, and fitness (Huey & Berrigan, 2001; Huey & Stevenson, 1979), and has long been regarded as the master environmental factor for fish (Brett 1971). Unsurprisingly, body temperatures experienced by fishes in the wild are commonly measured and analyzed in ecological studies (e.g., Donaldson et al., 2009; Hight & Lowe, 2007; Nakamura et al., 2020; Raby et al., 2018). Accurately measuring the body temperatures of fishes over time usually requires a temperature logger (or temperaturesensitive tag) to be gastrically or surgically implanted into the body cavity, which can be costly, time-consuming, and may negatively impact the fish (Dick et al. 2020; Jepsen et al., 2015; Thorstad et al., 2013). Some researchers have avoided using internal temperature loggers and assumed that fish body temperature is equivalent to the water temperatures they occupy. For example, Armstrong et al. (2016) assumed that water temperatures recorded by iButton temperature loggers (Dallas Semiconductor, USA) secured externally to adult Sockeye Salmon (Oncorhynchus nerka) were interchangeable 4 with their body temperatures. This assumption may be inaccurate as it takes time for a fish's body temperature to equilibrate with the water temperature, especially after a rapid change in water temperature, as seen when a fish dives from the surface of a lake to below the thermocline (Pépino et al., 2015; Negus & Bergstedt, 2012; Spigarelli et al., 1977). For example, Negus and Bergstedt (2012) found that after moving Lake Trout (Salvelinus namaycush, ranging in size from 817g to 2230g) from 15.6°C to 1.1°C water, it took about 40 minutes or longer for their body temperatures to stabilize with the water temperature. An alternative way to estimate fish body temperatures without using internal temperature loggers is to use Newton's law of cooling, which has been used in research on fishes to investigate the heat exchange between their bodies and ambient water (e.g., Azumaya & Ishida, 2005; Brill et al., 1994; Negus & Bergstedt, 2012; Stevens & Sutterlin, 1976). Newton’s law of cooling states that an object’s rate of temperature ௗ் change ( ) is proportional to the difference between its body temperature (ܾܶ) and the ௗ௧ ambient temperature (ܶܽ) (Davis et al., 2016; Gaur et al., 2019; Gockenbach & Schmidtke, 2009): Equation 1.1 ݀ܶ = ݇(ܶܽ − ܾܶ) ݀‫ݐ‬ The whole-body heat exchange coefficient ݇ (unit of time-1), represents the rate at which an object (such as the core of a fish’s body) warms or cools to match the ambient temperature (Negus & Bergstedt, 2012; Pépino et al., 2015; Stevens & Sutterlin, 1976). 5 Newton’s law of cooling can be applied to an object that is cooling (when ܶܽ < ܾܶ) or heating up (when ܶܽ > ܾܶ) (Gockenbach & Schmidtke, 2009). If we assume that ܾܶ = ܾܶ଴ when time t = 0, Equation 1.1 can be integrated, resulting in (Gockenbach & Schmidtke, 2009): Equation 1.2 ܾܶ௧ = ܶܽ + (ܾܶ଴ − ܶܽ)݁ ି௞௧ When applying Newton’s law of cooling to fish with effective countercurrent heat exchange systems, researchers have modified Equation 1.2 to account for the fish’s metabolic heat output and heat retention (Brill et al., 1994; Kitagawa et al., 2001): Equation 1.3 ܾܶ௧ = ܶܽ + ܶ݉ ܶ݉ ି௞(௧ି௅) + (ܾܶ଴ − ܶܽ − )݁ ݇ ݇ Where ܶ݉ is the rate of body temperature increase due to metabolic heat output, and ‫ ܮ‬is the time lag before a change in body temperature is seen (due to a fish’s metabolic heat output and heat preservation through countercurrent heat exchange systems). Versions of Equation 1.3 have also been applied to fish species that do not possess countercurrent exchange systems, such as Blue Sharks (Prionace glauca) (Kitagawa & Kimura, 2006), Whale Sharks (Rhincodon typus) (Nakamura et al., 2020), Chum Salmon (Oncorhynchus keta) (Azumaya & Ishida, 2005), and Brook Trout (Salvelinus fontinalis) (Pépino et al., 2015). However, it is unclear if it is necessary to 6 include metabolic heat output for fish lacking countercurrent exchange systems (Brill et al., 1994; Pépino et al., 2015). Coefficient ݇ may vary depending on the difference between the object’s (or fish’s) internal body temperature and the ambient temperature and whether the object (body) is cooling or warming (especially for fish with countercurrent exchange systems) (Çengal & Ghajar, 2010a; Felchhelm & Neill, 1982; Holland et al., 1992; Pépino et al., 2015; Spigarelli et al., 1977; Stevens & Fry, 1974). Slightly faster rates of body heating compared to body cooling have been observed in various fish species including Bigeye Tuna (Thunnus obesus), Bluegill, and Brown Trout (Salmo trutta), and is likely due to internal body heat production and differences in heart rate in varying water temperatures (Fechhelm & Neill, 1982; Holland et al., 1992; Spigarelli et al., 1977). This is not a consistent trend across all fish species and studies, however, with some studies suggesting non-significant differences in the rates of heating and cooling or faster rates of cooling than heating (Fechhelm & Neill, 1982; Pépino et al., 2015; Pettit & Beitinger, 1980). Previous studies have also shown that the log of ݇ is negatively and linearly related to the log of body mass in a variety of fish species. Generally, a fish with a larger mass has a thicker body wall and greater volume than a lighter individual with the same body length and shape (hence a lower body surface area to volume ratio) (Nakamura et al., 2020; Pépino et al., 2015). Additionally, as objects with a larger mass have greater heat capacities than objects with a smaller mass, high body mass fish have greater thermal inertia than low body mass fish (Nakamura et al., 2020; Sidebotham et al., 2015). 7 Due to the differences in surface-to-volume ratios and heat capacities, heavier bodies exchange heat with the surrounding environment slower than lighter bodies with similar sizes and shapes (Nakamura et al., 2020; Pépino et al., 2015). Limited published research has investigated the relationship between other body size parameters (such as condition factor or plumpness, length, body wall thickness, and body surface area to volume ratio) and the heat exchange coefficient. Spigarelli et al. (1977) examined the relationship between fish body temperature change half-times (‫ݐ‬1/2, the time it took for a fish's body temperature to heat or cool halfway to the water temperature) and various morphological size parameters (weight, length, girth, and condition factor) of five species of fish from Lake Michigan using regression analysis. They found that the model with log-weight as the predictor variable had the highest ܴ 2 value when compared to the regressions between ‫ݐ‬1/2 and the other morphological variables. Additionally, multiple regression models including log-mass plus all combinations of the other measured morphological variables did not improve the ܴ 2 value. Other published research including estimations of fish coefficients of heat exchange have cited Spigarelli et al. (1977) as justification for using fish mass as a predictor variable for ݇ (which is similar yet distinct from ‫ݐ‬1/2) instead of other size variables (Negus & Bergstedt, 2012). If a fish’s ݇ and ܶ݉ are known, Newton's law of cooling can be used to estimate body temperature based on the ambient temperatures experienced by the fish (or vice versa) (Pépino et al., 2015). However, limited research investigating ݇ and the relationship between recorded body temperatures of fishes and the water temperatures 8 they experience has been published, and this remains true for members of the Salmonidae family (e.g. Lake Trout, Negus & Bergstedt, 2012; Chum Salmon, Azumaya & Ishida, 2005; Brook Trout, Pépino et al., 2015). The relationship could be applicable in studies where body temperatures are required but external temperature logger tags are used instead of internal loggers, for example when using pop-off data storage tags or external "backpack" tags/loggers instead of gastrically or surgically inserted loggers (e.g., to reduce handling time or avoid potential complications associated with anesthetic use) (Cooke et al., 2011; Jepsen et al., 2015; Raby et al., 2017). Additionally, body temperature could theoretically be estimated without the use of any tags if the position of a fish in a known thermal gradient is tracked. Knowing how quickly the body temperature of a fish changes after a change in water temperature can help researchers understand how the physiology and behaviour of a fish may be impacted after varying lengths of exposure to unfavorable ambient temperatures. This chapter aimed to investigate the relationship between varying ambient water temperatures and body temperatures of adult Sockeye Salmon from Babine Lake. Using water and body temperatures recorded every 1.5 minutes for the duration of 24-hour thermal preference tests (undergone by 20 individual adult female Babine Lake Sockeye Salmon), the whole-body heat exchange coefficient ݇ was estimated using Newton’s law of cooling and non-linear mixed models. The relationship between ݇ and fish body size (mass and condition factor), a heating vs. cooling body, and the absolute difference between body and ambient temperature was explored. Additionally, metabolic heat output (ܶ݉) was estimated and evaluated. Model averaged ݇ and ܶ݉ estimates were 9 used to predict the body temperatures of externally tagged Sockeye Salmon while they passed through Babine Lake. Estimated Babine Lake body temperatures were used to address the objectives outlined in Chapter Two. 10 Methods Fish Capture and Data Collection Adult migrating female Sockeye Salmon were captured at the Babine River Counting fence, a semi-permanent structure located approximately 1km downriver from Nilkitkwa Lake, British Columbia (see Chapter Two, Figure 2.2). Fish were dip-netted from a fence trap box and immediately transferred to a fish trough with a constant flowthrough of river water. Fish were temporarily immobilized using Smith-Root electric fish handling gloves set to a current output of 5mA (Reid et al., 2019). Fish were weighed in a rubber mesh weigh bag and fork length was measured. Other parameters not relevant to Chapter One were measured and will be described in Chapter Two. The fish were tagged with two identical temperature loggers (DST micro-T, Star Oddi, Iceland) that recorded temperature every 1.5 minutes with an accuracy of 0.06°C. One of the loggers was inserted gastrically (non-surgically through the mouth) and the other was secured externally directly below the dorsal fin. The external loggers were secured using Peterson disks and two pins inserted through the tissue below the dorsal fin (Figure 1.1). 11 Figure 1.1. A female Sockeye Salmon externally and internally tagged with temperature data loggers before beginning a thermal preference test in the shuttle box. After sampling and tagging, the fish were placed in a river water-filled shuttle box tank (Loligo® Systems), where they underwent a 24-hour thermal preference test. A total of 20 thermal preference tests were completed, each with a different fish. Depending on the location of the fish in the system, the water heated or cooled at a rate of 2°C / hour. During each test, the external and internal temperature loggers recorded the internal body temperatures and ambient water temperatures experienced by every individual salmon every 1.5 minutes. After the tests were completed, the temperature loggers were removed for data download and the fish were released upriver from the fence to continue their migration. The shuttle box temperature logger data was used to estimate both ݇ and ܶ݉ in Chapter One (see “Data Analysis” section below) and the thermal preferences of the salmon in Chapter Two. Please see Chapter Two for a detailed description of the shuttle box system and the thermal preference tests. All fish handling, sampling, and tagging was 12 completed from mid-August to mid-October 2021 and followed procedures approved by the UNBC Animal Care and Use Committee and Fisheries and Oceans Canada (UNBC protocol number: 2021-05, DFO Licence Number: XR 237 2021). Data Analysis Equation 1.4, a slightly modified version of Equation 1.3 that assumes ‫( ܮ‬the time lag before body temperature starts to change after a change in ambient temperature) is negligible, was used to estimate ݇ and ܶ݉ of the adult female Sockeye Salmon that underwent thermal preference tests (Azumaya & Ishida, 2005; Pépino et al., 2015): Equation 1.4 ܾܶ௧ = ܶܽ + ܶ݉ ܶ݉ ି௞௧ ൰݁ + ൬ܾܶ௧ିଵ − ܶܽ − ݇ ݇ Note that ܾܶ଴ in Equation 1.3 is replaced by ܾܶ௧ିଵ in Equation 1.4 because body temperature was measured at multiple intervals of 1.5 min during the shuttle box tests, so that the body temperature at the start and end of an interval are given by ܾܶ௧ିଵ and ܾܶ௧ , respectively. The value of ܶܽ used for an interval is given by the average of ܶܽ at times ‫ – ݐ‬1 and ‫ݐ‬. The gastrically measured body temperatures and estimated ݇ coefficients were assumed to be consistent throughout the body. Equation 1.4 was selected over Equation 1.2 (a similar equation that assumes ܶ݉ is negligible) as it has been shown that fish lacking counter-current heat exchange systems do not instantaneously lose all internally produced body heat (Dean, 1976; Stevens & Sutterlin, 1976). Due to the data being nested within individuals (i.e., a hierarchical structure) and Equation 1.4 being non- 13 linear, non-linear mixed models were used to estimate ݇ and ܶ݉ (Oddi et al., 2019; Pépino et al., 2015; Pinheiro & Bates 2006). Coefficient ݇ was modeled as a function of heating/cooling (whether the body heated or cooled since ‫ ݐ‬− 1), the absolute temperature difference between fish body temperature at ‫ ݐ‬− 1 and the average water temperature between ‫ ݐ‬− 1 and ‫ݐ‬, and the body size of the fish. Since there were very few instances where body temperatures did not heat or cool between temperature recordings (less than 20 instances), these data points were removed prior to modeling. Since there is uncertainty in which body size measurement most notably affects ݇, two preliminary non-linear mixed models were fit including either fish mass or condition factor (an indicator of fish plumpness) and the other fixed effects (heating/cooling and absolute temperature difference). Condition factor was calculated with the following equation (Schreck & Moyle, 1990): Equation 1.5 ‫ݏݏܽܯ ݕ݀݋ܤ‬ ൰ × 100 ‫ = ݎ݋ݐܿܽܨ ݊݋݅ݐ݅݀݊݋ܥ‬൬ ‫ݐ݃݊݁ܮ ݇ݎ݋ܨ‬ℎଷ The two body morphometric candidate models were compared using Akaike Information Criterion (AIC) (Burnham & Anderson, 2002; Burnham & Anderson, 2004). The model including condition factor had a slightly lower AIC score (-70780.43 vs 70780.23 for mass), so condition factor was used in the subsequent model selection for ݇. The absolute temperature difference and condition factor variables, along with the mass variable (included in the morphometric candidate models), were standardized before 14 modelling (Schielzeth, 2010). Fish number (fish 1-20) was included as a random effect for the intercepts of ݇ and Tm. Therefore, the global model for ݇ was: Equation 1.6 ݇௜,௧ = (ߙ௞ + ߛ௜ ) + ߚ௛௘௔௧ ‫ݔ‬௛௘௔௧೔,೟ + ߚ௔௧ௗ ‫ݔ‬௔௧ௗ೔,೟ + ߚ௖ ‫ݔ‬௖೔ where ݇௜,௧ is the heat coefficient for individual ݅ during the time interval ‫ݐ ( ݐ‬௧ିଵ to ‫)ݐ‬, ߙ௞ is the intercept; ߛ௜ is the random deviation from the intercept for individual ݅, which is assumed to follow Normal൫0, ߪఊ ൯, ߚ௛௘௔௧ is the effect of heating (i.e. ‫ݔ‬௛௘௔௧೔,೟ = 1 when the body of individual ݅ warmed between ‫ ݐ‬− 1 and ‫ݐ‬, and 0 when cooled); ߚ௔௧ௗ is the slope for the effect of the standardized absolute temperature difference (‫ݔ‬௔௧ௗ ) of individual ݅ between ‫ ݐ‬− 1 and ‫ݐ‬, and ߚ௖ is the slope for the effect of standardized condition factor (‫ݔ‬௖ ) of individual ݅ (‫ݔ‬௖೔ ). The model used to estimate ܶ݉ was: Equation 1.7 ܶ݉௜ = ߙ்௠ + ߜ௜ where ܶ݉௜ is the rate of body temperature increase due to metabolic heat output by individual ݅, ߙ்௠ is the intercept, and δi is the random deviation from the intercept for individual ݅, which is assumed to follow Normal(0, ߪఋ ). The global model was fit to the data and model assumptions of residual homoscedasticity, normality and independence were checked graphically (Zuur et al., 2010). Assessments did not indicate violations of residual homoscedasticity and normality, but residuals were temporally autocorrelated. To account for temporal autocorrelation, an autocorrelation structure of order one for the 15 residuals was included. As published researched has estimated ݇ with equations including and excluding ܶ݉ (e.g. Pepino et al., 2015; Negus & Bergstedt 2012), two additional preliminary global models, one including ܶ݉ as a response variable and one with ܶ݉ set to 0, were fit and compared using AIC (Burnham & Anderson, 2002; Burnham & Anderson, 2004). The model including ܶ݉ as a response variable had a lower AIC score (-70780.43 vs -70374.05), so ܶ݉ was included in all models in the final candidate set. Following validation of the global model, a total of eight candidate models were fit to the data using maximum likelihood, including an intercept-only model and models containing all possible additive combinations of the absolute temperature difference, warming/cooling, and condition factor variables as fixed effects on ݇. To determine the most parsimonious model, AIC scores, ΔAIC values and AIC weights were calculated (Burnham & Anderson, 2004). As no singular model had an AIC weight greater than 0.95, models with cumulative AIC weights adding to at least 0.95 were averaged (Burnham & Anderson, 2004; Symonds & Moussalli, 2011). Before averaging, the weights of these models were re-scaled and the models were re-fit using restricted maximum likelihood. The relative variable importance of each fixed effect was calculated by summing the AIC weights of the models containing the effect in the full candidate models set (Burnham & Anderson, 2004). Models were fit to the data using package nlme (3.1-152, Pinheiro et al., 2021) in R (R Core Team, 2021). The averaged model was then used to predict the body temperatures of Sockeye Salmon passing through Babine Lake. These fish were externally tagged with a radio tag 16 (NTF-5-2, Lotek Wireless, Canada) and temperature/depth logger (LAT1100, Lotek Wireless, Canada; accuracy better than ± 0.2°C) that recorded ambient water temperature every 1.5 minutes, and their body weight and fork length were also measured during tagging. Further details on the tagging procedure can be seen in Chapter Two. The radio tagged fish only had external temperature loggers, so estimating and applying ݇ and ܶ݉ allowed us to predict the body temperatures of the radio tagged Sockeye Salmon (an important component of the calculations of effectiveness of behavioural thermoregulation performed in Chapter Two). The median (± median absolute deviation (MAD)) recorded ambient water temperature, predicted body temperature, and absolute difference between the ambient and predicted body temperatures (paired by fish and timestamp) while the fish were in Babine Lake were calculated and compared. Additionally, to investigate if greater and more rapid fluctuations in water temperature resulted in greater discrepancies between the body and water temperatures, the hourly average absolute difference in body and water temperatures were linearly modelled with the hourly average change in ambient water temperature experienced in Babine Lake: Equation 1.8 |ܶܽ − ܾܶ| ௜,௛ = (ߙ|்௔ି்௕| + ߛ௜ ) + ߚ‫ݔ‬௪௧௖ ௜,௛ + ߝ௜,௛ Where |ܶܽ − ܾܶ| is the hourly average absolute difference in body and water temperatures of individual ݅ during hour ℎ of their passage through Babine Lake, ߙ|்௔ି்௕| is the intercept, ߛ௜ is the random deviation from the intercept for individual ݅ 17 which is assumed to follow Normal൫0, ߪఊ ൯, ߚ is the slope, ‫ݔ‬௪௧௖ is the average rate of change in water temperature (°C / min) experienced by individual ݅ during hour ℎ, and ߝ௜,௛ is the residual error for individual ݅ during hour ℎ, assumed to follow Normal(0, ߪ). |ܶܽ − ܾܶ| and ‫ݔ‬௪௧௖ were square-root transformed prior to modelling to reduce residual heteroskedasticity and normalize their distribution. Additionally, an autocorrelation structure of order one was included due to the residuals being temporally autocorrelated. The hourly average change in water temperature was calculated as: Equation 1.9 ߂ܶܽ௜,௛,௠ ∑௜,௛ ൬ ൰ ߂‫ݐ‬௜,௛,௠ ‫ݔ‬௪௧௖ ௜,௛ = ݊௜,௛ Where ‫ݔ‬௪௧௖ is the average rate of change in ambient water temperature experienced by individual ݅ during hour ℎ of their time in Babine Lake, ߂ܶܽ is the difference in water temperature (ܶܽ) experienced by individual ݅ during hour ℎ between the current observation (݉, every 1.5 minutes) and the last observation (݉ − 1), ߂‫ ݐ‬is elapsed time (in minutes) since the last ܶܽ observation in hour ℎ (ܶܽ at ݉ – 1), and ݊ is the number of ambient water temperatures recorded for individual ݅ during hour ℎ. 18 Results The masses of the Sockeye Salmon that underwent thermal preference tests ranged from 1000g to 2200g (median: 1600g, median absolute deviation (MAD): 148.26g), and their estimated condition factors ranged from 0.78 – 1.15 (median: 0.98, MAD: 0.11). The median internal body temperature recorded during the thermal preference tests was 12.77°C (MAD: 2.73°C, range: 3.22°C to 19.75°C) and the median recorded external ambient temperature was 12.61°C (MAD: 2.92°C, range: 3.12°C to 19.89°C). The median absolute difference between the body and ambient recorded temperatures was 0.32°C (MAD: 0.28°C, range: 0.00°C to 2.44°C). Model Selection and Averaging The most parsimonious model for the coefficient of heat exchange, ݇, included absolute temperature difference as the singular fixed effect (Table 1.1). The secondranked model, with absolute temperature difference and warming/cooling as fixed effects, had a slightly smaller AIC weight than the top model (Table 1.1). Models with condition factor also had some support from the data (Table 1.1). The absolute temperature difference was present in all top four models, and its relative variable importance was 1.00. The relative variable importance of cooling/warming and condition factor was 0.46 and 0.31, respectively. The cumulative AIC weight of the top four models reached the 0.95 confidence set threshold and these models were used to compute model averaged predictions (Table 1.1). 19 Table 1.1. AIC scores and related statistics for candidate models for the heat coefficient, ݇. P is the number of parameters in the model, AIC is the AIC score, ΔAIC is the difference between the model AIC score and the top model AIC score, lik is the model likelihood, wAIC is the model AIC weight, and cwAIC is the cumulative AIC weight. In the fixed effects column, ܽ‫ ݀ݐ‬is the absolute temperature difference variable, ℎ݁ܽ‫ ݐ‬is the heating/cooling variable, and ܿ is condition factor. Fixed Effects P AIC ΔAIC lik wAIC cwAIC ܽ‫݀ݐ‬ 8 -70782.47 0.00 1.00 0.38 0.38 ܽ‫ ݀ݐ‬+ ℎ݁ܽ‫ݐ‬ 9 -70782.1 0.37 0.83 0.32 0.70 ܽ‫ ݀ݐ‬+ ܿ 9 -70780.8 1.67 0.43 0.17 0.86 ܽ‫ ݀ݐ‬+ ℎ݁ܽ‫ ݐ‬+ ܿ 10 -70780.43 2.04 0.36 0.14 1.00 ‫ݕ݈ܱ݊ ݐ݌݁ܿݎ݁ݐ݊ܫ‬ 7 -70722.1 60.37 0.00 0.00 1.00 ℎ݁ܽ‫ݐ‬ 8 -70720.62 61.85 0.00 0.00 1.00 ܿ 8 -70720.44 62.03 0.00 0.00 1.00 ℎ݁ܽ‫ ݐ‬+ ܿ 9 -70718.96 63.51 0.00 0.00 1.00 Model Averaged Estimates Model-averaged estimates revealed that absolute temperature difference had a positive effect on heat coefficient ݇ (with higher absolute temperature differences corresponding with a faster heat exchange; Figures 1.2 and 1.3), and its 95% confidence interval did not include zero (Figure 1.2). Coefficient ݇ was slightly higher when the body was warming compared to when the body was cooling (represented by the intercept 20 estimate) and its 95% confidence interval overlapped zero (Figure 1.2). Condition factor had a negative effect on ݇ (with higher condition factors resulting in slower heat exchange; Figures 1.2 and 1.3), but uncertainty about its effects were quite large, with the 95% confidence intervals including zero (Figure 1.2). The mean model-averaged ݇ values for each fish ranged from 0.07 min-1 to 0.10 min-1, and the median ݇ estimate was 0.08 min-1 (MAD: 0.003 min-1). The averaged estimate for the rate of body temperature increase due to metabolic heat, ܶ݉, was 0.007 min-1 (95% CI: 0.004 min-1 to 0.009 min1 ). Model-averaged predictions of body temperatures during the tests closely resembled the actual recorded body temperatures (median difference: 0.02°C, MAD: 0.02°C, range: 0.00°C to 0.41°C; Figure 1.4). Figure 1.2. Model averaged estimates (with 95% confidence intervals) of the intercept and fixed effects for the heat coefficient ݇ (min-1). 21 Figure 1.3. Model-averaged estimates of the heat coefficient, ݇, for a fish with the minimum (0.78), average (0.96), and maximum (1.15) condition factors measured for the shuttle box test fish. Figure A represents ݇ estimated with an absolute temperature difference of 0°C (the minimum observed absolute temperature difference in shuttle box tests), B represents ݇ estimated with an absolute temperature difference of 0.37°C (the average observed temperature difference in the shuttle box tests), and C represents ݇ estimated with an absolute temperature difference of 2.47°C (the maximum observed temperature difference in the shuttle box tests). 22 Figure 1.4. Recorded vs. model-averaged predicted body temperatures for the fish during their shuttle box tests. The orange line represents equal predicted and recorded temperatures. Visualizing Body Temperature Rate of Change Model-averaged predictions of the heat exchange model showed that a female Sockeye Salmon with a body temperature between 4°C to 12°C cooler or warmer than its stable ambient water temperature would take approximately 35 – 40 minutes to equilibrate with the ambient temperature (Figure 1.5). Despite minimal differences in the overall time it would take the body temperature to stabilize in the three different scenarios, the rate of body temperature change is initially faster when there is a larger difference between body temperature and water temperature (Figure 1.5). Rates of 23 temperature change slow down as the body temperature approaches that of the water temperature (Figure 1.5). Figure 1.5. Model-averaged predictions of body temperature over time of a Sockeye Salmon with a condition factor of 0.96. Plot A shows how the body temperature of a fish would warm after beginning at 8°C, 12°C, and 16°C and being placed in 20°C water. Plot B shows how the body temperature of a fish would cool after beginning at 20°C, 16°C, and 12°C and being placed in 8°C water. Sockeye Salmon Body and Water Temperatures in Babine Lake The masses of radio-tagged Sockeye Salmon ranged from 1000g to 3000g (median: 1500g, MAD: 296.5g), and their condition factors ranged from 0.76 to 1.64 (median: 1.01, MAD: 0.10). The median and minimum heat exchange coefficients ݇ 24 predicted for the fish in Babine Lake were nearly identical to the median and minimum values estimated for the fish in the shuttle box tests (median: 0.08 min-1; MAD: 0.002 min-1, minimum: 0.07 min-1). The maximum ݇ value for the fish in Babine Lake (0.13 min-1) was slightly higher than the maximum ݇ estimated for the shuttle box fish. The median temperature recorded by the external loggers on Sockeye Salmon in Babine Lake was 10.56°C (MAD: 3.90°C, range: 4.35°C - 17.35°C), slightly cooler and more variable than their model-averaged predictions of body temperatures (median: 10.68°C, MAD: 3.67°C, range: 4.73°C - 17.30°C). The median difference between the recorded water temperatures and model-averaged predictions of body temperatures was 0.18°C (MAD: 0.19°C, range: 0.00°C – 7.63°C). There was a positive and linear relationship between the mean hourly absolute difference of model-averaged predictions of body and recorded water temperatures and the hourly rate of change in water temperature (Figure 1.6). This relationship indicates greater differences between body temperature and ambient temperatures occurred when fish were experiencing more rapid and greater water temperature changes. 25 Figure 1.6. The relationship between the square-root transformed mean rate of water temperature change and square-root transformed mean absolute difference between the body temperatures of radio-tagged Sockeye Salmon in Babine Lake (model-averaged predictions) and the water temperatures they encountered. As mentioned in the methods section, these values were calculated for each recording (taken every 1.5 minutes) then averaged for the corresponding individual and hour. The orange line represents the linear relationship modelled between the variables when assuming zero residual error and a random deviation from the intercept of zero. The equation of the line has |ܶܽ − ܾܶ| representing the square rooted average absolute difference in body and water temperatures of an individual salmon, and ‫ݔ‬௪௧௖ representing the average rate of change in water temperature experienced by an individual during an hour in Babine Lake. The R2 value represents both the marginal and conditional R2. Figure 1.7 further illustrates how the greatest differences in body temperature and water temperature occurred when the tagged fish experienced relatively large and rapid changes in water temperature while passing through Babine Lake. An example of a rapid change in water temperature and how predicted body temperature differed from water temperature during this time can be seen in Figure 1.7 for fish ID 11 between September 26 8th at 18:00:00 and September 9th at 00:00:00. These rapid changes may have occurred when the fish were diving or ascending through the thermocline in Babine Lake. When the ambient temperature of the fish remained relatively constant, so did the fish’s predicted body temperatures, which was demonstrated by Fish ID 63 from 21:00:00 on August 30th to 04:00:00 on August 31st (Figure 1.7). Slower and smaller changes in ambient temperature also resulted in predicted body temperatures closely mirroring water temperatures, demonstrated by fish ID 63 on August 31st from 06:00:00 to 12:00:00 (Figure 1.7). Figure 1.7 illustrates how predicted body temperatures were slightly less variable than the experienced water temperatures in Babine Lake, as recorded water temperatures often exceeded predicted body temperatures at the peaks of rapid temperature changes and were cooler than predicted body temperatures at the bottom of the valleys of rapid temperature changes. 27 Figure 1.7. Twenty-four hour snapshot of water temperatures (recorded every 1.5 minutes) experienced by radio-tagged Sockeye Salmon ID 11 and 63 and their corresponding predicted body temperatures that occurred while passing through Babine Lake on their way to spawning grounds. 28 Discussion The primary goal of this chapter was to investigate the relationship between environmental water temperatures and adult Sockeye Salmon body temperatures using Newton’s law of cooling and estimating the fish’s coefficient of heat exchange, ݇, and the rate of body temperature increase due to metabolic heat, ܶ݉. My results suggest that ݇ varied primarily depending on the magnitude of the absolute temperature difference between the fish’s body and ambient water. It also minorly depended on a fish’s condition factor and whether its body was heating or cooling. Additionally, ܶ݉ appeared to be a small but notable factor in predicting body temperatures of Sockeye Salmon. The estimated values of ݇ for fish in the shuttle box tests (median: 0.08 min-1, range: 0.07 min-1 to 0.10 min-1) and in Babine Lake (median: 0.08 min-1, range: 0.07min-1 to 0.13 min-1) were very similar to published ݇ values for other members of the Salmonidae family. For example, Negus and Bergstedt (2012) estimated that the ݇ for Lake Trout ranged from 0.03 min-1 to 0.12 min-1, and Azumaya and Ishida (2005) found that the ݇ of adult Chum Salmon was 0.09 min-1. Pépino et al. (2015) found that the ݇ for Brook Trout ranged from approximately 0.14 min-1 to 0.17 min-1, depending on the absolute difference between body temperature and water temperature, which is similar but slightly higher than the estimated ݇ values for adult Sockeye Salmon (our study), Chum Salmon (Azumaya & Ishida, 2005) and Lake Trout (Negus & Bergstedt, 2012). The larger ݇ may be explained by the size differences of the fishes studied as the Brook Trout used in the Pépino et al. (2015) study ranged from about 200 g to 600 g, notably 29 smaller than the fish masses in our study that ranged from 1000 g to 2200 g, the Lake Trout studied by Negus and Bergstedt (2012) ranging from 817 g to 2230 g, and the Chum Salmon studied by Azumaya and Ishida (2005) ranging in fork length from 57 cm to 69 cm (similar and slightly longer than the lengths of Sockeye Salmon in our study). Holland et al. (1992) estimated the ݇ value for Bigeye Tuna, a fish species with countercurrent exchange systems and means of physiological thermoregulation, to be 0.03 min-1 when cooling and 2.41 min-1 when warming, which unsurprisingly are much different than the values estimated for Sockeye Salmon. The combination of the use of countercurrent heat exchange in cool water to preserve heat and slow cooling and the higher metabolic output of this species of Tuna likely contributed to the lower and higher ݇ values, respectively (Holland et al., 1992; Sidebotham et al., 2015). Condition factor, Body Size, and ࢑ Preliminary comparisons of models predicting ݇ as a function of body size parameters indicated that condition factor was a slightly better predictor of ݇ than those that included mass as a fixed effect. This was surprising as results from Spigarelli et al. (1977) suggested that fish mass was more strongly related to the time it took for the bodies of five freshwater fish species to heat or cool than other size parameters such as length, girth, and condition factor. Additionally, many other studies have found moderate or strong linear relationships between mass and ݇; however, they did not look at the relationship between other body size parameters and ݇ (e.g. Fecchelm & Neill, 1982; Nakamura et al., 2020; Negus & Bergstedt, 2012; Pépino et al., 2015). The relationship 30 between mass and ݇ can be explained by the fact that fishes with larger masses have slower rates of heat exchange due a thicker body wall and a lower surface-area-to-volume ratio (hence a greater thermal inertia) than lighter fishes with the same body shape (Nakamura et al., 2020; Pépino et al., 2015; Sidebotham et al., 2015). However, it is possible that condition factor, in this case Fulton’s condition factor, may be a better indicator of body wall thickness than mass alone. Fulton’s condition factor considers the mass of the fish relative to its length, and higher factors have been associated with a plumper body (Akpesse et al., 2022; Matondo et al., 2021; Treer & Babačić, 2017). As an object’s thermal resistance (the degree of resistance to heat flow) is usually directly related to its thickness (Çengal & Ghajar, 2010b; Mishra et al., 2019), a fish with a lower condition factor and an elongated, skinny body would have a shorter distance (and lower thermal resistance) between the water and their isothermal core than a plumper or thicker fish of the same mass. Consequently, the core of the low-condition fish would heat or cool at a faster rate than that of the high-condition fish. It is important to note that this relationship would only apply to individuals with similar masses, and perhaps both mass and condition factor should have been included in the models to better represent the thermal inertia of the fish. Despite the slightly higher support of a model with condition factor over one with mass, model selection indicated little support for the inclusion of condition factor when compared to other models in the candidate set. Heating, Cooling, and ࢑ There was a minimal difference between the estimated ݇ value for a warming or cooling body, with the warming ݇ being slightly higher. Pépino et al. (2015) also found 31 little to no difference between ݇ for heating and cooling Brook Trout and Kitagawa et al. (2001) found that ݇ was nearly the same when cooling and warming the bodies of Pacific Bluefin Tuna (Thunnus thynnus), a species of Tuna that is believed to not regulate body temperature using physiological means. A greater ݇ while the body is heating (and a more rapidly changing body temperature) than while cooling has been reported for a variety of other fish species. For Bigeye Tuna, the more rapid rate of heating than cooling is caused by their countercurrent heat exchange system that activate at cooler temperatures to retain body heat and disengage at warmer, more ideal temperatures to exchange heat more rapidly with the ambient water (Holland et al., 1992). However, many other fishes with different body heating and cooling rates, such as Bluegills (Fechhlem & Neill, 1982), Whale Sharks (Nakamura et al., 2020), and Smallmouth Bass (Micropterus dolomieu) (Weller et al., 1984), do not have countercurrent heat exchange systems. For these fishes, it has been hypothesized that a higher ݇ or rate of body temperature change occur 1) when bodies are warming due to heat produced internally (i.e. through metabolism) that slightly accelerates the body heating; 2) due to more rapid circulation and respiration in warm waters, which promotes faster heat exchange within tissues and via the gills; and 3) due to increased spontaneous movement, which can be triggered by encountering warmer ambient temperatures, and result in higher metabolic rates and more internally produced heat (Spigarelli et al., 1977). These relationships have been observed in reptiles and fishes (Chung et al., 2021; Nilsson et al., 2009; Sebacher & Grigg, 2001; Speers-Roesch et al., 2018). For example, Prystay et al. (2020) found that 32 the heart rates of adult Sockeye Salmon on spawning grounds increased by a mean 2.7 beats per minute for every 1°C increase in water temperature. Additionally, Nilsson et al. (2009) found that the respiration of five species of reef fishes was significantly impacted by water temperature, with notable resting rates of oxygen uptake (and therefore rate of gas exchange at the gills) occurring in warmer water temperatures. The metabolic rates (and the produced heat from metabolic processes) of various fish species have been observed to increase in warming water temperatures and decrease in colder temperatures (Chung et al., 2021; Speers-Roesch et al., 2018). For example, the metabolic rates of juvenile Atlantic Cod (Gadus morhua) from the southern coast of Norway increased with increasing water temperatures until reaching temperatures above 16°C (Chung et al., 2021). Similarly, Trudel and Welch (2011) found that water temperature positively correlated with the standard and total metabolic rates of Sockeye Salmon and Steelhead Salmon. Since my models accommodated for metabolic heat production in the ܶ݉ variable, differences in metabolic heat production while cooling or heating should not have impacted ݇. However, the slightly higher ݇ for warming could be explained by increased respiration rates and thus increased heat exchange through the gills that occurs when bodies warm (as seen by Nilsson et al., 2009; Seebacher & Grigg, 2001). Absolute Temperature Difference and ࢑ The absolute temperature difference between the body and water had the strongest effect of the fixed effects for ݇ that were included in the models. This result corroborates Pépino et al. (2015), who were the first to demonstrate that the absolute temperature difference between the body of a fish (Brook Trout) and the water positively affected ݇. 33 Pépino et al. (2015) noted that this relationship may to due to greater differences in water and body temperatures elevating the stress level of the fish, resulting in increased cardiovascular activity (e.g., respiration) and heat exchange. Effect of ࢀ࢓ on Body Temperature My results suggest that ܶ݉ had a small yet positive effect on the body temperatures of adult Sockeye Salmon. The estimated value of ܶ݉ of 0.007°C min-1 was similar to the value of ܶ݉ estimated by Azumaya and Ishida (2005) for Chum Salmon (0.01°C min-1). Despite this similarity, it was surprising that ܶ݉ appeared to have a notable impact on body temperature (the global model including ܶ݉ as a response variable had a lower AIC score than the global model without ܶ݉) as studies on fishes lacking countercurrent heat exchange systems have found ܶ݉ to have an insignificant impact on body temperature or have left out ܶ݉ when using Newton’s law of cooling to estimate body temperature. For example, Negus and Bergstedt (2012) did not consider ܶ݉ when estimating ݇ and body temperatures of Lake Trout, whereas Fecchlem and Neil (1982) found that the body temperatures of Bluegill, Blue Tilapia (Tilapia aurea), and Nile Tilapia (Tilapia nilotica) were negligibly impacted by their metabolically produced heat. Likewise, Nakamura et al. (2020) concluded that heat internally produced by Whale Sharks did not affect muscle temperature. Pépino et al. (2015) had variable results regarding the effects of ܶ݉ on the body temperatures of Brook Trout. They found that ܶ݉ had a small effect on body temperature when the body was heating but not when it was cooling. Perhaps the body size and shape of a fish alters the effect of Tm on body 34 temperature, with metabolically produced heat dissipating more quickly in flat, oval shaped fish (such as the Nile Tilapia and Bluegill) than in rounder fish such as Sockeye Salmon and Brook Trout, and not changing the overall body temperature of Whale Sharks due to their extremely high thermal inertia. A ܶ݉ that is large enough to affect body temperature during heating can be explained by fish exhibiting higher metabolic activity in warmer temperatures (as discussed earlier). Additionally, if the water temperature is warmer than the body temperature of a fish (i.e. a warming body) then metabolically produced body heat would not be lost to the surrounding environment and would enhance the rate of warming (Chung et al., 2021; Speers-Roesch et al., 2018; Spigarelli et al., 1977). In this study, we did not estimate ܶ݉ separately for cooling and warming, and the value estimated for ܶ݉ fell in the middle of the range of ܶ݉ values estimated by Pépino et al. (2015). It is possible that we would have had a similar result to Pépino et al. (2015) if we had estimated separate cooling and heating ܶ݉ values. Based on the lack of consensus regarding the effect of ܶ݉ on the body temperature of fish, future studies should investigate its importance before excluding it from analysis of heat transfer in fish. Predicted Body and Observed Ambient Temperatures in Babine Lake The predicted body temperatures experienced by adult Sockeye Salmon were similar to environmental water temperatures recorded while they were passing through Babine Lake. However, body temperatures were less variable than water temperatures, which is consistent with what has been found in other studies on fish (Nakamura et al. 35 2020; Pépino et al. 2015). There was also a positive and linear relationship between the average hourly temperature difference in estimated body temperature and water temperature in Babine Lake, and the average hourly rate of external temperature change. This result indicates that the greatest differences between body temperature and water temperature occurred after a large and rapid change in water temperature. These large differences could have occurred when a fish descended or ascended through the thermocline in Babine Lake, which is typically located at a depth of approximately 11m from May to October (Shortreed & Morton, 2000). Another important takeaway is that it would take an estimated 40 minutes for the internal body temperature of adult Sockeye Salmon to equilibrate after migrating into water temperatures 4°C to 12°C different from their initial body temperature. Similarly, Pépino et al. (2015) predicted that it would take approximately 20 to 30 minutes for the body temperature of Brook Trout to stabilize with the water temperature, and Negus and Bergstedt (2012) observed that it took Lake Trout up to approximately 40 minutes to do the same, depending on the magnitude of difference between the body and ambient water temperature and the size of the fish. Due to the delay until body temperature stabilizes with water temperature, a very brief exposure to an extreme or unfavourable water temperature would likely not be fully reflected by their body temperature. They may avoid or lessen the detrimental physiological effects that would occur if their body reached that extreme temperature since the effects of temperature on a fish depend on the magnitude and duration of temperature exposure (Rezende et al. 2014; Volkoff & Rønnestad, 2020). On the opposite end, if a fish experienced an unfavourable water 36 temperature long enough for their body to stabilize with it, it would take more than a brief exposure to favourable temperatures for their bodies to reach a more favourable temperature again. Additional Study Assumptions and Limitations The relationship between ݇ and additional body size parameters (e.g. girth), or between ݇ and activity or respiration level was not explored. Additionally, differences in ܶ݉ during body heating and cooling were not evaluated. Since previous research found that ܶ݉ negligibly or marginally impacted body temperature of fish lacking counter current heat exchange systems (e.g. Pépino et al., 2015), estimating ܶ݉ for Sockeye Salmon was considered a secondary focus of this chapter. As such, investigating relationships between ܶ݉ and fixed effects such as heating/cooling was not the focus and was not completed. Including these variables may have resulted in more accurate estimates of ݇ and ܶ݉, and better estimates of Sockeye Salmon body temperatures in Babine Lake. Another variable that was not tested but could impact ݇ is the body composition of the fish. Different body tissues have different heat capacities and conductivity, meaning the relative body tissue type percentage could influence how quickly the body exchanges heat with its surroundings (Faber & Garby, 1995; Kubb et al., 1980; Niell et al., 1976). For example, Niell et al. (1976) found that the red muscle of Skipjack Tuna (Katsuwonus pelamis) equilibrated with ambient water temperatures much slower than white muscle, the brain, and ventricle blood. Similarly, Kubb et al. (1980) found that the heart region of Smallmouth Bass exchanged heat with the surrounding 37 water notably faster than the gut region likely due to the lack of lateral insulation around the heart and its proximity to the gills. Faber and Garby (1995) found that the heat capacity of the body of mice was notably higher when they had higher body fat percentages, indicating mice with more fat would heat or cool more quickly. Estimating the relative percentage of different tissue types in a non-lethal study such as the present one may not be possible, however relative fat percentage was measured non-lethally using a microwave fat probe (see Chapter Two for details). The relationship between relative fat percentage and ݇ could be explored, however it would be important to consider that the body fat percentage of a salmon will change after sampling since Pacific salmon tend to consume body fat and replace lost tissues with water as they migrate to spawning grounds (Crossin & Hinch, 2005). Using a singular ݇ value for the entire body of a fish and assuming the temperature in the gut of the fish is representative of the whole-body temperature is a simplification. In reality, the body temperature of a fish varies throughout the body. For example, the body temperature of Salmon Sharks (Lamna ditropis) in the Gulf of Alaska varied greatly depending on where the temperature was measured in the body, with hearts having the coolest temperatures and spiral valves (part of the digestive tract) having the warmest temperatures (Anderson & Goldman, 2001). Additionally, different tissues and fluids in the bodies of Skipjack Tuna and Smallmouth Bass have been shown to equilibrate with the ambient water temperature at different rates (Kubb et al., 1980; Neill et al., 1976). Accounting for these differences, similar to incorporating different body 38 tissue compositions, would add excess complexity to the data collection and analysis that may not be necessary. Another assumption made was that the temperature of the water surrounding the entire body of the tagged Sockeye Salmon was the same as the temperature recorded by the external temperature loggers attached under their dorsal fin. This assumption may have been violated while the fish passed through Babine Lake, however if it was, differences in the temperature of the water surrounding the fish would likely have been very small. Trying to measure or account for possible fine-scale differences would require an additional temperature logger attached to another area on the body of the fish or incorporating habitat temperature distribution and fine-scale fish positional data into the analysis, both of which would add more complexity to the research. Furthermore, the accuracy of the estimates of ݇, ܶ݉, and body temperatures of the Sockeye Salmon passing through Babine Lake could have been improved by recording and analyzing temperature data at more frequent time intervals. Temperature in both the lab and tagged fish was recorded every 1.5 minutes, so any variability in body temperature and water temperature that occurred between time stamps was missed. The temperature logger response time (up to 8 seconds) and temperature accuracy ratings of the loggers (± 0.2°C) may have also led to inaccuracies in theses estimates. Additionally, I was only able to assess how ݇ varied with absolute differences in body and water temperature up to 2.47°C degrees in the shuttle box tests. If the relationship between ݇ and the absolute temperature difference is linear (which was assumed in the models of ݇), it should extrapolate well for larger temperature differences. 39 This linear relationship is supported by results found by Pépino et al. (2015), who found that ݇ increased linearly with increasing absolute differences in Brook trout body and water temperatures up to differences of 13°C (which was the greatest difference tested). Based on this, I believe it was reasonable to assume ݇ would linearly increase up to differences of 7.63°C, which was the greatest difference estimated for the Sockeye Salmon in Babine Lake. Applications and Potential for Future Research The estimated values of ݇ and ܶ݉ from this study could be applied to other projects interested in predicting body temperatures of adult Sockeye Salmon or other similarly sized and shaped salmonids. The water temperatures salmon experience could also be predicted based on internally recorded body temperatures. If doing the latter, the investigator should proceed with caution, as Pépino et al. (2015) found that using ݇ and known body temperatures of Brook Trout to predict the water temperatures they experienced was not entirely accurate due to different thermal experiences resulting in the same body temperatures. Additionally, others could follow the framework outlined in this chapter to predict values of ݇ and ܶ݉ for different species and age classes of fishes. The differences in body and water temperatures experienced by Babine Lake Sockeye Salmon in this study highlight a need to exhibit caution when equating water temperature to the body temperature of fish in thermally heterogenous environments. 40 References Akpesse, A. M. A., Kacou-Wodjé, C. N. E., Danielle, O. A. R., Diabaté, D., Coulibaly, T., Kissi, T. A. P., Koua, K. H., & Kouassi, K. P. (2022). Optimization of the production of black soldier flies Hermetia Illucens by controlling biological parameters in Côte d'Ivoire. Journal of Advances in Biology, 15, 11-19. https://doi.org/10.1371/journal.pone.0216160 Anderson, S. D., & Goldman, K. J. (2001). Temperature measurements from salmon sharks, Lamna ditropis, in Alaskan waters. Copeia, 2001(3), 794-796. https://doi. org/10.1643/0045-8511(2001)001[0794:TMFSSL]2.0.CO;2 Angilletta Jr, M. J. (2009). Thermal Adaption: A Theoretical and Empirical Synthesis. Oxford University Press. Armstrong, J. B., Ward, E. J., Schindler, D. E., & Lisi, P. J. (2016). Adaptive capacity at the northern front: sockeye salmon behaviourally thermoregulate during novel exposure to warm temperatures. Conservation Physiology, 4(1), cow039. https://doi.org/10.1093/conphys/cow039 Azumaya, T., & Ishida, Y. (2005). Mechanism of body cavity temperature regulation of chum salmon (Oncorhynchus keta) during homing migration in the North Pacific Ocean. Fisheries Oceanography, 14(2), 81-96. https://doi.org/10.1111/j.13652419.2004.00323.x Brett, J. R. (1971). Energetic responses of salmon to temperature. A study of some thermal relations in the physiology and freshwater ecology of sockeye salmon (Oncorhynchus nerka). American zoologist, 11(1), 99-113. https://doi.org/10. 1093/icb/11.1.99 Brill, R. W., Dewar, H., & Graham, J. B. (1994). Basic concepts relevant to heat transfer in fishes, and their use in measuring the physiological thermoregulatory abilities of tunas. Environmental Biology of Fishes, 40(2), 109-124. https://doi.org/10.1007/BF00002538 Burnham, K. P., & Anderson, D. R. (2002). Model selection and multimodel inference: A Practical Information-Theoretic Approach (2nd ed.). Springer-Verlag, New York. http://dx.doi.org/10.1007/b97636 Burnham, K. P., & Anderson, D. R. (2004). Multimodel inference: understanding AIC and BIC in model selection. Sociological Methods & Research, 33(2), 261-304. https://doi.org/10.1177/0049124104268644 41 Carey, F. G., Teal, J. M., Kanwisher, J. W., Lawson, K. D., & Beckett, J. S. (1971). Warm-bodied fish. American Zoologist, 11(1), 137-143. https://doi.org/10.1093/icb/11.1.137 Çengal, Y. A., & Ghajar, A. J. (2010a). Chapter 2: Heat Conduction Equation. Heat and mass transfer: Fundamentals & applications (4th ed.), (pp. 63 – 65). McGraw-Hill Education. Çengal, Y. A., & Ghajar, A. J. (2010b). Chapter 3: Steady Heat Conduction. Heat and mass transfer: Fundamentals & applications, (4th ed.), (pp. 139 – 159). McGrawHill Education. Chung, M. T., Jørgensen, K. E. M., Trueman, C. N., Knutsen, H., Jorde, P. E., & Grønkjær, P. (2021). First measurements of field metabolic rate in wild juvenile fishes show strong thermal sensitivity but variations between sympatric ecotypes. Oikos, 130(2), 287-299. https://doi.org/10.1111/oik.07647 Cooke, S. J., Woodley, C. M., Brad Eppard, M., Brown, R. S., & Nielsen, J. L. (2011). Advancing the surgical implantation of electronic tags in fish: a gap analysis and research agenda based on a review of trends in intracoelomic tagging effects studies. Reviews in Fish Biology and Fisheries, 21, 127-151. https://doi.org/ 10.1007/s11160-010-9193-3 Crossin, G. T., & 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(1), 184-191. https://doi.org/10.1577/FT04-076.1 Davis, L. J., Reiter, M., & Groom, J. D. (2016). Modelling temperature change downstream of forest harvest using Newton's law of cooling. Hydrological Processes, 30(6), 959-971. https://doi.org/10.1002/hyp.10641 Dean, J. M. (1976). Temperature of tissues in freshwater fishes. Transactions of the American Fisheries Society, 105(6), 709-711. https://doi.org/10.1577/15488659(1976)105<709:TOTIFF>2.0.CO;2 Dick, M., Patterson, D. A., Robinson, K. A., Eliason, E. J., Hinch, S. G., & Cooke, S. J. (2020). Adult sockeye salmon gastrically tagged near spawning grounds exhibit lower survival rates throughout the spawning period than externally tagged conspecifics. North American Journal of Fisheries Management, 40(4), 939-951. https://doi.org/10.1002/nafm.10455 42 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., & 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, 87(6), 480-490. https://doi.org/10.1139/Z09-032 Faber, P., & Garby, L. (1995). Fat content affects heat capacity: a study in mice. Acta Physiologica Scandinavica, 153(2), 185-187. https://doi.org/10.1111/j.17481716.1995.tb09850.x Fechhelm, R. G., & Neill, W. H. (1982). Predicting body-core temperature in fish subjected to fluctuating ambient temperature. Physiological Zoology, 55(3), 229239. https://doi.org/10.1086/physzool.55.3.30157887 Gaur, A., Ratra, J. S., Nakhva, D., & Aland, P. (2019). Newton's Law of cooling and its application to fin time related temperature of body. Journal of Emerging Technologies and Innovative Research, 6(5). Gockenbach, M., & Schmidtke, K. (2009). Newton’s law of heating and the heat equation. Involve, a Journal of Mathematics, 2(4), 419-437. https://doi.org/ 10.2140/involve.2009.2.419 Graham, J. B., & Dickson, K. A. (2001). Anatomical and physiological specializations for endothermy. Fish Physiology, 19, 121-165. https://doi.org/10.1016/S15465098(01)19005-9 Hight, B. V., & Lowe, C. G. (2007). Elevated body temperatures of adult female leopard sharks, Triakis semifasciata, while aggregating in shallow nearshore embayments: Evidence for behavioral thermoregulation?. Journal of Experimental Marine Biology and Ecology, 352(1), 114-128. https://doi.org/ 10.1016/j.jembe.2007.07.021 Holland, K. N., Brill, R. W., Chang, R. K., Sibert, J. R., & Fournier, D. A. (1992). Physiological and behavioural thermoregulation in bigeye tuna (Thunnus obesus). Nature, 358(6385), 410-412. https://doi.org/10.1038/358410a0 Huey, R. B., & Berrigan, D. (2001). Temperature, demography, and ectotherm fitness. The American Naturalist, 158(2), 204-210. https://doi.org/10.1086/321314 Huey, R. B., & Stevenson, R. D. (1979). Integrating thermal physiology and ecology of ectotherms: A discussion of approaches. American Zoologist, 19(1), 357-366. https://doi.org/10.1093/icb/19.1.357 43 Jepsen, N., Thorstad, E. B., Havn, T., & Lucas, M. C. (2015). The use of external electronic tags on fish: an evaluation of tag retention and tagging effects. Animal Biotelemetry, 3(1), 1-23. https://doi.org/10.1186/s40317-015-0086-z Kitagawa, T., & Kimura, S. (2006). An alternative heat-budget model relevant to heat transfer in fishes and its practical use for detecting their physiological thermoregulation. Zoological Science, 23(12), 1065-1071. https://doi.org/ 10.2108/zsj.23.1065 Kitagawa, T., Nakata, H., Kimura, S., & Tsuji, S. (2001). Thermoconservation mechanisms inferred from peritoneal cavity temperature in free-swimming Pacific bluefin tuna Thunnus thynnus orientalis. Marine Ecology Progress Series, 220, 253-263. https://doi.org/ 10.3354/meps220253 Kubb, R. N., Spotila, J. R., & Pendergast, D. R. (1980). Mechanisms of heat transfer and time-dependent modeling of body temperatures in the largemouth bass (Micropterus salmoides). Physiological Zoology, 53(2), 222-239. https://doi.org/ 10.1086/physzool.53.2.30152585 Masser, M. P., & Neill, W. H. (1986). Routes of heat transfer in two teleosts, Ictalurus punctatus and Lepomis macrochirus. Environmental Biology of Fishes, 16(4), 321-324. https://doi.org/10.1007/BF00842988 Matondo, B. N., Benitez, J. P., Dierckx, A., Renardy, S., Rollin, X., Colson, D., Baltus, L., Romain, V. R. M., & Ovidio, M. (2021). What are the best upland river characteristics for glass eel restocking practice?. Science of the Total Environment, 784, 147042. https://doi.org/10.1016/j.scitotenv.2021.147042 Mishra, R., Militky, J., & Venkataraman, M. (2019). Nanporous materials. Nanotechnology in Textiles: Theory and Application. (pp. 311 – 353). Woodhead Publishing. Nakamura, I., Matsumoto, R., & Sato, K. (2020). Body temperature stability in the whale shark, the world's largest fish. Journal of Experimental Biology, 223(11), jeb210286. https://doi.org/10.1242/jeb.210286 Negus, M. T., & Bergstedt, R. A. (2012). Rates of intraperitoneal temperature change in lake trout implanted with archival tags. Transactions of the American Fisheries Society, 141(2), 383-391. https://doi.org/10.1080/00028487.2012.664600 Neill, W. H., Chang, R. K., & Dizon, A. E. (1976). Magnitude and ecological implications of thermal inertia in skipjack tuna, Katsuwonus pelamis (Linnaeus). Environmental Biology of Fishes, 1, 61-80. https://doi.org/10.1007/BF00761729 44 Nilsson, G. E., Crawley, N., Lunde, I. G., & Munday, P. L. (2009). Elevated temperature reduces the respiratory scope of coral reef fishes. Global Change Biology, 15(6), 1405-1412. https://doi.org/10.1111/j.1365-2486.2008.01767.x Oddi, F. J., Miguez, F. E., Ghermandi, L., Bianchi, L. O., & Garibaldi, L. A. (2019). A nonlinear mixed‐effects modeling approach for ecological data: Using temporal dynamics of vegetation moisture as an example. Ecology and Evolution, 9(18), 10225-10240. https://doi.org/10.1002/ece3.5543 Heat transfer in fish: are short excursions between habitats a thermoregulatory behaviour to exploit resources in an unfavourable thermal environment?. Journal of Experimental Biology, 218(21), 3461-3467. https://doi.org/10.1242/jeb.126466 Pettit, M. J., & Beitinger, T. L. (1980). Thermal Responses of the South American Lungfish, Lepidosiren paradoxa. Copeia, 1980(1), 130-136. https://doi.org/ 10.2307/1444143 Pinheiro, J., & Bates, D. (2006). Mixed-effects models in S and S-PLUS. Springer New York. https://doi.org/10.1007/b98882 Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., & R Core Team. (2021). nlme: Linear and Nonlinear Mixed Effects Models. https://CRAN.Rproject.org/package=nlme Prystay, T. S., de Bruijn, R., Peiman, K. S., Hinch, S. G., Patterson, D. A., Farrell, A. P., Eliason, E. J., & Cooke, S. J. (2020). Cardiac performance of free-swimming wild sockeye salmon during the reproductive period. Integrative Organismal Biology, 2(1), obz031. https://doi.org/10.1093/iob/obz031 Raby, G. D., Johnson, T. B., Kessel, S. T., Stewart, T. J., & Fisk, A. T. (2017). A field test of the use of pop‐off data storage tags in freshwater fishes. Journal of Fish Biology, 91(6), 1623-1641. https://doi.org/10.1111/jfb.13476 Raby, G. D., Vandergoot, C. S., Hayden, T. A., Faust, M. D., Kraus, R. T., Dettmers, J. M., Cooke, S. J., Zhao, Y., Fisk, A. T., & Krueger, C. C. (2018). Does behavioural thermoregulation underlie seasonal movements in Lake Erie walleye?. Canadian Journal of Fisheries and Aquatic Sciences, 75(3), 488-496. https://doi.org/10.1139/cjfas-2017-014 R Core Team (2021). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.Rproject.org/ 45 Reid, C. H., Vandergoot, C. S., Midwood, J. D., Stevens, E. D., Bowker, J., & Cooke, S. J. (2019). On the electroimmobilization of fishes for research and practice: opportunities, challenges, and research needs. Fisheries, 44(12), 576-585. https://doi.org/10.1002/fsh.10307 Reynolds, W. W. (1979). Perspective and introduction to the symposium: thermoregulation in ectotherms. American Zoologist, 19(1), 193-194. https:// doi.org/10.1093/icb/19.1.193 Reynolds, W. W., & Casterlin, M. E. (1978). Estimation of cardiac output and stroke volume from thermal equilibration and heartbeat rates in fish. Hydrobiologia, 57(1), 49-52. https://doi.org/10.1007/BF00018625 Rezende, E. L., Castañeda, L. E., & Santos, M. (2014). Tolerance landscapes in thermal ecology. Functional Ecology, 28(4), 799-809. https://doi.org/10.1111/13652435.12268 Royer, M., Meyer, C., Royer, J., Maloney, K., Cardona, E., Blandino, C., Da Silva, G. F., Whittingham, K., & Holland, K. N. (2023). “Breath holding” as a thermoregulation strategy in the deep-diving scalloped hammerhead shark. Science, 380(6645), 651-655. https://doi.org/10.1126/science.add4445 Schielzeth, H. (2010). Simple means to improve the interpretability of regression coefficients. Methods in Ecology and Evolution, 1(2), 103-113. https://doi.org/ 10.1126/science.add4445 Schreck, C. B., & P. B. Moyle, P. B. (Ed.). (1990). Methods for Fish Biology. American Fisheries Society. https://doi.org/10.47886/9780913235584 Seebacher, F., & Grigg, G. (2001). Changes in heart rate are important for thermoregulation in the varanid lizard Varanus varius. Journal of Comparative Physiology B, 171, 395-400. https://doi.org/10.1007/s003600100188 Shortreed, K. R. S., & Morton, K. F. (2000). An assessment of the limnological status and productive capacity of Babine Lake, 25 years after the inception of the Babine Lake Development Project. Canadian Technical Report of Fisheries and Aquatic Sciences, 2316. Sidebotham, G. (2015). Modes of Heat Transfer: Lumped Capacity Systems and Overall Heat Transfer Coefficients. Heat Transfer Modeling: An Inductive Approach (pp. 31-60). SpringerLink: Springer International Publishing. 46 Speers-Roesch, B., Norin, T., Driedzic, W.R. (2018). The benefit of being still: Energy savings during winter dormancy in fish come from inactivity and the cold, not from metabolic rate depression. Proceedings of the Royal Society B, 285(1886), 110. https://doi.org/10.1098/rspb.2018.1593 Spigarelli, S. A., Thommes, M. M., & Beitinger, T. L. (1977). The influence of body weight on heating and cooling of selected Lake Michigan fishes. Comparative Biochemistry and Physiology Part A: Physiology, 56(1), 51-57. https://doi.org/10. 1016/0300-9629(77)90441-8 Stevens, E. D., & Fry, F. E. J. (1974). Heat transfer and body temperatures in nonthermoregulatory teleosts. Canadian Journal of Zoology, 52(9), 1137-1143. https://doi.org/10.1139/z74-152 Stevens, E. D., Lam, H. M., & Kendall, J. (1974). Vascular anatomy of the countercurrent heat exchanger of skipjack tuna. Journal of Experimental Biology, 61(1), 145-153. https://doi.org/10.1242/jeb.61.1.145 Stevens, E. D., & Sutterlin, A. M. (1976). Heat transfer between fish and ambient water. Journal of Experimental Biology, 65(1), 131-145. https://doi.org/10.1242/ jeb.65.1.131 Symonds, M. R., & Moussalli, A. (2011). A brief guide to model selection, multimodel inference and model averaging in behavioural ecology using Akaike’s information criterion. Behavioral Ecology and Sociobiology, 65, 13-21. https://doi.org/10.1007/s00265-010-1037-6 Thorstad, E. B., Rikardsen, A. H., Alp, A., & Økland, F. (2013). The use of electronic tags in fish research–an overview of fish telemetry methods. Turkish Journal of Fisheries and Aquatic Sciences, 13(5), 881-896. https://doi.org/10.4194/13032712-v13_5_13 Treer, T., & Babačić, H. (2017). The relationship between condition and form factors of the Adriatic fishes in the Zadar area. Croatian Journal of Fisheries, 75(4), 153155. https://doi.org/10.1515/cjf-2017-0019 Trudel, M., & Welch, D. W. (2005). Modeling the oxygen consumption rates in Pacific salmon and steelhead: model development. Transactions of the American Fisheries Society, 134(6), 1542-1561. https://doi.org/10.1577/T04-156.1 Volkoff, H., & Rønnestad, I. (2020). Effects of temperature on feeding and digestive processes in fish. Temperature, 7(4), 307-320. https://doi.org/10.1080/ 23328940.2020.1765950 47 Weller, D. E., Anderson, D. J., DeAngelis, D. L., & Coutant, C. C. (1984). Rates of heat exchange in largemouth bass: experiment and model. Physiological Zoology, 57(4), 413-427. https://doi.org/10.1086/physzool.57.4.30163343 Zuur, A. F., leno, E. N., & Elphick, C. S. (2010). A protocol for data exploration to avoid common statistical problems. Methods in Ecology and Evolution, 1(1), 3-14. https://doi.org/10.1111/j.2041-210X.2009.00001.x 48 CHAPTER TWO Thermal Preference and Behavioural Thermoregulation of Sockeye Salmon Introduction Behavioural thermoregulation is the process of regulating body temperature through behaviour (Reynolds, 1979). In environments with temperature gradients (or thermal heterogeneity), animals can behaviourally thermoregulate by moving to areas that are at their preferred temperature – the body temperature an animal aims to achieve when there are no ecological constraints (Reynolds, 1979; Ward et al., 2010). The thermal preference of an ectotherm is believed to be slightly cooler than the temperature that maximizes its fitness (Martin & Huey, 2008). These temperatures are still beneficial to fitness while minimizing the chance of exposure to unfavourably warm temperatures (which typically impact fitness more adversely than cooler temperatures differing similarly from optimal temperatures) due to imperfect thermoregulation (Martin & Huey, 2008). Additionally, behavioural thermoregulation can help organisms maintain more stable body temperatures in an environment with heterogeneous or fluctuating temperatures (Angilletta Jr, 2009; Goller et al., 2014). Behavioural thermoregulation has been reported in a wide range of taxa, from insects to mammals (e.g. Hanya et al., 2007; Honek & Martinkova, 2019), however it is most commonly associated with ectothermic animals (reptiles, invertebrates, amphibians, and fish) whose body temperature is largely dictated by their environment’s temperature 49 (Bicego et al., 2007). Well-known examples of behavioural thermoregulation include turtles sun basking on logs or cold-water fish species seeking out cool thermal refuges in lakes and rivers (e.g. Bertolo et al., 2011; Edwards & Blouin-Demers, 2007). Many freshwater, saltwater, and diadromous fish species exhibit thermoregulatory behaviour (e.g., Campana et al., 2011; Nordahl et al., 2018; Ward et al., 2010). When and why fishes behaviourally thermoregulate in the wild is not always clear, however some potential relationships between this behaviour and physiological and life-history characteristics have been identified by research. Thermoregulatory behaviour in some fishes is believed to be driven by rates of digestion and metabolism. Warmer waters can help the efficiency of digestion whereas cooler waters can slow metabolic rates and the usage of energy stores (Kitigawa et al., 2016; Papastamatiou et al., 2015; Roscoe et al., 2010). Blacktip Reef Sharks (Carcharhinus melanopterus), for example, move to shallow warmer waters during the day, likely to speed up digestion before evening feeding periods in cooler waters (Papastamatiou et al., 2015). Roscoe et al. (2010) found that adult female Sockeye Salmon (Oncorhynchus nerka) passing through the Anderson Watershed behaviourally thermoregulated to presumably reduce metabolic rates. Individuals with lower energy reserves occupied cooler water temperatures than those with larger reserves, and likely did so to preserve their limited amount of energy (Roscoe et al., 2010). It has also been hypothesized that fish behaviourally thermoregulate to manage reproductive maturation, gestation duration, and growth rates (Armstrong et al., 2013; Hight & Lowe, 2007; Newell & Quinn, 2005; Roscoe et al., 2010). Research on adult 50 Sockeye Salmon in Lake Washington showed that they avoided cold waters earlier in their migration (which could delay maturation) and occupied warmer waters as spawning season approached (Newell & Quinn, 2005). Similarly, Hight and Lowe (2007) observed that sexually mature female Leopard Sharks (Triakis semifasciata) tended to aggregate in warm shallow waters and hypothesized that this was to shorten their gestation periods. In juvenile fishes, thermoregulatory behaviour has been linked to faster growth rates. For example, Armstrong et al. (2013) found that the Juvenile Coho Salmon (Oncorhynchus kisutch) in Bear Creek, Alaska, behaviourally thermoregulated by following the shifting locations of warm waters patches throughout the creek to enhance their growth rate. Research has also suggested that fishes behaviourally thermoregulate to reduce development rates of disease caused by pathogens. Adult Sockeye Salmon in the Harrison River watershed likely behaviourally thermoregulate by remaining in areas of cool water in Harrison Lake while waiting to spawn, which may delay disease development that would be accelerated by warmer waters near the spawning grounds (Mathes et al., 2010). Similarly, migrating adult Cultus Lake Sockeye Salmon that held in the warmer waters of the Fraser River for longer periods of time before entering the cooler Cultus Lake had more severe infections of the parasite Parvicapsula minibicornis than the adults that entered the Lake earlier (Bradford et al., 2010). There is also evidence that warm water fishes, such as the Trinidadian Guppy (Poecilia reticulata), behaviourally thermoregulate by spending time in water that is hotter than the thermal tolerance of their parasites to prevent their development (Mohammed et al., 2016). 51 Furthermore, it has been reported that some cold-water freshwater fish species, such as Brook Trout (Salvelinus fontinalis) and Rainbow Trout (Oncorhynchus mykiss), behaviourally thermoregulate to avoid potentially lethal water temperatures or to recover from stress caused by warm temperatures. In these cases, behavioural thermoregulation is expressed by seeking out cool water thermal refuges or localized areas of cool water such as groundwater upwellings and springs (Baird & Krueger, 2003; Bertolo et al., 2010; Corey et al., 2019). Similar thermoregulatory behaviour is exhibited by adult Sockeye Salmon when they pass through lakes after migrating through warm and thermally stressful river waters, where they frequently descend to cooler, deeper waters (e.g. Katinic et al., 2015; Mathes et al., 2010; Newell & Quinn, 2005). Despite a large body of evidence showing that fishes often exhibit behaviours that presumably enable them to regulate body temperatures, not all fishes exhibit thermoregulatory behaviour when given the opportunity (Clark et al., 2022; Keefer et al., 2015). It has been hypothesized that ectotherms may behaviourally thermoregulate more efficiently when the benefits of thermoregulatory behaviour outweigh its costs (Huey & Slatkin, 1976). The benefits of behavioural thermoregulation, as discussed above, may increase when the thermal quality of the environment is low since the chance of randomly experiencing favourable temperatures decreases (Blouin-Demers & Nadeau, 2005). The costs of behavioural thermoregulation for fishes can include an increased risk of harvest or predation (Keefer et al., 2009; Nay et al., 2021), excess energy expenditure (Donaldson, 2008; Keffer et al., 2015), and missed foraging opportunities (Biro, 1998). 52 Temperature is only one of many factors that can influence the location of an ectotherm in its environment, and it can be difficult to tell when they are indeed behaviourally thermoregulating (Angilletta Jr., 2009; Clark et al., 2022). For example, Raby et al., (2018) had difficulty determining if Walleye (Sander vitreus) migrations between basins in Lake Erie were driven by behavioral thermoregulation or by prey distribution (or both). Vaudo and Heithaus (2013) found that microhabitat selection by juvenile Pink Whiprays (Himantura fai), Reticulate Whiprays (Himantura uarnak), and Shovelnose Rays (Glaucostegus typus) in Shark Bay, Australia, was most likely driven by shark predation despite resembling thermoregulatory behaviour. One of the simplest ways behavioural thermoregulation is measured in the literature is by comparing the body temperatures or ambient temperatures an ectotherm experiences to the available temperatures in their environment (Angilletta Jr., 2009). Keefer et al., (2018) used this method to measure the thermoregulatory behaviour exhibited by Columbia and Snake River Chinook Salmon (Oncorhynchus tshawytscha) and Steelhead Salmon (Oncorhynchus mykiss). They equated behavioural thermoregulation to differences between recorded fish body temperatures and the day’s average river temperature and found that the body temperatures of salmon migrating later in the year (coinciding with warmer river temperatures) were commonly cooler than the average river temperature by 2°C - 10°C, indicating behavioural thermoregulation via the use of cool water refuges. An issue with this method is that it does not account for the thermal preference of the study organism, making it more difficult to determine if exhibited behaviours are 53 thermoregulatory or caused by something else (Angilletta Jr., 2009). A more definitive way of measuring thermoregulatory behaviour is to compare body and available ambient temperatures to a lab-estimated thermal preference of the study species (ideally estimated for the same population and time of year of study) (Angilletta Jr., 2009; Row & BlouinDemers, 2006). Using the following equation, these three parameters can be compared to estimate if and how effectively an ectotherm behaviourally thermoregulates (Hertz et al., 1993): Equation 2.1 ‫= ܧ‬1− തതതത ܾ݀ തതത ݀݁ where ‫ ܧ‬represents the effectiveness of behavioural thermoregulation and ranges from −∞ to 1, with 0 representing a lack of behavioural thermoregulation (thermoconforming), 1 representing perfect thermoregulation, and negative values തതതത represents the representing avoidance of preferable temperatures (Hertz et al., 1993). ܾ݀ accuracy of thermoregulation and is the average absolute difference between the thermal preference and the body temperatures experienced by the study animal (Hertz et al., 1993). തതത ݀݁ represents the thermal quality of the environment and is the average absolute difference between the thermal preference of the study animal and the operative temperatures (Hertz et al., 1993). The operative temperature is a simplified measurement of the thermal environment and can be defined as the expected body temperature of the study organism in the study environment while at a steady-state and experiencing no heat 54 gain from metabolism or heat loss due to evaporation (Angilletta Jr., 2009; Bakken, 1992; Hertz et al., 1993). Despite numerous publications exploring the effectiveness of thermoregulatory behaviour of reptiles, to my knowledge, it has yet to be investigated in fish (apart from current research projects in the Freshwater Fish Ecology Lab at UNBC). Research on the behavioural thermoregulation of fishes has been limited to comparing environmental and body temperatures, where fish spending greater durations of time away from the average environmental temperature or experiencing temperatures deviating further from the average temperature are considered active thermoregulators (e.g., Armstrong et al., 2013; Campana et al., 2011; Keefer et al., 2018; Newell & Quinn, 2005). A recent review of published studies investigating the thermoregulatory behaviour of freshwater fishes found 77 relevant publications, most of which were on members of the Salmonidae family (Amat-Trigo et al., 2023). Behavioural thermoregulation of Salmonidae fishes is likely a popular research topic both because they are cold-water fish species, generally preferring cool waters, and ectothermic putting them at risk to climate change and warming water temperatures (Amat-Trigo et al., 2023; Myers et al., 2017; Mugwanya et al., 2022). Salmonids that are more effective thermoregulators could be more resilient to climate change–driven reduction of thermal habitat quality and increased temperature variability (Amat‐Trigo et al., 2023; Sunday et al., 2014). Knowing a fish’s thermal preference and how often they seek out and occupy these temperatures (i.e. behavioural thermoregulation), as well as where thermally favourable areas occur in 55 the environment can help researchers and managers understand the importance of those areas and what areas should be a priority for protection. The lakes that adult Sockeye Salmon pass through on their way to spawning grounds likely offer important thermoregulatory opportunities (Cox-Rogers et al., 2012; Mathes et al., 2010; Newell & Quinn, 2005). For example, Newell and Quinn (2005) found that adult Sockeye Salmon remained in Lake Washington for an average of 83 days during their upriver migration to spawning grounds, likely behaviourally thermoregulating by using the deeper, cooler lake water as a refuge from warmer river temperatures. Mathes et al. (2010) found that when the entrance to their spawning tributary was impassible due to drought, some Weaver Creek Sockeye Salmon moved upriver to cooler waters in Harrison Lake and remained there until the tributary was accessible again (indicating they used the lake to behaviourally thermoregulate). Despite the evident importance of lakes to adult anadromous salmonids, there is a lack of understanding of how they use and interact with lake habitats (Lennox et al., 2021). Additionally, there is still uncertainty regarding what (if any) physiological or genetic factors drive an individual salmon to behaviourally thermoregulate in lakes and if effective behavioural thermoregulation ultimately impacts fitness (Lennox et al., 2021). The primary goal of this chapter was to investigate if female adult Sockeye Salmon exhibited thermoregulatory behaviour while passing through Babine Lake, and if so, how effectively they behaviourally thermoregulated. Another goal of this chapter was to examine if there was a relationship between various physiological parameters of the fish and the effectiveness of behavioural thermoregulation. The parameters of focus were 56 the fish’s pathogen load, energy reserves, sexual maturity, and indicators of thermal stress since these factors are temperature dependent and have been hypothesized to impact the thermoregulatory behaviour of Sockeye Salmon (Larson et al., 2016; Mathes et al., 2010; Newell & Quinn, 2005; Roscoe et al., 2010). Since estimating the effectiveness of behavioural thermoregulation requires a measure of the study species’ (and ideally the study population’s) thermal preference, an additional goal of this chapter was to estimate the thermal preference of adult Babine Lake Sockeye Salmon in a controlled lab setting. Additionally, I investigated if effective behavioural thermoregulation impacted the spawning success of the Sockeye Salmon. 57 Methods Study Area The study area for this research was the Babine Lake Watershed located in the Skeena Region, British Columbia (BC). The primary area of interest in the watershed was Babine Lake, located north of Burns Lake, BC, and east of Smithers, BC (Figure 2.1). It is part of the Skeena River Watershed, the second most productive salmon watershed in BC (Gottesfeld et al., 2002; Wood, 2002). To reach Babine Lake, salmon migrate through the Skeena River, Babine River, Nilkitkwa Lake, and a small section of river known as Rainbow Alley for a total distance of approximately 500 km (Figure 2.1) (Hume & MacLellan, 2000). 58 Figure 2.1. Map of the Skeena River, Babine River, Nilkitkwa Lake, and Babine Lake. The small light blue portion of river between Nilkitkwa Lake and Babine Lake is known as Rainbow Alley. The river and lake shapefiles were retrieved from the BC Freshwater Atlas webpage in March 2023. The base maps are from ESRI (primary map) and OpenStreetMap (inset). 59 Babine Lake is one of the largest lakes in British Columbia (see Figure 2.2 for a map of Babine Lake and Figure 2.3 for a photo of the lake) and it is estimated that up to 90% of Skeena River Sockeye Salmon are reared in Babine Lake (Carr-Harris et al., 2015). There are over 20 spawning tributaries that flow into Babine Lake, however most Babine Lake Sockeye Salmon spawn in Pinkut Creek and Fulton River, which both have artificial spawning channels and water flow control (Figure 2.2) (Gottesfeld et al., 2002). Pierre Creek, Morrison Creek, Twain Creek, and Rainbow Alley through to the headwaters of Babine River have some of the highest numbers of returning adult Sockeye Salmon that are not reared in tributaries with artificial spawning channels in the Babine Lake Watershed (Gottesfeld et al., 2002). Babine Lake Sockeye Salmon support First Nation fisheries and are an integral part of the diet of Lake Babine Nation members (Doire & Macintyre, 2015; Lake Babine Nation, 2021). Additionally, the salmon support Canadian and United States commercial and recreational fisheries and the estimated yearly harvest of Babine Lake Sockeye Salmon totals more than one million Sockeye Salmon (Doire & Macintyre, 2015). 60 Figure 2.2. Map of the study area, including Babine River, Nilkitkwa Lake, Babine Lake, Morrison Creek, Fulton River, and Pinkut Creek. The small light blue portion of river between Nilkitkwa Lake and Babine Lake is known as Rainbow Alley. The tagging location shown on the map is also the location of the Babine River counting fence and where the thermal preference tests were conducted (See the “thermal preference tests” section below for details). The locations of the receiver station towers and the approximate locations of the temperature logger strings in 2021 are also shown (See the “Monitoring Water Temperatures in Babine Lake” and “Radio Telemetry and Fish Recovery” sections below for more details). 61 Figure 2.3. Babine Lake at the mouth of Pinkut Creek in September 2021. The red shapes in the water are adult Sockeye Salmon. Fish Capture, Sampling, and Tagging Fish were captured at the Babine River counting fence as described in the methods section of Chapter One. In addition to measuring the fork length (cm) and weight (g) of each Sockeye Salmon, their muscle fat content was measured using a Fish Fat Meter (FFM-992, Distell, UK). Two fat readings were taken from each fish, one above the lateral line 2.5 cm back from the gill plate and another above the lateral line 2.5 cm behind the location of the first reading. The average of the readings was used to estimate the gross somatic energy (GSE in units of MJ • kg-1) of each fish using an equation developed by Crossin & Hinch (2005). A 2 to 3 mm gill clip was taken from each fish and immediately placed in a vial with RNA-later, which were later analyzed in the lab to assess the individual’s relative infectious burden (RIB) and expression of 62 thermal stress. The clips were placed in a refrigerator for 24 hours, followed by being placed in a freezer (– 5°C) for up to 2 months in the field. After the field season, the samples were stored in a – 30°C freezer in the lab until processing. Twenty female Sockeye Salmon were internally and externally tagged with temperature loggers and underwent thermal preference tests, which occurred from August 18th, 2023, to October 11th, 2023 (see Chapter One and below for details). An additional 60 female Sockeye Salmon were caught and sampled between August 25th, 2023, and September 3rd, 2023 (10 each day) and were externally tagged with a uniquely coded radio tag (NTF-5-2, Lotek Wireless, Canada), and an archival temperature and depth logger (LAT1100, Lotek Wireless, Canada, temperature accuracy of ± 0.2°C and depth accuracy of ± 1 meter). The loggers recorded temperature and depth every 90 seconds. The tags and loggers were secured below the dorsal fin with pins and Peterson disks (see tags and tag placement in Figure 2.4). The fish were also externally tagged with T-tags (FD-94, Floy Tag & Mfg. Inc., USA). Total fish handling time ranged between 3 to 6 minutes. After sampling and tagging, the radio-tagged fish were released upriver of the counting fence to continue their migration towards and through Babine Lake (Figure 2.4). All fish handling, sampling, tagging and thermal preference trials were approved by the UNBC Animal Care and Use Committee and Fisheries and Oceans Canada (UNBC protocol number: 2021-05, DFO Licence Number: XR 237 2021). 63 Figure 2.4. A radio-tagged Sockeye Salmon in the Babine River after tagging on September 1st, 2021. Thermal Preference Tests Thermal preferences of Babine Lake Sockeye Salmon were estimated using a shuttle box system (Loligo Systems, Denmark). The shuttle box is an automated system that allows the test organism to control the temperature of the water it occupies and regulate its body temperature (Christensen et al., 2021). The system used in my study comprised two circular tanks (diameter: 1 meter) connected by a 25cm long open channel and covered by raised clear lids (to prevent fish from jumping out of the tanks). There was a cool tank and warm tank that were set to consistently differ by 2°C (Figure 2.5). The cool tank was set to 1°C colder than the river temperature at capture, and the warm tank was set to 1°C warmer than the river temperature at capture. The starting water 64 temperatures ranged from 6.56°C to 17.04°C and test start times ranged from 09:48 to 19:54. Before beginning each trial, the shuttle box was filled with approximately 325 litres of river water from the Babine River, and which tank was the cool and warm tank and what tank the fish began their trial in was decided by an online randomizer. The test began 30 minutes after the fish were placed in the system, allowing them to recover from sampling and tagging. During the tests, the fish could move freely between the tanks through the connecting channel and its location was monitored by a camera connected to the shuttle box software (Shuttlesoft, Loligo Systems, Denmark) on a computer. Temperature probes connected to the shuttle box system constantly monitored the temperatures in each tank. If the fish was in the cool tank, both tanks cooled at a rate of 2°C / hour, and if the fish was in the warm tank, both tanks warmed at a rate of 2°C / hour (Figure 2.5). Each completed thermal preference test lasted for 24 hours, except for one trial that lasted only 18 hours due to logger power supply issues. The shuttle box system was set up next to the Babine River counting fence inside a ShelterLogic Waterproof Portable Garage-In-A-Box to protect the system and prevent external light from disrupting test results. 65 Figure 2.5. Simplified schematic of the shuttle box. The camera tracks the location of this fish, informing the temperature control system if the fish is in the cool (blue) or warm (red) side of the shuttle box. The temperature control system warms both sides of the shuttle box if the fish is in the warm side (panel A) and cools both sides if the fish is in the cool side (panel B). 66 Monitoring Water Temperatures in Babine Lake In 2021, temperatures in Babine Lake were monitored by an array of 9 temperature logger strings operated by Fisheries and Oceans Canada (Figure 2.2). Throughout the study period, the strings recorded temperature from approximately 5 meters below the surface to the bottom of the lake. Depending on the logger and depth range, temperatures were recorded every 30 minutes or 60 minutes, as frequently as every 1.5 meters to as infrequently as every 50 meters deep (when measuring temperatures below 100 m). Radio Telemetry and Fish Recovery Radio receiver tower stations consisting of a receiver unit (SRX 1200, Lotek Wireless, Canada), power source, and an antenna raised approximately 2.5 meters above the receiver were set up 1 km downstream from the outlet of Babine Lake in Rainbow Alley, 50 meters upstream from the mouth of Pinkut Creek at the Pinkut River enumeration fence, and 1.5 km upriver from the mouth of Fulton River at the Fulton River Enumeration Fence (Figure 2.2). The receiver stations detected the radio tags, recording when the fish arrived at the outlet of Babine Lake and when they entered Pinkut Creek or Fulton River. Pinkut Creek and Fulton River were chosen as receiver station locations since the tagged fish were presumably part of the “mid-timing run” (comprising Fulton River, Pinkut Creek, Morrison Creek, Tahlo Creek, or upper Tahlo Creek populations), in which the majority of the fish migrate to Fulton River and Pinkut Creek (Gottesfeld et al., 2002). A receiver location was also intended to be set up near the 67 mouth of Morrison River, which would have detected tagged Sockeye Salmon migrating to Morrison River, Tahlo Creek, and Upper Tahlo Creek to spawn, however due to damaged equipment, this station was not established. Using the detections from the receiver stations and mobile radio telemetry unit (SRX 1200, Lotek Wireless, Canada), I located tagged Sockeye Salmon in the Fulton, Pinkut, and Morrison tributaries (Figure 2.2). I also used the mobile telemetry unit to check for tagged fish in some of the “early-run timing” spawning tributaries (Boucher Creek, Tsezakwa Creek, Five-mile Creek, and Nine-mile Creek), however no tags were detected. I was unable to check all possible spawning tributaries due to limited tributary accessibility, time, and resources. 68 Figure 2.6. Volunteer Spencer Smith using the mobile radio tracking unit (black box, wire, and antenna visible) to locate a radio-tagged fish in Fulton River in September 2021. Twenty-four spawned-out tagged Sockeye Salmon were recovered from spawning channels and rivers. Twenty-two of these fish were recovered from Fulton River, and two were recovered from Morrison River. One of the recovered fish from Fulton River was male, so the data from this fish was excluded from the analyses. Three and seven additional tagged fish were detected by the receiver stations at the mouth of Fulton River and Pinkut Creek, respectively, however they were not recovered. Three additional tagged fish appeared to spawn or die in Rainbow Alley, but their tags were not recovered. The locations where each dead tagged fish was found were recorded, and a picture of the 69 fish was taken. If the belly of the fish was intact (not disturbed by predators and not deteriorated), the approximate percentage of eggs remaining in the body was recorded as an indication of spawning success. The tags and loggers were removed from the fish, and the fish was returned to the location it was found. Gill Clip Processing The gill clip samples taken from the tagged Sockeye Salmon that were found post-spawn and those that underwent thermal preference tests were analyzed by the Molecular Genetics Lab, Pacific Biological Station, Nanaimo, BC, using the Salmon FitChip. The Salmon Fit-Chip is a microfluidics RT-qPCR (quantitative reverse transcription polymerase chain reaction) chip that when run using the Fluidigm BiomarkTm Platform allows researchers to assess the expression of up to 96 genes in 96 different tissue samples at the same time (Akbarzadeh et al., 2018; Fisheries and Oceans Canada, 2018; Miller et al., 2016). The chip contains biomarker panels for known types of stressors Sockeye Salmon experience (such as thermal, osmotic, and hypoxic stress). The panels include six to 12 biomarkers (host genes) for each stressor that, when expressed sufficiently, reveal and quantify the stressors experienced by a fish (Houde et al., 2019; communications with the Molecular Genetics Lab, Pacific Biological Station). In this study, I was interested in using the Salmon Fit-Chip to assess the infectious (pathogen) burden of the study fish and if they had experienced thermal stress during their upriver migration. RNA expression of 18 infectious agents (i.e. pathogens) known to infect Pacific Salmon were tested on the chip (Miller et al., 2016). The number of 70 RNA copies of each infectious agent in each sample were calculated by members of the Molecular Genetics Lab using the standard curve method (Bass et al., 2019; Ginzinger, 2002). The stressor biomarker panels were run as singletons (one copy of each biomarker) and the pathogen assays were run as duplicates. The host gene (stressor biomarker) expression results were normalized to the MRS_HTA_272 and MRS_FYB_241 genes. To evaluate the pathogen load of each Sockeye Salmon, their relative infectious burden (RIB) was calculated using the estimated number of RNA copies of each of the tested infectious agents (Bass et al., 2019): Equation 2.2 ௠ ܴ‫ = ܤܫ‬෍ ‫ܮ‬௜ ) ௜ୀଵ ‫ݔܽ݉ܮ‬௜ ( where ‫ܮ‬௜ is the load (i.e. number of RNA copies) of the ݅th infectious agent, ‫ݔܽ݉ܮ‬௜ is the load of the fish with the highest load within the population that tested positive for the ݅th infectious agent. This was summed across all infectious agents (݉) infecting the given individual. For the heat stress variable, the Molecular Genetics Lab classified the normalized biomarker panel results on a scale of 0 to 1 for each fish, where values closer to 1 indicated increased stress and values closer to 0 indicate little to no stress. The assigned values were based on pre-established values of biomarker expression indicating the extent to which the salmon has experienced heat stress over the last few days (Akbarzadeh et al., 2020; Akbarzadeh et al., 2021; Houde et al., 2019). It is important to 71 consider that for this variable, values below the lab-specific threshold of 0.5 may not be significantly different from each other. However, the relationship between the heat stress score for each fish and the river temperature at the time of their capture was strongly and linearly related, indicating that treating the heat stress score as a linear variable was adequate. The gill clips were also tested for other stressors including hypoxic stress (63% of tested fish exhibited notable hypoxic stress), viral disease development (67% of tested fish were likely in an active disease state), and “imminent mortality” (likely experienced death within 72 hours of sampling – as expected, 0% of tested fish likely died within 72 hours). These additional stressors were not included in this thesis for simplicity purposes, however they may be explored in the future. Thermal Preference Analysis Thermal preference was defined as the central 50% (the interquartile range, Q25 to Q75) of body temperatures experienced by the Sockeye Salmon in the shuttle box tests. A range was used instead of a singular median or mean value because some ectotherms regulate their body temperatures between lower and upper range values instead of a singular temperature (Hertz et al., 1993). The ranges, referred to as the “setpoint range” of the animal, usually include the central 50% or 80% of the body temperatures recorded in a thermal preference test (Angilletta Jr., 2009). Thermal preference was calculated for each hour of a fish’s shuttle box test. The data from two of the completed shuttle box tests were excluded from the analysis. Shuttle box test 12 was 72 excluded due to the low concentration of RNA extracted from the fish’s gill tissue sample (not enough to conduct the genomic analysis with, see Appendix Figure A.3 for this shuttle box test data), and shuttle box test 13 was excluded due to the fish staying in the cool tank consistently after the first few hours of the test (See Appendix Figure A.4). Additionally, the first hour of data from the thermal preference tests was excluded from the analyses due to the temperatures selected during this time being strongly, positively associated with the starting shuttle box water temperatures. This relationship became more moderate in test hour two and onwards. The relationship between thermal preference ranges (Q25 and Q75) and fixed effects of interest were investigated using Bayesian generalized linear multivariate mixed models. The five fixed effects tested were fish mass, GSE, heat stress, RIB, and day/night, a categorical variable where night was classified as the time between sunset and sunrise. All the fixed effects (excluding day/night) were standardized prior to modelling by subtracting the sample mean and dividing by the sample standard deviation. Therefore, the global model for thermal preference was: Equation 2.3 ܶ‫݂݁ݎ݌‬௜,௛(ொଶହ| ொ଻ହ) = ൫ߙ(ொଶହ| ொ଻ହ) + ߛ௜(ொଶହ| ொ଻ହ) ൯ + ߚ௛௘௔௧(ொଶହ| ொ଻ହ) ‫ݔ‬௛௘௔௧೔ + ߚோூ஻(ொଶହ| ொ଻ହ) ‫ݔ‬ோூ஻೔ + ߚ௠௔௦௦(ொଶହ| ொ଻ହ)‫ݔ‬௠௔௦௦೔ + ߚ௡௜௚௛௧(ொଶହ| ொ଻ହ) ‫ݔ‬௡௜௚௛௧೔,೓ + ߚீௌா(ொଶହ| ொ଻ହ) ‫ீݔ‬ௌா೔ where ܶ‫݂݁ݎ݌‬௜,௛(ொଶହ |ொ଻ହ) is the estimated thermal preference range 1st quartile (ܳ25) and 3rd quartile (ܳ75) of individual ݅ for hour ℎ of the thermal preference test, 73 ߙ൫ܳ25หܳ75൯ are the intercepts of ܳ25 and ܳ75; ߛ௜൫ܳ25หܳ75൯ are the random deviations of ܳ25 and ܳ75 from their respective intercepts, both assumed to follow Normal൫0, ߪఊ ൯, ߚ௛௘௔௧(ொଶହ| ொ଻ହ) are the slopes for the effect of heat stress (ℎ݁ܽ‫ )ݐ‬on ܳ25 and ܳ75; ‫ݔ‬௛௘௔௧೔ is the heat stress score of individual ݅, ߚோூ஻(ொଶହ| ொ଻ହ) are the slopes for the effect of RIB on ܳ25 and ܳ75; ‫ݔ‬ோூ஻೔ is the RIB of individual ݅, ߚ௠௔௦௦(ொଶହ| ொ଻ହ) are the slopes for the effect of mass on ܳ25 and ܳ75; ‫ݔ‬௠௔௦௦೔ is the mass of individual ݅, ߚ௡௜௚௛௧(ொଶହ| ொ଻ହ) are the effects of civil nighttime (when ‫ݔ‬௡௜௚௛௧೔ = 1 or nighttime) during hour ℎ on ܳ25 and ܳ75, ߚீௌா(ொଶହ| ொ଻ହ) are the slopes of the effect of GSE on ܳ25 and ܳ75, and ‫ீݔ‬ௌா೔ is the GSE of individual ݅. A total of 32 candidate models were fit, including an intercept only model (no fixed effects) and all possible additive combinations of the fixed effects. Fish ID was included as a random effect for the intercept in all models. The models were fit using a Bayesian approach based on Hamiltonian Monte Carlo using four chains and uninformative priors. Each chain was run for 4000 iterations, discarding the first 2000 iterations as burnin. To account for temporal autocorrelation in the residuals, an autocorrelation structure (AR1, first-order) was included in the models. The models were fit using Stan statistical software (V2.32, Stan Development Team 2023) through the brms package (V2.20.1, Bürkner 2017) in R (V4.2.3, R Core Team, 2022). Convergence of the chains was determined by checking that the R-hat statistic for all models/model parameters was < 1.1 (Muth et al., 2018). The effective sample sizes for all model parameters were greater than 1000, indicating enough samples from posterior 74 distributions to make reliable parameter estimates (Muth et al., 2018). Posterior predictive checks were completed for all models, and samples simulated from the posterior predictive distribution closely resembled the observed data. The predictive performance of the candidate models was estimated and compared by calculating the theoretical expected log pointwise predictive density (elpd) of each model (through efficient approximate leave-one-out (LOO) cross-validation and calculation of the LOO Information Criterion, LOOIC) (Vehtari et al., 2017; Vehtari, 2020). The models were ranked by their elpd values, where the best performing model had the highest density and the differences between the best model's elpd value and the elpd values for all other models were calculated (Silvua et al., 2022; Vehtari, 2020). Twenty-two of the models had differences of less than four, indicating that the difference in predictive power between these models was small (Silvua et al., 2022; Vehtari, 2020). Bayesian stacking weights were estimated for the candidate models by stacking their predictive distributions (Yao et al., 2018). The relative variable importance of each fixed effect was calculated as the sum of stacking weights of all models in the full candidate model set containing the fixed effect (Burnham & Anderson, 2004). To estimate the thermal preference ranges of the radio-tagged Sockeye Salmon, the models were run with the radio-tagged Sockeye Salmon (standardized) measurements, and the posterior predictive draws from the models were averaged based on their stacking weights. Bayesian stacking identifies the combination of candidate models (and stacking weights of these models) that optimizes model-averaged posterior predictive draws (Höge et al., 75 2020). As such, the predictions of models with stacking weights greater than zero were included in the thermal preference estimates. Effectiveness of Behavioural Thermoregulation Estimations A variation of the Hertz et al. (1993) equation of effectiveness of behavioural thermoregulation (Equation 2.1) was used to calculate the daily and nightly effectiveness of behavioural thermoregulation of each tagged fish (Blouin-Demers & Weatherhead, 2001) for every day they spent in Babine Lake: Equation 2.4 ‫ܧ‬௜,ௗ,௧ = തതത௜,ௗ,௛ − ܾ݀ തതതത௜,ௗ,௛ ) ∑௜,ௗ,௧(݀݁ ݊ௗ௘ തതതത೔,೏,೟ where ‫ ܧ‬is the average effectiveness of behavioural thermoregulation of individual ݅ on day ݀ during the time of day ‫( ݐ‬day or night). Equation 2.4 calculates ‫ ܧ‬on a continuous scale (from negative to positive infinity instead of being capped at one like Equation 2.1), where negative values indicate that the animal is avoiding their preferred temperature, zero representing an animal that is thermoconforming to the habitat, and positive values representing an animal that is thermoregulating. A higher E score indicates more effective behavioural thermoregulation, which occurs when തതത ܾ݀ത is low and തതത ݀݁ is high. The variable തതത ݀݁ is the thermal habitat quality (i.e. average absolute difference between a fish’s thermal preference and thermal habitat availability in Babine Lake) for each date ݀ തതതത is the accuracy of and hour of day ℎ for individual ݅ in Babine Lake. The variable ܾ݀ thermoregulation (i.e. average absolute difference between a fish’s thermal preference 76 and the temperatures experienced in Babine Lake) of individual ݅ in Babine Lake for the തതത – ܾ݀ തതതത for individual ݅ that occurred during date ݀ and date ݀ and hour ℎ. Values of ݀݁ time of day ‫ ݐ‬were summed and divided by the number ݊ of തതത ݀݁ (or തതതത ܾ݀) estimates for individual ݅ that occurred during date ݀ and time of day ‫ݐ‬. തതത) of Babine Lake, I first needed to To estimate the thermal habitat quality (݀݁ estimate the temperatures in Babine Lake at regular depth intervals for the duration of the study (i.e. every one meter) and then use these temperature by depth estimates along with estimates of water volume by depth to compute the weighted average temperature of the lake, where the weights were given by the volume of water at a given temperature. To estimate temperature by depth over the study, I used a generalized additive mixed model (GAMM) to model how temperature changed with depth and time in Babine Lake. The temperature and depth recordings taken by the nine temperature logger strings operated by Fisheries and Oceans Canada (described above in the “Monitoring Water Temperature in Babine Lake” section) were inputted into the model. The data from the second most northern temperature logger (Figure 2.2) was excluded from the model due to it recording some unexpectedly cool water temperatures around the 5-meter mark (cooler than temperatures recorded at 10 meters) that influenced predictions and resulted in non-sense patterns in the GAMM. The model fixed effects were depth (meters), day of the year, and the nearest hour of the temperature recording, and station ID was included as a random effect. A GAMM was used since it can account for non-linear relationships, such as the non-linear 77 relationship between temperature and depth in stratified lakes. A gamma distribution with log-link function was used in the model because the temperature data was continuous and always greater than zero, and because of the right-skewed distribution of the Babine Lake temperature data. Thin plate regression splines were used as smoothers for the depth and day of the year fixed effects, and a cyclic cubic regression spline was used as the smoother for hour (due to its cyclical nature) (Yang & Moyer, 2020). The model was fit with REML to avoid over-smoothing (Wood 2006) using the mgcv package (V1.9-0, Wood 2023) in R (V4.2.3, R Core Team, 2022). The model was used to predict a single hourly temperature for every meter deep between 0 and 55 meters (the deepest depth the tagged fish experienced), for every hour of the day (0 to 23), and for each day of the year between the first day a tagged fish entered Babine Lake (August 28th, 2021) and the last day a tagged fish exited Babine Lake (September 27th, 2021). To estimate water volume by temperature in Babine Lake, a historical (1964) bathymetric map of Babine Lake was retrieved from the British Columbia Ministry of Environment bathymetric map query and uploaded into ArcGIS Pro (ESRI). The depth lines were traced using the polyline feature and each line was assigned the corresponding depth value on the historical map (shore and islands counted as 0 m depth lines). The polyline layer was converted to a raster layer with a cell size of 1m2 (polyline to raster tool) and cells between the depth lines were interpolated using inverse distance weighted interpolating. The total surface area of cells falling into five-meter depth intervals were summed as shown in Table 2.1, up to a maximum depth of 55 m (due to 54 meters being 78 the deepest depth experienced by tagged fish in Babine Lake). The approximate volume of water found in each depth interval was calculated as follows: Equation 2.5 ܸ௥ = ‫ܣ‬௥ × ‫ܦ‬௥ Where ܸ is the approximate water volume (m3) found in depth range ‫ݎ‬, ‫ ܣ‬is the area (m2) of the depth range ‫ݎ‬, and ‫ ܦ‬is the depth (m) of the range ‫( ݎ‬5m intervals). The proportion of the total water volume of the lake (up to 55m) found in each of the depth intervals was calculated as interval water volume divided by total water volume up to 55m deep (Table 2.1). 79 Table 2.1. The estimated area, water volume, and proportion of the total Babine Lake water volume (above 55m depth profile) found in each five-meter depth interval. Depth Interval (m) Area (108·m2) Volume (108·m3) Volume Proportion 0-5 4.65 11.6 0.15 5 - 10 3.31 8.28 0.11 10 - 15 3.24 8.10 0.10 15 - 20 3.18 7.94 0.10 20 - 25 3.12 7.79 0.10 25 - 30 2.54 6.35 0.08 30 - 35 2.47 6.18 0.08 35 - 40 2.40 5.99 0.08 40 - 45 2.36 5.91 0.08 45 - 50 1.90 4.75 0.06 50 - 55 1.87 4.67 0.06 The hourly thermal habitat quality of Babine Lake for each fish while they were in the lake was calculated with the following equation: Equation 2.6 തതത௜,ௗ,௛ = ෍ ݀݁ ௜,ௗ,௛ |ܶ‫݂݁ݎ݌‬௜ − ܶௗ,௛,௥ | × ‫ܸ݌‬௥ തതത is the estimated average thermal habitat quality for individual fish ݅ during date where ݀݁ ݀ and hour ℎ, ܶ‫ ݂݁ݎ݌‬is the estimated thermal preference range of individual ݅, ܶ is the 80 average estimated Babine Lake temperature on date ݀, hour ℎ for depth range ‫ݎ‬, and ‫ܸ݌‬௥ is the proportion of Babine Lake water volume (down to 55 m) in depth range ‫ݎ‬. This equation treats each five-meter depth interval as a microhabitat of available temperatures while accounting for the “amount” of each microhabitat available (by calculating the weighted-by-volume average thermal habitat quality of each fish for every hour of their migration). തതതത) for each recovered fish for every hour The accuracy of thermoregulation (ܾ݀ they were in Babine Lake was calculated by the following equation: Equation 2.7 തതതത ܾ݀௜,ௗ,௛ = ∑௜,ௗ,௛ |ܶ‫݂݁ݎ݌‬௜ − ܾܶ௙,ௗ,௛,௠ | ்݊௕ where തതതത ܾ݀ is the hourly average estimated accuracy of behavioural thermoregulation during day ݀, hour ℎ by individual fish ݅, ܶ‫ ݂݁ݎ݌‬is the estimated thermal preference range of individual ݅, and ܾܶ is the estimated body temperature of individual ݅ during day ݀, hour ℎ, minute ݉ (to the nearest 1.5 minutes), and ݊ is the number of body temperatures of individual ݅ estimated during day ݀, hour ℎ. The body temperatures used in Equation 2.7 were predicted in Chapter One using models of the coefficient of heat exchange of the body of Sockeye Salmon and the recorded external temperatures they experienced in Babine Lake. To ensure only predicted body temperatures while the fish was migrating through Babine Lake were used in the analyses of effectiveness of behavioural thermoregulation, it was necessary to 81 estimate each recovered fish’s timing of entry into and exit from Babine Lake. To estimate the date and time each recovered tagged fish entered Babine Lake, their rate of travel (km/hour) from their release location at the Babine Fence to the Rainbow Alley receiver station was calculated using the following equation: Equation 2.8 ܶ‫݁ݐܴܽ ݈݁ݒܽݎ‬௙ = ‫ܦ‬௥௘௟௘௔௦௘ ௧௢ ௥௘௖௘௜௩௘௥ (‫ݐܦ‬௙௜௥௦௧ ௗ௘௧௘௖௧௜௢௡ − ‫ݐܦ‬௥௘௟௘௔௦௘ ) Where ݅ is the ID of the fish, ‫ ܦ‬is distance in km, and ‫ ݐܦ‬is date/time (to the nearest second). This rate was used to estimate the time it took individual fish ݅ to travel from the Rainbow Alley radio receiver to the inlet of Babine Lake. This time was added to the fish’s last detection timestamp at the Rainbow Alley receiver to estimate when they entered Babine Lake. The exit timestamp for Sockeye Salmon recovered in the Fulton tributary was the time in which they rose to and remained shallower than five meters in depth before their first detection at the Fulton receiver station (approximately two km upstream from the mouth of the Fulton River). The Babine Lake exit timestamp for the two fish recovered in Morrison River was when they rose to and remained shallower than 5 meters deep before their recovery. Relationship Between E and Covariates Generalized linear mixed models (GLMM; Bolker et al., 2009) were used to investigate the relationship between ‫ ܧ‬and covariates that have been hypothesized to impact thermoregulatory behaviour in fishes, including their GSE, the thermal stress 82 experienced in Babine River, and their RIB. Additionally, fish mass and daytime/nighttime were included as covariates as they are believed to influence temperature seeking and selection behaviour of some fishes. GSE, thermal stress, RIB, and mass were standardized prior to modeling. Models with all combinations of covariates (fixed effects) and an intercept only model were fit for a total of 32 candidate models. Fish ID was included as a random effect for the intercept in the models. To account for the temporal autocorrelation in the residuals, an autocorrelation structure of order one was included in the models. Models were fit with a maximum likelihood framework and the model residuals were homoscedastic. However, their distribution was slightly left-skewed. Data transformations were explored to normalize the distribution, however they did not notably improve the distribution of residuals so the un-transformed data was used in the analysis. Models were compared using AIC scores and AIC weights (Burnham & Anderson, 2004). Coefficient estimates and predictions of the models with weights adding up to at least 0.95 (i.e. the 95% confidence set for the best model) were averaged (Burnham & Anderson, 2004; Symonds & Moussalli, 2011). Additionally, to determine the relative importance of each covariate, the AIC weights of the models containing each fixed effect were summed (relative variable importance) (Burnham & Anderson, 2004). Model fitting and selection were conducted using packages nlme (V3.1-160, Pinheiro & Bates) and MuMIn (V1.47.5, Bartoń) in R (V4.2.3, R Core Team, 2022). 83 E and Spawning Success To investigate if effective thermoregulation influenced spawning success, the percentage of egg retention of each recovered tagged salmon post-spawn was estimated (approximated to 0%, 25%, 50%, 75%, and 100%). Salmon whose body cavities were intact (i.e. not ripped open due to predation or disintegrated) were included in this portion of the analysis (a total of 15 fish). Nearly all of the recovered fish were classified as having a 0% egg retention rate (few or no eggs remaining). Due to the small number of individuals that retained any eggs, I did not model the relationship between E and spawning success and present only a summary of the data. 84 Results Measured Body Parameters Overall, the masses, GSEs, RIBs, and heat stress values of the Sockeye Salmon that underwent thermal preference tests were similar to the values of the recovered radiotagged fish (Figure 2.7). A notable difference between the two groups was that the maximum RIB measured in the thermal preference fish was nearly twice as large as the maximum RIB measured in the recovered radio-tagged fish, however the median RIB of both groups were similar (Figure 2.7). Nine of the 18 tested infectious agents were detected in the study fish (median of four agents detected in each fish), and 76% of all the fish had notable heat stress (heat stress values reaching and exceeding the 0.50 significance threshold, measured in 83% of the recovered radio-tagged fish and 67% of the shuttle box tested fish). The lower percentage of shuttle box tested fish expressing notable heat stress and therefore wider distribution toward low heat stress values was likely because some of the tests occurred later in September and early October when Babine River temperatures were cooler (eight tests occurring after September 15th), whereas all radio-tagged fish and the majority of the shuttle box tested fish (12 individuals) were tagged in short period between late August and early September when temperatures were warmer. 85 Figure 2.7. Boxplots of the measured masses (g), gross somatic energies (GSE, MJ • kg-1), relative infectious burdens (RIB), and levels of heat stress of the shuttle box thermal preference tested fish and the recovered radio-tagged fish. The data from the salmon whose shuttle box tests were not included in the thermal preference models (test 12 and 13) are not included. The boxplots include the variable median (horizontal bar), interquartile range (Q25 – Q75, lower and upper bounds of the coloured boxes), whiskers (lowest and highest values outside of the interquartile range and within 1.5 x the interquartile range, vertical lines), and outliers (values extending past the interquartile range by more than ± 1.5 x the interquartile range, circles). Thermal Preference Results The median body temperature experienced by Sockeye Salmon in the shuttle box tests was 12.78°C (median absolute deviation or MAD: ± 2.79°C, median interquartile range or IQR: 10.92°C – 14.6°C), with the median temperatures experienced by each fish ranging from 8.18°C to 17.77°C. The coolest body temperature recorded during the shuttle box tests was 5.42°C and the warmest was 19.76°C (not including the shuttle box 86 tests that were excluded). Fifteen of the 32 candidate thermal preference models had stacking weights greater than 0 (Table 2.2). The model with the lowest LOOIC score had the fixed effects of heat stress and day/night, however we cannot confidently say this was the best model due to the small elpd differences between the candidate models. Most of the models had elpd differences of four or lower and all standard errors of the differences were large relative to their respective elpd differences, indicating similar performances and limited differences in predictive ability between the candidate models (Table 2.2, Sivula et al., 2022; Vehtari, 2020). 87 Table 2.2. The candidate models that had stacking weights greater than 0 (15 models), the difference in expected log predictive densities between these models and the model with the highest expected log predictive density (elpd Difference, model 5), and the standard error of these differences (Difference se). Model Stackin g Weight elpd Difference Difference se 1 Intercept 0.16 -8.61 6.01 2 RIB + Day/Night 0.15 -8.46 5.54 3 Heat Stress + Mass + Day/Night 0.13 -2.71 2.26 4 RIB + GSE 0.13 -7.40 5.59 5 Heat Stress + Day/Night 0.10 0.00 0.00 6 Heat Stress 0.06 -0.08 2.20 7 Heat Stress + Mass 0.06 -2.22 3.22 8 Heat Stress + Mass + GSE 0.05 -3.53 3.29 9 Heat Stress + Mass + GSE + Day/Night 0.05 -3.86 2.42 10 Heat Stress + GSE 0.03 -2.36 2.82 11 Heat Stress + RIB 0.03 -1.25 2.60 12 Heat Stress + RIB + Mass + GSE + Day/Night 0.02 -6.10 3.00 13 RIB + Mass 0.02 -9.60 6.20 14 Day/Night 0.01 -9.20 5.70 15 Heat Stress + GSE + Day/Night 0.01 -1.30 1.30 All of the tested fixed effects were present in at least five of the 15 candidate models with stacking weights greater than 0 (Table 2.2, Figure 2.8). The Q25 intercept 88 estimates in these models ranged from 12.62°C to 12.92°C, and the Q75 intercept estimates ranged from 13.09°C to 13.42°C. Heat stress was present in ten of these models, day/night was present in seven, both mass and GSE were present in six, and RIB was present in five of the models (Table 2.2, Figure 2.8). The relative variable importance of the fixed effects (calculated based on the stacking weights) were quite similar and moderate, with heat stress having the greatest importance of 0.54, followed by day/night with an importance of 0.46. Mass and RIB both had importance of 0.34 and 0.33, and GSE had the lowest importance of 0.29. Heat stress was also the predictive parameter with the greatest effect on estimated thermal preference range (Q25, Q75), and had a relatively narrow 95% credible interval (Figure 2.8). The effect of heat stress on thermal preference was positive in each of the top models including it and between 99.9% and 100% of posterior draws in these models estimated that heat stress had a positive effect on Q25 and Q75 (Figure 2.8). RIB, Mass, GSE, and day/night had smaller effects on thermal preference when present in the top candidate models and had 95% credible intervals that included zero (Figure 2.8). The effects of RIB and Night on thermal preference were negative, and GSE and mass had slightly positive effects (Figure 2.8). Although the 95% credible interval for the estimated effect of day/night included 0, a large percentage of the posterior samples from the models with non-zero stacking weights estimated that thermal preference was slightly lower during the nighttime (69.5% to 90.8% of the draws found nighttime to have a negative effect on thermal preference range, Figure 2.8). Between 53.6% to 95.9% of the posterior draws from the models including mass as a fixed effect (and had non-zero stacking weights) estimated mass to 89 have a positive effect on thermal preference, 53.1% to 80.5% of the samples from models including GSE estimated GSE to have a positive effect, and 47.3% to 88.8% of the posterior draws from models including RIB estimated RIB to have a negative effect on thermal preference. Figure 2.8. The mean effect size of the fixed parameters included in 15 thermal preference models with non-zero stacking weights. The effect sizes on both the lower (Q25) and upper (Q75) bounds of the thermal preference ranges are included. Horizontal orange and blue lines represent the 95% credible intervals of the fixed effect estimates. Vertical gray lines correspond to an effect size of zero. The overall median thermal preference range (Q25 to Q75) of the shuttle box fish fitted by the averaged posterior predictive distributions of the candidate models was 11.84°C to 12.32°C (MAD ± 0.33°C, 0.30°C). The median thermal preference range predicted for the recovered radio-tagged Sockeye Salmon (by the averaged posterior predictive distributions of the models) was 13.39°C and 13.89°C, respectively (MAD ± 90 0.35°C, 0.37°C). The predicted values of the thermal preference ranges varied from 12.05°C to 14.89°C for the lower Q25 value and from 12.48°C to 15.56°C for the upper Q75 value (Figure 2.9). The predicted thermal preference ranges were nearly identical during the daytime and nighttime (Figure 2.9). Figure 2.9. Violin plots of the predicted thermal preference ranges (Interquartile range – Q25, Q75) of the 23 recovered radio-tagged female Sockeye Salmon during the civil daytime and nighttime. The boxplots include the parameter median (horizontal bar), interquartile range (Q25 – Q75, lower and upper bounds of the coloured boxes), whiskers (lowest and highest values outside of the interquartile range and within 1.5 x the interquartile range, vertical lines), and outliers (values extending past the interquartile range by more than ± 1.5 x the interquartile range, circles). The solid colours represent the densities of thermal preference estimates. 91 Days and Depths Experienced in Babine Lake After tagging, I estimated that it took the recovered radio-tagged fish between 0.9 to 5.6 days to enter Babine Lake (median: 2.2 days, MAD: ± 0.5 days) (Figure 2.10). The fish spent an estimated median of 4.8 days in Babine Lake (MAD: ± 1.54 days), with fish ID 17 spending the fewest days in the lake (1.7 days) and fish ID 43 spending the most days in the lake (24.7 days) (Figure 2.10). Interestingly, despite spending the fewest days in Babine Lake, fish ID 17 took the longest to reach Babine Lake after being released (Figure 2.10). The estimated total time it took for the tagged Sockeye Salmon after release to enter their spawning tributaries ranged from 4.5 to 27.8 days (Figure 2.10). The tagged Sockeye Salmon reached a maximum depth of 53.4m and shallowest depth of 0m in Babine Lake (meaning that the base of their dorsal fins were within 0.1m from the water surface). The median depth experienced in Babine Lake by the tagged fish was 8.3m (MAD: ± 7.8m). Nearly all the recovered tagged fish made frequent, rapid, and deep descents to cooler water in Babine Lake (See Appendix Table A2, Appendix figures A6 & A7). 92 Figure 2.10. The date and time each recovered radio-tagged fish was released after tagging and the estimated time they entered and exited Babine Lake in 2021. The bars represent the duration each fish was assumed to be in Babine Lake. Temperatures in Babine Lake and Quality of Thermal Habitat (ࢊࢋ) While the recovered radio-tagged fish were in Babine Lake (from August 28th, 2021, until September 27th, 2021), the warmest lake temperature predicted by the GAMM was 17.31°C occurring at a depth of 0m on September 2nd, 2021 (Figure 2.11). The coolest predicted temperature in the shallowest 55 m of Babine Lake was 4.23°C occurring at a depth of 53m on September 1st, 2021. Throughout the month of September, the predicted Babine Lake water temperatures below 40m remained relative constant 93 (Figure 2.11). After September 2nd, 2021, the predicted temperatures of shallower water (especially between 0m and 20m) gradually cooled (Figure 2.11). The weighted-average hourly temperature of Babine Lake ranged from 9.69°C to 11.53°C and was always cooler than the predicted thermal preferences of the tagged fish. This indicates that more of the available habitat (or water volume) in Babine Lake was cooler than their thermal preference. Consequently, median hourly average quality of thermal habitat (݀݁) was 4.75°C (MAD: ± 0.26°C) during the time the recovered radiotagged fish migrated through Babine Lake (Figure 2.12). 94 Figure 2.11. The predicted hourly temperatures from 0 to 60 meters in Babine Lake from August 27th, 2021, to September 27th, 2021 (the time period recovered radio-tagged sockeye salmon were believed to be in Babine Lake). The temperatures were predicted with a GAMM model and eight stations of Fisheries and Oceans Canada temperature logger data. Accuracy of Behavioural Thermoregulation (ࢊ࢈) While in Babine Lake, 74% of the fish’s predicted body temperatures were cooler than their predicted thermal preference range, 23% were warmer, and 4% were within their thermal preference ranges. All of the recovered radio-tagged fish experienced temperatures within their predicted thermal preference range while in Babine Lake temporarily and the duration of time each fish exhibited body temperatures within their predicted thermal preference range varied from 3 minutes to 34.5 hours. The median hourly accuracy of behavioural thermoregulation (ܾ݀) score was smaller than the median 95 thermal quality of habitat score (Figure 2.12). However, the hourly accuracy of behavioural thermoregulation was more variable (MAD ± 2.80°C) and the poorest (i.e. largest value) accuracy of thermoregulation score was nearly twice as large as the poorest (i.e. largest value) thermal habitat quality score (Figure 2.12). Effectiveness of Behavioural Thermoregulation (ࡱ) The median hourly effectiveness of behavioural thermoregulation was moderate and positive (median 1.90, compared to the highest hourly ‫ ܧ‬of 5.76) indicating that the fish behaviourally thermoregulated in Babine Lake somewhat effectively (Figure 2.12). The effectiveness scores did notably vary between individuals and between the hourly scores for the same individual (MAD ± 2.68, Appendix Table A2). The fish also frequently achieved negative thermoregulatory effectiveness scores, which can be interpreted as avoidance of thermally favourable temperatures or poor behavioural thermoregulation (Figure 2.12, Appendix Table A2). Two of the recovered fish had slightly negative overall median ‫ ܧ‬estimates, while the rest (21 individuals) had positive overall median ‫ ܧ‬estimates (Appendix Table A2). 96 Figure 2.12. Violin plots of the estimated hourly thermal habitat quality (݀݁, °C), accuracy of thermoregulation (ܾ݀, °C), and effectiveness of behavioural thermoregulation (‫ )ܧ‬of the recovered radio-tagged female Sockeye Salmon while they were predicted to be in Babine Lake in August and September 2021. The boxplots include the median (horizontal bar), interquartile range (Q25 – Q75, lower and upper bounds of the coloured boxes), whiskers (lowest and highest values outside of the interquartile range and within 1.5 x the interquartile range, vertical lines), and outliers (values extending past the interquartile range by more than ± 1.5 x the interquartile range, circles). The solid colours represent the densities of estimates. Effectiveness of Behavioural Thermoregulation Models Out of the 32 candidate models of behavioural thermoregulation effectiveness, the top ranked model based on AIC had heat stress, RIB, mass, and day/night as the fixed effects, however this model had a relatively low AIC weight of 0.12 (Appendix Table A3). The second ranked model had heat stress, mass, and day/night as fixed effects, 97 followed by the model with heat stress, RIB, and mass as fixed effects, which had AIC weights of 0.11 and 0.09 respectively (Appendix Table A3). The rest of the candidate models had AIC weights less than 0.08, and the top 23 models had AIC weights adding to at least 0.95 (Appendix Table A3). Both the intercept only model and the global model were included in the top 23 models (Appendix Table A3). Heat stress had the greatest relative variable importance out of all the fixed effects, with models containing this variable having AIC weights totaling to 0.79 (Appendix Table A3). Mass had the next greatest relative variable importance (0.58), closely followed by day/night, that had relative importance of 0.51 (Appendix Table A3). RIB had a relative importance of 0.39, and GSE had the lowest relative importance of 0.23 (Appendix Table A3). The averaged model of effectiveness of behavioural thermoregulation indicated that heat stress had the greatest effect on ‫ ܧ‬per 1 unit change in standard deviation, followed by mass, RIB, and day/night (Figure 2.13, Figure 2.14). GSE had the smallest effect on ‫( ܧ‬Figure 2.13, Figure 2.14). Mass was negatively related to ‫ܧ‬, indicating that the larger tagged salmon may have been less effective behavioural thermoregulators (Figure 2.13, Figure 2.14). The other model-averaged fixed effects were positively related to ‫ܧ‬, indicating that fish with higher heat stress scores, RIBs, and GSEs may have behaviourally thermoregulated more effectively, and that the salmon were more effective thermoregulators during the nighttime (Figure 2.13, Figure 2.14). However, there is substantial uncertainty about these effects as their model-averaged had relatively wide confidence intervals that largely extended beyond zero (figure 2.13). 98 Figure 2.13. Model-averaged estimates and associated 95% confidence interval for the covariates present in the models included in the 95% confidence set for the best model of behavioural thermoregulation effectiveness (‫)ܧ‬. The expected ‫ ܧ‬for the daytime is represented by the intercept, which had a model average estimate of 1.40 (95% CI: 0.65 – 2.15). 99 Figure 2.14. Model-averaged predictions of the effectiveness of behavioural thermoregulation for a Sockeye Salmon with the following covariate values, based on the ranges of covariates measured in the radio-tagged and thermal preference test fish: A. Sockeye Salmon with the minimum (1000g) and maximum measured masses (2300g); B. Sockeye Salmon with the minimum (4.60MJ•kg-1) and maximum measured GSE (6.69MJ•kg-1); C. Sockeye Salmon with the minimum (0.06) and maximum (0.95) heat stress value; and D. Sockeye Salmon with the minimum (0.0) and maximum estimated RIB (1.93). Predictions were made for the two values of each variable for day and night, while keeping all other variables at their average value in the data set. Spawning Success and Effectiveness of Behavioural Thermoregulation The majority of the intact recovered radio-tagged Sockeye Salmon retained few to no eggs after spawning. Twelve fish were classified as retaining 0% of their eggs, two were classified as retaining 25% of their eggs, and two were classified as retaining 50% of their eggs (Figure 2.15). There may have been a negative relationship between the 100 overall median effectiveness of behavioural thermoregulation scores of the Sockeye Salmon in Babine Lake and their egg retention post-spawn. However this relationship was not investigated statistically due to the small sample size and the small number of recovered fish with 25% and 50% egg retention. All of the fish that retained 25% or 50% of their eggs after spawning had median effectiveness scores below two, whereas over half of the fish that retained few to no eggs (0% retention) had scores above two (Figure 2.15). The fish with the highest effectiveness of behavioural thermoregulation score had a 0% egg retention score and the fish with the lowest effectiveness score retained approximately 25% of their eggs (Figure 2.15). Figure 2.15. The approximate percentage of eggs retained after spawning and the median hourly effectiveness of behavioural thermoregulation scores of 15 of the recovered radio-tagged Sockeye Salmon. Eight recovered fish are excluded from this plot due to only their tag being recovered (with no body) or because their body cavity was not intact upon recovery. A darker orange circle indicates overlapping points. 101 Discussion In this chapter, I estimated the thermal preference ranges of adult female Babine Lake Sockeye Salmon and investigated how effectively they may have behaviourally thermoregulated while passing through Babine Lake. The median model-averaged fitted values of thermal preference range of the shuttle box salmon was 11.84°C to 12.32°C, and the predicted thermal preference range of the radio-tagged salmon in Babine Lake was 13.39°C to 13.89°C. While in Babine Lake, the radio-tagged salmon appeared to frequently and effectively behaviourally thermoregulate, however there was a lot of interand intra-individual variability in the effectiveness of behavioural thermoregulation (‫)ܧ‬ scores. There was substantial uncertainty about the relationships between ‫ ܧ‬and the indicators of heat stress, mass, GSE, and RIB of the Sockeye Salmon, and with the time of day (day or time), however heat stress exhibited the strongest relationship with ‫ܧ‬. Interestingly, all the fish exhibiting median hourly ‫ ܧ‬greater than two were found to have fully spawned their eggs. Thermal Preference Results The thermal preferences measured in lab and predicted for the radio-tagged Sockeye Salmon were very similar to previously estimated thermal preferences of juvenile Sockeye Salmon. Brett (1952) found that the thermal preferences of juvenile Issaquah Creek Sockeye Salmon ranged from 12.0°C to 14.0°C, a range that nearly encompasses the median observed shuttle box fish thermal preference range and the model-averaged predicted median thermal preference range of the recovered radio-tagged 102 Babine Lake Sockeye Salmon Unsurprisingly, the estimated thermal preference ranges were greater than the lethal lower temperature limit (0°C, four-day exposure) and less than the upper temperature limit (25.1°C, seven-day exposure) that have been estimated for juvenile Sockeye Salmon, and below the upper lethal temperature (22°C, 5-day exposure) identified for adult Sockeye Salmon (Brett, 1952; Servizi & Jensen, 1977). The thermal preference estimates were also notably cooler than the upper critical temperatures that were estimated for adult migrating Fraser River Sockeye Salmon from six different populations (20.8°C to 29.4°C, temperatures at which death can occur due to inability to use aerobic metabolism leading to cardiorespiratory collapse, Eliason et al., 2011). Additionally, the estimated ranges of thermal preferences of Babine Lake Sockeye Salmon were slightly cooler than the thermal optimum temperatures for cardiorespiratory ability that Eliason et al. (2011) estimated for six Fraser River Sockeye Salmon populations (14.5°C to 17.2°C). Assuming that Babine Lake Sockeye Salmon optimal temperatures are within the range measured for Fraser River Sockeye Salmon, an estimated thermal preference lower than an estimated thermal optimum is not surprising since the preferred temperature of an ectotherm is often cooler than its thermal optimum (Martin & Huey, 2008). An ectotherm behaviourally thermoregulating to its thermal optimum would inevitably experience unfavourable warmer temperatures due to environmental fluctuations and imperfect thermoregulation, a risk that can be reduced by thermoregulating to temperatures slightly below the optimum (i.e., to their thermal preference) (Martin & Huey, 2008). Therefore, the estimated thermal preference ranges 103 of Babine Lake Sockeye Salmon likely fit this cooler-than-optimal or sub-optimal is optimal pattern seen in ectotherms (Martin & Huey, 2008). The candidate thermal preference models had similar predictive performances, as indicated by the similar LOOIC scores and small elpd differences. The model similarities suggest that the inclusion of fixed effects heat stress, mass, RIB, GSE, and day/night did not notably improve model performance when compared to the intercept only model. This could also be interpreted as a lack of support for strong linear relationships between these parameters and the thermal preference of Sockeye Salmon. This was surprising as previous studies have found evidence to support relationships between the thermal preferences of fishes and their body size, infectious burden, thermal stress, energy reserves, and the time of day (Elmer et al., 2022; Lafrance et al., 2005; Mathes et al., 2010; Morita et al., 2009; Newell & Quinn, 2005). Fifteen of the candidate models had non-zero Bayesian stacking weights, and all of the tested fixed effects appeared in at least five of these models. When present in models, heat stress had a relatively large and positive effect size on thermal preference and a 95% credible interval excluding 0, indicating that it may have a notable impact despite the inconclusive results from the leave one out cross-validation. This positive and linear relationship could be due to the thermal stress variable, which measures the expression of a set of biomarker genes related to chronic temperature stress, being positively related to the water temperatures recently experienced by Sockeye Salmon (Akbarzadeh et al., 2018). The positive relationship between the heat stress variable and thermal preference may reflect some level of physiological acclimatization to river temperatures experienced during the approximate 104 three-to-five-week migration from the Skeena River mouth to the sampling site at the Babine River counting fence (Takagi & Smith, 1973). Indeed, studies have found positive relationships between the temperatures fish acclimate to before testing and their lab-estimated thermal preferences. For example, Cincotta and Stauffer (1984) found that the thermal preferences of six freshwater fish species from the Centrarchidae and Cyprinidae families generally increased when they were acclimated to warmer temperatures for five days. Kelsch and Neill (1990) saw a similar positive relationship between acclimation temperature and thermal preference for Bluegill (Lepomis macrochirus) and Blue Tilapia (Oreochromis aureus). However, a positive relationship between acclimation temperature and thermal preference is not universal. For example, Sauter et al. (2001) found that juvenile Chinook Salmon acclimated to warmer temperatures had a cooler thermal preference (likely to reduce energy expenditure after the warmer acclimation temperatures accelerated energy consumption), and Kita et al. (1996) saw no relationship between acclimation temperature and thermal preference of Marbled Rockfish (Sebastiscus marmoratus). A review by Johnson and Kelsch (1998) found that positive relationships between thermal preference and acclimation temperature are seen in fish species that experience large seasonal temperature fluctuations, such as Sockeye Salmon, but not in those that are from habitats with more stable temperature regimes. These results indicate that the relationship between thermal experience and thermal preference of fishes may be species and situationally specific. 105 The thermal preference models provided little to no support for a relationship between GSE and thermal preference. The models that included GSE as a fixed effect indicated that it may have had a slight positive relationship with thermal preference. A similar relationship was seen by Roscoe et al. (2010), who found that female adult Sockeye Salmon passing through lakes in the Seton-Anderson watershed with lower GSE densities tended to occupy slightly cooler temperatures than those with higher GSE densities. It may be more important for salmon with lower energy densities to occupy cooler temperatures since elevated water temperatures can result in more rapid depletion of their limited (and smaller) energy reserves (Plumb, 2018). A possible source of error in the GSE estimates in this study may have been introduced due to the accuracy range of the microwave fat meter used, which was quite wide compared to the fat readings taken (± 1.0% 95% confidence interval compared to fat readings of 0.5% to 2.1%) (Distell Fat Meter user manual, 2011). Since GSE was calculated with the fat readings, inaccuracies in the fat readings may have impacted the relationship between GSE and thermal preference seen in the models. Crossin and Hinch (2005) found a strong and linear relationship between lab-estimated GSE of adult Sockeye Salmon and natural-log transformed microwave fat readings as low as approximately 0.6%, which supports the validity of the fat meter readings taken in this study and their use in GSE estimates. In the future, fat meter readings for each fish could be repeated multiple times to check their accuracy, however this would add potentially unnecessary handling time. Similarly, the thermal preference models indicated that there may be a small, positive relationship between Sockeye Salmon mass and temperature preference 106 (although the support for this relationship was minimal). This result contradicts the trend of larger salmonids tending to occupy cooler waters than smaller salmonids (Barret & Armstrong, 2022; Morita et al., 2011). Larger individuals occupying cooler waters has been attributed to competition (Morgan & O'Sullivan, 2023) or smaller individuals prioritizing foraging and growth over energy preservation (Barret & Armstrong, 2022; Morita et al., 2011), both of which would not have impacted the thermal preference results in this study and may be why the trend was not observed. Bioenergetic models of adult Atlantic Salmon migration up the Lakselva River found that warmer water temperatures were more energetically taxing for smaller adult Atlantic Salmon than for larger individuals (Lennox et al. 2018). If this relationship applies to adult Sockeye Salmon, smaller individuals may gravitate to cooler temperatures to avoid the elevated energetic costs. The thermal preference models suggested that the thermal preferences of the adult Sockeye Salmon were nearly the same during the day and night, with nighttime thermal preferences potentially being marginally cooler. Differences in thermal preference during the daytime and nighttime have been reported in various fish species (Macnaughton et al., 2018; Vera et al., 2023). Vera et al. (2023) found that both Nile Tilapia (Oreochromis niloticus) and Zebrafish (Danio rerio) preferred warmer temperatures in a thermal gradient during the day when they were most active and cooler temperatures at night while resting. Contrarily, Macnaughton et al. (2018) found that juvenile Cutthroat Trout preferred warmer temperatures in shuttle box tests at night when their activity levels were reduced, although this result may have been due to an equipment issue. In the present 107 study, the distance each Sockeye Salmon moved per second during the shuttle box tests was recorded. The correlation between this velocity variable and thermal preference could be investigated in the future. The top thermal preference models also indicated that there may be a small and negative relationship between RIB and thermal preference (where higher RIB resulted in a slightly cooler thermal preference). This relationship could be explained by more heavily infected fish avoiding warm temperatures that could further accelerate infection or disease development. Sockeye Salmon that experience warmer temperatures while migrating may be more susceptible to disease development and increasing RIBs (Teffer et al., 2021; Gilhousen, 1990), and results found by Mathes et al. (2010) indicated that adult Weaver Creek Sockeye Salmon move to cooler waters to possibly prevent the expression of a Parvicapsula minibicornis, a common deleterious parasite. The weak support for the relationship between RIB and thermal preference found in my study may be related to the fact that RIB is a simplified value that does not account for the relative severity of each infectious agent (i.e. to what degree they may negatively impact the fish or trigger an immune response) or that infection does not necessarily equate to the afflicted fish experiencing symptoms of the corresponding disease (Bass et al., 2023). A better general indicator of the degree to which an individual is impacted by infection level may be the active viral disease development variable measured during the genomic analysis, which reveals if a Sockeye Salmon is experiencing an immune response to infectious agents (Miller et al., 2017). For simplicity purposes and due to a strong relationship being previously identified between temperature and the RIB of Sockeye 108 Salmon (Teffer et al., 2021), I decided to use RIB in this chapter as the indicator of infectious burden, however, the relationship between RIB, viral disease development, and thermal preference of the Sockeye Salmon will need to be explored in future analyses. One potential limitation of my thermal preference study is that the 24-hour shuttle box tests may not have been long enough to reveal an accurate thermal preference of the Sockeye Salmon. Christensen et al. (2021) stated that the proper time duration of a shuttle box trial should be long enough to “elicit a consistent pattern in the animals’ choice of environment”, and that this usually occurs in 24 hours or less. Despite the approximate 24-hour duration of the shuttle box test in this study, consistent patterns in the temperatures selected by the salmon were not always seen. For example, the fish in the fifth shuttle box test gradually chose warmer temperatures over the 24-hour test and the fish in the 19th test gradually chose cooler temperatures (See appendix figures A.2. and A.5). Both fish occasionally moved between the cool and warm tanks, suggesting that a lack of exploration (not “understanding” the system) did not cause the lack of consistent temperature pattern. Perhaps a stabilization of the temperatures selected by these fish may have been seen if their tests had been longer. In support of my approach, Harman et al. (2020) found that the duration of shuttle box temperature trials did not significantly impact the estimated thermal preference of Lake Whitefish (Coregonus clupeaformis). Macnaughton et al. (2018) found significant differences between the thermal preferences of Westslope Cutthroat Trout (Oncorhynchus clarkii lewisi) estimated by 12- and 24-hour shuttle box tests, however the differences were quite small and likely due to diurnal 109 differences in thermal preference, which I accounted for in my analyses. These results suggest that extending the duration of the shuttle box trials was unnecessary. Thermoregulation in Babine Lake The majority of Sockeye Salmon that successful reached the spawning grounds appeared to behaviourally thermoregulate in Babine Lake quite effectively, although they were far from exhaustive thermoregulators. The majority of the behavioural thermoregulation effectiveness scores were positive (indicating thermoregulation), although many of the salmon regularly exhibited negative and near-zero scores as well (suggesting avoidance or imperfect thermoregulation). This result corroborates other studies that have found evidence of adult Sockeye Salmon behaviourally thermoregulating in lakes from other watersheds, such as Little Togiak Lake (Wood River system, Alaska), Seton and Anderson Lakes (Seton-Anderson Watershed, British Columbia), and Lake Washington (Lake Washinton Watershed, Washington) (Armstrong et al., 2016; Newell & Quinn, 2005; Roscoe et al., 2010). These lakes, as well as Babine Lake, appear to provide adult Sockeye Salmon with important thermoregulatory opportunities. Despite the generally positive and moderately high effectiveness of behavioural thermoregulation scores, the recovered Sockeye Salmon infrequently experienced temperatures in their thermal preference range while in Babine Lake. The salmon made multiple rapid descents to deep, cooler waters and ascents to warmer, shallower waters while in Babine Lake, only briefly passing through waters temperatures in their estimated 110 thermal preference range (see Appendix figures A.6 and A.7). This behaviour has been exhibited by adult Sockeye Salmon in other lakes (Newell & Quinn, 2005; Roscoe et al., 2010) and may be influenced by trade-offs between temperature and other factors such as navigation and dissolved oxygen levels, energetic costs of behaviourally thermoregulating, or simply a result of imperfect thermoregulation. The benefits of behavioural thermoregulation are believed to increase when thermal habitat quality is poorer since the negative physiological effects increase in severity as temperatures become more unfavourable (Angilletta Jr., 2009; Huey & Slatkin, 1976). As a result, effective behavioural thermoregulation may be more prevalent in less thermally favourable environments. Chinook Salmon migrating in the Columbia River have been shown to only behaviourally thermoregulate when mainstem river temperatures reach or exceed 20°C (which is likely poor thermal habitat quality, Goniea et al., 2006). Perhaps the thermal habitat quality of Babine Lake was not poor enough to elicit extremely effective behavioural thermoregulation, resulting instead in the observed moderately effective and variable behavioural thermoregulation. It is important to consider that error may have been introduced into the calculations of effectiveness of behavioural thermoregulation through multiple avenues. First, the Sockeye Salmon body temperatures in Babine Lake (used to calculate ܾ݀) were estimated using the heat exchange model outlined in Chapter One. Any issues or inaccuracies with the model could have caused inaccurate estimates of ܾ݀, however this may not be a concern since the model predictions of body temperatures of the shuttle box fish closely matches the recorded internal temperatures. Similarly, any errors in the 111 thermal preference models could have led to inaccuracies in the estimated thermal preferences of the radio-tagged Sockeye Salmon and added error to the ܾ݀ and ݀݁ calculations. A key assumption of using the thermal preferences estimated at the Babine River fence for calculations of ݀݁ and ܾ݀ is that they did not change between the time of sampling and when the fish exited Babine Lake, and that the levels of the covariates included in the model as fixed effects (mass, RIB, heat stress, and GSE) measured at the time of capture were representative of the conditions influencing the fish thermoregulatory behaviour as they moved through Babine Lake. However, it is likely that these variables changed during the time it took the fish to traverse the lake (between 4.5 and 27.7 days). The infectious burden of adult salmon can change in a few weeks, as was observed in adult Coho Salmon by Chapman et. al. (2020), who found that Coho Salmon exhibited a notable increase in RIBs after migrating 75km to spawning grounds over periods of 2.9 to 37 days. Additionally, as briefly discussed in Chapter One, Sockeye Salmon use up their limited fat reserves to mature and fuel their migration (Crossin & Hinch, 2005). The used-up fat is largely replaced with water, meaning the masses of the fish may not decrease but their GSE density would (Crossin & Hinch, 2005). The usage of energy stores would continue while the Sockeye Salmon were in Babine Lake, impacting their GSE. Furthermore, the expression of genes associated with salmon heat stress have been found to decrease in Sockeye Salmon while they migrate through lakes (Elmer et al., 2023). It is likely that heat stress levels of the tagged Babine Lake Sockeye Salmon would have reduced after tagging since they passed through cool deep waters in Nilkitkwa Lake and Babine Lake. 112 Error was also likely introduced in the hourly quality of thermal habitat scores since the variable temperature profiles seen throughout Babine Lake had to be simplified into a singular temperature average for every 5m depth interval. Temperature-depth profiles vary throughout Babine Lake, with the South Arm typically having a deeper thermocline and the North Arm having a shallower thermocline (Shortreed & Morton, 2000). This means that when the Sockeye Salmon were in areas of the lake with average temperatures (relative to the temperatures throughout Babine Lake), the estimated available temperatures would have been more accurate than when they were in the cooler and warmer sections of the lake (North Arm, South Arm). Furthermore, calculations of thermal habitat quality operate under the assumption that all temperatures are equally accessible to the study animal. Although the data from the temperature loggers in the South Arm of Babine Lake were included, this area is unlikely to have been used by the tagged Sockeye Salmon compared to other areas of the lake since it is many 10s of km out of the way from their spawning grounds (Fulton River and Morrison Creek). Future studies could incorporate salmon location data and arm-specific temperatures to calculate more accurate ݀݁ scores. E and Measured Body Parameters The models of effectiveness of behavioural thermoregulation showed high uncertainty for the effects of Sockeye Salmon mass, energy reserves, relative infectious burden, time of day, or heat stress driving thermoregulation in Babine Lake. All of the models had low AIC weights (below 0.13), and the majority of the models run were included in the top model set (cumulative AICc weights adding to 0.95). The fixed effect 113 with the most support was heat stress, as it had high relative variable importance score, followed by mass, day/night, and RIB that had more moderate variable importance scores. However, all model-averaged fixed effect estimates had large confidence interval. The average effectiveness of behavioural thermoregulation models indicated that more thermally stressed Sockeye Salmon behaviourally thermoregulated more effectively in Babine Lake. Perhaps these fish allocated more effort towards occupying preferred temperatures in Babine Lake to avoid any further thermal stress. There are well established associations between heat stress and mortality of adult migrating salmon (e.g., Minke‐Martin et al., 2018; von Biela et al., 2020), and more thermally stressed salmon may be at a greater risk of these outcomes. Similarly, the averaged ‫ ܧ‬model indicated that salmon with higher infectious burdens behaviourally thermoregulated slightly more effectively than individuals with lower burdens. Elmer et al. (2023) saw a similar trend exhibited by adult Sockeye Salmon in Seton and Anderson lakes, where the individuals with higher relative infectious burdens were those that exhibited thermoregulatory behaviour. Thermoregulating more effectively could slow the development of disease or infectious agents that are impacted by temperature (Elmer et al. 2023) and may drive those with higher RIBs to behaviourally thermoregulated more effectively in Babine Lake to slow further development. The models of ‫ ܧ‬indicated that the Sockeye Salmon may have behaviourally thermoregulated more effectively during the night than during the day. This may have corresponded with the fish occupying slightly shallower waters with warmer temperatures (around their thermal preference) during the night and deeper, cooler waters 114 in the daytime. Roscoe et al. (2010) saw a similar behavioural pattern exhibited by some adult Sockeye Salmon in Anderson Lake, where they moved to warmer, shallower waters in the nighttime and cooler, deeper waters in the day. Perhaps this behaviour is influenced by navigation, since Sockeye Salmon likely use visual cues in lakes to help navigate and locate spawning grounds and may also use daylight to help with navigation through dams and fishways (Johnson & Groot, 1963; Naughton et al., 2005; Ueda et al., 1993). It is possible that the Sockeye Salmon allocated more effort to navigation during the day when it was light (making visual navigation easier) and allocated more effort to behavioural thermoregulation at night when visual navigation would be more difficult. Error may have been introduced in the models of the relationships between ‫ ܧ‬and measured fish variables since the same variables were used to model and estimate the thermal preference ranges of the tagged Sockeye Salmon. Fish that had higher heat stress scores, masses, and GSEs had higher predicted thermal preferences, and would have been assigned larger thermal habitat quality scores (due to cooler Babine Lake temperatures). In response, it would have been easier for them to achieve higher ‫ ܧ‬scores and augmented the positive relationship between heat stress and ‫( ܧ‬since this relationship was positive in both the thermal preference models and averaged effectiveness of thermoregulation models) and reduced the effect size of mass, RIB, GSE, and day/night in the averaged model (since they had differing positive and negative effects in the thermal preference models and averaged ‫ ܧ‬model). 115 E and Spawning Success An additional goal of this chapter was to investigate if effective behavioural thermoregulation in Babine Lake impacted the fitness, or spawning success, of adult female Sockeye Salmon. There may have been a positive relationship between spawning success and ‫ ܧ‬of the salmon (as indicated by all the fish with ‫ ܧ‬greater than two having 0% egg retention after spawning), however I did not model the relationship because nearly all the recovered fish had very high spawning success (or low egg retention). This may have been because the thermal habitat quality of Babine Lake was good enough that non-effective behavioural thermoregulation did not negatively impact fitness or spawning success. As mentioned earlier in the discussion section, most of Babine Lake was quite cool and the calculated average temperatures of the lake were within five degrees of the Sockeye Salmon estimated thermal preferences (indicated by the overall median ݀݁ of approximately 4.75). It would be interesting to investigate the relationship between ‫ ܧ‬and spawning success for Sockeye Salmon passing through lakes with lower thermal habitat qualities, or in systems where salmon do not have access to lakes on route to spawning ground. 116 Conclusion Additional Questions A missing component in the thermal preference and effectiveness of behavioural thermoregulation analyses I conducted was the reproductive maturity (or proximity to spawning) of the sampled Sockeye Salmon. This was originally intended to be included in the analysis as a fixed effect in the thermal preference and effectiveness of behavioural thermoregulation models, however I was not able to due to an issue with the blood samples taken to measure indicators of sexual maturation. Previous research has found that reproductive maturity impacted temperature selection and behavioural thermoregulation of Sockeye Salmon in lakes, and it would be interesting to see if this is true for Babine Lake Sockeye Salmon (Newell & Quinn, 2005; Roscoe et al., 2010). Some additional unanswered questions regarding the behaviour of Babine Lake Sockeye Salmon that could be investigated in the future include why adult Sockeye Salmon spend varying amounts of time in Babine Lake, and if Babine Lake Sockeye Salmon from different tributaries have different thermal preferences or exhibit differences in thermoregulatory behaviour. Additionally, it would be interesting to investigate if Sockeye Salmon behaviourally thermoregulate more effectively when habitat thermal quality is poorer (like some reptiles do) or exhibit effective thermoregulatory behavior in rivers and spawning grounds. 117 Research Implications Estimated thermal preferences of the Sockeye Salmon were mostly cooler than the mean daily Babine River temperature during salmon migration (15°C), and notably cooler than temperatures the river can reach during the hottest periods of the summer, where it can exceed 20.0°C (Stiff et al., 2015). After encountering warm and stressful river water temperatures, cooler waters (like those available in Babine Lake) can help salmon recover and avoid premature death (Bradford et al., 2010; Breau and Caissie, 2012; Kraskura et al., 2020). Babine Lake appeared to provide decent quality thermal habitat to the Sockeye Salmon with mean temperature usually within 5°C of the thermal preference ranges of the salmon, and had an abundance of temperatures well below documented critical and lethal temperature limits of Sockeye Salmon. Nearly all the recovered Sockeye Salmon had high spawning success, even if they did not exhibit highly effective thermoregulatory behaviour in Babine Lake – which in conjunction with the relatively good thermal habitat quality scores may indicate that the salmon did not need to actively thermoregulate in Babine Lake to experience favourable temperatures and the physiological benefits they may provide. Further support for the importance of Babine Lake to adult Sockeye Salmon is the number of days the fish spent in the lake. Even though the tagged salmon could have passed through Babine Lake to get to Fulton River or Morrison Creek in under two days (as illustrated by one of the tagged fish), they spent more than double this amount of time in the lake on average and up to three weeks in the lake. Babine Lake may also provide the fish relief from other stressors that they 118 could encounter in rivers (in addition to thermal stress), such as predation from Grizzly Bears and fast river currents (Bentley et al., 2014). Despite the relief Sockeye Salmon may get from warming river temperatures by accessing lakes, climate change has also been linked to increases in surface and hypolimnion temperatures of lakes, longer periods of lake stratification, and increases in total lake heat content (Desgué-Itier et al., 2023; O'Reilly et al., 2015; Šarović & Klaić, 2023). Although there is a lack of long-term monitoring of lake temperature profiles in British Columbia, a recent thesis on the limnological changes that have occurred in Cultus Lake, British Columbia found that climate change was driving surface temperatures to warm, total lake heat content to increase, and water quality in the lake to degrade (Sumka, 2017; Whijtkamp, 2011). There is also a lack of published data on longterm temperatures trends in Babine Lake, however Stiff et. al., (2015) found that mean river temperatures at the Babine River fence (slightly downriver from the outlets of Babine and Nilkitkwa Lakes) during Sockeye Salmon migration has warmed at a rate of 0.15°C per decade since the 1900s, indicating possible warming of Babine and Nilkitkwa Lakes as well. Warming lake temperatures can lead to depletion of dissolved oxygen (in part due to reduced gas solubility and vertical mixing) (Jane et al., 2021), potentially leading to a reduction in viable thermal and sufficiently oxygenated habitat in lakes (Sumka, 2017). Therefore, in the future, more effective thermoregulation may be needed for adult Sockeye Salmon to experience the benefits of lake thermal habitat and may become more prevalent and imperative as both rivers and lakes warm with climate change. 119 References Akbarzadeh, A., Günther, O. P., Houde, A. L., Li, S., Ming, T. J., Jeffries, K. M., Hinch, S. G., & Miller, K. M. (2018). Developing specific molecular biomarkers for thermal stress in salmonids. BMC genomics, 19, 1-28. https://doi.org/10.1186/s12864-018-5108-9 Akbarzadeh, A., Houde, A. L. S., Sutherland, B. J., Günther, O. P., & Miller, K. M. (2020). Identification of hypoxia-specific biomarkers in salmonids using RNAsequencing and validation using high-throughput qPCR. G3: Genes, Genomes, Genetics, 10(9), 3321-3336. https://doi.org/ 10.1534/g3.120.401487 Akbarzadeh, A., Selbie, D. T., Pon, L. B., & Miller, K. M. (2021). Endangered Cultus Lake sockeye salmon exhibit genomic evidence of hypoxic and thermal stresses while rearing in degrading freshwater lacustrine critical habitat. Conservation Physiology, 9(1), coab089. https://doi.org/10.1093/conphys/coab089 Amat‐Trigo, F., Andreou, D., Gillingham, P. K., & Britton, J. R. (2023). Behavioural thermoregulation in cold‐water freshwater fish: Innate resilience to climate warming?. Fish and Fisheries, 24(1), 187-195. https://doi.org/10.1111/faf.12720 Angilletta Jr, M. J. (2009). Thermal Adaption: A Theoretical and Empirical Synthesis. Oxford University Press. Armstrong, J. B., & Schindler, D. E. (2013). Going with the flow: Spatial distributions of juvenile coho salmon track an annually shifting mosaic of water temperature. Ecosystems, 16(8), 1429-1441. https://doi.org/10.1007/s10021-0139693-9 Armstrong, J. B., Ward, E. J., Schindler, D. E., & Lisi, P. J. (2016). Adaptive capacity at the northern front: sockeye salmon behaviourally thermoregulate during novel exposure to warm temperatures. Conservation Physiology, 4(1), cow039. https:// doi.org/10.1093/conphys/cow039 Baird, O. E., & Krueger, C. C. (2003). Behavioral thermoregulation of brook and rainbow trout: comparison of summer habitat use in an Adirondack River, New York. Transactions of the American Fisheries Society, 132(6), 1194-1206. https:// doi.org/10.1577/T02-127 Bakken, G. S. (1992). Measurement and application of operative and standard operative temperatures in ecology. American Zoologist, 32(2), 194-216. https://doi.org/ 10.1093/icb/32.2.194 120 Barrett, H. S., & Armstrong, J. B. (2022). Move, migrate, or tolerate: Quantifying three tactics for cold‐water fish coping with warm summers in a large river. Ecosphere, 13(6), e4095. https://doi.org/10.1002/ecs2.4095 Bartoń, K. (2023). MuMIn: Multi-Model Inference. R package version 1.47.5. https:// CRAN.R-project.org/package=MuMIn Bass, A. L., Bateman, A. W., Kaukinen, K. H., Li, S., Ming, T., Patterson, D. A., Hinch, S. H., & Miller, K. M. (2023). The spatial distribution of infectious agents in wild Pacific salmon along the British Columbia coast. Scientific Reports, 13(1), 5473. https://doi.org/10.1038/s41598-023-32583-8 Bass, A. L., Hinch, S. G., Teffer, A. K., Patterson, D. A., & Miller, K. M. (2019). Fisheries capture and infectious agents are associated with travel rate and survival of Chinook salmon during spawning migration. Fisheries Research, 209, 156166. https://doi.org/10.1016/j.fishres.2018.09.009 Bentley, K. T., Schindler, D. E., Cline, T. J., Armstrong, J. B., Macias, D., Ciepiela, L. R., & Hilborn, R. (2014). Predator avoidance during reproduction: diel movements by spawning sockeye salmon between stream and lake habitats. Journal of Animal Ecology, 83(6), 1478-1489. https://doi.org/10.1111/13652656.12223 Bertolo, A., Pépino, M., Adams, J., & Magnan, P. (2011). Behavioural thermoregulatory tactics in lacustrine brook charr, Salvelinus fontinalis. PLOS One, 6(4), e18603. https://doi.org/10.1371/journal.pone.0018603 Bicego, K. C., Barros, R. C., & Branco, L. G. (2007). Physiology of temperature regulation: comparative aspects. Comparative Biochemistry and Physiology Part A: Molecular & Integrative Physiology, 147(3), 616-639. https://doi.org/10.1016/ j.cbpa.2006.06.032 Biro, P. A. (1998). Staying cool: behavioral thermoregulation during summer by youngof-year brook trout in a lake. Transactions of the American Fisheries Society, 127(2), 212-222. https://doi.org/10.1577/1548-8659(1998)127< 0212:SCBTDS>2.0.CO;2 Bladon, A. J., Lewis, M., Bladon, E. K., Buckton, S. J., Corbett, S., Ewing, S. R., Hayes, M. P., Hitchcock, G. E., Knock., Lucas, C., McVeigh, A., Menéndez, R., Walker, J. M., Fayle, T., & Turner, E. C. (2020). How butterflies keep their cool: Physical and ecological traits influence thermoregulatory ability and population trends. Journal of Animal Ecology, 89(11), 2440-2450. https://doi.org/10.1111/13652656.13319 121 Blouin-Demers, G., & Nadeau, P. (2005). The cost–benefit model of thermoregulation does not predict lizard thermoregulatory behavior. Ecology, 86(3), 560-566. https://doi.org/10.1890/04-1403 Blouin-Demers, G., & Weatherhead, P. J. (2001). Thermal ecology of black rat snakes (Elaphe obsoleta) in a thermally challenging environment. Ecology, 82(11), 30253043. https://doi.org/10.1890/0012-9658(2001)082[3025:TEOBRS]2.0.CO;2 Bolker, B. M., Brooks, M. E., Clark, C. J., Geange, S. W., Poulsen, J. R., Stevens, M. H. H., & White, J. S. S. (2009). Generalized linear mixed models: A practical guide for ecology and evolution. Trends in Ecology & Evolution, 24(3), 127-135. https://doi.org/10.1016/j.tree.2008.10.008 Bradford, M. J., Lovy, J., Patterson, D. A., Speare, D. J., Bennett, W. R., Stobbart, A. R., & Tovey, C. P. (2010). Parvicapsula minibicornis infections in gill and kidney and the premature mortality of adult sockeye salmon (Oncorhynchus nerka) from Cultus Lake, British Columbia. Canadian Journal of Fisheries and Aquatic Sciences, 67(4), 673-683. http://doi.org/10.1139/F10-017 Breau, C., & Caissie, D. (2013). Adaptive management strategies to protect salmon (Salmo salar) under environmentally stressful conditions. DFO Canadian Science Advisory Secretariat, 2012/164. ii+14 p. Brett, J. R. (1952). Temperature tolerance in young Pacific salmon, genus Oncorhynchus. Journal of the Fisheries Board of Canada, 9(6), 265-323. https://doi.org/10.1139/ f52-016 Bürkner, P. (2017). brms: An R Package for Bayesian Multilevel Models Using Stan. Journal of Statistical Software, 80(1), 1–28. https://doi.org/10.18637/jss.v080.i01 Burnham, K. P., & Anderson, D. R. (2004). Multimodel inference: understanding AIC and BIC in model selection. Sociological Methods & Research, 33(2), 261-304. https://doi.org/10.1177/0049124104268644 Campana, S. E., Dorey, A., Fowler, M., Joyce, W., Wang, Z., Wright, D., & Yashayaev, I. (2011). Migration pathways, behavioural thermoregulation and overwintering grounds of blue sharks in the Northwest Atlantic. PLOS One, 6(2), e16854. https: //doi.org/10.1371/journal.pone.0016854 Carr-Harris, C., Gottesfeld, A. S., & Moore, J. W. (2015). Juvenile salmon usage of the Skeena River estuary. PLOS One, 10(3), e0118988. https://doi.org/10.1371/ journal.pone.0118988 122 Chapman, J. M., Teffer, A. K., Bass, A. L., Hinch, S. G., Patterson, D. A., Miller, K. M., & Cooke, S. J. (2020). Handling, infectious agents and physiological condition influence survival and post-release behaviour in migratory adult coho salmon after experimental displacement. Conservation Physiology, 8(1), coaa033. https:// doi.org/10.1093/conphys/coaa033 Christensen, E. A., Andersen, L. E., Bergsson, H., Steffensen, J. F., & Killen, S. S. (2021). Shuttle-box systems for studying preferred environmental ranges by aquatic animals. Conservation Physiology, 9(1), coab028. https://doi.org/10.1093/ conphys/coab028 Cincotta, D. A., & Stauffer, J. R. (1984). Temperature preference and avoidance studies of six North American freshwater fish species. Hydrobiologia, 109, 173-177. https://doi.org/10.1007/BF00011576 Clark, T. D., Scheuffele, H., Pratchett, M. S., & Skeeles, M. R. (2022). Behavioural temperature regulation is a low priority in a coral reef fish (Plectropomus leopardus): insights from a novel behavioural thermoregulation system. Journal of Experimental Biology, 225(18), jeb244212. https://doi.org/10.1242/jeb.244212 Corey, E., Linnansaari, T., Dugdale, S. J., Bergeron, N., Gendron, J. F., Lapointe, M., & Cunjak, R. A. (2020). Comparing the behavioural thermoregulation response to heat stress by Atlantic salmon parr (Salmo salar) in two rivers. Ecology of freshwater fish, 29(1), 50-62. https://doi.org/10.1111/eff.12487 Cox-Rogers, S. F., Spilsted, B., & Unit, N. C. S. A. (2012). Update assessment of sockeye salmon production from Babine Lake, British Columbia. DFO Canadian Science Advisory Secretariat, 2956: ix + 65 p. Crossin, G. T., & 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(1), 184-191. https://doi.org/10.1577/FT04-076.1 Desgué-Itier, O., Melo Vieira Soares, L., Anneville, O., Bouffard, D., Chanudet, V., Danis, P. A., Domaizon, I., Guillard, J., Mazure, T., Sharaf, N., Soulignac, F., Tran-Khac, V., Vinçon-Liete, B., & Jenny, J. P. (2023). Past and future climate change effects on the thermal regime and oxygen solubility of four peri-alpine lakes. Hydrology and Earth System Sciences, 27(3), 837-859. https://doi.org/10.5194/hess-27-837-2023 123 Distell.com. (n.d.). Distell fish fat meter FFM992 information guide pdf. https:// fishmeatfatmetertester.co.uk/product/distell-com-fish-fat-meter-model-ffm-992/ Doire, J., & Macintyre, D. (2015). 2015 Babine Lake Watershed Sockeye Smolt Population Estimation Project–Mark-Recapture. Lake Babine Nation Fisheries. https://data.skeenasalmon.info/dataset/004a41b1-b610-4c55-b93a2657d7da1d4c/resource/7e6ae895-f1ec-4ddd-8839f74c4a22b5c8/download/2015_babine_lake_watershed_sockeye_smolt_populatio n_estimation_project.pdf Donaldson, M. R. (2008). The physiology, behaviour, and fate of up-river migrating sockeye salmon (Oncorhynchus nerka) in relation to environmental conditions in the Fraser River, British Columbia [Master’s thesis, Carleton University]. Carleton University Institutional Repository. https://doi.org/10.22215/etd/200808359 Edwards, A. L., & Blouin-Demers, G. (2007). Thermoregulation as a function of thermal quality in a northern population of painted turtles, Chrysemys picta. Canadian Journal of Zoology, 85(4), 526-535. https://doi.org/10.1139/Z07-037 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., & Farrell, A. P. (2011). Differences in thermal tolerance among sockeye salmon populations. Science, 332(6025), 109112. https://doi.org/10.1126/science.1199158 Elmer, L. K., Moulton, D. L., Reid, A. J., Farrell, A. P., Patterson, D. A., Hendriks, B., Cooke, S. J., & Hinch, S. G. (2022). Thermal selection and delayed migration by adult sockeye salmon (Oncorhynchus nerka) following escape from simulated inriver fisheries capture. Fisheries Research, 251, 106321. https://doi.org/10.1016/ j.fishres.2022.106321 Fisheries and Oceans Canada. (2018). Developing a genomic tool (FIT-CHIP) for inseason information on salmon health. https://www.dfo-mpo.gc.ca/science/ partnerships-partenariats/research-recherche/grdi-irdg/projects-projets/007eng.html Lake Babine Nation. (2021). LBN Recreational Sockeye Fishery Management Plan. https://www.lakebabine.com/august-23-2021-periodic-recreational-fish-closurescoming-to-babine-river-lbn-recreational-sockeye-fishery-management-plan/ Gilhousen, P. (1990). Prespawning mortalities of sockeye salmon in the Fraser River system and possible causal factors. International Pacific Salmon Fisheries Commission Bulletin, 26, 58. 124 Ginzinger, D. G. (2002). Gene quantification using real-time quantitative PCR: An emerging technology hits the mainstream. Experimental Hematology, 30(6), 503512. https://doi.org/10.1016/S0301-472X(02)00806-8 Goetz, F. A., & Quinn, T. P. (2019). Behavioral thermoregulation by adult Chinook salmon (Oncorhynchus tsliawytscha) in estuary and freshwater habitats prior to spawning. Fishery Bulletin, 117(3), 258-275. https://doi.org/10.7755/FB.117.3.12 Goller, M., Goller, F., & French, S. S. (2014). A heterogeneous thermal environment enables remarkable behavioral thermoregulation in Uta stansburiana. Ecology and Evolution, 4(17), 3319-3329. https://doi.org/10.7755/FB.117.3.12 Goniea, T. M., Keefer, M. L., Bjornn, T. C., Peery, C. A., Bennett, D. H., & Stuehrenberg, L. C. (2006). 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(2), 408-419. https://doi.org/10.1577/T04-113.1 Gottesfeld, A. S., Rabnett, K. A., & Hall, P. E. (2002). Skeena stage i watershed-based fish sustainability plan: Conserving Skeena fish populations and their habitat. Skeena Fisheries Commission. Harman, A. A., Fuzzen, M., Stoa, L., Boreham, D., Manzon, R., Somers, C. M., & Wilson, J. Y. (2020). Evaluating tank acclimation and trial length for shuttle box temperature preference assays. Journal of Experimental Biology, 224(12): jeb233205. https://doi.org/10.1101/2020.07.21.214080 Hanya, G., Kiyono, M., & Hayaishi, S. (2007). Behavioral thermoregulation of wild Japanese macaques: Comparisons between two subpopulations. American Journal of Primatology: Official Journal of the American Society of Primatologists, 69(7), 802-815. https://doi.org/10.1002/ajp.20397 Hertz, P. E., Huey, R. B., & Stevenson, R. D. (1993). Evaluating temperature regulation by field-active ectotherms: The fallacy of the inappropriate question. The American Naturalist, 142(5), 796-818. https://doi.org/10.1086/285573 Hight, B. V., & Lowe, C. G. (2007). Elevated body temperatures of adult female leopard sharks, Triakis semifasciata, while aggregating in shallow nearshore embayments: Evidence for behavioral thermoregulation?. Journal of Experimental Marine Biology and Ecology, 352(1), 114-128. https://doi.org/ 10.1016/j.jembe.2007.07.021 Höge, M., Guthke, A., & Nowak, W. (2020). Bayesian model weighting: The many faces of model averaging. Water, 12(2), 309. https://doi.org/10.3390/w12020309 125 Honek, A., & Martinkova, Z. (2019). Behavioural thermoregulation hastens spring mating activity in Pyrrhocoris apterus (Heteroptera: Pyrrhocoridae). Journal of Thermal Biology, 84, 185-189. https://doi.org/10.1016/j.jtherbio.2019.07.013 Houde, A. L. S., Akbarzadeh, A., Günther, O. P., Li, S., Patterson, D. A., Farrell, A. P., Hinch, S. G., & Miller, K. M. (2019). Salmonid gene expression biomarkers indicative of physiological responses to changes in salinity and temperature, but not dissolved oxygen. Journal of Experimental Biology, 222(13), jeb198036. https://doi.org/10.1242/jeb.198036 Huey, R. B., & Slatkin, M. (1976). Cost and benefits of lizard thermoregulation. The Quarterly Review of Biology, 51(3), 363-384. https://doi.org/10.1086/409470 Hume, J. M., & MacLellan, S. G. (2000). An assessment of the juvenile sockeye salmon (Oncorhynchus nerka) populations of Babine Lake. Canadian Technical Report of Fisheries and Aquatic Sciences. 2327: 37 p. Jane, S. F., Hansen, G. J., Kraemer, B. M., Leavitt, P. R., Mincer, J. L., North, R. L., Pilla, R. M., Stetler, J. T., Williamson, C. E., Woolway, R. l., Arvola, L., Chandra, S., DeGasperi, L. C., Diemer, L., Dunalska, J., Erina, O., Flaim, G., Grossart, H. P., Hambright, K. D., Hein, C., ... & Rose, K. C. (2021). Widespread deoxygenation of temperate lakes. Nature, 594(7861), 66-70. https://doi.org/10.1038/s41586-021-03550-y Johnson, W. E., & Groot, C. (1963). Observations on the migration of young sockeye salmon (Oncorhynchus nerka) through a large, complex lake system. Journal of the Fisheries Board of Canada, 20(4), 919-938. https://doi.org/10.1139/f63-064 Johnson, J. A., & Kelsch, S. W. (1998). Effects of evolutionary thermal environment on temperature-preference relationships in fishes. Environmental Biology of Fishes, 53, 447-458. https://doi.org/10.1023/A:1007425215669 Katinic, P. J., Patterson, D. A., & Ydenberg, R. C. (2015). Thermal regime, predation danger and the early marine exit of sockeye salmon Oncorhynchus nerka. Journal of Fish Biology, 86(1), 276-287. https://doi.org/10.1111/jfb.12596 Keefer, M. L., Clabough, T. S., Jepson, M. A., Johnson, E. L., Peery, C. A., & Caudill, C. C. (2018). Thermal exposure of adult Chinook salmon and steelhead: Diverse behavioral strategies in a large and warming river system. PLOS One, 13(9), e0204274. https://doi.org/10.1371/journal.pone.0204274 126 Keefer, M. L., Clabough, T. S., Jepson, M. A., Naughton, G. P., Blubaugh, T. J., Joosten, D. C., & Caudill, C. C. (2015). Thermal exposure of adult Chinook salmon in the Willamette River basin. Journal of Thermal Biology, 48, 11-20. https://doi.org/ 10.1016/j.jtherbio.2014.12.002 Keefer, M. L., Peery, C. A., & 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(10), 1734-1747. https://doi.org/10.1139/F09-13 Kelsch, S. W., & Neill, W. H. (1990). Temperature preference versus acclimation in fishes: selection for changing metabolic optima. Transactions of the American Fisheries Society, 119(4), 601-610. https://doi.org/10.1577/15488659(1990)119<0601:TPVAIF>2.3.CO;2 Kita, J., Tsuchida, S., & Setoguma, T. (1996). Temperature preference and tolerance, and oxygen consumption of the marbled rockfish, Sebastiscus marmoratus. Marine Biology, 125, 467-471. https://doi.org/10.1007/BF00353259 Kitagawa, T., Hyodo, S., & Sato, K. (2016). Atmospheric depression-mediated water temperature changes affect the vertical movement of chum salmon Oncorhynchus keta. Marine Environmental Research, 119, 72-78. https://doi.org/ 10.1016/j.marenvres.2016.05.016 Kraskura, K., Hardison, E. A., Little, A. G., Dressler, T., Prystay, T. S., Hendriks, B., Farrell, A. P., Cooke, S. J., Patterson, D. A., & Eliason, E. J. (2020). Sex-specific differences in swimming, aerobic metabolism and recovery from exercise in adult coho salmon (Oncorhynchus kisutch) across ecologically relevant temperatures. Conservation Physiology, 9(1), coab016. https://doi.org/10.1093/conphys/ coab100 Lafrance, P., Castonguay, M., Chabot, D., & Audet, C. (2005). Ontogenetic changes in temperature preference of Atlantic cod. Journal of Fish Biology, 66(2), 553-567. https://doi.org/10.1111/j.0022-1112.2005.00623.x Larson, W. A., Lisi, P. J., Seeb, J. E., Seeb, L. W., & Schindler, D. E. (2016). Major histocompatibility complex diversity is positively associated with stream water temperatures in proximate populations of sockeye salmon. Journal of Evolutionary Biology, 29(9), 1846-1859. https://doi.org/10.1111/jeb.12926 127 Lennox, R. J., Eliason, E. J., Havn, T. B., Johansen, M. R., Thorstad, E. B., Cooke, S. J., Diserud, O. H., Whoriskey, F. G., Farrell, A. P., & Uglem, I. (2018). Bioenergetic consequences of warming rivers to adult Atlantic salmon Salmo salar during their spawning migration. Freshwater Biology, 63(11), 1381-1393. https://doi.org/ 10.1111/fwb.13166 Lennox, R. J., Pulg, U., Malley, B., Gabrielsen, S. E., Hanssen, E. M., Cooke, S. J., Birnie-Gauvin, K., Barlaup, B. T., & Vollset, K. W. (2021). The various ways that anadromous salmonids use lake habitats to complete their life history. Canadian Journal of Fisheries and Aquatic Sciences, 78(1), 90-100. https://doi.org/10.1139/ cjfas-2020-0225 Macnaughton, C. J., Kovachik, C., Charles, C., & Enders, E. C. (2018). Using the shuttle box experimental design to determine temperature preference for juvenile westslope cutthroat trout (Oncorhynchus clarkii lewisi). Conservation physiology, 6(1), coy018. https://doi.org/10.1093/conphys/coy018 Martin, T. L., & Huey, R. B. (2008). Why “suboptimal” is optimal: Jensen’s inequality and ectotherm thermal preferences. The American Naturalist, 171(3), E102-E118. https://doi.org/10.1086/527502 Mathes, M. T., Hinch, S. G., Cooke, S. J., Crossin, G. T., Patterson, D. A., Lotto, A. G., & 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(1), 70-84. https://doi.org/10.1139/F09-158 Miller, K. M., Gardner, I. A., Vanderstichel, R., Burnley, T., Angela, D., Li, S., Tabata, A., Kaukinen, K. H., Ming, T. J., & Ginther, N. G. (2016). Report on the performance evaluation of the Fluidigm BioMark platform for high-throughput microbe monitoring in salmon. DFO Canadian Science Advisory Secretariat, 2016/038. xi +282 p. Miller, K. M., Günther, O. P., Li, S., Kaukinen, K. H., & Ming, T. J. (2017). Molecular indices of viral disease development in wild migrating salmon. Conservation Physiology, 5(1), cox036. https://doi.org/10.1093/conphys/cox036 Minke‐Martin, V., Hinch, S. G., Braun, D. C., Burnett, N. J., Casselman, M. T., Eliason, E. J., & Middleton, C. T. (2018). Physiological condition and migratory experience affect fitness‐related outcomes in adult female sockeye salmon. Ecology of Freshwater Fish, 27(1), 296-309. https://doi.org/10.1111/eff.12347 128 Mohammed, R. S., Reynolds, M., James, J., Williams, C., Mohammed, A., Ramsubhag, A., van Oosterhout, C., & Cable, J. (2016). Getting into hot water: sick guppies frequent warmer thermal conditions. Oecologia, 181(3), 911-917. https://doi. org/10.1007/s00442-016-3598-1 Morgan, A. M., & O'Sullivan, A. M. (2023). Cooler, bigger; warmer, smaller: Fine‐scale thermal heterogeneity maps age class and species distribution in behaviourally thermoregulating salmonids. River Research and Applications, 39(2), 163-176. https://doi.org/10.1002/rra.4073 Morita, K., Fukuwaka, M. A., Tanimata, N., & Yamamura, O. (2010). Size‐dependent thermal preferences in a pelagic fish. Oikos, 119(8), 1265-1272. https://doi. org/10.1111/j.1600-0706.2009.18125.x Mugwanya, M., Dawood, M. A., Kimera, F., & Sewilam, H. (2022). Anthropogenic temperature fluctuations and their effect on aquaculture: A comprehensive review. Aquaculture and Fisheries, 7(3), 223-243. https://doi.org/10.1016/ j.aaf.2021.12.005 Muth, C., Oravecz, Z., & Gabry, J. (2018). User-friendly Bayesian regression modeling: A tutorial with rstanarm and shinystan. The Quantitative Methods for Psychology, 14(2), 99-119. https://doi.org/10.20982/tqmp.14.2.p099 Myers, B. J., Lynch, A. J., Bunnell, D. B., Chu, C., Falke, J. A., Kovach, R. P., Krabbenhoft, T. J., Kwak, T. J., & Paukert, C. P. (2017). Global synthesis of the documented and projected effects of climate change on inland fishes. Reviews in Fish Biology and Fisheries, 27, 339-361. https://doi.org/10.1007/s11160-0179476-z Naughton, G. P., Caudill, C. C., Keefer, M. L., Bjornn, T. C., Stuehrenberg, L. C., & 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(1), 30-47. https://doi.org/10.1139/f04-147 Nay, T. J., Johansen, J. L., Rummer, J. L., Steffensen, J. F., & Hoey, A. S. (2021). Species interactions alter the selection of thermal environment in a coral reef fish. Oecologia, 196(2), 363-371. https://doi.org/10.1007/s00442-021-04942-7 Newell, J. C., & 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(9), 1232-1239. https://doi.org/ 10.1139/z05-113 129 Nordahl, O., Tibblin, P., Koch-Schmidt, P., Berggren, H., Larsson, P., & Forsman, A. (2018). Sun-basking fish benefit from body temperatures that are higher than ambient water. Proceedings of the Royal Society B: Biological Sciences, 285(1879), 20180639. https://doi.org/10.1098/rspb.2018.0639 O'Reilly, C. M., Sharma, S., Gray, D. K., Hampton, S. E., Read, J. S., Rowley, R. J., Schneider, P., Lenters, J. D., McIntyre, P. B., Kraemer, B. M., Weyhenmeyer, G. A., Straile, D., Dong, B., Adrian, R., Allan, M. G., Anneville, O., Arvola, L., Austin, J., Bailey, J. L., Baron, J. S., ... & Zhang, G. (2015). Rapid and highly variable warming of lake surface waters around the globe. Geophysical Research Letters, 42(24), 10-773. https://doi.org/10.1002/2015GL066235 Papastamatiou, Y. P., Watanabe, Y. Y., Bradley, D., Dee, L. E., Weng, K., Lowe, C. G., & Caselle, J. E. (2015). Drivers of daily routines in an ectothermic marine predator: hunt warm, rest warmer?. PLOS One, 10(6), e0127807. https://doi.org/10.1371/journal.pone.0127807 Pinheiro J, Bates D, R Core Team (2022). nlme: Linear and Nonlinear Mixed Effects Models. R package version 3.1-160. https://CRAN.R-project.org/package=nlme Plumb, J. M. (2018). A bioenergetics evaluation of temperature‐dependent selection for the spawning phenology by Snake River fall Chinook salmon. Ecology and Evolution, 8(19), 9633-9645. https://doi.org/10.1002/ece3.4353 Podrabsky, J. E., Clelen, D., & Crawshaw, L. I. (2008). Temperature preference and reproductive fitness of the annual killifish Austrofundulus limnaeus exposed to constant and fluctuating temperatures. Journal of Comparative Physiology A, 194(4), 385-393. https://doi.org/10.1007/s00359-008-0313-7 R Core Team (2022). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.Rproject.org/ Raby, G. D., Vandergoot, C. S., Hayden, T. A., Faust, M. D., Kraus, R. T., Dettmers, J. M., Cooke, S. J., Zhao, Y., Fisk, A. T., & Krueger, C. C. (2018). Does behavioural thermoregulation underlie seasonal movements in Lake Erie walleye?. Canadian Journal of Fisheries and Aquatic Sciences, 75(3), 488-496. https://doi.org/10.1139/cjfas-2017-0145 Reynolds, W. W. (1979). Perspective and introduction to the symposium: Thermoregulation in ectotherms. American Zoologist, 19(1), 193-194. https:// doi.org/10.1093/icb/19.1.193 130 Roscoe, D. W., Hinch, S. G., Cooke, S. J., & Patterson, D. A. (2010). 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(1), 51-62. https://doi.org/10.1093/icb/19.1.193 Row, J. R., & Blouin-Demers, G. (2006). Thermal quality influences effectiveness of thermoregulation, habitat use, and behaviour in milk snakes. Oecologia, 148, 111. https://doi.org/10.1007/s00442-005-0350-7 Šarović, K., & Klaić, Z. B. (2023). Effect of climate change on water temperature and stratification of a small, temperate, Karstic Lake (Lake Kozjak, Croatia). Environmental Processes, 10(4), 49. https://doi.org/10.1007/s40710-023-00663-6 Sauter, S. T., Crawshaw, L. I., & Maule, A. G. (2001). Behavioral thermoregulation by juvenile spring and fall chinook salmon, Oncorhynchus tshawytscha, during smoltification. Environmental Biology of Fishes, 61, 295-304. https://doi.org/ 10.1023/A:1010849019677 Servizi, J.A., & Jensen, J.O.T. (1977). Progress Report: Resistance of Adult Sockeye Salmon to Acute Thermal Shock. International Pacific Salmon Fisheries Commission Progress Report, 34. https://www.psc.org/wpfd_file/ipsfcpr34/ Shortreed, K.S., and K.F. Morton. 2000. An assessment of the limnological status and productive capacity of Babine Lake, 25 years after the inception of the Babine Lake Development Project. Canadian Technical Report of Fisheries and Aquatic Sciences, 2316: 52 p. Sivula, T., Magnusson, M., Matamoros, A. A., & Vehtari, A. (2022). Uncertainty in Bayesian leave-one-out cross-validation based model comparison. arXiv:2008.10296v4. https://doi.org/10.48550/arXiv.2008.10296 Stan Development Team. (2023). Stan Modeling Language Users Guide and Reference Manual, 2.32. https://mc-stan.org Stiff, H.W., Hyatt, K.D., Hall, P., Finnegan, B., and Macintyre, D. 2015. Water temperature, river discharge, and adult Sockeye salmon migration observations in the Babine watershed, 1946-2014. Canadian Manuscript Report of Fisheries and Aquatic Sciences, 3053: vi + 169 p. Sumka, M. G. (2017). Climate change impacts on a eutrophying lake: Cultus Lake, British Columbia, Canada [Doctoral dissertation, University of British Columbia]. Vancouver: University of British Columbia Library. https://doi.org/10.14288/1.0354396 131 Sunday, J. M., Bates, A. E., Kearney, M. R., Colwell, R. K., Dulvy, N. K., Longino, J. T., & Huey, R. B. (2014). Thermal-safety margins and the necessity of thermoregulatory behavior across latitude and elevation. Proceedings of the National Academy of Sciences, 111(15), 5610-5615. https://doi.org/10.1073/ pnas.1316145111 Symonds, M. R., & Moussalli, A. (2011). A brief guide to model selection, multimodel inference and model averaging in behavioural ecology using Akaike’s information criterion. Behavioral Ecology and Sociobiology, 65, 13-21. https://doi.org/10.1007/s00265-010-1037-6 Takagi, K., & Smith, H. D. (1973). Timing and rate of migration of Babine sockeye stocks through the Skeena and Babine Rivers. Fisheries Research Board of Canada Technical Report, 419. Teffer, A. K., Hinch, S. G., Miller, K. M., Patterson, D. A., Bass, A. L., Cooke, S. J., Farrell, A. P., Beacham, T. D., Chapman, J. M., & Juanes, F. (2021). Host‐ pathogen‐environment interactions predict survival outcomes of adult sockeye salmon (Oncorhynchus nerka) released from fisheries. Molecular Ecology, 31(1), 134-160. https://doi.org/10.1111/mec.16214 Ueda, H., Kaeriyama, M., Mukasa, K., Urano, A., Kudo, H., Shoji, T., Tokumitsu, Y., Yamauchi, K., & Kurihara, K. (1998). Lacustrine sockeye salmon return straight to their natal area from open water using both visual and olfactory cues. Chemical Senses, 23(2), 207-212. https://doi.org/10.1093/chemse/23.2.207 Vaudo, J. J., & Heithaus, M. R. (2013). Microhabitat selection by marine mesoconsumers in a thermally heterogeneous habitat: behavioral thermoregulation or avoiding predation risk?. PLOS One, 8(4), e61907. https://doi.org/10.1371/journal.pone. 0061907 Vehtari, A., Gelman, A., & Gabry J. (2017). Practical Bayesian model evaluation using leave-one-out cross-validation and WAIC. Statistics and Computing, 27, 1413– 1432. https://doi.org/10.1007/s11222-016-9696-4 Vehtari, A. (2020, March 11). Cross-validation FAQ. Aki Vehtari: Model selection. https://avehtari.github.io/modelselection/CV-FAQ.html#1_What_is_crossvalidation von Biela, V. R., Bowen, L., McCormick, S. D., Carey, M. P., Donnelly, D. S., Waters, S., Regish, A. M., Laske, S. M., Brown, R. J., Larson, S., Zuray, S., & Zimmerman, C. E. (2020). Evidence of prevalent heat stress in Yukon River Chinook salmon. Canadian Journal of Fisheries and Aquatic Sciences, 77(12), 1878-1892. https://doi.org/10.1139/cjfas-2020-0209 132 Ward, A. J. W., Hensor, E. M. A., Webster, M. M., & Hart, P. J. B. (2010). Behavioural thermoregulation in two freshwater fish species. Journal of Fish Biology, 76(10), 2287-2298. https://doi.org/10.1111/j.1095-8649.2010.02576.x Whijtkamp, P.J. (2011). Interannual thermal-regime variability of two lakes in British Columbia, Canada: Implications for Climate Change. [Master’s thesis, Royal Roads University]. Library and Archives Canada. Wood, C. C. (2008). Managing biodiversity of Pacific salmon: lessons from the Skeena River sockeye salmon fishery in British Columbia. American Fisheries Society Symposium, 49, 349-364. Wood, S. N. (2006). Generalized Additive Models: An Introduction with R. (1st ed.). Chapman and Hall/CRC. Wood, S. (2023). mgcv: Mixed GAM Computation Vehicle with Automatic Smoothness Estimation. R package version 1.9-0. https://cran.r-project.org/web/packages/ mgcv/index.html Yang, G., & Moyer, D. L. (2020). Estimation of nonlinear water-quality trends in highfrequency monitoring data. Science of the Total Environment, 715, 136686. https://doi.org/10.1016/j.scitotenv.2020.136686 Yao, Y., Vehtari, A., Simpson, D., & Gelman, A. (2018). Using stacking to average Bayesian predictive distributions (with discussion). Bayesian Analysis, 13(3), 917–1007. https://doi.org/10.1214/17-BA1091 133 Appendix Figure A1. The interquartile range (Q25, Q75) of body temperatures experienced by shuttle box (SB) Sockeye Salmon (Fish) 1, 2, 3, and 4 hourly during their shuttle box test thermal preference tests, which occurred from August 18th, 2021 to August 26th, 2021. 134 Figure A2. The interquartile range (Q25, Q75) of body temperatures experienced by Shuttle Box (SB) Sockeye Salmon (Fish) 5, 6, 7, and 8 hourly during their shuttle box test thermal preference tests, which occurred from August 27th, 2021 to September 1st, 2021. 135 Figure A3. The interquartile range (Q25, Q75) of body temperatures experienced by Shuttle Box (SB) Sockeye Salmon (Fish) 9, 10, 11, and 12 hourly during their shuttle box test thermal preference tests, which occurred from September 1st, 2021 to September 14st, 2021. The data from SB Fish 12 was not included in the thermal preference models. 136 Figure A4. The interquartile range (Q25, Q75) of body temperatures experienced by Shuttle Box (SB) Sockeye Salmon (Fish) 13, 14, 15, and 16 hourly during their shuttle box test thermal preference tests, which occurred from September 1st, 2021 to September 14st, 2021. The data from SB Fish 13 was not included in the thermal preference models. 137 Figure A5. The interquartile range (Q25, Q75) of body temperatures experienced by Shuttle Box (SB) Sockeye Salmon (Fish) 17, 18, 19, and 20 hourly during their shuttle box test thermal preference tests, which occurred from September 24th, 2021 to October 12th, 2021. 138 Table A1. All candidate thermal preference models and their Bayesian stacking weights, the difference in expected log predictive densities between these models and the model with the highest expected log predictive density (elpd Difference, model 5th row from top), and the standard error of these differences (Difference se). The models are ordered based on their stacking weights, then by their elpd difference (for models with stacking weights of 0). Model Intercept RIB + Day/Night Heat Stress + Mass + Day/Night RIB + GSE Heat Stress + Day/Night Heat Stress Heat Stress + Mass Heat Stress + Mass + GSE Heat Stress + Mass + GSE + Day/Night Heat Stress + GSE Heat Stress + RIB Heat Stress + RIB + Mass + GSE + Day/Night RIB + Mass Day/Night Heat Stress + GSE + Day/Night RIB + Mass + Day/Night Mass + GSE + Day/Night RIB + GSE + Day/Night Mass + Day/Night RIB + Mass + GSE + Day/Night GSE RIB Mass GSE + Day/Night Mass + GSE RIB + Mass + GSE Heat Stress + RIB + Mass + GSE Heat Stress + RIB + GSE + Day/Night Heat Stress + RIB + Mass + Day/Night Heat Stress + RIB + Mass Heat Stress + RIB + GSE Heat Stress + RIB + Day/Night Stacking Weight 0.16 0.15 0.13 0.13 0.10 0.06 0.06 0.05 0.05 0.03 0.03 0.02 elpd Difference -8.60 -8.50 -2.70 -7.40 0.00 -0.10 -2.20 -3.50 -3.90 -2.40 -1.20 -6.10 Difference se 6.00 5.50 2.30 5.60 0.00 2.20 3.20 3.30 2.40 2.80 2.60 3.00 0.02 0.01 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -9.60 -9.20 -1.30 -11.60 -11.30 -11.20 -11.20 -10.60 -10.20 -9.70 -9.40 -9.10 -8.80 -8.50 -5.40 -5.40 -5.20 -3.30 -3.10 -2.30 6.20 5.70 1.30 6.60 6.00 6.20 6.20 5.40 5.90 6.20 6.10 5.50 5.40 5.20 3.70 2.20 2.70 3.10 2.80 1.60 139 Figure A.6. Top panel: The body temperatures (֯C) (predicted in Chapter One) experienced by the recovered Sockeye Salmon with radio tag ID 29 while it was in Babine Lake in 2021. Body temperatures were estimated every 1.5 minutes. The horizontal lines represent the thermal preference range (Q25, Q75) of the fish estimated by the averaged thermal preference model. Bottom panel: The depths experienced by the recovered Sockeye Salmon with radio tag ID 29 while in Babine Lake in 2021 (black line). The depths were recorded every 1.5 minutes by the fish's external logger. The colour gradient represents the temperatures predicted every 1 meter with the GAMM model in Chapter Two. The horizontal gray lines represent the thermal preference range (Q25, Q75) of the fish predicted by the averaged thermal preference model. 140 Figure A.7. Top panel: The body temperatures (֯C) (predicted in Chapter One) experienced by the recovered Sockeye Salmon with radio tag ID 43 while it was in Babine Lake in 2021. Body temperatures were estimated every 1.5 minutes. The horizontal lines represent the thermal preference range (Q25, Q75) of the fish estimated by the averaged thermal preference model. Bottom panel: The depths experienced by the recovered Sockeye Salmon with radio tag ID 43 while in Babine Lake in 2021 (black line). The depths were recorded every 1.5 minutes by the fish's external logger. The colour gradient represents the temperatures predicted every 1 meter with the GAMM model in Chapter Two. The horizontal gray lines represent the thermal preference range (Q25, Q75) of the fish predicted by the averaged thermal preference model. 141 Table A2. The median (MAD, min – max) depth and body temperature each recovered radiotagged Sockeye Salmon experienced in Babine Lake, and the same summary values for their hourly ݀݁, ܾ݀, and ‫ ܧ‬scores. I D Depth (m) Temp (°C) ݀݁(°C) ܾ݀(°C) ‫ܧ‬ 11 10.14 (1.97, 0.00 - 27.85) 8.59 (1.90, 6.29 - 15.19) 4.97 (0.06, 4.90 - 5.02) 5.35 (1.98, 0.00 - 7.60) -0.40 (1.92, -2.60 - 5.01) 13 8.51 (10.21, 0.00 - 34.48) 12.41 (2.33, 5.96 - 15.14) 4.61 (0.03, 4.54 - 4.67) 1.05 (0.77, 0.04 - 6.17) 3.55 (0.79, -1.55 - 4.52) 14 5.65 (8.06, 0.00 - 42.46) 13.73 (1.84, 4.74 - 15.46) 4.67 (0.05, 4.63 - 4.75) 0.98 (1.16, 0.00 - 8.52) 3.66 (1.20, -3.79 - 4.74) 16 7.68 (9.68, 0.00 - 46.92) 13.57 (2.08, 5.96 - 15.35) 4.56 (0.03, 4.53 - 4.61) 1.34 (1.25, 0.09 - 6.61) 3.19 (1.32, -2.06 - 4.48) 17 0.01 (0.02, 0.00 - 22.77) 14.87 (0.41, 9.23 - 15.50) 4.66 (0.05, 4.60 - 4.70) 1.10 (0.37, 0.37 - 3.65) 3.56 (0.38, 0.95 - 4.29) 19 5.13 (7.11, 0.00 - 29.88) 11.36 (2.89, 6.66 - 15.38) 4.69 (0.11, 4.46 - 4.85) 1.80 (2.34, 0.00 - 6.48) 2.92 (2.29, -1.76 - 4.74) 24 14.10 (5.86, 0.00 - 29.68) 11.23 (2.00, 6.42 - 15.14) 5.55 (0.16, 5.42 - 5.75) 3.70 (1.67, 0.00 - 8.10) 1.95 (1.56, -2.61 - 5.47) 29 6.08 (4.08, 0.00 - 26.18) 14.05 (1.29, 8.07 - 15.61) 4.29 (0.04, 4.25 - 4.40) 1.75 (0.53, 0.28 - 3.49) 2.54 (0.53, 0.81 - 4.01) 30 6.80 (5.30, 0.00 - 27.97) 10.27 (5.14, 5.83 - 15.50) 4.60 (0.03, 4.58 - 4.73) 2.85 (3.08, 0.26 - 7.26) 1.80 (3.00, -2.61 - 4.32) 32 6.41 (3.09, 0.00 - 28.39) 8.95 (1.89, 6.17 - 16.75) 4.81 (0.17, 4.65 - 5.14) 4.48 (1.83, 0.52 - 6.66) 0.30 (1.96, -1.91 - 4.51) 35 3.60 (4.69, 0.00 - 37.17) 13.41 (1.61, 6.71 - 15.48) 5.20 (0.04, 5.16 - 5.33) 1.15 (1.61, 0.00 - 6.75) 4.09 (1.50, -1.51 - 5.23) 38 7.94 (3.78, 0.00 - 25.69) 12.29 (3.06, 6.97 - 15.56) 4.80 (0.12, 4.68 - 5.07) 1.47 (1.79, 0.00 - 5.89) 3.38 (1.83, -1.19 - 5.02) 41 10.01 (3.93, 0.00 - 26.69) 9.26 (1.37, 6.54 - 15.31) 4.78 (0.12, 4.64 - 4.99) 4.01 (1.36, 0.45 - 6.29) 0.74 (1.27, -1.51 - 4.33) 142 Table A2 (continued). I D Depth (m) Temp (°C) ݀݁ (°C) ܾ݀ (°C) ‫ܧ‬ 43 13.48 (10.21, 0.00 - 53.43) 8.85 (4.09, 4.73 - 21.46) 4.27 (0.21, 3.24 - 4.52) 3.13 (2.92, 0.00 - 7.36) 1.07 (2.79, -3.64 - 4.40) 44 8.23 (4.08, 0.00 - 36.55) 10.32 (3.87, 5.88 - 15.64) 4.69 (0.11, 4.61 - 4.98) 3.06 (2.79, 0.12 - 7.04) 1.79 (2.60, -2.28 - 4.50) 47 6.60 (3.56, 0.00 - 22.01) 11.92 (5.22, 7.29 - 16.81) 4.77 (0.20, 4.58 - 5.11) 2.48 (2.46, 0.13 - 5.89) 2.30 (2.51, -1.24 - 4.74) 50 8.69 (4.69, 0.00 - 29.46) 10.28 (4.05, 5.02 - 15.73) 4.81 (0.11, 4.73 - 5.17) 3.43 (3.52, 0.00 - 8.30) 1.61 (3.60, -3.32 - 4.78) 54 7.39 (3.86, 0.00 - 20.76) 10.58 (2.73, 7.33 - 16.30) 5.76 (0.02, 5.53 - 5.79) 4.06 (2.56, 0.02 - 7.05) 1.68 (2.56, -1.37 - 5.76) 58 7.33 (3.52, 0.00 - 26.18) 11.03 (2.61, 5.94 - 15.90) 5.08 (0.16, 4.95 - 5.20) 1.99 (1.55, 0.34 - 6.06) 3.16 (1.52, -1.09 - 4.75) 59 12.03 (4.08, 0.00 - 46.52) 9.09 (2.32, 5.51 - 16.12) 5.00 (0.32, 4.72 - 5.41) 4.41 (2.39, 0.01 - 7.75) 0.67 (2.60, -2.74 - 4.72) 63 8.73 (4.84, 0.00 - 35.47) 9.34 (2.32, 5.07 - 15.70) 5.03 (0.20, 4.79 - 5.24) 3.86 (2.21, 0.92 - 7.48) 1.24 (2.15, -2.58 - 4.22) 66 14.94 (4.31, 0.00 - 28.00) 10.67 (1.79, 5.76 - 15.71) 4.75 (0.10, 4.61 - 5.30) 2.75 (1.70, 0.00 - 7.30) 2.05 (1.57, -2.17 - 4.63) 69 10.37 (3.48, 0.00 - 40.27) 8.05 (1.06, 5.25 - 16.27) 5.08 (0.16, 4.89 - 5.36) 5.52 (0.98, 0.18 - 7.73) -0.44 (0.95, -2.83 - 4.90) 143 Table A3. Candidate Effectiveness of Behavioural Thermoregulation (E) model AIC scores and relevant calculated values. P is the number of parameters in the model, AIC is the AIC score, ΔAIC is the difference between the model AIC score and the top model AIC score, lik is the model likelihood, wAIC is the model AIC weight, and cwAIC is the cumulative AIC weight. GSE is the estimated gross somatic energy density and RIB is the relative infectious burden. Fixed Effects P AIC ΔAIC lik wAIC cwAIC Heat Stress + RIB + Mass + Day/Night Heat Stress + Mass + Day/Night Heat Stress + RIB + Mass Heat Stress + Mass Heat Stress + Day/Night Heat Stress Heat Stress + RIB + Mass + GSE + Day/Night Heat Stress + Mass + GSE + Day/Night Heat Stress + RIB + Mass + GSE Heat Stress + RIB + Day/Night Heat Stress + Mass + GSE Day/Night Heat Stress + GSE + Day/Night Heat Stress + RIB Intercept Heat Stress + GSE Heat Stress + Mass RIB + Day/Night Heat Stress + RIB + GSE + Day/Night Mass GSE + Day/Night RIB Heat Stress + RIB + GSE RIB + Mass + Day/Night GSE RIB + Mass Mass + GSE + Day/Night RIB + GSE + Day/Night Mass + GSE RIB + GSE RIB + Mass + GSE + Day/Night RIB + Mass + GSE 8 7 7 6 6 5 9 8 8 7 7 5 7 6 4 6 6 6 8 5 6 5 7 7 5 6 7 7 6 6 8 7 1282.15 1282.34 1282.76 1282.91 1283.12 1283.80 1283.82 1284.12 1284.43 1284.44 1284.68 1284.78 1285.11 1285.17 1285.38 1285.80 1286.07 1286.39 1286.43 1286.61 1286.78 1287.02 1287.17 1287.33 1287.38 1287.90 1288.03 1288.38 1288.58 1289.02 1289.28 1289.84 0.00 0.19 0.61 0.76 0.97 1.65 1.67 1.97 2.28 2.29 2.53 2.63 2.97 3.02 3.23 3.65 3.92 4.24 4.28 4.46 4.63 4.87 5.02 5.18 5.23 5.75 5.88 6.23 6.43 6.87 7.13 7.69 1.00 0.91 0.74 0.69 0.62 0.44 0.43 0.37 0.32 0.32 0.28 0.27 0.23 0.22 0.20 0.16 0.14 0.12 0.12 0.11 0.10 0.09 0.08 0.08 0.07 0.06 0.05 0.04 0.04 0.03 0.03 0.02 0.12 0.11 0.09 0.08 0.07 0.05 0.05 0.04 0.04 0.04 0.03 0.03 0.03 0.03 0.02 0.02 0.02 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.00 0.00 0.00 0.00 0.12 0.23 0.32 0.40 0.47 0.52 0.58 0.62 0.66 0.70 0.73 0.76 0.79 0.82 0.84 0.86 0.88 0.89 0.90 0.92 0.93 0.94 0.95 0.96 0.97 0.97 0.98 0.99 0.99 0.99 1.00 1.00